School of Engineering
Civil Engineering
by
Stephen Grech B.E.& A (Hons)
2006
Acknowledgements
ACKNOWLEDGMENTS
I would like to express my gratitude to my supervisor, Prof. G.A.R. Parke, B.Sc.,
M.Sc., Ph.D., EurIng, CEng, MICE, FIStrucE, Professor of Structural Engineering,
Head of Civil Engineering, University of Surrey, who provided constructive
suggestions throughout the course of this research and showed support in my
work. I appreciate the time he dedicated.
I would also like to thank Prof. Alex Torpiano, BE&A., M.Sc., Ph.D., DIC, Eur.Ing.,
MIStruct.E., A.&C.E., Faculty of Architecture, University of Malta, for his advice and
for the interest he showed in our research.
Most of all I would like to thank my wife Carmen, my family and colleagues at
work. I am indebted to them for their moral support and encouragement
throughout my studies.
University of Surrey
Table of Contents
TABLE OF CONTENTS
LIST OF TABLES.......................................................................................................iv
LIST OF FIGURES.....................................................................................................iv
GLOSSARY OF TERMS .......................................................................................... viii
SYNOPSIS.................................................................................................................xi
chapter 1
INTRODUCTION................................................................................. 1
chapter 2
2.1.
Introduction ..........................................................................................................5
2.2.
2.3.
2.4.
chapter 3
3.1.
Introduction ........................................................................................................18
3.2.
3.2.1.
3.2.2.
3.3.
3.4.
3.5.
3.6.
Loading Parameters...........................................................................................29
3.7.
3.8.
chapter 4
RESULTS.......................................................................................... 35
4.1.
4.2.
4.3.
4.4.
4.5.
4.6.
4.7.
University of Surrey
ii
Table of Contents
4.7.1.
4.7.2.
4.7.3.
4.7.4.
4.8.
4.9.
chapter 5
CONCLUSIONS ............................................................................... 70
REFERENCES......................................................................................................... 76
BIBLIOGRAPHY ...................................................................................................... 79
appendix A .............................................................................................................. 81
appendix B............................................................................................................ 106
appendix C............................................................................................................ 133
appendix D............................................................................................................ 156
University of Surrey
iii
LIST OF TABLES
Table 4-1: Permutations used for finite element models....................................................46
Table 4-2: Comparison of slab capacity prediction using alternate theorems ..................53
LIST OF FIGURES
Figure 1.1: Typical application of transfer slab system at basement level ..........................2
Figure 1.2: Irregularities in load paths displayed in basement layout of Figure 1.1.............3
Figure 2.1: Typical yield lne pattern for a three-way spanning slab.....................................6
Figure 2.2: Nodal force distribution for three-way spanning slab (extracted from
Jones et. al., 1967) ...............................................................................................................8
Figure 2.3: Hillerborg strip apportionment for typical slab (extracted from Martin
et. al., 1989) ..........................................................................................................................9
Figure 2.4: Stipulated line load distribution width (extracted from Fig 3.6. as per
BS8110-1:1997)..................................................................................................................10
Figure 2.5: Westergaard loading scheme (extracted from Park et. al, 2000).....................12
Figure 2.6: Comparison of thin plate theory / elastic analysis / testing results
(Fenwick et. al., 1989) ........................................................................................................13
Figure 2.7: Loading / deflection curve for restrained reinforced concrete slab .................15
University of Surrey
iv
University of Surrey
Figure 4.1: Yield line formation adopted for analysis (Kennedy, 2003) .............................35
Figure 4.2: Varying mesh dimensions utilised for convergence test, with grid
spacing:..............................................................................................................................40
Figure 4.3: Finite element model convergence test ...........................................................41
Figure 4.4: Deflections of full slab finite element model.....................................................42
Figure 4.5: Deflections of quarter slab finite element model..............................................42
Figure 4.6: Principal stresses of full scale finite element model ........................................43
Figure 4.7: Principal stresses of quarter slab finite element model ...................................43
Figure 4.8: Strain distribution across trial section slab depth ............................................44
Figure 4.9: Stress distribution across trial section slab depth ...........................................45
Figure 4.10: Slab deflection with increased load increments for varying slab
reinforcement......................................................................................................................47
Figure 4.11: Plan - Cracking below wall line load ..............................................................48
Figure 4.12: Location of all regions experiencing cracking/crushing within slab
depth ..................................................................................................................................49
Figure 4.13: Plan - Development of concrete crushing......................................................49
Figure 4.14: Slab B torsional stress SXY at failure load (N/mm2) ........................................50
Figure 4.15: Displaced shape of slab B at failure ..............................................................51
Figure 4.16: Strain distribution across the depth of slab B following concrete
cracking ..............................................................................................................................52
University of Surrey
vi
Figure 4.17: Cartesian co-ordinate system and critical sections reported ........................54
Figure 4.18: Distribution of SY stresses along X-axis length at load factor 0.20 ................55
Figure 4.19: Main SY stresses for model A load increment 1.00 ........................................56
Figure 4.20: Transverse SX stresses for model A load increment 1.00 ..............................56
Figure 4.21: Main EY strains for model A load increment 1.00 ...........................................57
Figure 4.22: Main EY strains for model B for load increment 0.90.....................................58
Figure 4.23: Transverse EX strains in model B for load increment 0.90 .............................59
Figure 4.24: Distribution of SX stresses along X-axis length at load factor 0.90 ................60
Figure 4.25: Distribution of SY stresses along X-axis length at load factor 1.00 ................61
Figure 4.26: Distribution of SY stresses along Y-axis length at load factor 1.00 ................61
Figure 4.27: Main SY stresses in model C for load increment 0.90 ....................................62
Figure 4.28: Main EY strains in model C for load increment 1.00.......................................63
Figure 4.29: Main EY strains in model C for load increment 0.80.......................................64
Figure 4.30: Distribution of SY stresses along X-axis length at load factor 1.00 ................65
Figure 4.31: Reaction at slab support at failure .................................................................66
Figure 4.32: Tensile stresses in steel at failure along slab width .......................................67
Figure 4.33: Compressive stresses in top of concrete at failure along slab width ............68
University of Surrey
vii
Glossary of Terms
GLOSSARY OF TERMS
The following notation is used in this dissertation. Any other symbols not listed
below are defined in the main text. As a general rule, only one meaning is
attributed to each symbol, but in those cases where more than one meaning is
possible, then the correct one will be explained in the context in which it is used.
Beff
Di
Ee
EX
EY
fc
fcu
ft
fy
University of Surrey
viii
Glossary of Terms
SX
SY
SXY
plate thickness
tw
Poissons ratio
wudl
jw
University of Surrey
ix
Glossary of Terms
jx
jy
sn
t0
cp
University of Surrey
Synopsis
SYNOPSIS
Concrete slab design methods have been established based on uniformly
distributed loads, where most analysis methods and code requirements are
concerned with changes in slab geometry and support fixity. Current local
construction trends impose construction practices involving requisite of
minimisation of site works. Part of this scenario involves the reduction of shuttering
works and use of flat transfer slabs, to withstand overlying loadbearing wall line
loads, instead of adopting the use of a grillage of beams. Therefore, the manner in
which the concrete slabs are being utilised differs to the normal analysis models,
where distribution of the line loads becomes a critical aspect. Although
assimilations with flat slab behaviour can be made through resemblance to plate
action, the analytical model is complicated due to varying slab stiffness and due
to material non-linearity.
This study aims to carry out a review of slab analysis methods and code
provisions with relation to the application of line loads on slabs. An estimation of
the degree of slab distribution and the slab ultimate load capacity shall be
considered. A typical line load case for a one-way spanning slab with a given slab
span and thickness shall be used for all analysis methods. All methods and codes
shall be compared to a constructed finite element model of a typical transfer slab
with varying quantity of bottom reinforcement. Deficiencies of each analysis
method will be considered and attributes of the model variations shall be
examined in relation to the mode of failure.
University of Surrey
xi
Introduction
chapter 1 INTRODUCTION
The successful role of the structural engineer, to describe a physical reality in the
form of analytical models which result in a design solution, will depend greatly on
the complexity of the particular problem at hand and the accuracy of the guiding
principles. Each construction industry encounters particularities that evolve due to
the influencing factors involving material and labour resources. This has lead to
the consideration of this study through the authors role as a structural engineer
within the Maltese construction industry, where uncertainties involving the correct
adoption of slab design methods to current design issues have arisen in the local
construction industry.
The current construction norm, in the local residential building industry, involves
an architectural solution which amalgamates lower basement levels incorporating
parking facilities with cellular residential upper floor configurations up to heights of
ten floors. It is normal construction practice to adopt load-bearing masonry /
concrete slab construction for the repetitive residential levels, and either a
concrete frame solution for the parking levels or a load-bearing masonry
configuration based on the garage layout. The favourable use of load-bearing
construction in all levels results from the extrapolation of a design philosophy
usually adopted for low-rise residential construction. Clients and architects choose
to adopt the load-bearing construction since local contractors are familiar with this
system and it is still economically viable to adopt the same constructional solution
for medium rise buildings when compared to the use of a continuous concrete
frame.
From a structural perspective, although this form of load-bearing cellular
construction provides a very stiff composition in the upper floors, the nature of the
heavy loads produces critical load-path problems in the lower floors, where the
University of Surrey
Introduction
configuration of the plan layout is suddenly and radically modified. The design of
this horizontal interface, involving a transfer of the load path, is the critical aspect
which this study is attempting to analyse. Particularities in the local construction
industry have led to the development of the use of a concrete slab of uniform
thickness to transmit these intense line loads instead of a preferred grillage of
beams which directly transmits the line loads onto the supports. These
particularities involves the lack of skilled construction labour on the island, which
leads to the necessity to reduce shuttering and beam reinforcement works, and
the critical limitation of overall structural depth due to significant building height
restrictions imposed by local building authorities.
Introduction
line loads and the resulting resisting slab width which can be adopted for
simplified design purposes. Standard design methods usually deal only to a
limited extent with the application of line or point loads, and the reasoning
regarding the load distribution is often unclear. In addition, it is apparent that very
little development in the field of slab design has occurred in recent decades,
leading to the perhaps improper application of beam theory for design of slabs
which does not take into account the actual nature of the deformed slab
geometry.
Figure 1.2: Irregularities in load paths displayed in basement layout of Figure 1.1
(superimposed layout of ground floor plan shown blue)
The goal of this study is therefore to take a broad overview of recent research,
current slab theories and standard codes of practice with relation to the actual
slab mechanism and the application of these intense line loads on one-way slabs.
The actual stress state in the slab will be obtained by the application of analytical
models involving the use of a finite element model. The manner in which to
accurately model the finite element analysis shall be discussed with reference to
past modelling techniques, and a simplified finite element model shall be
University of Surrey
Introduction
proposed. Since the aim is to obtain an accurate picture of the slab mechanism,
material non-linearities shall be taken into consideration.
This study will deal primarily with the description of these analytical techniques for
a standard 6m x 12m one-way slab module with a single line load positioned
centrally and parallel to the spanning direction. Although typical slab loading
cases involve an assorted configuration of line loads, as displayed in Figure 1.2, a
single central line load has been chosen in order to obtain a clear understanding
of the basic loading case mechanism. Parameters of slab thickness, end
restraints and longitudinal / transverse stiffness ratio shall be kept constant with
the adoption of an isotropic pattern of reinforcement, which involves the use of an
equal area of steel in both directions. The major factor which is varied in this study
involves the quantity of reinforcement used within the slab, and this has been
chosen in order to gauge the influence of under / over reinforcement on the failure
mechanism. Consequently, this study shall also look into the actual mechanism
that occurs up to failure and how load distribution is effected as the slab deforms.
In addition, the stage at which failure is deemed to occur shall be considered and
compared with the possibility of continued slab deformation and the formation of
yield line propagation. The values of ultimate load obtained from standard slab
theory shall then be compared to the values obtained from a proposed finite
element model analysis.
A parallel study involving the influence of design variations on the slab load
distribution is also currently in progress in order to produce a complete
amalgamated research which looks into the effect of the varying parameters. This
parallel study looks into the utilisation of elastic grillage analyses to carry out a
greater number of permutations. Therefore the comparison of a similar isotropic
model shall be carried out in order to calibrate the results of the two studies.
University of Surrey
Literature Review
Introduction
The concrete slab has presented engineers with difficulties involving how to
accurately and appropriately carry out analysis in order to produce a simple and
rational design method. The curvature deflection pattern of a concrete slab is
associated with plate deformation involving an infinite number of indeterminate
strips with different curvatures. However, unlike steel plates, concrete slabs do not
behave elastically throughout the loading curve since materials act in a non-linear
manner and end restraints and reinforcement ratios have been found to affect the
collapse behaviour. In addition, steel reinforcement is usually positioned in an
orthogonal pattern which does not reflect the direction of principal stresses.
A century of research in slab design has attempted to produce design solutions
which take these factors into account in a relatively simple manner for frequent
use and also give safe, yet not too conservative, results. Consequently, due to the
complexity of this task, most research carried out dealt with uniform distributed
loading. Although much work was carried out, especially in the middle of the 20th
century, the slab analysis problem still remains as an unresolved debate in light of
the various design methods which were produced by the pioneers of slab design
theory. Although attempts were made to describe the action of slabs under
concentrated loading, mainly in the field of bridge deck design, it is therefore
relevant that we look at the manner in which these classical slab design theories
can be adopted for this particular study involving intense line loads.
University of Surrey
Literature Review
2.2.
The initial developments which led to the modern state of slab design was the
concept of the use of plastic design for the analysis of slabs, pioneered by
Professor K. W. Johansen, where the main method developed involved the yield
line theory (Johansen et. al., 1943). The concept of the theorem lies in the
assumption that the failure patterns form according to yield lines which
correspond to linear positions, where reinforcement is predicted to yield.
Consequently a failed section of the slab would consist of rigid elastic portions of
slabs rotating about plastic hinges which involve a finite plastic moment capacity
of the given section, thus ignoring any elastic deformations prior to failure. A
typical yield line pattern for a three-way spanning slab is displayed in Figure 2.1
below.
Figure 2.1: Typical yield lne pattern for a three-way spanning slab
(extracted from Jones et. al., 1967)
The first method of analysis developed was based on an energy or Work method
for calculating the failure load. The actual failure load of a slab, with given loading,
thickness and reinforcement, is estimated for different configurations of possible
yield line patterns which involve collapse mechanisms that are compatible with the
boundary conditions. A virtual work equation is set up for each different yield line
University of Surrey
Literature Review
configuration and the corresponding failure load is solved by taking the moment
capacity of the section at the yield line.
The correct yield line formation corresponds with the lowest value of the failure
load which relates to the failure mechanism that requires least work.
Consequently, the designer must be certain that all possible failure mechanisms
have been analysed and checked since this analytical technique is an upper
bound technique where failure loads obtained are equal or greater than the actual
failure load. Thus, the drawback of utilisation of this procedure involves the
necessity to be thorough in evaluation of all possible failure mechanisms, since it
may result in an unsafe over-estimation of the load capacity.
In addition, an early setback for this technique involved the realisation that
particular yield line geometries resulted in regions which are not always in overall
equilibrium when analysing moments along the yield lines to uphold the regions.
In 1922, Ingerselv presented a paper in which he derived the collapse modes for
slabs based on the assumption that regions were maintained in equilibrium by
assuming moments about the yield lines (Ingerselv, 1922). Whilst the method
provided correct solutions for symmetrical two-way slabs, actual testing on more
complicated asymmetrical cases had shown that this method produces false
results and this consequently led to the much debated question of what then is
the actual state of stress in the yield line. The yield line analysis technique however
never developed in a manner which answered this question due to Johansens
development of the concept of nodal forces involving the application of forces
acting at nodes to maintain equilibrium. These nodal forces varied in magnitude
and direction and were statically equivalent to the unknown twisting moments and
shears. They provide a useful tool by presenting information regarding which way
the layout should be adjusted in order to obtain the critical layout of a particular
mode.
University of Surrey
Literature Review
Figure 2.2: Nodal force distribution for three-way spanning slab (extracted from Jones et. al., 1967)
The yield line technique has provided a useful manner in which to obtain the
actual failure load for a given slab and correlates well with actual collapse test
results, as shall be discussed later. In some cases, the failure load given provides
a conservative result when compared to actual testing and this is considered to
be the result of inherent membrane actions occurring in the slab during failure.
This aspect is discussed in further detail below, but is not a shortcoming in the
yield line analysis since these forces are not considered when carrying out yield
line analysis. Its advantage is that it can be applied to all types of geometrical and
loading complications and shall therefore be applied to the intense line load case.
In the absence of actual physical testing in this study it shall be used to provide an
estimate of the slab failure load, as displayed in chapter 4, without reference to
the serviceability deflection requirements that are not critical to this study given the
substantial slab depths adopted.
Due to the previously stated complicated implications of an upper bound
technique involving endless permutations of collapse formations an intensive
evaluation process commenced in the 1950s which involved acquiring the
coincidental upper / lower bound solutions to obtain the true collapse load named
the exact solution. The most widely acknowledged methods for upper / lower
methods were consequently compared to find the exact solution for various
cases. However, despite the internationally sustained effort, these proved very
University of Surrey
Literature Review
difficult to establish and in the years that followed the engineering community
eventually leaned increasingly on the use of the Hillerborg strip method for slab
design. This method involves a design tool which provides a safe lower bound
method and does not require an iterative procedure, where reinforcement is
calculated from the loading and shape parameters (Hillerborg, 1996).
Figure 2.3: Hillerborg strip apportionment for typical slab (extracted from Martin et. al., 1989)
The Hillerborg strip method works on the principle of apportioning the loads
between sets of strips and consequently reducing the analysis of an infinitely
indeterminate plate into a series of statically determinate beams, as shown in
Figure 2.3. The strips are designed under the assumption that they act as oneway spanning slabs and reinforcement is calculated accordingly using beam
theory. Consequently, in this manner any apportionment scheme chosen by the
designer will satisfy the lower bound theorem since the structure is in equilibrium
throughout and all external loads are balanced, whilst at the same time the yield
University of Surrey
Literature Review
condition is not violated. The advantage of the strip method is that, contrary to the
yield line theory, a solution involving non-uniform reinforcement distribution is
possible and designed. Consequently, even though the lower bound theorem is
more conservative with regard to the slab capacity, compared to upper bound
solutions, it usually provides a more efficient solution due to possibility of
curtailment.
Figure 2.4: Stipulated line load distribution width (extracted from Fig 3.6. as per BS8110-1:1997)
The critical problem for the use of the strip method is to what degree a designer
should apportion loads onto strips and the actual widths of strips that are to be
assumed to resist the apportioned load. Whilst this problem is not critical for a
uniformly distributed load, it does create an anomaly for concentrated loading.
Since moment - curvature relationships are not used, the designer must take into
consideration a finite width of slab resisting the load in one-way action instead of
considering how the slab actually acts as a continuous plate to resist the load.
Text books and codes of practice provide ambiguous figures for distribution width
normally based on empirical prescriptions or on the designers judgement and
general experience.
University of Surrey
10
Literature Review
Solutions for intense line loading cases using the lower and upper bound
techniques are detailed and solved in chapter 4 and have been compared for
evaluation of slab capacities. Whilst the strip method provides a solution based on
an effective slab strip width, as per codes of practice, the yield line mechanism is
assumed to produce the actual collapse load based on the critical failure
mechanism and taking the whole of the slab geometry into consideration.
Comparison with a finite element model, providing the predicted yielding
sequence and loading distribution pattern, should give an insight into the
conservative degree of code BS8110-1:1997, based on the strip method, and the
ultimate failure load predicted by yield line analysis.
The design techniques discussed thus far form the basis of modern slab design
theory adopted in practice, which have been calibrated mainly in relation to
uniform loading cases. In contrast, there has been a limited quantity of work
carried out specifically on concentrated point and line loads which has made use
of an alternative analytical model involving thin plate theory, as successfully
developed by Timoshenko for steel plate design (Timoshenko et. al., 1959). This is
possible for application in concrete slab analysis due to similarities made in the
geometrical properties and elastic deflection characteristics, which therefore
provides information relating to the way the plate distributes load under
concentrated loading.
The most common field for this type of analysis has been developed in relation to
bridge deck analysis where attention has been given to analysis of wheel line
loading. The most important contribution was given in 1930 by Westergaard who
obtained expressions for distributive widths and moments mx and my resulting
from the application of point loads over a finite circular area on one-way and twoway slabs through an elastic analytical investigation (Westergaard, 1930). This
field of study however was not extended to line loading cases and predicting
values would have to be made on the basis of the dubious assumption of
superimposition of distribution widths and analysed moments.
University of Surrey
11
Literature Review
Figure 2.5: Westergaard loading scheme (extracted from Park et. al, 2000)
Specific studies that dealt with line load cases have been carried out by the BOCA
National Building Code (BOCA, 1996) and in publications by Woodring
(Woodring, 1963). The philosophy reached in their conclusions, following an
elastic analytical approach also, is to treat the line load case as an equivalent
uniform distributed load that gives the same magnitude of maximum moments
and shear, thus ignoring any effects of torsional considerations. The studies
however only considered line loads of a limited magnitude that were comparable
with the superimposed floor loading and therefore large expected variations in the
distribution pattern, with the intense line loads considered in this study, were not
encountered.
As outlined in the introductory section, the behaviour of reinforced concrete plates
is such that the application of elastic thin plate theory also produces unrealistic
results outside of the initial slab elastic range due to the capacity of sections to
further distribute loads after the commencement of cracking and even steel
yielding. This behaviour has been reported by Fenwick and Dickinson where
concentrated load tests were made and compared to thin plate theory (Fenwick
et. al., 1989). The results involving distribution of bending moments are as shown
in Figure 2.6, whilst estimates involving cracked section properties obtained
involving accompanying beam specimens were favourably compared, where a
difference in the results was attributed to the internal membrane forces. The
University of Surrey
12
Literature Review
description of boundary conditions of thin plates does however form the basis of
finite element analysis following years of development by mathematicians in this
field. For this study a finite element model was constructed and analysed using
the program Lusas, Version 13.6-2, where the parameters of the utilised model are
described in Chapter 3.
Figure 2.6: Comparison of thin plate theory / elastic analysis / testing results (Fenwick et. al., 1989)
2.3.
Physical Testing
By comparison with the actual physical testing carried out involving uniform load
application on one-way or two-way slabs, it is apparent that in general results
correlate very well with the yield line technique in relation to the obtained collapse
load analysis and failure pattern. This has been found as a common trend carried
out throughout the years where pioneering testing was carried out by the German
Reinforced Concrete Board, under the direction of C. Bach and O. Graf during the
period 1911-1925 (Park et. al., 2000). Results from this work indicated that there
was very good correlation in patterns and magnitude for various slab types and
have been reported by Johansen, where first yielding of the steel was at
approximately two-thirds of the ultimate load when distributed loading was applied
(Johansen, 1943).
Similar results were found following testing carried out in 1962 by the French
delegation to the European Concrete Committee, IRABA (Institute of Applied
University of Surrey
13
Literature Review
2.4.
The analytical models used to predict the slab behaviour discussed so far do not
take into account considerations involving the reserve that can be developed in
the slab through membrane action. Although a number of research studies on
membrane action in reinforced concrete slabs have been conducted, only
approximate ultimate strength theories have been developed. Consequently,
although it is evident that membrane action significantly increases the ultimate
load capacity, it proved to be difficult to incorporate the possible strength
enhancement within slab design applications.
University of Surrey
14
Literature Review
Figure 2.7: Loading / deflection curve for restrained reinforced concrete slab
(extracted from Park et. al, 2000)
As loading is increased the slab will deflect in a non-linear fashion since the
section properties will vary and transform from an uncracked state to a cracked
state. Point B on the curve represents the ultimate failure load, as is calculated by
yield line theory, where the steel has commenced plastic deformation. An outward
movement of the slab at its end supports is required in order for the slab to deflect
as the load progresses and consequently if the ends are restrained then
compressive arching action in the concrete will result. This stage is represented in
the loading curve between BC where a rapid deformation in the slab occurs
without increase in load until a full crack develops throughout the slab section
when the deflection of the slab is approximately equal to the slab depth at point C.
University of Surrey
15
Literature Review
From this point forward the slab acts as a catenary in plastic tensile membrane
action where the steel catches the slab from falling, continuing to support loads
that can be larger than the ultimate load, until the steel fractures at point D. It is
therefore essential that the steel extends across the whole length of the span and
is fully restrained at ends in order for tensile membrane action to occur. In
addition, whilst the pronounced occurrence of compressive membrane action
depends on the slab not being over-reinforced, tensile membrane range is more
extensive for heavily reinforced slabs due to the dependency on steel yielding.
Extensive testing by various sources has been carried out involving restrained
slabs under uniform loading as reported by Park, all of which confirm the
membrane action as described above (Park et. al., 2000). They also report
increases to the final collapse load by factors in the range of twice the ultimate
University of Surrey
16
Literature Review
failure load where initial yield line cracks are formed as predicted by yield line
analysis. Mathematical models have also been formulated by Park for both
compressive and tensile membrane actions where membrane forces are
estimated through use of modified concrete stress blocks and plastic tensile
action in both stages, as displayed in Figure 2.8 and Figure 2.9 respectively.
This behaviour has also been reported to occur in slabs with concentrated loads
as reported in tests carried out by Taylor, Aoki et. al. and Batchelor et.al., and is
therefore relevant to this study (Park et. al., 2000). In addition, this behaviour has
also been reported for slabs with edges that are free to move laterally, where outer
concrete regions restrain the slab from moving inwards at relatively large
deflections, resulting in an outer compression ring supporting tensile membrane
forces in the inner central region of the slab as displayed in Figure 2.10.
.
Figure 2.10: Membrane action development in unrestrained simply supported slabs
(extracted from Park et. al, 2000)
University of Surrey
17
Introduction
University of Surrey
18
Consequently, part of the study involves the research of past work involving both
linear and non-linear analysis and the parameters that need to be studied to
accurately model the slab.
3.2.
Previous Research
The following review details the progression from the conception of finite element,
based on flat plate design to current trends. Much of the work described below
deals primarily with the development of modelling and analysis techniques utilised
for reinforced concrete sections.
where,
University of Surrey
Et 3
12(1 v 2 )
Eqn 3.1
19
University of Surrey
20
University of Surrey
21
For nonlinear analysis the external loads cannot be directly related to the stress
distribution and a solution procedure is usually adopted in which the total load is
applied in a number of increments. Within each increment a linear prediction of
the nonlinear response is made. Subsequent iterative corrections are required in
order to restore equilibrium by the elimination of the residual or out of balance
forces. The iterative process ceases once the chosen convergence criteria and
required equilibrium state is achieved. A solution procedure which is commonly
adopted is the Newton-Raphson method. This involves an initial prediction of the
incremental solution based on the tangent stiffness from which incremental
displacements are derived. The manner in which the iterative corrections continue
to be derived varies according to the parameters of the analysis being carried out.
University of Surrey
22
A vital aspect of the modelling regime adopted involves the representation of the
reinforcement within the slab. One of the common approaches is the layered
model. An early example of its application involves the work of Lewinski and
Wojewodzki who demonstrated application of a nonlinear slab model combining
the modified stiffness approach and the layered model approach (Lewinski et. al.,
1991). Nonlinear concrete behaviour, such as cracking, and elastic-plastic steel
reinforcement deformation were also incorporated in addition to coupling between
extension and flexure. Several test analyses were conducted and compared with
experimental test data which showed that the proposed method is accurate for
evaluation of in-plane and out-of-plane effects in slabs.
Figure 3.3: Layered model of reinforced concrete slabs (extracted from Phuvoravan et. al., 2005)
University of Surrey
23
The layered approach to modelling involves dividing the concrete into a set of
layers, while the reinforcing steel is smeared into a layer between concrete layers.
The reinforcement layer has stiffness only in the direction of the reinforcement. In
this manner, the variation of stress along the slab thickness can be modelled.
Furthermore, since the three dimensional slab is simulated by a set of twodimensional layers, only a two-dimensional stress-strain relationship is required.
Layered modelling of RC slabs is simple, but provides an unrealistic
representation of the reinforcing steel and since real reinforcement is discrete,
only highly reinforced slabs can be appropriately modelled by the layered
approach.
The alternate common practice to modelling of reinforcement is the discrete
model (Phuvoravan et. al., 2005) where the concrete is modelled by three
dimensional solid elements while the reinforcing steel is modelled by truss
elements. The connectivity between a concrete node and a reinforcing steel node
is then achieved by either full bonding for the perfect bond case, or through a
spring element to simulate slip-bond relationships. Whilst the discrete method
provides more accurate results, it is however a more laborious procedure.
Figure 3.4: Discrete model of reinforced concrete slabs (extracted from Phuvoravan et. al., 2005)
24
dimensional solid elements (Barzegar et. al., 1994). Instead of either a smeared or
layered approach, a model for reinforcement independent of the finite element
nodal coordinates was presented by which an entire reinforcement cage is
automatically mapped into a mesh of solid concrete elements.
In addition, Phuvoravan and Sotelino derived a new finite element for nonlinear
analysis of reinforced concrete slab systems (Phuvoravan et. al., 2005). The new
element was a combination of the classic four-node Kirchoff shell element with
two-node Euler beam elements to simulate the steel reinforcement. Rigid links
were applied to couple the reinforcement to the concrete. Prior applications of
discrete reinforcement modelling incorporated reinforcement in a mesh of threedimensional solid elements. Replacing the three-dimensional mesh with a twodimensional mesh of modified Kirchoff elements allowed for a simple and efficient
manner of representing the system per element. The presented element shows a
better prediction of reinforced concrete flat slab behaviour than the general
layered shell element, and was verified with experimental data.
3.3.
The analytical package adopted for the non-linear analysis involves Lusas Version
13.6-2, where the selection criteria involved accessibility to the programme and
familiarity with its utilisation. Taking into account the above review of past works
and also the limitations associated with Lusas as a finite element modelling
programme for reinforced concrete, the following parameters were adopted.
3.4.
Geometrical Properties
The chosen dimensions of slab span and thickness used for the analytical model
were taken based on typical cases in the construction industry as experienced
locally and described in chapter 1. An aspect ratio of 1:2 was then chosen so as
University of Surrey
25
to enable the opportunity of studying the possibility of yield line propagation whilst
allowing for a load distribution pattern on a relatively large width of slab.
The analytical case to be considered involves a rectangular simply supported slab
with a central line load perpendicular to the longitudinal support edge,
representing a symmetrical case in geometry and loading. Therefore it is possible
to simplify the model by considering a quadrant of the whole slab and this has
been verified through tested preliminary cases, comparing the quadrant with the
full scale slab. It can be confirmed that no variations existed and that the chosen
boundary conditions were validated. The results of this positive preliminary
calibrating test are found in chapter 4. This step will allow an advantageous
analytical process since the iteration processes will be shortened whilst enabling
the use of a finer mesh. In addition, it will allow for the improved visualisation of
stresses through the thickness of the slab at longitudinal and transverse midspan. Figure 3.5 displays the line element spacing adopted, where a bisecting
mesh division between these elements was then selected.
University of Surrey
26
The chosen configuration was taken after testing the convergence of the selected
mesh spacing as is later described in chapter 4. The utilised geometry was found
to be within acceptable tolerance limits after carrying out an elastic finite element
model analysis and correlating these with exact elastic beam theory solutions.
3.5.
Supports Parameters
Since a quadrant of the whole slab was adopted it is essential that the action of a
bending element at mid-span be properly simulated, where bending stresses
should develop as would occur for a whole slab model. Consequently,
longitudinal and transverse mid-spans were fixed against rotations and horizontal
displacements in their respective directions. These parameters were proved to
correctly model the mid-span in the preliminary calibration models as outlined in
chapter 4.
University of Surrey
27
Supports along the slab edge simulate wall supports which allow for slippage of
the slab underside. Consequently, roller supports were placed along the outer line
edge and were utilised in order to provide vertical support only and allow for
lateral movement and rotation. This set-up ensured that no secondary effects
resulted from in-plane membrane stresses. It is also assumed that supports do
not displace vertically, as would occur in the case of beam supports. In addition,
the roller supports are modelled so as to contain both positive and negative
vertical reactions and consequently any tendency of slab uplift would be
restrained. Preliminary analysis shows that this occurrence is not a major influence
on the slab failure and therefore no attempt was made to correct support
characteristics to allow for slab uplift.
University of Surrey
28
3.6.
Loading Parameters
Two types of superimposed loads were defined in the finite element model and
have been placed on the model upper surface as uniformly distributed loads. The
slab self-weight was superimposed on the model taking a concrete slab density of
24 kN/m2. The line loading is then described through a representation of a central
wall line load applied as a uniform distributed load over a limited width equivalent
to the thickness of a typical 250mm thick load bearing masonry wall. The load
magnitude is varied accordingly in order to load the slab to failure and compared
with values as predicted in slab theory and stipulated codes of practice. Through
preliminary analysis, the critical load at which slab failure and model nonconvergence resulted was then used as the maximum load applied. A total of
twenty load increments were then used throughout the loading sequence, with
load increases of 5% carried out at each iterative step. The use of these small
increments was adopted in order to assist the model to converge in regions where
effect of material non-linearity sets in. In addition, a clearer picture of the slab
deformity resulting from load increase was also apparent from the analysis. The
relative load magnitudes attained are discussed in chapter 4.
3.7.
Figure 3.8: Non-linear scheme at element level slabs (extracted from Phuvoravan et. al., 2005)
University of Surrey
29
The use of non-linear analysis in the reinforced concrete slab basically refers
directly to the non-linear behaviour of the materials. Since the slab remains
geometrically stable and displays only small deflections due to the use of a thick
slab and small span-depth ratios, geometrical non-linearity is not an issue. As a
result, the Lagrangian integration scheme is adopted as a Lusas setting due to
the assumption of geometrical linearity. The loading is consequently taken to be
acting vertically throughout the loading sequence and not perpendicular to the
deforming plane of bending.
Lusas material parameters allows for concrete modelling through an option which
enables the portrayal of cracking / crushing propagation and through description
of the concrete stress-strain curves in both the compressive and tensile ranges as
shown below. The cracking / crushing tool will allow for the visualisation of
propagation of concrete failure in the model, where a distributed fracture pattern
will be permitted by setting a null value for the fracture energy capacity in the
material setting.
Figure 3.9: Concrete tensile / compressive stress-strain curve (extracted from Lusas Manual)
Since the study has been carried out on a purely analytical level, stress-strain
characteristics have
30
until the concrete cracks at a peak tensile capacity assumed to be 15% of the
compressive strength. The tensile stress-strain curve then develops into a
softening curve with an assumed limiting tensile strain (t0) of 0.004. The
compressive section is modelled on the stress block of BS8110-1:1997, taking
into account a calibrated C30 grade peak stress value of 13.4N/mm2, which
occurs at 0.002 compressive strain, and adding a dipping softening curve that
concludes once the concrete reaches a maximum strain of 0.0035. A Youngs
modulus value of 24.6kN/mm2 was also calculated on the assumption of use of
C30 concrete.
Figure 3.10: Concrete compressive stress-strain curve (as per fig. 2.1, BS8110-1:1997)
31
Figure 3.11: Steel stress-strain curve (as per fig. 2.2, BS8110-1:1997)
Stress
(N/mm2)
1.50e-2
438.1N/mm2
1.20e-1
503.8N/mm2
1.75e-3
350.5N/mm2
Strain
The chosen quantity of reinforcement has been taken so as to produce an underreinforced slab which should lead to a ductile failure behaviour and provide
indications of the yield line formations. The representation of the singly reinforced
University of Surrey
32
University of Surrey
33
3.8.
University of Surrey
34
Results
chapter 4 RESULTS
4.1.
Figure 4.1: Yield line formation adopted for analysis (Kennedy, 2003)
University of Surrey
35
Results
The solution for the failure load, using a yield line analysis, has then been
deduced from consideration of the load required to cause slab failure with
minimum work as per Eqn 4.1 and Eqn 4.2.
Use of Work Equation:
Ee = Di
Eqn 4.1
N . = m.l.
Eqn 4.2
This results in the adoption of the following solution for the applied moment on the
slab with relation to the applied loads and span:
M = f .wudl .L2 / 8
Eqn 4.3
Where:
M
wudl
W + S (kN/m)
Slab self-weight
University of Surrey
6.00 m
36
Results
Beff
gf
Variation between the two main methods for slab ultimate load prediction will
consequently differ with regards to the respective value of distribution width
adopted. A yield line analysis will assume the adoption of a continuous failure line
in order to induce the collapse of the slab, therefore taking a value of the full width
of 12.0m for the adopted model.
In comparison, use of the strip method shall entail the adoption of a finite strip
width. The actual width to be used is a contentious issue as has been described
in chapter 2, where an established dimension for large line loads is not readily
available. Since the best definition is found in clause 3.5.2.2a of BS8110-1:1997,
this value shall be adopted as a benchmark for a minimum allowable distribution
width permitted in codes, and is defined as:
Eqn 4.4
Where:
x
L/2
Beff
3.60m
3.85m
3.00m
Therefore:
0.25m
One should note that although this value is being use in order gauge the accuracy
of the strip method, it is not expected to relate accurately to the results of the finite
University of Surrey
37
Results
element analysis. The main reason for this is that the value of the distribution width
stated above is based on the assumption that the transverse slab reinforcement
involves a minimum percentage of reinforcement of 0.13% of the concrete gross
section, as stipulated in the BS8110-1:1997. Consequently, since the finite
element models involve permutations of isotropic reinforcement with transverse
steel areas ranging from 0.28% to 1.23%, it is expected that enhanced distributive
widths result.
The resulting wall line load capacities for each analysis permutation, using both
the yield line and strip methods, are listed in Table 4-2. These were calculated by
using Eqn 4.3 and estimating the value of W (wall line load), where the moment
capacity of the section was obtained by equilibrating the compressive and tensile
stress blocks and finding the corresponding lever arm. The resulting ultimate
loads and corresponding moment capacities of all sections where obtained
assuming a balanced section design, and it is therefore expected that steel
yielding occurs prior to any concrete crushing when the ultimate load is reached in
the finite element models.
4.2.
In order to gain confidence with the use of the Lusas finite element package and
to gauge the accuracy of the chosen mesh sizing, an initial model was utilised that
could be tested against known standard solutions. Since the basis of the problem
to be studied involves a one-way slab strip under uniform distributed loading, an
analogy with an elastic simply supported beam could be made. An acceptable
mesh size, for adoption in the finite element model, can be derived from a
convergence test involving the comparison of the deflection, deduced from elastic
theory, with those obtained from analysis of varying finite element mesh sizes.
University of Surrey
38
Results
Deflections for elastic simply supported beams can be obtained from text book
solution which, for a simply supported beam with a uniform load, is equal to
(Roark et. al., 1975):
5.w.L4
384.E.I
Eqn 4.5
where,
E
100 N/mm
6000 mm
University of Surrey
39
Results
(a)
(b)
(c)
(d)
(e)
Figure 4.2: Varying mesh dimensions utilised for convergence test, with grid spacing:
(a) 62.5mm; (b) 125mm; (c) 250mm; (d) 500mm.
(e) Section of slab analysed displaying restraint setup used to model full slab
Consequently, this deviation limit range and the associated mesh sizing are taken
to be acceptable for this particular study. It was therefore considered adequate to
use grid sizing of 125mm width close to the wall line load application, which also
matches with the width of the line load. The grid sizes are then enlarged to
250mm and 500mm away from the regions with concentrated stresses.
University of Surrey
40
Results
13.5
M i d sp an
Defl ect i on
(mm)
13.4
13.3
13.2
13.1
13.0
Elastic Analysis
12.9
12.8
12.7
12.6
62.5
2.5% Deviation
125
250
12.5
500
12.4
12.3
12.2
12.1
12.0
0.0
50.0
100.0
150.0
200.0
250.0
300.0
350.0
400.0
450.0
500.0
4.3.
University of Surrey
41
Results
After comparison of the plotted results, samples of which are displayed in Figure
4.4 to Figure 4.7, the distribution of the stresses and magnitudes of the loads
proved to be identical. The correct nature of the quarter slab restraints was
therefore confirmed. In addition, the reduction of the finite element model provided
extensive benefits in time and allowed for a greater number of permutations to be
carried out. Through comparison of the two models, it is estimated that at least
four hours was saved from each analysis.
University of Surrey
42
Results
4.4.
The first model used for this study was based on the typical transfer slab detail
adopted frequently in local construction, as described in the preliminary chapters.
These details involve the use of T25 reinforcement at 100mm spacing within a
400mm C30 concrete slab. The reinforcement was placed in an isotropic
University of Surrey
43
Results
In order to determine the acceptability of the model used, the strains and stresses
across the section depth were checked against known principles of beam theory.
Figure 4.8 and Figure 4.9 display the response of the above-mentioned slab whilst
the section is still behaving elastically prior to any cracking in the concrete, at
approximately 10% of the ultimate load. From these images it is apparent that the
strains retain a linear distribution along the depth with the neutral axis being
positioned in the section mid-depth, and consequently the fully bonded state of
the reinforcement layer is being adequately modelled.
University of Surrey
44
Results
The stress distribution then produces a good correlation with the selected material
properties and a clear definition of the steel and concrete regions. These result
from a distinct difference in the response to the experienced strains. As expected,
a concentrated band of high tensile stress is located at the position of the
reinforcement, which is balanced by a compressive region with a parabolic stress
distribution above the neutral axis. Consequently, the resulting stress block has
been successfully assimilated to that identified in codes of practice for reinforced
concrete beam design.
Following the confirmation of the correct nature of the model, the required steps
were taken, in the non-linear solution parameter, to ensure analysis convergence
and the full completion of all load increments, as described in the previous
chapter. At this stage, an acceptable degree of confidence in the first analysed
model, involving a steel area equivalent to T25-100 (4910mm2/m width) was
achieved and the results could then be adequately inspected.
University of Surrey
45
Results
A full description of the output of this initial analysis is described in Section 4.7.4.
In summary, the manner in which the slab behaved involved a non-ductile
response to the failure loads, where continuation of analysis was terminated once
high compressive stresses were experienced and extensive crushing of the
concrete occurred. At this final stage of the finite element analysis the steel layer
was still within elastic limits with no apparent yielding. Consequently, it is evident
that this study of failure sequence will depend heavily on the degree to which the
slab section is under reinforced, since the possible development of yield line
formation is obviously dependent on the formation of plastic hinges. Permutations
involving varying quantities of reinforcement were therefore analysed and these
were compared with predicted ultimate loads and failure modes to create a
complete picture of the influence of reinforcement percentages.
Table 4-1: Permutations used for finite element models
Model
Modelled Reinforcement
Equivalent Plate
Reference
(Both Directions)
Thickness
(per metre width)
1.13 mm
2.00 mm
3.14 mm
5.00 mm
The slab sections which have been analysed involve the variations described in
Table 4-1, whilst characteristics of material properties and finite element model
setup are kept constant.
University of Surrey
46
Results
4.5.
In order to provide an overview of the results of the various slab permutations, the
results of the load deflection response of the above-mentioned model cases
have been plotted and superimposed, as per Figure 4.10.
T12-100
T16-100
T20-100
T25-100
35
30
25
20
15
10
5
0
0.0
100.0
200.0
300.0
400.0
500.0
600.0
700.0
800.0
900.0
1000.0
Figure 4.10: Slab deflection with increased load increments for varying slab reinforcement
From the comparison of results displayed, the obvious difference in the slab
stiffness is immediately apparent, with increased deflection occurring for more
lightly reinforced models. Although the exact position at which the slab is
determined to be no longer serviceable is not clearly visible, clear kinks in curves
C and D are visible. These kinks have been related to distinct changes in the
respective compressive strain diagrams and the contemplated failure loads.
These aspects have been discussed in further detail for each model below.
Therefore, one result which is apparent at this stage is that there is a marked
difference in the manner in which models A and B fail when compared to models
C and D, possibly due to an enhanced degree of ductility during failure.
University of Surrey
47
Results
4.6.
University of Surrey
48
Results
Figure 4.12: Location of all regions experiencing cracking/crushing within slab depth
The initial cracking occurs when the load reaches a value of approximately 20% of
the ultimate value, with the tensile capacity exceeded in the central lower region of
the slab below the line load as expected. As the load is increased, further tensile
cracking develops in a concentric pattern away from the centre of the symmetrical
slab. In addition, extensive splitting of the concrete develops directly below the
line load as is apparent in Figure 4.11.
The first crushing in the concrete occurs in the upper section of the slab once an
increment of approximately 45% of the ultimate load is reached. It is visible in two
locations involving a central area, which develops in the same propagating pattern
as the central tensile cracking, and also in a region in contact with the support and
in line with the wall load. This second region is obviously a result of the increasing
shear stresses which develop at the support and these cracks eventually spread
throughout the line of the simple support.
University of Surrey
49
Results
Crushing of concrete also develops in a patch of the slab adjacent to the support
at approximately two metres away from the line load. This occurs at half the
ultimate load value and has been attributed to the torsional stresses that develop
in the slab. From comparison with the SXY torsional stress diagram of the same
slab shown in Figure 4.14, it is apparent that the location and general formation
pattern of the maximum torsional stresses occur in the same region.
The development of these torsional stresses adjacent to the support is attributed
to the displaced shape of the slab as a result of the presence of the line load. The
disparity in the load magnitude across the slab creates a considerable transverse
curvature across the slab, as displayed in Figure 4.15. This effect is magnified
adjacent to the support, with large variations in the relative rotation of the slab at
the support. There is an absence of theoretical knowledge regarding resistance of
slab to torsional effects. However, from the study of this particular slab it is
apparent that the development of load increments was not limited due to the
presence of these high torsional stresses, but rather as a result of the failure of the
material along the yield lines, as shall be discussed below. Consequently,
University of Surrey
50
Results
although it is unclear how these torsional stresses are being distributed along the
slab, they do not seem to be the source of slab failure.
University of Surrey
51
Results
Figure 4.16: Strain distribution across the depth of slab B following concrete cracking
4.7.
Failure Description
The magnitude of the failure load for each model type has been indicated in Table
4-2, where these values have been reasoned by consideration of the state of the
slab deformation. A major difficulty in the estimation of the failure load is actually
defining the level at which one considers that the slab is no longer performing its
function to withstand the load. As outlined previously in the literature review, this
stage is often reached prematurely when compared to the actual load required to
cause total collapse, where a defined failure load can be a fraction of this collapse
load. This topic will lead to the argument involving the acceptable serviceable
state of the slab with relation to the ultimate load state. Once limitations of
serviceability have been exceeded during loading, an increasing degradation of
the slab involving extensive cracking and crushing, may be used as the criterion
for the failure load rather than total collapse. This condition is related closely to the
degree of ductility displayed by slabs.
The criteria used to define failure will be the main focus of the model descriptions
below. Serviceability limitations associated with deflection restrictions are not an
issue for this particular study. Due to the large span to depth ratio of transfer
slabs, the resulting deflections are not of a critical nature since the maximum
University of Surrey
52
Results
deflection for all models does not exceed a recognized limiting value of span/200,
as is apparent in Figure 4.4.
Table 4-2: Comparison of slab capacity prediction using alternate theorems
Analysis Criteria
(Ultimate Wall Line Load kN/m)
Model
Slab Theory
Reference
Yield Line
Strip
Failure
Maximum
Theory
Method
Load
Analysis Load
A (T12)
340.4
109.2
393.8
437.5
B (T16)
690.7
221.6
562.5
625.0
C (T20)
1102.7
353.8
625.0
625.0
D (T25)
1659.4
532.4
760.0
950.0
Also of interest, in the model descriptions below, is the possibility to visualise the
propagation of the yield line formation. This will provide information regarding the
correct nature of the assumed yield line model for this particular slab system. In
addition, it will be possible to estimate effective load distribution widths based on
the stress distribution across the section. The co-ordinate system adopted for the
description of the stress state involves x-y axes based on the model grid, which
correlates with the direction of the isotropic reinforcement. It is therefore
considered that the stresses displayed are associated which those that would
exist should the steel layer be replaced by a grillage of reinforcement bars.
Appendices A to D have been compiled to give a picture of the growth of stresses
for models A to D respectively, for a selection of stepped load increments, along
critical sections chosen according to forecast yielding sections and support
positions as shown below. The points on the stress-length curves represent those
recorded by Lusas at mesh nodal positions, and therefore the values at
extremities represent the maximum stresses in the material outer fibres. In
University of Surrey
53
Results
addition, the state of the concrete strains, at critical load increments, has been
provided in order to explain the criteria used to define the ultimate load.
54
Results
Figure 4.18: Distribution of SY stresses along X-axis length at load factor 0.20
Once these irregularities occur, stresses are then distributed to adjacent regions
until the stresses in the material reach limits as defined in the material stress-strain
parameters. The concrete and steel of model A reach limiting stress values once a
393.8 kN/m line load, with an applied load factor 0.90, and this is therefore
conservatively taken to be the failure load. The following load increments cause an
increase in the yielded region of the main reinforcement both along the X-axis and
also underneath the line load. Consequently, it is apparent that a yield formation
occurs on a circular patch as a result of high stresses beneath the wall. The
pattern of yielding in the transverse direction extends up to a similar region but is
much more localised below the line load without extending transversely, as is
displayed in Figure 4.20.
Whilst the tensile stresses develop in a marked range of distribution, it is observed
that the compressive concrete stresses develop in a more uniform fashion. Once
the maximum compressive stress is attained, in the upper concrete fibres of the
section, this level of stress is kept at a regular upper limit by concrete crushing
action and subsequent releases onto adjacent areas. Consequently, this
maximum value of stress is spread both in plan and through the concrete depth to
produce the parabolic stress distribution previously described.
University of Surrey
55
Results
Values of strain of 0.0023 and 0.0013 occur in the concrete and steel respectively
at this maximum load increment. From previous trials it was found that continued
convergence and increased analysis of larger loads was found to be not possible.
Since it is evident that these values of strains relate only to the initial stages of
University of Surrey
56
Results
Although one would expect that this behaviour occur for axially restrained slabs,
the relatively small percentage of reinforcement compared with the slab thickness
allows yielding of the reinforcement to take place early in the loading sequence.
This would allow the development of catenary membrane action to occur in the
central region of the slab. The possibility for this form of behaviour to occur, even
University of Surrey
57
Results
for unrestrained simply supported slabs, has already been outlined in chapter 2,
as displayed in Figure 2.10. This hypothesis would however need to be verified
through either practical testing or carrying out alternate analysis on other finite
element software.
Figure 4.22: Main EY strains for model B for load increment 0.90
University of Surrey
58
Results
Figure 4.22: Main EY strains for model B for load increment 0.90
A common feature of all analysed models involves the localised manner in which
transverse x-direction stresses are distributed in the length of the slab along the xaxis. In contrast to the stresses in the main y-direction, transverse stresses reduce
drastically to about 50% of peak values at approximately 0.5m away from the wall
position and eventually become trivial at 2.0m away form the line load. Although
University of Surrey
59
Results
the peak stresses are of a large magnitude, they are also not sufficient to cause
yielding of the steel except in the case of model A which utilised T12
reinforcement. This limited distribution pattern has taken place even though an
isotropic arrangement of steel is assumed, and this would therefore tend to
suggest that slab distribution properties are not very sensitive to varying quantities
of transverse steel. These findings of distribution of stresses have also been
confirmed and agree with a parallel study being carried out by Peter Zammit using
a grillage analysis with varying permutations of transverse reinforcement.
Figure 4.24: Distribution of SX stresses along X-axis length at load factor 0.90
University of Surrey
60
Results
Figure 4.25: Distribution of SY stresses along X-axis length at load factor 1.00
Figure 4.26: Distribution of SY stresses along Y-axis length at load factor 1.00
It would be expected that the yield line formation process continue along the Xaxis through further redistribution of stresses at mid-span. Further yielding below
the wall along the Y-axis would most probably cease as a result of the stiffness
along the slab support edge and due to the limited values of bending stresses in
this location. One would also have to consider the degree of deformation that the
slab must undergo in order to reach a full yield line formation, since extensive
concrete cracking must occur beforehand.
University of Surrey
61
Results
Amongst the sections analysed, the T20 reinforcement model proved to be the
most problematic as regards to the use of Lusas finite element modelling.
Problems with application of increased load, due to model convergence failure,
resulted in the definition of a premature failure load which was close in magnitude
to the more lightly reinforced model B. The performance of the slab in the
deflection was shown to be in line with expected trends as displayed in Figure 4.4.
However, as previously mentioned, a sudden change in the slab deflection
occurred at load factor 0.90 which coincided with the first instance of the concrete
reaching its peak stress levels.
Throughout the rest of the loading sequence, the steel tensile stress increases to
values just below yielding levels, whilst it was also evident that the strains in the
section could have endured further deformation. Consequently, since this model
proved to develop into a problematic analysis, values regarding failure load
capacity should be treated with scepticism due to premature convergence
problems. The stresses and strains that resulted from sequential load increments
University of Surrey
62
Results
did however keep the same trends as previous models with regards to expected
stress levels. Consequently, results from this analysis will therefore still be utilised
for analysis of load distribution described below.
63
Results
Once this stage is surpassed, the slab continues to withstand further load
increments and increases by a further 20% prior to conclusion of the finite element
convergence process. As the load continues to grow, maximum compressive
stresses are distributed and spread across the slab section whilst steel yielding
eventually occurs at the maximum load increment. This process is coincidental
with a constant level of strain in the concrete and simultaneous growth of strain in
the steel. It should be noted that this behaviour was noted for the reinforcement in
the main x-direction, and therefore a potential of yield line propagation at midspan is therefore possible.
Observation of the transverse stress profiles reveal that the pattern of concrete
high compressive stress exists locally below the line load. Notably, the stress in
the steel involves an irregular state, probably due to the extensive concrete
cracking, and reaches values below yielding limits.
University of Surrey
64
Results
Figure 4.30: Distribution of SY stresses along X-axis length at load factor 1.00
4.8.
65
Results
the figures in the appendices. In addition, it should also be noted that the
unreinforced concrete shear capacity has been exceeded in all cases by a large
percentage at up to one metre away from the line load. Since it is not apparent
that the failure of the model was attributed to this occurrence, this result provides
a contentious issue which should be tackled by more detailed future research.
T12-100
T12 V
T16-100
T16V
T20-100
T20 V
T25-100
T25 V
1000
900
800
700
600
500
400
300
200
100
0
-100
-200
-300
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
5500
6000
Another point which is important to note is the uplift at the corner of the slab,
which resulted in all of the models and is visible by the negative reaction at the
end half metre of the slab width. This occurrence provides an inaccurate depiction
of unrestrained slab models which should not provide clamping action. It does
however indicate the expected negative reactions that result from the slab
curvature formed due to intense line loads. These reactions have also been
utilised to gauge the models state of equilibrium at failure. On comparison of
applied loads and vertical support reactions, it was found that the difference was
within reasonable limits. The average of all models was in the region of 10%.
University of Surrey
66
Results
4.9.
T12-100
T16-100
T20-100
T25-100
425
400
2
375
350
325
300
275
250
225
200
175
150
125
100
75
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
5500
6000
A similar exercise was carried out for the comparison of the failure stresses for the
four models, with the superimposition of stresses in the y-direction, corresponding
with the direction of the critical main reinforcement and the main one-way action
bending plane. Stresses in the steel and extreme concrete fibres at failure are
displayed in Figure 4.32 and Figure 4.33 respectively. It is immediately notable
that two distinct states occur, where an extremely haphazard stress state is found
up to half a metre away from the line load centre as a result of the non-linear
behaviour of the material. Away from the line load, whilst the concrete and steel
are still within their linear elastic range, the material stresses follow a regular
distribution pattern. The limiting stress values, from which non-linearity develops,
are consequently defined by the stress-strain curves as defined in chapter 3 with
the onset of extreme deformity at values of 350N/mm2, for steel first yield, and a
peak concrete compressive stress of 13N/mm2.
University of Surrey
67
Results
T12-100
T16-100
T20-100
T25-100
20.0
17.5
15.0
12.5
10.0
7.5
5.0
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
5500
6000
Figure 4.33: Compressive stresses in top of concrete at failure along slab width
It is evident that there is also a difference in the behaviour of the materials, with a
greater uniformity of concrete compressive stresses, across the slab width, when
compared to the steel stresses. However, of greater importance in relation to the
design aspect of transfer slabs, is the stress state existent in the reinforcement, in
the spanning direction being considered, and the response of the steel to the
applied line load. In this regard it is possible to comment on the recommended
design strip width involving a minimum of approximately two metres from the line
load. Although the quantity of transverse reinforcement has been varied in each
isotropic model, the transfer slab is found to have consistent distribution
characteristics throughout all models. It was found that at the stipulated two metre
distance the tensile stresses experienced a reduction in the range of 65% of the
peak stresses below the wall, whilst distribution continued up to the slab edge
with percentages of approximately 25%.
Consequently, the analysed load case, involving a symmetrical centrally located
line load, proved to display considerable load distribution characteristics that were
University of Surrey
68
Results
University of Surrey
69
Conclusions
chapter 5 CONCLUSIONS
The principal aim of this study was to establish capacities of slabs with varying
reinforcement under the effects of a continuous centralised line load. A study of
the existing methods of design has been established based on classical slab
theory and stipulated codes of practice. The main methods involve either the use
of yield line theory or the strip design method. It has been shown that, for a oneway spanning slab, a clear divergence in the estimation of ultimate loads exists
when these two methods are compared. This difference has been found to result
from the manner in which the line load distributive action is taken. Whilst a yield
line analysis produces a failure criterion which is based on assuming that the
whole slab width resists the load, the strip design method in contrast adopts a
finite width of slab to resist the load. The width utilised for this study has been
taken as stipulated as per BS 8110-1:1997, which is based on the assumption of
a slab with minimum 0.13% transverse distribution, unlike the more heavily
reinforced sections adopted in the comparative finite element models.
Therefore, it is evident from these hand calculation methods that the determination
of the intense line load magnitude will be dependent on the manner in which the
concentrated load is distributed. Since the applied slab load is not uniformly
distributed, it is expected that concentrations of stress would form close to the
load position. Consequently, although physical testing has proven to be in
agreement with yield line predictions, the degree to which a full yield line develops
across the whole width of the slab is expected to be limited by the stage at which
a finite portion of slab displays signs of failure. This leads to the importance of
making a distinction between the load at which total collapse occurs, as is
reflected by the total yield line formation, and the ultimate load to be used for
design purposes based on an unsightly condition of a slab portion.
University of Surrey
70
Conclusions
The use of a finite element model has consequently been formulated in order to
provide an insight into the mechanism involved with the slab failure, as discussed
above, and has provided a clear comparison of load capacities when compared
to traditional slab design techniques. The finite element model mesh adopted has
been successfully tested for convergence by using a simplified approach and
comparing displacements with those of exact solutions based on elastic theory
calculations. Since the analysis has involved the use of material non-linearity, any
further studies would need to carry out similar convergence tests on actual
physical testing so as to make a comprehensive test of both the mesh sizing and
the non-linear control parameters.
Another component of the finite element model, involving the quarter-slab
simplification and the associated restraint conditions, has been found to present a
correct representation of the full slab scenario. This simplification has allowed for
the execution of a greater number of analytical permutations and load increments.
The section adopted also proved to correctly model the expected reinforced
concrete beam stress block and incorporated a linear strain pattern across the
section up to the onset of cracking. The only anomaly of the adopted model,
which should be corrected in future studies, involves the use of a simplified layer
modelling approach to represent the existence of the reinforcement. An
inaccuracy has been introduced since the steel layer has provided stiffness in all
directions rather than only in the orthogonal reinforcement directions. It is however
not expected to have created too great an inaccuracy in the final outcome due to
relatively good correlation of stress distribution patterns, in the reinforcement
directions, when compared to a parallel study carried out by Peter Zammit
involving a grillage analysis of the same slab. In addition, the reinforcement
usually utilised in practice involves a closely spaced mesh which can be
assimilated closely to a continuous steel layer.
University of Surrey
71
Conclusions
University of Surrey
72
Conclusions
The mechanism which led to the description of the failure load therefore involved
the examination of the corresponding stresses and strains along the expected
yield line. Depending on the type of reinforcement used, the slab mid-span
section experienced either the development of tensile steel yielding or excessive
compressive stresses. It was observed for the more lightly reinforced T12 and T16
sections that a greater degree of ductility occurred where the critical aspect
involved the eventual yielding of the reinforcement. These models became
unstable once the first yield was reached and tensile strains became excessive,
where a yielding section of slab of width 0.50m from the wall existed. In contrast,
the T20 and T25 models were found to have a failure mechanism which was
limited by the excessive compressive strains that developed even whilst the steel
had not yet yielded.
The failure load was therefore linked to the physical state of the finite element
model in terms of excessive strains and growth in deflections experienced in the
slab. Since this was not based on actual physical observations relating to
propagation of acceptable crack widths, the identified failure load magnitudes
may therefore be subject to interpretation. What was however clearly evident was
the possibility of the slab to continue to withstand increased applied load and
remain in tact, which is in agreement with physically tested models carried out in
the past. The development of actual yield planes was evident in the stress
diagrams across both the predicted critical mid-span section and also along the
unpredicted plane beneath the wall line load. The mechanism that comes into
effect involves the shedding of load onto regions which have not yielded,
consequently leading to a propagating plane eventually creating a growing length
of cracked - crushed concrete and yielding reinforcement. Whilst this behaviour is
marked for the more ductile and lightly reinforced sections, it was also observed
for the T20 and T25 sections, where reinforcement eventually reached yielding
stresses below the line load and compressive strains continued to propagate. It
was however not possible to observe the full development of yield lines due to
University of Surrey
73
Conclusions
University of Surrey
74
Conclusions
width being formulated according to the worst case load distribution scenario
which involves the line load being placed at a slab edge. In addition, the failure
load was less than the yield line prediction, as is expected since full yield line
propagation did not form in these models.
Consequently, in light of the results gained through this study, it is apparent that
although the use of the stipulated effective design width produces a conservative
load capacity, further data is required to advise on the use of greater slab
distribution widths for moment capacity purposes. In addition, it is evident that the
slabs failure sequence and response to loading depends greatly on the quantity
of reinforcement in the slab and the resulting degree of slab ductility.
In closing, as a result of the observations made through this study a number of
recommendations can be made for further research study involving the
application of intense wall line loads on reinforced concrete slabs. These involve
the need to confirm the results by carrying out similar finite element analysis on
finite element packages, which are more adapt to analysis of reinforced concrete
than Lusas, and the calibration with actual physical testing. In addition, it would
also be ideal to gain a greater understanding of the distribution of shear and
torsional stresses within the concrete slab through studies which specifically
research these aspects.
University of Surrey
75
References
REFERENCES
Armer, G. S. T., Discussion of Reference Wood, R. H., The Reinforcement of
Slabs in Accordance with a Pre-Determined Field of Moments, Concrete, vol. 2,
pp. 319320, August 1968.
Barzegar, F. and Maddipudi, S., Generating Reinforcement in FE Modelling of
Concrete Structures, ASCE Journal of the Structural Division, vol. 120, pp. 1656
1662, May 1994.
BOCA National Building Code, 13th ed., Building Officials and Code Administrators
International, Country Clubs Hill, Illinois, 1996.
Brotchie, J. F. and Russell, J. J., Flat Plate Structures: Elastic-Plastic Analysis,
Journal of the American Concrete Institute, vol. 61, pp. 959996, August 1964.
BS 8110 (1997), Structural Use of Concrete. Part 1: Code of Practice for Design
and Construction, British Standard Institution.
Cope, R. J. and Vasudeva Rao P., Non-Linear Finite Element Analysis of Concrete
Slab Structures, Proc. Instn. Civ. Engrs., Part 2, 1977, Vol. 63, March, pp 159
179.
Fenwick, F. and Dickinson, A., Slabs Subject to Concentrated Loading, ACI
Structural Journal, Nov-Dec 1989, pp. 672-678.
Ingerselv, A. The strength of rectangular slabs, Journal Institution of Structural
Engineers, December 1922
University of Surrey
76
References
University of Surrey
77
References
University of Surrey
78
Bibliography
BIBLIOGRAPHY
ACI Committee 318 (2001), ACI 318-02: Building Code Requirements for
Structural Concrete, American Concrete Institute.
Burgoyre, C., Are Structures Being Repaired Unnecessarily, The Structural
Engineer, 6 January 2004, pp.22-26.
Cope, R. J. and Clark, L. A., Concrete Slabs Analysis and Design, Elsevier
Applied Science Publishers, 1984.
Eurocode 2 (1992),Design of Concrete Structures Part1-1: General Rules and
Rules for Buildings', Portland Cement Association.
Hillerborg, A., Strip Method Design Handbook, E & FN Spon, London, 1996.
Jones, L.L. and Wood, R.H.,Yield-Line Analysis of Slab, Thames & Hudson,
1967.
Kennedy, G. and Goodchild, C., Practical Yield Line Design, Reinforced Concrete
Council, British Cement Association, 2003.
Martin L.H., Croxton P.C.L. and Purkiss J.A., Structural Design in Concrete to
BS8110, Edward Arnold, London, 1989.
May, I.M. and Lodi, S.H., Deficiencies of the Normal Moment and Yield Criterion
for RC Slabs, Proceedings of the Institution of Civil Engineers, Structures and
Buildings 158, December 2005, Issue SB6, pp.371-380.
University of Surrey
79
Bibliography
McNeice, G. M. and Kemp, K. 0., Comparison of Finite Element and Unique Limit
Analysis Solutions for Certain Reinforced Concrete Slabs, ICE Paper 7201.
MSc Lecture Notes, Structural Mechanics and Finite Elements, University of
Surrey, Guildford, Surrey
MSc Lecture Notes, Space Structures, University of Surrey, Guildford, Surrey
Phillips, D. V. and Zienkiewicz, O. C., Finite Element Non-Linear Analysis of
Concrete Structures, Proc. Instn. Civ. Engrs., Part 2, 1976, Vol. 61, March, pp 59
88.
Wood, R. H., Plastic and Elastic Design of Slabs and Plates, Thames and
Hudson, London, 1961.
University of Surrey
80
appendix A
A.1: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.20
A.2: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.40
University of Surrey
81
appendix A
A.3: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.60
A.4: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.80
University of Surrey
82
appendix A
A.5: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.90
A.6: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 1.00
University of Surrey
83
appendix A
A.7: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.20
A.8: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.40
University of Surrey
84
appendix A
A.9: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.60
A.10: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.80
University of Surrey
85
appendix A
A.11: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.90
A.12: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 1.00
University of Surrey
86
appendix A
A.13: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.20
A.14: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.40
University of Surrey
87
appendix A
A.15: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.60
A.16: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.80
University of Surrey
88
appendix A
A.17: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.90
A.18: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 1.00
University of Surrey
89
appendix A
A.19: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.20
A.20: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.40
University of Surrey
90
appendix A
A.21: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.60
A.22: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.80
University of Surrey
91
appendix A
A.23: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.90
A.24: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 1.00
University of Surrey
92
appendix A
A.25: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.20
A.26: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.40
University of Surrey
93
appendix A
A.27: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.60
A.28: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.80
University of Surrey
94
appendix A
A.29: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.90
A.30: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 1.00
University of Surrey
95
appendix A
96
appendix A
97
appendix A
98
appendix A
99
appendix A
100
appendix A
101
appendix A
102
appendix A
103
appendix A
104
appendix A
105
appendix B
B.1: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.20
B.2: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.40
University of Surrey
106
appendix B
B.3: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.60
B.4: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.80
University of Surrey
107
appendix B
B.5: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.90
B.6: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 1.00
University of Surrey
108
appendix B
B.7: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.20
B.8: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.40
University of Surrey
109
appendix B
B.9: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.60
B.10: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.80
University of Surrey
110
appendix B
B.11: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.90
B.12: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 1.00
University of Surrey
111
appendix B
B.13: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.20
B.14: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.40
University of Surrey
112
appendix B
B.15: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.60
B.16: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.80
University of Surrey
113
appendix B
B.17: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.90
B.18: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 1.00
University of Surrey
114
appendix B
B.19: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.20
B.20: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.40
University of Surrey
115
appendix B
B.21: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.60
B.22: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.80
University of Surrey
116
appendix B
B.23: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.90
B.24: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 1.00
University of Surrey
117
appendix B
B.25: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.20
B.26: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.40
University of Surrey
118
appendix B
B.27: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.60
B.28: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.80
University of Surrey
119
appendix B
B.29: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.90
B.30: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 1.00
University of Surrey
120
appendix B
121
appendix B
122
appendix B
123
appendix B
124
appendix B
125
appendix B
126
appendix B
127
appendix B
128
appendix B
129
appendix B
130
appendix B
131
appendix B
132
appendix C
C.1: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.20
C.2: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.40
University of Surrey
133
appendix C
C.3: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.60
C.4: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.80
University of Surrey
134
appendix C
C.5: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.90
C.6: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 1.00
University of Surrey
135
appendix C
C.7: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.20
C.8: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.40
University of Surrey
136
appendix C
C.9: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.60
C.10: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.80
University of Surrey
137
appendix C
C.11: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.90
C.12: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 1.00
University of Surrey
138
appendix C
C.13: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.20
C.14: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.40
University of Surrey
139
appendix C
C.15: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.60
C.16: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.80
University of Surrey
140
appendix C
C.17: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.90
C.18: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 1.00
University of Surrey
141
appendix C
C.19: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.20
C.20: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.40
University of Surrey
142
appendix C
C.21: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.60
C.22: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.80
University of Surrey
143
appendix C
C.23: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.90
C.24: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 1.00
University of Surrey
144
appendix C
C.25: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.20
C.26: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.40
University of Surrey
145
appendix C
C.27: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.60
C.28: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.80
University of Surrey
146
appendix C
C.29: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.90
C.30: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 1.00
University of Surrey
147
appendix C
148
appendix C
149
appendix C
150
appendix C
151
appendix C
152
appendix C
153
appendix C
154
appendix C
155
appendix D
D.1: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.20
D.2: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.40
University of Surrey
156
appendix D
D.3: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.60
D.4: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.80
University of Surrey
157
appendix D
D.5: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 0.90
D.6: Sectional stresses sy (N/mm2) / Distance along x-axis (mm) Load increment 1.00
University of Surrey
158
appendix D
D.7: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.20
D.8: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.40
University of Surrey
159
appendix D
D.9: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.60
D.10: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.80
University of Surrey
160
appendix D
D.11: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 0.90
D.12: Sectional stresses sy (N/mm2) / Distance along y-axis (mm) Load increment 1.00
University of Surrey
161
appendix D
D.13: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.20
D.14: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.40
University of Surrey
162
appendix D
D.15: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.60
D.16: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.80
University of Surrey
163
appendix D
D.17: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 0.90
D.18: Sectional stresses sx (N/mm2) / Distance along x-axis (mm) Load increment 1.00
University of Surrey
164
appendix D
D.19: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.20
D.20: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.40
University of Surrey
165
appendix D
D.21: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.60
D.22: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.80
University of Surrey
166
appendix D
D.23: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 0.90
D.24: Sectional stresses sx (N/mm2) / Distance along y-axis (mm) Load increment 1.00
University of Surrey
167
appendix D
D.25: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.20
D.26: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.40
University of Surrey
168
appendix D
D.27: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.60
D.28: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.80
University of Surrey
169
appendix D
D.29: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 0.90
D.30: Reaction stresses Pz (N/mm2) / Distance along support (mm) Load increment 1.00
University of Surrey
170
appendix D
171
appendix D
172
appendix D
173
appendix D
174
appendix D
175
appendix D
176
appendix D
177
appendix D
178
appendix D
179
appendix D
180
appendix D
181
appendix D
182
appendix D
183
appendix D
184
appendix D
185
appendix D
186