Anda di halaman 1dari 11

New Algorithms for

Calculating Hertzian
Stresses,
Deformations, and
Contact Zone
Parameters
Emil W. Deeg
AMP Incorporated

ABSTRACT presented by H. Hertz in 1881 and extended in 1882 2 to


The complete set of assumptions directly or indirectly af- include a hardness definition. Summaries and examples are
fecting Hertzs contact stress theory is presented. A review found in many textbooks and handbooks.3-8 Application
of the derivation of his fundamental formulas leads to a to electrical connections is still under discussion.9-ll For
computer program that does not depend on elliptic inte- fiber-optic PC connections, Hertzs theory yields design
grals nor the conventional numerical method and the guidelines if calculations are restricted to the endfaces of
tabulated function of two coefficients introduced by Hertz. the contacting fibers.12 Analysis of a complete, 3-layer, ro-
A method permitting systematic characterization and de- tationally symmetric PC connection requires finite element
scription of combinations of two contacting bodies with methods. 13
individual, monotone sequentially, or randomly changing
principal radii of curvature is introduced. The validity of Although Hertzs theory provides only one approach to
Hertzs theory for describing fiber-optic PC connections is contact stress theory, its success in estimating allowable
tested through a comparison of specific features of the maximum or desired minimum local loads warrants a thor-
system with Hertzs assumptions. ough examination of its features. A particularly convincing
argument is its role in the optimization of design and use of
ball and roller bearings. One of the early reports on this
INTRODUCTION topic was given in 1901 by F. Heerwagen, who at the time
The ongoing miniaturization of electrical contacts and the was in charge of technology at a mine in Spain. In coopera-
introduction of fiber-optic physical contact (PC) connectors tion with an Austrian pump manufacturer, Heerwagen
triggered an examination of theories suitable for predicting reduced drastically the failure rate of ball bearings in his
local stresses and deformations at the point of contact. pumps. 14 A systematic, basic engineering study related
A generally accepted theory serving this purpose was to this topic, but addressing different applications, was

Copyright 2004 by Tyco Electronics Corporation. All rights reserved.

14 E.W. Deeg AMP Journal of Technology Vol. 2 November, 1992


published in 1921 by H. L. Whittemore and S. N. of the theory in its most general form, several computer
Petrenko.15 , a Hertzs theory has been used also for optimal algorithms were written requiring only input of material
selection of abrasives and pressure in grinding and constants, applied force and the geometric parameters of
polishing of brittle materials.16 the system. All use readily available software packages
without resorting to traditional computer programming.
At present, calculation of Hertzian stresses for ellipsoidal One of them is presented here.
surfaces relies on two functions of geometric parameters
1,2
presented in tabular form. A revised and extended table
REVIEW OF HERTZS CONTACT STRESS
was published by Heerwagen,14 rearranged by Whittemore15b
and is found in most of the references. 3-8 Use of these THEORY
numerically defined functions requires time-consuming, Basic Assumptions
rather tedious calculations of auxiliary quantities. To avoid
them, ellipsoidal surfaces in contact with each other are The decision to rejector accept an established theory for
frequently approximated by idealized geometries such as modelling a specific physical system requires a comparison
sphere-on-sphere, sphere-on-plane, cylinder-on-cylinder, of the features of the system with the complete set of as-
and cylinder-on-plane. Improper use of these simplifica- sumptions or axioms on which the theory is based. For the
tions can lead to erroneous prediction of the behavior of Hertzian contact stress theory, the fundamental
the modelled system. assumptions2,d are:
(a) at the point of contact the shape of each of the con-
There are also problems of a fundamental nature. The tacting surfaces can be described by a homogeneous
more-than-a-century-old Hertzian contact stress theory is quadratic polynomial in two variables;
being applied mostly based on secondary sources of the (b) both surfaces are ideally smooth;
kind mentioned above. By necessity, handbooks, textbooks,
and compendia of formulas can give only incomplete ac- (c) contact stresses and deformations satisfy the
counts of original publications. It is left to the user to con- differential equations for stress and strain of homo-
sult the sources in case of questions. For Hertzs contact geneous, isotropic, and elastic bodies in equilibrium;
stress theory this is difficult today. Although his collected (d) the stress disappears at great distance from the
works are available in leading libraries, the language bar- contact zone;
rier remains. But even engineers fluent in todays technical
(e) tangential stress components are zero at both
German will have to familiarize themselves with concepts
surfaces within and outside the contact zone;
and terminology of the late 19th century. One example is
Hertzs use of the term Druck for pressure as well as for a (f) normal stress components are zero at both surfaces
quantity today termed Kraft (= force). Furthermore, his outside the contact zone;
contemporaries were quite familiar with elliptic functions (g) the stressg integrated over the contact zone equals
and integrals. A case in point is Heerwagens study.14 Con- the force pushing the two bodies together;
sequently, Hertz left out details of derivations that now
require special effort by the reader to follow. Finally, the (h) the distance between the two bodies is zero within
but finite outside the contact zone;
collected works contain misprints that become obvious
when following Hertzs derivations step by step but will (i) in the absence of an external force, the contact zone
lead to errors if not recognized and corrected. Also of degenerates to a point.
interest, Hertz does not give explicit expressions for the
stress components inside the contacting bodies. Such for- While not directly identified in his papers, Hertz alerts the
mulas were derived later by M. T. Huber17, who also found reader to some assumptions generally accepted by his
that the intuitive stress distribution near the contact zone contemporaries. In the first paragraph of reference 1, he
given by Hertz2 is incorrect. Subsequent papers by considers the contact force to be of the nature of what
S. Fuchs18 and W. B. Morton et al.19 address the same would be now termed a (Dirac) d -distribution:
topic.c
In the theory of elasticity it is assumed that deformations are
caused by forces acting upon the interior and by forces acting upon
These considerations and the need to model stresses, de-
the surface of a body. Both kinds of forces may become infinitely
formations as well as shape and size of the contact zone in high in isolated, infinitely small portions of the body but so that
fiber-optic PC connections led to a study of Hertzs papers. their integrals over these portions are finite. If we then surround
The results are summarized below. To make efficient use the point of discontinuity by a closed surface that is very small

d
Reference 2, pp. 155-157 is preferred here because of the more concise formu-
a
This study distinguishes conceptually and experimentally between friction under lation than that given in reference 1. Secondary assumptions of importance are
load and maximum safe load. The mechanics of electrical connections make a identified in the text below.
e
similar distinction. Whittemores report, if consulted and interpreted properly, According to Hertz2, this assumption is the result of expanding around the
will add valuable information to the ongoing discussion. point of contact the equations of the contacting surfaces and neglecting higher
b
Whittemore credits Dr. L. B. Tuckerrnan for preparing the table of coefficients than second-order terms.
f
for Hertzs theory. To this statement should be added as commonly used by mechanical engineers
c
Although restricted to contacts sphere/sphere and sphere/p late, respectively, during the late 19th century.
g
these papers provide essential input to the ongoing discussion on applicability of Hertz calls this force Gesamtdruck (total pressure), which according to todays
Hertzs theory to connector mechanics. terminology leads to a dimensionally incorrect end result for the contact stress,

AMP Journal of Technology Vol. 2 November, 1992 E.W. Deeg 15


compared to the overall dimensions of the body but very large com- Fundamental Formulas h8
pared to the dimension of the part in which the forces act, we can
Hertz uses a Cartesian system of coordinates {x,y,z] with
consider the deformations outside of this surface independently
from those inside. At the outside the deformations depend on the
origin at the initial point of contact of the two bodies and
shape of the entire body, the distribution of the forces in general the z-axis oriented parallel to the applied force. This re-
and the finite integrals of the force components at the points of dis- quirement introduces an additional assumption if, for more
continuity. Inside they depend only on the distribution of the forces general reasons, the physical system is already described in
acting there: pressures and deformations are infinitely high differently oriented Cartesian coordinates or in a different
compared to those outside. coordinate system. The slanted fiber-optic PC connection is
an example. For optical reasons and because of its global
As is common in mathematical descriptions of physical sys- symmetry, the z-axis would be chosen to coincide with the
tems, the above assumptions are idealizations. The terms optical fiber axis. In contrast, the z-axis used for a Hertzian
ideally smooth, homogeneous, isotropic, elastic bodies, equi- analysis of the connection must be oriented parallel to the
librium, at great distance indicate this. For each concrete direction of the slant.
case, the degree of approximation of the physical system by
the theory must be considered and the effect on the results
assessed accordingly. Frequently, comparing qualities is
sufficient to determine the utility of the theory. If this Table 2. List of symbols. The third and fourth column show
approach is not possible or its feasibility appears question- those used in this paper.
able, quantitative comparisons are necessary. Examples of
such critical evaluations are already given by Hertz1,2 and
by Whittemore15. Neglecting these considerations has led
to misinterpretations and occasionally even unjustified
rejection of the theory.

For the fiber-optic PC connections discussed in references


12 and 13, such a comparison is summarized in Table 1.
The result is that two PC fiberends in contact with each
other meet Hertzs assumptions with satisfactory approxi-
mation. However, if the entire system comprising fiber,
adhesive layer and ferrule is to be analyzed, (c) and (g) are
not valid. Furthermore, if the connection is not aligned
axially, (a) and (e) may be violated. If one or both of the
connecting fibers are recessed, (a) is invalid. In the latter
case, the ferrules, not the fiber endfaces, establish the
initial contact, which may also cause violation of (i).

Table 1. Example for assessing applicability of Hertzs


contact stress theory. The fundamental assumptions of the
theory are compared with specific features of fiber-optic PC In the following each of the two bodies in contact with each
connections described in references 12 and 13, other is identified by the subscript i (= 1, 2). To distinguish
between the two principal radii of curvature of the surfaces
of the two bodies at the contact point, a second subscript j
(= 1,2) is introduced. The theory requires input of three
sets of data:
(1) a one-member set of the force p pushing the bodies
together;
(2) a four-member set of elastic constants, which within
the theory can be reduced to a two-member set;
(3) a five-member set of geometric parameters describ-
ing shape and relative orientation of the contacting
bodies.

The two elastic constants Youngs modulus E and Poissons


ratio v enter the final formulas in combination as the

h
For a summary of Symbols and abbreviations, see Table 2.

16 E.W. Deeg AMP Journal of Technology Vol. 2 November, 1992


Hertz coefficient i Introducing an auxiliary angle W through

Four geometric parameters characterize the shape of the


contacting surfaces. They are two principal radii of curva-
ture ril and ri2 for each body i with rij > 0 if the center of
curvature is inside body i. The theory uses the r ij in the yields
functions

To describe deformations of the two bodies under load,


The fifth geometric parameter is an angle w in the {x,y} Hertz assigns to each of them a separate Cartesian system,
plane which at infinity is attached to the corresponding body. Un-
der zero load the new systems coincide with the primary
system. A compressive force p applied parallel to the z-axis
causes a displacement of the secondary systems relative to
the primary system but so that, according to Hertzl, the
Within the same body, i.e. for the same index i, the planes plane z = 0 in each of them is infinitely close to the surfa ce
{z,ril] and {r, i2 } are normal to each other. of the corresponding body and, thus, can be taken as the
surface itself. The direction of the z-axis is the direction of
Hertz describes the surfaces by the normal to said surface.k

Because the two surfaces described by equation (3) are


pushed towards each other in the z-direction, the circum-
ference of the contact zone is an ellipse. The lengths of its
rotates the {x,y} coordinates so that C l = C2 = C and ob- semiaxes a and b are different from but their directions
tains for the distance e normal to the {x,y} plane between coincide with the directions of the semiaxes of an ellipse of
points {X l,yl} = (x2,y2}j equation (4), with e = constant. To link the force p and the
geometry of the contact ellipse, Hertz introduces a function
P(x,y,z), in reference 2 without explanation and in reference
1 through an electrostatic model]

From assumption (h) and the nature of e as a distance, it


follows that the conic described by equation (4) is an
ellipse, which includes the case of two contacting spheres,
i.e. RI = R2 = 0. Contact between cylinders requires
additional considerations and is discussed below. with u being the positive root of
With the abbreviations of equations (2), it is

k
These statements introduce the requirement that inside the contact zone
deformations in the z-direction are small compared to the body dimension in
this direction.
1
P is the potential of an ellipsoidal shell of zero thickness and zero extension in
i the z-direction (principal semiaxis c = 0), The justification given in reference 1
Hertz uses Kirchhoffs elastic constants. For this paper they were converted to
the now commonly accepted Youngs modulus and Poissons ratio. cannot hide its axiomatic introduction. A concise review of potentials of
j
This approach follows Hertzs earlier article. i In the later one, Hertz derives ellipsoids, still written in a form that provides access to their use by Hertzs
explicit equations for the surfaces yielding equations (5a) and (5b).2 contemporaries, is given by R von Mises.20

AMP Journal of Technology Vol. 2 November, 1992 E.W. Deeg 17


The body-specific material constants are then introduced The force p compresses the two bodies within the contact
by two auxiliary functions zone. This results in a deformation of the surfaces so that
two points located inside each body at {xl,yl}] = {x2,y2}
approach each other in the z-direction by

Some of their properties are discussed by Hertz.l


which with equation (4) yields
The functions P and IIi are used mostly to prove the valid-
ity of the theory within the set of assumptions (a) through
(i)m. A slightly modified P(x,y,0) yields the final expressions
for the shape of and the deformation and stress at the con- Comparing equations (1 lb) and (12b) yields
tact zone. To that end the displacements x i, h i and z i in the
directions x, y, and z, respectively, are written as

These equations contain as unknowns only the two


semiaxes of the circumference of the contact ellipse. Intro-
ducing the ratio k = b/a and setting l = b2 t2 in equations
(llc) and (11d) but l = a2 t2 in equation (11e) yields
and subsequently introduced in the corresponding terms
for the stress components. Of particular interest is the dif-
ference of the deformations in the z-direction in the plane
z = 0. Equation (l0b) yields

or and after some algebraic rewriting

with

where k is the root of the transcendental equation

m
These proofs can be carried out without difficulties and are not repeated here.
Frequently used is 2 P = O. To facilitate understanding of the features of the
theory this paper concentrates on the derivation of the final formulas.

18 E.W. Deeg AMP Journal of Technology Vol. 2 November, 1992


To find the quantity known today as Hertzian stress the infinite. In this case, a is determined by the global shape of
normal stress component ZZ is calculated for the contact the bodies and not by the conditions existing at the contact
zone where z = 0, which in equation (8b) yields u = 0. area. Thus, the cylinder/cylinder contact is actually outside
With the validity range of the theory.

To make the theory accessible to numerical application,


Hertz expresses a and b by

n
it is

where the dependency on the auxiliary angle W and the


integrals I(k) and J(k) of equations (15a) and (15b) are
included in the two coefficients f and g. Because of equa-
tion (16) they depend only on .W Hertz states in reference
2, page 182
For u = 0 the derivative P/z assumes the indeterminate
. . . the integrals in question can all be reduced to complete elliptic
form 0 ` Some rewriting involving the solution of the cu-
integrals of the first kind and their derivatives after the modulus.
bic equation (8b) and application of de lHospitals rule Thus, they can be found via Legendres tables without introducing
yields for the normal stress (ZZ)Z=0 within the contact zone new quadrature. However, the calculations are extensive and
p
therefore I calculated the table given below.

His table, which is included here in Table 3, contains to


three decimal points the coefficients f and g for ten values
of .W They are found in footnotes in references 1 (p. 164)
which for {x,y} = {0,0] is the Hertzian stress and 2 (p. 182); according to the latter, interpolation be-
tween these values will probably always offer sufficient
accuracy.

Hertz leaves it to the reader to convert the improper inte-


If the contact is formed by two cylinders, Hertz considers in grals in equations (14) and (15) to integrals with finite up-
reference 2, p. 186/87 the contact ellipse for a ` . This per limits. He also does not describe how he reduced these
requires also that p ` to maintain a finite value for the integrals named here H(k), I(k), and J(k) to complete ellip-
force per unit length. He finds B = (rl + r2)/2 with ri being tic integrals of the first kind and their derivatives after the
the finite radius of curvature of body i. It is also u << a2 modulus. An attempt to reconstruct his approach is shown
which permits a to be moved in front of the integral in in Appendix A. However, without knowing exactly the re-
equation (1le). Finally he replaces the indeterminate form sources Hertz used, it cannot be claimed that his method
p/a = ` / ` by an arbitrary constant 4p' /3 where p' is the was the same. One must keep in mind that compendia we
-
force per unit length of the cylinder. The formulas for the take for granted21 23 were not yet printed.
small semiaxis and the stress Z z are under these assump-
tions
NEW ALGORITHM FOR HERTZS CONTACT
STRESS THEORY

General Remarks
To use Hertzs theory in todays environment it is essential
and to describe by a condensed code the set of geometric pa-
rameters of the system to be modelled. This is particularly
important for systematic studies of ordered or random

The compression a would now become logarithmically


o
See Hertz'S introductory remarks to his first paper. A translation is given above
in the section on Basic Assumptions. We found that simulating a cylinder by
high but finite ratios r il /r i z in the regular formulas gives satisfactory approxima-
tions.
n P
Indices i = 1, 2 are left off here, The expressions apply equally to both bodies. A similar statement is found in reference 1, p, 164, footnote 1.

AMP Journal of Technology Vol. 2 November, 1992 E . W . D e e g 19


quadruple of principal radii
Table 3. Factors f and g for calculating semiaxes a and b of
the contact ellipse. The program described in this paper
yields values listed in columns f EWD and gEWD .

The angle ,w the fifth element of the set of geometric pa-


rameters, is included as index.

TO characterize monotone sequences of the rijj let

and

describe the contacting part of the surface of body 1, and

describe the contacting part of the surface of body 2. Each


ordered quadruple now depends on the integers m and
n, i. e.

sequences of the rij. Such sequences occur e.g. during de-


sign optimization or in determining the effect of variations
of processing conditions on shape and size of the contact and the q w ,m,n can be combined in a M x N matrix
zone and on Hertz stress. The methodology is described
below in the section on characterizing combinations of con-
tacting surfaces; examples are found in reference 12.
It is frequently desirable to have the central element of G
It is also essential to streamline the present method for serve as reference describing two spherical surfaces in con-
numerical evaluation of Hertzs formulas using personal tact. The remaining elements q w ,m,m would then describe
computers or workstations and software not requiring con- geometrically deviating combinations. Because of the form
ventional programming. A simple approach would be to in which the rij are expressed in equations (22a) and (22b),
write the table of coefficients f and g and the conventional this can be accomplished by choosing M and N as odd num-
method onto disk. More desirable would be to write com- bers. Depending on the values chosen for the members of
puter programs incorporating algorithms for complete the set {w , rl0 , r20 , D rll, D r12, D r21, D r22, M, N] various
elliptic integrals The most natural approach, however, combinations of ellipsoidal and spherical surfaces can be
would utilize Hertzs fundamental formulas directly and described.
integrate them without transformation to elliptic integrals.
This method and a corresponding computer program are For simulations of random variations of the surface shapes,
described below. the monotone increasing sequences for m and n in
equations (22a) and (22b) are replaced by randomized
Characterization of Combinations of Contacting sequences. The MATLAB program given in Appendix B
Surfaces serves this purpose. It generates four sets of random num-
At the point of contact the surface of each of the two bod- bers mj and nj, enters them into equations (22), computes
ies i (= 1, 2) is defined by two principal radii of curvature rij the rij and plots bar charts of their distribution.
with j (= 1, 2). Thus, regarding their shape the pair of
contacting bodies can be characterized by an ordered Direct Quadrature of Hertzs Basic Expressions
The earliest point in Hertzs derivations where modern
q
computational techniques could be applied without violat-
In connection with the work described here several such programs were written
and tested. References 21 through 24 were consulted and used. Of particular
ing his fundamental concepts is found in reference 1,
help was Hastings polynomial approximation of complete elliptic integrals. 25 pp. 164/65, or in reference 2, pp. 181 and 183. The

20 E.W. Deeg AMP Journal of Technology Vol. 2 November, 1992


corresponding equations above are (14) and (15). The printing of the entire program, the summary shown in
transformation Table 8 is convenient. It fits on an 8.5 x 11 sheet.

Table 4. Data input for the MathCAD program. The lines


separate the three sets of data. Note the changed symbols r ij
changes the integrals in equations (14a), (14b), (15a), and to rij etc.
(15b) to

For 0 < k # 1 these integrals converge absolutely. Their


values can be found e.g. by Simpsons rule or Rombergs Table 5. Definition of functions for converting primary input
modification of the trapezoidal rule.26 Both can be accessed data to intermediate quantities used by the program. The
through one of the MathCAD versions. With these, as well four-member set of material constants {El, E2, vI, v2] is
as other software packages, time-consuming, traditional reduced to the two-member set {01, 02], the five-member
programming can be avoided. MathCAD version 2.5 was set of geometric parameters {rl1, r12, r21, r22, w ] is re-
chosen here because of the transparency of its symbolic duced to the four-member set {R, Rl, R2, W ].
language, its ease of access and the very reasonable hard-
ware requirements.r The symbols used for the variables in
the program had to be adjusted to MathCAD terminology
and differ from those in the text. Both are included in
Table 2.

The program is listed in Tables 4 to 7. It follows essentially


the sequence of formulas above. Because the results of the
root operator are influenced by the seed value and to opti-
mize for speed, three different seed values Kl, K2, and K3
are pre-selected, each for a specific range of the auxiliary
angle W The conditional statements if [. . .] in the pro-
gram identify the proper range and, without operator
assistance, select the proper seed value. The program cal-
Table 6. Functions for computing the ratio k = b/a of the
culates semiaxes a and b of the contact ellipse, area Q of
the contact zone, compression a l and a 2 for each body, semiaxes of the contact ellipse. The lower part of the table
total compression a = a l + a 2, and Hertzian stress Zmax. shows the pre-selected seed values K1, K2, K3 and the
It also plots the circumference of the contact zone. If conditional statements for their optimal selection.
desired, other areas of interest, for instance those of fiber
core and fiber cladding 12, can be plotted in the same dia-
gram. For plots in Cartesian coordinates of the contact
zone the parameter t with 0 # t # 2 p and S : = 75,
s := 0..S, t s := 2 p s/S is introduced in

The value S := 75 gives satisfactory resolution of the plots


but can be reduced to increase speed. To avoid repeated

r
MathCAD 2.5 requires IBM PC, PC/XT, PC/AT or compatibles with 512K
RAM, floppy or hard disks and one of the common graphics adapters. Numeri-
cal coprocessor, although recommended, are not required. It works in MS-DOS
or PC-DOS and supports a wide range of common printers and plotters.

AMP Journal of Technology Vol. 2 November, 1992 E.W. Deeg 21


based. It shows also that for each concrete case the degree
Table 7. Definition of functions for data output. of approximation between the features of the physical sys-
tem to be modelled and the entire set of Hertzs assump-
tions must be compared. Neglecting these considerations
can lead to misinterpretation of theoretical predictions and
to unjustified rejection of the theory. The review of Hertzs
two original publications also reveals expressions that can
be written as computer algorithms requiring only input of
material constants, applied force, and geometric parame-
ters describing the contacting surfaces. The programs make
use of readily available software packages without resorting
to traditional computer programming and permit applica-
tion of the theory in its most general form. Among several
programs developed and tested for this purpose, one was
selected because of the transparency of its terminology,
Table 8. Summary of input, selected intermediate, and out- ease of access, and modest hardware requirements. Writ-
put data. The box marked PLOT contains the contour of the ten in MathCAD version 2.5, the program is described in
contact ellipse in the {x,y} plane according to equation (28). detail.

Optimizing design or studying the effect of variations of


manufacturing conditions on shape, size, deformation, and
stress in the contact zone requires large sequences of
systematically or randomly modified principal radii of cur-
vature of the contacting surfaces. Such sequences can be
characterized and described by a matrix whose elements
are ordered quadruples of the four radii of each individual
combination. For monotone sequences of the principal
radii of curvature, the elements are algebraically related
and by proper selection of their number the central
element of the matrix represents a sphere-on-sphere con-
tact which serves as reference.

REFERENCES
1. H. Hertz, ber die Berhrung fester elastischer
Krper, Gesammelte Werke (P. Lenard, ed.), Bd. 1,
(J.A. Barth, Leipzig, 1895) pp. 155-173. Originally pub-
lished in Journal f. d. reine u. angewandte Mathematik
92,156-171 (1881).

2. H. Hertz, ber die Berhrung fester elastischer Kr-


per und ber die Hrte, Gesammelte Werke
(P. Lenard, ed.), Bd. 1, (J.A. Barth, Leipzig, 1895) pp.
174-196. Originally published in Verhdlg. Ver. Bef.
Gewerbefl., Berlin, Nov. 1882.
SUMMARY
This paper could be subtitled What if Heinrich Hertz 3. A.E.H. Love, A Treatise on the Mathematical Theory of
would have had a personal computer? The high-speed Elasticity, 4th ed. (1927), (Dover, New York, 1944) p.
computing tools we are accustomed to did not exist when 195.
he published his theory about a century ago. To make his
theory accessible for engineering applications, he devel- 4. S.P. Timoshenko and J.N. Goodier, Theory of Elasticity
oped a numerical method with mathematical tools known 2nd ed. (McGraw-Hill, New York, 1951).
to his contemporaries. If Hertz had had access to todays
computing tools, he most likely would not have calculated 5. Htte, Des Ingenieurs Taschenbuch, Bd. I, 28. Auflage
the little table which, in most instances makes the rather (Ernst & Sohn, Berlin, 1955), p. 964.
extensive arithmetic superfluous (reference 1, p. 164).
6. R.C. Juvinall, Engineering Considerations of Stress,
A review of his papers leads to a clear identification of the Strain and Strength (McGraw-Hill, New York, 1967)
complete set of basic assumptions on which his theory is p. 371.

22 E.W. Deeg AMP Journal of Technology Vol. 2 November, 1992


7. R.J. Roark and W.C. Young, Formulas for Stress and 23. J. Hoel, Recueil de fomules et de tables numeriques
Strain, 5th ed. (McGraw-Hill, New York, 1982) Table (Gauthier-Villars, Paris, 1901).
33, p. 518.
24. L.V. King, On the Direct Numerical Calculation of Ellip-
8. W. Griffel, Handbook of Formulas for Stress and Strain tic Functions and Integrals (Cambridge Univ. Press,
(F. Ungar Pub]. Co., New York, 1966) pp. 238-244. Cambridge, 1924).

9. E.A. Kantner and L.D. Hobgood, Hertz stress as an 25. C. Hastings, Jr., Approximations for Digital Computers,
indicator of connector reliability, Connection Tech- (Princeton Univ. Press, Princeton).
nology, 5, 14-22 (March 1989).
26. L.W. Johnson and R.D. Riess, Numerical Analysis, 2nd
10. R. Mroczkowski, Concerning "Hertz Stress as a Con- ed., 323 (Addison-Wesley, Reading, MA, 1982).
nector Design Parameter, Order No. 82818 (AMP Incor-
porated, Harrisburg, PA, 1991).
Emil W. Deeg recently retired from his position as Project
11 H.S. Fluss, Hertzian stress as a predictor of contact Manager, Technology at AMP Incorporated in Harrisburg,
reliability, Connection Technology, 6, 12-21 (Dec. Pennsylvania. He is a consultant for the Company.
1990).
Dr. Deeg holds a physics diploma and a Dr. rer. nat.
12. E.W. Deeg and T. Bolhaar, Contact zone and Hertz- (magna cum laude) from Julius-Maximilians-Universitaet,
ian stress in fiber-optic connections with spherical or with thesis work at the Max-Planck-Institut fuer Silikat-
ellipsoidal fiber endfaces, AMP Journ. of Technol., 2, forschung in Wuerzburg, West Germany. As original con-
29-41 (1992). tributor and throughout his more than 25-year career as
R&D manager and executive for several international cor-
13. E.W. Deeg, Effect of elastic properties of ferrule ma- porations he authored or co-authored more than 70 articles
terials on fiber-optic physical contact (PC) connec- in physics, ceramics, glass science and engineering, and a
tions, AMP Journ. of Technol., 1,25-31 (1991). book on glass in the laboratory. He holds over 40 patents in
the same fields. Dr. Deeg served as a member of the Inter-
14. F. Heerwagen, Kugellager, Erfahrungen aus dem Be- national Commission on Glass (19631981, offices held), as
triebe und Beitrge zur Theorie. Zeitschr, Ver. Dtsch. a member of the International Commission for Optics
Ingen., 45, 1701-1705 (1901). (19641966), and as a consultant to the NASA Spacelab
Program (19711978). He is a fellow of the American Ce-
15. H.L. Whittemore and S.N. Petrenko, Friction and carry ramic Society, was inducted into the Hall of Fame of Engi-
ing capacity of ball and roller bearings, Technol. Paper neering, Science and Technology, and is listed in Whos
Natl. Bureau of Standards No. 201, 18 (Governmt. Who in the Worldj Whos Who in Finance and Industry, and
Printing Office, Washington, DC, 1921). other biographical publications. Dr. Deeg joined AMP in
1984 as Project Manager, Materials Engineering.
16. E.W. Deeg, Unpublished, 1970/76.

17. M.T. Huber, Zur Theorie der Berhrung fester elasti-


scher Krper, Ann. Phys., 14, 153-163 (1904). APPENDIX A
With q being the modulus, the complete elliptic integral of
18. S. Fuchs, Haupspannungstrajektorien bei der the first kind is
Berhrung einer Kugel mit einer Platte, Physikal.
Ztschrft. 14, 1282-1285 (1913).

19. W.B. Morton and L.J. Close, Notes on Hertzs theory


of the contact of elastic bodies, Phil. Msg. (Ser. 6), 43,
320-329 (1922). Setting q = 1 k 2 in equations (25) and (26) but q = 1
1 /k2 in (27), the denominator of integrals H(k), I(k) and
20. Ph. Frank and R. v. Mises, Die Differentialgleichungen J(k) is
und Integralgleichungen der Mechanik und Physik, Bd. 1,
pp. 608-611, 2nd. ed. (M. S. Rosenberg, New York,
1943). (Photolithographic copy of the original edition
published by F. Vieweg & Sohn, Braunschweig, 1930).
and the integrals in equations (26) and (27) assume the
21. E. Jahnke and F. Erode, Tables of Functions, 4th ed. form
(Dover, New York, 1945).

22. M. Abramowitz and I.A. Stegun, cd., Handbook of


Mathematical Functions (Dover, New York, 1972).

AMP Journal of Technology Vol. 2 November, 1992 E.W. Deeg 23


H(k) is already of the form K(q). Equation (A2) in (Al) APPENDIX B
yields MATLAB program for generating random distributions of
radii of curvatures r ij.
,
Ml= input(upper limit of m1 Ml = );
N1 =input(upper limit of n1 N1 = ,);
,
M2=input(upper limit of m2 M2 =, );
N2= input(upper limit of n2 N2 = );
,
xl =input(seed for m1-sequence xl = );
,
yl =input(seed for n1-sequence yl = );
,
x2=input(seed for m2-sequence x2 = ,);
and the first derivative of K(q) after the modulus is
y2= input(seed for n2-sequence y2 = );
,
rl0=input(base radius, endface #1 r10 = ,);
r20=input(base radius, endface #2 r20 = ,);
dr11 = input(r11 step size drll = );
,
dr12=input(r12 step size dr12 =, );
Comparing equations (A4) and (A5) yields the functional dr21 =input(r21 step size dr21 =,);
relationships for I(k) and J(k) as stated by Hertz: dr22=input(r22 step size dr22 = );
rand(uniform);
rand(seed,xl) ;ml= 1 +round(l0*rand( l, Ml));
rand(seed,yl);nl = 1 +round(l0*rand( l, Nl));
rand(seed,x2) ;m2=l+round(10* rand(l,M2));
rand(seed,y2) ;n2=l+round(10* rand(l,N2));
rll=rl0+(ml -( Ml+l)/2)*drl l,pause,
r12=r10+(m2-(M2+l)/2)*dr12,pause,
r21=r20+(nl-(Nl+l)/2)*dr21,pause,
r22=r20+ (n2-(N2+ l)/2)*dr22,pause,
subplot,
subplot(221),bar( rll),subplot(222),bar( r12),
subplot(223),bar( r21),subplot(224),bar(r22),pause

24 E.W. Deeg AMP Journal of Technology Vol. 2 November, 1992

Anda mungkin juga menyukai