You are on page 1of 99

Dissertation ETH Nr.

15615








Reinforced Concrete Slabs Compatibility Limit Design






A dissertation submitted to the
SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH
for the degree of
Doctor of Technical Sciences



presented by
Mario Nicola Monotti
dipl. Bauingenieur ETH
born 24. September 1975
Citizen of Cavigliano, TI



Accepted on the recommendation of
Prof. Dr. Peter Marti, examiner
Prof. Dr. Christopher Thomas Morley, co-examiner



2004

Foreword
Eine Theorie ist desto eindrucksvoller, je grsser die
Einfachheit ihrer Prmissen ist, je verschiedenartigere Dinge
sie verknpft und je weiter ihr Anwendungsbereich ist.
A. Einstein (1879-1955)


In the analysis of reinforced concrete structures, Einsteins citation may be referred to the theory
of plasticity. By modelling the material behaviour with a rigid and a plastic phase, the analysis
is focused on the ultimate limit state. The simplifying abstraction supports and enhances the
engineers insight and intentions particularly in the process of designing. The power of the
plastic approach is exemplified by its flexibility and simplicity.
In this thesis, plastic principles are applied to the study of reinforced concrete slabs.
Previous contributions and hypotheses on the structural behaviour of plates are discussed within
the rigorous framework of the theory plasticity aiming at a refinement of the plastic design of
reinforced concrete slabs. Novel results include a methodology to describe the slab behaviour
with curvilinear trajectories and a proposal for a design method focused on compatibility.
The present work is part of the research project Deformation Capacity of Structural
Concrete. Reviewing the application of plasticity theory to slabs in analogy to beams and
walls, it provides the basis for the extension of recent research on the deformation capacity of
tension elements and wall elements to slabs. Future studies will concentrate on experimental
verifications, on the evaluation of the shear capacity, and on detailing aspects.
The work has been carried out during my appointment as assistant and research associate of
Prof. Dr. Peter Marti at the Institute of Structural Engineering of the Swiss Federal Institute of
Technology (ETH), Zurich. Here, I would like to express my sincere thanks to Professor Marti
for introducing me into the structural engineering research and for his guidance, advice and
supervision during the past years. I am also most grateful to Prof. Dr. Christopher Thomas
Morley for his thorough review and his improvements of the thesis.
Many thanks to Nadia Scascighini, Andri Gerber and Gerald Prater for help on linguistic
aspects, and to my colleagues, in particular to Thomas Jger, for the interesting discussions.
Finally, special thanks to all the friends who shared my time in Zurich and in the Ticino, to
Tanja and to all my family: you were the best supporters!





Zurich, August 2004 Mario Monotti

Abstract
Based on the theory of plasticity this thesis develops a new design procedure for reinforced
concrete slabs the compatibility limit design method. The basic idea of this method is to
extend the typical design procedure for reinforced concrete beams and frames to slabs. For
beams and frames, the failure mechanisms indicate the global force flow because the plastic
hinges coincide with the points of zero shear force; the force flow within the beam or frame
segments defined by the points of zero shear force can be visualised using truss models or
corresponding stress fields and the segments detailing can be completed accordingly. For slabs,
static and kinematic considerations are normally applied in an unrelated way and hence, the
potential offered by the theory of plasticity is not fully utilised; by considering yield line
mechanisms and developing matching stress fields for slab segments defined by the yield lines
the compatibility limit design method attempts to overcome this situation.
After the introduction (Chapter 1) and a presentation of the fundamentals of the theory of
rigid-perfectly plastic bodies and its application to reinforced concrete (Chapter 2) Chapter 3 to
5 present the static, limit analysis and kinematic considerations underlying the compatibility
limit design method whose application is illustrated in Chapter 6 by means of a practical
example. Chapter 7 contains a summary, conclusions and recommendations for future studies.
The static considerations concentrate on the load transfer mechanisms in slabs and their
boundary conditions. Distributed and concentrated load transfer are differentiated. Distributed
load transfer is described by the generalised strip method, using general curved rather than
straight orthogonal beams. Concentrated load transfer occurs in strong bands or along shear
lines. Together with a set of stress fields describing a certain distributed load transfer within
individual slab segments strong bands and shear lines are the basic tools of a stress field
approach for slabs similar to that used for beams and frames.
The limit analysis considerations are based on a discussion of compatible states of stress
and deformation using the yield condition and the associated flow rule for orthogonally
reinforced concrete slabs. From a kinematic point of view, rigid parts, yield lines and yield
regions are differentiated. It is proposed to bypass the difficulties associated with yield regions
by introducing an approximate limit analysis, corresponding to enforcing yield line mechanisms
of unique sign; similar to the capacity design method used in earthquake engineering this can be
ensured by some local strengthening of the reinforcement.
The kinematic considerations illustrate the application of the work method and the
equilibrium method to yield line mechanisms. It is shown that the two methods are equivalent if
they are associated to a unique statical problem.
The practical application of the compatibility limit design method requires some
preliminary assumptions about the resistance distribution in the slab. Based on an intuitively
assumed yield line mechanism the required global resistances can be quantified and optimised.
In a second step, the stress field approach is employed to study the force flow within and
between the individual slab segments and to detect any local resistance deficits.
Kurzfassung
Auf der Grundlage der Plastizittstheorie wird eine neue Bemessungsmethode fr Stahlbeton-
platten entwickelt die Vertrglichkeitsmethode. Die Grundidee dieser Methode besteht darin,
die fr Stabtragwerke bliche Bemessungsmethode auf Platten zu bertragen. Die
Bruchmechanismen von Stabtragwerken zeigen den Kraftfluss im Grossen an, weil die
plastischen Gelenke den Querkraftnullpunkten entsprechen; der Kraftfluss im Kleinen, im
Inneren der durch die Querkraftnullpunkte begrenzten Elemente, kann mit Fachwerkmodellen
oder entsprechenden Spannungsfeldern untersucht werden, was eine passende konstruktive
Durchbildung ermglicht. Bei der Bemessung von Platten werden blicherweise statische und
kinematische Betrachtungen angestellt, die in keinem direkten Zusammenhang stehen, und das
Potential der Plastizittstheorie wird nicht ausgeschpft; mit der Vertrglichkeitsmethode wird
versucht, diese unbefriedigende Situation zu berwinden, indem ausgehend von angenommenen
Fliessgelenklinienmechanismen entsprechende vertrgliche Spannungsfelder in den einzelnen
durch die Fliessgelenklinien definierten Plattenteilen entwickelt werden.
Nach der Einleitung (Kapitel 1) und einer Zusammenstellung der Grundlagen der Theorie
starr- ideal plastischer Krper sowie deren Anwendung auf Stahlbeton (Kapitel 2) enthalten die
Kapitel 3 bis 5 die hinter der Vertrglichkeitsmethode stehenden statischen, grenztragfhigkeits-
theoretischen und kinematischen Betrachtungen. Die Anwendung der Vertrglichkeitsmethode
wird im Kapitel 6 anhand eines praktischen Beispiels illustriert, und Kapitel 7 enthlt eine
Zusammenfassung, Schlussfolgerungen sowie Empfehlungen fr weiterfrende Studien.
Im Zentrum der statischen Betrachtungen stehen der Kraftfluss in Platten sowie die
entsprechenden Randbedingungen. Dabei wird zwischen einer Lastabtragung ber verteilte und
konzentrierte Querkrfte unterschieden. Die Lastabtragung ber verteilte Querkrfte wird mit
der verallgemeinerte Streifenmethode untersucht, die allgemein gekrmmte statt gerade
orthogonale Koordinaten verwendet. Konzentrierte Querkrfte treten in versteckten Balken und
Schublinien auf. Zusammen mit einer Familie von Spannungsfeldern zur Beschreibung der
Lastabtragung in Plattenteilen ber verteilte Querkrfte ermglichen versteckte Balken und
Schublinien eine Spannungsfeldanalyse von Platten hnlich jener von Stabtragwerken.
Die grenztragfhigkeitstheoretischen Betrachtungen beziehen sich auf vertrgliche
Spannungs- und Verformungszustnde unter Zugrundelegung der Fliessbedingung und des
zugeordneten Fliessgesetzes fr orthogonal bewehrte Stahlbetonplatten. Kinematisch werden
starre Plattenteile, Fliessgelenklinien und Fliesszonen unterschieden. Um die mit Fliesszonen
verbundenen Schwierigkeiten zu umgehen wird vorgeschlagen, eine approximative
Grenztragfhigkeitsanalyse einzufhren, derart, dass nur positive oder negative Fliess-
gelenklinien auftreten; hnlich wie bei der Kapazittsbemessung im Erdbebeningenieurwesen
wird das Auftreten solcher Mechanismen durch rtliche Bewehrungsverstrkungen erzwungen.
Die kinematischen Betrachtungen sind der Anwendung der sogenanten Energie- und
Grenzgleichgewichtsmethoden auf Fliessgelenklinien gewidmet. Es wird gezeigt, dass die
beiden Methoden quivalent sind, wenn sie derselben statischen Problemstellung entsprechen.
Die praktische Anwendung der Vertrglichkeitsmethode setzt einige Annahmen ber die
Widerstandsverteilung in der Platte voraus. Von einem intuitiv angenommenen Fliess-
gelenklinienmechanismus ausgehend knnen die notwendigen Widerstnde bestimmt und
optimiert werden. In einem zweiten Schritt knnen dann allfllige lokale Widerstandsdefizite
durch Anwendung der Spannungsfeldanalyse endeckt und behoben werden.
Table of contents
1 Introduction 1
1.1 Defining the problem 1
1.2 Overview 2
1.3 Assumptions and limitations 4
2 Theory of plasticity 5
2.1 General 5
2.2 Rigid-perfectly plastic behaviour 5
2.3 Reinforced concrete 6
2.4 Yield condition and flow rule 8
2.5 Theorems of limit analysis 10
2.6 Limit analysis and design methods 11
3 Static method 13
3.1 General 13
3.2 Internal forces 13
3.2.1 Definition of internal forces 14
3.2.2 Stress field definition 14
3.3 Equilibrium 16
3.2.1 Orthogonal curvilinear coordinates 16
3.2.2 Sign convention 17
3.2.3 Equilibrium conditions 17
3.4 Load transfer 18
3.4.1 Distributed load transfer 18
3.4.2 Concentrated load transfer 20
3.4.3 Remarks 20
3.5 Boundary conditions 20
3.6 Lower-bound method 22
3.6.1 Generalised strip method 23
3.6.2 Elasticity 23
3.6.3 Strip method 24
3.6.4 Hencky-Prandtl solutions 25
3.6.5 General stress fields 25
3.6.6 Superposition principle 31
3.6.7 Stress field approach for slabs 33
3.7 Examples 34
3.7.1 Simply supported rectangular slab 34
3.7.2 Triangular slab with a free edge 38
3.8 Conclusions 40
4 Limit analysis 41
4.1 General 41
4.2 Yield condition and flow rule 41
4.2.1 Limit states 42
4.2.2 Yield condition 43
4.2.3 Flow rule 46
4.2.4 Discussion 48
4.3 Approximate limit analysis 49
4.4 Example application 50
4.5 Conclusions 51
5 Kinematic method 53
5.1 General 53
5.2 Failure mechanisms and limit analysis 54
5.3 Upper-bound method 55
5.3.1 Work method 55
5.3.2 Equilibrium method 59
5.4 Discussion 67
5.5 Conclusions 69
6 Compatibility limit design 71
6.1 General 71
6.2 Example application 71
6.2.1 Problem statement 72
6.2.2 Assumptions 72
6.2.3 Detailing 73
6.2.4 Kinematical analysis 74
6.2.5 Statical analysis 74
6.2.6 Reinforcement dimensioning 77
6.3 Discussion 78
6.4 Conclusions 80
7 Summary and conclusions 81
7.1 Summary 81
7.2 Conclusions 82
7.3 Recommendations for future studies 83
References 85
Notation 89
Curriculum vitae 91

1
1 Introduction
1.1 Defining the problem
The theory of plasticity provides a solid foundation for the ultimate limit state design of
reinforced concrete structures [64, 35, 50]. For beams and frames it is relatively easy to develop
coinciding static and kinematic (or lower- and upper-bound) solutions, i.e. to determine failure
mechanisms and matching stress fields at ultimate. For slabs, however, static and kinematic
analyses are usually performed in an unrelated way. It is the aim of this thesis to improve this
situation and to contribute to a better utilisation of the potential for reinforced concrete slab
design offered by the theory of plasticity.
As an introductory example consider the uniformly loaded beam shown in Fig. 1.1a,
clamped at x = 0 and simply supported at x = l.
The static analysis focuses on the force flow, i.e. the way the load is distributed between the
supports. The point of zero shear at x = x
0
in Fig. 1.1b subdivides the beam into two segments in
which the forces flow either to the left or to the right support. Shear force and moment
distribution follow from equilibrium, i.e. V = q (x
0
x) and M = q [(l
2
x
2
) / 2 x
0
(l x)]. For
l / 2 < x
0
< l the extreme positive and negative moments are located at x = x
0
(M(x
0
) = q(l x
0
)
2
/ 2)
and x = 0 ( M(0) = ql(l / 2 x
0
)); the limiting cases x
0
= l / 2 and x
0
= l correspond to a simply
supported beam and a cantilever, respectively, i.e. to statically determinate structures. In
Fig. 1.1b the statical indeterminacy represented by x
0
is reflected by a vertical shift and a
rotation about the end point M = 0 at x = l of the closing lines of the shear force and moment
diagrams, respectively. Assuming a uniform resistance M
u
against both positive and negative
bending moments and setting M(0) = M(x
0
) = M
u
one obtains x
0
= (2 2) l and
M
u
= ql
2
( 2 1)
2
/ 2; on the other hand, assuming an initially stress-free structure and a uniform
elastic bending stiffness, one gets x
0
= 5l / 8 and M(0) = ql
2
/ 8. Complete elastic unloading of
the corresponding elastic-plastic beam from the ultimate load q
u
= 2M
u
/ [( 2 1) l ]
2
would result
in a residual moment M(0) = M
u
+ 2M
u
/ [8 ( 2 1)
2
] = 0.457M
u
at the clamped end of the beam.
Generally, residual stresses depend on the entire loading and restraining history but they do not
affect the ultimate loads that can be carried by perfectly plastic systems.
The kinematic analysis considers the flexural failure mechanism shown in Fig. 1.1c.
Assuming a uniform resistance M
u
against both positive and negative bending moments the total
energy dissipation in the plastic hinges at x = 0 and at x = x
0
equals M
u
(2l x
0
)/ [x
0
(l x
0
)] and this
must be equal to the work ql / 2 of the externally applied forces, hence q = 2M
u
(2l x
0
)/ [lx
0
(l x
0
)].
The function q assumes the minimum q = 2M
u
/ [( 2 1) l ]
2
at x
0
= (2 2) l. It can be seen that
in general, the kinematic analysis results in unsafe estimates of the ultimate load.
For the assumed uniform bending resistance compatibility between statics and kinematics
requires that the plastic hinges coincide with the points of zero shear. Hence, the two free-body
diagrams shown in Fig. 1.1d can be drawn, resulting in the moment equilibrium equations
2M
u
q
2
0
x /2 = 0 and M
u
+ q(l x
0
)
2
/2 = 0, respectively, i.e. x
0
= (2 2)l and q
u
= 2M
u
/[( 2 1)l]
2
.
Note that contrary to the kinematic analysis x
0
is determined by equilibrium equations; no
differentiation process is necessary. The uniqueness of the solution can easily be verified by
means of Fig. 1.1b; by superimposing a residual stress state as indicated by the closing line of

Introduction
2
the moment diagram the yield condition for the positive bending moments would be violated;
similar arguments apply for the negative moments with a closing line rotation in the opposite
direction.
It can be seen that based on compatibility considerations the static and the kinematic
analyses are combined and made more efficient. Thus, the ultimate limit state design may start
from failure mechanism considerations indicating the global force flow and the required main
resistances, followed by a detailed analysis and design of the beam segments between the points
of zero shear force, typically based on truss models and corresponding stress fields as shown in
Fig. 1.1e.
Reinforced concrete slabs are frequently designed using moments determined according to
the theory of elastic plates or Hillerborgs strip method. Alternatively, yield line mechanisms
and approximate design procedures such as the equivalent frame method are applied. Generally,
the different methods are used independently of each other and their results may show
considerable discrepancies. Insufficient flexibility of the static as well as the kinematic
approaches is the reason for this incompatibility. By removing these limitations this thesis
attempts to extend the compatibility limit design considerations from reinforced concrete beams
and frames to reinforced concrete slabs.
1.2 Overview
Chapter 2 introduces the fundamentals of the theory of rigid-perfectly plastic bodies. Yield
condition and associated flow rule are presented in a general form and the theorems of limit
analysis are formulated based on the principles of virtual work and maximum energy
dissipation. A discussion on the applicability of limit analysis and design procedures to
reinforced concrete structures completes this chapter.
Chapters 3 to 5 present the static method, limit analysis and the kinematic method,
introducing basic tools, developing methods of analysis and illustrating their application by
means of two examples.
Chapter 3 concentrates on the load transfer mechanisms in slabs and their boundary
conditions. Different lower-bound methods are identified as particular forms of load transfer. In
order to obtain the necessary flexibility for compatibility limit designs stress fields for slab
segments characterised by distributed load transfer are developed. Together with shear lines and
strong bands such stress fields are the basic tools of a stress field approach for slabs similar to
that used for beams and frames.
Chapter 4 discusses compatible states of stress and deformation based on the yield condition
and the associated flow rule for orthogonally reinforced concrete slabs. From a kinematic point
of view, rigid parts, yield lines and yield regions are differentiated. Since yield regions are
difficult to deal with it is proposed to introduce an approximate limit analysis, corresponding to
enforcing yield line mechanisms of unique sign; similar to the capacity design method used in
earthquake engineering this is ensured by some local strengthening of the reinforcement.
Chapter 5 presents the basic principles of yield line analysis and illustrates the application
of the work method and of the equilibrium method. The relationships between the two methods
are discussed and it is shown that the equilibrium method can be interpreted as an expression of
compatibility limit design considerations (for a suitably defined approximate limit analysis
problem) similar to those underlying Fig. 1.1d.

Defining the problem Overview
3
Fig. 1.1: Statically indeterminate beam: a) system and loading; b) static analysis; c) kinematic
analysis; d) compatibility; e) truss model and stress fields.
a)
l
q
V
M
ql
ql /8
2
M
u
-M
u
b)
c)
e)
M
u
Closing line
x
z
x
V= 0
0
q
M
u
M
u
x
0
1
d)
q
x
0
x
0
l
M
u
qx
0
q x
0
l ( )
M
u
Introduction
4
Chapter 6 demonstrates the application of the compatibility limit design method by means
of a practical example, highlighting the importance of detailing considerations and providing
comparisons with previously derived solutions.
Chapter 7 contains a summary, conclusions and recommendations for future studies.
1.3 Assumptions and limitations
It is assumed that the load-deformation response of reinforced concrete can be idealised as rigid-
perfectly plastic. This requires a sufficiently ductile and appropriately anchored, distributed and
detailed reinforcement as well as adequate concrete cross-sections [64, 35].
Membrane action is neglected.
While the flow of distributed and concentrated shear forces is analysed in detail the
associated dimensioning and detailing is based on established procedures [33].
5
2 Theory of plasticity
2.1 General
The central task of structural engineers is the design of safe and economical structures. Safety
not only implies resistance to external actions, but also ductility and robustness. Collapse should
be preceded by perceivable deformations and in the case of failure, damage should not extend to
the whole structure. Load resistance and ductility are simplified and summarised in the form of
rigid-plastic behaviour on the basis of the theory of plasticity. Corresponding limit analysis
methods have been applied for a long time, implicitly or explicitly, to solve engineering
problems.
Only in the 1950s the theory of plasticity was established on a sound basis, deviating
radically in its approach from that of the theory of elasticity. By considering the elastic-plastic
behaviour the transition between the two theories was smoothened [53, 37]; however, the
complexity of elastic-plastic analyses and the problems related to the identification of the real
state of stress in a structure meant the results were largely of academic interest. In practice the
theories of elasticity and plasticity are used independently and with different purposes:
serviceability limit states are checked based on the theory of elasticity whereas ultimate limit
state checks and design are based on the theory of plasticity.
The present chapter gives a brief summary of the theory of plasticity, focusing on a rigid-
perfectly plastic material behaviour. Starting from one-dimensional problems (Chapter 2.2) the
analysis is extended to general systems (Chapter 2.4). The theorems of limit analysis (Chapter
2.5) are the basis of corresponding limit analysis and design methods (Chapter 2.6). A
comparison of reinforced concrete behaviour limited to one-dimensional problems with the
rigid-perfectly plastic model (Chapter 2.3) completes the discussion.
2.2 Rigid-perfectly plastic behaviour
In its basic and simplest form, the theory of plasticity assumes a rigid-perfectly plastic material
behaviour. The uniaxial test shown in Fig. 2.1a is examined by considering the stress-strain
curve depicted in Fig. 2.1b. The load is increased starting from point O, i.e. from a virgin state.
Up to the yield point = f
y
no deformations occur (OA). Once A is reached arbitrary
deformations in the load direction are possible without any change in stress; the stress-strain
curve extends toward D. At B the bar is unloaded. The stress-strain point moves parallel to OA
to point C; the strain , i.e. the plastic deformation of the bar, remains constant. Continuing the
experiment in the compression direction, a negative deformation occurs as soon as the yield
stress = f
y
is reached (point E ); for the sake of simplicity, the yield stresses in tension and
compression are assumed to be of equal magnitude. In the second plastic phase the stress-strain
point moves along EFI. Again, the opposite yield stress may be reached by reversing the load
direction. For example, starting from G, the curve GHJD is obtained.
Theory of plasticity
6
Fig. 2.1: Rigid-plastic material: a) uniaxial test; b) stress-strain curve; c) yield criterion.

A rigid-plastic material stores no energy. The work done by the external forces is
completely dissipated in plastic deformation. Denoting by F the applied load and by u& the
incremental plastic deformation of the bar corresponding to a strain rate & (Fig. 2.1a), the
energy balance over the whole body (volume V ) is given by

= = =
V
D dV u F W & & (2.1)
The energy dissipated per unit volume equals & = dD . Considering Fig. 2.1b, it is clear
that dD is a function of & , i.e. dD = dD( & ), and that the multiplication of & with a positive
factor k leads to a multiplication of dD with the same factor: ) ( ) ( & & dD k k dD = . Hence, dD is a
homogeneous function of first degree and from Eulers theorem on homogeneous functions one
gets

& d
dD d ) (
= (2.2)
The theory of rigid-perfectly plastic bodies concentrates on collapse loads. No statement is
made on deformations, which may grow in an uncontrolled manner as soon as the collapse load
is reached. According to this observation the stress-strain diagram of Fig. 2.1b is simplified to
Fig. 2.1c. Stresses with
y
f < are sustained without deformation. For
y
f = a plastic
strain increment & in the load direction is possible. States of stress with
y
f > are not
admissible. Introducing the yield function
y
f Y = ) ( the three cases are differentiated by
the sign of Y: Y < 0 represents the aplastic domain, Y = 0 defines the state of yielding and Y > 0
is not admissible. Possible strain increments & at collapse can be represented by outward
normal vectors to the yield boundaries.
It should be noted that the above analysis does not change if and & are replaced by F and
u& , i.e. instead of local stresses and strain increments generalised stresses (loads) and strain
(deformation) increments can be used [32].
2.3 Reinforced concrete
Reinforced concrete is composed of concrete and reinforcement. The concrete may be assumed
to work purely in compression whereas the reinforcement is predominantly subjected to tension.
For a large range of applications it is sufficient to consider the uniaxial response of either
material.
a)


b) c)
O
D
F G

O
A B
C
E
I
H
J
y

F
F
u
f
y
f
y
f
y
f
Rigid-perfectly plastic behaviour Reinforced concrete
7
Fig. 2.2: Reinforced concrete: a) stress-strain curve for uniaxially loaded concrete; b) stress-
strain curve for uniaxially loaded reinforcement; c) behaviour of reinforced concrete.

Concrete
Fig. 2.2a shows the stress-strain curve for the concrete in compression. After a more or less
elastic phase (OA), the concrete progressively loses its stiffness (AD) and reaches the maximum
compressive load (D). Unloading (BC ) is approximately elastic, parallel to OA. After the peak
load the system softens (DE ). The collapse of the specimen is characterised either by reaching
the deformation capacity of the concrete or by instability resulting from the equilibrium between
the elastic stored energy and the fracture energy [44]. It can be seen that concrete in
compression can be approximated as a rigid plastic material, assuming a conservative yield
strength f
c
.
The tensile strength of concrete is usually neglected. It should be noted that the behaviour of
concrete in tension is not plastic; after opening a crack (OF ) this crack has to be closed (FO)
before compressive stresses can be carried. A perfectly plastic behaviour would mean that the
stress-strain point would move parallel to the -axis, vertically down from F until reaching
= f
c
.
Reinforcement
The stress-strain curve for the reinforcement is depicted in Fig. 2.2b. Steel bars in tension
exhibit an initial elastic behaviour (OA), followed by a yield plateau, i.e. a yield point at = f
y

beyond which the strain increases without any change in stress (AB), and a strain-hardening
range (BE ) until rupture occurs at the tensile strength = f
u
, corresponding to a strain
u
(E ).
Unloading at any point of the stress-strain diagram occurs with approximately the same stiffness
as for the initial loading (CD). Neglecting elastic deformations and strain-hardening one arrives
at a rigid-perfectly plastic behaviour. Excluding instability problems, e.g. buckling, steel
exhibits a similar behaviour in compression.
a)
F
u
F
y
c
b
a
c)

f
c
O F
A
B
C
D
E

f
y
f
u
O
A B
C
D
E
b)
F
u
l
d
b
A
s

u
O
A
B
C
D
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . .
. .
. . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .
. . .
. .
1
2
3
F
cr
.
.
.
.
.
Theory of plasticity
8
Reinforced concrete
The behaviour of reinforced concrete is considered for the example of a simply supported beam
subjected to a concentrated load F in the middle of the span l (Fig. 2.2c). The beam has a
rectangular cross-section with effective depth d and width b. The reinforcements cross-
sectional area A
s
is constant over the whole length, and corresponds to a reinforcement ratio
= A
s
/ (bd ). The load F as well as the associated deflection u are considered as generalised
load and deformation, respectively. Alternatively, the bending moment and the rotation at
midspan may be chosen as generalised stress and strain, respectively. Under monotonous
loading from zero the beam first behaves as a homogeneous, linearly elastic member (OA). At
point A the specimen cracks (cracking load F = F
cr
). This produces a permanent change in the
structure. Moving towards B the beam still behaves approximately elastically, but with a
reduced stiffness (cracked behaviour). Upon yielding of the reinforcement (B) the specimen
exhibits a behaviour which may be considered as a plastic plateau, F = F
y
(BCD). Up to the
peak load (C) the hardening of the reinforcing steel compensates the softening of the concrete.
At D the concrete reaches its ultimate strain and the beam collapses.
The behaviour described above corresponds to the case of a moderately reinforced concrete
beam ( 0.5%). Increasing the reinforcement ratio the ultimate load increases, but the
ductility of the beam decreases. Curve a depicts the load-deformation curve corresponding to
balanced failure; yielding of the reinforcement occurs together with rupture of the concrete. In
this curve the phase BCD disappears. Conversely, reducing the reinforcement ratio, the ductility
increases and the ultimate load decreases. Curve b depicts the situation for a beam containing
minimum reinforcement ( 0.15%); the loads at cracking and yielding (points A and B)
coincide and BCD reflects the reinforcement behaviour. Any further reduction of the
reinforcement ratio would lead to a brittle failure as soon as the cracking load would be reached
(curve c).
The following derives from the above considerations:
The behaviour of reinforced concrete can be characterised by an elastic uncracked (line 1), an
elastic cracked (line 2) and a plastic range (line 3).
The plastic behaviour of reinforced concrete is strongly influenced by the reinforcement
ratio. A ductile failure results for low and moderate reinforcement ratios for which the
concrete crushes while the steel is yielding.
The deformation capacity of reinforced concrete elements depends on the ductility of the
reinforcement (for low and moderate reinforcement ratios) as well as on the ultimate concrete
strain (for high reinforcement ratios); whereas the latter can be improved by detailing
measures (e.g. by confinement) sufficient ductility properties of the reinforcement are of
utmost importance for the soundness of concrete structures.
2.4 Yield condition and flow rule
The plasticity considerations described above are generalised in the following to an n-
dimensional problem.
Generalised stresses
1
,
2
,,
n
and associated generalised strain increments
1
& ,
2
& ,,
n
&
of a structure are considered [32]. The values of
i
& are zero while the system remains rigid.
Finite values of
i
& correspond to an arbitrary collapse state.

Reinforced concrete Yield condition and flow rule
9
Fig. 2.3: Limit analysis: a) yield surface and plastic strain increment; b) singularities.


The rigid-perfectly plastic system does not store energy. Thus, the work of the external
forces equals the dissipated energy, i.e.
& = =W D (2.3)
Eq. (2.2) is generalised by
) (D grad = (2.4)
where D =
n n
& & & + + + ...
2 2 1 1
.
The generalised stresses at the ultimate limit state are assumed to be governed by the yield
condition
0 ) ( =
i
Y (2.5)
Depicting (2.5) in a (
1
,
2
,,
n
)-coordinate system a yield surface as shown in Fig. 2.3a is
obtained. Similar to the one-dimensional case the yield condition (2.5) isolates the aplastic
domain (enclosed area, Y(
i
) < 0) from the inadmissible region (Y(
i
) > 0).
Generalised strain increments compatible with a limit state are assumed to be outward-
directed, orthogonal to the yield surface, i.e.
Y grad k = & (2.6)
where & represents the strain increment vector and k denotes an arbitrary, non-negative factor.
Eq. (2.6) represents the normality condition or associated flow rule.
Since the generalised stress state
i
= 0 (i = 1, 2,,n) is admissible, the yield surface must
enclose the origin O of the coordinate system.
The yield surfaces must be at least weakly convex. Special cases are given by a plane
surface or a point of singularity [29] (Fig. 2.3b). If the yield surface contains plane parts (AB) an
infinite number of stress states correspond to the same strain increment. However, the value of
D is equal for all points of the plane part if the same strain increment is considered. At an apex
or edge [29] the generalised strain increment can lie arbitrarily between the normal directions
determined on the areas adjacent to the discontinuity (C ). In such a situation, a single stress
state corresponds to an infinite number of strain increments. However, D is still uniquely
determined by & .

a) b)
O
*
Y<
Y=
O
A B
C
0
0
AB
C
C
A-B
,
Theory of plasticity
10
2.5 Theorems of limit analysis
The theorems of limits analysis result from the application of the principle of virtual work to a
rigid-perfectly plastic system, whose behaviour is summarised by the principle of maximum
energy dissipation.
The theorems of limit analysis are credited to Gvozdev [13], Hill [17], Drucker, Greenberg
and Prager [6, 7] and Sayir and Ziegler [60].
Principle of virtual work
In an arbitrary mechanical system, the total work of the internal and external forces (including
any internal and external reactions as well as inertial forces) disappears for any admissible
virtual motion.
For static systems one gets
& & = u F (2.7)
where F and denote an equilibrium set of external and internal forces (generalised loads and
stresses), respectively, and u& and & denote an arbitrary associated compatibility set of external
and internal displacement increments (generalised deformation and strain increments),
respectively.
Principle of maximum energy dissipation
The stresses corresponding to given strain increments assume such values that the dissipation
becomes a maximum [65], i.e.
0 ) ( &
*
(2.8)
where is the actual generalised stress state at the yield surface corresponding to & and
*
is
any generalised stress at or within the yield surface (Fig. 2.3a).
Eq. (2.8) is trivially verified with the help of Fig. 2.3a, where the associated flow rule is
assumed to be valid. Alternatively, the principle of maximum energy dissipation may be
postulated and the associated flow rule follows from it [32].
Theorems of limit analysis
Excluding instability problems and combining the principle of virtual work with the principle of
maximum energy dissipation the following theorems are obtained:
Lower-bound theorem: any load F
s
corresponding to a statically admissible state of stress
everywhere at or below yield is not higher than the ultimate load F
u
.
Upper-bound theorem: any load F
k
resulting from considering a kinematically admissible state
of deformation, setting the work done by the external forces equal to the internal energy
dissipation, is not lower than the ultimate load F
u
.
Compatibility theorem: any load for which a complete solution, i.e. a statically admissible
state of stress everywhere at or below yield and a compatible, kinematically admissible state of
deformation can be found, is equal to the ultimate load.
A state of stress is statically admissible if it fulfils the equilibrium and static boundary
conditions. A state of deformation is kinematically admissible if it fulfils the kinematic relations
and boundary conditions. A state of stress and a state of deformation are compatible if they are
related via the associated flow rule (2.6). In a complete solution, the states of stress and
deformation only have to be compatible in the sense stated above.
Theorem of limit analysis Limit analysis and design methods
11
2.6 Limit analysis and design methods
By applying the lower- and upper-bound theorems according to the static and kinematic
methods of limit analysis lower- and upper-bound values F
s
and F
k
, respectively, are determined
for the ultimate load F
u
, see Fig. 2.4:

k u s
F F F (2.9)
The static method is based on the consideration of statically admissible stress fields which
nowhere violate the yield conditions of the system. The system stays aplastic; no statement can
be made about deformation and failure. The static method is suitable for design. The strength of
the system is determined following the flow of the forces in accordance with the selected
equilibrium solution.
The kinematic method is based on the consideration of kinematically admissible failure
mechanisms. In contrast to the static method, only the stresses at locations of plastic
deformation are of interest. No statement can be made about the flow of the forces and local
equilibrium. Rather than for the detailed design the kinematic method is suitable for the overall
analysis of structures.
By combining the static and the kinematic method based on the compatibility theorem, a
design approach is obtained which shall be called compatibility method. According to this
method complete solutions are considered, i.e. equilibrium solutions compatible with
kinematically admissible failure mechanisms. Complete solutions are strictly related to the yield
criteria assumed. By adjusting the ultimate resistance distribution in the system, equilibrium
solutions may transform into a mechanism and, on the other hand, failure mechanisms may be
made safe by associating an equilibrium solution and strengthening the weak zones. The
compatibility method corresponds to the method of capacity design used in earthquake
engineering. The design is based on a chosen failure mechanism, followed by supplying
sufficient strength to the rigid parts of the system.


Fig. 2.4: Relationships between limit analysis and design methods.
Compatibility
theorem
Lower-bound
theorem
Upper-bound
theorem
F
s
STATIC KINEMATIC COMPATIBILITY
F
u
F
k
"Design" "Analysis"

12


13
3 Static method
3.1 General
The lower-bound method for reinforced concrete beams [34] is based on truss models (strut and
tie models) or in a continuum form on stress fields. Dimensioning and detailing are
performed following the flow of the shear forces, considering truss models or associated stress
fields which describe the beams structural behaviour.
Reinforced concrete slabs are designed on the basis of an elastic analysis, the strip method
or, where available, a limit analysis solution. Each of these methods corresponds to a particular
application of the equilibrium conditions. The elastic analysis relates the stresses to the strains
and deflections by means of Hookes law and Kirchhoffs hypothesis; the equilibrium and the
boundary conditions evolve into a differential problem focused on the deflection function. In its
common form, the strip method reduces the slab to an orthogonal set of beams; this corresponds
to neglecting twisting moments in the equilibrium conditions. In limit analysis solutions the
equilibrium conditions are combined with the failure conditions according to a fixed
reinforcement distribution, i.e. the corresponding yield criteria and the associated flow rule
(Chapter 4); the known solutions are restricted to a few simple cases.
Basically, beams are special cases of slabs. However, when comparing the design methods
for beams and slabs, no direct link is recognised. The analyses concentrate on two different
tasks: the load transfer for beams, and the consequences of the load transfer, i.e. the moments,
for slabs. The lower-bound procedure for beams is more flexible and precise. Slab design
methods are an application of equilibrium considerations, but they do not match beam design
methods in their depth of modelling the structural behaviour.
The goal of the static or lower-bound method presented in the following is to provide the
foundation for a consistent design of reinforced concrete slabs analogous to that of beams. The
analysis focuses on the load transfer, whereby two types of loads are distinguished distributed
and concentrated. Distributed load transfer is governed by the generalised strip method, i.e. a
continuous truss model within and at the top and bottom surfaces of the slab, defined by the
principal shear and by the principal moment trajectories, respectively. Concentrated load
transfer corresponds to statical discontinuity lines. Finally, the stress field approach for beams is
extended to slabs, i.e. stress fields according to the generalised strip method and discontinuity
lines are put together like pieces of a puzzle to allow for the load transfer from the interior of a
slab to its supports.
3.2 Internal forces
The static analysis of slabs is performed with generalised stresses, i.e. stress resultants on
vertical strips of unit width. The definition of internal forces in arbitrary sections (Chapter 3.2.1)
is followed by the associated stress field definition (Chapter 3.2.2).
Static method
14
3.2.1 Definition of internal forces
A slab is a thin structural member bounded by two parallel planes, loaded perpendicularly to the
planes. The right-handed Cartesian coordinate system O(x,y,z), O being an arbitrary point on the
middle surface and z being vertical downwards, is introduced as the global system of reference.
The plane vector n in P, Fig. 3.1a, defines a vertical strip section, where the direction n is
normal and outwards to the strip considered. This direction and the directions t and z constitute
the local coordinate system P(n,t,z). At any point of the strip, e.g. at z = z
0
according to
Fig. 3.1a, a normal stress
n
and shear stresses
tn
and
zn
are acting. These stresses are positive
in the (positive) n-, t- and z-directions, respectively.
The internal forces and moments are obtained by integrating the stresses over the vertical
strip for a unit length in the t-direction and for h / 2 < z < h / 2, h being the slab thickness. They
are composed of the membrane forces, i.e. the axial force n
n
and the in-plane shear force n
tn

(neglected in the following, see Chapter 1.3)

=
2 /
2 /
h
h
n n
dz n ;

=
2 /
2 /
h
h
tn tn
dz n [kN / m] (3.1)
the shear force

=
2 /
2 /
h
h
zn n
dz v [kN / m] (3.2)
and the bending and twisting moments

=
2 /
2 /
h
h
n n
dz z m ;

=
2 /
2 /
h
h
tn tn
dz z m [kN] (3.3)
The internal forces and moments are indexed as the stress components: the first index
represents the direction of the stress and the second represents the section (normal vector)
considered. For coincidental stress and section directions only one index is used. Fig. 3.1b
represents the positive generalised stresses in P.
3.2.2 Stress field definition
Introducing generalised stresses, the stress state of the slab is defined by a plane shear vector
and a 2x2 moment tensor.
Two different sections n
1
and n
2
in P with internal forces v
n1
, m
n1
, m
tn1
and v
n2
, m
n2
, m
tn2
are
considered. An arbitrary section n
3
may be resolved into the sections n
1
and n
2
. Starting from a
unit strip length n
3
, the internal forces in this direction result from the sum of the forces on the
projected strips in the directions n
1
and n
2
(Fig. 3.1c). The shear force v
n3
is determined by

) cos(
) cos(
cos
) cos(
12 0
13 0 2
0
13 0 1
3

=
n n
n
v v
v (3.4)
where

12
12
1
2
0
sin
1
) cos ( tan

=
n
n
v
v
(3.5)
The moments m
n3
and m
tn3
are given by
13 1 13 1 3
2 sin 2 cos ) (
tn c n c n
m m m m m + + =
13 1 13 1 3
2 cos 2 sin ) (
tn c n tn
m m m m + =
(3.6)
Internal forces
15
Fig. 3.1: Slab section: a) stresses; b) stress resultants; c) stress transformation; d) Thales
circle for shear force transformation; e) Mohrs circle for moment transformation;
f) variation of v
n
; g) variation of m
n
and m
tn
.
a) b)

t
z
n

z
v

m
m
1
P
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
tn
zn
n
z
0
z
x
y
P
n
t
tn
n
n
c)
v
m
m
P
n
t
n
n3
3
3
3
1
v
m
m
n
t
n
n1
1
1
1
v
m
m
n t
tn
n
n2
2
2
2 2
n
3
n
3

13
.

23
.
= +
sin( )
23
sin( + )
13

23
sin( )
13
sin( + )
13

23
d) e)
.
.
.
.
v
n1
v
0
v
n2
v
n3

13

12

23
1
2
m
n
m
tn
1
2
m
c
Pole
m
1
m
2
r

/2

0
v
0
v
n
3 1

/2

1
m
1
m ,
n
3 1
m
tn
f) g)
2
m
n
m
tn
m
2
/4
n
3
n
n
n
n
n
3
n
2
n n n
2
n n
2
tn3
tn1
Static method
16
where

|
|
.
|

\
|

+ + =
1 2
2
2
2
1
2 1
2
1
n n
tn tn
n n c
m m
m m
m m m (3.7)
and
ij
indicates the clockwise angle between the directions i and j. For the case
12
= / 2, i.e.
orthogonal n
1
and n
2
, Eqs. (3.4) to (3.7) reduce to

13 2 13 1 3
sin cos
n n n
v v v + = (3.8)
13 1 13
2
2 13
2
1 3
2 sin sin cos
tn n n n
m m m m + + =
13 1 13 13 1 2 3
2 cos cos sin ) (
tn n n tn
m m m m + =
(3.9)
Figs. 3.1d and 3.1e represent (3.4) and (3.6) graphically. When denoting by v
n
one side
length of a right angled triangle, Eq. (3.4) indicates that different shear forces correspond to the
same base, i.e. the Thales circle diameter. This extreme shear is called the principal shear v
0
,
while its direction
0
is the principal shear direction [33]. Orthogonal to
0
, the circle tangent
determines v
0
= 0. Eq. (3.6) is the parametric equation of a circle in the (m
n
,m
tn
)-space with the
angle
13
(or 2
13
) as parameter. The circle is centred at C(m
c
,0) and has the radius
2 / 1 2
1
2
1
) ) ((
tn c n
m m m r + = . The moments in the direction n
3
are obtained by clockwise rotation
of the radius in (m
n1
,m
tn1
) by the angle 2
13
. This graphical method is attributed to Mohr [39].
The extreme values of the bending moment, i.e. the principal moments, result for m
nt
= 0 and are
given by m
1
= m
c
+ r and m
2
= m
c
r. The corresponding directions, i.e. the principal directions,
are indicated by
1
and
2
( =
1
+ / 2), respectively. The twisting moment has a maximum in
the directions
1
+ / 4 and
1
+ 3 / 4; m
tn
reaches the values r and is accompanied by the
bending moment m
n
= m
c
.
Figures 1f and 1g show the variation of the internal forces for 2 0
13
. The variation of
v
n
is harmonic with amplitude v
0
in the direction
0
and with period 2, see Fig. 3.1f. The
variation of m
n
is harmonic between m
1
in the direction
1
and m
2
in the direction
1
+ / 2 with
a period of . Finally, the variation of m
tn
has an amplitude (m
1
+ m
2
) / 2 and follows the m
n
-
curve with a phase difference of / 4, see Fig.3.1g. Any stress field is defined by five quantities.
By selecting the principal sections, these are distinguished by the statical values v
0
, m
1
, m
2
, and
the geometrical values
0
,
1
.
3.3 Equilibrium
The equilibrium conditions result from the comparison of the internal forces at two
neighbouring points, considering the effect of external loads, Chapter 3.3.3. Therefore, the
internal forces have to be defined with respect to a fixed system of reference, Chapter 3.3.2.
This is chosen in a general form using orthogonal curvilinear trajectories, Chapter 3.3.1.
3.3.1 Orthogonal curvilinear coordinates
A point P of a slab is defined in Cartesian coordinates by the intersection of the lines x = x
P
and
y = y
P
. These lines also define the y- and x-axes in P, which are positive in the increasing y
P
-
and x
P
-direction.
Similarly, a slab point in curvilinear coordinates is defined by the intersection of the curves
u = u
P
and v = v
P
, where u and v are functions of x and y:
Internal forces Equilibrium
17
) , ( ; ) , ( y x v v y x u u = = (3.10)
The curvilinear axes u and v in P are the curves v(x,y) = v
P
= const, and u(x,y) = u
P
= const,
positive in the increasing v
P
- and u
P
-direction, respectively. The orthogonality requirement for
the curvilinear coordinates is given by
0 = v u (3.11)
where ) / , / ( y x .
The u- and v-axis directions in P are specified introducing the unit vectors n and t tangential
to the u- and v-coordinates; P(n,t,z) constitutes a right-handed rectangular coordinate system.
With respect to O(x,y,z), the orientation of P(n,t,z) is indicated by the clockwise angle
between x and n.
In contrast to the Cartesian axes, curvilinear axes do not express a fixed length and a fixed
direction. A change du (dv) corresponds to an arc element length measured along the coordinate
line u (v) of du A dS
u u
= ( dv A dS
v v
= ) with

2 2
2
|
.
|

\
|

+ |
.
|

\
|

=
u
y
u
x
A
u
(
2 2
2
|
.
|

\
|

+ |
.
|

\
|

=
v
y
v
x
A
v
) (3.12)
Starting from the angle , given by

v A
y
u A
x
v u

= cos ;
v A
x
u A
y
v u

= sin (3.13)
a change du (dv) results in an n-direction variation of
du
v A
A
d
v
u

= ( dv
u A
A
d
u
v

= ) (3.14)
Combining (3.12) with (3.14) one obtains the radii of curvature of the curvilinear axes u and v

v A A
A
u A
v u
u
u u

1
;
u A A
A
v A
v u
v
v v

1
(3.15)
The notation introduced is summarised in Fig. 3.2a.
3.3.2 Sign convention
Previously, the internal forces and moments have been related to a local coordinate system of
the cross-section considered and the stress field at a point has been established from the stress
resultants on two different cross-sections. Equilibrium considerations require the definition of
internal forces corresponding to a fixed system of reference. Referring to the system (n,t,z) in P,
the sign convention assumed is defined in Fig. 3.2b. Shear forces are positive if related to
positive shear stresses in the z-direction and the moments are positive if related to positive
(negative) stresses in the slab portion z > 0 (z < 0). On the sections with negative n- and t-
directions, internal forces are in equilibrium with those in the positive one. Note that the
symmetry of the stress tensor (i.e. rotational equilibrium) leads to m
tn
= m
nt
.
3.3.3 Equilibrium conditions
In a curvilinear coordinate system (3.10) the stress field at the point P
1
(u
P1
,v
P1
) is compared with
that at the neighbouring point P
2
(u
P1
+ du, v
P1
+ dv). The shear forces v
n
and v
t
, the bending
moments m
n
and m
t
, and the twisting moments m
tn
and m
nt
act on the n- and t-sections in P
1
.
Static method
18
Moving an increment dS
u
along the u-axis, the n-direction rotates by (3.14
1
) and the internal
forces and moments are
du v v
u n n ,
+ ; du m m
u n n ,
+ ; du m m
u tn tn ,
+ (3.16)
Similarly, moving an increment dS
v
along the v-axis, the t-direction rotates by (3.14
2
) and the
internal forces and moments are
dv v v
v t t ,
+ ; dv m m
v t t ,
+ ; dv m m
v nt nt ,
+ (3.17)
A comma preceding an index indicates partial derivation with respect to the coordinate given as
index. The relations (3.16) and (3.17) define the stress field in P
2
.
With reference to Fig. 3.2c the slab element P
1
AP
2
B is considered. Assuming that the stress
in P
1
and P
2
does not change on the sides P
1
A, P
1
B and P
2
A, P
2
B, the following equilibrium
conditions can be formulated:
0
, ,
= + + + + q
v v
v v
v
n t
Sv t Su n
u



n t n
v
tn nt
u
Sv nt Su n
v m m m m m m = + + + + ) (
1
) (
1
, ,

(3.18)

t nt tn
v
n t
u
Sv t Su tn
v m m m m m m = + + + + ) (
1
) (
1
, ,


Eqs. (3.18) are the equilibrium conditions in curvilinear coordinates. The three equations
may be combined by inserting (3.18
2
) and (3.18
3
) into (3.18
1
). In Cartesian and in polar
coordinates, the equilibrium conditions result as special cases of (3.18) when considering u = x,
v = y , 1 /
u
= 1 /
v
= 0, and u
2
= r
2
= x
2
+ y
2
, v = = tan
1
( y / x ), 1 /
u
= 0,
v
= r , respectively.
3.4 Load transfer
Shear in slabs may arise in distributed [33] or concentrated form [4]. These two load transfer
modes are at the centre of the following analysis. In both cases, the requirements for the load
transfer are derived by considering the internal flow of forces in an infinitesimal region.
3.4.1 Distributed load transfer
As stated in Chapter 3.2.2, the shear forces v
n
and v
t
correspond to the principal shear force
5 . 0 2 2
0
) (
t n
v v v + = being transferred in the direction ) / ( tan
1
0 n t
v v

= to the n-axis, see Fig. 3.1d.


Denoting the principal shear trajectory by u
s
, it can be seen that along the length
S
u
dS a moment
increment

s
u
dS v m
0
= (3.19)
is required for equilibrium (Fig. 3.3a). As in a beam, moments are subordinated to the shear.
Eq. (3.18
2
) in the u
s
-v
s
-system of reference states that the moment change results from the
incremental values
s
u n
m
,
and
s
v nt
m
,
or from the trajectory geometry
S
u tn nt
m m / ) ( + and
S
v t n
m m / ) ( , see Chapter 3.6.


Equilibrium Load transfer
19
Fig. 3.2: Equilibrium analysis: a) orthogonal curvilinear coordinates, notation; b) sign
convention for stress resultants; c) slab element in orthogonal, curvilinear
coordinates.
v n
t
q
u
x
n t
v (u=u )
u

,u
.
P
n
t
z v m
tn n
m
n
v
m
nt
t
m
t
v m
tn n
m
n
v
m
nt
t
m
t
a)
b)
c)
.
.
x
y
z
u

v
m
tn
n
m
n
m
n,u
du
+
v
n,u
du
+
m
tn,u
du
+
v
m
tn
n
m
n
v
m
nt
t
m
t
m
nt
m
nt,v
dv
+
v
t
v
t,v
dv
+
m
t
m
t,v
dv
+
y
z
dS
v
dS
u
1 dS ( )
v
dS
u
v

+
1 dS ( )
u
dS
v
u

+
dS
u
dS
v
du
,v
dv
P
u (v=v )
P
x
y
x=x
P
y=y
P
P
1
P
2
A B
Static method
20
3.4.2 Concentrated load transfer
Assuming a concentrated shear force V being transferred within a narrow zone S of width t
s

between the slab regions A and B, Fig. 3.3b depicts a part of the shear trajectory as a free body
diagram. The internal forces (3.2) and (3.3) act between the shear strip and the adjacent slab
segments, A and B. Within the strip, the shear force V is accompanied by a concentrated bending
moment M and a strip-specific line load q . Equilibrium requires that
0
,
= + + q v v V
B
n
A
n t
(3.20)
V m m M
B
tn
A
tn t
= +
,
(3.21)
M m m
S
B
n
A
n
= ) ( (3.22)
Equation (3.21) allows for both shear line action and strong band action. A shear line results
in the case t
s
= 0, thus M = 0; concentrated shear forces are transferred thanks to a jump in the
twisting moments; some vertical reinforcement is required to provide the shear resistance. A
strong band results if
B
tn
A
tn
m m = . The strip t
s
has to be extra reinforced to resist the moment M.
Both load transfer modes may be exploited to transfer the shear forces from a slab segment A to
an adjacent segment B.
Setting V, q , M and t
s
equal to zero, the equations (3.20) to (3.22) reduce to the continuity
requirements for the stress fields in the slab segments A and B. The concentrated shear transfer
equations define in a general form the statical discontinuity in slabs: adjacent stress fields have
to provide continuity in the load transfer and in the moments.
3.4.3 Remarks
Distributed load transfer is inherently included in the equilibrium considerations of slabs. In
comparison to (3.18), Eq. (3.19) shows the essential structural behaviour; it plays a central role
in the development of continuous stress fields (Chapter 3.6.1).
Concentrated load transfer along lines of statical discontinuity represents a fundamental tool
of plastic analysis (Chapter 3.6.7). In the past, stress fields for slabs were first developed based
on the theory of elasticity [28]. Since elastic solutions correspond to a distributed load transfer
(Chapter 3.6.2) discontinuities occur only at slab boundaries (Chapter 3.5). Plastic solutions are
based on both load transfer modes, i.e. lines of discontinuity may also arise within the slab.
Johansen [21] first postulated the existence of internal shear zones. Hillerborg [18] mentioned
that discontinuity in the twisting moments generates a shear flow, but he considered only load
transfer in strong bands. The research on nodal forces in the 1960s (Chapter 5.3.2) recognised
the importance of shear zones without further developing them. Clyde [4] suggested that nodal
forces are real forces, identifying them as concentrated shear forces. Stress fields which
included statical discontinuities were developed by Rozvany [56], Morley [42, 43], Clyde [5],
Fox [9, 10] and Marti [36]. Recently, Meyboom and Marti [38] completed an analytical study
with an experimental verification of the static discontinuity behaviour.
3.5 Boundary conditions
Slab boundaries are statical discontinuities of the stress field. The statical requirements at slab
edges are an expression of a concentrated load transfer. S is a narrow boundary strip and the
forces with upper index B correspond to the reactions.
Load transfer Boundary conditions
21

Fig. 3.3: Shear transfer: a) principal shear v
0
and corresponding moment change m; b) strong
band or shear line.

Clamped edge
Along a clamped edge (Fig. 3.4a), the reaction forces equilibrate the internal boundary forces v
n
,
m
n
, m
tn
. When introducing a discontinuity, the reaction force values become redundant.
Simply supported edge
Replacing
B
n
v by the reaction force r
n
of a simply supported edge (Fig. 3.4b,
n
B
n
r v = ) and
neglecting the superfluous parameters
B
n
m ,
B
tn
m , M and q , Eqs. (3.20) to (3.22) are reduced to
) (
,t tn n n
m v r + = ;
tn
m V = ; 0 =
n
m (3.23)
Introducing principal moments into Eqs. (3.23
2
) and (3.23
3
) one obtains

) tan(
) tan(
1
1
1 2
n
n
m
m V

= = (3.24)
V corresponds to Johansens nodal force [21] when assigning the value of m
u
to the principal
moment m
2
.
Acting as a shear line, the simply supported edge has the ability to introduce into slabs a
twisting moment in addition to a shear force. Kirchhoff [28] first determined (3.23
1
) applying
the principle of virtual work for elastic slabs. Thomson and Tait [63] gave a statical explanation
of the behaviour of the simply supported edge. Exploiting St-Venants principle they replaced
the twisting moment at the edge by a force couple of continuous distribution. The corresponding
change m
tn,t
adds to the reaction force v
n
resulting from the shear.
Free edge
Free edges (Fig. 3.4c) have no reactions at the boundary, i.e. 0 = = =
B
tn
B
n
B
n
m m v . In the case of a
straight boundary (
S
) Eqs. (3.20) to (3.22) are reduced to
0
, ,
= + q m M v
t tn tt n
;
tn t
m M V =
,
; 0 =
n
m (3.25)
where q indicates a boundary line load.
S
m
tn
A
m
n
A
q
v
n
B
m
n
B
m
tn
B
n
V
M+
v
n
A
t
dM
V+dV
M
t
1
m
v
u
s

0
1
S
0 dS
u
n
t
s
a) b)
S

v
s
Static method
22
Fig. 3.4: Boundary conditions and boundary shear transfer: a) clamped edge; b) simply
supported edge; c) free edge; d) point support.

The free edge exhibits a shear line-strong band character. For M = 0 the free edge is similar
to a simply supported edge with r
n
= 0. With additional reinforcement the load carrying capacity
may be increased. The shear force transferred by the strong band adds to that of the shear line.
Point support
Loads may be transferred to a point support (Fig. 3.4d) in distributed or concentrated form.
Summing both contributions one obtains


+ =

0
0
lim rd v V R
r
r
k
i
i
(3.26)
where k indicates the number of concentrated load paths leading to the support and r is the slab
perimeter at a distance r from the point support, e.g. = 2 for a slabs internal support or
= / 2 for a corner support of a square slab.
3.6 Lower-bound method
Lower-bound approaches suggest criteria to eliminate the statical redundancy of structures.
Considering beams and frames, statically determinate truss models describe the equilibrium
state [55]. In a continuous form, using the stress field approach [45] internal forces are followed
by matching basic stress fields along statical discontinuities.
Slabs are characterised by an infinite internal statical indeterminacy. According to Chapter
3.4 loads in slabs may be transferred in distributed or in concentrated form. The statical
discontinuities define the boundaries of the segments within which loads are transferred in
distributed form.
n
1
r
n
t
n
m
tn
m
n
R

V
1
V
2
V
i
V
k
V V+dV
a) b)
c) d)
1
r
t
n
n
V V+dV
1
t
Boundary conditions Lower-bound method
23
Starting by considering the distributed load transfer, a statically determinate continuous
truss model adequate for slab analysis is defined with the principal shear force and the principal
moment trajectories. The shear field is determined with vertical equilibrium in the principal
shear force direction. Similarly, the moment field is obtained by formulating equilibrium in the
principal moment directions. Including curvilinear principal moment trajectories, the procedure
becomes an extension of Hillerborgs strip method [18], in which the principal moment sections
are limited to Cartesian coordinates, hence the designation generalised strip method.
On the basis of the new method, the lower-bound analysis of slabs is reviewed in Chapters
3.6.2 to 3.6.6, distinguishing an elastic approach, Hillerborgs method and the Hencky-Prandtl
solutions. Finally, the analysis of curvilinear trajectories on conical sections, together with the
superposition principle, leads to known limit analysis solutions. Similar to the stress field
approach for beams, a stress field approach for slabs is developed fitting the continuous stress
fields into the slab with connections along lines of discontinuity.
3.6.1 Generalised strip method
The generalised strip method results from a direct application of the distributed load transfer
requirements, cf. Chapter 3.4.1. Establishing equilibrium with respect to chosen principal shear
and moment trajectories, loads are firstly integrated to shear forces and secondly to moments.
Shear field
In the orthogonal curvilinear coordinate system u
s
, v
s
, with u
s
corresponding to the principal
shear trajectory, Eq. (3.18
1
) is reduced to
0
0
, 0
= + + q
v
v
s
s
v
Su

(3.27)
Assuming a certain shear flow, i.e. the geometry of the shear field, Eq. (3.27) can be
integrated. The boundary conditions of the shear problem require one value of v
0
along each
trajectory u
s
. Note that arbitrary load distributions are included in the analysis by considering q
as a function q(u
s
,v
s
). Concentrated loads correspond to the homogeneous problem q = 0.
Moment field
Introducing the principal moments m
1
and m
2
with their trajectories u
m
and v
m
, Eqs. (3.18
2
) and
(3.18
3
) are reduced to

m
m
m m
m
m
t
u
Sv n
v
Su
v m m m v m m m = + = + ) (
1
; ) (
1
1 2 , 2 2 1 , 1

(3.28)
The shear force components
m
n
v and
m
t
v of the shear field in the directions u
m
and v
m
,
respectively, are determined according to Eq. (3.8). Eqs. (3.28) generally involve a complex
analysis and do not always lead to an explicit solution; additionally, agreement with the slabs
boundary conditions is not always possible.
3.6.2 Elasticity
Most engineers design slabs according to elastic solutions. Basically, elastic solutions are not
simple to apply. However, a quick procedure results from the use of FE-software. In the
following, the generalised strip method is exploited to extend the elastic analysis of beams to
slabs.
Elasticity theory is based on Hookes law: ut tension sic vis (the extension is proportional
to the force): E = , where E is the modulus of elasticity of the material. During bending,
Static method
24
vertical cross-sections remain plane, so that they undergo only a rotation with respect to the
neutral axis (Bernoullis hypothesis); z = . The elongation of each fibre is proportional to
the distance z from the neutral axis. The proportionality constant is the curvature of the
deflection curve w. For small deflections in comparison to the span of the beam one gets
=
uu
w
,
, u being the bending direction. The bending moment results by integrating the
normal stresses acting on a cross-section, i.e.
uu
EIw M
,
= where EI is the flexural rigidity of
the beam. By introducing considerations of equilibrium
u
M V
,
= and
u
V q
,
= elastic beams
are described by the equation ) /(
,
EI q w
uuuu
= .
In a two-dimensional continuum the elastic stress-strain relation is given by
1
= (
1

2
) / E
and
2
= (
2

1
) / E, where denotes Poissons ratio. For a slab, assuming Bernoullis
hypothesis along the principal curvature trajectories u and v of the deflection function w, the
principal curvatures are given by
uu
w
, 1
= and
vv
w
, 2
= . Principal moments result by
integrating the stresses in unit vertical strips in u- and v-direction:
uu
Dw m
, 1
= and
vv
Dw m
, 2
= , respectively, D = Eh
3
/ [12 (1
2
)] being the flexural rigidity of the slab.
Introducing m
1
and m
2
into (3.28) and substituting the resulting shear components in Eq. (3.18
1
),
the elastic slab analysis simplifies to D q w / = .
Summarising, the generalised strip method extends the elastic analysis of beams to slabs by
considering the principal curvature trajectories of the deflection function w as elastic beams.
Within the whole slab loads are transferred in distributed form since the basic principal
trajectories extend over the whole slab to the supports.
3.6.3 Strip method
Hillerborgs method is based on the Cartesian system defined by the coordinates x and y. The
analysis starts by dividing the load into single portions referred to the coordinate directions
) , ( ) , ( ) , ( y x q y x q y x q
y x
+ = (3.29)
The shear field is defined by the components v
x
and v
y
. These result by integrating the respective
load portion as in a beam, i.e.
) ( ) , ( y C dx y x q v
sx x x
+ =

; ) ( ) , ( x C dy y x q v
sy y y
+ =

(3.30)
Eqs. (3.30) are additionally integrated to give the moment functions
) ( ) ( ) , (
2
y C y C x dx y x q m
mx sx x x
+ + =


) ( ) ( ) , (
2
x C x C y dy y x q m
my sy y y
+ + =


(3.31)
Hillerborgs method reduces the slab to two sets of beams at right angles to each other. Each
coordinate line (x = x
P
or y = y
P
) defines a beam. According to (3.30) and (3.31), on each beam a
shear integration constant, C
xs
(y
P
) or C
ys
(x
P
), and a moment integration constant, C
xm
(y
P
) or
C
ym
(x
P
), are available to fulfil the boundary conditions.
Concerning the shear field analysis, the load split (3.29) proposed by Hillerborg is
equivalent to the selection of the principal shear trajectories adopted in the generalised strip
method. Despite some analytical difficulties in the solution of (3.27) compared to (3.30), the
shear field analysis of the generalised strip method has the advantage of a deliberate choice of
the load path throughout the slab. In addition, the principal shear force allows a better control
against shear failure than the shear components. The main difference between Hillerborgs
method and the generalised strip method concerns the moment field analysis. In Hillerborgs
method principal moment trajectories are fixed in the Cartesian directions. By contrast, the
generalised strip method allows a free choice. With straight principal moment directions, the
geometrical load-carrying capacity resulting from curved principal sections (i.e. (m
1
m
2
) /
v
,
Lower-bound method
25
(m
2
m
1
) /
u
, cf. Eqs. (3.28)) disappears. Such geometrical contributions fill the gap between the
lower-bound solutions of the strip method and the complete solutions of slab limit analysis.
3.6.4 Hencky-Prandtl solutions
In contrast to the strip method, where the geometrical load transfer contribution is neglected
(1 /
x
= 1 /
y
= 0), the following analysis involves a load transfer with constant principal
moments
.
1
const m m
u
= = ; .
2
const m m
v
= = (3.32)
In an unknown net of principal moment trajectories Eqs. (3.28) are given by

m m
v n
m v / = ;
m m
u t
m v / = (3.33)
where
2 1
m m m = . In the same system of reference, Eq. (3.18
1
) together with (3.14) and
(3.15) result in the equilibrium condition
0 2
,
= + q m
m m
Sv Su
(3.34)
where the function (u
m
,v
m
) defines the moment net (Chapter 3.3.1), completing the stress field
of the slab. The boundary conditions of (3.34) have to determine the orientation of the principal
moments at the boundary.
In the special case q = 0 Eq. (3.34) is similar to the equations of the plane strain problem
[9, 14, 15, 16, 3]. The inhomogeneous problem requires a much more complex analysis [10].
Note that for m
1
= m
2
, hence 0 = m , the slab cannot carry any loads, and that q is proportional
to m .
3.6.5 General stress fields
The possibility of freely choosing principal trajectories, established in mathematical form by the
generalised strip method, relates the slab analysis to a flow field theory and its applications,
similar to electric flow fields, magnetic fields, water flow, membrane deflection [31, 57], natural
shapes of shells [54], Chladni-plates, etc.
The mathematical complexity of the solution of the generalised strip method equations as
well as the restrictions imposed by the boundary conditions hinder the development of stress
fields with curvilinear principal shear and moment trajectories. In this context a stress field
library is introduced as an intermediate step between the basic equations and the practical
statical analysis of slabs. Families of curvilinear coordinates, load distributions and boundary
conditions lead to a systematic approach in the equilibrium analysis of slabs. In the following,
the basic steps of the generalised strip method are applied for the analysis of curvilinear
coordinates given by conical sections for the case of a uniform load distribution.
Curvilinear coordinates and geometrical parameters
Consider the family of curvilinear coordinates

2 2
) , ( ky x y x u + = ;
k
l x y y x v

= ) / ( ) , ( (3.35)
represented in Fig. 3.5a as functions of the parameter k, l denoting a reference length.
The curves (3.35) fulfil the orthogonality condition (3.11). According to (3.12) and (3.15)
the infinitesimal arc length and the radii of curvature of the curvilinear axes are given by
Static method
26

2 2 2
y k x
dy ky dx x
ds
u
+
+
= ;
2 2 2
y k x
dy x dx ky
ds
v
+
+
= (3.36)
and

2 / 3 2 2 2
) (
) 1 ( 1
y k x
xy k k
u
+

;
2 / 3 2 2 2
2 2
) (
) ( 1
y k x
ky x k
v
+
+
=

(3.37)
respectively.
Shear field
For a load path along the u-coordinate with parameters k
s
(denoted by u
s
) and for a uniform load
distribution q, Eq. (3.27) reduces to
( )
( )
0
2 2 2
2 2 2
2 2
0 , 0 , 0
= + +
+
+
+ + y k x q
y k x
y k x k
v v y k v x
s
s
s s
y s x
(3.38)
Starting from the point P(x
P
,y
P
) and moving along the characteristics of the shear problem,
i.e. the curve
s
k
P
l x v x y ) / ( ) ( = with v
P
= v (x
P
,y
P
), (3.38) simplifies to

( )
0 ) / (
) / (
) / (
2 2 2 2
2 2 2 2
2 2 2
0 , 0
= + +
+
+
+
s
s
s
k
P s
k
P s
k
P s s
x
l x v k x q
l x v k x
l x v k x k
v v x (3.39)
Note that ( )dx v x y k v dv
y s x , 0 , 0 0
) / ( + = . Solving Eq. (3.39), the shear field is given by

( )

= + =

(
(

+
|
.
|

\
|
+ =
+
) 1 ( ln
) 1 (
1
2 2
0
1
2 2 2
0
s s
s
s
k
s s
k x q C y x v
k
k
q
x
l
C y k x v
s
(3.40)
To obtain (3.40), v
P
was replaced by
s
k
l x y

) / ( after integration. The integration constant C
s
is a
function of v
P
, i.e. C
s
= C
s
(v
P
). The shear force value on a point of each characteristic may be
arbitrary. For instance, requiring
0 0
v v = at x x = for 1
s
k , one obtains

(
(

+
+
|
|
.
|

\
|
+ |
.
|

\
|
=

+
1
2 / 1
2
2
2 2 2
0
1
s
s
k
s
k
q
x
x
y k x v
l
x
C
s
(3.41)
At (x,y) = (0,0), Eq. (3.4
1
) remains valid only if the shear force disappears (i.e. 0 =
s
C ). At this
characteristic intersection point different shear forces correspond to a load concentration, hence
a shear field singularity (see Eq. 3.26).
A similar analysis may also be formulated for a load path along the v-coordinate. Note that a
load transfer along closed trajectories (such as for positive k
s
, see Fig. 3.5a) is possible only if
they meet a support.
Moment field
The following analysis aims at developing a moment field in agreement with the shear field
determined above, having principal moment sections in the direction of the curvilinear
coordinates (3.35) with k = k
m
. Applying Eq. (3.8), Eq. (3.40) is split into the principal moment
sections:

( )


(
(

+
|
.
|

\
|
+

=
(
(

+
|
.
|

\
|
+
+
=
+
+
) 1 (
1
1
1
2 2 2
1
2 2 2
2 2
s
s
k
s
m
m s
t
s
k
s
m
m s
n
k
k
q
x
l
C
y k x
k k xy
v
k
q
x
l
C
y k x
y k k x
v
s
m
s
m
(3.42
1
)
Lower-bound method
27

( )
( )
( )

+
+
=

=
) 1 (
ln
1
ln
2 2 2
2 2 2
2 2
s
s
m
m
t
s
m
m
n
k
x q C
y k x
k xy
v
x q C
y k x
y k x
v
m
m
(3.42
2
)
Inserting (3.42
1
) into (3.28), one obtains

( )
( )
( ) ( )
( )
( )
( ) ( )

(
(

+
|
.
|

\
|
=
+

+ +
(
(

+
|
.
|

\
|
+ =
+
+
+ +
+
+
1
1
1
1
1 2
2 2 2
, 2 , 2
1
2 2
2 1
2 2 2
2 2
, 1 , 1
s
k
s m s
m
m m
y x m
s
k
s m s
m
m m
y m x
k
q
x
l
C k k xy m m
y k x
xy k k
m x m y k
k
q
x
l
C y k k x m m
y k x
y k x k
m y k m x
s
s
(3.43)
and, similarly, for (3.42
2
)

( )
( )
( ) ( )( )
( )
( )
( ) ( ) ( )

+ =
+

+
=
+
+
+ +
x q C k xy m m
y k x
xy k k
m x m y k
x q C y k x m m
y k x
y k x k
m y k m x
s m
m
m m
y x m
s m
m
m m
y m x
ln 1
1
ln
1 2
2 2 2
, 2 , 2
2 2
2 1
2 2 2
2 2
, 1 , 1
(3.44)
Eqs. (3.43) and (3.44) are the moment equilibrium requirements in curvilinear coordinates
resulting from a shear flow in the direction u
s
(k
s
) and a principal moment section in the
directions u
m
(k
m
) and v
m
(k
m
). Since both moments m
1
and m
2
are involved in the equilibrium
conditions with characteristics u
m
and v
m
, the analysis cannot be reduced to a one-dimensional
problem and a general solution cannot be formulated. Next, some particular solutions of (3.43)
are presented.
Rectangular slab segment
Assuming m
2
= 0 the equation system (3.43) becomes similar to the differential shear problem
(3.38). In a more general form, m
2
is set equal to a constant value m. Excluding the cases k
m
= 0
and k
m
= 1, which are analysed separately in the following sections, (3.43
2
) is solved with
respect to m
1
:
{ } 1 , 0
1 ) 1 (
) )( (
1
2 2 2
1

(
(

+
|
.
|

\
|

+
=
+
m
s
k
s
m m
m s m
k
k
q
x
l
C
k k
k k y k x
m m
s
(3.45)
Inserting (3.45) into (3.43
1
), one obtains

[ ] 0 ) 1 ( ) (
1
) 2 2 (
) 1 (
, ,
1
1
2
2 2 2
=
(
(

(
+ + |
.
|

\
|

(
(

+
|
.
|

\
|
+

+
+
+
s s y s m x s
k
m s
s
k
s s m s m m
m m
m
C k C y k C x
x
l
k k
k
q
x
l
C k k k k k
k k
y k x
s
s
(3.46)
For C
s
= 0, Eq. (3.46) is always fulfilled if 0 2 2
2
= +
s s m m m
k k k k k . This corresponds to a
shear distribution with parameter k
s
combined with principal moment sections with parameters

(

+ + = 1 14 1
4
1
2
s s s m
k k k k (3.47)
see Fig. 3.5b. The principal moments on the trajectories u
m
(k
m
) and v
m
(k
m
) are given by

1 2
2
2 2 2
1
+ +
+
=
m m
m
k k
y k x q
m m ; m m =
2
(3.48)
Static method
28
Exploiting Eqs. (3.9) the moment distribution is referred to Cartesian coordinates, i.e.

1 2
;
1 2
;
1 2
2 2
2 2
2
2
+ +
=
+ +
=
+ +
=
m m
m
xy
m m
m
y
m m
x
k k
xy k q
m
k k
y k q
m m
k k
x q
m m (3.49)
Assuming a boundary along the lines x = a or y = b, (3.23) leads to the boundary forces

) 1 (
) 1 (
) (
2
+ +
+
= =
m m
m
a a
k k
k qa
r q ;
) 1 (
) 1 (
) (
2
+ +
+
= =
m m
m m
b b
k k
k qbk
r q (3.50)
and the boundary moments

1 2
2
2
+ +
=
m m
a
k k
a q
m m ;
1 2
2
2 2
+ +
=
m m
m
b
k k
b k q
m m (3.51)
The shear forces transferred along such boundaries are given by

1 2
2
+ +
=
m m
m
a
k k
y a k q
V ;
1 2
2
+ +
=
m m
m
b
k k
x b k q
V (3.52)
According to Fig. 3.5c the conditions (3.50) to (3.52) for 2 / l k a
m
= and 2 / l b = describe
the boundary conditions of a uniformly loaded rectangular slab segment. Setting m
n
= 0 at the
boundaries, (3.51) corresponds to ] ) 1 ( 8 [ /
2 2 2
+ + =
m m m
k k l qk m , while the line loads
) /( ) 1 ( 4 l k k m q
m m a
+ = and ) /( ) 1 ( 4 l k k m q
m m b
+ = act at 2 / l k x
m
= and 2 / l y = , respectively.
According to (3.52) the boundaries act as shear lines transferring to each slab corner
m m
k m k F / 2 = . The boundary loads q are equivalent to the reactions q r
n
= , see Eq. (3.50).
Different boundary moments can be obtained by moving the system of reference in the plane of
the slab.
The strip method (k
m
= 0)
For k
m
= 0, u
m
and v
m
coincide with the Cartesian coordinates x and y and Eq. (3.43) reduces to
the familiar expressions v
x
= m
x,x
and v
y
= m
y,y
of the strip method, with

1 +
=
s
x
k
x q
v ;
1 +
=
s
s
y
k
y k q
v (3.53)
In (3.43) C
s
has been tacitly set equal to zero, i.e. the shear forces disappear at the origin of
the Cartesian coordinates. By comparing (3.40) and (3.53) it follows that a constant value of q
x

and q
y
in (3.29) corresponds to a principal shear distribution in the direction u
s
with
k
s
= (q q
x
) / q
x
= q
y
/ (q q
y
). In conclusion, the line k
m
= 0 in the k
s
-k
m
-diagram (Fig. 3.5b)
represents the strip method solution corresponding to a split of the external load into constant
portions.
Polar trajectories (k
m
= 1)
Considering k
m
= 1 and m
2
= m = constant in (3.43
2
), the moment equilibrium in the tangential
direction is trivially satisfied only if combined with a load distribution with k
s
= 1. This
statement agrees with (3.19), where principal trajectories aligned with the polar coordinates and
a constant moment distribution on the tangential section produce a moment change in the
tangential direction only, i.e. a radial shear transfer.
The analysis is performed by introducing polar coordinates, r
2
= x
2
+ y
2
; tan = y / x.
Thereby, the characteristic y(x) = v
P
x / l through P(x
P
,y
P
) with v
P
= v(x
P
,y
P
) corresponds to the
radial line =
P
= tan
1
(v
P
/ l ). Integrating (3.43
1
) for k
m
= k
s
= 1 on =
P
one obtains
m
r
C l C qr
m
m s
r
+ + + =
cos cos 6
2
2 2
; m m =

(3.54)
Lower-bound method
29
Fig. 3.5: General stress fields: a) family of orthogonal curvilinear coordinates; b) solutions
map; c) rectangular segment; d) annulus segment; e) trapezoidal segment.
1 0 1
k
v
u
x
y
a)
k 0
k
0
m
s
1
1
b)
c)
d
c
x
y
z
q
q
b
F
F F
F
q
a
0 = k
m
2
2 k
m
+ k
m
_
k
s
k
s
_
2
d)
a
b
b
a
q
a
q
b
D

B
A
B
C

A
b
c
a
x
y
z
m
r
b
F
B
F
A
b
m
r
a
F
C
a
F
D
r

m
m
e)
q

AB
b
c
a
x
y
z
m
b
m
a
r

q
Strip method
e ,
A
B
C
D
r
b
r
a
m
m
x
y
x
y
x
y
x
y
x
y
x
y
v 0 =
0
v 0 =
0
Static method
30
The integration constants C
s
and C
m
are functions of
P
: C
s
= C
s
(
P
), C
m
= C
m
(
P
). Since the
shear and moment characteristics coincide, C
s
is a constant in the shear-moment integration.
Polar trajectories emulate the previous case of the strip method, where each coordinate line

P
= const is an independent strip and two integration constants are available to fulfil the
boundary conditions. The coordinate centre is a singularity point of the shear as well as the
moment field; the values of the shear force and of the radial principal moments at this point
determine a symmetrical stress field distribution. In order to provide an efficient load transfer,
the geometrical load transfer rate ( r m/ ) has to improve the radial shear flow, i.e. shear flow in
positive (negative) r-direction requires m > 0 (m < 0). In the following, the stress field of an
annulus and a trapezoidal segment are explained as examples of the polar stress field family.
a) Annulus segment
Consider the annulus segment ABCD depicted in Fig. 3.5d. The geometry of the segment is
defined by the internal and the external supported boundary r = a and r = b, respectively, and
the arbitrary apex angle
AB
. The segment is subjected to a uniformly distributed load q. For the
boundary moments m
r
( r = a ) = m
a
, m
r
( r = b ) = m
b
and m

= m, the shear field (Eq. (3.40) with


k
s
= 1) and the moment field (Eqs. (3.54)) simplify to

|
|
.
|

\
|
= r
r
c q
v
2
0
2
(3.55)
( ) m
b a
m m
b a
q
r
ab c q qr
m
b a
r
+
(

+ + + =
6 2 6
2 2
; m m =

(3.56)
where
( ) ( ) ( )
(

=
b a
m m b m m a b a
q
b a q
c
3 3 2
6 ) (
2
(3.57)
Starting from r = c, the load on the segment area c r a is restricted by the reaction
r
a
= q (c
2
a
2
) / (2a) at the support CD and the load on the segment area b r c is restricted by
the reaction r
b
= q (b
2
c
2
) / (2b) at the support AB. Note that Eqs. (3.23) reduce to r
a
= v
0
(r = a),
and r
b
= v
0
( r = b ), respectively, AB and CD being principal moment trajectories.
For a = 0 the annulus segment is transformed into a circular sector. At r = 0, Eq. (3.55) is
indeterminate and the reaction r
a
degenerates to a concentrated support force of intensity
R
a
= q c
2

/2. This singularity disappears if c = 0, or, equivalently, if m = qb


2
/ 6+ m
b
.
b) Trapezoidal segment
Consider the trapezoidal segment ABCD depicted in Fig. 3.5e, having bases CD and AB at x = a
and x = b (0 < a < b), respectively, and oblique sides AD and BC at angles
A
and
B
from the x-
axis. The segment is subjected to a uniformly distributed load q. It is supported along the bases
AB and CD and free along AD and BC. In addition, the bending moments m, m
a
and m
b
act on
the sides AD and BC, CD and AB, respectively.
The stress field of the segment is obtained by adjusting Eq. (3.40) to k
s
= 1, and Eqs. (3.54)
to the boundary conditions m
x
( x = a ) = m
a
and m
x
( x = b ) = m
b
:

|
|
.
|

\
|
= r
r
c q
v

2
2
0
cos 2
(3.58)
( ) m
b a
m m
b a
q
r
ab c q qr
m
b a
r
+
(

+ + + =
6 cos cos 2 6
3 2
2 2

; m m =

(3.59)
with
( ) ( ) ( )
(

=
b a
m m b m m a b a
q
b a q
c
3 3 2
6 ) (
2
(3.60)
Lower-bound method
31
Verifying Eq. (3.41) for k
s
= 1 and C
s
= c
2
q / (2l
2
), the shear force vanishes at x = c. The load
on the segment areas c x a and b x c is transferred to the support CD
(v
x
( x = a ) = q ( c
2
a
2
) / (2a)) and AB (v
x
( x = b ) = q( b
2
c
2
)/ ( 2b )), respectively. According to
Eq. (3.23
1
) the reaction forces r
a
and r
b
are given by

a
m m
a
a c q
r
a
a

=
2 2
2
;
b
m m
b
c b q
r
b
b

+

=
2 2
2
(3.61)
The terms (m m
a
) / a and (m m
b
) / b are related to the twisting moments along the boundaries.
The supports act like shear lines transferring the forces V = ( m m
a
)tan
A
for x = a and
( m m
b
) tan
B
for x = b. To maintain equilibrium, hold-down forces F
A
= ( m m
b
) tan
A
and
F
B
= ( m m
b
) tan
B
, as well as support forces F
C
= ( m m
a
) tan
B
and F
D
= ( m m
a
)tan
A
are
required at the corners A, B, C, D. Note that the values of these forces result directly from the
intensity of the boundary moments and their orientation. If the boundary becomes a principal
moment direction (e.g. m
a
= m along CD), then the shear line character of the boundary
disappears.
For a = 0, the trapezoidal segment becomes triangular. The reaction r
a
degenerates to a
concentrated support load of intensity R = q c
2
/ 2 ( tan
A
+tan
B
). R = 0 if c = 0, or, equivalently,
if m = q b
2
/ 6+m
b
. For
b a
r r , , a rectangular segment is obtained, the polar coordinates
become Cartesian and the analysis is reduced to the strip method of analysis.
3.6.6 Superposition principle
Due to the linearity of the equilibrium conditions (3.18), the analysis of cases with complex
principal trajectories can be reduced to the sum of solutions with respect to a set of particular
curvilinear coordinates. In order to add different basic solutions, they have to be referred to a
common system of reference, generally the global system xyz. In the following, two examples
illustrate the superposition principle. The rectangular segment of Fig. 3.5c and two Hillerborg
solutions are combined to obtain Woods solution [67] and Nielsens solution [46]. Two
trapezoidal segments (Fig. 3.5e, a = 0) are combined to obtain a triangular segment with
arbitrary boundary conditions.
Rectangular segment
The rectangular slab segment depicted in Fig. 3.5c shows distributed as well as line loads.
Combining the stress field (3.49) with an equilibrium strip method solution one or the other load
form is privileged.
a) Woods solution [67]
Distributing the line loads on the boundaries of the rectangular slab segment (Fig. 3.5c) with
Hillerborgs method ( ) 1 /( ) 1 (
2
+ + + =
m m m x
k k k q q ; ) 1 /( ) 1 (
2
+ + + =
m m m m y
k k k qk q ), the stress field
superposition reduces to the solution of a corner-supported rectangular slab with spans
l k l
m x
= , l
y
= l and load ) 1 /(
2
+ + =

m m m
k k qk q , see Fig. 3.6a:

2
;
4
1
8
;
4
1
8
2
2
2
2
2 2
xy q
m
l
y
l q
m
l
x l q
m
xy
y
y
y
x
x
x

=
|
|
.
|

\
|
=
|
|
.
|

\
|
= (3.62)
The support force at each corner equals R = q
*
l
x
l
y
/ 4. Note that k
m
is a simple parameter and
no longer corresponds to the principal moment directions; correspondingly, the principal shear
trajectories are not given by (3.47).
b) Nielsens solution [46]
The rectangular slab segment of Fig. 3.5c is considered. Superimposing a Hillerborg stress field
to eliminate the boundary load
b
q and the distributed load q ( ) 1 /(
2
+ + =
m m x
k k q q ;
Static method
32
) 1 /( ) 1 (
2
+ + + =
m m m m y
k k k qk q ), the stress field of the corner supported rectangular slab segment with
spans l k l
m x
= , l
y
= l and a boundary load )) 1 ( 2 /(
2
+ + =
m m x m
k k l qk q at 2 /
x
l x = is obtained,
see Fig. 3.6b:

x
xy
y x
y
y x
l
xy q
m
l
y
l
l q
m m =
|
|
.
|

\
|
= = ;
4
1
4
; 0
2
2
2
(3.63)
The support force at each corner equals 2 /
y
l q .
Triangular segment
For a = 0, the trapezoidal segment assumes a triangular shape, transferring the loads in radial
beams between a corner and the opposite edge. The requirement of constant moments along the
free boundaries reduces the applicability of the segment, see Chapter 3.6.7. In the following,
more general boundary conditions are obtained by superimposing two trapezoidal segments
(a = 0) with different orientation in the plane of the slab.
Similar to Eq. (3.29), two sets of radial beams centred in A and B, respectively, transferring
the loads q
(A)
and q
(B)
, carry the total load q:

) ( ) ( B A
q q q + = (3.64)
Furthermore, the tangential moments m
(A)
and m
(B)
of the stress fields (A) and (B) result in the
total bending moment m along AB:

) ( ) ( B A
m m m + = (3.65)
Consider the triangular segment ABC shown in Fig. 3.6c. The geometry of the segment is
defined by the side lengths l
a
and l
b
of BC and AC, respectively, and the triangular height, h
a
.
ABC is uniformly loaded by the load q and it is supported along the boundaries AC and BC. In
addition, the bending moments m
a
, m
b
and m
c
act along BC, AC and AB, respectively.
Assuming A and B as shear flow centres, the load q and the boundary moment m
c
are split
arbitrarily into the portions q
(A)
, q
(B)
and m
(A)
, m
(B)
, respectively. Eqs. (3.64) and (3.65) reduce
the stress field analysis to that of the trapezoidal segment, see Figs. 3.6d and 3.6e. The stress
field sum involves a stress transformation. In contrast, the force flow is easily followed. The
reaction forces r
a
and r
b
along BC and AC, respectively, are given by

a a
b c
b
b
a a B
B b b
a
a c
a A
A b a
l h
m m
l
l
l h q
r r
h
m m
h q
r r

+ = =

+ = = 2
3
; 2
3
) (
) (
) (
) (
(3.66)
and the corner forces are given by

( ) ( )
( ) ( )
) ( ) ( ) ( ) (
) (
2
) (
) ( ) (
) (
) (
) ( ) (
tan ) ( tan ) (
tan
6
tan
6
B C b c A C a c B C A C C
A B a c b c
a a
b
a a B
B B A B B
B A b c a c
a
a
a a A
A A B A A
m m m m F F F
m m m m
l h
l
l h q
R F F
m m m m
h
l
l h q
R F F

+ = + =
+ + = =
+ + = =
(3.67)
Eq. (3.64) influences the shear flow of the basic segments, distinguishing zero shear lines at
) (
2 2
) (
2
) (
/ ) ( 2 3 /
A a c a A A
q m m h c x = = and
) (
2 2 2
) (
2
) (
/ ) ( 2 ) 3 /( ) (
B a c b a a B B
q m m l l h c x = = , hence
v
0
= 0 at the intersection point of these two lines in the resultant stress field. Eq. (3.65) does not
have any influence on the force flow.
In A and B the shear line force (3.23
2
) is added to the reactions R
A(A)
, and R
B(B)
. For the case
of vanishing reactions R
A(A)
or R
B(B)
, i.e. for
2
) (
/ ) ( 6
a b c A
h m m q = or ) /( ) ( 6
2 2 2
) ( a a b c b B
l h m m l q = ,
respectively, the segment gives an example of Nielsens nodal forces of type 2 [48, 50].
Lower-bound method
33
Fig.3.6: Superposition principle: a) point supported rectangular slab, uniform load
distribution, b) point supported rectangular slab, boundary line loads; c) triangular
segment; d) and e) basic trapezoidal segments (a = 0).
3.6.7 Stress field approach for slabs
The stress field approach for slabs aims at developing stress fields based on curvilinear
coordinates for the design of slabs with arbitrary shape and support layout. In addition to stress
fields corresponding to a continuous load transfer, statical discontinuities are introduced as new
elements. Similar to the stress field approach for beams [45], a limited number of stress fields
are considered. The different stress fields are combined in the plane of the slab. The analysis
focuses on the segment boundaries, where statical continuity has to be obtained.
c) d) e)
q
h
a
a) b)
x
y
z
q
R
x
l
R R
R
x
y
z
R
R R
R
q q
C
A
B
m
a
l ,
a
m
l
b
c
m ,
b
q
m
a
m
(A)
m
(A)
m
(B)
m
(B)
b
F
C
F
A
F
B
r
b
r
a
(A)
m
(B)
m m
(A)
q
(B)
r
b(A)
F
B(A)
F
C(A)
R
A(A)
r
b(B)
x
(B)
y
(B)
y
(A)
x
(A)
c
(A)
c
(B)
v 0 =
0
R
B(B)
F
C(B)
F
A(B)
y
l
x
l
y
l
*
v 0 =
0
C
A
B
C
A
B
Static method
34
The following analysis considers the basic segments resulting from the application of the
generalised strip method and the superposition principle. Of course, additional segments
developed in the same way could be added to the stress field library. Fitting the basic elements
in the plane of the slab, a load transfer scheme is selected. The segment geometry determines a
specific stress field. The conditions

B
n
A
n
r r = ;
B
n
A
n
m m = (3.68)
and

=
i
i
P
R 0 (3.69)
establish load transfer continuity and bending moment continuity along the sides between two
adjacent elements A and B, and vertical equilibrium at internal segment corners (
i
P
R denoting
the force (3.26) of segment i at node P), respectively. Considering (3.68
2
) as a boundary
condition of the basic segments (cf. Figs. 3.5 and 3.6), the analysis focuses on the force flow
between the segments. Loads concentrating at the segment boundaries are balanced with strong
bands or shear lines.
3.7 Examples
The theoretical lower-bound analysis is completed with two example applications a
rectangular and a triangular slab.
3.7.1 Simply supported rectangular slab
Consider a uniformly loaded, simply supported rectangular slab with side lengths l and 2l.
Strip method
The strip method of analysis is defined by the discontinuity lines and the load dispersion
indicated in the plane of the rectangular slab in Fig. 3.7. In the short span (y-direction) two
zones are distinguished the middle strips carry the whole load (zone 1); the edge strips (width
l / 2 ) carry only half of the total load (zone 2). The strips in the x-direction complete the load
transfer carrying the portion q / 2 in the boundary areas. The slab design involves bottom
reinforcement only. The total average moment equals 0.057ql
2
.
Generalised strip method
For k
m
= 2, Fig. 3.5c fits directly into the slab considered. According to Eq. (3.47), the shear
flow and moment trajectories are given by (3.35) with k
s
=2.5 and k
m
= 2, respectively, see
Fig. 3.8a. The principal shear force equals v
0
= q ( x
2
+ 6.25 y
2
)
1/2
/ 3.5, and the principal
moments are given by
m
u
m = q ( l
2
x
2
4y
2
) / 14 (Fig. 3.8b) and
m
v
m = ql
2
/ 14, respectively, where
the coordinates x, y define a Cartesian system with origin at the slab centre. Considering the
variation of m
1
between ql
2
/ 14 at the slab centre and ql
2
/ 14 at the slab corners, see Fig. 3.8b,
the total average moment equals 0.052ql
2
. Finally, the reaction forces are equal to 3ql / 7 and the
corner hold-down forces are equal to ql
2
/ 7.
Stress field approach
The transition from the generalised strip method to the stress field approach results as soon as
one single segment does not fit directly into the slabs shape. As an example application,

Lower-bound method Examples
35
Fig. 3.7: Simply supported rectangular slab, strip method analysis.
trapezoidal segments are fitted into the rectangular slab, taking into account the symmetry of the
statical problem, see Fig. 3.9a. The analysis attempts a direct load transfer to the supports,
making use of the curvature of the tangential moment trajectories.
Satisfying (3.68
2
) with constant bending moment m along AF, BE, CE, DF, EF, Fig. 3.9
aims at defining the range of possible equilibrium solutions. For l
EF
= l / 2 (i.e. a fixed segment
geometry), Fig. 3.9a investigates the influence of m (m > 0) on the force flow. Let F
triang
denote
the force resulting from BCE or AFD in E or in F, respectively, F
trap
= r
EF
l
EF
= forces of ABEF
or DFEC along EF, and F
tot
= 2F
triang
+ 2F
trap
. F
triang
, F
trap
and F
tot
decrease linearly with
increasing m. The range of the analysis is restricted by the geometrical condition c > 0 in each
stress field, c indicating the zero shear lines. For m = 9 ql
2
/ 128, the trapezoidal segments support
the triangular ones, giving F
tot
= 0. Investigating the combination of m and l
EF
resulting in
F
tot
= 0, Fig. 3.9b completes Fig. 3.9a by considering the influence of the basic segment
geometry. F
tot
= 0 is obtained only for l
EF
> 0.47l. The state m = 0.071ql
2
, l
EF
= 0.697l gives the
segment combination with the maximum m-value, i.e. the largest amount of force transferred
geometrically, making use of the curvature of the tangential trajectories of the polar principal
moment net.

Fig. 3.8: Simply supported rectangular slab: a) shear flow and moment field, b) corresponding
m
1
.
a)
x
y
b)
v
0
=
0
l [1000 m /ql
2
]
1
0
40
40
71
71
l /2
0.125
q
1 2 2
x
0.063
0.063
x
y
l /2
l /2
l /2 l /2 l
z
1 2
2
q
ql
2
m
y
ql
2
m y
ql
2
m
2
q
Static method
36
The values of m and l
EF
determine the stress field in the slab. Referring to Fig. 3.9, the states
(a) with largest geometrical load transfer (m = 0.071ql
2
, l
EF
= 0.697l), (c) with F
tot
= 0 in
Fig. 3.9a (m = 9ql
2
/ 128, l
EF
= 0.5l), and (e) with F
trap
= 0 in Fig. 3.9a (m = ql
2
/ 16, l
EF
= 0.5l ) are
analysed in Fig. 3.10 for comparison purposes. In each case the shear flow and the principal
moment trajectories as well as the radial moment are depicted in the plane of the slab.
Considering (3.60) for l
EF
= 0.697l and m = 0.071ql
2
one obtains c
ABEF
= c
DFEC
= a = 0.268l
and c
BCE
= c
AFD
= 0, i.e. the load is transferred radially from the lines HF (zero shear lines) to the
supports, (Fig. 3.10a). The radial moment varies between 0.071ql
2
in the middle of the slab and
0.120ql
2
at the corner of the trapezoidal segment (Fig. 3.10b). Accounting for the constant
moment m

= 0.071ql
2
, the total average moment equals 0.053ql
2
. The reaction force along AG
equals 0.429ql and along AI it is equal to 0.434ql. A hold-down force of 0.146ql
2
is required at
the corners.
For l
EF
= 0.5l and m = 9ql
2
/ 128, the shear flow varies as depicted in Fig. 3.10c. In the
triangular segment, equilibrium results in a zero shear line at the distance c = 0.217l from F.
The load portion ql
2
/ 32 is concentrated at the apex of the triangular segments. Introducing this
load with a strong band of strength M
x
= ql ( l
2
/ 16 x
2
) / 16 along EF, the trapezoidal segment
supports AFI through the reaction r
HF
= ql / 16. In Fig. 3.10d, the radial moment increases
beyond m to 0.104ql
2
in the vicinity of the apex of the triangular segments. The maximum
negative value of m
r
is reached at the corner of the trapezoidal segment, 0.158ql
2
. Accounting
for the constant moment m

= 9ql
2
/ 128, the total average moment equals 0.055ql
2
. Note that the
strong band moment is not considered in the average. The reaction force along AG equals
0.434ql and along AI it is equal to 0.438ql. Finally, a hold-down force of 0.152ql
2
is required at
the corners.
Neglecting the condition F
tot
= 0, the segment combination corresponds to a simply
supported rectangular slab with point supports at E and F. For m = ql
2
/ 16 and l
EF
= 0.5l, the
trapezoidal segment is similar to that of Fig. 3.10a, where the shear flows from EF to the
support. The triangular segment is similar to that of Fig. 3.10c, where the shear vanishing at a


Fig. 3.9: Simply supported rectangular slab, stress field approach a) internal force analysis;
b) geometrical analysis.
0.1
0.01
D
A
C
B
F E
l 0.5
0.75 l 0.5
m
ql
2
F
i
ql
2
F
tot
ql
2
2
c= 0
c= 0
F
tot
F
triang
F
trap
= 0
m
0.01
0.47
0.071
0.697
l
l
EF
a) b)
a c e
l
l 0.5
0.75 l
D
A
C
B
F E
x
y
Examples
37
Fig. 3.10: Particular solutions: a) shear flow and moment field for l
EF
= 0.697l and
m = 0.071ql
2
, b) corresponding m
r
; c) shear flow and moment field for l
EF
= 0.5l and
m = 9ql
2
/ 128, d) corresponding m
r
; e) shear flow and moment field for l
EF
= 0.5l and
m = ql
2
/ 16, f) corresponding m
r
; g) transfer of R
F
: shear lines FL; h) additional
moment distribution.
a)
I
A
H
G
l 0.5
l 0.349 l 0.651
c)
l 0.25 l 0.533
e)
v 0 =
0
l 0.25 0.5
l 0.217
F
71
0
40
r /8 =
HF
ql
R
F
=ql
2
/32
v
0
=
0
l 0.25
2
L
R
F
=ql
2
/24
R
F
v
0
=
0
v 0 =
0
R
F
=ql
2
/24
l 0.25 l 075 d d
0
40 80
I
A
H
G
I
A
H
G
I
A
H
G
l
0
40 80
40
0
40 80
0
40 80
40
0
40 80
40
x
x
x
y
y
y
y
M
m
x
=
d R
F
l
m
y
=
l R
F
d 4
141
28
70
63
108
104
71
158
31
120
42
b)
d)
f)
h) g)
[1000 m
r
/ql
2
]
[1000 m
r
/ql
2
]
[1000 m
r
/ql
2
]
x
F
F
F
Static method
38
distance 0.25l from F determines R
F
= ql
2
/ 24, see Fig. 3.10e. Fig. 3.10f shows the radial
moment variation in the two segments. The radial moment increases beyond m to 0.108ql
2
near
F in the triangular segment, and the extreme negative value is reached at the corner of the
trapezoidal segment, 0.141ql
2
. Accounting for the constant moment m

= ql
2
/ 16 the total
average moment equals 0.051ql
2
. The reaction forces are equal to 0.406ql and 0.417ql along AG
and AI, respectively, and hold-down forces of 0.135ql
2
are required at the corners. Considering
the fictitious reaction at F, Fig. 3.10g introduces a shear line transferring V = R
F
/ 2 between F
and L. Eqs. (3.20) to (3.22) correspond to a moment m
x
= d R
F
/ l and m
y
= R
F
l / ( 4d ) in the
regions LGHF and LFM, respectively (Fig. 3.10h). For d = 0 the shear line becomes a strong
band with a moment M
y
= R
F
l (1 2y / l) / 4, cf. Eq. (3.21). Finally, the total stress field of the
original problem is obtained by superimposing the stress field of Fig. 3.10e on that of
Fig. 3.10g. The total average moment is a function of d.
Discussion
The rectangular slab analysis has a long tradition, going back to the origins of the equilibrium
method (Chapter 5.3.2). Anticipating Johansens work [21], Ingerslev [19] investigated the
load-carrying capacity of a rectangular slab based on the equilibrium considerations suggested
in Fig. 3.9. First established by Nielsen [47], the stress field of Fig. 3.10a shows that Ingerslevs
result is correct only if
u u
m m 7 . 1 ' , where
u
m' is the negative bending resistance. For
u u
m m = '
the limit analysis problem of the rectangular slab has not been solved; Fig. 3.8 (Wood [67]) and
[49] give lower-bound approximations of the ultimate load. Neglecting limit analysis
requirements Fig. 3.10 extends Nielsens analysis considering more general triangular and
trapezoidal segments. The condition F
tot
= 0 corresponds to the application of Eq. (2.7) to all
mechanisms of the type shown in Fig. 3.9a; Nielsens solution (Fig. 3.10a) represents the
special case F
trap
= F
triang
= 0.
Comparing the total average moments of the different cases considered, Fig. 3.8 gives the
best solution (0.052ql
2
), followed by Fig. 3.10a (0.053ql
2
) and Fig. 3.10c (0.055ql
2
). The strip
method is less efficient (0.057ql
2
). For practical design, smooth moment distributions are
preferred. This criterion is better fulfilled by solutions based on curvilinear principal moment
sections.
3.7.2 Triangular slab with a free edge
Consider the uniformly loaded right-angled triangular slab ABC shown in Fig. 3.11a. The slab is
simply supported along the edges AB and BC with lengths 2l and 3l, respectively, and it is free
along AC.
Attempting to apply the strip method of analysis using strips parallel and perpendicular to
the free edge it is noted that a very large span results when approaching AC and that the
introduction of a strong band at this location is reasonable. Fig. 3.11a illustrates a possible load
split of q / 3 parallel to and 2q / 3 perpendicular to AC, and the resulting strong band load.
Next, the slab ABC is analysed with the stress field approach by dividing it into the
triangular segments BCD (segment 1) and ABD (segment 2), where D is the point of AC at
x = x
D
= l, see Fig. 3.11b. In both segments the points B and D replace the points A and B of
Fig. 3.6c. Introducing a constant bending moment m along BC and requiring r
b(B)
= 0, AC being
a free edge, the load q is split into the portions q
(B)
= 13m / ( 6l
2
) and q
(D)
= q + 13m / (6 l
2
) in the
two segments, cf. Eqs. (3.64) and (3.66
2
). Except for the forces F
D1
= ql
2
/ 3 + 4m / 3 and
F
D2
= 2ql
2
/ 3 + 3m / 4, the reactions (3.67) and (3.66
1
) coincide with the slab supports. Similar to
Fig. 3.9a, Fig. 3.11c analyses the influence of m on F
tot
= F
D1
+F
D2
. For m = 12ql
2
/ 25, F
tot
= 0
and ABD supports BCD with F
D
= 23ql
2
/ 75. Investigating the combination of m and x
D
resulting
in F
tot
= 0, Fig. 3.11d completes Fig. 3.11c for the whole range 0 < x
D
<3l. The maximum
moment (m = ql
2
/ 2) occurs for x
D
= 6l / 5, when F
D
between ABD and BCD equals the shear line
forces (3.24).
Examples
39
Fig. 3.11: Triangular slab with a free edge: a) strip method; b) stress field approach; c) internal
force analysis; d) geometrical analysis; e) shear flow for m = ql
2
/ 2, x
D
= 1.2l;
f) principal moment distribution.

Fig. 3.11e and Fig. 3.11f represent the shear field and the moment field in ABC for x
D
= 6l / 5
and m = ql
2
/ 2. The stress fields in ABD and BCD are determined according to the procedure of
Chapter 3.6.6, by superimposing a trapezoidal stress field centred at D with load 25q / 12, a = 0,
b = 6l / 5, m = ql
2
/ 2 and m
b
= 0 and another trapezoidal stress field centred at B with load
13q / 12, a = 0, b = 6l / 5, m = 0 and m
b
= ql
2
/ 2. The results are represented similarly to those of
standard finite element slab programs, using a quadratic grid with mesh length l / 5. In
Fig. 3.11e, the shear flow is indicated by arrows whose widths are proportional to the shear
forces. The forces flow from the free side AC and the corner B to the supports AB and BC. In
Fig. 3.11f, the principal moments and their directions are indicated by crosses. The cross arm
0.1
0.1
A
C
B
m
ql
2
F
D
ql
2 F
tot ql
2
3
c = 0
= 0
m
0.1
l
x
D
a) b)
l 2
x
y
l 3 A
C
B
l 2
x
y
D
2l l
1
2 3
q
3
q
2
A
C
13
ql
4
D
c =
D
0
x
D
= 1
1.2
0.5
c) d)
e) f)
ql ql
2
ql
2

2
1 F
tot
13 l
Static method
40
widths are proportional to the moment intensities and positive and negative values are
differentiated by grey and black colour, respectively. The positive principal moments are almost
constant with a maximum value ql
2
/ 2. Negative moments become larger near the supports. Note
that BD is a principal trajectory of the moment field.
Discussion
Contrary to the rectangular slab, in the triangular slab a shear force is transferred between the
basic segments in any case. When m is a maximum, this force becomes the shear line force
along AC.
Compared to the strip method of analysis, in the stress field of Figs. 3.11e and 3.11f the
edge AC does not involve particular reinforcement. If AC becomes a support, then three
trapezoidal segments (a = 0) will fit in the slab, each one corresponding to an edge. Intermediate
states are obtained when concentrating part of the load to the free edge, hence to A and C via a
strong band. Extra reinforcement along AC reduces the negative moments in the slab.
3.8 Conclusions
In analogy with the static analysis of beams, the static methods for slabs have been reorganised
and unified into the stress field approach by discussing the mechanisms of transverse shear
transfer.
Shear in slabs appears in distributed [kN / m] or in concentrated [kN] form involving a
distributed [kN] or a concentrated [kNm] moment change.
The flow of distributed shear forces is given by the generalised strip method. Following a
selected load path, i.e. fixed principal shear trajectories, distributed shear forces are determined
through vertical equilibrium. Arbitrary principal moment trajectories are selected; beam action
of the principal moment strips and geometrical contributions according to the curvatures of the
principal moment trajectories match the moment change required for the transfer of the shear
forces. Assuming a fixed load distribution and particular principal shear and principal moment
trajectories, the distributed force flow is developed for general slab segments of variable shape
and boundary conditions.
Concentrated shear forces flow along shear lines. The transfer of a concentrated shear force,
i.e. the required concentrated moment change, is produced through a beam action (strong band)
as well as through a discontinuity in the twisting moments.
Continuous stress fields, shear lines and strong bands are the basic tools of the stress field
approach for slabs. The idea underlying this approach is to develop a desired force flow.
Continuous stress fields are responsible for the load transfer within the individual slab segments
and shear lines and strong bands are introduced at their boundaries.
Limited to distributed shear forces, the usual static methods for slabs are special cases of the
generalised strip method; elasticity identifies the behaviour of the slab through elastic beams
spanning in the principal curvature directions of the deflection function; Hillerborgs strip
method constrains the principal moment directions to be straight; the Hencky-Prandtl solutions
correspond to a load transfer on the basis of geometrical contributions assuming fixed principal
moments; and known limit analysis solutions are related to curvilinear systems with particular
geometrical properties (e.g. conical sections).
41
4 Limit analysis
4.1 General
The first attempts at a slab limit analysis date back to the 1920s and 1930s [19, 21]. Supported
by experimental observations, it was assumed that slabs fail along yield lines. The limit analysis
consisted in determining the most critical yield line layout (see Chapter 5.3).
Attempts to solve the slab limit analysis problem on the basis of the theory of plasticity (see
Chapter 2.6) followed in the 1960s. The plasticity approaches started from the analysis of the
limit states following the yielding of the reinforcement. In the lower-bound analysis, the yield
condition found direct application for the design of the reinforcement [66]. The upper-bound
analysis based on the flow rule was found to be more general than the yield line theory
[37, 47, 26]. However, the kinematical research was essentially focused on a review of
Johansens theory [21]. The newly found failure modes did not correspond to the yield line
theory and the limit analysis problem could be completely solved only for some particular cases.
This raised doubts about the existence of complete solutions [8, 69] until the 1970s, when Fox
[9, 10] showed that rather simple loading problems may involve complex solutions. Parallel to
this development, the limit analysis of slabs split into clear upper- and lower-bound approaches.
This chapter summarises the fundamentals of slab limit analysis. Based on work by Save
[58] and Kemp [26], the following analysis starts from a statical yield condition and it uses the
flow rule to deduce how yield will occur. The yield surface is used to define the basic elements
of slab limit analysis and to illustrate the difficulty to determine matching upper- and lower-
bound analyses. These difficulties are mitigated by simplifying the limit analysis to a consistent
theory in agreement with a slab collapse involving yield line mechanisms. Simplified complete
solutions provide the basis for a new slab design method the compatibility limit design
method (see Chapter 6).
The following discussion is limited to homogeneous orthotropic slabs reinforced in the x-
and y-directions in both the top and bottom layers. However, the outlined procedure is valid for
an arbitrary reinforcement layout.
4.2 Yield condition and flow rule
For low reinforcement ratios, the slabs ultimate resistance depends primarily on the strength of
the reinforcement (Chapter 4.2.1). The yield condition describes all stress states for which the
reinforcement is yielding (Chapter 4.2.2). The transition from statics to kinematics is provided
by the flow rule (Chapter 4.2.3). Corresponding parameters of the generalised stresses m
x
, m
y
,
m
xy
are the generalised strains
x
& ,
y
& , 2
xy
& . Slabs may collapse along yield lines or at yield
points; the two cases are characterised by statical and kinematical redundancies, respectively
(Chapter 4.2.4).
Limit analysis
42
4.2.1 Limit states
Consider the simple case of pure bending of a slab strip containing bottom reinforcement with a
cross-sectional area per unit length a
s
at right angles to the x-axis (x-section), see Fig. 4.1a.
Assuming a perfectly plastic behaviour of the concrete and of the reinforcement (yield strengths
f
c
and f
y
), the stress distribution shown in Fig. 4.1b results in the ultimate resistance

=
2
1
2

d f m
c u
(4.1)
where d is the effective depth and ) /(
c y s
df f a = denotes the mechanical reinforcement ratio.
The bending moment in the section parallel to the reinforcement (y-direction) as well as the
twisting moments in the x- and y-directions must be equal to zero, because no steel is available
to develop a resistance.
The ultimate resistances in the x- and y-directions determine the limit states of the slab in an
arbitrary direction, see Chapter 3.2.2. Denoting by the clockwise angle between the
reinforcement direction and the vector orthogonal to the cross-section considered, Eqs. (3.9)
reduce to

2
cos
u n
m m
u
= ; sin cos
u tn
m m
u
= (4.2)
The consideration of the resistance in terms of generalised stresses assumes a continuous
material behaviour (i.e. a good distribution of the reinforcement), a constant in-plane resistance
of the concrete, and a negligible influence of the difference in the internal moment arm arising
from the variation of the compression zone width. If present, compressive reinforcement will
take part in resisting the compressive force and hence reduce the compressive force in the
concrete [50], see Fig. 4.1c; however, for low reinforcement ratios the bending resistance (4.1)
is only slightly increased.
Slabs generally include different reinforcement layers. Superimposing the above analysis
for each reinforcement direction i at the slab bottom, k i 1 , Eqs. (4.2) are generalised to

=
=
k
i
i
i
u n
m m
u
1
2
cos ;
i
k
i
i
i
u tn
m m
u
sin cos
1

=
= (4.3)
Eqs. (4.3) describe the positive slab resistance at ultimate. The negative resistance is determined
similarly, by considering the top reinforcement. Quantities referring to the top of the slab are
denoted by a prime. For example,
s
a is the cross-sectional area per unit length of the top
reinforcement (Fig. 4.1a) and
u
m is the negative bending resistance in the direction of the main
reinforcement.

Fig. 4.1: Slab element subjected to pure bending in the x-direction (reinforcement direction):
a) notation; b) plastic stress distribution, 0 =
s
a ; c) plastic stress distribution, 0
s
a .
d

a) c)
. . . . . .
x
z
y
a
s
m
u
1
m
u
a
s
f
y
f
c
d
z
x
a
s
f
y
f
c
a
s
f
y
b)
a
s
y
Yield condition and flow rule
43
4.2.2 Yield condition
The yield condition distinguishes safe stress states from those for which a given reinforcement
is not sufficient. The collapses of the bottom and top reinforcements define two regimes which
are first analysed separately. Then, positive and negative failures are considered together,
focusing on the regime change.
Positive regime
The positive regime is related to yielding bottom reinforcement. Negative failures are excluded
by constraining the top reinforcement to stay rigid.
A homogeneous orthotropic reinforced slab is considered with the ratio between the
ultimate bending resistances in the y- and the x-directions. In an arbitrary direction, the ultimate
resistances (Eqs. (4.3))

2 2
si cos n m m m
u u n
u
+ = ; sin cos ) 1 ( =
u tn
m m
u
(4.4)
are compared with the associated applied moments (Eqs. (3.9))
2 sin sin cos
2 2
yx y x n
m m m m + + = ; 2 cos cos sin ) (
yx x y tn
m m m m + = (4.5)
where denotes the clockwise angle between the x- and the n-directions.
A stress state (m
n
,m
tn
) is safe if
u
n n
m m and
u
tn tn
m m on any section n. Introducing the
parametric yield function
u
n n n
m m m Y =
+
) , ( [37], these requirements are expressed by
0
+
Y and 0
,

+

Y (4.6)
The yield condition (4.6) is based on a local stress definition. Since m
tn
= m
n,
/ 2, the
analysis focuses on the bending moment.
u
n
m and m
n
, depicted as functions of the parameter ,
must touch only tangentially to avoid violating the yield condition, see Fig. 4.2a. Considering
the variation of
u
n
m (curve a) and different stress states m
n
for 0 , the following holds:
curves
u
n
m and m
n
that do not coincide may touch in one point only (e.g. curve b at = );
different m
n
-curves may touch the
u
n
m -curve at the same point corresponding to the same
collapse (e.g. curves b and c at = );
for isotropic reinforcement ( = 1), the
u
n
m -curve becomes a straight line (curve d) and
collapse occurs in the first principal moment direction;
for orthotropic reinforcement ( 1 ), the orientation of the collapse section with respect to
the principal moment directions depends on the principal moment values.
The first statement follows directly from the similarity of the m
n
- and
u
n
m -curves. Collapse
is constrained to the yield-line direction, i.e the cross-section defined by the abscissa angle of
the contact point. If m
n
is identical with
u
n
m , then the collapse may occur in an arbitrary
direction.
Fig. 4.2b aims to define the set c of stress states compatible with collapse in the direction
= . Starting from the resistances m
x
= m
u
and m
y
= m
u
Mohrs circle for the resisting
moments is determined; Point N corresponds to collapse in the -direction. Any Mohrs circle
through N with a principal moment
u
m m
1
(
2 1
m m and 1 ) corresponds to an m
n
-curve
like c in Fig. 4.2a. The bending and twisting moments in the Cartesian directions are determined
by rotating the radius of N about the angle 2 in the anti-clockwise direction. Introducing the
bending moment m in the x-direction as parameter, the components of the moment tensor are
given by
( )

tan
,
tan
, , ,
2
m m m m
m m m m m
u u
u xy y x
(4.7)
where
u
m m . In the (m
n
,m
tn
)-plane the points (m
x
,m
yx
) of (4.7) reduce to a straight line, the X-
line, see Fig. 4.2b.
Limit analysis
44
The analysis is extended further by determining the states which generate positive collapse
in an arbitrary direction, 0 . Eliminating m and from (4.7) one obtains the conditions
0 ) )( ( :
2
=
+
y u x u xy
m m m m m Y ;
y u
x u
m m
m m

2
tan (4.8)
where
u x
m m and
u y
m m . In the (m
x
,m
y
,m
xy
)-space (4.8
1
) represents an elliptical cone with
as axis the line m
x
m
y
= m
u
(1) in the plane m
xy
= 0 and apex (m
x
,m
y
,m
xy
) = (m
u
, m
u
,0), see
Fig. 4.2c.
Eqs. (4.6) and (4.8
1
) define the positive regime of slabs in the local system of reference and
in the reinforcement directions (i.e. in the Cartesian directions), respectively. Using Eqs. (3.9),
the positive regime may also be expressed in the principal directions of the applied moments:
0 ) si (co ) cos (sin :
2
2 1 1
2
1
2
2 1
2
1
2
1
= + + +
+
u u u
m m m n s m m m m Y (4.9)
where
u
m m
1
,
u
m m
2
and
1
indicates the clockwise angle from the x-axis to the 1-axis.
Eq. (4.9) quantifies the dependence shown in Fig. 4.2a of the yield condition on the direction of
the principal moments relative to the reinforcement directions. This dependence vanishes in the
case of isotropic reinforcement, i.e.
0 ) )( ( :
2 1
=
+
m m m m Y
u u
(4.10)
where
u
m m m
2 1
, . The graphical representation of (4.9) in the (m
1
,m
2
)-plane depends on the
values of
1
[26]. Fig. 4.2d represents the positive regime for isotropic reinforcement (lines AB
and AC).
Comparing (4.6), (4.8
1
) and (4.9), each point on the
u
n
m -curves corresponds to a straight
cone surface line in Fig. 4.2c and to one of the boundary lines AB or AC in Fig. 4.2d, if the
reinforcement is isotropic. The cone apex in Fig. 4.2c and the point A in Fig. 4.2d indicate a
failure redundancy, involving potential collapse sections in arbitrary directions.
Negative regime
The negative regime is obtained analogously to the positive one. In a homogeneous orthotropic
reinforced slab with ratio between the ultimate bending resistances in the y- and the x-
directions, the ultimate resistances in the -section with respect to the x-axis are given by
sin cos ) 1 ( ; si cos
2 2
= + =
u tn u u n
m m n m m m
u u
(4.11)
The yield function ) ( ) , (
u
n n n
m m m Y + =

assesses in terms of (4.6) the safety of (4.5) in


the local system of reference.
The negative yield condition in the reinforcement directions is given by
0 ) )( ( :
2
= + +

y u x u xy
m m m m m Y (4.12)
where
u x
m m and
u y
m m . Using the principal moment directions one obtains
0 ) si (co ) cos (sin :
2
2 1 1
2
1
2
2 1
2
1
2
1
= + + + + +

u u u
m m m n s m m m m Y (4.13)
where
u
m m
1
and
u
m m
2
. Again, the yield condition depends upon the direction of the
principal moments relative to the reinforcement (angle
1
). With isotropic reinforcement (4.13)
simplifies to
0 ) )( ( :
2 1
= + +

m m m m Y
u u
(4.14)
where
u
m m m
2 1
, , see Fig. 4.2d (lines DE and DF).
The definition of a yield section results in a statical redundancy, defined by the set
( )

+
+ =

tan
,
tan
, , ,
2
m m m m
m m m m m
u u
u xy y x
(4.15)
Yield condition and flow rule
45
Fig. 4.2: Yield condition for orthotropic reinforced slabs: a) positive regime analysis in the
local system (n,t,z); b) Mohrs circles for applied and resisting moments; c) positive
regime analysis in the global system (x,y,z); d) positive and negative regime
isotropic reinforcement; e) positive and negative regime analysis in the local system;
f) positive and negative regime analysis in the global system.
a) d)
b) e)
m
n
m
tn
m
m
u

resistance
X
Y
m
n
u
m
tn
u
X Y
2
X-line
N
applied
f) c)

.
m
u
m
=
Y
+
m
xy
m
y
m
x
(2 )
xy
.
( )
y
( )
x
.
Y
+
.
Y
-

m
u
m
n

1 2

m
u

m
n
1

1
m

2
m
m
1
2
H D
G
A
m
u
m
u
B
F
C
E
Y
+
Y
-
( )
.
( )
.
1
2
u
1
=
2
m
y
( )
y
.
m
x
( )
x
.
m
xy
(2 )
xy
.
1 2
m
u

0
m
n
1

Yield-direction
a)
b)
d)
.
.
.
.
.
.
.
.
.
.
.
.
c)
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Limit analysis
46
with parameter m',
u
m m . The apex stress state (m'
u
,'m'
u
,0) indicates a failure
redundancy, being a point of Eq. (4.15) for any value of .
Regime change
The yield surface results from the intersection of the positive and negative regimes. The
following analysis highlights the stress field constraints resulting at a regime change.
Fig. 4.2e depicts the positive and negative resistance curves in an m
n
--diagram. The study
of stress states generating simultaneous positive and negative failures results in the following
statements:
the state of stress at a point of positive and negative collapse is defined by one of the angles

1
,
2
,
1
and
2
, where
1
,
2
are the positive and negative failure section directions and
1
,

2
are the principal moment directions;
for any positive collapse direction =
1
there is a negative collapse direction
=
2
= tan
1
{(m
u
+ m
u
) / [tan
1
(m
u
+ m
u
)]}. The angle =
2

1
is uniquely
determined;
among all admissible stress curves related to a positive or a negative collapse, the one tangent
to both
u
n
m and
u
n
m maximizes m = m
1
m
2
; m is a function of the collapse direction;
for isotropic slabs ( = = 1) the collapse directions are principal moment directions;
= / 2 and m = m
u
+ m
u
.
In analogy to Fig. 4.2c, Fig. 4.2f shows the two yield cones corresponding to the positive
and the negative yield condition in the (m
x
,m
y
,m
xy
)-space. The space enclosed by the two cones
defines the admissible states. The vertical plane through the points (m
u
,m
u
,0) and
(m
u
,m
u
,0) limits the positive and the negative regimes. The limit states for which the statical
requirements outlined in connection with Fig. 4.2e are valid correspond to points on the cones
intersection line.
Using principal moment directions is advantageous only in the case of isotropic
reinforcement. In the (m
1
,m
2
)-plane the positive and negative yield conditions define the square
AHDG (Fig. 4.2d). The transition between Y
+
and Y

is represented by the points G and H. At


these points, the principal moment values are fixed and the stress state depends only upon the
principal moment directions. It is implicit that positive and negative failure sections are
orthogonal.
4.2.3 Flow rule
The slab collapse behaviour is derived by applying the flow rule. A distinction is made between
yield line and yield point, depending on whether the collapse occurs in one or more directions.
The rate of dissipated energy results by multiplying the internal forces with the associated strain
rates.
Generalised strain rates
For the generalised stresses m
x
, m
y
, m
xy
, the associated generalised strain rates must be chosen so
as to obtain the specific rate of energy dissipated in plastic deformation when summing the
products of the corresponding generalised variables, Eq. (2.3):

xy xy y y x x
m m m dD & & & 2 + + = (4.16)
i.e. m
x
, m
y
, m
xy
are associated to
x
& ,
y
& ,
xy
& 2 , respectively.
The strain rate field has the nature of a plane tensor. The curvature rate
n
& and the rate of
twist
tn
& in an arbitrary direction n, forming an angle with the x-direction, are given by
Yield condition and flow rule
47



2 cos cos sin ) (
2 sin si cos
2 2
yx x y tn
yx y x n
n
& & & &
& & & &
+ =
+ + =
(4.17)
These equations may be graphically described by means of a Mohrs circle. There are two
sections in which the twist is equal to zero and the curvature rates are extremes. These values
are the principal curvature rates
1
& and
2
& and the corresponding directions are the principal
directions.
Yield lines
Applying the flow rule, Eq. (2.6), to the (weakly convex) lateral surfaces of the positive and
negative slab yield cones, one obtains

) , , ( ) , , (
) , , ( ) , , (
yx x u y u yx y x
yx x u y u yx y x
m m m m m k
m m m m m k
=
=

+ +


& & &
& & &

0
0

+
k
k
(4.18)
The curvature rates on a positive or negative collapse section in the n-direction defined by
the angle = with ultimate stress (4.7) or (4.15), respectively, are given by
0 ; 0
sin
1
2
= =

=
tn
u
n
k
m m
k

& &
0 ; 0
sin
1
2
= =
+
=
tn
u
n
k
m m
k

& &
(4.19)
The section n is a principal curvature section ( 0 =
tn
& ) and orthogonal to it, the principal
curvature vanishes ( 0 =
t
& ). From inspection of Fig. 4.2a it follows that the principal directions
of the moment tensor and those of the curvature rate tensor coincide only if m
1
= m
u
or
m
2
= m
u
. With isotropic reinforcement this is always the case, as evidenced by applying the
flow rule to the boundaries AB, AC, DE and DF in Fig. 4.2d.
Sections subjected to the yield moment and a rotation rate in the n-direction, dissipating the
rate of energy

y u x u n n
m m m dD
u
& & & + = = (4.20)
are called yield lines.
Yield points
In the following the flow rule is applied to the (strongly convex) singularity points of the yield
surface.
On the cone apex m
x
= m
u
, m
y
= m
u
and m
xy
= 0, Koiters flow rule [29] shows that the
strain-rate vector may lie anywhere on or inside the cone of the outward-pointing normal at the
singularity point (Fig. 4.2c). Hence, any combination of 0
1
& , 0
2
& with any principal
direction is possible. Note that every lateral line of the cone corresponds to a strain rate 0
1
& ,
0
2
= & in the yield direction. Thus, the positive singularity point corresponds to any number of
positive yield lines. The rate of energy dissipated equals

t n y u x u
m m m m dD
2 1
& & & & + = + = (4.21)
where n and t denote the principal curvature directions.
A similar analysis may be carried out for the negative singularity point (m
x
= m
u
,
m
y
= m
u
, m
xy
= 0). In this case, the principal curvature rates are both negative ( 0
1
& ,
0
2
& ).
Limit analysis
48
At a regime change, i.e. along the singularity line of the yield surface, a positive and a
negative failure section with curvature rate (4.18
1
) and (4.18
2
) are possible. The flow rule
permits any linear combination, with non-negative coefficients, of the two curvature rates, i.e.

+ +
+ = ) , , ( ) , , ( ) , , (
yx y x yx y x yx y x
k k & & & & & & & & & , 0 ,
+
k k (4.22)
The values and directions of the principal curvature rates following from (4.22) depend on the
magnitude of the coefficients k
+
and k

. The general statement is 0


2 1
& & . Finally, the rate of
energy dissipated at a point of the singularity line is given by
) ( ) (
+ + +
+ + + =
y u x u y u x u
m m k m m k dD & & & & (4.23)
Points subjected to more than one rotation rate dissipating the energy rate (4.21) or (4.23),
are called yield points.
4.2.4 Discussion
The yield surface and the associated flow rule describe the ultimate stress states and the
corresponding failure modes, establishing the relationships between statics and kinematics. For
regular parts of the yield surface, the statical and the kinematical approaches are equivalent.
However, the former or the latter approach is advantageous for strongly or weakly convex
singularities, respectively, such as an apex or a flat portion of the yield surface. Finally, a
collapse results when the points where the stress field is at the limit correspond to a valid
mechanism, i.e. when the statical and kinematical analyses converge to the same result.
The yield surface depicted in Fig. 4.2f is composed of flat portions connected by
singularity points.
The flat portions are confined to the lateral surfaces of the two yield cones. Each straight
line on the cones associates a particular yield line to a set of stress states, hence favouring a
kinematical approach. Since the strain rates show one vanishing principal curvature rate, the
slabs collapse surface will be developable. Yield line collapse mechanisms are governed by
the laws of motion of rigid bodies, see Chapter 5.2.
In the points of singularity of the yield surface apexes of the positive and the negative
yield cone and states describing the regime change the flow rule associates one stress state
with a set of strain tensors, hence favouring the statical approach. Being confined to one single
stress state, the apexes of the positive and the negative yield cone are statically irrelevant
because there cannot be any load transfer. By contrast, the states of stress on the singularity
line limit the load transfer capacity of the slab; once the positive and the negative yield strength
are reached the slab loses its redundancy and equilibrium conditions determine the stress field
(shear field and principal moment directions) within the yield regions developing in the load
transfer direction [16] (e.g. for isotropic reinforcement see Chapter 3.6.4).
Yield lines and yield regions are the basic tools of limit analysis. Failure mechanisms are
essentially fixed by yield lines; owing to kinematical or statical requirements, yield
regions develop locally between yield lines.
Kinematical yield regions result if the collapse considered includes positive and negative
failure intersections. Such a situation occurs for example at a re-entrant corner supported slab
section, seeFig.4.3a. Assuming an isotropic reinforcement the intersection between positive
and negative yield lines has to be orthogonal (static condition); on the other hand, a negative
yield line radiating from the re-entrant corner is required for kinematic admissibility. Statics
and kinematics match when developing a yield region, see Fig. 4.3b. By progressive rotation of
the failure sections the yield region transforms into an anticlastic surface between the rigid
collapsing segments of the slab.
Yield condition and flow rule Approximate limit analysis
49
Fig. 4.3: Yield region. Re-entrant corner supported slab: a) yield line failure mechanism,
b) kinematical yield region; statical yield region: c) central loaded simply
supported rectangular slab [9], d) uniformly loaded clamped square slab [10].
In statical yield regions the slab mobilises all geometrical reserves (Chapter 3.6.4) to
resist the applied loads. Referring to the studies of Fox, Fig. 4.3c and d depict the extension of
statical yield regions for the case of positive and negative isotropic reinforcement in a
centrally loaded, simply supported rectangular slab [9] and in a uniformly loaded clamped
square slab [10], respectively. Generally, statical yield regions occur in the vicinity of
supports.
4.3 Approximate limit analysis
The limit analysis of slabs is linked to two different failure regimes corresponding to the
yielding of the top and bottom reinforcements. Each regime lends itself to a kinematical
analysis, while the regime change favours a statical analysis. The interplay between kinematics
and statics obstructs the limit analysis, making it difficult to overcome the divergence between
upper- and lower-bound results.
a) b)
"Kinematical" yield region
c) d)
x
y
Q
"Statical" yield region
/4
x
y
"Statical" yield regions
Limit analysis
50
In order to simplify the analysis one may introduce an approximate limit analysis by
assuming a single regime as yield condition, constraining the top or the bottom reinforcement
to stay rigid. By limiting the failure modes to yield lines of one sign (see Chapter 5.2) the
approximate limit analysis favours the kinematical approach without excluding statical
considerations, hence a new limit analysis problem is defined. In order to emphasize the
difference to standard limit analysis problems, equilibrium solutions related to approximate
limit analysis problems are called upper-bound moment fields [41].
In Chapters 5.3.2 and 6, the approximate limit analysis provides the theoretical background
of the equilibrium method and the compatibility limit design method, respectively. Starting
with the kinematical analysis, the failure shape and the ultimate load are determined in a first
step, followed by a corresponding statical investigation.
4.4 Example application
The following application continues the analysis of the rectangular slab introduced in
Chapter 3.7.1 in order to illustrate the approximate limit analysis.
Fig. 4.4a reproduces the moment field distribution depicted in Fig. 3.10b. This is the radial
moment (m
1
) in the area of the triangular and the trapezoidal segments AFI and AGHF for
l
HF
= 0.697l /2. In both segments the tangential moment (m
2
) equals 0.071ql
2
. As basis for the
dimensioning of the reinforcement [66], the moments at the grid points (x, y) = (0.5l,0.25l )
are determined and represented in the normal moment variation diagram of Fig. 4.4b. On the
positive side, the m
n
-curves are tangent to the line m
n
= 0.071ql
2
, depending on the direction of
the second principal moment; an isotropic bottom reinforcement for the ultimate resistance
m
u
= 0.071ql
2
provides a safe and economical design (curve a). On the negative side, extreme
moments arise in the trapezoidal segment, in the direction of the internal boundary AF, when
approaching the corner A. The most efficient top reinforcement is obtained by reinforcing bars
orthogonal to AF (curve b). Using Cartesian directions and assuming a ratio = 0.25, the
required ultimate resistance in the x-direction equals
2
211 . 0 ql (curve c). Finally, using
isotropic reinforcement m
u
= 0.120ql
2
(curve d). The dimensioning may be refined by defining
regions in which the reinforcement content is adjusted.
Assuming an isotropic reinforcement with m
u
= 0.071ql
2
, one observes that the moment
field is composed of states at Boundary AC of Fig. 4.2.d, whereas along EF, the stress states
correspond to the singularity point A. Thus, each radial trajectory of the triangular and
trapezoidal segments is a potential yield line and points along EF are potential yield points of a
positive collapse mechanism. With reference to Fig. 4.4c, the yield lines AF, BE, CE, DF, and
EF and the yield points E and F produce a mechanism.
Since the moment field of Fig. 4.4a is composed of admissible stress states according to the
positive yield condition for isotropic reinforcement with resistance m
u
= 0.071ql
2
and since it is
compatible with the mechanism of Fig. 4.4c, the load q = m
u
/ (0.071l
2
) is the ultimate load.
Other layouts of the same yield line pattern define a new geometry of the trapezoidal and
triangular segments. For example, Fig. 3.10c shows the case with l
EF
= 0.5l. When comparing
the radial moment distribution (m
1
) with the yield line moment (m
2
= 0.070ql
2
) in Fig. 3.10d,
one observes that the radial moment values grow beyond the yield line moment, i.e. the stress
field no longer corresponds to a complete solution. As a rule, the complete solution will
correspond to the stress field with the most efficient load transfer and a more critical yield line
layout exists if a stress field associated to a certain yield line layout violates the yield condition
considered.
Approximate limit analysis Example application Conclusions
51
Fig. 4.4: Rectangular slab segment: a) statical analysis; b) yield condition; c) kinematical
analysis.
Assuming an isotropic reinforcement distribution corresponding to a resistance
2
071 . 0 ql m m
u u
= = (classical limit analysis), the stress states in region AJK of Fig. 4.4a are not
admissible. Using classical limit analysis one is not able to recognise the failure pattern of
Fig. 4.4c as the best yield line collapse. To make the solution admissible, the load transfer
within AJK would have to be realised by a Hencky-Prandtl solution (see Chapter 3.6.4).
Comparing the complex and unsolved classical limit analysis problem with the simple
approximate limit analysis problem, it can be stated that yield regions are local problems
resolvable with little additional reinforcement.
4.5 Conclusions
Caused by the yielding of the top or the bottom reinforcement, the slabs limit states are
distinguished as yield lines and yield points, depending on whether at failure one or more
rotation axes are possible. Allowing the combination of a particular failure mode with different
stress states, yield lines lend themselves to a kinematical approach. Conversely, allowing the
combination of a single stress state with different failure modes, the yield regions are best
investigated based on a statical approach. Generally, statical and kinematical analyses are
difficult to combine. Yield lines define the global failure pattern, while yield regions are
generally confined to small areas characterised by the union of yield lines of different sign
a)

m
n

211
120
53
71
b)
c)
d)
b)
c)
1
0.5
4/l
1.54/l
l 0.5
0.65 l 0.70 l
l 0.5
0.65 l
l 0.25
l 0.5
x
y
(71,71) (67,71) (0,71)
(50,71) (30,71) (11,71)
(0,71) (120,71)
1000 m / ql
2
1
m
2
(30,71)
71
0 40
0
40
40
H F I
G A
A B
C D
E F
a)
J
K
L P
O
N
M
H
I
G
A
L
P
O
N
M
1000 / ql
2
Limit analysis
52
(kinematical requirement) or by a mobilisation of all statical reserves (statical requirement).
Due to its complexity, standard limit analysis of slabs is of little practical interest.
To overcome the difficulties of standard limit analysis, it is suggested to select a single
failure regime as the governing yield condition, i.e. to prevent yielding in one reinforcement
layer by means of some extra reinforcement; yield line failures are limited to one sign.
Supplementing the kinematical study by statical considerations and thus establishing
compatibility, the approximate limit analysis problem is completely solved and the extra
reinforcement is quantified.
Consistent with the theory of plasticity and simpler than standard limit analysis, the
approximate limit analysis forms the basis of a review of Johansens equilibrium method
(Chapter 5.3.2) and of a new design method the compatibility limit design method
(Chapter 6).

53
5 Kinematic method
5.1 General
Kinematic methods mark the beginnings of the limit analysis of slabs. Starting from
experimental observations, Johansen [21] postulated that slabs collapse along straight lines. The
task of the analysis consists in determining the ultimate load and the failure mechanism.
Preceding the fundamentals of plasticity theory, Johansen suggested two alternative methods of
analysis: the work method and the equilibrium method.
The origin of the work method can be traced back at least to Galileo [12]. The application of
the work equation (which equates the dissipation rate to the work rate related to a collapse
increment) to slabs is suggested by the simplicity of the failure layout assumed. The yield line
pattern is defined with certain parameters and the ultimate state is determined by differentiating
the work equation with respect to these parameters.
The equilibrium method was pioneered by Ingerslev [19] and further developed by
Johansen [21]. The method derives from the study of internal forces on yield lines; shear
forces and twisting moments are accounted for by concentrated nodal forces at the yield
line end. Considering overall equilibrium of each rigid portion, the extent of the slab
failure regions is adjusted in order to obtain the same yield moment on the boundaries of each
region.
Almost twenty years after Johansens work, the yield line theory was discussed within the
context of the theory of plasticity, at the time limited to the upper- and lower-bound theorem,
see Chapter 2.5. Prager identified the yield line theory as the application of the upper-bound
theorem to reinforced concrete slabs [52]. During the 1960s, cases of breakdown between the
equilibrium method and the work method [67] renewed the interest in yield line analysis.
Reviewing Johansens work, attempts were made to find a remedy for the anomalies of the
equilibrium method; in particular new rules and limitations of its applicability were established.
These attempts can be classified into two types. The first establishes a direct transition between
the work method and the equilibrium method [40, 68]. The second follows Johansens
procedure, investigating the physical meaning of the nodal forces on the basis of statical
principles [25, 41, 48, 22]. In the 1970s, Clyde identified nodal forces as vertical shear at and
parallel to strength discontinuities [4].
Although the equilibrium method involves an upper-bound analysis, its procedure and
the nature of the nodal forces evidence a lower-bound character. However, the equilibrium
method is unable to provide complete solutions in the sense of the theory of plasticity in
general since it does not account for yield regions [58]. Yet, a solid foundation of Johansens
work within the theory of plasticity is established by considering the approximate limit
analysis introduced in Chapter 4.3; with failure mechanisms formed by yield lines of unique
sign the statics and kinematics become compatible and complete solutions are created, see
Fig. 2.4.
Kinematic method
54
5.2 Failure mechanisms and limit analysis
Failure mechanisms
Whereas equilibrium considerations are related to an infinitesimal neighbourhood, failure
mechanisms link the failure analysis to the whole slab taking into account geometry and
boundary conditions. The failure investigation has experimental origins. Observing the ultimate
behaviour of slab specimens, the collapse mechanism is assumed to be formed by rigid parts
rotating along yield lines [21]. The failure shape is derived intuitively by considering the laws
of the mechanics of rigid bodies:
the yield line between two parts of a slab and their axes of rotation intersect in a point
(Johansens 1
st
theorem, [21]);
the yield line layout or the rotation axes and the rotation ratios of the various parts define the
collapse mechanism (Johansens 2
nd
theorem, [21]).
For a slab supported at its edges, the axes of rotation are fixed along the edges. For a slab
segment supported by a column or at single points, the axis must pass over the column or over
the support points. Any change of the yield line layout results in a variation of the axes and the
ratios of the rotations of the different parts.
As exemplary applications, Figs. 5.1a and 5.1b depict two possible collapse shapes for a
quadrilateral slab simply supported along two adjacent sides and on a column. The number and
the sign of the yield lines, i.e. the yield line pattern, define the failure mode of the slab. A yield
line layout corresponds to fixed rotation axes and rotation ratios of the various collapsing parts.
Limit analysis
Yield lines constitute lines of discontinuity of the deformation field. With reference to Fig. 5.1c,
the relative rotation rate
n

&
of two adjacent collapsing pieces A and B corresponds to a jump
A
n
B
n
w w
, ,
& & in the slopes of B and A. The localized deformation leads to the generalised strain rate,
when distributing the rotation rate in a thin yield line zone of thickness b [2]: b
n n
/
1

&
& & = =
and 0
2
= =
t
& & , n and t denoting the yield line and its perpendicular section, respectively. Via
the flow rule, the strain rate vector of a yield line is associated with an infinity of stress states
resulting in the same dissipation rate, cf. Chapter 4.2; redundancy subordinates the statical
analysis to kinematical requirements.
According to yield line theory both positive and negative yield lines in arbitrary directions
may intersect at a point. However, for any yield line there is only one yield line of opposite sign
such that a compatible stress state can be associated at the intersection point, cf. Chapter 4.2.
Hence, except for special cases, intersections of positive and negative yield lines are not
possible from a limit analysis point of view; yield regions have to be considered in such cases.
Regarding yield line mechanisms it can be stated that although no compatible stress state can be
Fig. 5.1: Failure analysis: a) and b) yield line patterns for a quadrilateral slab simply supported
on two sides and one column; c) (positive) yield line deformation.
a) b) c)
n
A B

.
b n
n
t
A B
z
w
.
Failure mechanisms and limit analysis Upper-bound method
55
associated to positive and negative yield line intersections in general the work equation remains
unaffected since the integration width and hence the dissipation at the intersection point
vanishes.
Starting from yield line considerations and establishing compatibility with the statical
analysis, the approximate limit analysis (see Chapter 4.3) restricts yield line mechanisms to be
formed by yield lines of unique sign. There are exceptions to this rule, i.e. yield lines which may
be replaced by a boundary condition (e.g. along clamped edges) or which act like a support (e.g.
rotation axis over the column in Fig. 5.1b).
5.3 Upper-bound method
In plasticity theory, the kinematical analysis is governed by the upper-bound theorem and
complete solutions may be singled out with a statical analysis by applying the compatibility
theorem (see Fig. 2.4). In reinforced concrete slabs, the application of the upper-bound theorem
to yield line failures leads to the work method. Considering approximate limit analysis
problems, the application of the compatibility theorem corresponds to applying the equilibrium
method.
5.3.1 Work method
In a failure mechanism, the work produced by the external load q on a deflection w is given by

=
surface
dA w q W (5.1)
while the energy dissipation equals
M d D
lines yield
=

(5.2)
where and dM indicate the rotation vector and the moment vector on an infinitesimal yield
line portion, respectively.
In the case of yield line failures (see Chapter 5.2), Eq. (5.2) may be evaluated by means of
the normal method or the component vector method [23]. The normal method focuses on the
analysis of single yield lines determining the energy dissipation
n n n
l m
u
, where
n
,
u
n
m and l
n

are the yield line rotation, the ultimate bending moment and the yield line length, respectively.
The component vector method focuses on the analysis of the collapsing pieces, making use of
the invariance of the rotation rate. Referring to the global system of reference xy defined by the
reinforcement and to the ratio between the ultimate bending moments in x- and y-direction,
the energy dissipated along the boundary of a collapsing piece equals
y u y x u x
l m l m + . The
parameters
x
,
y
,
u
m ,
u
m and l
x
, l
y
designate the yield line rotation components, the yield
moments and the projection lengths of the boundary yield lines of the piece considered, all in
the x- and y-directions, respectively. The total dissipation is obtained by summing the
dissipation of each yield line or of each collapsing piece, respectively.
The approximation of the ultimate load (limit analysis) or the estimation of the required
amount of reinforcement (limit design) follows by equating Eq. (5.1) to Eq. (5.2).
For a fixed yield line pattern, the most critical failure mechanism is determined by defining
the yield line pattern with certain parameters and looking for the maximum ratio W / D.
Kinematic method
56
Examples
In the following the work method is applied to four examples. Starting with the analysis of an
isotropically reinforced slab, the application is extended to the case of orthotropic
reinforcement. The last two examples investigate the reliability of the work method by
comparing different upper-bound studies.
a) Rectangular slab
Consider the uniformly loaded simply supported rectangular slab with side lengths l and 2l
introduced in Chapter 3.7.1. For a given load intensity q, the work method is applied to estimate
the required plastic moment m
u
(isotropic reinforcement).
Fig. 5.2a depicts the failure mechanism on the basis of the upper-bound analysis. The
collapse shape divides the slab into two trapezoidal ( ABEF, CDEF ) and two triangular ( AFD,
BCE ) segments. Since the axes of rotation of the trapezoidal segments do not meet, the central
fracture line EF is parallel to the edges AB and CD. Considering the symmetry in the geometry,
the load and the resistance distribution, the yield line layout is defined by a single parameter, for
instance the length x of the line EF.
Assuming a unit deflection of line EF (see Fig. 4.4c), the work produced by the external
loads at failure equals ] 2 / 3 / ) 2 [( lx l x l q + .
During the collapse, the fracture lines EF and AF (or, equivalently, BE, CE and DF) rotate
by
EF
= 4 / l and )] 2 ( /[ ] ) 2 ( [
2 2
x l l l x l l
AF AF
+ = , respectively, where =
2
AF
l (5l
2
4lx + x
2
) /4.
Summing the contributions of each yield line, i.e.
AF u AF EF u EF
l m l m 4 + , the normal method
leads to a total dissipation of )] 2 /( 4 8 [ x l l m
u
+ . Alternatively, the energy dissipation may be
determined with the component vector method. Due to symmetry, only the pieces ABEF and
AFD are considered. The trapezoidal and the triangular segments rotate by 2/ l and 2 / (2l x)
respectively. The ultimate bending moments at the sections BE-EF-FA and AF-FD have the
resultants 2m
u
l and m
u
l in the directions BA and AD, respectively. Hence, the dissipated energy
equals 2(
ABEF
M
ABEF
+
AFD
M
AFD
) = m
u
[8 + 4l / (2l x)].
Equating the external work to the total dissipation, one obtains m
u
/ ql
2
= (4+ x / l) (2 x / l)/
/ [24 (5 2x / l )]. Fig. 5.2a depicts the variation of m
u
/ ql
2
as a function of x / l. The most critical
mechanism results for x = 0.697l, where m
u
= 0.071ql
2
.
b) Triangular slab with one free edge
Consider the triangular slab ABC simply supported along the edges AB and BC and free along
AC subjected to a uniformly distributed load q, see Fig. 5.2b. The edges AB and BC have
lengths of 3l and 2l, respectively. The following analysis aims at determining the ratio
between the ultimate bending moments in the y- and the x-direction, so as to obtain the highest
load carrying capacity for a fixed reinforcement amount.
The kinematical study is performed with a single positive yield line radiating from B. The
yield line pattern is defined by the distance x of point D from the boundary BC, see Fig. 5.2b.
When D undergoes a unit deflection, the work of the external load equals ql
2
. During collapse
the yield line BD rotates by
BD
= [4 (3l x )
2
+ 9x
2
]/ [6x (3l x) l
BD
], where 9 / ) 3 ( 4
2 2 2
x l x l
BD
+ = .
According to Eq. (4.4
1
) the bending resistance in the failure section equals
2 2 2 2
/ ) 9 /( ) 2 6 (
BD u BD u n
l x m l x l m m
u
+ = , where m
u
denotes the bending resistance in the x-
direction. The dissipation of energy obtained with the normal method is given by
)] 2 6 /( 3 ) 3 /( ) 3 ( 2 [ x l x x x l m l m
u BD n BD
u
+ = . Note that the twisting moment
) 3 /( ) 3 ( 2 ) 1 (
2
BD u tn
l x l x m m
u
= acting together with
u
n
m on the failure section does not
dissipate energy. The same results may be obtained with the component vector method. The
yield line rotation vectors of ABD and BCD are given by ) 0 , ) 2 6 /( 3 ( x l
ABD
= and
) / 1 , 0 ( x
BCD
= , respectively. The moment resultant on section BD of segment ABD is
) 3 / ) 3 ( 2 , ( x l m x m
u u ABD
= M and that of segment BCD is the reaction vector
ABD
M .
Finally,
ABD
M
ABD
+
BCD
M
BCD
leads to the same energy dissipation.
Upper-bound method
57
Fig. 5.2: Work method applied to uniformly loaded slabs: a) simply supported rectangular slab;
b) triangular slab with one free edge (orthotropic reinforcement); c) rectangular slab
with one free edge; d) square slab.
D
A
C
B
F E
/2 x
a)
l
l
2
0.010
0.697
A
C
B
l 2
D
3l
b)
c)
d)
0.071
x /2 l x
/2
l/2
m
u
m
u
x x
/ l x
3
0.10
1.20
0.50
0.53
1.46 0.96
= 1
= 0.5
= 2
q
q
ql
2
ql
2
m
/ l x
1.25

x
y
u
1
2
m ( + )
u

.
.
.
.
.
.
. . .
.
.
.
.
.
.
ql
2
=
ql
2
= a = 0.14l
ql
2
b = 0.30l a = 0.14l
a a
b
24
1
22.19
1
=
22.20
1
l
l
q q q
m
u
m
u
m
u
a
b
q
a
b
q
m
m
x
x
m
m
0.1
0.50 / b a
qab
m
u
0.60
0.68
= 0
0.5
= m m
u
m
u
m=
m
Kinematic method
58
External work and energy dissipation are equated and solved for the ratio between the
average resistance m
u
(1+) / 2 and the load portion ql
2
: m
u
(1 + ) / (2ql
2
) = 3x (3l x) (1 + ) /
/ [9x
2
+ 4 (3l x)
2
]. For a fixed value of , the most critical mechanism results when
m
u
(1 + ) / (2ql
2
) has a maximum. On the other hand, the highest load carrying capacity with the
smallest amount of reinforcement is given by the value of for which m
u
(1 + ) / (2ql
2
) at the
point of the most critical layout has a minimum. The diagram of Fig. 5.2b depicts this study
graphically. For = 1 the most critical layout corresponds to x = 1.2l. By increasing , x is
reduced, e.g. for = 2 one obtains x = 0.96l. Conversely, by reducing , x increases, e.g. for
= 0.5 one obtains x = 1.46l. The dotted curve connects the points corresponding to the most
critical layouts for different values of . This curve has a minimum for = 1, i.e. isotropic
reinforcement results in the most economical reinforcement layout.
c) Rectangular slab with one free edge, comparison of failure mechanisms
The next example considers a rectangular slab, simply supported along three edges and free
along the fourth. The side lengths are equal to a and b, respectively, where b refers to the free
edge. The slab is loaded by a uniformly distributed load q and a boundary moment m > 0 along
the free edge. The following study evaluates the required positive resistance m
u
(isotropic
reinforcement) as a function of the ratio a / b and m.
Symmetrical failure mechanisms formed by positive yield lines radiating from the doubly
supported corners are considered. The parameter x indicates the distance between the yield line
intersection and the support with length b. To account for the boundary change at x = a, the
cases x < a and x > a are differentiated, see Fig. 5.2c. The upper-bound study is summarised by
the diagram of Fig. 5.2c. Mechanisms x < a govern failure for slabs with a >> b. Vice versa,
mechanisms x > a occur for a << b. Paradoxically, two yield line layouts of the failure mode
considered correspond to the same ultimate load at the transition between the two modes when
keeping the value of m constant. For instance, the yield line layouts x = 0.84a and x = 1.14a
correspond to the same ultimate load q = 12.43m
u
/ ab, when a / b = 0.68 and m = 0. For
m = 0.5m
u
and a / b = 0.60, the yield line layout x = 0.55a and x = 0.66a result in
q = 12.16m
u
/ab. Only for m = m
u
and a / b = 0.50 the mode change corresponds to a single yield
line layout, i.e. x = a (q = 12m
u
/ ab).
The found paradox highlights a controversial aspect regarding statical and kinematical
boundary conditions. Depending on the failure mechanism considered, boundary loads produce
different work rates; hence they have a different influence on the ultimate load. In Fig. 5.2c the
boundary moment m corresponds to W = 0 for x < a and W = m (a /x 1) for x > a.
d) Square slab, comparison of failure mechanisms
A uniformly loaded, simply supported square slab with isotropic reinforcement corresponding
to an ultimate resistance m
u
is considered. The investigation of the slabs ultimate load is
improved by starting with a simple failure mechanism and increasing the complexity of the
yield line pattern progressively, see Fig. 5.2d. In the first mechanism, the yield line pattern is
fixed by symmetry. The ultimate load follows directly from the work equation, q = 24m
u
/ l
2
. The
second mechanism allows the corners to rise and rotate. The yield line layout is defined by the
distance a between a corner and the intersection of the yield line with the support. This
mechanism includes the first mechanism as special case when a = 0. The most critical layout is
given by a = 0.14l and corresponds to the ultimate load q = 22.20m
u
/ l
2
(7.5% less than the first
approximation). In contrast to the second mechanism, the third yield line pattern is not anchored
at the slabs centre. In addition to the parameter a, the geometry of the corner segments is fixed
by the parameter b. The previous two mechanisms are included as special cases in the third one.
The ultimate load q = 22.19m
u
/ l
2
corresponds to a = 0.14l and b = 0.30l.
Increasing the number of yield line parameters seems to constitute a straightforward
procedure to improve the upper-bound analysis [30]. However, referring to Fig. 5.2d, the first
mechanism agrees with Pragers solution [52], i.e. q = 24 m
u
/ l
2
is the correct ultimate load. How
Upper-bound method
59
is it possible to obtain a better upper-bound approximation of the ultimate load with the Maltese
cross mechanisms? The paradox disappears when noting that the Maltese cross mechanisms are
related to the collapse of a square slab with unrestrained corners whereas in Pragers solution
the corners are restrained, i.e. two different problems have been considered.
5.3.2 Equilibrium method
Starting from the compatibility theorem (see Chapter 2.5), the equilibrium method is related to
the complete solution of the approximate limit analysis problem (see Chapter 4.3). The
procedure and the limits of the equilibrium method are illustrated by means of practical
examples.
1) Approximate limit analysis
The approximate limit analysis investigates the ultimate behaviour of slabs with a fixed amount
of reinforcement in the bottom (or in the top) layer, and a negative (or positive) resistance
sufficient to prevent failure. Similar to the kinematical requirements necessary for the
admissibility of a yield line pattern (see Chapter 5.2), the following section formulates the
statical requirements for a permissible failure. If a yield line pattern satisfies the kinematical and
the statical requirements, then the validity of the ultimate solution depends only on the state of
stress within the failure segments. In the sense of a plasticity check, the ultimate stress field is
developed by considering the basic segments of the stress field approach (see Chapter 3.6.7) as
the failure pieces of the collapse mechanism. Solutions for isotropic reinforcement will be
discussed first, followed by an extension to orthotropic reinforcement by means of the affinity
theorem.
Isotropic slabs: basic considerations
Statical constraints for the collapse layout derive from equilibrium considerations in the narrow
yield line zone, assuming the ultimate state conditions outlined in Chapter 4.3. The analysis
concentrates on the shear flow across a yield line, distinguishing the cases of a distributed and a
concentrated load transfer (see Chapter 3.4).
Consider a yield line failure. Assuming a continuous stress field in the narrow yield line
zone, the collapse constrains one principal moment value to the ultimate resistance (m
1
= m
u
)
and its trajectory to the yield line direction (1/
v
= 0). Introducing these limitations into
Eqs. (3.28), the shear field on a yield line point is determined by

Su n
m v
, 1
= ; ) (
1
2 , 2 u
u
Sv t
m m m v + =

(5.3)
Eqs. (5.3) refer to the notation introduced by Fig. 5.3a. Since m
1
may not increase beyond m
u
, v
n

vanishes, i.e. failure sections in slabs identify principal shear trajectories or lines of zero shear
when v
t
= 0.
Provided that the slab is devoid of strong band reinforcement, concentrated load transfer
depends only on the moment distribution. Fig. 5.3b illustrates a shear line intersecting a yield
line at the angle in P. The axes n and t define a local system of reference in the shear line
direction, and m
n
and m
tn
are the internal moments acting outside the shear line. According to
(3.22), the bending moment is constant on both shear line sides. Assuming a continuous stress
field, the yield line moment (principal section) and the moment m
n
in the n-direction correspond
to a twisting moment (m
u
m
n
) / tan, see Fig. 5.3c. Finally, the shear force in P is determined by
the twisting moment change between the regions adjacent to the shear line (Eq. (3.21)):

tn
u n
P
m
m m
V +

=
tan
(5.4)
Kinematic method
60
Fig. 5.3: Shear transfer across yield lines: a) distributed shear transfer; b) concentrated shear
transfer; c) Mohrs circle for internal shear line moments.
Concentrated shear forces may also be due to a shear field singularity. Typical examples
include point loads or point reactions (see Eq. (3.26)). Combining a fictitious point load with a
fictitious reaction, singularities transfer a concentrated force across adjacent failure segments
(e.g. see Fig. 3.10c). However, for consistency of the approximate limit analysis problem shear
singularities have to be excluded; note that the internal forces at a point of singularity violate the
laws of transformation (Eqs. (3.4) to (3.7)) and hence the equilibrium requirements.
Since yield lines correspond to principal shear directions or lines of zero shear and adjacent
failure segments may exchange loads in a concentrated way only at shear lines, a collapse shape
is statically admissible if its failure segments together with the yield moment m
u
, zero shear
forces and available shear line forces on the boundaries provide equilibrium. To prove the
validity of the assumed ultimate load, it is necessary to check that the stress values do not
violate the slab resistance. The corresponding equilibrium analysis for the single failure
segments may be performed with the generalised strip method (Chapter 3.6.1).
Orthotropic slabs: the affinity theorem
In orthotropic slabs, principal moments and principal sections on yield lines are dependent on
each other (see Eq. (4.9)), not complying with the generalisation of Eqs. (5.3). In the following,
orthotropic slabs are reduced to isotropic slabs on the basis of the affinity theorem [21, 47].
Orthotropic slabs are obtained by stretching isotropic slabs in one reinforcement direction.
The affinity theorem defines the transformation of the geometry and of the stress distribution
maintaining equilibrium. With reference to Fig. 5.4a, Table 5.1 summarises the relationship
between the isotropic and the affine slab with a ratio between the bending resistances in the y-
and x-directions.
b)
m
n
m
tn
Pole
P

V
P
n
t
2

m
u m
n
m
tn
v
n
t
u
u

v
n
m
u
v
t
m
2
1/ = 0
a)
m
u
m
n
tan
m
tn
m
n
m
n
m
u
m
n
m
u
m
n
tan
c)
Upper-bound method
61
Consider an isotropic slab. Fig. 5.4b (left) depicts a yield line enclosing the angle with the
x-axis. On the yield line section, the stress resultants v
1
= 0, v
2
, m
1
= m
u
and m
2
(see Fig. 5.3a)
correspond to the shear vector (v
x
,v
y
) = (v
2
sin , v
2
cos) and the moment vector
(m
x
,m
y
,m
yx
) = (m
u
sin
2
+ m
2
cos
2
, m
u
cos
2
+ m
2
sin
2
, (m
u
m
2
) sin cos) in Cartesian directions.
Stretching the slab according to the coefficient , the yield line slope in the x-y-plane becomes
tan , and the affine stress resultants are ) , (
y x
v v and ) , , (
yx y x
m m m , see Table 5.1.
On the yield line section of the orthotropic slab the stress resultants are obtained from
considering the equilibrium of an infinitesimal triangular element generated by the Cartesian
axes and a yield line portion of unit length, see Fig. 5.4b (right):
sin cos ) 1 ( ; ) sin (cos ; 0
2 2
= + = =
u tn u n n
m m m m v
u u
(5.5)
where
2
tan 1 / tan cos + = ,
2
tan 1 / 1 sin + = and (n,t) denotes the system of
reference of the yield line section in the orthotropic slab. Similar to isotropic slabs, the shear
force on yield line sections of orthotropic slabs vanishes (see Eq. (5.5
1
)), whereas the internal
moments are the yield moments (see Eqs. (4.4)).
In analogy to Fig. 5.3b, Fig. 5.4c aims at determining the shear line forces in the case of
orthotropic reinforcement. Consider a yield line - shear line intersection with the angle .
Introducing the systems of reference n, t and n, t in the yield line and the shear line directions,
respectively, the yield moments and the shear line moments are fixed by (
u
n
m ,
u
tn
m ) and
(m
n
, m
tn
). By means of a Mohrs circle analysis, the twisting moment within the shear line is
found to be
u u
tn n n
m m m tan / ) (
'
. Thus, the twisting moment change between the regions
adjacent to the shear line determines the shear line force

n t tn
n n
P
m m
m m
V
u
u

+ +

=
tan
(5.6)
As in isotropic slabs, the statical admissibility of the yield line pattern in the case of orthotropic
reinforcement is established by Eqs. (5.5) and (5.6) by verifying the equilibrium of the failure
segments. The stress field associated to the failure mechanism is best obtained by the affinity
theorem, since in contrast to orthotropic reinforcement the yield line pattern for an isotropic slab
suggests the principal moment trajectories. Note that the affine transformation also applies to
the principal shear trajectories, but not to the principal moment trajectories.
2) The equilibrium method
The equilibrium method suggests a statical approach for the estimation of the slabs ultimate
load according to failure modes which differ from the optimum mode. Stress fields satisfying
equilibrium and compatibility with the failure mechanism considered replace the differentiation
process of the work method. The following discussion concentrates on isotropic slabs;
orthotropic problems can be treated by applying the affinity theorem.

Isotropic slab Orthotropic slab
Geometry: - length
) , ( y x
) , ( y x
- slope tan tan
Loads: - distributed load
q
q
- line load
q
)
2
sin
2
/(cos + q
- point load
Q
Q
Stress resultants: - shear force
) , (
y
v
x
v ) , (
y
v
x
v
- moment
) , , (
yx
m
y
m
x
m
) , , (
yx
m
y
m
x
m
Table 5.1: Affinity theorem: transformation laws.
Kinematic method
62

Fig. 5.4: Affinity theorem: a) transformation laws [47]; b) distributed shear transfer;
c) concentrated shear transfer.
y
x
y
x
a)
a
b
Q
q
q
Q
q

Isotropic Slab Orthotropic Slab


tan
q
tan
a
b
x
y
1
2
1
1
m
u
m
u
tan

1
x
y
t
n
m
m
m
y
m
xy
v
y
m
xy
v
x
x
m
2
2
v
v
2
2

m
y
m
xy
v
y
m
m
xy
v
x
x
m
m
tn
v
n

1 tan
2
+
tan
1 tan
2
+
1
sin

b)

sin
2
+
cos
sin cos

c)
m
n
m
tn
Pole
P

V
P
n
t
2

m m
t
n
nu
m
tnu
n
u
u
n m
n
u
m
n
tan
m
tn
u
m
n
u
m
n
tan
m
tn
u
m
n
m
t n
m
n
m
n
u
m
n
tan
m
tn
u
m
n
m
nu
m
tnu
cos
2
1
1
1
Upper-bound method
63
Assume a slab stress field satisfying equilibrium, being compatible with a kinematically
admissible state of deformation. Since equilibrium is satisfied, the work equation for the
collapse considered is inherently fulfilled. Eq. (2.7) gives the value of m
u
as a function of the
geometrical parameters defining the yield line pattern. The statical-kinematical compatibility
requirement (i.e. m
n
= m
1
= m
u
, 1 /
v
= 0, see Fig. 5.3a) ensures the validity of (5.3
1
) along yield
lines. Now, relating the yield line definition to the system of reference of the principal moment
trajectories, the condition 0
, 1
= =
Su n
m v introduced to avoid a violation of the yield condition
by the approximate limit analysis becomes the maximum condition of the work method. Note
that, in addition to v
n
= 0 and m
n
= m
u
on the failure sections, the equilibrium of the collapse
segments includes shear line forces (5.4). In conclusion, if a yield line layout satisfying the
statical and the kinematical requirements may be completed to give an equilibrium solution,
then the equilibrium method indicates the ultimate load for the failure considered.
Considering mechanisms in agreement with the approximate limit analysis (Chapter 5.2),
the equilibrium method splits kinematically permissible failures into two groups, i.e. compatible
and incompatible mechanisms. The most critical compatible mechanism corresponds to the
solution of the approximate limit analysis problem.
3) Examples
In the following the application of the equilibrium method is illustrated for the examples
introduced in Chapter 5.3.1.
a) Rectangular slab
Consider the rectangular slab and the failure shape illustrated in Fig. 5.2a. Fig. 5.5a depicts the
failure segments ABEF and BCE as free body diagrams. On the segments, the distributed load q
has the resultants 4 / ) 2 ( x l ql Q
ABEF
+ = and 4 / ) 2 ( x l ql Q
BCE
= at a distance ) 3 6 /( ) ( x l x l l + +
from AB and 6 / ) 2 ( x l from BC, respectively. For isotropic reinforcement and seeking
statically admissible yield line patterns, the stress resultants on the yield sections include only
the yield moments m
u
, see Chapter 5.3.2-1. Concentrated shear forces within the slab are not
required, since m
n
= m
u
and m
tn
= 0 (see Eq. (5.4)). Finally, the reactions r
AB
and r
BC
(including
available shear line forces) are unknown.
Combinations of x and q fulfilling the statical requirements are found providing equilibrium
of the failure segments. Avoiding the reaction forces, equilibrium is established by a moment
equation with respect to the supports: 0 12 / ) ( 2
2
= + l x ql lm
u
and 0 24 / ) 2 (
2
= x l ql lm
u
for
ABEF and BCE, respectively. Only the mechanism 2 / ) 13 5 ( l x = for
2
) 13 /( 96 l l m q
u
=
is statically admissible.
Figs. 3.10a and 3.10b prove the validity of the equilibrium method. In addition to
equilibrium and compatibility, the stress field fulfils the positive yield condition (
u
m m m
2 1
, )
giving the ultimate load of the approximate limit analysis problem (see Chapter 4.3).
With reference to Fig. 3.9 and to the stress field example of Figs. 3.10c and d, it is observed
that yield line layouts with 2 / ) 13 5 ( l x maintain equilibrium through a shear singularity in
E. However, according to the assumption of Chapter 5.3.2-1, such solutions are not admissible.
b) Triangular slab with one free edge
The application of the equilibrium method to orthotropic reinforced slabs is illustrated by the
example of the triangular slab shown in Fig. 5.2b.
Based on the failure mechanism formed by one yield line that radiates from the corner B to
the point D on AC, Fig. 5.5b illustrates the collapse segments as free body diagrams. The
distributed load on ABD and BCD is concentrated into the resultants Q
ABD
= q(3l
2
lx) and
Q
BCD
= qlx at a distance (6l 2x) / 9 from AB and x / 3 from BC, respectively. Considering an
orthotropic reinforcement with ratio between the y- and the x-direction bending resistances,
the stress resultants on BD are fixed by Eqs. (5.5): v
n
= 0, m
n
= m
u
[4(3l x)
2
+ 9x
2
]/ [9x
2
+ 4(3l x)
2
]
Kinematic method
64
Fig. 5.5: Equilibrium method: a) simply supported rectangular slab; b) triangular slab with one
free edge (orthotropic reinforcement).
and m
tn
= m
u
( 1)6x (3l x) / [9x
2
+ 4(3l x)
2
]. In D, Eq. (5.6) predicts the shear force
l x l x m V
u D
18 / ) 9 12 4 ( + = . Finally, r
AB
and r
BC
on AB and BC are unknown.
The equilibrium of ABD and BCD is formulated with moment equations about AB and BC,
i.e. 2ql(3l x)
2
/ 9 + m
u
[4(3l x)
2
9x(6l x)]/ 27l = 0 and qlx
2
/ 3+ m
u
[4(x
2
9l
2
)+ 9x
2
]/ 18l = 0. For
fixed , the yield line layout ) 9 4 /( ) 3 2 ( 6 = l x satisfies the statical requirements for
2
/ 2 l m q
u
= .
In the case of orthotropic reinforcement, the application of the generalised strip method
within the failure segments is hindered by the dependence between the principal sections and
the principal moment values. The difficulty is avoided by using the affinity theorem. For = 1,
the critical failure is given by x = 1.2l and q = 2m
u
/l
2
. Figs. 3.11e and 3.11f depict the stress
field of the isotropic case. Since the positive yield condition is never violated, the stress field is
the ultimate solution of the approximate limit analysis problem. The same may be stated for
orthotropic reinforcement, see Chapter 5.3.2-1; the affinity between the equilibrium method and
the work method may be checked in the particular cases selected in Fig. 5.2b: = 0.5,
x = 1.456l, m
u
/ ql
2
= 0.707 and = 2, x = 0.961l, m
u
/ ql
2
= 0.354.
c) Rectangular slab with one free edge
Consider the rectangular slab of Fig. 5.2c with m = 0 collapsing to a mechanism x > a. Fig. 5.6a
depicts the failure segments as free body diagrams. In addition to the load q and the yield
moments m
u
(isotropic reinforcement), the free edge AD involves the shear line forces m
u
b / (2x)
b)
Q
BCD
D
V
D
m
n
B
x
y
C
x /3
r
BC
D V
D
t
B
n
Q
ABD
A
6l x
9
2
r
AB
a)
u
m
tnu
m
nu
m
tnu
m
u
B A
E F
Q
ABEF
r
AB
( + ) l x
3
l
+ l x 6
m
u m
u
l 2
x
B
C
E
Q
BCE
r
BC
m
u
m
u
( )/6 l x 2 ( )/3 l x 2
l
x/3 2
Upper-bound method
65
Fig. 5.6: Rectangular slab with one free edge: a) equilibrium method; b) compatible stress field;
c) selected mechanisms for 0.62 a/b 0.87.
in E and F. Finally, equilibrium of the failure segments ABF (or CDE) and BCEF is given by
moment equations with respect to AB (or CD) and BC: x = a / 3 + (4a
2
+ 9b
2
)
1/2
/ 6 and
qab = 24m
u
/ [[4(a / b)
2
+ 9]
1/2
2a/ b]. The analysis is valid if 0 < a /b < 0.87 (note that a / b = 0.87
indicates the mechanism with x = a).
A stress field for the rectangular slab compatible with the failure mechanism considered is
obtained by fitting the segments of Figs. 3.5e and 3.6c into BCEF, and ABF, CDE, respectively.
The following study focuses on the segment BCEF. In relation to Fig. 5.6a, the stress field
(3.59) has parameters a x a =
~
, x b =
~
, = m
~
qa[(4a
2
+9b
2
)
1/2
2a]/ 24, 0
~
=
a
m and 0
~
=
b
m ,
(hence =
2 ~
c a[(4a
2
+9b
2
)
1/2
2a]/ 12+(x a)
2
), where ~ indicates values related to Fig. 3.5e.
Compatibility of the failure mechanism is ensured by the tangential moment, which equals m
u

over the whole area. In contrast, the radial moment does not influence the failure. Investigating
the m
r
-distribution, it is observed that the radial moment attains a maximum on c x
~ ~
= (zero
shear line). Including the condition
u r
m m in the analysis, the positive yield condition is
fulfilled only if 62 . 0 / b a . Fig. 5.6b depicts the stress field of BCEF for 62 . 0 / = b a . Since
m
r
= m

= m
u
on c x
~ ~
= , the zero shear line may act as a yield line generalising the failure shape
considered, see Fig. 5.2c. In conclusion, the solution of the approximate limit analysis problem
is given by a mechanism x > a, one of Fig. 5.6c and one with x < a in the ranges 62 . 0 / b a ,
87 . 0 / 62 . 0 b a and 87 . 0 / b a , respectively. Similar analysis holds also for 0 m on AD.

1
0.62 0.70 0.78 0.87
0.42
0.45
0.48
0.50
0.27
0.40
0.37
0.44
0.47 0.54 0.63
v
0
=
0
0.40
1
0.62
[1000 m
r
/qab]
83.5 60 20
20
20
60
20
37.3
37.3
b
x
a
q
q
q
m
u
m
u
m
u
b
x 2
m
u
b
x 2
m
u
b
x 2
m
u
b
x 2
a) b)
c)
0.40
r
BC
A B
B
C
C D
E
F
r
AB
r
CD
m
u
83.50 qab
=
1000
m
u
80.32 qab
=
1000
m
u
77.71 qab
=
1000
m
u
74.81 qab
=
1000
y
x
b
~
c
~
a
~
~
~
Kinematic method
66
d) Square slab
In the statical analysis, the axes of rotation of the failure segments act as supports. The work
method procedure allows a parametric definition of the rotation axes. In contrast, variable
supports impair the statical approach, since a variation of the boundary conditions changes the
nature of the equilibrium problem.
In Fig. 5.2d, a defines the rise point of the slab corners, whence the reactions of the edge
segments. For an arbitrary value of a (a / l), Fig. 5.7a summarises the values of the parameters b
(b / l ) and of the yield moment m
u
(m
u
/ (ql
2
)) which provide equilibrium of the basic segments;
e.g. for a = 0.1l one obtains b = 0.32l and m
u
/ ql
2
= 1 /22.35.
A stress field compatible with the failure shape considered is generated by fitting the
trapezoidal segments EID and DIBC, EFI and FGHI, and IHB and HAB into the regions EIBC,
EGHI and IHAB, respectively. In EIBC and EGHI a strong band redistributes the load
transferred between the segments EID-DIBC and EFI-FGHI along ID and FI; in IHAB, the
segment boundary BH acts as support, being the rotation axis of IHAB. Fig. 5.7b and Table 5.2
summarise the stress field resulting when considering a = 0.1l. The force concentrated in I
(5ql
2
/ 10
4
) is distributed on ID and IF (3ql / 10
3
) involving a strong band moment
M = 1.08ql
3
/ 10
5
. Since m
r
,m

ql
2
/ 22.35, the stress field corresponds to the complete
solution of the approximate limit analysis problem.
A similar analysis holds for l a 25 . 0 0 . The best yield line/support layout (a = 0.14l in
Fig. 5.7a) is obtained with vertical equilibrium of the segment ABIH, assuming that ABIH is
supported by the shear line forces along GH and BC at B and H, respectively. In the Maltese
cross mechanism (b = l / 2 a) the stress field of Fig. 5.7b fits into the failure segments
determining a shear singularity at the slabs centre (see Fig. 5.7c; a = 0.14l, m
u
/ ql
2
= 1 / 22.20);
for a = 0 this singularity disappears and the stress field simplifies to Pragers solution [52].
The failure analysis of the square slab with the Maltese cross mechanism or the equivalent
mechanism replacing a rising corner with a negative yield line along BH marked the beginning
of the nodal force research [67]. Without going into details, the stress fields of Figs. 5.7b and c
are considered in relation to Clydes analysis [4] aiming at indicating the meaning of the
invalid nodal forces. In Fig. 5.7b, HB acts as a support also when assuming the function of a
negative yield line. The nodal forces in B and H of the segment BIH correspond to the support
load along BH, i.e. the load balancing the shear line force along CB (segment EIBC ) and along
BH (segment ABH) in B and the load balancing the shear line force along GH (segment EGHI )
and along BH (segment ABH) in H. In Fig. 5.7c the segments EGH, EBC and EHAB are in
equilibrium when considering the load exchange at the singularity point E (see Chapter 5.3.2-1).
In both cases, the values of the invalid nodal forces result by considering vertical equilibrium
of the failure segments.


Parameters
Segments
a b m m
a
m
b

EID; EFI 0 0.08l ql
2
/ 22.35 ql
2
/ 22.90
DIBC;FGHI 0.11l 0.53l ql
2
/ 22.35 ql
2
/ 22.90 0
IHB 0 0 ql
2
/ 22.35 ql
2
/ 1200
HAB 0 0.07l 0 ql
2
/ 1200
Table 5.2: Stress field parameters for the trapezoidal segments involved in Fig. 5.7b.
Upper-bound method Discussion
67
Fig. 5.7: Square slab: a) equilibrium analysis; b) example of compatible stress field (a = 0.10l,
b = 0.32l, m
u
= ql
2
/ 22.35); c) stress field for the Maltese cross mechanism (a = 0.14l,
m
u
= ql
2
/ 22.20).
5.4 Discussion
As an application of the compatibility theorem, the equilibrium method lies between the
kinematical and statical methods (see Fig. 2.4). Since compatibility refers to an extension of the
yield surface, the equilibrium method clearly constitutes an upper-bound method. Through
equilibrium, the failure analysis matches a complete structural analysis. Defining the limits of
applicability of the equilibrium method, the following considerations aim at illustrating the
advantages of compatibility.
a)
0.25
0.25
0.32
a
b
0.0447
ql
2
m
u
x
0.25 a
0.040
0.14 / l / l
b)
c)
l 0.36 l 0.14
y
[1000 m
r
/ql
2
]
v
0
=
0
= 1.62 m
n 0
40
60.06
0
23.40
0
0 40
40
+ 8
+ 8
/ l
x
l 0.32 l 0.10
y
l 0.08
[1000 m
r
/ql
2
]
0
40
0
44.74 = 43.67 m
n
= 0.83 m
n
40
0
25.96
40
0
A B C
H
E
D
F G
I
A B C
H
D,E,F,I G
0.10 0.10
0.045
Kinematic method
68
For a given yield line pattern, the applicability of the equilibrium method depends on the
approximate limit analysis and on the compatibility between the kinematical and the statical
problems (boundary conditions).
The approximate limit analysis relates the failure to the collapse of a single reinforcement
layer. As stated in Chapter 5.2, yield lines on rotation axes (e.g. along clamped edges) identify
boundary conditions in the associated statical problem, hence are free in the sign. Further,
different slab problems may be connected along rotation axes (e.g. see Fig. 5.1b). This makes
the majority of yield line shapes compatible with the approximate limit analysis; only failures
involving yield lines of different sign within the slab are excluded (e.g. Fig. 4.3a).
The relationship between statical and kinematical boundary conditions is controversial. In a
kinematical analysis, statical boundary conditions may lose their meaning. Conversely, a failure
mechanism may produce statical ambiguities. Examples are given by the applications c and d in
Chapter 5.3. When considering the mechanism x < a in Fig. 5.2c, one observes that the
boundary moment m does not influence the kinematical analysis (see diagram in Fig. 5.2c); in
contrast, compatible stress fields are influenced by the values of m (e.g. Figs. 3.10a and b result
in a lower-bound solution if m = m
u
). In Fig. 5.2d, the parameter a determines the rotation axis
of the edge segments. Apparently, the corner segment is supported at the intersection points of
the supports with the rotation axes. However, an arbitrary point of a rotation axis may be
considered as a support without changing the upper-bound result, since forces along lines of
zero displacement do not produce work (e.g. Fig. 5.7b assumes the whole rotation axis BH as a
support). Compared to the kinematical study, the statical analysis is linked to a fixed limit
analysis problem and is sensitive to boundary condition changes. The equilibrium method
breaks down when the mechanism on the basis of the kinematical analysis results in a change to
the statical problem.
Fig. 5.8: Basic segment with polar trajectories: a) triangular segment; b) trapezoidal segment;
c) trapezoidal segment with one free edge; d) stress field combination along internal
boundaries; e) stress field combination along yield lines.
m
u
m
u
h
a) b) c)
d) e)
m
u m
u
m
u
a
b a
= q
m
u
6
=
3
q
a 2 /b
2
a 3
2
b +
m
u m
u
a
b a
m
u
12
=
2
q
b a 2 ab +
2

r
tan m
u

r
tan m
u

l
r=h
m
u
6
=
2
q
h
Fig. 3.6
Fig. 3.5 e
c
m
u
m
u
m
u
m
u
6
2
h
Discussion Conclusions
69
The equilibrium method involves a kinematical and a statical study complementary to each
other. The more the two analyses are performed simultaneously, the better the equilibrium
method.
As with the work method, the kinematical analysis determines the collapse mechanism
and the ultimate load of the slab. The equilibrium of the failure segments replaces the
differential procedure of the work method with a simpler calculation and gives additional
information about the force flow in the slab.
Along yield lines, the ultimate moments and the shear forces on failure sections isolate the
individual collapse segments as single problems. Considering the failure segments as basic
elements of the stress field approach, the statical analysis leads to a compatible equilibrium
solution. The slab stress field goes beyond the requirements of the work method. Its
determination outlines the slabs structural behaviour, proving the consistency of the failure
analysis (statical boundary conditions of the problem), improving the investigation of the most
critical failures (complete solution of the approximate limit analysis problem) and giving
information about the actual failure of the slab (development of plastic regions). Examples of
these considerations are given in the above applications: comparing Fig. 5.2d with Fig. 5.7b, the
slabs boundary conditions are not clear; in Fig. 5.6c the equilibrium method suggests the
ultimate failure; referring to the standard limit analysis problem (m
u
= m
u
), the stress fields of
Figs. 3.10a, 3.10b and 3.11e, 3.11f define the extension of yield regions in the slabs corners.
Combining the kinematical and the statical approaches the procedure of the equilibrium
method is facilitated. Figs. 5.8a to c give information on the ultimate load of failure elements
resulting from the trapezoidal segment introduced in Fig. 3.5e. Referring to the stress field
approach, different segments may be connected along internal boundaries or along yield lines.
The former case leads to failure segments with complex boundaries (e.g. see Fig. 5.8d) or to the
detection of new yield lines within a single failure segment (e.g. see Fig. 5.6c, the trapezoidal
segment of Fig. 5.8c constitutes a new yield line if 6 3 ab (a+b) (4a
2
+ab +b
2
)
3/2
< 0). In the
latter case different segments with known statical conditions are combined along yield lines
(e.g. several segments of Fig. 5.8a correspond to a uniformly loaded slab simply supported
along boundaries tangent to the circle r = h [21, 59], see Fig. 5.8e).
5.5 Conclusions
Fig. 5.9 reviews the kinematical analysis of slabs outlined in this chapter. The upper-bound
analysis is restricted to yield line failures. Such failure modes are in good agreement with
experiments, involve only few geometrical parameters, and correspond to an approximate limit
analysis problem if only positive or negative failures are enforced (see Chapter 4.3). Compared
to Fig. 2.4, the upper-bound theorem is replaced by the work method and the compatibility
theorem by the equilibrium method. The best yield line pattern of the approximate limit analysis
problem is found in the complete solution group. The equilibrium method singles out the most
critical yield line pattern of collapse mechanisms which agree with an approximate limit
analysis and correspond to a fixed statical problem. The work method applies to arbitrary failure
modes.
The compatibility theorem provides a solid foundation for Johansens equilibrium method,
removing the paradoxes related to the nodal forces. By means of the approximate limit analysis
the relation between the lower- and the upper-bound analyses is simplified. The equilibrium
method improves the global analysis of the work method; by substituting kinematical by statical
boundaries it provides a better understanding of the slabs behaviour. The kinematical analysis
favours a global consideration of the force flow before concentrating on local equilibrium. As
Kinematic method
70
an intermediate between statics and kinematics, methods of compatibility offer advantages for
the design of new as well as for the examination of existing structures.
On principle, the approximate limit analysis procedure outlined here for slabs can be
extended to arbitrary structures. The global failure analysis is followed by a local equilibrium
analysis (stress field analysis) of the individual failure segments, always observing and if
necessary extending the yield limits.

Fig. 5.9: Kinematic methods: summary.
STATIC KINEMATIC
COMPATIBILITY
Approximate limit analysis
Equilibrium method Work
method
Complete solutions
Yield line
failures
71
6 Compatibility limit design
6.1 General
The common design procedures for reinforced concrete structures focus on the dimensioning of
the main reinforcement. Detailing considerations typically follow in a second step. If only one
equilibrium state is available, i.e. for statically determinate structures, the detailing directly
depends on the stress field. In contrast, the stress field of redundant systems may be adjusted to
obtain a practical and economical reinforcement layout.
In the following, the slabs redundancy is exploited to improve its detailing. By means of a
practical example, the flexibility of the plastic solutions outlined in the previous chapters is
illustrated. An appropriate detailing shall ensure a plastic behaviour of the reinforced concrete
as well as a practical distribution of the reinforcement suitable for placing, for casting of the
concrete and for safe working conditions. Starting with a kinematical study, the analysis
proposes a design concept and basic dimensions of the slab. The detailing is completed in a
second step, by developing an equilibrium solution matching the design concept.
6.2 Example application
The following example is taken from a continuing education course on the application of the
theory of plasticity to reinforced concrete [64] which was recently used [62] for a numerical
comparison between elastic, elasto-plastic and plastic equilibrium solutions. While [64] includes
a lower-bound study (strip method) and an upper-bound check (work method) [62] reviews
numerical approaches. Together, the two studies summarise the generally accepted methods for
reinforced concrete slab design. This chapter illustrates the advances made with the present
thesis.
Fig. 6.1: Example application: rectangular slab supported on three edges.
7m
5m
q= 30.6 kN m /
2
x
y
z
f
c
= 21 N mm
f
y
= 460
h = 240 mm
/
2
N mm /
2
Compatibility limit design
72
6.2.1 Problem statement
Fig. 6.1 shows the geometrical and statical parameters of the slab problem considered. It
involves a 7m by 5m rectangular reinforced concrete slab with a thickness of 240mm, free along
one of its long edges, simply supported along one of its short edges and clamped at the
remaining two edges. The total load q = 30.6kN /m
2
includes a dead load of 6 kN / m
2
and
superimposed dead and live loads of 11 kN / m
2
using a global safety factor = 1.8. The material
properties of the concrete and of the reinforcement are given by a compressive strength
MPa f
c
21 = and a yield strength f
y
= 460MPa, respectively.
6.2.2 Assumptions
While Chapters 3 to 5 focused on the structural analysis, the following assumptions aim to
simplify the design procedure. They consist in considering internal moments as force couples
with a fixed lever arm and in assigning the shear transfer to the slab core between the flexural
forces. Thus, proposals for stress field design for a distributed [33] and a concentrated [33, 45]
load transfer are synthesised.
Consider a stress state defined by the shear vector (v
x
,v
y
) and the moment tensor
components (m
x
,m
y
,m
xy
), using Cartesian coordinates. Introducing a sandwich model of the slab
with a constant internal lever arm d
v
[33] leads to in-plane forces (n
x
,n
y
,n
xy
) = (m
x
/ d
v
,m
y
/ d
v
,m
xy
/ d
v
)
and ( m
x
/ d
v
, m
y
/ d
v
,m
xy
/ d
v
) in the bottom and top sandwich layers, respectively. Provided that
the nominal shear stress due to the principal shear force, v
0
/ d
v
(
2 2 2
0 y x
v v v + = ), does not exceed
the cracking shear stress
2 3 / 2
/ 76 . 0 1 . 0 mm N f
c cr
= = , shear in the sandwich core may be
resisted by mobilising the concrete tensile strength. For v
0
/ d
v
>
cr
, the core should be assumed
to be cracked. The tension force which equilibrates the diagonal compression in the core
horizontally is distributed equally to the bottom and top sandwich layers, resulting in forces of
(
x x y x
v v v v , ,
2 2
) / (2v
0
tan), where denotes the diagonal compressive stress field inclination in the
cracked core (typically, / 6 < < / 3). Starting from the total in-plane forces (n
x
,n
y
,n
xy
), the in-
plane reinforcement in the bottom and in the top layer is given by

y
yx
y
x
sx
f
n
f
n
a
tan
+ ;
tan
y
xy
y
y
sy
f
n
f
n
a + (6.1)

y
yx
y
x
sx
f
n
f
n
a
tan
+ ;
tan
y
xy
y
y
sy
f
n
f
n
a

+ (6.2)
where < < 0 . If v
0
/ d
v
exceeds
cr
, additional reinforcement is necessary in the transverse
direction:

y v
z
f d
v

tan
0
(6.3)
Eqs. (6.1) and (6.2) correspond to the positive and the negative yield condition, see Eq. (4.6).
Applying Eqs. (6.1) to (6.3) it has to be checked that the compressive strength of the bottom
(6.4
1
) and the top layer (6.4
2
) as well as of the core of the sandwich (6.5) is not exceeded:
) (
y x y sy y sx c
n n f a f a cf + + ; ) (
y x y sy y sx c
n n f a f a cf + + (6.4)
) cot (tan
0
+
v
c
d
v
f (6.5)
In a strong band the stress states are defined by a shear force V and a bending moment M.
Assuming a constant internal lever arm d
v
and a shear transfer via a diagonal compressive stress
Example application

73
field in the concrete, the bottom (top) stringer force N = M/ d
v
(N = M/ d
v
) and the horizontal
force (Vcot) in the slab core require a reinforcement cross-section of
) 0 (
2
cot
+
y y
s
f
V
f
N
A

(6.6)
in the strong band direction, and a transverse reinforcement ratio

y v
z
f b d
V

tan
(6.7)
where b and denote the strong band width (i.e. the width given by the strong band
reinforcement) and the diagonal compressive stress field inclination, respectively. Eqs. (6.6) and
(6.7) require a sufficient compressive strength in the horizontal and transverse directions, i.e.

b d h
V
b d h
N
f
v v
c
) ( 2
cot
) (
+

(6.8)

cos sin b d
V
f
v
c
(6.9)
For shear lines, the concentrated load transfer results from moment discontinuity and the
horizontal slab core force (Vcot) is introduced at the shear line ends. The shear line design
concentrates on reinforcement and the compressive strength check in the transverse direction.
Assuming b = h / 2, Eqs. (6.7) and (6.9) govern the shear line design.
Setting d
v
= 200mm and = / 4 and assuming that the concrete compressive strength is not
exceeded in the top and bottom layers as well as in the core of the sandwich, the following
analysis concentrates on the dimensioning of the reinforcement using Eqs. (6.1), (6.2), (6.3) and
(6.7).
6.2.3 Detailing
It is important to include detailing considerations from the beginning of a design.
The plastic behaviour of reinforced concrete requires a minimum reinforcement capable of
resisting the cracking stresses (see Fig. 2.2c) and suitable for a good concrete/reinforcement
interaction. In slabs, the ultimate moment m
u
must not be smaller than the cracking moment m
cr
,
i.e.
cr u
m m . The necessary reinforcement is usually indicated as a percentage of the concrete
cross-section, typically 0.15%; in our case, this requires a reinforcement cross-sectional area per
unit length of the slab of about 360mm
2
/ m. The minimum reinforcement concerns the
reinforcement layer in tension, i.e. the bottom reinforcement in the interior of the slab and the
top reinforcement near the clamped edges.
Apart from flexural failures, potential shear failures have to be considered, primarily close
to concentrated loads and reactions. Experimental investigations [38, 20] have shown that
transverse reinforcement significantly increases the ductility of slabs failing in shear. Typical
transverse reinforcements include closed stirrups, open stirrups with end hooks at the top,
lapped hairpins, nail-head-anchored stirrups or multiple headed studs. Some of these
reinforcements are particularly suited for relatively thin slabs; slabs with a thickness of about
400mm and more should preferably be equipped with a minimum transverse reinforcement in
any case.
The plastic behaviour of reinforced concrete is improved by a good distribution and a
correct anchorage of the reinforcement and by a confinement of highly compressed concrete
areas [44]. Minimum and maximum bar diameters and spacings as well as minimum slab
Compatibility limit design
74
thicknesses have to be respected and a thorough consideration of the working conditions during
execution is necessary.
Practical aspects are of paramount importance for the quality of the final construction. For
example, a slab thickness h > 120mm allows for two layers of reinforcement and a reinforcing
bar spacing of at least three times the maximum aggregate size permits an easy placing of the
concrete. In addition, a homogeneous reinforcement layout restricted to a few reinforcement
positions with standard bar spacings (100, 150, 200 or 250mm) and taking account of erection
requirements (e.g. adaptable in the length and avoiding superfluous joints) results in a uniform
structure, improves the practical work and facilitates quality control. Finally, since the
reinforcement has to act as a support for personnel and equipment during casting of the concrete
the reinforcing bar diameters should generally not be smaller than 10mm and their spacing
should not exceed 250mm.
6.2.4 Kinematical analysis
Starting from a fixed geometry and a given resistance distribution, the kinematical analysis is
commonly performed to investigate the load carrying capacity, i.e. as a check of design
computations or of existing structures. Considering the problem in an inverted order, i.e.
assessing the resistance distribution required in a structure to withstand a given load
configuration, the analysis permits to derive a suitable design concept. This chapter provides an
example of this procedure.
Fig. 6.2a suggests a failure mechanism for the slab problem considered. The yield line
pattern develops from the free edge, by introducing a 1m wide strong band with positive and
negative resistance M
u
. The rest of the slab is considered to be isotropically reinforced at the
bottom (resistance m
u
) and to have a negative resistance of 1.44m
u
along the clamped edges AB
and AG. Note that the ratio of 1.44 corresponds to reinforcing bars 12 and 10 at the same
spacing.
Looking for the values of m
u
and M
u
, the failure mechanism formed by the positive yield
lines AH, BH, EH, and the negative yield lines AB, AF is adjusted to resist the load
q = 30.6kN /m
2
, considering H to be free along CG (parameter x). On the basis of the
equilibrium method (see Chapter 5.3.2), considering the segments ABH, AHEF and BDEH as
free bodies one gets
q x M m q x M m q m
u u u u u
2 2
) 7 ( 17 . 1 4 ; 17 . 1 2 76 . 9 ; 67 . 2 44 . 2 = + = + = (6.10)
and hence m
u
= 33.4kN, M
u
= 149.3kNm and x = 4.2m, requiring a basic reinforcement of
m
u
/ (d
v
f
y
) = 364mm
2
/ m at the bottom of the slab, a clamping reinforcement of
1.44m
u
/ (d
v
f
y
) = 524mm
2
/ m and a strong band reinforcement of M
u
/ (d
v
f
y
) = 1623mm
2
. Bottom
bars 10@200mm provide 393mm
2
/ m, clamping bars 12@200mm provide 565mm
2
/ m and
eight bars 16 in the strong band provide 1608mm
2
. These values are approximately equal to
the required cross-sectional areas. The bottom reinforcement corresponds to the minimum
reinforcement; hence a very ductile behaviour and an economical design can be expected.
6.2.5 Statical analysis
The statical analysis completes the failure mechanism selected in Chapter 6.2.4 by providing a
compatible equilibrium solution. As suggested in Chapter 5.3.2, the failure segments are
analysed individually by means of the generalised strip method (see Chapters 3.6.1, 3.6.5 and
3.6.6).
ABH is a trapezoidal segment with parameters a = 0, b = 4m, m = 33.4kN and m
b
= 48.2kN.
Eq. (3.60) gives c = 0 and, according to (3.58) and (3.59), the stress field is defined by
Example application

75

Fig. 6.2: Structural analysis: a) kinematical analysis; b) shear field and shear line forces;
c) moment field.
a)
b)
c)
33.4 48.2
100 kN/m v 0 =
0
149.3
1m
x
M
1.44
M
A B
F D
G C
21.1
33.2
H
87.2
133.9 92.3 123.6
69.3
95.1
41.0
69.2
33.4/115.5
0.1/133.3
31.1/79.8
33.4/70.8
23.4/132.3
40.2/2.2 47.2/10.1 146.2/1.0
E
V
57.4
85.4
78.0 47.5
AB
V
CG
V
BC
V
AG
G
C
A
B
B
C
A
G
kN
kN
kN
kN
kN
kN
u u
m
u
m
u
1.44 m
u
[ ]
kN [ ]
Compatibility limit design
76
44 . 33 ; 10 . 5 44 . 33 ; 30 . 15
2
0
= = =

m r m r v
r
(6.11)
where r and are polar coordinates centred at H. In addition to the reactions r
AB
= 81.6kN / m
and m
AB
= 48.2kN, the edge AB acts as a shear line, transferring the forces 85.4kN and 57.4kN
in A and B, respectively.
BDEH and AHEF are divided along CG into the segments BCH, AHG and CDEH, GHEF,
respectively. BCH and AHG follow the analysis suggested by Figs. 3.6c to 3.6e, while CDFG is
considered with the strip method of analysis (see Chapter 3.6.3). Assuming constant bending
moments and shear forces between the segments BCH and CDEH, and AHG and GHEF,
respectively, the internal forces along CG are determined proving equilibrium of the different
segments. From the moment equation of BCH (AHG) with respect to BC (AG) it follows that
v
CH
= 7.06kN / m (v
GH
= 3.51kN / m). Distributing v
CH
(v
GH
) on CDEH (GHEF) as cantilever load
of the slab strip in the y-direction (i.e. q
y CDEH
= 7.1kN / m
2
, q
y GHEF
= 3.5kN / m
2
), moment
equilibrium of CDEH (GHEF ) with respect to CG requires m
CH
= 3.5kN (m
GH
= 1.8kN).
In BCH, the stress field is defined by the superposition of
)
cos
81 . 0
( 26 . 18
2
0
r
r
v =

; 44 . 33
cos
82 . 14
09 . 6
2
2
+ + =

r m
r
; 44 . 33 =

m (6.12)
and
)
cos
43 . 15
( 96 . 2
2
0
r
r
v =

2
2
cos
71 . 45
99 . 0 = r m
r
; 0 =

m (6.13)
Eqs. (6.12) and (6.13) refer to a polar system of coordinates centred at H ( = 0 for HC) and at
B ( = 0 for BC ), respectively. The stress field (6.12) gives the reaction forces r
BC
= 58.0kN / m
and R
H
= 21.1kN as well as the shear line force F
B
= 47.5kN. Similarly, Eq. (6.13) corresponds
to r
CH
= 7.1kN / m, R
B
= 32.2kN and F
H
= 21.1kN. Note that R
H
and F
H
add to zero, while r
CH

corresponds to the load exchange between BCH and CDEH. The negative sign of R
B
indicates a
downward force in B.
In AHG, the stress field is defined by the superposition of
)
cos
74 . 1
( 92 . 19
2
0
r
r
v =

; 44 . 33
cos
67 . 34
64 . 6
2
2
+ + =

r m
r
; 44 . 33 =

m (6.14)
and
)
cos
19 . 12
( 62 . 4
2
0
r
r
v =

2
2
cos
35 . 56
54 . 1 = r m
r
; 0 =

m (6.15)
Eqs. (6.14) and (6.15) refer to a polar system of coordinates centred at H ( = 0 for HG) and at
A ( = 0 for AG ), respectively. The stress field (6.14) gives the reaction forces r
AG
= 94.6kN / m
and R
H
= 33.2kN as well as the shear line force F
A
= 78.0kN. Similarly, (6.15) corresponds to
r
GH
= 3.5kN / m, R
A
= 58.9kN and F
H
= 33.2kN. Again, R
H
and F
H
add to zero, while r
GH

corresponds to the load exchange between AHG and GHEF. The negative sign of R
A
indicates a
downward force in A.
The stress field of CDFG is developed with the strip method of analysis, by considering the
load partition q
x
= 37.7kN / m
2
and q
y
= 7.1kN / m
2
in CDEH, and q
x
= 34.1kN / m
2
and
q
y
= 3.5kN / m
2
in GHEF. Strips are clamped at FG and simply supported at CD in the x-
direction, while they act as cantilevers in the y-direction. Centring the global system of
reference at D, the stress field in CDFG is given by
CDEH: ) 53 . 3 , 83 . 18 05 . 106 ( ) , ( ; ) 06 . 7 , 66 . 37 05 . 106 ( ) , (
2 2
y x x m m y x v v
y x y x
= = (6.16)
GHEF: ) 76 . 1 , 06 . 17 05 . 96 07 . 14 ( ) , ( ; ) 51 . 3 , 11 . 34 05 . 96 ( ) , (
2 2
y x x m m y x v v
y x y x
+ = = (6.17)
Example application

77
Finally, the reaction forces along CD and FG are equal to r
CD
= 106.1kN / m and
r
FG
= 142.7kN / m.
Summing up the edge reactions and the corner forces one obtains 106.1kN + 142.7kN +
+ 4m(58.0 + 94.6)kN/m + 7m81.6kN/m (47.5 + 32.2 + 85.4)kN (78.0 + 58.9 + 57.4)kN = 1071kN =
= 5m7m30.6kN / m
2
, i.e. vertical equilibrium is provided.
Figs. 6.2b and 6.2c summarise the results of the statical analysis, showing the extreme stress
values. The solution is compatible with the mechanism of Fig. 6.2a. Since the positive moments
increase beyond 33.4kN around H, the slab considered (with unchanged resistance distribution)
would probably collapse with a different mechanism.
6.2.6 Reinforcement dimensioning
Starting from the basic reinforcement layout selected with the aid of the kinematical analysis,
the slab resistances are locally increased to cover the stress states derived by the statical
analysis. Detailing considerations lead to a certain smoothing of the theoretical reinforcement
distribution. The reinforcement arrangement, including the reinforcement placing, is given in
Fig. 6.3; note that the bar numbers follow the placing sequence. The following section describes
its development.
Bending reinforcement
The basic reinforcement includes an isotropic mesh 10@200 at the bottom of the slab, a 1m
wide strong band with 816 along the free edge, and a clamping reinforcement 12@200 and
816 along AB, AG and GF, respectively (see Chapter 6.2.4). These reinforcements are
provided by positions 1 and 4 at the slab bottom, 2 at the strong band, and 8, 13 and 6 at the
clamped edges.
At the bottom of the slab the basic reinforcement extends over the whole length; it starts at
the clamped edges and is bent up and back into the slab at the simply supported and the free
edge. The straight end over the clamped edge ensures a good anchorage and the bending up at
the other end enables the reinforcement to function as shear line reinforcement (along the
simply supported edge), concrete confinement (along the free edge) and as splice and support
for the top reinforcement. Close to H position 4 is replaced by 3 (12@200) to resist stresses
beyond 33.4kN (see Fig. 6.2c).
The strong band reinforcement (position 2) follows the detailing of positions 3 and 4 at the
bottom of the slab. The required negative bending resistance is ensured by position 6; spliced
with the wall reinforcement along FG, position 6 extends into the slab to cover the negative
strong band moment, accounting for a development length of 40. Required for shear line
reinforcement along CG, position 7 together with the bends of position 1 improves the plastic
behaviour of the strong band by concrete confinement.
Similar to position 6, positions 8 and 13 provide the necessary negative resistance
determined by the kinematical study along GA and AB, respectively.
Considering the reinforcement positions 1, 2, 3, 4, 6, 7, 8 and 13, the slab resistances are
compared to the stress field developed in Chapter 6.2.5; weak regions are strengthened by
adequate extra reinforcement, paying attention to detailing aspects. Corresponding to the
clamping of the edges AB and AF and the twisting moment introduced by the simply supported
edge BD, Fig. 6.2c reveals negative moments extending far into the slab. Hence, the basic
reinforcement is completed with positions 5 (10@200) and 11 (10@200) and 12
(12@200) in the x- and y-directions, respectively. Spliced with the top part of position 1 along
DF, with the top parts of positions 2, 3 and 4 along BD and with the wall reinforcement along
the clamped edges, the top reinforcement generates a robust 200/ 200mm mesh, suitable for the
casting of the concrete.
Compatibility limit design
78
The basic reinforcement is sufficient for almost all stress states in the slab. Around A and B,
positions 9, 14 and 10 are added to resist the locally high negative moments. This extra
reinforcement fits between the basic one, reducing the mesh spacing to 100 / 100mm.
Shear reinforcement
While the kinematical analysis indicates the global shear flow, shear forces are determined in
detail by the statical analysis. This is summarised in Fig. 6.2b.
Up to m kN v / 152 76 . 0 200
0
= the core of the slab can be assumed to be uncracked and
no transverse reinforcement is required. Fig. 6.2b shows that indeed no transverse reinforcement
is required.
With 1m width, the strong band along the free edge is designed like a slab. Since the core is
uncracked (see Fig. 6.2b), transverse reinforcement is not necessary. As previously suggested, a
confinement of the strong band by means of positions 1 and 7 is recommended for an improved
plastic behaviour.
Making a conservative assumption about the shear line width (b = h / 2 = 120mm), transverse
reinforcement is required for a shear line force in excess of 18.2kN. It can be seen from
Fig. 6.2b that all the shear lines have to be reinforced. The required reinforcement along the
edges AG (V
max
= 78.2kN ) and AB (V
max
= 85.4kN ) could for example be realised with vertical
hairpins 12@200mm spliced with the wall reinforcement. Along BC (V
max
= 47.5kN ), the
resistance provided by the bent up bottom reinforcement (36.2kN ) has to be complemented with
some additional horizontal hairpins, at least in the region around B. Along CG (V
max
= 33.2kN ),
position 6 (10@200mm) is sufficient for resisting the shear forces. Note that according to
Eq. (6.7) transverse reinforcements 10 and 12 spaced at 200mm provide shear resistances of
36.2kN and 52.1kN, respectively.
Remarks
By applying an approximate limit analysis as outlined in Chapter 4.3, any reinforcement deficit
is generally constrained to one sandwich cover. Completing the reinforcement layout intuitively,
the local considerations suggested in Chapter 6.2.2 reduce to a simple check represented
graphically in Fig. 4.4b.
The reinforcement arrangement shown in Fig. 6.3 involves a reinforcement content of about
90kg / m
3
of concrete.
6.3 Discussion
As the positive yield condition is violated in the area around H adjacent to the strong band (see
Fig. 6.2c), the failure mechanism selected is not the most critical one. Presumably, the slab
considered (i.e. with the same reinforcement layout) would collapse as shown in Fig. 5.6c. The
complete solution of the approximate limit analysis problem would consider the basic
reinforcement required by the kinematical analysis, and provide some local strengthening to
eliminate the reinforcement insufficiency. Of course, the limit analysis problem changes by
varying the assumed reinforcement distribution. For instance, an edge beam along DF would
improve the failure pattern of Fig. 5.2a. Generally, the development of a complete solution of
the approximate limit analysis problem is only of academic interest; the possible reduction in
reinforcement is negligible compared to the additional work involved.
It can be stated that any failure shape which identifies a kinematical optimum corresponds
to a good design concept. On the other hand, failure shapes that do not correspond to a
Example application Discussion

79
Fig. 6.3: Reinforcement arrangement.
kinematical optimum generally involve shear singularities and so lead to reinforcement
concentrations (e.g. see Figs. 3.10e and f).
Comparing the stress field of Fig. 6.2 to that according to standard design methods, the
suggested solution fits between the elastic analysis [62] and static solutions independent of
kinematical considerations (e.g. strip method [64] or optimum solutions [62]). Considered as an
alternative to numerical computations, the efficiency of the method depends on the stress field
Top reinforcement
3512 200
12
2012 200
3 m
1 m
1 m
Bottom reinforcement
2 m
1 m
1.6m
0.4m
4.6m 0.6m 1.8m
8
13
3510 200 1
816 125 2
512 200 3
4 1510 200
2310 200 12 12 200 11
2510 200 5
1012 200 10
1210 200 9
14 910 200
3510 200 7
8 16 125 6
@
@
@
@
@
@
@
@ @
@
@
@
@
@
Compatibility limit design
80
library given in Chapter 3. Some difficulties occur in the reinforcement design, since a stress
field with variable principal moment trajectories (i.e. including the effect of the twisting
moments) does not match with a fixed reinforcement layout.
The amount of reinforcement involved in the reinforcement arrangement of Fig. 6.3 is
comparable with that according to elastic computations (81kg/ m
3
) but it is considerably higher
than that according to minimum reinforcement solutions (41kg / m
3
), see [62]. However,
computer analyses generally concentrate on the dimensioning of the main reinforcement and
neglect the reinforcement detailing which is left to the engineers experience. In contrast, the
procedure outlined here integrates detailing and dimensioning aspects, resulting in a practical
and detailed design.
6.4 Conclusions
In the example application considered, the individual plastic analysis components discussed in
the previous chapters are combined, leading from a design concept to a detailed structural
design consistent with detailing requirements.
The development of the design concept requires a simple analysis of global character
capable of establishing the force flow. The procedure involves some preliminary assumptions
(such as the resistance distribution in the slab, the strong band positions and the clamping ratios)
on the basis of an assumed failure mechanism. By means of the kinematical analysis, the
intuitively assumed mechanism is optimised and the required resistances are quantified. The
failure mechanism also indicates the force flow from the slab to the adjacent structural members
which may be designed simultaneously (see Chapter 5.3.2), and the required resistances permit
to select suitable combinations of slab thicknesses and basic reinforcement contents.
The detailed structural design involves the analysis of the force flow within the slab and the
determination of the required local resistances. The statical redundancy of a slab permits the
development of stress fields compatible with the assumed failure mechanism. The validity of
the preliminary kinematical analysis is maintained and the approximate limit analysis
considerations integrate optimum design criteria.
By adjusting the slab design to a selected failure mechanism, the present procedure may be
called a compatibility limit design method. Compared to pure lower-bound approaches such as
the strip method or statically admissible moment fields, the compatibility limit design method
integrates considerations of ultimate state deformations. Similar to an elastic analysis which is
suitable for serviceability verification, the compatibility limit design method could be developed
further to enable a check of the deformations involved in the plastic analysis, extending current
research on tension elements and beams to slabs [61, 24]. As a word of caution, however, the
advantage of the simple computations, essentially established by the stress field library, is
diminished by the difficulty involved in dimensioning and detailing the reinforcement; some
new research is certainly needed in this respect. Still, although being far from a complete
development the compatibility limit design method illustrates nicely the great flexibility and
power of plastic analysis when applied to reinforced concrete.



81
7 Summary and conclusions
7.1 Summary
This thesis deals with the application of the theory of plasticity to reinforced concrete slabs.
After the introduction (Chapter 1) and a presentation of the fundamentals of the theory of
plasticity (Chapter 2) it concentrates on the static method (Chapter 3), limit analysis (Chapter 4)
and the kinematic method (Chapter 5), step by step developing a new design procedure the
compatibility limit design method whose application is illustrated by means of a practical
example (Chapter 6).
The basic idea of the compatibility limit design method is to extend the typical design
procedure for reinforced concrete beams and frames to reinforced concrete slabs. For beams and
frames, the failure mechanisms indicate the global force flow because the plastic hinges identify
the zero shear points. The force flow within the individual beam or frame segments defined by
the zero shear points can then be visualised using truss models or corresponding stress fields
and the segments detailing can be completed accordingly. For slabs, the static and the
kinematic method are normally applied in an unrelated way and thus, the potential offered by
the theory of plasticity is not fully exploited. By considering yield line mechanisms and
developing matching stress fields for the individual slab segments defined by the yield lines the
compatibility limit design method attempts to overcome this unsatisfactory situation.
The presentation of the static method in Chapter 3 concentrates on the load transfer
mechanisms in slabs, differentiating between distributed and concentrated load transfer.
Distributed load transfer within slab segments is described by the generalised strip method,
using general curved rather than straight orthogonal beams. In addition to the normal beam
action with moment increments corresponding to the shear forces, this allows to recognise a
geometrical load transfer mechanism due to the curvature of the beams. Depending on the
selected beam geometry the statical redundancy of the slabs is replaced by geometrical
parameters. While solutions according to the theory of elastic plates refer to beams in the
principal curvature directions of the deflection function Hillerborgs strip method involves
straight beams, preventing geometrical load transfer. On the other hand, assuming constant
principal moments and a pure geometrical load transfer, Hencky-Prandtl solutions are obtained.
Stress fields of complete limit analysis solutions are related to particular curvilinear trajectories;
the case of polar trajectories is of considerable practical interest and is presented in detail.
Finally, it is demonstrated how stress fields can be adapted to different boundary conditions and
how they can be superimposed. Concentrated load transfer occurs via bending moment
concentrations in strong bands or due to twisting moment discontinuities along shear lines;
generally, strong band and shear line action may be combined along certain trajectories.
The discussion of limit analysis in Chapter 4 is based on a thorough evaluation of stress
states and the associated strain rates satisfying the yield condition of orthogonally reinforced
concrete slabs and associated flow rule. In the space of the bending and twisting moments with
respect to the reinforcement directions this yield condition corresponds to a yield surface
consisting of two intersecting elliptical cones, corresponding to positive and negative yield line
failures with yielding bottom and top reinforcement, respectively. Stress points within the yield
surface correspond to rigid states. Stress points on the cone surfaces correspond to uniaxial
Summary and conclusions
82
curvature rates and hence yield lines; since all points of a straight cone surface line are
associated with the same strain rates a kinematic approach is advantageous. Stress points on the
elliptical intersection line of the two cones correspond to combinations of positive and negative
curvature rates in two distinct directions; yield regions rather than yield lines develop under
such circumstances in general and a static approach is advantageous. The stress states
corresponding to the two cone apexes are compatible with arbitrary positive or negative
curvature rates, respectively. Stress points outside the yield surface are not admissible. The
existence of yield regions makes limit analysis solutions quite cumbersome in general. A very
effective simplification is obtained by keeping either the top or the bottom reinforcement rigid.
With such an approximate limit analysis which corresponds to the capacity design method used
in earthquake engineering, yield regions disappear and complete solutions with compatible
stress and strain rate fields can be developed.
The presentation of the kinematic method in Chapter 5 starts from a brief review of past
research. In particular, the development of the work method and the equilibrium method as well
as the nodal force debate of the 1960s are discussed. Then, the basic principles of yield line
analysis are introduced and the application of the work method and of the equilibrium method is
illustrated by means of four examples. It is shown that the equilibrium method can be
interpreted as an application of the compatibility theorem to a suitably defined approximate
limit analysis problem and that the work method and the equilibrium method are equivalent if
they are associated to a unique statical problem.
In Chapter 6, the application of the compatibility limit design method is illustrated by a
practical example. Starting from some preliminary assumptions about the resistance distribution
in the slab an intuitively assumed yield line mechanism is optimised and the required global
resistances are quantified. In a second step, the force flow within and between the individual
slab segments is studied based on the stress field approach developed in Chapter 3 in order to
detect any local resistance deficits. The importance of detailing considerations is emphasised
and comparisons with previously derived solutions are made.
7.2 Conclusions
Similar to beams and frames the shear forces provide the key to understanding the force flow in
slabs. Distributed and concentrated shear force transfer have to be differentiated. Distributed
load transfer can be described by the generalised strip method, using curved orthogonal
coordinates, allowing for the development of general stress fields. Concentrated load transfer
can be realised by strong bands and shear lines.
The investigation of the yield surface using the associated flow rule allows to determine
compatible states of stress and deformation. It can be recognised that a kinematic (static)
analysis is best suited for the weakly (strongly) convex parts of the yield surface corresponding
to yield lines (yield regions). By strengthening the top (bottom) reinforcement failure
mechanisms are restricted to positive (negative) yield lines and the associated approximate limit
analysis problem is simplified such that a complete solution can be developed.
Being based on an approximate limit analysis the compatibility limit design method for
reinforced concrete slabs is similar to the typical design procedure for reinforced concrete
beams and frames. It integrates kinematical, statical and detailing considerations and permits a
practical and economical design with modest computational effort.
Summary Conclusions Recommendations for future studies
83
7.3 Recommendations for future studies
The stress field library should be extended regarding load distribution, principal trajectories,
segment shapes and boundary conditions. Considering the basic segments as macro finite
elements a numerical implementation would greatly facilitate the application of the
compatibility limit design method.
The analysis presented here should be extended to skew reinforcement.
By making use of geometrical load transfer mechanisms stress fields corresponding to high
reinforcement contents in the rigid layer could be improved and the discrepancy between
approximate and classical complete limit analysis solutions could be reduced.
In line with previous work on one- and two-dimensional reinforced concrete elements
[35, 61, 1, 11, 27, 51, 24] recent research investigated the shear transfer mechanisms and the
deformation capacity of reinforced concrete slabs [38, 20]. This research should be combined
with the static and kinematic considerations outlined here.

84


85
References
[1] Alvarez, M., Einfluss des Verbundverhalten auf das Verformungsvermgen von
Stahlbeton, Institut fr Baustatik und Konstruktion, ETH Zrich, IBK Bericht Nr. 236,
Birkhuser Verlag, Basel, 1998, 182 pp.
[2] Braestrup, M.W., Yield lines in discs, plates and shells, Structural Research Laboratory,
Technical University of Denmark, Copenhagen, Report No. R 14, 1970, 54 pp.
[3] Carathodory, C., Schmidt, E.,ber die Hencky-Prandtlschen Kurven, Zeitschrift fr
Angewandte Mathematik und Mechanik, Vol. 3, 1923, pp. 468-475.
[4] Clyde, D.H., Nodal forces as real forces, IABSE Colloquium on Plasticity in
Reinforced Concrete, IABSE Final report, Copenhagen, Vol. 29, 1979, pp. 159-166.
[5] Clyde, D.H., Lower-bound moment field A new approach, personal correspondence
between P. Marti and D.H. Clyde, Apr. 24, 1997, 9 pp.
[6] Drucker, D.C., Greenberg, H.J., Prager, W., The safety factor of an elastic-plastic body
in plane strain, Journal of Applied Mechanics, ASME, Vol. 18, 1951, pp. 371-378.
[7] Drucker, D.C., Greenberg, H.J., Prager, W., Extended limit design theorems for
continuous media, Quarterly Journal of Applied Mathematics, Vol. 9, 1952, pp. 381-
389.
[8] Fox, E.N., The existence of exact solutions in limit analysis for homogeneous isotropic
plates of rigid perfectly-plastic material, Engineering plasticity, (Editors: J. Heyman and
F.A. Leckie) Cambridge University Press, 1968, pp.147-181.
[9] Fox, E.N., Limit analysis for plates: a simple loading problem involving a complex
exact solution, Philosophical Transactions of the Royal Society, London, Vol. 272,
Series A, 1972, pp. 463-492.
[10] Fox, E.N., Limit analysis for plates: the exact solution for a clamped square plate of
isotropic homogeneous material obeying the square yield criterion and loaded by uniform
pressure, Philosophical Transactions of the Royal Society, London, Vol. 277, Series A,
1974, pp. 121-155.
[11] Frst, A., Vorgespannte Betonzugglieder im Brckenbau, Institut fr Baustatik und
Konstruktion, ETH Zrich, IBK Bericht Nr. 267, Birkhuser Verlag, Basel, 2001, 124
pp.
[12] Galileo, G., Discorsi e dimostrazioni matematiche interno a due nuove scienze attente
alla meccanica e ai movimenti locali, Leyden, Elsevirs, 1638.
[13] Gvozdev, A.A., The determination of the value of the collapse load for statically
indeterminate systems undergoing plastic deformations, International Journal of
Mechanical Sciences, Vol. 1, 1960, pp. 322-335.
[14] Hencky, H., ber einige statisch bestimmte Flle des Gleichgewichts in plastischen
Krpern, Zeitschrift fr Angewandte Mathematik und Mechanik, Vol. 3, 1923, pp. 241-
251.
[15] Hencky, H., Zur Theorie plastischer Deformationen und der hierdurch im Material
hervorgerufenen Nebenspannungen, Zeitschrift fr Angewandte Mathematik und
Mechanik, Vol. 4, 1924, pp. 323-334.
[16] Hill, R., The mathematical theory of plasticity, Oxford, Clarendon Press, 1950, 356 pp.

86
[17] Hill, R., On the state of stress in a plastic-rigid body at the yield point, The
Philosophical Magazine, Vol. 42, 1951, pp. 868-875.
[18] Hillerborg, A., Strip method of design, Viewpoint, London, 1975, 256 pp.
[19] Ingerslev, ., The strength of rectangular slabs, Journal of the Institution of Civil
Engineers, Vol. 1, No. 1, Jan. 1923, pp. 3-14.
[20] Jaeger, T., Shear strength and deformation capacity of reinforced concrete slabs,
Sonderpublikation, 4
th
International Ph.D. Symposium in Civil Engineering, Technische
Universitt Mnchen, Universitt der Bundeswehr Mnchen, Springer VDI-Verlag,
Dsseldorf, Vol. 1, Sept. 19-21, 2002, pp. 280-296.
[21] Johansen, K.W., Yield-line theory, Cement and Concrete Association, London, 1962, 181
pp.
[22] Jones, L.L., The use of nodal force in the yield-line analysis, Magazine of Concrete
Research, Special Publication, May 1965, pp. 63-74.
[23] Jones, L.L., Wood, R.H., Yield-line analysis of slabs, Thames and Hudson, London,
1967, 405 pp.
[24] Kaufmann, W., Strength and deformations of structural concrete subjected to in-plane
shear and normal forces, Institut fr Baustatik und Konstruktion, ETH Zrich, IBK
Bericht Nr. 234, Birkhuser Verlag, Basel, 1998, 147 pp.
[25] Kemp, K.O., The evaluation of nodal and edge forces in the yield line theory,
Magazine of Concrete Research, Special Publication, May 1965, pp. 3-12.
[26] Kemp, K.O., The yield criterion for orthotropically reinforced concrete slabs,
International Journal of Mechanical Sciences, Vol.7, No. 11, Nov. 1965, pp. 737-746.
[27] Kenel, A., Biegetragverhalten und Mindestbewehrung von Stahlbetonbauteilen, Institut
fr Baustatik und Konstruktion, ETH Zrich, IBK Bericht Nr. 277, Birkhuser Verlag,
Basel, 2002, 114 pp.
[28] Kirchhoff, G.R., ber das Gleichgewicht und die Bewegung einer elastischen Scheibe,
A. L. Crelles Journal fr die reine und angewandte Mathematik, Berlin, Vol. 40, No.1,
1850, pp. 51-88.
[29] Koiter, W.T., Stress-strain relations, uniqueness and variational theorems for elastic-
plastic materials with a singular yield surface, Quarterly Journal of Applied
Mathematics, Vol. 11, 1953, pp. 350-354.
[30] Mansfield, E.H., An analysis of slabs supported along all edges, Concrete and
Constructional Engineering, Vol. 55, No. 9, 1960, pp. 333-340.
[31] Marcus, H., Die Theorie elastischer Gewebe und ihre Anwendung auf die Berechnung
biegsamer Platten, Julius Springer, Berlin, 2. Auflage, 1932, 368 pp.
[32] Marti, P., Zur plastischen Berechnung von Stahlbeton, Institut fr Baustatik und
Konstruktion, ETH Zrich, IBK Bericht Nr. 104, Birkhuser Verlag, Basel, 1980, 176
pp.
[33] Marti, P., Design of concrete slabs for transverse shear, ACI Structural Journal, Vol.
87, No.2, 1990, pp.180-190.
[34] Marti, P., How to treat shear in structural concrete, ACI Structural Journal, Vol. 96,
No.3, 1999, pp.408-414.
[35] Marti, P., Alvarez, M., Kaufmann, W., Sigrist, V., Tragverhalten von Stahlbeton, Institut
fr Baustatik und Konstruktion, ETH Zrich, IBK Publikation SP-008, Birkhuser
Verlag, Basel, 1999, 301 pp.

87
[36] Marti, P., Kraftfluss in Stahlbetonplatten, Beton- und Stahlbetonbau, Vol. 98, Heft 2,
Feb. 2003, pp. 85-93.
[37] Massonnet, C., Save, M. A., Calcul plastique des constructions, Centre Belgo-
Luxembourgeois dInformation de lAcier, Brussels, Belgium, 1961, 473 pp.
[38] Meyboom, J., Marti, P., Experimental investigation of shear diaphragms in reinforced
concrete slabs, Institut fr Baustatik und Konstruktion, ETH Zrich, IBK Bericht Nr.
243, Birkhuser Verlag, Basel, 2001, 165 pp.
[39] Mohr, O., ber die Darstellung des Spannungszustandesund des Deformations-
zustandes eines Krperelementes und ber die Anwendung derselben in der
Festigkeitslehre, Der Civilingenieur, Band XXVIII, 1882, pp. 113-156.
[40] Mllmann, H., On the nodal forces of the yield line theory, Bygningsstatiske
Meddelelser, Vol. 36, 1965, 24 pp.
[41] Morley, C.T., Equilibrium methods for least upper bound of rigid-plastic plates,
Magazine of Concrete Research, Special Publication, May 1965, pp.13-24.
[42] Morley, C.T., Equilibrium design solutions for torsionless grillages or Hillerborg slabs
under concentrated loads, Proceeding of the Institution of Civil Engineers, Vol. 81, Part.
2, Sept. 1986, pp. 447-460.
[43] Morley, C.T., Local couple transfer to a torsionless grillage, International Journal of
Mechanical Science, Vol. 37, No. 10, 1995, pp. 1067-1078.
[44] Muttoni, A., Die Anwendbarkeit der Plastizittstheorie in der Bemessung von Stahlbeton,
Institut fr Baustatik und Konstruktion, ETH Zrich, IBK Bericht Nr. 176, Birkhuser
Verlag, Basel, 1990, 158 pp.
[45] Muttoni, A., Schwartz, J., Thrlimann, B., Bemessung von Betontragwerken mit
Spannungsfeldern, Birkhuser Verlag, Basel, 1996, 145 pp.
[46] Nielsen, M.P., Exact solutions in the plastic plate theory, Bygningsstat. Medd.,Vol. 34,
No. 1, 1963, pp. 1-28.
[47] Nielsen, M.P., Limit analysis of reinforced concrete slabs, Copenhagen, Acta
Polytechnica Scandinavica, Civil Engineering and Building Construction Series, No. 26,
1964, 167 pp.
[48] Nielsen, M.P., A new nodal-force theory, Magazine of Concrete Research, Special
Publication, May 1965, pp. 25-30.
[49] Nielsen, M.P., Bach, J., A class of lower-bound solutions for rectangular slabs,
Bygnigngsstat. Medd., Vol. 50, No. 3, 1979, pp. 43-58.
[50] Nielsen, P.M., Limit analysis and concrete plasticity, Prentice-Hall Series in Civil
Engineering, Englewood Cliffs, New Jersey, 1984, 420 pp.
[51] Pfyl, T., Tragverhalten von Stahlfaserbeton, Institut fr Baustatik und Konstruktion,
ETH Zrich, IBK Bericht Nr. 277, Birkhuser Verlag, Basel, 2003, 139 pp.
[52] Prager, W., General theory of limit design, Proceedings of the Eighth International
Congress on Applied Mechanics, Istanbul 1952, Istanbul, 1956, Vol. 2, pp. 65-72.
[53] Prager, W., An introduction to plasticity, Addison-Wesley Publishing Company, Inc.,
Reading, Massachusetts U.S.A., 1959, 148 pp.
[54] Ramm, E., Schunck, E., Heinz Isler, Schalen, 3. Auflage, Hrsg. Gesellschaft fr
Ingenieurbaukunst, 2002, 111 pp.
[55] Ritter, W., Die Bauweise Hennebique, Schweizerische Bauzeitung, Vol. 17, Feb. 1899,
pp. 41-43, 49-52 and 59-61.
[56] Rozvany, G.I.N., Optimal design of flexural systems, Pergamon, Oxford, 1976, 246 pp.

88
[57] Saether, K., Flat plates with irregular column layouts Analysis, Journal of Structural
Engineering, ASCE, Vol. 120, No. ST5, May 1994, pp. 1563-1579.
[58] Save, M., A consistent limit-analysis theory for reinforced concrete slabs, Magazine of
Concrete Research, Vol.19, No. 58, Mar. 1967, pp.3-12.
[59] Sawczuk, A.T., Jaeger, T., Grenztragfhigkeits-Theorie der Platten, Springer-Verlag,
Berlin/Gttingen/Heidelberg, 1963, 522 pp.
[60] Sayir, M., Ziegler, H., Der Vertrglichkeitssatz der Plastizittstheorie und seine
Anwendung auf rumlich unstetige Felder, Zeitschrift fr Angewandte Mathematik und
Mechanik, Vol. 20, 1969, pp. 79-93.
[61] Sigrist, V., Zum Verformungsvermgen von Stahlbetontrgern, Institut fr Baustatik und
Konstruktion, ETH Zrich, IBK Bericht Nr. 210, Birkhuser Verlag, Basel, 1995, 159
pp.
[62] Steffen, P., Elastoplastische Dimensionierung von Stahlbetonplatten mittels Finiter
Bemessungselemente und Linearer Optimierung, ETH Zrich, IBK Bericht Nr. 220,
1996, 153pp.
[63] Thomson, W., Tait, P.G., Treatise on natural philosophy, Vol. I, Part II, No. 645-648,
New Edition, Cambridge University Press, 1883, pp. 188-193.
[64] Thrlimann, B., Marti, P., Pralong, J., Ritz, P., Zimmerli, B., Anwendung der
Plastizittstheorie auf Stahlbeton, bungsbeispiele zum Fortbildungskurs fr
Bauingenieure, Institut fr Baustatik und Konstruktion, ETH Zrich, 1984, 112pp.
[65] Von Mises, R., Mechanik der plastischen Formnderung von Kristallen, Zeitschrift fr
angewandte Mathematik und Mechanik, Vol. 8, 1928, pp. 161-185.
[66] Wolfensberger, R., Traglast und optimale Bemessung von Platten, Wildegg, Technische
Forschungs- und Beratungstelle der Zementindustrie, 1964, 119 pp.
[67] Wood, R.H., Plastic and elastic design of slabs and plates, Thames and Hudson, London,
1961, 344 pp.
[68] Wood, R.H., New techniques in nodal forces theory of slabs, Magazine of Concrete
Research, Special Publication, May 1965, pp.31-62.
[69] Wood, R.H., A partial failure of limit analysis for slabs, and the consequence for future
research, Magazine of Concrete Research, Vol.21, No. 67, Jun. 1969, pp.79-90.

89
Notation
General
x, y, z global coordinates
n, t, z local coordinates
r, polar coordinates
u, v curvilinear coordinates
A
u
, A
v
metric of u and v
dS arc element length
radius of curvature,
reinforcement ratio
C

integration constant
k positive factor, parameter
i, j number in a series
A, B, point, state
P point
, ,... angle
plane
safety factor
l span, length
b width
d effective depth
V volume
W work
D dissipation
Y yield function
F load
F load vector
F
s
lower-bound load
F
k
upper-bound load
F
u
ultimate load
(generalised) stress
(generalised) stress vector


(generalised) admissible stress vector
f
y
yield strength
f
u
tensile strength
f
c
compressive strength

cr
cracking shear stress
u deformation, elongation
u deformation vector
(generalised) strain

Beam
q distributed load
V shear force
M bending moment
I moment of inertia
E modulus of elasticity
M
u
ultimate moment
A
s
cross-sectional area of reinforcement
Slab
x geometrical variable
a, b,... length, coordinate
l
a
, l
b
, h
a
geometrical parameter
h thickness
d
v
internal moment arm
t
S
shear line width
,
ij
clockwise angle (from i to j)
clockwise angle (from x- to n-axis)
Q point load
q line load
q distributed load

n
,
tn
,
zn
stress components
n
n
, n
tn
membrane forces
v
n
shear force
v
0
principal shear force

0
principal shear direction
m
n
, m
tn
bending and twisting moment
m
1
, m
2
principal moments

1,

2
principal moment directions
m constant moment value
V concentrated shear force
M strong band moment
S shear line trajectory
r
n
reaction force
R point reaction
w deflection function
Poissons ratio
D flexural rigidity
resistance ratio
90
u
n
m ,
u
tn
m ultimate moments (orthotropic reinforcement)
m
u
ultimate moment (isotropic reinforcement)
M moment vector
yield line direction, compressive stress field inclination

n
,
nt
curvature components

1
,
2
principal curvatures
rotation angle
rotation vector
a
s
cross-sectional area of reinforcement per unit slab width
A
s
strong band reinforcement
Subscripts
x, y, z global axes
n, t local axes
r, polar axes
u, v curvilinear axes
s shear
m moment
S shear line
A, B,... point
cr cracking
l, r left, right
trap trapezoidal element
triang triangular element
tot total
Superscripts
A, B,... region identification
1, 2 segment identification
Symbols
variation
. rate (superscript)
top layer, negative, direction differentiation (superscript)
fixed parameter (superscript)
~ parameter differentiation (superscript)
+, positive, negative (superscript)
, derivation (subscript)
bar diameter
91
Curriculum vitae
Mario Nicola Monotti
born September 24, 1975
Cavigliano, Switzerland
Education
1990 - 1994 Liceo cantonale Locarno
Maturit tipo C
1994 - 1999 Studies at the Department of Civil Engineering
Swiss Federal Institute of Technology (ETH), Zrich
Diploma thesis: Verformungsverhalten von Biegezugzonen in
Sahlbetontrgern (Deformational behaviour of flexural tension zones in
reinforced concrete girders) supervised by Prof. Dr. P. Marti
1999 Dipl. Bauing. ETH
1999 - 2004 Assistant and research associate of Prof. Dr. P. Marti
Institute of Structural Engineering
Swiss Federal Institute of Technology (ETH), Zrich
Teaching assistant
1996 2001 Hydraulik (Hydraulics); Prof. Dr. W. Kinzelbach
1996 - 1998 Mechanik (Applied Mechanics); Prof. Dr. M. Sayir, Prof. Dr. J. Dual
1998 Hochbau (Building Structures); Prof. Dr. H. Bachmann
1998 - 2001 Stahlbeton (Structural Concrete); Prof. Dr. P. Marti
1999 Baustatik (Structural Analysis); Prof. Dr. P. Marti
1999 - 2001 Flchentragwerke (Plate and Shell Structures); Prof. Dr. P. Marti
2001 - 2003 Entwurf (Conceptual Design); Prof. Dr. P. Marti
Experimental research
1999 2000 Experiments on reinforced concrete frame corners constructed with integrated
formwork elements; Prof. Dr. P. Marti
Publications
2002 Monotti, M., Equilibrium Solutions for Reinforced Concrete Slabs,
Sonderpublikation, 4
th
International Ph.D. Symposium in Civil Engineering,
Technische Universitt Mnchen, Universitt der Bundeswehr Mnchen,
September 19-21, 2002, Springer VDI-Verlag, Dsseldorf, Vol. 2, pp. 56-63

You might also like