Fundamentals of Structural Geology provides a new working in related disciplines, including geophysics,
framework for the investigation of geological struc- rock mechanics, eld mapping, hydrogeology, petro-
tures by integrating eld mapping and mechanical leum and geotechnical engineering, and natural
analysis. It emphasizes the observational data, hazard mitigation. The book is supported by a
modern mapping technology, principles of contin- website (www.cambridge.org/ 0521839270) hosting
uum mechanics, and the mathematical and compu- images from the book, additional colour images,
tational skills, necessary to map, describe, model, student exercises and MATLAB scripts. Solutions to
and explain deformation in the Earths lithosphere the exercises are available to instructors.
quantitatively.
Assuming a basic knowledge of physical geology, dav i d p o l l a r d is the Morris Professor of Earth
introductory calculus, and physics, this advanced Sciences in the Department of Geological and
textbook builds on more traditional courses that Environmental Sciences at Stanford University
emphasize descriptive terminology, geometric tech- where he co-directs the program in Structural
niques, and kinematics. In a signicant departure Geology and Geomechanics. He and his students are
from conventional textbooks on the subject, differ- using quantitative eld data and principles of struc-
ential geometry is introduced and applied to quan- tural geology, combined with laboratory and com-
tify descriptions of geological structures. Differential puter modeling, to address questions about processes
geometry integrates the spatial information conven- of faulting, fracturing, and rock deformation. The
tionally found on maps with orientation data from research aims to understand how faults and fractures
stereograms to provide reproducible descriptions of evolve in the Earths crust; how they affect the ow
geological structures. By starting from the funda- of magma, groundwater, and hydrocarbons; and
mental conservation laws of mass and momentum, what role fractures play in earthquake generation
the constitutive laws of material behavior, and the and volcanic eruption
kinematic relationships for strain and rate of defor-
mation, the authors demonstrate the relevance of r a y m o n d f l e t c h e r is a Research Professor in
solid and uid mechanics to structural geology. the Department of Geosciences at the Pennsylvania
The constitutive relations used in the book are State University. He and his collaborators study the
sufciently elementary to enable students to gain continuous deformation of rock as in the emplace-
physical insight from analytical solutions, but are ment of mantled gneiss domes, rock folding, and
adequately realistic to provide compelling correla- basin and range necking. He also works on processes
tions to observational data. linking chemical aspects of mineral growth or disso-
This book offers a modern quantitative approach lution in rocks and deformation. Currently he is
to structural geology for advanced undergraduate studying folding near the base of ice sheets, and the
and graduate students and researchers in structural evolution of structures and rheological behavior of
geology and tectonics. It will also interest those composite rock masses.
Fundamentals of
Structural Geology
David D. Pollard
Stanford University
and
Raymond C. Fletcher
The Pennsylvania State University
University Printing House, Cambridge cb2 8bs, United Kingdom
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521839273
D. D. Pollard and R. C. Fletcher 2005
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2005
4th printing 2010
A catalogue record for this publication is available from the British Library
isbn 978-0-521-83927-3 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain, accurate
orappropriate.
Contents
Preface page ix
Acknowledgments xi
References 478
Index 497
Preface
Fundamentals of Structural Geology is a textbook that and a physics course covering mechanics and
emphasizes modern techniques of eld data heat. We consider these courses to be the essential
acquisition and analysis, the principles of contin- mathematical and scientic pre-requisites for a
uum mechanics, and the mathematical and com- course using this textbook. Elementary concepts
putational skills necessary to describe, model, of vector analysis, matrix theory, linear algebra,
and explain quantitatively the deformation of ordinary and partial differential equations, and
rock in Earths lithosphere. computer programming with Matlab are used
With precise location data now available from throughout, but are introduced in such a way that
the Global Positioning System (GPS) and powerful a formal course in these subjects, while helpful,
computer systems now transportable in a back- should not be considered a pre-requisite. The
pack, the quantity of reproducible eld data has authors view this textbook as appropriate for a
increased dramatically. These new data sets rst course in structural geology, but recognize
demand better methods for describing the geom- that many students will come to a course using
etry of structures, and we address this demand by this book after a traditional course that empha-
introducing the basic concepts of differential sizes the descriptive terminology, geometric tech-
geometry, which provide unambiguous descrip- niques, and kinematic concepts of the discipline.
tions of curved lineations and surfaces in three Although designed as a text for students, this
dimensions. Data sets from a variety of eld areas book also should be useful as a reference for
are provided via the textbook website to promote researchers in structural geology, and as an aid for
the practice of opening eld notebooks to the updating instructors and professionals who have
entire community of researchers, and as input for been exposed only to traditional courses and text-
student exercises (see below). books on the subject. Furthermore, this book
Textbooks in structural geology provide el- should be attractive to scientists in related dis-
ements of continuum mechanics (e.g. separate ciplines (geophysics, rock mechanics, tectonics,
chapters on stress and strain), but rarely are these geotechnical engineering, and petroleum engi-
concepts tied together with constitutive laws or neering) who are looking for a modern summary of
formulated into equations of motion or equilib- the fundamentals of structural geology. We encour-
rium to solve boundary or initial value problems. age students and professionals from these disci-
These textbooks largely beg the questions: what plines to learn about the modern methods and
methodology should one adopt to solve the prob- tools of structural geology so that they can effec-
lems of structural geology; and what are the fun- tively interact with geologists on multi-disciplinary
damental constructs that must be acknowledged projects.
and honored? These constructs are the conserva- One of the opportunities and challenges of
tion laws of mass, momentum, and energy, com- publishing a textbook in the twenty-rst century
bined with the constitutive laws for material is the fact that the printed volume is no longer the
behavior and the kinematic relationships for only vehicle for communication between authors
strain and rate of deformation. We use these con- and readers. Accordingly, we have prepared a
structs to build a rational methodology for the homepage for Fundamentals of Structural Geology
investigation of tectonic processes and their struc- that is available on the World Wide Web (www.
tural products. cambridge.org/0521839270) and provides the fol-
This textbook is designed for senior under- lowing supplementary materials for readers,
graduate students and graduate students who instructors, and students:
have taken an introductory physical geology
course, mathematics courses that include differ- Full color images for all outcrop photographs
ential and integral calculus in several variables, used in the text
x PREFACE
Full color images for key graphical results used ization environment and speed, and professional
in the text programmers have written applications such as
Supplementary outcrop photographs, maps, Matlab that provide most of the computational
and cross sections tools needed by structural geologists.
A repository for supplementary images con- For the authors of this textbook, it is not
tributed by readers sufcient to focus on understanding the struc-
Exercises for students that reinforce the con- tural history of the Earth as an arcane academic
cepts introduced in the text exercise. We believe that structural geologists
Data sets from eld mapping campaigns for use can make important contributions in natural
in the exercises resource recovery (including water, oil, gas, and
Solutions to the exercises for instructors with minerals), in the assessment of natural hazards
password protection (including earthquakes, landslides, and volcanic
Sample Matlab m-les for the exercises eruptions), and in the management of the en-
Sample Matlab m-les for recreation of graph- vironment (for example the long-term storage of
ical gures found in the text radioactive materials and the contamination of
A repository for exercises and Matlab m-les fractured aquifers by hazardous chemicals). It is
contributed by readers the authors hope that students and instructors
Errata alike will be as captivated as we have been by the
remarkable opportunities and challenges of struc-
With a laptop connected to the Web and an LCD tural geology. Great satisfaction in the practice of
projector instructors can use the color outcrop this science is achieved when one successfully
images in the classroom to illustrate geological brings together the beauty of the natural world
concepts, and run the m-les with their own and the physical world of continuum mechanics
choice of parameters for a dynamic demonstra- to achieve a better understanding of rock defor-
tion of the mechanical concepts. We envision mation and the development of structures. By
readers of the textbook having this website open doing so one contributes to the knowledge of
on their desktop to enhance their learning experi- Earths remarkable history and to the solution of
ence. Today desktop PCs provide the necessary important practical problems facing society
CPU power, 3D graphics cards provide the visual- today.
Acknowledgments
David Pollard would like to acknowledge four Taixu Bai, Laurent Maerten, Scott S. Young, Frantz
teachers who shaped his understanding of struc- Maerten, Stephan Bergbauer, Peter Eichhubl,
tural geology as an undergraduate and graduate PhillipG.Resor,KurtR.Sternlof,PatriciaE.Fiore,Ian
student. Donald B. McIntyre of Pomona College W. Mynatt, W. Ashley Grifth, Nicolas Bellahsen,
provided the spark that ignited his curiosity about Gaurav Chopra, and J. Ole Kaven.
the subject and put it in an historical context. David Pollard would like to thank John Suppe
Arvid M. Johnson of Stanford University intro- of Princeton University and Patience A. Cowie of
duced him to the tools of mechanics and to a ratio- Edinburgh University for hosting sabbaticals that
nal way to approach physical processes in the eld provided important time for development of the
and laboratory. John G. Ramsay of Imperial College materials presented here. He gratefully acknowl-
taught him how to measure deformation in edges the help of the staff of the Department of
outcrop and investigate the geometry and kin- Geological and Environmental Sciences and the
ematics of rock subject to ductile deformation. Branner Earth Sciences Library at Stanford Uni-
Neville J. Price of Imperial College introduced him versity. Also, he extends special thanks to the
to rock mechanics and the analysis of rock subject Seeley G. Mudd Science Library at Pomona College
to brittle deformation. These teachers provided a and to the National Cello Institute for providing an
diversity of viewpoints of structural geology that idyllic venue for preparation of the manuscript.
was fascinating as well as challenging, and the Raymond Fletcher would like to acknowledge
origins of many of the themes played out in this several people who contributed to his education
textbook can be traced directly to their classrooms. as a structural geologist. William F. Brace (MIT)
Arvid Johnsons role in the formative stages of awarded him a C in the undergraduate structural
work on the textbook was particularly important. geology course, giving useful incentive for further
David Pollard was privileged to study with stu- study of a subject that Bills treatment showed to
dents who were colleagues at Pomona College, consist of an intriguing combination of eld obser-
Stanford University, and Imperial College, and later vation and mechanical analysis. Bill Brace also gave
to work with students in a teaching and advisory excellent advice on what not to do as a Ph.D.
capacity at the University of Rochester, the US research project prior to the arrival at Brown
Geological Survey (Menlo Park), and Stanford University of his Ph.D. advisor William M. Chapple.
University. Many of these students have partici- Bill Chapple provided guidance in formulating a
pated in research that helped to shape the concepts tractable complete mechanical model for the
and methods described in this book. They include: emplacement of a gneiss dome and M. A. Jaswon
Atilla Aydin, Zeev Reches, Gary R. Holzhausen, John pointed him toward a method of analysis. Inter-
W. Cosgrove, Otto H. Muller, David R. Dockstader, action with Bill Chapple over many years contin-
Paul T. Delaney, Paul Segall, Jon H. Fink, J. Russell ued to enrich his experience. The foundation for
Dyer, Russell K. Davies, Laurie L. Erickson, Marie D. his understanding of continuum mechanics was
Jackson, Peter C. Wallmann, Stephen J. Martel, provided by the lucid presentation of this subject
Allan M. Rubin, Larry G. Mastin, Jon E. Olson, Sarah in a two-semester course at Brown University by
D. Saltzer, Scott S. Zeller, Andrew L. Thomas, Carl E. E. T. Onat. Arvid M. Johnson introduced him to the
Renshaw, Roland Brgmann, Pauline M. Mollema, disciplined mapping of small-scale structures in
Marco Antonellini, Haiqing Wu, Peter P. the eld interspersed with more freewheeling dis-
Christiansen,Stephen K. Matthi, JoshuaJ.Roering, cussions of mechanical modeling. Memorable dis-
J. Ramn Arrowsmith, George Hilley, Emanuel J. M. cussions over coffee and pastry with Bernard Hallet
Willemse, Michele L. Cooke, Elissa Koenig, Juliet G. continue to provide him with imaginative ideas,
Crider, W. Lansing Taylor, Simon A. Kattenhorn, such as treating the Basin-and-Range Province as a
xii ACKNOWLEDGMENTS
string of blood sausages. He has beneted from George Gazonas, Bill Kilsdonk, Frank Irwin,
and enjoyed collaborations with former graduate Duncan Mardon, and Tom Patton.
students Judi Chester, Russell Davies, Jon Fink,
Chapter 1
Mt. Hillers, southern Henry Mountains, UT. The mountain is The sciences do not try to explain, they hardly even try
cored by igneous rock and surrounded by upturned beds of to interpret, they mainly make models. By a model is
sandstone and shale. G. K. Gilbert coined the term meant a mathematical construct which, with the addi-
laccolite for these structures in the late 1870s and tion of certain verbal interpretations, describes
proposed models for this process of mountain building based observed phenomena. The justication of such a math-
on mechanical principles. Inset: Frontispiece from G. K. ematical construct is solely and precisely that it is
Gilberts Report on the Geology of the Henry Mountains expected to work (quote from John von Neumann;
(Gilbert, 1877). To the rear of this illustration the Gleick, 1987, p. 273).
sedimentary strata form the structural dome of
Mt. Ellsworth, and to the front the eroded remnant of the
dome represents the current topography of this mountain.
Photograph by D. D. Pollard.
2 MOTIVATIONS AND OPPORTUNITIES
I
n this chapter we motivate the study of struc- President of the Geological Society of Americas
tural geology by introducing selected topics Structural Geology and Tectonic Division, con-
that illustrate the extraordinary breadth of cluded:
interesting problems and important practical We are at the end of the era when an unquestioning
applications of this discipline. For example, we public belief in the benets of basic scientic research
use the Imperial Valley earthquake of 1979 along almost automatically led to increased budgets at the
the San Andreas Fault zone to describe tech- NSF (National Science Foundation) Program level.
niques for geological hazard analysis. In a second Already, NSF management and the Congress want to
example the lineaments visible in radar images hear arguments about how research, and especially
of Venus provide the data for investigating tec- new programs, will address important social issues:
tonic processes on a planet other than our own. environmental changes and hazards, exploitation,
This is followed by an investigation of normal waste, and recycling of natural resources, and the like
(Cowan, 1992).
faulting in a hydrocarbon reservoir under the
North Sea, off the coast of Norway, to introduce Thus, whether a career in the Earth sciences takes
an application to petroleum exploration and one to industry or to academia or to a government
production. Then we describe the pattern of laboratory, the structural geologist should know
small faults, veins, and solution surfaces from how to address problems of social importance. To
an exposure in southern France, an example that this end, we integrate aspects of active tectonics,
demonstrates the practice of structural geology engineering geology, and petroleum geology into
at the human scale. The concept of anticracks this book to show how structural geology can
that emerged from this academic investigation contribute to solving problems in these areas.
is now being used to help explain the origin of Most inhabitants of southern California are
huge earthquakes a hundred kilometers below familiar with earthquakes and the geological
Earths surface. Finally, we describe a mecha- hazard associated with living in an active tectonic
nism for mountain building that was discovered province, although the recurrence time of major
in the Henry Mountains of southern Utah in the events is great enough to instill a sense of com-
late nineteenth century by one of the pioneers of placency in many citizens. On the other hand,
structural geology, G. K. Gilbert. Earth scientists and government ofcials are
The frontispiece for this chapter is a photo- acutely aware that destructive earthquakes could
graph of Mt. Hillers in the southern Henry occur at any moment. Teams of scientists and
Mountains. Like all the photographs that appear engineers supported by federal and local govern-
as grayscale images in this book, a color image ments are monitoring the continuing activity of
of this photograph is available at the textbook the faults in this area and have tools in place to
website along with images of related exposures capture data from the next signicant event (Yeats
and scenes. These are presented as monitor et al., 1997).
resolution images for quick viewing with a web What are the data that these scientists and
browser or for LCD projection in the classroom for engineers are hoping to capture? Perhaps the
teaching purposes. most fundamental aspect of faulting is the fact
that the rock and soil on either side of the fault
slip past one another. There is relative motion of
1.1 Earthquake hazards in these two masses more or less parallel to the fault
surface. For example, Fig. 1.1 is a photograph
southern California taken across the trace of the Imperial Fault in the
Imperial Valley of southern California shortly
Academic researchers have learned that society after a magnitude 6.5 earthquake struck on
may not be content to continue funding the October 15, 1979. The vertical surface just behind
arcane studies of ancient rocks that have been the the observers feet is one surface of the fault
mainstay of the National Science Foundations exposed at the time of the earthquake. Relative to
Tectonics Program in the past. Darrel Cowan, then the ground on which the observer is standing, slip
1.1 EARTHQUAKE HAZARDS IN SOUTHERN CALIFORNIA 3
EA
fault to the location of the seismographs.
ST
The use of interferometric synthetic aperture
radar (InSAR) for the detection of ground dis- 111
HIG
highlighted in articles appearing in the early
HLI
1990s (Massonnet et al., 1993; Prescott, 1993;
NE
Zebker et al., 1994). The radar signal is transmitted 33o 15'
78
from a satellite to the ground surface where it is
reected back to the satellite and recorded as a 111
Brawley
set of pixels making up an image of the surface. Fault
Knowledge of the travel time and speed of the zone
CANAL
Im
0 km 10
pe
signal provide the information necessary to cal-
ria
l
culate the range, or distance, from the satellite
to each reective site on the surface. If the same 8
Fa
111
example before and after the earthquake, the dif-
u
lt
ference between the two images can be used to cal-
ATES
culate the component of the surface displacement UNITED ST
MEXICO Epicenter
directed toward the satellite. The resulting image 115o 45' 10/15/79
115o 30'
(Fig. 1.2e), called an interferogram, is similar to a
contour map of the displacement component on
Fig 1.3 Map of the region affected by the October 15,
which the white and black bands (called fringes) 1979, earthquake in southern California (Wosser et al.,
are the contours. The fault segments are shown as 1982). The epicenter is shown as a star in the lower right-
ne white lines superimposed on this image. By hand corner.
invoking a model (usually based on elasticity
theory) for the location and geometry of the fault
segments and the mechanical behavior of the the concepts and contributions from geophysics
rock mass, one may use this displacement dis- and geodesy to the study of faulting. In addition
tribution on Earths surface to calculate the important insights are attained from studying the
corresponding slip distribution on the fault. The effects of faulting on the geomorphology of the
abundance of data provides considerable con- landscape (Arrowsmith et al., 1996; Arrowsmith
straint on the unknown slip distribution below et al., 1998). The most comprehensive view of
Earths surface and very exciting avenues for new faults and the faulting process will come from an
research on faulting. integration of all these data and that integration
It should be obvious from this discussion that will be most effective in the context of building
the different disciplines contribute information well-constrained models.
that is based on different observations in different
locations and over different length and time 1.1.2 Conceptual and mechanical models
scales. Yet scientists from all three disciplines are for the 1979 earthquake rupture
studying the same physical phenomenon, fault- On October 15, 1979, the magnitude 6.5 earth-
ing, and they are using the same tools to build quake rupture began just south of the USMexico
their models, namely elasticity theory. In this border and spread approximately 35 km to the
textbook we focus on the geological data and the north into southern California (Fig. 1.3), breaking
models that are used to relate measurements ground along the trace of the Imperial Fault
of slip to fault behavior. On the other hand each (Johnson et al., 1982; Wosser et al., 1982). Many
discipline is providing important pieces of the agricultural features such as fence lines and
puzzle, so structural geologists should be aware of canals provided markers to measure the slip
6 MOTIVATIONS AND OPPORTUNITIES
Map view B
A A'
B'
u
~3 km/ s1 FR FR
A
10 ?
Depth
QP QP
20
(km)
Fig 1.4 Three views of a crustal-scale strike slip fault. Map on faults? Some answers to this question have
view illustrates the fault as a zone of deformation. Cross
come from research by scientists and engineers
section AA in the fault plane includes a contour map of the
over the past few decades, but much remains to be
slip (u) which goes to zero at the fault tipline and is greatest
near the hypocenter (star). Cross section BB perpendicular
understood.
to the fault plane suggests that slip mechanisms are frictional In the previous section we described how geol-
resistance (FR) in the upper part of the crust and localized ogists, geodesists, and geophysicists use models to
quasi-plastic flow (QP) in the lower part. The graph at the extrapolate information on displacements or
right indicates a linearly increasing resistance to shearing with accelerations from the locations where data are
depth to the brittleductile transition, and then a non-linear measured on the Earths surface to the fault in the
decreasing resistance to shearing with depth. Reprinted from sub-surface. These models help us to understand
Sibson (1989) with permission from Elsevier.
the behavior of faults where they cannot be
observed directly and they provide insights
across the fault trace. The farmers, homeowners, concerning earthquake faulting as a structural
businesses, and municipalities in the Imperial process. The faulting process is conceptualized at
Valley, mostly around the town of El Centro, sus- the crustal scale in Fig. 1.4 for a vertical fault with
tained over twenty million dollars in damage. strike slip motion (Sibson, 1989). Each view of this
Fortunately, there was no loss of life and few cata- conceptual fault model reveals different aspects
strophic failures of man-made structures in this of faulting at the crustal scale. The map view
event. On the other hand, earthquakes of similar shows a zone of fractures and deformation, rather
magnitude often are accompanied by many than two surfaces in contact. This suggests that
deaths in regions with less stringent building faults can be more complex than a single fracture
codes, or no building codes at all. These events and that shearing of material in a fault zone may
testify to the destructive power of earthquakes characterize the deformation rather than slip
and to the need to understand such hazards. between two surfaces. The vertical cross section
Because earthquakes are generated by sudden slip viewed parallel to the fault indicates that fric-
on faults, we need to understand the mechanisms tional resistance (labeled FR in Fig. 1.4) to slip on
and behaviors of faults in order to develop a fault operates to depths of perhaps 10 km and
informed hazard mitigation policy. Just what are plastic ow (labeled QP) is associated with dis-
the causes and consequences of dynamic rupture tributed shearing in a zone at deeper levels. Thus,
1.1 EARTHQUAKE HAZARDS IN SOUTHERN CALIFORNIA 7
the mechanisms of faulting may change with (a) Map view of rupture
depth as temperature and pressure increase, such 0 10 20 30 km
that brittle fracture and friction dominate at Distance along rupture
shallow depths and ductile ow dominates at Epicenter
Imperial Fault
greater depths. In this conceptual model the resis-
tance to shearing increases with depth to this Br
aw
le
transition and then decreases with depth. In a ver- y
Fa
ul
tical section viewed perpendicular to the fault N t
(AA), dynamic shearing begins at depth, near
the brittleductile transition and spreads out over Vertical cross sections of rupture
(b) Rupture time (s)
the fault surface at a velocity of about 3 km s1, 0
12
eventually reaching the Earths surface.
Hypocenter
The Imperial Valley earthquake is noteworthy 5 10
The mechanical model for the October 15, these dots and the hypocenter the model fault has
1979, earthquake event considers only the slipped and stopped, whereas to the north and
rupture along the Imperial Fault (Archuleta, above the 8-s contour the fault has not yet slipped.
1984). The lower four panels of Fig. 1.5 are graphs At any particular location on the model fault the
of different physical quantities calculated using slipping occurred over a period of time ranging
the model and plotted on a vertical planar section from a fraction of a second to almost 2 s as the
that approximates the more complex geometry of rupture front passed, and then slipping stopped.
the actual fault as suggested by the mapped trace Figures 1.5d and e show, respectively, two com-
in the rst panel. The model fault is about 12 km ponents of slip between the model fault surfaces
in depth (ordinate) and 35 km in length (abscissa). after the rupture has completed its propagation
The physical quantities (rupture time, slip dura- from the hypocenter to the northern termination.
tion, strike slip, and dip slip) are represented by Strike slip varies from 1.4 m near the bottom
contours of equal magnitude on these graphs. center of the fault to a few decimeters or less at
Together these panels provide a remarkable visu- the surface. The strike slip is zero along the south-
alization of the model slip event from the Earths ern portion of the fault at the surface and this is
surface to the bottom of the rupture. consistent with the observations shown on the
Figure 1.5b illustrates the position of the map in the rst panel. Note that the surface mea-
leading edge of the model rupture to the north of surements of slip, amounting to about 20 cm,
the point of rupture initiation, the hypocenter, at under-represent the slip at depth by a factor of
times measured in seconds after initiation. What eight or more. The model fault slipped much
happened to the south of the hypocenter is more at depth than at the surface. Dip slip is con-
ignored on these panels. At a given time, say 4 s, centrated near the surface at the northern end of
that portion of the fault between the hypocenter the model fault with magnitudes approaching
and the 4-s contour has slipped, while elsewhere a few decimeters. This is consistent with the
on the fault no slip has occurred. Clearly, slip on geological observations (see Fig. 1.1) that indicate
the model fault does not initiate everywhere the rocks on the northeastern side of this part of
simultaneously. Rather, the model rupture initi- the Imperial Fault went down relative to those
ated at a point, at the hypocentral depth of about on the southwestern side. The amount of dip slip
8 km. Then, the rupture front advanced rapidly to at the surface (up to about 20 cm) also is consis-
the north and less rapidly upward toward the tent with the eld observations. The relative
surface. The rupture took a total time of about 12 s motion on the Brawley Fault was also nearly pure
to spread the 35 km to the north end of the model dip slip with the northwestern side down. In fact,
fault. Thus, the average rupture velocity was about the region between the Imperial and Brawley
3 km s1 toward the north, approximately the Faults is a topographic depression occupied by a
speed of seismic shear waves. (usually) dry lake-bed. This suggests that the rela-
Figure 1.5c shows the total time that originally tive motion experienced during the 1979 earth-
adjacent particles on the two surfaces of the quake is typical of the recent geological history of
model fault were in relative motion. For example, this fault system.
along the contour labeled 1.6 s the two surfaces This mechanical model gives us a picture
slipped for a total time of less than 2 s. You might of active faulting that is reasonably consistent
nd this surprising given the fact that the total with the available surcial and seismic data from
duration of faulting was about 12 s. Clearly all the 1979 event. It informs our intuition about
parts of the model fault were not slipping at the the physical process of faulting and provides a
same time. This is illustrated in the previous panel glimpse into possible behavior along the Imperial
by the pattern of dots next to the 8-s contour. Fault at depth. Building models such as this one
These dots cover the relatively small portion of and using these models to understand the process
the fault that has already slipped and is still in the of faulting is an exciting area of research in which
process of slipping at the moment that the structural geologists can participate (Segall and
rupture front lies along the 8-s contour. Between Pollard, 1980; Aydin and Schultz, 1990; Cowie and
1.2 RADAR LINEAMENTS ON VENUS 9
Graben formed
above dike
0.0
2 Normal fault 0.2
Depth (km)
0.0 0.4
4 0.2
g 0.4
0.4
0.2
6 Contours of
Magma-filled 0.4 horizontal
dike 0.2 0.0 stress
0.0
8
5 0.0
0 5
tlf
0.0
f Distance (km)
rlf
Fig 1.8 Schematic block diagram of a graben bounded by
two normal faults and underlain by a dike (Rubin and Pollard,
50 km 1988; Koenig and Pollard, 1998). The horizontal stress
component induced by opening of the dike is contoured on
the front view showing a stress shadow (negative,
Fig 1.7 Left-looking F-MIDR 15s214 radar image of the
compressive stress) to either side of the dike and a stress
southeast quadrant of the volcanic edifice centered at 15S,
concentration (positive, tensile stress) near the dike tips.
215E. Surface structures are identified (Koenig and Pollard,
1998) as graben (g), fractures (f), terminal lava flows (tlf ),
and radial lava flows (rlf ). The graben and fractures
Horizontal stretching is caused by tensile stresses
radiate from the volcanic center in the upper left corner of
that tend to pull the rock apart. Therefore, one
the image.
needs to determine if dike opening at depth could
induce tensile stresses near the Earths surface,
reected off the fault surface on one side of the where normal faults bound the graben. Such as-
graben, returning a bright lineament, the fault sociations are found in volcanic regions on Earth,
surface on the other side of the graben lies in a including those in Iceland and Afar.
radar shadow and forms a dark lineament. Thus The mechanical model of this phenomenon is
the surface topography accounts for the closely based on principles that are formulated into a set
spaced pairs of bright and dark lines on the image. of mathematical equations known as the theory of
Figure 1.8 illustrates the conceptual model for elasticity (Timoshenko and Goodier, 1970). This
the development of graben over igneous dikes. theory and the relevant equations are described in
The pairs of normal faults bounding graben on detail later in this textbook. For the moment you
Venus are interpreted as having formed because of only need to know that these equations can, for
the local horizontal stretching of the rock imme- example, be solved to determine the stress
diately over or ahead of vertical dikes. Because the distribution in the rock mass surrounding a dike.
dikes apparently propagated upward and outward This formulation is called a boundary value problem
from a central magma chamber located under the because one prescribes the stresses on the bound-
summit of the edice at (15 S, 215E), the graben aries of a body and the governing equations of elas-
form a radial pattern on the anks of this volcano. ticity theory are used to calculate the stresses in
A couple of questions come to mind when the interior. In this case one boundary represents
thinking about the origin of graben as described in Earths surface, which is free of stress, and the
the previous paragraph. Does the opening of a dike other boundaries represent the dike walls that are
actually lead to stretching at the surface? If it does, subjected to stresses equal to the outward-directed
why should two normal faults form to either side of pressure of the magma. The magma pressure
the dike instead of one immediately over the dike? pushes the dike walls apart and distorts the sur-
These questions have been addressed by studying rounding rock mass, thereby inducing a change in
the physical relationships between normal faults the stress distribution that is not easy to imagine
and dikes (Rubin and Pollard, 1988; Rubin, 1990). without the aid of elasticity theory.
1.3 FAULTING IN A NORTH SEA HYDROCARBON RESERVOIR 11
On the front face of the block diagram in Fig. 1.8 After many years of observing geology departments,
values of the horizontal stress are shown using a our experience indicates that they have adjusted to the
contour plot. This illustrates quantitatively how the scientic revolution of recent decades, but have largely
opening of a model dike changes the stress eld in ignored the technical, economic, and social changes
the surrounding rock. To either side of the dike neg- that inuence the practice of geology and thereby
have ignored the professional lives of their students.
ative (compressive) stresses are induced that push
inward on any small element of rock. These Structural geologists can make important contri-
induced compressive stresses would tend to prevent butions to natural resource recovery, including
normal faults from forming because they increase water, oil, gas, and minerals. They can play key
the frictional resistance to sliding. However, imme- roles in the management of the environment, for
diately over the dike, positive (tensile) stresses are example in the long-term storage of radioactive
induced. These tensile stresses stretch the rock and materials and the contamination of fractured
could contribute to the formation of normal faults. aquifers by hazardous chemicals.
The distribution of horizontal stress at the surface One of the most interesting new areas of
has two maxima located symmetrically about the research in structural geology related to natural
plane of the model dike. These two maxima corre- resources is the investigation of folds and faults
late with the two normal faults that develop to using data from seismic reection surveys
either side of the dike plane. In this way the model (Kattenhorn and Pollard, 2001). These data are
has successfully addressed the questions we asked gathered by the petroleum industry to image
about the relationship between dikes and graben. hydrocarbon reservoirs (Sheriff and Geldart,
Does the success of the model provide all one 1995). To carry out a reection survey in a sedi-
needs to know to interpret the radar lineaments mentary basin, seismic waves are generated by
on Venus? Certainly it is supportive of the concept impulsively striking the surface of the Earth.
that dikes can induce slip on normal faults, but it Depending upon the depth of investigation
does not prove that dikes actually exist beneath required, these waves are generated using explo-
the graben. Some of the paired radial bright and sive charges, mechanically driven vibrators, or
dark lines on the radar image at (15S, 215E) can simply a hand-held hammer. Some of the waves
be traced down the anks of the volcano where travel down into the Earth, reect off sedimentary
they merge into a single bright radar lineament layers or other structures, and travel back to the
(labeled f on Fig. 1.7). These single lineaments surface where they are recorded by a string of
are interpreted as open ssures that formed as portable seismographs laid out along the survey
dikes neared the surface (Koenig and Pollard, line. The depth to different reecting horizons
1998). This interpretation is consistent with the can be computed by identifying the two-way travel
presence of dark lobate regions, interpreted as lava time (down and back) for each reection, and by
ows, emerging from near the distal end of some knowing the velocity for acoustic waves in the
ssures and extending outward in a radial direc- rock. A series of impulses is generated along the
tion (labeled tlf on Fig. 1.7). These ows provide survey line and reections are recorded at each
compelling evidence that dikes underlie the radial seismograph. The abundant data help to reduce
fractures and graben, and are the conduits for the the uncertainty in identifying and locating the
escape of magma from beneath the volcano. reecting horizons using a variety of data pro-
cessing techniques.
The output of this processing is a seismic
1.3 Faulting in a North Sea reection cross section (seismic section) of the Earth
immediately under the survey line. For example,
hydrocarbon reservoir consider the northern part of the North Sea
(Fig. 1.9a) on the Norwegian Continental Shelf
R. N. Farvolden and J. A. Cherry (1991) wondered if (Maerten et al., 2000; Maerten et al., 2002). Figure
geology departments are preparing their students 1.9b is a seismic section taken from approximately
for the twenty-rst century: 150 km west of the Norwegian coast (Faerseth
12 MOTIVATIONS AND OPPORTUNITIES
l
l
l l l l l l l l l l
l
l
l
l
l
l
l l
l
l l
l
l l
l
l l l l l
l
l l l l l l
l
l
l l l
l
l l l l
l
l
l
l l l l l
l l l l
l
l l
l l
rm
l l
l l l
o
l
l
Platform
Platf
l
l l
l
l l
l l
l l
gG
l
l
l l l l
l
a
l
l l
l
l l
l
Hord
l
Vikin
l l
l
l l
l l
l l l
1550
age, but do not extend into the overlying
1600
1650 Cretaceous strata, they are interpreted as having
1700 formed during Jurassic extension of the basin. The
1750
west-dipping fault interpreted in Fig. 1.9b cuts the
1800
1850 top of the Brent and extends slightly above the
1900 base of the Cretaceous. It also offsets one of
1950
the east-dipping faults. Because the west-dipping
2000
faults in this region systematically offset the east-
dipping faults, Faerseth and his colleagues con-
Fig 1.9 (a) Regional map showing the location of Oseberg clude that the west-dipping faults are younger.
Syd Field on the Norwegian Continental Shelf. A combination of factors, some related to
(b) Interpreted seismic data from the northern part of the improved equipment for acquiring such data and
North Sea approximately 150km west of the Norwegian
others related to improved software and hardware
coast. Two-way travel time plotted versus distance along
for processing such data, have changed the
section. Interpreted horizons include the base of the
Cretaceous, the top of the Heather Formation, and the top
quality and resolution of seismic imaging tech-
of the Brent Group. Reprinted from Maerten et al. (2002) nology dramatically in the past few decades.
and Faerseth et al. (1997) with permission from Elsevier. These developments are largely driven by the need
for oil and gas companies to improve their explo-
ration strategies through better images of sedi-
et al., 1997). For offshore surveys the seismographs mentary basins, and to improve their production
are towed in long strings behind a ship and the strategies through better characterizations of par-
sources for the acoustic waves are implosions ticular reservoirs within these basins. The older
created by submerged devices called air guns. The technique is essentially two dimensional, provid-
seismic section plots position from west to east ing an image of the strata on a vertical cross
along the survey line (see kilometer scale) versus section along the survey line. By capturing numer-
two-way travel time measured in milliseconds ous closely spaced parallel lines in two orthogonal
(ms). For a typical wave velocity of 2 km s1, the directions, enough information is gathered to
bottom of the section at 2000 ms 2 s represents a interpolate the reecting horizons between these
depth of about 2 km. The processed data appear as two-dimensional surveys and produce a three-
a series of bright and dark stripes representing dimensional image of the horizons within a
1.3 FAULTING IN A NORTH SEA HYDROCARBON RESERVOIR 13
The inset on Fig. 1.10 illustrates in cross producing the hydrocarbons from this reservoir
section how the horizontal component of separa- (Maerten et al., 1999, 2000). For example, if part of
tion (sometimes called the heave) and the dip of the target reservoir is separated from a well by a
the normal fault are represented by the width of sealing fault those hydrocarbons beyond the fault
the black stripes on maps. Greater stripe widths cannot be produced and another well must be
(gray on the inset) imply greater horizontal separ- drilled. If the second well is drilled and it turns
ations of two points that originally were adjacent out that the compartments do have a uid con-
to one another at the base of the Brent Formation. nection because the faults are not continuous, or
Two faults with the same dip are displayed as do not extend as far as originally interpreted, the
stripes with lesser and greater widths (horizontal unnecessary expense can be signicant. Drilling
separations) corresponding to lesser and greater one additional well in these offshore settings can
dip separation. Two faults with the same dip sep- be a multi-million dollar proposition. Clearly the
aration are displayed as stripes with lesser and design of the production strategy is dependent
greater widths corresponding to greater and upon accurate interpretation of the fault geome-
lesser dips. In regions where the dips of all the try and sealing properties. Here we demonstrate
normal faults are about the same, the widths of how models of fault slip distributions developed
the stripes on the structure map are used to infer by structural geologists can be helpful in making
the relative magnitude of dip separations. such design decisions.
Similarly, if the dip does not vary along the length The two compartments of the Brent reservoir
of a fault, the changing width of the stripe is used in the Oseberg Syd Field are labeled in Fig. 1.10 as
to infer changes in the distribution of dip separa- Omega North and South. Both are bounded to the
tion. west and east by major faults, presumed to be
From information gathered by drilling and sealing. Data from well No. 8, which intersects the
recovering core samples, measuring a variety of Brent Formation in Omega North, and data from
indicators in the wells, and running ow and pres- well No. 10, which intersects the formation in
sure tests in the sub-surface, it is believed that the Omega South, suggest that there is a uid pres-
rocks in the fault zones shown in Fig. 1.10 are rel- sure difference between the northern and south-
atively impermeable to uid ow compared to the ern compartments. However, the interpreted fault
unfaulted portions of the sandstone reservoir. pattern does not dene a complete compartment.
Therefore these are referred to as sealing faults. A The two northwestsoutheast striking faults just
number of mechanisms for changing the perme- to the southwest of Well No. 8 do not form a com-
ability of rock in a fault zone have been identied plete barrier between Omega South and Omega
(Jones et al., 1998). For example, the crushing of North. Perhaps there are other explanations for
sandstone grains during shearing in the fault the pressure difference, but the hypothesis inves-
zone can ll the pores with fragments, thereby tigated here is that these two faults are actually
decreasing the porosity and the permeability linked and thereby separate the reservoir into two
(Antonellini and Aydin, 1994, 1995; Antonellini et compartments.
al., 1994). Also, it is possible for very ne-grained This hypothesis was tested using a mechanical
clays to be dragged or injected into the fault zone model that relates fault geometry, rock properties,
to form a seal (Aydin and Eyal, 2002). If the faults and tectonic loading to fault slip (Crider and
are all of the sealing type, then they can divide the Pollard, 1998; Maerten et al., 2000; Crider, 2001).
reservoir into a number of isolated compartments The modeling method subdivides faults into many
with little or no uid communications between small triangular elements, as seen for a non-planar
compartments. The map shown in Fig. 1.10 sug- fault with irregular tipline in Fig. 1.11, and these
gests that the Brent Formation could be divided elements approximate the three-dimensional
into many compartments separated by the sealing geometry. The model is based on elasticity theory
faults. and the numerical method used to solve the gov-
The presence of fault-sealed compartments erning equations is called the boundary element
has important implications for locating wells and method (Crouch and Stareld, 1983). Each element
1.3 FAULTING IN A NORTH SEA HYDROCARBON RESERVOIR 15
B C
0 100 m
1 km
Slip
C
B
C
B
60
y 2
Quadrant 2 Quadrant 1 40
compression tension
Rhomb cavity 1
20
Left step { x
0 0
y
Vein Solution
Fault segment surface -20
-1
Quadrant 3 Quadrant 4 -40
tension compression
-2 -60
-2 -1 0 1 2
Fig 1.14 Schematic illustration of a left-lateral fault with x
veins and solution surfaces emanating from the fault tips in an
Fig 1.15 Contour map of the normal stress acting parallel
antisymmetric pattern (Fletcher and Pollard, 1981). Pairs of
to a fault modeled as a surface of displacement discontinuity
arrows indicate stretching or shortening of the rock
with uniform left-lateral slip (Crouch and Starfield, 1983).
associated with slip on the fault. This deformation is
Note the pattern of tensile stress (positive) in the first and
accommodated by the formation of the veins and solution
third quadrants, and compressive stress (negative) in the
surfaces, respectively.
second and fourth quadrants. Also note that the stress
magnitudes increase toward the model fault tips. See website
in the groundwater, the material to either side of for color image.
the solution surface moved inward. Note how the
veins and solution surfaces are distributed into
the four quadrants dened by a coordinate system Laboratory investigations have shown that the
centered on the fault with the x-axis parallel to the mechanical response of rock to stress is, in part,
trace of the fault. The conceptual model consists of dependent upon the magnitude of the normal
a system of cracks and anticracks distributed stress. For example, modest positive values of the
about the small fault in what is called an anti- normal stress (on the order of 1 to 10 MPa) corre-
symmetric conguration. The presence of veins late with extension of the rock mass and the devel-
and solution surfaces with this particular anti- opment of opening cracks. As the normal stress
symmetric arrangement is diagnostic of left-lateral becomes more compressive, pressure solution
slip on the fault. If the veins are in the second and may become an active deformation mechanism in
fourth quadrants, and the solution surfaces are in rocks with soluble components. Although the
the rst and third quadrants, the arrangement is exact magnitudes of the tensile and compressive
diagnostic of right-lateral slip. stresses necessary to induce these structures is
Our conceptual model for the relative motion not well known, it is clear that tension is neces-
of the surfaces of the cracks and anticracks (Fig. sary to induce opening cracks and compression is
1.14) suggests that the limestone in the rst and necessary to induce solution surfaces.
third quadrants extended parallel to the fault and The mechanical model we employ consists of a
the limestone in the second and fourth quadrants single two-dimensional fault (Fig. 1.15) with left-
contracted parallel to the fault. How do these lateral relative motion (Pollard and Segall, 1987).
deformations relate to the state of stress? Why do The normal stress acting on planes perpendicular
these structures initiate near the fault termina- to the model fault is calculated on a grid of points,
tions and propagate outward into these quad- and these values are contoured to produce the
rants? Why do the structures stop propagating at gure. The contours appear symmetric about the
a short distance from the fault? These and other model fault, but note that the stress is, in fact, dis-
questions can be addressed with a mechanical continuous across the fault surfaces, having the
model for the state of stress near a fault. same magnitude but opposite sign for adjacent
1.4 ANTICRACKS IN SOUTHERN FRANCE 19
points on the two surfaces. The stress, as indicated near the terminations of the faults. This is the
by the values associated with each contour, is posi- region of greatest stress concentration, and there-
tive (tensile) in the rst and third quadrants and fore is the locality where secondary structures are
negative (compressive) in the second and fourth most likely to form. Finally, the stress distribution
quadrants. Furthermore there is a stress concentra- offers an explanation for the limited extent of the
tion at the terminations of the model fault: the secondary structures. The stress decreases away
values associated with the contours increase from the fault tips toward much lower values at a
toward the fault tip and the spacing between con- distance that scales with the length of the fault.
tours decreases. In fact the stress becomes so great This decrease in stress is consistent with the ter-
very near the model fault tips that the contours mination of the veins and solution surfaces at
merge into a pattern that is no longer distinguish- modest distances from the fault tips. Although
able at the scale of this gure. Thus we have omitted this two-dimensional model provides important
the contouring in a small region around each tip. insights, additional understanding of the faulting
Figure 1.15 is an example of a plot prepared process may be achieved using three-dimensional
using MATLAB, the computational and graphics models (Willemse et al., 1996; Willemse, 1997;
engine that we employ throughout this textbook. Martel and Boger, 1998).
The m-file used to compute the values of the stress The authors became intrigued by the struc-
component at the grid points and to prepare the tures at Les Matelles over twenty years ago and, in
contour plot is available at the textbook website. the course of investigating models for their for-
There, a color version of the contour plot is view- mation, conceived of the concept of anticracks. At
able along with contour plots of other stress com- that time we had no practical applications for this
ponents. This procedure is followed throughout concept in mind. Nor, to our knowledge, did the
the textbook where grayscale gures are used to exposure at Les Matelles gure signicantly in the
reduce printing costs and color versions are avail- solution of any problem relevant to society. For us
able at the website. this was an academic exercise, motivated by a
To relate the veins and solution surfaces at the strong (and inexplicable) urge to understand these
Les Matelles outcrop to the left-lateral faults we structures, and nothing more. However, explana-
have to recognize that the veins and solution sur- tions for certain features of very deep and large-
faces are secondary structures and the faults are the magnitude earthquakes now utilize the concept of
primary structures. In other words, the veins and anticracks (Green and Burnley, 1989). Also, the evo-
solution surfaces formed in response to the stress lution of fault zones in limestone through a
changes in the rock mass as slip developed on the complex sequence of vein development, solution
faults. We can correlate the symmetry of these sec- surface development, and slip on solution surfaces
ondary structures with the symmetry of the stress has been documented and interpreted using the
eld about the model fault (Fig. 1.15). The veins are anticrack concept (Willemse et al., 1996). In addi-
cracks that are pulled open by tensile stresses and tion, the development of compaction bands in
therefore are associated with the eld of tensile porous sandstone has been explained using the
stress in quadrants 1 and 3. Conversely, the solu- anticrack concept (Mollema and Antonellini,
tion surfaces form in response to elevated com- 1996). These tabular zones of localized compaction
pressive stresses and therefore are associated with appear to propagate as anticracks in response to
the eld of compressive stress in quadrants 2 and 4. elevated compression, and they also have a
The correlation between the model and the signicant effect on the permeability of reservoirs
exposure observations is supported by the fact and aquifers in porous sandstone (Sternlof et al.,
that the veins and solution surfaces do not cross 2004). Perhaps the time and the taxpayers dollars
the fault surfaces where the model indicates a spent working on the seemingly arcane concept of
discontinuity (change in sign) of the stress. anticracks at the Earthquake Studies Branch of the
Furthermore, the model provides an explanation US Geological Survey can be justied in light of
for the initiation of these secondary structures these applications.
20 MOTIVATIONS AND OPPORTUNITIES
force to overcome the resistance to faulting. This correct according to this model for laccolith for-
does not preclude the lateral growth of a sill, but mation.
does preclude the development of the laccolith. The conceptual and mechanical models for-
From this relationship Gilbert inferred that short mulated by Gilbert for laccolith formation pro-
sills would be incapable of producing a structural vided him with a linear relationship between the
dome of the kind observed in the Henry diameter of laccoliths and their depth of burial.
Mountains. On the other hand if the magma were He devoted much of his time in the eld to gath-
able to spread far enough laterally as a sill, the ering data (Table 1.1) on the horizontal dimen-
driving force would equal the resisting force and, sions of laccoliths and their stratigraphic
in the next increment of growth, the fault would positions in order to test this relationship
develop and the overburden would begin to dis- (Gilbert, 1877, p. 86). Although the diameters
place upward to form a laccolith. given in this table are uncertain because of incom-
Solving the equilibrium equation for the plete exposure of the larger laccoliths in the
radius of the piston, Gilbert found: Henry Mountains, Gilbert came to the conclusion
that these data were consistent with the relation-
2dS
a (1.2) ship he had derived. None of the diameters of lac-
P m Pw
coliths from the upper zone exceed the diameters
He noted the linear relationship between piston of those from the lower zone. For its simplicity
radius, a, and depth of overburden, d, and inferred and the insight gained from it, this model is
that, for a given (positive) driving pressure, Pm remarkably successful.
Pw, and fault shear strength, S, magma must Many questions about laccolith formation
spread to a greater radius at a greater depth in cannot be addressed with Gilberts model. What
order for the overburden to be pushed upward are the feeder conduits? How does the sill propa-
along the fault. The second conjecture made gate laterally from this feeder to a diameter where
above (laccoliths with greater diameters should be the laccolith can form? At what rate did the
found lower in the stratigraphic sequence) is magma ow in these conduits? How might one
24 MOTIVATIONS AND OPPORTUNITIES
account for the resistance to bending of the strata tonic processes and their products over relevant
over the laccolith? How did the temperature vary length and time scales by combining observational
as the hot magma invaded the cold sedimentary data with both conceptual and mechanical
strata? We develop other models in later chapters models. For each of the examples given we
to address some of these questions. described the phenomenon and then offered a
Because the bottom contacts of the larger simple conceptual model that illustrates the key
igneous intrusions are not exposed, other con- elements of the physical process. Then we
ceptual models for the development of the struc- described a mechanical model used to understand
tural domes could be viable. Based on geological the physical process. A second objective for this
mapping of the Henry Mountain range and sur- chapter was to illustrate reductionism, a scientic
rounding plateau country, Hunt (1953) proposed methodology in which complex processes are
that each of the major mountain peaks is under- broken up into their (relative few) fundamental
lain by a cylindrical stock of igneous rock that elements in order to understand each element in
extends well below the bottom contacts that were isolation. Then, the elements are put back together
inferred by Gilbert. This hypothesis has been chal- to understand how the whole system operates. This
lenged, based on more recent geological mapping methodology is espoused throughout the book.
in the southern Henry Mountains and cross sec- Some of the topics in the introduction satisfy
tions that were constructed from new structural our objective of demonstrating the possible roles
data (Jackson and Pollard, 1988). The two hypothe- a structural geologist can take in society. For
ses for the shape of the magma chamber, stock or example, career opportunities may be available
laccolith, were debated in the literature and a undertaking geological hazard assessment with
variety of geological and geophysical evidence the US Geological Survey (USGS), being a member
brought to bear on the subject (Hunt, 1988b). of a production analysis team for an international
oil company, or participating in planetary explo-
ration with the National Aeronautics and Space
1.6 Concluding remarks Administration (NASA). We hope that these exam-
ples motivate students to learn more about the
One of our objectives for this chapter was to intro- fundamentals of structural geology by reading
duce a strategy for comprehending complex tec- this textbook.
Chapter 2
Structural geologists use a GPS receiver to determine the formed. Thus order in the sky became chaos. Random.
UTM coordinates of the point under their feet on this But even then . . . Even then, what Coyote did was evil,
outcrop of Aztec sandstone in the Valley of Fire, NV. but was there not a pattern, too, in the evil deed?
Photograph by D. D. Pollard. That had not been the time in Leaphorns life when
he had patience for the old metaphysics. He remem-
bered telling Haskie Jim about modern astronomy and
You think these raindrops are random? his uncle had the cosmic mechanics of gravity and velocity.
asked. And Leaphorn had been surprised. Hed said of Leaphorn had said something like Even so, you
course they were random. Didnt his uncle think they couldnt expect to nd anything except randomness
were random? in the way the rain fell. And Haskie Jim had watched
The stars, Haskie Jim said. We have a legend about the rain awhile, silently. And then he had said, and Joe
how First Man and First Woman, over by Huerfano Leaphorn still remembered not just the words but the
Mesa, had the stars in their blanket and were placing old mans face when he said them: I think from where
them carefully in the sky. And then Coyote grabbed we stand the rain seems random. If we could stand
the blanket and whirled it around and ung them into somewhere else, we would see the order in it.
the darkness and that is how the Milky Way was (Hillerman, 1990, pp. 21314.)
26 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
U
nlike the laboratory-bound physicist, vectors are dened with reference to a particular
structural geologists usually make mea- coordinate system. UTM coordinates may not be
surements and gather data in the eld, the best choice during a mapping campaign, and
where, because of poor exposures and the limited usually they are not the most convenient for mod-
view that erosion provides, structures may seem eling and data analysis. To change from one coor-
chaotically arranged. Yet, when each measure- dinate system to another one uses transformation
ment is tied to a particular geographic location equations, which make use of vector concepts and
and represented on a map, and when sets of mea- operations.
surements are organized by orientation and Because much of the structural information
plotted on an appropriate graph, systematic rela- recorded at exposures can be reduced to the ori-
tionships and patterns emerge that can be inter- entations in space of planar or linear elements, we
preted. In making such interpretations we have introduce the techniques to measure, record, and
achieved what Haskie Jim was alluding to, we have analyze these orientations. Special projections are
found where to stand to view the structures to used to depict a set of orientations on a at piece
bring order to what otherwise appears to be a of paper a common one being the stereographic
random phenomenon. projection. Here, the basic procedures for plotting
In pursuit of this objective we begin this orientation data on a stereogram are described in
chapter by nding our bearings in terms of the such a way that they can be implemented on a
coordinate systems commonly employed to locate computer. Combinations of structural maps and
outcrops in the eld and to construct maps of stereograms are used to visualize, respectively, the
geological structures. This includes an introduc- spatial distributions and the orientations of geo-
tion to the geographic coordinate systems used logical structures. Examples are provided to illus-
for the Global Positioning System (GPS), which is trate how this can be done.
becoming the standard tool for structural Finally we describe a modern mapping cam-
mapping (see frontispiece to Chapter 2). Also, we paign that utilizes GPS technology to create a
introduce the Universal Transverse Mercator precise structure contour map in a region where
(UTM) projection, used to project geographic four sets of faults intersect to disrupt and fold the
information from the curved surface of the Earth surrounding strata. With new technology the
onto a at piece of paper. Structural geologists mapping was accomplished in a fraction of the
should be capable of measuring the location of time required using traditional methods, with
key exposures where eld data are taken, and much greater control on the shapes of the
then constructing the map projections that deformed strata and on the fault offsets and slip
record the spatial distributions of these data. directions because of the abundance of quantita-
Structural maps should convey the geometry of tive data. The GPS technology is rapidly evolving
structures to other geologists in a form that is so we anticipate a new generation of structural
quantitatively precise and readily visualized and maps that will provide the impetus for more
analyzed. specic modeling of tectonic processes and the
A position vector uniquely determines the data better to constrain those models.
location of every exposure and every point at
which structural data are collected in the eld.
Vectors have many uses in structural geology, 2.1 Geographic coordinates and
from locating an exposure, to describing the
shape of folded strata, to representing quantities
map projections
such as displacement and velocity in the physical
laws that underlie the modeling of tectonic 2.1.1 Geographic coordinates: the
processes. The position vector is introduced here Global Positioning System (GPS)
along with some of the vector concepts and nota- The rst order of business for any mapping project
tions useful for structural mapping. Position is to locate oneself on Earths surface, either to
2.1 GEOGRAPHIC COORDINATES AND MAP PROJECTIONS 27
Semi-minor axis
(not very precise). Center
The convention in geodesy is to use a mathe-
matical model for the Earths shape that is an
oblate ellipsoid: a three-dimensional surface b
formed by rotating an ellipse about its minor axis.
If the semi-major and semi-minor axes of the
ellipse are called a and b, respectively, the com-
pression or attening of the ellipsoid at the poles
a
is dened as f (a b)/a. The rotation of the Earth
does cause some extension of the equatorial Semi-major axis
radius and some attening at the poles so a b,
and the semi-major axis is the radius of the equa- (b) WGS-84
torial circle of the ellipsoid. A standard model ellipsoid datum
used by geodesists is called the Geodetic Reference
System 1980 ellipsoid, or GRS-80 for short 3.6 m
(Hofmann-Wellenhof et al., 1997, p. 293). The Earth's
center of GRS-80 is located at the true center of the surface
Earth and the semi-minor axis is parallel to the
Earths axis of rotation (Fig. 2.2a). The length of +28.9 m
the semi-minor axis is b 6 356 752.3 m, and the
length of the semi-major axis, extending from the
center of the Earth to the equator, is a
6 378 137.0 m. Note that this ellipsoid is attened
only by 21 384.7 m (about 21 km) so f 0.003 352 8. Sea level
In other words the ellipsoid is only about 0.3% dif- Geoid datum
ferent than a perfect sphere. An early ellipsoidal
Different elevations of the
model for Earth was calculated in 1830 by Everest,
geology building at Stanford University
working in India, where he estimated a
6 377 276 m and b 6 356 075 m, so the attening Fig 2.2 Cross section of Earth. (a) Mathematical model
is f 0.003 324 4 (Richardus and Adler, 1972, p. 23). ellipsoid and physically defined geoid. Local gravitational
Different ellipsoids are used in different regions acceleration vector, g, defines down. (b) Elevations of
because they minimize discrepancies with the Geology Building at Stanford University relative to different
local mean sea level. datums.
To put the degreeminutesecond measures of
distance in perspective we use the semi-major axis
These odd values and the awkwardness of con-
of GRS-80 as the radius, R, of the equatorial circle,
verting degreesminutesseconds to standard dis-
and calculate the circumference as C 2R
tance units make this geographic coordinate
40 075 016.7 m. The following relationships are
system less than ideal, but it is conventional to use
found for distances along this equator:
it and conventions of this long standing (perhaps
1 degree 111 319.5 m three to four thousand years) are difcult to
1 minute 1 855.3 m abandon.
1 second 30.9 m
2.1 GEOGRAPHIC COORDINATES AND MAP PROJECTIONS 29
The third coordinate in the geographic system System datum for 1984 (WGS-84) and this is almost
is approximately parallel to a radial line from identical to the ellipsoid called GRS-80. However,
Earths center and is referred to as elevation. this ellipsoid may be separated from the mean sea
Often, elevation is referenced to a sea level datum, level geoid by a considerable distance. For
that is mean sea level as determined by tidal example, the WGS-84 ellipsoid is about 32.5 m
gauges. This is not an unreasonable choice, but it above the mean sea level geoid at Stanford
is subject to perturbations caused by water tem- University in central California (Fig. 2.2b). Thus,
perature variations and currents, and it is limited the reported elevation relative to the WGS-84
in applications to the coastline. So-called leveling datum for a location just outside the Geology
surveys are performed to determine elevations Building at Stanford University using a high-pre-
away from the coast as heights above (or below) cision GPS receiver is 3.6 m, but the building
another datum called the geoid. The geoid is a clearly is not below sea level! The elevation rela-
physically dened surface (Fig. 2.2a) that is every- tive to the mean sea level geoid is about 28.9 m.
where perpendicular to the local direction of the This difference is not due to errors in measure-
acceleration of gravity, g, so it is everywhere per- ment, but rather to use of the different datums.
pendicular to the direction we call down from A metric coordinate system for location on the
observations of falling objects. Leveling instru- Earths surface is becoming popular because of
ments (for example a bubble level) are capable of the accessibility of inexpensive receivers for the
very precise determinations of the local direction GPS and the fact that a metric system is easier to
of gravity and therefore can determine the shape manipulate. The GPS provides locations by mea-
of the geoid. For a uid Earth, the geoid would be suring the times of travel of radio signals from a
a perfect ellipsoid, but the Earths geoid has a set of satellites to a receiving antenna held at the
very irregular shape. These irregularities mean desired location (Hofmann-Wellenhof et al., 1997).
that the local direction of gravitational accelera- The travel time measurement requires very
tion (down) does not point directly toward the precise (atomic) clocks on the satellites and
center of the Earth. Indeed, this acceleration sophisticated electronic techniques for synchro-
varies from place to place, because of topography nizing the clock in the receiver to the satellite
and variable rock density. In some applications clocks. Using these clocks, the time of sending
the particular geoid chosen as the datum is the and receiving a radio signal from a particular
one that best approximates mean sea level at the satellite can be differenced to compute the travel
nearest coastline. time, t, for the signal from that satellite. Knowing
Thus, an important question to ask about el- the velocity of the radio signal, v, corrected for
evations is: what is the datum? Is it a mathemati- atmospheric delays, and the travel time, t, the dis-
cally dened ellipsoid or a physically dened tance from the antenna to that satellite is com-
geoid (see Fig. 2.2a)? This information should be puted as d vt. Knowing the positions of all the
provided on any map that depicts elevation. On satellites (provided by the government operators
many older maps in the continental United of the system) and the respective distances from
States, for example, the datum used is the the antenna to each satellite, the location of the
North American Datum for 1927 (NAD-27). This antenna can be computed. A minimum of four
coincides with the ellipsoid that closely approxi- satellite distances is required, three to determine
mates the geoid in North America, but the center location and a fourth to synchronize the clocks.
of this ellipsoid is about 100 m from the actual Any redundant data are used to rene the preci-
center of the Earth. Another familiar datum is the sion of the location.
North American Datum for 1983 (NAD-83) that Locations can be determined to better than
appears, for example, on many topographic and 1 m using portable receivers, and these are an
geologic maps published by the US Geological excellent choice for mapping most geological
Survey. Today the datum used for the Global structures (Chapter 2 Frontispiece). The two geol-
Positioning System (GPS) is the World Geodetic ogists shown on the frontispiece for this chapter
30 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
are standing on an exposure of Aztec sandstone in tion and then are unwrapped to a at sheet for
the Valley of Fire, Utah. The GPS antenna, seen display. The projection is done primarily for con-
just above the left-hand shoulder of one of the venience, to avoid having to use three-dimen-
geologists, is on a short mast extending from the sional representations such as globes. Ideally one
geologists backpack. In the geologists hands is would want the map projection to represent geo-
the computer used to enter data and control the graphic information without distortion. For
receiver. The most precise (geodetic grade) GPS example, any two curves of equal length on the
provide locations to about 0.01 m using equip- sphere (or ellipsoid) should project to two curves
ment transported in vehicles and set up over of equal length on the at map. This carto-
stable benchmarks. The more precise measure- graphic criterion is called equidistance because
ments are done using a technique of differential such a projection correctly represents distances.
corrections that reduces errors by comparing the The cartographic criterion called conformality
signal received by the roving antenna to the signal refers to the correct representation of shapes on
received at a nearby base station where the loca- the map and requires that angles on the sphere
tion is xed and well known. (or ellipsoid) project to the same angles on the
At the time of publication of this book the GPS map. The third cartographic criterion, equiva-
is the preferred technology for locating ones posi- lency, requires the correct representation of areas
tion on Earths surface while mapping geological from the sphere (or ellipsoid) to the map.
structures. Because this is a rapidly developing Unfortunately, it is generally not possible to
technology we have only described the rudimen- achieve undistorted representations that meet
tary features of the system and refer interested all three criteria simultaneously; so different
readers to books and manufacturers manuals for projections (apparently there are about two
details concerning usage, precision, and instru- hundred in use) are chosen for different purposes
mentation (Committee on the Future of the with the objective of minimizing the distortion
Global Positioning System, 1995; Hofmann- or honoring one or other of the three carto-
Wellenhof et al., 1997). graphic criteria (Richardus and Adler, 1972).
To illustrate the procedure of projection we
2.1.2 Map projections: the Universal consider a spherical datum of radius R, take the
Transverse Mercator (UTM) so-called perspective point (view point) as the
projection center of the sphere, C, and project points from
Different forms of map projections are used to the sphere onto a plane that is tangent to the
depict geographic information located on Earths sphere at the point O (Fig. 2.3a). The point O should
curved surface on a at piece of paper (Alpha et al., be centrally located within the region where
1988). This procedure is a necessary and crucial structural mapping is planned. The north pole of
part of making maps that accurately record and the sphere is labeled N. Points on the sphere are
convey the eld data of structural geology. In this projected to the plane along straight lines that
section we describe what is meant by map projec- emanate from the perspective point. This is called
tion, discuss some of the complications of this pro- the gnomonic projection and it is one member of
cedure, and provide some illustrative examples. In the class of azimuthal projections that provide
particular we describe the UTM projection, which images similar to what one would observe on a
is becoming the standard for geological and topo- photograph taken from space along a line of sight
graphical maps. coincident with the normal to the datum ellip-
A curved surface is taken as the datum, either soid or sphere at the point O (Richardus and Adler,
a spherical or an ellipsoidal model of Earth, and 1972, Chapter 4). Among the attributes of the gno-
the projection surface is taken as a plane, monic projection is the fact that directions from
cone, or cylinder that is tangent to the datum the point O are not distorted. Also, straight lines
(Richardus and Adler, 1972). The cone and cylin- from this point represent arcs of great circles (the
der are wrapped around the datum for projec- intersections of the sphere with planes that pass
2.1 GEOGRAPHIC COORDINATES AND MAP PROJECTIONS 31
12
W
0
126 o
oW
Northing
Equator
Equator
Easting
10 000 000 m
Central Y
(b) meridian
80o S
500 000 m
Equator X
on the Greenwich meridian . All forms of projection maps sections of the spherical or ellip-
this projection are limited in that an entire hemi- soidal datum onto a cylinder with its axis parallel
sphere cannot be projected: the boundary of the to the plane of the equator (Fig. 2.4a). The cylinder
hemisphere would plot at an innite distance is tangent to the ellipsoid along a particular line of
from the origin. longitude known as the central meridian. For this
One of the more common projections in use projection the equator and the central meridian
today is the Transverse Mercator (TM) projection in are straight lines whereas all other lines of longi-
which a spherical or ellipsoidal datum is projected tude and latitude are curved (Fig. 2.4b). Lines of
onto a right circular cylinder. The mathematical longitude to either side of the central line are
procedure used to accomplish the projection is concave toward this line. Lines of latitude, other
called conformal mapping (Hofmann-Wellenhof et than the equator, are concave toward their respec-
al., 1997, p. 287). As the name implies this trans- tive poles. Because this is a conformal projection
formation and others in this class preserve the the lines of longitude are everywhere orthogonal
angular relations of curves on the datum. The TM to the lines of latitude.
2.1 GEOGRAPHIC COORDINATES AND MAP PROJECTIONS 33
Taking a simple example for the sake of illus- these lines are projected as straight lines and they
tration consider the TM projection for a spherical are orthogonal to one another (Fig. 2.4c). Recall
datum of radius R. The Cartesian coordinates are that the other lines of longitude and latitude are
chosen so the X-axis is coincident with the equator projected as curved lines. The UTM grid is a
and Y-axis is coincident with the central meridian. Cartesian (rectangular) metric grid overlaid on
They are related to the latitude, , and longitude, this projection. The western edge of the UTM grid
, as (Richardus and Adler, 1972, p. 101): is a line drawn parallel to the central meridian but
500 000 m (about 4.5) to the west. The eastern
1
X R ln
2
1 cos cos
1 cos cos (2.3)
edge of the UTM grid is 500 000 m to the east of the
central meridian. Values along this axis are
Y R tan 1( cot sin )
referred to as false eastings because of the shift in
origin, and these values are measured parallel to
This projection is most appropriate where the the equator, starting at the western edge of the
mapped objects are organized along a particular UTM grid. Thus, the central line of longitude is at
meridian, or where the datum can be mapped sep- a false easting (X-coordinate value) of 500 000 m.
arately in narrow strips aligned with different Northing is measured in the northern hemi-
lines of longitude. sphere from the equator toward the north and
Most GPS receivers have onboard computers parallel to the central line of longitude. In the
that are capable of reporting locations using the southern hemisphere the zero northing is shifted
so-called Universal Transverse Mercator (UTM) projec- to the south, so the equator is at a false northing
tion, which is a modication of the TM projection. of 10 000 000 m. Note that for the depiction of
The UTM system was designed to meet a number zone 10 shown in Fig. 2.4c the easting axis is
of criteria including conformality to minimize stretched relative to the northing axis. At true
directional errors, a minimum number of zones, scale the zone in each hemisphere would be a very
limited errors in scale, a common referencing thin strip that is ten times taller than it is wide.
grid, and limited convergence of lines of longi- To specify the UTM coordinates of a particular
tude toward the poles (Richardus and Adler, 1972, point on Earths surface, the hemisphere and the
p. 138). The design of this system has proved to be zone number and the datum must be identied,
quite good so UTM coordinates are found on most along with the appropriate easting, northing, and
modern maps. To set up the UTM system the Earth elevation. Thus, for example, the UTM coordinates
is divided into 60 zones (the ancient Babylonians of a location just outside the Geology Building at
would like this choice!), each spanning 6 of lon- Stanford University are reported as follows:
gitude and each extending from 80 S to 84 N, and
Northern hemisphere
these zones are individually projected (e.g. Fig.
Zone M10
2.4c). The zones are numbered consecutively
Datum WGS-84
toward the east from M1, which spans from 180
Easting: 573 218.49 m
W to 174 W longitude, with a central line of lon-
Northing: 4 142 572.31 m
gitude at 177 W. Thus, for example, zones M10
Elevation: 3.6 m
through M19 span the United States from just off
the west coast (126 W line of longitude) to just off The horizontal precision of the measured coordi-
the east coast (66 W line of longitude). The central nates is about 3 cm for easting and northing and
meridian of zone M10 is 123 W and the central the vertical precision is about 10 cm. Recall that
meridian of zone M19 is 69 W. The projection the mean sea level elevation is about 28.9 m and
cylinder is tangent to the central meridian if the the negative elevation listed here is referenced to
datum is taken as a sphere. the WGS-84 ellipsoid (Fig. 2.2b).
To establish the UTM coordinates within a par- At the time of writing of this book the UTM
ticular zone, the central line of longitude and the projection is the preferred coordinate system for
equator are used as reference lines because both of the preparation of geological and structural
34 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
maps. Formulae to compute the coordinates on transformation from one coordinate system to
the UTM grid given geographic coordinates on an another because these are basic tools for struc-
ellipsoidal datum, and to calculate the geographic tural mapping.
coordinates on the ellipsoidal datum given the It is useful to understand the concept of
UTM coordinates have been derived (Richardus vectors in general, and to be familiar with specic
and Adler, 1972). These take several pages to write techniques for manipulating vectors, because
down and are not repeated here. Many GPS structural geologists employ them for modeling
systems have these formulae built in and make as well as mapping. This should come as no sur-
the computations at the push of a button. prise because the evolution of geologic structures
is primarily a physical process and the physical
laws that describe such a process are written as
2.2 Local coordinates and position vector equations. This section focuses on position
vectors, but also serves as a summary of some
vectors general vector concepts that we build upon in
later sections where, for example, we use vectors
Points and sets of points can be dened only relative to to characterize the shapes of folded geological sur-
(i.e., as functions of ) a coordinate system, never faces and to visualize velocity elds within a
absolutely. The coordinate system is the unavoidable deforming rock mass. Other general vector con-
residue of the eradication of the ego in that geomet- cepts are introduced as needed in later sections
rico-physical world which reason sifts from the given
and chapters.
using objectivity as its standard a nal scanty
token in this objective sphere that existence is only
given and can only be given as the intentional content 2.2.1 Locating data using local
of the processes of consciousness of a pure, sense- coordinates and position vectors
giving ego (Weyl, 1987). Geographical coordinates, based on the UTM
grid, may not be the best choice for mapping in
Students rst learn to deal with points and sets the eld, but their use is becoming universal for
of points described relative to a chosen coordi- the nal presentation of structural maps. For
nate system in elementary courses in mathemat- large-scale maps awkwardness arises because the
ics. This procedure was apparently conceived by central line of longitude for each UTM zone is
Ren Descartes in the early seventeenth century 500 000 m to the east of the origin (Fig. 2.4c), so a
and has become one of the most powerful tools typical easting would have six digits with meter
ever developed for scientists and engineers precision and eight digits if centimeter precision
(Davis and Hersh, 1986; Aczel, 2000). In honor of were required. Similar numbers of digits are
Descartes contribution the most familiar coordi- required for the northing unless, for example, the
nate system we use is referred to as the Cartesian location is near the equator in the northern
coordinate system. With this coordinate system and hemisphere. To reduce this cumbersome number
the concept of a position vector one can locate of digits it is practical to select a local origin
outcrops or the point where samples are col- within the mapped region. Local origins are also
lected in the eld relative to a local origin. In commonly employed in modeling and data analy-
other words the origin is located in the region sis. Furthermore, a local origin may be necessi-
being mapped or on the actual outcrop for very tated by the use of surveying equipment that
large-scale mapping, instead of at an arbitrary references locations to the instrument itself
point determined by a global projection such as rather than the UTM system. Topographical, geo-
the UTM grid. To put the map or data in a global logical, and structural maps can be prepared
context one transforms the position vectors from using a coordinate system consisting of the
the local coordinate system to the UTM system. easting, northing, and elevation for which the
Here we introduce the position vector, several datum is the local elevation of the instrument
local coordinate systems, and equations for the and the easting and northing are measured in the
2.2 LOCAL COORDINATES AND POSITION VECTORS 35
(a) (b) z
z
ez az
p
pz
ay
p
y
O
ey y
ex ax O
x c
px
(c) z x
z
py
pz z
(d)
P(x, y, z)
y z O
pz P p
p y
c
p
O x
y
O a
y
x x x b
Fig 2.6 The position vector, p. (a) Base vectors, e, and from scalar quantities, which have a magnitude,
direction angles, . (b) Components (px, py, pz) of the and possibly both positive and negative values,
position vector. (c) Rotation from the (x, y, z) coordinate but scalars lack a specic direction.
system to (x y z). (d) Translation of the coordinate system. In general, a vector may be written as a linear
combination of three base vectors and three quan-
dike introduced some uncertainty in locating the tities called the scalar components of the vector. For
contact, this procedure generally provided the position vectors we choose a set of three mutually
required precision. A portion of the rst photo- orthogonal base vectors (ex, ey, ez) each of unit
graph in the series is shown in Fig. 2.5b along with magnitude and each directed, respectively, from
the local coordinate system. the origin along the positive axes (Ox, Oy, Oz) of a
Every point in three-dimensional space is right-handed Cartesian coordinate system (Fig.
uniquely determined by a position vector, p, that 2.6a):
species the position of the point relative to a
ex (1, 0, 0), ey (0, 1, 0), ez (0, 0, 1) (2.4)
xed origin, O (Fig. 2.6a). The position vector is
visualized as an arrow extending from the origin These equations give the values of the three scalar
to the point. In other words the tail of this vector components for each unit base vector where it is
is at the origin and the head is at the designated understood that the respective components are
point. As with all vectors, the position vector has for the three coordinate directions (Ox, Oy, Oz).
both a magnitude and a direction. The length of Mutually orthogonal unit base vectors are said
the arrow is drawn proportional to the magni- to form an orthonormal basis (Malvern, 1969). In
tude, and the shaft and head of the arrow pre- some notations different symbols are used to
scribe the direction. In this text, vector quantities distinguish each base vector (e.g. i, j, k), but here
are written in boldface type to distinguish them a common symbol is used with subscripts to
2.2 LOCAL COORDINATES AND POSITION VECTORS 37
associate each base vector with a particular coor- orthogonal projections of v onto lines drawn
dinate direction. Most vector quantities used in parallel to the base vectors (ex, ey, ez), respectively.
this book, such as the position vector or velocity If the two points are coincident v 0 (0, 0, 0) is
vector, are free to take on any orientation and the zero vector.
therefore do not carry subscripts. The coordinate The meter (m) is the physical unit used to
axes, and therefore the base vectors, comprise a measure distance, and this is carried by the posi-
right-handed system. By right handed we mean tion vector components, not the base vectors.
that the axes (Ox, Oy, Oz) extend, respectively, in Thus one would write, for example, py 22.5 m
the same mutual orientations as the thumb, and |ey| 1. Each base vector has a magnitude of
index nger, and middle nger of the right hand one but does not carry a physical unit. The base
when these ngers are directed orthogonal to one vectors provide the directional information nec-
another. Base vectors need not be orthogonal to essary to compose a position vector. The three-
one another nor of unit magnitude, and Cartesian dimensional space dened in terms of position
coordinates can be left-handed, but these alterna- vector components using the basis (ex, ey, ez) is
tives are not employed in this book. referred to as Euclidean space (Lipschutz, 1969).
The scalar components of a vector are quanti- When the base vectors change direction
ties that can be visualized as the orthogonal pro- because of a rotation of the coordinate system
jections of the vector arrow onto lines in arbitrary about the origin (Fig. 2.6c), the components of the
but designated directions. In general, scalar com- position vector p for an arbitrary point P change,
ponents are written with the same symbol as the but the magnitude and direction of the position
vector, but the type is not bold and each compo- vector do not change. For example, the projection
nent has a subscript that identies the line pro- pz of p onto the rotated coordinate axis Oz is dif-
jected upon. For the position vector, p, the ferent from the projection pz onto the original
designated directions are parallel to the base coordinate axis Oz. Thus we say that (px, py, pz) are
vectors (Fig. 2.6b), in other words lines parallel to the components of p with respect to a particular
the respective coordinate axes. The three compo- basis (ex, ey, ez), and this basis must be dened in
nents of the position vector are written (px, py, pz). order to interpret the components. On the other
Thus, the equation for the position vector with hand if the base vectors and coordinate system are
the basis (ex, ey, ez) is written as the following translated to a new origin, O, but not rotated (Fig.
linear combination of scalar components and 2.6d), a new position vector, p, extends to the
base vectors: given point, P, and this vector has a different mag-
nitude and/or direction than p. In general, two
p pxex pyey pzez (2.5)
vectors are equal only if they have the same mag-
Although it is convenient to speak of a position nitude and direction, so the position vectors for
vector simply as p, and this notation is useful for the point P referred to different origins are not
writing compact vector equations, most calcula- equal, despite the fact that they dene the loca-
tions are done using the scalar components. tion of the same point. In this sense position
Apparently Descartes introduced the notion of vectors are so-called xed vectors (Malvern, 1969):
using vector components, dened with respect to they emanate from a particular point, the origin
a coordinate system in order to develop analytical of the coordinate system, and the location of this
geometry (Fung, 1969, p. 22). origin must be specied in order to dene a set of
An arbitrary vector, v, may have its tail at any position vectors. Position vectors are useful quan-
point (x1, y1, z1) and its head at any other point (x2, tities for dening the locations of points, but they
y2, z2) so this vector is written: lack the attributes of those vectors that we think
of as physical entities, such as force or velocity,
v (x2 x1)ex ( y2 y1)ey (z2 z1)ez
(2.6) which are independent of an arbitrarily dened
vxex vyey vzez
coordinate system.
The quantities (vx, vy, vz) are the scalar components The magnitude of any vector is calculated as
of the vector, which can be interpreted as the the square root of the sum of the squares of its
38 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
scalar components and this quantity is written Table 2.1. Position vectors for ship rock dike.
with the same symbol as the vector but not in
boldface type. For the position vector this rela- Counter, i px (m) py (m) qx (m) qy (m)
tionship may be deduced from Fig. 2.6b and the
theorem of Pythagoras. The squared length of the 1 3.9 15.7 3.9 15.7
position vector is p2 (pz)2 c2, but c2 (px)2 (py)2. 2 4.6 15.4 4.6 16.0
Thus, the magnitude, p, of the position vector p is: 3 5.8 15.4 5.8 16.0
4 7.0 15.3 7.0 16.1
p |p| p2x p2y p2z 0 (2.7) 5 8.1 15.2 8.1 16.0
The magnitude of a vector, written symbolically ... ... ... ... ...
using ||, is equivalent to the scaled length of the 2720 2900.2 6.1 2900.2 4.4
vector arrow, and is always greater than or equal 2721 2901.4 6.2 2901.4 4.7
to zero. The components of the position vector are 2722 2902.5 6.4 2902.5 5.0
proportional to the vector magnitude and the 2723 2903.7 6.5 2903.7 5.2
cosines of the angles that the vector makes with
2724 2904.8 5.9 2904.8 5.9
the positive coordinate axes (Fig. 2.6a):
px p cos x, py p cos y, pz p cos z (2.8)
think of the photograph-based map as the projec-
The set of three angles (x, y, z) are called the tion of the dike outcrop onto the (x, y)-datum plane
direction angles of the position vector, and the of the local coordinate system. Then, for example,
cosines of these angles are referred to as the direc- the position vector for the counter i 21, shown in
tion cosines. The direction angles are measured in Fig. 2.5b, is written in the form of (2.5) as:
the planes dened by the vector and the respec-
p pxex pyey (26.7 m)ex (18.2 m)ey (2.9)
tive coordinate axis, and each is taken as the
smaller of the two angles between the vector and Because all of the vectors used to quantify the dike
the positive coordinate axis. contact lie in the (x, y)-plane, the direction angle
At Ship Rock (Fig. 2.5a), the UTM coordinates of z /2 (Fig. 2.6a) and, for example, using (2.8) and
the local origin near the western (proximal) ter- i 21 we have:
mination of the middle dike are given above. The
local Cartesian coordinate system is oriented with
the z-axis vertical (upward) and the x-axis directed
py p cos y p cos
2
x p sin x
t(i) py(i) qy(i) |p(i) q(i)| (2.11) its nal thickness simply by opening a large crack.
The physical insights and new data gained through
This can be understood by considering the general precise quantitative mapping are the reward for
expression for the sum of two vectors: the effort required to learn and implement the
techniques.
if r v w, then rx vx wx,
ry vy wy, rz vz wz (2.12)
2.2.2 Transformation of position and
The difference of the two vectors may be thought basis vectors, and coordinate
of as r v (w). In other words the sum (or dif- systems
ference) of two vectors is a vector with scalar com- For the presentation and publication of maps and
ponents that are the sum (or difference) of the structural data sets it is appropriate to reference
respective components. This result can be gener- locations to the UTM grid (Fig. 2.4). For example,
alized to the sum or difference of any number of the local Cartesian coordinate system (x, y, z) used
vectors. For a given value of the counter the x-com- to dene the position vectors for the contact of
ponents of the position vectors from Table 2.1 are the northeastern Ship Rock dike (Fig. 2.7) can be
equal, and the z-components are zero for all these transformed to the UTM grid through a two-step
vectors, so the magnitude (2.7) of the vector dif- procedure starting with a clockwise rotation
ference reduces to the difference of the y-compo- through 34 about a vertical axis to align the x-axis
nents. The set of thickness values so determined is and y-axis with the easting and northing direc-
used in a mechanical model for the opening of the tions on the grid. This is followed by a translation
dike segments that leads to estimates for the stiff- of the local origin to the origin for the northern
ness of the host rock (Delaney and Pollard, 1981). hemisphere of zone 12, which is 694 000 m to the
Several segments of the northeastern Ship Rock west, 4 063 000 m to the south, and 1675 m down.
dike are shown on Fig. 2.7, a structure map using In general, changing (transforming) from one
local Cartesian coordinates (Fig. 2.5b). The precise rectangular coordinate system to another may
mapping method described above brings out a host involve three independent operations: a rotation
of structural features (Delaney and Pollard, 1981) about a xed origin (Fig. 2.6c) without changing
that otherwise would be unrecorded on typical the orthogonality or sense of the axes; a transla-
quadrangle-scale maps where thin constant-width tion of the origin (Fig. 2.6d) without changing the
lines only display the location and trace of dikes. orientation or sense of the axes; and a reection
Instead we note that the dike is composed of sub- that only changes the sense of the axes (right
parallel but offset segments forming an echelon handed to left handed or vice versa) while main-
array. Segments numbered 13 to 15 are offset by a taining their orientation and the origin. Because
few meters and have blunt, rounded terminations, we utilize only right-handed coordinate systems,
whereas segments 15 to 21 are offset by as much as the operation of reection is not considered
15 m and have more tapered terminations. The further. Here we illustrate rotation and transla-
contacts of overlapping segments have a distinct tion using the Ship Rock example and then gen-
asymmetry such that adjacent contacts curve eralize these operations for the transformation of
toward their respective terminations and distal vectors and rectangular coordinate systems in
contacts are relatively straight. At x 1000 m, on three dimensions.
segment 12, both contacts bulge outward to form The azimuth of the local x-axis at Ship Rock is
a putative volcanic neck; and at x 1310 m, on 56 measured clockwise from north in the hori-
segment 19, one contact bulges outward such that zontal plane (Fig. 2.5). Therefore, the transforma-
the total thickness almost doubles. Both of these tion to a new Cartesian coordinate system (x, y)
structures are associated with breccias that that shares the same origin but is aligned with the
suggest the local increases in thickness are related UTM grid axes is a clockwise rotation of 34 about
to fracturing and brecciation of the host rock and the vertical z-axis. This is a two-dimensional trans-
transport of the breccia by owing magma. At formation in which every point, P(x, y), referred to
these locations the dike apparently did not attain the old coordinate system on the map takes on
40 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
Meters
50
13
Km
9 Tmn 11 12
Km Thb
Tmb Thb Tmb 10 Tmb
83
Thb
7 Tmn Thb
8
0
800 900
16
14 Thb
15 75 Tmb
Thb
Thb Thb Tmn
Km
Tmn
Km
1000 1100
Km 18 19 87 80
Tmb
17 Thb
Thb 23
87 90
80 Km
82
1200 1300
Tmn Minette
Contact of dike with exposed host rock (km)
Thb Heterobreccia
Contact located to within 0.05 m
Contact approximately located; possible error greater that 0.1 m Tmb
Tmb Monobreccia
28 Strike and dip of beds
Mancos Shale
25 Strike and dip of joints
Km Meters
50
82 Strike of vertical joints
Km
9 Segment number 21 74 Thb
Thb Tmn
20 Thb 77
86
Km
Tmn
88
0
1400 1500
x x a x
y y b (2.18)
z z c Fig 2.9 Rotation of Cartesian coordinates in three
dimensions. (a) Direction angles relating old (x, y, z) and new
Here the constants (a, b, and c) are the coordinates
(x y z) axes. (b) Geometric interpretation of scalar product
of the origin, O, of the new system as written with of two arbitrary vectors, v and w.
respect to the old system. That is, the origin of the
old system is translated a distance a along x and a
distance b along y and a distance c along z from the two angles in the plane dened by Ox and Ox.
the origin of the old system (Fig. 2.6d). The inverse This direction angle is referred to as (x, x).
of this transformation is found algebraically by Similarly, the direction angle (y, x) relates the y-
rearranging the three equations to solve for x, y, axis to the x-axis, and the direction angle (z, x)
and z, respectively. relates the z-axis to the x-axis. Three direction
The two-dimensional rotational transforma- angles are dened similarly for the y-axis and for
tion (2.13) can be generalized to three dimensions the z-axis, bringing the total to nine angles.
if the old and new coordinate systems share a The old and new coordinate systems (Fig. 2.9a)
common origin. The coordinate axes (Ox, Oy, Oz) of have bases (ex, ey, ez) and (ex, ey, ez), respectively.
the old system are related to the coordinate axes To calculate the new basis vectors from the old
(Ox, Oy, Oz) of the new system using nine direction we need to dene the scalar product of two vectors.
angles. For example, in Fig. 2.9a the positive x-axis For two arbitrary vectors, v and w, the scalar
is related to the positive x-axis by the smaller of product is:
2.2 LOCAL COORDINATES AND POSITION VECTORS 43
v w |v||w| cos , for 0 The direction cosines and basis vectors may be
vxwx vywy vzwz (2.19) organized into a table that facilitates rotational
transformations in three dimensions:
Because the scalar product is written with a dot
ex ey ez
between the two vectors, it is referred to as the dot
product. Here is the smaller of the two angles ex mxx mxy mxz
(2.23)
between the lines directed parallel to v and w ey myx myy myz
measured in the plane dened by the two vectors ez mzx mzy mzz
(Fig. 2.9b), so the product |v|cos is the orthogo- The new and old basis vectors are listed in the rst
nal projection of v onto the line parallel to w. In row and rst column, respectively, and the direc-
other words |v|cos is the component of v in the tion cosines ll out the table such that their two
direction of w. Therefore, the scalar product may subscripts match the subscript of the basis vector
be interpreted geometrically as the product of the heading the row and the column, respectively.
component of v along w and the magnitude of w. Each new basis vector is composed of a linear
Alternatively, the scalar product may be inter- combination of the old basis vectors and the
preted as the product of the component of w direction cosines in the corresponding column of
along v and the magnitude of v. (2.23):
Two important relationships among the
vectors dening a basis are expressed con- ex mxxex myxey mzxez
veniently using the scalar product (2.19): ey mxyex myyey mzyez (2.24)
ex ex ey ey ez ez 1 ez mxzex myzey mzzez
ex ey ey ez ez ex 0 (2.20)
Each old basis vector is composed of a linear com-
These relations are interpreted by recalling that bination of the new basis vectors and the direc-
orthonormal basis vectors are of unit magnitude tion cosines in the corresponding row of (2.23):
and they are mutually orthogonal to one another.
ex mxxex mxyey mxzez
For the rotational transformation we want, for
example, to project ex, ey, and ez onto the line par- ey myxex myyey myzez (2.25)
allel to ex (Fig. 2.9a). These projections are accom- ez mzxex mzyey mzzez
plished using the scalar product (2.19) and
recalling that the base vectors are unit vectors: This is the inverse rotational transformation. The
basis vectors are unit vectors so (2.7) requires that
ex ex |ex||ex|cos (x, x) cos (x, x) mxx the sum of the squares of the direction cosines in
ey ex |ey||ex| cos (y, x) cos (y, x) myx (2.21) any row or column of (2.23) equal one. For
ez ex |ez||ex|cos (z, x) cos (z, x) mzx example:
Analogous procedures are used to project ex, ey, |ey| (mxy)2 (myy)2 (mzy)2 1,
and ez onto lines parallel to ey and ez and these so (mxy)2 (myy)2 (mzy)2 1 (2.26)
operations dene nine direction cosines that relate
The direction cosines as arranged in (2.23) com-
the old basis to the new basis:
prise a square matrix and the rotational transfor-
mxx cos (x, x), mxy cos (x, y), mxz cos (x, z) mation will be described in terms of matrix
operations later.
myx cos (y, x), myy cos (y, y), myz cos (y, z)
It was pointed out with respect to the two-
mzx cos (z, x), mzy cos (z, y), mzz cos (z, z)
dimensional form of the rotational transforma-
(2.22)
tion that basis vectors and components of
Here the double subscripts on the direction position vectors and coordinates of points all
cosines refer to the reference (old) axis and the transform using similar equations. This principle
transformed (new) axis, respectively. extends to three dimensions so we can construct
44 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
y orthogonal unit base vectors (ex, ey, ez) of the
r x2 y2, tan 1 , zz
x Cartesian system:
p pxex pyey pzez
The second line of (2.29) contains the correspond-
ing inverse transformation equations. Note that (r cos )ex (r sin )ey (z)ez (2.30)
46 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
The Cartesian coordinates dened in the rst line ing to the governing equations for heat conduc-
of (2.29) are the scalar components of the position tion with the boundary condition that the tem-
vector. These scalar components carry the units of perature far from the neck remains zero:
length (meters) as do the radius, r, and axial
length, z. The angle, , is measured in radians
which may be thought of as the ratio of arc length
IC: for t 0
T Tm at r R
T 0 at r R
(2.31)
to radial length, s/r. BC: at r T 0 for t 0
For modeling purposes it is helpful to reduce
the number of coordinates from three to two by A uniform temperature could be added every-
idealizing the structure as one that is perfectly where to account for the ambient temperature
cylindrical. For the volcanic neck at Ship Rock before the development of the volcanic neck.
(Fig. 2.10c), we would say that it has a similar The natural symmetry of some geological
geometry for any horizontal cross section within structures motivates use of a coordinate system
several tens of meters of the current outcrop, so that bears some resemblance to the cylindrical
we can ignore spatial variations in geometry, system, but admits very eccentric shapes. The map
material properties, and boundary conditions of dike segments (Fig. 2.7) suggests that an ellipti-
with the axial coordinate z. Then we focus on the cal cross section may be a good approximation for
two-dimensional polar coordinates (r, ) and dene some of these. Segment 16, for example, is 136 m
the problem in terms of these two coordinates long, has a maximum thickness of 3.4 m, and is
(Fig. 2.10d). Radial lines of constant and circles approximately symmetric about planes that pass
of constant r form an orthogonal network cen- through its middle parallel to its length and thick-
tered on the origin. The contact between the ness (Fig. 2.12a). The contact between the minette
minette of the neck and the Mancos Shale is and the Mancos Shale is relatively smooth and
dened as a particular circle, r R15 m, with a slowly tapers toward the distal terminations. We
radius that best approximates that of the neck. have scant information about the extent or shape
The alternative would be to dene the contact in of this segment with depth, or with height above
terms of the Cartesian coordinates as the circle (x2 the current outcrop. Unlike the necks which
y2)1/2 R, but this is mathematically more cum- stand well above the surrounding shale, the dike
bersome. segments reveal little of their three-dimensional
Similarly, the boundary (BC) and initial (IC) geometry. Over the few meters of local relief the
conditions for models of the volcanic neck are contact maintains a near vertical dip and this is
simplied using polar rather than Cartesian coor- consistent over the 55 m of elevation change from
dinates. For example consider a model for the con- one end of the dike to the other. Where exposed
ductive heat ow from the hot magma into the the terminations of individual segments are
cold host rock. The cylindrical geometry of the vol- steeply plunging. Based on this limited informa-
canic neck dictates a temperature eld that varies tion we adopt two-dimensional elliptical coordi-
spatially only in the radial coordinate direction, nates, which may be visualized as a set of confocal
thereby reducing the problem to one spatial ellipses and hyperbolae (Fig. 2.12b).
dimension. Because the governing equation for For reference consider the Cartesian axes (Ox,
heat conduction is written with temperature as Oy, Oz) with Oz perpendicular to the plane of inter-
the dependent variable, the initial conditions typ- est. We postulate that the dike segment has a
ically constrain the temperature eld at some similar geometry for any horizontal cross section
specied time and the boundary conditions con- within several tens of meters of the current
strain the temperature eld at specied locations outcrop, so we can ignore spatial variations in
on the surface(s) of the body. Here we consider geometry, material properties, and boundary con-
initial conditions at the time t 0, such that the ditions with the coordinate z. Then the basic ele-
temperature is uniform, Tm, throughout the neck ments of the two-dimensional elliptical system
and zero throughout the host rock. Then let the are an origin, O, the line Ox, a coordinate , a coor-
temperature eld, T(r, t) evolve with time accord- dinate , and a focal length, f, measured from the
2.2 LOCAL COORDINATES AND POSITION VECTORS 47
any circle centered at the origin would be repre- The specication of boundary conditions for prob-
sented by a constant value of r, each of the family lems related to dike segments is particularly
of confocal ellipses is represented by a constant simple using elliptical coordinates. For example,
value of . However, unlike the coordinate r, which if the magma pressure, pm, exerts traction, t, only
is a length measured in meters, the coordinate perpendicular to the contact, we write:
is just a number. Also, as each radial line in the
polar system is associated with a particular value
of the angle , each of the family of confocal
BC: on O
t pm
t 0
(2.36)
hyperbolae is associated with a particular value of Here t is the traction component acting on the
the angle . elliptical surface, O, in the -coordinate direction
48 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
cating that each joint nucleated at a particular The second and third lines contain the inverse
cavity and propagated vertically and laterally transformation equations. The position vector, p,
away from it. The three semi-axes of 14 cavities in used to locate an arbitrary point P(, , ) in spher-
a single layer were measured and provide the fol- ical coordinates (Fig. 2.11b) is written using the
lowing statistics: rst line of (2.38) as:
2.2 LOCAL COORDINATES AND POSITION VECTORS 49
3
used for a variety of different purposes in struc-
tural geology, so it is important to understand
x2 m12x1 m22x2 m32x3 m
i1
i2xi mi2xi (2.42)
written using indicial notation. Some of the basic x3 m13x1 m23x2 m33x3 m
i1
i3xi mi3xi
specied range. Because there is one free index as a table or array of numbers motivates consider-
which ranges from 1 to 3 in each of these equa- ation of the concept and mathematical properties
tions, each may be expanded into the three of a matrix. This is further motivated by the fact
equations. The forward and inverse rotational that the computational engine, MATLAB, used for
transformations for the components of the old the exercises and many of the graphical illustra-
and new position vectors ( pi, pj) and the old and tions in this text, treats all data sets as arrays of
new basis vectors (ei, ej) are constructed similarly. numbers and offers many useful functions that
The denition of the direction cosines in operate on matrices. Furthermore, many of the
terms of the base vectors is written using indicial constructs of continuum mechanics can be
notation and the scalar product (2.19) as: described and manipulated as matrices. As we
have just done with indicial notation, some of the
mij ei ej (2.44)
basic concepts of matrices are introduced here as
Here there are no repeated indices but both i and they specically relate to coordinate transforma-
j range from 1 to 3, so this equation expands into tions; others are introduced as needed through-
nine equations for the direction cosines. The out the text.
scalar product of two arbitrary vectors, v and w A matrix is a rectangular array of numbers
(Fig. 2.9b), is expressed using indicial notation as: with each element of the array designated by its
position in the array according to the row and
v w v i wi (2.45)
column number: an m by n matrix has m rows and
Here the repeated index requires summation of n columns. For example, consider the position
the product of the components over the range 1 to vectors that were used to describe the contact of
3. the dike at Ship Rock (Fig. 2.5). In general, a posi-
The facts that the base vectors are of unit mag- tion vector, p, is written using indicial notation
nitude and orthogonal to one another are and in expanded form as follows:
expressed by the two equations:
p piei p1e1 p2e2 p3e3 (2.50)
e1 e1 e2 e2 e3 e3 1
(2.46) The three components (p1, p2, p3) of the position
e1 e2 e2 e3 e3 e1 0
vector can be thought of as either a 1 by 3 row
These conditions motivate the denition of a new matrix or a 3 by 1 column matrix:
quantity, ij called the Kronecker delta that nds
considerable usage with indicial notation:
P1
1, if i j P [P1 P2 P3 ], or P P2 (2.51)
ij , for (i, j 1, 2, 3) (2.47)
0, if i j P3
The conditions of unit magnitude and orthogonal Notice that the single subscript for each vector
orientations (2.46) are succinctly written using component, pi, is replaced by a double subscript
the Kronecker delta: and that these subscripts refer, respectively, to the
row number and column number. There are no
ei ej ij (2.48)
restrictions on the number of elements in a row
Furthermore, the conditions that the squares of or column matrix, or on the relationships among
direction cosines in each row and each column of those elements, although our example happens to
(2.41) sum to one are written: use three elements that are components of a posi-
tion vector. Sometimes row and column matrices
mkimkj ij, mikmjk ij (2.49)
are referred to as vectors, but we restrict that
The rst equation applies to the direction cosines term to quantities in which the elements have the
in each column (sum over the rst index) and the properties of vectors.
second equation applies to each row (sum over the As a second example consider the set of
second index). direction cosines used to relate the old and new
The arrangement of direction cosines in (2.41) basis vectors for a rotational transformation of
2.2 LOCAL COORDINATES AND POSITION VECTORS 51
coordinates (2.41). Using indicial notation and in the two matrices and the sequence in which they
expanded form these direction cosines are can be multiplied. As an example, consider the 2
written: by 3 matrix, Q, and the 3 by 1 matrix, R:
m11 m12 m13
R1
mij m21
m31
m22
m32
m23
m33
(2.52)
Q Q1
Q1
Q2
Q2
Q3
Q3
, R R2
R3
(2.56)
M11 M12 M13
M M21
M31
M22
M32
M23
M33
(2.53) QR Q 11R11 Q 12R 21 Q 13R 31
Q 21R11 Q 22R 21 Q 23R 31
S (2.57)
Q 11 R 11 Q 12 R 12
the sequence QR if the number of columns of Q is
Q R Q 21 R21 Q 22 R 22 (2.54)
equal to the number of rows of R. Indicial notation
Q 31 R31 Q 32 R 32
provides a succinct way to describe the elements
This results in a row matrix of exactly the same of the matrix S (Malvern, 1969, p. 41):
order, 3 by 2. The addition (or subtraction) of
matrices requires that they be of the same order. i 1, . . . , m
The 3 by 2 matrix R is scaled by a constant Sij Q ik Rkj, for j 1, . . . , n (2.59)
numerical factor as follows: k 1, . . . , l
kR11 kR12
kR kR21 kR22 column of the matrix S. Because of the repeated
(2.55)
kR31 kR32
index k, each element is the sum of l terms. If the
number of rows of Q is the same as the number of
In general, the multiplication of a matrix by a con- columns of R, m n, the multiplication can
stant is accomplished by multiplying each proceed in the reverse sequence, RQ, but the two
element of the matrix by that constant. The divi- products are not equal, QR RQ.
sion of a matrix by a constant is equivalent to mul- Multiplication with square matrices is com-
tiplying by the reciprocal of that constant. monly encountered in applications to physical
The multiplication of two matrices follows problems and they have special properties that can
rules that are similar, in part, to those for the be illustrated using the rotational transformation
scalar product of two vectors (2.19), but there are equations. Recall the table (2.41) that relates the
important distinctions that restrict the order of old, xi, and new, xj, Cartesian coordinates using the
52 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
direction cosines, mij. These three quantities may describe the structures underfoot. Most geological
be written as two column matrices and one square structures may be idealized as three-dimensional
matrix: curved surfaces or curved lines. Examples are sur-
faces that truncate older formations such as a
X1 X1 M11 M12 M13 fault offsetting sedimentary bedding, or a dike
X X2 , X X2 , M M21 M22 M23 (2.60) that cuts across an igneous contact. Or there may
X3 X3 M31 M32 M33 be surfaces within a mass of rock dened by the
alignment of platy minerals, as in an igneous or
The inverse rotational transformation, xi mijxj, metamorphic foliation. Curvilinear structures
may be constructed directly by matrix multiplica- also may be composed of aligned mineral grains,
tion, X MX, because the number of columns of as in a metamorphic lineation. The intersection of
M is the same as the number of rows of X, and the two curved surfaces, for example the intersection
summation required by the indicial notation is of two faults, would dene a curved linear struc-
consistent with that of matrix multiplication: ture. Regardless of the specic nature of these
curved surfaces and lines we need techniques for
X1 M11 M12 M13 X1
measuring their orientations in the eld and
X2 M21 M22 M23 X2 for recording these orientations on a map at the
X3 M31 M32 M33 X3 position determined by the UTM coordinates.
(2.61)
M11X11 M12X21 M13X31 Techniques are introduced here along with a pro-
M21X11 M22X21 M23X31 jection that is useful for visualizing the relative
M31X11 M32X21 M33X31 orientations of such structures.
However, the forward rotational transformation, 2.3.1 Orientations of linear and planar
xj mijxi, is not represented as X MX, despite structural elements
the fact that the number of columns of M is the Most curved structural surfaces may be approxi-
same as the number of rows of X, because the sum- mated locally by a planar element that is tangential
mation implied by the indicial notation is not to the surface at the point of measurement.
consistent with matrix multiplication. Instead Similarly, most curvilinear structures may be
one must rst take the transpose of M, symbolized approximated locally by a linear element that is tan-
as M T, and then compute the product as X M TX: gential to the curve at the point of measurement.
What are actually recorded by the structural geol-
X 1 M11 M21 M31 X1 ogist at an exposure are the orientations of these
X 2 M12 M22 M32 X2 structural elements. The exposure photographs in
X 3 M13 M23 M33 X3 Fig. 2.14 show a number of geological structures
M11X11 M21X21 M31X31 that can be approximated in this way with planar
M12 X11 M22 X21 M32 X31 (2.62)
and linear elements. These exposures are from the
M13 X11 M23 X21 M33 X31 northern part of the San Rafael Swell in the
Colorado Plateau province of central Utah (Kelly,
Note that the transpose, M T, is found by inter- 1955). In Fig. 2.14a a member of the Chimney Rock
changing the rows and columns of M. fault system juxtaposes beds of limestone, silt-
stone, and mudstone of the Middle Jurassic
Carmel Formation (to the left) against the massive
2.3 Orientations of structural Jurassic Navajo Sandstone (to the right). Because
the Carmel Formation immediately overlies the
elements Navajo Sandstone in the normal stratigraphic
sequence, we deduce that the Carmel Formation
Given the complete UTM geographic coordinates exposed in this photograph has moved downward
for the position of a particular outcrop, the next on the fault relative to the Navajo Sandstone.
step in most structural studies is to measure and The fault pictured in Fig. 2.14a is a steeply
2.3 ORIENTATIONS OF STRUCTURAL ELEMENTS 53
structure. For the example illustrated in Fig. 2.14a element using strike and dip. The relations
we would refer to the strike of the fault, and the between dip direction and strike direction are:
fact that the fault is not simply a plane in space
d s 90, 0 d 90 (2.63)
would be left unspoken. To understand what
someone means by the strike of a fault, both Two special cases are noteworthy. For a horizontal
the nature of the surface within the fault zone planar element the dip is zero, so the strike and
and the size of the planar element should be dip direction are undened and we would write
identied. (UDF, 00) for the strike (or dip) direction and dip:
Knowing the strike of a planar element
s UDF d, d 0 (2.64)
enables one to identify the orientation of the line
of intersection of that element with a horizontal For a vertical planar element the dip is 90, but
plane. There are an innite number of elements there are two possible strike (and dip) directions,
that share the same strike, but vary in inclination, and either one is suitable. For example, the strike
so we have to dene a second angle that is a and dip of a vertical plane with line of strike
measure of this inclination in order to determine recorded in the eld as (136, 90) could equally well
uniquely the orientation of a particular planar be recorded as (316, 90).
element relative to the local geographic coordi- Next consider the orientations of linear ele-
nate system. This angle, d, is measured in a verti- ments that approximate structures such as the
cal plane that contains the dip direction, from the slickenlines in Fig. 2.14b. These also are dened
dip direction to the planar element in question with respect to the local geographic coordinate
(Fig. 2.15a). This angle is referred to as the dip of system composed of east, north, and up (Fig.
the planar element that approximates the struc- 2.15b). The rst step is to imagine a vertical plane
ture, and by convention two digits are used to that contains the linear element. Position your-
specify the dip. Thus, a dip of seven degrees would self over the structure at the exposure (or imagine
be written 07. Note that the angle measured from positioning yourself over the line segment) such
the horizontal plane to the planar element in ver- that the structure (line segment) is inclined
tical planes of other orientations would be downward in front of you. The plunge direction is
smaller than the dip. Such an angle is referred to the direction of your view in the horizontal plane.
as an apparent dip. Apparent dips are commonly The azimuth of the plunge direction, p, is some-
observed and measured in the eld where expo- times referred to as the trend of the linear
sures cut obliquely across structures. element and it is specied using three digits. The
Only two angles are necessary to reference a angle p measured in the vertical plane from the
planar element to the local geographic coordinate plunge direction down to the linear element is
system. Some use the strike and dip (s, d), dened as the plunge and it is specied using two
whereas others use the dip direction and dip (d, digits.
d). Because of the construction of some eld Two angles, the plunge direction and the
instruments it may be more convenient to plunge (p, p) are necessary to reference a linear
measure and record the strike. On the other element to the local geographic coordinate
hand, the dip direction and dip require the system and these would be recorded, for example,
identication of only one direction and that direc- as (356, 58) indicating a linear trend just a few
tion corresponds more directly to the inclination degrees west of north and plunging 58 in that
of the planar element that is measured. If the dip direction. Again, two special cases are noteworthy.
direction and dip were used, the number pair For a vertical linear element the plunge is 90, so
(085, 37) would indicate a planar element with a the plunge direction is undened:
dip direction just 5 to the north of east and an
p UDF, p 90 (2.65)
inclination of 37 in that same direction. The
degree symbol is left off for recording conve- This line would be recorded as (UDF, 90). For a hor-
nience. The number pair (355, 37), plus the right- izontal element the plunge is 0, so there are two
hand rule, would identify the same planar possible plunge directions and either one can be
56 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
used. Thus, the same horizontal linear element linear elements. For example, note the different
could be recorded as (022, 00) or (202, 00). symbols for joints and veins or dikes. In this way a
Some curvilinear structures lie in curved sur- lot of information is conveyed in a compact form
faces that are well-dened structures themselves. on the map, and structures can be related to one
For example, the slickenlines shown in Fig. 2.14b are another in terms of their locations.
in the surface of a fault. For these cases the strike
and dip of the planar element that represents the 2.3.2 Stereographic projection of
surface are recorded, along with an angle known as structural elements
the rake that can be measured on the exposure with It is useful to have a graphical means to visualize
a protractor (Fig. 2.15c). The rake is the angle, r, the attitudes of planar and linear structural ele-
measured in the plane of the element from the ments. The most effective tools for this purpose
strike direction down to the linear element. It is are a family of projections that create an image of
recorded using three digits and can vary from 000 the elements on a at piece of paper. The locations
(the linear element is parallel to the line of strike of of the structures are not recorded in this image,
the planar element) to just less than 180: but their orientations relative to the geographic
coordinate system are recorded. Here we intro-
000 r 180 (2.66)
duce one of these projections, the so-called stereo-
For r 090 the linear element is inclined graphic projection. Details concerning the use of
directly down the dip of the planar element. Of this projection, and other members of this family
course one could measure the plunge direction of projections, can be found in books devoted to
and plunge of this line, but often it is simpler to the subject (Phillips, 1954; Ragan, 1985; Marshak
measure the strike and dip of the planar element and Mitra, 1988). In some courses in structural
and the rake angle in this planar element. geology much of the students time is committed
Structural geologists refer to the attitude of a to the manipulation of these projections by hand
structure and by that they mean the orientation in and many of the geometric problems encountered
space, relative to the local geographic coordinate in eldwork are described in these reference
system, of the planar or linear element that works. Here we adopt the more analytical
approximates (is tangential to) the structure at the approach described by Goodman and Shi (1985,
point of measurement. Thus, the attitude of a p. 56) that avoids the tedium and inaccuracy of
fault at a particular location would be recorded as hand constructions. We also take advantage of the
the strike and dip, or the dip direction and dip. The visualization power of modern computer applica-
attitude of a slickenline on that fault would be tions for plotting quantitative eld data.
recorded as the plunge direction and the plunge. We begin by reviewing the basic concepts and
These measurements are represented on maps present the analytical expressions necessary to
using symbols and numbers placed at the appro- plot stereographic projections of planar and
priate location. Some of the symbols used on struc- linear structural elements. Consider a linear
tural maps are illustrated in Fig. 2.16, extracted element xed in space at the center, C, of a trans-
from a more extensive table of symbols in a parent sphere called the reference sphere (Fig. 2.17a).
manual of eld geology (Compton, 1962). For most Only the part of the linear element extending
of these symbols, longer line segments are drawn from C to the point P on the sphere is shown.
parallel to the strike direction so the azimuth can Points such as P are projected onto the equatorial
be determined with reference to the north direc- plane of this sphere and the intersection of the
tion on the map. Shorter line segments indicate sphere and this plane is called the reference circle.
the dip direction and arrows indicate the plunge Points on the reference circle represent the four
direction. Numbers set near the shorter line seg- compass directions (north, east, south, and west),
ments record the dip or plunge angle. The style of and the axis perpendicular to the equatorial plane
the line segments is used to distinguish different intersects the top of the sphere at the zenith, Z. In
structures that are approximated as planar or the view shown in this gure the sphere is rotated
2.3 ORIENTATIONS OF STRUCTURAL ELEMENTS 57
40
Contact, showing dip 90 Strike of vertical bedding;
stratigraphic tops to north (right)
90
Contact, vertical (left) and
63 overturned (right) Horizontal bedding
(a) N
Reference (b)
Projection Z sphere Stereonet
of point
Equatorial
N, y plane Great
circle
W C E, x
P W E
Reference Reference
S o circle
circle 33
Small =
Point, P circle Fp
Linear
element ap = 222o
S
N, y
(c) Z (d)
Equatorial Reference
Vertical circle
plane plane
y g
R
P Plunge C
direction W E, x
C Fp y ap
R P x
P
re ge
n
S
di lun
ct
P
PCP is the plunge, p, and we dene the angle at the center, C, of the transparent reference
CZP as . The distance from the center, C, to the sphere (Fig. 2.18a). For the sake of an example we
projected point is: take the dip direction and dip as (118, 26). The
intersection of the planar element with the
CP R tan (2.67)
sphere is a so-called great circle, because it is a
Because CP CZ R, the triangle CZPC is isosceles circle and because this circle has the largest pos-
and the angle ZPC also is . Therefore, the two sible radius of all those formed by planes of this
angles are related as: orientation intersecting the sphere. In fact it has
the same radius as the reference sphere, R. A
1
45 p (2.68) planar element not passing through the center of
2
the sphere also intersects the sphere to form a
Substituting for in the previous equation we circle, but this is called a small circle, because it has
write the distance of the projected point from the a radius that is less than the radius of the refer-
center as: ence sphere. Only the intersection of the planar
element with the lower hemisphere is drawn and
1
CP R tan 45 p
2 (2.69) a straight line marks the intersection of this
element with the equatorial plane. The stereo-
CP varies from R to 0 as the plunge angle varies graphic projection is constructed by connecting
from 0 to 90. lines of sight from the zenith, Z, to points such as
The next step is to determine the coordinates G on the half great circle. The line ZG intersects
of the projected point, P, relative to a Cartesian the equatorial plane at the point G, which is the
coordinate system with center at C and the x-axis projection of the point G. All possible lines ZG
and y-axis positive toward east and north respec- from the zenith to the half great circle intersect
tively (Fig. 2.17d). The point representing the the equatorial plane along a circular arc, and this
linear element is located along the radial line in arc is the stereographic projection of the half
the plunge direction at the distance CP from the great circle. Note that the radius of this circular
center. In the Cartesian system the coordinates of arc is greater than the radius of the reference
the point P are related to the angle , measured circle, R, unless the dip of the planar element is
counterclockwise from Ox to the line CP, as: zero. In this special case the half great circle rep-
resenting the planar element is coincident with
x CPcos ,y CPsin (2.70)
the reference circle.
Furthermore, the angle is related to the plunge The line of strike, s 28, connects the end
direction, p, as 90 p, so the coordinates of points of the projected half great circle at the ref-
the point P are: erence circle (Fig. 2.18b). The strike direction is
that direction viewed along the line of strike with
x CPsin p,y CPcos p (2.71)
the trace of the circular arc to the right. Any
Utilizing (2.69) to substitute for the distance CP, azimuth, such as the strike direction, s, or the
we have: dip direction, d, is measured from north clock-
wise around the reference circle. The dip angle,
1
x R tan 45 p sin p
2
d, is measured from the reference circle to the
circular arc along a radial line in the equatorial
(2.72) plane that is coincident with the dip direction.
1
y R tan 45 p cos p
2
Thus, planar elements with shallow dips project
as nearly complete half circular arcs lying close to
These are the equations used to plot the projec- the reference circle, whereas steeply dipping
tion of a linear element on a stereonet of radius R, planar elements project as nearly straight lines
given the azimuth of plunge, p, and angle of approaching the line of strike. Constructing the
plunge, p. circular arc by hand is accomplished on a sheet
Next consider a planar element xed in space of transparent material pined through the center
60 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
n
tio
dir rike
ec
St
(a) Zenith, Z (b) N as
Equatorial
plane Projection
Equatorial Projection of half
plane of half great great circle
N circle
e
trik
C
s
W E W C E
of
G
Planar
e
ad
Lin
S element
f
d D
G G
Half great circle dir ip
Sphere ec
tio
n
S
N, y
(c) Z (d) (x, y)
Vertical
plane n Q R'
Q R R Dip C b a
direction d
Q P k R E, x
W
C C h C
D
fd P d ip
Trace of P f ire
Trace of d ctio
projection n
planar
element Projection
Equatorial of half
plane S great circle
Fig 2.18 Stereographic projection of planar element. puter. Consider a vertical plane passing through
(a) Reference sphere. (b) Projection of half great circle onto the center of the sphere, C, and containing the
equatorial plane. (c) Vertical plane passing through center dip direction (Fig. 2.18c). The horizontal line is the
of reference sphere and containing the dip direction.
trace of the projection (equatorial) plane and the
(d) Equatorial plane with projection of whole circle.
line segment PQ is the trace of the planar
element. The point P is the projection of P onto
of the stereonet and marked with the north the equatorial plane from a perspective at the
direction at the top of the net. The overlay is then zenith, Z, and the point Q is the projection of the
rotated counterclockwise through the angle of point Q. The points P and Q lie on the projected
strike and the dip is scaled off from the reference circle and the line segment PQ is a diameter of
circle at east toward the center using the great that circle. An arc of this circle is the stereo-
circles of the net for a scale. The great circle so graphic projection of the half great circle, so we
identied is traced on the overlay and the overlay seek the center and radius of this projected circle
is rotated back to the original orientation to in order to plot it.
produce the graphical representation of the The angle PCP is the dip, d, of the planar
planar element shown in Fig. 2.18b. element and the angle CZP is dened as . By the
To avoid the tedious and inaccurate process of same argument leading to (2.68), the two angles
hand construction we seek analytical expressions are related as 45 d/2. Using this relation
for plotting the circular arc representing a planar and the triangle CZPC, the angle ZPC is related to
element on a stereographic projection by com- the dip as:
2.3 ORIENTATIONS OF STRUCTURAL ELEMENTS 61
1
angle ZPC 90 45 d (2.73) N
2
1
45 d angle CZC (2.74)
2
ad = 154o
Solving for the angle CZC we nd that this angle ap = 197o fd = 58o
must be equal to the angle of dip, d. The S
trigonometry of the right triangle CZCC provides fp = 50o
the distance, CC, from the center of the sphere to
Fig 2.19 Meridional stereographic net with projection of
the center of the projected circle and the radius of
linear element within planar element.
that circle, R:
R
CC R tan d, R (2.75)
cos d
The second step in each of these equations follows
As the dip of the planar element varies from 0 to from the fact that the angle is related to the dip
90, the distance CC varies from 0 to , and the direction, d, as 270 d. Substituting (2.75)
radius of the projected circle varies from R to . and (2.77) in (2.76) we have:
Thus, the projected arc of the half great circle
varies from being coincident with half the refer- x R tan d sin d (R cos d) cos
ence circle to being coincident with the straight (2.78)
y R tan d cos d (R cos d) sin
line of strike.
For plotting purposes a Cartesian coordinate These are the equations used to plot the projected
system is established with origin at the center, C, circle representing the orientation of a planar
of the equatorial plane and the x-axis and y-axis element on a stereonet of radius R, given the
are taken as positive toward east and north, azimuth of dip, d, and the angle of dip, d. To plot
respectively (Fig. 2.18d). The coordinates of points the whole circle one uses the range 0 2. To
on the projected circle are given by: restrict the plot to the circular arc lying within
the reference circle (Fig. 2.18d), in other words the
x h R cos
y k R sin 0 2 (2.76)
projection of the half great circle, the further con-
dition on the coordinates is (x2 y2)1/2 R.
Slickenlines lying in the plane of a fault (Fig.
Here (h, k) are the coordinates of the center, C, of 2.14) are idealized as linear elements contained
the projected circle and R is the radius. The point within planar elements on a stereographic pro-
C is at a distance CC on a radial line oriented at jection (Fig. 2.19). The linear element projects to
180 from the dip direction. The coordinates (h, k) the point P and falls on the great circle repre-
are related to the angle , measured counter- senting the planar element. The rake angle, r, is
clockwise from Ox to this radial line: measured from the point on the reference circle,
representing the strike direction of the planar
h CCcos CCsin d
element, along the great circle to the point P. The
(2.77)
k CCsin CCcos d example shown in Fig. 2.19 is for a linear element
62 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
N
Stereonet
ap = 298o
fp = 64o
P
W 90 o E
Q
ad = 118o
fd = 26o
S
Fig 2.20 Meridional stereographic net with projection of
linear element that is perpendicular to a planar element.
Rv/v
Equatorial Z, z
plane v
P
W
S C
P N, y
E, x
Reference
sphere
A
AP 2R sin (2.86)
(c) Sphere
Equatorial
It is a property of the Lambert equal area projec- plane
tion that the distance AP from the antipode to the
projected point, P, is equal to the distance AP.
Substituting for the angle using (2.68) we relate
C N, y
Q
this distance to the plunge angle:
xl
1
AP AP 2R sin 45 p
2 (2.87) w
Q(x, y, z)
z
Geometrically this step in the Lambert projection
can be accomplished by turning a circle with Fig 2.25 Lambert equal area projection of a linear
element. (a) Vertical plane containing the linear element.
center at A through the point in question onto the
(b) Equatorial projection plane. (c) Reference sphere.
projection plane (Ragan, 1985, p. 273).
66 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
The next step in the projection procedure con- An arbitrary point, Q (x, y, z), at the intersection of
siders the projection plane itself (Fig. 2.25b). a particular meridian and parallel corresponds to
Linear elements with zero plunge project onto the a linear element through the center of the sphere.
reference circle in this view, so (2.87) gives the This point projects onto the horizontal (x, y)-plane
radius of the reference circle as: at Q and the distance CQ (x2 y2)1/2. Further-
more, the cosine of the plunge angle, p, is CQ/R.
R 2R sin (45) R 2 (2.88) Using these relationships as shown in Fig. 2.25c,
the plunge direction and plunge are:
The arbitrarily oriented point P is associated with
a plunge direction, p, measured clockwise from 1
p tan 1( yx),
north. Taking a Cartesian coordinate system with 2
x- and y-axes coincident with east and north, p cos 1[(x2 y2)12R] (2.92)
respectively, the coordinates of the point P are:
Given the longitude and latitude of points on a par-
x APsin p, y APcos p (2.89) ticular meridian or parallel, the rst set of equa-
tions establishes the Cartesian coordinates and the
It is convenient to scale the reference circle of the second provides the plunge direction and plunge.
equal area projection so its radius, R, is equal to These two angles are used in (2.90) to project and
the radius of the reference sphere, R. This is plot the points. The longitude intervals between
accomplished by multiplying AP by 12 2. successive meridians is arbitrary and is taken as
Substituting for AP from (2.87) and scaling the 10 for the construction of Fig. 2.23. The range of
radius we have: longitudes is 0 180 to cover the lower hemi-
sphere. Similarly the latitude intervals between
1
x R 2 sin 45 p sin p
2
successive parallels is taken as 10, and the
range of latitudes is 90 90.
(2.90) Given the equations to construct the Schmidt
1
y R 2 sin 45 p cos p
2
net and to plot points representing linear ele-
ments on a fabric diagram, the most important
These are the coordinates of a point on the question is whether or not the distribution of
Schmidt net of radius R representing the orienta- points has a statistically signicant preferred ori-
tion of a linear element with plunge direction, p, entation (Kamb, 1959a). This question is addressed
and plunge angle, p. For foliations or other graphically by relating the area of the counter
planar fabrics the azimuth and plunge of the used in the construction of contours on the
normal, n and n, are substituted in these equa- diagram to the total number of points, N, in the
tions to prepare the fabric diagram. population. The counter is a circle with area Ac
The Schmidt net itself is plotted as the merid- that is some fraction, 0 f 1, of the total area An
ians and parallels of a sphere of radius R, oriented of the Schmidt net:
such that the poles, north and south, are on the
Ac r2
horizontal y-axis (Fig. 2.25c). Note that this illus- f (2.93)
An R2
tration of the sphere is rotated slightly about the
vertical z-axis so the equatorial plane is visible. The center of the counting circle is positioned at
Longitude angles, , are measured from the hori- every intersection of an r by r square grid laid over
zontal x-axis around the equatorial perimeter and the projection. The number of points within the
latitude angles, , are measured along a meridian counter is recorded for each intersection and
from the equator toward the poles. The equations these numbers are contoured. What distinguishes
relating longitude and latitude to the Cartesian this method from those mentioned earlier is the
coordinates are: choice of the radius, r, of the counting circle.
To understand how r is determined consider a
x R cos cos , y R sin , z R cos sin (2.91) set of N points that have statistically uniform ori-
2.3 ORIENTATIONS OF STRUCTURAL ELEMENTS 67
3R
entations over the entire net. In other words this r (2.96)
N 9
population has no preferred orientation. As the
counter is randomly positioned on the net the Given the radius of the Schmidt net, R, and the
number of points within the counter, n, varies. number of points, N, in the population (2.96) pro-
The distribution of n is a binomial distribution vides the radius of the counter and the distance
because the counting circle divides the popula- between the intersection points of the square grid
tion into two mutually exclusive sets: those inside that overlays the net. Using the counter and grid
the counter and those outside. On average the so dened one can construct the orientation
number of points in the counter will be fN density diagram to display graphically the statis-
because the area of the counter is a fraction f of tical signicance of a data set containing the ori-
the area of the net. The mean, m, and standard entations of a linear fabric or the normals to a
deviation, s, of this binomial distribution are planar fabric.
(Krumbein and Graybill, 1965, p. 102):
2.3.4 Field and model angles and
m f N, s [ f N(1 f )]12 (2.94)
analytical orientation statistics
Recall that the standard deviation is a measure of We turn now to the relationships between geo-
the spread of the distribution about the mean. As graphic angles and coordinates, and the angles and
the counter gets very small, f 0, m 0, and s coordinates used in data analysis and model con-
( fN)1/2. As the counter approaches the size of the struction. Recall that the orientations of planar
net, f 1, m N, and s 0. and linear elements that approximate geological
It is recommended (Kamb, 1959a) that the structures are measured in the eld using two geo-
radius of the counter be chosen such that m 3s. graphic angles, the azimuth, , and the inclina-
In other words, for the population with no pre- tion, . Different terms are associated with the
ferred orientation, the number of points within azimuth of strike (s), dip (d), plunge (p), and
the counter, on average, would be three times the normal (n), and appropriate subscripts distin-
standard deviation. The fabric diagrams can be guish these. Similarly, different subscripts for the
contoured at values of 0, 2s, 4s, 6s, etc. Contours inclination angle, , distinguish the dip (d),
drawn from counts using this prescription are plunge (p), and plunge of the normal (n). For any
very smooth. If such large counts cluster in one planar element the azimuth and plunge of the
region of the fabric diagram and produce closed normal line (pole) may be used to specify the orien-
contours with values greater than 3s, one can tation, so in fact we only need to consider how line
interpret the population as having a preferred ori- segments are oriented in three-dimensional space
entation. Substituting the expressions for the to account for the orientations of all linear and
mean and standard deviation from (2.94) into the planar elements that approximate geological struc-
condition m 3s and solving for f, we nd: tures. For data analysis and model computations it
is convenient to describe the orientation of any line
9
f (2.95) segment using three direction angles (x, y, z) that
N9
relate the line to a Cartesian coordinate system,
Note that a 1% counter area, sometimes chosen rather than the geographic system. Here we intro-
arbitrarily for the contouring of fabric diagrams, duce the relationships that transform eld data in
corresponds to N 891. This is an unusually large the geographic coordinate system to a Cartesian
number of points for fabric studies. Implemen- system using these direction angles. The textbook
tations using a 1% area and fewer than 891 points by Groshong (1999) provides additional discussion
are likely to produce irregular contours that have of these and other techniques of three-dimensional
no statistical signicance because the counter geometry as used in structural geology.
area is too small. Consider the orientation of the line segment OP
Substituting the ratio of areas for f in the pre- relative to the orthogonal geographic coordinate
vious equation, and solving for the radius of the system composed of the axes east, north, and up
counter, r, we have: (Fig. 2.26a). Regardless of their specic geological
68 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
OA OB OC
0 2, 0 (2.97) cos x , cos y , cos z (2.100)
2 OP OP OP
It is interesting to note that mining engineers These are the direction cosines for the line OP with
dene the normal to planar elements as the respect to the Cartesian coordinate system.
2.4 STRUCTURAL MAPPING USING GPS TECHNOLOGY 69
The transformation equations are constructed Note that the magnitude of the resultant
from the preceding trigonometric identities as vector does not have a unit value. The components
follows: of the resultant vector are found as the sums of
the respective components of the set of unit
OA OD
cos x sin cos vectors:
OD OP
OB OD n n n
cos y cos cos (2.101)
OD OP Ux u (i), U u (i), U u (i)
i1
x y
i1
y z
i1
z (2.105)
cos z
OC
OP
cos sin
2 The direction cosines of the resultant vector are
given by the ratios of the components to the mag-
In this way the direction cosines for the line OP are nitude of this vector:
calculated from the azimuth and plunge of the
linear element. The azimuth and plunge are cos x Ux U, cos y Uy U,
recovered from the direction cosines using: (2.106)
cos z Uz U
tan1 cos x
cos y, sin1[cos z] (2.102) The azimuth and plunge of the resultant vector
(mean direction) are found using (2.102). If the
To compute the full range for the azimuth, the unit vectors representing the direction data are
signs of both the numerator and the denominator widely scattered the magnitude U is small com-
in the arctangent function must be used. Most pared to n, whereas for tightly clustered data the
computer languages offer a function such as magnitude of U approaches n. The spherical vari-
ATAN2(XNUM, YNUM) that explicitly uses the two ance is dened (Davis, 1986, p. 334):
arguments with their signs.
The relationships we have just derived are used s 2s (n U)n (2.107)
to develop an analytical method for determining
the mean direction for a set of n linear elements This is a measure of the clustering of the direction
or normals to planar elements (Davis, 1986). data about the mean.
Consider each member of the set to be a unit vector,
u(i), where i 1 to n. For example, the line segment
OP (Fig. 2.26) could represent one such unit vector:
2.4 Structural mapping using GPS
u(i) ux(i)ex uy(i)ey uz(i)ez technology
(2.103)
u(i) [u2x (i) u2y (i) u2z (i)]12 1
2.4.1 The Chimney Rock fault array
The components of any vector are equal to the The Chimney Rock fault array crops out on the
vector magnitude times the respective direction northern San Rafael Swell (Fig. 2.27) and is
cosine (2.8), so in the case of a unit vector the com- exposed over an area of about 25 km2 where the
ponents are the direction cosines. Using this rela- local stratigraphy (Fig. 2.28) is composed of the
tionship and (2.101) the components of the unit Jurassic Navajo Sandstone and overlying Carmel
vector may be related to the azimuth, (i), and Formation (Maerten, 2000; Maerten et al., 2001;
plunge, (i), of the line element or normal. Davatzes and Aydin, 2003). The lower Carmel is
The mean direction for a set of linear elements predominantly shale, sandy shale, and limestone
or normals taken as unit vectors is dened as the beds. The top of the Navajo and three resistant
direction of the resultant vector, U: limestone layers in the lower Carmel provided
excellent marker horizons for mapping in this
U Uxex Uyey Uzez
(2.104) region and for determining the location, orienta-
U [U x2 U y2 U z2]12 tion, and offset on the faults. The traces of the
70 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
meters
0 15
1
Chimney Rock Alternating gray
2
00
fault array limestone and shale
9
0
30
90 0
00
60
Blue-gray limestone
0
0 0 (mapped layer)
12
60
0
30
1500 0 Blue sandy shale
0
Green
1800
30 River Yellowish calcareous
39o 0 sandstone
Sandy shale
0
210
600
e ll
Sw
el
Yellowish sandstone
a fa
fold
nR 900
5
Sa
ket
t
Wa San Rafael Alternating yellowish
limestone and shale
1200
Swell
Reddish limestone
Reddish shale and
111o sandstone alternating
with limestone beds
Henry Mountains Basin 0 10 20 30 0 Unconformity,
km weathered surface
Top Navajo
Sandstone
Fig 2.27 Structure contour map of San Rafael Swell region Cross-laminated
of south central Utah (Maerten, 2000). Contour interval is aeolian sandstone
300feet. Location of Chimney Rock fault array indicated with iron oxide (reddish)
within rectangle.
540 000 541 000 542 000 543 000 544 000 545 000
4 345 000
N
4 345 000
4 344 000
3o
1680
1660
4 344 000
1640
1700
fault
North
0
1700
0
1680
172
164
166
4 343 000
1720
4 343 000
Gla
1620
ss F
1700
ault
1680
0
0
166
164
170
0
66
rt F ault1
90
Sho
1620
4 342 000
16
4 342 000
1720
1660
1680
00
17
4341000
4 341 000
1640
Lit
tl eF
au
lt
10
1700
to the north
1660
er r y
1680
4 340 000
4 340 000
0
168
40
17
20
0
166
17
F re
00 nch
1640
17 ma
La S nF
al au
0
lt
0
168
Fa
166
ult
4 339 000
4 339 000
tion from the map that indicates it is in the north- The strike of the blue-gray limestone is every-
ern hemisphere, Zone M12, and the datum is where parallel to the structure contours so the
72 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
60 60
La Sal Little
Blueberry Fault Fault Fault
50 70
Rake ()
40 80o
Dip slip (m)
NE SW
30 90
20 100
?
10 110
0 120
0 500 1000 1500 2000 2500 3000
Distance from eastern tip (m)
Fig 2.31 Distribution of dip slip magnitude and rake of
ments in the TDC1TM data collector running Asset
slickenlines on Blueberry Fault of the Chimney Rock array
(Maerten, 2000).
SurveyorTM software that prompts the user for
particular field measurements using a data dic-
tionary. The receiver and data collector are
These data demonstrate that the Chimney carried in the field in a small backpack (fron-
Rock faults have a systematic variation of slip tispiece, Chapter 2) and the data are later down-
magnitude and direction and that the intersec- loaded to a laptop computer running the
tions of faults are associated with discontinuities PathFinder OfficeTM software. This software is
in both magnitude and direction. While one used to create a custom data dictionary, visualize
could characterize the Blueberry Fault as a the map data, process the coordinate data using a
normal fault with about 30 m of dip slip, this technique called differential correction, and
would not reveal the interesting slip variations export the final data to other software or to a
that provide important insights regarding the printer. Coordinate data can be collected at preset
mechanical behavior of this fault and its neigh- time intervals as the geologist walks throughout
bors. The data necessary to quantify these slip the field area, or at specific sites where particular
variations are obtainable using GPS technology. structures crop out.
The Asset SurveyorTM software enables the
2.4.2 GPS technology and mapping structural geologist to create a data dictionary
techniques specifically tailored to the structures and terrain
The combination of fault traces and structure in the region being mapped. In terms of recording
contours on Fig. 2.29 provide considerable insight quantitative data, keyed to locations, this is a
about the geometry of the structures in the significant improvement on the traditional geolo-
Chimney Rock region. The construction of this gist notebook. For this region the basic features
map was facilitated by a sub-meter precision GPS were the four distinctive sedimentary layers and
receiver combined with a data collector and a the faults, so a data dictionary could contain the
laptop computer (Maerten et al., 2001). This is a entries shown in Table 2.2 (Maerten et al., 2001).
remarkable tool for modern structural investiga- Each feature can be recorded at a particular site as
tions. Specifically, the ProXL SystemTM manufac- point data or at many sites along a traverse as
tured by Trimble Navigation Limited was used to line data. For example, as the geologist walks
receive the GPS signals and process the coordi- along the ledge (Fig. 2.28) formed by one of the
nate data to determine locations to within less resistant limestone layers (C1, C2, C3) or the top of
than 1 m, given sufficient satellite signals. These the Navajo Formation (Nav), the receiver can
data were combined with other field measure- collect location coordinates every few steps,
74 STRUCTURAL MAPPING TECHNIQUES AND TOOLS
Table 2.2. Data dictionary. to construct the map shown in Fig. 2.29 in fteen
days using the GPS system (Maerten et al., 2001). A
Features Attributes Options total of 14 863 locations (easting, northing, eleva-
tion) were collected to dene the tops of the four
Layer (point) Type C1, C2, C3, Nav sedimentary layers. The thickness of the local
Strike 0 to 360 stratigraphic section (Fig. 2.28) was used to extrap-
Dip 0 to 90 olate from a given measurement to the elevation
comments of the top of the blue-gray limestone at each loca-
Layer (line) Type C1, C2, C3, Nav tion. These combined data were used to construct
comment the structure contour map. In addition, a total of
Fault (point) Strike 0 to 360 1768 locations were collected to dene the traces
Dip 0 to 90 of the faults, and the orientations of these faults
Rake of 0 to 180 were measured at 537 locations. By precisely and
slickenline efciently digitizing the three-dimensional geom-
Surface Good, etry of the deformed layers and the faults in this
quality medium, region, the GPS mapping technology provides an
poor effective mapping tool for the structural geologist.
Fault offset 0 m to 100 m
Faulted layer C1, C2, C3, Nav
comments 2.5 Concluding remarks
Fault (line) Faulted layer C1, C2, C3, Nav
comments A number of new technologies are transforming
structural mapping. These include the Global
Positioning System (GPS), used to locate outcrops
creating a very dense data set along the traverse. and structures on Earths surface more rapidly and
For each feature there are a number of common precisely than one could have imagined when the
attributes such as the type of layer (C1 or Nav), authors of this textbook began their careers. High-
the orientation (strike and dip), and comments. resolution aerial photography and scanning tech-
The options determine such things as the names nologies enable one to create digital images with
of the layer types, or the magnitude of the offset pixel widths of a few centimeters on the ground.
on a fault. At particular sites, the orientation of a Airbourne Laser Swath Mapping (ALSM) produces
layer or a fault can be measured using a compass topographic maps with decimeter precision. These
and inclinometer, and these numbers are recorded technologies are remarkable and they are rapidly
as the strike and dip angles. The denitions of evolving. There is no doubt that this revolution in
these angles were given earlier in this chapter. our ability to image Earths surface will usher in a
A single geologist collected the data necessary new generation of structural maps and data.
Chapter 3
I
n the previous chapter we illustrated examples tations. It may be clear that a particular lineation
of geological surfaces, such as the top of the has a different plunge direction and plunge at dif-
Triassic Chinle Formation throughout the San ferent exposures, but how does the attitude vary
Rafael Swell in southern Utah, using structure from one location to the next? The lineation may
contours (Fig. 2.27). The more detailed shape of approximate a three-dimensional curve, so we
the top of the blue-gray limestone bed near the need to understand how to describe the shape of
bottom of the Carmel Formation in the Chimney such curves. The spatial variation of plunge direc-
Rock area is shown by the structure contours in tion and plunge can be inferred qualitatively from
Fig. 2.29. The frontispiece of this chapter shows the distribution of attitude symbols on a struc-
two aerial photographs of the exposed surfaces of tural map, but differential geometry provides the
Jurassic sandstone formations on the ank of the tools for the quantication and analysis of these
Waterpocket Fold that denes the southeastern spatial variations.
margin of the San Rafael Swell. Notice how the Relatively little use of differential geometry is
sandstone formations bend over the monoclinal found in the twentieth-century literature of
exure and also bend as the strike of the beds structural geology, despite the obvious need to
changes along the length of the fold. Monoclinal describe the complex shapes of curved lineations
exures are a continuing focus of structural inves- and surfaces, and the attractiveness of accom-
tigation in this region (Reches, 1978; Reches and plishing this in a quantitative manner (Mallet,
Johnson, 1978; Cooke et al., 2000; Johnson and 2002). This literature provides few examples that
Johnson, 2000). Clearly these surfaces are not we can use to illustrate the concepts introduced
planar, but what is their shape? A primary task for here. Furthermore, eld techniques for deducing
structural geologists is to describe and character- the shapes of curves or surfaces from scattered
ize such surfaces and this may be accomplished in location and orientation data are just now being
a mathematically rigorous and complete manner devised and tested. Therefore this chapter focuses
using concepts and tools from differential geometry, on the principles and methods of differential
the branch of mathematics that brings the power geometry that appear to have the greatest poten-
of vector calculus to geometry (Gauss, 1827). Here tial for application to structural geology. This
we review some of the elementary concepts of dif- introduction is meant to encourage the use of
ferential geometry that are helpful to quantify these principles and methods and thereby
the departure of geological surfaces from a plane provide, in the words of Reveil Netz (2000), good
(Mallet, 2002). starting points for truly productive discussion of
Structural data typically are gathered at scat- the geometry of geological structures.
tered exposures as point measurements and the Position vectors are used to describe points,
locations of these points should be identied curves, and surfaces in differential geometry.
using geographic or local coordinates and position Many other vector quantities, such as the tangent
vectors. The measured data include the local atti- and curvature vector, are derived from the posi-
tudes of planar and linear elements that approxi- tion vector and used extensively in this chapter.
mate, for example, a foliation (Figs. 2.24ac) or Therefore it is necessary to understand the basic
lineation (Figs. 2.24df) at the point of measure- concepts of vectors, and to be familiar with
ment (Cloos, 1946; Turner and Weiss, 1963). specic techniques for manipulating vectors as
Plotting the attitudes of a set of structural ele- introduced in the previous chapter before reading
ments on a stereographic projection enables one this chapter. For in-depth treatments of differen-
to compare the orientations of different members tial geometry that provide a rigorous mathemati-
of the set. While serving a useful purpose in their cal basis, the reader is referred to textbooks on the
own right, stereographic projections provide subject (Struik, 1961; Stoker, 1969). In particular,
an incomplete characterization of foliations and the book by Lipschutz (1969) is a source for much
lineations, because these projections lack any of the material in this chapter and provides many
information about the spatial variations of orien- useful exercises and worked examples.
3.1 THE CONCEPT AND DESCRIPTION OF LINEATIONS 77
(a) z (a) z
y
Arc of
circle
Fo
ld a
xis
Lineation
t = 2p
c y x
2p|b |
t=0
Cylindrically
folded surface
x |a |
(b)
(b) y z y
a
Arc of
circular helix
Fo
c ld a
xis
s Lineation
t
x x
t=0
Cylindrically
folded surface
Fig 3.5 (a) Circular helix, defined by position vector c, Fig 3.6 (a) Lineations on cylindrical fold with circular
with radius a and pitch b. (b) Special case of a circle where profile shape lie on arcs of circles. (b) Lineations oblique to
b 0. Arbitrary parameter is t; arc length is s. fold axis lie on arcs of a circular helix.
composed of two halves of a circular cylinder eations where they are continuous. Can the arc of
joined along the dashed line. This is just one of a circle or circular helix approximate a given lin-
the many possible shapes for a cylindrical fold. For eation, or is a more complex shape required? In
this special case, lineations that are perpendicu- part our inability to address this question is due
lar to the fold axis approximate the arc of a circle. to poor exposure and therefore inadequate data,
Lineations that are not perpendicular to the fold but in part it is due to the lack of quantitative
axis may approximate the arc of a circular helix tools to describe and analyze curved lines in three-
(Fig. 3.6b). Thus even these elementary examples dimensional space. Differential geometry pro-
of curved lines from differential geometry may vides these tools.
have application to geological lineations.
The lineations illustrated in Fig. 3.6 are shown 3.1.3 The unit tangent vector
as short tick marks on the folded surface. This is Recall that the local orientation of a curvilinear
in keeping with the fact that eld measurements structure at the exposure is measured as the ori-
usually are limited to orientations determined at entation of the line element that is tangential to
isolated exposures. Not much is known about the the lineation. It should come as no surprise, then,
continuity of lineations from exposure to expo- that we make use of the unit tangent vector, t, along
sure, or about the three-dimensional shapes of lin- a curve. Some care is needed to distinguish the
82 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
t
dc dc dt dc ds dc
ds dt ds dt dt dt | |
dc
dt
(3.8) c
y
This is a more general equation for the unit
tangent vector because it does not depend upon
the parameter being the arc length. Recalling x
that c is a vector function of a single real vari-
(b) y
able, t or s, how does one take the derivative of
such a function? The answer is that one takes the
t=/2
derivative of each component with respect to the t=-ex
variable and uses these as the components of a
new vector. For example, given c(t) as in (3.1), t=+ey
then: k= -e y /a
dc(t) dcx(t) dcy(t) dc (t) k=-ex/a x
e e z ez (3.9) t=0
dt dt x dt y dt
In the last step we used the fact that both vectors 3.1.4 The curvature vector and the
have unit magnitude, so the product of their scalar curvature
absolute values is one. Solving for the angle : Although the unit tangent vector is an important
geometric quantity for the characterization of
cos 1(t ez) cos 1[b(a2 b2)12] constant curves it is not one of the two fundamental prop-
(3.15) erties that uniquely determine the shape of a
curve. The rst of these fundamental geometric
As b goes to zero the helix collapses into a circle quantities is the curvature. The curvature vector, k,
on the (x, y)-plane and the angle goes to /2. This is dened for a natural representation of a curve,
special relationship is illustrated in Fig. 3.8b c(s), as the derivative of the unit tangent vector
where it is seen that t is perpendicular to the z- with respect to the natural parameter s. This deriv-
axis for all values of the arbitrary parameter t. ative is dened using the standard limiting pro-
For a lineation observed at exposure, such as cedure from calculus (Lipschutz, 1969, p. 62):
the slickenlines in Fig. 3.3, the orientation deter-
mined in the eld using geographic angles can be t(s s) t(s) dt
lim k(s) (3.17)
related to the unit tangent vector, t. Although s0 s ds
slickenlines may be curved, they are approxi- This denition, and the earlier denition of the
mated locally with linear elements whose orien- unit tangent vector (3.5), imply that the curvature
tation is measured using the plunge direction, p, vector is equivalent to the second derivative of
and plunge, p. We assert that such a linear the vector function c with respect to the natural
element is parallel to the unit tangent vector of a parameter: k(s) d2c/ds2. Therefore, denition of
three-dimensional curve that passes through the the curvature vector requires that the curve, c(s),
point of measurement, so it has the same direc- have a continuous second derivative over the
tion angles. Furthermore, because t is a unit interval of interest. In general, the curvature
vector, the scalar components are equivalent to vector is directed away from the curve on its
the direction cosines. We nd using (2.101): concave side (Fig. 3.8). In other words the curva-
ture vector points in the direction that the curve
tx cos x sin pcos p
is turning.
ty cos y cos pcos p (3.16)
The specic orientation of the curvature
tz cos z sin p
vector is determined by the fact that the tangent
In this way we relate the orientation data taken at vector is constant in magnitude; in fact it is a unit
scattered exposures on a continuous geological lin- vector. In general, if v is an arbitrary vector func-
eation to the components of the unit tangent tion such that |v| constant (not necessarily a
vectors at correlative points on a three-dimensional unit vector), then from the properties of the scalar
curve. product we have v v |v||v| cos 0 constant.
As an example, consider the lineation Differentiating this scalar product using the stan-
described by the plunge direction, p 222, and dard product rule (Selby, 1975):
plunge, p 33 (Fig. 2.17). Substituting these
dv dv dv
values into (3.16) we have tx 0.561, ty 0.623, v v 0,so v 0 (3.18)
dt dt dt
tz 0.545 and |t| 1. Recall that the geo-
graphic coordinates (east, north, up) correspond Given that neither v nor dv/dt is zero, a zero scalar
to the Cartesian coordinates (x, y, z), as shown in product implies that cos 0 and /2. Thus,
Fig. 2.15, to reconcile the three negative compo- the curvature vector is orthogonal to the unit
nents of the tangent vector with the orientation tangent vector.
of this lineation as plotted on the stereogram. Consider Fig. 3.9a to understand how the
Also note that these components combine to give magnitude of the curvature vector is related to
a unit magnitude, as expected for the tangent the change in orientation of the tangent vector
vector. with respect to position along a curve (Lipschutz,
3.1 THE CONCEPT AND DESCRIPTION OF LINEATIONS 85
z
| lim
s0
t(s s) t(s)
s | d
lim (s)
s0 s ds
c(s) (3.20)
c(s + s)
t(s + s) The left-hand side of this equation is the magni-
tude of the curvature vector (3.17), |k(s)| (s), a
y quantity known as the scalar curvature. From this
x relationship one can understand that the scalar
curvature is equivalent to the spatial rate of
(b) change of the orientation of the unit tangent
t(s)
vector with arc length along the curve. The scalar
curvature, (s), is called an intrinsic property of a
u curve (Lipschutz, 1969) because it is one of two
t(s + s) t(s) quantities that uniquely denes the shape of a
t (s + s) curve. Where the orientation of the unit tangent
vector changes more rapidly with position along
the curve, the curvature is greater. A point on the
curve where the curvature is zero is called an
Fig 3.9 Diagrams to define scalar curvature, . (a) Two inection point.
tangent vectors, t(s s) and t(s), on curve separated by arc For an arbitrary parametric representation of
length s. (b) Change in angle, , between two tangent a curve, c(t), the curvature vector and the scalar
vectors with respect to arc length s defines scalar curvature curvature are calculated using (Lipschutz, 1969,
in limit as arc length goes to zero. p. 65):
(3.21)
are shown with different orientations at their
respective points on the curve, separated by some That is, one rst calculates the unit tangent vector
small (differential) arc length s. Bringing the using (3.8) and then calculates the derivative of
tails of these two unit vectors together forms an that vector function. To calculate the scalar cur-
isosceles triangle (Fig. 3.9b) with sides of unit vature one takes the absolute value of the curva-
length, a base of length |t(s s) t(s)|, and a dif- ture vector.
ferential angle between the sides, . The general To calculate the curvature vector and the
relationship among the sides of equal length, a, scalar curvature for the circular helix (3.2) we take
the base, b, and the angle between the sides, , the derivative of the unit tangent vector (3.11) and
for an isosceles triangle is b/a 2 sin(/2). use (3.21) to nd:
Substituting b |t(ss) t(s)| and a 1 we
have, for small angles: k(t) a(a2 b2)1[(cos t)ex (sin t)ey],
a(a2 b2)1 (3.22)
|t(s s) t(s) | 2 sin (2) (3.19)
Note that the curvature vector lies in the (x, y)-
Here we have used the rst term of the series plane (there is no component in the z-direction), it
expansion for the sine function (Selby, 1975, is orthogonal to the tangent vector, and it points
86 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
t 0 2 32
(3.24) (b) y t = -0.5
k ex a ey a ex a ey a 7 = 0.707
= 1.414
This curvature vector is orthogonal to the circle
and directed toward the center of the circle (Fig. 6
3.8b). It is not a unit vector unless a 1. The scalar
curvature is inversely proportional to the radius
of the circle. 5
A positive scalar quantity called the radius of t =0
curvature, , is motivated by (3.23) and is dened =2
for an arbitrary parametric representation of a = 0.5
4
curve, c(t), as the reciprocal of the scalar curvature
(Lipschutz, 1969, p. 63):
1 3
(t) (3.25)
(t)
curve are identical at the point c(s). The circle C is The magnitude of the cross product is propor-
the one circle, of an innite number of different tional to the magnitudes of the two vectors and
circles that pass through this point, that has the the sine of the smaller angle, , between these
closest contact with the curve, and therefore pro- vectors measured in the plane that they dene:
vides a geometric visualization of the radius of
curvature. One can say that it is the best-tting |v w| |v||w|sin ,0 (3.29)
circle to the curve at that point. These general relationships for the cross (or
For the natural representation of a curve, c(s), vector) product are used extensively in differen-
recall that the curvature vector is the second tial geometry and are used here to calculate the
derivative of the position vector: k(s) d2c/ds2, so scalar curvature.
the scalar curvature is dened as (s) |d2c/ds2|. Again taking the circular helix (3.2) as an
Given this denition, one might be tempted to example we use (3.26) to calculate the scalar cur-
associate curvature with the second derivative of vature. The second derivative of the vector func-
a function, say y f(x), representing a plane curve. tion c(t) for the helix is:
In fact, this is commonly done in casual conversa-
tions where one says the rst derivative, dy/dx, is d2c
a(cos t)ex a(sin t)ey (3.30)
the slope of the curve, and the second derivative, dt2
d2y/dx2, is the curvature. However, such state-
The absolute value of the cross product of the two
ments involve special conditions and approxima-
derivatives is found using (3.27):
tions that are spelled out in the following
paragraphs. dc d2c
ab( sin t)ex ab( cos t)ey a2ez
If c c(t) is the arbitrary parametric represen- dt dt2
tation of a curve, as in (3.1), then a general
denition of the scalar curvature is (Lipschutz, |dc d2c
dt dt2 |
a(a2 b2)12 (3.31)
1969, p. 64):
Then the quotient in (3.26) is:
| || |
3
dc d2c dc
(t) (3.26) a(a2 b2)12
dt dt2 dt (t) a(a2 b2)1 (3.32)
[(a2 b2)12]3
When a curve is dened using a parameter other
then the arc length, s, the scalar curvature is not This is exactly the result found in (3.22): the scalar
simply the absolute value of the second derivative curvature of the circular helix is a constant
of the vector function c(t). Compare (3.26) to the depending upon both the radius and the pitch.
earlier denition of the scalar curvature (3.21): The general equation for the scalar curvature
both are useful denitions. (3.26) may be specialized for a plane curve, for
To calculate the scalar curvature using (3.26) example a curve that lies entirely within the (x, y)-
one uses the cross product of two vectors, say v and plane:
w, found from the components of these vectors as
c(t) cx(t)ex cy(t)ey (3.33)
(Selby, 1975, p. 556):
Here the vector function c(t) lacks any component
v w (vywz vzwy)ex (vzwx vxwz)ey
along the z-axis and the two non-zero components
(vxwy vywx)ez (3.27)
are functions of the arbitrary parameter, t.
The cross (or vector) product also may be evalu- Substituting into (3.26) we nd the scalar curva-
ated as the determinant of the matrix formed by ture for this parametric representation of the
the base vectors and components as follows: plane curve (Varberg and Purcell, 1992, p. 623):
ex vx wx
v w det ey vy wy (3.28) (t) | dcx d2cy dcy d2cx
dt dt2 dt dt2 |
dcx
dt
2
dcy
dt
2 32
ez vz wz (3.34)
88 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
Again we see that the curvature is not simply (x) 2(4x2 8x 5)32 (3.40)
related to the second derivative of the vector func-
For x t1, this is identical to the curvature of
tion c(t).
the parabola found in (3.36) using the parametric
The circular helix is not a planar curve, so we
representation as a function of t.
turn to the parametric representation of a para-
If the squared slope of the function, y f(x) in
bolic curve that lies entirely in the (x, y)-plane (Fig.
the denominator of (3.38) is small compared to
3.10b):
one, then the curvature may be approximated as:
c(t) (t 1)ex (t2 3)ey (3.35)
| |
2
2 y dy (3.41)
Taking the rst and second derivatives of the com- (x) ,for 1
x2 dx
ponents, the scalar curvature is found using
(3.34): For the parabola of Fig. 3.10b this approximation
|2 0| 2 gives a constant value, 2, for all x. This is exact
(t) 2 (3.36)
[(1) (2t)2]32 (1 4t2)32 at the base of the parabola where x 1 and the
slope is zero, but is in error by 182% where
For t 0.5 we nd 0.707 so 1.414 and a
0.707 at x 0.5 and the slope is 45o. For plane
circle of this radius is shown tangent to the
curves described by functions of the form y f(x),
parabola at this point in Fig. 3.10b. For t 0 the
with squared slopes that are not small compared
scalar curvature 2 and the radius of curvature
to one, (3.38) is the appropriate equation for scalar
is 0.5. A circle with this radius of curvature is
curvature. For the general parametric representa-
shown tangent to the point at the base of the
tion of a plane curve (3.34) is the appropriate equa-
parabola. Clearly there is a signicant change in
tion. For the general parametric representation of
curvature along a relatively short arc of the
a curve that is not conned to a plane (3.26) must
parabola.
be used.
We further specialize the representation of
the curve such that cx(x) x and cy(x) y. In other
words the arbitrary parameter in (3.33) is the x-
3.1.5 The unit principal normal vector
coordinate, and the derivatives of the two compo-
and binormal vector
We have already mentioned that the curvature
nents of c may be rewritten:
vector is not generally a unit vector, that it is
dcx dx d2c dcy dy d2cy d2y orthogonal to the unit tangent vector, and that it
1, 2x 0, , 2 2
dt dx dt dt dx dt dx is directed away from the curve on the concave
(3.37) side. As the curve c(s) passes through an inection
point (Fig. 3.11a), the curvature vector, k(s), goes to
Upon substitution into (3.34) the curvature takes zero magnitude and thereafter switches direction
the form often introduced in calculus textbooks to the other side of the curve. In the interest of
(Varberg and Purcell, 1992, p. 623): working with a geometric quantity that is less
erratic in both magnitude and direction, a unit
2 32
(x) | |
d2y
dx2
1
dy
dx
(3.38) vector is dened as parallel to the curvature
vector, but directed to remain continuous along
the curve wherever possible (Fig. 3.11b):
Again note that the scalar curvature is not simply
the absolute value of the second derivative of the k(s)
n(s) (3.42)
function y f(x). The parabolic plane curve illus- |k (s)|
trated in Fig. 3.10b may be written in this form using This vector is called the unit principal normal vector
x t1 and substituting for t in y t2 3 to nd: for the natural representation of the curve c(s)
(Lipschutz, 1969, p. 64). The choice of sign in the
y x2 2x 4 (3.39)
numerator is used to keep this normal vector from
The curvature for this function is found using switching direction arbitrarily from one side of
(3.38) as: the curve to the other at points of inection. For
3.1 THE CONCEPT AND DESCRIPTION OF LINEATIONS 89
(a) (a) z
z
b(t)
Inflection
y point n(t) t(t)
x
k(s)
c(t)
c(s) y
(b)
z x
(b)
y b(s)
x n(s) n(s)
Inflection
point t(s)
db/ds
c(s)
Fig 3.11 Diagrams to define unit principal normal z
(db/ds) n
vector, n. (a) Curve with curvature vectors, k(s). (b) Same c(s)
curve with unit principal normal vectors, n(s). y
x
example, in Fig. 3.11b the positive sign is used to
Fig 3.12 (a) Circular helix with moving trihedron defined
the left of the inection point so n(s) and k(s) are by unit tangent, principal normal, and binormal vectors (t, n,
similarly directed away from the concave side of b) all functions of the arbitrary parameter t. (b) Derivative of
the curve. However, to the right of the inection the binormal vector, db/ds, used to define the scalar
point the negative sign is used so n(s) and k(s) are torsion, .
oppositely directed. Although k(s) varies in mag-
nitude and remains directed away from the
concave side, n(s) is a unit vector directed consis- in the (x, y)-plane. However, as determined by
tently to one side of the curve. (3.11), the unit tangent vector for the circular
For an arbitrary parametric representation of helix is not in the (x, y)-plane unless b 0. Because
a curve, c(t), the unit principal normal vector is there are no inection points along the helix or
(Lipschutz, 1969, p. 66): the circle the choice of signs is arbitrary.
In order to identify the second property that
k(t) k(t) uniquely describes curves we again consider the
n(t) (3.43)
|k (t) | (t) natural parametric representation of a curve, c(s),
For the circular helix, using (3.22) in (3.43) with and dene a unit vector, b(s), called the unit binor-
the positive sign we have: mal vector, which is normal to the plane contain-
ing the unit tangent vector, t(s), and the unit
n(t) [( cos t)ex ( sin t)ey] (3.44)
principal normal vector, n(s):
Note that n(t) is independent of the pitch, b, of the
b(s) t(s) n(s) (3.45)
helix, so the principal normal vector given here is
the same as that for the circle. This unit vector is The three unit vectors [t(s), n(s), b(s)] form the so-
directed toward the z-axis and away from the called moving trihedron for the curve (Fig. 3.12b),
curve on its concave side for all values of the par- which can be thought of as traveling along the
ameter t (Fig. 3.12a). Also, note that this vector lies curve with change in arc length, s. The binormal
90 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
vector is directed following the conventional of the binormal vector, b, with arc length, s, but
right-hand rule for vector (cross) products: curl only that part of the change in orientation that
the ngers of your right hand from t toward n and projects onto the plane normal to the tangent
your thumb points in the direction of b. vector, t. In other words the torsion describes the
For an arbitrary parametric representation of component of rotation of the binormal vector
a curve, c(t), the unit binormal vector is the cross about the tangent line with change in position
product of the tangent and principal normal along the curve. The torsion is called an intrinsic
vectors, both written as functions of the parame- property of a curve because it serves, along with
ter t (Lipschutz, 1969, p. 68): the scalar curvature, to dene the shape of the
curve uniquely.
b(t) t(t) n(t) (3.46) For an arbitrary parametric representation of
Referring back to (3.29) and recalling that the the curve c(t) the torsion is (Lipschutz, 1969,
tangent vector, t(t), and the principal normal pp. 6970):
vector, n(t), are mutually orthogonal, the smaller
angle between them is /2, so sin 1.
Furthermore, both of these vectors have unit mag-
(t) | |
db
dt
dc
dt
n (3.49)
|
For the circular helix (Fig. 3.12a) the unit
|
2
dc d2c d3c dc d2c
binormal vector is found using (3.11) and (3.44) in (t) (3.50)
dt dt2 dt3 dt dt2
(3.46):
This property of the curve depends upon the arbi-
b(t) (a2 b2)12[b( sin t)ex b( cos t)ey aez]
trary representation of the curve having deriva-
(3.47) tives of order 3 or greater that are continuous. The
For the special case where b goes to zero, resulting numerator of (3.50) is sometimes written without
in a circle, the unit binormal vector is b(t) ez. The the scalar and vector product symbols and
circle lies in the (x, y)-plane, which also contains without the inner parenthesis. In this form it is
the tangent vector and the principal normal referred to as a triple scalar product. The following
vector, and the unit binormal vector is parallel to determinant provides a handy way to evaluate a
the z-axis. triple scalar product:
ux vx wx
3.1.6 The scalar torsion [uvw] u (v w) det uy vy wy (3.51)
The unit binormal vector is used to dene the uz vz wz
second intrinsic geometric property of curves,
namely the torsion. For the natural parametric rep- Here u, v, and w are arbitrary vectors and the
resentation of a curve, c(s), the torsion is dened determinant is composed of their components.
as (Lipschutz, 1969, p. 69): For the circular helix the torsion is calculated
using (3.49). Taking the derivative of the unit
db binormal vector (3.47) and (3.10) we have:
(s) n (3.48)
ds
db
(a2 b2)12[b( cos t)ex b( sin t)ey],
Because b(s) is a constant (unit) vector function, dt
the derivative db/ds is orthogonal to b(s), and
therefore lies in the plane containing t(s) and n(s) | |
dc
dt
(a2 b2)12 (3.52)
(Fig. 3.12b). The scalar product of db/ds and n
determines the component of the vector db/ds on Taking the scalar product of the ratio of these two
an axis for which n is the base vector. Thus, the quantities with the unit principal normal vector
torsion is a measure of the change in orientation (3.44), the torsion of the circular helix is:
3.2 THE CONCEPT AND DESCRIPTION OF CURVED SURFACES 91
(a2 b2)1[b( cos t)ex b( sin t)ey] must address the continuity and spatial variation
[ ( cos t)ex ( sin t)ey] b(a2 b2)1 (3.53) of these structures. In this way eld data can be
put in the context of differential geometry and
The torsion is constant and proportional to the one can begin to analyze the shapes of lineations
pitch, b. For b 0, 0 and the circular helix is in nature. Some of the tools developed to analyze
called right handed, that is the axis of the helix is curves are used to describe and analyze curved
parallel to the thumb of your right hand when surfaces, the subject of the next section of this
your ngers bend to follow the curve (Fig. 3.12a). chapter.
For b 0, 0 and the helix is left handed.
Because b 0 for the circle, (3.53) demonstrates
that the torsion is zero for that special case. A 3.2 The concept and description
curve with zero torsion lies entirely within a
single plane, and that plane contains both the
of curved surfaces
unit tangent vector and unit principal normal
vector. Such a curve is called a plane curve and the 3.2.1 Discrete, folded, and penetrative
torsion for all plane curves is zero (Lipschutz, geological surfaces
1969, p. 70). Curved surfaces are found in a wide variety of geo-
In summary, we have introduced the paramet- logical contexts making up many different kinds
ric representation of curves as dened by the of structures. For example, discrete geological sur-
vector function, c(t), where t is an arbitrary par- faces include faults (Fig. 3.1a), igneous contacts
ameter. The shape of such curves is determined by (Fig. 3.1b), and unconformities (Fig. 3.1c) all of
the curvature, (t), and the torsion, (t), both of which locally resemble a planar surface, but
which are scalar functions of position along the viewed more broadly are curved surfaces that can
curve. The curvature measures the change in ori- be characterized using the principles of differen-
entation of the unit tangent vector, t(t), and the tial geometry introduced in this section. In each
torsion measures the rotation of the binormal case the curvature of these surfaces has implica-
vector, b(t), about the tangent line. These two tions for their origin and the physical processes
vectors, along with the principal unit normal involved in their evolution. For example, faults
vector, n(t), form a mutually orthogonal set of unit may be curved because they evolved from discrete
vectors called the moving trihedron of the curved segments that do not lie in a plane (Segall and
line. Pollard, 1983b; Martel et al., 1988). Furthermore,
We have shown how the geographic angles the curvature of a fault surface may constrain the
used to measure the orientation of lineations at direction and magnitude of slip during an earth-
the exposure are related to the components of the quake (Carena and Suppe, 2002). The curvature of
unit tangent vector (3.16), and how to calculate the contact of an igneous dike may be used to
the curvature from the spatial variation of the deduce the stiffness of the surrounding host rock
tangent vector (3.26). To apply the concepts of and the distribution of magma pressure (Delaney
curved lines from differential geometry to curvi- and Pollard, 1981). The curvature of an angular
linear structures observed at exposure, lineations unconformity provides information about the
must be sufciently continuous so that the sedimentary processes that shaped that surface.
tangent, normal, and binormal vector functions To make the appropriate deductions about the
and their rst derivatives with respect to the physical processes involved in the formation of
natural parameter can be dened. Data that faults, igneous contacts, unconformities, and
would be suitable for such a study are available in other discrete geological surfaces one must quan-
the geological record, but the common measure- titatively characterize the shapes of these sur-
ments of lineation attitudes at discrete points are faces.
not sufcient. The tools we have introduced here Sedimentary and metamorphic layering com-
are suitable for an analysis of the spatial varia- monly is folded and the shapes of the surfaces of
tions of lineations, so the gathering of eld data the folded layers have been the subject of many
92 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
H
(a)
(a)
(b)
(b)
(b)
H
necessary to reproduce the surface, and they do plane, not a line. This leads to the denition of a
not provide adequate measures for quantitatively curved surface as a continuous vector function of
comparing the shapes of the different surfaces or two scalar variables (u, v), called the parameters of
the spatial variations in shape of a single surface. the surface, such that s s(u, v). The two parame-
Descriptions of folded surfaces using differential ters may be thought of as the coordinates of
geometry do not have these shortcomings. points on a plane, called the parameter plane, and
Metamorphic rocks composed of the system- those points map onto the surface according to
atic arrangement of layers of different rock types the vector function s(u, v). As the two parameters
(Fig. 2.24a), a set of sub-parallel fractures (Fig. vary, the heads of the successive position vectors
2.24b), or a set of similarly oriented platy mineral sweep out the curved surface in three-dimen-
grains (Fig. 2.24c) are said to have a penetrative sional space.
planar fabric (Turner and Weiss, 1963, p. 97). The To understand the analytical denition of a
name indicates that the constituents locally curved surface we begin by describing the coordi-
dene a surface that resembles a plane, but the nate systems used for the two-dimensional par-
normal to this planar fabric may systematically ameter plane and the three-dimensional curved
change orientation from exposure to exposure. In surface (Fig. 3.15). The two Cartesian axes (Ou, Ov)
these cases it may be possible to represent this and the associated base vectors (eu, ev) dene the
spatial variation in orientation with a set of parameter plane on which the two coordinates
curved surfaces such that a given surface is every- are the parameters u and v. The three Cartesian
where tangent to the locally planar fabric. axes (Ox, Oy, Oz) and the associated base vectors (ex,
Procedures to dene individual continuous sur- ey, ez) comprise the system for the curved surface.
faces in three-dimensional space from scattered The position vectors, s, for the curved surface can
measurements of a penetrative planar fabric have be written as a function of the two parameters
not been devised, so that is a noteworthy chal- (Lipschutz, 1969, p. 128):
lenge for structural geologists. With such a char-
s(u, v) sx(u, v)ex sy(u, v)ey sz(u, v)ez (3.54)
acterization in hand one could model fabrics
using continuum mechanics and test hypotheses The three scalar functions [sx(u,v), sy(u,v), sz(u,v)] are
concerning their orientation using differential the components of the vector function, s(u, v),
geometry. with respect to the base vectors (ex, ey, ez). These
functions, along with the base vectors, determine
3.2.2 Parametric representations of the position vectors for all points on the curved
curved surfaces surface. The vector equation (3.54) is called the
We began our discussion of curved lines by appeal- parametric representation of the surface. Compare
ing to the intuitive notion of a set of points the facts that a single variable parameterizes the
arranged in an orderly and continuous fashion to curved line (3.1) and a pair of variables parame-
form a curve in three-dimensional space. Because terizes the curved surface (3.54).
individual points in three-dimensional space are Any point in the parameter plane (Fig. 3.15a)
identied by their position vectors, p, this led to may be dened by a two-dimensional position
the denition of a curved line as a continuous vector w ueu vev with respect to the base
vector function of a single scalar variable t, called vectors eu and ev. Thus, the position vectors for the
the arbitrary parameter of the curve, such that c curved surface, s, in three-dimensional space
c(t). As t increases in value the heads of succes- are determined by a vector function of the two-
sive position vectors trace out the curved line. The dimensional vector variable, w, that is s s(w).
analytical description of curved surfaces also may However, because the components of the vector w
be approached from the intuitive notion of a set are the two parameters (u, v), we can speak of the
of points arranged in some continuous fashion in surface as a function of these two scalar parame-
three-dimensional space. However, sufciently ters, s s(u, v), and that is what we will do in the
close to any particular point, the neighboring following discussion. An individual point (uo, vo) in
points are distributed such that they resemble a the parameter plane (Fig. 3.15a) maps onto the
94 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
v = 0.7
p(0.8, 0.7)
v = constant
u = constant
c(u, vo ) u
Parameter
plane
y 1, 1
(b)
v-parameter
u-parameter curve
curve c(0.8, v)
x 2 z
c(u, 0.7)
s(u, v) +1
x 0
0 1
1 0 +1
c( u o ,v )
Fig 3.17 The same elliptic paraboloid shown in Fig. 3.16
but represented as a Monge patch with parameter plane
superimposed on the (x, y)-plane so u and v are equivalent to
x and y. (a) Parameter plane. (b) Wire-frame diagram of
elliptic paraboloid: the u- and v-parameter curves, c(u, vo) and
c(uo, v), are parabolas in (x, z)- and (y, z)-planes, respectively,
y
that define two of the sets of wires.
a surface that is the target for mapping may be corresponding point on the curved surface in
written: three-dimensional space.
s(x, y) xex yey g(x, y)ez (3.59)
3.2.3 The tangent plane, tangent vector,
Here we consider the (x, y)-plane as the horizontal and unit normal vector
plane of a local Cartesian coordinate system estab- We continue to use the more general parametric
lished in the region being mapped, and values of representation of a surface (3.54) to develop the
g(x, y) are the measured elevations of exposures of theoretical concepts necessary to characterize
the surface relative to the local origin. Similarly, surfaces, but recognize the Monge patch (3.58) as
the parameter plane can be superimposed on the a useful representation for mapping. Because the
UTM grid for the region in which case (3.58) is parametric representation of a surface, s(u, v),
transformed such that sx easting, sy northing, describes a vector function of two variable para-
and sz elevation. meters, there is a partial derivative associated
To gain further insights about the Monge with each parameter. To calculate the partial
patch consider the parameter plane (Fig. 3.17a) derivative s/u, for example, one takes the deriv-
with a rectangular grid of lines, u constant and ative with respect to u of each component of the
v constant, parallel to the coordinate axes and vector function while holding v constant, and
use the following parametric representation of a then uses these as the components of a new vector
curved surface which is in the form (3.58) of a function. Thus, the partial derivatives of s(u, v)
Monge patch (Lipschutz, 1969, p. 185): with respect to the two parameters are (Lipschutz,
1969, p. 126):
s(u, v) uex vey (u2 v2)ez (3.60)
s(u, v) sx sy s
Using the rst two components to eliminate the ex ey z ez
u u u u
parameters u and v from the expression for the (3.61)
s(u,v) sx sy sz
third component, and noting that the compo- ex ey ez
v v v v
nents of the vector are equivalent to the coordi-
nates, the equation z x2 y2 is found which is the Recall from (3.8) that for the natural representa-
same elliptic paraboloid illustrated in Fig. 3.16. tion of a curve, c(s), the derivative with respect to
However, in contrast to that representation, the u- the arc length, s, is the unit tangent vector.
and v-parameter curves for (3.60) are the intersec- Because the partial derivative s/u is taken with
tions of the surface with planes parallel to the v constant, this is equivalent to taking the deriv-
(x, z)-plane and the (y, z)-plane, respectively. ative along any one of the u-parameter curves, for
Note how three sets of curved lines are used to example, c(u, 0.7) as shown in Fig. 3.17b. Thus, the
represent the elliptic paraboloid in the wire- partial derivative, s/u, is a vector that is tangent
frame diagram (Fig. 3.17b). The intersections of to a u-parameter curve and points in the direction
the curved surface with planes parallel to the (x, z)- of increasing u. Similarly, s/v, is a vector that is
plane form a set of parabolas, as do the intersec- tangent to a v-parameter curve and points in the
tions of the curved surface with planes parallel to direction of increasing v. These tangent vectors
the (y, z)-plane. The intersections of the curved are not necessarily unit vectors because the u- and
surface with planes parallel to the (x, y)-plane form v-parameter curves are not necessarily the natural
a set of circles. A particular u-parameter curve, representations of these curves.
dened by the vector function c(u, 0.7), and a par- As an example consider the partial derivatives
ticular v-parameter curve, dened by c(0.8, v), are of the parametric representation for the elliptic
highlighted in Fig. 3.17b. The point (0.8, 0.7) on the paraboloid (3.60) illustrated in Fig. 3.17b:
parameter plane maps to the point p(0.8, 0.7) on s s
ex 2uez, ey 2vez (3.62)
the curved surface at the intersection of the two u v
highlighted curves. In this way every coordinate Note that the tangent vectors, s/u, for the u-
line and every point in the two-dimensional par- parameter curves lie in planes that are parallel to
ameter plane have a corresponding curve and a the (x, z)-plane. Consider the particular u-parameter
3.2 THE CONCEPT AND DESCRIPTION OF CURVED SURFACES 97
x, u = 0.8
8,
y, v = 0.7 y
p(
dc
the tangent plane. In this way the tangent vector ( cos )ex ( sin )ey
dt
to the surface s(u, v) at an arbitrary point in any
arbitrary direction is related to the partial deriva- (2uo cos 2vo sin 2t)ez (3.71)
tives of the parametric representation of the
At the arbitrary point p(uo, vo) the curve and its
curved surface at that point.
derivative are found by setting t 0 in these equa-
The following example provides geometric
tions. The tangent vectors anywhere on the ellip-
interpretations for the quantities du/dt and dv/dt
tic paraboloid (3.60) in the direction specied by
in (3.66) and further insight about the tangent
the direction cosines of the arbitrary line are cal-
vector T. Consider a straight line in the parameter
culated using (3.66) as:
plane through the arbitrary point p(uo, vo) and
having an arbitrary slope, m (Fig. 3.19c). The equa- T(u, v) (ex 2uez) cos (ey 2vez) sin (3.72)
tion for this line, given in the standard point-slope
form, is: This is equivalent to the tangent vector calculated
for the curved line in the preceding equation,
v vo m(u uo) (3.67) given our denitions of the parameters u and v as
To write this equation in parametric form con- functions of t.
sider the parameter, t, to be the coordinate mea- The orientation of the tangent plane (3.63) is
sured along the line from the arbitrary point uniquely determined by either of the two unit
p(uo, vo). The parameters u and v are related to the normal vectors to that plane. The choice between
parameter t by noting that the slope m tan , these two oppositely directed vectors is deter-
where is the angle from the positive u-axis to the mined by a right-hand rule: the unit normal
line, so: vector, N, makes a right-handed orthogonal
system with the two tangent vectors, s/u and
u uo t cos ,v vo t sin (3.68) s/v (Fig. 3.19b) and is dened as (Lipschutz, 1969,
By varying the angle, , lines with any orientation p. 158):
can be used to specify the direction of the tangent
s s
vector, T, at a point on the surface. The derivatives
u v
of u(t) and v(t) as used in the denition of the N (3.73)
tangent vector (3.66) are: |
s s
u v |
du
dt
dv
dt
cos , sin cos
2 (3.69) Recall that the vector (cross) product of two arbi-
These are the direction cosines of the angles trary vectors, v w, is normal to the plane con-
between the arbitrary line and the two coordinate taining v and w and that the thumb of your right
axes in the parameter plane. In (3.66) these two hand points in the direction of v w when your
quantities scale the tangent vectors for the u- and ngers curl from v toward w. Using (3.73) the unit
v-parameter curves to determine the tangent normal at any point on a surface can be calculated
vector, T, for the curve c[u(t), v(t)]. The ratio from its parametric representation s(u, v).
(dv/dt)/(du/dt) tan determines the direction of As an example consider the unit normal vector
the tangent line to this curve. for the elliptic paraboloid (3.60). Using (3.28) the
Returning to the example of the elliptic parab- vector product of the two tangent vectors is:
oloid (3.60), the straight line in the parameter
ex 1 0
plane (Fig. 3.19c) maps onto this surface as the s s
det ey 0 1
curve: u v
ez 2u 2v
c[u(t), v(t)] (uo t cos )ex (vo t sin )ey 2uex 2vey 1ez (3.74)
(u2o 2uot cos v2o
2vot sin t2)ez (3.70) Therefore, the unit normal vector is:
The tangent vector anywhere along this curve is: N (2uex 2vey 1ez)(4u2 4v2 1)1 2 (3.75)
100 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
(a)
Hackle
z
x
Main join
t
Fig 3.23 Oblique rendering of a portion of the joint fringe
in Fig. 3.22 from scanned data. Reprinted from Pollard et al.
(2004) with permission from The Geological Society of
London.
main joint break down zone hackle fundamental forms. These quantities are intro-
Change in N (o)
| |
12 (b) Surface with u- and v-parameter curves and parallelogram
dc dc dc
s dt dt approximating surface area between curves. Reprinted from
dt dt dt
a a Pollard et al. (2004) with permission from The Geological
b
E dt 2F dt dt G dt
du 2 du dv dv 2 12 Society of London.
dt (3.89)
a
The formula for dA is similar to (3.29) for the magnitude of the tangent vector to the v-parame-
absolute value of the cross product of two vectors. ter curves. In this case the vector has a magnitude
Therefore, the differential area of the parallelo- (c2 u2)1/2. From the general parametric represen-
gram can be calculated as: tation for a circular helix (3.2) we consider a par-
ticular example that has a radius uo and pitch c
dA |dc (u, vo) dc (uo, v) | | s s
u v
dudv | (Lipschutz, 1969, p. 63):
s s s s 12 c(t) (uo cos t)ex (uo sin t)ey (ct)ez (3.97)
dudv (3.92)
u v u v
Here t is the single arbitrary parameter for the
An identity relating scalar and vector products curve, and the tangent vector is:
can be used to convert this equation to an equa- dc
tion that involves the coefcients of the rst fun- (uo sin t)ex (uo cos t)ey (c)ez (3.98)
dt
damental form (Lipschutz, 1969, p. 10):
Comparing these equations to (3.77) and the
(a b) (c d) (a c)(b d) (a d)(b c) (3.93) second of (3.78) it is clear that each v-parameter
Substituting the appropriate vector cross prod- curve on the helicoid is a helix with radius uo and
pitch c. Because neither tangent vector (3.78) is
ucts into this identity yields:
zero but the coefcient F 0, the u- and v-parame-
s s
u v
s s
u v ter curves are orthogonal everywhere on the heli-
coidal surface. The set of straight lines and the set
of helixes cover the helicoidal surface with an
s s
u u s s
v v
s s
u v s s
u v orthogonal network.
The arc length of any curve on a surface is cal-
EG F2 (3.94) culated using (3.89) and for the helicoid this
Combining these results the differential area of reduces to:
the curved surface is dA [EG F2]12 dudv. t2
dt (c u ) dt
du 2 dv 2 12
Because EG F2 0, the square root of this quan- s 2 2 dt (3.99)
tity is a real number. The area of a curved surface t1
is found by integrating the differential area over
the surface: For a coordinate line v constant which maps
onto the helicoid as a u-parameter curve with u
A (EG F ) 2 12
dudv
(3.95)
t, we have du/dt 1, dv/dt 0. Taking the limits as
t1 b and t2 b we nd s 2b, just what is
The surface area is a function of the coefcients of expected for a straight line. For a coordinate line
the rst fundamental form and, in turn, of the u constant uo that maps onto the helicoid as a
parameters u and v. v-parameter curve (a helix) with v t, we have
Consider the helicoidal surface (3.77), which du/dt 0, dv/dt 1. Taking the limits as t1 0 and
we have identied as a geometric model, for twist t2 we nd the length of this segment of the
hackle (Figs. 3.22 and 3.23). The coefcients of the helix, s (u2o c2)1/2. Along the z-axis uo 0 and s
rst fundamental form are found using (3.85) c, which is the length of the surface (Fig. 3.21).
such that: The area of the patch of a helicoid that we take
as a fracture surface (Fig. 3.21) is found using the
E 1,F 0,G c2 u2 (3.96) coefcients of the rst fundamental form in (3.95)
In general E is the squared magnitude of the with the limits b u b and 0 v . From
tangent vector to the u-parameter curves. This symmetry this is equivalent to twice the area
vector for the helicoid is a unit vector that lies in using the range 0 u b such that:
the (x, y)-plane and is not a function of u, so the u- b
1
0.9 the twist hackle in the fringe region of a joint
0.8 (Fig. 3.22) the non-intuitive result is that the
0.7 p /180 surface area decreases as the number of frac-
p / 60
0.6 p / 20 tures increases (Fig. 3.27). On this gure each
p/6
0.5 curve corresponds to a different twist angle, .
0.4 For a twist angle of 1 ( /180) the surface area
0.3 of ten fractures is 99.5% of that for the single
0.2 fracture, only marginally less. However, for a
0.1 twist angle of 30 ( /6) the surface area of ten
0 fractures is 36.1% of that for the single fracture,
0 5 10 15 20 25 30 35 40 45 50 dramatically less. Because the energy required to
Number of partial fractures form a fracture in brittle materials scales with
the fracture surface area (Lawn and Wilshaw,
Fig 3.27 Graph of relative surface area versus number of
1975) this result shows that the breakdown of
partial fractures for helicoidal model of fringe fractures.
Relative surface area decreases with number of fractures. joints into hackle with helicoidal shapes is con-
Reprinted from Pollard et al. (1982) with permission from sistent with a condition of lesser energy
The Geological Society of America. expended during propagation.
1
a2 x2 dx xa2 x2 a2 ln (x a2 x2)
2
of the shape at any point on a continuous curved
surface. To understand how this is accomplished
(3.101) we focus on a very small part of the parameter
plane so lengths along the coordinate axes are
The area of the helicoidal fracture surface is: measured using the differential quantities du and
dv (Fig. 3.28a). Consider the arbitrary curve u
A b b2 c2 c2 ln b b2 c2
c (3.102) u(t), v v(t) in the parameter plane which maps to
the curve c[u(t), v(t)] on the curved surface. At an
As the spatial rate of twist goes to zero, 1/c 0, the arbitrary point along this curve the differential
area goes to that of a rectangular plane, A A0. tangent vector, dc, is dened by (3.83) and this
For unit half-width (b 1) and unit length ( c 1), vector lies in the tangent plane to the surface (Fig.
we have A0 2, and the normalized area, A/A0, 3.28b). The unit vector, N, at this arbitrary point is
increases slowly and non-linearly with rate of a function of the two parameters u and v, such
twist. that the differential is:
Using the dimensionless ratio b/c in (3.102), the
N N
area of the helicoidal fracture surface takes the dN du dv (3.104)
u v
form (Pollard et al., 2004):
The vector dN is a measure of the change in ori-
2 2
A c2 b b 1 ln b b 1 entation of N with position along the curve on the
c c c c surface and, in this sense, it is a measure of the
shape of the surface. Also, because N is constant in
(3.103)
magnitude, the vector dN is orthogonal to N and
This relationship demonstrates that the surface therefore lies in the tangent plane (Lipschutz,
area of n helicoidal fractures, each of half-width 1969).
b and length c, is less than the surface area of Although dc and dN both lie in the tangent
a single helicoidal fracture of half-width nb and plane of the surface (Fig. 3.28b), these vectors
108 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
(a) dv
Parameter plane II dN dc
N
u
N
du dv
v
s
u
s
du dv
v
p(u, v)
u = u(t),
N s
u u
du2 N s N s
u v v u
dudv
v = v(t) N s
v v
dv2 (3.105)
c[u(t), v(t)] dc
L N s
u u
, M
1 N s N s
2 u v v u
,
z v)
N
N s
v v (3.106)
(u,
p The quantities L, M, and N are functions of the two
y
Tangent parameters u and v, and are called the coefcients
x plane of the second fundamental form of the surface.
Do not confuse the scalar quantity N, and the
(c) Osculating
vector quantity N. The coefcients depend upon
plane N n(t)
the choice of parameters used to represent the
u surface, but the second fundamental form itself is
invariant with respect to this choice (Lipschutz,
c[u(t), v(t)] t(t) 1969), and in this sense II is a property of the
surface. Note that II characterizes the changing
shape of the surface in all directions at a particu-
z v) lar point and that the differential parameters du
u,
p( and dv dene the direction.
y k(t)
The coefcients of the second fundamental
x form (3.106) may be rewritten in a different way
that is useful for computations. The unit normal
Fig 3.28 Diagrams to define second fundamental form for vector, N, to the surface s(u, v) is perpendicular to
a surface. (a) Parameter plane with arbitrary curve uu(t), v the vectors that are tangent to the u- and v-par-
v(t). (b) Surface with curve, c[u(t), v(t)], unit normal vector, ameter curves (Fig. 3.19b). Therefore, for example,
N, differential tangent vector, dc, and differential normal the scalar product of N and the tangent vector
vector, dN. (c) Surface with osculating plane containing the s/u is zero, and this product may be expanded as
unit principal normal vector, n(t), and curvature vector, k(t). follows:
Reprinted from Pollard et al. (2004) with permission from
The Geological Society of London. s s 2s N s
N 0 N N 2
u u u u u u
(3.107)
are not necessarily parallel to one another. The
shape of the surface in the particular direction Rearranging the right-hand side of this expression
specied by dc is characterized by the scalar we have:
product of the two vectors, dN and dc, and this
product is used to dene the second fundamen-
tal form, II:
N
2s
u2
N s
u u
L
(3.108)
3.2 THE CONCEPT AND DESCRIPTION OF CURVED SURFACES 109
0, elliptic point
LN M2 0, parabolic point (3.110)
0, hyperbolic point
L M N 0,planar point
in (3.109) the coefcients of the second funda- curvature of this curve times the cosine of the
mental form for the helicoidal surface are: angle between n and N. If the osculating plane of
the curve contains the unit normal vector for the
L 0, M c c2 u2, N 0 (3.112)
surface, then n and N are parallel, and n . On
Using (3.110) the local shape of the helicoidal the other hand, if the osculating plane of the
surface is determined by the sign of LN M2 curve is parallel to the tangent plane for the
c2/(c2 u2) 0. With the exception of the case c surface, then n 0.
0, which describes a planar surface, the sign of The relationship given in (3.114) illustrates the
this quantity is negative, so every point on the fact that the curvature of an arbitrary curve at a
helicoidal surface is hyperbolic. point on a surface is greater than or equal to the
Recall from (3.21) that the shape of a curved normal curvature of the surface in the direction
line is characterized, in part, using the curvature of the curve at that point. Familiar examples are
vector, k, and the scalar curvature, |k|. For a the circles of latitude and longitude on a sphere
curve on a surface s(u, v) two analogous measures of radius R (Fig. 2.1a). The normal curvature in any
of shape are the normal curvature vector, kn, and direction at any point on the sphere is a constant,
the normal curvature, n. Both of these quantities n 1/R. Circles of longitude are the intersection
are dened by considering an arbitrary curve u of the sphere with planes that pass through the
u(t), v v(t) in the parameter plane (Fig. 3.28a) center and the poles. These curves have the same
which maps to the curve c[u(t), v(t)] on the surface radius as the sphere and therefore their curvature
(Fig. 3.28c). At a point on this curve the curvature is 1/R. However, circles of latitude, except the
vector k is a function of the parameter t; it lies in equatorial circle, have a lesser radius, () R,
the osculating plane of the curve; and it extends and therefore a greater curvature, 1 ().
away from the concave side of the curve. The oscu- These circles are the intersections of the sphere
lating plane of a curve is the plane that contains with planes that are parallel to the equatorial
both the unit tangent vector, t(t), and the unit prin- plane and do not pass through the center. The
cipal normal vector, n(t). The unit normal vector, unit principal normal vector, n(t), for these circles
N, is perpendicular to the surface; it is a function and the unit normal vector, N, for the sphere are
of the parameters u and v; and it may not lie in the not parallel. As the circles of latitude approach
osculating plane of the curve. The normal curva- the poles of the sphere, their radii of curvature
ture vector, kn, and the normal curvature, n, are approach zero, the osculating plane of the circle
dened in terms of k(t) and N(u, v) as: approaches the tangent plane of the sphere, and
the curvature of the circle becomes greater and
kn (k N)N,n k N (3.113)
greater. This exemplies the fact that curves on
Because N is a unit vector we understand from surfaces provide the direction in which the
(3.113) that n, is a scalar quantity equal to the normal curvature of the surface is measured, but
component of k along N. Also, kn is a vector of the normal curvature is not necessarily equal to
magnitude n with the same, or the opposite, the curvature of the curve.
direction as N. If one chooses two differently directed curves
Recall that the direction of the unit principal through the same point on a surface (Fig. 3.30a),
normal vector, n(t), is chosen for consistency along the respective values of the normal curvature, n,
the curve (Fig. 3.11). If we choose the direction of for the surface may be different. On the other
n(t) for the arbitrary curve c[u(t), v(t)] on the surface hand, the curvature, , at a point on a curve is a
s(u, v) (Fig. 3.28c) such that the angle, , between unique property of the curve. The normal curva-
n(t) and N(u, v) is in the range 0 /2, then ture, n, at a point on a curved surface varies in a
(Lipschutz, 1969): smooth and systematic manner with the direc-
tion of the tangent line through the point of inter-
n cos ,0 2 (3.114)
est, from a maximum value, 1, to a minimum
In other words, the normal curvature associated value, 2. These two values of normal curvature, 1
with a particular curve on a surface is equal to the and 2, are called the principal normal curvatures.
3.2 THE CONCEPT AND DESCRIPTION OF CURVED SURFACES 111
to 2
(a) N ne
n t li K 2
e h
ng wit 1.5
1
Ta rve
c u
1 2
Tangent line
Normal curvature
to
curve with K 3
n 0.5
a
0 4
v)
Ta rve
u, cu
ng w
z (
en ith
p 0.5 5
tl K1
in
y
e
1
to
x
1.5
(b) N
c(u, vo ) s 2
0 45 90 135 180 225 270 315 360
v
Alpha ()
T
c[u(t), v(t)] Uo Fig 3.31 Graph of normal curvature versus angle with
z examples for elliptical dome (1), parabolic antiform (2),
hyperbolic saddle (3), parabolic synform (4), and elliptical
s
y c(uo, v) basin (5). Reprinted from Pollard et al. (2004) with
u permission from The Geological Society of London.
x
Fig 3.30 Diagrams to define variation of normal curvature shapes illustrated in Fig. 3.29. We use the Monge
with direction at a point on a surface. (a) Angle measured
patch (3.58) where the (x, y)-plane is the horizontal
in the tangent plane from direction of maximum principal
plane to give the surface an orientation (up or
normal curvature, 1, to direction of normal curvature n.
(b) Angle o measured in the tangent plane from tangent to down) and relate it to geological structures. For
u-parameter curve to direction of principal normal the sake of this illustration we consider particular
curvature. cases where 1 2 1. The elliptic paraboloid
(3.60) in this context is analogous to the surfaces
of sedimentary beds in a basin-shaped structure
The variation of normal curvature with direc- (Fig. 3.17b). For points where both principal cur-
tion is of the same form for all surfaces with vatures are positive (concave upward) or negative
continuous second partial derivatives such that (concave downward) the shape is elliptical and the
(Lipschutz, 1969): structure is a basin or a dome. For points where
n 1 cos 2 2 sin 2 (3.115) one principal curvature is zero and the other is
positive or negative the shape is parabolic and the
This relationship is known as Eulers Theorem and structure is a cylindrical synform or antiform. For
the angle is measured in the tangent plane from points where the principal curvatures are of dif-
the direction of the tangent line corresponding ferent signs the shape is hyperbolic and the struc-
to the curvature 1 to that corresponding to n. ture is a saddle.
The directions of the tangent lines associated Because the normal curvature is a property of
with the extreme values of normal curvature are a surface at any point, we anticipate that it can be
called the principal directions of normal curvature written as a function of the fundamental forms.
and they are orthogonal. The derivation uses (3.21) to replace the curvature
The signs of the principal normal curvatures vector in (3.113) with the derivative of the tangent
are related to the shape and orientation of a vector written as a function of the arbitrary par-
surface in the vicinity of a point as dened in ameter t (Lipschutz, 1969):
(3.110). For example in Fig. 3.31 the distributions
of normal curvature n at a point as a function of
the angle are plotted for the fundamental
n k N dt
dt
N | |
dc
dt
(3.116)
112 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
Because the tangent vector, t, is perpendicular to dened in (3.77). The coefcients of the rst and
the normal vector, N, the derivative of their scalar second fundamental forms were derived from
product is zero: (3.85) and (3.106) above such that:
d dt dN 2c
(t N) N t 0, I (du)2 (c2 u2)(dv)2,II (dudv)
dt dt dt c2 u2
dt dN
so N t (3.117) (3.121)
dt dt
Using the last equation in (3.116) and substituting In the form of (3.119) the normal curvature for the
(3.8) for the tangent vector we have: helicoidal surface is:
n t
dN
dt /| |
dc
dt
dc dN
dt dt / dc dc
dt dt n
2c du dv
c2 u2 dt dt
(3.118) du 2 dv 2 (3.122)
(c2 u2)
dt dt
The derivatives of c and N with respect to t may be
rewritten as derivatives of s and N with respect to On a u-parameter curve u t, du/dt 1, and dv/dt
u and v using (3.83) and (3.104). Comparing the 0, so the normal curvature is n 0, just what one
results with the expressions for the coefcients of would expect for a straight line. Recall that such a
the fundamental forms, (3.85) and (3.106), leads to straight line is the generating line for the heli-
the denition of the normal curvature in terms of coidal surface when it is moved perpendicular to
these coefcients: itself and rotated about the z-axis which is the
mid-line of the helicoid (Fig. 3.21).
L(dudt)2 2M(dudt)(dvdt) N(dvdt)2 A v-parameter curve on the helicoidal surface
n
E(dudt)2 2F(dudt)(dvdt) G(dvdt)2 (Fig. 3.21) is a circular helix with radius uo and
(3.119) pitch c. On this curve v t, du/dt 0, and dv/dt 1,
From the discussion of (3.66) and Fig. 3.19 recall so the normal curvature is n 0. The curvature
that the ratio of the derivatives dv/dt and du/dt vector, k, for the helix is not zero (3.22) and is
determine the direction of the tangent line to the directed in the (x, y)-plane toward the z-axis. The
arbitrary curve c[u(t), v(t)] at the point in question, unit normal vector, N, for the helicoidal surface
so the normal curvature depends upon this direc- along a v-parameter curve is orthogonal to the cur-
tion. In addition the normal curvature depends vature vector, so the scalar product in (3.113)
upon the coefcients of the rst and second fun- denes the normal curvature as k N n 0. In
damental forms. other words the curvature vector for the helix
The normal curvature may be written in terms does not resolve any component onto the line
of the differentials, du and dv, using (3.119) normal to the helicoidal surface. This result is
(Lipschutz, 1969): non-intuitive because the v-parameter curve
clearly has a non-zero curvature, but the normal
L du2 2M dudv N dv2 II
n (3.120) curvature of the surface along the tangent line to
E du2 2F dudv G dv2 I
this curve is zero.
The ratio of the differentials, dv : du, determines
the direction of the tangent line to the arbitrary 3.2.7 Principal normal curvatures,
curve c[u(t), v(t)], and these differentials are Gaussian, and mean curvature
referred to as the direction numbers of the tangent Eulers Theorem (3.115) is used to calculate the
line. From (3.120) we understand that the normal normal curvature n in the direction of any line
curvature n at an arbitrary point on the surface tangent to a surface, s(u, v), given the principal
s(u, v) in the direction of this tangent line is equal normal curvatures, 1 and 2, at a point on the
to the ratio of the second to the rst fundamental surface. Here we describe how to calculate the
form. magnitudes of the two principal normal curva-
To illustrate the concept of normal curvature, tures and the principal directions. Recall from cal-
consider again the helicoidal surface (Fig. 3.21) as culus that the maximum and minimum values of
3.2 THE CONCEPT AND DESCRIPTION OF CURVED SURFACES 113
a function of two variables are found by setting for the coefcients of the fundamental forms,
the partial derivatives of the function to zero. (3.96) and (3.112), to nd tan o 1 c2 u2. Here
From (3.120) we note that the normal curvature, o is the angle measured in the tangent plane
n, is a function of the two differential quantities, to the surface from the tangent line for the u-
du and dv, which determine the direction of the parameter curve to the tangent line for the two
tangent line on the surface at a particular point. principal directions (Fig. 3.30b). Along the mid-
Thus, the derivative of n with respect to each of line, u 0, of the helicoidal surface (Fig. 3.21) the
these direction numbers is set equal to zero and tangent of the principal directions is equal to the
evaluated for the principal directions, identied spatial rate of twist, 1/c.
by duo and dvo: The magnitudes of the principal normal cur-
vatures, 1 and 2, are found by rearranging
n
du | (duo, dvo) 0,
n
dv | (duo, dvo) 0 (3.123) (3.124) to factor out the two differentials duo and
dvo:
We substitute n II/I, use the formula for the
derivative of a quotient, and write the partial (L oE)duo (M oF)dvo 0
(3.127)
derivatives of I and II in terms of their coefcients (M oF)duo (N oG)dvo 0
using (3.85) and (3.106). When the resulting
expressions are evaluated for the principal direc- These two linear equations have a simultaneous
tions, duo and dvo, the normal curvature takes on solution (duo, dvo) if the determinant of the
extreme values, o II/I, satisfying the following coefcients of the left side is zero. Expanding the
linear equations (Lipschutz, 1969): determinant produces a quadratic equation in o:
These equations have a simultaneous solution (1, For example, the magnitudes of the principal
o) if the determinant made up of the coefcients normal curvatures for the helicoidal surface (3.77)
on the left-hand side is zero. Expanding the deter- are found by substitution of (3.96) and (3.112) to
minant produces a quadratic equation in the nd 1, 2 c(c2 u2). The two principal curva-
direction numbers for the principal directions, tures are equal in magnitude and opposite in sign.
duo and dvo: They are independent of the parameter v, and so
are constant along any particular circular helix
(LF ME)(duo)2 (LG NE)duodvo
that is a v-parameter curve. Along the mid-line the
(MG NF)(dvo)2 0 (3.125)
principal curvatures are equal in magnitude to
The ratio of the direction numbers, dvo : duo tan the spatial rate of twist, 1/c. For a given rate of
o, and this determines the angles, o and o /2, twist the principal curvatures decrease in magni-
between the tangent to the u-parameter curve and tude with distance, u, from the mid-line.
the tangent to the principal directions (Fig. 3.30b). Equation (3.128) may have two real and
Dividing (3.125) by (duo)2 and substituting for unequal roots, 1 and 2, or two real and equal
the ratio of direction numbers, we nd a qua- non-zero roots, or two zero roots (Lipschutz, 1969).
dratic equation in tan o: The second case pertains to elliptical points (Fig.
3.29a) at which the normal curvature is non-zero
(MG NF) tan 2o (LG NE) tan o
but the same in all directions and the ratios of
(LF ME) 0 (3.126)
respective fundamental coefcients are constant:
The principal directions for the normal curvature
L M N
at a point on a surface are found from (3.126) n constant (3.129)
E F G
using the standard formula for the solution of a
quadratic equation. Using the helicoidal surface This is referred to as an umbilical point. The third
(3.77) as an example, the principal directions of case is the planar point where the normal curva-
the normal curvature are found by substituting ture is zero in all directions.
114 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
ize geological structures. A keyword search of surface that are everywhere parallel to the local
GEOREF, a geological reference data base in July direction of relative motion, and then use the cur-
2003 using the exact phrase differential geome- vature and torsion to characterize the shapes of
try uncovered only fourteen references from these curves.
1963 to 2003. Of these ten are applications to To illustrate the fact that slickenlines do vary
other geoscience subjects such as gravity and systematically with position on a fault, and to
geodesy, seismology, hydrogeology, and paleon- appreciate some of the challenges inherent to the
tology. The four remaining consider the three- investigation of supercial lineations we turn to a
dimensional geometry of folds, the deformation data set from the Chimney Rock fault array (Fig.
of Earths surface due to neotectonic crustal 2.29). The four sets of faults in this region are dis-
motion (Zakarevicius, 2000; Grachev et al., 2001), played on the structure contour map constructed
and the normal curvature of geological surfaces on the base of the Carmel Formation. Note, for
(Bergbauer and Pollard, 2003). A search on the example, that individual contours on this map are
phrase Gaussian curvature again uncovered truncated by the Frenchman Fault. When traced
fourteen references including three on the esti- to the north across the fault the sense of step is
mation of strain and the prediction of fractures consistently to the east. However, the magnitude
within folds (Lisle, 1994, 2000; Ozkaya, 2002). of the step decreases toward both terminations of
This section includes examples that provide this fault. This change in step magnitude suggests
insights into how differential geometry can be that the magnitude of the slip decreases from the
applied to problems in structural geology. These mid-section of the Frenchman Fault toward the
examples are works in progress and we expect terminations and, indeed, the slip must go to zero
more details to emerge during on-going studies. at the terminations by denition.
None-the-less we hope that these examples will The distribution of dip slip is plotted versus
encourage others to apply differential geometry position along the trace of the Blueberry Fault in
to structural problems. Fig. 2.31. Note how the magnitude of the dip slip
(gray boxes) increases from zero at the eastern ter-
3.3.1 Characterizing the shapes of mination to more than 30 m near the middle of
lineations on discrete surfaces the fault. The slip distribution is not continuous,
Lineations are found on discrete geological sur- but jumps abruptly where members of other fault
faces such as faults and intrusive contacts (Fig. sets intersect the Blueberry Fault. The distribu-
3.3). These supercial lineations typically are tion of slickenline rake (black diamonds) is
aligned on an exposure such that multiple mea- plotted versus position along the trace of the
surements of the orientations of linear elements Blueberry Fault. The rakes are approximately 90
would have a standard deviation of few degrees. (down dip) near the northeastern and southwest-
However, faults and intrusive contacts may pass ern terminations of the fault, but decrease more
through rocks with different mechanical proper- or less systematically to about 70 as one
ties, they may be inuenced mechanically by adja- approaches the intersections with the La Sal and
cent faults or intrusions, and they may be curved Little Faults. Across these faults the rakes abruptly
surfaces. Thus one should expect supercial lin- increase. The rakes are systematically greater than
eations to vary in orientation over the surfaces on 90 (inclined toward the east) between the inter-
which they are found. On a fault, for example, sections with the La Sal and Little Faults, and less
slickenlines should form a systematic pattern that than 90 (inclined toward the west) on the distal
reects the relative motion of the two surfaces sides of these intersections. These changes reect
during frictional sliding. Of course the direction the mechanical interaction of the faults and are
of relative motion at a point may change as a fault consistent with elastic models of this interaction
develops, leading to overprinting of slickenlines (Maerten, 2000).
with different orientations. Where overprinting is Exposure of the faults at Chimney Rock are
not an issue, one should be able to dene a set of adequate to document the lateral variation in
three-dimensional curves lying on the fault rake of the slickenlines over a distance of almost
116 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
Small ~N
hut D 26
C C- B 20
B-
A2
Hinge line
A1
(a) (b)
100
z (m)
8 50
0
0 100 200 300 400 500 600
100
2000
z (m)
7 50
0
0 100 200 300 400 500 600
100
z (m)
6 50
1500 0
0 100 200 300 400 500 600
100
z (m)
5 50
0
0 100 200 300 400 500 600
100
4
z (m)
1000 50
0
y (m)
3 50
0
0 100 200 300 400 500 600
500 100
2
z (m)
50
0
0 100 200 300 400 500 600
100
z (m)
1 50
0
0 0 100 200 300 400 500 600
z 0 500 x (m)
x (m)
Fig 3.35 (a) Map view of Emigrant Gap anticline with 2529 not exposed. Eight cross sections of the folded
GPS survey points (Bergbauer, 2002). (b) Eight profiles of surface were used to constrain the shape of the
fold shape constrained by GPS data. surface (Fig. 3.35b). From these cross sections one
can infer that the anticline is asymmetric, with
folded surface could be sampled only in the south- the west ank exhibiting steeper dips than the
ern part of the map area, and the hinge only is east ank, and the hinge is rounded. The cross sec-
exposed and sampled near the northern termina- tions were constructed using GPS points located
tion of the outcrop. The GPS data were used to within 150 m of each cross section. Cross sections
create a digital model of the surface, despite the 1 through 6 are weakly constrained near the fold
limitation that the A1 sandstone is exposed only hinge, whereas cross sections 7 and 8 are weakly
over about 25% of the map area. Clearly, sufcient constrained on the fold limbs.
data control is lacking to make conclusive state- A model surface was constructed by interpola-
ments about the geometry where the surface is tion and ltering (Bergbauer and Pollard, 2004)
118 CHARACTERIZING STRUCTURES USING DIFFERENTIAL GEOMETRY
2.5
1500 1500
2
Curvature (m1)
y (m)
y (m)
1.5
1000 1000
1
0.5
500 0 500
0.5
Hinge line
Crest line
0 1 0
0 500 0 500
x (m) x (m)
Fig 3.36 (a) Map view of Emigrant Gap anticline with
contours of minimum principal normal curvature, 2, and tick parallel to the trend of the fold hinge, and
marks parallel to direction of minimum curvature (Bergbauer, signicantly change directions only across the
2002). (b) Contours of maximum principal normal curvature, gentle surface undulations. The directions of
1, and tick marks parallel to direction of maximum maximum curvature (white ticks, Fig. 3.36b) are
curvature. approximately perpendicular to the fold hinge.
It is standard practice in structural geology to
that is approximately cylindrical in shape and idealize folds as cylindrical structures. The prin-
exhibits a smoothly rounded hinge region. Gentle cipal normal curvatures provide a quantitative
surface undulations, which trend obliquely to the measure of the departure from a cylindrical
hinge line, are superimposed on the more or less shape. For a surface to be perfectly cylindrical one
cylindrical shape. Apart from these gentle undu- principal curvature must be zero everywhere, and
lations, the limbs of the modeled surface are the other principal curvature must have the same
approximately planar. Fig. 3.36 shows the distrib- distribution on every cross section taken perpen-
utions of the maximum, 1, and minimum, 2, dicular to the fold axis. Not only is the minimum
principal normal curvatures. Values of the principal curvature non-zero across the modeled
minimum curvature range from 1 e3 to 1 e3 bedding surface (Fig. 3.36a), the maximum curva-
m1 and values of the maximum curvature range ture distributions differ from one cross section to
from 1. 7 e4 to 3.8 e3 m1. Areas with elevated another (Fig. 3.36b). Discrimination of cylindrical
magnitudes of minimum curvature trend and non-cylindrical areas across the surface is
obliquely across the fold due to the gentle surface possible using the categorization depicted in Fig.
undulations. The directions of minimum curva- 3.32. Based on the signs of the Gaussian and mean
ture, shown as white ticks in Fig. 3.36a, are sub- curvatures at every grid point, the surface can be
3.4 CONCLUDING REMARKS 119
decomposed into areas that are locally shaped like tive geometry and stereographic projection did in
one or the other of the six different characteristic the second half of the twentieth century. This is
shapes (Bergbauer and Pollard, 2003). Performing not the intention of the authors. We view differ-
this analysis on the modeled bedding surface ential geometry as the appropriate mathematical
shows that the surface is composed primarily of machinery to characterize structures, but this
domal and saddle-like areas, which reect the characterization is just one step in an investiga-
gentle undulations superimposed on the broader tion which ultimately must include consideration
fold shape. The modeled bedding surface does of the constitutive properties of rock and models
not contain any cylindrically shaped areas of of deformation based on the equations of motion
signicant extent. We suggest that the characteri- (Guiton et al., 2003).
zation of folded surfaces using differential geom- Justications for learning differential geome-
etry will provide new insights concerning the try are several. The structures encountered in
process of folding (Fisher and Wiklerson, 2000; Earths crust are three dimensional with spatial
Lisle, 2000; Bergbauer and Pollard, 2004). variations in size and shape that only can be
accounted for using a geometry that involves the
spatial derivatives of such things as orientation and
3.4 Concluding remarks curvature. Plotting orientation data on a stereo-
graphic projection eliminates the opportunity
The objective of this chapter is to introduce struc- to visualize and analyze these spatial changes.
tural geologists to the elementary concepts of dif- Furthermore, to proceed with modeling one needs
ferential geometry that serve to characterize to write boundary conditions that refer explicitly
curves (lineations) and surfaces in three-dimen- to geometry of surfaces. Finally, we now have
sional space. One could imagine that these con- precise eld data on the three-dimensional shape
cepts and the tools that follow from them might of surfaces from new technology such as GPS and
capture the attention of structural geologists, we need to know how to describe these surfaces and
much as the concepts and tools related to descrip- how to compare them to a model result.
Chapter 4
Earths crust under the state of Texas being lifted by a crane. In physical science a rst essential step in the direction
Scaling laws demonstrate that the good state of Texas is of learning any subject is to nd principles of numeri-
utterly incapable of self-support. Reprinted from Hubbert cal reckoning and practical methods for measuring
(1945) by permission of the AAPG whose permission is some quality connected with it. I often say that when
required for further use. you can measure what you are speaking about, and
express it in numbers you know something about it;
but when you cannot measure it, when you cannot
express it in numbers your knowledge is of a meager
and unsatisfactory kind: it may be the beginning of
knowledge, but you have scarcely, in your thoughts,
advanced to the stage of science, whatever the matter
may be (Thomson, 1891).
4.1 PHYSICAL QUANTITIES AND THE CONTINUUM 121
I
n an insightful article about research in tools to understand the scaling of structural phe-
geology in the early twentieth century M. King nomena, and to set up scaled laboratory experi-
Hubbert refers to Sir William Thomson (Lord ments to model the development of structures.
Kelvin) as the Patron Saint of geologists, includ-
ing himself, who espouse a quantitative agenda,
and he cites this quotation from Thomson as their 4.1 Physical quantities and
guiding credo (Hubbert, 1974). Thomson is not
advocating numeration purely for the sake of col-
the continuum
lecting numbers; rather this is a call to measure
relevant physical quantities and express them as 4.1.1 Fundamental and derived quantities
numbers. Thomson was a physicist, not a geolo- Structural geology is concerned with deformation
gist, but Hubbert recognized the importance of of rock and this is largely a physical process,
quantication in the geological sciences and was although chemical processes can play important
a leader among geologists of his generation in this roles. Most of the physical quantities we use in
regard (Hubbert, 1972). In part, Hubbert was react- this textbook can be described in terms of four
ing to the popularity of descriptive taxonomy for fundamental quantities for mechanical systems,
geologists of the twentieth century, structural namely length, mass, time, and temperature.
geologists being no exception: an introductory Associated with each fundamental quantity are
textbook published in 1987 provides a glossary of actual objects (e.g. a cylinder of platinumiridium
terms with over 350 entries that beginning stu- alloy residing at Svres, France, and assigned a
dents might be expected to master (Dennis, 1987). mass of one kilogram), or devices with prescribed
In a playful reaction to the plethora of terms for procedures (e.g. a device to measure the duration
intrusive forms at mid-century Charles Hunt com- of 9 192 631 770 periods of radiation correspond-
ments on the feeder to the Trachyte Mesa laccol- ing to transitions of the cesium-133 atom and
ith: Because the form has certain resemblances to assigned a time of one second; Mechtly, 1973).
the woody structure of the cane cactus the name These are used as standards to dene the quanti-
cactolith might be used and dened as a quasi-hor- ties, and copies of the standards are used for
izontal chonolith composed of anastomosing duc- everyday measurement. For example, one would
toliths whose distal ends curl like a harpolith, thin compare an unknown mass to a copy of the stan-
like a sphenolith, or bulge discordantly like an dard kilogram using a balance. For structural
akmolith or ethmolith (Hunt, 1953). geologists most measurements are made using
Here we introduce some of the concepts and classical physical principles that predate relativ-
the tools necessary to practice structural geology ity, quantum mechanics, and the physics of
in a manner that Hubbert would have understood atomic and sub-atomic particles. As two modern
and Thomson would have appreciated. We begin physicists point out:
this chapter by dening the basic physical quanti- Observations are formulated in the language of classi-
ties used to describe and measure Earth struc- cal physics because that is the language used to record
tures, and agree on their units of measure. This measurements with macroscopic instruments. That
leads to a discussion of the continuum, the math- statement does not imply that the measuring instru-
ematical idealization that forms the basis for ments follow classical physics instead of quantum
most of our thinking about the spatial and tem- physics, a wrong opinion some writers ascribe incor-
poral variations of the relevant physical quanti- rectly to Bohr. Instead our statement implies that the
ties. These so-called eld quantities are dened at special nature, in particular the larger size, of measur-
every point in the continuum and are inferred to ing instruments allows the description of their behav-
ior in classical terms (Feshbach and Weisskopf, 1988).
be measurable in the rock mass. Next we consider
physical dimensions and explain how dimen- Such measurements determine a numerical value
sional analysis is used to check the consistency of for the physical quantity.
equations and to construct graphs of physical It is not the measured number itself that is
quantities. Dimensional analysis provides the useful, but rather that number in combination
122 PHYSICAL QUANTITIES, FIELDS, DIMENSIONS, AND SCALING
Note that proper names for the SI units, such as Unit weight
newton and kelvin, are not capitalized, but the 1m = 26670 N m3
1
corresponding units themselves, N and K, are cap-
italized. The unit of stress follows from the
2
concept of a force per unit area:
such a column, but we will evaluate those lateral Table 4.2. Selected SI prefixes and symbols.
constraints in a later chapter. A handy rule of
thumb is: the vertical stress in the Earths crust Prefix Symbol Multiple of
due to the weight of overlying rock increases with
depth at a rate of about 25 million pascals per giga G 109
thousand meters. mega M 106
A few units, not part of the ofcial SI system, kilo k 103
are in such common usage in the geological liter- deci d 101
ature that we refer to them throughout the text. centi c 102
The annum, a, is used for one year when measur- milli m 103
ing the age (time before present) of rocks and min-
micro 106
nano n 109
erals. The unit degree Celsius, C, is equivalent to
the unit kelvin, but the scales are offset such that
the number of degrees Celsius is less than the tables to facilitate unit conversion. These conver-
number of kelvin by the constant 273.15. Some sions take the form of the common examples
important derived quantities are made up of shown in Table 4.3.
ratios of fundamental quantities in which the
units cancel out. For example, one measure of 4.1.3 The material continuum
deformation called stretch, is dened as the nal Geometric and physical quantities used in struc-
length of a material line segment divided by its tural geology (e.g. strike and dip, mass density,
original length, so the stretch is devoid of units. stretch, displacement) usually are measured at
Because angles are dened as ratios of circular arc scattered locations or isolated exposures and the
lengths to radial lengths, they too are devoid of values so obtained commonly are used to charac-
units. However, it is customary to assign the unit terize a volume of rock that surrounds each
radian to angles. location. For example, in their monograph on
metamorphic tectonites Turner and Weiss (1963)
4.1.2 SI prefixes and conversion factors emphasize that one of the foundations of struc-
One of the beauties of the SI system is the ease tural analysis, as conceived by Bruno Sander in
with which quantities are manipulated in simple the second half of the twentieth century (Sander,
powers of ten by placing different prexes on the 1970), is the concept that a deformed rock mass is
units (Table 4.2). Other prexes exist that extend separable into volumes of statistically homoge-
the range of values both upward and downward, neous fabric that are investigated independently.
but these are less commonly called for in struc- One of the principal tools of such an investigation
tural geology. The prexes and symbols from is the stereonet on which orientation data are
Table 4.2 are attached to the front of the respec- plotted, devoid of any connection to location. This
tive unit or symbol. For example, using the viewpoint begs the question: how does the physi-
symbols k and M the rule of thumb stated in the cal quantity under investigation vary from one
previous section says that the vertical stress volume to an adjacent volume? The spatial varia-
increases with depth at a rate of 25 megapascals tion of physical quantities is unapproachable
per kilometer or 25 MPa km1. Because the ages using this method, in part because it avoids the
of rock formations typically fall in the range of use of calculus and the underlying principles of
millions of years the units are written mega- that mathematical discipline.
annum (Ma). The alternative, advocated here, is to embrace
Because there are many examples of archaic calculus and use it to investigate how physical
units in the literature of structural geology, one quantities such as the poles to planar elements,
needs to be procient converting to the SI system. temperature, velocity, and stress vary in space and
The CRC Handbook of Chemistry and Physics (Lide, time as structures evolve. In this context physical
2004) and The International System of Units (Mechtly, quantities are dened at a mathematical point by
1973) are useful references that contain extensive a limiting process in which an element of the
4.1 PHYSICAL QUANTITIES AND THE CONTINUUM 125
Table 4.3. Selected conversions from archaic to abandon this level of understanding until they had no
SI units. alternative. However, the development of physics in
the twentieth century has been a progressive move-
From To Multiply by ment away from visualizable models and toward
abstract mathematical models.
To convert length Physicists now talk of elds in a much more
inch meter 2.54102 abstract way than Faraday or Maxwell did. A eld is
foot meter 3.048 101 now thought of as a way of assigning numbers to a
mile meter 1.609 344103 region of space, much as a temperature map assigns a
temperature to every point on the earths surface.
To convert mass Although this description makes a eld seem very
kgf s2 m1 kilogram 9.806 65 abstract, it proves to be a very rich way of talking
pound mass kilogram 4.535 924101 about nature. In fact, physicists today talk about elds
(lbm) in exactly the same way as they talk about material
To convert time objects (Gregory, 1990).
hour second 3.60 103
annum second 3.153 6107 This viewpoint was of tremendous value to physi-
To convert temperature cists and engineers throughout the twentieth
Celsius kelvin T(K)T(C) century, but few structural geologists adopted
273.15 this perspective.
Fahrenheit kelvin T(K)(5/9)[T(F) Perhaps our most familiar experience with the
459.67] concept of a continuum comes with the realiza-
To convert force tion that given any two real numbers one can
kilogram newton 9.806 65 choose another that falls between the rst two.
force (kgf) Because of this property the set of real numbers is
pound newton 4.448 222 called continuous, or we would say it forms a con-
force (lbf) tinuum. This concept is applied every time we con-
dyne newton 1105 struct a graph of a continuous function and give a
scale to the ordinate and abscissa. We know that
To convert pressure, traction, or stress
atm pascal 1.01105 we can choose any scale for the axes and the func-
bar pascal 1.00 105 tion will plot without gaps. Because physicists
dyne/cm 2
pascal 1101 assert, based on intuition, that time and space can
2
lbf/in (psi) pascal 6.894 757103 be represented by real numbers, it is natural to
think of time and space as continuous. The fertile
To convert angle
imaginations of mathematicians have come up
degree radian 3.141 59/180
with functions that are discontinuous and some
of these have applications in structural geology.
For example, faults may be modeled as a surface
material is shrunk down about that point. We of discontinuity in a function for the displace-
understand that such a viewpoint cannot be taken ment eld. None-the-less, the displacement eld
literally if the element becomes too small (e.g. is adequately represented as continuous in the
smaller than a single pore in a sandstone), but the rock surrounding the fault.
denition provides the necessary mathematical Structural geologists seek to describe the
properties to interpret and explain geologic struc- motion of particles in a rock mass as it deforms
tures in which these point quantities vary contin- under the action of prescribed forces. It would be
uously in space and time. In other words these are useful to assign material properties or calculate
eld quantities dened in a material continuum. physical quantities at arbitrary points within
We feel we understand something when we can the rock mass. For this endeavor we construct a
picture how the wheels and levers must t together in material continuum for which the mass density,
order for it to work. Physicists were reluctant to momentum, and energy are well dened at every
126 PHYSICAL QUANTITIES, FIELDS, DIMENSIONS, AND SCALING
5000
Rock mass in the material continuum is:
density
4000
3000
lim
n
Mn
Vn
(4.9)
abstraction or idealization of nature and we must of force times displacement. Thus, when analyz-
address its limitations. As John Wheeler reminds ing the relationships among various quantities it
us in his forward to The Continuum: A Critical is instructive to consider the dimensions of those
Examination of the Foundation of Analysis by quantities. In this section we introduce the
Hermann Weyl: dimensions commonly encountered in mechani-
cal processes.
For the advancing army of physics, battling for many a
Dimensional analysis is a useful tool for
decade with heat and sound, elds and particles, gravi-
working with and understanding theoretical con-
tation and spacetime geometry, the cavalry of mathe-
matics, galloping out ahead, provided what it thought structs in all of science and engineering. The
to be the rationale for the real number system. En- paper by M. K. Hubbert (1937) puts the use of
counter with the quantum has taught us, however, dimensional analysis in a geological context and
that we acquire our knowledge in bits; that the contin- relies on the methods put forward in the book by
uum is forever beyond our reach. Yet for daily work the P. W. Bridgman (1931). In this section we use
concept of the continuum has been and will continue dimensional analysis to understand whether a
to be as indispensable for physics as it is for mathemat- given equation, which reportedly describes some
ics. In either eld of endeavor, in any given enterprise, aspect of rock deformation, is consistent from a
we can adopt the continuum and give up absolute dimensional point of view. If not, the equation is
rigor, or adopt rigor and give up the continuum, but
invalid and should be discarded. Then we intro-
we cant pursue both approaches at the same time in
duce the technique for plotting dimensionless
the same application (Weyl, 1987).
graphs and illustrate why this is the preferred
We refer to mathematical points at which density method to present scientic results.
is dened in a continuum mechanical model of a
faulted rock mass. Those points and the 4.2.1 Dimensionally homogeneous
motions or material properties attributed to equations
them, must be thought of as representative of a The dimensions of the fundamental mechanical
nite piece of rock, perhaps several cubic cen- quantities (length, mass, time, and temperature)
timeters in volume. are given as: L, M, T, and respectively. The
dimensions of derived quantities are composed of
products and powers of these fundamental
4.2 Physical dimensions and dimensions. Reading {} has dimensions of we
have, for example:
dimensional analysis
area, A{}L2 (4.10)
One can express physical quantities in terms of volume, V{}L3 (4.11)
many different units of measure and a particular displacement, u{}L (4.12)
quantity can take on very different numerical
values under the different systems of units. For velocity, v{}L T 1 (4.13)
example, 1 m 2.85 103 printer points 3.94 acceleration, a{}L T 2 (4.14)
101 inches 3.28 100 feet 4.97 103 furlongs
mass density, {}M L 3 (4.15)
6.21 104 miles. However, the underlying
physics must be independent of the choice of force, F{}M L T 2 (4.16)
units: it cant depend on the length of the Kings stress, {}M L 1 T2 (4.17)
Foot! This leads us to the concept that there is
thermal expansion, {}1 (4.18)
something more fundamental than the units
attached to a physical quantity and this is the phys- stretch, S{}L L 1 L0 1 (4.19)
ical dimension of that quantity. Regardless of the
3.141 592 65, . . . , {}1 (4.20)
units chosen, for example, for mechanical work
(newton meter, pound force foot, or dyne cen- Both the stretch and the angle are dimensionless
timeter) this physical quantity has the dimensions physical quantities, but we use the symbol l
128 PHYSICAL QUANTITIES, FIELDS, DIMENSIONS, AND SCALING
{}
event, or process. This procedure enables one to a
t
2 (Tw Ta)
check whether an equation might be in error by a w
being dimensionally inconsistent. This is probably
M L3 L2T1T
the rst thing you should do when confronted (1)(1)( )
M L3 M L3 1
with an unfamiliar equation, especially a very
complicated one. For example, suppose that you (M L3)(M1 L3)(1)() L2 T1 T
are reading the geological literature and you come
M0 L0 0 L2 T 0 L2 L (4.28)
across the following equation for the change in
elevation, E, of the seaoor due to cooling and Note that the sum or difference of two terms with
thermal contraction of oceanic crust (Fig. 4.3) as it the same dimensions can be shortened to a single
is transported away from a mid-ocean ridge by term with those same dimensions, so ( ) is
plate motion (Davis and Lister, 1974): written as (). Also, exponents are added for terms
that are multiplied, and these may cancel to
t
E a
2 (Tw Ta) (4.21) produce a dimensionless term, as in 1 0 1.
a w After simplifying the right-hand side we nd
4.2 PHYSICAL DIMENSIONS AND DIMENSIONAL ANALYSIS 129
0.2 m
0.4 m 2
1
G
(192 m),
x (m) 1
so 2 0.002 (4.36)
192 m 0.0 +192 m G
uG x2
1 2 (4.38)
2 a(1 ) a
0.2
0.8 kt/a2 = 0.00
Next we consider a model for heat conduction
Temperature, T/Tm
The model (Fig. 4.6b) is based on the solution contact, x/a 1, changes to Tm/2. As time increases,
for conduction of heat in one dimension, x, away temperatures in the model dike decrease and
from the tabular region a x a in an innite those in the immediate surroundings increase
body with homogeneous thermal properties and then decrease. For t/a2 5, corresponding to
(Lovering, 1935, 1936; Jaeger, 1957). The tabular a time of about 58 days for a dike 1 m thick, the
region is taken as a model for a dike of thickness temperature has dropped to about 20% of its
2a. The only thermal property is the diffusivity, , initial value and risen to a comparable tempera-
which typically has values near 1 106 m2 s1 for ture in the immediate surroundings.
rocks (Lee and Delaney, 1987). The initial condi-
tions (IC) on the temperature, T, are dened at the
arbitrary time, t 0, and the boundary condition 4.3 Dimensionless groups and the
(BC) on the temperature is dened at an innite
distance from the model dike:
scaling of structural processes
IC: for t 0,T Tm for a x a
Models of geologic structures provide insights
IC: for t 0, T 0 for a x a (4.39) about deformation in Earths crust, some of
which come from studying dimensionless groups
BC: at x , T 0 for all t
of variables. In this section we explore examples of
Here Tm is the initial temperature throughout the these dimensionless groups and show how they
model dike, the initial temperature is zero every- are used to understand the scaling of structural
where else, and very far from the model dike the processes. The direct method to identify dimension-
temperature remains zero for all times. A con- less groups considers the governing differential
stant ambient temperature, Ta, may be added to equations for the process and manipulates these
the solution for all positions and times. to isolate the dimensionless groups (Bird et al.,
The distribution of temperature in space, x, 1960, pp. 107, 185, 338). While the direct method
and time, t, normalized by the initial tempera- is preferred, it is not applicable if the governing
ture, Tm, is (Carslaw and Jaeger, 1959): equations are unknown or in doubt. In Chapter 12
we discuss the procedure for selecting the gov-
T(x, t) 1
Tm 2
erf
ax
2 t
erf
ax
2 t
,
erning equations and general boundary condi-
tions of a problem. We begin our discussion of the
x , t 0 (4.40) direct method with the bending of sedimentary
layers over a laccolith and then consider the ow
The function erf() is the error function whose of magma through a sill.
values are tabulated in reference books (Carslaw Next we introduce the Rayleigh method of
and Jaeger, 1959). Recalling from (4.26) that the dimensional analysis (Brodkey and Hershey, 1988)
dimensions of thermal diffusivity are L2 T1, the with an example that addresses the buoyant rise
terms in parentheses in (4.40) are dimensionless of salt through a sedimentary basin in an intru-
so the equation is dimensionally homogeneous. sive form called a diapir. This method does not
In Figure 4.6c the normalized temperature dis- rely on knowledge of the governing equations for
tribution is plotted as a function of distance from viscous ow, but does require a complete knowl-
the centerline of the dike using values of t/a2 as edge of all the variables relevant to the process.
a parameter that is a proxy for time. In this way, Underlying this method is a theorem introduced
for a given diffusivity and dike thickness, each by Buckingham (1915) and based on the necessity
curve represents the distribution of temperature for equations that describe physical processes to
for a particular snapshot in time. Note that the be dimensionally homogeneous. The Rayleigh
initial temperature eld is portrayed by the line method itself cannot assure one that the dimen-
labeled t/a2 0, and the eld for subsequent sionless groups so determined are correct, and an
times has successively greater values of this par- erroneous result will be found (with no warning)
ameter. In the rst instant the temperature at the if the number of variables is too few or too many.
4.3 DIMENSIONLESS GROUPS AND SCALING 133
Johnson, 1973; Koch et al., 1981), but these effects Next one substitutes the normalized variables
are ignored in order to introduce dimensional and differential operator into the differential
analysis using a simple model. These effects, in equation:
part, may account for the atter top of the
1 d4 u d4u*z 12p
Trachyte Mesa laccolith (Fig. 4.7a) as compared to (uou*z) 4o (4.44)
L d(x*)
4 4 L d(x*)4 BH3
the laboratory model laccolith with a single layer
(Fig. 4.7c). The nal step is to rearrange the equation to
The governing differential equation for the group the constants into a single dimensionless
bending plate model is presented without deriva- group:
tion (Johnson, 1970), because our purpose is to
demonstrate the direct method of dimensional
analysis. This equation is based upon simplifying
d4u*z
d(x*) 4
12
p L4
B uoH3 (4.45)
postulates about the kinematics of bending that The left-hand side of (4.45) is the dimensionless
are valid if the layer is thin compared to its length differential operator acting on the dimensionless
(Timoshenko and Woinowsky-Krieger, 1959). The dependent variable. The term in square brackets
differential equation for the vertical deection, on the right-hand side is the dimensionless group
uz, of the middle surface of the layer is: we have identied for this differential equation.
In some contexts dimensionless groups are
d4uz 12p
(4.41) referred to as the scale factors. Note that the
dx4 BH3
dimensionless group identied in (4.45) contains
The rst step in the analysis is to identify the vari- the elastic stiffness of the bent layer, B, the
ables, and to understand their roles in the physi- length, L, and height, H, of the layer, and the net
cal process. Here, the spatial coordinate, x, is the upward pressure, p, acting on the layer. The
only independent variable, and the vertical powers to which these quantities are raised in the
deection, uz, is the only dependent variable. A dimensionless group inform us about the relative
solution to (4.41) is uz f (x), a function that sensitivity of the deection, uz u*u z o
, to varia-
describes the distribution of deection with posi- tions in these physical quantities. For example,
tion along the layer. the deection scales directly with the fourth
The next step is to make the variables dimen- power of the length, L. Thus, all else being equal,
sionless (to normalize them) by dividing each by a two layers that differ in length by a factor of two
characteristic value of a quantity with the same would differ in deection by a factor of sixteen.
dimensions. The natural choice for normalizing x Changing the height also has a dramatic effect on
is L, the length of the layer in the x-direction. For the bending whereas changing the rock stiffness
a characteristic vertical deection, we choose the or the net upward pressure by a comparable
value at the center of the plate, uo uz (x 0). These factor has relatively little effect because the stiff-
choices are arbitrary, but are motivated by the ness, B, and the net pressure, p, enter the dimen-
geometry and the symmetry of the problem. The sionless group to the rst power. If the height is
normalized variables are written with a super- doubled, the deection decreases by a factor of
script *: eight, but if the stiffness is doubled the deection
is decreased by a factor of two. Similarly, doubling
x u the net pressure increases the deection by a
x* , u*z z (4.42)
L uo factor of two.
The differential operator in (4.41) also must be In this manner one can assess the importance
normalized. In this case d4/dx4 {} L4, and this of different physical quantities for the outcome of
operator is normalized using the length of the a tectonic process. Interestingly, this assessment
layer: does not require one to solve the differential equa-
tion (4.41). By determining the sensitivity of the
d4 d4 dependent variable to the various parameters that
L4 4 (4.43)
d(x*)4 dx affect that variable, one can design a strategy for
4.3 DIMENSIONLESS GROUPS AND SCALING 135
The governing equation for the ow of viscous malized by a physical quantity that shares the
magma in a sill is derived from two general prin- same dimensions. It is customary in uid dynam-
ciples, conservation of mass and conservation of ics to select a characteristic length and a charac-
momentum. The rst of these dictates that vx teristic velocity for this purpose. Here the only
cannot vary in the direction of ow, but it can characteristic length is the width of the conduit,
vary across the conduit in the z-direction, and it W. We select the velocity, vo, at the center of the
can vary in time. Thus, conditions on the velocity conduit to be characteristic. This is the maximum
components are: velocity, but the selection is arbitrary so we could
have selected the average velocity. The normalized
vx f (z, t) only,vy 0 vz (4.46)
variables are dened as:
The velocity component, vx, is one of the depen- x z v
x* , z* , t* o t ,
dent variables of this problem. W W W
The second principle, conservation of momen- v p po
v*x x , p* (4.48)
tum, introduces the forces acting on volume ele- vo v o2
ments of the magma. The decrease in pressure The characteristic velocity and distance are used
from one side of the element to the other intro- in the ratio vo/W to dene a dimensionless time.
duces a net force in the direction of this pressure Also, a reference pressure, po, is subtracted from
decrease. The pressure, p, is the second dependent the pressure and then the combination v 2o is used
variable in this problem and it can vary with posi- to normalize this reduced pressure. The reference
tion, x, along the direction of ow, and with time. pressure could be that at the entrance to the sill.
Viscous drag introduces another force on the The differential operators are normalized as
volume element and this is proportional to the follows:
viscosity, . The gravitational force acts in the ver-
tical direction, and therefore does not contribute W
, W ,
to ow in the horizontal conduit, and is ignored. t* vo t x* x
Under the restriction of constant density and vis- 2 2
W2 2 (4.49)
cosity, the pressure forces and viscous forces are (z*)2 z
capable of producing accelerations in the magma
The normalized variables (4.48) and differential
described by the equation:
operators (4.49) are substituted into the governing
vx p 2v equation (4.47) to nd:
2x (4.47)
t x z
vo 1
(v v* ) ( v 2o p* po)
This is a special case of the more general, W t* o x W x*
three-dimensional equations of motion called the 1 2
2 (v v* ) (4.50)
NavierStokes equations, developed by Navier in W (z*)2 o x
1822 (Bird et al., 1960, p. 81). The left-hand side of
Bringing the constants outside the derivatives
this equation is the mass per unit volume times
and eliminating the derivative of the constant ref-
the acceleration (time derivative of the velocity).
erence pressure this equation becomes:
The right-hand side is the sum of the pressure and
viscous forces per unit volume. In essence this vo2 v*x v 2 p* vo 2v*x
o (4.51)
equation is a specialized expression of Newtons W t* W x* W 2 (z*)2
Second Law of Motion written in the order ma F,
Note that the two combinations of physical con-
where m is the mass, a is the acceleration, and F is
stants in this equation are dimensional; they both
the net force acting on a uid element.
have dimensions of force per unit volume; and
There are two dependent variables, velocity
each is associated with the magnitude of a differ-
and pressure, and three independent variables,
ent force acting in the ow system:
the x- and z-coordinates and time in (4.47). In addi-
tion, there are two uid constants, mass density vo2
inertial force per unit volume (4.52)
and viscosity. Each of the variables must be nor- W
4.3 DIMENSIONLESS GROUPS AND SCALING 137
(a) are the density of the sphere, s, and the density
of the host uid, f . Since the sphere is rising
because of buoyancy, another parameter of this
problem must be the gravitational acceleration, g.
The velocity elds inside and outside the sphere
are complex, but here we focus only on the veloc-
ity of the sphere, v, relative to the static host at a
great distance from the sphere, and consider that
to be the dependent variable. For this conceptual
model we postulate that the ow is steady, so the
(b) velocity is constant and time does not enter the
rf , hf problem. Also we postulate that the ow is
isothermal, so heat transfer from the body to the
surroundings is ignored. We do not specify any
v distance scale that would place boundaries on the
rs , hs size of the surrounding uid mass. Conceptually,
the body rises forever in a host uid of innite
g
extent. Finally, the direction of rise is tacitly
assumed to be in the opposite direction of the
gravitational acceleration, so no coordinate axes
2R are explicitly required to dene this problem.
Despite all of the simplifying postulates made
in the previous paragraph, we have identied six
Fig 4.9 (a) Schematic diagram of salt diapers. (b) Stokes quantities that apparently affect the velocity of
model for a viscous sphere rising in a viscous fluid. Schematic
the sphere. In an experimental approach to this
diagram reprinted from Trusheim (1960) by permission of
problem, each quantity would be systematically
the AAPG whose permission is required for further use.
varied, as all others are held constant, in order to
discover their relationships. The number of exper-
and brittle solid that bent and fractured in iments would appear to be daunting; however,
response to the advancing magma. Masses of salt dimensional analysis helps to reduce the number
(or magma) whose rise is largely accommodated of variables for experimentation. The rst step is
by ductile ow of the surrounding rock are called to list all the physical quantities and identify their
diapirs. We illustrate the Rayleigh method of dimensions:
dimensional analysis in the context of diapirs by
radius of sphere, R{}L (4.56)
considering the slow rise of a buoyant viscous
sphere in another viscous uid of greater density relative velocity of sphere, v{}L T 1 (4.57)
(Fig. 4.9b). The solution by C. G. Stokes dates to the density of host fluid, f {}M L 3 (4.58)
middle of the nineteenth century and has found
density of sphere, s {}M L 3 (4.59)
innumerable applications in engineering and
science (White, 1974, p. 211). However, we analyze viscosity of host uid,f {}M L 1 T 1 (4.60)
this problem without the benet of the governing viscosity of sphere, s {}M L 1 T 1 (4.61)
equations or their solution by employing dimen-
acceleration of gravity, g{}L T 2 (4.62)
sional analysis.
A postulate, born out by the Stokes solution, is There are seven quantities in the three dimen-
that the rising body of viscous uid maintains a sions: length (L), mass (M), and time (T). We may
spherical form. Therefore the size and shape of reduce this number by making the additional
this body is completely specied by its radius, R. assumption that the densities and the accelera-
The viscosity of the sphere, s, and the viscosity of tion of gravity enter only through the difference
the host uid, f, are both considered constant, as in specic weights of the two uids:
4.3 DIMENSIONLESS GROUPS AND SCALING 139
difference in specic weights, groups and the exponents for these can be used to
determine the other three exponents.
g ( f s )g{}M L 2 T2 (4.63)
We choose the exponents c and e, and solve for
Thus, the revised list includes only ve indepen- the other exponents in terms of these:
dent quantities. Furthermore, the quantity mass
d c e
density times gravitational acceleration is a
measure of one of the forces acting in ow b 2c d e
regimes such as the rising diapir: 2c (c e) e c (4.69)
a b 2c d e
g gravitational force per unit volume (4.64)
( c) 2c (c e) e 2c
Recall that two other forces, inertial and viscous,
The exponents a, b, and d are removed from (4.65)
were dened in (4.52) and (4.53).
by substitution:
The Rayleigh method takes each of the ve
R2(g) c s e
independent quantities and raises it to an R2cvc ( g)c (ce) es
unknown integral or fractional exponent, here f v f f
given by the symbols a through e. An objective of constant (4.70)
the analysis is to determine these exponents and
The terms in square brackets are the two dimen-
use them to identify the dimensionless groups.
sionless groups for this process. The dimensional
The quantities, raised to these unknown powers,
analysis provides no additional information
are multiplied together and it is asserted that
about the values of the exponents. That informa-
their product is equal to a constant:
tion is discovered through laboratory experimen-
Ra v b ( g) c fd se constant (4.65) tation. However, we now have only two quantities
to work with instead of the original ve, so the
The appropriate dimensional expressions are sub-
design of the necessary experiments is greatly
stituted for the physical quantities in (4.65):
simplied.
La (L T1)b (M L2 T2)c (M L1 T1)d (M L1 T1)e The rst term in square brackets in (4.70) con-
La(Lb Tb )(Mc L2c T2c )(Md Ld Td )(Me Le Te) tains two measures of force per unit volume of
1 (4.66) sphere. The gravitational force per unit volume
(4.64) is proportional to the density difference
Because the product of the quantities raised to the between the host uid and the sphere, and to the
unknown powers is a constant, the product of the acceleration of gravity. The viscous force per unit
dimensional terms raised to these powers must be volume (4.53) is proportional to the viscosity of
dimensionless: the product must be equal to one. the host uid and the relative velocity, and
This implies that the product of each dimensional inversely proportional to the square of the sphere
term raised to the given powers is equal to one: radius. The rise of the sphere can thus be seen as
LaLbL2cLdLe Lab2cde 1 dependent upon a competition between the grav-
itational and viscous forces. Considering the
Mc Md Me Mcde 1 (4.67)
powers to which the variables are raised, we note
Tb T2c T d Te Tb2cde 1 that the relative velocity is most sensitive to
From (4.67) we conclude that the sum of the expo- changes in the radius of the sphere.
nents for each dimensional term is zero: The second dimensionless group identied in
(4.70) is the ratio of viscosities for the sphere and
a b 2c d e 0 host. Because the viscosity of the sphere might be
cde0 (4.68) either zero (an open hole) or innite (a rigid body),
b 2c d e 0 it is advisable to use the following dimensionless
group:
By this procedure we have reformulated the ve
e
unknown exponents into three equations. This f s
(4.71)
suggests that there are only two dimensionless f s
140 PHYSICAL QUANTITIES, FIELDS, DIMENSIONS, AND SCALING
b
quantity. The theorem states: ng
Lim
e
Lim
The number of dimensionless groups for a particular Th
ick
b
physical process is equal to the number of variables nes
,H
less the number of dimensions represented in those s, H
ss
ne
variables.
ick
Th
Viscosity, h Thickness, H
4.3.4 Dimension analysis applied to the Dxx
folding process Amplitude, A
A fundamental question in structural geology
concerns the length scale of structures compris-
ing an array. A notable example is that of an array Wavelength, L
of folds (Fig. 4.10a), but the question pertains to
many other structures. Convenient measures of
length scales for folds include the distance along Fig 4.10 Fold styles, terminology, and modeling.
a particular surface from hinge to hinge and the (a) Photograph of folds in metamorphic rocks. (b) Sketch of a
thickness between adjacent surfaces (Fig. 4.10b). portion of a fold illustrating the hinges and limbs, and
In a deformed terrain, folds will generally occur at suggesting field measurements of arc length and thickness.
many scales; from single layers a few millimeters (c) Model geometry and parameters described in the text.
Photograph by R. C. Fletcher.
in thickness, H, with arc lengths, La, of a few cen-
timeters to composite rock layers several kilo-
meters in thickness having arc lengths of ten or
more kilometers. adjacent surface traces bounding a fold are likely
The simpler example we address here is a to vary, as are thicknesses at different positions
folded layer embedded in deformed metamorphic along the fold limbs. However, average values
rock (Fig. 4.10a). Hinge-to-hinge arc lengths for two from multiple measurements along a train of
4.3 DIMENSIONLESS GROUPS AND SCALING 141
folds provide quantitative estimates of the regu- within this region at an instant of time, with the
larity of folding. The arc length is commonly, but aim of analyzing the rate of change of quantities
incorrectly, called the fold wavelength (Fig. 4.10b). of interest.
Our analysis considers the folded quartz vein to be The contrasting possibilities of interest are: (i)
isolated. What we principally mean by this is a negligible positive (or negative) value of the rate
that the folding is independent of that in nearby of change of fold amplitude, dA/dt, in which case
layers. That is, we do not see other nearby layers the layer will undergo nearly uniform thickening;
that are folded in concert with them. and (ii) a large positive value of dA/dt, correspond-
Note that the original vein thickness was not ing to the marked folding or buckling of the layer.
uniform and that this has affected the regularity To think about that, we have to imagine, in a con-
of the folding, but the arc length to thickness crete fashion, what the properties of the rocks
ratios fall within a modest range. What might involved are under the conditions that the folds
this regularity mean? Clearly, it cannot easily be formed. Clearly, they have deformed in a more or
ascribed to something built into the rock prior less continuous fashion. Although it takes a
to folding, since the regularity is present only in rather large leap, we might assume, for simplicity,
this layer. We must then ascribe it to some mech- they behave like other stiff, but still deformable,
anism inherent in the process of folding itself. uid-like media with which we are familiar, and
Before discovering an explanation, and, as an aid treat them as viscous uids. We then suppose the
to this, we consider a perfectly regular, periodic layer and medium have viscosities and 1,
structure of the same sort (Fig. 4.10c). This con- respectively.
sists of a layer embedded in a uniform medium, We now have a reasonably clear idea of a
with the upper and lower surfaces of the layer in model that might address some aspect of the
the form of in-phase sinusoidal surfaces with folding process so we turn to dimensional analy-
wavelength, L, and amplitude, A. The innite, per- sis. The quantities involved in the model, with
fectly periodic fold train is drawn with a low limb their dimensions, are:
dip.
The truly periodic fold train is an idealization rate of change in fold amplitude,
of the geometric form of a fold train and is used dAdt{}L T 1 (4.72)
here to isolate a small volume of rock containing bulk rate of shortening, Dxx {}T 1 (4.73)
a single fold from the remainder of the layer. fold amplitude, A{}L (4.74)
Because of the periodicity, the two vertical planes
layer thickness, H{}L (4.75)
in Fig. 4.10c are mirror planes of symmetry. For
wavelength, L{}L (4.76)
such a plane, two conditions apply. First, in the
deformation, which will be idealized to uniform layer viscosity, {}M L 1 T 1 (4.77)
layer-parallel shortening, except for the ow asso- medium viscosity, 1 {}M L 1 T 1 (4.78)
ciated with the folding, a particle cannot pass
through a mirror plane. Second, the shear stress There are seven physical quantities involving
must vanish at a mirror plane. three dimensions, M, L, and T, so the Buckingham
We model the case of a segment of the layer Theorem indicates that there are four dimen-
containing a single troughcresttrough fold sionless groups. Since M occurs only in the vis-
with a horizontal span given by the wavelength, L, cosities, one dimensionless group must be the
and an arc length, La L. To remind ourselves of viscosity ratio
the conditions that must apply at them, we 1
replace the bounding mirror planes by rigid R (4.79)
platens with smooth, frictionless, vertical sur-
faces. We suppose that these approach each other It is useful to choose groups that are relatively
at a rate corresponding to a rate of deformation, simple and that have a concrete physical or geo-
Dxx. As a natural starting point, we shall be con- metrical interpretation. An appealing choice is
cerned only in the elds of velocity and stress the aspect ratio, layer thickness to wavelength,
142 PHYSICAL QUANTITIES, FIELDS, DIMENSIONS, AND SCALING
which we write in the form: We term this ratio the dimensionless, or normal-
ized, rate of growth.
k 2
H
L
(4.80) We now write (4.83) as a function of the other
dimensionless groups:
Since 2/L is the wavenumber, we may call
dAdt
(4.80) a dimensionless or normalized wavenum- f (k, A, R) (4.84)
A |Dxx |
ber. The aspect ratio, H/L, is appealing because it
immediately brings to mind the data on the ratio This states that the rate of growth, normalized by
of thickness to arc length, H/La, which we obtained the rate of shortening, may be expressed, for any
to express fold regularity. Including the wave- pair of viscous materials for which the viscosity
number in (4.80) anticipates special results of ratio is given, as a function of the aspect ratio H/L
folding theory. or the dimensionless wavenumber, k, and the
As a third dimensionless group, we take the limb dip, A. If we are correct in supposing that
maximum slope of the folded surface, or the limb the folds, or the positions of fold hinges, are estab-
dip of the fold: lished when the deections of the layer are still
small, we may then limit ourselves to discovering
tan A 2
A
L
(4.81) something about the growth rate under the
restriction A 1. In that case this dimension-
Here is the maximum dip angle. With the aspect less group makes a negligible contribution to the
ratio, k, and the dip, , we may readily visualize relation and write (4.84):
the fold form. One might wonder why the wave- dAdt
length, L, rather than the arc length, La, has been q(k, R) (4.85)
A |Dxx |
chosen. The implied reason is that we have tacitly
assumed that in the process of folding the thick- Here q designates the unknown function.
ness/arc length ratios, which clearly involve a If we further suppose that the relative rate of
random element, are selected during the initial growth of the fold takes place rapidly, so that the
phase of folding, when the folds have low aspect ratio remains the same, k may be treated as
limb dip. In the geometric idealization of such a a constant. Hence, the function on the right-hand
fold the arc length approximates the wavelength, side will be a constant for a given pair of materi-
La L. als, R, and a given aspect ratio, k. As we will show
Only one dimensionless group remains to be later, k will vary according to k(t) k(0) exp(2Dxxt),
set. This must involve the rate of change of ampli- where k(0) is its initial value. But, if q is a constant,
tude, dA/dt, and the rate of deformation, Dxx, we may integrate the above equation to obtain the
which have so far been left out. Excluding the vis- description of fold amplitude growth:
cosities, only two quantities contain the time, so A(k, t) A(k, 0) exp(q |Dxx |t) (4.86)
the dimensionless group will have the form:
Here, we are led to think about initial fold ampli-
dAdt tude. Does this mean, somehow, that the regular
(4.82)
X |Dxx | length scale of folds in the fold train is pre-
Here X is a quantity with the dimension of length, gured in the form of the layer? The answer is yes,
and must be selected from the possibilities H, L, in the sense that we can hardly expect to get some-
and A. Since the anticipated result is that dA/dt thing, i.e. folding, if the layer is uniform in thick-
0, and shortening implies Dxx 0, we take its ness and has perfectly planar surfaces to begin
absolute value to avoid negative dimensionless with. However, the answer is no, if we suppose that
groups. We choose X A, so the dimensionless the initial layer form contains a kind of perfect
group is: although very greatly diminished template of the
nal fold train form. What is present initially is
dAdt imperfection in the form of a gentle irregular, or
(4.83)
A |Dxx | random, waviness of the layer surfaces.
4.4 SCALED LABORATORY MODELS 143
We know that such waviness can be broken results that we will obtain in a later chapter, this
up, mathematically, into its Fourier wavelength relation is
components. Here, we shall suppose merely for
the sake of simplicity in visualization and analy- Ld 2
(4.87)
sis, that such waviness can be thought of as an H (6R)1 3
innite sum or superposition of two-dimensional
A given value for (L/H)mean, implies a particular vis-
cylindrical sinusoidal waves with axes normal to
cosity ratio R 1/. Issues arise in judging the
a single direction of layer shortening, each with
validity of this result, including whether it is
a different wavelength, L. As with many other
appropriate to treat rocks under the conditions of
wave phenomena, we suppose, further, that each
deformation as viscous uids. However, it is satis-
component behaves independent of all the
fying to obtain information on the fundamental
others, at least in approximation. Then, the equa-
properties of rock by dimensional analysis. Such
tion for A(k, t) may be thought of as describing the
properties clearly cannot be inferred from eld
growth or amplication, of each component, as a
observations alone.
function of its k-value. If then, q(k, R) varies in a
suitable manner, and, in particular, if it has a
single maximum at some value of k kd corre-
sponding to the ratio Ld/H, the amplication of 4.4 Scaled laboratory models
the fold components will be selective. Here the
subscript d refers to the dominant wavelength. It seems inevitable that model experiments coupled
That is, as time or layer shortening goes on, the with theoretical analysis of the dynamics of tectonic
component at Ld/H will receive the greatest processes will contribute greatly to a sound, coherent
amplication, and the shape of the layer will be theory of structural geology and tectonics. By running
scale models of tectonic events, one may ultimately
dominated by a superposition of fold components
hope to separate the physically possible from the phys-
with L/H at or near this value. In this manner, the
ically impossible hypotheses, and the former may be
regularity seen in the nal conguration will be studied in detail to illustrate tectonic processes to an
established. extent not otherwise possible (Ramberg, 1967).
When the folded form of the layer, which will
inherit substantial irregularity from the random- Both the length scale and the time scale for many
ness in amplitude and phase of the initial wavi- tectonic processes make direct observation impos-
ness, reaches some stage in its development, the sible. In terms of length, we have no difculty
independent growth of individual wavelength observing the surface of the Earth at the neces-
components will cease, and the form established sary scale, but observations are extremely limited
at that point, as in the positions of fold hinges, at depth. Mines and wells are few and far between,
will be locked in. If folding or buckling accom- and modern imaging technologies (e.g. three-
plishes a shortening of the span of the layer with dimensional seismic reection), while vastly
minor changes in its thickness, the spacing of improved, typically provide data only from loca-
hinges along the layer, in terms of arc length, will tions of interest to the oil and gas industry. In
tend to be preserved. Thus, the data collected from terms of the time scale, some tectonic processes
our fold train may be referred back to the time of take millions of years to develop and their charac-
cessation of selective amplication. teristic rates prevent most attempts to monitor or
It is reasonable to suggest that the mean value investigate the phenomena directly. If, as Hans
of L/H provides an estimate of the value Ld/H, or the Ramberg suggests, we can make models in the lab-
so-called dominant wavelength/thickness ratio. If we oratory that reproduce these processes, we can
knew the form of the function q(k, R) we might directly observe the model structures as they
then nd Ld/H as a function of the viscosity ratio, develop and gain important insights. Rambergs
R, and thus use the L/H data to estimate the ratio opinion was written at a time when numerical
of the viscosities of the medium and the layer at models of tectonic processes were still under
the time the folding occurred. To anticipate development, and these now offer an alternative
144 PHYSICAL QUANTITIES, FIELDS, DIMENSIONS, AND SCALING
make a visual model of this outcrop that honors experiment. The time is measured from some arbi-
the geometry of the prototype using image pro- trary moment, often at the initiation of the
cessing software (Fig. 4.11b). Here the wavelength process, and the model ratio for time is dened as:
and height are halved in value, so Lm 106 mm and
Tm
Hm 6 mm. The model ratio, Lr, for the lengths is: Tr (4.91)
Tp
Lm 1
Lr (4.88) Given this ratio, one can compare the prototype
Lp 2
and a model at corresponding times during the
Any other length that we measure on this outcrop development of the process. We say that the pro-
is related to the corresponding length in the totype and the model are kinematically similar if
model by this same ratio. For example, for the they are geometrically similar at every corre-
height of the gray band we have: sponding time over the duration of the process. By
corresponding time we mean a time for the pro-
Hm 1
Lr (4.89) totype process and a time for the model process
Hp 2
that are related by the model ratio for time.
The prototype and model are geometrically similar. Kinematic similarity can be understood in
The image of the outcrop shown in Fig. 4.11a is terms of two motion pictures, one of the model
scaled down to a smaller size to t on the page of and the other of the prototype. Lets say the
this book. Thus, the lens cap displayed on the model ratio for time is Tr 3.1 1011, so each
gure is not 50 mm in diameter, but it represents model second represents one thousand prototype
50 mm on the actual exposure. The photographic years. The camera recording the model process
image is, itself, a geometrically scaled model of shoots at a speed of one frame per second and the
the exposure and the process of reproduction of camera recording the prototype process shoots at
this image maintains geometric similarity. a speed of one frame per thousand years. If each
Another model of the exposure is shown in successive pair of frames of the two motion pic-
Fig. 4.11c. Is this geometrically similar? The eld tures is geometrically similar, the two processes
of view has changed somewhat, but the image also are kinematically similar. The corresponding
seems to have many similarities to the prototype frames may have different length scales, but
shown in Fig. 4.11a. On the other hand, a careful lengths throughout the prototype and model
inspection reveals that the fold shape is distorted obey the model ratio for lengths.
relative to the prototype. We measure the wave- Turning to a geological example, it is likely
length in this model as Lm 106 mm and the that the layers shown in Fig. 4.11a had lesser
height as Hm 12 mm. Comparing the model amplitudes and greater wavelengths at an earlier
ratios we nd: time in the folding process. In Fig. 4.12 three dif-
ferent stages in the hypothetical one million year
Lm 1 H 1
Lr, m Lr (4.90) development of the prototype fold are illustrated,
Lp 2 Hp 1
assuming that the deformation conserved
The scaling in the horizontal direction is the same volume and that the shortening in the direction
as that used to produce Fig. 4.11b, but lengths of the measured wavelength is simply the recip-
measured in the vertical direction are the same as rocal of the elongation perpendicular to this
those in the prototype. Thus, the height of the direction. We start to record the process at an
layer in the prototype and model are identical. arbitrary time (0 s) shown in Fig. 4.12a. Five
This model does not preserve geometric simi- hundred thousand years into the process the pro-
larity. Comparing the lens cap in the three gures totype fold would look like the image in Fig.
conrms the scaling relations. 4.12b, and at the end of the one million years the
Tectonic processes may be very slow to develop, fold would have attained the shape observed in
but they are not static. Thus, the relative time outcrop today (Fig. 4.12c).
scales for the model, Tm, and the prototype, Tp, must For a kinematically similar model of this
be considered carefully when designing a model folding process using the model ratio proposed
146 PHYSICAL QUANTITIES, FIELDS, DIMENSIONS, AND SCALING
(b) Time, t1 Lm M T
Lr, m Mr, m Tr (4.92)
Lp Mp Tp
m Mm L3
m follows that any ratio of two different forces in
mass density, Mr L3
r (4.96)
p Mp L3
p the prototype must equal the corresponding ratio
of those different forces in the model. For
Fm Mm Lm T2
force, m
Mr Lr T2
r (4.97) example, considering a process in which inertial,
Fp Mp Lp T2
p viscous and gravitational forces are present, we
nd:
m Mm L1
m Tm
2
stress, Mr L1
r Tr
2 (4.98)
p Mp Lp Tp
1 2 Fmi Fpi
, or Rem Rep (4.100)
Fmv Fpv
The derived quantities are reduced to their equiv-
alent fundamental quantities and these are used Fmi Fpi
, or Frm Frp (4.101)
to dene the model ratios. Fmg Fpg
Notice that the model ratio for forces contains
the ratios for length, time, and mass. Length and Recall that Re is the dimensionless group called
time are considered through their respective the Reynolds Number, dened in (4.55). Dynamic
model ratios to assure geometric and kinematic similarity requires that the Reynolds Numbers for
similarity. Mass usually is not considered inde- the model and prototype be identical. These
pendently, but rather through the analysis of the numbers can be evaluated using a characteristic
force ratio, which is used to evaluate the dynamic length, wo, a characteristic velocity, vo, the density,
similarity between a model and the prototype. To , and the viscosity, , for both the model and pro-
assure similarity one must identify all of the dif- totype. The Froude Number, Fr, measures the rela-
ferent forces acting in the tectonic process under tive importance of inertial and gravitational
investigation. For example, in our study of the forces. From (4.52) and (4.64) we have:
dimensionless groups for ow of magma in a
v2o
conduit (Fig. 4.8) and rise of a salt diapir (Fig. 4.9) Froude Number Fr ,
gwo
we identied inertial forces (4.52) caused by the
change in velocity with time, viscous forces (4.53) inertial force
(4.102)
caused by the drag of magma against the side of gravitational force
the conduit, and gravitational forces (4.64)
The Froude Numbers for the model and prototype
related to a density contrast. A model and proto-
can be evaluated in terms of characteristic veloci-
type that are geometrically and kinematically
ties and lengths, and the acceleration of gravity,
similar, are said to be dynamically similar if the
to assure dynamic similarity.
model ratios for all of the forces acting on any two
This example points out a second important
corresponding particles are equal (Ramberg,
role for dimensionless groups in structural
1967, p. 4):
geology. We have already shown how useful they
Fmi Fmv Fmg are for understanding and interpreting the equa-
Fme
Fr (4.99) tions that govern mathematical models of tec-
Fpi Fpv Fpg Fpe
tonic processes. Now we see that they are useful in
the design of scaled laboratory model experi-
Here the subscripts m and p refer to the model
ments of these processes. Besides inertial, viscous,
and prototype as before, and the subscripts i, v, g,
and gravitational forces, there are likely to be
and e refer to inertial, viscous, gravitational, and
forces associated with spatial gradients in pres-
elastic forces, respectively. There may be other
sure or stress:
forces that should be considered.
It would be a daunting task to evaluate the p
forces acting on all corresponding particles for a pressure (stress) force per unit volume
wo
prototype and a model, but fortunately a simpler (4.103)
procedure usually is adequate. Because the ratios
of particular forces in the prototype and model all Here p refers to a characteristic change in pres-
must be equal to a common model ratio, Fr, it sure (or stress) from one location to another. These
148 PHYSICAL QUANTITIES, FIELDS, DIMENSIONS, AND SCALING
pwo
Stokes Number St ,
vow 2o
pressure force
(4.104)
viscous force
g
Ramberg Number Ra ,
vo w2o
gravitational force
(4.105)
viscous force
g
Smoluchowski Number Sm ,
pwo
gravitational force
(4.106)
pressure force
(b)
4 5
Ramberg has referred to the ratio of gravitation to
viscous forces as a nameless ratio but we 3 6
suggest the Ramberg Number in recognition of the
major contribution he made to tectonic model- 2
1 7
ing. Ramberg suggested the name for the
Smoluchowski Number because of the two papers 8
published by M. Smoluchowski in which a theory
for buckling of elastic layers on a viscous founda-
tion is developed along with associated model
Fig 4.13 Centrifuge models and equipment (Ramberg,
experiments (Smoluchowski, 1909). The Stokes
1967). (a) Model of salt dome development. (b) Centrifuge
Number is named for one of the pioneers in uid
used to increase the body force: 1, model in centrifuge cup;
dynamics who solved some of the classic prob- 2, stroboscopic light reflector; 3, camera; 4, camera
lems for slow viscous ow. For strict dynamic sim- electronics; 5, stroboscope; 6, temperature and speed
ilarity each of these numbers for the prototype control unit; 7, motor; 8, refrigerator.
must be equal to the corresponding number for
the model.
The procedure for tectonic model design may 2.16 103 kg m3
be further simplied when some of the character- g 9.8 m s2
istic forces are much smaller than others. For
example, inertial forces typically are much The radius of the prototype salt dome ranges
smaller than either the viscous or gravitational from 1.5 to 3 km, and the velocity ranges from
forces, so the Reynolds and Froude Numbers are 1 cm a1 to 1 m a1. These values are inferred
very small. For example, in the rise of a salt dome from eld observations and interpretations of
the relevant parameters take on the following geological records. The density and viscosity are
ranges (Ramberg, 1967, pp. 489): values estimated from laboratory measurements.
Density is well known, but viscosity is poorly
1.5 103 m wo 3 103 m constrained because it can vary over many orders
of magnitude depending upon temperature, pres-
3 1010 m s1 vo 3 107 m s1
sure, and chemical environment. Based on the
1 106 N s m2 1 1016 N s m2 (4.107) ranges of values given above, the Reynolds
4.4 SCALED LABORATORY MODELS 149
Thus, the model length and time ratios are not mg mLm Lm
(4.111)
independent. Solving for the model time: pg pLp Lp
150 PHYSICAL QUANTITIES, FIELDS, DIMENSIONS, AND SCALING
Here the ratio of densities in the model and pro- to increase the body force in the model by placing
totype is approximately one. This analysis leads to it in a centrifuge (Fig. 4.13b) (Ramberg, 1967;
the conclusion that the strength ratio must scale Dixon and Summers, 1985). For example, if the
as the length ratio. Using a reasonable range of dominant forces operating in the prototype are
model ratios for length (106 Lr 1) we conclude pressure, gravitational, and viscous forces, the
that the model ratios for strength must be in this Stokes (4.104), Ramberg (4.105), and Smolu-
same range (106 r 1). The upper end of this chowski (4.106) Numbers must be examined and
range corresponds to laboratory experiments on shown to be equivalent to the corresponding
meter-scale prototypes such as small folds and numbers for the model. Thus, comparing the
these can be effectively modeled with relatively Smoluchowski Numbers for the model and pro-
strong materials (Ramberg, 1963). At the lower totype, one must show that:
end of this range the experimenter must try to
model processes with length scales of kilometers
or tens of kilometers in the laboratory. If gravi-
g wo
p m
g wo
p p
(4.112)
tational forces induce the deformation, very weak If the gravitational force as measured by g is
materials are required for the model to meet the approximately the same in the prototype and
constraint imposed by the model ratio for model and the model ratio for lengths is 106,
strengths. then the model ratio for pressure (stress,
M. King Hubbert (1945) brought this point to strength) also must be 106 and this is not easily
the attention of geologists in an article entitled attained. On the other hand if the body force in
Strength of the Earth that examined the appar- the model can be articially increased in a cen-
ent contradiction between the great strength of trifuge, such that the model ratio for accelera-
a hand sample of rock and the modest strength tions is 103, then the model ratio for strength
of a huge rock mass containing innumerable need only be 103. Materials with strengths
folds and faults that witness to its apparent required to meet this constraint are readily
weakness. To drive home this point Hubbert pro- obtained for laboratory experiments. Similar con-
posed the operation illustrated in the fron- clusions about the appropriate viscosity for
tispiece to this chapter in which the state of model materials are found by examining the
Texas is lifted by a huge crane. Of course this Ramberg Number.
would be impossible to implement, so he consid-
ered a laboratory model of such an operation. By
considering the scaling of this model he con- 4.5 Concluding remarks
cluded that it would be impossible under the
existing force of gravity. In this chapter we introduced the material contin-
Consequently, if we tried to lift such a block in the uum, a construct that has produced astounding
manner indicated . . . the eyebolts would pull out; if we results in both fundamental physics and applied
should support it on a pair of saw horses, its middle engineering, including solid deformation, uid
would collapse; were we to place it upon a horizontal ow, and heat transport. Most human-made objects,
table, its sides would fall off. In fact, to lift it at all from automobiles to spacecraft, from bridges to
would require the use of a scoop shovel . . . The dams, from golf clubs to bicycles, are designed using
inescapable conclusion, therefore, is that the good continuum mechanical principles. Furthermore,
state of Texas is utterly incapable of self-support many of these objects are built on or with machines
(Hubbert, 1945).
that were, themselves, designed using these princi-
The scaling of model experiments has proven ples. The success of this way of thought should be
to be problematic because materials that ow beyond dispute, but curiously the application of
and fracture under their own weight are rare, yet continuum mechanics to structural geology lags
large masses of rock subject to gravitational considerably behind applications in other scientic
forces do just that. A solution to this dilemma is and engineering disciplines. This textbook is
4.5 CONCLUDING REMARKS 151
designed to encourage students of structural groups, and the scaling of tectonic processes. These
geology to use the concept of the material contin- should be readily available tools to take in the eld
uum. We also reviewed some bread and butter as mapping progresses and questions arise, and to
techniques for the practicing structural geologist use at the desk when puzzling over the result of a
including dimensional analysis, dimensionless computation.
Chapter 5
A clear separation between geometrical and dynamic deformation which can be considered independently
considerations was maintained by Becker, the of force, mass, chemical composition, elasticity, heat,
American geologist, at a very early date, and he magnetism and electricity; and it is of greatest use to
referred to the English physicist Thomson (Lord Kelvin), science for such properties to be considered as a rst
who says very clearly: We can see, therefore, that there step (Sander, 1970, p. 12).
are many attributes of movement, displacement, and
DEFORMATION AND FLOW 153
M
ost structural geologists think about (a)
deformation and ow in an inverse
problem mode: from the nal state back
toward the initial state of the deformed body of
rock. For example, the lower right photograph in
the frontispiece for this chapter shows deformed
oids with elliptical shapes: ratios of long to short
axes are about 1.56 (Cloos, 1947, 1971). The upper
left photograph shows nearly undeformed oids
with approximately circular shapes: ratios of long
to short axes are about 1.16. One can think of the
deformed oids being transformed back toward
an initial state much like that of the nearly unde-
formed oids. In other words the ellipsoidal par-
ticles become nearly spherical. This viewpoint is
natural since the primary observational data of
structural geologists are eld observations of
(b)
deformed rock. y
The above admonition by the Austrian geolo- P(x,y)
gist Bruno Sander, written in 1948 (Sander, 1970), b
a x
suggests that we set aside much of the physics in O
our initial study of rock deformation and focus ex-
clusively on kinematics. Becker and Sander are
prominent participants in the early history of
a
structural geology. Their advice is followed to this
date by an influential school of structural geolo- Fig 5.1 (a) Broken and extended fossil belemnite with
gists, to which the present authors do not sub- quartz, q, and calcite, c, filling. (b) The deformed shape of an
scribe (Fletcher and Pollard, 1999; Pollard, 2000). initial unit circle with radius of one undeformed belemnite
is an ellipse with semi-axes a and b. The trace of cleavage
Instead, we take kinematics as an integral part
(lineation) is parallel to the long axis of this strain ellipse and
of a complete mechanical analysis, including
the orientation and stretch of the belemnite are represented
enough physics to formulate a well-posed problem. by the radial line OP. Reprinted from Badoux (1963) with
Therefore we devote this chapter to the subject of permission of Universit de Lausanne.
kinematics and succeeding chapters to the other
elements of a complete mechanics.
Other types of observations enter a synthesis such processes as faulting, folding, or mountain
leading to understanding of a process of deforma- building at plate margins. Field observations of
tion and its products. For example, controlled many sorts, including the deformed shapes of
laboratory experiments produce detailed infor- objects such as oids may be selected to constrain
mation on the behavior of rock materials under a forward model built upon a synthesis of much
load, ranging from brittle failure at modest tem- previous information (Nur et al., 1986; Pachell,
perature and pressure to slow creeping ow at et al., 2003).
high temperature and pressure. Moreover, theor- From the more traditional perspective the
etical study of deformation and failure of a wide result of a process of deformation, such as one of
range of materials and human-made structures the deformed objects shown in Fig. 5.1 and in the
over the last two centuries has provided a rened frontispiece to this chapter, is described, and the
picture of the processes involved and the laws gov- relationship between its present and initial forms
erning them. The considerable insight and infor- is worked out. Applied to one or more such
mation provided by these additional sources tend objects, this procedure may lead to an estimate
to favor the formulation of a forward problem for of the strain of the rock containing them. The
154 DEFORMATION AND FLOW
procedure chiey involves considerations of geom- direction in which the rock has undergone the
etry, although assumptions must be made as to the greatest extension (Wood and Oertel, 1980). A
initial forms of the objects. However, as the above belemnite oriented parallel to the cleavage would
quotation suggests, no reference is made to the show the maximum value of the ratio l/l0, provid-
underlying physical process that resulted in the ing a measure of the bulk rock deformation in that
deformation. We begin this chapter by working direction.
through a few examples. Interpretation of the Estimation of the maximum stretch from the
examples involves many simplifying assumptions information available at the exposure, requires
that set aside complicating factors, which we other assumptions. One is that the stretch of the
return to after developing the other parts of a com- belemnite was equal to that of any other linear
plete mechanics in subsequent chapters. material line in the rock with the same orienta-
tion, for example, one lying along the line l on
the gure, or one much farther from the belem-
5.1 Rock deformation: some nite. This assumption implies homogeneity of defor-
mation within some volume of rock containing
observations and a simple the belemnite. In detail, the belemnite itself has
description not undergone a homogeneous deformation, but
has broken into pieces that have separated by
5.1.1 Deformed belemnite means of a process involving precipitation of
The closely lined segments of the stretched belem- material into the gaps. The belemnite might have
nite in Fig. 5.1a were initially connected to form an been more resistant to extension than the sur-
intact fossil (Badoux, 1963). First quartz and later rounding rock, and thus its stretch would under-
calcite was precipitated into the gaps between the estimate that of the rock.
segments as stretching occurred. Measuring the We have introduced the notion of homo-
nal span occupied by the segments, l, and initial geneous deformation in a simple way. As we show
length, l0, yields a measure of the elongation of a later in this chapter the technical denition of a
material line, the belemnite or its centerline. An deformation is the transformation that relates
assumption is that the segments did not stretch: the positions of particles in a body in one state to
the belemnite fractured into pieces that behaved as that in another. Here we speak of these two states
rigid objects embedded in a deformable medium, as the initial state and the nal state. A particle is
the rock containing it. The nal length, measured an element of mass within the rock of small
from the gure, is l 153/60", measured with a dimensions relative to the length scale of interest.
scale marked off in 60ths of an inch, and the sum Here, the only length scale is the length of the
of the segment lengths is l0 103/60". The ratio l/l0 belemnite, so small is taken relative to that. The
1.48 termed the stretch of a material line is a position of a particle is given by its position vector
dimensionless measure of the elongation. Thus, between a coordinate origin and the particle. In
any units may be used for the measurements. The this more formal context homogeneity is dened
use of fossil belemnites to estimate strain goes back as a linear transformation between positions in the
at least to the late-nineteenth century (Cloos, 1946; initial and nal states.
Hossain, 1979). The intent here is to estimate, from the elonga-
The motivation for study of the belemnite is not tion of the belemnite and the angle between it and
for its sake alone, but for what it might tell us cleavage, the maximum elongation in the plane of
about the deformation of the rock containing it. cleavage. We also assume, for this example, that
The belemnite lies in the bedding plane and is the plane of cleavage is normal to the bed contain-
inclined at an angle to cleavage, represented by ing the belemnite. The data given provide no basis
the parallel line segments in the gure. Much for this assumption, but allow us to consider only
research has shown that the plane of cleavage is the plane of particles making up the bedding
normal to the direction in the rock that has under- surface and thereby reduce the problem to two
gone the greatest shortening, and contains the dimensions.
5.1 EXAMPLES OF ROCK DEFORMATION 155
Diablerets Nappe
UH
NW
Ultra-Helvetic nappes SE
Morcles Nappe
Top of Cretaceous
Thrust contact
Top of Jurassic
Unit circle
Strain elipse (XZ)
0 1 2
km
Aiguilles Rouges Massif
3.
3.46
18
1.9
2.
4
37 728 E
2.
2.0
2. 69 B
9
5 91
3.4
5.
16 .31
4
630 E+F
728 H
4.4
85 B" 5.39 58-2
l0
(b)
l
Fig 5.3 (a) Folded quartz vein in schist; the lens cap is 5 cm
in diameter. (b) Sketch of fold with initial arc lenth, lo, and
final length, l. Photograph by R. C. Fletcher.
layer interfaces. To change them to this form, a the belemnite of Fig. 5.1 in the initial and nal
spatial distribution of relative motion between states. We refer these positions to coordinate axes
particles occurred. Each elliptical section in the xed in the bedding plane, with origin xed to a
gure represents the strain ellipse for some rock particular particle, e.g. the tip of the belemnite.
volume obtained by combining many oid shape Under the postulate of homogeneous deforma-
measurements, assuming homogeneous deforma- tion, the nal coordinates of a particle, x and y,
tion over the sample area, and that these objects, may be related to its initial coordinates, denoted
as the belemnite in Fig. 5.1, underwent the same X and Y, by linear equations describing a linear
deformation as the rock. If oids were stiffer than transformation:
the matrix around them, they would have a strain x x(X, Y ) AX BY
less than that of the bulk rock. The numbers given (5.3)
y y(X, Y ) CX DY
are the long to short axial ratios, a quantitative
measure of the strain. Since the initial ratio would Here A, B, C, and D are constants. The mathemati-
have been unity for spherical oids, a ratio of 4 cal statement x x(X, Y), where x occurs on both
represents an increase in the long diameter by a sides, is common in continuum mechanics and
factor of 2 and a decrease in the short diameter by other branches of physics. The x on the left-hand
a factor of 12 . side denotes the numerical value of the function
Data on the distribution of the strain in the x on the right-hand side for any pair of arguments
fold provide a constraint in addition to layer X, Y. To use another letter means that our descrip-
shape, on the deformation that took place. For tion will involve many additional letters and that
example, we might suppose that the fold was we will have to keep track of what they refer to,
formed when the rock layer was bent into a form e.g. if we had written x f(X, Y), we would have to
of roughly a semi-circle with little change in remember that f went with x. This way, the expres-
thickness. The ratio of the diameter to the cir- sion is self-referential. The inverse transforma-
cumference of a semi-circle, 2/, would then cor- tion, between the nal and initial positions, is:
respond to the bulk shortening of the rock mass X X (x, y) ax by
containing it. The square of the reciprocal of it, (5.4)
Y Y(x, y) cx dy
2.46, might then correspond to an oid axial ratio.
This value falls within the range of ratios given. The constants a, b, c, and d may be expressed in
Since bending implies extension at the upper terms of A, B, C, and D by solving (5.3) for X and Y
surface of the layer and contraction at the lower and comparing coefcients. We obtain:
surface, and such a distribution is not indicated
a D(AD BC)
by the data, this model for the folding is not con-
b B(AD BC)
sistent with the deformation of the olites. (5.5)
c C(AD BC)
The assumption that the oids maintain con-
d A(AD BC)
stant volume allowed Cloos to determine that
material lines normal to the plane of section kept Likewise, we nd:
their initial length (Cloos, 1947, 1971). The area of
A d(ad bc)
an ellipse is ab, where a and b are the lengths of
B b(ad bc)
the semi-axes, and (4/3)abc is the volume of an (5.6)
C c(ad bc)
ellipsoid, where c is the length of the plane-
D a(ad bc)
normal semi-axis. Scale the dimensions so that
abc a30 1, where a0 1 is the radius of the initial The particle at the origin of coordinates stays
sphere. Then, if b 1/a, c 1, and a material line in there.
the normal direction has not changed its length. In a homogeneous transformation, a straight
material line is transformed into another one
5.1.3 Linear transformations with different position, length, and orientation. A
Consider the positions of particles, such as those unit circle in the undeformed state, X 2Y 2 1, is
that might lie in the bedding plane occupied by transformed into an ellipse. We used this result,
158 DEFORMATION AND FLOW
(b)
Trilateration station
displacement with 2s error elipse
Trilateration station held fixed
2 m displacement
HAWAII
19o 30'
0 10 km
19o 20'
PACIFIC
OCEAN
155o 30' 155o 20' 155o 10' 155o 00' 154o 50'
0 8 16 km
N (a)
Chindamora batholith - rock types
Tonalite
Granodiorite
0 50 100 150 km Adamellite
Western adamellitic
granite
Greenstone belt
17o 30' S
0 10 km
Granitic batholiths
o
Greenstone belts 31 15' E
(b)
Finite strain
Fig 5.8 Array of batholiths in the Archean craton of
Zimbabwe. Reprinted from Ramsay (1989) with permission
from Elsevier.
1
showing the short and long axis are demonstrated 2
for each population. All xenolith axial lengths are
1
2
changed from their initial values.
Ramsay proposed an ingenious model to 2
explain the shape variation of the xenoliths in the 0 5 10 km
12
1
Chindamora batholith. The model addresses the
radial increase of deformation outward and suc-
1 strain elipse X/Y = X/Z
cessive emplacement of magma at the batholith 2
composite strains, 1 from greenstone
center. He supposed that the deformation of a xenoliths, 2 from tonalite xenoliths
xenolith begins when it is captured near the
solidication front in mostly crystallized material Fig 5.9 (a) Compositional zoning of the Chindamora
that is strong, but still hot and plastic. The xeno- batholith. (b) Distribution of representative elliptical sections
lith is then deformed with its host as further for local populations of xenoliths. Where two sections are
magma is injected into the center of the intru- shown, the one with greater eccentricity is for country rock
sion, inating it. He presents a quantitative model xenoliths embedded in xenoliths of the prior solidified
magmatic phase, that with smaller eccentricity is for the
of pluton ination, which he treats as an expand-
cognate xenoliths. Reprinted from Ramsay (1989) with
ing spherical or hemispherical body forcibly
permission from Elsevier.
deforming the surrounding greenstone.
Ination by means of magma supply from a
dike-like or columnar conduit does not seem a and the hemispherical ination model then appro-
plausible mechanism for the vast array of domical ximates the ow within the domes (Ramsay, 1989).
batholiths of the Zimbabwe greenstone belt. In an
alternative model, an entire layer underlying the 5.2.4 Inflation of a spherical shell: a
greenstone might have undergone melting or kinematic model
partial melting. Organized upwelling of domical Consider the ination of a spherical shell, or
masses of this layer might then have occurred (Fig. hollow sphere. Ination would be attributed to the
5.8). Ramsay makes a suggestion along these lines action of an internal pressure, but we are not
5.2 KINEMATIC MODELS, VELOCITY MODELS, AND DEFORMATION 163
=2
r
sional, depending only on radius. The object is the 4
r1
volume contained between two concentric spheri- 2 .0
r=
cal surfaces. We may readily grasp the nature of the R
R
deformation from Fig. 5.10a. Focus on two bound-
=1
R
.2
ing spherical surfaces, with initial radii R and R1. 1=
1
Let R1 be the radius of the surface occupied by the
2 1 0 1 2
xenolith when it starts to deform and R be a nearby
radius in the solid shell. The shell is assumed to (b)
have the same volume after deformation, when its
inner and outer radii are r1 and r, respectively. Then:
3
4 4
(r 3 r 31) (R3 R31),
3 3
or r 3 r 31 R3 R31 (5.8)
2
Since we are concerned with the deformation
of the xenolith, a body of very small dimensions
relative to the pluton, we let R R1 R. Since R
R and r r, we can throw out terms in 1
which R or r are squared or cubed. For example:
r 3 r 31 (r1 r)3 r 31
r 31 3r 21r 3r1(r)2 (r)3 r31 0
2 1 0 1 2
3r21r (5.9)
With such approximation, substitution into (5.8) Fig 5.10 (a) Expanding sphere (or hemisphere) model for
pluton emplacement (Ramsay, 1989). (b) Elliptical principal
yields:
cross sections of initial spheres with ratio of final radius to
radius at incorporation.
r R2
r21r R21R,or 21 (5.10)
R r1
semi-axes, a and c:
Written in terms of innitesimal quantities, the
relation is exact: dr 2c c
(5.13)
dr R21 r1d 2a a
(5.11)
dR r21 Eliminating d between these:
To describe the deformation of the xenolith, c dr R1
(5.14)
consider the small elements shown in Fig. 5.10a. a dR r1
Since we have not described how to treat the Substitution from (5.11) gives the desired result:
deformation of a sphere to an ellipsoid we take a
1 3
c R1
somewhat rougher approach. Suppose the initial (5.15)
a r1
spherical xenolith is just enclosed by the element.
Then the ratio of the elements dimensions must This states that the current position of the xeno-
be unity, or: lith, r1, and its shape, given by the ratio c/a, deter-
mine the initial position, R1. Xenolith cross
dR
1 (5.12) sections for values r1/R1 1, 1.5, 2, 2.5, and 3 are
R1d shown in Fig. 5.10b.
In the nal state, we suppose the element just How may this description of a single xenolith
encloses an ellipsoid with long and short principal be used to support or refute the notion that the
164 DEFORMATION AND FLOW
7 (a)
15 8
6 1
9
Axial ratio, a/c
10 0.8
10 5
1 2 3 4 t 11 0.6
15 0.4
5
a 20 0.2
gd 25
wg
1 0
0 5 10 15 20 25 30
Final radius, r1 0.2
0.4
Fig 5.11 Plot of xenolith axial ratio, a/c, versus final radius,
r1, for the Chindamora pluton. Data symbols refer to t, 0.6
tonalite; gd, granodiorite; a, adamellite; wg, western granite. 0.8
Reprinted from Ramsay (1989) with permission from Elsevier.
1
mechanism of pluton emplacement is balloon- 1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
ing? Ramsay plots the axial ratio, a/c, versus the
(b)
nal radius of the corresponding xenolith popu- 1
lation (Fig. 5.11). The points are scattered, but lie
0.8
in relatively localized regions. If the capture
radius, R1, is used as a parameter, each value of it 0.6
corresponds to a curve in r1, a/c-space. Only certain 0.4
of these curves will sweep through the regions for
0.2
each intrusive rock type. The range in values of R1
for these curves then corresponds to the position 0
of the partial solidication front at which the 0.2
xenolith is frozen into the expanding plastic shell. 0.4
The model of an inating pluton with spheri-
0.6
cal symmetry is a kinematic model because it pre-
scribes the motion of the particles within the 0.8
body from the initial to nal states. No informa- 1
tion is required beyond the spherical symmetry
during ination, and the concept of a variable 1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
radius at which xenoliths are incorporated into
Fig 5.12 A buoyant viscous sphere rising in a viscous fluid
the deforming shell. Given the t between the (Lamb, 1945). (a) Velocity vectors. (b) Streamlines followed
strain data from the eld and the model relation by particles. Surfaces across which particles do not move,
(Fig. 5.11), the model is a viable one. Can you and within which the fluid is confined during the rise, are
propose further tests of it? toroidal, like the external surface of a donut; the traces of
several such surfaces are shown.
5.2.5 Internal deformation in a rising
spherical diaper country rock. The model has also been used to
A second model of pluton emplacement has been simulate the rise of hot, buoyant diapirs, or
popular for over fty years (Grout, 1945; White- mantle plumes, through the mantle (Anderson,
head and Luther, 1975). Earlier studies used labo- 1975; Whitehead and Luther, 1975; Ribe and
ratory models and the principles of model scaling Christensen, 1999).
(Hubbert, 1937) discussed in Chapter 4. The model This model is not simply a kinematic model,
consists of the rise of an approximately spherical as the explicit involvement of materials of well-
mass of viscous uid, representing the pluton, in characterized behavior and properties, i.e. viscos-
another viscous uid, representing hot, plastic ity and density, and , and the acceleration of
5.2 KINEMATIC MODELS, VELOCITY MODELS, AND DEFORMATION 165
gravity, g, indicates. As we shall see, the velocity By symmetry, particles in the sphere move in
eld, although expressed by (5.19), is not easy to vertical planes through its center, so that it will
conceive of in an ad hoc manner. sufce to consider a description conned to one of
The remarkable solution of Stokes for a rigid these planes, the (x, y)-plane. Accordingly, the
sphere rising or sinking in a viscous uid was mathematical dependence of the solution is two
modied by others to treat the rise of a viscous dimensional. The velocity of a particle, referred to
sphere (Lamb, 1945). A sphere of viscous uid with the spatial coordinates, is independent of time.
density 1, intrusion viscosity 1, and radius R will The components of velocity, vx and vy , are:
rise in a viscous medium with density and vis-
2 2
cosity if the density difference, 1, is posi- r x xy
vx C 2 1 ,vy C (5.19)
tive. Experiment and theory show that the sphere R R r2
maintains its shape if the rate of rise is slow
where
enough (Fig. 5.12). We use the results to illustrate
the motion within the sphere, but specically to
determine the deformation within it and internal
r (x2 y2), C
1 ( 1)gR2
3 2 31
structures that might form between initial and
nal states, separated by an interval of rise. The velocity eld can be represented by the
The steady, or time-independent, rate of rise of stream function, a scalar quantity whose contours
the sphere is: are the streamlines (Fig. 5.12b):
2
r y
V
3
2 ( 1)gR2
( 1)
(2 31) (5.16) (x, y) C 1
R
r2 sin 2, sin
r
(5.20)
You are likely to be familiar with the special case Here is the angle in the section from the hori-
of a rigid sphere, 1 , sinking in a viscous uid, zontal. In the gure, contours of the stream
which may be used to estimate the rate of settling function at equal interval are plotted. The veloc-
of sediment particles in water: ity vector (Fig. 5.12a) is tangent to the stream-
lines and the speed, or magnitude of the velocity
V
9
2 ( 1)gR2
(5.17) vector, is inversely proportional to the contour
spacing. Thus, the maximum speed, relative to
The other limiting case applies, for example, to an origin at the center of the sphere, occurs at
the rise of a gas bubble, 1 0, in a viscous uid the surface of the sphere at its equator.
such as a basaltic magma: Because the ow within the sphere is steady in a
reference frame with origin at the sphere center,
V
3
1 ( 1)gR2
(5.18) particles remain on the streamlines as the
sphere rises. The steady internal motion is driven
The speed of rising or sinking only varies by a by the same steady rate of recovery of gravita-
factor of 3/2 in going from a rigid sphere (5.17) to tional potential energy that drives the rise of the
one that has low viscosity relative to its surround- sphere. The external ow is also steady if
ings, 1/ 1, (5.18). referred to the coordinate system rising with the
Position within the sphere is referred to sphere.
Cartesian coordinates that are xed at its center, To determine the nal, or current, position of
with coordinate x in the vertical direction of any particle in the sphere, for a given initial posi-
motion, and y and z in the horizontal plane. As the tion, we must follow it over the course of the inter-
ow continues, these positions will be occupied by val of rise using the relations (5.19). This is done
different particles. We may continue to think of the numerically, since, while the velocity at any posi-
spatial coordinates x, y, and z as giving the current tion is constant, particles move along paths
position of a particle which occupied an initial through the sphere along which the velocity
position with coordinates X, Y, and Z. We may not changes. If x(t; X, Y ) and y(t; X, Y ) are the coordi-
be able to specify the initial coordinates usefully. nates of the current position of a particle that
166 DEFORMATION AND FLOW
x(0; X, Y ) X 0.5
(5.21)
y(0; X, Y ) Y
(a) (a)
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0.2 0.2
0.4 0.4
0.6 0.6
0.8 0.8
1 1
1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1 1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
(b) (b)
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0.2 0.2
0.4 0.4
0.6 0.6
0.8 0.8
1 1
1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1 1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
Fig 5.14 A buoyant viscous sphere rising in a viscous fluid Fig 5.15 A buoyant viscous sphere rising in a viscous fluid
(Lamb, 1945). (a) Initial array of circular material lines in (Lamb, 1945). (a) Initial traces of equally spaced horizontal
vertical diametral plane of sphere. (b) Deformed final state material planes. (b) Deformed traces after rise of sphere of
array after rise of one sphere diameter. one diameter.
hemispherical volume, and a scheme adapted to a Thus, the stretching of a thin layer at the margin
rising sub-spherical intrusion might be worked of the intrusion must be the same as that of a
out. In any case, the more elaborate model should thin layer in the adjacent wall rock. The thin-
follow Stokes example and use a complete ning shown by the deformed initial spheres near
mechanics. the surface must substantially under-estimate the
Thinning of the wall rock at the top of the thinning at the contact with the overlying uid
sphere can be determined from the Stokes model. medium, since a large fraction of this has been
A boundary condition was imposed in the model greatly thinned to make room for the material
that required the uids to stick at their interface. represented by the grid. To estimate the thinning
168 DEFORMATION AND FLOW
at the contact, we use a method similar to that deformed internal contacts, and the stretching of
used to nd the ratio of deformed ellipsoid axes in the surrounding country rock. If these features
the spherical shell model. conform to the predictions of the model, they
Consider a pill-shaped element of material serve to support it. If they do not, a markedly dif-
whose upper surface coincides with the surface of ferent model must then be formulated. Either
the sphere at its apex. The element is taken to be outcome would advance our knowledge of the
sufciently small that the surface of the intrusion process of pluton emplacement.
is approximated by the horizontal tangent plane.
In the initial state, before any rise of the sphere,
the radius of the pill is Y and its thickness is X. 5.3 Relation between deformation
The volume of the pill is V (Y )2X. After some
amount of rise, in the nal state, the radius and
and velocity fields
thickness of the pill are y and x, and its volume
is v (y)2x. Since the volumes are equal, y/Y 5.3.1 Chevron folds
(x/X)1/2 . The nal form of the pill may be The folds, seen in cross section in Fig. 5.16, are
found by tracking the position of the particle with called chevron folds because of their straight limbs
initial position (R, Y ) on the tangent plane, or and narrow, sharp hinges (Ryan and Smith, 1998).
with initial position (R X, 0) on the vertical axis A chevron is composed of two stripes that meet
of the sphere. For the former, the velocity away at a sharp angle with the apex generally up, as
from the center point is: seen in insignia of rank on military uniforms.
Prominent quartz veins, in the form of saddle
d Cy d(y) C
vy(R, y) (y) , or dt (5.24) reefs occur at the crests of the folds. Many more
dt R y R
veins are present, including those along the limbs
Integration from t0, and between the limits Y and parallel to bedding, and in fault zones whose
and y corresponding to the initial and nal posi- location appears to be controlled in part by the
tions of the particle at the end of the pill diame- prior presence of the fold structure. Important,
ter, yields: also, is the remarkable dike that is emplaced
along the hinge surface of the anticline. The
ln
y
Y
C
R
y
t*, so exp
Y
C
R
t*
(5.25) quartz veins contain high-grade gold ore, so that
the process that formed them and determined
their distribution is of much commercial interest.
Here t* is the time of rise yielding the desired hor-
Here we consider how the folds might have
izontal stretching of the pill-shaped element. A
formed.
reduction in pill thickness by a factor of 0.1 cor-
Evidence for slip between layers in the form of
responds to an increase in pill radius of
slickenlines is found in association with the bed-
10 3.2, or:
parallel quartz veins. Slipping layers generally
C consist of many individual beds and are 10 m in
t* ln 10 1.15 (5.26)
R thickness. The length of the fold limbs in the
example from Bendigo, Australia (Ryan and
The distance of rise of the sphere is:
Smith, 1998), is about 300 to 400 m or more, so
100 m
NOVA SCOTIA
GOLDENVILLE
Discovery 1861
Grade 15g/t
Total Production 6.7 million grams
Au to 300 m 6.5 million grams
Depth 325 m (unknown 0 cm 3
below this depth)
CENTRAL VICTORIA
BENDIGO (Great Extended Hustlers) Fig 5.17 Chevron folds in phyllitearenite layers, Poudre
Discovery 1851 Canyon, Colorado. Photograph by R. C. Fletcher.
Grade 15g l1
Total Production 32 million grams
Au to 300 m 0.5 million grams
Depth 1200 m+
(a) y
H
d
L
(b) d
Step 1 Step 0
u
Step 2 d
r =R
r sin u
the faults. The present model affords a simple (X, Y )
example of such a situation, the faults being the u R
layer-parallel or bed-parallel slip surfaces. Since R sin
rock bodies may deform by the relative motion of
x
many approximately rigid elements, down to indi- R cos
vidual grains or grain fragments, this concept of
bulk strain has many applications. In the present r cosu
case, the slip across each interface is:
Fig 5.22 Rigid-body rotation about the origin through the
u H( tan tan 0) (5.30) positive, anticlockwise angle .
x r cos , y r sin
(5.35)
X R cos , Y R sin
X sin Y
1
cos
x FxxX
Fxx = 1 Fxx = 0.5
1 (5.39)
y FyyY Y
Fxx Fig 5.24 Initial chevrons (Fxx 1) and flattened chevrons
(Fxx 0.5). Circular markers become strain ellipses.
Here Fxx and Fyy are constants. The initial and nal
fold forms are two-dimensional cylindrical forms
with generator or axis parallel to the z-axis.
First consider what happens to a rectangular We may compute a relationship between the
element of area in the (x, y)-plane whose initial initial and nal dips, 0 and , or the ratio of the
corners are the points (0, 0), (0, X), (X, Y ), and (0, Y ). principal axes of the strain ellipse from the trans-
Its area is XY. In the deformed state, the element formation (5.39). Without imposing the restric-
remains a rectangle, with corners at (0, 0), (0,x), (x, tion of constant area:
y), and (0, y) and area:
b Fyy S
Thus, by choosing Fyy 1/Fxx in (5.39) we prescribe , Fxx (5.41)
a Fxx S0
that the cross-sectional area of the body, or any
part of it, remains the same. Referring to a broad Here b and a are the vertical and horizontal semi-
range of structure cross sections, structural geol- axes of the strain ellipse. In contrast to the ratio
ogists refer to this circumstance as one in which S/S0 obtained in the rotating rigid layer model,
area is conserved. Keep in mind that conservation of this quantity does describe a strain: here the
area is NOT a law of nature, such as conservation of change in length of a horizontal material line
mass. Does the De Sitter model conserve area? between the initial and nal states.
Application of the transformation (5.39) with The type of folding, or the fold mechanism, of
Fxx 1, to the conguration in Fig. 5.20, results in the homogeneous attening model is often
a tighter fold. An example is shown for Fxx 0.5. termed passive folding (Donath and Parker, 1974),
Fig. 5.24 shows a model of this type for Fxx 0.5, since it corresponds to a situation in which the
which starts with a chevron seed fold with a limb mechanical properties of layers or interfaces play
dip of 15. Several circles in the initial congura- no active role. It is as though the layers were com-
tion are deformed into ellipses. By our hypothesis posed of materials with the same isotropic
for the relationship between deformation and mechanical properties, which allowed for a stiff
cleavage, the model cleavage is oriented vertically, uid-like behavior, and surfaces of easy slip were
normal to the short axis of the ellipse. Application not present. One may also apply this model to dif-
of (5.39) requires a computation for individual ferent seed fold forms, such as that dened by a
lines, circles, and other loci. set of sinusoidal surfaces (Fig. 5.25). Since the
176 DEFORMATION AND FLOW
(a) 1 (5.42)
(b) y (1) X sin (1) Y
cos (1)
1 (1) (5.43)
y y
(c) Fxx =0.5, Fxy =0 F (2)
xx
Fxy =0, Fyy =1/Fxx
Substituting from (5.42) we have:
Fyx = 0, Fyy =1/Fxx 1 1 (5.44)
y X sin (1) Y
Fxx
(2) cos (1)
Fig 5.25 Folds with sinusoidal initial and final surface forms
produced by passive folding: (a) the seed fold, (b) flattening, Notice that the forms of the coefcients are:
(c) a combination of flattening and shearing. The coefficients
defining the deformations are given.
x Fxx X FxyY
(5.45)
y Fyx X FyyY
with such vertical cleavage, and with evidence for nal state, but of the continuous progression of
inter-layer slip, we might then conclude that our states between them. The term progressive deforma-
model is appropriate. If the cleavage were not ver- tion is often used by structural geologists to denote
tical except within the hinge region where sym- such a progression, whether or not it can be
metry requires it another kinematic model must described in detail (Ramsay and Huber, 1983, 1987).
be concocted. We do not claim that geometric, That is, any rock mass may be viewed as having
evolutionary, and kinematic models are the nal undergone a progressive deformation, the end
goal of our study, but they do provide a means of result of which is the suite of structures and
organizing observations and sorting out hypothe- deformed objects that we see in the exposure. On
ses as to folding mechanism. the other hand, the term progressive deformation
does not appear in the literature or texts of contin-
uum mechanics, which also deal with the defor-
5.4 Velocity fields: the mation of materials, including rocks. This term
seems to be the special invention of structural
instantaneous state of motion geologists. In the context of our discussion, the
description of a progressive deformation would
We have described the production of the limb of a appear to consist of a specication of the four
chevron fold in terms of the homogeneous defor- coefcients Fxx(t), Fxy(t), Fyx(t), and Fyy(t) as functions
mation, as described by the transformation from of time.
initial to nal coordinates of the particles in the Consider the rate of change in the positions of
body through relations of the form (5.45). The particles in the body. This description refers
quantities Fxx, Fxy, Fyx, and Fyy are the components explicitly to the physical variable time t, and thus
of a second-order tensor, the deformation gradi- leads toward a consideration of the physical
ent tensor. In the models so far developed, the processes responsible for folding. Taking the
coefcients have been obtained for two simple derivatives of the expressions in (5.45) with
kinds of deformation. Then, we obtained the respect to time:
coefcients for a sequence of two deformations in
dx dFxx dFxy
succession, one of each kind. This was consistent X Y
dt dt dt
with two observed features of chevron folds: inter- (5.48)
dy dFyx dFyy
layer slip and cleavage. However, this composite X Y
dt dt dt
model seems articial. It suggests a process in
which slip surfaces are initiated between rigid Since X and Y are the initial coordinates of the par-
layers and a fold forms in this manner up to a ticle, we do not operate on these. But the rates of
certain dip, at which point a completely different change of the current coordinates of the particles
mechanism of folding sets in. The model describes are their velocity components, or:
neither the initiation of slip nor its cessation. The
opposite sequence might also be considered. This dx dy
vx, vy (5.49)
composite process is plausible if the two mecha- dt dt
nisms operated during episodes separated in
The velocity eld at any time will depend on the
time, in which the conditions were markedly dif-
current conditions, properties, and forces
ferent. For example, lower temperature and pres-
applied to the body. Such a description is gener-
sure and the presence of uids might have been
ally developed in terms of position and time in
associated with an episode of inter-layer slip and
the body, so that in the two-dimensional case
higher temperature and pressure with an episode
under consideration:
of homogeneous deformation.
Since a parameter such as changes continu- vx vx(x, y, t, material properties, applied forces)
ously, the description obtained might be inter- vy vy(x, y, t, material properties, applied forces)
preted not only as the description of an initial and (5.50)
178 DEFORMATION AND FLOW
Here, x and y are used to describe the current coor- Lxx(Fxx X FxyY) Lxy(Fyx X FyyY)
dinates of some particle in the body, but they are
also used to describe position in space. The use of Collecting terms in X and Y and carrying out the
these quantities to describe current position of same operation using vy :
particular particles and position within the body dFxx
creates no problem. For example, the velocity at a Lxx Fxx Lxy Fyx
dt
particular position is an attribute of the particle dFxy
currently occupying this position. The mathemat- Lxx Fxy Lxy Fyy
dt
ical description of the variation in velocity with dFyx (5.53)
position may be smoothly varying. While we may Lyx Fxx Lyy Fyx
dt
formally evaluate the function elsewhere than dFyy
within the body, e.g. within a hole in the body, it Lyx Fxy Lyy Fyy
dt
is clear that the values of quantities there have no
physical signicance. To illustrate how we think about velocity elds of
The velocity eld within the body at a particu- the simple type (5.51) and how we incorporate
lar instant of time will depend upon the physical them into (5.53) and solve for the state of defor-
state of the body, its properties, and the forces mation, we go back to our two chevron fold
applied to it. Thus, rather than the current state models. We show how we can combine the two
of deformation, with reference to some initial models into one that allows the mechanisms of
conguration, it is this that we must look to if we attening and inter-layer slip and layer rotation to
want to understand the process producing the go on simultaneously. The key concept is that the
structure of interest. We will examine this rela- separate velocity elds for the two mechanisms at
tionship in this textbook. An example given in any instant are additive.
this chapter is that of the Stokes solution. Initial and nal positions of particles for
However, we also want to know how the history of attening are related by:
the instantaneous state of motion, or the velocity x Fxx X,y FyyY (5.54)
eld, gives rise to the current or nal structure;
and within the limited context of kinematics that The set of differential equations (5.53) is incom-
is something we may accomplish in this chapter. plete without a set of initial conditions. Initially,
In the present case of a homogeneous defor- the coordinates of particle position x and y are just
mation, as in a single limb of an idealized chevron equal to X and Y, so:
fold, the deformation is given by the coefcients
Fxx(0) 1,Fxy(0) 0,
Fxx, Fxy, Fyx, and Fyy. Moreover, the velocity eld (5.55)
Fyx(0) 0,Fyy(0) 1
within the fold limb where we again consider
the smooth equivalent of the rigid layer model
To have (5.54) at any time during the folding
can be written:
process, we must require that the coefcients Fxy
vx Lxxx Lxy y and Fyx are always zero. Examination of (5.53) indi-
(5.51)
vy Lyxx Lyy y cates that the quantities Lxy and Lyx must always be
zero, and (5.53) reduces to:
Here the coefcients are uniform in the limb.
Here, x and y denote both the spatial coordinates dFxx dFyy
Lxx Fxx, L yy Fyy (5.56)
and the current positions of particles. Combining dt dt
(5.45) and (5.48) through (5.51), we obtain relations
The equivalent velocity eld is:
for the rates of change of the components Fxx,. . . :
vx Lxx x,vy Lyy y (5.57)
vx
dx
dt
dFxx
dt
X
dFxy
dt
Y
To integrate (5.56) with the initial conditions
Lxxx Lxy y (5.52) (5.55), we need to know how Lxx and Lyy vary with
5.4 VELOCITY FIELDS: THE INSTANTANEOUS STATE OF MOTION 179
(5.60)
1 d
1
y Fyx X FyyY X sin Y vx x tan
cos dt
(5.63)
If this applies throughout the folding process, we
1
d
must require Fxy(t) 0. Since examination of the vy x(1 tan 2 ) y tan
dt
form (5.53) indicates that this requires that Lxy
Velocity elds for 0 and 22.5 are shown in
vanish, (5.53) reduces to:
Fig. 5.27. The result for 45 is the same as that
dFxx shown in Fig. 5.26 for the attening model! That
Lxx Fxx
dt this must be so is seen by substituting tan 1 for
dFyx 45 in (5.63).
Lyx Fxx Lyy Fyx (5.61)
dt The velocity eld must change as the dip
dFyy changes, and you can perhaps visualize how
Lyy Fyy
dt these examples correspond to a combination of
In (5.60), Fyy 1/Fxx, and the restriction Lyy Lxx smoothed inter-layer slip and simultaneous rigid
also holds. We may use (5.60) to compute the rotation of the entire limb. To aid visualization,
coefcients for the velocity eld in (5.61): we decompose the velocity eld into these two
component parts. The rigid-body rotation will
d
Lxx Lyy tan always have the same form, but the part corre-
dt (5.62) sponding to the sliding will vary in magnitude
d with limb dip. For example, this decomposition
Lyx (1 tan2 )
dt is done for 22.5 in Fig. 5.28. The decomposi-
These expressions depend upon the rate of change tion of (5.63) is obtained by noting that the rigid-
of limb dip with time, and a question immediately body rotation (Fig. 5.28b) is expressed as:
180 DEFORMATION AND FLOW
1
d (a)
(vx)rotation y
dt
(5.64)
1
d
(vy)rotation x
dt
1 dS
(1)
1 (1) been anticipated, is:
v (S, 0)
S dt S x
1 (1)
L S tan
d
(5.66)
1 dS
S dt
L(1)
xx Lxx Lxx
(2) (5.68)
S xx dt
A simple combined model is one in which the
For model 2: individual contributions are in constant ratio:
(2)
1 dS
S dt
L(2)
xx (5.67) L(1)
xx
Lxx
1R
, L(2)
xx
RLxx
1R
L(2)
,R xx
L(1)
xx
(5.69)
5.4 VELOCITY FIELDS: THE INSTANTANEOUS STATE OF MOTION 181
(a)
vy
1
1R
1
tan
tan x y Lxx (5.70)
Combining (5.73) and (5.74) with the expressions
(1)
for Lxx (2)
and Lxx :
182 DEFORMATION AND FLOW
y 60
1 50
x
0 R=0
d
40 1
tan d 2
The innitesimal material line is represented which applies when the square of the extension is
by the vector dX, and therefore by its length dS much less than the extension and much less than
[(dX)2 (dY)2]1/2 and direction , measured coun- one. Because the innitesimal strain is dened in
terclockwise from the positive X-axis such that the terms of the displacement gradients we write the
components are (Fig. 5.34b): quadratic elongation (5.85) in terms of the dis-
placement gradients using (5.82). For small strains
dX dS cos , dY dS sin (5.83) we dont have to pay attention to the distinction
The vector dx may be similarly expressed in terms between position (x, y) and position (X, Y). Then,
of quantities ds and : the derivative operators with respect to initial
coordinates, /X and /Y, may be replaced by the
dx ds cos ,dy ds sin (5.84) operators /x and /y. If the displacement gradi-
2 2 1/2 ents all are much less than one, we may approxi-
Here ds [(dx) (dy) ] is the length of the ma-
mate the quadratic elongation by discarding
terial line in the deformed state.
quantities in multiples of the gradients such as
Deformation, as we have seen examples of it,
(ux/x)2 or (ux/y) (uy/x) to nd:
involves strain and rotation. A familiar aspect of
strain is a change in the length of a material line
which may be quantied using the square of the
stretch, (ds/dS)2, a dimensionless quantity referred
n
1 ux uy
2 x
y
1 ux uy
2 x
y
cos 2
to as the quadratic elongation. For an arbitrarily ori-
ented material line segment of innitesimal
1 u y ux
2 x
y
sin 2 (5.87)
2
ds (dx)2 (dy)2
ux u y 1 u y ux
dS (dS)2 xx , yy , xy yx
x y 2 x y
(Fxx cos Fxy sin )2
(5.88)
(Fyx cos Fyy sin )2
(5.85) Then, the extension at a point in the direction
12 (Fxx
2 F2 F2 F2 )
xy yx yy may be written in terms of these components by
substitution into (5.87):
12 (Fxx
2 F 2 F 2 F 2 ) cos 2
xy yx yy
(Fxx Fxy Fyx Fyy) sin 2 n 12 (xx yy) 12 (xx yy) cos 2 xy sin 2
(5.89)
The last two lines are found using the standard
double angle formulae (Selby, 1975). This is an The two-dimensional innitesimal strain tensor com-
exact description of change in length in two ponents form a symmetric array because xy yx:
dimensions at a point in terms of the deformation
gradients and the orientation of the material line
in the initial state.
xx
yx
xy
yy (5.90)
Another useful measure of change in line If all its components vanish, no innitesimal
length is the extension, dened as n (ds dS)/dS material line element at the point under consid-
(ds/dS) 1. The extension is related to the qua- eration will undergo a change in length.
dratic elongation as: Turning now to the concept of rotation, a
rigid-body rotation in the neighborhood of a par-
2
ds
(1 n)2 1 2n 2n 1 2n (5.86) ticle is illustrated in Fig. 5.35. An arbitrarily ori-
dS
ented innitesimal material line represented by
In the last step we have utilized the approxima- the vector dX in the initial state is rotated
tion taken to dene the so-called innitesimal strain through the angle , and there represented by
5.5 GENERAL RESULTS 187
Y, y
sin 16 3
dS cos ( + v)
cos 1 12 2 1 (5.94)
dx
u dX
displacement gradients are related to the
ds
ux ux
dS cos
X, x x
uy
y
uy
0
0 (5.95)
x y
Fig 5.35 Pure rotation through the positive,
counterclockwise, angle . Infinitesimal vectors dX
and dx lie along a material line segment in the initial and final Combining (5.88) for the innitesimal strain
states, respectively. The vectors emanate from the same and (5.95) for the innitesimal pure rotation, the
particle fixed at the origin. The vectors may be represented displacement gradients may be expanded into
in terms of polar coordinates (dS, ) and (ds, ), respectively, symmetric and antisymmetric parts:
where dS ds.
ux 1 uy ux
x 2 x y
dx. The material line does not change length, so 1 uy ux uy
that ds dS. If all such material lines behave this
2 x y y
way, the local deformation is a pure rotation. From
Fig. 5.35:
1 uy ux
dx dS cos ( ) 0
(5.91) 2 x y
dS( cos cos sin sin ) 1 uy ux
0
Repeating these steps for dy and associating the 2 x y
trigonometric functions of the angle of rotation,
with the components of Fij we have:
Fxx cos , Fxy sin
xx
yx
xy
yy
0
0 (5.96)
(5.92)
Fyx sin , Fyy cos Here the innitesimal strain is symmetric and the
innitesimal rotation is antisymmetric.
Substitution of these expressions into (5.85)
When the strain and rotation are small, or
conrms the condition ds/dS 1.
innitesimal in the sense that squares of quanti-
The displacement gradients for pure rotation
ties can be discarded, special results apply. The
are obtained by substituting (5.91) and (5.92) into
deformation can be broken up into a rotation and
(5.80) and (5.82):
a strain, applied in either order, since the contri-
ux u butions are additive:
1 cos , x sin
X Y
u y
X
uy
sin , 1 cos
Y
(5.93) Fxx
Fyx
Fxy
Fyy
1 xx
yx
xy
1 yy (5.97)
If the rotation is small, 1, then the series As we shall see in later chapters, small strains
expansion of the trigonometric functions gives: and rotations are signicant in elasticity and its
188 DEFORMATION AND FLOW
application in models for the formation of joints, In the deformed state the vectors have the compo-
dikes, faults, or other structures. nents:
Notice that there are two applications of the
dx1 (Fxx cos Fxy sin)dS
innitesimal concept in the preceding discus-
sion. The rst is introduced in calculus and con- dy1 (Fyx cos Fyy sin )dS
cerns the treatment of a material line element (5.100)
dx2 (Fxx sin Fxy cos )dS
of arbitrary orientation that is represented by
the innitesimal vector dX with components dy2 (Fyx sin Fyy cos )dS
(dX, dY) in the initial state and dx with compo-
The cosine of the angle between these two vectors
nents (dx, dy) in the deformed state. These
is obtained by forming their scalar product:
vectors are contained within a neighborhood of
the particle or spatial point from which they dx1 dx2 (dx1)(dx2) (dy1)(dy2)
emanate that is sufciently small so one may
ignore the non-linear terms in the Taylor series (dx1)2 (dy1)2 (dx2)2 (dy2)2 cos
expansion about that particle or point. For (5.101)
example we have:
In the second line is the angle between the
x
X
x 1 2x
dx dX dY 2 X2 (dX)
Y
2 vectors. Substituting from (5.100) we have:
2
2x
XY
2x
(dX)(dY) 2 (dY )2
Y (5.98)
cos
1
2
Fxx
2 F 2 F 2 F 2 sin 2
xy yx yy
For (5.79) only the rst-derivative terms were kept.
The resulting measures of strain at a point, such
(Fxx Fxy Fyx Fyy) cos 2
as the quadratic elongation (5.85), are exact and
involve no approximations and therefore no
1 2
(F F 2 F 2 F 2 )
2 xx xy yx yy
errors of analysis. The second application is in the 1 2
simplifying approximation of innitesimal strain (Fxx 2 F 2 F 2 ) cos 2
Fxy yx yy
2
and rotation. If n 1 and 1, we discard
12
terms higher than rst-order in these quantities. (Fxx Fxy Fyx Fyy) sin 2
This introduces errors of analysis which may or
may not be tolerable (see Section 5.5.3).
An aspect of strain that is quite distinct from
1 2
(F F 2 F 2 F 2 )
2 xx xy yx yy
the change in length of material lines is shear. 1 2
Shear is a measure of the change in angle between (Fxx 2 F 2 F 2 ) cos 2
Fxy yx yy
2
two initially perpendicular material line ele-
12 1
ments (Fig. 5.36). Consider two initial innitesimal (Fxx Fxy Fyx Fyy) sin 2
(5.102)
vectors directed at right-angles to each other: dX1
with components (dX1, dY1) and dX2 with compo-
The change in angle between the two vectors is
nents (dX2, dY2). The rst innitesimal vector is ori-
the angle of shear, or /2 , where:
ented at the arbitrary angle . Since we are only
interested in the angle between the material line
elements in the deformed state, both vectors are
given the same length dS. In the initial state the
cos cos
2
sin (5.103)
vectors have the components: Considering only the angle change between the
two orthogonal material lines for the case 0,
dX1 dS cos ,dY1 dS sin the components of the two representative vectors
(5.99)
dX2 dS sin ,dY2 dS cos (5.99) reduce to:
5.5 GENERAL RESULTS 189
xk xk
vector X with components Xi (X1, X2, X3) in the CIJ (5.111)
XI XJ
initial state that is located by position vector x with
components xi (x1, x2, x3) in the current state. An A counterpart to this tensor, which is used to
arbitrarily oriented material line extending from compute (dS)2 using partial derivatives of the
that particle in the initial state along the innites- initial coordinates with respect to the current
imal vector, dX, with components (dX1, dX2, dX3) is coordinates, is referred to as the Cauchy deforma-
translated, rotated, and stretched to lie along the tion tensor (Malvern, 1969).
innitesimal vector, dx, with components (dx1, In the Lagrangian formulation the change in
dx2, dx3) in the current state. Innitesimal refers to the squared length of the material line is related
the fact that, if the vector were not innitesimal in to the deformation gradients as:
length, the material line element would be curved
in the current state and the vector dx would not
provide an adequate representation. In other
(ds)2 (dS)2 dXI
xk xk
XI XJ
IJ dXJ (5.112)
answer means that the strain should be described displacement gradients referred to the initial
using the Lagrangian strain tensor (5.113) or its coordinates. Under these conditions the differ-
Eulerian counterpart. To address this question ences between partial derivatives taken with
one can, for example, compare the Lagrangian respect to the initial coordinates and those taken
and innitesimal strains for a given set of defor- with respect to the current coordinates are negli-
mation or displacement gradients and calculate gible, so the distinctions made here between the
the error. Whether or not that error is acceptable two sets of coordinates are ignored and the strain
depends upon the application. If one is looking for components are written:
order of magnitude results from a calculation,
large errors may be tolerable. If one is working
under strict engineering guidelines, only very
ij
1 ui uj
2 xj xi (5.118)
1 ui uj X Y
ij (5.116)
2 Xj Xi
The components of the Lagrangian strain (5.115)
The error introduced by using (5.116) instead of are:
(5.115) may be dened as:
Exx 0.30, Exy 0, Eyx 0, Eyy 0.18
Eij ij (5.120)
eij 100 (5.117)
Eij
The components of the innitesimal strain (5.116)
This evaluates the error, eij, on a component by are:
component basis.
If squares and products of the displacement xx 0.25, xy 0, yx 0, yy 0.20
gradients are small enough compared with the (5.121)
gradients themselves to result in tolerable errors The errors in the two normal components of
then the last term on the right-hand side of (5.115) strain are found using (5.117):
may be dropped. What is left are the components
of the innitesimal strain written in terms of exx 17%, eyy 11% (5.122)
192 DEFORMATION AND FLOW
38o50'
ux ux (y, z), uy 0, uz 0 (5.123)
Fig 5.37 Map of the horizontal displacement of twelve
monuments near Pt. Arena, CA, during the 1906 San The displacement of monument 5, adjacent to the
Francisco earthquake (Lawson, 1908; Pollard and Segall, fault zone, was about 2.5 m and the displacement
1987). of monument 1, at a distance of 13.5 km from the
fault zone, was about 1 m. Using these values, and
an estimated depth of faulting of 12.5 km, we esti-
In contrast the errors associated with the less mate the displacement gradients as:
deformed sample are 7% and 4%, respectively.
Whether or not these errors are tolerable depends ux ux 1.5 m
1.1 104
upon the magnitude of errors associated with y y 13.5 103 m
data collection and the application of the analysis. ux ux 2.5 m
2.0 104
As a second example consider the displace- z z 12.5 103 m
ment eld associated with faulting during the (5.124)
great San Francisco earthquake of 1906 along the
San Andreas Fault (Fig. 5.37). In the vicinity of Although the slip on the fault was several meters,
Point Arena near the northern-most trace of the the displacement gradients are so small that
fault the relative horizontal displacements of squares and products of these gradients may be
monuments were calculated from triangulation neglected. This conclusion admits use of innites-
surveys taken before and after the earthquake imal strains and suggests that linear elasticity
(Lawson, 1908). The displacement vectors on this theory would be an appropriate tool to investigate
map demonstrate that slip on the fault was right the faulting process (Pollard and Segall, 1987).
lateral, that the offset across the fault zone was
about 4 m, and that Earths surface displaced
during the earthquake at least 15 km away from
5.6 Concluding remarks
the fault by about 1 m. Can we use innitesimal
strains to characterize this deformation? In this chapter, we examined deformed objects in
We choose a Cartesian coordinate system and rocks belemnites, folded veins, concretions, and
reference frame on the trace of the fault with the xenoliths and indicated how a quantitative
x-axis horizontal and parallel to the trace, the y- measure of strain might be obtained from mea-
5.6 CONCLUDING REMARKS 193
surements of them. Several examples demonstrate between observations of geometry and strain and
that strain typically has a continuous variation detailed models of structural evolution, we con-
within structures, providing one constraint on the sidered geometric and ad hoc kinematic models of
process of formation. More detailed treatment of chevron folds. We study a mechanical model for
practical methods of strain estimation from their origin in a later chapter. The kinematic
deformed geological objects are given in other model presented here provides a simple example
textbooks of structural geology (Ramsay and with which to illustrate the formal treatment of
Huber, 1983). Interpretation of deformed xenoliths strain and rotation as integrals that follow a par-
in the Chindamora batholith by means of an ticle moving through a temporally and possibly
appealing, but ad hoc, kinematic model illustrates spatially varying velocity eld. In the rising viscous
a methodology aimed at constraining the mecha- sphere example, strain ellipses were computed by
nism of intrusion by means of strain measure- following dense sets of particles by numerical
ments. In contrast, we showed how a simple means. The chapter concluded with a more formal
steady-state distribution of velocity in a rising treatment of plane deformation in two dimen-
diapir, provided by the complete mechanical sions, and descriptions of deformation and strain
model afforded by the Stokes solution, might be in three dimensions. In this discussion we point
used to follow the positions of particles, thus gen- out how to evaluate the errors expected when one
erating a detailed picture of the evolution of the chooses to employ the innitesimal strain in the
strain distribution in the body. As another bridge analysis of problems in structural geology.
Chapter 6
Photoelastic image of maximum shear stress contours in The concept of stress is the heart of our subject. It is
grains of model rock. Stress is concentrated at grain the unique way continuum mechanics has for specify-
contacts. Inset: photoelastic image of three circular disks ing the interaction between one part of a material
with point contact loads. Reprinted from Gallagher et al. body and another (Fung, 1969, p. 41).
(1974) with permission from Elsevier.
FORCE, TRACTION, AND STRESS 195
I
n this chapter we dene the relationships (a) y (b) y
F
among forces, tractions, and stresses. One of
f
the rst concepts encountered in a physics
a
class is that of the resultant force, F, acting on a
particle with mass, m, and the associated linear r
x z x
acceleration, a, of that particle in the direction m
that the force acts (Fig. 6.1a). For a rigid body (Fig. z T
Point
6.1b) one considers, for example, the resultant
torque, , about the axis z, due to the force, f,
acting at position, r, and the associated angular (c) t(n)
acceleration. For a deformable body the traction n
vector, t(n), is a measure of force per unit area
acting on the surface of a body (Fig. 6.1c), where
the surface has an orientation specied by the
outward unit normal vector, n. This surface can t(n) n
Area
be the exterior boundary of a rock mass or an
imagined surface within the rock mass. The trac-
tion vector is dened at a point on such a surface
Syy
in a limiting process as the area of a small
element of this surface shrinks toward zero about Syz Syx
(d) Sxy
the point. Similarly, if one imagines a small
Szy
cubical element of a given orientation at a point Szx S xx
within a body (Fig. 6.1d), one can dene the trac- Sxz
tions acting on all six sides as the volume shrinks y Szz
toward zero. The components of the traction on
each side dene the components of the stress Volume
acting on the cubical element with reference to
the chosen coordinate system, and this collection
of forces per unit area is referred to as a tensor x
quantity. Normal and shear components of the
stress tensor are directed perpendicular and par- z
allel to the sides of the cubical element, respec-
tively. In this hierarchy of concepts the force
vector acts on a point mass; the torque vector acts Fig 6.1 Force, torque, traction, and stress are illustrated
about an axis; the force per unit area, or traction as: (a) force vector, F, acting on a particle of mass, m, with
vector, acts on a surface element; and the set of corresponding acceleration vector, a; (b) torque vector, ,
forces per unit area, or stress tensor, acts on a due to force, f, acting at position vector, r; (c) traction
volume element. vector, t(n), acting on surface element with outward normal
The traction and stress are of interest to the n; (d) stress tensor, ij, acting on volume element with edges
structural geologist because structures develop as parallel to coordinate axes.
the rocks of the Earths crust strain and ow, and
the distribution of this deformation is related
to the stresses acting within the rock mass and tities with spatial and temporal variations. For
the tractions acting on its surfaces. In this and example, the frontispiece for this chapter is a pho-
later chapters we show how the concepts of trac- tographic visualization of the distribution of
tion and stress can be applied to understand the shear stress in the grains of a model sandstone
origin and evolution of geological structures. In (Gallager et al., 1974). An understanding of the pos-
most natural examples the traction and stress sible variations of these elds in the Earth is of
vary with position and time: they are eld quan- fundamental importance to structural geologists.
196 FORCE, TRACTION, AND STRESS
b dV ex bx dV ey by dV ez bz dV forces, fi.
V V V V
(6.1)
volume with sides aligned with the coordinate
Here is the mass density and (bx, by, bz) are the system as in Fig. 6.2b), the volume integral may be
Cartesian components of b, and both the density evaluated as a triple integral with appropriate
and body force may vary with the spatial coordi- ranges for the limits in the three coordinates:
nates. The resultant body force is the vector sum of z2 y2 x2
the body forces acting on all innitesimal ele-
ments within the nite volume V. The density and b dx dy dz
z1 y1 x1
(6.2)
the body force per unit mass may vary in time, in
which case (6.1) is considered to represent a given As indicated in (6.1) each component may be
instant in time. In principle, if the spatial varia- written as a separate triple integral and these are
tions of b are known as functions of the three added to compute the resultant body force.
coordinates, and if the shape of the nite volume In general we want to characterize the body
is relatively simple (for example a rectangular force at any arbitrary point P in the deformable
6.1 CONCEPTS OF FORCE AND TRACTION 197
solid or owing uid and this requires an ideal- Taylor, 2003). For a Cartesian coordinate system
ization of the rock mass as a material continuum. with z-axis vertical and directed upward (Fig. 6.2b)
For this purpose we consider a sequence of nite the components of the body force per unit mass
volumes, Vi, each with resultant body force, fi, are:
and average density, i, and each containing the
point in question (Fig. 6.2b). The nite volumes are bx 0, by 0, bz g* (6.4)
ordered from largest, i 1, to smallest i n, such
that the volume approaches zero as n . In this For the rectangular body of volume V shown in
limit the largest dimension of the sample Fig. 6.2b with side lengths x x2 x1, y y2 y1,
approaches zero, so the volume converges to the and z z2 z1 the resultant body force magni-
point P, not to a surface or a curve containing the tude is:
point (Malvern, 1969). Furthermore, the resultant z2 y2 x2
ically illustrated in an enlarged view (Fig. 6.3, The French mathematician Augustine-Louis
inset) by a collection of arrows with their tails or Cauchy (17891857) apparently rst contemplated
heads on a small patch of the surface of area A this ratio and proposed that it approaches a
that contains the point P. These forces account for denite value in this limit (Fung, 1969). The trac-
the mechanical action of the rock mass on the tion vector sometimes is referred to as the stress
positive side of the patch and they may be repre- vector, but the stress is a different construct so
sented by a resultant force, f, acting at the cen- this choice of words is confusing and should be
troid, P, of the patch, and a resultant torque, , avoided.
acting about an axis through the centroid. The We have written the traction vector as t(n) in
resultant force and torque are not restricted as to (6.7) to emphasize that this quantity is a function
direction. In some contexts the resultant torque is of the orientation of the surface upon which it
referred to as the resultant moment. acts. This orientation is specied by the vector n of
In a thought experiment the rock mass is ide- unit magnitude, directed outward and normal to
alized as a material continuum and we consider a the surface (Fig. 6.4a). An innite number of sur-
sequence of patches of the designated surface, S, faces (S1, S2, S3, . . .) with different normal curva-
with nite areas, Ai, each associated with a dif- tures may be constructed through the point P, all
ferent resultant force, fi, and each containing the having the same normal, n, at that point. The trac-
point P. The patches are ordered from largest, tion vectors acting on all of these surfaces at P are
i 1, to smallest i n, such that their surface areas identical in magnitude and direction. On the
approach zero as n . In this limit the largest other hand two surfaces may be constructed
dimension of the patch approaches zero, so the through the same point P with different normals,
patch converges to the point P and not to a curve n(1) and n(2) (Fig. 6.4b). The traction vectors, t[n(1)]
6.1 CONCEPTS OF FORCE AND TRACTION 199
(a) t(n) tion vector for that force. The magnitude of this
position vector is the distance from the centroid
to the point of application of the force on the
n surface, and in the limit this distance goes to zero,
so Cauchys proposal has an intuitive appeal. Note
that a resultant torque may exist on a surface of
P nite area. For example, as a layer of sedimentary
rock is bent during folding most cross-sectional
Surface surfaces of the layer are loaded by forces that con-
S1 tribute to a net torque. This torque is called a
bending moment and these moments play a
prominent role in theories of bending and folding
S2
S3 (Timoshenko, 1958; Timoshenko and Woinowsky-
Krieger, 1959). The theory of coupled stresses,
which will not be considered further in this text-
book, admits the possibility of a continuous dis-
(b) t[n(2)] tribution of torques per unit area, with a limit at
n(1)
a point on a surface that is different from zero
(Malvern, 1969).
t[n(1)] n(2) The traction at a point on a surface is a
measure of the force per unit area imparted by the
material from one side of the surface to the ma-
Surface terial on the other side. This concept can be stated
succinctly as (Fung, 1969, p. 51):
P
t[n(2)] t[n(1)] (6.9)
S1
Here it is understood that n(1) and n(2) are unit
normals at the same point on a surface viewed
S2 from opposite sides, so n(1) n(2). To interpret
(6.9) consider Fig. 6.5a in which a rock mass is
Fig 6.4 Relations among surfaces Si through point P and divided into two parts, 1 and 2, by the surface, S,
traction vector at point P. (a) Surfaces with different
illustrated here in cross section. The outward unit
curvatures but same outward unit normal, n, have the same
normal vector for part 1 is n(1), and that for part 2
traction. (b) Surfaces with different normals have different
tractions regardless of their curvature.
is n(2): these are oppositely directed vectors. At the
point P the rock of part 2 exerts a traction, t[n(1)],
on part 1. As drawn, part 2 is pulling on part 1 at
an angle that is somewhat oblique to n(1). Now,
and t[n(2)], acting on these surfaces are different imagine removing part 2 and replacing the
in magnitude and direction. mechanical action of part 2 on part 1 with the
In a thought experiment similar to that appropriate distribution of tractions. If this oper-
leading to (6.7) the ratio of resultant torque to ation were done accurately, according to Cauchy,
surface area is considered (Fig. 6.3). Cauchy appar- nothing about the mechanical state of part 1
ently contemplated this ratio and proposed that would change. Now consider the same surface
the limiting value approaches zero: with part 1 removed (Fig. 6.5b). What traction
would have to be applied at point P to replace the
lim
n
n
An
0 (6.8) mechanical action of part 1 on part 2? According
to Cauchys concept, we would have to apply a
Torque is the vector product of the force acting at traction, here called t[n(2)], of magnitude equal to
a given distance from the centroid and the posi- t[n(1)], but oppositely directed.
200 FORCE, TRACTION, AND STRESS
1.0
springs (Fig. 6.7a) applies the same displacement,
0.8
u, to each constituent, the ratio of total force to
0.6 effective spring constant, F/K, must be the same as
0.4 the ratio of any individual force to the individual
0.2 spring constant, Fi /Ki. For the purpose of under-
0 standing the denition of the traction vector, we
0 10 20 30 40 50 60 70 80 90 100 substitute the appropriate traction, multiplied by
Number of grains the surface area on which it acts, for the forces
acting on the constituents to nd:
Fig 6.7 Model for grain-scale heterogeneity of rock.
(a) Bed of springs with different stiffnesses accounts for
tDW ti DWi
u (6.12)
grains and pores. (b) Partial traction normalized by total K Ki
traction approaches a unit value as number of grains
Here t is the total traction on the platen of area DW,
increases. Reprinted from Amadei and Stephansson (1997)
and the ti are the individual tractions that replace
with kind permission of Springer Science and Business Media.
the mechanical action of the platen on each con-
stituent of area DWi. Because they have a negligible
behavior of the individual constituents of the spring constant, those constituents representing
sandstone at the grain scale (Amadei and pores carry a negligible traction regardless of
Stephansson, 1997, p. 50). A row of N grains and their width. However, for a given displacement,
pores is shown in two dimensions where the this equation demonstrates that the traction
individual widths, Wi, combine to a total width, acting on particular constituents increases in pro-
W. The height, H, before loading, and the depth portion to decreases in width. Also, the traction
out of this plane, D, are constants. The stiffness of increases in proportion to increases in the spring
each constituent is approximated with spring constant. In other words, thinner constituents of
constant, Ki, and the force acting on each spring is the same stiffness carry greater traction, and
Fi. The spring constant for pores would be very stiffer constituents of the same width carry
small compared with that for grains. The greater traction.
loading system is a rigid platen that imposes the Consider an area that is only a part of the total
same displacement, u, on all the springs. area, and the partial traction, tp, transmitted from
6.1 CONCEPTS OF FORCE AND TRACTION 203
the platen to the constituents under this partial area would vary slowly and approximately contin-
area. The partial traction is dened as: uously throughout the rock. The area dened by
this procedure is referred to as the representative
n n
i i
tp im i1 1mN tions by Cauchy in 1823 and 1827 (Malvern, 1969).
, (6.15)
t n N mnN The orientation of the patch is determined by
im
Wi
i1
Ki
the outward-directed unit normal vector n,
which makes direction angles (x, y, z) with the
As the partial area approaches the total area in
coordinate axes. Each angle is taken as the smaller
size, that is as m goes to 1 and n goes to N, the ratio
of the two in the plane containing n and the
tp /t goes to 1. In other words, the partial traction
respective coordinate axis. Because n is a unit
approaches the value of the total traction, as
vector, the components (nx, ny, nz) are the direction
expected. For partial areas less than the total area,
cosines:
the ratio may differ from one, and this difference
is a measure of the variability in traction intro- nx cos x,ny cos y,nz cos z (6.16)
duced by the heterogeneity in stiffness and width
Recalling the denition of the magnitude of a
of the individual constituents.
vector (2.7), these components are related as:
An example of many calculations of the ratio
tp /t for different sample areas from a model rock (nx)2 (ny)2 (nz)2 1 (6.17)
is shown in Fig. 6.7b (Amadei and Stephansson,
The areas of the three orthogonal sides of this
1997, p. 50). For small numbers of constituents in
element are found by projection of A onto the
the partial sample the scatter of the ratio about 1
coordinate planes such that:
is signicant, but the inclusion of 50 constituents
reduces the scatter to less than 10%. Since typical Ax Anx, Ay Any, Az Anz (6.18)
grains in medium sandstone are less than
0.50 mm in diameter, an area on the order of For this derivation n points into the positive
10 mm2 would be large enough to average out octant, but the resulting equations actually apply
most of the variability in partial traction. Using to all orientations on n. These relationships are
this area the ratio of resultant force to surface used below to derive equations that describe the
204 FORCE, TRACTION, AND STRESS
0.2 x2 y2 z2
1 (6.23)
0.1 t1 t2 t3
t1
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
For tractions scaled as t t/1MPa m2 this is an
ellipsoid drawn in physical space with coordinate
tx(n) axes x, y, and z. The semi-major, semi-intermedi-
ate, and semi-minor axes of the ellipsoid (6.23)
have lengths that are the positive square roots of
(b)
1 t1, t2, and t3, respectively. The trace of the traction-
Physical space n director surface for the two-dimensional example
0.9 using (6.22) is illustrated in Fig. 6.10b. The tangent
0.8 plane to the traction-director surface at an arbi-
trary point (x0, y0, z0) is:
0.7
0.6 (x0, y0, 0)
xx0 yy0 zz0
1 (6.24)
Tangent plane
t1 t2 t3
0.5
y
)
t(n
0.4
lt
h
lle
0.3
pa
surface
0.1 xnx yny znz
1 (6.25)
0 h h h
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
x This plane is parallel to that on which the traction
vector t(n) acts. The tangent plane (6.24) and the
Fig 6.10 (a) Plot of the trace of a traction ellipsoid in first plane dened by (6.25) are identical if:
quadrant of the (tx, ty)-plane in traction space (MPa) with a
particular traction vector, t(n). (b) Plot of the trace of the hx0 hy hz
t1 , t 0, t 0 (6.26)
traction-director surface in the (x, y)-plane of physical space. nx 2 ny 3 nz
The tangent plane at the point of intersection of a line
parallel to t(n) is parallel to the plane on which this traction Substituting these expressions for the magni-
acts. tudes of t1, t2, and t3 into (6.20) we have:
origin through the point (x0, y0, z0) is parallel to enigmatic. This makes it challenging to develop an
this traction vector (Fig. 6.10b). intuitive understanding for stress by simply
The unit normal vector, n, determines the ori- observing or mapping geologic structures.
entation of the tangent plane to the traction- However, certain structures can provide com-
director surface at the point (x0, y0, z0) and this pelling data, because they have a simple geometri-
plane is parallel to the surface on which the trac- cal relationship to some aspect of the stress eld.
tion vector t(n) acts. Given the magnitudes of the Some of the best examples are vertical igneous
special tractions t1, t2, and t3 and the components dikes, formed as magma was injected into frac-
of t(n), the components of n from (6.20) are: tures that tend to be oriented perpendicular to the
tx(n) ty(n) tz(n) direction of least horizontal compressive stress. In
nx , ny , nz (6.28) such a case, a set of curves drawn parallel to the
t1 t2 t3
pattern of dikes provides a map of the orientation
The following generalizations can be made about of the stress trajectories. Here we use a map
the traction vector at a point by studying the trac- pattern of dikes to introduce the concept of stress
tion ellipsoid and the traction-director surface: and to relate this to the traction vector.
Many vertical dikes crop out in the Raton Basin
1. The traction vector t(n) varies continuously in
of southeastern Colorado. The photograph shown
magnitude and direction as the surface on
in Fig. 6.11 includes an outcrop of a large dike
which it acts changes orientation.
trending northward from near the base of West
2. The special tractions t 1, t 2, and t 3 act on three
Spanish Peak. Many of the dikes of this region are
mutually orthogonal planes and have zero tan-
more resistant to erosion than the sedimentary
gential components, so they are perpendicular
host rock and crop out as prominent vertical walls
to the respective plane on which they act.
capping long ridges that form a radial pattern
3. The magnitudes of the tractions t1, t2, and t3 are
about the peak (Johnson, 1961, 1968). The radial
represented by the lengths of the semi-axes of
dikes are classied based on their composition
the traction ellipsoid: they are equivalent to
and Fig. 6.12a shows only those dikes of syenite
the extreme values of the normal component
and syenodiorite composition. The inference is
of the traction at a point.
that rocks of similar composition represent one
4. The traction t(n) is not parallel to the unit
period of magmatic activity and, perhaps, one
normal vector n except on surfaces where this
regional stress eld.
traction is equal to either t 1, t 2, or t 3.
The sedimentary rocks of the Raton Basin (Fig.
Other facts about the traction vector are described 6.12a) form the broad La Veta syncline with a
after we introduce the stress tensor in the next steeply dipping western limb that abuts the older
section. The traction vector provides the link rocks of the Sangre de Cristo Mountains, and a
between the concept of a distributed surface force gently dipping eastern limb that merges laterally
and the stress. It also provides the means to with the sub-horizontal formations of the Great
describe boundary conditions in terms of distrib- Plains. The sedimentary rocks folded into the Le
utions of forces acting on the internal or external Veta syncline are cut by the igneous rock and
surfaces of the material continuum. some deformed and metamorphosed sedimen-
tary rock making up West and East Spanish Peaks.
These impressive mountains rise almost 2 km
6.2 Concept and analysis of stress above the topography of the surrounding plain
and the pattern of igneous dikes seems to radiate
The shape of a deformed fossil (Fig. 5.1), the offset from West Peak. It was this systematic map
of a marker horizon across a fault (Fig. 2.14), and a pattern that led Helmer Od (1957) to propose a
multitude of other geological structures (Ramsay correspondence between the dike pattern and the
and Huber, 1983) provide direct evidence relevant stress distribution at the time of dike formation.
to the kinematics of deformation; however, eld He suggested that the dike pattern should corre-
evidence relevant to the state of stress is more spond to the pattern of stress trajectories.
208 FORCE, TRACTION, AND STRESS
(a) y
ne
syncli
Radial dikes 37 30' (a)
85
ta
b
Sangre de Cristo Mountains
7
La Ve
(b) sxx
Magma chamber Host rock
y y
t(a) = p
x
t(d ) C
A x
syx
(c)
t(b) t(c)
sxy
Potential
B dike path
traction vectors and note that stress component pression is positive and tension is negative. Shear
arrows always are drawn in pairs. The normal stress components directed as shown in Fig. 6.13c
stress component yy is dened using a similar are given negative signs. This choice is motivated
argument. by the fact that the normal stress typically is com-
The combination of the two tangential traction pressive at depth in the Earth, because it is related
components, ty(a) and ty(c), in Fig. 6.13a are used to the weight of the overlying rock. A disincentive
along with the condition (6.30) to dene xy as a for choosing this convention is the awkward
shear component of stress. This quantity is repre- result that components of the displacement
sented schematically as two arrows drawn parallel vector, when related to the stress components, are
to sides a and c, with lengths and directions con- found to be positive in the negative coordinate
sistent with the respective traction components directions. This is at variance with standard prac-
(Fig. 6.13c). The two subscripts are chosen because tice for dening vector components. Students of
this stress component is acting on two opposing structural geology should be able to work with
planes with normals parallel to the x-axis and it is both sign conventions, because both are used
acting in an orientation parallel to the y-axis. The throughout the relevant literature.
arrows have open heads, but only half of the head A simple procedure for remembering how to
is drawn to distinguish the shear stress from the draw all eight arrows representing the stress com-
normal stress. The shear stress component yx is ponents in two dimensions is to consider the
dened using a similar argument. outward normal vector to the plane under con-
There is an apparent contradiction in alge- sideration. If that normal points in the positive
braic signs between traction components and coordinate direction (as on sides a and b, Fig. 6.13),
stress components. For example, one of the arrows a positive stress component arrow should point in
for xx points in the positive x-direction and the the positive coordinate direction. If that normal
other arrow points in the negative x-direction (Fig. points in the negative coordinate direction (sides
6.13b). This stress component is made up of two c and d), then a positive arrow should point in the
traction components that have opposite signs (Fig. negative coordinate direction. This produces a set
6.13a). On the other hand, we need to give a single of arrows consistent with the sign convention for
algebraic sign to the stress component. The tradi- positive stress components (Means, 1976).
tional choice, motivated by considering the sign By extension of the reasoning behind the
conventions for the traction components point- denition of the Cartesian components of stress
ing into the rst quadrant on sides a and b (Fig. in two dimensions (Fig. 6.13), there are nine
6.13a), is to assign positive signs to stress compo- components of stress in three dimensions (Fig.
nents that correspond to these positive traction 6.14), one normal component and two shear
components. This makes a tensile normal stress, rep- components acting on each of the three pairs of
resented by outward directed arrows, a positive opposing sides of the cubical element. Arrows
quantity, whereas a compressive normal stress, rep- representing these nine components are shown
resented by inward directed arrows, is a negative acting on the visible sides of the element. These
quantity. Similarly, the two shear stresses are pos- components are drawn in their positive orienta-
itive as drawn in Fig. 6.13c. This is the convention tions and it is understood that arrows drawn in
used in most of the physics and engineering liter- the opposite directions on the opposing sides of
ature. Because many of the concepts and analysis the element make up the hidden member of
methods we use in structural geology are taken each pair.
from that literature, this is an attractive choice The nine components of stress are not inde-
and we will use it throughout this book, unless pendent. This is understood here by treating the
otherwise stated. cubical element in Fig. 6.14 as nite in size with
Most practitioners of soil and rock mechanics side lengths x, y, and z; considering the state
(Jaeger and Cook, 1979), as well as many structural of stress to be homogeneous throughout; and
geologists and geophysicists, use the opposite con- ignoring the effects of body forces. We imagine
vention for signs of the stress components: com- cutting the element free of the surroundings and
6.2 CONCEPT AND ANALYSIS OF STRESS 211
xxyz, xyyz, xzyz xx xy xz
yx yy yz (6.35)
on x 12 x (6.31)
zx zy zz
On the negative x-side the equivalent stress com- The normal stress components are placed along
ponents give rise to negative forces of the same the main diagonal in this matrix and the equiva-
magnitudes in the x-, y-, and z-directions: lent shear stress components are placed in sym-
xxyz, xyyz, xzyz metric locations about this diagonal. The rows of
this matrix contain, respectively, the stress com-
on x 12 x (6.32) ponents on the x-, y-, and z-sides of the cubical
element pictured in Fig. 6.14.
Similar arguments for the y- and z-sides demon- Other notations for the stress components are
strate that the forces in the three coordinate direc- found in the literature. For example, the symbol
tions exactly balance because, for a homogeneous may replace for all shear stresses (e.g. xy, yz, zx)
stress state, the corresponding force components to distinguish normal and shear stress compo-
on opposing sides of the element are equal and nents (Timoshenko and Goodier, 1970). Or, the
opposite. The condition of force equilibrium normal stress components may have only one sub-
places no constraints on the stress components. script (e.g. x, y, z) because the subscripts are
However, consideration of moment equi- identical (Jaeger and Cook, 1979). To accommo-
librium does constrain some of the stress date the use of indicial notation, the Cartesian
212 FORCE, TRACTION, AND STRESS
11 12 13
21 22 23 (6.37)
31 32 33
also have ignored moments due to distributed
Recalling the onin convention for subscripts body or surface couples because most problems in
the rst row contains the stress components structural geology have found satisfactory corre-
on the sides of the cubical element perpendicular spondence to nature without invoking coupled
to the x1-axis and these components are directed, stresses (Malvern, 1969). The constraint on the
respectively, in the x1-, x2-, and x3-directions. shear stress components (6.34) that leads to a sym-
Unlike the displacement, velocity, or accelera- metric matrix of stress components (6.35) is lost in
tion the physical quantity we call the stress is not the presence of coupled stresses.
dened at a point in the continuum by a single Some problems in structural geology can be
vector with three components. In fact it takes six idealized using cylindrical symmetry about an
traction vectors, three of which are independent axis in three dimensions. The map view of the
because of (6.9), acting on the orthogonal sides of dike pattern at Spanish Peaks is a possible
the cubical element to determine the nine stress example (Fig. 6.12a) because the dikes appear to
components. The state of stress at a point in the con- radiate from a point near the center of West
tinuum is completely dened by the nine compo- Spanish Peak. A practical example would be the
nents of (6.35), six of which are independent cylindrical hole cut by a drilling rig to produce
because of (6.34). Stress is referred to as a second- water or hydrocarbons from porous formations at
order tensor to distinguish it from vectors, which are depth. For these and other problems it is useful to
rst-order tensors, and scalars, which are zero- dene the stress components in terms of a cylin-
order tensors. To complete the denition of stress drical coordinate system (Fig. 6.15) which is com-
at a point, one would include the appropriate unit posed of a cylindrical axis, Oz, a perpendicular
of measure, along with the three coordinates of the axis, Ox, a radial distance, r, and a counterclock-
point. If the stress state is a function of time then wise angle, , from the Ox-axis to the radial line.
the denition would include the appropriate time. The six independent stress components (rr,
We reduced the nine stress components to six r, rz, , z, zz) are dened on a small element
by postulating static equilibrium for the nite that has sides parallel to radial lines and concen-
cubical element, no body forces, and a homo- tric circles with centers at the origin. These are
geneous state of stress. In Chapter 7, after intro- called the cylindrical components of stress. Shear
ducing the conservation laws, we show that (6.34) components on adjacent faces must be of equal
is not so restricted, but applies to problems of uid magnitude to prevent angular accelerations:
dynamics in which elements experience both
r r, rz zr, z z (6.38)
linear and angular accelerations, to deformation
in the presence of body forces, and to hetero- Note that the onin convention for subscripts is
geneous stress states in solids and uids. Here we followed: r acts on the sides with normals
6.2 CONCEPT AND ANALYSIS OF STRESS 213
Fig 6.16 Relations among traction vector and stress fx [tx(n)]A xxAx yxAy zxAz 0
tensor components. (a) Cauchy tetrahedron with traction (6.39)
t(n) acting on surface with outward unit normal n and stress
Note that the component of force due to the trac-
components acting on coordinate planes. (b) Two-
dimensional relations among traction vector and stress tion component tx(n) is balanced by that due to the
tensor components. stress components acting in the x-direction on the
x-, y-, and z-sides of the element. Rearranging (6.39)
using (6.18), and following similar arguments for
parallel to the radial, or r-direction, and in the the balance of forces in the y- and z-directions, we
circumferential, or -direction. nd three equations that collectively are known
as Cauchys Formula:
6.2.2 Cauchys formula tx(n) xxnx yxny zxnz
We nd important relationships among the trac-
ty(n) xynx yyny zynz (6.40)
tion and stress components by studying the
tz(n) xznx yzny zznz
Cauchy tetrahedron (Fig. 6.16a). Recall that the
inclined side of this element is given an area A Cauchys Formula instructs us that the traction
and the areas of the orthogonal sides (Ax, Ay, Az) vector, t(n), on a surface of any orientation
are found by projection of A onto the coordinate (dened by the outward unit normal vector n)
planes (6.18). Unlike the case illustrated in Fig. through a given point, is completely determined
6.8b, here the traction vectors acting on the by the nine (six independent) components of the
orthogonal sides are not constrained to be paral- stress tensor at that point. We derived Cauchys
lel to the respective normal vectors. In other Formula ignoring body forces and postulating
214 FORCE, TRACTION, AND STRESS
t1(n) 11 21 31 n1
t2(n) 12 22 32 n2 (6.42) x
t3(n) 13 23 33 n3
Fig 6.17 Volume element with one surface coincident with
Note that the matrix of stress components is the surface of a body. Components of the traction vector, t(n),
transpose of (6.37). The matrix representation acting on this surface are related to certain stress
(6.42) emphasizes the fact that the stress tensor components acting on the element.
can be thought of as a linear operator that gives
the traction vector as a function of the unit
normal vector. The components of stress are con- tx(n) yx xy, ty(n) yy,
sistent with the denition of a second-order tensor
tz(n) yz zy (6.43)
quantity in that they associate a vector (the trac-
tion) with any direction in space as determined by The other components of the stress tensor, namely
the respective direction cosines (components of xx, xz zx, and zz, are not determined by the
the unit normal vector). traction acting on this boundary. These compo-
Cauchys Formula relates the tractions, acting nents may be calculated by solving a boundary
as boundary conditions for models of geologic struc- value problem, but are not given by the boundary
tures, to the stress components on a cubical condition itself.
element adjacent to that boundary. For example, Cauchys Formula given by (6.40) reduces to
consider the body shown in Fig. 6.17 and a small two dimensions for conditions of plane deforma-
element that has one side coincident with the tion. For example, taking the (x, y)-plane as the
boundary of the body. For convenience we choose plane of interest (Fig. 6.16b) we have:
a coordinate system with the y-axis normal to that
nz 0, xz 0 zx, yz 0 zy (6.44)
side, and the other two axes parallel to the edges
of that side. The outward unit normal vector to The two out-of-plane shear stresses must be
the boundary of the body at that point, has com- zero by denition, and the out-of-plane normal
ponents nx 0, ny 1, and nz 0, so the traction stress which generally is not zero is eliminated
and stress components are related according to from (6.40) because the direction cosine, nz,
Cauchys Formula: is zero. The other two direction cosines are
6.2 CONCEPT AND ANALYSIS OF STRESS 215
Cauchys Formula (6.40) is used to write (6.48) in (6.52) by dividing each component of n (t n) by
terms of the stress components: |ts|.
The two-dimensional resolution in the (x, y)-
tn xxn2x yyn2y zzn2z 2xynxny 2yznynz
plane of the x- and y-components of the traction t
2zxnznx (6.49) onto the n- and s-axes can be derived by inspection
Given the state of stress and the orientation of a of Fig. 6.16b:
surface at a point, (6.49) is used to calculate the
tn txcos x tysin x
normal component of the traction vector. (6.54)
ts txsin x tycos x
We use the second term on the right-hand side
of (6.46) to calculate the tangential (shear) compo- Here the (n, s)-coordinate axes are arranged with n
nent of t. The magnitude of the vector t n is the perpendicular and outward from the surface on
magnitude of this component: which t acts and s tangential to this surface. Both
n and s are in the (x, y)-plane and s is directed such
|t n | |t ||n | sin |t | sin |ts | (6.50)
that n is to the right when looking in positive s (a
Given the range 0 , the quantity |t| sin is right-hand rule). Given these two reference direc-
always positive. Having specied only one refer- tions, tn and ts may be either positive or negative.
ence direction, n, we cannot distinguish positive The two-dimensional forms of Cauchys Formula,
and negative signs for the tangential component (6.45), are used to write (6.54) in terms of the stress
of t. However, the vector t n is directed perpen- components:
dicular to the plane dened by t and n, and lies in
tn xx cos 2x yy sin 2x 2xy sin x cos x
the surface on which t acts because it is perpen-
dicular to n. Recall that a vector product may be ts (xx yy) sin x cos x (6.55)
written in terms of the components as: xy( cos 2x sin 2x)
t n (tynz tzny)ex (tznx txnz)ey Given a two-dimensional state of stress in the (y, z)-
(txny tynx)ez (6.51) or the (z, x)-plane, (6.55) may be used with appro-
priate exchange of subscripts.
Similarly, the second vector product of (6.46) may
be written using (6.51) as:
n (t n) [(1 n2x )tx nxnyty nxnztz]ex
6.2.4 Principal values and principal axes
of normal stress
[nxnytx (1 n2y )ty nynztz]ey In Section 6.1.4 we derived the equation for the
[nznxtx nznyty (1 n2z )tz ]ez traction ellipsoid (6.21) by postulating without
(6.52) derivation that three orthogonal surfaces at a
point could be so oriented that the traction vector
This vector is the resolution of the traction t onto acting on each is directed parallel to the respec-
the surface with outward unit normal vector n. tive normal vector. In other words the tangential
Cauchys Formula (6.40) may be used to write component of the traction on each orthogonal
(6.52) in terms of the stress components. The mag- surface is identically zero. The tractions acting on
nitude of the tangential (shear) component of t these three surfaces (t1, t2, and t3) correspond to
may be calculated as |n (t n)| using (6.52) or, the semi-axes of the traction ellipsoid (Fig. 6.9)
noting in Fig. 6.18 that |tn| and |ts| are the and are ordered such that t1 t2 t3. Here we
lengths of the sides of a right triangle and |t| is derive these relationships in terms of the state of
the length of the hypotenuse, one may use the fol- stress at a point by equating the normal compo-
lowing (positive) square root: nent of the traction vector to the normal stress.
|ts | |t |2 |tn |2 We show that the normal stress takes on extreme
(6.53)
values in three orthogonal directions and dene
The direction cosines of the vector component of each of these as a principal normal stress. The prin-
t tangential to the surface are calculated from cipal normal stresses play important roles in
6.2 CONCEPT AND ANALYSIS OF STRESS 217
theories of failure, fracture, and faulting of rock, This cubic equation for the unknown normal stress,
topics that we consider in later chapters. nn, has three real roots (Bell, 1920), and these are
We begin by considering an arbitrary state of the three principal stresses 1, 2, 3 which are
stress at a point dened by six independent com- ordered such that 1 2 3. Standard algebraic
ponents and ask if a plane through this point techniques exist to solve such a cubic equation
exists upon which the shear component of the (Selby, 1975) and to show for all possible values of
traction vector vanishes. Referring to Fig. 6.18, the the stress components that the three roots are real.
traction vector, t, would be parallel to the outward Given the values of the principal stresses each
unit normal vector, n, and the only stress compo- may be substituted, successively, for the normal
nent associated with this plane would be the stress, nn, in (6.57) and the three components of
normal stress, nn. The components of the traction the unit normal vector for each principal stress
vector acting on this plane in the arbitrarily axis may be determined (Fig. 6.19a). The three
chosen Cartesian coordinate system would be: components of the unit normal vectors that are
parallel to the principal axes are written:
tx(n) nnnx, ty(n) nnny,
(nx1, ny1, nz1) components of n(1)
tz(n) nnnz (6.56)
(nx2, ny2, nz2) components of n(2) (6.60)
These same traction components are related to (nx3, ny3, nz3) components of n(3)
the stress components through Cauchys Formula.
Substituting the right-hand sides of (6.56) for the Here, for example, the component nz1 is the direc-
traction components in (6.40), and rearranging tion cosine used to project the base vector for the
yields three linear equations for the unknown z-axis onto the axis parallel to the direction of 1.
normal stress, nn, and the three unknown com- To show that the principal axes are orthogonal
ponents of the unit normal vector (nx, ny, nz). We use nn 1 in (6.57) and multiply each equation
include (6.17) to make a set of four equations in by the respective components of n(2), (Jaeger and
four unknowns: Cook, 1979, p. 20):
| |
xx nn yx zx
Adding the three equations (6.61); then adding
xy yy nn zy 0 (6.58)
the three equations (6.62); and nally subtracting
xz yz xx nn
the second sum from the rst and invoking the
Expanding the determinant and invoking the symmetry condition (6.34) we nd:
symmetry of the stress tensor (6.34) we have: (1 2)(nx1nx2 ny1ny2 nz1nz2) 0 (6.63)
(a) z (a)
n(3) n(2)
z
0.2
0.1
0
0.1 1
0.2
0.5 0.5
(z, 1) 0
x
y y 0 0.5
0.5 1
n(1) O
(x, 1)
(y, 1) (b)
x
z
z 0.5
(b)
0
0.5
2
s2 1.5
s3 1 1
0.5 0.5
0
0 0.5 x
y 0.5 1
1 1.5
2
s1 y
Fig 6.20 (a) Traction ellipsoid for state of stress with
principal axes parallel to coordinate axes. (b) Traction
ellipsoid for state of stress defined in (6.68) with principal
axes not parallel to coordinate axes.
x
1 0 0 unequal in this example the traction vectors trace
0 2 0 (6.64) out an ellipsoid with three unequal axes which
0 0 3 are coincident with the coordinate axes (Fig.
6.2 CONCEPT AND ANALYSIS OF STRESS 219
6.20a). The complete stress analysis, including three roots that are the so-called eigenvalues or, as
principal stresses, trajectories, and all compo- we refer to them in this context, the principal
nents, is shown in color on the textbook website. values 1, 2, 3. The eigenvalues of an asymmet-
Equations (6.57) may be solved using matrix ric matrix can be real or complex, but the special
algebra and this is a classic problem called an case of a symmetric matrix always yields real
eigenvalue problem, which has many applications eigenvalues (Gere and Weaver, 1965), so the sym-
in the sciences and engineering (Gere and Weaver, metric matrix of stress components always yields
1965). Here we refer to this as a principal value real principal stresses. Real eigenvalues may be
problem since we are looking for the principal positive, negative, or zero and likewise the princi-
values of stress, and (6.57) is written: pal stresses may be positive (tension), negative
(compression), or zero subject only to the con-
xx yx zx nx nx straint that 1 2 3.
xy yy zy ny ny (6.65) Each eigenvalue of a matrix is associated with
nn
xz yz zz nz nz a vector called the eigenvector. For the matrix of
A solution to (6.65) is the vector n with compo- stress components the eigenvectors are the unit
nents (nx, ny, nz), which determines the orientation normal vectors for the planes on which the prin-
of a principal plane with respect to the Cartesian cipal stresses act. Having determined the values of
coordinates (x, y, z). The product of the matrix of the three principal stresses from solving (6.59),
stress components and the unit normal compo- these are substituted separately into (6.67) to
nents is equal to the product of a scalar, nn, and obtain three simultaneous equations to be solved
the unit normal components, where this scalar is for the components of the normal vector n(1), n(2),
the unknown principal stress. Rearranging (6.65) or n(3), associated with each principal stress. In
we have the following homogeneous equation: general, the eigenvectors for a symmetric matrix
are orthogonal to one another (Gere and Weaver,
1965).
xx yx zx 1 0 0 nx 0
xy yy zy nn 0 1 0 ny 0 As an example, consider a point in a contin-
xz yz zz 0 0 1 nz 0 uum where the following symmetric matrix pro-
vides the stress components referred to a specied
(6.66)
Cartesian coordinate system (MPa):
This equation is called homogeneous because the
2 34 1
right-hand side is the null vector. The second xx xy xz 4
3 1
matrix on the left-hand side is called the identity yx yy yz 4 1 2 (6.68)
1 1 1
matrix because multiplication of a matrix of the zx zy zz 4 2 2
xx nn yx zx nx 0
1 2.5080, 2 0.8261, 3 0.1659 (6.69)
xy yy nn zy ny 0
xz yz zz nn nz 0 The corresponding components of the principal
vectors are:
(6.67)
One solution to (6.67) is the null vector n, referred nx1 0.8398 nx2 0.5256
to as a trivial solution, but this is of no interest for ny1 0.4930 , ny2 0.6333 ,
the physical problem we are considering. Non- nz1 nz2 0.5681
0.2273
trivial solutions exist only if the determinant of
the coefcient matrix is zero, as stated in (6.58), nx3 0.1361
and this leads to the cubic equation (6.59) in the ny3 0.5965 (6.70)
unknown normal stress, nn. This equation has nz3 0.7909
220 FORCE, TRACTION, AND STRESS
xx xy 0 principal
yx yy 0 (6.71) stress
1
0 0 zz trajectory
x
The two in-plane principal stresses are
(Timoshenko and Goodier, 1970):
1 1
2 (xx yy) [4 (xx yy) xy]
2 2 12
(b)
1 1
(6.72)
2 (xx yy) [4 (xx yy) xy]
2 2 12
1 2xy
1 tan 1 (6.73)
2 xx yy
principal stress, 3 (maximum compressive the orientation of a coordinate system. Rather, the
stress). The triangulated surfaces are model fault constitutive law and strength of a particular ma-
surfaces for a hydrocarbon reservoir in the North terial should be a property of the material itself
Sea and the principal stresses were used to predict and the ambient conditions of temperature and
the location and orientation of small faults in this pressure.
reservoir (Maerten et al., 2002).
The orientation of the Cartesian coordinate 6.2.5 Maximum shear stresses
system in the preceding discussion of principal Given the principal normal stresses (1, 2, 3) and
stresses was arbitrary. Any orientation could be their orientations, one can calculate the variation
chosen and the resulting principal stresses and in the shear traction magnitude, |ts|, with the ori-
principal axes would be identical. In other words entation, n, of the plane on which it acts. We
the three roots of the cubic equation (6.59) are the equate |ts| to the magnitude of the shear stress
same, regardless of coordinate system, so the |ns| and seek orientations of the planes on which
coefcients of this equation must not vary for a the shear stress attains extreme values because
given state of stress. Rewriting this cubic equation these quantities play important roles in rock
by collecting the stress components into constant deformation, particularly faulting. As shown in
coefcients we have: Fig. 6.18 and written in (6.53) the magnitudes of
3nn I1 2nn I2nn I3 0 (6.74) the traction vector, t, and the stresses ns and nn
are related such that:
The invariant coefcients are:
2ns |t |2 2nn (6.77)
I1 xx yy zz
Here we have equated the normal traction com-
I2 xxyy yyzz zzxx 2xy 2yz 2zx
ponent, tn, to the normal stress nn acting on the
I3 xxyyzz 2xyyzzx xxyz2 yyzx2 zzxy2 plane of interest. To write the quantities on the
(6.75) right-hand side of (6.77) in terms of the principal
stresses a Cartesian coordinate system is chosen
These combinations of the Cartesian stress com- with positive x-, y-, and z-axes in the directions of
ponents are referred to as the stress invariants. the unit vectors n(1), n(2), and n(3) respectively
The denitions of the stress invariants in (Fig. 6.22a). Cauchys Formula, (6.40), then reduces
terms of the Cartesian components are reduced to to tx 1nx, ty 2ny, tz 3nz and the squared
denitions in terms of the principal stresses by magnitude of the traction vector is the sum of the
rotating the coordinate system until it aligns with squared components:
the principal axes. This is equivalent to setting the
shear stress components to zero and equating the |t |2 12n2x 22n 2y 32n 2z (6.78)
normal components to the principal stresses: The squared normal stress is taken from (6.49)
I1 1 2 3 with the shear stress components equal to zero
and the normal stress components equal to the
I2 12 23 32 (6.76)
principal stresses:
I3 123
2nn (1n2x 2n2y 3n2z )2 (6.79)
Note in particular that the sum of the three
normal stress components is invariant. This has a Substituting (6.78) and (6.79) into (6.77) we have:
rather simple interpretation as three times the
ns2 12n 2x 22n2y 32 n2z (1n2x 2n2y 3n2z )2
mean normal stress. The other invariants do not
(6.80)
have such simple interpretations but all are
employed in the development of constitutive laws When this equation is expanded there are terms
for isotropic materials and in theories of failure in each of the principal stresses that can be
because, it is argued, such laws and theories rearranged using (6.17) and the following example
should not depend upon an arbitrary choice for (Fung, 1965):
222 FORCE, TRACTION, AND STRESS
ns
values of the shear stress are found by taking the y
derivatives of (6.83) with respect to nx and ny:
dns
2ns (3 1)22nx(1 2n2x ) [(1 2)2
dnx x
(2 3)2 (3 1)2]2nxn2y
Fig 6.22 (a) Volume element oriented such that principal
dns (6.84)
stresses act on sides. (b) Volume element rotated such that
2ns ( 2 3 )22n y(1 2ny ) [(1 2)
2 2
dny the maximum shear stress acts on sides.
(2 3)2 (3 1)2]n2x 2ny
The right-hand sides of (6.84) are zero if nx 0 ny, (6.83) such that |ns | 12 |(2 3)|. Substituting
but this refers to a principal plane on which the the components into (6.79) and equating the
shear stress is zero (minimum). normal traction component to the normal stress,
Taking nx 0 the right-hand side of the rst of nn, we nd the magnitude of the normal stress
(6.84) is zero and the second reduces to: acting on these planes is |nn | 12 |(2 3)|. The
steps of this paragraph are repeated taking nx 0
dns
2ns (2 3)22ny(1 2n2y ) 0 (6.85) to nd a second set of components for n and then
dny
the entire analysis is repeated after eliminating nx
For 2 3 0, the second term in parentheses or ny from (6.82) to nd a third set of components.
must be zero, and this requires n2y 12. Using (6.17) All three sets are given in the Table 6.1 along with
we nd n2z 12 and conclude that the components the magnitudes of the maximum shear stresses
of n for the planes carrying this shear stress and the magnitudes of the normal stresses on
are: nx 0, ny 12, and nz 12. These compo- these planes.
nents dene four planes that contain the x-axis The shear stress we have identied as 12 |1 3 |
(the direction of 1) and bisect the y- and z-axes in Table 6.1 always is the greatest in magnitude,
(the directions of 2 and 3), (Fig. 6.22b). The mag- because 1 2 3, but the order of the other
nitude of the shear stress acting on these planes two depends upon the particular values of the prin-
is found by substituting the components into cipal normal stresses. The maximum shear stresses
6.2 CONCEPT AND ANALYSIS OF STRESS 223
nx ny nz |ns| |nn| H
D
0 12 12 1
|2 3 | 1
|2 3 | G
2 2 F I
12 0 12 1
|1 3 | 1
|1 3 | 2az
2 2 2ax
2ay snn
12 12 0 1
|1 2 | 1
|1 2 |
2 2 s3 B s2 C A s1
y s2
stresses. The circle with a center at A traces out are entirely to the right of the origin, the normal
coordinate pairs (nn, |ns|) from (1, 0) to ( 2, 0) as stresses on all possible planes are positive so they
the angle 2x varies from 0 to for surfaces that are tensions. If the circles are to the left of the
contain the z-axis and the principal direction n(3) origin, the normal stresses are negative so they
such that z /2 (see Fig. 6.23b). Similar state- are compressions. If the circles straddle the
ments follow for the circles with centres at B and origin, then some planes carry tensile stresses and
C that trace out coordinate pairs from (2, 0) to others carry compressive stresses. The greatest
(3, 0) as 2y varies from 0 to and x /2, and shear stress on a plane that contains the y-axis and
from (3, 0) to (1, 0) as 2z varies from 0 to and n(2) is found on the circle with center at C where
y /2. The center of the circle at C corresponds 2z 90. In other words, in physical space this
to the normal stress (1 3)/2 and the radius of shear stress acts on a plane oriented at 45 to the
that circle is the magnitude of maximum shear principal directions n(1) and n(3).
stress (1 3)/2. The other stresses from Table 6.1
are related to the centers and radii of the circles 6.2.7 Variation of stress components
centered at A and B. with orientation of the coordinate
We now identify where, in Mohr space, coordi- system
nate pairs (nn, |ns|) would plot that act on planes Given the stress components referred to a
oblique to all three coordinate axes. The normal Cartesian coordinate system with a particular ori-
vector, n, to the arbitrarily oriented plane in phys- entation, it is useful to calculate the stress compo-
ical space (Fig. 6.23b) extends from the origin to nents referred to a Cartesian system with another
the perimeter of a unit sphere. The direction orientation. In other words the two coordinate
angles for this vector are x, y, and z. The cone systems are related by a rotation about a common
swept out by rotating about the Ox-axis with con- origin. This procedure is somewhat similar to the
stant angle x intersects the unit sphere along the transformation of coordinates by rotation that was
dashed circle DE. The coordinate pairs (nn, |ns|) described in Chapter 2, but here the equations are
acting on surfaces with these orientations are rep- different because the relative surface areas upon
resented in Mohr space along the dashed circle which the stress components act must be taken
DE with center at B. Similarly, the cone about into consideration (Jaeger and Cook, 1979,
the Oy-axis with constant angle y traces the pp. 245). It should come as no surprise that we
dashed circle FG on the unit sphere and coordi- employ Cauchys Formula in the derivation.
nate pairs (nn, |ns|) acting on these surfaces are The given stress components are referred to
represented in Mohr space along the dashed circle axes of the rst coordinate system (x, y, z) and we
FG with center at C. Finally, the cone for con- seek the corresponding stress components referred
stant angle z traces the dashed circle HI on the to axes of a second coordinate system (x, y, z ).
unit sphere and coordinate pairs (nn, |ns|) are Basis vectors from the common origin, O, and
represented in Mohr space along the dashed circle directed along the positive axes Ox, Oy, and Oz are
HI with center at A. The common intersection used in the transformation equations and these are
of the three dashed circles in Mohr space provides dened using the direction cosines of the angles
the normal stress and the magnitude of the shear between the respective coordinate axes (Fig. 6.24a):
stress on the oblique plane with normal vector n.
ex mxxex myxey mzxez
All possible coordinate pairs fall in the gray region
ey mxyex myyey mzyez (6.86)
between the three half-circles.
ez mxzex myzey mzzez
From the Mohr diagram (Fig. 6.23a) we observe
that the principal stresses act on orthogonal The double subscripts on the direction cosines mij
planes. For example, the point (1, 0) in Mohr refer to the reference axis and the transformed
space is associated with the double angle 2z , axis, respectively. For example, myx is the cosine of
so in physical space we have z /2. Because the the angle (y, x ).
principal stresses plot on the abscissa, the princi- The normal stress component, xx, acts on the
pal planes carry no shear stress. If the Mohr circles plane with normal ex and in a direction parallel to
6.2 CONCEPT AND ANALYSIS OF STRESS 225
(a) z xx t(ex) ex
xxm2xx yym2yx zzm2zx 2xymxxmyx
z 2yzmyxmzx 2zxmzxmxx (6.88)
y
ez The shear stress component acting on this plane
ey and in a direction parallel to the y-axis is found by
(z, x) resolving the traction vector t(ex) onto the Oy-axis.
y In other words it is the scalar product of the trac-
ex
x tion vector and the unit normal vector ey. The trac-
O tion vector components are given in (6.87) and the
(x, x) components of ey are given in (6.86) so the shear
(y, x)
x stress is:
xy t(ex) ey
(b) y
xxmxxmxy yymyxmyy zzmzxmzy
y
xy(mxymyx myymxx)
syy
x yz(myymzx mzymyx)
syx
zx(mzymxx mxymzx) (6.89)
ax
sxy Resolving the traction vector t(ex) onto the Oz-
sxx axis determines the other shear stress component
x
acting on this plane.
Following a similar procedure for planes with
normal vectors ey and ez the other four transfor-
mation equations are (Jaeger and Cook, 1979,
pp. 246):
this basis vector. This stress component is found by 2xymxzmyz 2yzmyzmzz 2zxmzzmxz
resolving the traction vector t(ex) acting on this (6.91)
plane onto the Ox axis. In other words the normal
stress is the scalar product of the traction vector yz xxmxymxz yymyymyz zzmzymzz
and the basis vector ex. The traction vector compo-
xy(mxymyz mxzmyy)
nents are found from Cauchys Formula, (6.40), by
noting that the normal to the plane of interest is ex: yz(myymzz myzmzy)
zx(mzymxz mzzmxy) (6.92)
tx(ex) xxmxx yxmyx zxmzx
ty(ex) xymxx yymyx zymzx (6.87)
zx xxmxzmxx yymyzmyx zzmzzmzx
tz(ex) xzmxx yzmyx zzmzx
xy(mxzmyx mxxmyz)
Using these components of the traction vector yz(myzmzx myxmzz)
and the components of ex from (6.86), the normal
zx(mzzmxx mzxmxz) (6.93)
stress is:
226 FORCE, TRACTION, AND STRESS
Although these equations look complex, a few a given point in the body determined by the polar
minutes study reveals patterns that reect the coordinates r and (Fig. 6.15). One can imagine
fact that the three normal components transform sliding the origin of coordinates to the point in
in a similar fashion, as do the three shear compo- question so the geometry is much like that in Fig.
nents. Furthermore the sequence of subscripts is 6.24b where the coordinates (x, y) are oriented in
systematic and repetitive. the same direction as the coordinates (r, ). If we
The transformation equations for stress com- take the angle x and the angle to be equal, the
ponents, (6.88) through (6.93), are more general transformation from Cartesian stress components
than one might suppose. Equations of this same to polar stress components is identical to (6.95)
form are, for example, appropriate for the trans- after substituting for x and making the appro-
formation of the innitesimal strain compo- priate substitutions for the stress components:
nents and other physical quantities (Fung, 1969,
rr xx cos 2x yy sin 2x 2xy sin x cos x
pp. 325). These quantities are collectively known
as tensors of rank two. They are dened by two char- xx sin 2x yy cos 2x 2xy sin x cos x
acteristics: r (xx yy) sin x cos x
xy( cos 2x sin 2x) (6.96)
1. they are composed of nine components in the
(x, y, z)-coordinate system; and The inverse transformation, from polar to
2. they transform to the (x, y, z)-coordinate Cartesian stress components, is found by rst
system following equations of the form pro- exchanging the subscripts in (6.96) and then
vided above. changing the sign of x. In this case only the sign
of terms containing sin x change.
The transformation of stress components in two
dimensions follows from the three-dimensional
expressions given above. For example consider the 6.2.8 An example: stress analysis at the
(x, y)-coordinate system transformed to the (x, y)- grain scale
coordinate system by a rotation about the out-of- The individual sand grains of Fig. 6.6a are in
plane z-axis through an angle x, measured from Ox contact over small areas between the pores, and
to Ox (Fig. 6.24b). The following relationships are the force that one grain exerts on another is
helpful: entirely transmitted through that area of contact.
The relatively great forces and small contact areas
combine to produce signicant concentrations
mxx mxy mxz cos x sin x 0
myx myy myz sin x cos x 0 (6.94) and complex distributions of stress within indi-
mzx mzy mzz 0 0 1 vidual grains. The spatial variation of the
maximum shear stress within grains can be visu-
From these relationships the transformation alized using a technique called photoelasticity
equations for the normal and shear stress compo- (Frocht, 1948). The frontispiece for this chapter is
nents are: a photograph from a photoelastic model experi-
ment that illustrates the shear stress distribution
xx xx cos 2x yy sin 2x 2xy sin x cos x induced in simulated grains of a porous sand-
stone by the forces acting on the contacts (Price,
yy xx sin 2x yy cos 2x 2xy sin x cos x
1966; Gallagher et al., 1974). Each grain in this
xy (xx yy) sin x cos x model is cut from a sheet of transparent and opti-
xy( cos 2x sin 2x) (6.95) cally isotropic material, such as polycarbonate.
The pattern of black bands in the photograph of
Similar equations enable a two-dimensional these grains is equivalent to a contour map of
transformation in the other coordinate planes. the magnitude of the maximum shear stress at
The transformation equations from Cartesian each point.
to polar stress components, or vice versa, follow A somewhat simpler picture emerges if we con-
from those just derived. Consider the stress state at sider only one grain loaded by opposed forces of
6.3 STATE OF STRESS IN THE EARTH 227
0.6
0.4
2.5 xx
t
2P (R y)x2 (R y)x2 1
r 41
r 42
d (6.97)
2
0.2 2P (R y)3 (R y)3 1
yy (6.98)
0
t r 41 r 42 d
1.5
0.2 2P (R y)2x (R y)2x
xy (6.99)
0.4 1 t r 41 r 42
described a state of stress that provides a good ref- Stress states at greater depths must be extrapo-
erence state for investigations of problems in lated from these data, inferred from studying
structural geology: data recorded on seismographs during earth-
quakes, or calculated from models. Seismic data
It is possible to imagine a condition in which the indicate the orientations of the principal stresses
lateral pressure from all sides increases steadily with
in the vicinity of a signicant earthquake
depth, so as to be everywhere equal to the vertical.
(Engelder, 1993). Typically these events range
This will not often happen in nature, but it forms a
convenient standard of reference, and may be dened
from a few kilometers depth to a few tens of kilo-
as the standard state (Anderson, 1951, pp. 13, 148). meters, so this method extends our knowledge of
the stress state throughout much of the Earths
From this description we understand that the crust. The magnitudes of the stresses are not
normal stress components (pressure) are equal determined by this method, and there can be con-
and the shear stress components are zero. In other siderable uncertainty about the orientations
words this is an isotropic state of stress and the mag- (McKenzie, 1969). None-the-less, these so-called
nitude is determined by the weight of overlying fault-plane solutions have proved to be very valuable
rocks. In this section we dene a state of stress in compiling maps of the principal stress orienta-
that is consistent with Andersons concept and tions (Zoback, 1992).
then describe data from eld measurements that In most in-situ stress measurement data sets
show typical variations from this state with the vertical normal stress ranges from zero at the
depth. Two techniques for measuring the state of surface to about 50 MPa at 2 km depth (Brown and
stress at shallow depths are described and the Hoek, 1978), more or less following a linear distri-
data from such tests are summarized. Finally, we bution (Fig. 6.26a). This is consistent with the ver-
provide examples at both the outcrop and the tical normal stress being related simply to the
crustal scale that illustrate how tectonic states of weight of the overlying rock mass. On the other
stress act to supplement the standard state, and hand, the horizontal components of normal stress
cause different styles of deformation. vary in a less systematic fashion with depth
Can stress be measured in the Earth? The mea- (Fig. 6.26b), possibly reecting differing tectonic
surement is not direct in the sense that one mea- loading conditions. Here, the ratio of horizontal
sures a distance directly with a ruler. Instead, to vertical stress is shown to be widely scattered
calculations using measured values of other phys- near the surface and converging toward values of
ical quantities and/or a model are required. one or less at depths greater than 2 km.
Several techniques for so-called in-situ stress mea- The direction of principal stress and its varia-
surement have been developed and used at expo- tion in map view across continents, plate bound-
sures and in boreholes and mines (Engelder, 1993). aries, and other tectonic features may be compiled
Amadei and Stephansson (1997) describe these from the point measurements. Usually these data
techniques and also document much of the avail- are presented in terms of the direction of the most
able data. These data have been used by mining compressive normal stress acting in the horizontal
engineers in the design of underground openings, plane, near the Earths surface (Fig. 6.27). In this
and by civil engineers in the design of founda- particular gure the authors have focused on
tions for dams and other large construction pro- California and the orientation of this stress near
jects. As structural geologists, our interest in the the San Andreas Fault system (Zoback et al., 1987).
state of stress stems from the fact that the evolu- They nd that the direction of the greatest com-
tion of geologic structures depends upon the tem- pression is remarkably consistent across this
poral and spatial variations of stress. region, being more or less from northeast to south-
Because in-situ stress measurement techniques west. This direction varies from somewhat oblique
require direct access to the rock mass, stress mea- to nearly perpendicular to the trace of the San
surements have only been made at shallow depths Andreas Fault zone. Where the most compressive
in the crust, typically less than a few kilometers. stress is oblique to the fault, one can think of this
6.3 STATE OF STRESS IN THE EARTH 229
1000 xy y x y z zy zx x z 0 (6.100)
Fx Fy 0,Fz g*
1500
Here g* is the weight per unit volume of rock.
Australia Positive z is upward from the surface, so the
2000
USA normal stress components are negative (compres-
Canada sive) below the surface, and they increase in mag-
Scandinavia
2500 South Africa nitude linearly with depth for a constant unit
Other regions weight.
3000 Values for the unit weight of rock vary from
0 10 20 30 40 50 60 70
Vertical stress, sv (MPa) about 2.0 104 to 3.5 104 N m3 with a pure
quartzite having a unit weight of 2.65 104 N m3
(b) (Daly et al., 1966). Summarizing many individual
0
measurements of the vertical stress component,
zz, a value of 0.0265 MPa m1 is reported for data
from around the world at depths ranging from
500
100 to 3000 m (McGarr and Gay, 1978), a value of
Depth below surface (m)
Idaho
42o
Oregon
Nevada
Kla California
Ca
ma
sc
th
ad
Mo
eM
u
ou
nta
nta
ins
ins
40o
Sie
Sa
rra
n
Ne
Gr
An
vad
ea
dr
t
ea
38o
Ha
s
yw
M
ar
SAN FRANCISCO
ou
d
nt
ai
ns
CALAVERAS Ne
va
Va
Ca d a
ll
lifo
ey
Gr
Sa orio
rn
eg
COALINGA ia
n
Co
as
t
36o
OCK
GARL
Ra
A
ng
Mojave Desert
Ari
zon
B CAJON PASS
a
CARRIZO PLAIN Ran ges Eastern
C Traverse Transverse
SANTA BARBARA Ranges
o Borehole
34 elongation
Focal Sa
Pe
n SA
n
mechanism NEWPORT-INGLEWOOD Ja N
nin
c AN
in
su
Hydraulic to D
R
la
S
ng
Geological
es
32o
124o 122o 120o 118o 116o 114o
D
Fig 6.27 Directions of greatest compressive horizontal
stress from various measurement techniques near the San
0
xx yy zz g dz
(6.101)
Andreas Fault system in California. Reprinted with
permission from Zoback et al. (1987), copyright 1987. xy yx yz zy zx xz 0
Jaeger and Cook refer to this as Heims rule after
depth is referred to as a lithostatic state of stress. For the Swiss geologist Albert Heim (Jaeger and Cook,
a variable density, , and variable gravitational 1979, p. 371). For the investigation of many struc-
acceleration, g, with respect to depth, the stress tural problems, it is not necessary to consider the
components in a body of rock at depth, z D, continuous variations in these quantities. Rather,
may follow the relationship: an average or representative unit weight can be
6.3 STATE OF STRESS IN THE EARTH 231
(a) 1 (b)
Sxx /Rg*b 1
b Syy /Rg*b
b
0 0
1 1
y/b
y/b
2 2
3 3
4 4
0 1 2 3 4 0 1 2 3 4
x/b x/b
(c) 1
b Sxy /Rg*b contours of shear stress are quite complex in
pattern, but a 0-contour extends from the ridge
0 top vertically downward, indicating that the prin-
cipal stress directions are vertical and horizontal
1 along that symmetry line. The small values of
shear stress relative to the two normal stresses
y/b
(z 0), along with the gravitational force compo- (a) Vertical section
nents per unit volume, Fi, are given by:
Fluid injection
xx yy ; zz g*z
1 zz
xy yx yz zy zx xz 0 (6.104)
Pressure Hydraulic
Fx Fy 0, Fz g* D
transducer fracture
This state of stress is referred to as a state of perfect
connement.
The relationship between the vertical and hori-
zontal components of stress (6.104) depends upon
Packer
an elastic property called Poissons ratio, . For this
discussion it is sufcient to understand that SV
Poissons ratio is a dimensionless number with
values that range between 0 and 1/2. For 0.25, the SH
value used for the models illustrated in Fig. 6.30, the
horizontal stress is one-third of the vertical stress
according to (6.104). For 1/3, the horizontal stress
is one-half of the vertical stress, and as goes to 1/2, Sh
the vertical and horizontal stress components (b) Horizontal section SH
become equal. Thus, Andersons standard state is
consistent with an elastic, perfectly constrained
rock mass only if Poissons ratio is equal to 1/2.
Wellbore
u
6.3.2 Measurement of in-situ stress:
hydraulic fractures and wellbore
breakouts Hydraulic
Two common methods for stress measurement at
fracture
depth in the Earths crust involve data taken from
wellbores. The rst method is based upon per-
turbing the local state of stress near a wellbore by
increasing the internal uid pressure until the Fig 6.31 Schematic illustrations of hydraulic fracture
wall of the wellbore fractures (Fig. 6.31). Because generation from a wellbore. (a) Vertical cross section in the
the fracture is induced by uid pressure, this is plane of the fracture and containing the wellbore.
(b) Horizontal cross section through the fracture and
referred to as the hydraulic fracturing method. The
wellbore. State of stress in absence of fracture and wellbore
objective is to determine the magnitudes and ori-
is (SV, SH, Sh).
entations of the three principal stresses at the site
of the measurement, so this is referred to as the in-
situ stress. Here we introduce the elementary con- surface, it is presumed that one principal stress is
cepts and theory behind these tests. Amadei and vertical. The magnitude of this compressive stress
Stephansson describe hydraulic methods for is called SV, and it is aligned with the vertical axis
stress determination in more detail, and evaluate of the wellbore (Fig. 6.31a). Furthermore, it is pre-
more general conditions for these tests (Amadei sumed that the magnitude of this stress is deter-
and Stephansson, 1997). Other techniques for esti- mined by the average unit weight, g*, of the
mating stress in the Earths crust involve the overlying rock and the depth, D:
interpretation of earthquake data (Hanks, 1977;
SV g*D (6.105)
Scholz, 1990) and the interpretation of geological
structures (Zoback et al., 1989; Zoback, 1992). Given SV, the problem is reduced to nding the
Based on proximity of the Earths traction-free magnitudes of the greater and lesser principal
6.3 STATE OF STRESS IN THE EARTH 235
O x 1 R 2
R 2
rr (SH Sh) 1 P
2 r r
Circular Remote stress
2 4
1 R R
R hole r >> R, u = 0o (SH Sh) 1 4 3 cos 2
2 r r
Fig 6.33 Geometry and stress components for the Kirsh 6.108)
solution to the elastic boundary value problem of a circular
hole in an infinite body (Jaeger and Cook, 1979). 1 R 2
R 4
r (SH Sh) 1 2 3 sin 2
2 r r
(6.109)
system with origin at the center of the hole (Fig.
6.33). For the sake of mathematical convenience
2 2
1 R R
the elastic body extends to an innite distance (SH Sh) 1 P
2 r r
from the hole, but the solution approximates
4
nite bodies that extend to distances that are 1 R
(SH Sh) 1 3 cos 2 (6.110)
great compared to R. The internal boundary con- 2 r
ditions for this problem are specied in terms of
the traction acting on the edge of the hole: According to Jaeger and Cook this is, perhaps, the
most important solution for the discipline of rock
BC: on r R,tn P,ts 0 (6.106) mechanics (Jaeger and Cook, 1979, pp. 249), and it
has been used for a number of important applica-
This traction presses against the hole boundary tions in structural geology as well. These rather
with a magnitude equal to P, and has no shear complicated looking equations can be reduced to
component, so this is equivalent to a static uid some simple relationships between the stress
pressure. The remote boundary conditions are components and the pressure in the hole by ana-
written in terms of the stress components decay- lyzing how the stresses are distributed about the
ing to some uniform values at innite distances: hole.
We learn from the Kirsh solution how the
stresses are distributed with radial distance from
rr 12 (SH Sh) 12 (SH Sh)cos 2 the hole. For example, in Fig. 6.34a we plot the
r
BC: as , r 21 (SH Sh)sin12
1
radial and circumferential stress components for
R 2 (SH Sh) 2 (SH Sh)cos 2 the case of uniaxial remote compression. These
stress components are normalized by the magni-
(6.107)
tude of the remote stress, SH, and are plotted as
For 0 the stress components are rr SH, r they are distributed along the radial line, 0o,
0, and Sh. In other words this is a biaxial from the edge of the hole, r/R 1, to a distance r/R
state of compressive stress with the radial compo- 5. The shear stress is zero along this line of sym-
nent being the greatest compression, SH, and the metry. For uniaxial compression the radial stress
circumferential component being the least com- is zero at the edge of the hole, it increases slightly
pression, Sh. SH and Sh are the magnitudes of the to a tensile maximum, and then steadily
remote principal stresses, because the shear stress decreases toward the remote compressive value
is zero for this orientation. rr/SH 1. The circumferential stress at the edge
A solution to this elastic boundary value of the hole is tensile and of the same magnitude
6.3 STATE OF STRESS IN THE EARTH 237
r r P pressure, Pc:
r
At 1 and 0, : r 0
R For T, P Pc 3Sh SH T (6.113)
P SH 3Sh
(6.111) By measuring the breakdown pressure during the
hydraulic fracturing procedure, and by measur-
For a given state of remote biaxial compres- ing the tensile strength in the laboratory for a
sion, the pressures necessary to induce a circum- sample of the formation being fractured, two of
ferential tension at these points are: the four quantities in this equation can be deter-
mined (Scheidegger, 1962).
P 3Sh SH (6.112)
Because the appropriate tensile strength is
To initiate an opening fracture in otherwise that for the rock at the in-situ conditions of stress,
unfractured rock, the local tensile stress must temperature, etc., and because these conditions
equal the tensile strength, T, and the pressure nec- may be difcult to reproduce in the laboratory, an
238 FORCE, TRACTION, AND STRESS
alternative method usually is employed (Brede- ular to the direction of Sh, so the fracture orienta-
hoeft et al., 1976). During the second (and subse- tion determines the orientation of SH and Sh. The
quent) cycles of uid injection, the pressure other principal stress is presumed to be vertical.
versus time record is monitored and then ana- Alternatively, one can use the breakdown pres-
lyzed carefully to identify a change in slope of the sure, Pc, and a measure of the tensile strength, T,
rising pressure that signals the reopening of the in the determination of SH.
fracture. The elastic model suggests that the re- The second wellbore procedure provides a
opening pressure, Pr, is that necessary to increase direct determination of the orientation of the in-
the circumferential stress just slightly above zero: situ stresses, SH and Sh. As with the hydraulic frac-
turing method, the simplest interpretations
For 0,P Pr 3Sh SH (6.114)
depend upon a vertical wellbore and the pre-
This interpretation of the pressure versus time sumption that one of the principal stresses is ver-
record implies that the in-situ tensile strength is: tical. A variety of instruments, including the
borehole camera, dipmeter, acoustic televiewer,
T P c Pr (6.115)
and electrical resistance microscanner are capa-
The nal measure of pressure used to estimate ble of measuring the shape of the wellbore
the in-situ stresses is that just sufcient to hold the (Amadei and Stephansson, 1997, p. 308). Although
developed fracture open against the least com- the drilling bit is designed to cut a cylindrical
pressive horizontal stress, Sh. The relationship hole, the records from these instruments demon-
between this pressure and stress is found from a strate that sections of some wellbores are not
different solution to the elastic boundary value cylindrical, but instead have systematic increases
problem, because the geometry now is that of a in radii along two diametrically opposed zones
crack, not a circular hole. The opening, u, of a (Fig. 6.35). These zones are referred to as wellbore
crack in an elastic body is proportional to the dif- breakouts because it is inferred that the hole was
ference between the internal pressure, P, that enlarged by the breakage of rock, due to a local
forces the walls apart, and the remote compres- stress concentration, and the subsequent spalling
sive stress, here taken as Sh, that pushes them of the rock fragments into the wellbore.
together (Pollard and Segall, 1987): The geometry of the zones of broken rock
associated with a wellbore breakout suggest that
u P Sh (6.116)
these are not a result of a single fracture extend-
Thus, the pressure and the stress must be just ing perpendicular to the wellbore, as in the
about equal as the crack starts to open, or as it hydraulic fracturing procedure (Fig. 6.31). Instead,
closes. Just as the fracture surfaces come together, it has been proposed that a set of shear fractures
a second knee in the pressure record indicates the oriented oblique to the wellbore (Fig. 6.35), or a set
shut-in pressure, Ps, and this is interpreted as equal of opening fractures oriented parallel to the well-
in magnitude to the least compressive stress: bore, is responsible for the fragmentation of the
rock (Zoback, 1985; Zheng et al., 1989). In either
For u 0,P Ps Sh (6.117)
case the stress concentration induced by drilling
Given the depth of overburden, D, and its the hole is held responsible for the fracturing.
average unit weight, g*, and the two pressures, Ps Once the rock is fractured, ow of the drilling
and Pr, read from the pressure versus time record, uid carries the fragments away, leaving the open
one can calculate all three in-situ principal stresses breakout.
as: The Kirsh solution for the elastic boundary
value problem of a circular hole subject to inter-
SV g*D
nal pressure and remote biaxial compressive
S h Ps (6.118)
stresses provides the equations necessary for an
SH 3Ps Pc T 3Ps Pr
elementary analysis of the stress concentration
This interpretation presumes that a vertical frac- that may cause breakouts. To assure that
ture has propagated in a plane that is perpendic- hydraulic fractures have not initiated, and that
6.3 STATE OF STRESS IN THE EARTH 239
rr P
r 3 one would expect individual dikes to be curved,
At 1 and , : r 0
R 2 2 reecting propagation along a curved stress tra-
P 3SH Sh
jectory. The radial dike set in the Spanish Peaks
(6.120) region (Fig. 6.12a) presumably reects variations
240 FORCE, TRACTION, AND STRESS
0 100
RO km
CK
Y
M
O
UN
TA
IN
CA
S
NA
DI
AN
BA
SI
N
LEGEND
Well location number
Major population
Minor population
108o
116o 112o
Fig 6.36 Map of western Canada with directions of least (SH Sh)d2
0.5 (6.121)
horizontal compressive stress inferred from wellbore Pr2o
breakouts. Reprinted from Bell and Babcock (1986) with
For the model result shown in Fig. 6.37 the orien-
permission of Elsevier.
tation of the greatest regional compression is 082o
and the estimated difference between the princi-
in the orientation of the stress eld, so these vari- pal values is (SH Sh) 0.03P. Given an estimated
ations can be deduced from the pattern of dikes. upper bound for the magma pressure of
The traces of dikes are correlated (Fig. 6.37) to the P 100 MPa, based on the strength of rock sur-
trajectories of maximum compressive stress deter- rounding the magma chamber, the regional stress
mined from the solution to a boundary value difference is less than 5 MPa. For comparison an
problem of elasticity theory (Pollard and Muller, average value of 2.1 MPa is reported for 26 mea-
1976; Muller and Pollard, 1977). For comparison surements of horizontal principal stress differ-
purposes the trajectories and dike traces are pre- ence from seven wells in the Piceance Basin of
sented in alternating vertical strips across the Colorado using the hydraulic fracturing tech-
map. Here the correlation is surprisingly good nique (Bredehoeft et al., 1976).
and there is a unique relationship among the The analysis used to estimate the regional
values for the remote principal horizontal stress stress difference is based upon the solution for an
components, SH and Sh, the magma pressure, P, elastic body with a pressurized hole (the magma
the radius of the magma chamber, ro, and the dis- chamber), a rigid boundary (the Sangre de Cristo
tance from the magma chamber to the Sangre de range), and a regional stress eld. No account is
Cristo Mountain front, d: made of the perturbation of the stress eld by the
6.4 CONCLUDING REMARKS 241
37o 30'
Stress
trajectories
Radial
dikes
0 5 km
Distance
37o 15'
105o 00' 105o 45'
Fig 6.37 Map of Spanish Peaks region of southeastern relationship one can map the chains of cinder
Colorado with traces of dikes (dotted lines) and trajectories
cones just like the exposures of dikes at the
of greatest compressive horizontal stress (short solid lines)
Spanish Peaks and use these data to estimate the
from an elastic solution. Reprinted from Muller and Pollard
(1977) with permission of Birkhanser-Verlag. horizontal principal stress orientations along vol-
canic arcs like the Aleutians (Fig. 6.38). Note that
the estimated motion vectors for the Pacic plate
emplacement of the dikes, yet each dike will relative to the North American plate are sub-par-
perturb the local stress eld and could thereby allel to the maximum horizontal compression
inuence the propagation path of the next dike inferred from the cinder cones. Apparently the
(Meriaux and Lister, 2002). Also, this analysis convergence direction of the two plates is
neglects the inuence of pre-existing fractures, reected in the compression direction within the
which may guide the propogating dike (Ziv and volcanoes near the Earths surface. This informa-
Rubin, 2000), and the effects of topography (Fialko tion is crucial to an understanding of the struc-
and Rubin, 1999). tural and volcanic history of such regions.
Andersons concept has been applied to the
injection of dikes under active volcanoes in the
Aleutian Arc (Nakamura, 1977; Nakamura et al.,
6.4 Concluding remarks
1977). A direct relationship between chains of
cinder cones on large volcanoes and the orienta- In the Newtonian context of rigid-body dynamics
tion of sub-surface dikes was inferred by suggest- it is understood that forces are associated with
ing that dikes provided the conduits for ow of accelerations. On the other hand, in the context of
magma to the surface of the volcano. With this a deformable solid the stress is associated with
242 FORCE, TRACTION, AND STRESS
1000 km
Structural block diagram of a part of the Penninic Alps, language of mathematics. The goal is to construct a
Switzerland (Argand, 1911).
working model of the universe out of commonplace
notions: ideas of number and order and measures of
At the heart of all this calculation lies the deeply held time and distance. With such a working model, we
conviction that natural phenomena are, in essence, can leap ahead in time and predict what the other-
the consequence of just a small number of physical wise opaque future has in store for us (Peterson,
laws, and that these laws are best expressed in the 1993).
244 CONSERVATION OF MASS AND MOMENTUM
I
n the context of structural geology we can con- moments during rare events such as earthquakes.
struct a working model of mountain building In the absence of appreciable accelerations, con-
from those small number of physical laws and servation of momentum requires that the resul-
then leap backward in time and understand the tant forces and torques acting on a rock mass are
development of geological structures such as those negligible. This condition ensures that the linear
depicted in the structural block diagram (Chapter and angular momenta are nearly constant with
7, frontispiece) of a part of the Penninic Alps of respect to time. These requirements are funda-
Switzerland constructed by Emile Argand and pub- mental and should be examined at an early stage
lished in 1911 (Argand, 1911). This is one of the ear- of any program of modeling geologic structures.
liest published block diagrams in the literature of Our review of concepts introduced in the
structural geology (McIntyre and Weiss, 1956; typical college physics class leaves us short of
Howarth, 1999) and it illustrates what was known having all of the necessary tools of mechanics to
in the early part of the twentieth century about analyze tectonic processes. Given our human time
one of the most interesting and complex regions of scale and the poor resolution of our eyes for dis-
folding and faulting in that mountain chain. cerning small changes in the shapes of objects,
Among the small number of physical laws that rock does seem quite rigid. One might think that
can be employed to understand tectonic processes the dynamics of rigid bodies would be as far as we
and their structural products are those of mass, have to go to tackle problems of rock deformation
momentum, and energy conservation. Newtons in the Earth. Indeed, the early practitioners of
Second Law of Motion, for example, relating force, plate tectonics conceptualized plates of the litho-
mass, and linear acceleration is embodied in a sphere as thin rigid masses slowing moving over a
generalized statement of momentum conserva- mobile aesthenosphere. Similarly, the typical mid-
tion. Kinetics is the branch of mechanics that con- twentieth century structural geology textbook
siders the action of forces and torques on particles introduced students to faulting with diagrams
and rigid bodies, and their resulting accelera- showing rigid blocks moving relative to one
tions. This should be familiar ground for students another (Billings, 1972), and laboratory exercises
recently exposed to the mechanics section of a utilized painted wooden blocks to illustrate the
college physics course. However, we offer a short patterns of offset strata (Fig. 7.1). While these
review for those who would benet from a second gures and blocks may be instructive guides to
encounter with these topics. This review serves understanding the map patterns of faulted strata,
another important purpose. Textbooks in struc- the perceptive students might ask: what happens
tural geology typically fail to make a clear con- near the end of a fault? Clearly a rigid block model
nection between the material taught in the could not provide a satisfactory answer, because
mechanics section of an introductory college the model fault has no end.
physics course and the mechanical concepts Moving beyond the rigid block models, it is nec-
employed in analyzing geological structures. By essary to consider a continuous and deformable
making that connection explicit students are pre- body of rock and to broaden our perspective to con-
pared to use the mechanics effectively and with a strain explicitly how mass and momentum are
condence that stems from understanding the conserved throughout such a body. Here we postu-
fundamentals of the discipline. late that temperature changes and chemical
We begin with the concepts of linear and changes within the body are negligible, so the
angular momentum as treated in particle dynamics model is isothermal and isochemical. If heat ow and
and generalize this to rigid-body dynamics. While chemical reactions play an important role, than
the methods of classical dynamics have applica- conservation of energy and conservation of chem-
tion in many familiar human endeavors (from ical species must be included. With these limita-
tracking satellites to playing snooker) where accel- tions in mind, conservation of mass leads to the
erations are key to understanding, the rock masses equation of continuity and the conservation of
that comprise Earths crust do not experience momentum leads to the equations of motion for the
appreciable accelerations, apart from those brief material continuum. In turn, the equations of
7.1 PARTICLE DYNAMICS 245
Particle
of mass m
Acceleration, a
y
Resultant f(n)
x force, F
(b) (b) z
f(1)
f(2)
Particle
of mass m
Velocity, v
y
Fig 7.1 Rigid structural block diagrams of normal fault
Linear f(n)
x momentum, p
cutting inclined sedimentary strata (a) Faulted blocks.
(b) Left-hand block eroded to remove fault scarp. Reprinted
Fig 7.2 (a) Particle of mass, m, with set of forces, f(i),
from Billings (1972) by permission of Peavson Education, Inc.,
resultant force, F, and acceleration, a. (b) Particle of mass, m,
Upper Saddle River, NJ.
with set of forces, f(i), linear momentum, p, and velocity, v.
motion are specialized for the linear elastic solid with respect to a Cartesian coordinate system (x, y,
and the linear viscous uid, two of the most useful z) and kinematic quantities, such as velocity and
material behaviors in studies of structural geology. acceleration, are dened with respect to an iner-
Finally, we consider the equations of motion for a tial frame of reference (Fig. 7.2). An inertial frame of
body that experiences negligible accelerations reference is xed in space or moving with a constant
and derive the equations of equilibrium. velocity relative to the distant stars. Newtons
Second Law describes the relationship between
the acceleration of such a particle, relative to an
7.1 Particle dynamics inertial frame of reference, and the forces acting
upon it:
7.1.1 Force and linear momentum
F ma (7.1)
Consider the motion of a single particle under the
action of applied forces (Resnick and Halliday, Here the vector a is the acceleration of the particle,
1977, Chapter 5). In this idealized context a parti- the vector F is the resultant force acting on the par-
cle is considered to be an isolated quantity of mass ticle, and the scalar m is the mass of the particle.
residing at a point: it has neither a size nor a shape. What Newton meant by resultant force, F, is
Furthermore the mass is postulated to be constant the vector summation of all the forces f(1), f(2),
with respect to time, so conservation of mass is . . . , f(n) acting on the particle (Fig. 7.2a). Here the
satised by denition. The particle is positioned numbers in parentheses identify the different
246 CONSERVATION OF MASS AND MOMENTUM
(7.2)
p mv (7.4) z A
fco
Just as the force and acceleration have the same s
direction, the linear momentum and velocity E
have the same direction because their respective n B
components are in the same ratio, for example
px/py vx/vy. f
O y
Given constant resultant force acting on a par-
r
ticle of known mass, one can calculate its acceler-
fsin
ation using (7.1). The instantaneous velocity of the D
particle is found using the fact that the time rate
x C
of change of velocity is the acceleration; so one Particle of
can integrate a constant acceleration over time to mass m
calculate the changing velocity. In differential
Fig 7.4 Schematic diagram to define the torque, , with
form dv adt, and integrating:
respect to the origin, O, of a Cartesian coordinate system as
dv a dt,so v at v(0) (7.5)
the vector product of the radial position vector, r, and the
net force, f, acting on a particle of mass, m.
Here r and f are the magnitudes of the radial and the particular location of the origin of the coordi-
force vectors, respectively, and f sin is the com- nate system, O, and inertial frame of reference
ponent of f along the line DE drawn perpendicular (Fig. 7.4). That is, we speak of the torque with
to the radial vector and in the plane containing r respect to the origin, or the angular momentum
and f (Fig. 7.4). Thus, force acting parallel to the with respect to the origin. If we were to move the
position vector produces no torque and force location of the origin these quantities would
acting at right-angles to the position vector will change in magnitude and direction. Further-
contribute all of its magnitude to the torque. more, the particle is not tied to the origin, so it
The angular momentum, , is dened in terms will move along a path dened by the line of
of the position vector of the particle, r, and the action of the applied force, f, rather than spin
linear momentum as: about an axis of rotation parallel to n. The line
dened by n is not an axis of rotation for the par-
r mv r p (7.9) ticle per se, but rather it is the line along which
Because the linear momentum acts in the same the torque and angular momentum are directed.
direction as the force, the angular momentum, ,
acts in the same direction as the torque, (Fig. 7.4).
That is, acts along the normal n with a direction 7.2 Rigid-body dynamics and
determined by the right-hand convention. The statics
magnitude of the angular momentum is:
x
Here it is understood that V is the total mass of
the body and iVi is the mass of each volume
element. The total mass, V, contained in B does
r* not change with time: it is conserved. The small
volumes, Vi, do not change with time because
the body is rigid throughout. The individual den-
ri sities, i, are uniform within each element but
may vary spatially throughout the body. On the
other hand, these individual densities do not vary
in time: there is no mass transport from one
Vi, ri
volume element to another. In the limit as Vi
goes to zero this collection of volume elements is
equivalent to a system of particles, and the
Rigid body, B motion of each element is governed by the re-
lationships reviewed in the previous section.
Fig 7.5 Schematic diagram to define the center of mass, Depending upon the forces that are applied to
r*, of a rigid body made up of m volume elements, Vi, of this rigid body an individual volume element may
mass density, i. move in a complex manner that involves both
translations and rotations with respect to the ref-
erence frame. Any two elements, however, do not
southwest and interpreted to be continuous. In move relative to one another, because the body is
Fig. 7.5 we schematically represent Argands block rigid by denition. There is one position in this
diagram as a rigid body called B, which has an body, called the center of mass (Fig. 7.5), that moves
average density, , and total volume, V, neither of in the same manner as a single particle of con-
which change with time. You might wonder how centrated mass, V, when subject to the resultant
the folds and faults could have formed if the block of all forces applied to the body. The center of
were rigid and, of course, the answer is that they mass is quite special in that it behaves like the par-
could not. However, the center of mass can be cal- ticles we dealt with in the previous section, but
culated for an arbitrary instant in time during the there is only one such center of mass in any rigid
deformation using the method introduced here. body. The center of mass has a position vector, r*,
Furthermore, Argands example serves to empha- that is (Resnick and Halliday, 1977, Chapter 9,
size that this is not a vacuous exercise in mathe- p. 164):
matical physics, but rather a crucial step toward m
understanding the deformation of Earths crust. r* V r V
i1
i i i (7.13)
The rigid body of Fig. 7.5 is lled with m (not to
be confused with the mass, m) small volume ele- In other words, the product of the position vector
ments of rock, Vi, each having a uniform density, for the center of mass of the body and the total
i. Each volume element is located with a position mass is equal to the vector sum of the product of
vector, ri, radial to the origin, O, of the (x, y, z)- the position vectors for each volume element and
coordinate system and inertial frame of reference. their respective masses.
Fixing a coordinate system at some location in or The coordinates of the center of mass of the
on the Earth usually provides a suitable inertial body are the individual components of the pos-
frame of reference. The average density of the ition vector r*. For example, the x-coordinate of r*
body is related to the individual densities and is the component x*:
250 CONSERVATION OF MASS AND MOMENTUM
d2 d2 d2
1V1 2r1 2V2 2r2 mVm 2rm torques acting on the body and is derived in
dt dt dt
Section 7.2.4.
F (s) F (b) (7.18)
7.2.3 Evaluation of surface and body
Note that the body force acts on all m elements of forces
the body, but the n surface forces act at particular We replace the mechanical action of the exterior
locations on the exterior surface of the body (Fig. rock mass on the body schematically in Fig. 7.6
7.6). with a set of surface forces, f1, f2, . . . , fn, acting at
Because neither the volume nor the average discrete points. The resultant of these n surface
density is a function of time for the rigid body, the forces is their vector sum which acts at the center
rst time derivative of the left-hand side of (7.15) is: of mass of the body:
d n
V r* Vv* P
dt
(7.19) F(s) fj
j1
(7.22)
related to the individual densities and volumes of momentum is neither created nor destroyed
its constituent elements using (7.15). spontaneously, but changes with time in strict
Taking the time derivative of the right-hand accordance with the action of these forces.
side of (7.15) with volume and density of each If the angular momentum does not change
element considered constant gives (7.16), the appreciably with time one postulates that it is
vector sum of the linear momenta for the m el- exactly zero for modeling purposes. The conserva-
ements. The cross products of the respective posi- tion of angular momentum then requires that the
tion vectors and linear momenta are the angular resultant of all external torques is zero:
momenta of the elements, and the vector sum of n
d
these is: if
dt
0, then [r f ] r* Vg* 0
j1
j j (7.31)
r1 1V1v1 r2 2V2v2 . . . rm mVmvm
This is the second condition used to establish the
(7.27)
static equilibrium of a rigid body. The other condi-
From (7.11) the time derivative of the angular tion is (7.21). Note that the angular momentum of
momentum is the torque, so the resultant torque, individual elements of the body may change with
T, acting on the rigid body is: time, but the angular momentum of the body as
d d a whole, with respect to an inertial frame of ref-
(r 1V1v1) (r2 2V2v2) erence, is constant in time if the resultant torque
dt 1 dt
d is zero.
(rm mVmvm) T (7.28)
dt
This summation includes both the torques caused
7.2.5 Static equilibrium in integral form
for the rigid continuum
by external forces acting on the body and those
When contemplating the equilibrium of a body of
caused by internal forces acting among the el-
rock it would rarely be practical to think about a
ements. Newtons Third Law is invoked to justify
large number of volume elements, each with a par-
ignoring torques caused by internal forces
ticular density or momentum. While this device
(Resnick and Halliday, 1977).
has clear pedagogical advantages for introducing
What remains after eliminating the contribu-
the conservation laws for a rigid body, a more prag-
tions of the internal forces are the torques caused
matic approach is to invoke the continuum and
by each of the n surface forces, fj, applied at pos-
let the physical quantities under discussion be
itions, rj, and the torque caused by the uniform
dened at every point of the body. Then summa-
gravitational body force, Vg*, acting at the center
tion over the number of elements is replaced by
of mass, r* (Fig. 7.7):
integration. One of founders of structural geology
n
in the nineteenth century, Grove Karl Gilbert (Fig.
[r f ] r * Vg* T(s) T(b)
j1
j j (7.29)
7.8a), developed his mechanical model for laccol-
This is the resultant external torque with respect ith formation (Fig. 1.18) using this approach to
to the origin, O, and inertial frame of reference. equilibrium. The concept of equilibrium, whether
From (7.11) the time rate of change of the angular used in the consideration of mountain building or
momentum, , of the body as a whole is equal to of erosion of the landscape was central to Gilberts
the resultant torque caused by all external forces method of investigation (Pyne, 1980).
acting on the body: The concept of a center of mass is extended to
n
the continuum by considering and V to be the
d
dt
j1
[rj fj ] r* Vg* (7.30) average density and total volume, neither of
which change with time because the body is rigid.
This equation expresses another of those small The mass of this body is conserved by denition.
number of physical laws mentioned at the begin- An integral over the volume of the body is derived
ning of the chapter, in this case the law of conserva- from (7.13) by letting the element volumes, Vi,
tion of angular momentum. For a rigid body, subject shrink to an innitesimal size, dV, so the position
to given surface forces and gravity, angular vector of the center of mass is:
254 CONSERVATION OF MASS AND MOMENTUM
1
r* r dV
V (7.32) (a)
V
F(s) t(n) dS (7.33)
S
n
S
F(b) g*dV (7.34)
traction vectors, t(n), acting on surface elements, S, and a
distribution of weights per unit volume, g*, acting on
V
volume elements, V.
This integral expression represents the resultant
body force due to gravity, F(b), and it may be
thought of as acting at the center of mass of the In both cases the torque is found by considering
body. the location of each surface and volume element
The surface and body forces may produce in terms of the position vector, r, and forming the
torques about the origin of the coordinate system. cross product of the position vector with the force.
7.2 RIGID-BODY DYNAMICS AND STATICS 255
T(s) [ r t (n)] dS (7.35)
S
T(b) [r g*] dV (7.36)
V
[r t(n) ] dS (r g*) dV 0 (7.38) Fig 7.9 (a) M. King Hubbert: photograph reproduced with
permission of the School of Earth Sciences, Stanford
S V
University. (b) Vertical cross section through idealized fold
All of the physical quantities in these equations and thrust mountain belt. Reprinted from Hubbert (1951)
are dened at each and every point in the contin- with permission of The Geological Society of America.
uum. None of these quantities are functions of
time, but the traction and density may vary spa-
understanding for a variety of empirically well-known
tially and must be integrable functions of the
geologic structures (Hubbert, 1951).
spatial coordinates. Gravitational acceleration is
taken as uniform in space and constant in time. In this paper Hubbert provides the motivation for
laboratory experiments using the sandbox technique.
7.2.6 An example: Appalachian fold and Similar techniques using model materials such as
thrust mountain belt sand, clay, and plaster continue to provide insights
In 1951 M. King Hubbert (Fig. 7.9a) published a concerning the development of crustal-scale struc-
paper in the Geological Society of America tures to this day (Fossen and Gabrielsen, 1996;
Bulletin entitled Mechanical basis for certain Wang and Davis, 1996; Guglielmo et al., 2000;
familiar geologic structures. The abstract reads Ackermann et al., 2001; Cobbold et al., 2001; McClay
as follows: and Bonora, 2001). The sandbox used by Hubbert
contained a rigid platen that produced a set of
A simple experiment with loose sand shows that this
thrust faults when moved laterally (Fig. 7.10). The
material exhibits faulting under deformational stress
in a manner remarkably similar to rocks. Moreover, motion of the platen apparently increased the hor-
the sand experiment is amenable to theoretical analy- izontal compressive stress while the vertical stress
sis with good agreement between predicted and remained essentially that caused by gravity. These
observed behavior. The same theoretical treatment, stress changes led to the development of the faults
with slight modication, is also applicable to the apparently analogous to those in fold and thrust
behavior of rocks, and appears to afford a basis of mountain belts.
256 CONSERVATION OF MASS AND MOMENTUM
BC: on x 0,0 y W,
Tz(L) Rg*H H Tz(R) H z 0; tx CL g*z
(7.39)
SB BC: on x L,0 y W,
H z 0; tx g*z
CB
The tectonic traction is uniformly distributed over
Fig 7.11 Free-body diagrams for fold and thrust mountain the left side whereas the lithostatic traction
belt. (a) One-sided, active thrust concept with unbalanced increases linearly with depth. Both parts of this
tectonic traction, CL, and lithostatic tractions, g *z. traction push against the body and are positive in
(b) Lithostatic tractions on cross sections in (y, z)-plane. sign (note z is negative below the surface). The
(c) Additional loading necessary for static equilibrium tractions are drawn as arrows pointed at the mid-
includes shear tractions SB on the base, and TZ on the sides, section of the body, but it should be recognized
and normal traction CB on the base.
that the traction is uniformly distributed over the
entire side of the free body. The traction on the
The explanation based on an active thrust right side pushes against the body in the negative
apparently was satisfactory for many structural x-direction, so its sign is negative.
geologists of the day; however, as Hubbert pointed The top surface of the free body is traction
out, the distribution of surface and body forces free and on the bottom surface the shear traction
258 CONSERVATION OF MASS AND MOMENTUM
proposed by Hubbert is taken as uniform with a The shear traction is equal to the tectonic traction
magnitude, SB: times the ratio of the height to the length of the
body.
BC: on 0 x L, 0 y W, z 0; tx 0 The only y-components of the tractions acting
BC: on 0 x L, 0 y W, (7.40) on the free body (Fig. 7.11b) are those related to the
z H; tx SB lithostatic loading and these exactly balance one
another. The z-components of the surface trac-
On the front and back sides of the free body we tions and the body forces are shown on the free
follow Hubberts suggestion that there are no body in Fig. 7.11c. The presence of a shear traction
shear tractions: on the bottom of the body raises a question about
BC: on 0 x L,y 0, the possibility of a shear traction acting in the z-
H z 0;tx 0 direction on the left and right sides. For the
(7.41) moment we do not explicitly dene these trac-
BC: on 0 x L,y W, tions, because their distributions and magnitudes
H z 0;tx 0 are best addressed in terms of the conservation of
angular momentum, but account for them as
These conditions would apply for a linear moun-
follows:
tain belt.
For equilibrium the integral of the x-compo- BC: on x 0,0 y W,
nents of the tractions over the surfaces on which H z 0;Tz Tz(L)
they act must sum to zero (7.37): (7.45)
BC: on x L,0 y W,
0 W 0 W
H z 0;Tz Tz(R)
(CL g*z) dy dz (g*z) dy dz
H 0
W L
H 0
Here Tz(L) and Tz(R) are unspecied functions rep-
(S ) dx dy 0
resenting the traction components in the z-direc-
B (7.42) tion on the left- and right-hand sides of the body.
0 0
The top surface is traction free, but the bottom
Because the integrands are constant the inner surface is subject to a uniform traction in the z-
denite integrals over the limits 0 to W with direction of magnitude CB:
respect to y and 0 to L with respect to x produce
BC: on 0 x L,0 y W,
the constants W and L, respectively. The remain-
z 0;Tz 0
ing integrals are evaluated over their respective
(7.46)
limits and the resulting expression is simplied as BC: on 0 x L,0 y W,
follows: z H;Tz CB
| | |
0 0 W
1 1 On the front and back sides of the body the sym-
W (CLz g*z2) W g*z2 L SB y 0
2 H 2 H 0
metry of the loading dictates that there are no
CLWH SBWL (7.43)
shear tractions in the z-direction:
The net force caused by the tectonic loading on
BC: on 0 x L,y 0,
the left side of the free body is exactly balanced by
H z 0; Tz 0
the net force caused by the resisting shear loading
on the bottom of the body. The quantitative rela- BC: on 0 x L,y W,
tionship between the magnitudes of the tectonic H z 0;Tz 0 (7.47)
traction on the left side, CL, and the resisting shear
traction on the bottom, SB, is: Referring to (7.37), we integrate the z-compo-
nents of the tractions over the surfaces on which
SB CL
H
L
(7.44) they act, integrate the body force over the volume,
and set the sum to zero:
7.2 RIGID-BODY DYNAMICS AND STATICS 259
0 W 0 W W L
z
Tz(L) dy dz Tz(R) dy dz C dx dy
B
(a) L/2
H 0 H 0 0 0
Center
0 W L
of mass
(g*) dx dy dz 0 (7.48)
H 0 0 rg*V
H/2 r* b*
We postulate that Tz(L) and Tz(R) produce net O r x
surface forces that are equal in magnitude and
CB
oppositely directed, so they exactly balance and
the rst two integrals cancel one another.
Carrying out the integration, we have: (b)
z
CL L
CB g*H (7.49)
the two vectors. Thus, we evaluate the torque pro- (g* sin *) dx dy dz 0
duced by the normal traction on the base of the (7.51)
free body, and that due to the weight acting at the 12 CBWL2 12 g*L2WH 0
center of gravity as (Fig. 7.12a):
From (7.49) we know that CB g*H, so these
On 0 x L,0 y W,z 0, torques exactly balance.
Next consider the torques produced by the tec-
or r x; Tz CB,sin sin (2)
tonic traction acting on the left side of the body
(7.50) and the shear traction acting on the right side
At x 12L, y 12W,z 12H, (Fig. 7.12b). The shear traction on any vertical
surface must be zero where that surface intersects
L
or r r* ;g g*,sin sin * the traction-free surface of the Earth. Thus, we
2 sin *
postulate that the distribution of shear traction
The torque caused by the bottom traction is nega- on the right side varies linearly from zero at the
tive (it acts out of the page in the negative y- top to a value SR at the bottom:
260 CONSERVATION OF MASS AND MOMENTUM
sin
scribed tectonic boundary conditions, this analy-
L
zCL sin (2) dy dz
sis does not consider the deformation of the body
0 0 0 0
itself. For structural geologists the deformation of
z
SR 1 sin () dy dz 0
H
(7.53) a rock mass is the central focus, so we turn now to
conservation of mass and momentum in a
CLWH SRLW 0 deformable body of rock idealized as a material
continuum. These relationships underlie the
Using (7.44) we nd SR is related to the magnitude more specialized equations that we use later in
of the basal shear traction as: the textbook to model rock deformation during
folding, faulting, fabric development, and the for-
SR CL
H
L
SB (7.54) mation of many other geological structures.
We note here that both of the conservation
The shear traction on the right side in the lower laws are developed without regard for particular
right corner (x L, z 0) is equal to the shear trac- material properties other than mass density, so
tion on the bottom, because the respective shear they apply quite generally to the entire spectrum
stresses must be equal. The shear tractions on the of rock deformation, from silicate magmas that
left- and right-hand sides are: ow like a uid to rocks that deform like a mal-
leable solid to rocks that deform like an elastic
Tz(L) Tz(R) SB 1 z
H (7.55)
solid. We do restrict our attention to conditions
where temperature uctuations and the associ-
ated changes in density and ow of heat can be
This completes the equilibrium analysis of
ignored. In addition, we ignore chemical reac-
Hubberts Appalachian free body. This, or the
tions and the associated changes in density and
analogous procedure using relationships that
concentration of chemical species. Thus, the rela-
include the time rate of change of the linear and
tionships derived here apply, strictly speaking, to
angular momentum, is a vital step in any model-
a material that is isothermal and isochemical.
ing program to conrm that the prescribed trac-
Although few tectonic processes would obey these
tions and body forces are consistent with the
restricted conditions in detail, we show by
conservation of momentum. Without such
example that they do not preclude gaining con-
conrmation, any discussion of the development
siderable insight into those processes.
of structures lacks the necessary foundation in
mechanics to be credible, and it should be
7.3.1 Referential and spatial descriptions
ignored.
of motion
Recall from the discussion of kinematics in
Chapter 5 that two different sets of coordinates
may be used to describe the positions of particles in
a deforming rock mass: (X, Y, Z) are the coordinates
7.3 THE DEFORMABLE CONTINUUM 261
x X ux
10 cm
x X u,or y Y uy (7.56)
z Z uz
x x(X, Y, Z, t)
motion of particles from the initial state would be
x x(X, t),or y y(X, Y, Z, t) (7.57) a function x(X, t) that describes the continuous
z z(X, Y, Z, t)
change of position for every particle, each origi-
nally at a particular point X, for all times from t
Recall that the x (or x, y, and z) on the left-hand
0 to t the current time. In other words, the paths
sides of these equations is the value of the func-
followed by the particles are traced out by the
tion whereas the same symbols on the right-hand
position vectors x according to (7.57). We do not
sides signify the function itself. After some
know this functional relationship for the particles
unknown duration of time, slip ceased on the fault
near the fault (Fig. 7.13), so must be content with
and the conguration of particles attained that
a two-state description of the proposed elastic
seen in the exposure today (Fig. 7.13a). The position
deformation. However, elastic models could be
vector x locates the particle in the current state
investigated that would track the particles as they
that was located by the position vector X in the
accelerated from their initial positions, attained
initial state. This is called the referential descrip-
some peak velocity, and decelerated to their
tion of motion because it refers back to the initial
current positions. Given such a function that is
state. It also is called the Lagrangian description of
differentiable with respect to time one may calcu-
motion after the Italian mathematician Joseph-
late the velocity, v, of an arbitrary particle as the
Louis Lagrange (17361813), (Fig. 7.14). The coordi-
vector function G:
nates (X, Y, Z) are called the Lagrangian coordinates,
or sometimes the material coordinates because they
describe the initial positions of material particles.
v
x
|
t X
G(X, t) (7.58)
7.3 THE DEFORMABLE CONTINUUM 263
Here the subscript after the vertical bar indicates may suggest directions and magnitudes of rela-
x also is a function of the independent variable X, tive velocities. In this context current may be
which is held constant during partial differentia- taken within the time frame of active deforma-
tion with respect to time. The material coordi- tion and certainly does not mean the present day.
nates are held constant because we want to track As an example, consider another exposure
the velocity of a particular particle. For this from the Sierra Nevada of California where a
reason the partial derivative in (7.58) is referred to mac dike about 30 cm thick cuts the metamor-
as the material time derivative and the resulting phic host rock (Fig. 7.15a). The igneous rock is com-
velocity is called the particle velocity (Malvern, posed of a very ne-grained black groundmass
1969). To understand this viewpoint an analogy is interspersed with lath-shaped white phenocrysts
drawn to an observer, stationed at the origin of (probably feldspar) that are arranged in an evoca-
the material coordinates (X, Y) in Fig. 7.13, who tive pattern across the dike (Fig. 7.15b). Near the
measures the time rate of change of the particle at two contacts with the host rock the phenocrysts
every position along its path from X to x as the tend to have their long axes parallel to the
fault slips. contact, whereas near the mid-line of the dike the
Given a function x(X, t) that is twice differen- long axes are perpendicular to the contact.
tiable with respect to time, one may obtain the Furthermore, there appears to be a greater con-
particle acceleration, a: centration of phenocrysts near the mid-line. The
geological inference is that magma from an
a | |
2x
v
G(X, t)
t2 X t X t
(7.59) unknown source forced open this fracture and
owed through it for some period of time in the
Again, we do not know the function x(X, t) for the direction that the pencil is pointing. The phe-
particles near the fault (Fig. 7.13), but an elastic nocrysts were organized into a systematic pattern
model would provide the particle accelerations that eventually became frozen in place as
according to (7.59). The particle velocity (7.58) and enough heat was lost to solidify the magma. The
particle acceleration (7.59) are kinematic quanti- mechanical inference is that the injection of
ties employed in elastic solid mechanics. magma into the dike can be described by the ow
We turn now to the ow of a viscous uid, of a viscous uid containing a number of lath-
which is viewed with the current state taken as shaped solid objects.
the reference. Under given loading conditions a We adopt a spatial description of motion which
particular rate of deformation is associated with takes the coordinates (x, y, z), and time, t, as the
the particle at each coordinate position in the independent variables (Malvern, 1969). These
current state and the initial conguration of coordinates describe positions in space such that
those particles is not prescribed. Upon unloading the point (x, y) in Fig. 7.15b is associated with a par-
each particle simply remains at rest at its current ticle of a particular phenocryst at a given time,
position. A special case is that in which the veloc- but would be associated with a particle of the
ity eld does not change with time and the rate of ground mass or a different phenocryst at a later
deformation is constant at any position in the time as the magma ows through that xed point.
ow. This is referred to as a steady state of ow for This viewpoint brings attention to given points in
the viscous uid. In the geological context, a space rather then to given particles. With each
nearly steady state may exist for some period of successive instant in time the particle that was at
time, preceded and followed by periods of accel- a given position may move away and a new parti-
erating or decelerating ow. For the structural cle may move into the eld of view, but the posi-
geologists standing on an exposure of igneous tion in space remains xed. This is called the
rock the frozen pattern of aligned xenoliths, spatial description of motion. It also is called the
phenocrysts, or vesicles may suggest how this Eulerian description of motion after the Swiss
material once owed. Similarly, for an exposure of mathematician Leonhard Euler (170783), (Fig.
highly deformed metamorphic rock, although 7.16). The coordinates (x, y, z) are called the Eulerian
ow has long since ceased, the folds or boudinage coordinates and also the spatial coordinates.
264 CONSERVATION OF MASS AND MOMENTUM
(b)
x y
(x, y)
x v
In contrast to the referential description of model magma would be described at every posi-
motion, particle paths are not a primary feature tion as a function of time, providing the vector
of the spatial description, and the initial positions function:
of particles usually are not dened. For example,
v g(x, t) (7.60)
the initial positions of the phenocrysts in the
mac dike (Fig. 7.15) are unknown. A complete The quantity dened in (7.60) is called the local
spatial description of motion would be a function velocity because it refers to the velocity at a given
that describes the velocity, v, at every position, x, location. Note that both (7.58) and (7.60) are equa-
for all times from t 0 to t the current state. In tions for velocity, v, so the particle velocity at a
our eld example we might consider a steady particular location and time is the same as the
velocity eld of a viscous uid containing lath- local velocity in that location at that time.
shaped solid objects within a conduit of xed However, the functions G(X, t) and g(x, t) are dif-
width equal to that of the dike. The velocity of the ferent because the former describes the velocity of
7.3 THE DEFORMABLE CONTINUUM 265
v
|
g(x, t)
t x t
(7.61)
z
| | | | |
z x
z y
s r x y s r y x s r
(7.63)
The quantity dened in (7.61) is referred to as the Making the appropriate associations for the quan-
local rate of change of velocity (Malvern, 1969) and tities in the vector function (7.62) we have:
not as the acceleration because it does not neces-
sarily dene the particle acceleration (7.59) at the
position x and time t. To appreciate this apparent
v
|
v x
| | | |
v t
t X x t t X t x t X
(7.64)
contradiction recall the velocity distribution The left-hand side is the material time derivative
(5.19) within a rising viscous sphere (Fig. 5.12). of the velocity which is, by (7.59), the particle
This is an example of steady ow, which means that acceleration, a. The rst partial derivative on the
all of the kinematic quantities are constant in right-hand side may be calculated from the local
time at every point. Thus, at every current pos- velocity (7.60), which is given. The second partial
ition x the time derivative of velocity as dened in derivative is the particle velocity (7.58), but that is
(7.61) is identically zero. On the other hand the equivalent to the local velocity, v, dened in (7.60).
266 CONSERVATION OF MASS AND MOMENTUM
The third partial derivative is the local rate of of uid mechanics, as developed using the spatial
change of velocity (7.61), which may be calculated description of motion, the operator (7.67) is used to
from (7.60). The fourth partial derivative evaluates calculate the rate of change of material properties
to one. that play roles in the fundamental principles of
Using the above interpretations for the partial conservation of mass and momentum. Apparently
derivatives in (7.64) and rearranging them, we the concepts embodied in (7.67) can be traced back
have an expression for the particle acceleration to publications of Euler (Fig. 7.16) in 1770 and
written in terms of the local velocity using a Lagrange (Fig. 7.14) in 1783 (Malvern, 1969).
spatial description of motion: The material time derivative operator (7.67)
may be applied to scalar, vector (7.65), or tensor
a |
v
v
v
t x x t| (7.65) functions that describe a property of the material
in terms of the spatial coordinates, x, and time, t
The rst term on the right-hand side is the local (Malvern, 1969). Suppose the mass density is
rate of change of velocity at the current position known as (x, t). The material time derivative
x. What distinguishes this term from the particle of this scalar quantity is:
acceleration at that position is the second term,
which is a product of the local velocity and the
spatial derivative of velocity at the current time t.
D
|
Dt t X
v
x t| (7.68)
The second term may be interpreted as the rate of The rst term on the right-hand side of (7.68)
change of velocity due to the ow of material at describes the local rate of change of density at the
velocity v through a spatially varying velocity current position x. The second term describes the
eld. For steady ow, such as that within the rate of change of density at the current time t due
rising viscous sphere (Fig. 5.12), the rst term on to the ow of material at velocity v with a spatially
the right-hand side of (7.65) is zero, so particle varying density eld. In terms of the velocity com-
accelerations are entirely due to the spatial varia- ponents (7.68) is:
tions in velocity as described by the second term. D
For example, the x-component of acceleration vx vy vz (7.69)
Dt t x y z
from (7.65) is written:
Here is it understood that the velocity compo-
v v v v nents are known functions of the current position
ax x vx x vy x vz x (7.66)
t x y z and time (7.60). Furthermore, the mass density is
The components ay and az follow by change of sub- a known function of the current position and
scripts. Here is it understood that the velocity time, so the partial derivatives are taken with the
components are known functions of the current appropriate independent variables held constant.
position and time, v g(x, t), so the partial deriva- Other scalar properties of the material are oper-
tives are taken with the appropriate independent ated upon and interpreted similarly.
variables held constant as indicated in (7.65).
The operation characterized in (7.65) for calcu- 7.3.2 Conservation of mass: the
lating the particle acceleration from the local equation of continuity
velocity may be generalized to calculate the What constraints must be imposed to assure that
material time derivative of any quantity associated the relative motions of particles in a deforming
with the material, given a spatial description of rock mass obey the fundamental law of mass con-
its kinematics: servation? To address this question we adopt the
spatial description of motion and consider a xed
D
|
Dt t X t x |
v
x t | (7.67) volume element within a deforming material con-
tinuum (Fig. 7.17). The center of the element is at
The operator D/Dt is also referred to as the substan- the arbitrary point specied by the position vector
tial derivative, but material time derivative is more x with components (x, y, z) which are the current
descriptive of its role. For example in the context coordinates. The sides of the element are parallel
7.3 THE DEFORMABLE CONTINUUM 267
rate of rate of rate of or line. The partial derivative of the x-component
mass mass mass (7.70) of the mass ux, vx, with respect to x is dened in
increase in out this limit as:
lim ( vx) |xx2 ( vx) |xx2
In other words, mass may enter and leave the ( v ) (7.73)
x x n x
element and the total mass of the element may
change with time, but mass is neither created nor Equating (7.71) and (7.72), dividing through by the
destroyed within the element. We assume that volume, and using (7.73) we have:
mass is not converted to energy in the processes
that generate geological structures (Wilczek, ( vx) (7.74)
t x
2004).
The rate of accumulation of mass in the Both the rate of change of density and the spatial
element, the left-hand side of (7.70), is measured derivative of the component of mass ux are
by the temporal derivative of density evaluated at evaluated at the current location x in this one-
x and multiplied by the volume, xyz: dimensional description of mass conservation.
Conservation of mass during motion in one
x y z (7.71) coordinate direction (7.74) is, perhaps, more inter-
t
esting than one might surmise. Because both
The right-hand side of (7.70) is accounted for using density and velocity may be functions of x, the
the mass ux per unit volume. This vector quan- derivative of their product is:
268 CONSERVATION OF MASS AND MOMENTUM
d vx
r = uniform
u ex ey ez
x y
z
(uxex uyey uzez)
u u y uz
x x (7.77)
x y z
dx
(b) r Note that this operation is similar to the scalar
product of two vectors: each partial derivative
dr operates on the respective component of the
vx = uniform
vector and the resulting sum is a scalar quantity.
Using the del operator, the rate of change of
x
x d x/2 x + d x/2 density may be written (Bird et al., 1960):
Fig 7.18 Graphs to illustrate the two terms on the right- ( v) (7.78)
hand side of (7.75) that account for the mass conservation in t
one-dimensional flow. (a) Rate of change of density as a This scalar equation is a spatial description of the
function of a spatial change in velocity with uniform density. conservation of mass because it describes
(b) Rate of change of density as a function of a spatial change
changes at a xed point in space and both the
in density with uniform velocity.
density and the velocity components are
expressed as functions of the spatial coordinates.
v The relationship in (7.78) is called the equation of
x vx (7.75)
t x x continuity.
In component form the continuity equation
The rst term in parentheses describes the rate of
(7.78) is written:
change of density if the density does not vary with
x, but the velocity does vary with x (Fig. 7.18a). For
example, where the velocity increases with x, the
t
x
( vx) ( vy) ( vz)
y z (7.79)
material in the vicinity of x is stretched, so the
density there decreases with time in proportion to Because both the density and the velocity com-
vx/x. The second term in parentheses describes ponents may be functions of the spatial coordi-
the rate of change of density if the velocity does nates, the partial derivatives in (7.79) expand as
not vary with x, but the density does vary with x follows:
(Fig. 7.18b). Where the density increases with x,
motion in the x-direction carries material with
lesser density into the vicinity of x, so the density
t
vx vy vz
x y z
vx vy vz
x y
z
there decreases with time in proportion to vx and (7.80)
/x. Density at a xed point in a material con-
tinuum can change by either one (or both) of Notice that the sum of the left-hand side and the terms
these two independent mechanisms while mass is in the second parentheses on the right-hand side are
conserved. the material time derivative of density as dened in
The one-dimensional relationship (7.74) is gen- (7.69). Therefore we can write the continuity equation
eralized for mass uxes through all six sides of the as (Malvern, 1969):
element using the differential operator, , (called
del) on the mass ux vector, v. When operating
on a vector quantity is dened as the vector (Bird
D
Dt
vx vy vz
x y z
v (7.81)
et al., 1960):
Here the operation v is called the divergence of
ex ey ez (7.76) the velocity vector eld which sometimes is
x y z
written div v. The product of the density and the
7.3 THE DEFORMABLE CONTINUUM 269
pressible material, D/Dt 0. For a body that is rate of rate of
homogeneous with respect to density and incom- momentum momentum
pressible: increase in
rate of resultant
0, and 0 (7.83) momentum of all (7.84)
x y z t
out forces
These are the most constrained conditions for the
material continuum subject to conservation of Each term in (7.84) is taken per unit volume. The
mass: the mass density is uniform in space and center of the element is at an arbitrary point
constant in time. specied by the position vector x with compo-
nents (x, y, z) which are the current coordinates.
The sides of the element are parallel to the coor-
7.3.3 Conservation of linear momentum: dinate axes, and the lengths of the sides are x, y,
the equations of motion and z. The momentum associated with the point
Conservation of linear momentum for the rigid at the center of the element is a vector function
body is described by (7.20). Here we derive the of the current location and time, v(x, t), with
270 CONSERVATION OF MASS AND MOMENTUM
components in all three coordinate directions. lim vx( vx) |xx2 vx( vx) |xx2
v ( v ) (7.87)
The rate of increase of the momentum is: x x x n x
Multiplying both sides of (7.87) by the volume and
( v) (7.85)
t comparing this to (7.86), we see how the negative
This vector accounts for the left-hand side of of the left-hand side of (7.87) accounts for one com-
(7.84). ponent of the rate of x-momentum change. The
As material moves through the xed element components of the rate of x-momentum change
(Fig. 7.19), momentum is carried in and out paral- through the front and back of the element, and
lel to the three coordinate directions in proportion through the bottom and top of the element, are
to the respective velocity components. This trans- similarly dened so the net rate of change of x
port of momentum is measured by the momentum momentum is:
ux per unit volume, which is the product of the v ( v )y z x vy( vx)x z y
velocity and the momentum, v( v). For example, x x x y
on the left-hand side of the element we have vx(vx), vz( vx)x y z (7.88)
where and vx are evaluated at x x/2. This quan- z
tity times the area, yz, is the momentum ux Here it is understood that the derivatives are eval-
through the left-hand side, vx(vx)yz. Similarly, uated at the point x. The net rate of change of y
the momentum ux through the right-hand side is and z momentum each have three components
vx(vx)yz, where and vx are evaluated at x x/2. that are found using a similar procedure.
We account for the difference between the rate of The one-dimensional relationship (7.88) for the
momentum in through the left side and the rate rate of change of momentum is generalized to
out through the right-hand side as: three dimensions using the differential operator
on the momentum ux, v( v). Because the
vx( vx) |xx2y z vx( vx) |xx2y z (7.86) momentum ux is a product of two vectors,
referred to as a dyadic product, this operation is
Because the velocity components vy and vz are par- somewhat different from that dened in (7.77).
allel to these sides, they cannot contribute to this The dyadic product is a special form of second-
part of the rate of momentum change. On the rank tensor with nine components. For example,
other hand, vy can carry x momentum, vx the dyadic product of the two arbitrary vectors, u
through the front and back sides of the element, and w is (Bird et al., 1960):
and vz can carry x momentum, vx through the
uxwx uxwy uxwz
bottom and top of the element. The pairs of
uw uywx uywy u y wz (7.89)
arrows normal to the sides of the element in Fig.
u z wx uzwy u z wz
7.19 are meant to represent these three uxes of x
momentum across the element. The order of the Using the del operator (7.76) on this dyadic
two velocity vector components is in keeping with product we have:
an onin subscript convention: e.g. vy(vz)xz is
the momentum ux on a side with normal par-
allel to the y-coordinate of the momentum in
uw
uxwx uywx uzwx
x
y
z
ex
the z-direction. uxwy u ywy uzwy
To convert the nite difference in (7.86) to a ey (7.90)
x y z
partial derivative a set of n elements (Fig. 7.19) is
considered with successively smaller volumes
that contain x and approach zero in the limit as n
uxwz uywz uzwz
x
y
z
ez
so the elements converge on the central point
at x. In this limit the partial derivative of the The rate of change of momentum per unit
momentum ux, vx(vx), with respect to x is: volume is found by dividing (7.88) and its coun-
7.3 THE DEFORMABLE CONTINUUM 271
surface force (7.94) is generalized to three dimen-
[v (v)] v ( v ) v ( v ) v ( v ) e sions using the differential operator (7.76) on
x x x y y x z z x x
the stress tensor, , following the procedure intro-
v ( v ) v ( v ) v ( v ) e
x x y y y y z z y y duced in (7.91):
v ( v ) v ( v ) v ( v ) e
x x z y y z z z z z
xx yx zx
x
y
z x
e
(7.91) xy yy zy
x
y
z y
e
This vector accounts for the rst and second terms
on the right-hand side of (7.84).
The nal term to evaluate in (7.84) is the
xz yx zz
x
y
z z
e
(7.95)
resultant force. The resultant surface force in This vector is the resultant surface force per unit
the x-coordinate direction is related to the x- volume. The body force per unit volume acting on
components of the tractions acting on the sides of the element is taken as:
the xed volume element (Fig. 7.20). It is conven-
g* (7.96)
tional to use the equivalent stress components
instead of the traction components to account for This vector accounts for the other part of the last
the surface forces. Thus, for example, the net term of (7.84).
surface force associated with the normal stress Collecting terms from (7.85), (7.91), (7.95), and
component, xx, on the left- and right-hand sides (7.96) we have a statement of the conservation of
of the element is: linear momentum (Bird et al., 1960, p. 78):
272 CONSERVATION OF MASS AND MOMENTUM
v vy v
(v) [v (v)] g* (7.97) v x x vx v x z
t x y z
y Surface, S
x
dS
n
Volume, V
x
t(n)
dV
rv(x, t)
rg*
Fig 7.22 Schematic diagram to define the conservation of
linear momentum for a fixed volume element based upon the
integration of the tractions, t(n), acting on surface elements,
S, over the surface, S, and the integration of the unit
Fig 7.21 Portrait of the French mathematician Augustine-
weights, g*, acting on volume elements V, over the
Louis Cauchy who was born in Paris, France, in 1789
volume, V.
(OConnor and Robertson, 2004). Cauchys First Law of
Motion is given in (7.104) through (7.106).
the surface integral is transformed to a volume tions, t(n), acting on the surface S. The cross
integral using the divergence theorem (Malvern, product of the position vector and the traction, x
1969, p. 200). Reynolds transport theorem is used t, is integrated over the surface as in (7.35) to
to move the material time derivative inside the nd the resultant torque due to surface forces. A
integral on the left-hand side (Malvern, 1969, distribution of body forces per unit volume is rep-
p. 210). Because the integrals refer to any arbitrary resented by the unit weight, g*, and the cross
volume the integrand on the left-hand side must product x g* is integrated over the volume as in
equal the sum of the integrands on the right at (7.36) to nd the resultant torque due to body
each point in the material continuum. The result- forces. Newtons Third Law is invoked to argue
ing equation is Cauchys First Law of Motion that the forces due to internal interactions are
(7.102). equal, opposite, and collinear so they produce no
resultant torque on the body.
7.3.4 Conservation of angular The underlying postulate is that the material
momentum: symmetry of the time derivative of the total angular momentum is
stress tensor equal to the vector sum of the resultant torques:
Conservation of angular momentum for the ma-
terial continuum is described using the momen- D
(x v) dV (x t) dS (x g*) dV
tum principle, given in the previous quotation, Dt
V S V
with the word angular inserted before momen- (7.108)
tum and the words all external torques substi-
tuted for all external forces. A similar statement, The material time derivative is used because the
applied to a single particle, was expressed in (7.11) angular momentum must be that associated with
and again in (7.30) for the rigid body. One might particles.
suppose, based on what we have derived in the pre- The next step in the derivation involves writing
vious section for conservation of linear momen- the vector cross products in terms of the vector
tum that the conservation of angular momentum components. Recall from (3.27) that the vector
in a deforming continuum would lead to a set of product u of two arbitrary vectors, v and w, is
equations relating density, velocity, stress, and dened in terms of their Cartesian components as:
gravitational acceleration in a form analogous to u v w (vywz vzwy)ex (vzwx vxwz)ey
Cauchys First Law of Motion (7.102). However, a
(vxwy vywx)ez (7.109)
derivation based on extending this concept to the
continuum (Malvern, 1969, p. 215) leads to a This expansion of (7.108) would involve a large
remarkably simple set of equations, relating only number of terms, but is condensed using indicial
the shear stress components to one another such notation and the permutation symbol epqr:
that the stress tensor is symmetric. The derivation
and its implications are described here. 0, when any two indices are equal;
In the context of a material continuum con- 1, when indices are (x, y, z),
sider a set of particles of given total mass occupy- epqr
( y, z, x), or (z, x, y);
ing a volume, V, with bounding surface, S, at a 1, when indices are (x, z, y),
given instant in time, t (Fig. 7.22). For the particle
(y, x, z), or (z, y, x).
at x, the momentum per unit volume, v, is a
(7.110)
function of the current coordinates and time, so
the angular momentum per unit volume at that Employing (7.110) the components of the vector
location, dened as x v, also is a function of x product u are written:
and t. This quantity is integrated over the volume
up epqrvqwr (7.111)
to obtain the total angular momentum. The body
is acted upon by a distribution of surface forces Here it is understood that the indices p, q, and r
per unit area represented by the equivalent trac- range over x, y, and z and that repeated indices on
7.3 THE DEFORMABLE CONTINUUM 275
the right-hand side imply summation over that referred to as the divergence theorem or Gausss
range. Each component of u is composed of nine theorem (Malvern, 1969, p. 200). Written in terms of
terms but eight of these are reduced to zero by the components of an arbitrary vector, u, this
(7.110) and the remaining terms are those found in theorem is:
(7.109).
The integral equation for the conservation of u
uini dS i dV (7.116)
angular momentum (7.108) is rewritten using the xi
S V
vector components and the permutation symbol On the left-hand side the component of the vector
as: u directed normal to the surface S is integrated
over that surface. On the right-hand side the diver-
D
e x v dV epqr xqtr dS epqr xqg*r dV gence of the vector u is integrated over the volume
Dt pqr q r
V S V V bounded by that surface. Applying the diver-
(7.112) gence theorem to (7.115) we have:
The left-hand side of (7.112) can be rearranged xqsr
epqr xqsrns dS epqr dV
using Reynolds transport theorem (Malvern, 1969, xs
S V
p. 210) which applies to any scalar, vector, or
tensor function. Given such a function, Q, of
the current coordinates and time, the material
epqr xq
V
sr
xs
sr
xq
xsdV
D DQ
Q dV dV (7.113) This transformation enables us to consider the
Dt Dt
V V right-hand side of (7.112) as a single volume inte-
gral:
Therefore the left-hand side of (7.112) may be
written:
sr
epqr xq g*r qr dV (7.118)
xs
D D
e x v dV epqr (xqvr) dV V
Dt pqr q r Dt The last term in square brackets follows from the
V V
fact that qssr qr.
V
Dv
epqr xq r vr
Dt
Dxq
Dt
dV Using (7.114) and (7.118) the rearranged and
transformed integral equation for the conserva-
tion of angular momentum (7.112) becomes:
Dvr
epqr xq dV (7.114)
Dt
Dvr
sr
V
In the second step Dxq/Dt vq by denition and epqr xq dV epqr xq g*r qr dV
Dt xs
V V
epqrvrvq 0 using (7.110).
(7.119)
In the rst term on the right-hand side of
(7.112) Cauchys Formula, tr srns, is used to
Buried within this equation is Cauchys First Law
replace the traction vector components with the
of Motion (7.103), such that the left-hand side of
stress tensor components, sr, where ns are the
(7.119) exactly cancels the terms in parentheses on
components of the outward unit normal to
the right-hand side leaving:
the surface S:
epqrqr dV 0
(7.120)
epqr xqtr dS epqr xqsr ns dS (7.115) V
7.4.1 Field equations for the linear acknowledged in applications of the linear theory
isotropic elastic solid and could prove to be problematic.
The Lagrangian (material) coordinates X are Further simplications of the equations of
taken as independent variables. Because there are motion (7.103) are achieved by considering the
only two states to compare, and it is the displace- material derivatives of the kinematic quantities.
ment that measures the change in position of any For example, components of the particle velocity
particle from the reference to the current state, are linearized as:
the equations of motion (7.103) must be written in
terms of the displacement u rather than the Dux ux u u u u
vx vx x vy x vz x x
velocity v. Because the equations of motion (7.103) Dt t x y z t
are written in terms of a spatial description of (7.125)
motion, conservation of momentum is satised
in the current conguration of the deforming Here products of the velocity components and the
body. To evaluate the conditions under which displacement gradients must be small compared
conservation of momentum would be satised in to the time rate of change of displacement.
the reference state, the stress tensor is written in Similarly, components of the particle acceleration
a form other than that introduced by Cauchy. are linearized as:
Consideration of this so-called PiolaKirchhoff
stress tensor is beyond the scope of this text so we Dvx vx v v v v 2u
ax vx x vy x vz x x 2x
refer the interested reader to other sources for Dt t x y z t t
the details of the evaluation (Fung, 1965; (7.126)
Malvern, 1969). In summary, equations of motion
of the same form as (7.103) are written for the ref- Here products of velocity components and the
erence state using the PiolaKirchhoff stress velocity gradients must be small compared to the
tensor. There equations may be transformed to time rate of change of velocity.
those using the Cauchy stress tensor, but terms Given the simplications and linearizations
appear that may be approximated, for example, described above, the equations of motion in the
as: reference state are written:
ux u u The right-hand side of (7.127) contains spatial
1 x x xx (7.124) gradients of the Cauchy stress components
x xx y yx z zx
which are eliminated in favor of the displace-
Here products of displacement gradients with ments by rst relating the displacement gradi-
respect to the spatial coordinates, xi, and stress ents to the strain components and then relating
components must be small compared to the the strains to the stresses. These steps involve
stress components. This approximation is rarely additional linearizations that are consistent
278 CONSERVATION OF MASS AND MOMENTUM
Eij
1 ui uj uk uk
2 Xj Xi Xi Xj (7.128)
2 2 2
ux 1 ux u y uz ux
Exx
X 2 X X X X
Exy
1 ux uy
2 Y X
1 ux ux uy uy uz uz
2 X Y X Y X Y
1 ux uy Fig 7.23 Disputed portrait of the English natural scientist
(7.129) Robert Hooke who was born in 1635 on the Isle of Wight,
2 Y X
England (see Phillip Ball, Nature, v. 433, p. 197, Jan. 2005).
Note that the non-linear terms are products of the The linear form of Hookes Law for the isotropic elastic
material is given in (7.131).
displacement gradients and these must be small
compared to the displacement gradients them-
selves. With approximations such as these the
innitesimal strain tensor, ij, is related to the dis- Newton. Chapter 8 is devoted to a discussion of the
placement gradients as: measurement of elastic material properties and
various forms of Hookes Law. The general linear
ij
1 ui uj
2 Xj Xi (7.130) form is simplied here for an isotropic material:
one in which the elastic constants are not depen-
A nite strain tensor also may be written using dent upon direction. For such a material the
the spatial description of motion and when a stressstrain relationships are:
similar linearization is carried out an equation of
the same form as (7.130) emerges in which the ij 2Gij kkij (7.131)
partial derivatives are taken with respect to the Here G is the elastic shear modulus and is Lams
Eulerian (spatial) coordinates, xi. Typically, the constant. These are properties of the material and
distinction between these two descriptions of the they have the same units and dimensions as
kinematics is overlooked in applications of linear stress. Expanding (7.131), typical normal and
elasticity. shear components of stress are related to the
For the linear elastic material the stress com- innitesimal strains as:
ponents are related to the innitesimal strain
components using constitutive equations called xx 2Gxx (xx yy zz)
Hookes Law after Robert Hooke (Fig. 7.23) the (7.132)
xy 2Gxy
English polymath and contemporary of Isaac
7.4 ELASTIC AND VISCOUS FIELD EQUATIONS 279
u ux uy uz Claude Louis Marie Henri Navier who was born in Dijon,
xx 2G x France, in 1785 (OConnor and Robertson, 2004). His name,
X X Y Z
(7.133) along with that of Stokes, is associated with the velocity
ux uy equations of motion for the viscous fluid (7.170).
xy G
Y X
2u i 2u i 2 u k 2 u x 2 u y 2uz
G (G ) g*i (7.134) g*x (7.135)
t2 XkXk XiXk X2 XY XZ
These are known as Naviers displacement equa- 2 u y 2u y 2u y 2u y
tions of motion (Fung, 1969, p. 261) after the 2
G (G )
t X2 Y2 Z2
French mathematician and scholar of engineer-
ing science Claude Louis Marie Henri Navier
(17851836), (Fig. 7.24). Apparently they were
2ux 2uy 2uz
XY Y2 YZ
g*y (7.136)
In component form Naviers equations of The independent variables are the three material
motion (7.134) are: coordinates and time (X, Y, Z, t) and the dependent
280 CONSERVATION OF MASS AND MOMENTUM
variables (unknowns) are the three displacement conserved. Typically, however, the structural geol-
components (ux, uy, uz). A solution would be three ogist is not confronted with data in the eld that
equations for the displacement components as directly constrain the particle velocity or acceler-
functions of the material coordinates and time. ation. Rather it is the displacement from some
The three equations, (7.135)(7.137), are solved for inferred initial conguration to the current
an elastic body of prescribed geometry subject to conguration, as in the opening of a dike or the
boundary conditions dened at every point on the slip on a fault. Therefore, the structural geologist
exterior and interior boundaries in terms of the generally approaches problems related to the
three displacement components. Derivatives of development of structures from a point of view in
the displacement components with respect to which the left-hand sides of (7.134) are set to zero
time provide the particle velocity as in (7.125) and and Naviers displacement equations of motion
the particle acceleration as in (7.126). Derivatives become:
of the displacement components with respect to
the material coordinates (7.130) provide the inni- 2 u i 2 uk
G (G ) g*i 0 (7.138)
tesimal strain components, and the isotropic XkXk XiXk
forms of Hookes Law (7.131) provide the stress
The motion of particles is described by the dis-
components. In this way all of the relevant physi-
placement from the reference state to the current
cal quantities are accounted for as functions of
state, while the details of the path followed, veloc-
the material coordinates and time, and the fun-
ities, and accelerations are ignored.
damental laws of conservation of mass and
The restrictions imposed to derive (7.138) put
momentum are obeyed.
the problem in the realm of quasi-static equilib-
The most familiar applications of solutions to
rium. We use the prex quasi because this is
the dynamic equations of elasticity (7.134) are to
not a problem of a static rigid body, but rather
bodies set in motion by sudden loading or unload-
one in which the body deforms and the relative
ing, for example from an explosion or the impact
displacement of particles is accounted for by the
of two bodies in motion, that generate waves
strain eld. In component form the quasi-static
within the elastic solid (Achenbach, 1973), or from
versions of Naviers displacement equations are
the rapid propagation of a fracture (Freund, 1979).
identical to (7.135)(7.137) with the left-hand
Well-known geological examples include the
sides set to zero. A solution to these three equa-
motion immediately following an explosive vol-
tions would be three equations for the displace-
canic eruption, the impact of a meteor (Melosh,
ment components as functions of the material
1989), or the rupture of a fault (Li, 1987; Kostrov
coordinates. The strains are computed from the
and Das, 1988; Scholz, 1990). In these cases rock
displacements using the kinematic equations
particles close to the impulsive event are set in
(7.130) and the stresses follow from Hookes Law
motion rst, while the rest of the body is unaf-
(7.131).
fected. Seismic waves propagate outward from the
For some problems in structural geology it is
source with speeds on the order of a few kilo-
more appropriate to formulate the elastic bound-
meters per second and set the rest of the rock
ary value problem in terms of the stress compo-
mass in motion (Aki and Richards, 1980).
nents. Taking the equations of motion as (7.127)
and supposing that the products of mass density
7.4.2 Quasi-static equilibrium for the and linearized accelerations are insignicant
linear isotropic elastic solid compared to the gradients in stress and the body
The equations of motion (7.134) for the isotropic
forces per unit volume, we have:
and isothermal linear elastic material place no
restrictions on the magnitudes of the velocity or ji
acceleration of any particle in the continuum. g*i 0 (7.139)
Xj
They describe a material that may be accelerating
or decelerating, but these changes are always These equations of quasi-static equilibrium are
related to the appropriate forces, so momentum is expanded in component form as:
7.4 ELASTIC AND VISCOUS FIELD EQUATIONS 281
2 2 0 (7.143) 1
Y2 X XY xx (yy zz)
E xx
2yy 2zz 2yz
2 2 0 (7.144) 1
Z2 Y YZ xy (7.150)
E xy
2zz 2xx 2
2
2 2 zx 0 (7.145) Refer to Chapter 8 for a discussion of the mea-
X Z ZX
surement of E and , and for equations relating
2xx
YZ X
yz zx xy
X
Y
Z
0
(7.146) the isotropic elastic constants, only two of which
are independent.
282 CONSERVATION OF MASS AND MOMENTUM
The derivation of the compatibility equations boundary conditions are dened as traction dis-
in terms of the stress components is given in tributions on the internal and external bound-
detail elsewhere (Malvern, 1969, p. 502). In short, aries of the body. Because of the inherent difculty
the stress components are substituted for the of solving the compatibility equations in addition
strain components using (7.149) and the equilib- to those of equilibrium the number of analytical
rium equations (7.139) are used to simplify the solutions for three-dimensional problems with
resulting equations and nd the so-called stress components as the dependent variable is
BeltramiMichell compatibility equations for the small. The more common approach is to take the
isothermal and isotropic linear elastic material. three displacement components as the dependent
In the most general form of these equations there variables and solve Naviers equations of motion
are partial derivatives of the body force per unit (7.138). On the other hand many approaches and
volume with respect to the material coordinates. solutions exist for two-dimensional problems,
To be consistent with our development thus far examples of which are presented in Chapter 8.
the body force is taken as mass density times grav-
itational acceleration, g*, which is postulated to
be uniform in space and constant in time. Under 7.4.3 Equations of motion for the linear
these conditions terms containing the body force isotropic viscous fluid
drop out of the BeltramiMichell compatibility Fluid ow is investigated by focusing attention on
equations which reduce to: the current state of the body and ignoring what-
ever might have been described as an initial or ref-
1 2
2ij 0 (7.151) erence state. Therefore the spatial description of
1 iXj
motion is adopted and we take the Eulerian
As mentioned above, these six equations repre- (spatial) coordinates and time (x, y, z, t) as the inde-
sent only three independent conditions. These pendent variables. This is one of several ways in
compatibility equations and the equilibrium which the analysis of uids differs from that of
equations (7.139) form a complete set for three- elastic solids, for which the referential descrip-
dimensional problems in elastic theory where the tion of motion is adopted. Instead of choosing the
six stress components are the dependent vari- displacement components of particles, the veloc-
ables. ity components at given positions are chosen as
Expanding typical members of (7.151) in com- dependent variables. Rather than the innitesi-
ponent form we have, for example: mal strains, the fundamental kinematic quanti-
ties used to describe the deformation are the
2
2
2 2
2 2 xx
X Y Z
1 2
( yy zz) 0
1 X2 xx
components of the rate of deformation tensor
(Malvern, 1969, p. 145):
2
2
2 2
2 2
X Y Z xy
1 2
( xy zz) 0
1 XY xx Dij
1 vi vj
2 xj xi (7.153)
(7.152)
Note both the similarity and the difference
Note that these relationships depend entirely between this quantity and the strain as dened in
upon second derivatives of stress components (7.130). The innitesimal strain, ij, is proportional
with respect to the material coordinates. to partial derivatives of the displacement compo-
Therefore functions for the stress components nents, ui, with respect to the material coordinates,
that are constant or linear in the material coordi- Xi, whereas the rate of deformation, Dij, is propor-
nates will automatically satisfy the compatibility tional to partial derivatives of the velocity compo-
conditions (Timoshenko and Goodier, 1970, nents, vi, with respect to the spatial coordinates,
Chapter 9). Such functions are solutions to the xi. Furthermore, the rate of deformation is not
three-dimensional elastic problem if they satisfy limited to small velocity gradients in the manner
the equilibrium conditions (7.139) and the pre- that the innitesimal strain is limited to small dis-
scribed boundary conditions. In these cases the placement gradients.
7.4 ELASTIC AND VISCOUS FIELD EQUATIONS 283
1
2 (1 3), 12 (2 3), 12 (1 2) (7.154)
thermodynamic pressure and a linear function of In other words the uid is incompressible. Another
the rate of deformation tensor. Any uid obeying possibility is that the bulk viscosity (7.161) is iden-
this general linear form or simplications of it is tically zero:
referred to as a Newtonian viscous uid because of
Newtons insightful investigations of viscous ow. 0,so 23 and p p (7.164)
The general linear form is simplied here for an
isotropic uid: one in which the viscous constants Apparently this was suggested by Stokes and so it
are not dependent upon direction. For such a uid is referred to as the Stokes condition. If either (7.164)
the stressrate of deformation relationships are: or (7.163) is satised the constitutive law is
reduced to one material constant, and coinciden-
ij pij 2 Dij Dkkij (7.160) tally from (7.161) the mean normal pressure is
Here and are the two material constants that equal to the thermodynamic pressure.
characterize the viscosity of the uid. Note both the Employing the Stokes condition the constitu-
similarity and the difference between this constitu- tive law for the linear and isotropic viscous uid
tive law and Hookes Law for the elastic solid (7.131). is:
The last two terms on the right-hand side are of the
same form, but the rate of deformation replaces the ij pij 2 Dij 23 Dkkij (7.165)
innitesimal strain and the constants have a differ-
In the absence of any gradients in velocity the
ent meaning. Here the additional term is that con-
normal stress components are equal to the nega-
taining the thermodynamic pressure, p. For the
tive of the thermodynamic pressure and the stress
uid at rest the normal stress components are equal
state is isotropic. In the analysis of geologic struc-
to the negative of the thermodynamic pressure.
tures it is commonly postulated that the rock
Two further simplications of the constitutive
mass is incompressible: in other words the mass
of the law for the Newtonian viscous uid (7.160)
density is constant. Using this constraint as
lead to a reduction of the number of material con-
described by (7.163) the constitutive law (7.165)
stants to one. Both of these follow from the rela-
becomes:
tionship between the mean normal pressure, p
and the thermodynamic pressure, p (Malvern, ij pij 2 Dij (7.166)
1969, p. 299):
Expanding (7.166) in component form, typical
2
p p Dkk p Dkk
3
(7.161)
stress components are:
initial value problems. It is these problems, and links between the concepts of calculus and
their solutions, that constitute the essential physics as taught in typical undergraduate
ingredients of model studies in structural courses and the concepts of continuum mechan-
geology. In this chapter we have provided the ics as required to practice structural geology.
Chapter 8
Elastic deformation
Experiment Theory
() ()
() ()
Comparison of displacement field components near an edge The conceptual success of the [innitesimal theory of
dislocation from phase images of experiments (left column) elasticity and the linear theory of viscosity] is perhaps
and from anisotropic elastic theory (right column). (a) and the broadest we know in science: in terms of them we
(b) displacement component ux parallel to bottom edge of face, explain, and in varying amount control, our
image. (c) and (d) displacement component uy parallel to left daily environment: winds and tides, earthquakes and
edge of image. Photograph reproduced from (Hytch et al., sounds, structures and mechanisms, sailing and ying,
2003) with kind permission of Martin J. Hytch. heat and light (Truesdell and Noll, 1965).
288 ELASTIC DEFORMATION
I
n this chapter we describe how the elastic prop- this theory has made to structural geology.
erties of rock are measured in the laboratory Next, the measurement techniques used to
and provide tables of numbers representing the determine elastic properties in the laboratory and
range of values for different rock types. However, at engineering eld sites are described. This leads
the need to understand and measure the resis- to a discussion of how elastic properties vary with
tance to deformation of rocks goes well beyond the size of the rock mass being tested. Con-
the simple accumulation of numbers in hand- siderations of scale effects are closely linked to
books of rock properties. To paraphrase Truesdell those of heterogeneity with respect to elastic
and Noll (1965), the aim of structural geology is to properties. To build intuition a solution to a
construct mathematical models that enable us, boundary value problem is examined for a circu-
from use of knowledge gathered in a few observa- lar inclusion with different elastic stiffness and
tions, to predict by logical processes the outcomes compressibility than the surrounding material.
in many other circumstances. To analyze a geo- Finally, we consider how the elastic properties of
logic structure one must choose the appropriate rock may vary with orientation, in other words
boundary or initial value problem to serve as a how anisotropic is rock with respect to elastic
mechanical model. To formulate such a problem properties? After reviewing Hookes Law for
one must postulate a particular mechanical anisotropic materials in general, and providing
behavior. That is, one must say exactly what the some representative values of the elastic moduli
relationship is between the stress acting within a from laboratory measurements, we describe a
material and some measure of the deformation, solution to a boundary value problem for an
usually strain or rate of deformation. These orthotropic elastic material. While it is well
relationships are called constitutive equations. For known that rock masses at the scale of Earths
example, researchers studying the displacement crust can be heterogeneous and anisotropic to
eld around an edge dislocation in silicon as some degree, it is noteworthy how well isotropic
revealed by electron microscopy (Chapter 8, fron- and homogeneous elastic models correspond to
tispiece) postulated an anisotropic linear elastic measured deformation. This is illustrated using
constitutive law and calculated model displace- data on surface displacements during the 1999
ments that are remarkably similar to those Hector Mine earthquake.
observed.
We begin this chapter by describing the defor-
mation of the Mancos Shale associated with the
emplacement of a basaltic dike near Ship Rock, 8.1 Estimating rock properties
New Mexico, about 30 million years ago. In this
example the shale was compressed as magma
from geological field tests
pressure forced the dike open, so the thickness of
the dike gives us a measure of the resistance to To determine how rock masses resist deformation
deformation of the surrounding rock. Next we one must conceive an experiment and build or
introduce the formal concept of linear elasticity identify a testing apparatus that controls the
as rst envisioned by Robert Hooke in 1676, along applied loads and facilitates measurement of the
with the constitutive equations that connect resulting deformation. In the context of elastic
stress to strain for idealized elastic materials. solids the experiment is designed to determine a
These constitutive equations are central to the quantitative relationship between the stress and
innitesimal theory of an elastic continuum. As the strain. Here we focus on such experiments,
Truesdell and Noll (1965) suggest in the opening conducted in the eld rather than in a laboratory,
quotation for this chapter, elastic theory has because this enables one to estimate the elastic
played an enormous role in the development of properties of rock at length scales of kilometers
science and engineering. Some examples are which are relevant to the geological structures
reviewed to illustrate the type of contributions under consideration.
8.1 ESTIMATING ROCK PROPERTIES FROM FIELD TESTS 289
New Mexico
emplacement, so the top of the Ship Rock edice
(composed of Tertiary tuff breccia, Ttb) was prob-
ably within a few hundred meters of the old land Kmu
surface at the time lava erupted from the central
conduit of this volcano. Because the exposed dikes
range up to 10 km in outcrop length it seems likely
that some of them also erupted to feed lava ows. 0 1 2 3 4 km
Today approximately 1 km of sedimentary rock
lies below the ground surface and overlie a crys- 108o 50'
talline basement of Precambrian age. The limb Fig 8.1 (a) Oblique aerial photograph of Ship Rock, NM
dips on a gentle syncline that cuts across the (photograph by D. L. Baars). Height of edifice is about 600 m
northeastern dike are less than 5 and these dips and length of the northeastern dike in the foreground is
are typical for the exposed sedimentary strata about 3 km. (b) Map of region near Ship Rock: Kmu,
over much of the area. Cretaceous Mancos Shale; Tmn, Tertiary minette; Ttb,
The basic idea behind our use of dikes as natural Tertiary tuff breccia (Delaney and Pollard, 1981).
testing machines is that magma pressure in dikes
pushed against and elastically compressed the sur-
rounding sedimentary rock mass. Our objective is Mancos Shale as four springs that resist the com-
to estimate the resistance of the sedimentary rock pression (Fig. 8.2a). The mechanical relationship
to this loading. For the sake of this example we ide- between applied force, F, and the relative displace-
alize the dike walls as two rigid plates and the ment, u, of the two ends of an elastic spring is:
290 ELASTIC DEFORMATION
Fig 8.2 Spring and block model for dike. (a) Before dike T
emplacement compression, C, acts across the prospective (P C)WL K (8.2)
2
dike plane; K/2 is the spring constant. (b) Pressure, P,
compresses springs and model dike opens with displacement, The equivalent relationship between stress and
u, in both directions.
strain is found by rearranging this equation:
F Ku (8.1) (P C)
K T2
W L
E
T2
L (8.3)
The spring constant, K, measures the stiffness, or Here the left-hand side is the applied stress and
resistance, to deformation. The springs are con- the right-hand side is a constant, E, times the
nected to two outer rigid plates, initially loaded by resultant strain associated with opening of the
a compressive stress of magnitude, C, represent- idealized dike. Recall that normal strain is a
ing the tectonic loading at the time of dike change in length divided by the original length of
emplacement. This compressive stress is transmit- a line element. Here the change in the original
ted through the springs (which have shortened spring length, L, is given by T/2, so the term in
accordingly), thereby holding the two inner plates parentheses on the right-hand side is the normal
tightly pressed together. strain. The constant, E, is called Youngs modulus
The plates have a length, L, the outcrop length of elasticity. It measures the resistance of a ma-
of the dike, and a width (out of the plane of view) terial to change in length (strain) under an
given by W, so their area is WL. Before dike applied normal stress.
emplacement, the distance, L, separates the inner The northeastern dike at Ship Rock has an
and outer plates. At this point in our thought outcrop length L 2900 m and an average thick-
experiment the two outer plates are xed in ness T 2.3 m. Using (8.3) the ratio of driving pres-
place and thereafter are not allowed to move, so sure to Youngs modulus is estimated as:
we can measure the relative displacement of the
spring ends as the plates are pushed apart. We (P C) T
0.0004 (8.4)
refer to the distance between the inner plates as E 2L
8.1 ESTIMATING ROCK PROPERTIES FROM FIELD TESTS 291
By measuring dike length and thickness, and Table 8.1. Geologic eld tests using Ship Rock
nding independent estimates for the magma dikes.
pressure and compressive stress, we can use (8.4)
to estimate Youngs modulus for the rock mass Dike name Host rock E (GPa)
deformed by the dike.
The geological evidence suggests that the Northeastern Mancos Shale 5
current outcrop of the northeastern dike was Southern Mancos Shale 4
about 1 km below the ground surface at the time
Small Mancos Shale 12
of dike intrusion, and we assume that the dike
erupted at that surface. We take m 2600 kg m3 multiple dike segments yields a Youngs modulus
and r 2400 kg m3 as the magma and host rock of about 2 GPa, not very different from the single
densities, g* 9.8 m s2 as the acceleration of fracture estimate (Table 8.1). If the dike did not
gravity, and D 103 m as the height of the magma propagate to the surface, the pressure would not
column and thickness of overburden. The magma necessarily be the hydrostatic value used above.
pressure and compressive stress in the Mancos Furthermore, the dike is unlikely to be two dimen-
Shale at the depth of the current outcrop are cal- sional and the effect of the three-dimensional
culated as P mgD 25.5 MPa and C r gD 23.5 form on opening should be addressed using solu-
MPa. Therefore, the driving pressure was (P C) tions to a three-dimensional boundary value
2 MPa and Youngs modulus is estimated from problem of elasticity.
(8.4) as E 5 GPa. The southern dike has a length
of about 9 km and an average thickness of about 8.1.2 A generalized geological field
10 m, and the small dike just to the south of the method for estimating rock
northeastern dike has a length of about 1 km and properties
an average thickness of about 0.5 m. These data Geological eld methods for estimating rock
provide additional estimates of Youngs modulus properties are useful for two reasons. Most
(see Table 8.1). We do not suggest that these obvious is the fact that many geological struc-
values of Youngs modulus necessarily are repre- tures are too large to submit to laboratory study;
sentative of other large rock masses, and they cer- their size puts them beyond the capability of
tainly are not known with the same precision human engineering in terms of their length scale.
expected for laboratory measurements. On the Furthermore, if time is an important variable,
other hand, we believe that the method described then laboratory tests are incapable of duplicating
here is important and should be applied to other geological time scales. Note that we are not refer-
igneous dikes. ring here to scaled model experiments that seek
The northeastern dike at Ship Rock is not con- to simulate geological processes (see Chapter 4),
tinuous along the outcrop, but is divided into 35 but rather to the measurement of the physical
echelon segments separated by Mancos Shale properties of rock. Second, the physical and chem-
(Delaney and Pollard, 1981). Also, the thickness of ical conditions under which many geological
the dike is partly attributable to erosion of the structures have formed is unknown, and even if
dike wall by the owing magma. Apparently these conditions could be deciphered, they may
the hot magma caused thermal fracturing of the not be reproducible in the laboratory. Thus we are
Mancos Shale at the contact and this fractured faced with a considerable challenge to determine
rock was locally removed by the magma. Clearly the properties of rock masses at depth in the
such a process is not included in the elastic model Earth at geological length and time scales.
of fracture dilation, so thickness measurements One approach to this problem is to use tests
should be corrected accordingly. Furthermore, under natural conditions and at natural length
one might want to make a small correction for and time scales. That is, let nature do the experi-
shrinkage of the igneous rock as it cools from ment and look for an appropriate way to interpret
magmatic temperatures. Using a numerical solu- what has been done. The method is summarized
tion to the elastic boundary value problem for in six steps as follows:
292 ELASTIC DEFORMATION
1. Identify a natural experiment run under the gists rock hammer hits an exposure. If the expo-
conditions and at the scale of interest. sure is fresh granite, the hammer springs back
2. Use mapping techniques in the eld to charac- quickly with a high-pitched ringing and vibrates
terize the structures, identify the lithologies, vigorously for a second or two. The rock also gives
and measure the relevant geometric parame- off a sharp audible report. Perhaps a few small
ters. chips of rock or metal shoot out from the point of
3. Infer the sequence of deformation and appro- impact if the blow is particularly aggressive, but
priate boundary conditions for loading and lighter blows permanently deform neither the
displacement (or other relevant physical quan- hammer nor the rock. Both rock and hammer
tities) from the eld data. return nearly to their shapes just before the
4. Set up and solve, or borrow from the literature, impact. If you held your hand on the exposure
the appropriate mechanical problem (usually a near the point of the hammer blow you would feel
boundary or initial value problem from con- vibrations in the granite, a result of waves propa-
tinuum mechanics). gating out from the point of impact. Other types
5. Derive from the solution to this problem an of rock respond in a similar fashion, but with
equation for the physical property of interest in some quantitative differences. For example, sand-
terms of the measured and/or inferred quanti- stone might respond with a duller sound and the
ties. hammer will not seem to jump back as quickly.
6. Use the derived equation and available data to When materials return essentially to their origi-
estimate the physical property. nal shape after the applied loading is removed, we
say they are elastic.
This method is not limited to rocks with elastic A linear relationship between force and exten-
properties, but is generally applicable to any ma- sion is credited to the English natural philosopher
terial behavior described by a well-dened consti- Robert Hooke (16351703) who published a state-
tutive law. Nor is it limited to dikes, but is ment on the subject in 1676 in the following
generally applicable to any geological structure remarkable form (Gordon, 1976):
that can be modeled using continuum mechanics.
ceiiinosssttuu
da
E(tangent) f (ea) (8.7)
Axial stress
sa = f/A
linear dea
elastic
non-linear
loading elastic The local slope is the rst derivative of the axial
stress with respect to the axial extension and is a
sa function of the extension.
unloading
r e Because extension is dimensionless, Youngs
o
m tiff ea
compression tension
tic
las
~e
loading
steel (spring), E 200 GPa
copper, E 110 GPa
tic
elastic
ine
x
s
y
and yy xx zz xx so yy zz (E)xx. s
t
Consider successively applying a normal stress in
each coordinate direction and accounting for all
x
of the possible normal strains. A consequence of
the assumption of small strains and rotations is
that the strain state is independent of the order in
which the stresses are applied to the body.
Therefore, these strain states can be superimposed
Fig 8.6 Element used to define relation between shear
and we have: stress and shear strain for linear elastic material
1 (Timoshenko and Goodier, 1970). (a) Tension and
xx (yy zz)
E xx
compression of equal magnitudes. (b) Shear stress on
element oriented at 45 to that in (a).
yy
1
(zz xx)
E yy (8.12)
on a diagonal boundary with outward unit
zz
1
(xx yy)
E zz normal, n, oriented at 45 are:
yy xx. In other words the extension in the The stress components appear as dependent
x-direction is equal to the contraction in the variables in (8.19) to be calculated from the strain
y-direction. The element inclined at 45 experi- components (independent variables) and elastic
ences only a shear strain. This follows from the moduli. In contrast, the strain components appear
fact that the strains transform just as the stress as dependent variables in (8.18) to be calculated
components (Fung, 1969): from the stress components (independent vari-
ables) and elastic moduli. Some have suggested that
xx xx cos 2 45o xx sin 2 45o 0 the possibility of treating either stress or strain as
the dependent variables implies that there is no
xy (yy xx) sin 45o cos 45o xx (8.14)
cause and effect relationship between stress and
The nal step is to relate the shear stress and shear strain (Marrett and Peacock, 1999). This notion is
strain on the inclined element. From (8.12) where contrary to the concept of Newtonian mechanics in
yy xx and zz 0, we have: which force is described as the causative agent and
acceleration is the resulting effect.
xx
1
yy
E xx
1
E xx (8.15) The natural extension of Newtons concept to
the elastic continuum is to view stress or traction
But we have just learned, from resolving stresses as the causative agent and strain or displacement
and strains on the inclined element, that xx as the resulting effect. Thus (8.18) and (8.19) simply
xy and xx xy, so: show how stress components can be calculated
from strain components, or vice versa, and
1 nothing more profound is implied concerning the
xy xy (8.16) physical framework of Newtonian cause and effect.
E
The constitutive equations for the linear and
Thus, the shear strain is proportional to the shear isotropic elastic solid are used extensively in the
stress acting on the inclined element and the con- analysis of geologic structures, often with little
stant of proportionality is (1 )/E. experimental justication. Ideally, the rocks
The relationship between shear stress and would be sampled in the eld area, brought back
innitesimal shear strain (8.16) can be generalized to the laboratory, and tested to reveal their
for all of the components in three dimensions to mechanical properties. Usually, the facilities and
give: funding for such testing is not available. Even if
1 1 1 rock samples were tested rigorously, would the
xy xy,yz yz,zx zx
E E E measured properties correspond to those millions
(8.17) of years ago when the geologic structures under
Equations (8.12) and (8.17) are a statement of investigation actually formed? Identifying those
Hookes Law for the isotropic elastic material. conditions is a challenging problem for structural
These six equations are written using indicial geologists. More often than not, justication for
notation as: using the elastic model comes a posteriori, after
1 the solution to the boundary value problem pro-
ij ij kkij (8.18)
E E vides a compelling correlation to eld observa-
Given the innitesimal strain components, (8.18) tions. Examples of such correlations are found
can be rearranged as: throughout this book and in the cited literature.
ij
E
1
ij
E
(1 )(1 2) kk ij
(8.19) 8.2.3 Relations among elastic moduli for
isotropic materials
It is clear from (8.18) that when all the stress com- To characterize further the isotropic and linear
ponents are identically zero, all the strain com- elastic material we derive the relation between
ponents are zero. This describes the initial, volumetric strain and pressure. The innitesimal
unloaded state of the elastic body. volumetric strain is given by the change in volume
298 ELASTIC DEFORMATION
of an element divided by the original volume, The sum of the normal stress components also is
V/V. To understand how this quantity is related to invariant for any rotation of the coordinate
the innitesimal strain components consider a system. For an isotropic state of compressive stress
sphere of radius L in the initial state that deforms the negative of the uniform normal stress is the
to an ellipsoid with half axial lengths L1, L2, and L3 static pressure, po xx yy zz. Some use
in the current state. Recall that the normal the phrase hydrostatic pressure or hydrostatic
innitesimal strain of a line element is equivalent compression for this quantity, but this should be
to the extension of that line element. Thus, the avoided, as the subject here is the deformation of
normal strain of the line that becomes the major elastic solids, not of water. Because the volumetric
axis of the ellipsoid is 1 (L1 L)/L, so L1 (1 1) L. strain (8.22) is the sum of the three normal strain
The other axial lengths are similarly related to the components, we add these as dened in (8.12) and
principal strains. The volume of the sphere is Vs use (8.23) with m p0 to nd:
4 3 4
3 L and that of the ellipsoid is Ve 3 (L1L2L3)
1 2 3(1 2) 1
which may be approximated as: kk kk m p0 (8.24)
E E K
Ve 43 (1 1)(1 2)(1 3)L3 Here K is called the bulk modulus, which relates the
43 (1 1 2 3)L3 (8.20) innitesimal volumetric strain to the pressure for
an isotropic state of stress.
The approximation is obtained by neglecting
Considering Youngs modulus and Poissons
products of the principal strains because they are
ratio as the two independent moduli of the
very small compared to one or compared to the
isotropic elastic material, and using (8.24), the
strains themselves.
bulk modulus is written:
It is one of the basic attributes of the innites-
imal strain tensor that the sum of the normal E
K (8.25)
strains is invariant for any rotation of the orthog- 3(1 2)
onal coordinate axes, so we can write the volume
Note that the bulk modulus, K, approaches an
of the ellipsoid:
innite value as Poissons ratio approaches 0.5.
Ve 43 (1 xx yy zz)L3 (8.21) This is consistent with the characterization of
such materials as being incompressible. Most
Substituting for the volumes of the ellipsoid and
liquids are nearly incompressible, whereas gases
sphere in the equation for the volumetric strain,
are highly compressible. Very porous rocks are
we have:
somewhat compressible, whereas rocks with low
V Ve Vs 43 (1 xx yy zz) 43 porosity tend to be less compressible. For the per-
V Vs 4
3
fectly compressible material, 0, so the bulk
xx yy zz kk (8.22) modulus is K E/3.
The elastic shear modulus, G, is used to relate
Note that the last term in (8.19) is proportional to
shear stress to shear strain, as in (8.17), from
the volumetric strain. Thus, each normal stress
which we have:
component is related to the corresponding
normal strain component and the volumetric E
G (8.26)
strain. Also, for the perfectly compressible mater- 2(1 )
ial, 0, each normal stress is simply propor-
The factor of one-half appears because we are
tional to the corresponding normal strain,
using the tensor convention to dene shear strain.
because the last term in (8.19) is zero.
This shear strain is one-half the magnitude of the
Next we relate the volumetric strain to the
so-called engineering shear strain and the shear
mean normal stress, m, dened as the average of
modulus was originally dened using the engi-
the three normal stress components:
neering convention (Fung, 1969). For perfectly
compressible material, 0, so G E/2, and for
m 13 (xx yy zz) 13 kk (8.23)
incompressible material, 1/2, so G E/3.
8.3 QUASI-STATIC DISPLACEMENT BOUNDARY VALUE PROBLEMS 299
Gabriel Lam (17951870) introduced another Sokolnikoff, 1956; Timoshenko and Goodier,
elastic constant, , in 1852 (Fung, 1969) and this is 1970; Barber, 1992).
related to Youngs modulus and Poissons ratio as: The mathematical structure of the theory of
elasticity makes it possible to solve boundary
Ev
(8.27) value problems of elasticity using displacement,
(1 )(1 2)
traction, or mixed boundary conditions. This fact
Poissons ratio is related to Lams constant as: led to a debate about whether boundary displace-
ments cause the elastic solid to deform (Marrett
(8.28) and Peacock, 1999; Tikoff and Wojtal, 1999;
2(G )
Peacock and Marrett, 2000; Pollard, 2000). In the
The constant characterizes the spectrum of authors opinion the opportunity to employ dis-
behaviors from perfectly compressible materials, placement boundary conditions is no more than
0, for which 0, to incompressible materials, a consequence of the underlying mathematical
1/2, for which . Taken together, and G relationships among traction and displacement
sometimes are called Lams constants. Equations components. In a Newtonian context one would
(8.19) may be written in terms of Lams constants seek an explanation for the specied displace-
using (8.26) and (8.27) such that: ments in terms of forces applied exterior to the
model boundary, and these forces would be the
ij 2Gij kkij (8.29)
ultimate causative agents for the resulting defor-
The relatively simple behavior of the isotropic and mation in the interior. On the other hand, where
linear elastic solid makes it ideally suited for tractions are the prescribed boundary conditions
analysis in continuum mechanical models. One one may refer to the associated forces as the
can solve problems that involve quite complex causative agents for the deformation.
geometries and boundary conditions by restrict-
ing the material behavior in this way, and the 8.3.1 Two-dimensional plane strain
solutions provide important insights about the solutions for cylindrical structures
origins of some geologic structures. All structures in the Earth are three dimensional,
but there are circumstances in which it is appro-
priate to ignore some of the components of stress,
8.3 Quasi-static displacement strain, and displacement. This has practical impli-
cations because the mathematical complexity of
boundary value problems the boundary value problem is greatly reduced.
One circumstance involves structures that are
One may formulate a problem in elasticity very long in one dimension relative to their size in
theory in terms of the displacement compo- the other two dimensions. If the geometry of such
nents, or in terms of the stress components, as a structure does not change signicantly along its
the dependent variables. In this section we take length, it may be described as a cylindrical structure.
the equations of motion written in terms of the A common geological example would be the
displacement components as the dependent surface of a sedimentary layer, folded into a shape
variables. A specic example is provided using that may be approximated by moving the fold axis
the two-dimensional solution for an edge dislo- through space without changing its orientation
cation which has surprisingly broad applica- (Fig. 8.7a). Other common examples include the
tions in structural geology, ranging from surfaces of blade-like dikes in volcanic rift zones
micrometer-scale defects in mineral grains to and of vertical joints conned between two hori-
plate-bounding faults at continental margins zontal sedimentary layers (Fig. 8.7b).
(Weertman and Weertman, 1964; Weertman, The special case of deformation that applies to
1996). Textbooks on elasticity theory provide two-dimensional cylindrical structures is called
many other useful solutions for displacement plane strain. Here the (x, y)-plane is taken as the
boundary value problems (Muskhelishvili, 1954; plane of interest (Fig. 8.7), and we postulate that
300 ELASTIC DEFORMATION
ux f1(x, y), uy f2(x, y), uz 0 (8.30) The out-of-plane shear stress components are
zero: yz 0 and zx 0.
These conditions demand that neither the geom-
etry nor the loading conditions change along the 8.3.2 A complete solution in two
z-axis. Based upon these displacement conditions, dimensions: the edge dislocation
and ignoring body forces, the quasi-static form of
Naviers displacement equations of motion (7.138) The dislocation is an object worthy of study. Its exis-
reduce to: tence permits metals to be plastically deformed with
ease, a circumstance upon which our modern
2ux 2u 2 u y
(2G ) 2
G 2x (G ) 0 (8.31) technology is so dependent. . . . The dislocation also
x y xy
permits nonmetallic crystalline materials to be
2uy 2uy 2ux
G 2 (2G ) 2 (G ) 0 (8.32) plastically deformed. . . . Thus the dislocation plays
x y xy a commanding role in those grandest of all
These equations govern the spatial distribution of deformations on earth: the upheavals that have
the two displacements in the (x, y)-plane. Note that produced the mountain ranges and the continents
here and in what follows we have replaced the themselves (Weertman and Weertman, 1964, p. 1) .
8.3 QUASI-STATIC DISPLACEMENT BOUNDARY VALUE PROBLEMS 301
(a) ux
b
2
tan 1
y
x
G
2G
xy
x2 y2
(8.36)
r = 5b
Dislocation
Elastic
region
b
uy
2
G
2(2G )
ln
x2 y2
C
core
Glide x
G
2G
y2
x2 y2
(8.37)
0.5 tion line (Fig. 8.10b, ux term 1). Here the radius is
0.4
ux term 1 taken as r 10b, outside the dislocation core. Note
0.3
how the contribution from this term to the nor-
0.2 uy term 1
malized displacement, ux/b, is zero at 0 and
0.1
ux term 2 decreases linearly with to a minimum of 0.5 at
0
0.1
, the position of the glide plane. Then, the
0.2 uy term 2 normalized displacement jumps discontinuously
0.3 to 0.5 across the glide plane and thereafter
0.4 decreases linearly to zero at 2. The relative
0.5 displacement, or displacement discontinuity,
across the glide plane (x 0, y 0) is equal to the
50 100 150 200 250 300 350
magnitude of the Burgers vector, in this case a unit
(o)
value. The other three terms in the displacement
Fig 8.10 (a) Schematic illustration of edge dislocation and equations are necessary to satisfy the governing
dislocation core inside of which the deformation is inelastic. equilibrium equations, but do not contribute to
(b) Plot of terms in the displacement equations, (8.36) and the displacement discontinuity.
(8.37), versus angle defined in (a). The kinematic equations (8.33) are used to
determine the strain components:
The elastic solution for the edge dislocation is bx 2(G )(x2 y2)
premised on the conditions that the deformation xy (8.40)
2 (2G )(x2 y2)2
may be taken as quasi-static and plain strain. The
solution to the Naviers displacement equations of The denominators for each of the strain compo-
motion, (8.31) and (8.32), is (Weertman and nents is proportional to r 4 where r (x2 y2)1/2 is
Weertman, 1964, p. 36): the radial distance from the dislocation in the
304 ELASTIC DEFORMATION
y/b
0, ij ). This is mathematically correct, but is 0
non-physical, so we restrict attention to the region
500
outside the dislocation core (r 5b), where the
strain components are nite.
Hookes Law for plane strain conditions (8.34) 1000
bx 2G(G )(x2 y2) 400
xy (8.43) 200
2 (2G )(x2 y2)2
0 0
y/b
within the crust so these pairings of edge disloca- 124o W 120o W 116o W
tions become models for thrust and normal faults
as well. 42o N
0 200 km
8.3.3 A compelling example of elastic 40o N
deformation: the Hector Mine
earthquake
To provide a specic example at the kilometer- 38o N
scale and to motivate the further application of
linear elastic models to structural problems we 36o N
describe the surface displacement eld for the
magnitude 7.1 Hector Mine earthquake that 34o N
occurred on October 16, 1999, in southern
California (Fig. 8.13). This earthquake is associated
with the earlier Landers earthquake (Sieh et al.,
1993), a magnitude 7.5 event that occurred on
June 28, 1992. Both earthquakes occurred in a A B
desert region to the northeast of the San Andreas
Fault. The desert terrain provided an excellent
opportunity to observe and measure the rupture
traces at the surface and to map the details of the northeast br
shear zones and the crustal-scale deformation
La
vic
E
st We
Sandwell, 1998).
llio Bu
M
D'
n llio
es
ul Fa
ite
ul
E'
ul
1993). The radar signal is transmitted from a satel- Mapped surface rupture
t
A'
lite (in this case ERS-1 from an altitude of 785 km) Surface trace of model
to the ground surface where it is reected back to
the satellite and recorded as a set of pixels Fig 8.13 Location map for the Hector Mine earthquake,
making up an image of the surface, each pixel showing California with box indicating region of earthquake
representing an area of about 100 m2 on the rupture, expanded in lower part of figure. Reprinted from
Maerten et al. (2005) with permission of the Seismological
ground. Knowledge of the travel time and speed
Society of America.
of the signal provides the information necessary
8.3 QUASI-STATIC DISPLACEMENT BOUNDARY VALUE PROBLEMS 307
2
2
y x
2
2 (xx yy) 0 (8.54) Note that the shear stress is zero and the con-
stants play no role in the functions for the stress
The compatibility equation for plane strain and
components. A particular region in the (x, y)-plane
constant body forces reduces to this harmonic
(shaded rectangle in Fig. 8.17a) is illustrated with
equation for the sum of the two in-plane normal
the traction boundary conditions (8.45) corre-
stress components.
sponding to this solution. The upper surface is
Solutions for the three in-plane-stress compo-
traction free and therefore could represent
nents must satisfy equilibrium (8.46)(8.47) and
Earths surface. If the material is incompressible,
compatibility (8.54). A method for solving these
0.5, this Airy stress function gives the isotropic
equations was proposed by G. B. Airy in 1862
state of stress equivalent to Andersons standard
(Timoshenko and Goodier, 1970, p. 32) and the
state: xx yy zz g*y. For values of Poissons
function used in this method, (x, y), commonly is
ratio less than 0.5, the out-of-plane-normal stress,
referred to as the Airy stress function. The stress
zz 2g*y, is somewhat less than the in-plane-
components are written in terms of this stress
normal stresses. For the perfectly compressible
function as:
material, 0, the out-of-plane-normal stress is
2 2 zero.
xx g *y,yy 2 g *y,
y 2 x The region of interest in the (x, y)-plane is arbi-
2 (8.55) trarily chosen, but the mathematical solution for
xy the state of stress applies everywhere in the plane.
xy
This can lead to states of stress that satisfy the gov-
Direct substitution demonstrates that these func- erning equation (8.56), but are not appropriate for
tions satisfy the equilibrium equations, (8.46) the physical problem under consideration. For
(8.47), and substitution into the compatibility example, if the region of interest in Fig. 8.17a were
equation (8.54) yields the biharmonic partial dif- to be extended upward to positive values of y, the
ferential equation: resulting normal stresses would be tensile (posi-
4 4 4 tive), where y is positive. This would not represent
2 2 2 4 0 (8.56)
x 4 x y y stresses induced by the weight of the body and
8.4 QUASI-STATIC TRACTION BOUNDARY VALUE PROBLEMS 311
(a) y L
y
tx = 0, ty = 0 x
tx = 0, ty = 0 x
tx = rg*y tx = rg*y
ty = 0 H ty = 0
tx = rg*y H tx = CL + rg*y
ty = Cy ty = Cy
tx = 0, ty = rg*H
thrust faulting in fold and thrust mountain belts Note that the strains scale with the same con-
(Hafner, 1951). The in-plane stress components are stant, C, that scales the stresses. Also, the strains
found using (8.55): are inversely proportional to Youngs modulus
and directly proportional to a factor that includes
xx Cx g *y,yy g *y,xy Cy (8.60) Poissons ratio. Although there is no tectonic
normal stress in the y-direction, there is a tectonic
Both normal stress components include a contri-
normal strain in this direction, induced by the tec-
bution from the weight of the material that
tonic normal stress in the x-direction. The region
increases in compression with depth (negative y).
x 0 (Fig. 8.18) is one of horizontal tectonic exten-
The tectonic part of the horizontal normal stress
sion (xx 0), so the vertical strain is a contraction
varies linearly in the x-direction and is zero along
(yy 0). The region x 0 is one of horizontal tec-
the (x 0)-coordinate axis. The tectonic shear
tonic contraction (xx 0), so the vertical strain is
stress varies linearly in the y-direction from zero
an extension (yy 0). In both regions the strains
along the (y 0)-coordinate axis, taken here as
increase in magnitude linearly from x 0.
the traction-free representation of the Earths
The tectonic displacement eld is found by
surface. The only regions of interest in the (x, y)-
integrating the kinematic equations (8.33) after
plane are in the third and fourth quadrants of the
substituting (8.61):
coordinate system, where y 0 and the gravita-
tional normal stresses are compressive (negative).
For a rectangular region in the fourth quadrant
(x 0), the tectonic normal stress is tensile and
ux xxdx E 2
1 2 1 2
Cx f1(y) C1 (8.62)
the deformation that would be experienced by the Substituting for the displacement components
elastic material if gravity were turned off in the from (8.62) and (8.63) and for the shear strain from
initial state and then turned on for the nal state. (8.61) we nd:
For the Earth, the force of gravity is not turned on
and off. Therefore, most problems in structural
geology are better posed by comparing an initial
df2(x) df1( y)
dx
dy
2
E
Cy 2
1
E
Cy 0
state with gravitation loading, to a nal state with (8.65)
gravitation loading plus the appropriate tectonic
forces. In this context one should ignore the con- The rst term is only a function of x and the last
tributions of gravitational loading to the strain three terms are only functions of y so:
and displacement elds.
Substituting (8.60) for the stress components
in Hookes Law, without the term g*y, the in-
df2(x)
dx
C3 1
dy
df ( y) 2
E
Cy 2
1
E
Cy
plane tectonic strain components are: (8.66)
1 2 2
xx Cx,yy Cx, Rearranging and integrating:
E E
xy
1
E
Cy (8.61)
f2(x) C3dx C3x C4 (8.67)
8.4 QUASI-STATIC TRACTION BOUNDARY VALUE PROBLEMS 313
are zero: zz (rr ), rz 0, z 0. Taking
f1( y)
2 2
E
Cy C3 dy
the cylindrical z-axis and the reference axis Ox as
horizontal, perpendicular to the direction of the
gravitational body force, F, near Earths surface,
E 2
2 2 1 2
Cy C3 y C5 (8.68) the components of the body force are:
The functions f1(y) and f2(x) are substituted into Fr g * sin ,F g* cos ,Fz 0 (8.72)
(8.62) and (8.63), and the constants are combined For these conditions the equilibrium equations
to give the displacement components: are (Timoshenko and Goodier, 1970):
ux E 2
1 2 1 2
Cx
E
2 2 1 2
2
Cy C3 y C6 rr 1 r rr
r
r
r
Fr 0 (8.73)
2 F 0 (8.74)
uy Cxy C3x C7 (8.70) r r r
E
The equilibrium equations are satised, as may be
The last two terms in each displacement equation shown by substitution, with polar stress compo-
have a special interpretation. The constants, C6 nents related to an Airy stress function, (r, ), as
and C7, represent a rigid translation of the region of follows (Timoshenko and Goodier, 1970):
interest (Fig. 8.19a). The two terms with the con-
stant C3 represent a rigid rotation of the region 1 1 2
rr g*r sin (8.75)
about the origin (Fig. 8.19b). These constants do r r r2 2
314 ELASTIC DEFORMATION
1 (a)
r (8.76)
r r su r sru srr
2 suu
g*r sin (8.77) Fu
r2
Fr
To solve a particular problem the stress compo-
nents must satisfy the specied traction boundary
conditions following Cauchys Formula.
Hookes Law is dened by noting that one may
rotate the Cartesian coordinates (x, y) about the z-
axis to coincide with the polar coordinates (r, ). r u
Then, the strainstress relationships in polar
coordinates are written by exchanging each x
z
Cartesian subscript in (8.48)(8.50) with its polar
counterpart:
(b) (c)
1
rr [rr(1 2) (1 )] (8.78)
E
1
[(1 2) rr (1 )] (8.79)
E
1
r (8.80)
E r
1
u
u
r r (8.82)
(f) (g)
r
1 ur
2r
u
r u
r (8.83)
2 1 1 2
r2 r r r2 2 2 1 1 2
r2 r r r2 2
0 terms from Michells general solution in polar coordinates
(8.71) (Michell, 1899).
(8.84)
Solutions to this equation for stress states that In 1899, J. H. Michell derived a generalized
satisfy particular boundary conditions are Airy stress function in polar coordinates, from
reviewed in textbooks on elasticity theory which many particular solutions can be extracted
(Timoshenko and Goodier, 1970; Barber, 1992). (Michell, 1899; Timoshenko and Goodier, 1970):
8.4 QUASI-STATIC TRACTION BOUNDARY VALUE PROBLEMS 315
1
cr
2 1
cos (d1r 3 c1r1 d1r ln r) sin
D = r sin u
(a r
n2
n
n b r n2 a rn b rn2) cos n
n n n 2R
(c r
n2
n
nd
nr
n2 crn d rn2) sin n
n n suu
(8.85)
For r >> R,
For example, the rst three terms of (8.85) solve srr = suu
srr
the problem of symmetric loading of a hollow = rg*r sin u
cylinder (Fig. 8.20b), which is used for in-situ deter-
Fig 8.21 Schematic illustration of elastic half-space with
mination of elastic moduli in boreholes. The
half-cylindrical cut from traction-free surface used as model
fourth term is used to solve the problem of an for a valley.
elastic half-space with a distributed constant
normal traction (Fig. 8.20c). This nds many appli-
cations to loading of Earths surface. The last term and the z-axis is parallel to the valley. The shape of
on the rst line provides a pure shear stress as in the valley is idealized as a circular cylinder and it
a twisted rod (Fig. 8.20d). Terms on the second and is very long in the z-direction compared to the
third lines are used to solve the problem of a point radius, R. The geometry in the plane of interest
force in an innite body (Fig. 8.20e). Combinations is that of the half-space, r sin () 0, with a semi-
of point forces have been applied as earthquake circular cut that removes the elastic material
source mechanisms and the integration of the where r R.
point force solution over a boundary is a standard The surface of the elastic half-space with the
technique for developing solutions to new prob- cut is traction free. Thus, the conditions on the
lems. The rst term on the third line is used below three segments of the boundary are:
to solve the problem of a cylindrical valley loaded r R, 2
by gravity (Fig. 8.20f). The terms on the fourth line BC: on r R, , tr 0, t 0
with n 2 are the solution for a cylindrical hole in
r R, 2 (8.86)
a body with a uniform normal stress at innity
(Fig. 8.20g). This has been called the most impor- Prior to the valley being incised the polar stress
tant problem in rock mechanics (Jaeger and Cook, components are given by (8.75)(8.77) without the
1979, pp. 249) and apparently was rst obtained by terms containing the Airy stress function:
B. Kirsch (1898; Timoshenko and Goodier, 1970,
rr g *r sin ,r 0 (8.87)
p. 90). It has been applied to the stress distribution
around boreholes and tunnels, and to a variety of Because the valley provides only a local perturba-
in-situ stress measurement techniques. tion of this stress state the boundary conditions in
the remote eld are:
8.4.4 The stress state induced by gravity
rr g * r sin
near a valley
As an example of a particular solution taken from BC: as r and sin 0, g * r sin
Michells generalized solution (8.85), consider the r 0
state of stress near a valley excised into an other- (8.88)
wise featureless (planar) terrain (Fig. 8.21). The The normal stress components are equal com-
polar coordinate system (r, ) is chosen such that pressions that increase in magnitude linearly
the x- and z-axes are parallel to the planar surface with depth, D r sin.
316 ELASTIC DEFORMATION
R 2 3500
rr (g* r sin ) ,r 0, 0 5
r 4000
(8.90)
4500
5000 0
Everywhere on the circular boundary, r R, the 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
not perturbed by the valley, so it remains com- 109o 45' 109o 30'
Span CITY OF MOAB
pressive and increases in magnitude linearly with ish V
alley
Salt
depth below the horizontal surface. Contours of Antic
line
(Coll
the total circumferential stress are straight hori- apse
38
Kings Bottom d)
o 30'
zontal lines throughout the half-space including Sa
lt
Sy
ncl
the vicinity of the valley (Fig. 8.22c). The total cir- ine
Cane Creek
cumferential stress at the valley bottom is: Salt Anticline
Colorado
(r R, 3 2) g*R (8.91)
103
River
o 30'
In other words, the normal stress acting in the
horizontal direction is compressive, and scales in Gray
s Pa
sture
Salt
magnitude with the average unit weight of the
30'
Sync
line
38 o
surrounding rock mass and with the vertical dis- MEANDER
tance from the valley rim to the valley oor. ANTICLINE
38
Because the polar component of total shear stress
o
15
is zero, the stress trajectories are radial lines (least 110o
'
compressive stress) and circumferential curves
Gr
(greatest compressive stress). ee
n
Ri
ve
One of the more interesting structures associ- r
2
The rst type is characterized as broad upright
38 o
109
A Outlines of
folds that follow major drainages and, although 3 Needles
o
Miles
fault
45'
the folding tends to die out upward, it may deect 0 5 205.5 4
zone
strata all the way to the valley rim, up to 600 m B
C 5
Miles
D 6 38o
above the valley oor. The second type is smaller,
sometimes with straighter limbs and tighter Home Springs
Anticline
hinges, and usually is conned to thin-bedded 7
stress trajectories to the side of the valley, with (a) Uniaxial compression
greatest compression approximately vertical and applied by machine
least compression approximately horizontal is Hydraulic
consistent with the normal faulting. Lead piston head
wires Force
transducer
Hardened steel
8.5 Elastic properties from platen
specimen ends are not perfectly parallel, bending (a) Axial extension, ea
can be induced. Also, depending on differences in 0.003 0.002 0.001 0
0
the elastic properties of the platen and specimen,
and on the friction at their contacts, the specimen
25
Table 8.2. Rock mechanics laboratory tests for Table 8.3. Rock mechanics laboratory tests for
apparent Youngs modulus (GPa). apparent Poissons ratio.
granite is given in Fig. 8.26a (Obert and Duvall, A few generalizations can be made from the
1967). The set of line segments drawn between apparent elastic moduli given in Table 8.2. A par-
successive data points on this graph form a nearly ticular rock type is not associated with a particu-
straight line. In contrast Colorado granite displays lar modulus. This should come as no surprise
a distinctly non-linear behavior. The straight-line because rock classication schemes are based on
segments are not parallel, but these data could be ranges of values of chemical composition, miner-
used to calculate a tangent modulus. alogy, and texture. You have seen (Fig. 8.26b) that
Both granite specimens (Fig. 8.26a) failed by even samples from the same rock mass can exhibit
fracturing shortly after the most compressive different moduli. The metamorphic and igneous
stress shown on the graph was imposed. The rocks typically have greater values than the clastic
results we have given do not include possible data sedimentary rocks, although strongly weathered
on unloading to lesser compressive stresses, which granite can be less stiff than well-indurated sand-
would test whether the deformation of these stone. Speaking qualitatively, we would describe
granite specimens was reversible before fracturing rocks with values around 100 GPa as being very
began. Unloading data are shown in Fig. 8.26b for stiff, whereas rocks with values around 1 GPa
two limestone specimens (Obert and Duvall, 1967). would be termed very soft. From Table 8.2 we draw
One shows a nearly linear relationship between the following rule of thumb: laboratory specimens of
stress and extension, a relatively large modulus, rock have Youngs moduli that range from about 1 to
and approximately reversible behavior. The other 100 GPa with a typical value of about 50 GPa. Rocks
is distinctly non-linear, has lesser moduli, and does are very stiff relative to our experience with softer
not follow the same path in loading and unload- elastic materials like rubber, but they are elastic
ing. This non-linear behavior is attributed to the none-the-less.
presence of abundant microcracks in the second of When cylindrical specimens of rock are short-
the limestone specimens. ened axially in a uniaxial compression test (Fig.
8.25), they respond by expanding perpendicular
8.5.2 Apparent Youngs modulus and to the applied load. The negative ratio of this per-
Poissons ratio pendicular extension to the axial extension is
Representative values of the apparent Youngs Poissons ratio (8.10). For most rock specimens
modulus for selected metamorphic, igneous, and Poissons ratio is approximately constant only
sedimentary rock types are given in Table 8.2 over restricted ranges of loading, time, and speci-
(Bieniawski, 1984). Youngs modulus and the men size. None-the-less, representative values for
tangent modulus are not distinguished in these the same suite of rock types selected for Table 8.2
data. This table is not meant to be complete: one are given in Table 8.3 (Bieniawski, 1984).
can turn to handbooks that have more extensive Unlike values of Youngs modulus that vary
tabulations (Clark, 1966). over two orders of magnitude, values of Poissons
322 ELASTIC DEFORMATION
Table 8.4. Laboratory and engineering field tests for Youngs modulus (GPa).
BC: ur(r R, ) ur(r R, ), Remarkably, the principal stresses within the
(8.93) inclusion are simply proportional to their respec-
u(r R, ) u(r R, )
tive values in the remote eld and the propor-
It follows that the traction acting on the inclusion tionality constant is the same for each principal
is equal and opposite to the traction acting on the stress. The magnitude of the stress within the
surroundings at every point of the contact. This inclusion changes with the ratio of the shear
implies that certain components of the stress are moduli, k, but this quantity is always positive, so
continuous across the interface: the sign of the internal stress is always the same
as the sign of the respective remote stress. Thus, a
rr(r R, ) rr(r R, ), remote tension induces tension within the inclu-
(8.94) sion, and a remote compression induces compres-
r (r R, ) r (r R, )
sion within the inclusion.
We consider the case of biaxial remote stresses, 1r Recalling that k Gi/Gs, the above equations
and 2r , acting at innite distance in the x- and y- can be used to determine how the stress state
directions, respectively: changes within the inclusion as a function of this
ratio, again for the case where both Poissons
BC: at x , y : ratios are equal to 14 :
(8.95)
xx r1, xy 0 yx, yy r2
i1 i2 0,for Gi Gs
(8.99)
Because the shear stress components are zero in
the remote eld the normal components are prin- i1 32 r1 and i2 32 r2,for Gi Gs
cipal stresses. In general the stress in a softer inclusion is lesser
A remarkable and non-intuitive result found in magnitude than that in the surroundings, and
by solving this boundary value problem is that the the stress in a stiffer inclusion is greater. In the
state of stress in the inclusion is homogeneous. limit as the stiffness of the inclusion goes to zero
The principal stresses within the inclusion, 1i and the behavior changes to that of a circular hole
2i , are ( Jaeger and Cook, 1979, pp. 262): in elastic surroundings. The greatest stress that
can be induced within a stiffer inclusion is an
[k(s 2) i] k(s 1) r
ixx i1 increase over the remotely applied stress by a
2(2k i 1) (ks 1) 1
factor of 32.
[k(s 2) i 2] k (s 1) r Outside the inclusion the state of stress is het-
2 (8.96)
2(2k i 1) (ks 1) erogeneous but it changes with distance from the
[k(s 2) i 2] k(s 1) r inclusion to approach the remote state of stress.
iyy i2 1 The stress components in the surroundings are
2(2k i 1) (ks 1)
(Jaeger and Cook, 1979, p. 250):
[k(s 2) i] k (s 1) r
(8.97) 1 BR2
2(2k i 1) (ks 1) 2 rr ( r1 r2) 1 2
2 r
The homogeneity of stress and strain inside the
inclusion is true for all possible values of the
elastic moduli. The stress and strain also are
1
2AR2 3CR4
( r1 r2) 1 2 4 cos 2
2 r r
homogeneous within an elliptical inclusion, and (8.100)
within a three-dimensional ellipsoidal inclusion
(Muskhelishvili, 1954; Eshelby, 1957).
1
BR2
( r1 r2) 1 2
2 r
For the sake of easily interpretable results, con-
sider the special case where vi 0.25 vs, so i 2
s. The stress everywhere inside the inclusion is:
1
2 r
3CR4
( r1 r2) 1 4 cos 2
(8.101)
i1
3k
r , i2
2k 1 1
3k
r
2k 1 2
(8.98)
1
2 r
AR2 3CR4
r ( r1 r2) 1 2 4 sin 2
r (8.102)
8.6 ELASTIC HETEROGENEITY AND ANISOTROPY 325
The stress components are functions of the polar applied stress, and the stress is zero on the inter-
coordinates, r and , and the constants A, B, and C face that is parallel to the applied stress. In this
in these equations are related to the elastic con- sense the stiff inclusion creates a stress shadow in
stants as: the surrounding softer material. The stiffer inclu-
sion carries more of the applied load than adja-
2(1 k) 1 k(s 1)
A ;B i ; cent regions by up to a factor of 3/2. The adjacent
k s 1 2k i 1
regions carry less of the applied load by up to a
k1
C (8.103) factor of 1/2. These results provide a useful bench-
k s 1
mark for assessing the effect of stiffer or softer
Consider the two points just outside the inclu- rock masses on the state of stress.
sion, where the r R , and the polar angles are Distributions of the Cartesian stress compo-
0 and /2. There we calculate the circumfer- nents are illustrated in Fig. 8.29 for a uniaxial
ential stress using (8.101). Again, consider the case stress in the remote eld r1 1 and a softer inclu-
where the two Poissons ratios are equal to 14 , so sion with shear modulus Gi 10 GPa and Poissons
the relevant constants are A 2(1 k)(1 2k), ratio i 14 embedded in stiffer surroundings with
B 12 A, and C 12 A. Using these coordinates shear modulus Gs 30 GPa and Poissons ratio s
1
and constants we nd: 4 . Note that the stress components are uniform
within the inclusion and that both the y-compo-
(r R, 0) 1k
r
1 2k 1
3
r
1 2k 2 nent of normal stress and the shear stress are zero.
The x-component of normal stress within the
(8.104) softer inclusion is diminished in value relative to
the remote stress. The contour patterns outside the
(r R, 2) 3
r
1 2k 1
1k
r
1 2k 2 inclusion are quite complex, but they clearly illus-
trate the facts that perturbations due to the inclu-
(8.105) sion are symmetric, they decrease with distance,
and they are negligible at radial distances r 5R.
Because the coefcients that multiply the remote
stresses are just interchanged for the two points,
8.6.2 Generalized Hookes Law for
consider only a uniaxial stress, 1, acting along
anisotropic rocks and minerals
the x-axis:
The most general linear relationship among the
components of strain and stress is one in which
r1, for Gi Gs
(r R, 0) 0, for Gi Gs (8.106) each of the nine components of strain, ij, is lin-
1 r early related to the nine components of stress, ij.
2 1, for Gi Gs
For example, given a Cartesian coordinate system
with orthogonal axes x, y, and z, the equation for
3 r1, for Gi Gs
r R,
2 r1, for Gi Gs (8.107)
0, for Gi Gs
the normal strain component, exx, may be written
as (Nye, 1985, Chapter VIII):
xx s1111xx s1112xy s1113xz
If the inclusion is much softer than the sur-
roundings, Gi Gs, the circumferential stress s1121yx s1122yy s1123yz
component is similar to that for an open circular
s1131zx s1132zy s1133zz (8.108)
hole: on the interface that is perpendicular to the
applied stress the local stress is equal in magni- The sijkl are constants of proportionality and the
tude but opposite in sign, and there is a stress con- four subscripts on each s correspond to the two sub-
centration factor of 3 on the interface that is scripts of the strain component e, followed by the
parallel to the applied stress. two subscripts of the stress component , with sub-
If the inclusion is much stiffer than the sur- scripts (1, 2, 3) corresponding to (x, y, z), respec-
roundings, Gi Gs, there is a diminution factor of tively. There are eight more equations of this form,
1/2 on the interface that is perpendicular to the each one linearly relating a component of strain to
326 ELASTIC DEFORMATION
(a)4 xx
material is one in which relatively small stresses
1.6
cause relatively large strains. In other words, the
3 1.5 respective constants for a compliant material are
2 1..4 greater than those for a stiff material.
1..3 The units and dimensions of compliance are
1
1..2 the inverse of those for stress:
0 1..1
compliance, s[ ] N1m2 Pa1
1.0 (8.109)
-1 compliance, s{ } M 1L T 2
0..9
-2 The symmetry of the innitesimal strain com-
0..8
-3
ponents and of the stress components enables a
0..7
reduction in the number of independent compli-
0..6
-4
-4 -2 0 2 4 ances to 36 according to:
(b) yy
4 sijkl sjikl and sijkl sijlk (8.110)
0.15
3 0.10 Therefore, the linear anisotropic elastic material
2
0.05 is one in which each of the six independent com-
0.00 ponents of strain is linearly related to the six inde-
1
-0.05 pendent components of stress (Lekhnitskii, 1963;
0 -0.10 Jaeger and Cook, 1979):
-0.15 xx s11xx s12yy s13zz s14yz
-1
-0.20
-2
s15zx s16xy (8.111)
-0.25
-3 -0.30 yy s21xx s22yy s23zz s24yz
-0.35
-4 s25zx s26xy (8.112)
-4 -2 0 2 4
(c) xy
4
zz s31xx s32yy s33zz s34yz
3 0..2 s35zx s36xy (8.113)
2
0..1
yz s41xx s42yy s43zz s44yz
1
0.0 s45zx s46xy (8.114)
0
Fig 8.29 Contour maps of stress components near The sij form a matrix of constants of proportionality
cylindrical inclusion. (a) Normal stress, xx. (b) Normal which also are referred to as compliances. Here the
stress, yy. (c) Shear stress, xy. subscript notation for the compliances has been
simplied, based on the order of listing the strain
the nine components of stress, and each one and stress components. The rst subscript corre-
having nine proportionality constants. Altogether sponds to the rank of the strain component in the
there are 81 constants that are referred to as com- column on the left-hand side. The second subscript
pliances. These constants make up a fourth-rank corresponds to the rank of the stress component in
tensor quantity (Nye, 1985, p. 133). A compliant each row on the right-hand side. For isothermal and
8.6 ELASTIC HETEROGENEITY AND ANISOTROPY 327
reversible deformation, considerations of the common minerals is less than 21 because of their
elastic strain energy (Nye, 1985, p. 136) prove that symmetry. There are seven systems of crystal sym-
this matrix of compliances is symmetric: metry (triclinic, monoclinic, rhombic, tetragonal,
trigonal, hexagonal, or cubic) and specic symme-
sij sji (8.117)
try classes within each system (Lekhnitskii, 1963,
This symmetry further reduces the number of pp. 2632). Triclinic crystals have 21 different com-
independent compliances to 21, so the matrix of pliances, but as the symmetry increases the
compliances can be represented in the following number of compliances, as referred to crystallo-
manner: graphic reference axes, decreases. For example,
some trigonal crystals have only six independent
s11 s12 s13 s14 s15 s16 compliances (Nye, 1985, table 9):
s22 s23 s24 s25 s26
s11 s22, s12, s44 s55, s33, s13 s23,
s33 s34 s35 s36
(8.118) s14 s24 2s56, s66 2(s11 s12) (8.120)
s44 s45 s46
s55 s56 All hexagonal crystals have ve independent
s66 compliances:
It is understood that the terms below the diagonal s11 s22, s12, s44 s55, s33,
are equivalent to their counterparts above the s13 s23, s66 2(s11 s12) (8.121)
diagonal. Cubic crystals have three independent compli-
It is intuitive that solids will extend in the ances:
direction of an applied tensile stress. Further- s11 s22 s33, s12 s13 s23, s44 s55 s66
more, we have experience with solids (e.g. a (8.122)
rubber band) that contract in the directions per-
pendicular to an applied tension. Earlier in this The compliances not listed in these relationships
chapter we dened Poissons ratio as the elastic are identically zero.
property of an isotropic material that related such Measured values for the compliances of a few
lateral contractions to the extension in the direc- well-known minerals from the trigonal, hexag-
tion of the applied tensile stress. However, the onal, and cubic systems are given in Table 8.5
general anisotropic elastic material exhibits (Birch, 1966; Nye, 1985). The units are 1011N1/m2
behavior that is not so intuitive. For example, if a (1011 Pa1) and all refer to room temperature
uniaxial tensile stress is applied along the z-axis, unless otherwise noted.
the only stress component is zz, yet all six inde- Recall that the range of Youngs moduli for
pendent strain components would be non-zero: rock is from about 109 to 1011 Pa, so the inverse
Youngs moduli would range from 1011 to 109
xx s13zz,yy s23zz,zz s33zz, Pa1, roughly the same range as that of the values
(8.119)
yz s43zz,zx s53zz,xy s63zz of compliance in this table. Note that the miner-
als from each crystal system span a similar range
It is expected that a rectangular block of this
of compliances with the exception of ice, which is
material would extend parallel to the tension and
extraordinarily compliant. Also, minerals of the
contract perpendicular to this tension. However,
same composition, but different systems, such
there are shear strains induced by the tensile
as - and -quartz, have different compliances.
stress such that the block would have non-rectan-
Usually, greater temperatures correspond to
gular sides in the loaded state.
greater compliances, as in the case of halite.
Finally, minerals known for their hardness,
8.6.3 Compliances for anisotropic such as diamond, have very low compliances.
minerals and rocks Because rock is made up of many different
Minerals are natural examples of crystalline solids minerals, and these minerals are anisotropic with
that are anisotropic with respect to elastic proper- respect to elastic properties, it is natural to
ties; however, the number of compliances for many suppose that rock would be anisotropic. However,
328 ELASTIC DEFORMATION
in many examples the individual mineral grains that there are ve independent compliances for
are arranged with random orientations. Taking a rocks with this form of symmetry and that the
sample of such a rock that is large compared to compliances have the same relationship to one
the grain size, the elastic properties would be another as in the hexagonal crystal class. Rocks
approximately isotropic. Calculation procedures with these elastic properties are referred to being
have been derived to estimate the average transversally isotropic.
isotropic elastic constants for a rock from knowl- The strainstress relations for the transversally
edge of the elastic properties and abundances of isotropic rock can be written using variants of the
each mineral (Hearmon, 1961). more familiar elastic constants, Youngs modulus
Some rocks at the hand-sample scale are and Poissons ratio as follows:
anisotropic with respect to elastic properties
1
because the constituent grains are not randomly xx (xx yy) zz (8.127)
E E
oriented. For example, slates have a strong pre-
ferred orientation of the platy minerals. Such 1
yy (yy xx) zz (8.128)
rocks are likely to have an axis of elastic symme- E E
try perpendicular to the plane of the mineral
1
fabric. Within the plane of the fabric the rock may zz ( yy) (8.129)
E zz E xx
be approximately isotropic. Such a material is
described as having a plane of isotropy, or an axis 1 1 2(1 )
yz , , xy (8.130)
(perpendicular to this plane) of rotational sym- G yz zx G zx xy E
metry (Lekhnitskii, 1963, p. 24) and the linear
Here E and are Youngs modulus and Poissons
strainstress equations reduce to:
ratio for any direction of applied normal stress
within the plane of isotropy. The constants E and
xx s11xx s12yy s13zz (8.123)
are Youngs modulus and Poissons ratio for an
yy s12xx s11yy s13zz (8.124) applied normal stress along the axis of symmetry.
The constant G is the shear modulus for an
zz s13xx s13yy s33zz (8.125)
applied shear stress in any plane that contains the
yz s44yz, zx s44zx, xy 2(s11 s12)xy axis of symmetry.
For plane strain deformation (8.30) with uz 0
(8.126)
consider the possibility that there are two
Here the z-axis is the axis of rotational symmetry orthogonal axes of elastic symmetry in the (x, y)-
and the (x, y)-plane is the plane of isotropy. Notice plane. These conditions require four compliances
8.6 ELASTIC HETEROGENEITY AND ANISOTROPY 329
in contrast to the two elastic moduli for the Table 8.6. Youngs modulus of a few rocks in
isotropic material (8.48)(8.50) and dene the so- orthogonal directions (GPa).
called orthotropic solid:
Rock type Perpendicular Parallel Parallel
xx s11xx s12yy, yy s12xx s22yy, (A) (B)
xy s66xy (8.131)
Gneiss 18.6 23.1 12.4
In terms of Youngs moduli and Poissons ratios, Marble 49.3 62.7 71.7
these equations are written: Granite 30.4 27.4 44.2
1 1 Limestone 33.4 41.0 37.2
xx 21 , 12 xx yy, Sandstone 6.0 6.7 8.8
E1 xx E2 yy yy E1 E2
1
Sandstone 7.1 10.6 11.2
xy xy (8.132) Oil shale 12.4 21.4
G
Oil shale 21.1 33.2
What appear to be ve constants are only four
because the two Poissons ratios are related:
21 12 E circular hole (Fig. 8.30) in an orthotropic solid
, so 12 21 1 (8.133) loaded by a remote uniaxial stress ( Jaeger and
E2 E1 E2
Cook, 1979, pp. 2979). This is a two-dimensional,
E1 and 12 are Youngs modulus and Poissons ratio plane strain solution, so the hole represents a
for an applied normal stress along the x-axis, long cylindrical opening perpendicular to the
whereas E2 and 21 are Youngs modulus and (x, y)-plane. The in-plane stress components are
Poissons ratio for an applied normal stress along related to the Airy stress function by:
the y-axis. G is the shear modulus in the (x, y)-
plane. 2 2 2
xx 2
,yy 2 ,xy (8.136)
For isotropic materials, the number of inde- y x xy
pendent compliances reduces to two because:
Substituting these equations into the constitutive
s11 s22 s33, s12 s13 s23, equations for the orthotropic elastic material
s44 s55 s66 2(s11 s12) (8.134) (8.131):
How these compliances relate to the isotropic 2 2 2 2
elastic constants introduced earlier can be xx s11 2
s12 2 ,yy s12 2 s22 2 ,
y x y x
deduced from the strainstress relationships 2
written using the compliances and (8.134), and xy s66 (8.137)
xy
comparing these to (8.18), (Nye, 1985, p. 143):
Substituting the constitutive equations into the
1 1 compatibility equation written in terms of the
s11 , s12 , 2(s11 s12) (8.135)
E E G strain components, we have:
Data on anisotropic elastic properties for rock 4 4 4
samples are presented in Table 8.6 for measure- s22 (s66 2s12) 2 2 s11 4 0 (8.138)
x4 x y y
ments of the modulus of elasticity (Youngs
modulus) perpendicular to bedding or foliation (if Dividing through by s22 we dene the following
any) and in two orthogonal directions parallel to constants:
bedding or foliation (Obert and Duvall, 1967, p. 486). s11 s 2s12
12 C1, 66 1 2 C2
s22 s22
8.6.4 Deformation of an anisotropic
(8.139)
elastic body
To assess the importance of elastic anisotropy The compatibility equation can be rearranged as
during deformation of a rock mass we consider a follows:
330 ELASTIC DEFORMATION
2 2 2 2 problem are given in terms of yet more constants
2
1 2 2
2 2 0 (8.140) that are related to the constants 1 and 2 as
x y x y
follows:
Note that the sum and the product of the con-
stants 1 and 2 are related to the compliances 1 1 1
(8.139), so one must solve these two equations 1 0.094,2 2 0.261
1 1 2 1
simultaneously to nd these relations. Multi-
plying the second of (8.139) by 1 we have: (8.145)
12 C1, 21 12 C21 (8.141) Now consider the circular hole in an innite
anisotropic body with uniaxial stress, 1, at an
Subtracting the second equation from the rst innite distance acting in the Ox-direction (Fig.
equation we nd the following quadratic equa- 8.30). The x- and y-axes are symmetry axes with
tion: respect to the anisotropy. At the edge of the hole,
the circumferential normal stress is (Jaeger and
21 C21 C1 0 (8.142)
Cook, 1979, pp. 298):
The two solutions for this quadratic equation are:
(1 1)(1 2)(1 1 2 12 2 cos 2)
1 1 (1 21 21 cos 2)(1 22 22 cos 2)
2 C2 2 C 2 4C1,
1
1 2 where C 22 4C1 0
(8.143) (8.146)
150
0
20 40 60 80 100 120 140 160
100
-1
50
-2
Position Theta (degree) 0
0o 15o 30o 45o 60o 75o 90o
Fig 8.31 Plot of circumferential stress component, , at
Inclination, b, of cleavage to specimen axis ()
edge of cylindrical hole in orthotropic elastic material with
uniaxial remote stress, 1r. Three different cases are:
isotropic (triangles), loading parallel to E1 (diamonds), and Fig 8.32 Plot of strength versus inclination of cleavage for
loading parallel to E2 (squares). triaxial tests of Martinsburg Shale for three different
confining plessures: , 3.5 Mpa; , 10.5 Mpa; , 35 Mpa.
Reprinted from Donath (1961) with permission of The
The normalized circumferential stress (Fig. Geological Society of America.
8.31) for the isotropic material varies from 1 at
0 to 3 at /2. In comparison, the normal- of rocks are much more important in failure
ized circumferential stress for the anisotropic under inhomogeneous stresses than effects of
material dened above, and loaded parallel to the anisotropy of elasticity (Jaeger and Cook, 1979,
direction of greater Youngs modulus, varies from pp. 299). For example, data on the anisotropy of
0.737 at 0 to 3.475 at /2. Thus, the breaking strength under triaxial compression for
local stress acting perpendicular to the remotely samples of Martinsburg Shale are shown in Fig.
applied stress is diminished somewhat less than 8.32 where values range over one order of magni-
that for the isotropic material, and the local stress tude depending upon the orientation of the cleav-
acting parallel to the applied stress is concen- age (Donath, 1961). The greatest strength is
trated somewhat more. If the remote loading is measured when the cleavage is perpendicular to
parallel to the lesser Youngs modulus of this the compression and the least when the cleavage
anisotropic material, the circumferential stress is at an angle of about 30o to the compression.
varies from 1.443 at 0 to 2.722 at /2.
Thus, the local stress is diminished somewhat
more than that for the isotropic material, and
8.7 Concluding remarks
concentrated somewhat less. For the degree of
anisotropy used here, the differences in stress con- Since the earliest investigations of geologic struc-
centration and diminution relative to the tures, geologists have used qualitative observa-
isotropic material are modest. Using the isotropic tions and everyday words to interpret how rocks
solution to model this anisotropic rock does not behave as structures evolve. For example, two com-
change the qualitative nature of the stress distri- monplace words, competent and incompetent, have
bution and the sign changes are preserved from been used to explain the style and relative magni-
one side of the hole to the other. tude of deformation. Being colloquial expres-
In assessing the effect of elastic anisotropy sions, these words have been used quite freely to
Jaeger and Cook conclude that the ratio E1/E2 2 describe the mechanical behaviors of rock: com-
as used in Fig. 8.31 is rather extreme, so it is prob- petent implying less easily deformed and incom-
ably true that the effects of anistropy of strength petent more readily deformed under a given set of
332 ELASTIC DEFORMATION
forces. While these words are, perhaps, appropri- stress and strain. To understand the tectonic
ate for ordinary conversations about structures, processes that lead to the development of geolog-
they lack the formal denition that might make ical structures one must choose an appropriate
them useful in a quantitative analysis. In this boundary or initial value problem to serve as a
chapter we have introduced the concept of elas- mechanical model. To formulate such a problem
ticity and the elastic properties of rock. These one must postulate a particular mechanical
properties are useful in quantitative investiga- behavior. Here we have chosen Hookes Law for the
tions of geologic structures because they provide linear elastic solid; in Chapter 10 we choose
precise measures of the relationship between Newtons relation for the linear viscous uid.
Chapter 9
Brittle behavior
Jointed limestone bed at Lilstock Beach on the southern of direction, truly wonderful . . . The observer becomes
coast of the Bristol Channel, England. Photograph by D. D. so absorbed with the scene that he unconsciously
Pollard. begins to feel as if the rocks under and around him
were in process of being illimitably cleft from north to
An admirer of Nature may be excused becoming enrap- southas if the earths crust were in course of splitting
tured when he takes a view from any of these noble up from one pole to the other; and he only rids himself
terraces [in County Clare, Ireland]. Looking north, or of the feeling to become bewildered with the question,
south, his eyes are riveted on vast surfaces of gray lime- as to what mysterious agent produced the singular
stone rocks, split up to an extent, and with a regularity phenomenon he is contemplating (King, 1875).
334 BRITTLE BEHAVIOR
s2 = s3 s1 = s2
Fig 9.1 Schematic representation of brittle failure styles in formation of faults or joints. On the other hand
triaxial tests. (a) Extension test. (b)(e) Compression test laboratory tests may produce useful values of
with confining pressure increasing to the right. Reprinted material properties such as elastic stiffness, fric-
from Griggs and Handin (1960a) with permission of The
tional strength, or fracture toughness over a range
Geological Society of America.
of conning pressures, temperatures, and strain
rates. Although laboratory triaxial experiments
Jaeger and Cook comment on the naming of may provide few insights about the processes of
inclined fractures in compression test specimens faulting or jointing, knowledge of the material
(Fig. 9.1c and 9.2b) as follows: properties obtained from them is vital for the
Griggs and Handin (1960) . . . call it [the inclined frac- development of models of faulting and jointing.
ture] a fault because of its correspondence with geo- Ultimately the laboratory specimens depicted
logical faulting, and they have been followed by many in the rst three columns of Fig. 9.1 break apart
writers; however, it seems preferable to conne the into two or more fragments because extension or
term fault to the geological context and to retain the shear fractures propagate across the rock cylin-
term shear fracture in the experimental context
ders, which thereby loose their ability to support
(Jaeger and Cook, 1979, pp. 92).
the applied compressive stresses. On the other
Here we choose to follow the recommendation of hand, joints and faults are contained within
Jaeger and Cook so the two phenomena, shear Earths crust, so any behavior of laboratory speci-
fractures in laboratory specimens and faults in mens that is dependent upon the cylindrical geom-
Earths crust, are clearly distinguished. Further- etry or the properties of the testing apparatus in
more, we extend this recommendation to other contact with the specimen is unlikely to be dupli-
styles of localized deformation. For example, cated in nature. For example, consider the joints
extension fracture and splitting fracture are terms and faults in the inter-layered limestone and shale
appropriate for laboratory fractures that open, beds of the Blue Lias Formation at Lilstock Beach.
whereas we conne the term joint to the geologi- These are exposed in cross section (Fig. 9.3) along
cal context. What may seem like a semantic detail the steep wave-cut cliff just inland from the expo-
is important because it emphasizes the fact that sure (Chapter 9, frontispiece) that reveals the joint
laboratory tests rarely duplicate the length and traces on a single limestone bedding surface.
time scales, the materials, or the conditions of In this cliff exposure vertical joints are seen to
336 BRITTLE BEHAVIOR
Fig 9.2 Examples of brittle failure in triaxial test specimens Laboratory experiments other than the stan-
of Ohtawa basalt. (a) 0.1 MPa confining pressure. (b) 49 MPa dard triaxial test are motivated by eld observa-
confining pressure. (c) 98 MPa confining pressure. Reprinted
tions of joints and faults. For example, exposures
from Hoshino et al. (1972) with permission of The
that reveal the tips of joints (Chapter 9, fronti-
Geological Survey of Japan.
spiece) raise the question: what is the resistance
to propagation of a single opening fracture in
fracture the limestone beds but terminate at or rock? To understand how the concentrated tensile
near interfaces with adjacent shale beds. Clearly a stresses along the advancing fracture tip are
given limestone bed did not break along a single resisted by rock strength one may conduct an
fracture: multiple parallel joints formed with a experiment (Fig. 9.4a) where control of the oppo-
regular spacing and ultimately several differently sitely directed displacement of the two pins
oriented sets of joints developed. Two faults crop enables one to monitor fracture propagation and
out in this exposure that cut across and locally measure the so-called fracture toughness of the rock
disrupt the otherwise nearly horizontal bedding. specimen (Atkinson and Meredith, 1987). Speci-
Unlike a single open joint, each fault is composed mens made up of layers of the same or different
of multiple closely spaced surfaces and each materials enable the researcher to investigate
surface is in intimate contact with another opening fractures that stop or propagate across
surface. Members of this set of faults do not termi- layer interfaces (Renshaw and Pollard, 1995;
nate within the exposure so their extent is McConaughy and Engelder, 2001). Exposures of
unknown, but detailed knowledge of the stratigra- faults such as those at Lilstock Beach (Fig. 9.3)
phy has been used to estimate a few meters to a few raise the question: what is the resistance to
tens of meters of slip (Engelder and Peacock, 2001). sliding of one specimen in direct contact with
The juxtaposition of limestone and shale beds, the another of the same or dissimilar rock type?
magnitude of the slip, the rotation of the fractured Laboratory studies (Fig. 9.4b) of frictional strength
limestone blocks, and the owage of the shale are help to address this question. While fracture
some of the characteristics of these faults that are toughness or friction experiments focus attention
not duplicated in laboratory experiments. on mechanical aspects of opening fractures or
9.2 STRENGTH OF LABORATORY SAMPLES 337
F
9.2 Strength of laboratory samples
a negative slope and the inelastic mechanisms the stressstrain curve led researchers to try to
operating there are responsible for a degradation capture the complete stressstrain curve and to under-
of the load-carrying capacity as the sample contin- stand what inuence the testing machine might
ues to shorten and the failure process develops. have on the specimen behavior during failure
The sample is said to be in a brittle state because it is (Hudson et al., 1972; Jaeger and Cook, 1979, p. 177).
accumulating permanent strain as the magnitude Consider the schematic illustration of a
of the compressive stress decreases. Unloading testing machine (Fig. 8.25) and note that it is com-
during this brittle deformation along the dashed posed of two basic components: an hydraulic actua-
path results in a larger permanent strain, ep, and tor (including piston and cylinder, rod, and platen)
reloading returns approximately to the original for applying a force to the specimen and a reaction
curve. Tests typically end at a strain limited by the frame (tie-bars and cross-heads) for supporting this
apparatus (e.g. point D) or complete failure of the force. When uid pressure is increased in the
sample. actuator the piston is driven downward applying
a compression to the specimen and shortening it.
9.2.1 Soft and stiff testing machines The downward directed force is transmitted
Given a well-designed testing machine, the testing through the specimen to the lower cross-head.
procedure to determine strength is straightfor- The actuator also applies an equivalent upward
ward: load the sample until it starts to lose its directed force to the upper cross-head so the tie-
capacity to carry the load, and record the extreme bars are placed in tension and elongate, while the
value of stress. However, to understand the pro- cross-head bends. Both the actuator and the reac-
cess of failure one must understand the mechani- tion frame deform and are capable of storing
cal interplay between the sample and the testing elastic energy. If the mechanical parts of the
machine. It turns out that the machines can play testing machine combine to be very much stiffer
a strong, and even dominant, role in the outcome than the rock specimen, the machine would
of mechanical property tests and experiments. deform very little and we could ignore its role in
Indeed, the particular action of these machines the test. However, specimens are explosively dis-
has strongly inuenced our understanding of how integrated in some testing machines because
rock specimens behave when they are fracturing, the energy stored in the machine is released into
so we need to understand this action and ask if the specimen even though no uid is pumped
it might duplicate the behavior of the Earth into the actuator to do additional work. In other
during natural deformation events. This action, as words, the system composed of the machine and
we learn in this section, depends upon whether the specimen self-destructs.
the machine is soft or stiff relative to the rock To understand this process the initial defor-
sample. mation of the specimen is idealized (Fig. 9.6a)
For many discussions of extension and shear with a spring of constant stiffness, Cs. The
fracturing in the literature of rock mechanics the machine frame is idealized with two springs of
stressstrain curve after failure initiates is not stiffness Cm/2 in parallel and tied together with an
shown because the testing machine was incapable upper and lower rigid bar. All of these springs are
of tracking these quantities accurately during the tied to a rigid basal support which is taken as the
rapid failure event. Shortly after reaching the peak reference frame for measuring displacements.
stress (Fig. 9.5, point C), the data stream was ter- The vertical y-axis is positive downward so forces
minated because the specimen disintegrated into and displacement acting downward are positive.
a pile of rock chips and dust with a loud bang. As Two applied forces idealize the actuator: Fs acting
early as 1943 it was clear that explosive disinte- downward on the specimen spring and Fm acting
gration was not necessarily a natural behavior, but upward on the rigid bar joining the two machine
rather elastic energy is stored in the cylinder and springs (Fig. 9.6b). These forces are identical in
in the machine . . . and the release of this energy magnitude and opposite in direction, so Fs Fm.
causes the breakdown of the cylinder (Whitney, Under the applied force, Fs, the specimen spring
1943). Curiosity about the post-peak-stress part of contracts and its upper end moves toward the
340 BRITTLE BEHAVIOR
Consider a test in which loading has pro- are those that are not prevented by the boundary
gressed along the elastic portion of the forcedis- conditions, such as the xed lower bar. Applying
placement curve until Fs F (Fig. 9.7a). For the this principle to the system illustrated in Fig. 9.7
next small increment of force, F, the displace- we add the two equations in (9.5) and set the
ment increases by u and the increment of work results equal to zero:
done by the force on the specimen spring is:
Wm Ws 12 Cm | f(us) | (u)2 0 (9.6)
W Fu 12 Fu (9.3)
The system is in static equilibrium only if the
W is called the virtual work and u is called the machine stiffness, Cm, and the magnitude of the
virtual displacement. We use (9.3) to calculate the specimen stiffness, |f (us)|, are identical.
virtual work associated with the specimen and To understand the implications of (9.6) we
testing machine in the post-peak stress regime. In examine the two cases where equilibrium is not
that regime, suppose the actuator applies forces attained. First, consider the machine to be stiffer
Fs P and Fm P so the system is in equilibrium. than the specimen, so Cm |f (us)| and area 2 is
Imagine replacing the actuator with a perfectly greater than area 3 (Fig. 9.7a, b). For this condi-
rigid bar that transmits this force and couples the tion, the work that the machine spring can do on
specimen and machine springs so they must dis- the rest of the system is not sufcient to meet the
place by the same amount (Fig. 9.6c). If a down- requirements of the work done on the specimen
ward virtual displacement, u, is imposed on the spring. To achieve the downward displacement,
upper horizontal bar (Fig. 9.6d), the forces within u, work must be done by some external force, for
the springs change as: example the hydraulic pump supplies more uid
to the actuator, thereby increasing the applied
Fm Cmu
(9.4) force. These testing conditions are described as
Fs f (us)u | f(us) |u, post-peak stress stable because the system will remain in static
Using (9.3) the virtual work for the machine and equilibrium unless some external action is taken.
specimen are: If this external action is carefully controlled, the
test can proceed under stable conditions and the
complete stressstrain curve will be captured.
Wm Pu 12 Cm(u)2
(9.5) Now suppose the machine is softer than the
Ws Pu 12 | f(us)|(u)2 specimen spring, so Cm |f (us)| and area 2 is less
than area 4 (Fig. 9.7a, c). The machine spring can
Note that u is small (less than unity), so (u)2 is do more work than the specimen spring requires;
very small. Thus, the leading terms in (9.5) are so if the distance between the upper bar and the
larger than the second terms, so Wm is a negative specimen spring is not changed, the machine will
quantity and Ws is positive. That is, work is done drive the top of the specimen downward at an
by the machine on the specimen. accelerating rate. For an actual testing machine,
The next step in the mechanical analysis is to this could lead to a catastrophic result, and the
determine if the system is stable by asking: what stressstrain information for the remainder of the
are the conditions under which the system will test would be lost. These testing conditions are
spontaneously evolve to a different state? Such a described as unstable because the system self-
spontaneous response could represent cata- destructs unless some external action is taken, for
strophic failure of the specimen. To address this example the hydraulic pump drains uid out of
question we apply the principle of virtual work: for the actuator, thus decreasing the applied force.
any possible virtual displacement of an elastic This must happen at a sufcient rate for the system
body in equilibrium, the total virtual work done to remain in equilibrium. Another solution is to
by the internal forces, the body forces, and the control the test using the displacement of the
surface forces must vanish (Venkatraman and specimen, us, instead of the force from the actua-
Patel, 1970, p. 120). Possible virtual displacements tor. Modern testing machines use servo-control
342 BRITTLE BEHAVIOR
Table 9.1. Rock mechanics laboratory tests for Table 9.2. Rock mechanics laboratory tests for
uniaxial tensile strength (MPa). uniaxial compressive strength (MPa).
region that has a more homogeneous state of along the axis of the specimen and record the
stress. magnitude of its minimum (most compressive)
A data set (Table 9.1) for uniaxial tensile value (9.9). Friction between the end platens and
strengths (Bieniawski, 1984) is given for the same the sample may constrain the sample to expand
suite of rocks used to tabulate elastic properties laterally in concert with the platens. The resulting
(refer to Tables 8.2 and 8.3). stress state near the ends can be non-uniform and
The uniaxial tensile strengths range over about not uniaxial. The effects of the end platens on the
one order of magnitude among all rock types tab- local stress state and the consequences for style
ulated, and even among different samples of the and localization of deformation in uniaxial com-
same lithology. The tensile strengths for crys- pression have been investigated (Peng, 1971; Peng
talline rocks can be as small as those for clastic sed- and Johnson, 1972). Proposed methods to com-
imentary rocks, but typically they are somewhat pensate for this effect include matching the
greater. We conclude that laboratory specimens of elastic constants of the end platens to the sample
rock have uniaxial tensile strengths that range and inserting materials between the ends and the
from about 2 to 40 MPa with a typical value of platens to reduce the friction. These and other
about 10 MPa. renements of testing procedures have resulted in
Comparing the apparent Youngs moduli from more accurate determinations of uniaxial com-
Table 8.2 to the tensile strengths of Table 9.1, pressive strengths (Bieniawski and Bernede, 1979).
notice that the Youngs moduli are several thou- Selected uniaxial compressive strengths are
sand times greater than the tensile strengths. given in Table 9.2 (Bieniawski, 1984).
Using 1 : 4000 as representative of the ratio of These uniaxial compressive strengths range
tensile strength to Youngs modulus, the value of over about one order of magnitude. Crystalline
the axial extension at failure would be rocks tend to have greater compressive strengths
ea(max) Tu E 2.5 104. A practical conse- than clastic sedimentary rocks, but some granites
quence of this great difference between Youngs are weaker than some shales. We conclude that
modulus and the uniaxial tensile strength is that laboratory specimens of rock have uniaxial com-
axial extensions in uniaxial test samples are very pressive strengths that range from about 30 to
small. In most cases the deformation is approxi- 350 MPa with a typical value of about 150 MPa.
mately elastic for these small extensions. We con- Comparing Tables 9.1 and 9.2 we draw another
clude that rocks have an elastic stiffness on the conclusion: typical laboratory samples of rock
order of 103 times the tensile strength under these subject to uniaxial loading are weaker in tension
conditions. than in compression by about one order of magni-
An apparatus for conducting uniaxial com- tude. This has profound implications for the devel-
pressive tests is shown in Fig. 8.25. Again, the opment of brittle deformation in Earths crust and
objective is to induce a uniform normal stress its interpretation by structural geologists.
344 BRITTLE BEHAVIOR
3
=s
the loss of load-carrying capacity; they simply
2
=s
assert that this happens at a certain stress level.
1
s
9.2.3 Polyaxial strength, stress invariants, Cu
and stress deviation
s1 = Pc = s2
Below the traction-free surface of the Earth the
state of stress may be approximated by Andersons
C t
standard state, introduced in Chapter 6, in which
the principal stresses are equal in magnitude
(1 2 3) and become more compressive lin-
early with depth. This isotropic state of stress is
likely to be supplemented in regions of tectonic
activity by stresses that result in a polyaxial state of
Fig 9.10 Plot in principal stress space with uniaxial tensile
stress: the principal stresses have different magni-
and compressive strengths, Tu and Cu, and triaxial
tudes, none of which is likely to be zero. In prin-
compressive strength, Ct.
cipal stress space (Fig. 9.10), the uniaxial tensile
strength plots as a point (Tu, 0, 0) on the positive 1-
axis and the uniaxial compressive strength plots the strength. We dene a criterion for failure as a
as a point (0, 0, Cu) on the negative 3-axis. The function of the principal stresses that represents
paths representing these tests are straight lines the failure surface and write this function in terms
from the origin. A third path lies in the plane (1 of the stress invariants as dened in Chapter 6. Recall
2) and extends into the octant where all prin- that the invariants have the special property that
cipal stresses are compressive along the straight they do not change magnitude with the orienta-
line 1 2 3. This represents the isotropic com- tion of the Cartesian coordinate system. Because
pressive loading envisioned by Anderson and physical properties of a material, such as strength,
usually is followed initially in laboratory triax- should not depend upon the arbitrary orientation
ial tests. Then, two of the principal stresses are of a coordinate system, it is natural to dene such
held constant and equal to what is called the properties in terms of the stress invariants (Jaeger
conning pressure, Pc, while the third principal and Cook, 1979). Criteria for failure under uniaxial
stress becomes more compressive until failure at conditions are equivalent to placing limiting
the point (Pc, Pc, Ct), the triaxial compres- values on the rst invariant of the stress tensor,
sive strength (see next section). Strength in the I1 1 2 3, when only one principal stress is
context of a polyaxial state of stress is represented non-zero:
by a surface, called the failure surface, that passes
through these three points. The complete failure I1 T0,2 0 3 (9.12)
surface separates possible states of stress between
I1 C0,1 0 2 (9.13)
the origin and the surface from impossible states
of stress on and beyond the surface. Note that this pre-supposes that the initiation of
The combination of principal stresses (1, 2, failure depends only upon the current homo-
3) at any point on the failure surface represents geneous state of stress. That is, failure does not
346 BRITTLE BEHAVIOR
depend upon stress gradients or the history of In practice, strength has been dened using
loading. many different combinations of the stress compo-
For a polyaxial state of stress it is commonly nents as dictated by the many different kinds of
assumed that failure is primarily related to the dis- testing machines and sample congurations that
tortion of a material, whereas changes in volume have been invented. Some of these are based on a
are of secondary importance. Recall from our dis- mathematical premise, such as dependence upon
cussion of the bulk modulus of elasticity in Chapter invariants of stress or of stress deviation, whereas
8 that volume change is proportional to the mean others are based on physical arguments for the
normal stress, m 13 (1 2 3) 13 (I1). This is causes and mechanisms of failure. Usually these
the normal stress that acts on the octahedral strength criteria may be characterized as some
planes, the eight planes with normals that are functional relationship among the components
equally inclined to the directions of principal of principal stress (Jaeger and Cook, 1979):
stress. The vertices of the octahedron dened by
1 f ( 2, 3) (9.18)
these planes lie on the principal stress axes.
Therefore, the mean normal stress is sometimes Each of these functions denes a failure surface in
referred to as the octahedral normal stress. The mean principal stress space. These criteria do not take
normal stress is subtracted from each normal possible spatial gradients in the stress eld into
stress component to dene the components of the account, so they must be applied on a point-by-
stress deviation tensor as follows: point basis in a heterogeneous eld of stress.
sxx xx m, syy yy m, szz zz m,
9.2.4 Triaxial strength, confining
syz yz, szx zx, sxy xy (9.14) pressure, and pore pressure
Note that the shear stress components are identi- Largely because of the technical difculties in the
cal to the shear stress deviation components. design of a true polyaxial apparatus, the most
The invariants of stress deviation are (Jaeger common procedure used in rock mechanics is the
and Cook, 1979): triaxial test. The conditions imposed are illustrated
in principal stress space (Fig. 9.10) and the appar-
J1 sxx syy szz 0 atus is shown in a schematic cross section (Fig.
J2 (sxxsyy syyszz szzsxx) s2xy s2yz s2zx 9.11). More complete and precise engineering
(9.15)
J3 sxxsyyszz 2sxy syzszx sxxs2yz drawings are available (Griggs and Handin,
syys2zx szzs2xy 1960b). The apparatus itself is placed between the
The second invariant of stress deviation is related two platens of a testing machine (Fig. 8.25), which
to the shear stress acting on the octahedral planes provides the axial load. A key feature of the appar-
which is called the octahedral shear stress, o: atus is a port for supplying uid, under pressure
called the conning pressure, Pc, to the region
1
o [(1 2)2 (2 3)2 (3 1)2 ] 12 between the inner wall of the pressure vessel and
3
the cylindrical jacket surrounding the rock
2
J (9.16) sample. The jacket, often rubber or malleable
3 2
metal like copper, is impermeable to uids and
For materials that failure due to shearing one pos- more easily deformed than the sample itself. A
sible criterion using stress deviation invariants is separate port can supply a different uid, under
based on the octahedral shear stress attaining a pressure called the pore pressure, Pp, directly to the
critical value, Oo, taken as a constant (Jaeger and sample surface, and thence to the internal pores
Cook, 1979): of the rock. Because Pp Pc, the pore uid does not
inate the jacket and stays within the rock pores.
Oo max (o) (9.17)
A furnace can be attached to the pressure vessel to
Here Oo is the octahedral shear strength of the mate- heat the sample and the vessel to a designated
rial (Hobbs, 1962). temperature. Here we examine the role that
9.2 STRENGTH OF LABORATORY SAMPLES 347
Force applied stress, a r, but they change in shape for non-
by testing machine isotropic stress states, and these shape changes
are believed to be related to failure. Recall,
Piston head
however, that an anisotropic elastic material will
Force transducer distort under isotropic loading (Chapter 8), so
Pore fluid this presumption is appropriate, at best, for
Pressure, Pp
samples that are isotropic with respect to elastic
Platen
properties.
Seal Positive differential stress corresponds to
extension tests and negative differential stress
Deformable corresponds to compression tests (Fig. 9.1). Extreme
jacket
values of the differential stress are recorded as the
Rock sample differential strength for extension, De, and compres-
sion, Dc, respectively:
Confining fluid
pressure, Pc
De max (),1 Pc, 2 Pc 3
Pressure vessel (9.21)
Dc |min () |,1 Pc 2, 3 Pc
(9.22)
Testing machine frame
Some experimentalists use the extreme value of
Fig 9.11 Schematic illustration of a triaxial testing the axial stress as the measure of strength in tri-
apparatus. axial tests (Jaeger and Cook, 1979):
s3 (MPa)
conning pressure, to a localized shear fracture
oblique to the sample axis at the intermediate
conning pressure, to a broad network of oblique
shear fractures at the highest conning pressure.
600 Clearly the magnitude of the conning pressure
plays an important role in determining the mech-
anisms that govern the strength of this rock.
800 An important concept concerning the
Maze strength of soils that are saturated with water or
Maze other uid was introduced by Karl Terzaghi in
Furukawa 1923 and later found application in studies of
1000
rock, concrete, and other porous and permeable
Fig 9.12 Plot of triaxial strength data in principal stress
solids (Terzaghi, 1943). The concept depends upon
space for three sandstones from Tertiary basins of Japan
the uid lling all the pores of the material and
(Hoshino et al., 1972).
these pores must be homogeneously and perva-
sively distributed throughout at a scale that is
The uniaxial compressive strengths for the small compared to the scale of interest. Further-
three sandstones range from 46 to 113 to 122 MPa, more the pores must be interconnected in such a
so there is considerable variability in strength way that local changes in uid pressure during
even among samples representing the same lithol- deformation are rapidly equilibrated by ow
ogy. The two samples of Maze sandstone, taken through the network of pores. Finally, the concept
from the same formation, differ by almost a factor is purely mechanical, so the uid must not react
of three. Presumably these differences reect chemically with the solid. Under these condi-
subtle differences in the constituents of these tions, Terzaghi discovered that the deformation
rocks that are not reected in their lithologic and and failure of soil samples in the laboratory
formation names. The triaxial compressive depended upon the so-called effective stress state, as
strengths, Ct 3, increase as conning pres- opposed to the stress state as ordinarily dened
sure, Pc 1, increases. Note that the general (Nur and Byerlee, 1971).
form of the experimental data for a particular The effective stress state is related to the stress
suite of samples approximates a linear relation- state by adding the pore uid pressure to the
ship between the principal stress components. normal stress components:
Later in this chapter we introduce a failure cri-
terion, called the Coulomb criterion that is con- xx Pp, yy yy Pp, zz zz Pp,
xx
sistent with a linear relationship between the yz yz, zx zx, xy
xy (9.24)
principal stress components, 1 and 3, at failure.
In Figure 9.2 samples of Ohtawa basalt are Do not confuse the components of effective stress
shown after deformation in the same triaxial (9.24) with the components of stress deviation (9.14)
testing apparatus used in the previous example dened by subtracting the mean normal stress
(Hoshino et al., 1972). These particular samples from each of the normal stress components. One
were deformed at conning pressures of 0.1, 49, can visualize the role of pore pressure by consi-
and 98 MPa, respectively, while all other condi- dering the typical stress state to be compressive
9.2 STRENGTH OF LABORATORY SAMPLES 349
Table 9.3. Triaxial test results on Berea Taking the conning and pore pressures to 500 MPa
Sandstone (MPa). resulted in a differential strength of 63 MPa, within
the range of values for the tests at lower pressures.
Pc (Pp 0) Dc Pc Pp Dc Clearly, the pore pressure serves to neutralize the
role of conning pressure in increasing the differ-
0 72 0 72 ential strength of Berea Sandstone.
50 159 50 82 Consider the triaxial strength test data on
100 242,248 100 78
Berea Sandstone plotted in principal stress space
150 150 75
(Fig. 9.13a). Conning pressures were varied from
200 418,432 200 60,64
0 to 200 MPa, representative of depths to about
500 500 63
8 km. The tests were conducted at room tempera-
ture and pore pressures from 0 to 175 MPa. For
(negative) at depth. There, the normal stress com- each condition of conning and pore pressure, a
ponents act inward on the boundaries of a repre- data point represents the principal stress state at
sentative element of porous rock, whereas the uid failure. For a given maximum principal stress
pressure pushes outward in all directions from (1 Pc), say 200 MPa, the minimum principal
within the pores. In this way the pore pressure stress (3 a) at failure is less compressive as
serves to counterbalance the compressive stress. the pore pressure increases. This variation in
Since the pore pressure acts equally in all directions strength is considerably greater than the varia-
it does not inuence the shear stress components. tion between samples tested at the same pore
Two experimental observations, carried out pressure, so it is deemed to be signicant. It is not
using triaxial testing procedures, led Terzaghi to possible to summarize these data with a single
the concept of effective stress. When the axial and line or curve.
radial stress are of equal magnitude, the sample is If Terzaghis concept has merit, the strength of
subjected to an externally imposed isotropic state Berea Sandstone at a given effective conning
of stress, a r Pc. As this isotropic state of pressure should be constant. The effective conning
stress changes, the volume of the sample changes. pressure is dened as the difference between the
On the other hand if the pore pressure is conning pressure and the pore pressure:
increased in magnitude at the same rate as this
Pc Pc Pp (9.25)
isotropic stress so Pc Pp, Terzaghi noted that the
volume of the sample did not change appreciably. In Figure 9.13b the data for Berea Sandstone are
Thus, the volume change is related to the effective plotted in effective principal stress space. On the
conning pressure, Pc Pp. abscissa the maximum principal effective stress is
The second observation of Terzaghi was that plotted, and this is equivalent to the negative of
the strength of laboratory samples increased the effective conning pressure, 1 Pc. The
signicantly with conning pressure, but did not minimum effective stress, 3, is plotted on the
increase appreciably if the pore pressure was ordinate. All data points at a given effective
increased in concert with the conning pressure. conning pressure have essentially the same
We use experimental data for Berea Sandstone in strength, thereby verifying Terzaghis concept.
Table 9.3 to illustrate this phenomenon (Handin et Furthermore, the collection of data points repre-
al., 1963). senting effective stress states at failure collapse
For this porous and permeable sandstone the approximately (presumably within the experi-
differential strength, Dc, increased from 72 to over mental error) onto a straight line. This fact sug-
400 MPa with increases in conning pressure from gests that a single failure surface can be dened in
0 to 200 MPa and zero pore pressure. With pore terms of the effective principal stresses.
pressure equal to conning pressure the differen- The pore pressure also effects the transition
tial strength varied non-systematically over the from brittle to ductile behavior (Handin et al.,
range 60 to 82 MPa. These changes are probably 1963). This is illustrated (Fig. 9.14) on axial stress
within experimental error of being constant. versus axial strain curves for Indiana Limestone at
350 BRITTLE BEHAVIOR
44 480
(a) s1 = s2 = Pc (MPa)
250 200 150 100 50 0 0.0
0
s3 (MPa)
41.4
22 241
Pore pressure
0 MPa
50 MPa
500
100 MPa
55.2
600
150 MPa
175 MPa 62.1
700
0
250 200 150 100 50 0
0
0 2.54 5.08 7.62 104
100 Piston displacement (m)
= s3
s 1= s 2
200 Fig 9.14 Plot of axial force versus piston displacement for
triaxial test of Indiana Limestone at a constant confining
s3 (MPa)
s | |
Wx
sin
Wy cos
tan ,0 90 (9.27)
W
The absolute value is taken to assure a positive (c)
sign. In this context the angle is referred to as
the angle of friction.
A conceptual model for laboratory friction
experiments includes a surface in contact with an B
C
object, sometimes called the slider, and a spring
Force, F
A E
that represents the elastic behavior of the testing
machine (Fig. 9.15b). As the right end of the spring
moves with a steady velocity v imposed by the
testing machine, the force F in the spring D
increases linearly with the displacement d of the
point at the right-hand end of the spring, while
the slider remains xed, ds 0. When the equality
Displacement, d
in (9.26) is satised the slider begins to displace
and the plot of force versus displacement becomes Fig 9.15 Schematic illustrations of concepts related to
non-linear (Fig. 9.15c). The behavior of this system friction. (a) Block on inclined plane slides under the action of
in the simplest cases can be described either as gravity. (b) Block on horizontal plane slides under action of
stable sliding or as stick slip sliding (Byerlee, 1978; force transmitted through a spring. (c) Plot of force versus
Dieterich, 1981). In stable sliding, the slider accel- displacement shows stable sliding (OABC) and stick slip
erates gradually to a constant velocity equal to v (OABDE). Reprinted from Byerlee (1978) with permission of
Birkhanser-Verlag.
and maintains this speed with a constant spring
352 BRITTLE BEHAVIOR
x
force (path OABC). On the other hand, stick slip
behavior involves the sudden acceleration of the
slider with concurrent drop in the spring force.
5.17 3.45 1.72 0
Eventually the force recovers and such events may Normal traction, tn (MPa)
be repeated many times during a single experi-
ment (path OABDE, etc.). This behavior has been Fig 9.16 Plot of shear traction versus normal traction for
suggested as the laboratory counterpart of the laboratory friction tests on sliding surfaces in marble (A),
mechanism responsible for earthquakes (Brace trachyte (B), trachyte with smoother surfaces (C), and
and Byerlee, 1966). sandstone (D). Reprinted from Jaeger and Cook (1979) with
The actual conguration of specimen and the kind permission of Mrs. Jennifer D. Cook.
testing machine for laboratory friction experi-
ments is quite variable (Jaeger and Cook, 1979):
F dW (constant sliding velocity) (9.28)
some are not much more sophisticated than the
conceptual model of a slider on a surface (Fig. Here d is the dynamic friction. The dynamic fric-
9.15b). A common design utilizes the standard tri- tion typically is less than the static friction.
axial test with a diagonal saw cut across cylindri- Although static friction, s, is fundamentally
cal specimens (Fig. 9.11). Another design involves dened in terms of the applied forces (9.26), one
pushing a block of rock between two adjacent may divide through by the apparent contact area
blocks (Fig. 9.4b), and still another (Tullis and and write a comparable relationship for the static
Tullis, 1986; Tullis, 1988) employs the counter- friction in terms of the shear traction, ts, and
rotation of two hollow cylinders. Reported values normal traction, tn, acting on the surfaces:
of the static friction do not vary widely. In Table
9.4 a few representative values are recorded
|ts | stn, tn 0 (sliding initiates) (9.29)
(Jaeger and Cook, 1979). The absolute value is used on the left-hand side
The rule of thumb we take from these data is because the relationship should not depend upon
that values of the static friction for common rock the arbitrary sign of the shear traction. Recall that
types typically range from about 0.5 to 0.8, with a positive normal traction pulls on the surface,
0.6 being a good general estimate. Apparently whereas a negative traction pushes against it. The
there is some ambiguity in the reported values of normal traction must be zero or negative to insure
the static friction because some researchers use that the two surfaces stay in contact.
the force required to initiate sliding (point A, Fig. Laboratory data using an apparatus similar in
9.15c), others use the greatest value (point B), and design to that shown in Fig. 9.4b approximate a
still others use the value after a stable sliding at a linear relationship between the shear and normal
constant velocity is achieved (point C). For the case tractions as sliding initiates for marble (A), tra-
of sliding at a constant velocity a second value of chyte (B), trachyte with smoother surfaces (C), and
friction usually is dened as: sandstone (D), (Fig. 9.16). The surfaces of the test
9.2 STRENGTH OF LABORATORY SAMPLES 353
specimens are described as moderately rough Table 9.5. Frictional strength and coefcient of
and the behavior is believed to be representative friction.
of many rock types ( Jaeger and Cook, 1979). Note
that the intercept on the shear traction axis is not Rock type Sf (MPa) c
zero for any of the data sets, so a more general
linear relationship than (9.29) is required: Marble 1.10 0.75
Trachyte 0.41 0.68
|ts | Sf ctn,tn 0 (sliding initiates) (9.30) Gabbro 0.38 0.66
Granite 0.31 0.64
The intercept on the ordinate, Sf , is the frictional
Sandstone 0.28 0.51
strength in the absence of any normal traction and
the slope c is the coefcient of friction. This linear
relationship is identical in form to Coulombs cri- on the magnitude of the normal traction acting on
terion for the shear strength of intact solids which the sliding surface (Byerlee, 1978). The rst class
is described later in this chapter (9.38). The includes normal traction conditions to 5 MPa and
Coulomb criterion applies to the interior of a con- would therefore relate to very shallow conditions
tinuous solid rather than to discrete surfaces in in the Earth, typically less than a few hundred
frictional contact. The mathematical similarity of meters depth. This is the environment of the engi-
these equations should not obscure the fact that neering geologist and civil engineer (Barton, 1973).
the former describes a friction experiment where At moderate normal tractions, from 5 to
sliding is induced along two surfaces in contact, 100 MPa, many laboratory results for maximum
whereas the latter describes the initiation of a friction (point B on Fig. 9.15c) plot close to a line
shear fracture in an otherwise unbroken solid. with zero intercept and a slope equal to a static fric-
Laboratory experiments also have demonstrated tion of 0.85 (Fig. 9.17a). Rock types include sand-
that friction is dependent upon the velocity of stone, graywacke, limestone, quartzite, gneiss,
sliding and the time of contact (Dieterich, 1979a, granite, granodiorite, and gabbro. These results are
b, 1981; Kilgore et al., 1993). applicable to underground excavations, well-bore
Given data that dene the linear relationship problems, and sliding on faults down to about 4 km
(9.30), the static friction, s, is derived from the depth. Apparently the friction is not highly depen-
frictional strength and coefcient of friction as dent on lithology (with a few exceptions), nor is it
follows: particularly dependent on the roughness of the
|t s |S sliding surfaces. The data set for maximum friction
s f c, in the range from 100 MPa up to about
tn tn (9.31)
tn 0 (sliding initiates) 1500 MPa is applicable to sliding on faults at
depths from 4 to perhaps 60 km (Fig. 9.17b). A linear
For values of the normal traction less than or com- relationship with an intercept of 50 MPa and a
parable to the frictional strength, the rst term on slope equal to a coefcient of friction of about 0.6
the right-hand side makes a signicant contribu- ts much of the data (excluding materials such as
tion to the static friction. Under these conditions montmorillonite, vermiculite, and illite). For
it is not appropriate to equate the coefcient of normal traction magnitudes greater than a few
friction and the static friction. If the normal trac- hundred MPa, the coefcient of friction is essen-
tion is much greater than the frictional strength, tially equal to the static friction (9.31).
the rst term becomes insignicant and the The frictional behavior of most rocks in the
coefcient of friction approaches the value of the ranges of normal traction specied in Fig. 9.17a and
static friction. Some laboratory values for the fric- b can be characterized as follows (Byerlee, 1978):
tional strength and the coefcient of friction
from (9.30) are reported in Table 9.5. |ts | 0.85tn, 5 tn 100 MPa
Laboratory data on the friction of rock surfaces |ts | 50 MPa 0.6tn, (9.32)
may be categorized into three broad classes based 100 tn 2000 MPa
354 BRITTLE BEHAVIOR
Load
accompanied by a zone of microcracking and
other grain-scale damage. One moment in this
steady-state process of fracture propagation is 1
illustrated schematically in Fig. 9.18c. The damage
zone (hachured region) develops in front of the
fracture tip and is left behind as a wake along
the side of the fracture surfaces. The spatial Rock
density of the microcracking is greater near the 0
macroscopic fracture surfaces (double hachured
region). Displacement
The phenomenon observed in these fracture (b)
propagation experiments is unlikely to occur in a
material with homogeneous strength. Rather, the 0
material would break at the point of greatest stress
concentration (in this case the notch tip) and the
fracture would propagate with little or no damage
zone. Because most rocks are highly hetero- 1
geneous at the grain scale, there are likely to be
many weak points where inelastic deformation
can proceed at local stress levels less than that at
the point of greatest stress concentration. The
microscopic deformation mechanisms active in
these damage zones include the growth of existing 2
microcracks within mineral grains (Fig. 9.19a), the
nucleation of microcracks at aws within grains
(Fig. 9.19b), the opening of grain boundaries (Fig.
9.19c), and the shearing of grain boundaries (Fig.
9.19c) (Friedman et al., 1972; Hoagland et al., 1973; (c)
Peck et al., 1985; Labuz et al., 1987). The roughness
of the fracture surfaces corresponds in part to the
grain size, because the macroscopic fracture prop-
agates both around and through individual
grains, seeking the path of least resistance.
Consider an opening fracture (Fig. 9.20) that Fig 9.18 Double cantilever beam testing procedure for
has not propagated through a laboratory speci- opening fracture propagation in Salem Limestone. (a) Plot of
men. The specimen is subjected to an applied load versus displacement (inset illustrates testing apparatus).
stress, a, that is less than the uniaxial tensile (b) Drawings of samples at different loads showing
strength, Tu. We would like to know the value of development of microcrack damage zone. (c) Schematic
applied stress required to initiate and continue illustration of steady-state fracture propagation with damage
zone. Reprinted from Hoagland et al. (1973) with permission
fracture propagation. Put another way, what is
of Springer-Verlag.
the resistance of a rock to fracture propagation?
356 BRITTLE BEHAVIOR
Microcrack y
syy
Flaw x
x
(c) (d)
Grain 1 2b
2a
Grain 2
12
a
yy a ,for x a,y 0 (9.33) for many loading congurations, the relationship
2x
between the local stress and the distance is the
Here, 2a is the fracture length and x is distance same: the local (crack tip) stress, yy, varies approx-
away from the tip in the plane of the fracture. imately as one over the square root of the distance
For many sample and fracture geometries, and from the tip. This is referred to as the near-tip stress
9.3 BRITTLE FAILURE IN A FIELD OF HOMOGENEOUS STRESS 357
distribution. As the fracture tip is approached, x Table 9.6. Fracture toughness (MPa m1/2).
becomes smaller and smaller relative to 2a, so yy
becomes larger and larger relative to a. This indi- Rock type From To
cates that the stress is highly concentrated near
the fracture tip and it is this stress concentration Granite 1.66 3.52
(Fig. 9.20b) that promotes fracture propagation. Basalt 0.99 3.75
Because the near-tip stress has the same depen- Quartzite 1.31 2.10
dence on distance for all opening fractures, we Marble 0.87 1.49
can redirect our attention to the other quantities Limestone 0.86 1.65
in (9.33) and group them into a new parameter Sandstone 0.34 2.66
called the stress intensity, KI:
Shale 0.17 2.61
K I a a (9.34)
values are for tests conducted at room temperature
Here the subscript I stands for mode I fracture and and atmospheric pressure.
indicates that the relative displacements of the The rule of thumb we take from this data set
fracture surfaces produce an opening motion. is: values of fracture toughness for common rock
Later we consider the other two modes of fracture types tested at room temperature and atmos-
in which the relative displacements of the frac- pheric pressure range from about 0.1 to 4.0 with a
ture surfaces produce a shearing motion. For representative value of 1.0 MPa m1/2. These con-
other fracture geometries and other arrange- cepts have found applications to hydraulic frac-
ments of the applied loads, equations comparable turing of wellbores (Rummel, 1987).
to (9.34) are tabulated in engineering handbooks
(Tada et al., 1973).
The stress intensity is a measure of the magni- 9.3 Brittle failure in a field of
tude of the local stresses anywhere in the fracture
tip region. Laboratory experiments have shown
homogeneous stress
that fracture propagation depends on this local
stress eld and that one can write a propagation Field observations and laboratory tests serve to
criterion in terms of the stress intensity reaching motivate the development of a theory for the
a critical value (Atkinson, 1987). The criterion is: failure of rock samples subject to stress states that
resolve both compression and shear across poten-
K IC K I (at propagation) (9.35)
tial fracture surfaces. It is anticipated that a
Here, K IC is called the critical stress intensity or theory for shear strength will help to explain the
fracture toughness. The units and dimensions of development of shear fractures in laboratory spec-
fracture toughness may be worked out from (9.34) imens and, perhaps, be useful in extrapolating
and are: laboratory data to faults in Earths crust. Given
such a theory, and the appropriate data from eld
fracture toughness [ ] MPa m12
(9.36) observations and laboratory experiments, the
KIC {} M L 12 T2
structural geologist should be in a position to esti-
Fracture toughness is a property that measures mate the magnitude of the stresses at the time of
the resistance of a particular material to the prop- faulting. This would provide a sound physical
agation of a fracture. As such it should be inde- basis for interpreting the geologic history of
pendent of fracture size or geometry, but it may faulted rock masses.
depend on such things as the temperature,
conning pressure, and chemical environment. 9.3.1 Coulombs concept of failure in
A variety of laboratory procedures for measur- shear
ing fracture toughness have been devised (Atkinson Conceptually one might postulate that shear frac-
and Meredith, 1987), and representative values for turing is caused by shear stresses and that the sign
selected rock types are given in Table 9.6. These of the shear stress is irrelevant to the strength. For
358 BRITTLE BEHAVIOR
Because the normal stress usually is compressive Fig 9.22 Map of aftershocks following the 1979
(negative) and the pore pressure is a positive Homestead Valley earthquake and the Coulomb stress on
number, the addition of pore pressure mitigates planes parallel to the main fault at a depth of 3 km. Reprinted
the effect of the compression. In other words, the from King et al. (1994) with permission of the Seismological
Society of America.
greater the pore pressure the lesser the shear
stress required to induce shear fractures.
The Coulomb criterion (9.38) is used to dene
stress is viewed as the dependent variable and
the Coulomb stress, C:
would be calculated using the shear and normal
C |s | i n, stress acting on the potential shear fracture
(9.40) planes.
1 Tu and C S0
Two other important applications of the
The Coulomb stress may be used to compare dif- Coulomb stress (9.40) are to relate aftershock dis-
ferent potential shear fracture orientations in a tributions to stress changes after major earth-
deforming rock mass and assess which is closer to quakes, and to assess the likelihood of one
failure. These could be differently oriented planes earthquake event triggering another. For example,
at the same location subject to the same principal the spatial distribution of aftershocks following
stress state, or two planes at different locations the 1979 Homestead Valley earthquake (Fig. 9.22)
subject to different stress states. The potential have been shown to correlate with the distribution
fracture surfaces could be in different rock types of Coulomb stress change (Stein and Lisowski,
with different coefcients of internal friction and 1983). Here the contours represent equal values of
different pore pressures. In all cases the Coulomb C on planes parallel to the main fault at a depth
360 BRITTLE BEHAVIOR
Rearranging to solve for the critical Coulomb Table 9.7. Rock mechanics laboratory tests for
angles: the coefcient of internal friction.
1
C tan 1
2
1
i
(9.45) Rock Type i C ()
3
s
=
the calculated directions of intermediate princi-
1
s
pal stress (Fig. 9.24b) suggests that the Coulomb Stress states
Failure not possible
criterion provides a good correlation at many surface because of
locations. failure
must not be tensile. Then, separating the princi- (9.50) to account for pore uid pressure, the re-
pal stresses we nd: lationship remains linear and compares favorably
to data (Fig. 9.13b) from triaxial compression tests
1[(1 2i)12 i] 3[(1 2i)12 i] 0 (9.52) on Berea Sandstone (Handin et al., 1963).
Rearranging (9.50) to put it in a similar form:
1[(1 2i)12 i] 3[(1 2i)12 i] 2S0 9.4 Brittle failure in a field of
(9.53) heterogeneous stress
These two expressions must coincide at the point
in stress space where the failure surface and the The Coulomb criterion, reviewed in the previous
restriction on the stress state are both satised. section, asserts that solids subject to a homoge-
Eliminating 3 by adding (9.52) and (9.53), and neous stress state lose their load-carrying capacity
then substituting for 1 in (9.52) we nd: when a certain combination of shear and normal
stress acting on potential shear fractures reaches
3 S0[(1 2i)12 i] 12 Cu (9.54) a critical value. The Coulomb criterion is cali-
brated using laboratory measurements of the uni-
Thus, the restriction n 0 on the potential shear axial compressive strength and the coefcient of
fracture truncates the failure surface where 3 is internal friction. Although this criterion provides
equal to the negative of half the uniaxial compres- a reasonable t to some laboratory data and has
sive strength (Fig. 9.25). For greater values of 3, it been applied to numerous practical problems
has been suggested that the uniaxial tensile with some degree of success, it does not explicitly
strength would limit the stress state such that address development of the fractures themselves
(Paul, 1961): in the heterogeneous stress state induced by the
fractures. Here we introduce methods for calcu-
1 Tu,for 1 3 12 Cu (9.55) lating the perturbed stress eld around holes and
cracks and use these to investigate brittle failure
In this way the intersection of the failure surface in a eld of heterogeneous stress (Anderson, 1951;
with the (1, 3)-plane is composed of two straight Hubbert, 1951; Chinnery, 1961, 1963, 1966; Cooke
lines. The lightly shaded region (Fig. 9.25) below and Underwood, 2001; Bourne, 2003).
and to the right of these lines is off limits because
of shear or tensile failure. 9.4.1 The boundary value problem of
A number of laboratory studies have compared C. E. Inglis
data on the stress state at failure during triaxial One of the most cited articles in the literature of
tests to the Coulomb criterion. These studies fracture mechanics was presented by C. E. Inglis in
include those that focus on the effects of 1913 to the Royal Institute of Naval Architects in
conning pressure, temperature, deformation England (Inglis, 1913). The title of his article,
rate, and pore pressure on the strength of a wide Stresses in a plate due to the presence of cracks
variety of rocks (Handin and Hager, 1957, 1958; and sharp corners, might seem somewhat
Handin et al., 1963). As an example, we return to abstract for a group of naval architects, but one of
the data selected from a study of the mechanical the principal causes of ship disasters in those days
properties of sedimentary rocks from Tertiary was the growth of fractures in plates making up
basins of Japan (Hoshino et al., 1972). Recall that in the hulls of ships. To address this problem Inglis
this study the rocks were deformed in triaxial solved the elastic boundary value problem for an
compression tests at room temperature with no elliptical hole with major diameter, 2a, and minor
pore pressure. When plotted in principal stress diameter, 2b, in an elastic plate (Fig. 9.26).
space (Fig. 9.12) these data approximate a linear At the time Inglis took up this problem, engi-
relationship of the form (9.50). Using the effective neers knew that holes (e.g. those cut in deck plates
principal stresses, 1 1 Pp, 3 3 Pp, in to make hatches) would alter the local state of
9.4 BRITTLE FAILURE IN A FIELD OF HETEROGENEOUS STRESS 365
r
syy (app) ~ syy pro rata was not a good rule of thumb for these
calculations.
Inglis solved this problem analytically and to
Elastic plate help make the mathematical problem tractable
syy (x = a, y = 0)
y he considered a plate that is innite in extent.
After solving the boundary value problem for the
syy (ave) innite plate it is possible to show that this solu-
tion closely approximates a nite plate that is only
2b x several times bigger than the major diameter of
ts tn the hole. For a remote boundary condition on the
plate, he specied that a uniform normal stress,
2a ryy, acts perpendicular to the major diameter:
BC: as (x2 y2)12 ,xx 0,
xy 0,yy ryy (9.57)
be the applied stress times the cross-sectional area Note that the local stress is proportional to the
of the plate, WD. At the mid-section of the plate, y remote stress, ryy, and is related to the shape of
0, the cross-sectional area is reduced to (W 2a)D. the hole through the ratio of major to minor
The engineers did not know how to calculate the diameters. Comparing (9.56) to (9.59) it is clear
stress acting across the mid-section, so they simply that the average stress can be a very poor estimate
dealt with the average value, yy (ave). Equilibrium of the maximum stress. Inglis related this result
of forces requires that the same total force is trans- to the fracture of ship hulls by noting:
mitted across any section of the plate, so
yy(app) WD yy(ave) (W 2a)D. Solving for the When a/b 1,000, the tension at x a, y 0 is 2,001
average stress: times the mean tension. The ellipse in this latter case
would appear as a ne crack, and a very small pull
yy(ave) yy(app) W
W 2a
, y 0 (9.56)
applied to the plate across the crack would set up a
tension at the ends sufcient to start a tear in the
material (Inglis, 1913).
The Latin phrase used by the engineers to describe
this method, and perhaps to obscure their lack This solution to the elastic boundary value
of an exact calculation, was pro rata. In other problem provided, for the rst time, a quantita-
words the ratio of average stress to applied stress tive prediction of the great increase in stress at
was the same as the ratio of plate width to notches, corners, and crack tips and thereby led to
reduced plate width. Inglis was able to show that an understanding of why such geometric features
366 BRITTLE BEHAVIOR
posed a great danger for engineering structures. cracks with which Mr. Inglis paper deals are math-
According to (9.59) a remotely applied stress that ematical and not real ones (Inglis, 1913).
is well within the normal safety factors for hull
Despite such vacuous criticism, the solution to
plates could be amplied locally by the presence
the boundary value problem derived by Inglis
of a crack to values well above that predicted by
became a cornerstone of one of the most success-
(9.56), and perhaps well above the strength of the
ful engineering endeavors of the twentieth
material.
century, the development of engineering fracture
For practical purposes (9.59) suggests that the
mechanics (Lawn and Wilshaw, 1975; Kanninen
concentration of tensile stress near the ends of
and Popelar, 1985; Anderson, 1995).
cracks in engineering structures can locally break
the material apart and create a propagating frac-
9.4.2 Fracture initiation at Griffith flaws
ture that might lead to catastrophic failure of the
A. A. Grifth addressed the concept of fracture in-
structure. A plausible propagation criterion
itiation in solids in two articles in the early 1920s.
would be that the stress at the fracture tip reaches
He is best known for his analysis of fractures in
the uniaxial tensile strength, Tu, of the material at
terms of a macroscopic energy balance, but in his
that point: yy (x a, y 0) Tu. Substituting
second paper he addressed the local stress con-
this condition into (9.59) and rearranging to solve
centration near the ends of crack-like holes and
for the remote stress at propagation, we have
related this to fracture initiation (Grifth, 1921,
b 1924). Perhaps the most important contribution
ryy Tu , for b a (9.60)
2a of Grifths research was his demonstration that
solids contain sub-microscopic aws that act to
By measuring the uniaxial tensile strength of a increase the tensile stress locally and thereby in-
material in the laboratory (Table 9.1) and knowing itiate tensile fracture growth. The presence of
the geometry of any crack that might exist in the unseen aws was a non-intuitive concept, consid-
structure, an engineer could use (9.60) to place ering that Grifth was working on laboratory
limits on the applied stress in order to prevent glass specimens that appeared nearly awless
further fracture propagation. Although a vast under the microscope.
improvement over the pro rata method, this Grifth began his research with the assertion
approach is somewhat impractical because of the that the analysis of C. E. Inglis (Inglis, 1913) for
difculty of measuring the minor axis of cracks. the state of stress around holes and notches in
This difculty may be addressed by characterizing elastic plates could be used to estimate the intrin-
the stress concentration in terms of the stress sic tensile strength of solids. By intrinsic strength
intensity factor (9.33), which is only a function of Grifth meant the greatest stress that the solid
crack length. could endure before rupture of the bonds
Although the paper of Inglis contains a care- between the atoms, ions, or molecules that hold
fully constructed analysis and a thorough investi- the solid together. Apparently, he envisioned the
gation of the results in practical terms, he met bonds starting to break in the vicinity of a stress
with considerable resistance to his new concepts. concentration created by a aw, so the breaking
A prominent engineer wrote in the proceedings of bonds formed a more-or-less planar displace-
the Royal Institute: ment discontinuity, perhaps along a lattice plane
I regret to say that, although I have studied them with if the solid were crystalline. Grifth used the
Professor Loves book on elasticity at my elbow, I nd stress (9.59) at the end of a very eccentric ellipti-
that the notations, which are doubtless very con- cal hole (b a) in an elastic plate (Fig. 9.26) to
venient for the discussion of vibrations and other com- model the aw and used the radius of curvature,
plicated subjects, are so elaborate that I have not been rc b2/a, to characterize the local shape of the
able to do more than apply some simple tests to the aw:
results, and do not now feel satised that the math-
a 12
ematical deductions are fair representations of practi-
yy (x a, y 0) 2 ryy , for rc a (9.61)
cal cases; in other words, the holes, corners, and rc
9.4 BRITTLE FAILURE IN A FIELD OF HETEROGENEOUS STRESS 367
Grifth suggested that the appropriate radius of large compared to molecular dimensions. These
curvature in (9.61) would be the intermolecular Grifth aws serve to concentrate the stress and
spacing for the solid, so rc 5 1010 m, and he thereby weaken the material so that the macro-
determined the value of the product ryya12 by scopic measures of strength are much less than
experimentation in order to estimate the intrinsic the intrinsic strength. Fossils, clasts of different
tensile strength. lithologies, pore cavities, grain boundaries, and
Grifth introduced a small scratch of known microcracks are some of the many possible aws
length, 2a, onto the surface of cylindrical tubes or that can provide the stress concentration neces-
spherical bulbs of glass with a glass cutter. Then sary to initiate fractures in rock.
he pressurized the vessels with gas until they
burst, and recorded the ultimate pressure. The 9.4.3 Griffiths criteria for brittle failure
relationship between the gas pressure and the In the second of his two classic papers on fracture
remote stress component acting across the Grifth explored the problem of fracture initia-
scratch, ryy, was determined from the boundary tion and failure under a biaxial remote stress,
value problem for internal pressure in elastic using the solution of Inglis for an elliptical cavity
tubes and spheres. Grifth discovered that the in an elastic material (Inglis, 1913; Grifth, 1924).
product, ryya12 2.63 105 N m32 was nearly Here we review Grifths analysis, but, to make it
constant for the scratch lengths he tested, from 4 more relevant to geological applications, we
to 23 mm. Substituting into (9.61) Grifth evalu- include uid pressure acting on the cavity walls.
ated the stress at the tip of the incipient fracture The limiting case of an elliptical cavity with very
as 2.3 104 MPa and suggested this was an esti- great aspect ratio, a/b 1, is used to model a
mate for the intrinsic strength, Ti. However, this crack-like aw. Flaws of all possible orientations
estimate is very great relative to measured uni- relative to the remote principal stress axes are
axial tensile strengths which typically are of the examined, and the one with the greatest induced
order 101 to 102 MPa (Table 9.1). tensile stress is identied as the most danger-
To understand this puzzling result, Grifth ous. When this tensile stress equals the uniaxial
divided the estimated intrinsic strength by tensile strength the initiation of a tensile fracture
Youngs modulus for glass, E 6.2 104 MPa, to is predicted and this condition is assumed to be
estimate a strain of 0.37 at failure. He realized that coincident with failure of the material. This pro-
Hookes Law probably would not hold at these cedure is used to dene a failure surface in prin-
levels of strain and he knew that the concepts of cipal stress space and that surface is compared to
an elastic continuum were on shaky ground at the laboratory data.
molecular scale. Therefore he inferred that the The boundary conditions at an innite dis-
intrinsic tensile strength would be somewhat less tance from the elliptical hole (Fig. 9.27) consist of
than the value calculated above. Grifth sug- uniformly distributed normal stresses that are the
gested that a reasonable order-of-magnitude value principal stresses, 1 and 3, directed along the x-
was: and y-axes, respectively:
E BC: at x2 y2 ,xx 1,
yy(x a, y 0) Ti 6 103 MPa (9.62)
10
xy 0,yy 3 (9.63)
This value is greater than the uniaxial tensile
Recall from the discussion of elliptical coordi-
strengths for glass specimens, the strongest of
nates in Chapter 2 that constant denes a
which Grifth measured at Tu 1.7 102 MPa, and
family of confocal ellipses with 0 being the par-
many specimens had uniaxial tensile strengths as
ticular ellipse designated as the surface of the
much as two orders of magnitude less than the
hole. The boundary conditions there consist of a
intrinsic strength. These results inspired the con-
uniform normal traction of magnitude P and no
clusion mentioned above: brittle solids such as
shear traction:
glass must contain a myriad of aws, too small to
be detected by the optical microscope, but very BC: on 0, tn P, ts 0 (9.64)
368 BRITTLE BEHAVIOR
s3
y
t P 2a
b
1 C
2a
b
1
(9.66)
9.5 Fracture propagation and fault 9.5.1 Propagation of joints, veins, and
dikes
growth Fractures are idealized as two surfaces with
mirror image geometry that are in contact in
Criteria for failure in a homogeneous stress the initial unloaded state and are bounded in
state, such as the Coulomb criterion (9.50), extent by a common curve called the tipline.
inform us about the limiting stress conditions Sufciently close to the tipline the shape of this
that are obtained at failure, but do not explicitly curve is approximately straight and the surfaces
include the structure (e.g. a shear fracture) that are approximately planar. We adopt a Cartesian
is associated with the process of failure. Criteria coordinate system with the y-axis normal to the
for failure in a heterogeneous stress state, such plane of the surfaces and the z-axis parallel to the
as the Grifth criteria, (9.69) and (9.70), include tipline (Fig. 9.30). Upon loading of the elastic
the stress concentrating structure and describe body the fracture surfaces move relative to one
the initiation of cracking, but do not address the another and this motion may be classied
propagation of the crack or other possible events according to the coordinate directions we have
and mechanisms that may be involved in the chosen (Kanninen and Popelar, 1985). Relative
evolution of structures such as joints and faults. motion in the y-coordinate direction is referred
By combining solutions to elastic boundary value to as Mode I or the opening mode; relative motion
problems with principles of fracture mechanics in the x-coordinate direction is mode II or the
one can explore the processes of fracturing and sliding mode; and relative motion in the z-coordi-
faulting in rock from the initiation stage through nate direction is mode III or the tearing mode.
a stage of propagation or development to the Mode I is associated with geologic structures
eventual cessation of tectonic activity as the such as joints, veins, and dikes, whereas modes II
structure attains the size and conguration we and III are associated with shear fractures and
observe in exposure today. faults. Modes II and III both involve a shearing
The evolution of structures in brittle rock is a motion of the surfaces with the former being per-
large topic with interesting examples that are too pendicular to the tipline and the latter being par-
numerous to be described in detail here. Instead allel to the tipline. This classication may appear
we focus on a few examples that are meant to arbitrary, but because of the symmetry of the
provide a summary of the methodology of inves- fracture tip the elastic stress elds near the tip
tigation and some insight concerning the kinds of are uniquely distinguished by these modes, and
results one might expect. The rst example is the each mode is associated with a different style of
propagation of opening fractures such as joints, propagation.
veins, and dikes. Earlier in this chapter we There are many solutions in the literature for
described the loading conditions necessary for linear elastic problems that involve fractures (Sih,
propagation in terms of the stress intensity at the 1973; Tada et al., 1973). A common feature of these
fracture tip reaching the fracture toughness of is that the stress components in the vicinity of the
the rock (9.35). Here we show how the loading con- fracture tip are distributed in a way that depends
ditions determine the path that the fracture largely upon the fracture mode (Irwin, 1957;
follows as it propagates. The second example is the Williams, 1957). In contrast, the magnitudes of
growth of faults in granite and in sandstone. In the stress components depend upon the fracture
the granite it is the propagation of opening frac- geometry away from the tipline and the loading
tures to link adjacent sheared joints that enables conditions. This interesting and perhaps non-
the faults to grow in length. In the sandstone it is intuitive fact comes about because the near-tip
the clustering of deformation bands that enables stress distributions are dominated by the local
the fault to grow in thickness. Through these geometry of the fracture tips, which are taken as
examples of fault growth we make the point that identical for all modes, and by the relative motion
faulting is a process that can involve several dif- of the fracture surfaces, which are uniquely dis-
ferent physical mechanisms. tinguished by the modes (Fig. 9.30).
372 BRITTLE BEHAVIOR
yz
K III
sin (2)
(2r)12 cos (2) (9.74)
polar coordinate system (r, ) with origin at the tip The term KIII is the mode III stress intensity factor.
(Fig. 9.30). What we mean by the vicinity of the One may summarize the equations for the near-
fracture tip (Fig. 9.20) is a distance r 0.01a, tip stress components as follows:
where 2a is a characteristic length of the fracture ij [KI fij() KIIgij() KIIIhij()](2r)12 (9.75)
(Pollard and Segall, 1987). When analytical solu-
tions for the plane strain stress components are Here the indices i and j range over x, y, and z. This
expanded about the fracture tip and higher-order clearly demonstrates the separation of stress mag-
terms in powers of r are eliminated, what remains nitude (intensity) from stress distribution. Note
is proportional to r 1/2 and contains trigonomet- that several of the trigonometric functions in
ric functions of of order unity. For pure mode I (9.75) are zero.
fracture tips, the stress components in the (x, y)- Because the stress intensity and the inverse
plane are approximated as: square root of the radial distance are common to
all stress components (9.75), the components are
xx cos (2)[1 sin (2) sin (32)] most easily compared by plotting the trigonomet-
yy KI cos (2)[1 sin (2) sin (32)] ric functions over the range , from
xy (2r)12 sin (2) cos (2) cos (32) one fracture surface around to the other (Fig.
(9.72) 9.31). Positive values are associated with tensile
K I is the mode I stress intensity, which has units and normal stresses. Some instructive results con-
dimensions given in (9.36) and has a value that cerning geologic structures are found in these
depends upon the fracture geometry and loading plots. For mode I the function fyy is proportional to
conditions. Stress intensity factors are tabulated the normal stress component, yy, acting perpen-
in engineering handbooks (Tada et al., 1973). dicular to the plane of the opening fracture. One
For pure mode II fracture tips, the components might expect this stress component to have a
are approximated as: maximum value just ahead of the fracture tip in
the plane of the fracture, 0. Instead, yy has
xx sin (2)[2 cos (2) cos (32)]
two equal maxima to either side of the fracture
yy KII sin (2)[cos (2) cos (32)]
(2r)12 cos (2)[1 sin (2) sin (32)] plane (Fig. 9.31a). These maxima may, in part,
xy explain the cloud of microcracks that develop
(9.73) around the opening fracture tip in the laboratory
9.5 FRACTURE PROPAGATION AND FAULT GROWTH 373
Cracks
Fissure
Cracks
Fig 9.32 Map of open fissures and normal faults in vicinity of the fracture where it is perturbed, and
Keanakakoi ash deposit associated with the 1974 fissure this perturbation is greatest in the near-tip stress
eruption on the southwestern rift of Kilauea Volcano. eld as dened in (9.75). Before the fracture devel-
Numbers are horizontal (H) and vertical (V) separations in ops, the stress eld is uniform throughout the
millimetres across the cracks. Reprinted from Pollard et al. body and equal to the remote stress. The symme-
(1983) with permission of Elsevier.
try of the opening displacements for straight
joints and dikes suggests that these fractures form
is consistent with the formations of veins and solu- symmetrically with respect to the remote princi-
tion surfaces there. Because the graph in Fig. 9.31b pal stress axes. The greatest mode I stress intensity
is for right-lateral slip, one must change the sign is achieved if the fracture is perpendicular to the
of gxx to apply the result to the Languedoc faults. least compressive (most tensile) remote principal
The traces of many joints, veins, and dikes in stress, so that is the preferred orientation.
outcrop are remarkably straight (Fig. 2.12) or Furthermore, the near-tip stress eld tends to
gently curved (Chapter 9, frontispiece). In a rock guide the fracture tip into an orientation that is
mass that is isotropic with respect to fracture symmetric with the remote principal stress axes
toughness, the direction of propagation for these (Cotterell and Rice, 1980).
mode I fractures is determined by the stress eld, For the pure opening fracture (Fig. 9.33a) there
and the systematic traces suggest that the princi- is no resolved remote shear stress on the plane of
pal stress axes were uniformly oriented or the fracture. For the near-tip stress eld (9.72), we
smoothly varying on the length scale of the frac- have K I 0, but fyx 0 at 0 (Fig. 9.31a) so the
tures. Here we distinguish two regions, one far near-tip shear stress yx 0 on the next increment
from the fracture and the other very near the frac- of growth in the plane of the fracture. Under these
ture tip relative to the in-plane fracture length. conditions the mode I fracture is predicted to
We describe these as the remote stress eld and propagate along a straight path. However, if this
the near-tip stress eld, respectively. The remote fracture propagates straight into a region with
stress eld is thought of as uniform, except in the differently oriented remote principal stress axes,
9.5 FRACTURE PROPAGATION AND FAULT GROWTH 375
No Resolved Resolved
resolved mode II mode III
shear shear shear
(a) (b)
(c) (d)
(e) (f)
DB
0 1 2
m
Fig 9.38 Schematic illustration of development of faults in
porous sandstones of the San Rafael Desert (Aydin, 1978; faults with oblique fractures (Martel et al., 1988;
Aydin and Johnson, 1978). (a) Single shearing deformation Martel and Pollard, 1989). These fault zones typi-
band (DB) offsets marker horizon by a few millimeters. (b) cally are 0.5 to 3 m in width, reecting the spacing
Deformation bands cluster into a zone with offset of several inherited from the original joint set. The most
centimeters. (c) A slip surface develops on one margin of a
prominent and earliest formed fractures within
zone with offset of several meters. Reprinted from Davatzes
the zone strike at an acute counterclockwise
and Aydin (2003) with permission of Elsevier.
angle to the two bounding surfaces of the zone
and do not cross-cut those boundaries (Fig. 9.37e).
on one surface and compressive on the other. The Left-lateral offset of older structures is localized
splay fractures extend no more than a few meters on the former faults that bound these zones and
from the fault surfaces and near their ends strike can be as great as 10 m. In contrast to the
from 15 to 60 counterclockwise from the trend of mylonitic textures found in the faults (Fig. 9.37c),
the faults. the same epidote, chlorite, and quartz assemblage
Where two echelon fault segments are in the boundary faults is characterized by cata-
arranged with a left step (looking toward the tip, clastic textures. The fault zones can be up to 1 km
the next segment is to the left), splay fractures in length and are composed of segments a few
emanating from both segments may link to the tens of meters in length joined end-to-end at steps
adjacent segment and some may open to form and bends. This segmentation reects the geome-
rhomb-shaped cavities up to 10 cm wide (Fig. try of the earlier formed joint set and its domains.
9.37d). These extensional steps range from a few The granitic rock outside the two bounding sur-
centimeters to approximately 1 m in width faces of the fault zone is fractured near the steps
(normal distance between the segments) and are and bends, but elsewhere the fracturing is strictly
up to 2 m in length (overlap of the segments). The conned to the zone. Many of the internal frac-
slip on fault segments may be transferred to adja- tures show evidence only for opening but some
cent segments in a succession along strike by this also are sheared. Right-lateral deformation was
linkage mechanism such that faults more than accommodated on kink bands (Davies and
100 m in trace length are formed. The length of Pollard, 1986). The nal stage identied in the
faults apparently increased not by shear fracture development of the faults is the side-to-side
propagation, but rather by the end-to-end linkage linkage of adjacent fault zones and faults with
of echelon segments through opening splay frac- oblique fractures to form a compound fault zone
tures in extensional steps. At some locations near (Fig. 9.37f) about 10 m in width and several kilo-
the contact with the younger Mono Creek Granite, meters long (Martel, 1990). These zones offset
right steps display well-developed ductile fabrics older structures by as much as 100 m in a left-
with a foliation oriented oblique to the fault seg- lateral sense.
ments and these also may serve to transfer slip The second example considers fault develop-
(Brgmann and Pollard, 1992, 1994). ment in porous sandstone (Aydin, 1978; Aydin and
The next stage in the development of faults in Johnson, 1978). These faults were identied,
the Lake Edison Granodiorite (Fig. 9.36c) involved mapped, and described from outcrops of Entrada
the side-to-side linkage of parallel and adjacent and Navajo Sandstones in the San Rafael Desert of
9.5 FRACTURE PROPAGATION AND FAULT GROWTH 381
(a) (b)
zone itself but locally vary in orientation (Fig. together. It is conceivable that the slip surfaces in
9.39b). The traces of bands are nearly straight on sandstone propagate in this way, but they require
cross sections parallel to the direction of offset, the previous development of the zone of defor-
but form a wavy pattern on cross sections perpen- mation bands in order to nucleate.
dicular to the offset. Lenses of sandstone may be The development of faults in granitic rock is
preserved between wavy bands but the bands not expected to conform to the example presented
rarely appear to cross one another. The thickness here unless a pre-existing set of weak surfaces is
of a zone increases simply by the addition of defor- present. If such anisotropy is absent, or if more
mation bands. The relative shearing displacement than one set exists, we would anticipate a different
across a zone is the sum of that across the outcome. For porous sandstone the sequence from
members and may be as great as 25 to 30 cm for a deformation band to slip surface described above
zone containing 100 bands. is not the only possibility. For example, in the pres-
The third stage in the development of these ence of a set of pre-existing joints the mechanisms
faults is the localization of discrete surfaces (Fig. involved in the evolution of faults in sandstone is
9.38c) with slickenlines and striations indicating quite different (Flodin and Aydin, 2004; Meyers
slip. These slip surfaces are found on the margins and Aydin, 2004). Furthermore, for a different
of thick zones of deformation bands (Fig. 9.39c) lithology, such as interbedded limestone and
and they accommodate several decimeters to shale, the mechanisms are different than those
several meters of offset. There is a reduction in described for granite or sandstone (Peacock, 1991;
grain size and porosity within a few millimeters Peacock and Sanderson, 1991, 1994; Willemse et al.,
of the slip surfaces and the surfaces themselves 1997; Cooke, 1997). Some phenomena described
may be highly polished relative to the sandstone. here may be explained using quasi-static elastic
Investigations at other localities suggest that slip models; others may require solutions for dynamic
surfaces nucleate in small patches which link to elastic problems (Rice, 1980; Poliakov et al., 2002).
form an anastomosing network (Shipton and
Cowie, 2001).
This review of the physical mechanisms
9.6 Concluding remarks
involved in faulting of granitic rock in the Sierra
Nevada (Fig. 9.36) and of sandstone in the San The dominant behavior of rock in Earths upper
Rafael Desert (Fig. 9.38) serves to make the point crust is elastic and brittle at scales that range from
that faulting is a more complicated process than that of mineral grains to the crust itself. Brittle
the development of opening fractures such as the deformation is manifest in rock structures includ-
veins or dikes described in the previous section. ing microcracks, joints, veins, dikes, deformation
Multiple physical mechanisms are involved and bands, compaction bands, and faults. In the lab-
these may differ depending upon the rock type oratory, extension and shear fractures form as
and tectonic setting (Brgmann et al., 1994). The loading conditions reach the strength of samples.
fault zones in granite begin to form with rela- While these laboratory experiments do not neces-
tively little internal deformation except for a sarily reproduce the mechanisms responsible for
single set of oblique fractures. The fault zones in outcrop-scale structures, they do provide impor-
sandstone only form by the accumulation of tant data regarding strength, friction, and frac-
bands accounting for considerable internal defor- ture toughness. The strength of rock samples
mation. The thickness of zones in granite is deter- typically increases with increasing conning pres-
mined by the original spacing of joints whereas sure and decreases with increasing pore pressure.
the thickness of zones in sandstone is determined The fact that tensile strength is about one order of
by how many deformation bands cluster together. magnitude less than compressive strength helps
In neither case is the growth of these faults to rationalize the abundance of opening fractures
readily conceptualized in terms of the propaga- in the crust despite the nominal compressive
tion of a fracture tipline and the increase in area stress regime due to the overburden weight.
of two surfaces that formerly were bonded Grifths concept of stress concentration at aws
9.6 CONCLUDING REMARKS 383
and the tools of linear elastic fracture mechanics play important roles in the development of faults.
are usefully employed to understand the initia- These and other contributions represent a con-
tion, propagation, and pattern development of siderable achievement in understanding the
opening fractures. The Coulomb criterion for phenomena of faulting (Sibson, 1986, 1987, 1989),
shear fracture initiation has found many insight- and a comparable body of literature has advanced
ful applications to faulting and earthquake phe- our understanding of the mechanics of faulting
nomena. On the other hand, exposure studies of (Rudnicki, 1977, 1979, 1980; Rice, 1980, 1992), but
faults reveal that stress and material heterogen- the research to integrate these different views of
eity not incorporated into the Coulomb criterion faulting stands before us.
Chapter 10
Viscous flow
A
s the above quotation indicates, over one
hundred years ago, the rst director of the
United States Geological Survey, Clarence
King, hired a scientist to determine the viscosity of
rock. Searching through Kings account of the
survey of the fortieth parallel, we nd no mention
of Questions bearing directly on the viscosity of
rock masses . . . King did not write much, and so
it appears his thoughts on this subject and the (b)
observations that motivated them may have been
lost. King undoubtedly looked at rock masses that
contained fold structures like those shown on the
frontispiece for this chapter and in Fig. 10.1a,
which would have suggested to him that rocks
underwent continuous ow. Barus worked at ele-
vated temperature with steel and other sub-
stances, but not, to our knowledge, with rock. The
rst experimental studies in which rock was
deformed in a continuous manner were those of
Adams and Nicholson (Adams and Nicholson,
1901). Although interesting results were obtained,
research on the rheological behavior of rocks Fig 10.1 (a) Multi-layer folds (Moine Formation, Scotland).
apparently stopped until Griggs took it up in the (b) Multi-layer folds in alternating dark carbonatelight
1930s, working with technology developed by anhydrite, now gypsum, annual layer pairs (Castile Formation,
Bridgman, the 1946 Nobel Prize winner in physics Permian Basin, near Carlsbad, New Mexico). Photographs by
for his studies of the behavior of materials at high D. V. Wiltschko and K. Cruickshank.
pressure (Griggs, 1939). Examples of ordinary
viscous uid include cooking oils, lubricating ice, as in glaciers, and of salt, has been treated as
oils, molasses and honey, asphalt, molten glass, though these materials behave as viscous uids,
and magmas with modest volume fractions of and numerous researchers have modeled mantle
crystals and bubbles. Direct experience with some convection as ow in a viscous uid. The rst
of these materials reveals the large temperature detailed models of rock folding treat the buckling
dependence of their viscosity. of a viscous layer (Ramberg, 1960; Biot, 1961;
The use of the constitutive relations for an Chapple, 1968; Dietrich, 1969; Dieterich and
isotropic Newtonian viscous uid, dened below, Carter, 1969; Dieterich and Onat, 1969).
as a model for the rheological behavior of rock in Newton (1687) addressed the denition of a
mathematical models of rock deformation has a viscous uid in The Principia: The resistance
long history. Haskell (1937) used it for homo- which arises from the lack of lubricity in the parts
geneous crust and mantle in a model used to in- of a uid other things being equal is propor-
terpret measurements of glacial rebound. Cathles tional to the velocity by which the parts of the
(1975) treats the same phenomenon by represent- uid are being separated from each other. The
ing the crust and mantle as a series of layers with statement is a bit hard to interpret, but we
different viscosity, thus dealing approximately imagine Newton may have had two parallel plates
with the large effects of temperature and compo- containing a uid in mind, and the velocity is that
sition on rheological behavior. The slow ow of of one plate relative to the other in the direction
386 VISCOUS FLOW
velocity or stress on x. Ignoring the weight of the Dxx [xx (yy zz)][2G(1 )]
uid itself and that of the rigid plate, only the
stress component xy is non-zero. Then, in the Dyy [yy (zz xx)] [2G(1 )]
absence of accelerations and gravity, Cauchys Dzz [zz (xx yy)] [2G(1 )]
Laws of Motion (Chapter 7) reduce to the single (10.8)
stress equilibrium equation: Dyz yz 2G
Dxy xy 2 (10.7) The constitutive relations for a viscous uid
have the same form as those for an incompress-
Here is a material constant, the viscosity, with ible elastic solid, and the kinematic equations for
dimensions of M L1 T1. Other things being equal, the rate of deformation in terms of velocity gra-
the shear stress is proportional to the velocity gra- dients have the same form as the kinematic equa-
dient and the proportionality constant is twice tions for innitesimal strain in terms of
the Newtonian viscosity. displacement gradients. Thus, there is a com-
This experimental result is a special case of plete formal equivalence between the equations
the full set of constitutive relations for an governing the deformation of an elastic solid and
isotropic uid. A general linear relation between those governing the ow of a viscous uid. A
the innitesimal strain tensor and the stress solution to a problem for the deformation of an
tensor for an isotropic elastic solid was given in elastic body is associated with an equivalent
Chapter 8. The relations desired here may be solution to a problem for ow of a viscous body.
obtained from these by replacing the compo- This equivalence is described through the
nents of innitesimal strain with the compo- Correspondence Principle of Maurice Biot, and it
nents of the rate of deformation tensor, and extends to all viscoelastic substances, whose
writing G and for the shear modulus, G, and behavior combines elastic and viscous responses
Poissons ratio, : (Biot, 1965).
388 VISCOUS FLOW
10.3 Plane and antiplane flow The two constitutive relations are:
The stress distribution is statically determinate in All other deviatoric stress components are zero.
this case. That is, it only depends on the stress The form of (10.26) takes care of signs, since Dxy
boundary conditions and the equilibrium equa- and xy must have the same sign whatever the
tions. This means that we might consider any kind value of n 1. If, as here, both are positive, we may
of uid owing downhill provided its properties write, more simply:
only vary in the y-direction. Only Dxy is non-zero,
and for the uniform viscous uid: Dxy Bxy
n (10.27)
n1
another example, thrusting and folding of rock of 2Bh y
n1
vx c gh sin () n 1 (10.29)
the Pyrenees Mountains (Fig. 10.4a) is envisioned h
as a result of the upwelling of lower crust or
Here c is a constant of integration. Ignoring the pos-
mantle, here called the asthenolith (Choukroune
sibility of the uid sliding on its supporting plane,
and Seguret, 1973). If the lower crust is much less
we take the velocity to vanish at the base, y 0, so c
viscous because of its composition than the
equals the term in braces, and the nal result is:
underlying mantle, downslope motion of the
crust off such a high might be accomplished
n1
chiey by a glacier-like ow. If we apply the veloc- 2Bh y
vx [gh sin ()]n 1 1 (10.30)
ity eld (10.25) to an initially upright fold form, n1 h
390 VISCOUS FLOW
(a)
Arch re
imed ssu
ian pre
upward
about 40 km
(b)
y
y
x x
Fig 10.4 (a) Schematic diagram suggesting the mechanism
for thrusting and folding in the Pyrenees Mountains. If a mass of rock did ow in approximately
(b) Deformation of initial upright folds by glacier-like this fashion, and was cut by erosion, measure-
gravitational flow. Reprinted from Choukroune and Seguret ment of the shear strain distribution would allow
(1973) with permission of John Wiley & Sons. one to estimate n. Given the velocity eld, which
in this case is steady, we may determine the dis-
tribution of strain in the body for a given amount
The term in the rst braces is the velocity at the of surface displacement (Fig. 10.5b). Total dis-
surface of the layer, y h, since the second term placement is different for each prole in order to
equals 1 when y h. For given material properties, separate them. The deformed shapes of small
n and B, one may evaluate this velocity. Alterna- circles initially situated along the y/h-axis were
tively, we may measure glacier thicknesses, slopes, computed from a series of initial particle pos-
and surface velocities to estimate these proper- itions. Because of the gradient in shear rate, the
ties, in the spirit of a eld experiment. Notice that resulting loci only approximate ellipses. For
the surface velocity is proportional to hn 1. example, see the strongly deformed circle at the
An effect of the stress exponent, n, may be seen base of the sheet for n 10. Procedures in Chapter
by plotting the dimensionless ratio vx/vx(h) against 5 provide exact results: since the velocity gradi-
y/h for various n (Fig. 10.5a). As n increases, shear- ent tensor for a particle is constant as it moves
ing becomes strongly concentrated downward along a line at constant height y, the displace-
with increasing shear stress. For example, velocity ment gradient tensor may be computed by direct
proles through the thickness of a glacier may be integration.
measured by melting a vertical borehole and
placing a cable in it (Raymond, 1971; Meier et al., 10.3.3 Flow between approaching or
1974). After an interval of time, the cable may be separating rigid plates
re-occupied by sliding a melting device along it. The ow of uid outward from approaching rigid
Measurement of the inclination of the cable as a plates has been in the literature of structural
function of inextensible wire length from the geology for several decades (Jaeger, 1964a). It has
surface then allows the prole to be determined been applied to deformation in an orogenic belt
by integration from the surface position. between lithospheric plates whose relative motion
10.3 PLANE AND ANTIPLANE FLOW 391
(a) 1 (a)
0.8
0.6
y/h
n=1
0.4
3 (b)
x
0.2 10
V +V
100
0
0 0.2 0.4 0.6 0.8 1
vx(y)/vx(h)
y
0 2L
(b) 1
0.8
2h
0.6
n=1
y/h
3
0.4
10
Fig 10.6 (a) Conceptual model for deformation in an
0.2
orogenic belt between lithospheric plates with normal and
tangential relative motion; reprinted from Robin and Cruden
0 (1994) with permission of Elsevier. (b) Axes and parameters
0 0.2 0.4 0.6 0.6 1.0 1.2 for flow between approaching parallel plates (Jaeger, 1964).
x/h
vx(x,y) vx(x, y)
vy(x,y) vy(x, y) (10.32)
vx(x, y) vx(x, y)
(x, h) dx
1 problem for ow between approaching parallel
yy(ave) yy (10.46)
L plates.
0
A schematic illustration of the model
Substituting from the second of (10.45) we nd: conguration (Fig. 10.8) has a feature termed a
backstop. We do not wish to motivate and justify
yy(ave) 23 A L2 VL2h3 (10.47) this peculiar feature exhaustively here, but some
discussion seems necessary. An accretionary
We note that the approach velocity, V, of the wedge is composed of sediments and other ma-
plates scales with the applied stress and h3, terials scraping off a subducting plate, and the
whereas it is inversely proportional to the viscos- analogy has been drawn between this process and
ity and L2. pushing a mass of material across a horizontal or
Robin and Cruden (1994) use this ow eld to tilting surface as between a snowplow mounted at
simulate the deformation in a slab of material the front of a truck and the road surface. The rel-
lying between two lithospheric plates (Fig. 10.7a) ative motion is inverted here so the subducting
that have a component of normal motion as well plate moves, while the plow stays xed. The wedge
as tangential relative motion, i.e. the deformation of material on the subducting plate moves in a
termed transpression. In this case, the region direction opposite to that of subduction. The
shown in Fig. 10.7b is imagined as vertical, with backstop is identied with the plow and taken to
material in it extruding out from the top to form be vertical. No plow is present towards the rear of
a kind of mountain belt. The trans- part of the a natural accretionary wedge, so this must stand
motion is shearing across the region, but in and in as an approximation for some other feature.
out of the plane. The shearing motion may be Progressing from the toe (Fig. 10.8), an accre-
superposed on the -pression motion derived tionary wedge has a topographic culmination, to
here, since the governing equations are linear. the rear of which it loses topography. The reason
Although the resulting velocity eld is easy to for this is that the subducting plate piles up ma-
visualize, the distribution of strain and rotation terial as long as it exerts a shearing motion at the
in the uid slab is complicated; the results must base of the wedge, but once the plate surface con-
be computed and puzzled out. tacts lower, hotter material of negligible strength,
394 VISCOUS FLOW
30
6 km/my
(b) Austroalpine
Jura Mtns. Swiss foreland nappes
0 0
Depth (km)
e
on
30 WEDGE
rz
30
ea
sh
tro
Moho S 40
re
50
~5 km
/my
(c) L/4
Erosion A
we focus on the prominent seaward side of the
Backstop
The rst step is to drop the x-derivative in the subducting plate at the backstop are not provided.
second equation in (10.15) and write: Indeed, no conditions at a backstop are specied
in the formulation of this problem! Rigorously, we
yy y g (10.48)
cannot even speak of its presence because it has
Integrating this equation with the condition that not entered the formulation and analysis.
the normal stress vanish at the top of the wedge, At this point, a complete approximate solution
approximated as yy(x, h) 0, yields: for stress and velocity in the wedge is expressed in
terms of the unknown prole h(x,t). To obtain this,
yy g( y h),h h(x, t) (10.49) we write the relation for conservation of mass
between vertical surfaces at x and x dx for wedge
In the kinematic relations (10.12), derivatives of vx material of uniform and constant density:
occur in Dxy and Dxx, and that in Dxy is larger, by
h J
hypothesis, so Dxy Dxx. Discarding the derivative dx Jx(x) Jx(x dx) x dx (10.55)
t x
vy x in Dxy, we have:
Here Jx is the volume ux across a vertical surface
Dxy 12 (vx y) (10.50) per unit strike length, or:
h
From (10.14) we conclude that xy |xx yy|.
Using these approximations in the rst equilib-
Jx(x) vx(x, y)dy (10.56)
0
rium equation (10.15), the result (10.49) yields:
Substituting (10.54) with the condition vx(0) V
xy yy h into (10.56) and integrating:
xx g (10.51)
y x x x
gh3 h
Integrating, and using the requirement that the Jx Vh (10.57)
3 x
shear traction vanish at the upper surface,
approximated by xy(h) 0, we obtain: Because the wedge slopes toward its toe, h/x 0,
h the rst term is positive, signifying a ux toward
xy g ( y h) (10.52) the toe. The second negative term represents the
x
motion of the subducting plate relative to a xed
We may now use (10.14), (10.50), and (10.52) to backstop.
write: Substituting (10.57) into (10.55), yields a partial
vx g h differential equation in h:
( y h) (10.53)
y x
Integration gives:
h gh3 h
t x 3 x
Vh
(10.58)
vx
g h y2
x 2
hy vx(0) (10.54) We look for a steady-state solution, in which the
form of the wedge is unchanged, and set ht 0.
If the viscous wedge does not slip at its base, Integrating the resulting ordinary differential
vx(0) V . You may wonder how we are able to equation yields:
move the wedge along with our plow with this no-
gh3 h
slip condition. The notion is that the uid is Vh C (10.59)
3 x
scraped off right at the backstop, but the mass
of uid is not sliding over the surface as a whole. This equation satises the condition that the
As in the analysis of the ow between wedge has a nite width, L, that is h(L)0, if C 0.
approaching rigid plates (Section 10.3.3), end con- Factoring out an h, integrating again, and re-
ditions on the wedge are ignored. Thus, the applying this condition gives (Emerman and
details as to how the material is scraped off the Turcotte, 1983):
396 VISCOUS FLOW
(b)
10.4 Viscous flow in layers:
h(x)/h(0)
mullions and folds
h
9VL
g
1
x
L
(10.60) 10.4.1 Biharmonic equation and its
solution
The proles of the wedges in Fig. 10.9 have the In Chapter 8 we introduced the three components
form specied by (10.60), where h in both cases is of stress in the (x, y)-plane written as partial deriv-
taken as the total vertical height above the planar atives of the Airy stress function, (x,y):
base. The solution applies to a nite wedge of any
2 2 2
cross-sectional area A, where: xx 2
, xy ,yy 2 (10.63)
y xy x
L
4 13
243VL These expressions satisfy the two-dimensional
A h(x)dx (10.61)
64g stress equilibrium equations (10.15) and thus
0
The maximum wedge thickness at x 0 from replace three unknown stress components in a
(10.60) is: problem of plane ow with a single unknown
function from which they may be derived.
13
9VL The plane ow kinematic relations (10.12) are
h0 (10.62)
g three equations in two unknowns, vx and vy.
10.4 VISCOUS FLOW IN LAYERS: MULLIONS AND FOLDS 397
2 2 1 2 2 Mullion structure was the rst geological struc-
ture to be investigated using the nite-element
y2 x2 4 y2 x2
method (Dieterich and Onat, 1969). Here, we will
2
2
1 2
xy 2 xy
0 (10.66) learn something about its formation from a
simple analytical study. This structure, illustrated
at scales of centimeters and kilometers in
In the present case, we suppose that the viscosity, Fig. 10.10, consists of a lobe-and-cusp morphology
, is not a function of position so we obtain: of interfaces between different rock types and is
4 4 4 interpreted to have formed during interface-
2 2 2 4 0 (10.67) parallel shortening of large magnitude. Lobate
y 4 y x x
forms are developed in the stiffer rock with the
This is the biharmonic equation, often written in the softer, or less viscous, rock-lling cusps. One
condensed form: might suppose that the surfaces of the more
viscous rock layers were puckered, so as to
4 2(2) 0 (10.68)
undergo little surface-parallel shortening. If so,
Having reduced the general set of equations for one could lay out the surface trace in a straight
plane ow, (10.12)(10.15), to a single equation in line and use this to estimate the original length of
one unknown function, (x, y), the next step is to the interface. Then, as done for the folded vein in
set up some boundary value problems of interest Fig. 5.3, one could estimate the amount of bulk
and solve them. shortening.
398 VISCOUS FLOW
D(1)
xy 0 Dxy
(2) y (x, t) A(t) cos (t) x (10.72)
From (10.69) and the constitutive relations (10.14) Both the amplitude and wavenumber are func-
we have: tions of time, t. The interface in the case of mul-
lions shown in Fig. 10.10a is not periodic, nor does
xx yy) 41 ( xx yy) 42 Dxx
( (1) (2)
(10.72) describe the lobe-and-cusp asymmetry of
(10.70)
xy xy xy 0
(1) (2) the mullion interface. However, a sum of wave-
length components provides the possibility of
The stress components yy and xy are the same in extending the solution to cover these variations.
both half-spaces, as required by the continuity By considering a periodic interface form, we
of traction at the planar interface. The surface- isolate a region of interest that is a single wave-
parallel normal stress must be different unless length wide. The vertical planes at x L/2 may
the viscosities are the same. be treated as though they were the frictionless
Taking the coordinate origin at the interface, planar surfaces of two rigid platens, between
the velocity components, with the origin xed, are: which the two media with their sinusoidal inter-
v (1)
face shorten. Because the shear stress must vanish
x v x Dxxx
(2)
(10.71) on a mirror plane of symmetry, these planes have
v (1)
y v y Dyy y Dxx y
(2)
the property of a frictionless surface.
These satisfy the condition that adjacent particles The local slope with maximum at x L/4 of
in the two viscous uids remain neighbors as magnitude A is:
10.4 VISCOUS FLOW IN LAYERS: MULLIONS AND FOLDS 399
tan x ( A) sin x (10.73) ponents of stress (Fig. 10.11). We might also have
used the components with respect to the coordi-
Taking A 1, we use the approximations:
nate axes x and y. The present choice is more gen-
tan 33 ( A) sin x erally useful; for example, if the interface were
sin 36 ( A) sin x (10.74) frictionless, the shear component of the traction
would be required to vanish. It would then be nec-
cos 1 22 1
essary to write the velocity boundary conditions
We seek a solution accurate to terms proportional in terms of normal and tangential components
to A. In the derivation, if terms proportional to also, since slip on a frictionless interface would
(A)2 or higher are encountered, they are deleted. exclude a condition being set on the tangential
We cannot hope to simulate the development of a component of velocity.
large-amplitude mullion (Fig. 10.10a) with this Expressing the tractions in terms of the stress
approximation. Instead, we study mullion initia- components referred to the local coordinate axes,
tion and the inception of an asymmetric lobe-and- n and s:
cusp structure.
t (1)
n nn, t n nn
(1) (2) (2)
For the wavy interface, the basic-state solution (10.79)
no longer satises all boundary conditions at the s ns , t s ns
t (1) (1) (2) (2)
interface to the desired accuracy, as we now In terms of the stress components, the boundary
demonstrate by introduction of a correction term conditions are:
in the form of additional sets of velocity, stress,
nn (x, ) nn (x, )
(1) (2)
and rate of deformation components. The veloc- (10.80)
ity components in the two media are the sum of ns (x, ) ns (x, )
(1) (2)
The stress and rate of deformation components nn xx sin 2 2xy sin cos yy cos 2
are composed of similar sums. Suppose that the
interface is welded, so velocity components of ns (xx yy) sin cos xy( cos 2 sin 2)
adjacent particles across it are equal: (10.81)
~ ~(2)
x (x, ) v x (x, )
v (1) No correction to the basic-state solution is nec-
(10.77)
~
v (1) ) ~
v (2) essary if the interface is perfectly planar (A 0);
y (x, y (x, )
that is, the perturbing solution is identically zero.
The boundary conditions on the normal and Therefore we postulate that the components of
shear components of the traction on the interface stress and velocity in the perturbing solution are
are: themselves proportional to A. Since xy 0, we
t (1) have xy ~ xy and the term xy in the rst of
s (x, ) t s (x, )
(2)
(10.78) (10.82) can be discarded because it is a product of
t (1)
n (x, ) t n (x, )
(2)
two terms, both of which are proportional to A.
The tractions will be evaluated for each medium Applying these conditions, and the forms (10.74),
in terms of the interface geometry and the com- the stress boundary conditions (10.82) become:
400 VISCOUS FLOW
yy (x, ) yy (x, )
~
(1) ~ (2) The solution to (10.88) may be written:
~
(1) (1)
xy (x, ) ( xx yy)[(A) sin x] (10.83) F( y) 2{[a b(y1)]ey [c d(y 1)]ey}
~ xy (x, ) ( xx yy)[( A) sin x]
(2) (2)
(10.89)
The second relation in (10.83) is consistent with Here a, b, c, and d are arbitrary coefcients whose
the hypothesis that the perturbing ow is propor- values are xed by the boundary conditions. The
tional to A. Indeed, any information that the term ey is multiplied by a b to give the resulting
interface is not planar, and, hence, that a per- stress components a symmetric form. The factor
turbing ow must be present, comes only from 2 is added to make the expressions for the veloc-
this condition. ity components simpler in form. The stress com-
The second boundary condition in (10.83) ponents associated with the perturbing ow are
determines the x-dependence of the required solu- found from (10.63):
tion of the biharmonic equation (10.67). Two
~
xx 2{[a b(y 1)]ey
pieces of information are supplied. First, the con-
dition must be satised for all values of x on the [c d(y1)]e y}cosx
nearly planar interface y (x). This suggests the ~
yy 2{[a b(y1)]ey (10.90)
possibility that the function (x, y) and, from it, [c d(y 1)]ey }cos x
the components of stress and velocity, might be
~
xy 2[(a by)ey(c dy)e y] sin x
written in a separable form:
(x, y) F( y)G(x) (10.84) From the constitutive relations (10.14) and the
kinematic relations (10.12) for plane ow we have:
Since ~xy 2 xy, the second condition in ~x ~
v xx ~
yy
(10.83) could be satised if G(x) cos x, giving: [(a by)ey
x 4
(c dy)e y] cosx (10.91)
~
xy(x, ) (dFdy)y sin x (dFdy)y0 sin x
(10.85) Integrating (10.91) the x-component of velocity is:
must be distinct, there are eight arbitrary con- such particles are not dissolved away or eroded,
stants to be xed using boundary conditions. nor is new material added to the interface. Using
These constants are: a1, b1, c1, d1, a2, b2, c2, d2, the solution for the coefcients, (10.95) and
where subscripts 1 and 2 refer to the upper and (10.96), and the expressions for the velocity com-
lower half-spaces, respectively. At the interface ponents, (10.92) and (10.93), we may evaluate the
there are only four boundary conditions, (10.77) velocity vectors for a large number of particles on
and (10.83), thus indicating that we have not yet the interface, y Acos x, and increment their
specied four boundary conditions elsewhere. positions by multiplying the velocity vectors by a
Considering the y-component of velocity (10.93) in small quantity t. The velocity components in the
the upper half-space, y 0, notice that the lower layer are:
coefcients a1 and b1 multiply terms proportional
to ey, which grow in magnitude without limit as x Dxxx a2(1 y)e sin x
v (2) y
(10.97)
y increases. Since a relatively gentle waviness at v (2)
y Dxx y a2ye cos x
y
the interface cannot be expected to produce an This includes the basic state and the perturbing
ever-increasing ow away from the interface, we ows. If the viscosity of the lower medium is
take a1 b1 0. Considering the same situation in greater than that of the upper, 2 1, then a2 0
the lower half-space, we conclude that c2 d2 0. for shortening, Dxx 0. It is useful to re-cast the
The remaining four coefcients are xed by expressions (10.97) with (10.95) and (10.96) into a
application of the four boundary conditions at dimensionless form, using |Dxx|L as the charac-
the interface. Begin with the condition on the teristic velocity:
continuity of the shear stress, (10.86). Using
(10.90), and substituting y in the functions con-
tained in the expressions ~ ~ (2)
xy and xy , we obtain:
(1)
v (2)
x
|Dxx|L
x 1 1R
Sgn (Dxx)
L 1R 2 A
L
(2)
Recall our hypothesis that the perturbing ow is vy y 1 1R 2 A
Sgn (Dxx)
proportional to A. This can be the case only if the |Dxx|L L 1R L
coefcients themselves are proportional to this
quantity. Since (A) cos x~A, we expand the yeycos x
exponential functions to rst order in this quan- Here the dimensionless quantities are 2A/L A
tity: e 1 and e 1. Retaining only and R 2/1. If the half-spaces are undergoing
terms linearly proportional to A we nd shortening, Sgn(Dxx) 1, but for interface-
1c1 2a2 2(2 1)Dxx(A). The condition on parallel extension, Sgn(Dxx) 1.
continuity of vx, (10.77), treated in the same way, Figure 10.12a shows the perturbing part of the
yields c1 a2. Combining these equations we have: velocity eld in the lower, more viscous half-space
c1 a2 2[(1 2)(1 2)]Dxx A (10.95) for R 10. Notice how the perturbing ow is con-
centrated near the surface and has zero vertical
The remaining two boundary conditions on the velocity there. The horizontal ow away from the
normal component of stress and the vertical com- crest of the sinusoidal perturbation in surface
ponent of velocity yield: shape broadens the lobes and tightens the cusps
in the interface form. This is kinematically
c1 d1 0 a2 b2 (10.96)
amplied in the basic state of uniform shortening.
From (10.95), if the viscosities of the two media are Figure 10.12b shows contours of the perturbing
equal, the perturbing ow vanishes. part of the horizontal normal stress. A horizontal
We consider the deformation of the interface zero contour lies at y L/, below which this
by following particles on it. We postulate that stress component changes sign, from tensile to
402 VISCOUS FLOW
(a)
(a)
(b)
(b)
Fig 10.13 (a) Profiles from mullion model computed by
0.9 0.9
incrementing positions of particles. (b) Profiles computed
0.5 using the interface evolution equations.
0.5 0.5
present solution, such components are linearly lation for dA/dt. In fact, if the missing term is
independent, i.e. non-interacting. However, for a included, the 2 in the second equation must be
periodic perturbation, we may analyze how the cancelled so (10.105) over-estimates the rate at
incipient mullion structure evolves at larger which the lobe-and-cusp structure develops. The
amplitude and interface slope while keeping the more accurate result is shown in Fig. 10.13b,
mathematics manageable. A step in this direction which is computed using this method.
is to substitute the solution into the interface To improve upon the analysis of stress, veloc-
evolution equation (10.102). ity, morphological development, and strain
Imagine that the initially sinusoidal interface within mullion structures, one may appeal to
develops into a form better approximated by: numerical codes (Dieterich and Onat, 1969), or
obtain more accurate results by extending the
(x, t) A cos x Acos 2x (10.103) analysis used here to higher orders of approxi-
Here the initial amplitudes are A(0) A0 and mation (Johnson and Fletcher, 1994). The rela-
A(0) 0. The asymmetric lobe-and-cusp modi- tively simple method of analysis and results
cation to the sinusoidal form is represented by the provided here offers insight that helps us to
second term. When velocity components are substi- understand how the lobe-and-cusp morphology
tuted in (10.102), we retain contributions propor- develops. In essence, the cusps occur due to a
tional to cos 2x. In the expansion of v (2)
y (x, ), we
small compressive stress concentration offset by a
retain terms in a2, that is those proportional to stress reduction in the material that extends
(A)2. Substituting (10.103) and (10.98) into (10.102): outward into the less viscous uid (Fig. 10.12b).
We may follow the strain distribution as well as
dA
dt
cos x A sin x x
d
dt
dA
dt
cos 2x the interface morphology using the present
results, or to see to what extent a tendency to pre-
d serve arc length along the interface exists. Both
A sin 2 x 2x
dt the kinematics and the stress distribution
suggest that a variety of minor structures might
Dxx(A cos x A cos 2x) a2(A cos x) cos x
be formed close to the interface in a region of
(Dxxx a2 sin x)(A sin x2 Asin 2x) marked strain variation.
(1.104)
10.4.3 Folding of a single viscous layer
We nd, according to this computation, that A is
The method of analysis developed in the preced-
proportional to (A)2. The product a2(A) also is
ing section may be applied to the folding of a
proportional to (A)2. Since d dt Dxx, the
more viscous layer isolated in a less viscous
terms proportional to x on both sides cancel.
medium, giving insight into how folds such as
Then, since this relation must be satised at all x,
those in Fig. 10.1a and 10.14 are initiated. The
the terms in cos x and cos 2x on each side must
latter shows a train of folds in a single layer of
be equal, requiring:
coarse K-feldspar/quartz leucosome sandwiched
dAdt Dxx A between two layers of quartzplagioclasebiotite
dAdt Dxx A 2[(1 R) (1 R)]Dxx A( A)
gneiss of somewhat different composition. The
layer is irregular, because either it formed that
(10.105)
way, in part, or because it was possibly stretched
Here we use the relation cos 2x cos2x sin2x in an irregular fashion before it was then folded
and have substituted for a2. These are a coupled in layer-parallel shortening. Of particular interest
set of two equations in the shape of the interface is the rough regularity in the arc lengths between
expressed in terms of a primary sinusoidal form fold hinges. Other features, such as the tendency
and a rst-harmonic form with one-half the wave- to nd quartz lling in the tightly appressed
length. Because we have not obtained a complete hinges and the fold forms are also of interest. The
solution accurate to terms proportional to (A)2, lobe-and-cusp forms seen in the mullion struc-
there will generally be terms missing in the re- tures are also approximately seen in the forms of
404 VISCOUS FLOW
(a)
(b)
(a) z' A
In-phase
perturbation y
x
0 2h
z'' A
=
L
Fold form
(b)
+ (x, y) (x, y)
(L/2 + x, y) (L/2 x, y)
Pinch-and-swell
form
To exploit the symmetry seen in Fig. 10.16b, the mullion problem. However, we refer them to a
take the coordinate origin for the layer at its mid- coordinate system in which the x-coordinate is the
plane. The velocity vector is shown for four par- same as that for the layer, but the origin is moved
ticles in symmetric relation to each other. Vertical up an amount h to the mean interface position.
mirror planes of symmetry are present at at x 0, Then, the velocity components in the upper half-
L/2, and L/2, with centers of symmetry at x space are:
L/4 and y 0. The y-component of the velocity ~ x (c1 d1y)e
v (1) y sin x
vector for the perturbing ow in the layer then (10.113)
satises ~ v y(x, y) ~
v y(x, y), so ~v y is an even func- ~ y [c1 d1(y 1)]e
v (1) y cos x
tion of x. We accordingly use the general form The stress components in the upper half-space are:
(10.93). The center of symmetry requires
~
v y(L2 x, y) ~ v y(x, y). Using these conditions ~ xx 21[c1 d1(y 1)]e
(1) y cos x
~ v x(x, h) ~
~ v (1)
x (x, 0)
v y {a(ey ey) b[y(ey ey) (10.115)
~ ~
v y(x, h) v y (x, 0)
(1)
(ey ey)] }cos x (10.110)
Recalling that the origin for the upper half-space
Thus, ~ v y is an even function of y, and symmetry
is taken at the mean base of this medium to sim-
reduces the number of arbitrary coefcients from
plify algebraic expressions, these conditions yield:
four to two. The x-component of perturbing veloc-
ity is: a(eh eh) bh(eh eh) c1
a(eh eh) b[h(eh eh) (eh eh)]
~
v x [a(ey ey) by(ey ey)] sin x
(c1 d1) (10.116)
(10.111)
Continuity of the normal and shear tractions
The perturbing stress components in the layer are: requires:
~
yy(x, h) ~ yy (x, 0)
(1)
~
xx 2{a(ey ey) b[y(ey ey)
~
xy(x, h) 4Dxx( A) sin x
(ey ey)]} cos x
~ (1)(x, 0) 4 D ( A) sin x
xy 1 xx (10.117)
~
yy 2{a(ey ey) b[y(ey ey) (10.112)
These yield:
(ey ey)]} cos x
2{a(eh eh) b[h(eh eh) (eheh)])
~
xy 2[a(ey ey) by(ey ey)] sin x 21(c1 d1) 2[a(eh eh)
(10.118)
bh(eheh)] 4Dxx A
In the upper half-space, the perturbing velocity
and stress components are the same as those in 21c141DxxA
10.4 VISCOUS FLOW IN LAYERS: MULLIONS AND FOLDS 407
Sgn(Dxx)(2 + q)
R cosh h) (1 R)Dxx A
1/10
Here R 1 . Solving for the constants we
nd:
(a) 18 (b) 30
16
25
14
20
12
2 + qd
L d /H
15
10
10
8
5
6
4 0
2 1.5 1 0.5 0 2 1.5 1 0.5 0
log10(R) log10(R)
q2 / k2 2R
6
Rk
k (10.126)
Ld/H and qd are plotted in Fig. 10.18 from a numer-
ical evaluation of (10.125) and the approxima-
tions (10.127). The solutions for Ld/H and qd do not
Typically, only the rst two terms are given in the admit of values less than 2 for Ld/H, or of more
so-called thin-plate approximation (Biot, 1961), than moderately strong folding instability with
but the third term improves the approximation qd 10 for Ld/H less than about 10. Thus, the folds
(Fig. 10.17). The expression (10.126) sums two in Fig. 10.14 cannot be interpreted from these
effects. The part of the curve descending as L/H relations. It is necessary to follow amplication of
decreases is dominated by the term k2/6 in the fold components as layer thickening continues,
denominator, and is tied to the bending resis- and the value of L/H of a component decreases.
tance of the layer; the part of the curve descend- This is done by numerically integrating (10.124)
ing as L/H increases is dominated by the terms R using:
and is associated with resistance of the medium to 1
layer deection. At the dominant wavelength, the d ln ( A) Sgn(Dxx)(2 q)dt (10.128)
A
two effects are balanced to give a maximum
growth rate. with an initial value (0) A(0). The result for R
Using only the two terms of (10.126) we have: 1/20 at several values of layer thickening is shown
10.4 VISCOUS FLOW IN LAYERS: MULLIONS AND FOLDS 409
(a)
20
102
18
Amplification, l(t)A(t)/l(0)A(0)
16
14
Frequency
12
101 10
8
6
4
2
100
100 101 102 0
log10 (L/H) 2 3 4 5 6 7 8 9 10
Larc/H
(b)
Fig 10.19 Amplification of A as a function of L/H with 50
layer thickening H/H0 from 1.2 to 2.4 in increments of 0.2.
Trajectories of individual components are for initial values 45
L/H 2, 4, 6, . . . , 18, 20, 30, 40, 60, 80, and 100.
40
Cumulative number
35
in Fig. 10.19. The value Lmax/H of the component
30
with the maximum amplication is Ld/H at H/H0
1.1 and 6.5 at H/H0 2. Curves of amplication 25
versus L/H for individual fold components are the
20
lines rising to the left.
The above relations allow interpretation of 15
data collected from fold trains of the sort shown in 10
Fig. 10.14, if the layer and surrounding medium
may be approximated as uniform viscous uids. 5
The data set for this prole and others from the 0
same fold train is presented as a histogram of fold 100 Larc/H 101
arc length to thickness ratio and also a cumulative
frequency distribution in Fig. 10.20. An interpreta- Fig 10.20 Frequency distribution of Larc/H for fold train, an
tion of the data to estimate the viscosity ratio and example of which is shown in Fig. 10.14: (a) histogram, (b)
in this case the amount of uniform layer thicken- cumulative frequency.
ing H/H0 may be based on the variations of
maximum amplication and values Lmax/H at Lmax/H. Despite the notation, Larc is not a wave-
which it occurs, as given in Fig. 10.21. Details of the length but simply, in this case, twice the hinge-to-
argument are given elsewhere (Sherwin and hinge arc length in a fold train, a quantity that has
Chapple, 1968). Simulations of fold train formed a random distribution. For the fold train from
by combining wavelength components of specied which Fig. 10.14 comes, the average value is
amplitude derived from an initial white rough- Larc H 4.92. Thus, values of R and H/H0 will lie
ness amplitude spectrum (all components have along a contour close to that for the 5 contour in
the same initial slope) showed that the mean of the Fig. 10.21. From an assessment of the degree of reg-
fold arc length thickness ratio, Larc/H, is a good esti- ularity of folding in terms of the dispersion (stan-
mate of the value at maximum amplication, dard deviation/mean) of the distribution shown in
410 VISCOUS FLOW
3.5
currently active accretionary wedges like those of
200
40
60
400
0
2.4 1000
30
10
Taiwan and the Himalaya (Hilley and Strecker,
20 2004) and the Olympic Mountains of Washington
2.2 4 State (Batt et al., 2001; Brandon, 2004), or in
15 extinct wedges. Here, we formulate and analyze a
2.0 model for ow, deformation, and stress in the
4.5
H/H0
take place on both sides of this. A very simple wedge divided by its mean or maximum thick-
modication of the present model may be made ness. The second dimensionless group is the ratio
to account for this, but we avoid the temptation to of the magnitude of underplating to that of sub-
introduce it in order to achieve the simplest duction drag. The third dimensionless group is
approximation to wedge dynamics. Moreover, the the ratio of the lithostatic stress to a quantity
effect of any such modication to a model is best with the dimensions of stress that is proportional
appreciated by rst studying the behavior before to the viscosity and K. The parameter K is associ-
it is introduced. ated with thinning of the wedge by erosion: it is
Prior to solving (10.134), we consider the not a rate of deformation, but has the same
special restriction to the case of a steady-state dimensions as one (time)1. The third dimension-
wedge, in which the amount of material lost by less group may be interpreted as the ratio of the
erosion and outward ow is just balanced by the relative outward ux of material by gravity-driven
inward ow, and the balance is achieved by time- glacier ow within the wedge to that by erosion.
independent regimens of inux, underplating, Note that gH/2 is a rate of deformation. Each of
and erosion. For the present model, this requires these dimensionless groups, then, has a clear and
that the vertical ux of material in the interval of concrete connection with major aspects of wedge
interest, 0 x L/4, and at the upper surface, is form and dynamics, and a connection with mea-
just balanced by erosion. We use a rate of erosion surable quantities such as wedge form, the ampli-
proportional to the local topographic relief tude of topographic relief, and rate of erosion or
which, in view of the sinusoidal variation of exhumation, or erosional ux. These parameters
topography, is described by: also establish the deformation within the wedge
and the strain observed in rocks exposed at the
(d Adt)erosion KA (10.135) surface.
Before solving for the coefcients, we obtain
In the steady state, this loss is just balanced by the one other result that emphasizes the simplicity of
ux associated with vy(x, H) over the same interval the model and, by inference, that of the dynamics
in x. This gives the condition: of a natural accretionary wedge when viewed at a
suitably large length scale and increment of time.
K Ass U{[a b(k 1)]ek [c d(k 1)]e k} 0 The velocity and stress distributions may be
(10.136) obtained if we know the stream function, ,
Here Ass is the amplitude of the steady-state topo- where vx y and vy x. For this case:
graphic relief.
Substitution of (10.136) into the fourth equa- UH(1H){[a b(y1)]ey
tion in (10.134) yields the set of equations: [c d(y 1)]ey} sin x (10.139)
0. 3
0. 2
2py/L
0. 1
0.0
0.0 0. 5 1.0 1. 5 2.0 2. 5 3.0
2px/L
Fig 10.23 Exact and approximate (higher in each pair)
streamlines for a circulating cell model.
for (k 3)(21). In the present case, k /10
and 1.2, and 0.15.
In Fig. 10.24, streamlines are plotted for four
Using the Olympic accretionary wedge as an pairs of and with k /10. In these gures, the
example (Fig. 10.8a), we may roughly estimate the small A/H 1 surface amplitude is not shown
model parameters (Batt et al., 2001; Brandon, and the upper surface is taken as nominally plane.
2004). A rate of erosion 1 km Ma1 is associated Figure 10.24a with 0.1 and 1 is a possible
with a topographic amplitude A 1.5 km, giving approximation for the Olympic accretionary
an erosion constant K 2/3 Ma1. Using a density wedge. If 0, strong horizontal extension
2700 kg m3, acceleration of gravity g 9.8 m occurs at the upper surface, with large magnitude
s2, and H 20 km, we obtain 1.2, much larger shearing of opposite sign corresponding to the
than the value used in Fig. 10.23. The ratio W/U inward and outward uxes of material in the
can be estimated from interpretation of observa- wedge ow. Vigorous underplating (Fig. 10.24c)
tions. Strain measured from rock at exposure eliminates much of the shearing associated with
gives no indication of extension, which is favored basal drag. Vigorous erosion (Fig. 10.24d) elimi-
by larger values of , i.e. of underplating. This pro- nates the near-surface extension.
vides an upper limit on . While this estimate may The nite deformation may be computed by
be established by examining model streamlines, it integration to determine the displacement gradi-
is simpler to examine the rate of deformation and ent tensor Fij along particle trajectories. In plane
strain of a particle underplated at the backstop. ow the evolution equations reduce to:
The approximate solution for the velocity compo- DFxx Dt Dxx Fxx (Dxy z)Fyx
nents provides convenient closed-form results:
DFxy Dt Dxx Fxy (Dxy z)Fyy
(10.147)
Dxx(0, y)
U
1
3
2k2
k
1
2y
y2
k (10.146)
DFyx Dt (Dyx z)Fxx DyyFyx
DFyy Dt (Dyx z)Fxy DyyFyy
Using L= 400 km, or twice the width of the back-
stop-to-toe distance, the magnitude of the rate of Here z 12(vy x vx y). Since the velocity
deformation, U, is 0.5 1014 s1 times the value components are known, integration may be
of U expressed in cm a1. The horizontal rate of carried out numerically. A simpler means of
deformation is a minimum (greatest negative obtaining an illustration of the strain ellipses is to
value) at the base of the wedge, and increases follow the deformation of an initial circle of par-
upwards. Near-surface extension occurs for any ticles by incrementing their positions for a succes-
greater than a value which gives zero at y k, or sion of small time steps, but this operation does
10.4 VISCOUS FLOW IN LAYERS: MULLIONS AND FOLDS 415
dFyy dvy 1 1
(c) vy Dyy Fyy F , or dFyy dvy
dy dy yy Fyy vy
(10.149)
0.4
0.3
2py/L
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1
2px/L
Fig 10.25 Strain ellipses for particles rising along geneous functions of the components of stress.
trajectories at 0, 0.2, 0.4, 0.6, and 0.8 of the half-cell width
We then write:
for the case 0.1, 1.
0 2Dxy a66xy
0 0.2 0.4 0.6 0.8 1
z/H Here b11 a11 a213 / a33 and b12 a12 a13a23 / a33. Evo-
Fig 10.26 Quadratic elongation (a/b) for particles rising king incompressibility, we obtain b11 b12 0 and
along the backstop, for 1 and 0.1, 0.15, 0.2, 0.4, and b12 b22 0, so b22 b12 b11. Substituting for the
1.
two independent constants in (10.156) the consti-
tutive relations for plane ow are:
ai1 ai2 ai3 0, i 1, 2, 3, 4, 5, 6 (10.153)
Dxx (xxyy) 4n
The number of independent coefcients for a
Dyy (xx yy) 4n (10.157)
general anisotropic viscous uid is reduced from
21 to 15. Dxy xy 2s
Following the discussion of anisotropic elastic
Here n and s are the principal viscosities for
materials with certain symmetry restrictions, we
plane ow in the (x, y)-plane (10.154). For plane
write for an orthorhombic material with three
ow in other symmetry planes, the numerical
mutually perpendicular mirror planes of symme-
values of these quantities would change.
try, initially setting aside the conditions (10.153):
We may further suppose that the material of
interest is transversely isotropic, with equivalent
Dxx a11xx a12yy a13zz
directions x and z, so that only the y-axis is a
Dyy a12xx a22yy a23zz unique direction within the material. The
coefcients in (10.154) then have the additional
Dzz a13xx a23yy a33zz
(10.154) restrictions:
2Dyz a44yz
a23 a12, a33 a11, a44 a66, a55 2(a11a13)
2Dzx a55zx
(10.158)
2Dxy a66xy
The last condition arises from isotropy in the (x, z)-
This simpler form of the constitutive relations plane, i.e. that the constants do not change for
only applies when the axes x, y, and z are principal such a material when the coordinate axes are
axes of anisotropy, xed in material directions changed by an arbitrary rotation about the y-axis.
coinciding with the intersections of pairs of An incompressible, transversely isotropic aniso-
mirror planes in this case. tropic viscous uid is then described by only two
We will investigate a plane ow problem for an parameters, the principal viscosities n and s of
incompressible anisotropic viscous uid. One (10.157).
requirement for plane ow is that the plane of Such a material may be thought of as an
418 VISCOUS FLOW
1
approximation to a layered medium made up of f1 f2
s (10.165)
alternating layers of stiff and soft isotropic 1 2
viscous uid when the scale of the ow of interest
The ratio of the principal viscosities is:
is much greater than the layer thickness, as in the
case of chevron folds. Suppose the layers have n f f
thicknesses h1 and h2 and viscosities 1 and 2, and m ( f11 f22) 1 2 (10.166)
s 1 2
take the axes x and z parallel to layering so the y-
axis is normal to the layering. The bulk properties This is a measure of the degree of anisotropy.
of this composite material are obtained as follows. Inspection shows that m 1. It may be shown that
In layer-parallel extension, the rates of extension the maximum of m occurs at f1 f2 0.5.
in the two layer types must be equal. In the two in- The full constitutive relations for the trans-
plane directions: versely isotropic anisotropic uid, in principal
coordinates, are:
xx D xx Dxx, D zz D zz Dzz
D (1) (2) (1) (2) (10.159)
For simplicity, consider plane ow in the (x, y)- Dxx xx 13 (xx yy zz) 2n
plane, so that for example: Dyy yy 13 (xx yy zz) 2n
(10.167)
Dzz zz 13 (xx yy zz) 2n
D(1)
zz (1) 1
zz 3 ( (1) (1)
xx yy zz ) 21 0
Dyz yz 2s, Dzx zx 2n, Dxy xy 2s
zz ( xx yy) 2
(1) (10.160)
(1)
1
f2
2
xy
2
xy
2s
fold limbs, an exact interpretation of the macro-
scopic solution may be made to give the quantities
for component layers or surfaces.
(10.164)
As in Chapter 5, consider the evolution of sym-
Rearranging, the bulk viscosity in layer-parallel metric, periodic chevron folds formed by shorten-
shear is: ing normal to their axial planes (Fig. 10.27). The
10.5 FLOW OF ANISOTROPIC VISCOUS FLUIDS 419
d d
tan u tan (1 tan 2) 2(Dxy Dxx tan )
dt dt
u
x (10.172)
(10.174)
normal to the fold axial plane. The constitutive
relations for the homogeneous material in the This relation may be integrated numerically to
left-dipping fold limb are then, by transformation follow the evolution in limb dip with shortening.
of coordinates: As in Chapter 5, it is necessary to introduce a peri-
odic set of seed folds with initial limb dip 0.
4nDxx [(1 m) (1 m) cos 4 ]sxx
Examples of fold evolution in terms of limb dip,
(1 m) sin 4sxy , for several values of the viscosity ratio, m n/s,
4nDxy (1m) sin 4sxx are given in Fig. 10.29 for m 1, 1.5, 2, 3, 4, 10, and
. Note that fold growth corresponds to a decrease
[(1 m) (1 m) cos 4]sxy (10.168)
in fold span and so progress is from right to left on
Here the stress components are the deviatoric this gure. For the isotropic uid, m 1, fold
stress components dened as: growth is the passive kinematic amplication of
420 VISCOUS FLOW
tan u 30 2 3 4 10
du 1.5
20
u 1
x 10
1
0
Fig 10.28 Change in angle and length of layer segment 0.5 0.6 0.7 0.8 0.9 1
after a time increment dt. Fold span (S/S0)
Rheological behavior
Pressure solution in sandstone from the Olympic No mathematical theory can completely describe the
accretionary wedge, Olympic Mountains, NW Washington complex world around us. Every theory is aimed at a
State, USA. The horizontal dimension is 10 mm. certain class of phenomena, formulates their essential
Photograph by J. M. Rahl. features, and disregards what is of minor importance.
The theory meets its limits of applicability where a dis-
regarded inuence becomes important (Flugge, 1967).
422 RHEOLOGICAL BEHAVIOR
11.1 Departures from linear In one instance (Groshong, 1975), layer thick-
ening associated with a well-developed set of folds
viscous flow in a limestone layer embedded in shale was only
10%. Interpretation of data for these folds
L
aboratory determinations of steady-state rock (Fletcher, 1974) is consistent with the type of non-
creep and eld observations indicate that linear relation observed in the laboratory for
ductile rocks may not be well-approximated creep of carbonate rock. It has been suggested that
by homogeneous, isotropic, incompressible, and the limited range in fold arc length/thickness
linear viscous uids of uniform viscosity. In this ratios 4 to 6 is a manifestation of highly non-
chapter, we consider other constitutive relations linear behavior (Smith, 1977, 1979). Further evi-
that broaden the range in rock behavior described dence for markedly non-linear behavior is offered
and provide a basis for understanding some of the by examples of necking in layer-parallel exten-
differences arising. Because the linear Newtonian sion, which cannot occur in a Newtonian layer.
viscous uid is the simplest material that under- Alternatively, interpretation of the relative vis-
goes large permanent deformation, the formula- cosities of component rocks in a deformed con-
tion and analysis of models using it and their glomerate has been offered as evidence of linear
application to interpret a set of eld observations viscous behavior (Treagus and Treagus, 2002).
is always a useful rst step (see Chapter 10). The Motivated by these questions, we study necking
results obtained establish a benchmark from and folding of layers of non-linear power-law uid
which to understand differences in behavior asso- in this chapter.
ciated with other constitutive relations. Whether Inhomogeneity in rock masses occurs at a wide
the new model results in a large or subtle contrast range in scale, from less than a grain diameter to
in behavior relative to an already well-understood the tens or hundreds of kilometers appropriate to
viscous model, we will achieve a better under- the rst-order dynamics of crustal deformation. A
standing of the reasons for the differences and a useful postulate is that at the scale of interest the
greater condence in data interpretation. rock may be treated as a continuum whose behav-
For example, initiation of a regular train of ior is approximated by constitutive relations con-
folds by selective amplication was studied for the taining only a few rheological parameters. The
Newtonian viscous layer in Chapter 10. Data from question then arises as to how these parameters
natural fold populations yield a limited range in might be estimated from the three major deter-
mean fold arc length/thickness ratios of about 4 to minants of the bulk behavior: the volume frac-
6. Interpretation of the data in terms of the tions of the signicant mechanically distinct
folding of a Newtonian layer implies: (i) a modest components, their individual constitutive re-
layer to host viscosity ratio, 1/R, of about 10 to 20; lations, and the phase geometry of the composite
(ii) a weak folding instability, so that a large layer- material. Answers to this question would provide
parallel shortening is required to establish a a better understanding of the degree of complex-
regular fold train, e.g. a doubling of layer thick- ity of rocks and a basis for accepting or maintain-
ness. On the other hand, laboratory studies of ing doubts about the approximation of a locally
steady-state rock creep suggest that much larger homogeneous continuum with spatially varying
contrasts in effective viscosity should be common properties. In this chapter, we use elementary
(Table 11.1), with equivalent 1/R values of about methods to analyze composites made up of two
100 to 1000 or more, and that constitutive rela- isotropic viscous components. Composite materi-
tions for rock creep should generally be non- als must generally exhibit anisotropy in their rhe-
linear. Do the laboratory results extrapolate to ological behavior, with isotropic behavior being
natural rock behavior at much smaller rates of only a special case. We therefore include consid-
deformation and lower temperatures? Answers to eration of anisotropy.
this question not only affect conclusions as to Anisotropic materials such as foliated or
small-scale folding, but also the modeling of large- layered rock may have nearly uniform properties
scale crustal deformation. with respect to their principal axes of anisotropy.
11.2 BOUDINAGE AND THE NON-LINEAR POWER-LAW FLUID 423
11.2.1 Plane-sections-remain-plane
analysis for a free plate
Let the material be the isotropic incompressible
power-law uid already treated in the glacier-like
ow of Chapter 10. Consider a free plate of cylin-
drical form whose thickness varies as in Fig. 11.2.
It is not necessary to assume such a restricted
form, but we do suppose in the present case that
layer thickness varies slowly with length and that
(b)
the thickness variation is small relative to the
mean thickness. For this periodic shape, these
requirements are:
A 1 and A h 1 (11.1)
A ~
dH dH
[Hn1 (n 1)Hn2H~ ] C (11.8)
dt dt
x Expansion and separation of mean and perturb-
2h
0 ing parts gives:
n1 dH
H C
dt
~
1 dH dH
~
(n 1)H 0
L H dt dt
(11.9)
~
dH
Fig 11.2 Sinusoidal pinch-and-swell perturbation. (n 1)Dxx H
~
dt
I2 12 DijDij B2J(n1)
2
1
2 sijsij B J2
2 n
(11.17)
J2 (I2B2)1n
(ref)
sxy /sxy
Using (11.17) in (11.13) we have:
sij B1nI(n1)
2
2nD
ij (11.18)
n =infinite,
It is useful to make a comparison between the rigid-plastic solid
relations for the Newtonian viscous uid, the 1
power-law uid, and a third material, the
rigid, perfectly plastic solid. For simplicity, the 10
results for plane ow are given, for which 3
sxx syy 12 (xx yy), sxy xy. All materials are
incompressible so the rates of deformation are n =1
related as:
1n
(11.21) xy Dxy
where J2 14 (xx yy)2 2xy (ref ) (11.23)
(ref )
xy Dxy
random waviness. Selective amplication may plane and only the normal, principal components
then only occur when the constitutive rela- of sij are non-zero, yielding:
tions of a non-linear material can be linearized
2 D~ xx s~xx (n 1)2nJ2 sxx (sxx s~xx syy s~yy szz s~zz)
about a basic state of ow, since only then do the
wavelength components in the perturbing ow 2 D~ yy s~yy (n 1)2nJ2 syy (sxx s~xx syy s~yy szz s~zz)
not interact. Linear independence also requires
2 D~ xy s~xy (11.27)
that linearization of the boundary conditions,
when only terms proportional to A are re- Here D~ xx D~ yy 0 . Because of the proportionality
tained, is accurate, as in our analysis of viscous of components of the rate of deformation and
folding. deviatoric stress in the isotropic relations for the
Here, the basic state is taken as uniform layer- basic state in (11.26), the basic-state deviatoric
parallel extension or shortening, which simply stress components and J2 may be replaced by
changes the sign of Dxx. To emphasize this, we basic-state rate of deformation components and
write: I2 12 Dkl Dkl such that:
Dxx Sgn(Dxx) |Dxx | (11.24) 2 2 2 2
I2 12 (D xx D yy D zz) D xx(1 2 ) (11.28)
Since we have not studied folding in layer-parallel
shortening for a non-linear uid layer, we also The quantity Dzz Dxx is the ratio of the basic-
consider that here. We shall see that other types state rate of deformation component parallel to
of instability in the shortening or extension of a the axis of the shape perturbation to that acting
layer occur; for example shortening of a soft layer in the plane of the layer and normal to the axis.
between stiff media produces mullions (Smith, The plane ow case is then 0.
1975). Because restrictions are placed on the com-
To add interest to the analysis, consider a basic- ponents of the rate of deformation, it is more con-
state ow in which the principal rates of defor- venient to use the inverse relations (11.18). The
mation, Dxx, Dyy, Dzz, are all non-zero, with axes x linearized form is obtained in indicial notation:
and z in the plane of the layer and y normal to it.
The perturbing ow is restricted to a plane ow
associated with a cylindrical component in the
(n1) 2n
sij s~ij B1n I 2
1
(n 1) ~
D D (D D
2nI2 kl kl ij
~
ij)
(11.29)
shape perturbation whose axis lies parallel to the
principal axis of the basic state, . From (11.29) we extract the basic-state and pertur-
To linearize the constitutive relations, it is con- bation relations:
venient initially to carry out the computation
using indicial notation: sij B1n I 2(n1)2n Dij
(n 1) (11.30)
(n1)2
1 s~ij B1nI(n1) ~
2n D
ij
~
D D D
Dij Dij B (sklskl 2skl s~kl)
~
(sij s~ij) 2
2nI2 ij kl kl
2
In the present case, we introduce two restrictions:
BJ2(n1)2 1
(n 1)skl s~kl
2J2
(sij s~ij) (i) the only non-zero components are the principal
components Dxx, Dyy, and Dzz; and (ii) the perturb-
(11.25)
ing ow is plane, with non-zero components
~ ~ ~
From the second line of (11.25) we nd: D xx Dyy, Dxy .
The relations (11.30) give for the
perturbing ow:
Dij BJ2(n1)2 sij
D~ ij BJ2(n1)2 s~ij
(n 1)
2J2
sijskl s~kl
(11.26)
s~xx 2 1
(n 1) ~
D (D Dyy) D
2nI2 xx xx xx
(n 1)
We now reduce the second set of six equations, s~yy 2 1 ~
D (D Dxx) D (11.31)
2nI2 yy yy yy
where indices i and j take values 1, 2, and 3, for the
restrictions of interest: the perturbing ow is s~xy 2 D
~
xy
11.2 BOUDINAGE AND THE NON-LINEAR POWER-LAW FLUID 429
d4V 1 1
d2V
(n1) 2n
Here 2 B1nI 2 and is the effective viscos- 2 1 2 2 4V 0 (11.38)
ity (11.14) in the basic state, which is uniform within dy4 n x ny dy
the layer. These may be further condensed to: The solution may be written:
s~xx 2
1 ~
D
nx xx V (a cos y b sin y)ey
(c cos y d sin y)ey (11.39)
1 ~
s~yy 2 D (11.32) Here:
ny yy
n, 1 n
1 1
s~xy 2 D
~
xy
(11.40)
Here the inverse of nx and ny are dened: 3(n 1) 2
1
n n 1
1 (n 1) 4(1 2)
1 D (D Dyy)
nx 2nI2 xx xx
(11.33) To obtain the expression for n in (11.40) we used
1 (n 1) the relations (11.28) and the following:
1 D (D Dyy)
ny 2nI2 yy xx
2 1 1
(11.41)
The stress and deviatoric stress components are n nx ny
related by:
From (11.36) and (11.40), the expressions for
~ xx
2
1 ~
D p~
nx xx
the velocity components are:
~ yy 2
1 ~
D p~ (11.34)
v~x a (n^ 1)b cos y
ny yy
b (n^ 1) a sin y ey
2 Dxy ~
~ xy
c (n^ 1)d cos y
Here p~ 13 (~ xx ~ yy ~ zz)
is the perturbing part
of the pressure, or the negative of the mean
d (n^ 1)c sin y ey sin x
normal perturbing stress. For plane ow, the con- v
~ [(a cos y b sin y)e y
y
dition of incompressibility is:
(c cos y d sin y)e y ] cos x (11.42)
~ ~
v~x v~y
Dxx Dyy 0 (11.35)
x x From the third equation in (11.32) and (11.42) the
This is automatically satised by the separable stress component ~ xy may be obtained, and from
expressions: the stress equilibrium equations, the components
~ xx and ~ yy are obtained by differentiation and
1 dV integration:
v~x sin x
dy (11.36)
~ 22
xy a (n 1)b cos y
v~x V cos x
Substituting (11.36) and (11.32), using (11.34), into b (n 1)a sin y ey
the stress equilibrium equations (10.15):
c (n 1)d cos y
~xx ~xy
x
y
0 d (n 1)c sin y e y sin x
(11.37)
~xy ~yy ~ 2[(a cos y b sin y) ey
0 yy
x y
(c cos y d sin y) ey ] cos x (11.43)
Eliminating between the two equilibrium equa-
p~
tions, we obtain an ordinary differential equation The even more complicated expressions for
in V V(y): the layer-parallel normal component
~
xx and the
430 RHEOLOGICAL BEHAVIOR
pressure p~ are not used in solving for the velocity 22 a (n 1)b cos h (e h eh)
(11.45)
v~y [a cos y (ey ey)
n(2 )(1 R)
(1 Q ) 2(nsin1)k [(1 Q )(e
(b sin y)(ey ey)]cos x
2 2 k e k) 2Q (ek ek)]
The stress components are:
Sgn(Dxx)|Dxx|A
xy
~ 22 a (n 1)b cos y (ey ey)
q(k; n, n1, R, )Sgn (Dxx) |Dxx| A (11.51)
b (n 1)a sin y (ey ey) sin x Here:
~
yy 2[a cos y(ey ey) 1 1 1
Q R (11.52)
b sin y(ey ey)] cos x (11.46)
Using the same approximations as in the analysis The principal argument k 2 (HL) of the func-
for the buckling of a viscous layer, we obtain four tion q species the wavelength to layer thickness
relations from the boundary conditions at the ratio of the cylindrical sinusoidal perturbation.
upper sinusoidal surface: The rst three dimensionless parameters describe
the rheological behavior of the layer and medium:
a (n 1)b cos h (eh eh) n is the stress exponent of the layer, n1 that of the
medium, and R is the ratio of the effective viscos-
b (n 1)a sin h (eh eh) ity of the medium to that of the layer. Finally,
describes the basic-state ow to which the pinch-
1 c1 (n 1)d1 (11.47) and-swell component is responding. Since Dxx is
greater than zero for axis-normal extension, 0
a cos h(eh eh) b sin h(eh eh) c1 implies an additional axis-parallel extension, and
(11.48) 0 implies axis-parallel shortening. Note that
11.2 BOUDINAGE AND THE NON-LINEAR POWER-LAW FLUID 431
the effective stress exponent of the layer, n, is a identical to (11.51) except for a single change from
function of both n and (11.40); the stress expo- a positive to a negative sign in the denominator of
nent n1 is hidden in the quantity Q. The function the second term in braces, just after the quantity
q may be termed the relative rate of amplication (1 Q2). This is simply a consequence of a symme-
factor; it is dimensionless. try in folding opposite to that expressed by (11.44),
Since Dyy Dxx, the layer thickness, wave- or, in folding, v~y(x, y) v~y(x, y). Regular mullion
length, and wavenumber satisfy the relations: structures (Fig. 10.10b) forming in shortening of a
soft layer between stiff half-spaces are treated by
dH dL d assigning values of R 1 in either of these two
Dyy H, Dxx L, Dxx (11.53) relations. Mullions may be approximately sym-
dt dt dt
metric about the layer mid-plane, in which case
What seem like only moderate extensions of the (11.51) applies, or less commonly, asymmetric, in
model for initiation of folding of a single layer of which case the relation with the sign change
viscous uid embedded in a viscous medium, here would apply. Given a set of natural pinch-and-
for the complementary case of necking, lead to a swell structures, folds, or mullions, leading ques-
somewhat daunting range in behavior. If the layer tions would be whether the present model could
and medium are non-linear power-law uids, two produce them and what ranges of parameters are
stress exponents are introduced, n and n1, and if necessary.
the basic-state ow has a component of stretching As in the viscous folding problem, it is useful
or shortening parallel to the perturbation axis, a to expand the relative rate of amplication factor,
feature not uncommon in natural deformation, q, to low-order terms in k. The result is:
the parameter enters. In contrast, the behavior in
the viscous folding model only varies with the vis- n(2 )(1 R)
q (1 ) (11.54)
cosity ratio R. Thus, from a single parameter space, k2 2Q Qk 2
2 1
we must now consider a four-dimensional space! 6 k 3 n
In interpreting a set of data from trains of
The free-plate result is obtained by setting Q R
natural folds or pinch-and-swell structures, all
0:
four parameters may play a signicant role. Thus,
the relation (11.51), although relatively compli-
cated, allows for a full study of the variation in
behavior with the four parameters, the viscosity
1
(q)free plate (1 ) n 1
2
1 121 k
2
necking in layer extension, we know that we must The last expression corresponds to the plane-sec-
consider n 1. In most examples presented here, tions-remain-plane approximation (11.10) but now
the parameters are reduced to two by setting n1 contains information on the effect of additional
1, the Newtonian viscous limit, and 0. An shortening ( 0) or extension ( 0) along the
exception is the result for mullions in a layer for axis of the perturbation. The second line indicates
which computed results are obtained later (Fig. a wavenumber, k, dependent deviation from the
11.12), which only form if n1 1, and for which the plane sections result. The plane sections approxi-
dependence on n1 is key. Because the parameter mation for q, the last line in (11.55), is contoured
Q ~ nn1 plays a signicant role in the relation in (Fig. 11.5) as a function of the intrinsic stress
(11.51), restriction to n1 1 bypasses interesting exponent, n, and the rate of deformation ratio, .
behavior (Smith, 1977). This shows a substantial decrease in instability, as
Perhaps remarkably, the result for amplica- measured by the relative rate of amplication q,
tion of a fold component is given by a relation away from a plane ow basic state, 0. This
432 RHEOLOGICAL BEHAVIOR
4 1000 10
n = 10, n1 = 1
0
0.001
3
5
100 R = 0.01
log10(n)
q
2
40 0.1
20
q =10 0
1
4
2
1
0
0
1 0 +1 5
j 100 101 102
L/H
Fig 11.5 Contours of q for plane-sections-remain-plane
approximation for the free plate in (n, )-space. Fig 11.6 q-spectra for n 10 and R 0, 0.001, 0.01, and
0.1; heavy lines show the exact results, light lines the
approximation for k 1.
decrease is less for extension parallel to the pinch-
and-swell axis ( 0) than for shortening ( 0)
parallel to the axis. For example, if n 101, q 100 the amplication, and thus the regularity in an
for plane ow, but only q 8 for an axial rate of array of structures. The q-spectra for R 0.05 and
extension one-half that of the axis-normal exten- n 10, 100, and 10 000, and (Fig. 11.7) show mul-
sion, that is for 0.5. For axis-normal contrac- tiple maxima that become prominent for n 100.
tion, 0.5 and the relative rate of amplication These arise because of the increasing dominance
factor is q 3. Thus, the model suggests that pinch- in the variation of q with k (11.51) of the sinusoidal
and-swell structures are favored in nearly plane term, sin [k1 (1n)] sin k, n 1 over the
deformation. exponential term, exp ( k 1n ) 1, n 1. The
The dependence on L/H, where H 2h is the transition between two modes of necking or
thickness of the layer, is shown by plotting q folding, one at modest stress exponent n, and one
versus L/H (Fig.11.6) for necking in plane ow for n at large stress exponent n 10100, is also tied to
10, n1 1, and several values of R. The expansion this transition in dependence. The second mode is
for k 1 yields a good approximation (dashed one of resonance folding or necking in which:
lines) down to L/H 10. While the stress exponent The competent layer does not act mechanically
n 10 is large compared with values estimated in as a coherent unit, but, instead, the irregularities
laboratory experiments, the necking instability is on one interface produce motions that deform
relatively weak, so that a large basic-state stretch the other and vice versa (Smith, 1979). This
is required to get signicant amplication. For behavior is associated with the dominance of
example, an amplication of 10 at the dominant sinusoidal versus exponential variation in veloc-
wavelength at which q is a maximum, for qd 9 ity and stress components in the y-direction,
would require a stretch S exp(ln 10qd) 1.29. which result in the dependences of q on k just
We denote the variation of q with wavelength noted. It is only shown for k 1, or for L/H 6,
or wavenumber the q-spectrum. From it the pos- when the sinusoidal dependence in k begins to
ition of maximum is at the dominant wave- become apparent.
length/thickness ratio, Ld/H, and the relative The coherent layer mode is exemplied by
sharpness of the peak indicates how selective is the pure folding mode of a single viscous layer, as
11.2 BOUDINAGE AND THE NON-LINEAR POWER-LAW FLUID 433
20
L/H = 0.8
10000
15
10 100
n =10 n = 1, L/H = 4
5
3
0
q
5
L/H = 4
10
15
20
100 L/H 101
n = 1, L/H = 2
Fig 11.7 Multiple peaks in qq(L/H) as a function of n
10, 100, 10 000 and the n approximation for R0.05, n1
3, 0. Velocity fields for the two maxima at L/H4 and
0.8 and the minimum at L/H4/3 are shown in Fig. 11.9.
4 (a)
400
3 200
4.1
100
40 2
log10(n)
20 qd = 10
2
Ld/H = 4.5
5
1 6
10 8
20 15
(b)
0
0
3 2 1
log10(R)
log10(n1)
suggestive of a continuous ductile necking.
Detailed study to support or refute this suggestion
6
has not been carried out by us. None-the-less, this 2 5
hypothesis provides a useful example for the
present discussion.
In Fig. 11.11b, we have constructed the rectan- Ld /H = 6 10
2 8
gles approximating the initial boudin form for 1
the entire structure, supposing the central neck
to be a later feature, and for the two individual
boudins or pinch-and-swell structures. Their
0.2 0.6 1.0 1.4 1.8
aspect ratios are, for the entire structure, 4.2, and log10(R)
for the individuals, 2.3 on the left and 2.7 on the
right. The value for the entire structure is consis- Fig 11.12 qd and Ld/H for mullion structures for n 3 and
tent with model values of Ld/H for n 1, but n1 2.
those for the individual structures are not. It
might be suggested that the latter reect the ten-
dency for a mechanically isolated segment to The strength of the instability is small unless the
divide in two sub-equal segments if it becomes host has a very large stress exponent, so that
unstable with respect to necking, in this case mullion structures such as those seen in Fig. 10.2b
chiey by faulting. The present model applies may be inferred to represent large amounts of
only to a continuous layer or a segment with very layer-parallel shortening. This might explain the
large aspect ratio, and does not apply directly to signicant difference between the span-to-thick-
a process involving discrete faults (but see the ness ratio of these structures and the values of
next section). Ld/H 6.
The present model may also be applied to the
initiation of the regular mullion structures 11.2.4 A model for large-scale crustal
shown in Fig. 10.2b. The lobe-and-cusp morphol- necking
ogy indicates that the layer has the lower effective Necking may manifest itself at crustal and litho-
viscosity, or R 1. These structures are produced spheric scales, on the Earth, on other planets and
in layer-parallel shortening, so that Sgn ( Dxx ) 1. on the moon of Jupiter, Ganymede (Fink and
Using these and plane ow ( 0), (11.51) then pro- Fletcher, 1981; Collins et al., 1998; Patel et al., 1999;
vides a relation describing the growth or decay of Dombard and McKinnon, 2001). The strikingly
the pinch-and-swell perturbations whose selective regular succession of basins and ranges that form
amplication give rise to the mullions. Appreci- the dominant structure in the Basin and Range
able instability requires that the host be non- Province of the western United States (Fig. 11.13)
linear. We thus assign a typical stress exponent have been interpreted as the result of necking of
(Table 11.1) to the layer, n 3, and determine the the strong brittle layer of the crust (Fletcher and
variation of qd and Ld/H in (n1, R)-space (Fig. 11.12). Hallet, 1983). The crust is broken up into segments
Note that n1 2. This gure is read in the same approximately 30 km in width (Fig. 11.14). These
manner as Fig. 11.10, with values of n and n1 deter- are superposed on a subtler necking at a scale
mining Ld/H and qd, the d again denoting the 100 km in which the strong upper layer of the
dominant or most rapidly amplifying component. mantle lithosphere plays a signicant role (Zuber
11.2 BOUDINAGE AND THE NON-LINEAR POWER-LAW FLUID 437
10
9
8
7
Frequency
6
5
4
3
2
1
0
10 15 20 25 30 35 40 45 50 55 60
Spacing between ranges (km)
The appropriate form of the solution for may be Combining (11.62) and (11.67):
written:
b 2 |Dxx |A S
cos k
Sgn(Dxx) sin k
1
2
K
|Dxx |
(b cos y d sin y)cos x (11.62)
k
(11.68)
~ (x, 0)
~ (1) (x, 0)
The perturbing stress components are gener- yy yy
(11.69)
ated from the yield condition and, by use of the ~ (x, 0) 2K Sgn(D )( A) sin x
xy xx
Airy stress function, from the equations of stress
xy (x, 0) 41Dxx(A) sin x
~ (1)
equilibrium. The choice of the constant factors in
(11.62) is merely to provide convenient expres- Substituting from (11.66) and (11.62) together with
sions for both the stress and velocity components. expressions for a viscous half-space (Chapter 10)
From the ow law in (11.22): into (11.69) gives:
v
~
x [(a d by) sin y (c b dy) cos y] sin x
dA
dt
S
D xx A 2 S (1 R cos k sin k)
Rk
v
~
y [(a by) cos y (c dy) sin y] cos x
Sgn(Dxx) sin2k |Dxx |A
(11.66)
of shortening, since the shear strength K of the fault scarps (Andrews and Hanks, 1985) supposes
brittle upper crust is independent of rate. At slow that the ux of material is proportional to slope.
rates of extension, R will be smaller, and the Conservation of mass, excluding differences
degree of instability will be larger. The dimen- in density between rock and sediments, then
sionless group S gH2K is the ratio of a litho- requires that the rate of change in elevation be
static stress to the stress difference at yield in the proportional to the negative of the divergence of
plastic layer. A more realistic model for the the volume ux. Any adjustment towards an iso-
strength of the crust would introduce a mean static state, often included in models for the
stress dependence on the strength. Effectively, inlling of sedimentary basins and concomitant
only a cohesion-like quantity is used here, erosion at their periphery, is already included in
although its value may be assigned to account for the solution for the velocity eld. Applying
a mean value of the strength of the brittle crustal Cullings model on a component-by-component
layer that is dependent on the thickness H. basis, we obtain:
Further discussion of this issue may be found in
the papers cited earlier. The third dimensionless
group is k 2/(L/H).
dA
dt surface
P2A Mk2 |Dxx |A
1 ment. More simply, if the rate of reduction by
(1 2 sin 2 k) A 2 1 sin k A
R erosion is 1, 10, 100 m Ma1, or 1 km Ma1 for a
log10(R)
maximum value of Ld/H. A question of principal 40 6
interest is how the existence of the necking insta-
bility and its length scale constrains the dimen- Ld/H = 4.6 5
100
sionless groups S, R, and M. Alternatively, we may
2
ask which a priori estimates of these quantities, as 200
of M above, would be consistent with necking
400
at the basin-and-range scale 30 km. Overlap
between these two sets of values would support
the model. 1000
An estimate for mean strength of the brittle 3
2 1 0 1 2
crust in extension, using a friction angle of 30 and log10(M)
no cohesion is K gH/3. Including cohesion, K
will be larger than this by a multiple greater than Fig 11.15 Contours of qd and Ld/H for the crustal necking
one and certainly less than two for H 10 km. model in (R, M)-space for S1.
Thus, we expect 3/2 S 3/4, and for 1 1019 to
1021 Pa s1 and Dxx 1015 s1, R 0.0002 to 0.02.
The latter values are remarkably small relative to behavior (Fletcher and Hallet, 1983), but compara-
effective viscosity ratios obtained from interpreta- ble to the present one, provide a better t.
tion of natural fold data; if Dxx 1014 s1, R
0.0020.2. With these estimates in mind, we may
illustrate application of the model to the basin- 11.3 Coupling of viscous flow and
and-range structures by computing Ld/H, qd, and macroscopic diffusional
(A/A)d for S 1 over a wide range in R and M. transport
Contours of these quantities, determined numeri-
cally, are shown in Fig. 11.15. Under a wide range of conditions, pressure solu-
The regularity of basin-and-range structure tion is a signicant mechanism of deformation in
(Fig. 11.12) suggests a strong necking instability, so rocks that contain a large volume fraction of mod-
that as previously discussed for folding (Chapter erately soluble minerals such as calcite and quartz.
10) we require qd 20. This quantity depends Pressure solution taking place at the grain scale
chiey on R (Fig. 11.15) and the constraint requires (Chapter 11, frontispiece) can result in pervasive
R 0.05. We may use the mean value of the mea- ductile deformation. For example, in this sand-
sured spans as an estimate of Ld, since the amount stone, dissolution occurs on grain surfaces and on
of extension associated with the later-stage basin- surfaces of the dark anastamosing solution seams
and-range structure is modest. Since the smallest approximately normal to the direction of maxi-
values of Ld/H are 4.6 to 4.8, this implies H mum compression, horizontal in the gure.
6.5 km. Although this is smaller than the typical Dissolved material diffuses to grain surfaces sub-
depths to the brittleductile transition in conti- jected to the intermediate or least compression
nental crust of 10 to 15 km, it may be reasonable and precipitates there, often in the form of bers,
in view of the higher thermal gradient in this as here. Precipitation occurred onto the ends of
region. We thus conclude that the present simple the bers as they incipiently pulled away from the
model is a plausible one for the initiation of basin- sub-horizontal surfaces of the grains. The rate of
and-range structure. Other models with a more deformation is determined by the rate of diffusive
realistic characterization of crustal rheological transport from sites of dissolution to those of
11.3 VISCOUS FLOW AND MACROSCOPIC DIFFUSIONAL TRANSPORT 441
Dxx
1
( yy) D
D 2
2
x y
2
2 (xx yy)
(11.80)
4 xx
1
Dyy (xx yy) D
4
(11.76)
where
f c0V02
4RT
1 The dimensions of are [ ] M1 L3 T.
Dxy xy
2 Combining (11.78) through (11.80) and (11.76)
Here twice D is the rate of dilatation, referring to yields the desired constitutive relations for the
the rate of change in the area of an element in the plane ow, including the diffusional transport of
plane of ow. the soluble mineral component. An inhomo-
In the case of pressure solution, the rate of geneous distribution of pressure or mean stress in
dilatation, twice D, is associated with dissolution, a viscous medium deforming in plane ow has
transport, and precipitation of the soluble, volu- been used to assess dissolution, diffusion, and pre-
metrically dominant, mineral in the rock. A cipitation within it (Stephansson, 1974), where
further assumption, point 4, is that there is no the divergence of the gradient in mean stress is
porosity within the material to act as a sink, nor used as a measure of dilatation. However, as we
can porosity be created by dissolution. Thus, dis- have learned, the Airy stress function, , in such
solution or precipitation must be balanced by a material satises the biharmonic equation, so
transport that at the scale of interest is by diffu- that:
sion along intergranular uid lms. Then: 2 2 2 2 2 2
( yy) 0
x2 y2 xx
x2 y2 x2 y2
J Jy
2D x V0 (11.77)
x y (11.81)
11.3 VISCOUS FLOW AND MACROSCOPIC DIFFUSIONAL TRANSPORT 443
Thus, in the present formulation, such a stress dis- The rate of dilatation is:
tribution would give no dilatation, even though
vx vy
the gradient in mean stress is not zero. The re- 2D (1 2)(mey ney) cos x
x y
lation (11.81) is obtained by substituting the stress
(11.86)
components into the constitutive relations for an
incompressible viscous uid, and these in turn into
the equation of compatibility. Repeating this 11.3.2 Necking of a power-law layer
here, but with the present constitutive relations, embedded in a viscous medium
we obtain: with macroscopic transport
To illustrate some aspects of the behavior of per-
2
2 2 2 2 vasive pressure solution (Fletcher, 1982), we
2
2 1 2 2
2 0 (11.82)
x y x y return to the study of boudinage, which serves as
a focus of interest and attention in this chapter.
To treat examples of deformation and diffu-
Consider the necking of a power-law uid layer in
sional transport in layered rock congurations,
a medium of the type formulated here. Pressure
we again seek a solution that is separable in x and
solution does not take place within the layer. The
y. This is:
relations derived from the boundary conditions
(x, y)
2
2
[a b(ly 1)]ey mey
are nearly the same as in (11.47) through (11.50),
but the right-hand sides are replaced according to:
[c d(ly 1)]ey ney cos x vsx(x, h) (c1 g1) sin x
~
1 v y(x, h) (c1 d1 g1) cos x
~
where 1 (11.83)
22
~yy (x, h) 21(c1 d1 g1) cos x (11.87)
The four terms containing the constants a, b, c,
~xy(x, h) 21(c1 g1) sin x
and d in this expression are identical to those used
in Chapter 10, and satisfy the biharmonic equa- 4 Dxx(A) (1 1 )
tion. The remaining two terms containing m and
Here we have also used the restriction to a plane
n yield a non-zero dilatation rate, 2D. The quantity
basic-state ow. Since there are now ve co-
has the dimensions of a length squared, and
efcients to be xed, the additional one in this
we write L* (2).
case being g1, the equivalent of n in the relations
The stress components derived from (11.83)
(11.83) to (11.86), another boundary condition is
are:
required. This is the vanishing of the normal com-
xx 2 {[a b(y 1)]ey [c d(y 1)]ey ponent of the diffusional ux at the interface.
2(mey ney)} cos x While there is a basic state of uniform exten-
sion, the rate of dilatation in the case of a layer in
yy 2 {[a b(y 1)]ey [c d(y 1)]ey which diffusion does not occur is zero. Since there
(mey ney)} cos x (11.84) is a jump in layer-parallel normal stress, and
hence mean stress, across the mediumlayer
xy 2 {[(a by)]ey (c dy)ey interface, the condition of zero diffusional ux
(mey ney)} sin x may not correspond to what is seen in a natural
deformation. In extension, the mean stress in the
The velocity components derived using (11.76) and layer is less than that in the medium, and the
(11.80) are: component of a soluble mineral in the medium
vx [(a by)ey (c dy)ey mey would tend to diffuse into the stiff layer and pre-
ney] sin x
cipitate. If such a situation were posited, the addi-
(11.85) tional boundary condition would be on the
vy {[a b(y 1)]ey [c d(y 1)]ey
continuity of the normal ux at the interface, but
(mey ney)} cos x for simplicity, we exclude such diffusion.
444 RHEOLOGICAL BEHAVIOR
2
enters, but parameters and , which also refer to
12 10
the medium, are not subscripted. At any layer
stress exponent, n, qd increases with L*/H and the
qd = 1000
1 increase is signicant for large n at which the
400
5 200 necking instability is otherwise moderate to
10 100 strong. This effect saturates as L*/H becomes
log10(L*/H)
20
40 comparable to Ld/H, or at approximately L*/H 10.
0 That is, as transport becomes efcient, the driving
gradient itself is lowered, so Ld/H and qd reach lim-
10 3.5 iting values and then do not change as L*/H
1 4 increases further.
At large n, or the plastic layer limit, the insta-
10 8 6 5 Ld /H = 4.5 bility becomes very large. Since much boudinage
occurs by separation of the stiff layer into discrete
2 segments by faults or shear zones, we might antic-
0 1 2 3 4
log10(n) ipate that this behavior is indeed characteristic of
layers forming this type of boudin. The present
model would thus be expected to provide a satis-
Fig 11.18 qd and Ld/H contoured in (n, L*/H)-space for
factory model for any initial phase of continuous
necking for viscosity ratio R 0.05.
necking that might have preceded through-going
faulting, as in the basin-and-range model.
The effect of diffusional mass transport in the
In the absence of basic-state transport, the con-
medium on necking is further illustrated by plot-
dition of zero normal ux at the interface, to rst
ting (Fig. 11.19) deformed grids and velocity elds
order in slope A, is:
for cases of no diffusion (L*/H 0) and vigorous dif-
1 ~ (1) ~ (1) fusion (L*/H 3.2). Only the grid deformation due
~(1)
J y (x, 0) ( xx ) 0 (11.88)
y 2 xx (x,0)
to the perturbing ow is shown; in (Fig. 11.19a),
the apparent thickening in the swell of the layer
To the present approximation, this yields the con-
or in the medium adjacent to the pinch would
dition:
be offset by the uniform basic-state extension.
1 Without diffusion (L*/H 0), the medium does not
d1 ( 2 1)g1 0 (11.89)
2 thin appreciably over the swell nor thicken over
the pinch or neck, but with diffusion (L*/H 3.2),
The ve equations obtained by combining
these effects are marked. Notice that the elements
(11.87) and (11.89) are solved numerically for the
above the swell also shrink, while those over the
constants a, b, c1, d1, and g1. The added complexity
neck undergo a positive dilatation. A third effect,
relative to the necking problem considered above
associated with a partial reduction of the mechan-
suggests that closed-form results may be laborious
ical constraint from the medium in the case of dif-
to obtain and may not easily provide physical
fusion, is that the perturbing deformation is
insight. We do not attempt this here.
markedly reduced away from the layer. The last
The effect on necking of dilatation mediated
effect is emphasized in the perturbing velocity
by diffusional transport in the host is shown by
elds (Fig. 11.19b); the medium velocity being
contouring qd and Ld/H in (n, L*/H)-space at xed vis-
much greater in the case of no diffusion except in
cosity ratio R 0.05 (Fig. 11.18) using:
the close vicinity of the layer.
A scalar measure of the strength of diffusive
L* 2 1
(11.90) transport is afforded by the ratio of the net rate of
H H2
diffusional volume transport towards the neck
This is a suitable dimensionless group for the per unit depth across a surface at x L/4 to the
present problem. Here, the medium viscosity transport by the perturbing ow through the
11.3 VISCOUS FLOW AND MACROSCOPIC DIFFUSIONAL TRANSPORT 445
(a) D =0 D =10
(b)
(c)
1 5 4
3
2.5
2
log10(L*/H)
0 1.5
1
Fig 11.19 (a) Grids with initially square elements
deformed using only the perturbing velocity without diffusion 0.5
(L*/H 0) and with vigorous diffusion (L*/H3.2). 0.2
(b) Velocity fields for the perturbing flow for the same cases 0.1
1 0.05
as in (a). (c) Ratio of the net diffusional volume flux per unit
depth across the vertical surface at xL/4 to the volume flux 0.01
per unit depth across this surface within the layer from the
perturbing flow only, contoured in (n, L*/H)-space.
0 1 2 3 4
log10(n)
446 RHEOLOGICAL BEHAVIOR
upper half of the necking layer towards the swell As a more concrete example, the sandstone in
(Fig.11.19c). This number may be greater than the frontispiece is made up of larger grains,
unity because the diffusional ux must also chiey of quartz, embedded in a ne-grained
counter the drag of material in the medium matrix rich in phyllosilicates and quartz, two
toward the swell. This suggests that a contribu- components that might be expected to have had
tion of slip at the layermedium contact, which distinct mechanical behavior. The larger grains
would diminish such drag, may enhance the insta- might have behaved as approximately rigid el-
bility further; the effect is not studied here. ements in a weak matrix. Evidently, though, the
large grains may be dissolved. The large grains are
elongate parallel to cleavage. Discontinuous
11.4 Continuum properties of seams depleted in ne-grained quartz, rich in
phyllosilicates, and aligned along cleavage are a
composite materials prominent if volumetrically minor additional
component of the rock. The presence of these and
Excluding volcanic glasses, rocks are made up of the elongate grains argue against isotropy of the
grains or crystals, and contain cracks and pores constitutive behavior, either in rheological or
that in situ will be lled with liquid or gas. At a elastic behavior. In shear parallel to cleavage, the
larger scale, rock masses are heterogeneous, from seams would possibly contribute to the deforma-
the relatively simple case of a layered sedimentary tion to a degree far outweighing their volumetric
rock to the more complex internal structure of a fraction. How might we quantify these features in
strongly deformed metamorphic rock. Treatment producing a model for the bulk behavior?
of these materials as continuous media, the
justication for which we have previously set 11.4.1 Voight and Reuss estimates for
forth, still leaves open questions. How may we esti- the bulk viscosity of a composite
mate the bulk properties of rocks relevant to of two viscous fluids
certain behavior, such as elastic deformation or The simple procedures for estimating the bulk
ow, from the properties of the major component properties of composite materials used here do
materials and their geometric conguration? As not always conform to the methodology so far pre-
the conguration of its mechanical components sented, in which we advocate the formulation and
changes in a ductily deforming rock, can the evol- solution of boundary value problems. In analyz-
ution in macroscopic properties be determined? ing these problems, the judicious use of approxi-
These questions relate to the correlation of mation is useful and often necessary, but
detailed observations of rock composition and conditions of traction continuity and continuity
structure with their constitutive behavior when of displacement or velocity, if appropriate, are
the deformation of interest took place. For honored. In the estimation procedures intro-
example, it is often remarked that the rheological duced here, however, this use of approximation
behavior of a rock in ductile deformation is deter- seems to have been taken too far!
mined by its weakest component, such as quartz As an example, imagine a composite material
in granite. Can we nd support for this idea made up of two isotropic linear viscous uids. For
through mechanical analysis of such a composite deniteness, imagine a conguration like that of
material? Such questions may be answered by sandstone, in which the quartz grains are
direct laboratory experiment. But these are replaced by one viscous uid and the ne-grained
sufciently expensive in time and resources to matrix by the other. Suppose that grain contacts
warrant discovering methods of estimation are welded: no slip or separation occurs along
whose reliability may be tested against particular them. Also, set aside the processes of dissolution
experimental results. Further, we need a concep- and precipitation that are central to the actual
tual basis with which to systematize experimental deformation of this rock. If there were a higher
results and to think about natural deformation. fraction of quartz grains, they would form more
In this section, we examine the simplest methods. complex interconnected bodies of this com-
11.4 CONTINUUM PROPERTIES OF COMPOSITE MATERIALS 447
ponent rather than isolated masses surrounded viscosities 1 and 2 and volume fractions f1 and
by the matrix component. The homogeneous f2 1 f1, the component Dxx will have the values
mechanical elements of the composite will then in the two components:
consist of isolated bodies to rather complex inter- 1 1
digitated bodies consisting of amalgamations of D (1)
xx s , D (2)
xx s (11.91)
21 xx 22 xx
grains composed of one or the other component.
By our hypothesis, that each component may be The macroscopic value of this component is then:
approximated by an isotropic viscous uid, the Dxx f1D (1)
xx f2 D xx
(2)
that determining the stress and velocity distribu- f1 f 2
1
10 n V ,s R (11.97)
9 Our present aim is to construct estimates for an
isotropic composite material, so this association of
Bulk viscosity/Lower viscosity
8
the Voight and Reuss estimates with the principal
Normalized
7 Voight Paul
viscosities of a layered, anisotropic material would
estimate seem no more than fortuitous. However, it does
6 give us a strong clue as to a signicant factor: the
VRH
Paul soft
5 cubical
effect of the two-component geometry of the com-
inclusions posite. By considering the layered conguration,
4 we observe that in layer-parallel shortening or
extension, both layer types support the bulk stress.
3
Paul stiff Thus, if one component has a very much larger vis-
cubical Reuss
2 inclusions cosity, its effect on the principal viscosity n will be
large, if not dominant. The component with
1 higher viscosity in this case may then make up the
0 0.2 0.4 0.6 0.8 1
f1 load-bearing framework, with intervening layers
of much smaller viscosity supporting little load.
Fig 11.20 Estimates of bulk viscosity. VRH refers to the On the other hand, in layer-parallel shear, the resis-
VoightReussHill estimate. tance, or lack of it, may be principally associated
with the low-viscosity component, with sheets of
the high-viscosity component acting as isolated
The Voight estimate is an upper bound on the vis-
inclusions. Indeed, both components support the
cosity of the composite, a result also proved by
same load, in terms of layer-parallel shear stress.
Paul (1960).
Thus, as might have been guessed beforehand, the
A simple way of showing that the two estimates
geometry of the composite may have a great range
are bounds and here we must be careful not to
even if it corresponds to isotropic bulk behavior,
claim we have obtained a proof ! is to consider the
and this will have a major effect on the bulk vis-
case when one component is rigid, or has innite
cosity. As suggested here, two principal types of
viscosity. The Voight estimate indicates that no
conguration exist, those in which the high-vis-
matter how small a volume of this material is con-
cosity component forms a connected load-bearing
tained in the composite, the composite itself will
framework, and those in which it is present as iso-
have innite viscosity. The Reuss estimate indi-
lated inclusions surrounded by a matrix of the low-
cates that no matter how small an amount of the
viscosity component (Handy, 1994). We turn to
material with nite viscosity is present, the vis-
another simple method of estimation that
cosity of the composite will be nite. With regard
accounts for this difference.
to the supposition that the rheological behavior of
a composite material is determined by the low-
viscosity component in it, the Reuss and Voight 11.4.2 Paul estimates for an isotropic
estimates say quite opposite things. The Voight viscous composite
and Reuss bounds for the bulk viscosity are shown The lack of dependence on the geometry of the
in Fig. 11.20 for materials consisting of two viscous conguration inherent in the Voight and Reuss
uids with 2 101 as a function of the volume estimates is not likely to be wholly appealing to
fraction of the uid with the lower viscosity, f1. the structural geologist. A simple method for
The Voight and Reuss bounds on the viscosity obtaining estimates that takes something of the
of an isotropic composite may also be recognized internal conguration into account has been
as the exact principal viscosities of a composite proposed (Paul, 1960). This makes use of the
made up of alternating layers of the two uids, as strength-of-materials plane-sections-remain-plane
used in Chapter 10 to model the gross mechanics approximation used in our initial analysis of
of chevron folds, with: necking. These estimates show the expected asym-
11.4 CONTINUUM PROPERTIES OF COMPOSITE MATERIALS 449
(a) (b)
f 1/3 f 123 (1)
xx (1 f 1 ) xx xx (11.100)
23 (2)
1 Here (1)
P is a Paul estimate with the 1-component
as an inclusion in a matrix of the 2-component.
Fig 11.21 Central sections of composite cubes used in the Performing the indicated algebra:
Paul estimate for macroscopic viscosity for a volume fraction
f 0.25. (1)
P f 123 (1 f 123)(2 1)
(11.103)
1 f 131 [ f 1 (1 f 1 )(2 1)](1 f 1 )
23 23 13
cosity 1 undergo a rate of extension: VRH 1 V R
(11.105)
1 (1) 1 (2) 1 2 1 1
Dxx
(11.99)
31 xx 32 xx
This estimate may be chosen if we have no infor-
To conform to the macroscopic applied stress, mation on the geometry of the composite, other
the average over the unit area of the section must than that it is consistent with bulk isotropic
be: behavior, or that the sample of interest is selected
450 RHEOLOGICAL BEHAVIOR
randomly from a population with variable geom- anisotropic behavior, given known or postulated
etry but given volume fractions of the two com- behavior of the components. The estimates dis-
ponents. The bounds provide information on the cussed here may also be applied to non-linear
possible range of variation. materials, although computation of the Paul esti-
Another estimate may be obtained by placing mates is then rather complicated. In the special
Paul elements in series, so that the softer material case that the two materials are power-law uids
is neither always an inclusion in the stiffer ma- and the stress exponents are equal, the computa-
terial or vice versa. The latter is more likely if f1 tion becomes quite simple.
is large, so we weight its contribution by f1 and
that of the former by 1f1, obtaining:
P (2) (1) 11.5 Anisotropic fluids and
f1 P (1 f1) P (11.106)
1 1 1 internal instability
All six estimates are plotted in Fig. 11.20 as a
function of the low-viscosity component volume In the treatment of large-scale structures, such as
fraction, f1, for the case 2/1 10. Note that the accretionary wedges, it has been usual to approx-
Voight and Reuss estimates manage to stay imate the rheological behavior, however varied
outside of the tangle of the other four estimates, within the structure, as isotropic. At the scale of
which supports their nature as bounds. The an exposure, when layered and foliated rocks
largest difference in estimates occurs for higher deform in a ductile manner, the resulting struc-
volume fractions of the high-viscosity component, tures generally imply strong anisotropy, but to
1f1 0.5. The difference between the Paul esti- introduce such behavior into models for deforma-
mates, for which explicit congurations of the tion at a much larger scale, it would be necessary
composite are associated, give us an impression of to specify something of the initial disposition of
the joint effect of conguration and the variables layering and to keep track of it during the defor-
f1 and 2/1. For example, a relative bulk viscosity mation. Further, with the development of small-
of 5 corresponds to f1 0.25 for the high-viscosity scale structures, such as folds, which break up
inclusion in a low-viscosity matrix conguration, the layering or foliation within volumes large rel-
but to f1 0.5 for the low-viscosity inclusion in a ative to the scale of folding, the bulk behavior at
high-viscosity matrix conguration, so that a this scale will exhibit a smaller degree of
range in volume fractions may be offset by the anisotropy. Thus, the use of isotropic constitutive
nature of the composite conguration. Again, at relations for deformation at the largest scales may
the same volume fraction, a substantial range in be justied as an approximation.
relative bulk viscosity may be achieved by chang- There is still a strong motivation to study the
ing the conguration. consequences of anisotropic behavior in this and
As a rock consisting of two or more major many other situations. For example, the forma-
mechanical components deforms, its internal tion of structures such as folds and internal boud-
geometry will change. If it began as an assemblage inage in rocks that are foliated, layered, or
of equant grains or mineral aggregates or a more otherwise anisotropic in their bulk behavior may
complex intermeshing conguration still consis- be modeled as an instability in an anisotropic
tent with bulk isotropic behavior, deformation medium. The results may be used to interpret
will tend to produce a material with anisotropic arrays of natural structures, many of which are
properties. Instead of treating idealized congura- quite complex in terms of spatial variation in
tions by the Paul method, the actual congura- length scales, structural morphology, and the
tion of the rock might be sampled and a intensity of structural development in a rock that
plane-sections-remain-plane analysis undertaken, might otherwise have been imagined to be ini-
both in extension parallel to geometrically esti- tially an approximately homogeneous mass of
mated principal directions and in shear. This layered sedimentary or metamorphic rock,
would provide a method of assessing the bulk gneiss, or schist.
11.5 ANISOTROPIC FLUIDS AND INTERNAL INSTABILITY 451
11.5.1 Deformation of an anisotropic we would initially explore the scales and pat-
viscous fluid terns of the deformation and inferred constitu-
Modeling of ductile deformation of a rock volume tive behavior in terms of these components.
is appropriately treated using the constitutive rela-
Consider an anisotropic linear viscous uid
tions for a continuum that captures the bulk
described by relations introduced in Chapter 10:
rheological behavior at the scale of interest.
Understanding that this approximation applies to 1 1
Dxx s ,Dxy s (11.107)
a composite material, we realize that suitable con- 2n xx 2s xy
stitutive relations may be needed to describe an- Here x and y are local axes in the material paral-
isotropic behavior, as for a layered or foliated lel to the principal axes of anisotropy. Consider
material, a composite mass with a strong shape continuous variations in three quantities: , the
fabric, or a polycrystalline material with lattice small deviation of the x-axis from the x-axis of
preferred orientation, or a material in which all of xed reference axes x and y, and the two principal
these elements contribute to rheological ani- viscosities n and s. Write the latter as:
sotropy. Further, all of these factors and others,
such as composition, will contribute to inhomo- n n(1 n),s s(1 s) (11.108)
geneity. In some cases, such inhomogeneity as in a
volume of metamorphic or sedimentary rock may Here n and s are spatially varying fractional devi-
be approximated by a set of a discrete layers or ations from the mean values. Referred to x and y,
bodies of other shape, e.g. inclusions, which are the constitutive relations are:
themselves homogeneous. This is useful when the
4nDxx [(1 m) (1 m) cos 4]sxx (1 m) sin 4 sxy
contrast in mean properties of the discrete bodies
is much greater than that exhibited by internal 4nDxy (1 m) sin 4 sxx [(1 m) (1 m) cos 4]sxy
heterogeneity. Types of structures arising in this (11.109)
case, such as folds and pinch-and-swell structures,
Here m n s.
have been treated earlier in this chapter and in
tion gives the distribution or eld. 2 2 2
3. Physical intuition is most readily obtained by 4( m 1) 2 sxy 2 s
y 2 x xy xx
focusing on the behavior of the individual com-
ponents.
4. To interpret a naturally deformed rock mass,
2
2n 2n
y 2
2 sxx 2m
x 2s
xy
sxy
(11.111)
452 RHEOLOGICAL BEHAVIOR
The perturbation in principal axis orientation A cylindrical surface to which the principal
and those in the principal viscosities occur in sep- axes are tangent and normal that may be associ-
arate terms, because any term containing a ated with (11.113) is:
product of a perturbing quantity is deleted in the
(x, y) y A cos (x y) (11.114)
linearization. They may thus be thought of as
giving separate effects even though for any Here A and x tan . We may think
volume element in the uid both are required to of the trace of a cylindrical surface (11.114) as a
specify the local rheological behavior. We might foliation surface; its mean height, y, above some
thus imagine one limiting situation in which the reference level, which denes, in part, a material
principal viscosities were perfectly homogeneous surface in the uid, is treated differently from the
but the principal axis orientation varied, and independent coordinate y. An example of a set of
another in which the opposite was true. This sep- foliation surfaces of the form (11.114) is shown in
aration is no longer possible when the dimen- Fig. 11.22 for the case 45.
sionless quantities describing them are both too The solution that we seek starts with the par-
large for the linearized equations to hold. ticular solution to (11.111), where only the rst
In contrast to the many examples of boundary term on the right-hand side is considered.
and initial value problems discussed in this text, Substituting (11.113) into that term, it is then clear
we now consider an initial value problem for an that sin(x y), and we obtain:
unbounded volume of material. The initial value
(1 2)(4m 1)[( 2 1)sxy2 sxx] sin (x y)
description now refers to the functions:
[1 (2m 1) 2 4]
(x, y), n n(x, y), s s(x, y) (11.112) (11.115)
Fig 11.22 Single component with 45. The Fig 11.23 Symmetric internal boudinage is the sum of two
maximum slope is chosen large enough so that the structure cylindrical perturbations with 45 and45.
is visually distinct.
11.23, which combines a perturbation of the sort Focusing on amplitude rather than slope, we then
shown in Fig. 11.22 with another of equal ampli- examine the dimensionless quantity:
tude and phase, but with or of opposite sign,
1 dA
45. In pure extension or shortening, the com- q(; m, dxx)
I2 A dt
ponents will behave in a symmetric manner, but
if there is a component of foliation-parallel shear, 4(m 1)[2 dxy (1 2)dxx]
d xx
they will not and the independent evolution of [1 2(2m 1) 2 4]
the two parts becomes obvious.
(11.122)
The relations (11.120) may be integrated
numerically to follow the evolution of the struc- Because d2xx d2xy 1,
the dependence on the basic
ture, as long as the slope remains small, here not state reduces to the single variable dxx.
on a single interface but throughout a volume In contrast to cases in which a layer of nite
of uid. We now use the above results to illus- thickness H undergoes folding or necking, there is
trate structures that might be produced in an no dependence on an absolute length scale, since
454 RHEOLOGICAL BEHAVIOR
1
0.8
Dxx /(Dxx2 + D 2xy)1/2
0.6
0
0.4 0.05
0.2 0.1
0 0.2
0.2
0.4
0.4
0.6
0.6
0.8
0.8
1
80 60 40 20 0 20 40 60 80
Band orientation (o)
lock-in by a mechanism as yet not identied to orientation for a given bulk shortening might
exist before decay sets in. Potential instability afford some indication of the strength of
exists in combined foliation-parallel extension anisotropy as described by the parameter m the
and shear, but the strength of instability in the basic-state deformation, and the character of the
present case is weak. Continued amplication initial perturbation.
with shear takes place in a basic-state ow with dxx
0.6, and aggregate amplication will outweigh
later decay for any value dxx 0.
11.6 Concluding remarks
By integrating the relations (11.120) through a
nite deformation, we may determine the ampli- This chapter has introduced a few of the factors
tude and orientation history of a component. If an which might be treated in more realistic models
initial set of perturbations with random initial of rock deformation: (i) rheological non-linearity;
slope, A, and phase, and chosen at equal intervals (ii) dilatation at a macroscopic scale mediated by
in orientation to foliation are followed, the nal diffusional transport; (iii) bulk properties of a
form of foliation planes may be computed as a composite material; and (iv) instability in the
summation. An example is shown in Fig. 11.25 for deformation of an anisotropic material. The
foliation-normal shortening. The initial shape of models formulated and analyzed are relatively
the rock element was square. Fold axial planes are simple and to a substantial degree build on ma-
not perfectly normal to the direction of shorten- terial presented in earlier chapters. The rst two
ing and they are nite in extent, even as seen in topics were studied in application to the initia-
this nite segment of the medium. They are tion of pinch-and-swell structures in extension, as
nearly normal to the direction of shortening, as well as to folds and mullions, to suggest the pos-
expected from the rate of shortening maximum sibility of the study of a wide range of behavior in
there (Fig. 11.24). structures such as folds or boudinage. Overall,
At present, little quantitative interpretation though, the presentation is one of brief sketches
of structures in foliated and multi-layer rocks has of restricted subject matter and methods. While
been carried out, and this eld of study remains these sketches often supply means of carrying out
open. Figure 11.24 suggests that the continuity the modeling of naturally deformed rock, they
of axial planes and the distribution in their also provide a means of gaining physical insight.
Chapter 12
Black Mesa
Maiden Creek
Oblique aerial photograph of the eastern slope of Mt. and how they work. (2) Then one can nd exactly
Hillers in the Henry Mountains, Utah. The Black Mesa where each part is located in the system in relation to
laccolith and Maiden Creek sill form the black, cliff-forming all the other parts and how they interact with each
outcrops. other. These two approaches are unlikely, by them-
selves, to reveal exactly how the system works. (3) To do
this one must also study the behavior of the system
Three main approaches are needed to unscramble a and its components while interfering very delicately
complicated system. (1) One can take it apart and char- with its various parts, to see what effect such alter-
acterize all the isolated bits what they are made of ations have on behavior at all levels (Crick, 1988).
12.1 IDEALIZATION OF FIELD OBSERVATIONS 457
N
ature, or what we might call natural 7.8). The frontispiece for this chapter is an oblique
reality, can appear to our senses as a very aerial photograph of the eastern ank of Mt.
complicated system when we view geo- Hillers showing laccolithic intrusions at Black
logical structures in outcrop. Francis Crick (1988) Mesa and Trachyte Mesa. Gilbert was one of the
also faced a complicated system when viewing the masters of scientic methodology in the practice
constituents of living cells. He suggests in his of geology, and we have much to learn from study-
book What Mad Pursuit that one should rst char- ing the way he worked. We give an example of the
acterize all the parts of the system and then procedure of idealization, recognizing that this
understand their geometric relationships. This is will not provide one with all the knowledge
what we attempt to do as structural geologists required to implement the procedure. Indeed, to
when mapping structures in the eld. In Chapters perfect this procedure requires considerable ex-
2 and 3 some of the useful tools for mapping were perience mapping structures in the eld, and in-
described and the principles of differential geom- timate familiarity with continuum mechanics.
etry were introduced to provide the fundamental Nevertheless, we introduce the procedure now in
basis for characterization of structures. Unlike the a specic, and perhaps narrowly focused, manner
cell in a test tube, many parts of typical geological using the example of Gilberts discovery of laccol-
structures are inaccessible because of limited ites (now called laccoliths) and his research into
exposure, and usually there are limited data on the origins of the Henry Mountains (Fig. 1.16). It is
the temporal development of structures. Crick assumed that the reader is familiar with the intro-
then suggests one must study the system as a duction to Gilberts research provided in Chapter
whole to understand how it behaves when various 1 and with his conceptual model for laccoliths
parts are perturbed. This step is possible when the (Fig. 1.17).
system is a cell in a test tube, but generally it is
impossible when the system is a rock mass larger 12.1.1 G. K. Gilberts field observations
than a cubic meter. Consequently, we turn to in the Henry Mountains
laboratory and mathematical models of geologic Gilbert arrived on the western edge of the Henry
structures to carry out this step in Cricks pre- Mountains in mid August of 1875 by horseback.
scription for scientic inquiry. Models are con- Intrigued by the range of mountains he saw to the
structed with specic features and attributes of east, Gilbert wondered if they might be a good
the rock mass, and the model system is studied to place to consider the question of whether volcanic
understand how it works. mountains were just piles of lava ows or perhaps
This chapter focuses on the development of the result of doming by injection of magma at
mathematical models of geologic structures. In depth. On August 18, 1875, Gilbert observed the
particular we consider the idealization of observed sedimentary rocks circling the base of Mt. Ells-
structures and the selection of general boundary worth and dipping away from the summit (Fig.
conditions for model development. At the end of 12.1). The next day he speculated that Mt. Hillers
this chapter we present a methodology for the must be capped with trachyte (igneous rock) and
practice of structural geology that summarizes the again observed the sedimentary units dipping
underlying concepts that have been described and steeply around the base of the mountain in con-
utilized throughout this book. trast to their nearly horizontal aspect further
from the mountain.
On August 21st Gilbert approached the base of
Mt. Hillers and recorded in words and a sketch
12.1 Idealization of field (Fig. 12.2) the scene that he observed:
observations
Camp 38 is on the SE base of Hillers. We have found
several minute springs in skirting the mountain and
We begin this chapter by describing how G. K. this one barely sufces us. It cannot be depended on as
Gilbert (1877) idealized the laccolithic intrusions permanent . . . The rock which rises toward Hillers
in the Henry Mountains of southeastern Utah (Fig. from the south is the B cliff (Ferron Sandstone). It is
458 MODEL DEVELOPMENT AND METHODOLOGY
model results should diminish. If these parame- possible models for these structures. Here we give
ters and conditions were included, the model may examples, all based upon the structure that
approach the actual behavior of the natural Gilbert set out to investigate over a hundred years
process more closely. On the other hand, the addi- ago in the Henry Mountains, and show how dif-
tional parameters and conditions may be less well ferent questions, posed in the eld about the
constrained by actual data. A master of this kind same structure, lead to different choices for the
of analysis, Art Lachenbruch, describes this context of modeling, that is to different general
dilemma in the following way: boundary conditions.
In selecting a formal model there is a trade-off; as its Because tectonic processes are multifaceted,
complexity increases, we can usually conclude less often involving solid and uid deformation, heat
from it with more condence. In this paper we select a and mass transport, and chemical reactions, rarely
simple model and attempt to conclude a lot from it, is there a single choice of general boundary condi-
while recognizing that the literal application of these tions that enables one to address all the interesting
conclusions is questionable, but that the insight is questions that arise from eld observations.
likely to be useful (Lachenbruch, 1973). Furthermore, we are limited in our understanding
It is relatively easy, given the power of modern of how these various processes are coupled
computers and the sophisticated tools available together, and the solution methods for coupled
for modeling complex mechanical systems, to problems may be difcult to implement. Finally,
create a model of a geological process that is itself the solutions to coupled problems may be so
too complex to understand, or that is too poorly complex that they are difcult to understand and
constrained by data. In hindsight it is clear that therefore are not readily applied to the simple
Gilberts piston laccolith was simple enough to questions asked in the eld. For all of these reasons
give him insight, and practical in the sense that we choose to break up tectonic processes into com-
the necessary eld data could be gathered to test ponent parts following the methodology des-
the hypothesis derived from the model. cribed in the quote at the beginning of this chapter
In general terms the procedure we have (Crick, 1988). Each part is analyzed according to a
described in this section is not new. Indeed, it can particular choice of general boundary conditions.
be traced back to the reductionist methodology Thus, the selection of general boundary conditions
rst articulated by Descartes in 1619 (Davis and is inuenced by practical necessity and guided by
Hersh, 1986). It is one of the cornerstones of an underlying reductionist philosophy of science.
modern scientic research. Once we have made the choice to work within
a particular area of continuum mechanics such as
plate theory, elasticity theory, viscous uid
12.2 Selection of general boundary mechanics, or heat conduction, the most general
differential equations for that area usually are
conditions simplied by eliminating certain variables or
terms from these equations. This elimination may
General boundary conditions dene the context be based on eld observations that justify, for
of a problem, so the appropriate theoretical prin- example, the reduction of a three-dimensional
ciples, fundamental laws, governing equations, structure to a two-dimensional approximation for
and computational tools can be brought to bear that structure. Or, for example, the elimination
on nding a solution. The selection of general may be based on a choice to ignore the initial and
boundary conditions is not a simple matter, and nal phases of deformation during which a ow
no easily written prescription can address all the eld changes in time, so a steady-state approxima-
variations and complexities one is likely to tion can be used to describe the well-developed
encounter. This procedure requires an under- ow eld that existed between these phases. In
standing of both the eld observations that serve other cases dimensional analysis is used to quan-
to characterize the structures, and the physical tify the relative importance of terms in the gover-
principles and scaling relationships that underlie ning equations and eliminate those that are
12.2 SELECTION OF GENERAL BOUNDARY CONDITIONS 463
negligible. In this section we describe a variety of with some power of its depth, but I am unable to say
simplication procedures, and the methods that what power. So far as I am aware, neither mathematical
are employed to make and to justify these choices. analysis nor experimentation has been directed to the
The procedure for idealization, described in the problem in question. According to Rankine the resis-
previous section, and the selection and simpli- tances of exure of similar cross-sections (of elastic
beams) are as their breadths and as the squares of their
cation of general boundary conditions are not
depths (Applied Mechanics, page 316), and it is possi-
independent. The former is described as a eld pro-
ble that the same law applies to the resistances which
cedure and it should, for the most part, be under- continuous strata oppose to the uplifts of domes. But it
taken during mapping or data collection, when the appears more probable that the greater complexity of
decisions can be evaluated by direct observations. the strains developed in the formation of domes causes
The latter is described as an ofce procedure, and it the depth to enter into the formula with a higher power
is likely to be undertaken with a continuum than second (Gilbert, 1877).
mechanics textbook in hand. In fact most struc-
tural geologists, including Gilbert, intertwine To address questions about the shape of the
these two procedures. Apparently Gilbert carried domed strata and their resistance to bending a
an engineering handbook into the eld and pre- number of models have been proposed since
sumably used it to help conceptualize and set up Gilberts report based on a theory for the
problems (Pyne, 1980). An idealization may be deection of thin plates (Johnson, 1970; Pollard
designed specically to take advantage of a known and Johnson, 1973; Koch et al., 1981; Kerr and
solution, for example in elasticity theory or uid Pollard, 1998).
mechanics. Later, having found a solution for a The context, or general boundary conditions,
more realistic set of parameters, certain cons- for the analysis of thin plates is the engineering
traints imposed in the former idealization may be discipline called strength of materials or plate theory.
relaxed. In this way an analysis may progress, for Reference textbooks are available on this subject
example, from one employing a linear and iso- from the engineering literature (Timoshenko,
tropic viscous uid (Chapter 10), to an anisotropic 1958; Timoshenko and Woinowsky-Krieger, 1959).
viscous uid, to a uid with more complex rheology In Section 4.3.1 the dimensionless group for plate
(Chapter 11). This give-and-take relationship bending was derived, (4.45), and showed that dis-
between the idealization of eld observations and placements scale with the fourth power of the
the selection of general boundary conditions for length and inversely with the elastic stiffness and
modeling is vital to the development of effective the third power of the thickness. Thus, bending
models and the analysis of structural data. under distributed loads is most sensitive to the
geometry of the plate.
12.2.1 Plate theory: what controls the To apply plate theory to the laccolith problem
shape of the domed strata over a the sedimentary strata overlying the laccolith are
laccolith? idealized as a stack of n thin plates of thickness, hi
It is clear that Gilbert understood the strata over (Fig. 12.6a). Particular plates may or may not cor-
laccoliths in the Henry Mountains did not behave respond to stratigraphic units, but each behaves
as rigid pistons (Fig. 1.18) because his conceptual as an independent mechanical unit, capable of
model that preceded the piston laccolith shows slipping relative to adjacent mechanical units
the strata bent into a domed shape (Fig. 12.5). He along bedding-plane faults. Unlike Gilberts rigid
noted that resistance to uplift for the piston piston, these plates deform according to linear
model is proportional to the overburden thick- elastic relationships among the stress and inni-
ness and speculated in his report of 1877 that tesimal strain components dened by Hookes
resistance to bending might increase with some Law (Chapter 8). For plates that are isotropic and
greater power of the overburden thickness. homogeneous with respect to elastic properties,
the two relevant material constants for each plate
I am led by the analogy of allied problems in mechanics are Youngs modulus, Ei, and Poissons ratio, i.
to assume that the resistance of the body of strata varies The plates are bent by upward directed forces due
464 MODEL DEVELOPMENT AND METHODOLOGY
Normalized deflection
at the distal edge of each plate are that both the
0.030
deection and the slope are zero.
A uniform distribution of pressure, P con- 0.025
stant, is dened as the difference between the
magma and lithostatic pressures, Pm Pw. For 0.020
plates that are able to slide freely over one another
(no shear tractions across the interfaces), the 0.015
exural rigidity is the sum of the individual
0.010
exural rigidities, and we refer to this as the effec-
tive flexural rigidity, Re (Pollard and Johnson, 1973):
0.005
n Eih3i
Re
i1 12(1 vi )
2
(12.2) 0
0 0.2 0.4 0.6 0.8 1
The plates are numbered from 1 to n starting at Normalized position, x/a
the base of the stack (Fig. 12.6); hi is the thickness Fig 12.7 Plot of normalized vertical displacement
of the ith plate; and Youngs modulus and (deflection) versus position for bending elastic plates with
Poissons ratio for this plate are Ei and i, respec- uniform distributed loads. Because of symmetry only half the
tively. distribution is shown.
The solution to (12.1) for the deection of any
one of the uniformly loaded elliptical plates is
(Love, 1944; Timoshenko and Woinowsky-Krieger, sure and to the fourth power of the half-length, a,
1959): so it is considerably more sensitive to changes in
the length of the plate, than to changes in pres-
2 2 2
x y sure.
1
(Pm Pw) a c The deection is illustrated in Fig. 12.7 where
w (12.3)
8Re 3 2 3 the dimensionless deection is plotted versus the
dimensionless distance from the center of the
a4 a2c2 c4
bent plate for both the circular and anticlinal
The two end-member cases simplify this two- plans. Only positive values of x/a are shown
dimensional problem to one spatial dimension because the solution is symmetric about x/a 0.
where the deection distributions are: All deection distributions for the general ellipti-
cal plan shape would fall between these two so
(Pm Pw) 4
w (a 2a2x2 x4)(anticlinal) the graph illustrates the entire spectrum of
24Re deections for this model. The form of the model
(12.4) deection is double hinged, that is a distal
concave upwards hinge surrounds the edge of the
(Pm Pw) 4
w (a 2a2x2 x4) (circular) (12.5) dome and gradually changes to a limb of nearly
64Re
constant dip where the curvature reverses sign to
The deection distribution along any cross form the concave downward hinge marking the
section in the (x, z)-plane is the same for the anti- center of the dome. This form is qualitatively
clinal plan shape. The deection distribution similar to the form envisioned by Gilbert in one of
along any radial line from the origin is the same his idealizations of a laccolith (Fig. 12.5) and to the
for the circular plan shape. Note that the deec- form illustrated in his restored section of Mt.
tion is directly proportional to the driving pres- Ellsworth (Chapter 1, frontispiece).
466 MODEL DEVELOPMENT AND METHODOLOGY
For a comparison of the model deection dis- opportunities for the development of bedding-
tributions to eld data we turn to maps and cross plane faults and shear zones within softer layers
sections for the southern Henry Mountains that would act to delaminate the overburden into
(Jackson and Pollard, 1988, 1990). On the geologic thinner mechanical units. Bedding-plane fault
and structural map of Mt. Holmes (Fig. 12.8) note zones are exposed within the sandstones and at
that the sedimentary strata are continuous over formational contacts (Fig. 12.9), indicating that
the top of the dome, and the beds dip more or less the overburden behaved in these places as a stack
in radial directions around the mountain. of mechanical units that could slip over one
Furthermore, the magnitude of the dips is small another. Measurements of bedding-plane faults at
far from the mountain, less than 10; increases to Mt. Holmes demonstrate that they are spaced
over 20 on the steep anks, and decreases to less from 150 to 200 m apart, so this may be a typical
than 10 near the top. The cross section along the thickness of the mechanical units (Jackson and
line AA (Fig. 12.9) is based on the bedding atti- Pollard, 1988). In this way the total overburden
tudes measured at exposures around the moun- thickness of about 4 km could be reduced to an
tain, and the geologic and topographic maps. This effective thickness, he, that was considerably less
cross section reveals a form of bending that is than the diameter of the domes.
similar to the plate model (Fig. 12.7). In addition The effective thickness of a stack of bending
to the doubly hinged structure the cross section plates is equal to the thickness of a single plate
reveals a gently dipping peripheral limb that that would resist bending just as much as the
extends 34 km beyond the lower hinge for all entire stack. For mechanical units with the same
three of the southern Henry Mountains (Jackson elastic properties, E and , that are able to slide
and Pollard, 1990). This does not correlate with freely over one another, the exural rigidity
the simple plate model, and may indicate the pres- reduces to the following simple form:
ence of underlying sills and smaller laccoliths n
E Eh3e
around the anks of the central laccoliths. Re
12(1 2) i1
h 3
i
12(1 2)
(12.6)
The correlations between the plate model
deections and the cross-sectional shapes for Mt. Note that the resistance to bending in this equa-
Ellsworth and Mt. Hillers are more difcult to tion scales with the cube of the effective thick-
assess because the upper parts of these domes are ness. Thus, if the effective thickness is less than
eroded. Furthermore, whereas maximum limb the total thickness, there can be a profound effect
dips of 20 at Mt. Holmes are just within the range on the resistance to bending. For example, if h
permitted for the application of plate theory, the 4 km, then h3 64 km. However, if this overburden
dips are between 50 and 55 at Mt. Ellsworth, and is delaminated into four equal mechanical units
between 75 and 85 at Mt. Hillers. These intrusive by bedding-plane faults, then hi 1 km each, n 4,
structures have developed beyond the stage where and h3e 4. Delamination into four mechanical
elastic plate theory should be applied. However, if units reduces the resistance to bending by a factor
these domes passed through an earlier stage of of 16. Because bedding-plane faults would offer
development when, according to Gilberts con- frictional resistance to sliding and, as pointed out
ceptual model, the sedimentary overburden was in the next section, these faults are unlikely to
domed much like that at Mt. Holmes, plate theory develop over the entire laccolith, this analysis
would be applicable for this earlier stage. over-estimates the reduction in exural rigidity.
The diameters of the southern Henry None-the-less the mechanical effect of bedding-
Mountains domes, 2a 10 to 14 km, are only two plane faults on the development of laccoliths is
to three times the total overburden thickness, h likely to be signicant.
4 km, yet applications of plate theory require a In summary, the plate theory provides insight
ratio of diameter to thickness greater than about concerning the resistance to bending of sedi-
eight. On the other hand the sedimentary section mentary strata over laccoliths, and the solution
is composed of a multitude of sandstone, silt- for uniform loading gives a deection shape that
stone, and shale beds, and this bedding provides is similar to the early stages of doming at
12.2 SELECTION OF GENERAL BOUNDARY CONDITIONS 467
Fig 12.8 Map of Mt. Holmes in the southern Henry Mountains, UT. Black is diorite porphyry; white (N) is Navaho
sandstone. Reprinted from Jackson and Pollard (1988) with permission of The Geological Society of America.
468 MODEL DEVELOPMENT AND METHODOLOGY
Bedding-plane
faults
A
Fig 12.9 Cross section of Mt. Holmes identified as AA sliding over one another (Jackson and Pollard,
on Fig. 12.8. Note bedding-plane faults. Reprinted from 1988). Plate theory indicates the importance of
Jackson and Pollard (1988) with permission of The this delamination mechanism in reducing the
Geological Society of America.
effective thickness of the overburden and thereby
reducing the resistance to bending so the laccol-
Mt. Holmes. To answer the question posed at the ith can grow in amplitude. However, these
beginning of this section, the resistance to mechanical units and the boundary conditions
bending apparently exerts an important control between them are prescribed in setting up a
on the shape of the domed strata, and this resis- problem in plate theory, so one cannot address
tance is strongly dependent upon the thickness the questions posed in the title of this section.
of the mechanical units in the sedimentary Instead one can use a model based on Gilberts
sequence. The shape of the domed strata also concept that the earliest stage in the development
depends upon the pressure distribution in the of laccoliths is the insinuation of a thin horizon-
laccolith, the distribution of slip between the tal sill of magma between the strata (Fig. 12.10a).
mechanical units, and the nature of the boundary When the horizontal dimension, 2a, of such a sill
conditions at the distal edge of the laccolith, all of is less than the depth, d, the resistance to bending
which have been investigated using plate theory of the thick overburden is so great that opening of
(Pollard and Johnson, 1973; Koch et al., 1981; Kerr the sill is accommodated primarily by the elastic
and Pollard, 1998). Additional insights have been compression of the surrounding rock, both above
gained by addressing the silllaccolith transition and below the sill. The dashed lines in Fig. 12.10a
using elasticity theory (Zenzri and Keer, 2001). schematically represent how bedding planes
would displace upward and downward to accom-
12.2.2 Elasticity theory: how and where modate opening of the sill. Note that the dis-
do bedding-plane faults form placements decrease away from the sill and are
over sills? negligible at Earths surface.
Field observations of bedding-plane faults within The state of stress on horizontal planes around
strata overlying the laccoliths at the three south- a model sill are calculated to determine where
ern Henry Mountains provide evidence that the bedding-plane faulting might initiate (Jackson
overburden was sub-divided into mechanical and Pollard, 1990). The context, or general bound-
units on the order of 200 m thick and capable of ary conditions, for this analysis is elasticity theory
12.2 SELECTION OF GENERAL BOUNDARY CONDITIONS 469
12.11c), this interaction is signicant and the laccolith (Fig. 4.12). The context, or general bound-
regions of positive Coulomb stress extend from ary conditions, for the analysis of magma ow is
the surface to the sill tips, and these regions that branch of uid mechanics devoted to the ow
extend laterally to a distance of almost 4 km from of uids with mechanical properties that include
the point immediately over the center of the sill. viscosity and strength. Reference textbooks on
The overburden out to this distance is susceptible uid mechanics provide the background for
to the development of bedding-plane faults, model development and include solutions for
except for the region immediately over the center ow of materials with a variety of physical prop-
of the sill where enhanced vertical compression erties in conduits of various shapes (Lamb, 1945;
would prevent frictional slip. Delamination is pre- Schlichting, 1979; Landau and Lifshitz, 1960;
dicted to develop above the distal margin of the White, 1974).
sill and not immediately over its center. This dis- In Section 4.3.2 we used dimensional analysis
tribution of bedding-plane faulting would to identify Reynolds Number (4.55) as the scale
promote the formation of laccoliths with at tops factor for viscous ow in conduits (Reynolds, 1883;
and monoclinal bending over their periphery White, 1974). This dimensionless group is a ratio of
(Koch et al., 1981). inertial to viscous forces in the owing uid and is
From the elasticity theory we understand that proportional to the conduit width, uid density,
the stress perturbation associated with the lateral and characteristic velocity, and inversely propor-
growth of a sill enhances the shear stresses and tional to the Newtonian viscosity. Reynolds lab-
lowers the normal compressive stresses on hori- oratory experiments (Fig. 4.8c) demonstrated that
zontal bedding planes in such a way that faulting the ow regime is laminar for numbers less than
is likely on weak bedding planes above the advanc- about 2000. The apparent viscosities of silicate
ing sill tip. As the sill approaches a length equal to liquids (magma) have been measured for a wide
the overburden thickness, the delamination variety of chemical compositions and water con-
spreads all the way to Earths surface and we would tents over the range of melting temperatures
anticipate a transition to laccolithic bending, (Shaw, 1963, 1969; Shaw et al., 1968; Murase and
accommodated by sliding of the mechanical units McBirney, 1973; McBirney and Murase, 1984; Ryan
over one another along bedding-plane faults. and Blevins, 1987). Given the great viscosity of
Elasticity theory and the Coulomb failure criterion most magmas and the modest velocities for ow in
explain how and where these faults form, and this sills the regime is likely to be laminar.
prediction is consistent with eld observations of For the purpose of a simple example we con-
bedding-plane faults at Mt. Holmes in the Henry sider an isothermal uid, the model magma,
Mountains (Fig. 12.9). owing in a tabular conduit of length 2a and
height 2h, surrounded by rigid host rock (Fig.
12.2.3 Viscous fluid mechanics: how 12.12a). A feeder dike at the center of the sill sup-
rapidly can magma flow into sills? plies the magma. Of course hot magma emplaced
The questions asked in the previous two sections into cold sedimentary rock will loose heat to the
focus attention on host rock deformation during surroundings (Lovering, 1935, 1936; Jaeger, 1957,
sill and laccolith formation. The mechanical role 1964b), but rock is a good insulator, so it is not
of the magma is reduced to providing a pressure unreasonable to postulate that the temperature
on the stack of plates or on the elastic material change is insignicant over the time required for
surrounding the sill. In other words the mechani- the sill to propagate to the transition length for
cal action of the magma is replaced by the appro- laccolith formation. In Section 4.2.2 we reviewed
priate distribution of tractions in the form of a the solution for conductive heat loss from a
boundary condition. This is an effective way to tabular intrusion of magma and showed that
simplify the mechanical system composed of both emplacement times on the order of a few days are
injecting magma and deforming host rock, but it consistent with this postulate (Delaney and
necessarily means that one cannot address ques- Pollard, 1981, 1982). As the previous sections indi-
tions about the rate of development of the sill or cate, the sill grows in length and thickness as
472 MODEL DEVELOPMENT AND METHODOLOGY
vx f(z), vy 0 vz (12.10)
(b) z
Because vy and vz are both zero, the continuity
equation (Section 7.3.2) insures that vx does not
vary in the x-direction. Because we postulate that
Pf vx x Pt all cross sections parallel to the (x, z)-plane are
2h
identical, vx does not vary in the y-direction.
Furthermore, the steady-state condition requires
h
that vx is not a function of time.
With these constraints on the gravitational
a
and velocity vectors, the NavierStokes equations
for the pressure and velocity distributions reduce
Fig 12.12 Idealized model for flow of viscous magma in a
sill. (a) Geometry of sill conduit with feeder dike. (b) Velocity to:
profile in model sill. p d2v p p
2x 0, 0, g* 0
x dz y z
magma is injected from the feeder dike. Here the (12.11)
ow is determined for conduits with particular
The rst equation governs the rate of ow in the x-
ratios of thickness to length, representing differ-
direction and the distribution of this velocity
ent moments during development of the sill. The
from the top to the bottom of the model sill. The
solution does not describe the progressive devel-
second and third equations, respectively, require
opment from an early stage to a later stage in this
the pressure to be constant in the y-coordinate
process and neglects the vertical component of
direction, and the pressure to vary linearly in the
velocity. We also neglect any consideration of the
vertical z-direction in proportion to the unit
propagation mechanism or ow near the tip of
weight of the magma. The distribution of velocity
the conduit, which is postulated to advance with
from the bottom to the top of the model sill is
the same velocity as the magma.
found by integration with a no-slip boundary con-
For the purpose of this example we postulate
dition at the conduit walls:
that the model magma has a constant density, ,
and behaves like an isotropic Newtonian viscous
uid with constant viscosity, . In other words
this uid obeys the Stokes condition described in
vx
h2 p
2 x z2
z2
1 2 vx(max) 1 2
h h (12.12)
Section 7.4.3 and the constitutive law given in The maximum velocity is at the center, z 0, and
(7.166). The viscosity of magma increases dramati- is proportional to the square of the conduit thick-
cally as temperature decreases, but we have ness and the pressure gradient in the ow direc-
already postulated an isothermal ow, so it is con- tion, and inversely proportional to the
sistent to ignore changes in viscosity. Under these Newtonian viscosity. The distribution of velocity
conditions Cauchys Laws of Motion reduce to the across the conduit is symmetric and parabolic
NavierStokes equations (7.170) which describe (Fig. 12.12b).
12.2 SELECTION OF GENERAL BOUNDARY CONDITIONS 473
vx(max)
2
h2 P t Pf
a (12.13) 1.E-03
1.E-04
Here 2h and 2a are the thickness and length of the 1 MPa s
sill, whereas Pt and Pf are the magma pressures at 1.E-05 100 MPa s
the tip and the feeder. For the Henry Mountains 10 000 MPa s
1.E-06
0 200 400 600 800 1000 1200 1400
examples, the sills that developed into the larger
laccoliths formed at about 4 km depth where the Sill length (m)
lithostatic pressure would be about Pw 100 MPa. Fig 12.13 Plot of magma velocity versus sill length for
The magma pressure in excess of the lithostatic three different viscosities.
pressure at this depth has been estimated to range
from 30 to 70 MPa, based on the density of the
magma, the density stratication of the host rock, would range between about 105 and 1 m s1. This
and the depth to the source of the magma implies that sills would take between about 103 s
(Johnson and Pollard, 1973). Taking 50 MPa as rep- (25 minutes) and 108 s (5 years) to develop to
resentative of this excess pressure, the total the transition stage. While the time scale is not
magma pressure at the feeder of the sill would be very well constrained, it clearly is a very short
Pf 150 MPa. We postulate that the magma pres- time compared to geologic eons.
sure would decrease to the lithostatic pressure at The results we have just obtained ignore the
the tip of the sill, so Pt 100 MPa. To be consistent deformation of the rock into which the magma is
with the previous section, we use a thickness to injected. A number of papers have treated the
length ratio of h/a 1/200. If the viscosity is taken coupled problem of host rock deformation and
as 102 MPa s, a value representative of silica- magma ow during dike and sill emplacement
rich magma (Johnson and Pollard, 1973), a rela- (Spence and Turcotte, 1985; Lister, 1990; Lister and
tively short sill, a 200 m, would have a maximum Kerr, 1991; Rubin, 1995). Comparing the energy
velocity of about 0.001 ms1 (Fig. 12.13, square consumed by fracturing, which is taken as inde-
symbols). For this sill nearing the transition to lac- pendent of the intrusion length, and that con-
colithic bending of the overburden, a 1500 m, sumed by viscous ow, which is taken as
the lesser pressure gradient would tend to increasing linearly with intrusion length, it is
decrease the velocity, but this is more than com- clear that viscous dissipation will dominate
pensated for by the greater sill thickness, so the beyond some critical length. For typical laboratory
maximum velocity would be about 0.01 m s1. values of fracture energy this length is on the
Magma viscosity is very sensitive to tempera- order of 1 m. However, it has been suggested that
ture and water content, and the constitutive the fracture energy may not be constant and that
properties of the magma are likely to be non- the region of inelastic deformation at the tip of a
Newtonian (Johnson and Pollard, 1973). These dike or sill may increase in size with the length of
factors are not well constrained for the intrusions the intrusion (Rubin, 1993). Under these condi-
in the Henry Mountains, so the value of viscosity tions the fracture energy should not necessarily be
chosen above could be in error by a couple orders neglected, particularly where geological evidence
of magnitude. Based on the linear relationship supports the development of such large regions of
between viscosity and velocity found above, this inelastic deformation (Delaney et al., 1986).
would imply differences in velocity of a couple As sills propagate laterally and then bulge
orders of magnitude. To illustrate this range of upward to form laccoliths, magma must be con-
behaviors we plot velocities for 10 and for 104 tinually injected from below. This injection of
MPas in Fig. 12.13. From this plot we conclude that magma results in the transport of heat from a
the maximum ow velocity for a viscous magma source at greater depth into and throughout the
474 MODEL DEVELOPMENT AND METHODOLOGY
shallow intrusion. To address questions about the 8.15). The outcome of a structural investigation is
cooling of the magma in tabular intrusions, and judged to be successful if there is a compelling
heating of the host rock, a number of models have correspondence between the views provided by
been investigated, some that consider heat ow observation and modeling.
from a stagnant magma (Lovering, 1935, 1936; At times these different views of structural
Jaeger, 1957, 1964b; Irvine, 1970), and others that geology seem too disparate to be reconciled. The
consider the transport of heat within and from a one is the world of boots and backpacks, rock saws
owing magma (Delaney and Pollard, 1982; and microscopes, maps and photographs, com-
Habert and De Saint-Blanquat, 2004). The context, passes and measuring tapes. The other is the
or general boundary conditions, for these analy- world of vectors and tensors, material continua
ses is the subject of heat transfer, and there are and differential equations, keyboards and com-
many useful textbooks on this subject (Carslaw puters, graphs and numbers. For some practition-
and Jaeger, 1959; Bird et al., 1960). ers of structural geology a choice is made at an
early stage in their education that closes the door
on one of these worlds in favor of the other. One
12.3 A methodology for the objective of this textbook is to encourage struc-
tural geologists to integrate these two worlds and
practice of structural geology discover the benets of both.
A methodology for the practice of structural
Newtons axiomatic framework allowed him to pursue geology that integrates observations and modeling
a strategy in which he could construct a simplied,
is illustrated in Fig. 12.14 as a set of ten stepping
idealized mathematical model of the physical system
stones on a path with the suggestion that this be
he wanted to probe in this case, the solar system.
Using mathematics, Newton could work out the conse-
traversed in a counterclockwise sense starting at
quences of certain actions and compare them with the top. Of course scientic investigations are
measurements and empirical observations. That com- rarely this well organized or rationalized. Instead
parison, in turn, would suggest ways in which the of beginning with a eld observation one may be
model could be adjusted and rened to achieve even inspired to investigate a new problem while
greater realism. In essence, this strategy of maintain- reading a textbook on uid mechanics or contem-
ing a right interplay between mathematical analysis plating the solution to an elastic boundary value
and physical experience afforded a marvellously pro- problem displayed as velocity vectors on a com-
ductive way of using mathematics to explain the work- puter screen. Regardless of the source of inspira-
ings of nature. Revolutionary in Newtons time, this
tion one could argue that seeking the map or
kind of approach is taken for granted in modern
photograph or measurement in the eld to conrm
research (Peterson, 1993).
that the phenomenon in question occurs naturally
The material presented in this textbook reveals in Earths crust is a pre-requisite to launch an inves-
two distinct views of structural geology. On the tigation. Thus, we nd ourselves at the top of the
one hand we have described observations of struc- diagram and suggest that this step of seeking data
tures: for example views through a microscope of in the eld should benet from the techniques of
tiny spherical objects that were deformed into eld-based structural geology described in Chapter
ellipsoidal shapes as rocks were contorted into the 2. Also, for the quantitative characterization of
South Mountain fold (Fig. 5.4), or photographs of structures we advocate the use of differential
exposures that show successive stages in the devel- geometry, some of which is described in Chapter 3.
opment of strike slip faults in granitic rock of the Once launched along the path (Fig. 12.14) it is
Sierra Nevada (Fig. 9.37). On the other hand we common to step off and move directly to a step-
have described models of structures: for example ping stone out of sequence, or even to backtrack
a plot of principal stress trajectories associated along the path. None-the-less this simple diagram
with a pattern of dikes around the Spanish Peaks provides a way to organize the primary scientic
(Fig. 6.37), or a plot of displacement vectors asso- procedures employed by a structural geologist
ciated with the 1999 Hector Mine earthquake (Fig. and to understand how they might relate to one
12.3 A METHODOLOGY FOR THE PRACTICE OF STRUCTURAL GEOLOGY 475
Field observations
Maps
Photographs
Measurements
Descriptions
Visualization techniques
General boundary conditions
Graphs
Solid mechanics
Contour maps
Fluid mechanics
3D projections
Heat transport
Animations
Solutions
Fundamental relationships Stress
Conservation laws Displacement
Laws of motion Velocity
Constitutive laws Temperature
Fig 12.14 A methodology for the practice of structural arguably, the most challenging steps to describe
geology. and, therefore, to teach. It is generally not possi-
ble to include all of the features of a structure in
another. For example, in Chapter 12 we have a model and those that are included usually are
focused on two of the stepping stones near the idealized. Idealization is given as the second step
beginning of the path: idealization of observed along the path and is a transition from observa-
structures and selection of general boundary con- tions to a model. It is largely done in the eld
ditions for model development. These are, during the process of mapping and measurement.
476 MODEL DEVELOPMENT AND METHODOLOGY
Here one postulates rst-order cause-and-effect viscous ow problem. Then the specic boundary
relationships and uses these to exclude observa- and/or initial conditions are selected to provide
tions that are deemed to be irrelevant. the constants of integration necessary to solve the
The third step involves the selection of the governing equations. Sometimes, the specic
branch of continuum mechanics that provides boundary conditions may be constrained by eld
the general boundary conditions for modeling. measurements, thereby establishing a direct link
Here the scaling arguments introduced in with the stepping stone at the top of the diagram.
Chapter 4 should prove helpful, and the concepts In other cases the specic boundary conditions
of strain, rate of deformation, traction, and stress are chosen arbitrarily, just to see what the
described in Chapters 5 and 6 must be understood. outcome might be, and thereby learn about the
Although some sub-disciplines of continuum behavior of the model structure. We have illus-
mechanics are spelled out on the third stepping trated both of these approaches to modeling with
stone, this list is not meant to be exclusive or to examples throughout the text. In Chapters 8, 9,
imply that coupled problems, for example of uid 10, and 11 we have employed methods that
ow and solid deformation, are unimportant. achieve analytical solutions to particular prob-
Indeed, the level of understanding of coupled lems in elasticity theory or uid mechanics. In the
problems among structural geologists and the simpler cases these solutions are derived, but
modeling resources now available in terms of others are taken from the vast literature on these
both hardware and software suggest that this will subjects without derivation to illustrate a partic-
be a very fruitful area of investigation in the ular point or concept in the context of the
twenty-rst century. seventh stepping stone. In a few cases solutions
After selecting the theoretical context for an were obtained by numerical methods but these
investigation one should focus on the fundamen- methods are not even introduced here, much less
tal relationships, two of which are described in developed from rst principles. The nite-differ-
Chapter 7 as the conservation laws of mass and ence method (FDM), nite-element method (FEM),
momentum. This fourth step along the path com- and boundary element method (BEM) have been
monly is overlooked, but it provides a touchstone exploited by engineers and scientists to nd
that makes clear how the Laws of Motion have numerical solutions to boundary value problems
been formulated and how the constitutive laws (Timoshenko and Goodier, 1970; Crouch and
are employed to reduce these to a specic set of Stareld, 1983; Hughes, 1987).
governing equations identied with the fth step Whether found using analytical or numerical
along the path. At this point one has made an methods, the solutions themselves are the eighth
explicit commitment to a material behavior and step in this methodology and they provide the
perhaps simplied the equations of motion to eld quantities such as stress, displacement,
exclude temporal or spatial variations in material velocity, and temperature as functions of the
properties. In this textbook we give considerable spatial coordinates and time. Only in rare
attention to the linear elastic material as instances are the distributions in space and time
described in Chapter 8 and the linear viscous of these quantities so simple as to be understand-
material as described in Chapter 10. One of the able from an analytical equation. Therefore, one
most commonly employed governing equations in employs visualization techniques, the ninth step-
the elastic context is the biharmonic equation for ping stone, that range from two-dimensional
the stress function in two dimensions. In the graphs and contour plots to three-dimensional
viscous context the NavierStokes equations are projections and animations. Throughout this text
the benchmark. we have used Matlab to aid in visualization.
In the sixth step one chooses a particular Because the scripts that produce these illustra-
model geometry, which may reduce the govern- tions are available, visualization is a dynamic
ing equations to two or even one spatial dimen- process in which the reader can adjust the bound-
sion. Here one may eliminate time altogether, as ary conditions or values of the parameters and see
in a quasi-static elastic problem or steady-state the response.
12.4 CONCLUDING REMARKS 477
Achenbach, J. D., 1973, Wave Propagation in Elastic Solids. Microstructure of deformation bands in porous
New York: American Elsevier. sandstones at Arches National Park, Utah. Journal of
Ackermann, R. V., Schlische, R. W., and Withjack, M. O., Structural Geology 16: 94159.
2001, The geometric and statistical evolution of Archuleta, R., 1984, A faulting model for the 1979
normal fault systems: an experimental study of the Imperial Valley earthquake. Journal of Geophysical
effects of mechanical layer thickness on scaling Research 89: 455985.
laws. Journal of Structural Geology 23: 180319. Argand, E., 1911, Les nappes de recouvrement des Alpes
Aczel, A. D., 2000, Mystery of the Aleph: Mathematics, the Pennines et leurs prolongements structuraux.
Kabbalah, and the Search for Infinity. New York: Four Beitrge zur geologischen Karte der Schweiz 31: 125.
Walls, Eight Windows. Arrowsmith, J. R., Pollard, D. D., and Rhodes, D. D.,
Adams, F. and Nicholson, J., 1901, Experimental investi- 1996, Hillslope development in areas of active tec-
gation into the flow of marble: Philosophical tonics. Journal of Geophysical Research, B, Solid Earth and
Transactions of the Royal Society of London 195: 363401. Planets 101(3): 625575.
Aki, K. and Richards, P. G., 1980, Quantitative Seismology: Arrowsmith, J. R., Rhodes, D. D., and Pollard, D. D.,
Theory and Methods. San Francisco: W. H. Freeman. 1998, Morphologic dating of scarps formed by
Albritton, C. C., Jr., 1980, The Abyss of Time: Changing repeated slip events along the San Andreas Fault,
Conceptions of the Earths Antiquity after the Sixteenth Carrizo Plain, California. Journal of Geophysical
Century. San Francisco: Freeman, Cooper and Co. Research, B, Solid Earth and Planets 103(5): 10 14160.
Alpha, T. R., Vigil, J. R., and Buchholz, L., 1988, Map Atkinson, B. K., 1987, Fracture Mechanics of Rock. London:
Projections Via Hyper Card, Open File Report 88364. Academic Press.
Menlo, CA: US Geological Survey. Atkinson, B. K. and Meredith, P. G., 1987, Experimental
Amadei, B. and Stephansson, O., 1997, Rock Stress and its fracture mechanics data for rocks and minerals. In
Measurement. London: Chapman and Hall. B. K. Atkinson, ed., Fracture Mechanics of Rock. London:
Anderson, D., 1975, Chemical plumes in the mantle. Academic Press, pp. 477525.
Geological Society of America Bulletin 86: 15931600. Aydin, A., 1977, Faulting in sandstone. Unpublished
Anderson, E. M., 1936, The dynamics of the formation Ph.D. thesis, Stanford University.
of cone-sheets, ring-dykes, and caldron-subsidences. Aydin, A., 1978, Small faults formed as deformation
Proceedings of the Royal Society of Edinburgh 56: 12856. bands in sandstone. Pure and Applied Geophysics 116:
Anderson, E. M., 1951, The Dynamics of Faulting and Dyke 91330.
Formation with Application to Britain. Edinburgh: Aydin, A., 2000, Fractures, faults, and hydrocarbon
Oliver and Boyd. entrapment, migration and flow. Marine and
Anderson, E. M., 1972, The Dynamics of Faulting and Dyke Petroleum Geology 17(7): 797814.
Formation with Applications to Britain. New York: Aydin, A. and DeGraff, J. M., 1988, Evolution of polyg-
Hafner. onal fracture patterns in lava flows. Science 239:
Anderson, T. L., 1995, Fracture Mechanics. Boca Raton: 4716.
CRC Press. Aydin, A. and Eyal, Y., 2002, Anatomy of a normal fault
Andrews, D. J. and Hanks, T. C., 1985, Scarp degrada- with shale smear: implications for fault seal.
tion by linear diffusion: inverse solution for age. American Association of Petroleum Geologists Bulletin
Journal of Geophysical Research 90: 10 193208. 86(8): 136781.
Antonellini, M. and Aydin, A., 1994, Effect of faulting Aydin, A. and Johnson, A. M., 1978, Development of
on fluid flow in porous sandstones: petrophysical faults as zones of deformation bands and as slip sur-
properties. American Association of Petroleum Geologists faces in sandstone. Pure and Applied Geophysics 116:
Bulletin 78: 33577. 93142.
Antonellini, M. and Aydin, A., 1995, Effect of faulting Aydin, A. and Schultz, R. A., 1990, Effect of mechanical
on fluid flow in porous sandstones: geometry and interaction on the development of strike-slip faults
spatial distribution. American Association of Petroleum with echelon patterns. Journal of Structural Geology 12:
Geologists Bulletin 79: 64271. 1239.
Antonellini, M. A., Aydin, A., and Pollard, D. D., 1994, Badoux, H., 1963, Les blemnites tronones de
REFERENCES 479
Leytron (Valais). Bulletin de Gologie, Minralogie, Bieniawski, Z. T., 1984, Rock Mechanics Design in Mining
Gophysique et du Muse Gologique de lUniversit de and Tunneling. Rotterdam: A. A. Balkema.
Lausanne 138: 17. Bieniawski, Z. T. and Bernede, M. J., 1979, Suggested
Bahat, D., 1991, Tectonofractography. New York: Springer- methods for determining the uniaxial compresssive
Verlag. strength and deformability of rock materials.
Bai, T. and Pollard, D. D., 2000a, Closely spaced frac- International Journal of Rock Mechanics and Mining
tures in layered rocks: initiation mechanism and Sciences and Geomechanics Abstracts 16(2): 13540.
propagation kinematics. Journal of Structural Geology Billings, M. P., 1972, Structural Geology. Englewood
22: 140925. Cliffs, NJ: Prentice-Hall.
Bai, T. and Pollard, D. D., 2000b, Fracture spacing in Biot, M. A., 1961, Theory of folding of stratified visco-
layered rocks: a new explanation based on the stress elastic media and its implications in tectonics and
transition. Journal of Structural Geology 22: 4357. orogenesis. Geological Society of America Bulletin 72:
Bai, T., Pollard, D. D., and Gao, H., 2000, Explanation 1595620.
for fracture spacing in layered materials. Nature 403: Biot, M. A., 1965, Mechanics of Incremental Deformation.
7536. New York: John Wiley and Sons.
Barber, J. R., 1992, Elasticity. Dordrecht: Kluwer Birch, F., 1966, Compressibility: elastic constants. In
Academic. S. P. Clark, ed., Handbook of Physical Constants. New
Barton, N., 1973, Review of a new shear-strength cri- York: The Geological Society of America,
terion for rock joints. Engineering Geology 7: 287332. pp. 97173.
Barus, C., 1891, The viscosity of solids. US Geological Bird, R. B., Stewart, W. E., and Lightfoot, E. N., 1960,
Survey Bulletin 73. Transport Phenomena. New York: John Wiley and Sons.
Batchelor, G. K., 1967, An Introduction to Fluid Mechanics. Bott, M. H. P., 1959, The mechanics of oblique slip fault-
Cambridge: Cambridge University Press. ing. Geological Magazine 96: 10917.
Batt, G. E., Brandon, M. T., Farley, K. A., and Roden-Tice, Bourne, S. J., 2003, Contrast of elastic properties
M., 2001, Tectonic synthesis of the Olympic between rock layers as a mechanism for the initia-
Mountains segment of the Cascadia wedge, using tion and orientation of tensile failure under remote
two-dimensional thermal and kinematic modeling compression. Journal of Geophysical Research 108:
of thermochronological ages. Journal of Geophysical 10.1029/2001JB001725.
Research 106: 26 73146. Bourne, S. J. and Willemse, E. J. M., 2001, Elastic stress
Beckmann, P., 1971, A History of (pi). New York: Dorset control on the pattern of tensile fracturing around a
Press. small fault network at Nash Point, UK. Journal of
Bell, J. S. and Babcock, E. A., 1986, The stress regime of Structural Geology 23: 175370.
the western Canadian Basin and implications for Bowden, F. P. and Tabor, D., 1950, The Friction and
hydrocarbon production. Bulletin of the Canadian Lubrication of Solids. Oxford: Clarendon Press.
Petroleum Geologists 34: 36478. Brace, W. F., 1964, Brittle fracture of rocks. In W. R.
Bell, R. J. T., 1920, An Elementary Treatise on Coordinate Judd, ed., State of Stress in the Earths Crust. New York:
Geometry of Three Dimensions. London: Macmillan. Elsevier, pp. 11174.
Bergbauer, S., 2002, The use of curvature for the analy- Brace, W. F. and Bombolakis, E. G., 1963, A note on
ses of folding and fracturing with application to the brittle crack growth in compression. Journal of
Emigrant Gap Anticline, Wyoming. Unpublished Geophysical Research 68: 370913.
Ph.D. thesis, Stanford University. Brace, W. F. and Byerlee, J. D., 1966, Stick slip as a mech-
Bergbauer, S. and Martel, S. J., 1999, Formation of anism for earthquakes. Science 153: 9902.
joints in cooling plutons. Journal of Structural Geology Brandon, M. T., 2004, The Cascadia subduction wedge:
21(7): 82135. the role of accretion, uplift, and erosion. In B. A. van
Bergbauer, S. and Pollard, D. D., 2003, How to calculate der Pluijm and S. Marshak, eds., Earth Structure: An
normal curvatures of sampled geological surfaces. Introduction to Structural Geology and Tectonics. New
Journal of Structural Geology 25: 27789. York: W. W. Norton, pp. 56674.
Bergbauer, S. and Pollard, D. D., 2004, A new concep- Braunstein, J. and OBrien, G. D., 1968, Diapirism and
tual foldfracture model including prefolding Diapirs: Memoir. Tulsa, OK: American Association of
joints, based on the Emigrant Gap anticline, Petroleum Geologists.
Wyoming. Geological Society of America Bulletin 116: Bredehoeft, J. D., Wolff, R. G., Keys, W. S., and Shuter, E.,
294307. 1976, Hydraulic fracturing to determine the regional
480 REFERENCES
in situ stress field, Piceance Basin, Colorado. Chandrasekhar, S., 1979, Beauty and the quest for
Geological Society of America Bulletin 87(2): 2508. beauty in science. Physics Today July: 2530.
Bridgman, P. W., 1931, Dimensional Analysis. New Haven, Chapple, W. M., 1968, A mathematical theory of finite-
CT: Yale University Press. amplitude rock-folding. Geological Society of America
Brodkey, R. S. and Hershey, H. C., 1988, Transport Bulletin 79: 4768.
Phenomena: A Unified Approach. New York: McGraw- Chapple, W. M., 1978, Mechanics of thin-skinned fold-
Hill. and-thrust belts. Geological Society of America Bulletin
Brown, E. T. and Hoek, E., 1978, Trends in relationships 89: 118998.
between measured in situ stresses and depth. Chinnery, M. A., 1961, The deformation of the ground
International Journal of Rock Mechanics and Mining around surface faults. Bulletin of the Seismological
Science 15: 21115. Society of America 51: 35572.
Buckingham, E., 1915, Model experiments and the Chinnery, M. A., 1963, The stress changes that accom-
forms of empirical equations. Transactions of the pany strike-slip faulting. Bulletin of the Seismological
American Society of Mechanical Engineers 37: 26392. Society of America 53: 92132.
Brgmann, R. and Pollard, D. D., 1992, Influence of the Chinnery, M. A., 1966, Secondary faulting: I and II.
state of stress on the brittleductile transition in Canadian Journal of Earth Sciences 3: 16390.
granitic rock: evidence from fault steps in the Sierra Choukroune, P. and Seguret, M., 1973, Tectonics of the
Nevada, California. Geology 20: 64548. Pyrenees: role of compression and gravity. In K. A. De
Brgmann, R. and Pollard, D. D., 1994, Strain accom- Jong and R. Scholten, eds., Gravity and Tectonics. New
modation about strikeslip fault discontinuities in York: John Wiley and Sons, pp. 14156.
granitic rock under brittle-to-ductile conditions. Clark, S. P. J., 1966, Handbook of Physical Constants. New
Journal of Structural Geology 16: 165574. York: The Geological Society of America.
Brgmann, R., Pollard, D. D., and Martel, S. J., 1994, Slip Cloos, E., 1946, Lineation: A Critical Review and Annotated
distributions on faults: effects of stress gradients, Bibliography, Geological Society of America, Memoir 18.
inelastic deformation, heterogeneous host-rock stiff- Ann Arbor, MI: Edwards Brothers.
ness, and fault interaction. Journal of Structural Cloos, E., 1947, Oolite deformation in the South
Geology 16: 167590. Mountain Fold, Maryland. Geological Society of America
Byerlee, J., 1978, Friction of rocks. Pure and Applied Bulletin 58: 843918.
Geophysics 116: 61526. Cloos, E., 1971, Microtectonics along the western edge
Calladine, C. R. and Drucker, D. C., 1962, Nested sur- of the Blue Ridge, Maryland and Virginia. Baltimore,
faces of constant rate of dissipation in creep. Journal MD: The Johns Hopkins University Press.
of Applied Mathematics 29: 7984. Cobbold, P. R., Cosgrove, J. W., and Summers, J. M., 1971,
Carena, S. and Suppe, J., 2002, Three-dimensional Development of internal structures in deformed
imaging of active structures using earthquake after- anisotropic rocks. Tectonophysics 12: 2353.
shocks: the Northridge Thrust, California. Journal of Cobbold, P. R., Durand, S., and Mourgues, R., 2001,
Structural Geology 24(4): 887904. Sandbox modelling of thrust wedges with fluid-
Carslaw, H. S. and Jaeger, J. C., 1959, Conduction of Heat assisted detachments. Tectonophysics 334: 24558.
in Solids. Oxford: Clarendon Press. Cocco, M. and Rice, J. R., 2002, Pore pressure and poro-
Carter, N. L., Friedman, M., Logan, J. M., and Stearns, elasticity effects in Coulomb stress analysis of earth-
D. W., 1981, Mechanical Behavior of Crustal Rocks (The quake interactions. Journal of Geophysical Research 107:
Handin Volume), Geophysical Monograph. Washington, 117.
DC: American Geophysical Union, p. 326. Collins, G. C., Head, J. W., III, and Pappalardo, R. T.,
Cartwright, J. A., Trudgill, B. D., and Mansfield, C. S., 1998, The role of extensional instability in creating
1995, Fault growth by segment linkage: an explana- Ganymede grooved terrain: insights from Galileo
tion for scatter in maximum displacement and trace high-resolution stereo imaging. Geophysical Research
length data from the Canyonlands grabens of SE Letters 25: 2336.
Utah. Journal of Structural Geology 17: 131926. Committee on the Future of the Global Positioning
Cathles, L. M., 1975, The Viscosity of the Earths Mantle. System, 1995, The Global Positioning System: A Shared
Princeton, New Jersey: Princeton University Press. National Asset. Recommendations for Technical
Cathles, L. M., 1977, An analysis of the cooling of intru- Improvements and Enhancements. Washington, DC:
sions by ground-water convection which includes Aeronautics and Space Engineering Board, National
boiling. Economic Geology 72: 80426. Research Council, National Academy Press.
REFERENCES 481
Compton, R. R., 1962, Manual of Field Geology. New York: echelon normal faults. Journal of Geophysical Research,
John Wiley and Sons. B, Solid Earth and Planets 103(10): 24 37391.
Cooke, M., Mollema, P., Pollard, D. D., and Aydin, A., Crouch, S. L. and Starfield, A. M., 1983, Boundary Element
2000, Interlayer slip and joint localization in East Methods in Solid Mechanics: with Applications in Rock
Kaibab Monocline, Utah: field evidence and results Mechanics and Geological Engineering. London: George
from numerical modeling. In J. W. Cosgrove and Allen and Unwin.
M. S. Ameen, eds., Forced Folds and Fractures, Cruikshank, K. M. and Aydin, A., 1995, Unweaving the
Special Publication. London: Geological Society, joints in Entrada Sandstone, Arches National Park,
pp. 2349. Utah, USA. Journal of Structural Geology 17: 40921.
Cooke, M. and Pollard, D. D., 1996, Fracture propaga- Culling, W. E. H., 1960, Analytical theory of erosion.
tion paths under mixed mode loading within rectan- Journal of Geology 68: 33644.
gular blocks of polymethyl methacrylate. Journal of Dana, J. D., 1847a, Geological results of the Earths con-
Geophysical Research 101: 3387400. traction. American Journal of Science 3: 17688.
Cooke, M. L., 1997, Fracture localization along faults Dana, J. D., 1847b, Geological results of the Earths con-
with spatially varying friction. Journal of Geophysical traction. American Journal of Science 4: 8892.
Research, B, Solid Earth and Planets 102(10): 22 42534. Dahlen, F. A., Suppe, J., and Davis, D., 1984, Mechanics
Cooke, M. L. and Underwood, C. A., 2001, Fracture ter- of fold-and-thrust belts and accretionary wedges:
mination and step-over at bedding interfaces due to cohesive Coulomb theory. Journal of Geophysical
frictional slip and interface opening. Journal of Research 89: 10 087101.
Structural Geology 23(23): 22338. Daly, R. A., Manger, G. E., and Clark, S. P. J., 1966,
Cotterell, B. and Rice, J. R., 1980, Slightly curved or Density of rocks. In S. P. J. Clark, ed., Handbook of
kinked cracks. International Journal of Fracture 16: Physical Constants. New York: Geological Society of
15569. America, pp. 1926.
Coulomb, C. A., 1773, Sur une application des rgles de Davatzes, N. C. and Aydin, A., 2003, Overprinting fault-
maximis et minimis quelques problmes de sta- ing mechanisms in high porosity sandstones of SE
tique relatifs larchitecture. Acad. Roy. des Sciences, Utah. Journal of Structural Geology 25: 1795813.
Mmoires de math. et de physique par divers savants 7: Davies, R. K. and Pollard, D. D., 1986, Relations between
34382. left-lateral strike-slip faults and right-lateral mono-
Couples, G., 1977, Stress and shear fracture (fault) pat- clinal kink bands in granodiorite, Mt. Abbot
terns resulting from a suite of complicated bound- Quadrangle, Sierra Nevada, California. Pure and
ary conditions with applications to the Wind River Applied Geophysics 124: 177201.
Mountains. Pure and Applied Geophysics 115: 11333. Davis, D., Suppe, J., and Dahlen, F. A., 1983, Mechanics
Couples, G. D. and Lewis, H., 1998, Lateral variations of of fold-and-thrust belts and accretionary wedges.
strain in experimental forced folds. Tectonophysics Journal of Geophysical Research 88: 115372.
295: 7991. Davis, E. E. and Lister, C. R. B., 1974, Fundamentals of
Cowan, D., 1992, Chairpersons message. Structural ridge crest topography. Earth and Planetary Science
Geology and Tectonics Division Newsletter. Boulder CO: Letters 21(4): 40513.
Geological Society of America, p. 1. Davis, G. H. and Reynolds, S. J., 1996, Structural Geology
Cowie, P. A. and Scholz, C. H., 1992, Growth of faults by of Rocks and Regions. New York: John Wiley and Sons.
accumulation of seismic slip. Journal of Geophysical Davis, J. C., 1986, Statistics and Data Analysis in Geology.
Research 97: 11 08595. New York: John Wiley and Sons.
Cowie, P. A. and Shipton, Z. K., 1998, Fault tip displace- Davis, P. J. and Hersh, R., 1986, Descartes Dream. Boston,
ment gradients and process zone dimensions. MA: Houghton Mifflin.
Journal of Structural Geology 20: 98397. Dawers, N. H., Anders, M. H., and Scholz, C. H., 1993,
Crick, F., 1988, What Mad Pursuit: A Personal View of Growth of normal faults: displacement-length
Scientific Discovery. New York: Basic Books. scaling. Geology 21: 110710.
Crider, J. G., 2001, Oblique slip and the geometry of De Sitter, L. U., 1964, Structural Geology. New York:
normal-fault linkage: mechanics and a case study McGraw-Hill Book Company.
from the Basin and Range in Oregon. Journal of DeGraff, J. M. and Aydin, A., 1987, Surface morphology
Structural Geology 23(12): 19972009. of columnar joints and its significance to mechanics
Crider, J. G. and Pollard, D. D., 1998, Fault linkage: and directions of joint growth. Geological Society of
three-dimensional mechanical interaction between America Bulletin 99: 60517.
482 REFERENCES
Delaney, P. T., 1990, Deep magma body beneath the equipment, model construction techniques and rhe-
summit and rift zones of Kilauea Volcano, Hawaii. ology of model properties. Journal of Structural Geology
Science 247: 1265372. 7: 83102.
Delaney, P. T., Denlinger, R. P., Lisowski, M., et al., 1998, Dombard, A. J. and McKinnon, W. B., 2001, Formation
Volcanic spreading at Kilauea, 19761996. Journal of of grooved terrain on Ganymede: extensional insta-
Geophysical Research 103: 18 00323. bility mediated by cold superplastic creep. Icarus 154:
Delaney, P. T., Miklius, A., Arnadottir, T., Okamura, 32136.
A. T., and Sako, M. K., 1993, Motion of Kilauea Donath, F. A., 1961, Experimental study of shear failure
Volcano during sustained eruption from the Puu Oo in anisotropic rocks. Geological Society of America
and Kupaianaha vents, 19831991. Journal of Bulletin 72: 98590.
Geophysical Research 98: 17 80120. Donath, F. A. and Parker, R. B., 1974, Folds and folding.
Delaney, P. T. and Pollard, D. D., 1981, Deformation of Geological Society of America Bulletin 75: 4562.
host rocks and flow of magma during growth of Duba, A. G., Durham, W. B., Handin, J. W., and Wang,
minette dikes and breccia-bearing intrusions near H. F., 1990, The BrittleDuctile Transition in Rocks (The
Ship Rock, New Mexico. US Geological Survey Heard Volume), Geophysical Monograph. Washington,
Professional Paper 1202, pp. 161. DC: American Geophysical Union, p. 243.
Delaney, P. T. and Pollard, D. D., 1982, Solidification of Emerman, S. and Turcotte, D. L., 1983, A fluid model
basaltic magma during flow in a dike. American for the shape of accretionary wedges. Earth and
Journal of Science 282: 85685. Planetary Science Letters 63: 37984.
Delaney, P. T., Pollard, D. D., Ziony, J. I., and McKee, Enever, J. R., Cornet, F., and Roegiers, J. C., 1992, ISRM
E. H., 1986, Field relations between dikes and joints: commission on interpretation of hydraulic fracture
emplacement processes and paleostress analysis. records. International Journal of Rock Mechanics and
Journal of Geophysical Research 91(B5): 492038. Mining Science 29: 6972.
Dennis, J. G., 1987, Structural Geology: An Introduction. Engelder, T., 1993, Stress Regimes in the Lithosphere.
Dubuque, IA: William C. Brown. Princeton, NJ: Princeton University Press.
Dieterich, J. H., 1969, Origin of cleavage in folded Engelder, T., and Peacock, D. C. P., 2001, Joint develop-
rocks. American Journal of Science 267: 15565. ment normal to regional compression during flex-
Dieterich, J. H., 1979a, Modeling of rock friction: 1. ural-flow folding: the Lilstock buttress anticline,
Experimental results and constitutive equations. Somerset, England. Journal of Structural Geology 23:
Journal of Geophysical Research 84: 21618. 25977.
Dieterich, J. H., 1979b, Modeling of rock friction: 2. Erdogan, F. and Sih, G. C., 1963, On the crack extension
Simulation of preseismic slip. Journal of Geophysical in plates under plane loading and transverse shear.
Research 84(B5): 216975. Journal of Basic Engineering, Transactions ASME 85:
Dieterich, J. H., 1981, Constitutive properties of faults 51927.
with simulated gouge. In N. L. Carter, M. Friedman, Eshbach, O. W., 1961, Handbook of Engineering
J. M. Logan and D. W. Stearns, eds., Mechanical Fundamentals, Handbook Series. New York: John Wiley
Behavior of Crustal Rocks: the Handin Volume. and Sons.
Washington, DC: American Geophysical Union, Eshelby, J. D., 1957, The determination of the elastic
pp. 10320. field of an ellipsoidal inclusion, and related prob-
Dieterich, J. H., 1986, A model for the nucleation of lems. Proceedings of the Royal Society of London Series A
earthquake slip. Geophysical Monograph 37: 3747. 241: 37696.
Dieterich, J. H., 1994, A constitutive law for rate of Evans, B. and Kohlstedt, D. L., 1995, Rheology of rocks.
earthquake production and its application to earth- In T. J. Ahrens, ed., Handbook of Physical Properties of
quake clustering. Journal of Geophysical Research, B, Rocks. Washington, DC: American Geophysical
Solid Earth and Planets 99(2): 260118. Union, pp. 14865.
Dieterich, J. H. and Carter, N. L., 1969, Stress-history of Faerseth, R. B., Knudsen, B.-E., Liljedahl, T., Midbe,
folding. American Journal of Science, 267: 12954. P. S., and Sderstrm, B., 1997, Oblique rifting and
Dieterich, J. H. and Onat, E. T., 1969, Slow finite defor- sequential faulting in the Jurassic development of
mations of viscous solids. Journal of Geophysical the northern North Sea. Journal of Structural Geology
Research 74: 20818. 19(10): 1285302.
Dixon, J. M. and Summers, J. M., 1985, Recent develop- Fairhurst, C., 1964, Measurement of in situ rock
ments in centrifuge modelling of tectonic processes: stresses with particular references to hydraulic frac-
REFERENCES 483
turing. Rock Mechanics and Engineering Geology 2: Flodin, E. and Aydin, A., 2004, Faults with asymmetric
12947. damage zones in sandstone, Valley of Fire State Park,
Farvolden, R. N. and Cherry, J. A., 1991, Opinion: are southern Nevada. Journal of Structural Geology 26:
geology departments prepared for the 21st century? 9838.
Geology May: 419. Flugge, W., 1967, Viscoelasticity. London: Blaisdell.
Feehan, J. G. and Brandon, M. T., 1999, Contribution of Fossen, H. and Gabrielsen, R. H., 1996, Experimental
ductile flow to exhumation of low Thigh P meta- modeling of extensional fault systems by use of
morphic rocks: San JuanCascade Nappes, NW plaster. Journal of Structural Geology 18: 67387.
Washington State. Journal of Geophysical Research 104: Freund, L. B., 1979, The mechanics of dynamic shear
10 883901. crack propagation. Journal of Geophysical Research 84:
Feshbach, H. and Weisskopf, V. F., 1988, Ask a foolish 2199209.
question. Physics Today October: 9. Friedel, J., 1964, Dislocations. Oxford: Pergamon Press.
Fialko, Y. A., Khazan, Y., and Simons, M., 2001, Friedman, M., Handin, J., and Alani, G., 1972, Fracture-
Deformation due to a pressurized horizontal circu- surface energy of rocks. International Journal of Rock
lar crack in an elastic half-space, with applications Mechanics and Mining Science 9: 75766.
to volcano geodesy. Geophysical Journal International Frocht, M. M., 1948, Photoelasticity. New York: John
146: 18190. Wiley and Sons.
Fialko, Y. A. and Rubin, A. M., 1999, What controls the Fung, Y. C., 1965, Foundations of Solid Mechanics, Prentice-
along-strike slopes of volcanic soft zones? Journal of Hall International Series in Dynamics. Englewood Cliffs,
Geophysical Research 104: 20 00720. NJ: Prentice-Hall.
Fialko, Y. A., Sandwell, D. T., Agrew, D., Simons, M., Fung, Y. C., 1969, A First Course in Continuum Mechanics.
Shearer, P., and Minster, B., 2002, Deformation on Englewood Cliffs, NJ: Prentice-Hall.
nearby faults induced by the 1999 Hector Mine Gallagher, J. J., Friedman, M., Handin, J., and Sowers,
earthquake. Science 297: 185862. G. M., 1974, Experimental studies relating to
Fink, J. H. and Fletcher, R. C., 1981, A Mechanical Analysis microfracture in sandstone. Tectonophysics 21:
of Extensional Instability on Ganymede, Technical 20347.
Memorandum 84211. Washington, DC: National Gauss, K. F., 1827, Karl Friedrich Gauss, general investi-
Aeronautics and Space Administration, pp. 513. gations of curved surfaces. In J. C. Morehead, ed.,
Fisher, M. P. and Wiklerson, M. S., 2000, Predicting the Investigations of Curved Surfaces of 1827 and 1825 (pub-
orientation of joints from fold shape: results of lished in 1902). Princeton, NJ: Princeton University
pseudo-three-dimensional modeling and curvature Library.
analysis. Geology 28: 1518. Ge, H. and Jackson, M. P. A., 1998, Physical modeling
Fisher, R. A., 1953, Dispersion of a sphere. Proceedings of of structures formed by salt withdrawal: implica-
the Royal Society of London Series A 217: 295. tions for deformation caused by slat dissolution.
Fletcher, R. C., 1974, Wavelength selection in the American Association of Petroleum Geologists Bulletin
folding of a single layer with power-law rheology. 82: 22850.
American Journal of Science 274: 102943. Gere, J. M. and Weaver, W. J., 1965, Matrix Algebra for
Fletcher, R. C., 1982, Coupling of diffusional mass Engineers. New York: Van Nostrand Reinhold.
transport and deformation in a tight rock. Gilbert, G. K., 1877, Report on the Geology of the Henry
Tectonophysics 83: 27591. Mountains, US Geographical and Geological Survey of the
Fletcher, R. C. and Hallet, B., 1983, Unstable extension Rocky Mountain Region. Washington, DC: US
of lithosphere: a mechanical model for basin-and- Government Printing Office.
range structure. Journal of Geophysical Research 83: Gleick, J., 1987, Chaos: Making a New Science. New York:
27591. Viking Penguin.
Fletcher, R. C. and Pollard, D. D., 1981, Anticrack model Goodman, R. E., 1980, Introduction to Rock Mechanics.
for pressure solution surfaces. Geology 9: 41924. New York: John Wiley and Sons.
Fletcher, R. C. and Pollard, D. D., 1999, Can we under- Goodman, R. E. and Shi, G.-H., 1985, Block Theory and Its
stand structural and tectonic processes and their Application to Rock Engineering. Englewood Cliffs, NJ:
products without a complete mechanics? Journal of Prentice-Hall.
Structural Geology 21: 107188. Gordon, J. E., 1976, The New Science of Strong Materials.
Fleuty, M. J., 1964, The description of folds. Proceedings Princeton, NJ: Princeton University Press.
of the Geologists Association 75(4): 46192. Grachev, A. F., Magnitsky, V. A., Mukhamediev, S. A., and
484 REFERENCES
Nikolaev, V. A., 2001, The effect of neotectonic move- extensional diapirs overprinted by compression.
ments on the gradients and curvatures of the north- American Association of Petroleum Geologists Bulletin 84:
ern Eurasian lithosphere surface. IzvestiyaRussian 1095108.
Academy of Sciences, Physics of the Solid Earth, 37: Guiton, M. L. E., Sassi, W., Leroy, Y. M., and Gauthier,
89106. B. D. M., 2003, Mechanical constraints on the
Green, H. W. and Burnley, P. C., 1989, The mechanism chronology of fracture activation in folded Devonian
of failure responsible for deep-focus earthquakes. sandstone of the western Moroccan Anti-Atlas.
Nature 341: 7337. Journal of Structural Geology 25: 131730.
Gregory, B., 1990, Inventing Reality: Physics as Language. Gutscher, M., Kukowski, N., Malavielle, J., and
New York: John Wiley and Sons. Lallemand, S., 1998, Material transfer in accretion-
Griffith, A. A., 1921, The phenomena of rupture and ary wedges from analysis of a systematic series of
flow in solids. Philosophical Transactions of the Royal analog experiments. Journal of Structural Geology 20:
Society of London Series A 221: 16398. 40716.
Griffith, A. A., 1924, The theory of rupture. In C. B. Habert, G. and De Saint-Blanquat, M., 2004, Rate of
Biezeno and J. M. Burgers, eds., First International construction of the Black Mesa bysmalith, Henry
Congress on Applied Mechanics. Delft: J. Waltman, Mountains, Utah. In C. Breitkreuz and N. Petford,
pp. 5563. eds., Physical Geology of High-Level Magmatic Systems.
Griggs, D. T., 1939, Creep of rocks. Journal of Geology 47: London: Geological Society Special Publications 234,
22551. pp. 163174.
Griggs, D. T. and Handin, J., 1960a, Observations on Hafner, W., 1951, Stress distributions and faulting.
fracture and an hypothesis of earthquakes. In D.T. Geological Society of America Bulletin 62: 37398.
Griggs and J. Handin, eds., Rock Deformation, Memoir Hhner, G. and Spencer, N., 1998, Rubbing and scrub-
79. New York: Geological Society of America, bing. Physics Today 51(9): 227.
pp. 34764. Haimson, B. C., 1968, Hydraulic fracturing in porous
Griggs, D. T. and Handin, J., 1960b, Rock Deformation, and nonporous rock and its potential for determin-
Memoir 79. New York: Geological Society of America, ing in situ stresses at great depth. Unpublished Ph.D.
p. 382. thesis, University of Wisconsin.
Grosfils, E. B. and Head, J. W., 1994, The global distribu- Haimson, B. C. and Fairhurst, C., 1967, Initiation and
tion of giant radiating dike swarms on Venus: impli- extension of hydraulic fractures in rocks. Society of
cations for the global stress state. Geophysical Research Petroleum Engineers Journal September: 31018.
Letters 21: 7014. Handin, J. and Hager, R. V., 1957, Experimental deforma-
Groshong, R. H. Jr., 1975, Strain, fractures, and pressure tion of sedimentary rocks under confining pressure:
solution in natural single layer folds. Geological tests at room temperature on dry samples. American
Society of America Bulletin 86: 136376. Association of Petroleum Geologists Bulletin 41: 150.
Groshong, R. H. Jr., 1999, 3-D Structural Geology: A Handin, J. and Hager, R. V., 1958, Experimental defor-
Practical Guide to Surface and Subsurface Map mation of sedimentary rocks under confining pres-
Interpretation. Berlin: Springer-Verlag. sure: tests at high temperature. American Association
Gross, M. R., 1993, The origin and spacing of cross of Petroleum Geologists Bulletin 42: 2892934.
joints: examples from Monterey Formation Santa Handin, J., Hager, R. V., Friedman, M., and Feather, J. N.,
Barbara Coastline, California. Journal of Structural 1963, Experimental deformation of sedimentary
Geology 15: 77351. rocks under confining pressure: pore pressure tests.
Gross, M. R., Fischer, M. P., Engelder, T., and Greenfield, American Association of Petroleum Geologists Bulletin 47:
R. J., 1995, Factors controlling joint spacing in 71755.
interbedded sedimentary rocks: integrating numeri- Handy, M. R., 1994, Flow laws for rocks containing
cal models with field observations from the two non-linear viscous phases: a phenomenological
Monterey Formation, USA. In M. S. Ameen, ed., approach. Journal of Structural Geology 16: 287301.
Fractography: Fracture Topography as a Tool in Fracture Hanks, T. C., 1977, Earthquake stress drops, ambient
Mechanics and Stress Analysis. London: Geological tectonic stresses and stresses that drive plate
Society, pp. 21533. motions. Pure and Applied Geophysics 115: 44158.
Grout, F. F., 1945, Scale models of structures related to Happel, J. and Brenner, H., 1965, Low Reynolds Number
batholiths. American Journal of Science 111: 26084. Hydrodynamics: With Special Applications to Particulate
Guglielmo, G., Vendeville, B. C., and Jackson, M. P. A., Media. Englewood Cliffs, NJ: Prentice-Hall.
2000, 3-D visualization and isochore analysis of Harris, R. A. and Segall, P., 1987, Detection of a locked
REFERENCES 485
zone at depth on the Parkfield, California, segment Pretoria: Natural Mechanical Engineering Research
of the San Andreas fault. Journal of Geophysical Institute, CSIR.
Research 92(B8): 794562. Hoek, E. and Bieniawski, Z. T., 1965, Brittle fracture
Harris, R. A. and Simpson, R. W., 1998, Suppression of propagation in rock under compression. International
large earthquakes by stress shadows: a comparison Journal of Fracture Mechanics 1: 13755.
of Coulomb and rate-and-state failure. Journal of Hofmann-Wellenhof, B., Lichtenegger, H., and Collins,
Geophysical Research 103: 24 43951. J., 1997, Global Positioning System. New York: Springer-
Haskell, N. H., 1937, The viscosity of the aestheno- Verlag.
sphere. American Journal of Science 33: 228. Hoshino, K., Koide, H., Inami, K., Iwamura, S., and
Hausman, D. M., 1998, Causal Asymmetries. Cambridge: Mitsui, S., 1972, Mechanical Properties of Japanese Tertiary
Cambridge University Press. Sedimentary Rocks Under High Confining Pressures, Report
Head, J. W., Crumpler, L. S., Aubele, J. C., Guest, J. E., and 244. Kawasaki: Geological Survey of Japan.
Saunders, R. S., 1992, Venus volcanism: classification Hossain, K. M., 1979, Determination of strain from
of volcanic features and structures, associations, and stretched belemnites. Tectonophysics 60: 27988.
global distributions from Magellan data. Journal of Howarth, R. J., 1999, Measurement, portrayal and
Geophysical Research 97: 13 15397. analysis of orientation data and the origins of early
Heard, H. C., 1960, Transition from brittle fracture to modern structural geology (16701967). Proceedings of
ductile flow in Solenhofen limestone as a function the Geologists Association 110: 273309.
of temperature, confining pressure, and interstitial Hubbert, M. K., 1937, Theory of scale models as applied
fluid pressure. In D. T. Griggs and J. Handin, eds., to the study of geologic structures. Geological Society
Rock Deformation. Boulder, CO: Geological Society of of America Bulletin 48: 1459520.
America, pp. 193226. Hubbert, M. K., 1945, Strength of the Earth. American
Hearmon, R. F. S., 1961, An Introduction to Applied Association of Petroleum Geologists Bulletin 29: 163053.
Anisotropic Elasticity. Oxford: Oxford University Press. Hubbert, M. K., 1951, Mechanical basis for certain
Herget, G., 1993, Rock stresses and rock stress monitor- familiar geologic structures. Geological Society of
ing in Canada. In J. A. Hudson, ed., Rock Testing and America Bulletin 62: 35572.
Site Characterization: Comprehensive Rock Engineering. Hubbert, M. K., 1972, Structural Geology. New York:
Oxford: Pergamon Press, pp. 47396. Hafner.
Hill, R., 1965, A self-consistent mechanics of composite Hubbert, M. K., 1974, Is being quantitative sufficient?
materials. Journal of the Mechanics and Physics of Solids In D. F. Merriam, ed., The Impact of Quantification on
13: 21322. Geology. Syracuse, NY: Syracuse University, pp. 2749.
Hillerman, T., 1990, Coyote Waits. New York: Harper and Hubbert, M. K. and Willis, D. G., 1957, Mechanics of
Row. hydraulic fracturing. Petroleum, Transactions AIME 210:
Hilley, G. E. and Strecker, M. R., 2004, Steady-state 15366.
erosion of critical Coulomb wedges with applica- Hudleston, P. J., 1973, Fold morphology and some geo-
tions to Taiwan and the Himalaya. Journal of metrical implications of theories of fold develop-
Geophysical Research 109: doi:10.1029/2002JB002284. ment. Tectonophysics 16(12): 146.
Hilley, G. E., Strecker, M. R., and Ramos, V. A., 2004, Hudleston, P. J., 1992, A comparison between glacial
Growth and erosion of fold-and-thrust belts, with an movement and thrust sheet or nappe emplacement
application to the Aconcagua Fold-and-Thrust Belt, and associated structures. In S. Mitra and G. W.
Argentina. Journal of Geophysical Research 109: Fisher, eds., Structural Geology of Fold and Thrust Belts.
doi:10.1029/2002JB002282. Baltimore, MD: The Johns Hopkins University Press,
Hirth, J. P. and Lothe, J., 1982, Theory of Dislocations. New pp. 8191.
York: John Wiley and Sons. Hudson, J. A., Crouch, S. L., and Fairhurst, C., 1972,
Hoagland, R. G., Hahn, G. T., and Rosenfield, A. R., 1973, Soft, stiff and servo-controlled testing machines: a
Influence of microstructure on fracture propagation review with reference to rock failure. Engineering
in rock. Rock Mechanics 5: 77106. Geology 6: 1559.
Hobbs, D. W., 1962, The strength of coal under biaxial Hughes, T. J. R., 1987, The Finite Element Method.
compression. Colliery Engineering 39: 28590. Englewood Cliffs, NJ: Prentice-Hall.
Hodgson, R. A., 1961, Classification of structures on Hunt, C. B., 1953, Geology and Geography of the Henry
joint surfaces. American Journal of Science 259: Mountains Region, Utah, US Geological Survey Professional
493502. Paper. Washington, DC: US Government Printing
Hoek, E., 1965, Rock Fracture Under Static Stress Conditions. Office.
486 REFERENCES
Hunt, C. B., 1988a, Geology of the Henry Mountains, Utah, failure in rings of rock loaded in diametral tension
as Recorded in the Notebooks of G. K. Gilbert, 187576. or compression. British Journal of Applied Physics 17:
Boulder, CO: The Geological Society of America. 68592.
Hunt, C. B., 1988b, The laccolith-stock controversy: new Johnson, A. M., 1970, Physical Processes in Geology. San
results from the southern Henry Mountains, Utah. Francisco, CA: Freeman, Cooper and Co.
Discussion and reply. Geological Society of America Johnson, A. M. and Fletcher, R. C., 1994, Folding of
Bulletin 100: 16579. Viscous Layers. New York: Columbia University Press.
Huntoon, P. W., 1982, The Meander anticline, Johnson, A. M. and Pollard, D. D., 1973, Mechanics of
Canyonlands, Utah: an unloading structure result- growth of some laccolithic intrusions in the Henry
ing from horizontal gliding on salt. Geological Society Mtns, Utah: Part I, Field observations, Gilberts
of America Bulletin 93: 94150. model, physical properties and flow of magma.
Hurlbut, C. S. and Griggs, D. T., 1939, Igneous rocks of Tectonophysics 18: 261309.
the Highwood Mountains, Montana. Geological Society Johnson, C. E., Rohahn, C., and Sharp, R. V., 1982,
of America Bulletin 50: 1032112. Introduction, The Imperial Valley, California,
Hytch, M. J., Putaux, J.-L., and Penisson, J.-M., 2003, Earthquake of October 15, 1979, Professional Paper
Measurement of the displacement field of disloca- 1254. Washington, DC: United States Geological
tions to 0.03 by electron microscopy. Nature 423(15 Survey, pp. 13.
May): 2703. Johnson, K. M. and Johnson, A. M., 2000, Localization of
Inglis, C. E., 1913, Stresses in a plate due to the pres- layer-parallel faults in San Rafael swell, Utah, and
ence of cracks and sharp corners. Royal Institute of other monoclinal folds. Journal of Structural Geology
Naval Architects Transactions 55: 21930. 22: 145568.
Irvine, T. N., 1970, Heat transfer during solidification Johnson, R. B., 1961, Patterns and origin of radial dike
of layered intrusions: I. Sheets and sills. Canadian swarms associated with West Spanish Peak and Dike
Journal of Earth Science 7: 103161. Mountain, south-central Colorado. Geological Society
Irwin, G. R., 1957, Analysis of stresses and strains near of America Bulletin 72: 57990.
the end of a crack traversing a plate. Journal of Johnson, R. B., 1968, Geology of the Igneous Rocks of the
Applied Mechanics 24: 3614. Spanish Peaks Region, Colorado, Professional Paper
Irwin, G. R., 1958, Fracture. In S. Flugge, ed., 594G. Menlo, CA: US Geological Survey, pp. 147.
Encyclopedia of Physics. Berlin: Springer-Verlag, Jones, G., Fisher, Q. J., and Knipe, R. J., 1998, Faulting,
pp. 55190. Fault Sealing and Fluid Flow in Hydrocarbon Reservoirs,
Jackson, M. D. and Pollard, D. D., 1988, The laccolith- Special Publication 147. London: Geological Society,
stock controversy: new results from the southern p. 319.
Henry Mountains, Utah. Geological Society of America Jonsson, S., Zebker, H. A., Segall, P., and Amelung, F.,
Bulletin 100: 11739. 2002, Fault slip distribution of the 1999 M-w 7.1
Jackson, M. D. and Pollard, D. D., 1990, Flexure and Hector Mine, California, earthquake, estimated from
faulting of sedimentary host rocks during growth of satellite radar and GPS measurements. Bulletin of the
igneous domes, Henry Mountains, Utah. Journal of Seismological Society of America 92: 137789.
Structural Geology 12: 185206. Kamb, W. B., 1959a, Theory of preferred crystal orienta-
Jaeger, J. C., 1957, The temperature in the neighbor- tion developed by crystallization under stress.
hood of a cooling intrusive sheet. American Journal of Journal of Geology 67: 15370.
Science 255: 30618. Kamb, W. B., 1959b, Ice petrofabric observations from
Jaeger, J. C., 1964a, Elasticity, Fracture and Flow with Blue Glacier, Washington, in relation to theory and
Engineering and Geological Applications. London: experiment. Journal of Geophysical Research 64:
Methuen. 1891909.
Jaeger, J. C., 1964b, Temperatures outside a cooling Kanninen, M. F. and Popelar, C. H., 1985, Advanced
intrusive sheet. American Journal of Science 257: 4454. Fracture Mechanics. New York: Oxford University Press.
Jaeger, J. C. and Cook, N. G. W., 1979, Fundamentals of Kattenhorn, S. A., Aydin, A., and Pollard, D. D., 2000,
Rock Mechanics. London: Chapman and Hall. Joints at high angles to normal fault strike: an expla-
Jaeger, J. C. and Hoskins, E. R., 1966a, Rock failure nation using 3-D numerical models of fault-per-
under the confined Brazilian test. Journal of turbed stress fields. Journal of Structural Geology 22(1):
Geophysical Research 71: 26519. 123.
Jaeger, J. C. and Hoskins, E. R., 1966b, Stresses and Kattenhorn, S. A. and Pollard, D. D., 2001, Integrating
REFERENCES 487
3-D seismic data, field analogs, and mechanical for oceanic spreading centers. Journal of Geophysical
models in the analysis of segmented normal faults Research 78: 3395417.
in the Wytch Farm oil field, southern England, Lamb, H., 1945, Hydrodynamics. New York: Dover.
United Kingdom. American Association of Petroleum Landau, L. D. and Lifshitz, E. M., 1960, Fluid Mechanics.
Geologists Bulletin 85(7): 1183210. Reading: Addison-Wesley.
Kelly, V. C., 1955, Monoclines of the Colorado Plateau. Lawn, B. R. and Wilshaw, T. R., 1975, Fracture of Brittle
Geological Society of America Bulletin 66: 789804. Solids. Cambridge: Cambridge University Press.
Kerr, A. D. and Pollard, D. D., 1998, Toward more realis- Lawson, A. C., 1908, The California Earthquake of April, 18,
tic formulations for the analysis of laccoliths. Journal 1906, Report of the State Earthquake Investigation
of Structural Geology 20: 178393. Committee. Washington, DC: Carnegie Institute of
Kilgore, B. D., Blanpied, M. L., and Dieterich, J. H., 1993, Washington, p. 641.
Velocity dependent friction of granite over a wide Lee, T.-C. and Delaney, P. T., 1987, Frictional heating
range of conditions. Geophysical Research Letters 20(10): and pore pressure rise due to fault slip. Geophysical
9036. Journal of the Royal Astronomical Society 88: 56991.
Kind, D. and Quinn, T., 1999, Metrology: quo vadis? Lekhnitskii, S. G., 1963, Theory of Elasticity of an
Physics Today (Sixteenth Annual Buyers Guide) 52: Anisotropic Elastic Body. San Francisco: Holden-Day.
BG13BG15. Li, V. C., 1987, Mechanics of shear rupture applied to
King, G. C. P., Stein, R. S., and Lin, J., 1994, Static stress earthquake zones. In B. K. Atkinson, ed., Fracture
changes and the triggering of earthquakes. Bulletin of Mechanics of Rock. San Diego, CA: Academic Press.
the Seismological Society of America 84: 93553. Lide, D. R., 2004, CRC Handbook of Chemistry and Physics.
King, W., 1875, Report on the superinduced divisional Boca Raton, FL: Chemical Rubber Company Press, Inc.
structure of rocks, called jointing, and its relation to Linker, M. F. and Dieterich, J. H., 1992, Effects of vari-
slaty cleavage. Royal Irish Academy Transactions 25: able normal stress on rock friction: observations and
60562. constitutive equations. Journal of Geophysical Research,
Kirby, S. H. and Kronenberg, A. K., 1987, Rheology of the B, Solid Earth and Planets 97(4): 492340.
lithosphere: selected topics. Reviews of Geophysics 25: Lipschutz, M. M., 1969, Theory and Problems of Differential
121944. Geometry, Schaums Outline. New York: McGraw-Hill.
Kirsch, G., 1898, Die Theorie der Elastizitt und die Lisle, R. J., 1994, Detection of zones of abnormal strains
Bedrfnisse der Festigkeitslehre. Zeitschrift des Verlines in structures using Gaussian curvature analysis.
Deutscher Ingenieure 42: 797807. American Association of Petroleum Geologists Bulletin 78:
Koch, F. G., Johnson, A. M., and Pollard, D. D., 1981, 181119.
Monoclinal bending of strata over laccolithic intru- Lisle, R. J., 2000, Predicting patterns of strain from
sions. Tectonophysics 74: T21T31. three-dimensional fold geometries: neutral surface
Koenig, E. and Pollard, D. D., 1998, Mapping and mod- folds and forced folds. In J. W. Cosgrove and M. S.
eling of radial fracture patterns on Venus. Journal of Ameen, eds., Forced Folds and Fractures, Special
Geophysical Research 103: 15 183202. Publication 169. London: Geological Society
Kostrov, B. V. and Das, S., 1988, Principles of Earthquake pp. 21321.
Source Mechanics. Cambridge: Cambridge University Lister, J. R., 1990, Buoyancy-driven fluid fracture.
Press. Journal of Fluid Mechanics 210: 26380.
Krantz, W. R., 1988, Multiple fault sets and three- Lister, J. R. and Kerr, R. C., 1991, Fluid-mechanical
dimensional strain: theory and application. Journal models of crack propagation and their application
of Structural Geology 10: 22537. to magma transport in dykes. Journal of Geophysical
Kranz, R. L., 1983, Microcracks in rocks: a review. Research 96: 10 04977.
Tectonophysics 100: 44980. Liu, S. and Dixon, J. M., 1990, Centrifuge modelling of
Krumbein, W. C. and Graybill, F. A., 1965, An thrust faulting: strain partitioning and the sequence
Introduction to Statistical Models in Geology. New York: of thrusting in duplex structures. In R. J. Knipe, ed.,
McGraw-Hill. Deformation Mechanisms, Rheology and Tectonics, Special
Labuz, J. F., Shah, S. P., and Dowding, C. H., 1987, The Publication 54. London: Geological Society,
fracture process zone in granite: evidence and effect. pp. 43144.
International Journal of Rock Mechanics and Mining Liu, S. and Dixon, J. M., 1991, Centrifuge modelling of
Science 24: 23546. thrust faulting: structural variation along strike in
Lachenbruch, A. H., 1973, A simple mechanical model fold-thrust belts. Tectonophysics 188: 3962.
488 REFERENCES
Lockner, D. A. and Beeler, N. M., 2003, Rock failure and Marshak, S. and Mitra, G., 1988, Basic Methods of
earthquakes. In W. H. K. Lee, H. Kanamori, P. C. Structural Geology. Englewood Cliffs, NJ: Prentice-Hall.
Jennings, and C. Kisslinger, eds., International Martel, S. and Boger, W. A., 1998, Geometry and
Handbook of Earthquake and Engineering Seismology, mechanics of secondary fracturing around small
Part A. Amsterdam: Academic Press, pp. 50538. three-dimensional faults in granitic rock. Journal of
Love, A. E. H., 1944, A Treatise on the Mathematical Geophysical Research 103: 21 299314.
Theory of Elasticity. New York: Dover. Martel, S., Pollard, D. D., and Segall, P., 1988,
Lovering, T. S., 1935, Theory of heat conduction applied Development of simple strike-slip fault zones in
to geological problems. Geological Society of America granitic rock, Mount Abbot quadrangle, Sierra
Bulletin 46: 6994. Nevada, California. Geological Society of America
Lovering, T. S., 1936, Heat conduction in dissimilar Bulletin 99: 145165.
rocks and the use of thermal models. Geological Martel, S. J., 1990, Formation of compound strike-slip
Society of America Bulletin 47: 87100. fault zones, Mount Abbot Quadrangle, California.
Maerten, F., Resor, P., Maerten, L., and Pollard, D. D., Journal of Structural Geology 12(7): 86982.
2005, Inverting for slip on three-dimensional fault Martel, S. J. and Pollard, D. D., 1989, Mechanisms of slip
surfaces using angular dislocations. Bulletin of the and fracture along small faults and simple strike-slip
Seismological Society of America in press. fault zones in granitic rock. Journal of Geophysical
Maerten, L., 2000, Variation in slip on intersecting Research 94: 941728.
normal faults: implications for paleostress inversion. Massonnet, D., Rossi, M., Carmona, C., et al., 1993, The
Journal of Geophysical Research 105: 25 55365. displacement field of the Landers earthquake
Maerten, L., Gillespie, P., and Pollard, D. D., 2002, mapped by radar interferometry. Nature 364: 13842.
Effects of local stress perturbation on secondary McBirney, A. R. and Murase, T., 1984, Rheological prop-
fault development. Journal of Structural Geology 24: erties of magmas. Annual Reviews of Earth and
14553. Planetary Sciences 12: 33757.
Maerten, L., Pollard, D. D., and Karpuz, R., 2000, How to McClay, K. and Bonora, M., 2001, Analog models of
constrain 3D fault continuity and linkage using restraining stepovers in strike-slip fault systems.
reflection seismic data: a geomechanical approach. American Association of Petroleum Geologists Bulletin 85:
American Association of Petroleum Geologists Bulletin 84: 23360.
131124. McClay, K. R., Waltham, D. A., Scott, A. D., and
Maerten, L., Pollard, D. D., and Maerten, F., 2001, Abousetta, A. A., 1991, Physical and seismic model-
Digital mapping of three-dimensional structures of ing of listric fault geometries. In A. M. Roberts, G.
the Chimney Rock fault system, central Utah. Journal Yielding, and B. Freeman, eds., The Geometry of Normal
of Structural Geology 23: 58592. Faults, Special Publication 56. London: Geological
Maerten, L., Willemse, E. J. M., Pollard, D. D., and Society, pp. 2319.
Rawnsley, K., 1999, Slip distributions on intersecting McClay, K. R. and White, M., 1995, Analogue models of
normal faults. Journal of Strurctural Geology 21: 25971. orthogonal and oblique rifting. Marine and Petroleum
Mallet, J. L., 2002, Geomodeling. Oxford: Oxford Geology 12: 13751.
University Press. McClintock, F. A. and Walsh, J. B., 1962, Friction on
Malvern, L. E., 1969, Introduction to the Mechanics of a Griffith cracks in rocks under pressure. In Fourth US
Continuous Medium. Upper Saddle River, NJ: Prentice- Congressional Applied Mechanics, Proceedings. Berkeley,
Hall. CA: American Society of Mechanical Engineers,
Manduit, T. and Brun, J. P., 1998, Growth fault/rollover pp. 101521.
systems: birth, growth, and decay. Journal of McConaughy, D. T. and Engelder, T., 2001, Joint initia-
Geophysical Research 103: 18 11936. tion in bedded clastic rocks. Journal of Structural
Mardon, D., 1988, Localization of pressure solution Geology 23(23): 20321.
and the formation of discrete solution seams. McGarr, A. and Gay, N. C., 1978, State of stress in the
Unpublished Ph.D. thesis, Texas A&M University. Earths crust. Annual Reviews of Earth and Planetary
Margenau, H., 1977, The Nature of Physical Reality: A Sciences 6: 40536.
Philosophy of Modern Physics. Woodbridge, CT: Ox Bow McGill, G. E. and Stromquist, A. W., 1979, The grabens
Press. of Canyonlands National Park, Utah: Geometry,
Marrett, R. and Peacock, D. C. P., 1999, Strain and stress. mechanics, and kinematics. Journal of Geophysical
Journal of Structural Geology 21: 105763. Research 84: 454763.
REFERENCES 489
McIntyre, D. B. and Weiss, L. E., 1956, Construction of the North Anatolian Fault with static stress changes.
block diagrams to scale in orthographic projection. Geophysical Journal International 154(2): 47182.
Proceedings of the Geological Association 67: 14255. Muller, O. H. and Pollard, D. D., 1977, The stress state
McKenzie, D. P., 1969, The relation between fault plane near Spanish Peaks, Colorado, determined from a
solutions for earthquakes and the directions of the dike pattern. Pure and Applied Geophysics 115: 6985.
principal stresses. Bulletin of the Seismological Society of Murase, T. and McBirney, A. R., 1973, Properties of
America 50: 595601. some common igneous rocks and their melts at high
Means, W. D., 1976, Stress and Strain. New York: temperatures. Geological Society of America Bulletin 84:
Springer-Verlag. 356392.
Mechtly, E. A., 1973, The International System of Units: Murray, J. R. and Segall, P., 2002, Testing the time-pre-
Physical Constants and Conversion Factors, Special dictable earthquake recurrence model by direct
Publication SP-7012. Washington, DC: National measurement of strain accumulation and release.
Aeronautics and Space Administration. Nature 28: 28791.
Meier, M. F., Kamb, B., Allen, C. R., and Sharp, R. P., Murray, J. R., Segall, P., Cervelli, P., Prescott, W., and
1974, Flow of Blue Glacier, Olympic Mountains, Svarc, J., 2001, Inversion of GPS data for spatially
Washington, USA. Journal of Glaciology 13: 187212. variable slip-rate on the San Andreas fault near
Melosh, H. J., 1989, Impact Cratering: A Geologic Process. Parkfield, CA. Geophysical Research Letters 28: 35962.
New York: Oxford University Press. Muskhelishvili, N. I., 1954, Some Basic Problems of the
Meriaux, C. and Lister, J. R., 2002, Calculation of dike Mathematical Theory of Elasticity, tr. J. R. M. Radok,
trajectories from volcanic centers. Journal of Leyden: Noordhoff.
Geophysical Research 107(B4): doi: 10.1029/2001JB000 Nakamura, K., 1977, Volcanoes as possible indicators of
436. tectonic stress orientation: principal and proposal.
Meyers, R. D. and Aydin, A., 2004, The evolution of Journal of Volcanology and Geothermal Research 2: 116.
faults formed by shearing across joint zones in sand- Nakamura, K., Jacob, K. H., and Davies, J. N., 1977,
stone. Journal of Structural Geology 26(5): 94766. Volcanoes as possible indicators of tectonic stress
Michell, J. H., 1899, On the direct determination of orientation: Aleutians and Alaska. Pure and Applied
stress in an elastic solid, with application to the Geophysics 115: 87112.
theory of plates. Proceedings of the London Mathematical Narr, W. and Suppe, J., 1991, Joint spacing in sedimen-
Society 31: 10024. tary rocks. Journal of Structural Geology 13(9): 103748.
Mogi, K., 1958, Relations between the eruptions of Nelson, R. A., 2003, Guide for metric practice. Physics
various volcanoes and the deformations of the Today 56(8): BG15BG16.
ground surfaces around them. Bulletin of Earthquake Nemat-Nasser, S. and Horii, H., 1982, Compression-
Research Institute, University of Tokyo 36: 99134. induced nonplanar crack extension with application
Mohr, O., 1882, ber die Darstellung des to splitting, exfoliation, and rockburst. Journal of
Spannungszustandes eines Krperelements. Geophysical Research 87: 680521.
Civilingenieure 28: 11356. Netz, R., 2000, The origins of mathematical physics:
Mohr, P. J. and Taylor, B. N., 2003, The fundamental new light on an old question. Physics Today June: 327.
physical constants. Physics Today 56(8): BG6BG13. Neurath, C. and Smith, R. B., 1982, The effect of ma-
Mollema, P. N. and Antonellini, M. A., 1996, terial properties on growth rates of folding and
Compaction bands: a structural analog for anti- boudinage: experiments with wax models. Journal of
mode I cracks in aeolian sandstone. Tectonophysics Structural Geology 4: 21529.
267: 20928. Newton, I., 1999, The Principia. Berkeley, CA: University
Moore, J. M. and Schultz, R. A., 1999, Processes of fault- of California Press. (Originally published 1687,
ing in jointed rocks of Canyonlands National Park, Cambridge: Cambridge University Press.)
Utah. Geological Society of America Bulletin 111: Nicholson, R. and Ejiofor, I. B., 1987, The three-
80822. dimensional morphology of arrays of echelon and
Morris, A. P., Ferrill, D. A., and Henderson, D. B., 1996, sigmoidal, mineral-filled fractures: data from north
Slip tendency analysis and fault reactivation. Geology Cornwall. Journal of the Geological Society, London 144:
24: 2758. 7983.
Muller, J. R., Aydin, A., and Maerten, F., 2003, Nur, A. and Byerlee, J. D., 1971, An exact effective stress
Investigating the transition between the 1967 law for elastic deformation of rock with fluids.
Mudurnu Valley and 1999 Izmit earthquakes along Journal of Geophysical Research 76: 641419.
490 REFERENCES
Nur, A., Ron, H., and Scotti, O., 1986, Fault mechanics Topographic wavelengths of Ganymede groove lanes
and kinematics of block rotation. Geology 14: 7469. from Fourier analysis of Galileo images. Journal of
Nye, J. F., 1953, The flow law of ice from measurements Geophysical Research, E, Planets 104: 24 05774.
made in glacier tunnels, laboratory experiments, Paterson, M. S., 1978, Experimental Rock Deformation: the
and the Jungfrau borehole experiment. Proceedings of Brittle Field. New York: Springer-Verlag.
the Royal Society of London, Series A 219: 47789. Patton, T. L., Logan, J. M., and Friedman, M., 1998,
Nye, J. F., 1985, Physical Properties of Crystals. Oxford: Experimentally generated normal faults in single-
Clarendon Press. layer and multilayer limestone specimens at confin-
Obert, L. and Duvall, W. I., 1967, Rock Mechanics and the ing pressure. Tectonophysics 295: 5377.
Design of Structures in Rock. New York: John Wiley and Paul, B., 1960, Prediction of elastic constants of multi-
Sons. phase materials. Transactions of the Metallurgical
OConnor, J. J. and Robertson, E. F., 2004, The MacTutor Society of AIME 218: 3641.
History of Mathematics Archive. St. Andrews, Fife: Paul, B., 1961, Modification of the CoulombMohr
School of Mathematics and Statistics, University of theory of fracture. Journal of Applied Mechanics 28:
St. Andrews. 25968.
Od, H., 1957, Mechanical analysis of the dike pattern Peacock, D. C. P., 1991, Displacement and segment
of the Spanish Peaks area, Colorado. Geological Society linkage strike-slip fault zones. Journal of Structural
of America Bulletin 68: 56776. Geology 13: 102535.
Okada, Y., 1985, Surface deformation due to shear and Peacock, D. C. P. and Marrett, R., 2000, Strain and stress:
tensile faults in a half-space. Bulletin of the reply. Journal of Structural Geology 22: 136978.
Seismological Society of America 75: 113554. Peacock, D. C. P. and Sanderson, D. J., 1991,
Olson, J. and Pollard, D. D., 1989, Inferring palaeo- Displacements, segment linkage and relay ramps in
stresses from natural fracture patterns: a new normal fault zones. Journal of Structural Geology 13:
method. Geology 17: 3458. 72133.
Olson, J. E., 1993, Joint pattern development: effects of Peacock, D. C. P. and Sanderson, D. J., 1994, Geometry
subcritical crack growth and mechanical crack inter- and development of relay ramps in normal fault
action. Journal of Geophysical Research 98(B7): systems. American Association of Petroleum Geologists
12 25165. Bulletin 78: 14765.
Olson, J. E. and Pollard, D. D., 1991, The initiation and Pearl, J., 2000, Causality: Models, Reasoning, and Inference.
growth of en chelon veins. Journal of Structural Geology Cambridge: Cambridge University Press.
13(5): 595608. Peck, L., Barton, C. C., and Gordon, R. B., 1985,
Oppenheimer, D. H., Reasenberg, P. A., and Simpson, Microstructure and the resistance of rock to tensile
R. W., 1988, Fault plane solutions for the 1984 fracture. Journal of Geophysical Research 90: 11 53346.
Morgan Hill, California, earthquake sequence: evi- Peng, S. and Johnson, A. M., 1972, Crack growth and
dence for the state of stress on the Calaveras fault. faulting in cylindrical specimens of Chelmsford
Journal of Geophysical Research 93: 900726. granite. International Journal of Rock Mechanics and
Owen, S., Segall, P., Freymueller, J., et al., 1995, Rapid Mining Science 9: 3786.
deformation of the south flank of Kilauea Volcano, Peng, S. D., 1971, Stresses within elastic circular cylin-
Hawaii. Science 267: 132832. ders loaded uniaxially and triaxially. International
Ozkaya, S. I., 2002, QUADRO: a program to estimate Journal of Rock Mechanics and Mining Science 8:
principal curvatures of folds. Computers and 339432.
Geosciences 28: 46772. Peterson, I., 1993, Newtons Clock: Chaos in the Solar
Pachell, M. A., Evans, J. P., and Taylor, W. L., 2003, System. New York: Freeman.
Kilometer-scale kinking of crystalline rocks in a Petit, J.-P. and Barquins, M., 1988, Can natural faults
transpressive convergent setting, Central Sierra propagate under mode II conditions? Tectonics 7:
Nevada, California. Geological Society of America 124356.
Bulletin 115: 81731. Petit, J.-P. and Mattauer, M., 1995, Palaeostress superim-
Pan, E., Amadei, B., and Savage, W. Z., 1995, position deduced from mesoscale structures in lime-
Gravitational and tectonic stresses in anisotropic stone: the Matelles exposure, Languedoc, France.
rock with irregular topography. International Journal Journal of Structural Geology 17(2): 24556.
of Rock Mechanics and Mining Science 32: 20114. Phillips, F. C., 1954, The Use of Stereographic Projection in
Patel, J., Pappalardo, R. T., Head, J. W., III, et al., 1999, Structural Geology. London: Edward Arnold.
REFERENCES 491
Pinto da Cunha, A., 1990, Scale Effects in Rock Masses. Prager, W., 1961, Introduction to the Mechanics of Continua.
Rotterdam: A. A. Balkema. Boston, MA: Ginn and Co.
Poirier, J.-P., 1985, Creep of Crystals: High-Temperature Prescott, W., 1993, Seeing earthquakes from afar.
Deformation Processes in Metals, Ceramics and Minerals. Nature 364: 1001.
London: Cambridge University Press. Price, E. J. and Brgmann, R., 2002, Interactions
Poliakov, A. N. B., Dmowska, R., and Rice, J. R., 2002, between the Landers and Hector Mine, California,
Dynamic shear rupture interactions with fault earthquakes from space geodesy, boundary element
bends and off-axis secondary faulting. Journal of modeling, and time-dependent friction. Bulletin of the
Geophysical Research 107(B11): no. 2295, doi:10. Seismological Society of America 92: 145069.
1029/2001JB000572. Price, E. J. and Sandwell, D. T., 1998, Small-scale defor-
Pollard, D. D., 2000, Strain and stress: discussion. mations associated with the 1992 Landers,
Journal of Structural Geology 22: 135967. California, earthquake mapped by synthetic aper-
Pollard, D. D. and Aydin, A., 1984, Propagation and ture radar interferometry phase gradients. Journal of
linkage of oceanic ridge segments. Journal of Geophysical Research 103: 27 00116.
Geophysical Research 89(B12): 10 01728. Price, N., 1966, Fault and Joint Development in Brittle and
Pollard, D. D. and Aydin, A., 1988, Progress in under- Semi-Brittle Rock. Oxford: Pergamon Press.
standing jointing over the past century. Geological Price, N. J. and Cosgrove, J. W., 1990, Analysis of
Society of America Bulletin 100: 1181204. Geological Structures. Cambridge: Cambridge
Pollard, D. D., Bergbauer, S., and Mynatt, I., 2004, Using University Press.
differential geometry to characterize and analyze Price, R. A., 1973, Large-scale gravitational flow of
the morphology of joints. In J. W. Cosgrove and T. supracrustal rocks, southern Canadian Rockies. In
Engelder, eds., The Initiation, Propagation, and Arrest of K. A. De Jong and R. Scholten, eds., Gravity and
Joints and Other Fractures, Special Publication 231. Tectonics. New York: John Wiley and Sons,
London: Geological Society, pp. 15382. pp. 491502.
Pollard, D. D., Delaney, P. T., Duffield, W. A., Endo, E. T., Pyne, S. J., 1980, Grove Karl Gilbert: A Great Engine of
and Okamura, A. T., 1983, Surface deformation in Research. Austin, TX: University of Texas Press.
volcanic rift zones. Tectonophysics 94: 54184. Ragan, D. M., 1985, Structural Geology: An Introduction to
Pollard, D. D. and Johnson, A. M., 1973, Mechanics of Geometrical Techniques. New York: John Wiley and
growth of some laccolithic intrusions in the Henry Sons.
Mountains, Utah, II. Tectonophysics 18: 31154. Ramberg, H., 1960, Relationships between length of
Pollard, D. D. and Muller, O. H., 1976, The effect of gra- arc and thickness of ptygmatically folded veins.
dients in regional stress and magma pressure on the American Journal of Science 258: 3646.
form of sheet intrusions in cross section. Journal of Ramberg, H., 1963, Fluid dynamics of viscous buckling
Geophysical Research 81: 97584. applicable to folding of layered rocks. American
Pollard, D. D., Muller, O. H., and Dockstader, D. R., 1975, Association of Petroleum Geologists Bulletin 47: 484505.
The form and growth of fingered sheet intrusions. Ramberg, H., 1967, Gravity, Deformation and the Earths
Geological Society of America Bulletin 86: 35163. Crust. London: Academic Press.
Pollard, D. D., 2000, Strain and stress: Discussion. Ramsay, J. G., 1967, Folding and Fracturing of Rocks. New
Journal of Structural Geology 22: 135967. York: McGraw-Hill.
Pollard, D. D. and Segall, P., 1987, Theoretical displace- Ramsay, J. G. and Huber, M. I., 1987, The Techniques of
ments and stresses near fractures in rock: with Modern Structural Geology, Vol. 2, Folds and Fractures.
applications to faults, joints, veins, dikes, and solu- London: Academic Press.
tion surfaces. In B. K. Atkinson, ed., Fracture Mechanics Ramsay, J. G., 1989, Emplacement kinematics of a
of Rock. London: Academic Press, pp. 277349. granite diapir: the Chindamora batholith,
Pollard, D. D., Segall, P., and Delaney, P. T., 1982, Zimbabwe. Journal of Structural Geology 11: 191210.
Formation and interpretation of dilatant echelon Ramsay, J. G. and Huber, M. I., 1983, The Techniques of
cracks. Geological Society of America Bulletin 93: Modern Structural Geology, Vol. 1, Strain Analysis.
1291303. London: Academic Press.
Potter, D. B. and McGill, G. E., 1978, Valley anticlines Ranalli, G., 1987, Rheology of the Earth: Deformation and
of the Needles District, Canyonlands National Flow Processes in Geophysics and Geodynamics. London:
Park, Utah. Geological Society of America Bulletin 89: Allen and Unwin.
95260. Rawnsley, K. D., Peacock, D. C. P., Rives, T., and
492 REFERENCES
Petit, J.-P., 1998, Joints in the Mesozoic sediments ics of fracture. In H. Liebowitz, ed., Fracture: An
around the Bristol Channel Basin. Journal of Advanced Treatise, Vol. 2. New York: Academic Press,
Structural Geology 20: 164161. pp. 191311.
Raymond, C. F., 1971, Flow in a transverse section of Rice, J. R., 1980, The mechanics of earthquake rupture.
Athabasca glacier, Alberta, Canada. Journal of In A. M. Dziewonski and E. Boschi, eds., Physics of the
Glaciology 10: 558. Earths Interior. Delft: North Holland, pp. 555649.
Reches, Z., 1978, Development of monoclines: Part I. Rice, J. R., 1992, Fault stress states, pre-pressure distrib-
Structure of the Palisades Creek branch of the East utions, and the weakness of the San Andreas fault.
Kaibab monocline, Grand Canyon, Arizona. Geological In B. Evans and T. F. Wong, eds., Fault Mechanics and
Society of America Memoir 151: 23571. Transport Properties of Rocks. New York: Academic
Reches, Z., 1983, Faulting of rocks in three-dimen- Press, pp. 43559.
sional strain fields: II. Theoretical analysis. Richardus, P. and Adler, R. K., 1972, Map Projections For
Tectonophysics 95: 13356. Geodesists, Cartographers and Geographers. Amsterdam:
Reches, Z. and Dieterich, J. H., 1983, Faulting of rocks North-Holland.
in three-dimensional strain fields: I. Failure of rocks Rispoli, R., 1981, Stress fields about strike-slip faults
in polyaxial, servo-control experiments. inferred from stylolites and tension gashes.
Tectonophysics 95: 11132. Tectonophysics 75: T29T36.
Reches, Z. and Johnson, A. M., 1978, Development of Roberts, A., 2001, Curvature attributes and their appli-
monoclines: Part II. Theoretical analysis of mono- cation to 3D interpreted horizons. First Break 19:
clines. Geological Society of America Memoir 151: 85100.
273311. Robin, P.-Y. and Cruden, A. R., 1994, Strain and vorticity
Renshaw, C. E., 2000, Fracture spatial density and the patterns in ideally ductile transpression zones.
anisotropic connectivity of fracture networks. Journal of Structural Geology 16: 44766.
Geophysical Monograph 122: 20311. Robinson, L. H., 1959, The effect of pore and confining
Renshaw, C. E. and Harvey, C. F., 1994, Propagation pressure on the failure process in sedimentary rock.
velocity of a natural hydraulic fracture in a poroelas- Quarterly of the Colorado School of Mines 54: 17799.
tic medium. Journal of Geophysical Research, B, Solid Rubin, A. M., 1990, A comparison of rift-zone tectonics
Earth and Planets 99(11): 21 65977. in Iceland and Hawaii. Bulletin of Volcanology 52(4):
Renshaw, C. E. and Park, J. C., 1997, Effect of mechani- 30219.
cal interactions on the scaling of fracture length Rubin, A. M., 1993, Tensile fracture of rock at high con-
and aperture. Nature 386(6624): 4824. fining pressure: implications for dike propagation.
Renshaw, C. E. and Pollard, D. D., 1994, Numerical sim- Journal of Geophysical Research 98: 15 91935.
ulation of fracture set formation: a fracture mechan- Rubin, A. M., 1995, Propagation of magma-filled cracks.
ics model consistent with experimental Annual Reviews of Earth and Planetary Sciences 23:
observations. Journal of Geophysical Research, B, Solid 287336.
Earth and Planets 99(5): 935972. Rubin, A. M. and Pollard, D. D., 1988, Dike-induced
Renshaw, C. E. and Pollard, D. D., 1995, An experimen- faulting in rift zones of Iceland and Afar. Geology 16:
tally verified criterion for propagation across 41317.
unbonded frictional interfaces in brittle, linear Rudnicki, J. W., 1977, The inception of faulting in a
elastic materials. International Journal of Rock rock mass with a weakened zone. Journal of
Mechanics and Mining Science 32(3): 23749. Geophysical Research 82: 84454.
Resnick, R. and Halliday, D., 1977, Physics, part 1, 3rd Rudnicki, J. W., 1979, Rotation of principal stress axes
edn. New York: John Wiley and Sons. caused by faulting. Geophysical Research Letters 6:
Reynolds, O., 1883, An experimental investigation of 1358.
the circumstances which determine whether the Rudnicki, J. W., 1980, Fracture mechanics applied to
motion of water shall be direct or sinuous, and the the Earths crust. Annual Reviews of Earth and
laws of resistance in parallel channels. Philosophical Planetary Sciences 8: 489525.
Transactions of the Royal Society of London 174: 93582. Rummel, F., 1987, Fracture mechanics approach to
Ribe, N. M. and Christensen, U. R., 1999, The dynamical hydraulic fracturing stress measurements. In B. K.
origin of Hawaiian volcanism. Earth and Planetary Atkinson, ed., Fracture Mechanics of Rock, Academic Press
Science Letters 171: 51731. Geology Series. London: Academic Press, pp. 21739.
Rice, J. R., 1968, Mathematical analysis in the mechan- Rutter, E. H., 1983, Pressure solution in nature, theory
REFERENCES 493
and experiment. Journal of the Geological Society of Shaw, H. R., 1963, ObsidianH2O viscosities at 1000 and
London 140: 72540. 2000 bars in the temperature range 700 to 900.
Ryan, M. P. and Blevins, J. Y. K., 1987, The Viscosity of Journal of Geophysical Research 68: 633743.
Synthetic and Natural Silicate Melts and Glasses at High Shaw, H. R., 1969, Rheology of basalt in the melting
Temperatures and 1 bar (105 Pascals) Pressure and at range. Journal of Petrology 10: 51035.
Higher Pressures, Bulletin 1764. Denver, CO: US Shaw, H. R., Peck, D. L., Wright, T. L., and Okamura, R.,
Geological Survey. 1968, The viscosity of basaltic magma: an analysis of
Ryan, R. J. and Smith, P. K., 1998, A review of the field measurements in Makaopuhi lave lake, Hawaii.
mesothermal gold deposits of the Meguma Group, American Journal of Science 266: 22564.
Nova Scotia, Canada. Ore Geology Reviews 13: 15383. Sheriff, R. E. and Geldart, L. P., 1995, Exploration
Sander, B., 1970, An Introduction to the Study of Fabrics of Seismology. New York: Cambridge University Press.
Geological Bodies. Oxford: Pergamon Press. Sherwin, J.-A. and Chapple, W. M., 1968, Wavelengths
Sanford, A. R., 1959, Analytical and experimental study of single layer folds: a comparison between theory
of simple geologic structures. Geological Society of and observation. American Journal of Science 266:
America Bulletin 70: 1952. 16779.
Savage, J. C., Prescott, W. H., Lisowski, M., and King, N., Shipton, Z. K. and Cowie, P. A., 2001, Damage zone and
1979, Deformation across the Salton trough, slip-surface evolution over micron to km scales in
California, 19731977. Journal of Geophysical Research high-porosity Navajo sandstone, Utah. Journal of
84: 306979. Structural Geology 23: 182544.
Savage, W. Z. and Swolfs, H. S., 1986, Tectonic and gravi- Sibson, R. H., 1986, Earthquakes and rock deformation
tational stresses in long symmetric ridges and in crustal fault zones. Annual Reviews of Earth and
valleys. Journal of Geophysical Research 91: 367785. Planetary Sciences 14: 14975.
Savage, W. Z., Swolfs, H. S., and Powers, P. S., 1985, Sibson, R. H., 1987, Earthquake rupturing as a mineral-
Gravitational stresses in long symmetric ridges and izing agent in hydrothermal systems. Geology 15:
valleys. International Journal of Rock Mechanics and 7014.
Mining Science 22: 291302. Sibson, R. H., 1989, Earthquake faulting as a structural
Scheidegger, A. E., 1962, Stresses in the Earths crust as process. Journal of Structural Geology 11(1/2): 114.
determined from hydraulic fracturing data. Geologie Sieh, K., Jones, L., Hauksson, E., et al., 1993, Near-field
und Bauwesen 27: 4553. investigations of the Landers earthquake sequence,
Schlichting, H., 1979, Boundary-Layer Theory. New York: April to July 1992. Science 260: 1716.
McGraw-Hill. Sih, G. C., 1973, Handbook of Stress Intensity Factors.
Scholz, C. H., 1990, The Mechanics of Earthquakes and Bethlehem, PA: Institute of Fracture and Solid
Faulting. Cambridge: Cambridge University Press. Mechanics, Lehigh University.
Segall, P. and Harris, R. A., 1986, Slip deficit on the San Smith, R. B., 1975, Unified theory of the onset of
Andreas Fault at Parkfield, California, as revealed by folding, boudinage, and mullion structure. Geological
inversion of geodetic data. Science 233: 140913. Society of America Bulletin 86: 16019.
Segall, P., McKee, E. H., Martel, S. J., and Turrin, B. D., Smith, R. B., 1977, Formation of folds, boudinage, and
1990, Cretaceous age of fractures in the Sierra mullions in non-Newtonian materials. Geological
Nevada batholith, California. Geology 18: 124851. Society of America Bulletin 88: 31220.
Segall, P. and Pollard, D. D., 1980, Mechanics of discon- Smith, R. B., 1979, The folding of a strongly non-
tinuous faults. Journal of Geophysical Research 85: Newtonian layer. American Journal of Science 279:
433750. 27287.
Segall, P. and Pollard, D. D., 1983a, Joint formation in Smoluchowski, M., 1909, ber ein gewisses
granitic rock of the Sierra Nevada. Geological Society of Stabilittsproblem der Elastizittslehre und dessen
America Bulletin 94: 56375. Beziehung zur Entstehung von Faltengebirgen. Anz.
Segall, P. and Pollard, D. D., 1983b, Nucleation and Akad. Wiss. Krakau, Math. Naturw. 2: 3 and 727.
growth of strike slip faults in granite. Journal of Sokolnikoff, I. S., 1956, Mathematical Theory of Elasticity.
Geophysical Research 88(B1): 55568. New York: McGraw-Hill.
Segall, P. and Simpson, C., 1986, Nucleation of ductile Sokoutis, D., 1987, Finite strain effects in experimental
shear zones on dilatant fractures. Geology 14: 569. mullions. Journal of Structural Geology 9: 23342.
Selby, S. M., 1975, Standard Mathematical Tables. Sommer, E., 1969, Formation of fracture lances in
Cleveland, OH: CRC Press, p. 756. glass. Engineering Fracture Mechanics 1: 53946.
494 REFERENCES
Sosa, E. and Tooley, M., 1993, Causation. Oxford: Oxford echelon fractures in rock: implications from labora-
University Press, p. 252. tory and numerical experiments. Journal of Structural
Spence, D. A. and Turcotte, D. L., 1985, Magma-driven Geology 15: 32334.
propagation of cracks. Journal of Geophysical Research Thomson, W., 1891, Electrical units of measurement:
90: 57580. lecture delivered at the Institution of Civil Engineers
Stein, R. S., Dieterich, J. H., and Barka, A. A., 1996, Role on May 3, 1883. Nature Series, Popular Lectures and
of stress triggering in earthquake migration on the Addresses, Vol. 1, Constitution of Matter. London:
North Anatolian Fault. Physics and Chemistry of the MacMillan, pp. 80134.
Earth 21: 22530. Tikoff, B. and S. F. Wojtal, 1999, Displacement of
Stein, R. S. and Lisowski, M., 1983, The 1979 Homestead control of geologic structures. Journal of Structural
Valley earthquake sequence, California: control of Geology 21: 95967.
aftershocks and postseismic deformation. Journal of Timoshenko, S., 1953, History of Strength of Materials.
Geophysical Research 88: 647790. New York: McGraw-Hill.
Stephansson, O., 1974, Stress-induced diffusion during Timoshenko, S., 1958, Strength of Materials. I. Elementary
folding. Tectonophysics 22: 23351. Theory and Problems. Princeton, NJ: Van Nostrand.
Stephansson, O., 1993, Rock stress in the Timoshenko, S. and Goodier, J. N., 1970, Theory of
Fennoscandian shield. In J. A. Hudson, ed., Elasticity. New York: McGraw-Hill.
Comprehensive Rock Engineering. Oxford: Pergamon Timoshenko, S. and Woinowsky-Krieger, S., 1959, Theory
Press, pp. 44559. of Plates and Shells. New York: McGraw-Hill.
Sternlof, K. R., Chapin, J. R., Pollard, D. D., and Timoshenko, S. and Young, D. H., 1968, Elements of the
Durlofsky, L. J., 2004, Permeability effects of defor- Strength of Materials. Princeton, NJ: Van Nostrand.
mation band arrays in sandstone. American Treagus, S. H. and Lisle, R. J., 1997, Do principal sur-
Association of Petroleum Geologists Bulletin 88(9): faces of stress and strain always exist? Journal of
131529. Structural Geology 19: 9971010.
Stoker, J. J., 1969, Differential Geometry. New York: Wiley- Treagus, S. H. and Treagus, J. E., 2002, Studies of strain
Interscience. and rheology of conglomerates. Journal of Structural
Struik, D. J., 1961, Lectures on Classical Differential Geology 24: 154167.
Geometry. Reading, MA: Addison-Wesley. Treiman, J., Kendrick, K., Bryant, W., Rockwell, T., and
Tada, H., Paris, P. C., and Irwin, G. R., 1973, The Stress McGill, S., 2002, Primary surface rupture associated
Analysis of Cracks Handbook. Hellertown, PA: Del with the Mw 7.1 16 October 1999 Hector Mine earth-
Research Corp. quake, San Bernardinno County, California. Bulletin
Tapponnier, P. and Brace, W. F., 1976, Development of of the Seismological Society of America 92: 1171191.
stress-induced microcracks in Westerly granite. Truesdell, C., 1961, Stages in the development of the
International Journal of Rock Mechanics and Mining concept of stress. In Problems of Continuum Mechanics.
Science 13: 10312. Philadelphia, PA: Society for Industrial and Applied
Taylor, W. L. and Pollard, D. D., 2000, Estimation of in Mathematics, pp. 55664.
situ permeability of deformation bands in porous Truesdell, C. and Noll, W., 1965, The non-linear field
sandstone, Valley of Fire, Nevada. Water Resources theories of mechanics. In S. Flugge, ed., Encyclopedia
Research 36(9): 2595606. of Physics, Vol. 3, No. 1. Berlin: Springer-Verlag.
Taylor, W. L., Pollard, D. D., and Aydin, A., 1999, Fluid Trusheim, F., 1960, Mechanism of salt migration in
flow in discrete joint sets: field observations and northern Germany. American Association of Petroleum
numerical simulations. Journal of Geophysical Research, Geologists Bulletin 44: 151940.
B, Solid Earth and Planets 104(12): 28 98329 006. Tullis, T. E., 1988, Rock friction constitutive behavior
Terzaghi, K., 1943, Theoretical Soil Mechanics. New York: from laboratory experiments and its implications
John Wiley and Sons. for an earthquake prediction field monitoring
Thatcher, W. and Bonilla, M., 1989, Earthquake fault program. Pure and Applied Geophysics 126: 55588.
slip estimation from geologic, geodetic and seismo- Tullis, T. E. and Tullis, J., 1986, Experimental rock
logical observations: implications for earthquake deformation techniques. In B. E. Hobbs and H. C.
mechanics and fault segmentation, US Geological Heard, eds., Mineral and Rock Deformation: Laboratory
Survey, Open File Report 89315. Menlo Park, CA: Studies, The Paterson Volume, Geophysical Monograph.
pp. 38699. Washington, DC: American Geophysical Union,
Thomas, A. L. and Pollard, D. D., 1993, The geometry of pp. 297324.
REFERENCES 495
Turcotte, D. L. and Schubert, G., 1982, Geodynamics: tation and patterns of wing cracks and solution sur-
Applications of Continuum Physics to Geological Problems. faces at the tips of a sliding flaw or fault. Journal of
New York: John Wiley and Sons. Geophysical Research 103: 242738.
Turner, F. J. and Weiss, L. E., 1963, Structural Analysis of Willemse, E. J. M., Pollard, D. D., and Aydin, A., 1996,
Metamorphic Tectonites. New York: McGraw-Hill. Three-dimensional analyses of slip distributions on
Twiss, R. J. and Moores, E. M., 1992, Structural Geology. normal fault arrays with consequences for fault
New York: W. H. Freeman and Co. scaling. Journal of Structural Geology 18(23): 295309.
Van Dyke, M., 1982, An Album of Fluid Motion. Stanford, Williams, M. L., 1957, On the stress distribution at the
CA: The Parabolic Press. base of a stationary crack. Journal of Applied
Varberg, D. and Purcell, E. J., 1992, Calculus with Mechanics, Transactions ASME 24: 10914.
Analytical Geometry. Englewood Cliffs, NJ: Prentice- Wilson, G. and Cosgrove, J. W., 1982, Introduction to
Hall. Small-Scale Geological Structures. London: George Allen
Vendeville, B. C., Ge, H., and Jackson, M. P. A., 1995, and Unwin.
Scale models of salt tectonics during basement- Wiltschko, D. V. and Sutton, S. J., 1982, Deformation by
involved extension. Petroleum Geosciences 1: 17983. overburden of a coarse quartzite conglomerate.
Venkatraman, B. and Patel, S. A., 1970, Structural Journal of Geology 90: 72533.
Mechanics with Introduction to Elasticity and Plasticity. Wong, T.-F., 1982a, Effects of temperature and pressure
New York: McGraw-Hill. on failure and post-failure behavior of Westerly
Wang, W.-H. and Davis, D. M., 1996, Sandbox model granite. Mechanics of Materials 1: 317.
simulation of forearc evolution and noncritical Wong, T.-F., 1982b, Micromechanics of faulting in
wedges. Journal of Geophysical Research 101: 11 32939. Westerly granite. International Journal of Rock
Wallace, R. E., 1951, Geometry of shearing stress and Mechanics and Mining Sciences 19: 4964.
relation to faulting. The Journal of Geology 59: 11830. Wood, D. S. and Oertel, G., 1980, Deformation in the
Weertman, J., 1996, Dislocation Based Fracture Mechanics. Cambrian slate belt of Wales. Journal of Geology 88:
River Edge, NJ: World Scientific. 30926.
Weertman, J. and Weertman, J. R., 1964, Elementary Woodford, A. O., 1956, What is Geologic Truth? Journal
Dislocation Theory. New York: Macmillan. of Geological Education 4(1): 58.
Weinberger, R., 2001, Joint nucleation in layered rocks Woodworth, J. B., 1896, On the fracture system of
with non-uniform distribution of cavities. Journal of joints, with remarks on certain great fractures.
Structural Geology 23: 124154. Boston Society of Natural Historical Proceedings 27:
Weyl, H., 1987, The Continuum: A Critical Examination of 16383.
the Foundation of Analysis. Kirksville, MO: The Thomas Wosser, T. D., Campi, D. E., Fovince, M. A., and Smith,
Jefferson University Press, p. 94. W. H., 1982, Damage to Engineered Structures in
White, F. M., 1974, Viscous Fluid Flow. New York: California, The Imperial Valley, California, Earthquake of
McGraw-Hill. October 15, 1979, Professional Paper 1254. Washington,
Whitehead, J. A. Jr. and Luther D. S., 1975, Dynamics of DC: US Geological Survey.
laboratory diapir and plume models. Journal of Wu, H. and Pollard, D. D., 1995, An experimental study
Geophysical Research 80: 70517. of the relationship between joint spacing and layer
Whitney, C. S., 1943, Discussion on paper by V. P. thickness. Journal of Structural Geology 17(6): 887905.
Jensen. Journal of the American Concrete Institute 39: Yeats, R. S., Sieh, K., and Allen, C. R., 1997, The Geology of
548. Earthquakes. New York: Oxford University Press.
Wilczek, F., 2004, Whence the force F ma? II: Zakarevicius, A., 2000, Skaitmeniniai vertikaliuju
Rationalizations. Physics Today V: 1011. zemes plutos deformaciju rodikliai. Geodezija ir
Willemse, E. J. M., 1997, Segmented normal faults: cor- Kartografija [Digital values of vertical crustal defor-
respondence between three-dimensional mechanical mations. Geodesy and Cartography] 26: 579.
models and field data. Journal of Geophysical Research Zebker, H. A., Rosen, P. A., Goldstein, R. M., Gabriel, A.,
102(1): 67592. and Werner, C. L., 1994, On the derivation of co-
Willemse, E. J. M., Peacock, D. C. P., and Aydin, A., 1997, seismic displacement fields using differential radar
Nucleation and growth of strike-slip faults in lime- interferometry: the Landers earthquake. Journal of
stones from Somerset, UK. Journal of Structural Geology Geophysical Research 99: 19 61743.
19: 146177. Zenzri, H. and Keer, L. M., 2001, Mechanical analyses of
Willemse, E. J. M. and Pollard, D. D., 1998, On the orien- the emplacement of laccoliths and lopoliths. Journal
496 REFERENCES
of Geophysical Research, B, Solid Earth and Planets 106(7): New evidence on the state of stress of the San
13 78192. Andreas fault system. Science 238: 110511.
Zheng, Z., Kemeny, J., and Cook, N. G. W., 1989, Zoback, M. L., 1992, First- and second-order patterns of
Analysis of borehole breakouts. Journal of Geophysical stress in the lithosphere: the world stress map
Research 94: 717182. project. Journal of Geophysical Research 97: 1170328.
Ziv, A. and Rubin, A. M., 2000, Stability of dike intru- Zoback, M. L., Zoback, M. D., Adams, J., et al., 1989,
sion along preexisting fractures. Journal of Geophysical Global patterns of tectonic stress. Nature 341: 2918.
Research 105(3): 594761. Zuber, M. T., Parmentier, E. M., and Fletcher, R. C., 1986,
Zoback, M. D., 1985, Well bore breakouts and in-situ Extension of continental lithosphere: a model for
stress. Journal of Geophysical Research 90: 552330. two scales of basin and range deformation. Journal of
Zoback, M. D., Zoback, M. L., Mount, V. S., et al., 1987, Geophysical Research 91: 482638.
Index