You are on page 1of 160

Research Collection

Report

Influence of axial tension on the structural response of


reinforced concrete slab strips

Author(s):
Galmarini, Andreas

Publication Date:
2014-10

Permanent Link:
https://doi.org/10.3929/ethz-a-010262954

Rights / License:
In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more
information please consult the Terms of use.

ETH Library
i bk
Institut für Baustatik und Konstruktion, ETH Zürich

Influence of Axial Tension on the Structural Response


of Reinforced Concrete Slab Strips

Andreas Galmarini

IBK Bericht Nr. 358, Oktober 2014


KEYWORDS:
Reinforced concrete, tension, flexure, shear, slab strip, analytical model, shear failure, system failure,
funicular polygon

Dieses Werk ist urheberrechtlich geschützt. Die dadurch begründeten Rechte, insbesondere die
der Übersetzung, des Nachdrucks, des Vortrags, der Entnahme von Abbildungen und Tabellen,
der Funksendung, der Mikroverfilmung oder der Vervielfältigung auf anderen Wegen und der
Speicherung in Datenverarbeitungsanlagen, bleiben, auch bei nur auszugsweiser Verwertung,
vorbehalten. Eine Vervielfältigung dieses Werkes oder von Teilen dieses Werkes ist auch im Einzelfall nur
in den Grenzen der gesetzlichen Bestimmungen des Urheberrechtsgesetzes in der jeweils geltenden
Fassung zulässig. Sie ist grundsätzlich vergütungspflichtig. Zuwiderhandlungen unterliegen den
Strafbestimmungen des Urheberrechts.

Andreas Galmarini:
Influence of Axial Tension on the Structural Response of Reinforced Concrete Slab Strips

Bericht IBK Nr. 358, Oktober 2014

© 2014 Institut für Baustatik und Konstruktion der ETH Zürich, Zürich

Gedruckt auf säurefreiem Papier


Printed in Switzerland

Sie finden das Verzeichnis der IBK-Publikationen auf unserer Homepage unter:
The catalogue of IBK publications is available on our homepage at:
www.ibk.ethz.ch/publications

Die meisten Berichte von Nr. 270 bis Nr. 333 sind auch noch in gedruckter Form unter Angabe der ISBN-Nr. erhältlich bei:
Most reports from No. 270 to No. 333 can still be purchased in printed form by indicating the ISBN number from:

AVA Verlagsauslieferung AG
Centralweg 16
CH-8910 Affoltern am Albis

Tel. ++41 44 762 42 00


Fax ++41 44 762 42 10
e-mail: avainfo@ava.ch

Berichte ab Nr. 334 sind nur noch in elektronischer Form verfügbar. Sie finden die entsprechenden Dateien in der
e-collection der ETH Bibliothek unter http://e-collection.library.ethz.ch oder über die Links auf unserer Homepage.
Reports from No. 334 onwards are only available in electronic form. The respective files can be found in the e-collection
of the ETH Library at http://e-collection.library.ethz.ch or through the links on our homepage.
Influence of Axial Tension on the
Structural Response of
Reinforced Concrete Slab Strips

Andreas Galmarini

Institute of Structural Engineering


Swiss Federal Institute of Technology

Zürich
October 2014
Preface

After gaining practical experience at the Major Bridges Department of an international engineering
company for several years I came back to the Swiss Federal Institute of Technology to learn more
about large-scale testing and to extend my skills with the accumulated knowledge and the philosophy
at the Chair of Theory of Structures and Structural Concrete.
Unexpectedly long time passed before, finally, the possibility to carry out my own series of
large-scale tests materialised. Without doubt, the test planning, practical work in the laboratory, the
observation of the specimen behaviour during the tests and the analysis and interpretation of the data
collected formed the highlight of my time as a research associate. The first test series opened our
eyes to how profoundly axial tension influences the structural response and to how we are tricked by
our experience and routine with structures without axial tension. These insights lead to both a predic-
tion competition for four of the tests and to a second test series. The prediction competition con-
firmed that the addition of an axial tensile force to even a simple structural system causes many
engineers and scientists to have difficulties in its assessment. As a consequence, it was decided that I
abandon my original research topic “Design of Box Caisson Foundations for Vessel Impact” in
favour of the present work that will hopefully aid engineers and scientists confronted with slender
reinforced concrete structures subjected to axial tension and transverse load to quickly gain an under-
standing of the particularities of such systems. This thesis gives both an overview of the influence of
axial tension on the structural response and provides tools for structural analysis.
Working in a research group such as the one headed by Professor Peter Marti that I was lucky
to be part of, leads to many memorable experiences, valuable friendships and a fruitful technical
exchange. This exchange undeniably has a significant impact on the research work of the group
members. And although it is not possible to pin-point all the influences and contributions, I would
like to specifically thank Stephan Etter and Sebastian Villiger (test set-up, rigid-perfectly plastic
model) and Simon Zweidler (elastic shear stress distribution, many theoretical issues) for sharing
their ideas, Hans Seelhofer for introducing me to many aspects of the group’s philosophy, Clare
Burns for reviewing the manuscript and Daniel Locher for the fantastic collaboration during the two
test series and continuous support during the writing of this thesis. Patrick Stähli (Concretum), Ursina
Jenny (KIBAG), and Oliver Kübler kindly supported me with their knowledge on materials and
statistics and Michael Faber with his experience in the academic world.
I am grateful to Professor Peter Marti for giving me the opportunity to carry out large-scale
tests and write this thesis as well as for his openness in sharing his visions and his immense technical
experience and understanding of structural engineering. I also wish to thank my co-examiners Profes-
sor Dan Kuchma and Professor Linh Cao Hoang for their effort and flexibility in reviewing this work.
Not everything was easy and straight forward during my time in academia, but at all times my
wife Nathalie was a strong and loving support and our kids Nina, Livio and Kaj a source of great joy.
To them I am deeply grateful.

Zürich, July 2014 Andreas Galmarini


.
Abstract

This thesis aims at contributing to a better understanding of the structural response of reinforced
concrete slab strips subjected to axial tension and transverse load, focusing on the influence of axial
tension on the structural response. Simple, consistent models – based on mechanical principles and
the assumption that plane sections remain plane – are presented, allowing a realistic description of
the structural response of reinforced concrete slab strips under axial tension and transverse load.
The first part of the thesis covers material properties and the characteristics of the stress-strain
behaviour of concrete and reinforcing steel as well as fundamental aspects of the bond behaviour.
The characteristics of the section behaviour are discussed in terms of moment-axial force strength
envelopes and moment-curvature relations.
The second part covers the theoretical behaviour of reinforced concrete slab strips. Taking a
propped cantilever under uniform transverse load as example, three models of increasing complexity
are presented: rigid-perfectly plastic model, differential equation model and numerical integration.
Selected aspects of modelling are examined and the parameters influencing the flexural behaviour are
investigated with a special focus on the influence of an axial tensile force. Under the assumption that
flexural cracks prevent axial tensile stresses but not the transfer of shear stresses, the shear stress
distribution prior to shear cracking is investigated and a cracking shear force is defined considering
the elastic shear flow for the transformed section. The influences of different models and of axial
tension on the uniform transverse load causing shear cracking in a propped cantilever are examined.
A model subdividing the system into intact sectors and a shear-failed sector is proposed for the
estimation of the structural response after the occurrence of a shear failure. The transition from the
behaviour of the intact system to the behaviour following a shear failure and the influence of the
length of the shear-failed sector are discussed.
The third part of this thesis covers the validation of the models presented. The test set-up and
procedure, the test specimens and the main results of an experimental campaign comprising twelve
large-scale tests on reinforced concrete slab strips carried out at the Institute of Structural Engineer-
ing (IBK) of the Swiss Federal Institute of Technology (ETH) are summarised. The models intro-
duced in the second part of the thesis are adapted to the test set-up and selected results are presented.
Finally, the model results are compared with the experimental data in terms of material, section and
system behaviour and the accuracy and limitations of the models are discussed.
The fourth part summarises the work of the first three parts and gives an overview over the
insights gained, namely that: (i) differential equation models provide a realistic description of the
structural response of reinforced concrete slab strips under axial tension and transverse load; (ii) axial
tension increases the ultimate resistance of the system; (iii) the system stiffness increases with
increasing axial tension; (iv) that shear failure does not cause system failure in systems under axial
tension (provided, that the reinforcement is detailed such that the tension can be sustained in the
shear-failed zone); and (v) the accuracy of a shear failure prediction is considerably smaller than the
accuracy of the prediction of the flexural behaviour.
Kurzfassung

Die vorliegende Arbeit soll zu einem besseren Verständnis des Tragverhaltens von Stahlbeton-
Plattenstreifen unter Längszug und Querbelastung beitragen. Basierend auf den Grundlagen der
Mechanik und der Annahme eben bleibender Querschnitte werden einfache, konsistente Modelle
präsentiert, welche eine realistische Beschreibung des Tragverhaltens von Stahlbeton-Plattenstreifen
unter Längszug und Querbelastung ermöglichen.
Im ersten Teil werden die Materialeigenschaften von Beton und Bewehrungsstahl, deren
Spannungs-Dehnungsbeziehungen sowie grundlegende Aspekte des Verbundverhaltens behandelt.
Zudem werden Momenten-Normalkraft-Interaktionsdiagramme und Momenten-Krümmungs-
Beziehungen als Charakteristika des Querschnittsverhaltens eines Stahlbeton-Plattenstreifens
analysiert.
Der zweite Teil behandelt das theoretische Verhalten von Stahlbeton-Plattenstreifen. Am Bei-
spiel eines am Ende gestützten Kragarms werden drei Modelle von zunehmender Komplexität vorge-
stellt: ein starr-ideal plastisches Modell, ein Differentialgleichungsmodell und die numerische
Integration. Ausgewählte Aspekte der Modellbildung werden diskutiert, und die Parameter, welche
das Biegeverhalten beeinflussen, werden erörtert. Dabei wird ein besonderes Augenmerk auf den
Einfluss einer Längszugkraft gelegt. Ausgehend von der Annahme, dass Biegerisse axiale Zugspan-
nungen verhindern, jedoch nicht die Übertragung von Schubspannungen, wird die Schubspannungs-
verteilung vor der Bildung von Schubrissen basierend auf dem elastischen Schubfluss am ideellen
Querschnitt untersucht und eine Rissschubkraft definiert. Die Einflüsse von Modellbildung und
Längszug auf die Querbelastung, unter welcher am gestützten Kragarm Schubrisse auftreten, werden
diskutiert. Für die Abschätzung des Tragverhaltens nach einem Schubbruch wird ein Modell präsen-
tiert, welches auf der Unterteilung des ursprünglichen Systems in intakte Bereiche und einen Schub-
bruchbereich beruht. Der Übergang vom ursprünglichen Tragverhalten in das Nachbruchtrag-
verhalten und der Einfluss der Länge des Schubbruchbereichs werden beleuchtet.
Die Validierung der vorgestellten Modelle ist Gegenstand des dritten Teils. Die Versuchs-
anlage, der Versuchsablauf, die Versuchskörper sowie ausgewählte Resultate einer zwölf Versuche
umfassenden experimentellen Untersuchung von Stahlbeton-Plattenstreifen unter Längszug und
Querbelastung am Institut für Baustatik und Konstruktion (IBK) der ETH werden zusammenfassend
vorgestellt, wie auch die Anpassung der Modelle an die Versuchsbedingungen und die Resultate
dieser Modelle. Die Modellresultate werden mit den Versuchsresultaten hinsichtlich der Material-
eigenschaften, des Querschnittsverhaltens und des Systemtragverhaltens verglichen, und daraus
werden Schlüsse über die Modellgenauigkeit und die Begrenzungen der Modelle abgeleitet.
Der vierte Teil beinhaltet eine Zusammenfassung der ersten drei Teile sowie eine Übersicht
der gewonnenen Erkenntnisse, namentlich: (i) Differentialgleichungsmodelle vermögen das Biege-
tragverhalten von Stahlbetonplattenstreifen unter Längszug und Biegung realistisch zu beschreiben;
(ii) Längszug vergrössert den Systemwiderstand; (iii) die Systemsteifigkeit vergrössert sich mit
zunehmendem Längszug; (iv) ein Schubbruch verursacht kein Systemversagen in Systemen unter
Längszug (vorausgesetzt, die konstruktive Durchbildung der Bewehrung ermöglicht, dass die Längs-
zugkraft durch den Schubbruchbereich geleitet werden kann); (v) die Genauigkeit einer Schubbruch-
vorhersage ist wesentlich kleiner als die Genauigkeit der Vorhersage des Biegetragverhaltens.
Table of Contents

1 Introduction 1
1.1 Context 1
1.2 Objective and Scope 1
1.3 Overview 2
1.4 Limitations 3

2 Material Behaviour and Interaction 5


2.1 General Considerations 5
2.2 Concrete 7
2.2.1 Standard Compressive Strength Tests 7
2.2.2 Behaviour of Concrete under Uniaxial Compression 8
2.2.3 Behaviour of Concrete under Uniaxial Tension 11
2.2.4 Behaviour of Concrete under Multiaxial Loads 12
2.2.5 Empirical Relations 14
2.2.6 Models 16
2.3 Reinforcing Steel 20
2.3.1 Standard Tension Tests 20
2.3.2 Characteristics 21
2.3.3 Empirical Relations 22
2.3.4 Models 23
2.4 Interface between Concrete and Reinforcement / Bond 25
2.4.1 Standard Tests 25
2.4.2 Characteristics 26
2.4.3 Empirical Relations 27
2.4.4 Models 27

3 Section Behaviour 31
3.1 General 31
3.2 Moment-Axial Force Strength Envelope 32
3.3 Moment-Curvature Relation 35
3.3.1 Phase Model for Moment-Curvature Relations 36
3.4 Particular Aspects 41

i
3.4.1 Contribution of Reinforcing Steel 41
3.4.2 Opposite Sign of Moment and Curvature 42
3.4.3 Influence of Loading History 43
3.4.4 Examples of Moment-Curvature Relations 44
3.4.5 Non-Monotonic Loading History 46
3.5 Rigid-Perfectly Plastic Behaviour 48
3.6 Summary 49

4 Flexural Behaviour of RC Slab Strips: Theoretical Modelling 51


4.1 General 51
4.2 Differential Equations 52
4.2.1 Without Axial Force 52
4.2.2 With Axial Force 57
4.3 Rigid-Perfectly Plastic Model 67
4.3.1 Without Axial Force 67
4.3.2 With Axial Tension 68
4.4 Funicular Polygon 71
4.5 Numerical Integration 72
4.5.1 Without Axial Force 73
4.5.2 With Axial Tension 73
4.6 Comparison 78
4.6.1 Without Axial Force 78
4.6.2 With Axial Tension 79
4.7 Influences on the Flexural Behaviour 81
4.7.1 General 81
4.7.2 Axial Tension 86
4.8 Summary 88

5 Shear Behaviour of RC Slab Strips: Theoretical Modelling 91


5.1 General 91
5.2 Shear Stress Distribution Prior to Shear Cracking 93
5.2.1 Transformed Section Model 93
5.2.2 Neighbouring Sections Model 96
5.2.3 Section Behaviour 97
5.2.4 Variation of Section Behaviour 99
5.2.5 Non-Monotonic Loading History 100
5.2.6 Structural System 101

ii
5.2.7 Influence of Axial Tension on the Shear Cracking Load 105
5.3 Consequences of a Shear Failure 106
5.3.1 Behaviour of Shear-Failed Sector 106
5.3.2 Behaviour of Intact Sector 107
5.4 Summary 110

6 Behaviour of RC Slab Strips: Experimental Investigations 111


6.1 Test Set-Up and Test Procedure 111
6.2 Test Specimens 113
6.3 Test Results 115
6.4 Predictions 118
6.4.1 Differential Equation Model 120
6.4.2 Rigid-Perfectly Plastic Model 122
6.5 Comparison 123
6.5.1 Material 123
6.5.2 Section Behaviour 124
6.5.3 System Behaviour 125
6.6 Interpretation 130

7 Summary and Conclusions 131


7.1 Summary 131
7.2 Conclusions 132
7.2.1 General Conclusions 132
7.2.2 Influence of Axial Tension on the Structural Response of RC Slab Strips 134
7.3 Recommendations for Future Research 134

Appendix A – Coefficients for Differential Equation Solutions 135

Appendix B – Cracking Shear Force Relations 136

Literature 137

Notation 143

iii
Context

1 Introduction

1.1 Context
Reinforced concrete slab strips are among the most commonly used members in reinforced concrete
structures. As part of a larger structural system slab strips can be subjected to axial tension (mem-
brane action) in addition to loads perpendicular to their centre plane, such as self-weight, superim-
posed dead and live loads. Examples are stress-ribbons, parts of box girders and parts of folded plate
structures. The combination of the global and local force flow can result in large axial tensile forces
occurring concurrently with significant moments and shear forces.
Due to the scarce experimental data found in literature concerning the combination of axial
tension, flexure and shear, an experimental campaign consisting of two series of large-scale tests was
carried out at the laboratory of the Institute of Structural Engineering (IBK) of the Swiss Federal
Institute of Technology (ETH) between 2011 and 2013 (Galmarini et al. 2013; Locher et al. 2014).
In the wake of the first series, a prediction competition was initiated (Galmarini et al. 2012).
The submitted predictions and the number of teams that retreated because they could not find a
satisfying solution showed the need for a compilation of the theoretical background as well as for
simple models for identifying the main parameters and checking the plausibility of the results from
complex finite element models.
Modelling a structure for structural analysis implies simplifying a physical reality to such an
extent that it can be analysed analytically or numerically. Due to the uncertainties of the input param-
eters, the structural behaviour can never be calculated exactly, regardless of the model sophistication.
It can only be estimated. A good model should describe the relevant characteristics of the structural
behaviour well enough to enable solving the problem at hand or studying a particular aspect of the
behaviour, while being as simple as possible. Bearing this in mind, the present thesis provides a
framework of models with different degrees of complexity for the theoretical investigation of the
structural response of reinforced concrete slab strips subjected to axial tension and transverse load.

1.2 Objective and Scope


This thesis aims at contributing to a better understanding of the structural response of reinforced
concrete slab strips subjected to axial tension and transverse load. In particular, the influence of axial
tension on the structural response shall be characterised.
Simple, consistent models – based on mechanical principles and the assumption that plane sec-
tions remain plane – will be presented for the flexural and shear behaviour of reinforced concrete slab
strips. These models with different levels of complexity will serve as basis for identifying the main
parameters influencing the behaviour and for characterising the influence of axial tension. The mod-
els will be compared and validated with results from the experimental campaign mentioned above in
order to clarify their accuracy and their limitations.

1
Introduction

1.3 Overview
The first part of this thesis covers material properties and the basic aspects of section behaviour. In
Chapter 2 the characteristics of the stress-strain behaviour of concrete and reinforcing steel are exam-
ined and models for describing this behaviour are introduced. Fundamental aspects of the bond
behaviour are reviewed. In Chapter 3 the moment-axial force strength envelope and the moment-
curvature relations, which characterise the section behaviour are discussed. Thereby the assumptions
of plane sections remaining plane and the sectional fibres being independent from one another are
adopted.
The second part covers the theoretical behaviour of reinforced concrete slab strips. Taking a
propped cantilever under uniform transverse load as an example, three models of increasing com-
plexity are presented in Chapter 4: a rigid-perfectly plastic model, a differential equation model and a
numerical integration model. Selected aspects of modelling are examined and the parameters influ-
encing the flexural behaviour are investigated with a special focus on the influence of the axial
tensile force. Following a brief discussion on the different contributions to the shear transfer in a
cracked reinforced concrete member, the interconnection between flexure and shear is highlighted in
Chapter 5. Under the assumption that flexural cracks prevent axial tensile stresses but do not prevent
the transfer of shear stresses, the shear stress distribution prior to shear cracking is investigated and a
cracking shear force is defined based on the elastic shear flow in the transformed section. The influ-
ence of the model and the axial tension on the uniform transverse load causing shear cracking in a
propped cantilever is examined. A model that subdivides the system into intact sectors and a shear-
failed sector is proposed for the estimation of the structural response after the occurrence of a shear
failure. The transition from the behaviour of the intact system to the behaviour following a shear
failure and the influence of the length of the shear-failed sector are discussed.
The third part of this thesis covers the validation of the models presented in the second part.
The model results are compared with measured results from the experimental campaign. In Chapter 6,
the test set-up, test procedure, test specimens and the main results from a series of twelve large-scale
tests on reinforced concrete slab strips carried out at the Swiss Federal Institute of Technology are
summarised. Then, the models introduced in the second part of the thesis are adapted to the static
system and the loading conditions of the test set-up and selected prediction results are presented.
Finally, the model results are compared with the experimental data in terms of material, section and
system behaviour and the accuracy and the limitations of the models are discussed.
The fourth part, Chapter 7, summarises the work of the first three parts, providing an overview
of the insights gained, and concludes with a set of recommendations for future research.

2
Limitations

1.4 Limitations
Throughout this thesis, the structures are assumed to be initially stress-free. Time-dependent effects
such as creep, shrinkage and relaxation are disregarded, except for a brief examination of the influ-
ence of taking into account the static vs. the dynamic properties of the reinforcing steel. Short-term
static loading is assumed, excluding dynamic or cyclic loads.
All models presented in Chapters 3 to 5 are based on the assumption of plane sections remain-
ing plane. Consequently, the local force flow at supports and load application points is not treated.
The slab strips are assumed to have symmetric cross-sections, a constant geometry and uni-
form material properties over the entire length. No membrane actions other than axial tension are
considered, except for the discussion on the section behaviour with axial compression.
As a slab strip is equivalent to a beam with a rectangular cross-section, and in order to ease the
comparison with the test results presented in Chapter 6, the stress resultants are given as for a beam
(not per unit slab width).
It is investigated how a shear failure influences the structural response of a system with rein-
forced concrete slab strips under axial tension. However, the shear resistance of the reinforced con-
crete slab strips itself is not subject of the present thesis.

3
Introduction

4
General Considerations

2 Material Behaviour and Interaction

Many advantages of the use of reinforced concrete for structural members originate in the situational
combination of two materials with diverging characteristics: concrete and reinforcing steel. Concrete
is a moderately priced material that is poured into a formwork where it hardens and reaches small
tensile and moderate compressive strengths. Reinforcing steel is an expensive material with elevated
tensile and compressive strengths that is used in form of bars placed in the formwork before casting
the concrete. The behaviour of reinforced concrete depends on the behaviour of the individual mate-
rials and of their aggregation. In the following, the main structural properties of both the concrete and
the reinforcing steel relevant for structural members subjected to plane stress states are outlined and
selected numerical descriptions of their behaviour are introduced, see Fig. 2.1. Finally, aspects of the
interaction between concrete and reinforcement are discussed.

Fig. 2.1 – Uniaxial behaviour of concrete and reinforcing steel: (a) typical stress-strain behaviour; .
(b) measured behaviour and numerical idealisations.

2.1 General Considerations


Materials can be characterised in many ways. For their structural use, they are often characterised by
their stress-strain behaviour, weight and durability.
Typically, the stress-strain behaviour is gained from standardised laboratory tests. The condi-
tions in the laboratory differ from those in real structures and so the use of properties measured in the
laboratory to model the behaviour in real structures is a first simplification. Furthermore, the proper-
ties measured are unique for the sample tested and to some degree influenced by the test set-up and
procedure. As it is impossible to test every single element that is used in a construction, it is neces-
sary to derive generic properties from the test results. This is a next simplification. Based on the
generic properties, idealised analytical or numerical descriptions of the behaviour are found and used

5
Material Behaviour and Interaction

in structural analysis to approximate and evaluate the behaviour of structures and/or their members.
Nowadays, it is possible to approximate the real behaviour of a structure in great detail. However, the
more detailed the desired results, the more complex the model and the more parameters have to be
implemented and set to appropriate values. Advanced models are time consuming and demanding in
operator skills and computing power. In practice, often, the focus is on the overall understanding of
the structure and the identification of the key aspects rather than on the subtleties of the detailed
behaviour. Furthermore, the resources available for the structural design and analysis are limited. In
these circumstances it is important to have models at hand that are as simple as possible while still
able to provide the necessary insights on the behaviour of the structure investigated.

Fig. 2.2 – Stress-strain behaviour: (a) elastic; (b) elastic-plastic; (c) strain-softening; key properties:
(d) bilinear; (e) rigid-perfectly plastic, hysteretic; (f) advanced models.

Fig. 2.2 illustrates some basic aspects of stress-strain behaviour and of possible idealisations.
Diagram (a) shows elastic behaviour characterised by the unique relationship between stress and
strain, by the complete reversibility of deformations and by the absence of energy dissipation. Elastic
responses can be further differentiated as linear (Curve I) or non-linear (Curve II). The shaded area
under Curve II represents the strain energy per unit volume dU, corresponding to the energy stored in
an elastic body. It is defined as the integral of the stresses over the corresponding strains. In elastic-
plastic behaviour as illustrated in Diagram (b) the deformations are not entirely reversible and the
unloading curve (Curve III) differs from the loading curve. The area enclosed by the loading curve
and the unloading curve from a given point corresponds to the dissipated energy per unit volume dD.
The strength fu of a material and the corresponding strain ε(fu) are key values of a stress-strain curve.
Often, stress-strain curves are practically linear close to the origin and the stress-strain relation can be
described by the modulus of elasticity E, also called Young’s modulus. These properties are indicated
in Diagram (c). The behaviour where the stress reduces with increasing strain is called strain-
softening (IV). Strain softening sections of measured stress-strain curves usually not only reflect the
material behaviour but also the response of the entire test set-up including the test specimen. Bilinear
relations (Curve V) as shown in Diagram (d) are commonly used idealisations. Special cases of
bilinear relations, such as elastic-perfectly plastic (Curve VI) and rigid-perfectly plastic (Curve VII)
relations, are shown in Diagram (e). When plastic materials are unloaded and reloaded, the reloading
path is generally not identical to the unloading path. In a bilinear model, the paths are different, if

6
Concrete

unloading reaches the opposite limit. Depending on the material, the behaviour and thereby the limits
can be different (Curve VIII) or identical (Curve IX) in tension and compression. In a bilinear model,
the loading, unloading and reloading paths are parallel. The loop described by the unloading and
reloading curves is called hysteresis. The area enclosed represents the hysteretic dissipation per unit
volume dH. In hysteretic behaviour the stress-strain state depends not only on the current loading
state but also on the loading history. The modelling of hysteretic behaviour therefore requires a
memory function. Discontinuities (Curve X) and abrupt changes from inclination to declination
(Curve XI) in behaviour models may lead to numerical challenges in their application, see Diagram
(f).
Linear elastic models can be sufficient for investigating the structural behaviour under small to
moderate loads. Rigid-perfectly plastic models often are sufficient when only the ultimate resistance
of a structure is of interest. With bilinear models most aspects can be approximated closely, save the
post-peak behaviour. The idealisations described here for the stress-strain behaviour can also be
usefully applied for sectional behaviour or even for the behaviour of entire structural components.

2.2 Concrete
Concrete is made from aggregate, Portland cement, water and often additives. The constituents are
mixed and filled into formworks in a viscous state. Through hydration, the cement hardens and binds
the aggregate. Even though concrete has been known and used for centuries, the process of hardening
and the interaction of aggregate and cement under load are not understood down to the last details.
From an engineering perspective, however, the behaviour is generally described sufficiently well
assuming concrete as a homogeneous material with average properties. Much of the following brief
description is based on the publications by Wight and MacGregor (2011), Etter (2012), Kauf-
mann (1998) and Seelhofer (2009).

2.2.1 Standard Compressive Strength Tests

Usually, the properties of plain concrete are derived from standardised compressive strength tests.
Common test specimens are cubes with a side length of 15 cm and cylinders with a diameter of
d = 15 cm and a height of l = 30 cm. A typical test set-up and a test cylinder before testing and after
failure is shown in Fig. 2.3(a) to (c).The recordings include at least the compressive strength fc but
often also the modulus of elasticity Ec at low stress levels and the crushing strain εcu. Entire stress-
strain curves are mostly recorded for research purposes only.
While the concrete compressive strength can be determined with sufficient accuracy with
basic testing and measuring equipment, the accurate determination of the concrete stiffness is more
delicate, as the flexibility of the entire test set-up including the test specimen has to be considered,
see Fig. 2.3(d) and (e). The height of the test specimen should be at least 2 to 2½ times the diameter
(or shortest side length for prismatic specimens) in order to avoid an artificial increase in strength and
stiffness by the restraining effect of the base plates of the testing equipment on the interface surfaces
of the test specimen. This effect is one of the main reasons why the strength determined by cube
tests fc,cube is higher than the cylinder strength fcc.
When the stress-strain relation after reaching the strength is of interest, the loading must be
deformation controlled and the stiffness of the test set-up and the slenderness of the test specimen
must be considered, see Fig. 2.3(f). If the elastic energy released from these parts is larger than the
energy dissipated in the fracture zone, brittle failure will be observed and no information on the
strain-softening can be gained. The introduction of a control parameter based on a combination of
force and deformation can alleviate this difficulty (Okubo & Nishimatsu 1985).

7
Material Behaviour and Interaction

(a) (b) (c)

Fig. 2.3 – Concrete compressive strength tests: (a) test set-up; (b) equipped test specimen prior to testing;
(c) test specimen after failure (cylinder G2-2 from Galmarini et al. (2013)); (d) static system and
measuring lengths; (e) stress-strain recordings in function of measuring length (Etter 2012);
(f) stress-strain recordings in function of the test specimen slenderness (data from (Rokugo & Ko-
yanagi 1992)).

For the determination of concrete properties in existing structures, cores are drilled from the
structure. Care must be taken, not to damage the matrix during drilling, transport and preparation of
such test specimens. The cores are typically smaller than the standard cylinders for newly made
concrete. Their diameter and height should, however, not be smaller than three times the largest
aggregate diameter to be representative for the concrete in question.

2.2.2 Behaviour of Concrete under Uniaxial Compression

The behaviour of concrete is nonlinear and ductile to some extent, despite its constituents being of
linear elastic and of brittle nature. Hsu (1963) identified microcracking as the source of this behav-
iour. The cracks can be grouped into bond cracks which occur along the interface between aggregate
and mortar and mortar cracks which cross the latter. In higher strength concrete, in lightweight con-
crete or where unsuitable aggregate was used, cracks can also be observed in the aggregates. In
uniaxially loaded concrete, the crack types described above develop in a specific order as illustrated
in Fig. 2.4(a):
Even before loading, shrinkage of the mortar during hydration restrained by the aggregates
and temperature differences between mortar and aggregate cause bond cracks. Such bond cracks have
little influence on the load bearing behaviour and the stress-strain curve remains essentially linear up
to 30 to 40 % of the compressive strength.
Due to the difference in stiffness between mortar and aggregate, a further increase in com-
pression will lead to bond cracks caused by stress concentrations. The accompanying load redistribu-
tion is the reason for the gradual levelling of the stress-strain curve.

8
Concrete

The discontinuity limit (Newman & Newman 1971) marks the onset of mortar cracks caused
by tensile stresses in the mortar transverse to the direction of loading. This limit usually lies between
50 to 60 % of the compressive strength. Cracking is still stable at this stage. With increasing load
these cracks connect and form a continuous pattern causing the slope of the stress-strain curve to
drop.
Usually at around 75 to 90 % of the compressive strength (in the example illustrated at 94 %)
the crack pattern has densified to such an extent that there no longer is enough intact mortar and
crack propagation turns unstable. This critical stress level marks the transition between stable and
unstable crack propagation and coincides with the minimum volumetric strain εc,vol = 2εc1,2 – εc3. The
critical stress is also called creep rupture strength, see also Fig. 2.5(e). Concrete under sustained
loading above the critical stress level fails. Concrete under cyclic uniaxial compressive stresses has a
shake-down limit approximately equal to the critical stress.

Fig. 2.4 – Behaviour of concrete under uniaxial compression: (a) and (b) stress-strain curves, cracking stages
and transverse strain relations (data from (van Mier 1984)); (c) and (d) fracture zone, strain soften-
ing and fracture energy per unit volume.

For low and medium stresses, the Poisson’s ratio ν = -εc1,2 / εc3 is constant at 0.15 to 0.20.
Mortar cracking causes ν to increase so that concrete under stresses close to its compressive strength
regains its unloaded volume.
The strain softening observed in tests for strains higher than the crushing strain is dependent
on the test set-up and the geometry of the test specimen in addition to the concrete tested. Conse-
quently, the recorded curve represents a system property and not a material property. Sigrist (1995)
introduced the specific fracture energy (per unit volume) UcF, as corresponding material property.
The latter is derived from the recordings assuming that the specimen is divided into a fractured zone
of a certain height lF sandwiched between the intact end zones. This is a crude simplification, as in

9
Material Behaviour and Interaction

reality there is no clear transition from one zone to the other, see Fig. 2.3(c) and Fig. 2.4(c). There-
fore, the definition of the height of the fracture zone remains somewhat arbitrary. Based on experi-
mental observations Sigrist proposed the height to be lF = 2d for cylinders or 2.26A1/2 for prismatic
test specimens (A being the cross-sectional area). More recent investigations suggest that the height
of the fracture zone is variable within the test specimen (smaller in the centre, larger towards the
outside) and that the average height of the fracture zone could be as little as 0.6 to 1.3A1/2
(Etter 2012).
The properties of concrete vary within a wide range and the range is continuously widened by
the ongoing research and development. In general, concrete with a higher strength has a higher
modulus of elasticity and a sharper stress drop-off after the compressive strength is reached. Main
factors influencing the properties are:
The type and quantity of cement has a strong influence on the strength, the hydration heat and on
how fast the strength develops.
An increased water/cement ratio results in a higher porosity of the hardened concrete which lowers
the strength and can increase the freeze-thaw resistance.
A portion of the cement can be replaced by other materials or chemicals in liquid form can be added
to the constituents before mixing. These additives are used to influence the properties of the wet or
hardened concrete.

Fig. 2.5 – Behaviour of concrete under uniaxial compression: (a) influence of aggregate (Carrasquilio et
al. 1981); (b) influence of strain-rate on compressive strength (data from 64 authors compiled by
Ammann (1983), Bischoff and Perry (1991) and Pajak (2011)); (c) strain-rate effect (Grote et
al. 2001); (d) behaviour of concrete at different ages (Etter et al. 2012); (e) and (f) strength and
creep of concrete under sustained compression loads (Rüsch 1960).

Well-graded aggregate reduces porosity and increases strength. Crushed aggregate tends to
result in a slight increase in stiffness and strength, see Fig. 2.5(a). The strength of the aggregate
should be such that failure occurs in the mortar rather than in the aggregate. Otherwise the failure of
the concrete will be more brittle. Lightweight aggregates are used to save weight. They typically
result in lower strength concrete.

10
Concrete

The curing conditions (moisture, temperature, mechanical strain) have a strong influence on
the development of the concrete strength. Inappropriate conditions can lead to damage in the micro-
structure of the concrete and thus severely impair the concrete strength.
Time influences the properties and the behaviour of concrete in multiple aspects. The con-
crete strength is commonly measured at 28 days. The hydration process, however, continues and with
it the strength increases, see Fig. 2.5(d). Hardening and drying of the concrete are coupled with
shrinkage. Under sustained loading, concrete creeps and under sustained strain it relaxes, see
Fig. 2.5(e) and (f). The effects described have a strong influence in a first period whereafter they
asymptotically tend towards a final value.
When a load is applied rapidly and only over a short time (e.g. earthquake, impact, blast) the
concrete strength can increase considerably. This effect is called strain-rate effect. Strain-rate test
results, especially concerning the modulus of elasticity and the crushing strain, are influenced by the
test set-up and the test method. The results compiled in Fig. 2.5(b) and (c) suggest the compressive
strength to increase proportionally to the logarithm of the strain-rate increase. Most likely, the crush-
ing strain also increases, as does the secant modulus while the initial elastic modulus remains
unaffected (Bischoff & Perry 1991; CEB 1988). The relative strength increase due to strain-rate is
lower for higher strength concrete than for normal strength concrete.

2.2.3 Behaviour of Concrete under Uniaxial Tension

The tensile strength fct of concrete amounts to only a fraction of its compressive strength, commonly
8 to 15 %, and its scatter is larger than found for of the compressive strength.

Fig. 2.6 – Behaviour of concrete under uniaxial tension: (a) direct tension test; (b) double punch test;
(c) splitting cylinder test; (d) bending or modulus of rupture test; (e) comparison of tensile strength
test results, mean value and standard deviation (data from (Alvarez 1998; Stoffel & Marti 1997));
(f) load-deflection recording and test specimen with fictious crack; (g) stress-strain behaviour
outside fracture zone; (h) stress-crack opening relation.

11
Material Behaviour and Interaction

Direct tension tests (Fig. 2.6(a)) are demanding and therefore primarily carried out for material
research. More often, the tensile strength is experimentally determined by indirect testing, see
Fig. 2.6(b) to (d). These tests are less demanding to execute but require assumptions for the stress
state within the specimen at failure in order to calculate the tensile strength from the measured load.
The method as well as the size and geometry of the test specimen influence the recorded tensile
strength, see Fig. 2.6(e).
The stress-strain curve in tension (Fig. 2.6(f)) is practically linear in a first part with the same
modulus of elasticity as in compression. With increasing stress, the stiffness drops due to microcrack-
ing. At failure, a single through-crack forms. The quasi-brittle failure with pronounced strain soften-
ing for strains higher than the tensile strain εct can only be observed in very stiff testing machines
with sensitive measuring equipment and precise control devices. The quasi-brittle behaviour is likely
caused by the mortar bridging the crack. Similar to the softening behaviour in compression the slope
is dependent on the stiffness of the test-setup and the length of the test specimen. In order to derive a
material property, it is assumed that after the peak load is reached, the material in the zones away
from the crack unloads elastically (Fig. 2.6(g)), while the deformations localise in the fracture pro-
cess zone, comprising the crack and its immediate vicinity. Assuming that the initial length of this
fracture process zone is zero, the fracture behaviour can be described by a stress-crack opening
relationship, see Fig. 2.6(h). In this case, the area under the stress-crack opening curve, the specific
fracture energy (per unit area) GF, is the material property. Alternatively, a finite length can be
assumed as an additional material property.
The tensile strength is influenced by the same factors as the compressive strength. The tensile
strength, and with it the bond and shear strengths, develop faster than the compressive strength. The
strain-rate effect is more pronounced for the tensile strength than for the compressive strength and
can increase the tensile strength by several times its value.

2.2.4 Behaviour of Concrete under Multiaxial Loads

Experimental investigation of the behaviour of concrete under arbitrary triaxial loading states
(σ1 ≠ σ2 ≠ σ3) is very demanding. Niwa et al. (1967) and van Mier (1986) contain examples of such
studies. A key challenge is the reduction of the transverse restraint at the loaded concrete surfaces.
The use of rod-platens for the load transfer has been found to mitigate the issue. For the same reason,
rod-platens in combination with cube or plate type test specimens are widely used to investigate the
behaviour of concrete under biaxial loading (σ1 = 0, σ2 ≠ σ3), see Fig. 2.7 (a) to (c). Kupfer (1973)
and Nelissen (1972) contain examples of extensive campaigns on biaxial loading states. The most
common triaxial loading states investigated are those with identical values for two axes (σ1 = σ2 ≠ σ3),
see Fig. 2.7(d). Thereto, cylindrical test specimens are loaded in hydraulic triax cells. Typically, the
hydraulic pressure in the cell is increased to a certain level in a first step (hydrostatic loading state
σ1 = σ2 = σ3), and then maintained, while the load on the cylinder bases is varied, see Fig. 2.7(e).
Usually, the test specimens are coated to avoid pore pressure in the concrete.
The stress-strain curves for plane stress states (biaxial loading) shown in Fig. 2.7(b) illustrate
that the behaviour is similar to that under uniaxial compression. Transverse compression increases
the strength by up to around 30 % for a stress ratio of σ2 / σ3 ≈ 1/2. For higher stress ratios, the
strength increase drops to around 15 % for σ2 / σ3 = 1. The elastic modulus increases slightly with
increasing lateral compression, while the crushing strain does not differ significantly from the value
found under uniaxial compression. Lateral tension reduces strength and crushing strain significantly,
while its effect on the elastic modulus is negligible. Kaufmann (1998) noted that after tensile failure
in biaxial tension-compression has occurred, the compressive stresses parallel to the cracks can be
further increased until the specimen fails in compression at values somewhat lower than the uniaxial
compressive strength. Nelissen (1972) noted that the strength increase percentage for a given stress
ratio is not significantly influenced by the loading path or by the concrete strength but that the maxi-
mum strain was influenced by the loading path. His results also show the influence of the loading
duration on strength and strain to be similar for biaxial loading and uniaxial loading.

12
Concrete

Fig. 2.7 – Concrete under multiaxial loading: (a) biaxial tests on plate specimens; (b) stress-strain curves for
biaxial loading (data from (Hussein & Marzouk 2000)); (c) concrete strength for biaxial loading and
typical failure modes (data from (Hussein & Marzouk 2000; Kupfer 1973; Nelissen 1972; Niwa et
al. 1967)); (d) triax tests on cylinder specimens; (e) stress-strain curves for triaxial loading (data
from (Xie et al. 1995)); (f) concrete strength for triax loading (data compiled by (Seelhofer 2009;
Jiang et al. 1991)).

The stress-strain curves for triaxial stress states shown in Fig. 2.7(e) illustrate that the behav-
iour up to the maximum compression is similar to the behaviour under uniaxial compression. Already
a small amount of transverse compression increases the strength significantly compared to uniaxial
tests. With high transverse compression, the strength is multiplied, see Fig. 2.7(f). The crushing strain
equally increases with lateral compression and the strain softening at large strains decreases, making
the concrete become more ductile. The initial modulus of elasticity remains virtually unaffected by
lateral compression. After failure, concrete under lateral compression has a significant residual
strength.

13
Material Behaviour and Interaction

2.2.5 Empirical Relations

Often, only the uniaxial compressive strength is determined experimentally. The other properties are
then derived by empirical relations. A multitude of such relations exist, a number of which are re-
viewed in Popovics (1970). A set of relations is given in the following and compared in Fig. 2.8 with
the relations given in Model Code 2010 (fib 2010) and with experimental data collected over the
years at the Chair of Theory of Structures and Structural Concrete of the Institute of Structural Engi-
neering (IBK) at the Swiss Federal Institute of Technology (Alvarez 1998; Ernst & Marti 1998;
Eskola 1996; Etter et al. 2009; Etter et al. 2012; Fehlmann et al. 2011; Fürst & Marti 1999; Galmarini
et al. 2013; Jäger & Marti 2006; Kaufmann & Marti 1996; Kenel & Marti 1996; Locher et al. 2014;
Meyboom & Marti 2001; Pfyl & Marti 2001; Seefeld-Ebert et al. in press; Sigrist & Marti 1993;
Stoffel & Marti 1997; Trümpi-Althaus 2006).

Fig. 2.8 – Concrete properties in function of the cylinder compressive strength: (a) cube compressive strength;
(b) splitting tensile strength; (c) specified strength vs. effective strength with 5 % fractile range
(data from (Nowak & Szerszen 2003) and 11600 cube tests from 2008 - 2012 provided by the Swiss
industry); (d) elastic modulus; (e) crushing strain; (f) variation of concrete properties over time
(fib 2010). Experimental data (mean values) from concrete tests at the IBK.

The cube strength of concrete relates linearly to the cylinder strength:

f c,cube = 1.15 f cc [MPa] (2.1)

Fig. 2.8(a) shows the above relation to be adequate for the vast majority of results. The splitting
tensile strength can be derived as

f cts = 0.3 f cc2 3 [MPa] (2.2)

14
Concrete

Fig. 2.8(b) shows the above relation to be adequate for the vast majority of results. The lower values
for high strength concrete stipulated by the Model Code are not supported by the IBK experimental
data. The elastic modulus can be derived as

Ec = k E f cc
13
[MPa] (2.3)

where kE depends on the type of aggregate and varies from 7∙103 to 12∙103. Fig. 2.8 (d) shows
kE = 10∙103 to yield acceptable results albeit some significant outliers. The correlation of compressive
strength and elastic modulus is not as strong as for the tensile strength. The crushing strain can be
estimated as

εcu ≈ 0.55 f cc
13
[‰] and [MPa] (2.4)

Fig. 2.8(e) shows the large scatter of the crushing strain. Part of the scatter of the test values in
Fig. 2.8(b), (c) and (e) can be explained by the different influence of the age on the properties, see
Fig. 2.8(f), and the wide range of age of the test specimens. The Model Code relations for splitting
tensile strength and crushing strain are shown for both characteristic and mean compressive strengths.
The test values are mean values.
The Poisson’s ration can be taken as vc = 0.2. Sigrist (1995) states typical values of the specific
fracture energy per unit volume UcF to range between 0.06 and 0.12 MPa and recommends the use of
0.08 MPa for calculations (associated to a certain length of the fracture zone, see Section 2.2.2). The
Model Code proposes to estimate the specific fracture energy (per unit area) GF as

GF = 73 f cc0.18 [N/m] and [MPa] (2.5)

As Fig. 2.8 (c) shows, the characteristic strength of concrete delivered to a building site fck,eff
generally corresponds to the characteristic concrete strength specified fck,spec. The mean strength,
however, can deviate significantly from the assumption

= f ck + 8 [MPa]
f cm (2.6)

stipulated by the Model Code. The standard deviation of the concrete strength seems to be related to
the production facilities and traditions and varies independent of the compressive strength. Data from
the Swiss industry shows the mean compressive strength to be higher than specified for the most
common concrete types (5 % for C25/30, 10 % for C30/37, and 3 % for C35/45), while the US values
are slightly lower than specified. The data from Switzerland shows a decrease in the standard devia-
tion for high strength concrete while the opposite is true for the US data. In Switzerland, due to
building activity, cement production capacity and ambient temperature, the concrete strength is
slightly lower in summer than in winter. Due to differences in casting, vibrating and curing, the
properties of the concrete in a structural member may differ from the properties measured on a test
specimen from the same batch. Furthermore, the concrete properties vary within the structural mem-
ber. In a slab for example, the concrete strength in the upper layers of the slab is 10 to 20 % less than
the concrete strength in the lower layers (Suprenant 1995).

15
Material Behaviour and Interaction

2.2.6 Models

A large number of formulas for the numerical description of the stress-strain relationship exist.
Reviews can be found in Popovics (1970) and Chang and Mander (1994). Here, only a selection is
presented. Time and duration effects (aging, creep, relaxation, shrinkage) will not be considered.
Their inclusion leads to cumbersome iterative calculations even for the sectional behaviour. For more
details see Etter (2012).

Fig. 2.9 – Models of the stress-strain relations of concrete in compression with input parameters and compari-
son with experimental data from (Galmarini et al. 2013): (a) linear elastic; (b) rigid-perfectly
plastic; (c) compression chord model; (d) Collins and Mitchell; (e) Hognestad; (f) Model Code.

Linear Elastic
The simplest model, often adequate for investigating serviceability issues, includes only a single
parameter, the elastic modulus Ec. The stress-strain relation is described as

σc = Ec εc for εc < f ct Ec
(2.7)
σc =0 otherwise

The tensile strength is generally neglected, fct = 0.

16
Concrete

Rigid-perfectly plastic
Another simple model, often adequate for inverstigating ultimate capacity issues, also includes only a
single parameter, the nominal compressive strength fc. The stress-strain relation is described as

σc =− f c for ε c < 0
(2.8)
0 ≥ σc ≥ − f c otherwise

The tensile strength is neglected. The effective compressive strength fc ranges between 0.45fcc
and 1.0fcc, the reduction accounting for the influence of transverse strain and the model simplification.

Compression Chord Model


In his study of structural compression members combining old and young concrete Etter (2012)
modified Sigrist’s (1995) Compression Chord Model (CCM) and extended it to account for multi-
axial loads with the following set of equations. Up to the crushing strain, the stress-strain relation
follows Sargin’s (1971) proposal:

kσ ζ − ζ 2
σc = − f cc for − εcu ≤ εc ≤ 0
1 + (kσ − 2)ζ
−εc
where ζ = (2.9)
εcu
Ec εcu
and kσ =
f cc

For strains beyond the crushing strain, a linear softening is assumed:

σc = − f cc + (εc + εcu ) EcF for − εcR ≤ εc ≤ −εcu


Ec
where EcF =
1
1−
χcF
(2.10)
f2
and χcF = cc
2 EcU cF
f cc
and εcR =εcu −
EcF

The influence of multiaxial loading on the strength is modelled with the modified Coulomb failure
criterion

τ + σ tan ϕ − c = 0
(2.11)
with σ ≤ f ct

17
Material Behaviour and Interaction

For concrete, reasonable results are found setting the angle of internal friction ϕ and the cohesion c to

3
tan ϕ =
4
(2.12)
1 − sin ϕ f cc
=c = f cc
2cos ϕ 4

The strength is independent of the intermediate main stress and can be derived based on the maxi-
mum principal stress σc1 as

1 + sin ϕ
f cc3 =f cc − k3σ c1 ≤ 4 f cc where k3 = =4 (2.13)
1 − sin ϕ

The basic relation leads to an overestimation of the strength for high transverse compression. There-
fore, the triaxial compressive strength is limited to four times the uniaxial compressive strength. The
increase of the crushing strain due to lateral compression is estimated as

 σ 
εcu 3 =
εcu 1 − k3 k3ε c1  where k3ε =
3 5 (2.14)
 f cc 

For dimensioning, k3ε = 3 yields conservative results. Here, k3ε = 5 is used. The assumption that the
increasing macrocrack formation after the compressive strength is reached reduces the cohesion to
zero without changing the angle of internal friction leads to a conservative estimate for the residual
strength of concrete under transverse compression:

f cR 3 =−k3σc1 (2.15)

The softening stability coefficient and the residual strain are estimated as

f cc2 εcu χcF


χ= ⋅ =
cF 3
2 EcU cF εcu 3 σc1 (2.16)
1 − k3 k3ε
f cc

f cc  1 
εcR 3 =
εcu 3 +  − 1 (2.17)
Ec  χcF 3 

Following Kaufmann (1998), the influence of transverse tensile strain on the compressive strength of
reinforced concrete can be estimated as

f cc2/3
=fc3 ≤ f cc (2.18)
0.4 + 30ε1

18
Concrete

The tensile strength is generally neglected. If not, the tensile stress-strain relation is assumed inde-
pendent of the multiaxial stress state and modelled using the Model Code proposal εct = 0.15 ‰ and
Sargin’s equations:

kσt ζ t − ζ t2
=σc f ct for 0 < εc ≤ εct
1 + (kσt − 2)ζ t
ε
where ζ t = c (2.19)
εct
Ec εct
and kσt =
f ct

Hognestad
Based on an analysis of the results from tests on prismatic specimens loaded by a compressive force
with a constant eccentricity, Hognestad (1951) proposed a stress-strain relation for concrete in com-
pression zones that is parabolic up to an effective compressive strength followed by a linear softening.

(
σc = − f c 2ζ − ζ 2 ) for − εcu ≤ εc ≤ 0

 ε + εcu 
σc = − f c 1 + 0.15 c  for − εcR ≤ εc < −εcu
 εcR − εcu 
−ε
where ζ = c (2.20)
εcu
2f
and εcu = c
Ec
and εcR = 0.0038
and f c = 0.85 f cc

Collins and Mitchell


Based on a mathematical relation proposed by Popovics (1971), and modified by Thorenfeldt et
al. (1987), Collins and Mitchell (1997) suggested the stress-strain relation to be described as:

nc ζ
σc =− f cc
nc − 1 + ζ nc kc
−εc
where ζ =
εcu
Ec (2.21)
and nc =
f cc
Ec −
εcu
1 for ζ ≤ 1
and kc = 
0.67 + f cc 62 ≥ 1 otherwise

19
Material Behaviour and Interaction

Model Code
The Model Code (fib 2010) recommends the following stress-strain relation:

kσ ζ − ζ 2
σc = − f cc for − εcR ≤ εc ≤ 0
1 + (kσ − 2)ζ
−εc
where ζ = (2.22)
εcu
Ec εcu
and kσ =
f cc

The values of Ec, εcu and εcR are given in tables as a function of fcc. Linear interpolation can be
used, if necessary. For tensile stresses a bilinear stress-strain relation is given:

f ct
σc= Ec εc for 0 ≤ εc ≤ 0.9
Ec
(2.23)
 0.00015 − εc  f ct
=
σc f ct 1 − 0.1  for 0.9 ≤ εc ≤ 0.00015
 0.00015 − 0.9 f ct Ec  Ec

2.3 Reinforcing Steel


Steel is by far the most common material used to reinforce concrete. It is used in form of deformed
bars of circular cross-section with rolled-on lugs or ribs to enhance its bond properties (see
Section 2.4) or, less frequently, in form of wires or seven-wire strands that are pre-tensioned or post-
tensioned. Only the first form is discussed in the following.
In order to ensure weldability of the reinforcing bars they are typically made of steel with a
low carbon content. Even though reinforcing bars are weldable, the use of welding should be consid-
ered carefully as it changes the material properties of the bars concerned. In particular, welding
reduces the ductility of the bars and thereby often also of the entire structural member. Steel used for
reinforcing bars is usually of higher strength than steel used for pure steel structures. The increase in
strength is achieved by alloying (micro alloyed steel) or by a change of the microstructure (hot-rolled
or cold-worked steel). In Switzerland, today, hot-rolled and cold-worked bars (only for small diame-
ters) are commonly used.

2.3.1 Standard Tension Tests

The standard tension test consists of a bar sample clamped at its ends to the testing machine by
means of hydraulic shoes (Fig. 2.10(a)) and fitted with an extensometer (Fig. 2.10(b)). Care must be
taken to have a smooth transition between the clamped lengths and the free length of the sample and
to not induce bending moments in the sample at the clamped ends. There is no standard length for the
bar specimen. The sample is tensioned at a constant rate and eventually torn apart, while the distance
covered by the movable part of the testing machine, the deformation of the extensometer and the
force. Are measured

20
Reinforcing Steel

(a) (b) (c)

Fig. 2.10 – Reinforcing steel tests: (a) testing machine (b) test specimen with an extensometer fitted prior to
testing; (c) specimen with necking zone after rupture.

Since many years, at the IBK, the loading of the bar has been interrupted during the yield plat-
eau and close to the maximum force for two minutes each to record the influence of the loading rate,
see downward spikes in the stress-strain recordings of Fig. 2.11(a) and (b).

2.3.2 Characteristics

There are two types of reinforcing steel: steel with a yield plateau, see Fig. 2.11(a), and steel without,
see Fig. 2.11(b). Both types initially behave linear elastically. The yield strength fsy is defined by an
increase in strain under constant stress. Material scientists distinguish between an upper and a lower
yield strength, where the first value designates the peak value prior to the plateau and the second a
lower value in the plateau. This distinction generally is circumstantial in structural engineering,
where only the lower value is used. The yield plateau, also called Lüders plateau, is caused by the
yielding of the weakest sector of the test specimen. This sector yields until it reaches its hardening
phase. With this hardening, the sector in question ceases to be the weakest and another sector begins
to yield and so on, until the entire bar reaches the hardening phase at εsh. The initial strain hardening
modulus Esh is significantly lower than the elastic modulus Es and the slope of the stress-strain curve
reduces to zero with increasing stress up to the tensile strength fsu. For larger strains, the distinction
has to be made between a damaged zone with a strain softening behaviour and the remainder of the
bar unloading quasi elastically, similar to the behaviour of concrete specimens described previously.
The strain softening recorded depends on the extensometer length and position. It is not a material
property but the result of the reduction of the cross-sectional area in the roughly two diameter long
damaged zone, see Fig. 2.10(c). This reduction is referred to as necking. For this reason, material
scientists distinguish between true stresses and strains corresponding to the force measured divided
by the actual cross-sectional area and engineering stresses and strains corresponding to the force
measured divided by the initial or even the nominal cross-sectional area (as used in the present thesis).
The use of the nominal diameter as a basis is a reason for the deviations of the elastic moduli record-
ed in tests from the material property, which is approximately Es = 205 GPa for all common types of
reinforcing steel. For steels without a yield plateau, the stress at which the plastic portion of the strain
has reached 2 ‰ (fs0.2) is used as an equivalent to the yield strength, see Fig. 2.11(e).
The behaviour of steel in compression is identical to the behaviour in tension when viewed in
true stresses and strains (Dodd & Restrepo-Posada 1995). For high compressive stresses buckling
complicates direct testing. The different cross-sectional area changes for tension (reduction) and
compression (increase) result in higher engineering stresses in compression than in tension for the
same absolute strain. The difference is negligible in the elastic phase but notable in the strain-
hardening phase.
For unloading, the stress-strain relation is linear for stress reductions in the order of 0.85fsy to
1.35fsy (Dodd & Restrepo-Posada 1995). For larger stress reductions, the stiffness is increasingly
reduced, see Fig. 2.11(c). This stiffness reduction is known as the Bauschinger effect.

21
Material Behaviour and Interaction

Fig. 2.11 – Behaviour of reinforcing steel under uniaxial tension: (a) and (b) stress-strain curves for micro-
alloyed, hot-rolled and cold-worked steel (data from (Etter et al. 2012)); (c) stress-strain curve with
cyclic loading history (based on data presented in (Lowes 1999)); (d) schematic stress-strain
diagram for micro-alloyed, heat-treated, low-carbon or hot-rolled steel; (e) schematic stress-strain
diagram for cold-worked or high-carbon steel; (f) strain-rate effect on the yield and ultimate
strength (data from multiple authors compiled by (Malvar 1998)).

The strain-rate increases the yield strength more than the tensile strength, see Fig. 2.11(f), but
does not significantly influence the strain at tensile strength (Manjoine 1944; Malvar 1998). The
influence of the strain-rate is smaller for reinforcing steel than for concrete. Repeated loading impairs
the tensile resistance of reinforcing steel (fatigue). Temperature also influences steel properties. Low
temperatures increase the yield strength, tensile strength and elastic modulus while decreasing the
strain at tensile strength and fatigue resistance. High temperatures decrease the yield strength, tensile
strength and elastic modulus, while increasing the strain at tensile strength. If not protected properly,
ordinary reinforcing steel corrodes. The corrosion rate depends on the environment and the steel.
Corrosion leads to a reduction in cross-sectional area and thereby of a bar’s strength.
The stress state in reinforcing steel in reinforced concrete members is predominantly uniaxial.
The behaviour of steel under multiaxial loads is therefore not relevant in the present context.

2.3.3 Empirical Relations

The ratio of tensile and yield strength and the strain at tensile strength are important measures of the
ductility of the reinforcement. For current Swiss reinforcing bars fsu / fsy ranges between 1.05 and 1.4
and εsu between 20 ‰ and 200 ‰.

22
Reinforcing Steel

The stress-strain relation in compression can be derived from the stress-strain curve in tension
as (Dodd & Restrepo-Posada 1995)

(1 + εtension )
2
σcomp
s = −σtension
s s

εtension (2.24)
εcomp =
− s

1 + εtension
s
s

A distinction between tension and compression is, however, not necessary for most applications in
reinforced concrete because the reinforcement does not reach the hardening phase in compression.
The same authors relate the unloading modulus to the maximum strain (per direction) reached:

Es ,unloading 1
= 0.82 + (2.25)
Es 5.55 + 1000ln(1 + ε s ,max )

2.3.4 Models

In the present study, the difference between tension and compression as well as the reduction of the
unloading modulus and the Bauschinger effect are neglected. A selection of models is visualised and
compared to experimental data in Fig. 2.12. The respective equations are given in the following.

Fig. 2.12 – Models of the stress-strain relations of reinforcing steel with input parameters and comparison with
experimental data from (Galmarini et al. 2013): (a) linear elastic; (b) bilinear; (c) rigid-perfectly
plastic; (d) hot-rolled steel; (e) trilinear (hot-rolled steel with k = 1); (f) cold-worked steel.

23
Material Behaviour and Interaction

Linear Elastic
The range is not limited as for concrete.

σ s = Es ε s (2.26)

Bilinear
In a bilinear model, the elastic phase changes to a linear strain hardening phase at the yield strength:

σ s= Es ε s for ε s ≤ ε sy
εs
=σs
εs
(f sy + Esh  ε s − ε sy  ) for ε sy < ε s ≤ ε su
(2.27)
f su − f sy
where Esh =
ε su − ε sy

Rigid-Perfectly Plastic
For the plastic model, a nominal yield stress fs0, defining the average stress in the reinforcing steel, is
to be defined. The nominal yield stress ranges between the yield strength and the tensile strength
(fsy ≤ fs0 ≤ fsu). The nominal yield stress can be different for tension and compression. For design
purposes the nominal yield stress is usually taken as fs0 = fsy. Here the nominal yield stress is taken as
fs0 = fsu,stat (fsy,dyn < fsu,stat < fsu,dyn).

σ s =− f s 0 for ε s < 0
=σs 0 =for ε s 0 (2.28)
=σs f s0 for ε s > 0

Hot-Rolled Steel
Mander (1983) developed a model providing an accurate description of the strain-hardening phase:

σ s= Es ε s for ε s ≤ ε sy
εs
σs f sy for ε sy < ε s ≤ ε sh
εs

εs   ε − εs  
k

( )
(2.29)
=σs  f su − f su − f sy  su   for ε sh < ε s ≤ ε su
εs   ε su − ε sh  
 
ε − ε sh
where k = Esh su
f su − f sy

With k = 1 the equations describe a trilinear relation.

24
Interface between Concrete and Reinforcement / Bond

Cold-Worked Steel
The following equation can be used to describe the stress-strain relation of cold-worked steel
(Cosenza et al. 1993):

k
σ σ σ 
ε s= s + 0.002 s  s 
Es σ s  f sy 
 ε − f su Es  (2.30)
ln  su 
where k =  
0.002
(
ln f su f sy )

2.4 Interface between Concrete and Reinforcement / Bond


The interaction and transfer of force between embedded reinforcement and concrete is called bond.
Bond influences the spacing and width of transverse cracks as well as the rotation capacity of plastic
hinge regions in reinforced concrete members. Bond enables bars to be anchored and lapped without
further mechanical connection.
Despite the importance of bond, there is no widely accepted standard test or model describing
the phenomenon. Test results exhibit large scatter, even within a single test series. Even empirical
relations are therefore to be treated with caution.

2.4.1 Standard Tests

Contrary to concrete and reinforcing steel, there is no widely used standard for the experimental
investigation of bond behaviour. Often, so-called pull-out tests are used. A typical set-up is illustrated
in Fig. 2.13(a).

Fig. 2.13 – Bond behaviour: (a) pull-out test and bond shear stress distribution; (b) bond shear stress-slip
recordings (data from (Eligehausen et al. 1982)); (c) schematic diagram of the deformed concrete
around a reinforcing bar (based on (Goto 1971)).

In pull-out tests, a tensile force is applied to a reinforcing bar embedded over a short length in
a block of concrete. The displacement of the bar end protruding the concrete block on the side oppo-
site the force application is measured. The displacement is referred to as bond slip δb. Division of the
force F by the circumference of the bar and the embedment length lb results in an average bond shear
stress τb which is plotted against the slip, see Fig. 2.13(b). The concrete block usually has a square or

25
Material Behaviour and Interaction

circular cross section and often extends beyond the embedment sector to reduce the influence of the
restraining effect of the reaction plate on the bond zone. In most cases, the embedment length is
small (3 to 5 times the diameter of the reinforcing bar diameter) in order to limit the stress variation
in the bond zone and to ensure that the reinforcing bar remains in an elastic state throughout the test.

Pull-out tests are criticised for the concrete being in compression, whereas in real structures the
concrete is in tension where bond is especially important (e.g. tension zone in bending members).
Whether the so-called more realistic bond tests, where the investigated bond zone is part of a simpli-
fied structural member (ACI Committee 408 2003) provide better results, remains doubtful in view
of the complex stress state and the assumptions required to interpret the results. For tension zones in
bending members with a constant moment, the tests consisting of a reinforcing bar embedded central-
ly in a concrete prism of square or circular cross-section over a longer length and a tensile force is
applied directly to either side of the bar (Goto 1971) seem more appropriate.

2.4.2 Characteristics

Bond is commonly represented in terms of a shear stress acting on the nominal circumference of a
reinforcing bar.

Fig. 2.14 – Bond behaviour: (a) bond shear strength under cyclic loading (data from (Eligehausen et al. 1983))
(b) reduction of bond shear stress upon yielding of the reinforcing bar (data from (Shima et
al. 1987)); (c) strain rate influence on bond shear strength (data from (Hjorth 1976)).

Based on their experimental and analytical studies, Lutz and Gergely (1967) conclude that
three different mechanisms contribute to the load transfer between bar and concrete: Chemical adhe-
sion, friction and mechanical interaction between concrete and steel (bearing of the ribs against the
concrete). After initial slip, the latter is dominant for deformed reinforcing bars. Tests with epoxy
coated bars indicate, however, that the contribution of friction cannot be neglected.
Goto (1971) noted radial cracks at a slope of approximately 60 degrees to the axis of the bar,
see Fig. 2.13(c), forming at minimal slip. These cracks result in the formation of concrete cones
essentially loaded in compression. They do not seem to cause a notable reduction of the bond. At
higher loads, cracks parallel to the bar axis propagate from the bar-concrete interface outwards.
These cracks are a result of hoop stresses in the concrete cones. If the concrete cover or the spacing
of the bars is too small, these cracks propagate to the next bar or the surface and lead to a splitting
failure with loss of bond. Lutz and Gergely (1967) observed the formation of concrete wedges
crushed to powder in front of the ribs extending up to two times the height of the rib; these zones
moves with the ribs. The authors also noted that this crushing occurs only for ribs with face angles
larger than 40 degrees and that for other bars or after the crushing zone is established significant slip
with friction occurs. Furthermore, they observed that the bond stiffness for unloading and reloading

26
Interface between Concrete and Reinforcement / Bond

was substantially higher than the initial one. However, this seems true only as long as the loading
does not change sign. For cyclic loading with stress reversal bond deteriorates rapidly when a previ-
ous loading cycle goes beyond around 80 % of the peak bond stress (Eligehausen et al. 1982), see
Fig. 2.14(a). Shima et al. (1987) discovered that bond is significantly reduced upon yielding of the
reinforcing bar in tension, see Fig. 2.14(b), which can be attributed to the bar’s cross-section con-
tracting and to the steel evading further loading through plastic deformation. In the case of yielding
in compression, an increase of bond shear is reported (Lowes 1999).
The main factors influencing the bond shear strength τbu are:
The mechanical properties of the concrete. The higher the concrete compressive strength,
the higher the bond shear strength.
The geometry of the reinforcing bar. The height, face angle and spacing of the ribs signifi-
cantly influece the bond shear strength and stiffness. For three bars with the same diameter but dif-
ferent rib geometry Goto found the bond stiffness to vary by a factor of 4.5. The bond shear strength
increases slightly with the bar diameter.
The volume of concrete around the bars, the presence of confinement reinforcement and of
lateral pressure all increase the resistance against lateral dilatation and thereby increase the bond
shear strength, see Fig. 2.13(b).
The orientation of the bar (Rehm 1957a) and the surface condition of the bar as well as the
casting conditions also influence the bond behaviour.
The influence of the strain-rate is notable but not as strong as for the concrete or steel strength.
Based on his experimental investigation, Hjorth (1976) noted that the strain-rate does not influence
the shape of the bond shear stress-slip relation. Only, a translation of the curve towards higher bond
shear stresses occurs for an increase in strain-rate. This increase in bond shear strength is due to the
increase in concrete strength. As the increase vanishes after very short loading durations, the strain-
rate effect can be neglected for most practical issues.

2.4.3 Empirical Relations

The bond shear strength is related to the concrete compressive strength. The exponent in the
relation varies between ¼ and 1, depending on the author.
The spacing of the ribs sR is commonly in the range of 40 to 70 % of the bar diameter.

2.4.4 Models

In view of the current state of knowledge, models will be able to give an order of magnitude
for bond issues but cannot, in general, be considered to give very precise estimates. A multitude of
models are presented in a report by the fib (2000). Alvarez (1998) investigated the behaviour of
reinforced concrete tension members with different models of bond and steel. Two models, a simple
one and a more complex one are illustrated in the following.

27
Material Behaviour and Interaction

Fig. 2.15 – Models of the bond shear stress-slip relations with input parameters: (a) Model Code (good bond
conditions, unconfined concrete, no transverse pressure, away from transverse crack); (b) Sigrist.

Model Code
For good bond conditions and unconfined concrete, bond behaviour is described by:

0.4
δ 
=
τb kd ks k p τbu  b  for 0 ≤ δb ≤ δb1
 δb1 
=
τb kd ks k p τbu for δb1 < δb ≤ δb 2
 δ − δb 2 
=τb kd k s k p  τbu − (τbu − τbf ) b  for δb 2 < δb ≤ δb3
 δb 3 − δb 2 
=τb kd k s k p τbf for δb3 < δb
where
x (2.31)
=
kd ≤ 1 with x being the distance from a transverse crack

( 2 − fct f sy )
2
 ε −ε 
s sy
−5 
 ε −ε 
=
ks 0.15 + 0.85e  sh sy  with ε s ≥ ε sy yielding of steel
σc,transverse
kp =
1.0 − tanh(20 ) with σc,transverse ≤ 0 transv. pressure
f cc
τbu = 2.5 f cc1/2 ; τbf = 0.4τbu ; δb1 = 1 mm; δb 2 = 2 mm; δb3 = sR pull-out failure
τbu =3.13 f cc1/4 ; τbf =0; δb1 =1.75 f cc−5/8 ; δb 2 =δb1; δb3 =1.2δb1 splitting failure

Where there is no yield plateau, εsu is used instead of εsh. Pull-out failure is governing when the
concrete cover is larger than 5.Ø and the bar spacing is larger than 10.Ø, where Ø is the diameter of
the reinforcing bar in question.
Unloading within the rising and horizontal part of the relation is defined as linear with a slope
of 200 MPa/mm.

28
Interface between Concrete and Reinforcement / Bond

Sigrist
In order to find analytical solutions for the second order differential equation describing the inter-
action of reinforcement and concrete (Rehm 1957b), Sigrist (1995) proposed a stepped, rigid-
perfectly plastic bond shear stress-slip relationship:

τb = 2 f ct = 0.6 f cc2/3 for 0 ≤ δb ≤ δb (ε s = ε sy )


(2.32)
τb = f ct = 0.3 f cc2/3 for δb (ε s = ε sy ) < δb

The model is valid only for pull-out type failure.

29
Material Behaviour and Interaction

30
General

3 Section Behaviour

3.1 General
The behaviour of a reinforced concrete beam or slab strip can be investigated based on the assump-
tions that plane sections remain plane and that the behaviour of any fibre of the section is independ-
ent of the neighbouring fibres. Under these assumptions, the stress at any point of an arbitrary cross-
section is defined unambiguously by the strain at that point, the respective material model and, where
required by the latter, the strain history. Integration over the section then yields the section stress
resultants. The procedure described is illustrated in Fig. 3.1 for the case of axial force and bending
about the transverse axis.

Fig. 3.1 – Section behaviour: (a) section and material widths; (b) estimation of the stress resultants correspond-
ing to a given strain plane based on material laws and geometry.

The integration of the stresses is conveniently done separately for each material:

( Fs1 + Fs 2 ) + F=c ∫− h /2 σs (ε0 , χ, z )bs ( z )dz + ∫− h /2 σc (ε0 , χ, z )bc ( z )dz


h /2 h /2
N (ε0 , χ=
)
2 (3.1)
∑ zsi Fsi + zc F=c
h /2 h /2
M (ε0 , χ=
)
i =1
∫ − h /2
σ s (ε0 , χ, z )bs ( z ) zdz + ∫
− h /2
σc (ε0 , χ, z )bc ( z ) zdz

For biaxial bending, the same procedure can be used. As in this case the strain also varies with y, the
integration has to go over both axes.
Each strain plane defines a specific point in the moment-axial force space (M-N). However, a
certain point in the M-N space can correspond to more than one strain plane or to no strain plane at
all. The former is generally the case for points inside the moment-axial force strength envelope, see
Fig. 3.2(a), the latter for points outside. Often, the strain plane is described by the axial strain ε0 and
the curvatures χy and χz (for uniaxial bending with only one curvature, the index is omitted). The
values for axial strain and curvature corresponding to the strength envelope in Fig. 3.2(a) are shown
in Fig. 3.2(b).

31
Section Behaviour

Fig. 3.2 – Section behaviour: (a) moment-axial force strength envelope; (b) axial strain and curvature corre-
sponding to Mu(Nu); (c) moment-curvature relation and corresponding axial strain.

For a constant axial force, there is a unique relation between moment and curvature, as can be
seen in Fig. 3.2(c). Running along the moment-curvature curve in Diagram (c) corresponds to depart-
ing from the axial force axis at N = 0.96 MN in Diagram (a) along a horizontal until the strength
envelope is reached and then returning towards the axial force axis on the same horizontal. Section
failure determines which point on the horizontal can be reached.
The models presented in the following assume initially stress-free sections and disregard time
related effects such as creep, shrinkage and relaxation. The influence of time related effects increases
with the fraction of concrete in a cross-section under compression and with the magnitude of the
compressive stresses. A detailed discussion of these aspects is found in Etter (2012).

3.2 Moment-Axial Force Strength Envelope


For many applications, the strength envelope is the most important characteristic of a section. Note
that the envelope has a three-dimensional lemonlike shape if biaxial bending is considered. In slabs,
bending about the axis perpendicular to the middle plane is usually negligible and will therefore not
be considered further in the present document.
Whether the tensile resistance of concrete should be taken into account or not depends on the
loading history; if for example substantial axial tension is applied to the section prior to the applica-
tion of a bending moment, the section will be entirely cracked and there are no tensile stresses in the
concrete at the cracks, whereas if the moment is applied first and the axial tension second, there can
be regions of continuous tensile stresses in the concrete. Because cracking occurs at discrete loca-
tions there are always regions of concrete with tensile stresses between two cracks. The influence of
these regions on the behaviour can be treated with the Tension Chord Model (Marti et al. 1998;
Alvarez 1998). Thereby the assumption of plane sections remaining plane is softened and only ful-
filled at discrete locations. Generally, the tensile part of the stress-strain relation of concrete has little
influence on strength envelopes and is commonly neglected. It is also neglected in the following
diagrams. Envelope curves can also be drawn for other states such as cracking or yielding.
The integration of the stresses over the section generally needs to be carried out numerically.
Numerical methods are also required to determine envelopes and moment-curvature relationships.
Numerous simplifications have been developed over time to reduce the complexity of the calcula-
tions and to find approximations that allow an analytical treatment. Significant steps of increasing
simplification are illustrated in Fig. 3.3 and Fig. 3.5.

32
Moment-Axial Force Strength Envelope

Fig. 3.3 – Section behaviour – simplified estimation methods: (a) neglecting the spatial distribution of rein-
forcing steel; (b) assuming a stress block for the concrete; (c) assuming yielding of the reinforce-
ment and neglecting the reinforcing steel on compression side.

A first simplification is the concentration of the reinforcing steel area at points corresponding
to the centroids of the reinforcing bars on either face, see Fig. 3.3(a) and Curve (a) in Fig. 3.5. The
width of the concrete is thereby assumed constant and the steel stresses in the steel points are reduced
by the respective concrete stresses. Thereby the geometrical complexity of the calculations is greatly
reduced, generally without significant impact on the moment-axial force strength envelope.
The next simplification is the stress block method. Instead of integrating the stresses allocated
to the strains through the concrete stress-strain relationship an idealised description of the concrete
stress distribution in the entire compression zone is used to compute the contribution of the concrete
to the axial force and moment, see Fig. 3.3(b) and Curve (b) in Fig. 3.5. Such idealisations were
developed with the aim of simplifying the calculations for the behaviour of compression zones close
to failure and usually prescribe the strain at the compression edge of a section εc,lim and a nominal
compressive stress fc (≤ fcc) to be taken into account. Simple geometric shapes, such as rectangles and
parabolas, are used. The stress block method is useful for calculating moment-axial force strength
envelopes, but of limited value for the investigation of the section behaviour under small forces. As
an example, the stress block model by Mattock et al. (1961), see Fig. 3.4, is used here.

Fig. 3.4 – Stress block model for rectangular compression zones according to Mattock et al. (1961).

Eq. (3.2) shows how stress block models simplify the computation of the forces corresponding
to a strain plane. The prescription of the strain at the compression edge of the section makes the

33
Section Behaviour

iterations to find the maximum moment for a given axial force obsolete, which greatly reduces the
calculation effort required for a moment-axial force strength envelope.

N (ε0 , χ) =Fs1 + Fs 2 + Fc =σ s ( ε0 , χ, z =−e ) as + σ s ( ε0 , χ, z =e ) as − f c bαx


(3.2)
M (ε0 ,=χ) eFs1 − eFs 2 + ( h 2 − βx ) Fc

Idealisations with a sudden increase in the concrete stress and regions where negative strains
do not result in concrete stresses (strains between 0 and (α - 1)∙3 ‰ in the example) cause a step in
the M-N diagram (at Nu ≈ -650 kN in Curve (b) in Fig. 3.5). Stress block models are generally not
suited for large compressive forces without or with small eccentricities (x > h in the example). Addi-
tional rules for the stress block are required for these cases. Alternatively, linear interpolation be-
tween the last valid point defined by the stress block model (Point 7 in Fig. 3.5) and the point of pure
compressive strength (Point 8) can be used. For slab strips, this region is usually not of interest.
As a further simplification, the stress in the reinforcing steel can be assumed as elastic or taken
as equal to the yield or tensile strength. Where the behaviour close to the flexural strength for a
section without axial force is of interest, reasonable results can be achieved even neglecting the
reinforcement on the compression side of the section, see Fig. 3.3(c) and Curve (c) in Fig. 3.5. This is
especially true for the frequent case of sections with a low degree of reinforcement on the compres-
sion side.

Fig. 3.5 – Section behaviour – comparison of moment-axial force strength envelopes and corresponding
curvatures according to different methods.

The overall shape of the moment-axial force strength envelope is convex. Exceptions can be
caused by material behaviour models with re-entrant corners (in the example, the yield plateau and
strain-hardening of the steel is the origin of the dent in the moment-axial force strength envelope at
Nu ≈ 3000 kN), or by strain restrictions in the material model or strain-softening of the material (in
the example, strain-softening of the concrete is the origin of the non-convexity at Nu ≈ -7000 kN).

34
Moment-Curvature Relation

Due to the general convexity of the envelope, linear interpolation between two known points on the
envelope results in a – in general conservative – approximation of the envelope between these points.
The quality of the approximation depends on the number of points and the selection of the relevant
strain planes. Typical points with corresponding strain planes and the resulting approximation are
illustrated in Fig. 3.5 (Curve (b’)). For the example, considering only points 1, 5, 6 and 8 would
already result in a good approximation.
Further subtleties and particular aspects are discussed in Section 3.4.

3.3 Moment-Curvature Relation


The moment-curvature relation (M-χ) is an often used characteristic of a section. The moment-
curvature relation varies with the axial force applied. The slope (stiffness of the section) prior to the
peak is used for deflection calculations. The post-peak behaviour is relevant when moments are
redistributed in a structural system.

Fig. 3.6 – Moment-curvature relation in function of the axial force: (a) and (b) positive moments at small to
moderate curvatures; (c) and (d) positive moments at large curvatures; (e) positive and negative
moments at small curvatures.

For cross-sections symmetric with respect to their transverse axes, the moment curvature rela-
tion is symmetric with respect to the origin. Asymmetry in slab strips is typically caused by opposite

35
Section Behaviour

faces having a different reinforcing ratio (different diameter and/or spacing) or cover. In the follow-
ing, only positive curvatures are discussed.
The moment-curvature relation of a certain cross-section is found numerically by iterating the
axial strain until a given axial force is achieved for a chosen curvature. The latter is increased step-
wise in order to gain the desired part of the moment-curvature relation. Fig. 3.6 shows the moment-
curvature relation in function of the axial force for the example defined in Fig. 3.1 (geometry) and
Fig. 3.5 (material).
In the family of curves depicted in Fig. 3.6, different features can be observed: Initially, the
moment increases until reaching a first peak followed by a decline of varying slope and magnitude.
The initial stiffness ranges from EII for cases with axial compression to EIs for cases with axial
tension (diagrams (a) and (b)). In cases with moderate axial tension the moment increase continues
with a stiffness around EIII. Except for axial forces close to the uniaxial strength, the decline is fol-
lowed by another rise, which continues until the cross-section fails (diagrams (c) and (d)). In some
cases, the moment at failure is higher than the moment at the first peak (curve for N = 3 MN in
Diagram (c)). For high compressive forces and high reinforcing ratios the moment drops to the
negative range after peaking. The absolute value of the moment at failure can be larger than the
maximum moment (curves for N = -7 to -5.5 MN in Diagram (d)). For axial tensile forces smaller
than half the pure tensile strength of the section, the moment drops to a negative value after the last
peak (curves for N = 0.5 to 1.5 MN in Diagram (c)). The moment reached does not change signifi-
cantly when the curvature is further increased (not depicted). This behaviour is caused by the rupture
of the reinforcement on one face. After this event, the remaining reinforcement bears the entire axial
tension which results in a negative moment due to the eccentricity of this reinforcement. Theoretical-
ly, a similar phenomenon occurs for small compressive forces. However, as the remaining active
reinforcing bars are under significant compression, they will, in practical cases, evade the load
through buckling. Therefore, the decline following the second peak of these relations is not shown
(Diagram (d)).

3.3.1 Phase Model for Moment-Curvature Relations

For checking detailed calculations or for preliminary estimates, simplified moment-curvature rela-
tions can be useful. Simplified moment-curvature relations can be derived from the combination of a
number of phases where the relation is based on different assumptions. Depending on the loading
history and the axial force, the moment-curvature behaviour does not go through all phases. For some
phases, the assumptions yielding the most accurate results depend on the specific geometry and
material combination and the range of axial forces of interest. A set of equations describing four
phases of the moment-curvature behaviour of the example cross-section at moderate axial tension is
presented in the following. The equations are valid only for cross-sections symmetrical about their
transverse axis. The strain planes and associated stress states for the four phases are illustrated in
Fig. 3.7.

Fig. 3.7 – Four phases of moment-curvature behaviour: (a) Phase I – elastic-uncracked; (b) Phase II – elastic,
steel only; (c) Phase III – elastic-cracked; (d) Phase IV – plastic.

36
Moment-Curvature Relation

The following ratios are used to simplify the equations representative for the different phases:

Es as E s as
ns = , ρ= , ω = ρn = (3.3)
Ec bh Ec bh

Phase I: Elastic-Uncracked
For the elastic-uncracked phase, both the concrete and reinforcing steel are assumed to be elastic
materials.

N ( ε0 ) =ε0 Ec bh 1 + 2ρ ( ns − 1) 

bh3  24e ρ ( ns − 1) 
2
M ( χ ) =χEI I where EI I =Ec 1 +  (3.4)
12  h2 
− f cc χh f χh
+ ≤ ε0 ≤ ct −
Ec 2 Ec 2

The two equations for the axial force N and the moment M are independent of one another and
can be solved in both directions (ε0, χ -> N, M or N, M -> ε0, χ) without iteration. The limit values
(onset of cracking) are defined by:

N cr 0 f ct bh 1 + 2ρ ( ns − 1) 
=

bh 2  24e2ρ ( ns − 1) 
=
M cr 0 f ct 1 +  (3.5)
6  h2 
 N  f ct − f cc
=
M cr M cr 0 1 −  for N cr ≥ N ≥ bh 1 + 2ρ ( ns − 1) 
 N cr 0  2

In the moment-axial force plane the cracking limit is a straight line.


Once a cross-section is cracked, it cannot return to the elastic-uncracked phase as long as there
are tensile stresses. For high compressive stresses over a large fraction of the cross-section, the above
relations loose accuracy due to the nonlinearity of the concrete behaviour.

37
Section Behaviour

Phase II: Elastic, Steel Only


For the elastic phase where solely the reinforcing steel is activated, the reinforcing steel is assumed to
be an elastic material.

N ( ε0 ) =2ε0 Es as

M ( χ ) =χEI s where EI s =2 Es as e 2 (3.6)


χh
≤ ε0 ≤ ε sy − eχ
2

The two equations for the axial force N and the moment M are independent of one another and can be
solved in both directions (ε0, χ -> N, M or N, M -> ε0, χ) without iteration. The transition to the
elastic-cracked phase or to the plastic phase is defined by:

2e 2 N f sy as
=
M trans N =
and χtrans for 0 ≤ N ≤
h E s as h 1 2+e h
(3.7)
2 f sy as − N
( )
f sy as
M=
trans e 2 f sy as − N and χ=
trans for < N ≤ 2 f sy as
2 E s as e 1 2+e h

In the moment-axial force plane the transition values form two straight lines.
For higher axial forces the reinforcing steel is in the hardening phase and is no longer
adequately represented by an elastic law.

Phase III: Elastic-Cracked


For the elastic-cracked phase, both reinforcing steel and concrete in compression are assumed to be
elastic materials.

 χh  1 ε0  
2
N ( ε=
0 , χ ) Ec bh  ε 0 2ω −  −  
 2  2 χh  

Ec bh3 12e2 ω  ε0  1 ε0  
2
M ( ε0 , χ ) =χ  + 1 +  −   (3.8)
6  h2
  χh  2 χh  

− f cc χh χh
≤ ε0 − ≤ 0 and − ≤ ε0 ≤ ε sy − χe
Ec 2 2

The two equations for the axial force N and the moment M are coupled. For a given axial force N the
axial strain ε0 can be expressed as a function of the curvature

 1 2 N 
ε0 ( χ, N ) = χh  2ω + − 4ω2 + 2ω −  (3.9)
 2 χh Ec bh 

such that the moment can be expressed as a function of the curvature only. The function is nonlinear
for N ≠ 0. With the above equations the concrete contribution is overestimated when the height of the
compression zone is larger than the cover of the reinforcing steel, because the respective section area
is counted both for the concrete and the reinforcing steel.

38
Moment-Curvature Relation

In the case of a moderate axial tensile force, a good approximation for the moment is found
neglecting the axial force term in (3.9):

( χ ) M trans + ( χ − χtrans ) EI II
M=

Ec bh3 12e2 ω
2
1 
=
where EI II
 2 + (1 + ξ )  − ξ   (3.10)
6  h 2  

1  1 
and ξ= + 2ω 1 − 1 + 
2  2ω 

Phase IV: Plastic


In the plastic phase, the concrete contribution is taken into account in the form of a stress block,
while the reinforcement on the tension side is assumed to be yielding and the reinforcing steel on the
compression side is assumed to be elastic. With the stress block definition of Fig. 3.4 the relevant
equations are

1 ε 
ε0 , χ ) as  f sy + Es ( ε0 − χe )  − f c bhα  − 0 
N (=
 2 χh 
α  1 ε   1 ε 
ε0 , χ ) as e  f sy − Es ( ε0 − χe )  + f c bh 2  − 0  1 − α  − 0  
M (= (3.11)
2  2 χh    2 χh  
χh
−0.003 ≤ ε0 − ≤ 0 and ε sy − χe ≤ ε0 ≤ ε sy + χe
2

The two equations for the axial force N and the moment M are coupled. For a given axial force N the
axial strain ε0 can be expressed as a function of the curvature

α
( )
N − as f sy − Es χe + f c bh
ε0 ( χ, N ) = 2
(3.12)
f c bα
+ E s as
χ

such that the moment can be expressed as a nonlinear function of the curvature only. With the above
equations the concrete contribution is overestimated when the height of the compression zone is
larger than the cover of the reinforcing steel, because the respective section area is counted both for
the concrete and the reinforcing steel. Where relevant, this can be avoided by substituting Es by
(Es - Ec).

39
Section Behaviour

Comparison
For the example defined in Fig. 3.1 (geometry) and Fig. 3.5 (material properties), Fig. 3.8(a) shows
the regions defined by the boundary conditions of the above phase descriptions in the moment-axial
force space as well as the transitions defined by the intersection of the curves of different phases for
the same axial force. For the plastic phase (IV), the stress block factor is taken as α = 0.85.
Fig. 3.8(b) shows four moment-curvature relations for the example section. The axial forces
are thereby chosen such that the stress in the reinforcing steel caused by the axial force alone
amounts to σs(N) = 0, 150, 300, 450 MPa. For comparison, the strength envelope and the moment-
curvature relations determined numerically with the material models of Fig. 3.5 are also given.

Fig. 3.8 – Phase model of section behaviour: (a) moment-axial force diagram; (b) moment-curvature relation.

It is seen that the phase descriptions given do not cover the entire domain within the strength
envelope for moderate axial tension. Between Phase III (elastic-cracked) and Phase IV (plastic) there
is a range of curvatures for which no moment is defined. The region is small for axial forces from
N = 2114 to 1307 kN and then increases with decreasing axial tension. The state not defined is the
one where the concrete behaves essentially elastic and the tension reinforcement is yielding. The
omission of this phase leads to an overestimation of the stiffness in the respective section of the
moment-curvature relation. For axial tension, the entire domain covered by Phase I (elastic-
uncracked) is also covered by either Phase II (elastic, steel only) or Phase III (elastic-cracked). This
is necessary, as there has to be a definition of how the moment-curvature relation continues after
cracking. The direct transition from Phase II (elastic, steel only) to Phase IV (plastic) is characterised
by a jump in curvature (curve N = 2867 kN in Fig. 3.8 (b)). The axial strain is discontinuous at the
transitions either from Phase I or to Phase IV. The first discontinuity is caused by a change of physi-
cal state, whereas the second is a consequence of the simplifications of the phase model.
The Phase Model strength predictions generally compare well with the numerical model, with
a slight overestimation for very low axial tension (curve for N = 0 in Fig. 3.8 (b)) and an underesti-
mation for high axial tension, where the pure steel section is dominating (curve for N = 2867 kN with
the moment at large curvatures plotted for a reduced χ-scale to the right of the regular diagram).
Except for the latter case, the curvature at maximum moment also compares well with the numerical
model.
It is concluded from the comparison, that the phase model represents the moment-curvature
behaviour up to the flexural strength well and contains the most relevant characteristic phases for
sections and axial tension, where the pure steel section is not governing. The post-peak behaviour of
a section is not well represented by the phase description presented.

40
Particular Aspects

3.4 Particular Aspects

3.4.1 Contribution of Reinforcing Steel

Both concrete and reinforcing steel contribute to the resistance of a section. In a rigid-perfectly
plastic model the strengths of the material sections are additive, see Section 3.5. In more refined
models, the contributions are additive only for a common strain plane. Because the peak stress for
concrete occurs at a strain an order of magnitude smaller than the strain at the peak stress for steel
and the concrete stress subsequently reduces to zero long before the steel stress peaks the combined
resistance is smaller than the sum of the individual resistances. The combined resistance curve has
two peaks in the compression domain, see Fig. 3.10(c). In pure compression, the first peak is higher
for common reinforcing ratios and material properties.
A similar phenomenon can also be observed for sections under combined axial force and mo-
ment, as illustrated in Fig. 3.9 for two example sections that differ only in the eccentricity of the
reinforcement. The figure shows the moment-axial force strength envelopes, the corresponding
curvatures and moment-curvature relations for selected axial forces, both for the composite section as
well as for the cases, where the section consists only of reinforcing steel (at the same position as in
the composite section) or the section consists entirely of concrete. The moment-axial force strength
envelopes for the composite section and for the concrete only section were found by a routine that
increased the curvature for a given axial force until the corresponding moment ceased to increase.
The envelope for the steel-only section was found by running through the strain planes with the strain
of the outermost steel held to ± εsu.

Fig. 3.9 – Moment-axial force-curvature relations – reinforced concrete as well as concrete and steel only:
(a) and (d) moment-axial force strength; (b) and (e) axial force-curvature; (c) and (f) moment-
curvature for selected axial forces. Identical sections except for cover: (a) to (c) e = 35 mm,
(d) to (f) e = 80 mm.

All moment-axial force strength envelopes are symmetrical about the axial force axis. The
envelopes of the steel-only sections are also symmetrical about the moment axis. The envelope of the

41
Section Behaviour

concrete-only section does not vary with the eccentricity of the reinforcing steel. The envelope of the
steel-only section keeps the same limits on the axial force axis but differs on the moment axis. The
flexural strengths vary approximately linearly with the eccentricity. The envelope of the composite
section also keeps the same limits on the axial force axis and becomes wider on the moment axis, but
the form of the envelope changes with varying eccentricity. Over a range of axial forces the flexural
strength of the steel-only section is larger than the one of the composite section. This implies that the
envelope of the composite section depicted is not the true envelope, because no part of a true enve-
lope must lie inside the envelope of one of its components.
The curvatures corresponding to the flexural strength in function of the axial force (Fig. 3.9(b)
and (e)) show that the moment maxima for the steel-only sections occur at much higher curvatures
than the maxima of the composite section and of the concrete only section. The difference decreases
with increasing eccentricity of the reinforcement, but remains significant. The combination of curva-
ture and axial strain corresponding to the moment-axial force strength envelope for the composite
section with e = 35 mm (Fig. 3.9(a)) is shown in Fig. 3.10(b).
The moment-curvature relations for selected axial forces (Fig. 3.9(c) and (f)) indicate the rea-
son for the strength envelopes of the composite section depicted in Fig. 3.9(a) and (d) partly lying
inside the envelopes for the steel only: The moment-curvature relations contain multiple peaks and it
is not always the first one that is the highest. The routine used to find the envelope for the composite
section, however, assumes the first peak value to be the flexural strength. Which peak is higher
depends on the axial force, the geometry and the material properties. For moderate to high axial
compression forces and common reinforcing ratios the first peak is governing. The limitation of the
steel strains, as it is sometimes implemented in commercial software, limits the curvature and conse-
quently reduces the flexural strength where the last peak is governing. A limitation of the compres-
sive steel strain can, however, be a reasonable way of taking into account buckling of the reinforcing
bars.
The moment-curvature relations at small curvatures (left part of Fig. 3.9(c)) illustrate that the
contributions of steel and concrete cannot be computed as the sum of the steel-only and the con-
crete-only sections, but have to be determined for a common strain plane. For high curvatures (right
part of Fig. 3.9(c)) the contribution of the concrete is negligible and the moment-curvature relation of
the composite section is almost identical to the relation of the steel-only section. Because the latter is
identical for axial forces of opposite sign with identical magnitude, the corresponding moment-
curvature relations of the composite section become almost identical for high curvatures, despite
marked differences for small curvatures. The first peak of the moment-curvature relation for a com-
pressive force is higher than the peak for the corresponding tensile force.
The moment-curvature relation for an axial tensile force of N = 3 MN (Fig. 3.9(c)) and the cor-
responding contributions of concrete and steel are discussed in detail in Section 3.4.4.

3.4.2 Opposite Sign of Moment and Curvature

Generally, a moment has the same sign as the corresponding curvature. In special cases, however,
negative moments can occur for positive curvatures and vice versa, as mentioned in Section 3.3 and
shown in Fig. 3.6.
For high compression these negative moments can be larger in magnitude than the maximum
positive moment for the same axial force. The minimum moments reached for high axial compres-
sion result in a bulge in the moment-axial force strength envelope, as shown in Fig. 3.10(a) (dash-
dotted line). The combination of negative moment and positive curvature is a sustainable state only
under special circumstances where the deformation is imposed. For common cases, where the load is
imposed, the combination mentioned is a transient state during failure. Commonly, moment-axial
force envelopes are interpreted/used under the assumption that the sign of moment and corresponding
curvature are identical. For these two reasons, the envelopes shown and discussed in the following do
not include the bulge.

42
Particular Aspects

3.4.3 Influence of Loading History

A point on the moment-axial force strength envelope can be reached by different load histories.
When non-reversible material models are used, the unambiguous allocation of a stress resultant
vector to a strain plane is lost. The allocation then depends on the loading history.
Fig. 3.10 shows three load paths. They were designed to find the largest negative moment for a
given axial compression. The compressive force was chosen, such that the largest negative moment is
situated on the bulge described in the previous section. In the first history (bold dashed line) the
moment is increased at zero axial force until the desired curvature is reached. Then, the axial strain is
increased while keeping the curvature constant. In the second history (thin dashed line), the axial
compression is increased at zero moment until the strain on the upper face has reached its final value.
Then, the negative strain on the lower face is further increased, while the strain on the upper face is
kept constant. In the third history (bold continuous line), the axial compression is increased at zero
moment to its final value. Then, the curvature is increased, while keeping the axial force constant. In
the first two cases, the compressive strains increase monotonically, such that the concrete stresses in
every point at every time lie on the stress-strain curve defined in Fig. 3.5. In the third case, a part of
the section unloads, such that the concrete stresses at some points lie below the stress-strain curve.
The evolution of stresses and strains in the section for this case is discussed in more detail in Sec-
tion 3.4.4.

Fig. 3.10 – Particular aspects of the section behaviour: (a) moment-axial force strength envelope and three load
histories; (b) axial strain and curvature corresponding to the moment-axial force strength envelope;
(c) axial force in function of a uniform strain across the section; Details of the three load histories:
(d) moment-curvature relations; (e) axial strain versus curvature; (f) axial force versus axial strain.

It is seen that the first two load histories pass through the same point on the bulge of the mo-
ment-axial force strength envelope and through the same point in the axial strain-curvature space.
The negative moment reached for the selected axial force is larger than the one reached with the third
history. The opposite is true for maximum positive moments: the envelope of the flexural strength

43
Section Behaviour

found under the condition that all concrete stresses lie on the stress-strain curve (thin dashed line in
Diagram (a)) is located inside the envelope found by increasing the curvature for a constant axial
force (continuous line). Consequently, the moment-axial force strength envelope depends on the
loading history assumed.

3.4.4 Examples of Moment-Curvature Relations

For two axial forces, Fig. 3.11 shows the moment-curvature relation with the contributions of the
steel and the concrete, the corresponding axial strain and the internal forces with their position as
well as selected strain planes with corresponding steel and concrete stress distributions. The two
relations, one for axial tension and one for axial compression, are characteristic for the particular
aspects discussed in the previous sections.

Axial Tension
The first example is one with axial tension (N = 3 MN) and a moment-curvature relation containing
two peaks, the second being the highest (the same moment-curvature relation is also shown in
Fig. 3.9(c)).
At zero curvature the section is cracked; consequently, the concrete stress is zero throughout
the section, and the reinforcing steel is in the elastic range (see Point 1 in Fig. 3.11(a) and corre-
sponding strain plane and stress distributions in Fig. 3.11(c)). With increasing curvature, the upper
reinforcement is unloaded in the elastic range, while the steel stresses in the lower reinforcement are
increased in the elastic range. At Point 2 the lower reinforcement reaches the yield plateau and the
moment increase with the curvature continues with a flatter slope (reduction of stiffness) up to
Point 3, where the strain at the top edge of the section is zero. At this point, the concrete begins to
contribute to the section resistance (appearance of Fc at the top of Fig. 3.11(b)) and the stiffness of
the section is increased until the upper reinforcement reaches the yield plateau. At Point 4 both the
upper and lower reinforcement are yielding. Up to this point, the concrete behaves almost linear-
elastically. With increasing curvature, the moment increases only slightly (stiffness close to zero) as
the steel runs through the yield plateau. At Point 5 the lower reinforcement reaches the hardening
phase and the concrete at the upper edge reaches the compressive strength. With increasing curvature,
the uppermost concrete crushes and the corresponding stresses decrease. The total contribution of the
concrete still increases, however, as the height of the compressive zone increases further. Its moment
contribution peaks when the strain-softening phase of the uppermost concrete is concluded (concrete
stress zero) and the height of the compressive zone is maximal. The resulting moment increase with
the curvature of the composite section is higher than in the previous phase. The total moment peaks
at Point 6. When the curvature is further increased, the height of the compressive zone slightly de-
creases and its centroid moves downwards (reduction of lever arm), resulting in a reduction of the
moment contribution of the concrete. The top fibres of the upper reinforcing steel unload and the
lower reinforcing steel is in the strain hardening phase, such that the moment contribution of the steel
increases with increasing curvature. The total moment has a local minimum at Point 7. At Point 8 the
moment reaches the second peak and the maximum. The lower edge of the lower reinforcing steel is
at rupture strain εsu. The lower part of the upper reinforcement is in the strain hardening phase in
tension while the upper part is yielding in compression. Note that the calculation model neglects the
Bauschinger effect (Section 2.3.2) for the unloading steel. The inclusion of this effect would only
have a marginal influence on the stress distribution and the results shown.

44
Particular Aspects

Fig. 3.11 – Moment-curvature relations – contribution of concrete and reinforcing steel for selected axial forces:
(a) and (d) moment-curvature and corresponding axial strain; (b) and (e) magnitude and position of
the internal stress resultants; (c) and (f) characteristic strain planes with corresponding concrete and
steel stress distributions.

45
Section Behaviour

Axial Compression
The second example has a high axial compression (N = - 6.64 MN) and a moment-curvature relation
where the maximum moment is smaller than the negative moment reached afterwards (the same
moment-curvature relation is also shown in Fig. 3.10(d)).
At zero curvature, the section is under a uniform compressive strain slightly smaller than the
crushing strain (Point 1 in Fig. 3.11(d)). With increasing curvature, the moment contributions of both
the concrete and the steel increase almost linearly. As the strain at the upper edge of the section
approximates the crushing strain, the moment contribution of the concrete becomes distinctively
nonlinear and reaches its maximum when the strain at the upper edge equals the crushing strain.
Crushing of the concrete at the upper edge causes a reduction of the concrete contribution to the axial
compression and a downwards movement of its centroid. Consequently, the moment contribution of
the concrete is also reduced and even turns negative for larger curvatures. As the moment contribu-
tion of the steel continues to rise independently of the concrete, the total maximum moment (Point 2)
is reached after the maximum of the concrete contribution. At Point 3 the upper reinforcing steel
reaches the yield plateau and the linear increase of the moment contribution of the steel gradually
changes to a decrease with increasing curvature. At a certain curvature, the reduction of the concrete
contribution to axial compression can no longer be compensated by an increase of the steel contribu-
tion and the section fails (Point 4).
Taking the axial compression as sole criterion, the section at zero curvature could also be un-
der a uniform compressive strain larger than the concrete crushing strain. In this case, however, when
the curvature is changed the concrete stresses throughout the section decrease. A new state of equilib-
rium is only found in cases where an increase of the steel contribution can outweigh the decrease of
the concrete contribution. Often, this is not possible and in the cases where a strain plane exists that
fulfils the aforementioned condition, the negative strains in the reinforcing steel are such that the bars
will evade the loading by buckling.

3.4.5 Non-Monotonic Loading History

The previous sections of this chapter discuss the case of monotonically increasing loads or curvatures.
In real structures, loading is generally non-monotonic. Occasionally, even load reversals occur, for
example the case of earthquakes or differential settlements. More often a structural member is loaded
and partly unloaded a number of times. Traffic on a bridge is a typical example. In statically indeter-
minate systems with elements characterised by softening behaviour, redistribution can cause parts of
the structure to unload even under monotonically increasing loads.
Fig. 3.12(a) compares the moment-curvature relations of a monotonically increasing and de-
creasing curvature (grey line) for three cases, where the curvature is increased to a certain value and
then decreased until failure occurs. The material models do not contain the Bauschinger effect (see
Section 2.3.2). Still, it is seen that the behaviour under load reversal depends on the maximum curva-
ture reached prior to unloading, that the unloading behaviour becomes softer the higher the curvature
reached during loading and that the maximum negative moment reached during unloading is smaller
than the one reached with a monotonically decreasing curvature.
Fig. 3.12(b) compares the moment-curvature relation of a monotonically increasing curvature
(grey line) with the behaviour under unloading-reloading-cycles at increasing curvatures. It is seen
that the cycles do not have a significant influence on the moment-curvature relation. In particular, the
maximum moment and the curvature at failure are virtually unaffected. The hystereses formed by
unloading-reloading increase with the curvature but remain small even for large curvatures. The
unloading becomes softer as the moment decreases.
The initial unloading stiffness depends on the strain state prior to unloading, see Fig. 3.12(c),
and can be estimated considering a curvature increment Δχ, see Fig. 3.12(d).

46
Particular Aspects

Fig. 3.12 – Particular aspects of section behaviour – non-monotonic loading history: (a) moment-curvature
relations for reversed loading; (b) moment-curvature relation with unloading sequences and the
unloading stiffness at selected points; (c) strain plane and corresponding stresses and force compo-
nents prior to unloading; (d) simplified model for an unloading step.

Because the axial force remains constant, there generally is an axial strain increment Δε0 asso-
ciated to a curvature increment Δχ. Consequently, Δε0 can be determined with the condition that the
sum of the axial force contributions must be equal to zero (ΔN = ∑ΔF = 0). With the incremental
strain plane defined, the moment increment corresponding to the curvature increment can be comput-
ed. The unloading stiffness is equal to the ratio of moment increment and curvature increment. The
unloading of the concrete and steel is represented well with linear elastic models. If the height x and
position of the compressive zone are kept the same as for the initial strain state and the stress gradient
over this zone is neglected, the unloading stiffness can be estimated. For cracked sections, where the
largest compressive strains exceed the concrete crushing residual strain εcR (see Eq. (2.10)), the axial
force contribution of the concrete is

 ε 2 + ε0  εcR
∆Fc = ∆ε( z∆c ) Ec xb =  ∆ε0 + ∆χ cR  Ec b (3.13)
 χ  χ

47
Section Behaviour

The relation between axial strain increment and curvature increment then follows as

∆χ εcR 2 + ε0
∆ε= ⋅
0
χ 2χns as (3.14)
1+
εcR b

and the unloading stiffness becomes

   
∆M  ε (ε 2 + ε )
2
1  
EI unl ( ε0 , χ=
) = Ec b cR cR 3 0 1 − 2χn a  + 2e 2 ns as  (3.15)
∆χ  χ  1+ s s  
  ε  
  cR b  

The initial unloading stiffness decreases with increasing curvature but is never less than the
stiffness of the reinforcing steel alone (EIunl ≥ EIs). Prior to cracking, the unloading stiffness is equal
to the uncracked stiffness (EIunl = EII). The unloading stiffness cannot be larger than the uncracked
stiffness.
An unloading-reloading cycle is reasonably well represented by a linear elastic relation with
the initial unloading stiffness as long as no reinforcing steel reaches the opposite yield strain. Where
this is the case, the hysteresis of an unloading-reloading cycle can become significant.

3.5 Rigid-Perfectly Plastic Behaviour


When rigid-perfectly plastic material models are used for both the reinforcing steel and concrete, one
single parameter characterising the strain plane is sufficient to determine the stress resultants M and
N. Conveniently, this parameter is chosen as the height of the neutral axis (x in Fig. 3.13). Due to the
rigid-perfectly plastic material models, only the sign, but not the magnitude of the rotation of the
strain plane (curvature) around the neutral axis, influences the stress resultants. The moment-axial
force strength envelope, see Fig. 3.13(c), is found by running the neutral axis over the height of the
section (x = 0 to h) and integrating the corresponding stresses. The strength envelope is an upper
bound solution when fc = fcc and fs0 ≥ fsy, except for high axial tension where the reinforcing steel is in
the hardening phase.
The moment-curvature relation of a rigid-perfectly plastic section is defined by a single
parameter, namely the plastic moment Mpl that depends on the axial force (for the example section
Mpl = 172, 136, 94 and 28.7 kNm for N = 0, 956, 1911 and 2867 kN, respectively, see Fig. 3.13(d)).
The concentration of the reinforcing steel in its centroids, see Fig. 3.13(b), further simplifies
the calculations. In this case, when the location of the neutral axis coincides with a reinforcing steel
centroid the corresponding steel stress σs can take any value between fc – fs0 and fs0, which results in a
straight line sector of the moment-axial force strength envelope. This simplification has little impact
on the moment-axial force strength envelope (dash dotted lines in Fig. 3.13(c)). However, in applica-
tions requiring an unambiguous relation between the height of the neutral axis and the stress result-
ants, the simplification is not suitable.

48
Summary

Fig. 3.13 – Rigid-perfectly plastic model: (a) and (b) strain plane and stress distribution; (c) moment-axial force
strength envelope and corresponding distance x between the neutral axis and the upper face;
(d) moment-curvature relation for selected axial forces.

Often, a rigid-perfectly plastic behaviour is assumed because it simplifies the structural analy-
sis and not because of the complexity of the section behaviour calculations. In these cases, the plastic
moment Mpl is taken as the peak value of the relevant moment-curvature relation (dotted grey line for
N = 1.91 MN in Fig. 3.13(d)).

3.6 Summary
Based on the assumptions of plane sections remaining plane and the behaviour of any fibre of the
section being independent of the neighbouring fibres, the stress at any point of an arbitrary ross-
section is defined unambiguously by the strain at that point, the respective material model and the
strain history. The stress resultants corresponding to a strain plane are found by integration of the
stresses over the section. The integration generally has to be carried out numerically. Simplifications
such as the stress block method or the assumption of linear elastic materials allow analytical solu-
tions for simple cross-sectional geometries. The analytical solutions are useful for verifying the
numerical integration results, and for many practical problems they yield results of sufficient accura-
cy, rendering numerical integration unnecessary.
The two most important characteristics of a section usually are the moment-axial force strength
envelope and the moment-curvature relation for a certain axial force. The moment-axial force
strength envelope consists of the maximum moments of the moment-curvature relations for all axial
forces in-between the pure compressive strength of the section and its pure tensile strength.

49
Section Behaviour

Moment-curvature relations generally contain multiple peaks. The moment-curvature behav-


iour at high curvatures is governed by the steel. Especially with hot-rolled reinforcing steel in a
section of limited height very high curvatures are reached prior to section failure. In cases the maxi-
mum moment occurs at such high curvatures, the maximum moment might not be reached in a real
structure, because the system does not have the deformation capacity required. A shift in the govern-
ing peak with changing axial force can lead to non-convexities in the moment-axial force strength
envelope. Computational routines that concentrate on the first peak do not necessarily yield the
strength envelope, but a conservative approximation. For curvatures larger than the curvature corre-
sponding to the maximum moment, negative moments can occur. For high compressive forces, these
negative moments can be larger in magnitude than the corresponding maximum moment. These
moments of opposite sign to the corresponding curvature are relevant only in very few practical cases
and are usually neglected for the moment-axial force strength envelope. The maximum moment for a
certain axial force depends on the loading history. The moment-axial force strength envelope for a
certain load history can therefore vary from the envelope corresponding to another load history.
Often, the envelopes are based on load histories resulting in a monotonic strain increase (with posi-
tive or negative sign) in every point of the section. As the variation of the envelope caused by differ-
ent load histories is generally small, it can be neglected in most cases.
Similar to the loading history, initial residual strains and time dependent effects such as creep,
shrinkage and relaxation influence the section behaviour. This is not treated in the present thesis.

50
General

4 Flexural Behaviour of RC Slab Strips:


Theoretical Modelling

4.1 General
Modelling a structure for structural analysis means simplifying a physical reality to such an extent
that it can be analysed analytically or numerically. At the time of the structural analysis, the physical
structure may already exist (for example verification of an existing structure prior to its renovation)
or not (for example design of a new structure). Even for an existing structure, it is in general not
possible to determine the values of all relevant input parameters in detail. Some are impossible to
determine, such as the strain history and the residual stresses and others can practically only be
determined randomly at discrete locations and points in time, such as the material properties and the
geometry. Common simplifications are the assumption of (sectorwise) constant or linear geometry
and uniform material properties. A frequent and significant simplification concerns the support
conditions. These are often modelled as rigidly restrained or frictionless supports of zero length.
The degree of sophistication of a model is coupled with the number of input parameters, the
variability of their values, the complexity of the calculations and with the cost of setting up, operating
and interpreting the model. When a model is used for design or verification of a real structure, the
sophistication has a natural limit in the available possibilities to determine the input parameter values.
Most of the time, however, the goal of modelling is not to achieve the highest possible sophistication
but to find the simplest model that describes the relevant characteristics of the structure accurately
enough to investigate and solve the problem at hand. In design and verification, the solution should
be practical to implement on site and conservative. In science, the solution should describe a certain
phenomenon with a certain accuracy.
Analytical models are usually the most suitable to investigate the nature of a system and to
identify the key challenges as well as the order of magnitude of the main parameters. They require
engineering judgement and a certain theoretical background and are typically custom made for a
specific structure or type of structure. They require only a limited number of parameters and imply
significant simplifications. Typically, the solutions found with an analytical model are exact solutions
within the framework of the model.
Finite element models are numerical models suitable to investigate details of a structure and
narrow down the range of reasonable parameter values. Nowadays, finite element models are imple-
mented in software packages that support the generation of a model based on CAD drawings with
little input from the user. The software can typically be operated without a deeper technical back-
ground. However, the correct application and interpretation of the results for non-standard situations
can require significant experience and theoretical knowledge. A challenge is to identify when a
situation is non-standard. As for all numerical methods, numerical issues influence the results and the
solutions found are approximate solutions within the framework of the model.
Often, analytical models can be extended by the application of numerical methods so that some
restrictions can be lifted and the degree of simplification can be reduced. For example, a variable
stiffness can be taken into account with a numerical integration of a differential equation that has an
analytical solution for constant stiffness only.
For reinforced concrete beams under transverse loads but without axial force, the prediction of
the flexural stiffness and the influence of this stiffness, which can vary along the beam and with the
load, on the deformation and the force flow has been the subject of many scientific investigations. An

51
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

early example is Branson (1968), more recent examples are Hoang and Nielsen (1996), Hag-
sten (2010) and Burns (2012).
In the following, the flexural behaviour of a reinforced concrete slab strip is investigated using
different analytical models. A propped cantilever subjected to a uniform transverse load and an axial
force is chosen as reference system because it contains regions with positive and negative moments
and has one degree of static indeterminacy, such that the redistribution of stress resultants can be
investigated. The system is illustrated in Fig. 4.1.

Fig. 4.1 – Propped cantilever: (a) static system and loading; (b) slab strip properties.

In the reference case, the slab strip has a cross-section symmetric about the y-axis. As a slab
strip is equivalent to a beam with a rectangular cross-section, and in order to ease the comparison
with the test results presented in Chapter 6 the stress resultants are given as for a beam (and not per
unit slab width).

4.2 Differential Equations


Differential equations are powerful tools to investigate the structural behaviour of static systems.
Unfortunately, analytical solutions are generally only found for cases with constant coefficients and
the subdivision of a system into sectors rapidly increases the complexity of the calculations.
The static system of a propped cantilever can be described by differential equations if the beam
stiffness EI is, at least sectorwise, constant. When the moment reaches a limit value for which the
section is assumed to be perfectly plastic the boundary conditions of the differential equations and
the coefficients of the equations change. There are different equations for cases without and with
axial force. Both cases are presented in the following. The beam stiffness is thereby assumed con-
stant over the entire length of the beam. For cases with sectorwise constant stiffness, each sector has
to be treated individually, taking into account the entire system.

4.2.1 Without Axial Force

Fig. 4.2 shows the static systems for different load levels as well as results for an example. For a
monotonically increasing transverse load q, the first change of the static system occurs when the load
reaches the yield load qy, defined by the formation of a plastic hinge at the fixed support. The load
can be further increased up to the ultimate load qu, where an additional plastic hinge forms in the
span. At this point the load cannot be increased further. Rather, the deformations can be increased
without a change in the external and internal forces.
The maxima of the deflection, slope and the moment occur at different locations, depending on
the load level. For the deflections, the difference between the maximum and the value at midspan
(x = L/2) is very small, see Fig. 4.2(e). The difference is larger in the case of the moment, but still
small compared to the magnitude of the moments. For the sake of simplicity, the values at midspan
will be used for diagrams and comparisons.

52
Differential Equations

Differential Equation and General Solution


The case without axial force is described by the differential equation

( EIw′′ )′′ = q (4.1)


EIw′′′′ q=
= for EI const

and its solution including derivatives can be written as

q
( x)
w= x 4 + c1 x3 + c2 x 2 + c3 x + c4
24 EI
q 3
w′ ( x )= x + 3c1 x 2 + 2c2 x + c3
6 EI
(4.2)
q 2 M
w′′ ( x ) = x + 6c1 x + 2c2 = −χ ( x ) = −
2 EI EI
q V
w′′′ ( x ) = x + 6c1 = −
EI EI

The coefficients c1 to c4 depend on the boundary conditions.

Fig. 4.2 – Propped cantilever: (a) definitions and static system at different load levels; (b) deformation and
stress resultants; (c) q-M relations; (d) q-V relations; (e) load-deformation relation.

53
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Boundary Conditions and Coefficients: Initial Conditions


With the boundary conditions w(0) = w′(0) = w(L) = w″(L) = 0 the solution of the differential equa-
tion and its derivatives are:

( x)
w=
q
48EI
(
2 x 4 − 5Lx3 + 3L2 x 2 )
w′ (=
x)
q
48EI
(
8 x3 − 15Lx 2 + 6 L2 x )
(4.3)
q
(
M ( x ) = −4 x 2 + 5Lx − L2
8
)
( x ) q  L − x 
5
=
V
8 

Equations (4.3) show that the deflection and the stress resultants are defined by a shape func-
tion multiplied by the transverse loading q. In addition, the shape functions for the deflection and
slope are divided by the stiffness EI. Consequently, the maxima of the functions are at the same
positions independent of the loading. The maximum deflection is given by:

qL4
= w=
wmax ( x 0.578=
L ) 0.00542 (4.4)
EI

In absolute values, the moment at the fixed support is larger than the maximum moment in the span:

1 2
M max
= −M ( x =
− M min = 0) = qL
8
(4.5)
 5  9
M=
max M=
x =L qL2
 8  128

For beams with a symmetric cross-section or where the negative plastic moment resistance at
the support is not larger than 16/9 of the positive plastic moment resistance in the span, the yield load
is defined by the onset of yielding at the fixed support and amounts to:

8M pl
qy = (4.6)
L2

Boundary Conditions and Coefficients: Plastic Moment at Fixed Support


With the boundary conditions w(0) = w(L) = w″(L) = 0 and M(0) = -Mpl the solution of the differen-
tial equation and its derivatives are:

54
Differential Equations

w( x)
=
1
48EI
( ( )
2qx 4 − q y + 4q Lx3 + 3q y L2 x 2 + 2 q − q y L3 x ( ) )
w′ ( x )
=
1
48 EI
( ( )
8qx3 − 3 q y + 4q Lx 2 + 6q y L2 x + 2 q − q y L3 ( ) )
(4.7)
1
( (
M ( x ) = −4qx 2 + q y + 4q Lx − q y L2
8
) )
q y + 4q
V ( x)
= L − qx
8

Rearranging the terms shows that the solution is composed of two functions. One is a shape
function multiplied by the load q and the other is a different shape function multiplied by the yield
load qy. Deflection and slope are inversely proportional to the stiffness EI.

w( x)
=
1 
48 EI 
( ) (
2q x 4 − 2 Lx3 + L3 x − q y Lx3 − 3L2 x 2 + 2 L3 x 
 )
w′ ( x )
=
1 
48 EI  ( ) (
2q 4 x3 − 6 Lx 2 + L3 − q y 3Lx 2 − 6 L2 x + 2 L3 
 )
(4.8)
M (=
x)
1
8  ( )
4q − x 2 + Lx + q y Lx − L2 
 ( )
L  L
V ( x )= q  − x  + q y
2  8

Consequently, the locations of the maximum deflection, slope and moment move with changing load.
The maximum moment is given by:

( )
2
 4 + q y q  L2 4q − q y
M max =
 x L =  ⋅ (4.9)
 8  128 q

The ultimate load qu is reached when the maximum moment equals the plastic moment re-
sistance of the section, provided the plastic moment resistance in the span is at least 9/16 of the
plastic moment resistance at the support:

qu qy =
(
1 + 2ξ 1 + 1 + 1 ξ
where ξ
)
M pl , span
4 M pl
(4.10)
3+ 8 M pl
=qu q =
y 11.66 2 =
for ξ 1
4 L

Both the yield load and the ultimate load depend only on the resistance of the section and are
independent of the stiffness.
When the load and/or the deformations are increased following the formation of a plastic hinge,
a rotation occurs at the hinge. The rotation is equal to the slope at the fixed support or the jump in the
slope in the span, see Fig. 4.2(b). As the slope is the integral of the curvature, any rotation requires an
infinite curvature. Consequently, the results cannot be used directly to assess the impact of a limited
curvature capacity of a section. To work around this problem one may assume a certain hinge length
with constant curvature.

55
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Softening at Plastic Hinges


The moment-curvature relation of the example beam softens after the peak, see Fig. 4.1. A possibility
to include this aspect is to introduce rotational springs at the locations of the plastic hinges discussed
previously. The springs are rigid until Mpl is reached; thereafter they are characterised by
Mspring = Mθ + kθ, where k is the spring stiffness and Mθ is the offset at θ = 0 (for the first sector
Mθ = Mpl). For softening behaviour k is negative. At the fixed support, the spring rotation corresponds
to the slope of the beam (θ = w′(0)).
Introducing such a spring at the fixed support, the coefficients in Eq. (4.2) are:

M pl qL  5kL 
+ 1 + 
L 2  12 EI  qL2 5qL (4.11)
c1 =
− c2 =
− − 3Lc1 c3 = + 2 L2 c1 c4 =
0
6 EI + 2kL 4 EI 24 EI

With the location of the maximum moment

M pl L 5kL 
+ 1 + 
qL 2  12 EI 
x ( M max ) = (4.12)
kL
1+
3EI

the maximum load is found. In the example, this load amounts to qu = 31.9 kN/m, which is slightly
less than the maximum load found for perfectly plastic hinges (qu = 33.7 kN/m). The corresponding
deflection is slightly higher for the case with springs (wmax = 142 mm and 137 mm), see Fig. 4.3.

Fig. 4.3 – Propped cantilever with springs (bold lines) instead of perfectly plastic hinges (thin lines): (a) static
systems and definitions; (b) comparison of deformations and stress resultants; (c) comparison of
q-M relations; (d) comparison of q-V relations; (e) comparison of load-deformation relations.

56
Differential Equations

If the post-peak softening in the moment-curvature relation is followed by a hardening phase


the above procedure can be applied phase-wise. The procedure can also be extended to account for
multiple hinges. The beam is subdivided into sectors with separate solutions for the differential
equation. Boundary and continuity conditions can be written in the form A∙C = R, where C is a vector
consisting of the coefficients and A is a matrix consisting of the multipliers of the coefficients in the
boundary conditions. R is a vector consisting of all the terms in the boundary conditions without
coefficients. The coefficient vector C can then be computed as C = A-1R, as shown in Eq. (4.13) for
the example.
In the example case, two hinges can form, which requires two sectors and eight constants
(c11 - c14 for the first sector and c21 - c24 for the second). The boundary condition w1(0) = 0 yields
c14 = 0. The other coefficients are computed from the remaining conditions w2(L2) = M2(L2) = 0,
w1(L1) = w2(0), M1(L1) = M2(0), M1(0) = –Mθ – kw1′(0), M2(0) = M1(0)∙(1 – L1/L) + qL1L2/2 and
M2(0) = ξMθ + ζk[w1′(L1) – w2′(0)], where ξMθ and ζk designate the initial moment and the stiffness of
the spring in the span. When the M-θ relation consists of multiple sectors (the example has two
sectors), the values of Mθ, k, ξ and ζ change depending on the sectors of the spring relations and their
combination.

−1
 − qL42 24 EI 
 c11   0 0 0 L32 L22 L2 1
 
c   0 
 − qL22 4 EI 
 12  
0 0 3L2 1 0 0
 
 c13   L13 L12 L1 0 0 0 −1

 − qL1 24 EI
4

    
 c21  =  3L1 1 0 0 −1 0 0  − qL1 4 EI
2  (4.13)
c   0 −2
  
 22   k EI 0 0 0 0  − M θ EI 
c23   0 L1 L − 1 0 0 1 0 0  − qL1L2 4 EI 
   2   
c24  3L1 2 L1 1 0 2 EI ζk −1 0  −ξM θ ζk − qL13 6 EI 

For springs with a softening behaviour the deformation cannot be increased further under a
constant load qu as in a model with perfectly plastic behaviour. In a deformation controlled system,
the load drops to a local minimum and rises again when both springs are in the hardening phase. The
system fails when the rotation capacity of the spring(s) is used up. If the moment at this rotation is
smaller than the initial moment of the spring, the load q at system failure is smaller than qu. If the
same system is force controlled, failure occurs when the load equals qu.
For the example illustrated in Fig. 4.3 the results of the spring model differ little from those of
the model with perfectly plastic hinges up to the maximum load qu. The most significant difference is
the reduction of 9 % of the maximum shear force. Following the peak, the spring model predicts
immediate failure for load controlled systems, whereas the model with perfectly plastic hinges pre-
dicts increasing deformation. For deformation controlled systems, the spring model also predicts
increasing deformations, but at a reduced load level.

4.2.2 With Axial Force

The static systems for different load levels and the results for the example beam are illustrated in
Fig. 4.4. For a monotonically increasing transverse load q and a constant tension T, the first change in
the static system occurs when the load reaches the yield load qy1 defined by the formation of a plastic
hinge at the fixed support. The load can be further increased up to the load qy2 where an additional
plastic hinge forms in the span. Contrary to the case without axial force, the load can be increased
further. Thereby the plastic hinge in the span extends such that an entire sector of the beam is
yielding.

57
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

As for the case without axial force, the maxima of the deflection, slope and the moment occur
at different locations depending on the load level. For the deflections, the difference between the
maximum and the value at midspan (x = L/2) is very small, see Fig. 4.4(e). For the moment, the
difference between the maximum and the value at midspan is small for small loads and decreases to
zero for loads larger than the yield load qy2. For the sake of simplicity, the values at midspan will be
used for diagrams and comparisons.
For systems with an axial force it is essential to take the deformations into account for the
equilibrium formulation. Consequently, stress resultants acting on the beam (N, V, M) and stress
resultants referring to the initial (undeformed) reference system (T, V0, M0) have to be distinguished.

N   1 w′ 0   T 
   − w′ 1 0   V 
V =   0  (4.14)
M  M 
 − w 0 1  
  0

An axial tension T reduces the moment acting on the beam, except at the supports where the
deformation is zero. The opposite is true for an axial compression.

Differential Equation and General Solution


The case with axial force is described by the differential equation

( EIw′′ )′′ − (Tw′ )′ =


q
(4.15)
EIw′′′′ −=
Tw′′ q for= =
EI const and T const

The equations are valid for both axial tension and compression. In the following, only the case with
axial tension will be discussed. The solution including derivatives can be written as

qx 2
w( x) =− + c1 cosh( κx) + c2 sinh( κx) + c3 x + c4
2T
qx
w′ ( x ) = − + κc1 sinh( κx) + κc2 cosh( κx) + c3
T
q
M ( x ) = 2 − T [ c1 cosh( κx) + c2 sinh( κx) ] =χ ( x ) EI =− w′′ ( x ) EI (4.16)
κ

V ( x ) =−T κ [ c1 sinh( κx) + c2 cosh( κx) ] =− w′′′ ( x ) EI

T
where κ 2 =
EI

The coefficients c1 to c4 depend on the boundary conditions. In the case of compression, the solution
comprises harmonic functions in place of hyperbolic functions.
The example in Fig. 4.4 demonstrates that the moment and shear force acting on the beam
(black lines) are significantly smaller than the moment and shear force referring to the initial refer-
ence system (grey lines). The difference increases with increasing transverse loading. For the axial
force, the difference is small.

58
Differential Equations

Fig. 4.4 – Propped cantilever under axial tension: (a) definitions and static systems at different load levels;
(b) deformation and stress resultants; (c) q-M relation; (d) q-V relation; (e) load-deformation
relation.

59
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Boundary Conditions and Coefficients: Initial Conditions


With the boundary conditions w(0) = w′(0) = w(L) = w″(L) = 0 the solution of the differential equa-
tion and its derivatives are:

qL2   x 2 
w ( x=
) −   + A[ cosh( κx) − 1] + B [sinh( κx) − κx ]
2T   L  
qL  2 x 
w′ ( x=
) − + Aλ sinh( κx) + Bλ [ cosh( κx) − 1]
2T  L 
 1 L2 
M (=x ) q  2 − [ A cosh( κx) + B sinh( κx) ]
 κ 2 
qλ L (4.17)
V ( x) =− { A sinh( κx) + B cosh( κx)}
2
2 [ λ − sinh(λ) ]
+ sinh(λ)
where A = λ 2
with λ = κL
λ cosh(λ ) − sinh(λ )
2 [ cosh(λ ) − 1]
− cosh(λ )
B= λ2
λ cosh(λ ) − sinh(λ)

Eq. (4.17) shows that the deflection and the stress resultants are defined by a shape function
multiplied by the transverse load q. The shape function depends on the ratio of axial tension to stiff-
ness. Contrary to the case without axial force, the stress resultants depend on the stiffness.
The extreme values of the moment are:

 1 L2 
M min= M ( =
x 0=) M B= q  2 − A
κ
 2  
(4.18)
 arctanh ( − B A )   1 L2
 

M max =M x=  =q  2− A −B 
2 2

 κ  κ
 2 

The moment at the fixed support is always larger in magnitude than the maximum moment in
the span, –Mmin > Mmax. Consequently, the first yield load can be determined.

2κ 2 M pl
q y1 = (4.19)
λ2 A − 2

The first yield load increases with increasing tension T and plastic moment Mpl and with de-
creasing stiffness EI and length L.

60
Differential Equations

Boundary Conditions and Coefficients: Plastic Moment at Fixed Support


With the boundary conditions w(0) = w(L) = w″(L) = 0 and M(0) = -Mpl the solution of the differen-
tial equation and its derivatives are:

q  x 2 cosh( κx) − 1 1 − cosh(λ) Lx 


w( x)= − + + sinh( κx ) +  −
T  2 κ2 κ 2 sinh(λ ) 2 
M pl  sinh( κx) x 
− 1 − cosh( κx) + − 
T  tanh(λ ) L 
q sinh( κx) 1 − cosh(λ ) L
w′ ( x )= − x + + cosh( κx) +  − 
T κ κ sinh(λ ) 2
M pl κ  cosh( κx) 1  (4.20)
− − sinh( κx) + − 
T  tanh(λ ) λ 
q  1 − cosh(λ)   sinh( κx) 
M (=
x) 2 
1 − cosh( κx) − sinh( κx)  − M pl cosh( κx) − 
κ  sinh(λ)   tanh(λ) 
−q  1 − cosh(λ)   cosh( κx) 
V ( x)
= sinh( κx) + cosh( κx)  − M pl κ sinh( κx) − 
κ  sinh(λ)   tanh(λ) 

The solution is composed of one part that depends on the transverse load q and one part that
depends on the plastic moment Mpl at the fixed support. Therefore, the locations of the maximum
deflection, slope and moment vary with the load. The maximum moment occurs at a position x1:

 
arctanh  coth ( λ ) − 
1


sinh ( λ ) 1 + κ 2 M pl q 
 ( ) (4.21)
= =
M max M ( x1 ) where x1
κ

The second yield load is reached when a plastic hinge forms in the span. It can be determined
iteratively. The following formulation converges rapidly:

ξ + cosh ( ( )) tanh (λ )
κx1 q(y 2 )
i −1

( ( ))
sinh κx1 q(y 2 )
i −1

q(y 2) = κ 2 M pl (4.22)
i

1 − cosh ( κx ( q( ) ) ) − sinh ( κx ( q( ) ) )
i −1 1 − cosh ( λ ) i −1
sinh ( λ )
1 y2 1 y2

Similar to the first yield load, the second yield load qy2 increases with increasing tension T and
plastic moment Mpl and with decreasing stiffness EI and length L.

61
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Boundary Conditions and Coefficients: Plastic Moments at Fixed Support and in Span
When there are two plastic hinges, the beam is subdivided into sectors, which are analysed individu-
ally. For the propped cantilever, there are three sectors. The first sector begins at the fixed support
and ends at the plastic hinge zone in the span. The second sector comprises the plastic hinge zone in
the span and the third sector starts at the end of that zone and ends at the support of the cantilever, see
Fig. 4.4. The first and the third sector can be described using the solution for a beam sector with
boundary conditions as shown in Fig. 4.5.

Fig. 4.5 – Beam sector with definitions and boundary conditions.

The coefficients of the general solution can be written as

2(
1 − cosh ( λi ) ) + M l cosh ( λi ) − M r
q q
− Ml
ci1 κ=2
ci 2 κ
T T sinh ( λi )
(4.23)
qL2i
− M l + M r − T ( wl − wr )
ci= 2 ci=
3 4 wl − ci1
TLi

The maximum moment depends on the transverse loading q and occurs at

 
q − κ2 M r
arctanh  coth ( λi ) − 

x1 =

 (
sinh ( λi ) q − κ 2 M l ) 

(4.24)

The boundary conditions for the first sector are: w(0) = 0, M(0) = -Mpl, w(L1) = wpl(L1) and
M(L1) = ξMpl; similarly for the third sector: w(0) = wpl(L-L3), M(0) = ξMpl and w(L3) = w″(L3) = 0.
The deflection wpl is taken from the description of the second sector. As the maximum moment Mmax
cannot be larger than the plastic moment Mpl the equation for x1 is used to find the lengths of the first
and third sector (L1, L3) for a given transverse load q. The procedure is iterative.
The second sector is characterised by yielding of the beam. The deflection is found solving the
moment equilibrium condition M(x) = ξMpl = M0(x) – Tw(x). The slope is given by the derivative of
the deflection. The shear force is zero because the moment is constant.

( )
qx 2 − M pl L ( L − x ) − ξM pl
(
w x, M = ξM pl = ) T
w pl ( x )
=

q ( L 2 − x ) + M pl L
(
w′ x, M =ξM pl = ) T
(4.25)

M ( x) =ξM pl , χ ( x ) = q T
V ( x) = 0

The stiffness only influences the results in the first and third sector. The axial tension, the
transverse load and the plastic moments have an influence in all three sectors.

62
Differential Equations

Softening at Plastic Hinges


Similar to the case without axial force, springs can be introduced at the locations where the moment
reaches –Mpl or ξMpl. In the example, the first hinge forms at the fixed support. Introducing a spring,
the boundary conditions are w(0) = w(L) = w″(L) = 0 and M(0) = Mθ – kw′(0) and the coefficients of
the differential equation solution are:

q T κ 1 Lκ2  M θ
 + − + + q
− c1 cosh ( λ )
T κ 2  k sinh ( λ ) L 2  k
c2 T κ
2
c1 =
T
+
κ

1 sinh ( λ )
k tanh ( λ ) L (4.26)

qL 1  c1
c3 =− + c4 =
−c1
T  2 Lκ2  L

The maximum moment occurs at

arctanh ( −c2 c1 )
x1 = (4.27)
κ

When the maximum moment reaches ξMpl, the beam is subdivided into two sectors with the
boundary conditions w1(0) = w2(L2) = w2″(L2) = 0, M1(0) = –Mθ – kw1′(0), w1(L1) = w2(0),
M1(L1) = M2(0), M1(L1) = M1,0(L1) – Tw1(L1) and M1(L1) = ξMθ + ζk[w1′(L1) – w2′(0)]. Based on these
conditions, the coefficients of the solution of the differential equation are found by means of a matrix
inversion and multiplication, see Eq. (A.1). If the M-θ relation consists of multiple sectors (the exam-
ple has two sectors), the values of Mθ, k, ξ and ζ change according to the sectors of the spring relation
and combinations thereof.
The maximum moment in each sector can be found using Eq. (4.27). As soon as the maximum
moment in one of the sectors reaches ξMpl, the sector is divided into two new sectors and the coeffi-
cients have to be computed adding the continuity conditions w(x–) = w(x+), M(x–) = M(x+),
M(x) = M0(x) – Tw(x) and M(x) = ξMθ + ζk[w′(x–) – w′(x+)]. The calculations become increasingly
cumbersome as the size of the matrix increases and the more sectors the M-θ-relations contains.
The results are visualised in Fig. 4.6 for the example system up to the load q = 88.6 kN/m at
which a third hinge forms. The results of the model with perfectly plastic hinges are shown for com-
parison.
The hinge in the span forms at a load marginally smaller and a deflection slightly larger than in
the model with perfectly plastic behaviour (qy2 = 68.3 kN/m and 69.0 kN/m, wL/2 = 176 mm and
172 mm, respectively). The most significant difference up to the formation of the second hinge is the
reduction of the moment M (30 %) and shear force V (20 %) at Support B.
The difference in the load-deformation relation becomes more pronounced when the load is in-
creased above qy2. The model with springs is softer than the model with perfectly plastic behaviour;
the deformation at midspan at the occurrence of the third hinge equals 253 mm compared to 239 mm
for the same load in the model with perfectly plastic behaviour. The difference in shear force at
Support B is reduced as the load increases, both in absolute and relative values.

63
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Fig. 4.6 – Propped cantilever under axial tension with springs (bold lines) instead of perfectly plastic hinges
(thin lines): (a) static systems; (b) comparison of deformations and stress resultants; (c) comparison
of q-M relations; (d) comparison of q-V relations; (e) comparison of load-deformation relations.

64
Differential Equations

The differences in the deformation and stress resultant distributions shown in Fig. 4.6(b) are
the most interesting: the deflection line has a distinct kink at the spring in the span corresponding to a
slope discontinuity at this location, while the deflection line is smooth and the slope continuous for
the model with perfectly plastic behaviour. Consequently, the shear force distribution also contains a
discontinuity for the model with springs. The moment and curvature distributions take on a festoon-
like form for the model with springs, while the moment distribution is continuous and the curvature
distribution contains two discontinuities and a sector with a constant curvature in the central part of
the span for the model with perfectly plastic behaviour. The upward peak of the moment distribution
for the maximum load determined with the spring model can be interpreted as the effect of a point
load acting in the negative z direction (upwards in Fig. 4.6(a)). The point load corresponds to the
deviation force generated by the kink in the deflection line of a member under axial tension.

Mixed Transverse Loading


In real structures uniform loads are often combined with point loads Q. In tests, uniform loads are
often simulated with a number of point loads. A certain amount of uniform load is unavoidable,
unless the self-weight of the structure is neglected.
In order to use the differential equation solutions for a beam with points loads, the beam has to
be subdivided into sectors. The continuity conditions are w(x–) = w(x+), M(x–) = M(x+),
V(x–) + Tw′(x–) – Q = V(x+) + Tw′(x+) and in the elastic phase w′(x–) = w′(x+) or at a plastic hinge
M(x) = ξMpl. Using the boundary and continuity conditions, the coefficients of the differential equa-
tion solutions are found by matrix operations, as described previously.
In Fig. 4.8 the results are shown for a single point load Q at midspan combined with a uniform
load g = 3 kN/m. Thereby, the point load represents half of a uniform load. An additional quarter of
the uniform load acts at each support; this contribution only influences the reactions at the supports
which are not shown in the figure. For comparison, an equivalent uniform load qeq = g + 2Q/L is used.
In the example, the hinge at the fixed support forms at an equivalent load and a deflection sig-
nificantly higher than for a uniform load (34.2 kN/m and 62 mm versus 23.9 kN/m and 47 mm). The
hinge in the span forms shortly after the first hinge and at an equivalent load significantly smaller
than for a uniform load (41.8 kN/m and 79 mm versus 69.0 kN/m and 110 mm). For equivalent loads
equal to or higher than the uniform load causing the formation of the hinge in the span, the deflection
at midspan is identical. The influence of the ratio of the point load to the uniform load on the yield
load are illustrated in Fig. 4.7(a) and (b). It can be seen that a significant difference in yield loads is
unavoidable whenever there is a substantial point load.

Fig. 4.7 – Influence of the ratio of point load to uniform load on the results: (a) equivalent yield loads as a
function of the ratio Q/g; (b) ratio of yield loads as a function of the uniform load contribution;
(c) equivalence for wL/2 and VB as a function of the load.

65
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Fig. 4.8 – Propped cantilever under axial tension and mixed transverse loading (bold lines) – comparison with
uniform transverse loading (thin lines): (a) static systems and definitions; (b) deformation and stress
resultants; (c) q-M relations; (d) q-V relations; (e) load-deformation relations.

The most significant differences caused by the simulation of a uniform load by point loads are
found in the moment and shear force distributions illustrated in Fig. 4.8(b). The moment distribution

66
Rigid-Perfectly Plastic Model

contains a distinct peak under the point load, whereas the moment distribution under uniform load
has an almost constant moment over the entire central part of the span. The shear force is larger in the
span and smaller close to the supports where the extreme values are found under uniform loading.
The differences in shear force increase with the load, see Fig. 4.8(d).
If point loads are used to simulate uniform loads in testing care should be taken in transferring
the results, as an equivalence requires different point loads in function of the load level and depends
on the parameter used to define equivalence. Fig. 4.7(c) illustrates the equivalent uniform load for the
deflection at midspan wL/2 (from wL/2(qeq) = wL/2(q) one finds qeq(wL/2) / q(wL/2)) and for the shear force
at the fixed support VB. In some cases, equivalence might not be reached at all. In the example, this is
the case for shear forces at the fixed support larger than 79 kN.

4.3 Rigid-Perfectly Plastic Model


Rigid-perfectly plastic models can be efficient tools for investigating the resistance of a structural
system. They are generally not suitable for investigating structural serviceability issues because they
usually do not provide useful results for small loads.

4.3.1 Without Axial Force

Rigid-perfectly plastic models result in an ultimate load for which the capacity of the structural
system is reached and do not give an indication of the corresponding deformations, see Fig. 4.9.

Fig. 4.9 – Rigid-perfectly plastic model of a propped cantilever without axial force: (a) definitions and static
system; (b) deformation and stress resultants for ultimate load; (c) load-deformation relation.

67
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

4.3.2 With Axial Tension

Contrary to the case without axial force, a rigid-perfectly plastic model yields well defined results for
the stress resultants and the deformation for loads above qy2 (occurrence of a plastic hinge both at the
support and in the span) when the system is loaded with an axial force in addition to the transverse
load. For smaller loads, the deflection is zero and the stress resultants can only be defined within a
certain range, see Fig. 4.10.

Initial Conditions
Prior to the formation of the plastic hinges at the fixed support and in the span, the beam remains in
its initial shape (w = w′ = 0) and consequently, the axial force has no influence on the results, except
for the interaction of the plastic moment and the axial force, see Chapter 3.
The stress state is not determined unambiguously but confined by the condition that the mo-
ment must not exceed the plastic moment at any point (-Mpl ≤ M(x) ≤ ξMpl). The larger the transverse
load the narrower the range of possible stress states. A first plastic hinge forms at a load not less than

 8M pl

 2
q y1 = min  L (4.28)
 8ξM pl

 L2

The corresponding range of moments and shear forces is illustrated in Fig. 4.10(b) and described by

=
w ( x ) w=
′( x) 0

− x 2 + Lx L−x
M ( x) q
= + MB
2 L
(4.29)
L  M
V ( x )= q  − x  − B
2  L
 2ξM pl 1  8ξM pl
where − M pl ≤ M B ≤ qL2  −  and q ≥
 qL2 2 L2
 

In the case where a first hinge is assumed to occur in the span when the transverse load equals
qy1 the hinge is located at midspan and moves towards the simple support as the load is further in-
creased.
Each formation of a plastic hinge reduces the degree of static indeterminacy by one. In the ex-
ample, the first hinge renders the system statically determinate.

68
Rigid-Perfectly Plastic Model

Fig. 4.10 – Rigid-perfectly plastic model of a propped cantilever under axial tension: (a) definitions and static
systems at different load levels; (b) deformation and stress resultants; (c) q-M relation;
(d) q-V relation; (e) load-deformation relation.

69
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Plastic Moment at Fixed Support and in Span


The formation of a second hinge turns the system into a mechanism. With a mechanism movements
and thus deflections occur. In axially loaded systems, deflections trigger second order effects. Axial
tension has a stabilising effect, while axial compression is destabilising. Without axial force, the
movements in the mechanism are neither defined nor limited and the load causing the formation of
the second hinge qy2 is equivalent to the ultimate load qu, see Fig. 4.9. With axial tension, the defor-
mations are defined and the load can be increased above qy2. The second yield load qy2 can be com-
puted as:

qy2 2 M=
1 + 2ξ 1 + 1 + 1 ξ (
where ξ
)
M pl , span
pl 2
L M pl
(4.30)
3+ 8
qy2 2M=
pl for ξ 1
L2

The location of the hinge in the span is:

 
L 1 
=
x1 1+
( )
(4.31)
2  1 + 2ξ 1 + 1 + 1 ξ 
 

In the presence of axial tension, the hinge expands with increasing transverse loading and
divides the span into three sectors. The lengths of the sectors are found setting the shear force to zero
(V = V0 – Tw′ = 0; for the first sector: V(L1) = V0(L1) – Tw(L1)/L1 = V0(L1) – [M0(L1) – Mpl,span]/L1 = 0):

2 M pl 2ξM pl
=L1 (1 + ξ )=
; L3 (4.32)
q q

The rigid beam sectors (1 and 3) connect the supports with the plastic sector (2) as a straight
line. For the first sector, the deformation and stress resultants are given by:

x
w( x) = w ( L1 )
L1
w ( L1 )
w′ ( x ) =
L1
(4.33)
− x 2 + Lx x−L x
M ( x) = q + M pl − T w ( L1 )
2 L L1
L  M pl T
V ( x )= q  − x  + − w ( L1 )
2  L L1

The deformation and stress resultants in the second sector are defined by Eq. (4.25). Conse-
quently, from the moment of plastification at midspan, the transverse load-deflection relation (q-w2)
at midspan is identical to the relation found by means of differential equations.
Both moment and shear force are independent of T. The shear force at the fixed support equals:

V=
B =
qL1 2 (1 + ξ ) M pl q (4.34)

70
Funicular Polygon

Softening at Plastic Hinges and Mixed Transverse Loading


The rigid-perfectly plastic model can be modified to include softening at the hinges and/or for mixed
transverse loading. The principles for these modifications are similar to those presented for the dif-
ferential equation model.

4.4 Funicular Polygon


In the presence of axial tension, a propped cantilever could bear load even without any moment
resistance (Mpl = 0). When there is no moment resistance, there is no difference between a propped
cantilever, a simple beam and a beam with fixed supports on both sides. In all systems, the defor-
mation is identical to the funicular polygon. For a constant self-weight and no additional transverse
load (g(s) = const, q(x) ≠ const, where s is the coordinate along the deformed deflection line), the
funicular polygon is described by a hyperbolic function called catenary. If the self-weight is neglect-
ed and the transverse load with respect to the initial reference system is constant (q(x) = const), the
funicular polygon changes to a quadratic parabola. The difference between the functions is small.
Where the assumption of a parabola is adequate, the deformations and stress resultants are given by:

q Lx − x 2
w ( x=
) ⋅
T 2
qL 
′( x)
w=  − x (4.35)
T2 
M ( x ) = 0, χ ( x ) = q T
V ( x) = 0

The maximum deflection occurs at midspan and is proportional to the transverse load and in-
versely proportional to the axial tension. The results are illustrated in Fig. 4.11.

Fig. 4.11 – Funicular polygon: (a) definitions and static system; (b) deflection and slope;
(c) load-deformation relation.

71
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

4.5 Numerical Integration


By numerical double integration of the curvature over the length of a beam arbitrary moment-
curvature relations can be taken into account. In statically determinate systems the moment distribu-
tion is known and the integration is straight forward. In statically indeterminate systems, the solution
is found iteratively. The moment distribution is estimated and improved based on the outcome of the
integration of the corresponding curvatures.
As for the differential equations, the integration of curvatures can no longer be continued
without modification once the moment resistance (or the first moment maximum in cases where the
resistance is reached after a local moment minimum, see Section 3.4.1) is reached and a plastic hinge
forms. The modification consists in the introduction of a hinge with a length lp and a constant curva-
ture. This can also be interpreted as a rotational spring at the location of the plastic hinge. The proce-
dure corresponds to a splitting-up of the entire moment-curvature relation into a shortened moment-
curvature relation ending at the moment resistance (or the first moment maximum) Mpl and a mo-
ment-rotation relation beginning at Mpl, see Fig. 4.13(a). The rotational spring is rigid up to the
moment resistance (or first moment maximum) followed by a moment-rotation relation that is esti-
mated with

) l p χ ( M ) − χ pl 
θ ( M= for χ > χ pl (4.36)

The spring stiffness is inversely proportional to the hinge length. The effective hinge length is
a system property and can in many cases be estimated by means of stress fields and the tensile stress-
es in the reinforcing steel (Tension Chord Model) as discussed by Sigrist (1995). Sigrist found a
hinge length equal to the effective depth of the section to be an adequate assumption for beams with
marked flanges and without axial force. The model was, however, developed for cases where the
tensile reinforcement yields prior to the crushing of the concrete compressive zone and cannot be
used without modification for cases where concrete crushing is dominant. For the following exam-
ples, the hinge length is assumed to be equal to the section height, as the tests (Galmarini et al. 2013)
showed the plastic zone to extend over a larger length than the effective depth.
When the moment at the fixed support decreases with increasing load the moment in the sec-
tions adjacent to the support is reduced, too. This unloading is treated assuming a linear elastic be-
haviour with an unloading stiffness depending on the largest curvature reached at a given section as
described in Section 3.4.5.
For the example calculations, the moment at the fixed support was varied until the computed
deflection at the simple support was smaller than a tolerance value of 1 μm. The deflection at the
simple support is very sensitive to changes of the moment at the fixed support. The load was there-
fore increased in small increments such that the moment at the fixed support did not vary much from
step to step.
The length of the integration segments can influence the results significantly. Especially in re-
gions with large curvature gradients the segment length must be small. In the example, a standard
segment length of Δx = 100 mm was used with a reduction down to Δx = 10 mm close to the fixed
support.

72
Numerical Integration

4.5.1 Without Axial Force

Fig. 4.12 shows the results of a numerical integration for the example system without axial force. It is
seen that the influence of the nonlinear moment-curvature relation of the beam section becomes more
pronounced as the load increases. As a consequence of the softening of the moment-curvature rela-
tion following the moment resistance the deflection at midspan increases suddenly for q = qy (Dia-
gram (e)). At the same load the moments and shear forces at the support change (Diagrams (c)
and (d)). In the example configuration the moment at the fixed support is close to the local minimum
when the moment resistance in the span is reached (q = qu). The load cannot be increased further
after this point because the increase in moment at high curvatures at the support does not outweigh
the reduction of the moment in the span.

Fig. 4.12 – Propped cantilever, numeric integration: (a) definitions, moment-curvature and moment-rotation
relations; (b) deformation and stress resultants; (c) q-M relation; (d) q-V relation; (e) load-
deformation relation.

4.5.2 With Axial Tension

The sensitivity of the numerical integration procedure described previously increases with the axial
tension as any difference in deflection has an impact on the moment M through the contribution of
the axial tension (M = M0 – Tw). Therefore, smaller load steps than for the case without axial force
were chosen.

73
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Fig. 4.13 – Propped cantilever under axial tension, numeric integration: (a) definitions and division of moment-
curvature relation; (b) deformation and stress resultants; (c) q-M relation; (d) q-V relation;
(e) load-deformation relation.

74
Numerical Integration

Fig. 4.13 shows the results of a numerical integration for the example system with an axial ten-
sile force corresponding to a uniform tensile stress in the reinforcing steel of 300 MPa. The nonline-
arity of the beam’s moment-curvature relation becomes very visible at higher loads. When comparing
the moment distribution with the curvature distribution for q = 100 kN/m in Fig. 4.13(b), it is obvious
which parts of the beam are in which sector of the moment-curvature relation.
The influence of softening and the local minimum in the moment-curvature relation on the
load-deflection behaviour (Fig. 4.13(e)) and on the moments and shear forces at the supports
(Fig. 4.13(c) and (d)) is less pronounced than in the case without axial force and the transitions are
less abrupt. This can be explained by the contribution of the axial tension to the moment increasing
with increasing deflections.
Because the moment-curvature curve levels off towards the moment resistance and the curva-
ture at the moment resistance is comparatively large, the moment resistance in the span is reached
only with a high transverse loading; the curvature at the moment resistance is χpl = 68 mrad/m. The
curvature in a plastic sector is χ = q/T (Eq. (4.25)). The formation of a plastic hinge in the span oc-
curs at a transverse load qy2 ≥ Tχpl = 130 kN/m.

Finite Dimension of Supports


So far, the supports were assumed to be of infinitesimal length (LB = 0). In reality, the supports have
a certain length. Usually, that length is small compared to the span. Often, the supports consist of a
steel structure or they are monolithic connections in reinforced concrete that are idealised as supports
in the static model.
For a propped cantilever, especially the dimension of the fixed support is of interest, as a dis-
tributed support reaction reduces the moment MB at the support and because it leads to a wider stretch
of large negative moments and corresponding curvatures. The modified static system of the example
is shown in Fig. 4.14(a). The distributed support reaction is calculated as

qL M ( 0 )
− − Tw′ ( 0 )
qB = 2 L
(4.37)
 L 
LB  1 − B 
 2L 

When a hinge forms at x = 0, there will be a concentrated reaction force caused by the kink in
the deflection line in combination with the tensile force in addition to the distributed reaction. The
results of a numerical integration taking into account the distributed reaction are shown in Fig. 4.14
(bold line) and compared to the case with an infinitesimal support width (long-dashed line). It can be
seen that taking into account the distributed reaction results in an increase of the load causing the
formation of a hinge at the support. Prior to the formation of the support hinge and after the local
moment minimum has been reached in this hinge, the behaviour is virtually identical to the behaviour
of the system with an infinitesimal support width. The governing shear force at the support
(Fig. 4.14(b)) seems reduced, however; the largest contribution to this difference comes from the
change in the location of the relevant section (x = 0 for an infinitesimal support and x = LB otherwise).

75
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Fig. 4.14 – Propped cantilever under axial tension, numerical integration, distributed reaction at the fixed
support, both with (short-dashed line) and without (bold line) considering a modified moment-
curvature behaviour at the support: (a) definitions; (b) deformation and stress resultants;
(c) q-M relation; (d) q-V relation; (e) load-deformation relation.

76
Numerical Integration

The distributed support reaction acts as a transverse pressure on the compression zone of the
beam. As discussed in Chapter 2.2, transverse pressure modifies the concrete behaviour, which in
turn influences the beam’s section behaviour treated in Chapter 3. For the example case, the influ-
ence is illustrated in Fig. 4.15. Most significantly, the transverse pressure caused by the support
reaction leads to an increase of the concrete compressive strength and to an increase of the moment
resistance of the section. If the increase of the moment resistance is large enough, the hinge at the
support will not form at x = 0 but at the transition to the free span (x = LB) and the distributed support
reaction is given by:

q L  Tw ( LB ) + M ( LB )
q=  + 1 − (4.38)
LB ( L − LB )
B
2  LB 

As the support reaction and therewith the transverse pressure change with the load, the section
behaviour of the beam changes with the load. The detailed consideration of this requires an iterative
approach, a good data management and enough computing power. In view of the simplifications
made for the hinge, however, it is not adequate to track the changes in behaviour too closely. Rather,
the transverse pressure at the formation of the hinge can be estimated using Equations (4.37) and
(4.38) and the corresponding moment-curvature relation can be used for the beam above the support
for all loads. The results of a numerical integration using this approach are visualised in Fig. 4.14
(short-dashed line) and compared to the case with an infinitesimal support width (long-dashed line)
and to the case with a distributed reaction and a constant moment-curvature behaviour throughout the
beam (bold line).

Fig. 4.15 – Influence of a transverse pressure on the concrete behaviour and on the section behaviour:
(a) stress-strain curves; (b) moment-axial force strength envelopes and corresponding curvatures;
(c) moment-curvature relations.

The load at which a hinge forms at the support is highest if both the distributed reaction and
the corresponding modified section behaviour are considered. The load-deflection behaviour is
identical to the other cases up to the formation of the hinge and slightly stiffer than for the other cases
at higher loads. At the formation of the hinge at the fixed support the shear force at the fixed support
is slightly higher than in the other cases, while the shear force at the simple support is identical for all
cases throughout the range of loads shown. The distributions of deformation and stress resultants
(Fig. 4.14(b)) illustrate how the influence of the model modifications at the fixed support decreases
as the distance to the fixed support increases. The general form of the distributions and the relations
between load and stress resultants respectively deformation are not altered significantly by the modi-
fications.

77
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

4.6 Comparison
So far in this chapter, different methods for investigating the behaviour of a beam under axial tension
and transverse loading have been presented. In the following, the methods are compared. For the sake
of clarity, only the basic cases of the differential equation model (perfectly plastic behaviour of the
hinges) are included.
The initial static system, the span (L) and the loads (q, T) are identical for all models. The
funicular polygon does not require additional parameters but is only applicable for cases with axial
tension. The rigid-perfectly plastic models have the least parameters. For symmetrical cross-sections,
they require a value for the plastic moment of the beam section (Mpl). For asymmetric cross-sections,
two values are required (Mpl and ξMpl). The differential equation models require the beam stiffness
(EI) as additional parameter. Note that the stiffness of asymmetric cross-sections is generally differ-
ent for positive and negative bending, a case not treated here. The numerical integration requires a
moment-curvature relation instead of plastic moments and stiffnesses. Additionally, for cases with
local minima in the moment-curvature relation, a hinge length (lp) is required.
The complexity and cost of the calculations increase with the number of parameters and the
checking of the results becomes increasingly difficult. Numerical models usually require an analyti-
cal model for the verification of the results. Depending on the goal of an investigation, different
levels of complexity might be sufficient.

4.6.1 Without Axial Force

Fig. 4.16 shows the results of the rigid-perfectly plastic model (dash dotted line), the differential
equation model (continuous line) and the numerical integration (dashed line). The first corresponds to
a rigid-perfectly plastic moment-curvature behaviour of the beam sections, the second to an elastic-
perfectly plastic behaviour and the third to a nonlinear plastic behaviour, see Fig. 4.16(a).
The rigid-perfectly plastic model yields the same ultimate load as the differential equation
model and the same moment and shear force distribution at that load. However, the rigid-perfectly
plastic model does not yield any information on the corresponding deformations and gives no infor-
mation for loads less than the ultimate load.
The results of the numerical integration and of the differential equation model differ only mar-
ginally up to the formation of the hinge at the fixed support (q = 23 kN/m). For higher loads, the
numerical integration yields larger deformations and a redistribution of the moments (smaller
moment at the fixed support, larger moment in the span) and shear forces (smaller shear force at the
fixed support, larger shear force at the simple support).

78
Comparison

Fig. 4.16 – Propped cantilever – comparison of models: (a) definitions and moment-curvature relations;
(b) deformation and stress resultants; (c) q-M relations; (d) q-V relations; (e) load-deformation
relation.

The ultimate load resulting from the numerical integration is smaller than the ultimate load re-
sulting from the rigid-perfectly plastic model or the differential equation model. In addition, once the
ultimate load is reached, the numerical integration predicts a sudden failure while the load can be
sustained with increasing deformation in the other models.

4.6.2 With Axial Tension

Fig. 4.17 shows the results of the rigid-perfectly plastic model (dash dotted line), the differential
equation model (continuous line), the numerical integration (dashed line) and the funicular polygon
(short dashed line). The first corresponds to a rigid-perfectly plastic moment-curvature behaviour of
the beam sections, the second to an elastic-perfectly plastic behaviour and the third to a nonlinear
plastic behaviour, see Fig. 4.17(a). The funicular polygon corresponds to a section with vanishing
plastic moment and/or vanishing stiffness.

79
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Fig. 4.17 – Propped cantilever under axial tension – comparison of models: (a) definitions; (b) deformation and
stress resultants; (c) q-M relation; (d) q-V relation; (e) load-deformation relation.

The funicular polygon features a linear increase of the midspan deflection with the load
(Fig. 4.17(e)). The deflection starts to increase at an infinitesimal load. The rigid-perfectly plastic
model predicts the same linear increase but with an offset. A comparison of the relevant equations for

80
Influences on the Flexural Behaviour

perfectly plastic behaviour, Eq. (4.25), and for the funicular polygon, Eq. (4.35), shows that the
perfectly plastic behaviour is composed of the funicular polygon plus a term depending on the plastic
moment(s). The resulting offset is independent of the transverse load. For the midspan deflection the
offset in terms of the transverse load is given by Eq. (4.30) and amounts to 18 kN/m for the example
case. In terms of midspan deflection, the offset amounts to ΔwL/2 = Mpl/T ∙ (1/2 + ξ). The load-
deformation relations resulting from other models lie in-between the relations defined by the funicu-
lar polygon and the rigid- perfectly plastic model when the plastic moment is taken as the maximum
moment of the moment-curvature relation. Consequently, the possible range of solutions decreases
with increasing tension and increases with increasing maximum moment. Another possibility to
check more complex calculations is that the curvature in the span cannot be larger than the curvature
corresponding to the funicular polygon (Fig. 4.17(b) and Eq. (4.25)): χ ≤ q / T.
For any model, the moment M0 with respect to the initial centroidal axis takes on values in-
between those resulting from the rigid-perfectly plastic model and the funicular polygon (Fig. 4.17(b)
and (c)). The moment M acting on the beam cannot be larger than the respective plastic moment at
any point of the beam under any load. The moment can, however, be larger than the corresponding
moment from the rigid-perfectly plastic model in regions where the latter is smaller than the plastic
moment. In the example, the moment distributions differ most close to the supports, while the general
form of the distribution only differs little. The hinge at the fixed support forms at a higher load in the
numerical integration than in the differential equation model (26.7 kN/m versus 23.9 kN/m). In the
rigid-perfectly plastic model, the corresponding load is not defined. Under increasing loads, the
moment at the support is reduced with the numerical integration, while it remains constant for the
differential equation model. In the rigid-perfectly plastic model, the load at which there is a hinge at
both the fixed support and in the span (18 kN/m) is smaller than the load at which the first hinge
forms in the other models. In the differential equation model, the hinge in the span forms at 69 kN/m
and expands under increasing load with a significantly higher curvature in the hinge and a curvature
discontinuity at the borders of the hinge sector. In the numerical integration the curvature gradually
reaches values close to those corresponding to the funicular polygon in the central part of the span.
The formation of an actual hinge occurs only at a load higher than the range illustrated.
For loads q ≥ qy2 the difference in the shear force V0 referring to the initial reference system is
constant and depends on the plastic moment at the fixed support. While the difference in V0 is com-
paratively small, the difference between the models is the largest for the shear force V acting on the
beam. The differences are mainly caused by the slope w′ taking on different distributions depending
on the model. It is worth noting that the shear force from the rigid-perfectly plastic model is not
always the largest (Fig. 4.17(b) and (d)).

4.7 Influences on the Flexural Behaviour


In the following, the influence of a number of parameters is investigated for a propped cantilever
under axial tension and transverse loading (Fig. 4.1) with the differential equation model assuming
perfectly plastic hinge behaviour. The beam is assumed to have a symmetric cross-section.

4.7.1 General

For a given static system, the beam stiffness EI, the plastic moment Mpl, the tension T and the span
length L are the dominating parameters. The influence of these parameters on the yield loads qy1 and
qy2 and the corresponding values of deflection wL/2 (at midspan) and shear force VB (at fixed support)
is shown in Fig. 4.18. The reference case is defined by the parameter values of the example used
previously, see Fig. 4.4. One parameter is varied at a time, while all other parameters maintain their
reference values.

81
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

Fig. 4.18 – Influence of main parameters on the yield loads qy, corresponding deflection wL/2 and shear force VB:
(a) stiffness EI; (b) plastic moment Mpl; (c) span length L; (d) tension T.

Increasing the stiffness reduces the yield loads as well as the corresponding deflections and
shear forces. The reduction is important for small stiffnesses and moderate to negligible for large
stiffnesses. The yield loads and the corresponding deflections and shear forces increase linearly with
the plastic moment. The yield loads decrease with the span length; initially, the decrease is pro-
nounced. For larger span lengths the decrease becomes small for the first yield load and practically
vanishes for the second yield load. The corresponding deflections increase rapidly with the span

82
Influences on the Flexural Behaviour

length. The corresponding shear forces initially decrease with the span length. For span lengths larger
than approximately 5 m the decrease of the shear force corresponding to the first yield load levels off,
while the decrease of the shear force corresponding to the second yield load changes to a significant
increase. The yield loads as well as the corresponding shear forces increase with the axial tension.
The deflection corresponding to the first yield load decreases with the axial tension while the deflec-
tion corresponding to the second yield load increases.
The ratios between the first and second yield load and between the corresponding deflections
and shear forces vary with the parameter value varied, except for the plastic moment, for which the
ratios are constant.
Often, more than one parameter is variable within a certain range. Fig. 4.19(a) is an example of
a diagram illustrating the influence of two parameters (span length and axial tension). Another possi-
bility is the use of a new parameter that combines a number of the original parameters. As the differ-
ential equation solutions already contain two such parameters, namely κ2 = T/EI (Fig. 4.19(b)) and
λ = κL, these seem a natural choice for such combining parameters. The standardisation of axes is
another means to condense diagrams, see Fig. 4.19(c). However, the more condensed a diagram, the
less obvious the interpretation and use become (a good example is found in Fig. 4.22, where Dia-
gram (c) shows the ratio of the first yield loads to tend towards infinity, while Diagram (d) shows
that the absolute value of the first yield load actually tends towards zero).

Fig. 4.19 – Yield loads as a function of different parameters: (a) span length and tensile force;
(b) κ (stiffness and tensile force) and span length; (c) λ (tensile force, stiffness and span length).

In differential equation models, the beam section behaviour is linearly elastic up to the plastic
moment. In reality, the section behaviour is highly non-linear and can contain local minima after
reaching the first maximum moment. Furthermore, the latter is not equal to the flexural strength in all
cases.
The stiffness that results in the same passive work W = ∫Mχdx under a certain load would be an
appropriate choice for the stiffness. Unfortunately, such an equivalent stiffness varies with the load
and it can only be determined with a numerical integration. On one hand, the use of a differential
equation model is limited when a numerical integration has already been carried out and on the other
hand the resulting equivalent stiffness depends on a number of numerical factors, such as step width
and integration method. Furthermore, for loads higher than the first yield load, the dissipation in the
hinge has to be taken into account as well, with W = ∫Mχdx + ∑Mθ. As the hinge behaviour itself
depends on assumptions, the equivalent stiffness would be highly dependent on that assumption,
which limits the practical value of such a calculation. Rather, it makes sense to narrow down the
range of possible values and make an informed choice within this range, see Fig. 4.20. For a given
moment-curvature relation, the maximum stiffness is easily found. Often it is the initial stiffness
(relation for N = 0 in Fig. 4.20(a)). In other cases the value is found at a point before the moment
increase levels off towards the maximum moment (steepest tangent from the origin, see relations for
N = 956 and 1911 kN in Fig. 4.20(a)). A lower bound for the stiffness is the maximum moment
divided by the corresponding curvature (relations for N = 0 and 1911 kN in Fig. 4.20(a)) or the

83
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

stiffness of the steel alone (relation for N = 956 kN in Fig. 4.20(a)). The relation for a very high axial
tension (N = 2867 kN in Fig. 4.20(a)) shows that the above considerations do not always result in
reasonable limit values. The stiffness of the steel only is an upper limit, but for the lower limit, the
first moment maximum, rather than the absolute maximum, divided by the corresponding curvature
should be used.

Fig. 4.20 – Beam bending properties: (a) stiffness; (b) plastic moment; (c) input for differential equation.

The plastic moment is limited by the maximum moment and the local minimum of the moment,
see Fig. 4.20(b). For the case with very high axial tension, where the maximum moment occurs at a
very large curvature, it makes sense to consider using first maximum as the plastic moment in sys-
tems where the large curvatures associated with the maximum moment are not likely to be reached.
The influence of the variation of the stiffness and the plastic moment within the limits de-
scribed above on the results of the differential equation model is visualised in Fig. 4.21. Generally,
the influences of a variation of stiffness and plastic moment as shown in Fig. 4.18 are found again in
the left diagrams of Fig. 4.21. The importance of the influence is, however, variable with the load q.
For the case without axial tension, the stiffness only has an influence on the deflection. The stiffness
also does not have a significant influence on the moment at the fixed support in the case with axial
tension. The influence of the stiffness on the moment at midspan vanishes for high loads. The influ-
ence of the plastic moment on the shear forces and on the deflection increases with the load up to a
certain point after which the influence remains constant. This is in line with the findings in Fig. 4.17,
where the results of the differential equation model tend towards the results of the rigid-perfectly
plastic model with increasing load: as the rigid-perfectly plastic model only depends on the plastic
moment, the influence of the stiffness must vanish for high loads and as the results of the rigid-
perfectly plastic model are composed of a term proportional to the load and a constant term depend-
ing on the plastic moment, the influence of the plastic moment must be constant for high loads.
As can be seen from the diagrams on the right hand side of Fig. 4.21, the results of the numeri-
cal integration meander within the range given by the limits for the stiffness and the plastic moment.
The numerical results reach the borders of the range in the M-q-diagram and come close to the bor-
ders in the q-wL/2-diagram. A comparatively large spread with the numerical results located in the
centre third is found for the shear force at the fixed support for the case with axial tension and loads
close to the first yield load. For higher loads, the shear forces determined numerically tend towards
the upper bound of the corresponding range resulting from the differential equation model. The
numerical results for the case without axial force are located outside the range spanned by the stiff-
ness and plastic moment variation because only a symmetric variation of the plastic moment Mpl
(ξ = 1) was considered. As the numerical results nevertheless remain within the continuation of the
borders of the respective parameter this is not investigated further.

84
Influences on the Flexural Behaviour

Fig. 4.21 – Influence of choice of beam bending properties: (a) q-M relation; (b) q-V relation; (c) q-wL/2 relation.
left: variation of plastic moment (dark hatch) and stiffness (light hatch) from basic case (thick line);
right: numerical integration compared to range by combined variation of plastic moment and
stiffness.

85
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

4.7.2 Axial Tension

The present investigation focuses on the influence of the axial tension on the behaviour of a propped
cantilever. As discussed in Chapter 3, the stiffness EI and the plastic moment Mpl are properties of
the beam section and depend on the axial force N. For a given beam section, a moment-axial force
strength envelope and the corresponding curvature values, see Fig. 4.22(a), can be calculated based
on a choice of material models. Taking the stiffness as EI = Mu/χ(Mu), it can be described as a func-
tion of the axial force. In a wide range of small to moderate axial tension, the moment and the corre-
sponding curvature vary almost linearly with the axial force or are close to constant, see Fig. 4.22(b).
Taking the plastic moment equal to the flexural strength Mpl = Mu and using linearised functions for
the relations of the curvature and the flexural strength, the yield loads and the corresponding defor-
mations and stress resultants can be expressed as functions of T, if the variation of T with the slope is
neglected (see Eq. (4.14)). The simplification N(x) = T = const simplifies the calculations without
significant loss of accuracy as N differs only slightly from T over the length of the beam, even for
high loads q, see Fig. 4.4(b), Fig. 4.10(b) and Fig. 4.13(b). The respective diagrams are found in
Fig. 4.22(c) - (f).

Fig. 4.22 – Influence of axial tension on the yield loads and corresponding deflections and shear forces:
(a) moment-axial force strength envelope and corresponding curvature; (b) linearisation of the flex-
ural strength and curvature relations; (c) increase of yield loads compared to case without axial
tension; (d) yield loads qy; (e) deflection wL/2; (f) shear force VB.

The influence of the stiffness and plastic moment (see Fig. 4.20) on the results from a differen-
tial equation model for different tensile forces T is illustrated in Fig. 4.23. The spread of the stiffness
increases with the axial tension, while the spread of the plastic moment decreases. Consequently, the
spread of the results can be expected to be largest for moderate loads where both the stiffness and the
plastic moment have a significant influence. For the load deflection relation (Fig. 4.23(d)) this expec-
tation comes true. Furthermore, the spread is smaller with increasing tension. The variation of the
stiffness only influences the moment at the fixed support when the variation is large (relation for
T = 2867 kN in Fig. 4.23(a)). The moment at midspan is affected by a variation of the stiffness as
soon as there is a tensile force present (Fig. 4.23(b)). The combined influence of a variation of stiff-

86
Influences on the Flexural Behaviour

ness and plastic moment is largest for the shear force at the fixed support, see Fig. 4.23(c). The
spread increases with the tension and is largest for loads between the first and second yield load. For
the highest tension shown (T = 2867 kN), the maximum spread is more than two thirds of the mean
value, compared to 5 % for the case without axial force.

Fig. 4.23 – Influence of axial tension and variation of the results depending on the stiffness and plastic moment:
(a) moment at the fixed support; (b) moment at midspan; (c) shear force at fixed support;
(d) deflection at midspan.

87
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

4.8 Summary
Modelling a structure for structural analysis means simplifying a physical reality to such an extent
that it can be analysed analytically or numerically. The degree of sophistication of a model is coupled
with the number of input parameters, the variability of their values, the complexity of the calculations
and with the cost of setting up, operating and interpreting the model. The exact value of most param-
eters is either unknown or known only within a certain range. Consequently, the structural behaviour
can never be calculated exactly, regardless of the model sophistication, it can only be estimated. A
good model should describe the relevant characteristics of the structural behaviour well enough to
enable solving the problem at hand or studying a particular aspect of its behaviour, while being as
simple as possible.
Using the example of a propped cantilever under axial tension and transverse loading, different
modelling techniques have been presented and discussed: differential equations, rigid-perfectly
plastic models and numerical integration. In all models, in order to gain meaningful results, the
equilibrium conditions have to be formulated for the deformed system when axial forces are present.
There are distinct differential equations for the cases with and without axial force. Closed-form
solutions are only found in special cases and for beams with sectorwise constant stiffnesses. Every
change in the system, such as the formation of a hinge, entails a new solution. The solutions differ in
their coefficients and/or in the number of sectors with individual solutions (and sets of coefficients).
Once the solution or the set of solutions for a certain structural system and loading is found, a wide
range of aspects of structural behaviour can be efficiently investigated.
Rigid-perfectly plastic models are the simplest models but they yield only a limited amount of
information on the structural behaviour. For cases without axial force, only the load under which the
system transforms into a mechanism is found. For cases with axial force, information on the defor-
mation is found for loads larger than required to turn the system into a mechanism stabilised by the
tensile force.
The challenges of a numerical integration are (i) to develop a calculation set-up that can handle
sharp changes in stiffness and non-monotonous stiffness increases; and (ii) to identify a suitable
parameter and its starting value such that the numerical procedure converges towards a solution. The
calculation set-ups are often customised for a specific structural system, given beam properties and
loading range and their adaptation for many different cases is tedious. The numerical integration
allows a detailed investigation of the consequences of a variation of the initial parameters within a
limited range.
In general, when the (first) maximum moment is reached and the load is further increased (or
sustained in the case without axial force), a kink forms in the deflection line. The kink is associated
with a discontinuity of the slope, which in turn is associated with an infinite curvature over an infini-
tesimal length. If the moment-curvature behaviour softens after the moment peak (i.e. for all cases
except perfectly plastic behaviour), the underlying theory with plane sections remaining plane ceases
to give realistic results and other models, for example stress field models, have to be used for the
estimation of the structural response under increasing loads. Alternatively, a work-around in the form
of a virtual hinge length can be used to continue with the original models. As a consequence of the
hinge length the curvature in the hinge is finite. The curvature corresponds to an associated moment
in the moment-curvature relation and ordinary procedures can be used.
In the case without axial tension, softening after the peak moment reduces the ultimate load
(bearing capacity of the system), because the flexural strength cannot be activated simultaneously at
the fixed support and in the span. In the case with axial tension, a perfectly plastic behaviour of the
hinge in the span results in an expansion of the hinge. Within the hinge the rigid-perfectly plastic and
the differential equation models are identical and the results can be interpreted as the sum of a fu-
nicular polygon contribution and a bending contribution. A hinge with softening behaviour in the
span leads to a festoon-like moment distribution and a multitude of hinges forming in the span as the
load increases.

88
Summary

A more detailed modelling of the support with a finite support width and a distributed support
reaction (and possibly by taking into account the influence of the transverse pressure on the moment-
curvature behaviour) has a moderate effect on the stress resultants close to the support and a negligi-
ble effect on the general distribution of deformations and stress resultants and on the load-deflection
curve.
The main influencing parameters for the behaviour of a given structural system under axial
tension are (i) the tensile force itself; (ii) the plastic moment(s); and (iii) the stiffness. For a given
beam section the plastic moment and the stiffness depend on the axial force and the variability of the
results can be expressed as a function of the axial force only. For example, an increasing axial force
results in increasing transverse loads required for the formation of hinges, an increasing system
stiffness and a reduction of the shear forces. The influence of the tensile force is more pronounced
the higher the transverse load. For the propped cantilever investigated, the influence of the variability
of the stiffness is significant primarily for transverse loads between the first and second yield load.
For higher loads, the influence vanishes. Consequently, the results of the differential equation and the
rigid-perfectly plastic models converge for higher loads, see Fig. 4.24. The shear force at the fixed
support is the most sensitive to a variation of the plastic moment and stiffness. The deflection at
midspan is the least sensitive.

Fig. 4.24 – Influence of axial tension and variation of the results for different models: (a) shear force at fixed
support; (b) load-deflection relation. Hatch: differential equation; thin lines: rigid-perfectly plastic;
bold line: numerical integration for T = 1911 kN.

With increasing transverse load the load-deflection relation approaches a parallel to the linear
load-deflection relation corresponding to the funicular polygon for a given tensile force.
Care should be taken when simulating uniform loads with discrete point loads. Even though
the load-deflection relations match closely, the differences in the distributions of stress resultants and
their magnitude at a specific location can vary significantly.

89
Flexural Behaviour of RC Slab Strips:
Theoretical Modelling

90
General

5 Shear Behaviour of RC Slab Strips:


Theoretical Modelling

5.1 General
As emphasised in Chapter 4, the shear forces and moments are inherently linked. The shear force is
the derivative of the moment or the variation of the moment over a certain length of a beam. Fig. 5.1
illustrates the shear force distribution in a propped cantilever and its change with the transverse load
for the cases with and without axial tension.

Fig. 5.1 – Shear forces in a propped cantilever: (a) system with and without tensile force; (b) deformation and
stress resultants; (c) q-V relations; (d) q-w relations.
Note: underlying model = differential equation model.

91
Shear Behaviour of RC Slab Strips:
Theoretical Modelling

The distributions shown in Fig. 5.1(b) correspond to the yield load (continuous line, thin for
T = 0, bold for T = 1.9 MN), to the ultimate load of the case without axial tension (dash-dotted line,
for T = 0 only) and to the second yield load of the case with axial tension (dashed line, for
T = 1.9 MN only). It can be seen that the largest shear forces occur at the fixed support independently
of the load q and of the axial tension T. For loads up to the yield load of the case without axial ten-
sion, the deformation adjacent to the fixed support and the shear force at the support of both cases
match closely. For higher loads, the axial tension reduces the increase of the shear force at the sup-
port with increasing q. Furthermore, a difference between the shear force referring to the undeformed
reference system and the shear force acting on the beam at the fixed support develops and increases
with q. The difference is caused by the rotation of the beam in the hinge at the support. The slope w′
in connection with the axial tension acting on the beam also causes the shear force to decay faster
with increasing distance from a support than in the case without axial tension. Higher shear forces are
reached in the case with axial tension than in the case without, but they occur at significantly higher
transverse loads. In the absence of axial tension the attainable shear forces are limited by the flexural
strength of the beam.
Because of the interconnection of moment and shear it is essential to model the flexural behav-
iour of a structure adequately when the shear capacity is a concern.
The shear resistance of reinforced concrete members has been the subject of extensive research
for more than a century and yet to this day there is no mechanical model capable of predicting the
shear resistance accurately over a wide range of cross-sections and stress resultants for members
without shear reinforcement. Key reasons therefore are the many influencing factors and the fact that
the relative importance of these influencing factors differs depending on the situation. The matter is
further complicated by the fact that the shear resistance depends on the tensile strength of concrete
that in itself is challenging to determine (see Section 2.2.3) and scatters significantly. In design, the
issue is often avoided by providing shear reinforcement in critical regions or throughout the member.
In members without shear reinforcement, shear failure is caused by a crack – usually inclined
away from the nearest support or concentrated load – that prevents forces transverse to the member
axis from being transferred from the lower part of the member to the upper part, see Fig. 5.2(a). More
frequent than a pure shear failure is the combination of the opening of the shear crack and crushing in
the zone above the crack, see Fig. 5.2(b). In members with shear reinforcement, this reinforcement
takes over the transfer of vertical forces upon opening of the shear crack and failure is caused either
by rupture of the shear reinforcement or crushing of the concrete in the compression diagonal, see
Fig. 5.2(c).

Fig. 5.2 – Shear cracks: (a) pure shear failure; (b) combined shear flexure failure; (c) shear reinforcement and
compression diagonal in concrete; (d) shear crack originating in flexural crack; (e) shear crack with
central horizontal origin.

Shear cracks are often the inclined continuation of flexural cracks, see Fig. 5.2(d), but shear
cracks can also occur independently of the flexural cracks. There are examples of shear cracks form-
ing horizontally close to the centre of the section from where they expand diagonally towards the free
surfaces, see Fig. 5.2(e).
The forming of a shear crack is not equal to failure; different mechanisms enable a force trans-
fer in and around a crack even in the absence of shear reinforcement: (i) force flow in the uncracked
part of the member (usually the compression zone), (ii) aggregate interlock in the crack and (iii)
dowel action of the ordinary reinforcement. Fenwick and Paulay (1968) concluded from their exper-
imental campaign that aggregate interlock was by far the most important factor for the force transfer
followed by the compression zone and the dowel action. The latter was noted to vary significantly,

92
Shear Stress Distribution Prior to Shear Cracking

depending on the position of the respective reinforcement within the section and the casting condi-
tions.
A crack initiates when the maximum principal stress σ1 (see Fig. 5.3(e)) reaches the concrete
tensile strength. The crack expands perpendicularly to the direction of this principal stress. Prior to
cracking, the behaviour of concrete is approximately linearly elastic.
The experimental evidence that shear cracks form as continuations of flexural cracks or as in-
dependent cracks and that aggregate interlock enables a significant shear transfer across the crack
suggests that disregarding the influence of flexural cracks on the shear flow (flexural cracks only
prevent axial tensile stresses but do not prevent the transfer of shear stresses) might be a reasonable
simplification estimating the shear stress distribution prior to the formation of shear cracks. Such a
shear stress distribution is presented in Section 5.2.
Cracking causes a redistribution of the stresses. This redistribution depends on many parame-
ters. Where redistribution is not possible, the member fails in shear over a certain length if the load is
not reduced. Such a local shear failure is not in all cases equal to the failure of the entire member.
The influence of a local shear failure on the overall structural response of a propped cantilever under
tension is discussed in Section 5.3.
The shear capacity of a cracked member (i.e. after redistribution) is not subject of the present
work. For an introduction, the reader is referred to Collins and Mitchell (1997). Ehmann (2003)
contains an in-depth review of relevant models and experimental data. A widely known model for the
shear capacity of a cracked member was developed by Vecchio and Collins (1986). Based on this
model and a number of experimental results, Adebar and He (1994) observed that a small to moder-
ate axial tension only moderately reduces the shear capacity of a cracked member.

5.2 Shear Stress Distribution Prior to Shear Cracking


In the mid-nineteenth century Jourawski (1856) published an analytical relation of the axial (flexural)
stresses and the shear stresses in a homogeneous prismatic member subjected to flexure and shear
based on equilibrium conditions. Around 1900 the equilibrium conditions were used to determine the
shear stresses for a discontinuous distribution of axial flexural stresses assumed for a reinforced
concrete section with bending cracks (Mörsch 1902). The shear stress τ is thereby assumed to be
variable only with the height over the section and constant over the width. A number of researchers
later investigated the influence of flexural cracks on the shear flow. Based on a review of the research
at the time, Collins and Mitchell (1997) concluded that “Mörsch’s predicted distribution was fairly
accurate after all”.

5.2.1 Transformed Section Model

For a section as shown in Fig. 5.3(a) equilibrium of the free body leads to:

zs
bτ d=
x ∫ b dσ x ( z ) dz (5.1)
zbot

As the axial stress σx distribution for a given moment M can be expressed as a function f (of
the geometry, material and axial force N) multiplied by the moment, and as the shear force V is the
derivative of the moment, the shear flow bτ at a given height can be expressed as:

93
Shear Behaviour of RC Slab Strips:
Theoretical Modelling

=σ x( z ) f ( geometry, material, N ) ⋅ M

dM
=V (5.2)
dx
zs
bτ V ∫ b ⋅ f ( geometry, material, N ) dz
=
zbot

The shear flow distribution is a function scaled linearly with the shear force. For a homogene-
ous section of linear elastic material and a constant axial force the function f can be derived analyti-
cally as a function of the width b and the moment of inertia Iy. In this case the shear flow is

zs
VS
bτ = where S( zs ) ∫ b( z ) z dz (5.3)
Iy zbot

Fig. 5.3(a) shows the shear stress distribution for a rectangular section and Fig. 5.3(b) the dis-
tribution for the case where the width of the section is doubled in the upper part. It can be seen that
the changed section modifies the shear flow distribution and that a jump in width causes a jump in
the shear stress distribution. The shear stress distribution implies a significant shear stress at the
lower edge of the flange where the shear stress must be zero. This shows a limitation of the method.
When a section consists of materials with different stiffnesses, the shear flow distribution can be
derived based on a transformed section where the width bn is the sum of the widths of the individual
materials weighted by the ratio of their stiffness and the reference stiffness bn = (∑biEi)/Eref .

VSn zs
bτ = where Sn ( zs ) ∫ bn ( z ) z dz (5.4)
I yn zbot

When the entire width of the section at a given height consists of the same material, the shear
stress is found by the division of the shear flow by the width b. The example in Fig. 5.3(c), where the
stiffness of the upper part of the section is twice the stiffness of the lower part, shows that the result-
ing shear flow is identical to the one in Fig. 5.3(b) (the transformed section is identical), while the
shear stresses differ in the upper part (the width is different).
When the width consists of more than one material, the distribution of the shear flow to the
different materials has to be estimated. Here, the contribution of each material is assumed to be
proportional to the width multiplied with the initial elastic modulus (for the concrete
τc = bτ / (bc + nsbs)). This distribution is somewhat arbitrary. It is, however, of little importance as (i)
it rarely affects the peak values (in the majority of cases, they are located either in the compression
zone or in the central zone between the reinforcement layers); and (ii) the stress distribution close to
the reinforcement is dominated by the force transfer between the reinforcing steel and the concrete.
When stress-strain relations are nonlinear the transformed section can usually only be deter-
mined numerically. Thereby, the tangential stiffness Etan, see Fig. 5.3(d), corresponding to the strain
at a given height ε(z) is to be used.

94
Shear Stress Distribution Prior to Shear Cracking

Fig. 5.3 – Shear stress distribution: (a) rectangular section with notation and free body diagram; (b) T-section;
(c) section with different materials; (d) stress-strain relations and corresponding tangential stiffness
for concrete and steel; (e) Mohr’s stress circle; (f) example.

Fig. 5.3(f) illustrates the procedure for the example section defined in Fig. 3.1 for a given set
of stress resultants N, M, V. As the transverse stresses σz can usually be neglected in reinforced
concrete slab strips, the principal stresses σ1, σ2 and their angle ϕ to the horizontal, see Fig. 5.3(e),
are given by:

σx σ2x
σ1,2
= ± + τ2
2 4
(5.5)

tan 2ϕ =
σx

The procedure does not account for the interaction of the axial and shear stresses in connection
with the stress-strain relation. This simplification reduces the complexity of the calculations signifi-
cantly. Under the assumption that the highest principal stress cannot be larger than the concrete
tensile strength fct, the maximum shear stress in a section can in most cases not be larger than the
tensile strength due to the regions of constant shear and zero axial stress (where the section is cracked
in flexure). This stress state is shown with continuous lines as a Mohr’s stress circle in Fig. 5.3(e).
When a shear stress equal to the tensile strength is concurrent with an axial stress equal to the crush-

95
Shear Behaviour of RC Slab Strips:
Theoretical Modelling

ing strength fc, the principal stress σ2 is larger than the crushing strength (dotted lines in Fig. 5.3(e)).
Using the empirical relation for the tensile strength (Eq. (2.2)) the principal stress σ2 becomes:

−2 3
−σ2 1 + 1 + 0.36 f c
= (5.6)
fc 2

The resulting difference (2 % for fc = 10 MPa, 0.5 % for fc = 80 MPa) is negligible in the light
of the scatter of the material properties. For smaller axial stresses the difference is larger, however at
smaller stresses the tangential stiffness does not vary significantly so this difference is acceptable, too.
The difference between the shear flow above and below the reinforcement corresponds to the
shear force that has to be transferred to the reinforcement through bond shear. Assuming again that
the maximum shear stress is equal to the tensile strength, the maximum bond stress can be computed
as:

b f ct
τ b,max = (5.7)
nbar π Ø

For a given bond shear capacity and number of bars nbar or bar spacing sy the minimum bar di-
ameter Ø required for bond not to govern can be determined. For reinforcement in a single layer and
Sigrist’s bond model (Eq. (2.32)), the minimum diameter is given by:

b
Ø≥ for beams
2π nbar
(5.8)
sy
Ø≥ for slab strips

Note that Sigist’s model only considers pull-out type failure. Where splitting is governing the
bond shear capacity is lower and, consequently, the minimum bar diameter larger.

5.2.2 Neighbouring Sections Model

In the later editions of his classic “Der Eisenbetonbau” (1923; 1929), Mörsch pointed out that the
procedure using a transformed section does not result in an exact solution, but merely an approxima-
tion for cases where the height of the neutral axis zc varies with changing moment. He suggested that
a better approximation is found based on the axial stress distribution of two neighbouring sections,
see Fig. 5.4. The shear stress is determined assuming a uniform shear stress τ over Δx between the
sections and setting the shear force at a given height equal to the difference of the horizontal forces
below, as illustrated in Fig. 5.4(b).

zs b 2
∫ ∫ ∆σ x dy dz
zbot −b 2 (5.9)
bτ =
∆x

The question of the distribution of the shear flow at heights with mixed materials is not pur-
sued further here. Rather, a distribution based on the widths weighted with the initial elastic modulus
is used.

96
Shear Stress Distribution Prior to Shear Cracking

Fig. 5.4 – Shear stress distribution based on the difference of the axial stresses in neighbouring sections:
(a) stresses and strains in neighbouring sections; (b) horizontal equilibrium and heights of neutral
axis; (c) shear stress distribution.

The distance Δx between the sections is related to the moment increment ΔM and the shear
force V by the equilibrium condition:

∆M
∆x = (5.10)
V

Fig. 5.4(c) shows that the difference between the resulting shear stress distribution (hatched
area) and the stress distribution determined with the procedure based on the transformed section
(short-dashed line) is small, even for a significant difference in moment (ΔM/M = 9 %). A smaller
difference in moment results in a smaller difference. Because taking into account two sections simul-
taneously complicates the calculations without a significant change in results, the procedure based on
transformed sections is used here.
Comparing the neighbouring section model to the transformed section model it becomes clear
that the neighbouring section model goes over into the transformed section model when the distance
Δx is infinitesimal; Equation (5.9) becomes (5.1) and (5.10) becomes (5.2-2). Consequently, in
practice, the calculation procedure can be set up according to either model depending on which
procedure fits better into the overall calculation set-up.

5.2.3 Section Behaviour

Fig. 5.5 illustrates the variation of the shear flow and of the cracking shear force Vcr with increasing
curvature for the example section (see Fig. 3.1) with an axial tension of N = 1911 kN.
Fig. 5.5(a) and (b) as well as the strain plane and the axial stress distributions in Fig. 5.5(e)
show the details of the flexural behaviour in the same way as Fig. 3.11. Fig. 5.5(b) zooms in on the
initial part of Fig. 5.5(a) and shows the shear force at the onset of shear cracking. As previously, the
latter is assumed to occur when τmax = fct. The cracking shear force for curvatures larger than the
curvature at the maximum moment is usually not of interest, because in most structures plastic hinges
form at the supports or discrete loading points, where the transverse load is resisted directly through a
fan type stress field rather than being transferred as shear stresses, or the hinges form at regions with
vanishing shear forces.
There are three well-defined sectors in the cracking shear force-curvature relation. Initially
(between points 1 and 2 in Fig. 5.5(b)), there is no compression zone and the transformed section
consists of the (elastic) reinforcing steel alone. Only the central part in-between the two reinforce-
ment layers carries the shear flow and the cracking shear force is constant throughout the sector. At
higher curvatures the transformed section is enlarged by the contribution of the compression zone.
Except for the cover at the tension side, the composite section carries the shear flow and the cracking
shear force increases with the height of the compression zone. The onset of yielding of the rein-
forcement on the tension side marks the transition to the last sector. When the entire reinforcement

97
Shear Behaviour of RC Slab Strips:
Theoretical Modelling

on the tension side is in the yield plateau (Point 4) the transformed section consists only of contribu-
tions of the compression zone and the adjacent reinforcement. The shear flow is limited to the area
between the compression face and the reinforcement on the same side and the cracking shear force is
significantly reduced. With increasing curvature, the concrete compressive strength is reached within
the section and the contribution of the compression zone to the transformed section and to the height
carrying the shear flow is reduced, see Point 6. The corresponding cracking shear force decreases
with increasing curvature.

Fig. 5.5 – Moment and cracking shear force as function of the curvature: (a) moment-curvature relation,
contributions of concrete and reinforcing steel, corresponding axial strain; (b) cracking shear force
and moment at small to moderate curvatures; (c) magnitude and position of the resultants of internal
axial forces; (d) states of concrete and reinforcing steel at small to moderate curvatures;
(e) characteristic strain planes with corresponding transformed sections and stress distributions.

Strictly keeping to the principles described initially would result in a negative width of the
transformed section for negative strains larger than the crushing strain (i.e. where the concrete is
softening). The shear flow distribution would contain both positive and negative parts. The cracking
shear force would drop more rapidly after Point 5 in Fig. 5.5(b) and turn negative before the flexural
strength is reached. However, the shear flow in the crushing zone is driven by the unloading of the
concrete adjacent to the crushing zone and is not related to the overall shear flow. As the crushing

98
Shear Stress Distribution Prior to Shear Cracking

zone (and the adjacent concrete) are distributed over a certain length Δx, pure section analysis used as
basis for the present model cannot adequately describe the phenomena involved. Disregarding the
softening zone as proposed here is therefore merely a work-around.
Simplifications as introduced for the flexural section behaviour (Section 3.2) can also be used
for the estimation of the shear behaviour of a section. For example, the concentration of the reinforc-
ing steel area at points corresponding to the centroids of the reinforcing bars leads to abrupt changes
in the shear flow at the heights of the reinforcement layers instead of the curved functions shown
here. As the latter are arbitrary approximations anyway, the simplification does not result in any
significant loss of information or accuracy. When the flexural behaviour is described by a phase
model, where each phase is defined by analytical expressions (Section 3.3.1), the corresponding
transformed sections and shear flow can equally be expressed analytically.

5.2.4 Variation of Section Behaviour

The drop in the cracking shear force curve in Fig. 5.5(b) (prior to Point 4) is caused by the reinforc-
ing steel yielding and is less abrupt for cold-worked steel, see Fig. 5.6 (measured values for cold-
worked steel: dash-dotted line; model results for cold-worked steel with example steel characteristics:
dash-double dotted line). The drop is not significantly reduced when a stress-strain relation without
plateau is used for the reinforcing steel (dashed line) because the strain-hardening modulus is much
smaller than the initial elastic modulus (760 MPa versus 215 GPa in the example).

Fig. 5.6 – Influence of reinforcing steel on moment-cracking shear force relation: (a) stress-strain relations for
different models of reinforcing steel; (b) cracking shear force-curvature relations; (c) moment-
cracking shear force relations.

Fig. 5.7(a) shows the cracking shear force-curvature and moment-curvature relations corre-
sponding to four different axial forces for the example section. The relations are only shown only up
to the flexural strength (or the first moment maximum in case of N = 2867 kN). All relations for axial
tension start with the same cracking shear force. Provided the reinforcement does not yield prior to
the activation of the compression zone, the higher the axial tension (N = 956 kN and 1911 kN), the
longer the cracking shear force remains constant. Afterwards, the cracking shear force increases
significantly over a small curvature increase. Thereafter, the cracking shear force increases moderate-
ly with increasing curvature or even decreases slightly before the reinforcing steel at the tension side
yields. The yielding causes the cracking shear force to drop to values lower than the initial value. In
cases where the reinforcement yields prior to the activation of the compression zone (N = 2867 kN),
the cracking shear force drops abruptly to a significantly lower level at which it remains until the
compression zone is activated or strain hardening begins. The relation for the case without axial force
starts at a value significantly higher than for the sections with axial tension. Flexural cracking causes
a slight initial drop. The cracking shear force continues to decrease slightly with increasing curvature.
When the moment approaches the flexural strength, the decrease becomes more pronounced. The

99
Shear Behaviour of RC Slab Strips:
Theoretical Modelling

states of the reinforcing steel and the concrete for the cases N = 0 and N = 2867 kN are shown in
Appendix B.

Fig. 5.7 – Cracking shear force relations for selected axial forces: (a) moment-curvature and cracking shear
force-curvature relations; (b) moment-cracking shear force relations.

A comparison of the cracking shear force as a function of the moment for different axial forces
(Fig. 5.7(b)) shows that a small to moderate axial tension does not significantly change the maximum
cracking shear force. However, with increasing tension the moment (and curvature) range over which
a high cracking shear force is maintained decreases. Axial tension furthermore causes the cracking
shear force for small to moderate moments (at small curvatures) to be significantly smaller than for
the case without axial force. In all cases, yielding of the reinforcing steel causes a significant reduc-
tion of the cracking shear force. When yielding of the reinforcing steel causes a reduction of the
moment and the moment (and the cracking shear force) recovers at higher curvatures, the drop is
reduced in load controlled systems, because the entire range of curvatures corresponding to the local
moment minimum will be concentrated in a single point and result in a jump the curvature relation.
An increase in the eccentricity of the reinforcement increases the initial cracking shear force
and the maximum cracking shear force.

5.2.5 Non-Monotonic Loading History

In the previous section, is was shown how the cracking shear force changes with a monotonically
increasing curvature or moment. In real structures the loading is generally non-monotonic. The
influence of non-monotonic loading on the flexural behaviour was discussed in Section 3.4.5.
Fig. 5.8 shows a non-monotonic loading history (Diagram (a)) for the example section with an axial
tension of N = 1911 kN. Diagram (b) illustrates the results as a function of the curvature. The corre-
sponding cracking shear force-moment relation is shown in Diagram (c).
It is seen that an unloading in flexure at curvatures higher than the one corresponding to the
maximum cracking shear force (χ = 21 mrad/m in Fig. 5.8(b)) increases the cracking shear force
abruptly to a value close to the maximum. This increase is caused by the reinforcement that was
yielding turning elastic again. Depending on the curvature at the beginning of the unloading, the
cracking shear force remains at the same level throughout unloading or gradually decreases to the
cracking shear force corresponding to the initial state, where only the reinforcement contributes. The
plateau with the high cracking shear force in Fig. 5.8(c) widens as the moment at the beginning of the
unloading increases. In the domain of interest (curvatures smaller than the one corresponding to the
maximum moment) reloading closely follows the unloading path and regains the relation correspond-
ing to monotonic loading.
When the curvature at the beginning of the unloading is smaller than the curvature correspond-
ing to the maximum cracking shear force, the difference between the unloading / reloading curve and

100
Shear Stress Distribution Prior to Shear Cracking

the curve for monotonic loading is negligible (for smaller curvatures at the beginning of unloading
there is no difference at all).

Fig. 5.8 – Cracking shear force for non-monotonic loading history: (a) history; (b) cracking shear force and
moment as function of the curvature; (c) cracking shear force-moment relation.

5.2.6 Structural System

Once both the flexural and the shear cracking behaviour of a section under a given axial force (or for
a range of axial forces, when not constant) have been determined, the shear cracking load of a system
can be determined. Fig. 5.9 shows the moment-shear force combinations for selected loads q in a
propped cantilever for a case with and without a tensile force T.
It can be seen that the M-V-combination curve scales linearly with q up to the yield load qy(1).
For higher loads, the right part of the curve (corresponding to the negative moments towards the
fixed support) translates towards higher shear forces, while the left part roughly continues to scale
with q up to the second yield load qy2 (with tensile force) or the ultimate load qu (without axial force).
In the case with a tensile force, the load can be increased further. Thereby, the right part of the curves
continues its translation, while the left part is stretched horizontally.

101
Shear Behaviour of RC Slab Strips:
Theoretical Modelling

Fig. 5.9 – Shear force in a propped cantilever: (a) system with and without tensile force; (b) M-V combinations
for selected loads. Differential equation model.

In the case without axial force (right side of Fig. 5.9) the flexural capacity (q = qu) of the sys-
tem is reached without crossing the shear cracking limit curve and a bending failure (BF) can be
expected. In the case with tension (left side of Fig. 5.9), the shear cracking limit curve is crossed at a
load slightly less than the first yield load qy1 and shear cracking (SC) is to be expected. When the
shear cracks influence the flexural behaviour of the cracked region, the form of the M-V-combination
curves changes (the curves shown for qy1, qy2 and q = 100 kN/m are based on the original behaviour).
A model that allows to estimate how the form changes is presented in Section 5.3.
Whether shear cracking is an issue or not, does not only depend on the section and the axial
force, but also on the system. On the right side of Fig. 5.9(b) the variation of the M-V-combination
curve at ultimate load due to a change of the span by 1.2 m is shown: If the span is reduced, shear
cracking is likely, if the span is enlarged the flexural capacity of the system is governing.
In the immediate vicinity of the supports the shear forces are resisted by strut action. Stress
fields are a suitable method to investigate this local behaviour. Commonly, in the absence of a de-
tailed investigation, the transverse loads within a 45° angle from the edge of the support are assumed
to be resisted directly and the shear resistance is verified at a distance d/2 from the edge of the sup-
port (d being the distance between the tension reinforcement and the compressive face of the member,
for the example section defined in Fig. 3.1 d = 135 mm).
Fig. 5.10 shows aspects of the structural response of the example system under tension deter-
mined by numerical integration. In Diagram (c) M-V-combination curves for q increasing from zero
to 125 kN/m are shown for x = 0, 135 and 350 mm, together with the corresponding cracking shear
force curves. In Diagram (d) the distributions of the shear force and the cracking shear force close to
the fixed support are illustrated for selected loads.

102
Shear Stress Distribution Prior to Shear Cracking

Fig. 5.10 – Shear force in a propped cantilever based on numerical integration: (a) system; (b) deflection,
moment and shear force distribution for selected loads; (c) M-V relations at x = 0, 135, 350 mm for
0 ≤ q ≤ 125 kN/m; (d) shear force and cracking shear force distribution close to Support B for
selected loads.

As shown in Fig. 5.10(d), the margin between the actual shear force and the cracking shear
force increases rapidly with the distance from the support at the load at which the curves meet at
x = 0 (q = 23.7 kN/m). At the formation of a hinge at the fixed support (q = 26.7 kN/m), the intersec-
tion of the two curves is located at x = 82 mm. If the governing section is located beyond this dis-
tance, no shear cracking is expected up to this load. When the load is further increased, the moment
in the hinge at the fixed support drops to a minimum (at q = 49.8 kN/m). The decrease in the moment
at the support causes a flexural unloading of the adjacent beam sector subjected to negative moments.
Flexural unloading increases the cracking shear force close to the fixed support. The shear force
initially drops together with the hinge moment, but starts to increase again before the hinge moment
reaches its minimum. When the moment in the hinge starts to increase again, it also increases the
moments in the immediate vicinity of the support (see curve for x = 135 mm in Fig. 5.10(c)) but
continues to decrease the moments further away (see curve for x = 350 mm in Fig. 5.10(c)). The
continuing decrease of the moment (and curvature) causes the sector with a reduced cracking shear

103
Shear Behaviour of RC Slab Strips:
Theoretical Modelling

force (only the reinforcing steel active in flexure and only the intermediate concrete active for the
shear flow) to move closer to the support. At the same time, the shear force is continuously increas-
ing. Consequently, shear cracking will initiate at some distance from the fixed support (in the exam-
ple at approximately x = 330 mm and q = 129 kN/m).
Considering that the drop in cracking shear force at moments close to the maximum moment is
caused by yielding of the reinforcement on the tension side of the section, it can be questioned
whether the assumption of a direct load transfer within a 45° angle is reasonable; a fan type stress
field spreading from the support requires the reinforcement on the opposite side (i.e. the tension side)
to be loaded with bond shear increasing with the distance from the support. However, when the
corresponding reinforcing steel is yielding (yield plateau for hot-rolled steel), the stress in the steel
remains constant and no additional load can be carried. As a result, the reinforcing steel would be
strained to its hardening phase, where it is capable again of resisting the increasing tension caused by
the bond shear. For reinforcing steel in the hardening phase, both the bond shear capacity and the
stiffness are reduced compared to the elastic phase. A reduced bond shear capacity increases the
length over which a given bond force is transferred and thereby increases the opening angle of the
fan and the distance from the support of the governing section. The straining of the reinforcing steel
and its reduced stiffness in the hardening phase increase the rotation at the support, which in turn
increases the slope of the beam and thus reduces the shear force. When the reinforcing steel enters
the hardening phase after the yield plateau, the cracking shear force increases.
So far, the sections were considered initially free of residual stresses and time depending
effects were disregarded. These effects do, however, influence the moment or curvature at which the
reinforcing steel on the tension side yields and as a consequence where the cracking shear force drops.

Fig. 5.11 – Comparison of shear forces and the cracking shear force in a propped cantilever based on numerical
integration and based on a differential equation model: M-V relations at x = 0, 135, 350 mm for
0 ≤ q ≤ 125 kN/m.

The combination of moment and shear force at a given distance from the fixed support
depends on the model used, as illustrated in Fig. 5.11, where the numerical integration results (black
lines) are compared to the results obtained with a differential equation model (grey lines). In statical-
ly indeterminate systems that allow for a redistribution of loads, the shear cracking resistance de-
pends on the actual load and on the loading history. Both depend on the model and consequently, the
onset of shear cracking is predicted for a different load and at a different location (it is assumed that
the governing section is sufficiently distant from the support, so that no shear cracking occurs before
the maximum moment at the fixed support is reached; in the differential equation model, the load at
the onset of shear cracking is q = 103 kN/m and the location x = 285 mm). Both models predict shear

104
Shear Stress Distribution Prior to Shear Cracking

cracking to initiate not at the support but at a distance of around 1.5 times the height of the section
and in the central part of the section.
In the absence of shear reinforcement, the shear flow redistributes to the adjacent concrete.
The diagram for q = 125 kN/m in Fig. 5.10(d) indicates that there indeed might be some spare capaci-
ty (vertical distance between the shear force curve and the cracking shear force curve) in the adjacent
sections (i.e. to the left and to the right of the point where the shear force curve touches the cracking
shear force curve). With increasing load, the crack will propagate towards the fixed support and
when the redistribution capacity is used up, the crack will suddenly propagate to the support. Consid-
ering the crack origin at Δx ≈ 3h / 2 and Δz ≈ h / 2 the crack is expected to be rather flat
(arctan(1/3) = 18°).

5.2.7 Influence of Axial Tension on the Shear Cracking Load

If the immediate vicinity of the support is not governing, shear cracking occurs at some distance from
the support, where the shear force is equal to the initial cracking shear force. As both shear force and
moment increase towards the support, the governing combination occurs at the end of the phase,
where only the steel resists flexure (see Section 3.3.1 and Fig. 5.12(a)).
While the initial cracking shear force can be estimated by Vcr = 2ebfct, the governing moment
Mtrans is given by Eq. (3.7). This moment is shown in Fig. 5.12(b) together with the moment strength
Mu for the example section. For axial forces N ≥ 2350 kN the reinforcing steel on the tension side
yields before the concrete compression zone is activated and the corresponding cracking shear force
drops, rather than increases, as discussed in Section 5.2.4.

Fig. 5.12 – Influence of axial tension on the shear cracking load: (a) cracking shear force and corresponding
moment; (b) moment strength and maximum moment of the steel only phase as afunction of the
axial tension; (c) shear cracking load-axial tension relation. Note: underlying model = differential
equation model.

Knowing the governing combination of shear force and moment, the corresponding transverse
load q can be found for a given system and axial tension. The procedure is iterative, because both the
load and the position x where the critical M-V combination occurs are unknown.
The shear cracking load qVcr is computed for the propped cantilever example, assuming
EI = Mu / χu and Mpl = Mu. The results are shown in Fig. 5.12(c) for axial tension forces
50 kN ≤ N ≤ 2350 kN. The shear cracking load qVcr increases with increasing axial tension. It can be
seen that shear cracking occurs after the central part of the beam begins to yield for axial tension
forces smaller than N = 1910 kN and before the the central part yields for higher axial tension forces.
If the beam’s stiffness is limited to at least the stiffness of the reinforcing steel only (EI ≥ EIs), shear
cracking only occurs after the central part begins to yield and at a lower transverse load (thin lines in
Fig. 5.12(c)).

105
Shear Behaviour of RC Slab Strips:
Theoretical Modelling

5.3 Consequences of a Shear Failure


The structural behaviour of a system with a member that has failed in shear can be investigated by
subdividing the member concerned into a shear-failed sector and an intact sector, as illustrated in
Fig. 5.13. Each sector can then be modelled separately, taking into account the relevant boundary and
continuity conditions. Different models can be used for the different sectors. In the following, the
behaviour is investigated based on the differential equation model (Section 4.2).

Fig. 5.13 – Shear failure in a propped cantilever under axial tension and small transverse loads:
(a) static system prior to shear failure; (b) static system, sectors and continuity conditions.

5.3.1 Behaviour of the Shear-Failed Sector

Disregarding possible stresses in the shear crack, disregarding the dowel action of the reinforcement
and neglecting the shear flow in the compression zone at the end of a shear crack a lower limit of the
residual capacity of a structure is found. Under these assumptions, the (beam) length L0 over which
the shear crack spreads can no longer bear any shear forces. In a propped cantilever under axial
tension and uniform transverse loads, the lower crack end is located at the fixed support. Consequent-
ly, V(0 < x < L0) = 0.
Considering that the shear force is the derivative of the moment and that the shear force is zero,
the moment must be constant within the shear-failed length L0. If L0 is smaller than the distance
between the section of zero moment and the support, the moment, or a portion of the moment, at the
end of the crack M(x = L0) can be maintained. Where L0 is larger than the distance between the sec-
tion of zero moment and the support, the moment ML0 within L0 is zero. As for the case with a con-
stant plastic moment in the span (Eq. (4.25)), the deflection and slope can be derived based on the
moment equilibrium condition M(x) = ML0 = M0(x) - Tw(x) and the behaviour of the beam within L0 is
described by:

x  q ( L − x ) ML0 
w ( x, M = M L 0 ) =  +  where − M pl ≤ M L 0 ≤ 0
T 2 L 
q ( L 2 − x ) + ML0 L
w′ ( = L0 )
x, M M=
T
(5.11)
M ( x) =
ML0 , χ ( x ) =
−q T

V ( x) = 0

When ML0 is zero, the behaviour corresponds to the behaviour of a funicular polygon.

106
Consequences of a Shear Failure

5.3.2 Behaviour of the Intact Sector

For the intact sector of the beam, two cases have to be distinguished: If the transverse loads are small
enough for the entire sector to remain elastic, the sector is defined by one single differential equation.
The coefficients of its solution (Eq. (4.16)) can be determined according to Fig. 4.5 and Eq. (4.23).
For ML0 = 0 the coefficients are:

q −q  L − L0 
=c1 =c2 tanh  κ 
κT2
κT
2
 2 
(5.12)
q  L20  q  L0 1 
c=
3  L − L 0+
2T  L 
 c=
4 
T 2
(
L − L0 − 2 
κ 
)
 

The maximum moment occurs at x = (L + L0) / 2 and amounts to:

q   L − L 0 
M max =2 1 − cosh −1  κ
2  
(5.13)
κ  

If parts of the intact sector are yielding, the sector has to be subdivided as illustrated in
Fig. 5.14 and each sector solved individually, taking into account the respective boundary and/or
continuity conditions. If the beam was yielding in the span prior to the shear failure (Fig. 5.14(a)), the
length of the sector at the simple support (Sector 3) remains unaffected by the shear failure at the
fixed support. In this case, the influence of the shear failure on the moment and shear force is negli-
gible as can be seen in Fig. 5.15. If additionally ML0 = 0, the first sector of the intact part takes on the
same length as the third sector. In other cases, the lengths of the sectors have to be determined itera-
tively, so that the maximum moments in Sectors 1 and 3 occur at the transition to Sector 2.

Fig. 5.14 – Shear failure in a propped cantilever under axial tension and high transverse loads:
(a) static system and sectors prior to shear failure; (b) static system and sectors after shear failure.

107
Shear Behaviour of RC Slab Strips:
Theoretical Modelling

The behaviour of the yielding sector (Sector 2) is described replacing Mpl by ML0 (except for
the terms ξMpl that remain unaltered) in Eq. (4.25). For ML0 = 0 the behaviour in the yielding sector is
described by:

qx ( L − x ) 2 − ξM pl
w( x) =
T
q
w′ ( x )
= ( L 2 − x)
T
(5.14)
M ( x) =
ξM pl , χ ( x ) =
q T

V ( x) = 0

The validity of the differential equation solution presented above is limited by the extent of the
yielding sector (Sector 2 in Fig. 5.14): the sum of the lengths of the shear failure sector and the first
sector of the intact part of the beam must not be larger than the length of the first sector prior to shear
failure. In cases where the length of the first sector prior to failure is larger, the beam would unload
from the plastic state which, is not covered by the equations presented.
The loading state prior to shear failure does not have an influence on the state after shear fail-
ure when L0 is sufficiently small: Consequently, load-deflection, -moment and -shear relations can be
computed for the two states independently. A shear failure then represents a jump from one relation
to the other. Due to the problem of unloading, the jump is only exact in load-controlled systems.
Fig. 5.15 illustrates the influence of a shear failure on the structural response of a propped cantilever.
The distributions in Fig. 5.15(b) show an influence of shear failure primarily between the fixed
support and the beginning of the yielding sector of the beam. The more the length of the shear-failed
sector approaches the distance between support and the point of zero moment prior to the shear
failure, the smaller the difference in the moment and shear force distributions in the intact part of the
beam. The shear force in the intact part adjacent to the shear-failed sector is reduced by the shear
failure. However, as the shear crack typically originates in the zone, where only the reinforcement is
active and the cracking shear force is constant for smaller curvatures, the governing cracking shear
force will remain the same. Consequently, new shear cracks are likely to form and the length of the
shear-failed sector increases when the load q is increased.
The load-moment relations (Fig. 5.15(c)) show that a shear failure has virtually no effect on
the moment at midspan. Obviously, the effect is larger on the moment at the fixed support, where the
moment drops to zero. Similarly, the load-shear force relations (Fig. 5.15(d)) show that a shear
failure has virtually no effect on the shear force at the simple support, despite the shear force at the
fixed support being reduced to zero.
A shear failure softens the load-deflection behaviour ((Fig. 5.15(e)). At higher loads, the tan-
gential stiffness of the system remains virtually identical and even the system with shear failure is
significantly stiffer than the funicular polygon. Taking into account a possible moment in the shear-
failed sector results in relations between the shown curves.
A comparison of Fig. 5.15 and Fig. 4.6 shows that a shear failure has a similar effect as the
softening of the hinge at the fixed support.

108
Consequences of a Shear Failure

Fig. 5.15 – Shear failure in a propped cantilever under axial tension: (a) static system and definitions;
(b) deformation and stress resultants; (c) q-M relation; (d) q-V relation; (e) load-deformation
relation. Note: underlying model = differential equation model.

109
Shear Behaviour of RC Slab Strips:
Theoretical Modelling

5.4 Summary
Because of the interconnection of moment and shear it is essential to model the flexural behaviour of
a structure adequately when the shear capacity is a concern.
The shear resistance of reinforced concrete members has been the subject of extensive research
for more than a century and yet to this day there is no mechanical model capable of predicting the
shear resistance accurately over a wide range of cross-sections and stress resultants for members
without shear reinforcement.
The formation of a shear crack is not equal to failure; different mechanisms enable a force
transfer in and around a crack even in the absence of shear reinforcement: (i) force flow in the
uncracked part of the member (usually the compression zone), (ii) aggregate interlock in the crack
and (iii) dowel action of the ordinary reinforcement. Shear failure, however, is always accompanied
by shear cracks. A crack initiates when the maximum principal stress reaches the concrete tensile
strength. The crack expands perpendicularly to the direction of this principal stress. Prior to cracking,
the concrete behaviour is approximately linearly elastic. Assuming that flexural cracks only prevent
axial tensile stresses, but do not prevent the transfer of shear stresses (due to aggregate interlock), the
shear stress distribution prior to shear cracking can be found using the elastic shear flow theory on a
transformed section. The cracking shear force for a given section depends on the moment and axial
force applied. For a symmetric section under axial tension, the cracking shear force is constant at a
value around Vcr = 2ebfct for moments that are resisted by the reinforcing steel only. The cracking
shear force increases rapidly as the concrete compression zone builds up and then drops below the
initial value when the reinforcing steel on the tension side starts to yield. If the drop coincides with a
local moment minimum, the drop is not governing for load controlled systems. The cracking shear
force depends on the loading history and can be sensitive to residual strains. Furthermore, the crack-
ing shear force can be limited by the bond shear capacity for low longitudinal reinforcing ratios. The
model presented assumes that the full bond shear stress capacity can be mobilised (pull-out type
failure). In cases where splitting failure occurs, the bond shear stress capacity is significantly reduced.
Depending on the longitudinal reinforcing ratio, this reduction also reduces the cracking shear force.
In a propped cantilever under axial tension, shear cracking can initiate in the central part of the
section at a distance from the fixed support. In the example shown, the corresponding load increases
with the axial tension in the range of small to moderate axial tension. Cracking will not necessarily
immediately result in shear failure. When failure occurs, the crack has expanded to the fixed support
and the static system has changed. In a propped cantilever without axial force, shear failure is equal
to system failure. In the presence of an axial tension, the deflections will increase and the system will
continue to resist increasing loads.

110
Test Set-Up and Test Procedure

6 Behaviour of RC Slab Strips:


Experimental Investigations

In 2012 and 2013 two series of large-scale tests on reinforced concrete slab strips were carried out at
the Institute of Structural Engineering (IBK) of the Swiss Federal Institute of Technology (ETH) in
Zürich in order to gain insights into the structural behaviour of multi-span slab strips under axial
tension and transverse loading. A total of 12 slab strips with different reinforcing ratios and arrange-
ments and under different tensile forces were tested. In the following sections, the test set-up, the test
procedure, the test specimens and the main results are described. For a detailed description and
additional information, the reader is referred to the comprehensive test reports (Galmarini et al. 2013;
Locher et al. 2014). Further, the test data is compared to the results of calculations based on the
models described in the previous chapters. Finally, the main conclusions are summarised.

6.1 Test Set-Up and Test Procedure


A simply supported beam with a cantilever was chosen as the basic static system for the tests, see
Fig. 6.1(a). The system is statically determinate, hence the stress resultants are unambiguously
defined by the loads and the deflections. Furthermore, the system allows for a redistribution of
moments. A uniform load in the span was simulated by three point loads Q, and the moment at
Support B was controlled by the application of a point load P at the end of the cantilever (C). For
practical reasons, in the actual test set-up a rod was added to the basic system to the right side of C,
and the tensile force T was applied at Support A, see Fig. 6.1(b).

Fig. 6.1 – Basic static system: (a) conceptual system; (b) actual system used for tests.

The actual test set-up consisted of a reaction structure, the test specimen, different hydraulic
systems with jacks for applying the loads and various measuring equipment. An overview of the
reaction structure with the test specimen and the jacks is shown in Fig. 6.2 together with selected
details of the test specimen-reaction structure connections.
The test specimens were suspended from the reaction structure at Supports A and B and con-
nected to shear walls by means of a double pin-jointed rod in Axis C and a double pin-jointed 2.5
MN hydraulic jack at Support A. At supports A and C where multiple elements were joined, a cus-
tom-made hinge with a single common bolt was used in order to avoid eccentricities both initially
and during the tests, where significant deformation of the test specimens occurred. At the suspended
supports at A and B multiple pin joints were used to allow free horizontal movement of the test
specimen without the introduction of secondary moments. For safety reasons and to avoid measure-

111
Behaviour of RC Slab Strips:
Experimental Investigations

ment results marred by the deformation of the reaction structure, the steel reaction structure was
designed to be very stiff and all bolted connections were fully prestressed.
The transverse loads Q in the span were applied by loading yokes as line loads over the width
of the specimen. The loading yokes were pulled towards the strong floor by two hydraulic jacks each.
The line load from the yoke was distributed by a 100 mm wide steel plate fixed to the top of the test
specimens.

Fig. 6.2 – Test set-up: (a) general arrangement; (b) plan view; (c) test specimen; (d) detail of Support B;
(e) connection at the ends of the test specimen. Dimensions in mm.

112
Test Specimens

Between the first and second test series, the pin joints at A and C were fitted with an additional
polymer sliding surface between bolt and outermost clevis brackets in order to further reduce friction.
Three independent hydraulic systems were used: one for the tensile force T, one for the trans-
verse load P and one for the transverse loads Q. The six jacks for the loads Q (two jacks per load Q)
were coupled to ensure that they were all working with the same pressure, as were the two jacks for
load P. The pressure in the hydraulic systems for the loads P and Q was controlled by manually
operated pumps. A motor-driven pump, combined with a spring valve, allowed for the pressure in the
T-jack to be gradually increased to a desired level at which the pressure automatically was kept
constant for the remainder of the test.
In order to increase the total stroke at the loading yokes beyond that of the hydraulic jacks, a
set of additional hollow piston jacks was fitted under the strong floor. These jacks operated individu-
ally or in pairs by means of a separate hand pump. When operated, these jacks pulled out the piston
rod of the corresponding jacks at the loading yoke, while increasing the loads Q. By repeated use, the
total stroke could be increased to over 700 mm.
Continuous measurements by means of extensometers, load and pressure cells were supple-
mented by manual measurements of the deflection line and average strains on the surface of selected
parts of the test specimen. The load increase was interrupted at pre-defined intervals to take these
manual measurements. Besides manual measurements characteristic crack data (location, length and
width) was collected, and the cracks were highlighted and photographed.
The tests typically lasted two days. On the first day, reference measurements were taken. Then
the tensile force T was applied in a first load step and maintained constant, while the vertical loads
were increased stepwise. The loads in the span (Q) and at the cantilevering end (P) were increased
proportionally to a level where both steel and (cracked) concrete were still in the elastic phase. The
loads were kept constant overnight. On the second day the loads P and Q were further increased,
until the formation of a plastic hinge was observed at Support B. Then, the hydraulic circuit for
load P was closed so that the deflection at C remained constant, and only load Q increased further.
Load Q was increased in steps until failure either in shear or flexure was observed. After a series of
manual measurements, the load was increased until either the forces in the steel rods at Support B or
the midspan deflection reached their limiting values. In the second test series, some tests were
stopped after specimen failure due to time constraints.
In the tests, the tensile force T was chosen so that the maximum steel stress caused by the ten-
sile force was approximately half the steel tensile strength, respectively σs = 0, 150 or 300 MPa.

6.2 Test Specimens


The test specimens consisted of slab strips 0.6 m wide, 0.2 m high and 8.1 m long. The main rein-
forcement consisted of six continuous longitudinal top and bottom reinforcing bars. The longitudinal
bars protruded from the concrete and had threaded ends to connect to the reaction structure and to the
jack for the tensile force (Fig. 6.2(e)). In the first test series, all longitudinal bars had a diameter of
26 mm and were located 35 mm above and below the centre plane, see Fig. 6.3(a). In the second test
series, the longitudinal bars were located 67 mm above and below the centre plane and had varying
diameters (26, 20 and 14 mm), see Fig. 6.3(b). The longitudinal reinforcement was complemented by
stirrups to prevent failure of the specimen during transport or due to the local force flow at the sup-
ports and load application points, see Fig. 6.3(c). In the first test series, one specimen contained
additional transverse reinforcing bars (T3 in Fig. 6.3(a)), and two specimens contained stirrups at
different intervals as shear reinforcement (T4 and T5 in Fig. 6.3(a)). An overview of the reinforce-
ment arrangement of the test specimens is given in Tab. 6.1.
Cast-in lifting lugs were used for transportation of the specimens within the laboratory by an
overhead crane.

113
Behaviour of RC Slab Strips:
Experimental Investigations

Careful curing and handling resulted in close to uncracked test specimens at the beginning of
the test.

Fig. 6.3 – Test specimens – reinforcement: (a) cross-sections of the first test series; (b) cross-sections of the
second test series; (c) basic arrangement. Dimensions in mm.

Within each test series, all reinforcement bars of the same diameter were produced in the same
lot. The test specimens of each series were cast at the same time using the same batch of concrete.
The test specimens were cast inside the laboratory where they were also cured and prepared for the
tests. The specimens were cast at least three months before testing, so that the time-dependent varia-
tion of the concrete properties was negligible (as confirmed by material tests). Tab. 6.2 shows a
summary of the main properties of the construction materials.

Tab. 6.1 – Test specimens – overview.


Test Diameter Eccentricity Transverse Stirrups Weight Age at Tensile force
longitudinal longitudinal reinforcement beginning
reinforcement reinforcement G of test σs(T) T.
[mm] [mm] [kN] [d] [MPa] [kN]
T0 26 35 none local 26.8 98 0 0
T1 26 35 none local 26.0 128 300 1911
T2 26 35 none local 26.0 138 150 956
T3 26 35 Ø22, s = 200 mm local 26.7 153 300 1911
T4 26 35 none s = 100 mm 26.6 174 300 1911
T5 26 35 none s = 200 mm 26.7 183 300 1911
T6 26 67 none local 27.0 120 300 1911
T7 20 67 none local 24.6 127 300 1131
T8 20 67 none local 25.4 134 150 565
T9 14 67 none local 24.6 149 300 554
T10 14 67 none local 24.2 156 150 277
T11 14 67 none local 24.9 104 0 0

114
Test Results

Tab. 6.2 – Test specimens – material properties.


Material Property Unit 1st test series 2nd test series
Concrete Cylinder compressive strength fcc [N/mm2] 41.2 40.3
Splitting tensile strength fcts [N/mm2] 03.2 03.1
Modulus of elasticity Ec [kN/mm2] 32.4 33.8
Crushing strain εcu [‰] 01.8 01.6
Longitudinal Dynamic yield strength fsy,dyn [N/mm2] 527 496
reinforcement Dynamic tensile strength fsu,dyn [N/mm2] 632 603
Ø26 Strain at peak stress Agt [‰] 140 123
Longitudinal Dynamic yield strength fsy,dyn [N/mm2] 507
reinforcement Dynamic tensile strength fsu,dyn [N/mm2] 605
Ø20 Strain at peak stress Agt [‰] 119
Longitudinal Dynamic yield strength fsy,dyn [N/mm2] 536
reinforcement Dynamic tensile strength fsu,dyn [N/mm2] 643
Ø14 Strain at peak stress Agt [‰] 114
Stirrup Dynamic yield strength fsy,dyn [N/mm2] 517
reinforcement Dynamic tensile strength fsu,dyn [N/mm2] 585
Ø8 Strain at peak stress Agt [‰] 69

6.3 Test Results


The load-deflection characteristics of all tests are illustrated in Fig. 6.4.

Fig. 6.4 – Selected test results: (a) Q-w2; (b) P-wC.

115
Behaviour of RC Slab Strips:
Experimental Investigations

The key results of the two test series are summarized in Tab. 6.3.

Tab. 6.3 – Summary of main test results.


Test Failure / Max Load Deflection Stress resultants
Type Place T Q P w2 wC MB M2,0 M2 VB,0– N3 - B V3 - B M6.15 M7.05
[kN] [kN] [kN] [mm] [mm] [kNm] [kNm] [kNm] [kN] [kN] [kN] [kNm] [kNm]
T0 FS B+ 0 64 103 142 27 -189 125 125 -1381) - - - -
S 2 0 67 56 230 31 -110 172 172 -130 - - - -
T1 MB,max 1911 122 145 148 31 -150 317 34 -218 - - - -
S 3 - B 1913 149 136 188 32 -130 407 48 -256 1928 -103 -12 -
max 1916 165 115 223 32 -94 476 49 -275 - - - -
T2 MB,max 956 68 13 97 43 -164 149 56 -139 - - - -
S 3 - B 949 146 111 316 42 -127 399 100 -250 983 -134 30 -
T3 MB,max 1912 110 144 131 42 -106 306 55 -195 - - - -
S 3 - B 1915 170 137 222 41 -98 488 63 -283 1930 -108 25 -
max 1917 253 113 357 39 -61 756 73 -403 1983 -43 17 -
T4 MB,max 1911 242 147 336 35 -132 688 45 -397 - - - -
max 1912 253 139 351 35 -119 727 55 -411 1965 -139 22 -
T5 MB,max 1911 96 148 113 39 -123 253 37 -175 - - - -
max 1909 253 133 355 38 -100 737 59 -409 1968 -143 32 -
T6 MB,max 1913 69 170 58 47 -132 169 58 -137 - - - -
S 3 - B 1912 175 155 204 44 -114 497 107 -294 1933 -154 58 -
max 1919 249 137 319 43 -86 731 119 -400 1960 -143 44 -
T7 MB,max 1136 170 91 372 43 -75 499 76 -280 - - - -
S 3 - B 1138 192 89 426 43 -72 566 81 -312 1186 -119 50 -52
T8 MB,max 571 142 87 494 45 -114 396 114 -243 - - - -
S 3 - B 574 147 87 507 45 -114 410 119 -250 610 -149 49 -87
T9 MB,max 557 90 56 388 45 -62 264 48 -157 - - - -
F B 561 107 53 473 45 -57 320 54 -183 593 -76 25 -43
T10 MB,max 280 60 50 374 45 -76 169 64 -115 - - - -
F B 283 77 46 531 45 -69 221 71 -138 303 -88 23 -56
max 282 79 0 685 48 11 269 76 -131 314 -46 58 18
T11 max 2 36 46 245 25 -93 87 86 -80 9 -72 -10 -75
FS B+ 6 34 40 516 25 -83 86 82 -76 12 -75 -8 -76
MB,max 4 27 55 574 50 -107 55 53 -70 - - - -
Failure types: F: flexural failure; S: shear failure; FS combined flexural and shear failure.
1)
VB+ = 115 kN

Test specimen T5 was left in its position and attached to the reaction structure after the end of
the first test series (test T5). The measuring equipment was dismantled. Prior to the second test series,
the measuring equipment was reinstalled with some modifications. A reduced test with specimen T5
was carried out to confirm correct operation of the equipment. Under a constant axial tension of
T = 956 kN, the specimen was loaded transversely with loads up to Q = 180 kN and a maximum
deflection of w2 = 407 mm. Only few manual measurements were taken and the crack characteristics
were not documented. During the test various sensors had to be replaced, so that no consistent set of
data was available in the end. Furthermore, the initial conditions were significantly different from the
regular tests, which is why it was decided not to include the data in the test report. However, the test
still demonstrated that minimal shear reinforcement was sufficient to prevent shear failure under the
loads tested.

116
Test Results

The key test results can be summarized as follows:


- The specimens without stirrups and with moderate to high longitudinal reinforcing ratios,
which were loaded with an axial tensile force (T1, T2, T3, T6, T7, T8) failed in shear in the
region 3 - B. The failure generally was brittle without warning; however, shear cracks formed
prior to the failure (except for T6). The specimen deformations and the widths of the flexural
cracks on the bottom face of the specimen between the supports and on the top face of the
specimen over Support B were very large before failure.
- The load decrease following the abovementioned shear failures mentioned above was less than
15 % in the first test series and less than 25 % in the second test series. Subsequently it was
possible to increase the load beyond the load at shear failure (due to time constraints, Speci-
mens T7 and T8 were not loaded further after shear failure). The shear failures did not limit the
system resistance.
- The starting point of the shear failure zones located between Yoke 3 and Support B varied, and
there was no direct relation to the local flexural moment. Failure initiated between the two
longitudinal reinforcement layers.
- The test specimens with stirrups (T4, T5) did not fail in shear under the applied loads (which
were limited by the capacity of the test set-up).
- The specimens with a low longitudinal reinforcement ratio and loaded with an axial tensile
force (T9 and T10) failed due to rupture of the top longitudinal reinforcement bars above Sup-
port B. This failure was caused by large plastic deformations. The shear capacity of these spec-
imens was not reached.
- For specimens with the same longitudinal reinforcement ratio, a larger axial tensile force
caused shear failure to occur at a lower shear force but at a higher vertical load Q.
- The global load-deformation behaviour of the test specimens with identical longitudinal rein-
forcement arrangement and loaded with the same axial force (T1, T3, T4 and T5) was almost
identical. In specimens that experienced shear failure, the deformations did increase with re-
spect to the specimens that did not experience shear failure. However, as the load was in-
creased further the deformation behaviour of all test specimens started to align again. Shear
failure caused a local increase in the slope of the specimen deflection in the vicinity of the fail-
ure location.
- Under large loads and deformations, plastic hinges formed over Support B and under the three
loading yokes, provided the specimens did not fail beforehand. In these cases the global load-
bearing behaviour corresponded to a funicular polygon. This behaviour was observed at least
partially for all specimens loaded with an axial tensile force and was more pronounced at larg-
er axial tensile forces.
- At low load levels, the global load-deformation behaviour of the test specimens with the same
longitudinal reinforcement ratio was practically independent of the magnitude of the axial
force. Only at larger load levels and after the plastic moment Mpl was reached at Support B, the
specimens behaved more stiffly at higher axial tensile forces.
- The specimens exhibited increasingly stiffer load-deformation behaviour as the axial tensile
force and the longitudinal reinforcement ration increased. A larger eccentricity of the rein-
forcement under otherwise equal conditions also resulted in larger specimen stiffness.
- Increasing axial tensile forces and deflections resulted in smaller the shear forces and span
moments in the specimens.
- The specimens subjected to an axial force showed ductile global behaviour and reached very
large deformations (deflections between 1/38 and 1/13 of the span at failure). The deflections
at failure tended to increase with decreasing axial tensile force.

117
Behaviour of RC Slab Strips:
Experimental Investigations

6.4 Predictions
Because the test set-up is statically determinate, the distribution of the stress resultants with respect to
the undeformed reference system (M0, V0, T) can be calculated for any given combination of loads.
Fig. 6.5 shows the loads (Figure (b)) and the corresponding moment components (Figure (d)).
The deflection line (Fig. 6.5(c)) is required to calculate the stress resultants acting on the beam
(M, V, N), as illustrated for the moment distribution in Fig. 6.5(e). In the experiment, the deflection
line can be measured. For the prediction, the deflection line is estimated using a model. The models
presented in Chapter 4 have to be adapted to the actual static system and loading of the experiment.

Fig. 6.5 – Predictions: (a) test set-up; (b) static system and loading; (c) deflection line; (d) moment compo-
nents; (e) moment distribution. Dimensions in mm.

All models require some parameters describing the section behaviour. The estimated detailed
section behaviour of the test specimens is illustrated in Fig. 6.6. The parameters of the simplified
section behaviour characteristics used for the rigid-perfectly plastic models and for the differential
equation models are listed in Tab. 6.4.

118
Predictions

Fig. 6.6 – Section behaviour: (a) moment-axial force strength envelope; (b) axial force-curvature correspond-
ing to flexural strength; (c) and (d) moment-curvature relations for selected axial forces.
Bold lines: dynamic steel properties; thin lines: static steel properties.

Fig. 6.6(a) shows that the choice of the steel properties (static or dynamic) can increase the
section strength significantly for moderate compression and for tension. The initial stiffness, however,
is not affected significantly by the choice of static or dynamic properties (Fig. 6.6(c) and (d)). For the
predictions, the static properties are used, unless noted otherwise.

Tab. 6.4 – Test specimens – static section behaviour characteristics.


Series 1st test series 2nd test series
Specimen T0 T2 ,T1, T3, T6 T7 T8 T9 T10 T11
T4, T5
Mpl [kNm] 150.5 119 80.2 123 72.5 109 39.3 57.4 74.4
EI [kNm2] 4800 3750 2400 5500 3300 4600 1650 2550 3350

119
Behaviour of RC Slab Strips:
Experimental Investigations

The predictions are based on theoretical loading sequences and load values. The transverse
load at the end of the cantilever is defined as:

wc
P = α + βQ + T [kN], [mm] (6.1)
2000

The coefficients α and β are listed in Tab. 6.5, together with the load Q, at which the jacks at C
were blocked and the deflection wC was kept constant.

Tab. 6.5 – Predictions – load parameters.


Series 1st test series 2nd test series
Test T0 T2 ,T1, T2, T6 T7 T8 T9 T10 T11
T4, T5
α [kN] 5.58 5.58 5.58 7.0 7.0 7.0 7.0 7.0 7.0
β [-] 1.44 1.44 1.44 1.686 1.5 1.567 1.676 1.684 1.582
Q(wC = const) [kN] 64 68 63 69 40 45 17 18 22

6.4.1 Differential Equation Model

As illustrated in Fig. 6.7, the differential equation model consists of 5 sectors, with the first one
starting at the steel/concrete interface towards A and the last ending at the steel/concrete interface
towards C. The sector limits are given by the positions of the point loads Q and of support B.

Fig. 6.7 – Predictions: (a) initial static system and loads; (b) sectors; (c) static systems at higher loads.

The point loads are taken into account in the transition conditions. Prior to the formation of a
plastic hinge, the transition conditions at the axes 1, 2 and 3 are wl = wr, Ml = Mr, wʹl = wʹr and

120
Predictions

Vl – Q = Vr. After the formation of a plastic hinge at the transition in question the conditions change
to wl = wr, Ml = Mr, Mr = Mpl and Vl + Twʹl – Q = Vr + Twʹr. Note that two of the four conditions
remain unaffected by the formation of the hinge. At Support B, the transition conditions prior to the
formation of a plastic hinge are wl = 0, wr = 0, Ml = Mr, and wʹl = wʹr. With a plastic hinge at Support
B, the condition regarding the slope is substituted by Mr = –Mpl. The sequence of plastic hinge for-
mation is shown in Fig. 6.7(c).
True to the original intent of the loading sequence, the deflection at C is fixed as the plastic
hinge at B forms. This reduces the number of different cases but also the accuracy of the load-
deflection relation of the end of the cantilever.
Shear cracking is assumed to occur when the shear force in the central part of Sector 4 (be-
tween Axes 3 and B) reaches the initial cracking shear force. This corresponds to the state, in which
flexure is resisted by the reinforcement only; this is the case for T2 in Fig. 6.8. Further, shear crack-
ing is also assumed to occur when the shear force at a distance of 180 mm from Axis B, and prior to
the formation of a plastic hinge at B, is larger than the minimum cracking shear force. This corre-
sponds to the state in which only the steel is resisting flexure, and the reinforcement on the tension
side is yielding; this is the case for T6, T7 and T8 in Fig. 6.8. The onset of shear cracking is assumed
to result in shear failure.
The behaviour following shear failure is estimated using the same principles as those used for
the intact beam. The only difference is that the plastic moment for the hinges at 3 and B is set to zero.
The effect of shear failure on the deflection line can be seen in the last three diagrams of Fig. 6.7(c).
Fig. 6.8 shows the predictions for the tests with axial tension. For each test, the upper line indi-
cates the behaviour without shear failure and the lower line the behaviour after shear failure. Bold
lines designate the predicted behaviour including the transition from one to the other behaviour.

Fig. 6.8 – Predictions with differential equation model: (a) Q-w2; (b) P-wC.

The cracking shear force depends linearly on the concrete tensile strength. The tensile strength
is the concrete property with the largest spread. The coefficient of variation of the experimental

121
Behaviour of RC Slab Strips:
Experimental Investigations

values is around 10%. A reduction of Vcr by 10% results in a reduction of the shear cracking load QVcr
by far more than 10% (thin horizontal lines for T1 and T2 in Fig. 6.8(a)). When the relevant shear
forces at the formation of a hinge at Yoke 3 are only slightly smaller than the cracking shear force, a
reduction of the cracking shear force will change the prediction from no shear cracking at all to shear
cracking for a certain load.
The shear cracking load QVcr depends on whether the static or dynamic properties of the rein-
forcing steel are used: using the dynamic steel properties, the plastic moment at Support B increases
and consequently, the shear forces in Sector 4 (3 to B) also increase. As the reinforcing steel proper-
ties do not affect the initial cracking shear force Vcr, the shear cracking load QVcr is reduced.

6.4.2 Rigid-Perfectly Plastic Model

The load-deflection relations resulting from a rigid-perfectly plastic model are shown in Fig. 6.9. The
plastic moments are taken as equal to the flexural strength of the corresponding section (see Tab. 6.4).
The dead load is modelled as concentrated loads in the axes A, 1, 2, 3, B and C.

Fig. 6.9 – Predictions with rigid-perfectly plastic model: (a) Q-w2; (b) P-wC.

Shear failure is assumed to occur when the shear force equals the shear resistance estimated by
an upper bound solution based on a mechanism suggested by Nielsen and Hoang (2011):

 1+ a h 2 − a h f sy , stat a 
 ( )
=Vu + ρz =bhf c where f c 0.6 f cc (6.2)
 2 f c h 
 

The parameter a = 1.05 m represents the clearance between the loading plate of Yoke 3 and the
bearing plate of Support B. ρz is the geometric stirrup reinforcement ratio. It is important to note that

122
Comparison

this mechanism-based shear resistance model takes into account a certain length of the beam. This is
markedly different form the section-based models used in this document.

6.5 Comparison
The comparisons presented here follow the workflow of the model prediction; material behaviour,
section behaviour and behaviour of the entire system.

6.5.1 Material

In Fig. 6.10 the experimental stress-strain curves are overlaid with the material models used for the
predictions. For the concrete, the compression chord model presented in Section 2.2.6 is used, and for
the reinforcing steel, the model for hot-rolled steel presented in Section 2.3.4 is used.

Fig. 6.10 – Comparison of material behaviour: (a) concrete, 1st test series; (b) concrete, 2nd test series;
(c) reinforcing steel Ø26, 1st test series; (d) reinforcing steel Ø26, 2nd test series; (e) reinforcing steel
Ø20, 2nd test series; (f) reinforcing steel Ø14, 2nd test series.

It can be seen that the material models generally agree well with the experimental curves, ex-
cept for the magnitude of the residual stress in the concrete after crushing, which is neglected in the
model used. How well the steel model with static properties (shown only in Fig. 6.10(c), thin line)
compares to the real behaviour can only be verified in discrete points. Nevertheless, it seems plausi-
ble that the model with static properties would not fit much worse than the one with dynamic proper-
ties in the remainder of the curves.

123
Behaviour of RC Slab Strips:
Experimental Investigations

6.5.2 Section Behaviour

The predicted moment-curvature relations are compared to the experimental values of the first and
second test series in Fig. 6.11 and Fig. 6.12, respectively.
The experimental values are derived from the continuous force measurements, the manual
deflection measurements (M = M0 – Tw) and from the manually measured strains at the top and
bottom surfaces of the test specimens (χ = Δε / Δz). While the moments are determined from meas-
urements at the same position, the curvatures are determined from mean strains over 300 mm and
then linearly interpolated for the value at the corresponding position of the moment values. As the
curvature gradient close to axes 1, 2, 3 and B is large at higher loads Q and as the measured mean
strains are influenced by the discrete location of the cracks, a certain spread of the experimental
values is unavoidable. Experimental values corresponding to unloading states (absolute value of a
moment for a load step smaller than the absolute value of the corresponding moment in the previous
load step) are not plotted. Test T1 is not included in the comparison due to an uncertainty of the
load P caused by an omission during testing, see Galmarini et al. (2013).
In the first test series, the initial distance between the measurement bolts was recorded after the
test specimen had been placed in its position for testing and had deformed under its own weight. In
the second test series, the initial distance was measured when the test specimen was supported at
several positions and virtually undeformed. Also, the height of the test specimens corresponded
better to the theoretical value and varied less in the second test series. Therefore, the spread of the
experimental values of the first series is larger than that for the second test series.

Fig. 6.11 – Moment-curvature relations and measurements, 1st test series: (a) T3, T4 and T5; (b) T0; (c) T2.

Because the range of the measuring devices was limited and because the mean strains could
not be measured across the steel plates at Support B, there are no experimental values for large
curvatures.

124
Comparison

Fig. 6.12 – Moment-curvature relations and measurements, 2nd test series: (a) T6; (b) T8; (c) T7; (d) T11;
(e) T10; (f) T9.

The predicted moment-curvature relations generally compare well with the experimental val-
ues. The difference due to using either static or dynamic steel properties is smaller than the general
spread of the results.
Some experimental values are significantly higher than the predictions with dynamic steel
properties, which indicates that the actual position of the reinforcement did not correspond exactly to
its theoretical position and/or that the concrete dimensions were larger than the theoretical values.
The latter was confirmed by measurements of the test specimens prior to testing: thicknesses up to
208 mm were measured.

6.5.3 System Behaviour

The load deflection characteristics determined by the differential equation model and by the rigid-
perfectly plastic model are compared to the measured load-deflection behaviour in Fig. 6.13 (1st test
series) and Fig. 6.14 (2nd test series).
Figs. 6.15 and 6.16 show a more detailed comparison of the results of the differential equation
model with those of tests T3 and T5. These two tests are selected, because the effect of shear failure
can be seen clearly in the difference of the results and because they indicate the possible variation in
results for close to identical specimens.

125
Behaviour of RC Slab Strips:
Experimental Investigations

Fig. 6.13 – Comparison of system behaviour, 1st test series: (a) Q-w2; (b) P-wC.

Fig. 6.14 – Comparison of system behaviour, 2nd test series: (a) Q-w2; (b) P-wC.

126
Comparison

Fig. 6.15 – Comparison of system behaviour, tests T3 and T5: (a) Q-w; (b) Q-M.

127
Behaviour of RC Slab Strips:
Experimental Investigations

Fig. 6.16 – Comparison of system behaviour, tests T3 and T5: (a) deflection lines; (b) moment distribution;
(c) shear force distribution.

128
Comparison

The load-deflection curves for the specimen midspan (Fig. 6.13(a)), determined by the differ-
ential equation model, generally agree well with the experimental data. For the tests of the second
test series with smaller-diameter longitudinal reinforcing bars, the specimens are predicted to behave
less stiffly than was the case in the actual tests. The influence of a shear failure is predicted well for
the tests with 26 mm reinforcing bars but overestimated for the tests with 20 mm reinforcing bars.
The predictions of the behaviour at the end of the cantilever (P-wC, Fig. 6.13(b)) correspond reasona-
bly well with the experimental data. The prediction does, however, underestimate the maximum
values of both load and deflection (because of a reduced number of cases included in the model, see
Section 6.4.1). The tests with shear failure are correctly predicted. The load at shear failure is pre-
dicted reasonably well for the first test series but grossly underestimated for the second test series,
see also Tab. 6.6.
The detailed comparisons for tests T3 and T5 (Fig. 6.15 and Fig. 6.16) confirm that the overall
behaviour can be well described with the differential equation model. The comparison of the load-
moment relations (Fig. 6.15(b)) shows that the plastic moment at Support B is underestimated signif-
icantly, while the plastic moment in the span is overestimated. The moment distribution (Fig. 6.16(b))
differs correspondingly. As the moment gradient between Axis 3 and B is the same, the shear force
distribution (Fig. 6.16(c)) in this sector correspond closely nevertheless. The detailed comparison
also confirms that the influence of a shear failure is overestimated.

Tab. 6.6 – Comparison of shear failure loads.


Model 1st test series 2nd test series
Test T2 T1 T3 T4, T5 T6 T7 T8 T9 T10
Vcr [kN] 139 55 47 253 45 222
Vu [kN] 133 > 435 130
Vmax1) [kN] 198 134 205 121 182 65.5 95.7
QVcr [kN] DEQ 129 -/ 1872) - 33 20 - - -
QVu [kN] DEQ 117 177 - 115 - 122.5 - -
QVu [kN] RPP 90 123 - 83 -/ 1183) 95 - -
QSF [kN] test 146 149 170 - 175 192 147 - -
Models: DEQ: differential equation model; RPP: rigid-perfectly plastic model.
1)
Vmax = 2Mpl / 1.2 m.
2)
load for 90% Vcr.
3)
dynamic reinforcing steel properties.

The load-deflection curves for the specimen midspan (Fig. 6.14(a)), determined by the rigid-
perfectly plastic model, generally correspond well with the experimental data at higher loads. For the
tests of the second test series, the specimens with smaller-diameter longitudinal reinforcing bars are
predicted to behave less stiffly than was the case in the actual tests. The predictions of the behaviour
at the end of the cantilever (P-wC, Fig. 6.14(b)) correspond reasonably well with the experimental
data. The capacities of the specimens tested without axial force applied (T0 and T11) are underesti-
mated significantly. The load at shear failure is underestimated, see Tab. 6.6.
The shear failure is modelled as load controlled. In the experiments, the failure occurred de-
formation controlled.
The plastic hinges are modelled to occur in the axes of Support B and the loading yokes. In
reality, the hinges form adjacent to the support or loading structure in cases where the hinge is caused
by concrete crushing (all tests except T9, T10 and T11). Similarly, the shear-failed sector is modelled
to extend from axis to axis, while in reality the extent of it is smaller. Both simplifications contribute
to the underestimation of the moment at Support B.

129
Behaviour of RC Slab Strips:
Experimental Investigations

6.6 Interpretation
Interpretations and conclusions of the comparison of the predictions and the experimental data are:
- Rigid-perfectly plastic models describe the general load-deflection behaviour at high loads
sufficiently well.
- Differential equation models accurately describe the load-deflection behaviour for the full
range of loads. Consequently, moment and shear force distributions are also well described.
- Shear failure does not necessarily result in system failure in systems under axial tension.
- Modelling the shear-failed sector as a funicular polygon and introducing hinges with Mpl = 0 at
both ends of the sector is a simple method and results in a conservative estimate of the influ-
ence of shear failure.
- The cracking shear force model presented in Section 5.2 provides an explanation of why in
some of the specimens shear failure initiated in the span and not at a support. However, the
model is difficult to handle and can result in unreasonably low cracking shear force estimates
for sections in which the reinforcement on the tension side is yielding.
- According to the cracking shear force model, shear cracks outside the concrete compression
zone initiate at an angle of 45 degrees with respect to the beam axis. In test T1, however,
cracks developed parallel to the beam axis, and for T3, T6, T7 and T8 the crack angle was less
than 45 degrees. As no longitudinal compression was present in the specimens it is conceivable
that concurrent transverse tensile stresses influence the principal stresses and their direction.
Bond shear is accompanied by tensile hoop stresses, see for example Tepfers (1979). If rein-
forcing bars are sufficiently close to each other, the individual tensile hoops combine into a
single tensile hoop around multiple bars. The vertical distance between the reinforcement
layers being small compared to the bar diameter of specimens with large diameter bars, tensile
hoop stresses are a likely source of tensile stresses transverse to the beam axis.
- Longitudinal cracks on the upper and lower face of T6 at high loads indicate that the bond of
the longitudinal reinforcement is reduced by splitting (which is not considered in the model).
- Shear failure can be caused by the failure of a compression strut or arch. This needs to be
checked in addition to the cracking shear force and cracking shear forces larger than the shear
failure forces determined by an upper bound mechanism shall not be taken into account.
- Rigid-perfectly plastic models underestimate the load causing shear failure because the slope
of the beam in the governing sector is predicted too small (and consequently, the shear force
acting on the beam is too large).
- For the static system investigated, the combination of an upper bound solution for the shear
resistance and shear forces determined by a differential equation model results in a reasonable
estimation of the shear failure load.
- The prediction of the shear failure load is highly sensitive to changes in the flexural behaviour
(for example in the case of T7, changing the reinforcing steel properties from static to dynamic
results in shear failure being predicted where no shear failure was predicted before).
- The maximum possible shear force in a beam sector is given by the sum of the plastic
moments at its ends divided by the sector length (all values with positive sign). In cases where
the maximum shear force is smaller than the minimum of cracking shear force and shear
resistance, shear failure will not occur.
- Significant deviations from the theoretical values of the effective moment-curvature relation
and the plastic moment must be expected in a structure. The deviations are caused by the dif-
ference in geometry and, especially in the design of new structures for which no material tests
are available, by the material properties. Due to the sensitivity of the shear failure prediction to
such deviations, a cautious assessment is advisable.

130
Summary

7 Summary and Conclusions

7.1 Summary
This thesis aims at contributing to a better understanding of the structural response of reinforced
concrete slab strips subjected to axial tension and transverse loads. The main focus is on the influ-
ence of the axial tension on the structural response. Simple section based models are presented that
allow (i) a realistic assessment of the load-deformation behaviour; (ii) a reasonable estimation of the
moment and shear force distribution; and (iii) the identification of the governing parameters and the
assessment of their influence on the structural behaviour in the conceptual design phase.
The first part of the thesis (Chapters 2 and 3) covers material properties and basic aspects of
section behaviour.
In Chapter 2, the characteristics of the stress-stain behaviour are examined. Based on a litera-
ture review, it is shown how the properties of concrete and reinforcing steel are determined experi-
mentally, what the relevant parameters are and how these parameters are influenced by different
conditions. This is supplemented by a number of useful empirical relations connecting different
parameters. For both materials an overview of models with different levels of complexity for describ-
ing the stress-strain behaviour is given. A similar overview for the bond between concrete and rein-
forcement concludes the chapter.
In Chapter 3, the characteristics of section behaviour are discussed with moment-axial force
strength envelopes and moment-curvature relations. The principles for their derivation with the
assumptions of plane sections remaining plane and of the section’s fibres being independent from one
another are outlined. Different simplifications for the numerical integration of the stresses over the
section are presented and their impact on the moment-axial force strength envelope is illustrated for
an example section. A simple model for moment-curvature relations with analytical expressions for a
number of distinct phases and a model based on rigid-perfectly plastic behaviour are presented and
their results compared to the ones found by numerical integration. As particular aspects of the section
behaviour, the contribution of reinforcing steel, the occurrence of situations where the moment and
corresponding curvature have opposite signs, the influence of the loading history in general and of
non-monotonic load histories in particular are discussed in detail.
The second part of the thesis (Chapters 4 and 5) covers the theoretical modelling of the behav-
iour of reinforced concrete slab strips.
Chapter 4 begins with an overview of different methods for modelling the flexural behaviour
of structural systems with reinforced concrete slab strips and introduces a propped cantilever with a
uniform transverse load as an example for discussing selected models later in the chapter. As a first
model, differential equation solutions are presented for both the case with and without axial force.
The influence of softening at the hinges as well as the effect of mixed transverse loading are investi-
gated. Next, a model based on rigid-perfectly plastic behaviour is explained. By means of a model
requiring numerical integration the influence of the support dimensions and of the transverse pressure
in the concrete at the supports is discussed. The results of the different models for the propped canti-
lever are compared and the factors influencing the flexural behaviour are investigated with a special
focus on the influence of an axial tensile force.

131
Summary and Conclusions

The first section of Chapter 5 points out the interconnection between flexure and shear, estab-
lishes a connection between shear cracks and shear failure and lists the different contributions to the
shear transfer in a reinforced concrete member that are commonly included in attempts to estimate
the shear capacity of a cracked member.
Under the assumption that flexural cracks prevent axial tensile stresses but do not prevent the
transfer of shear stresses, the shear stress distribution prior to shear cracking is discussed in the
second section of Chapter 5 using the elastic shear flow theory for a transformed section. The influ-
ence of different reinforcing steel models and of the load history on the cracking shear force are
presented. The cracking shear force model is combined with a numerical integration or a differential
equation model and applied to a propped cantilever. For the same propped cantilever, the influence of
axial tension on the transverse load that causes shear cracking is investigated.
The consequences of shear failure on the structural response of a system are the subject of the
third section of Chapter 5. A model subdividing the system into intact sectors and a shear-failed
sector is proposed for estimating the structural response after shear failure. The transition from the
behaviour of the intact system to the behaviour following a shear failure and the influence of the
length of the shear-failed sector are discussed.
The third part of the thesis (Chapter 6) contains the validation of the presented models by
means of a comparison of the model results with the measured results from an experimental
campaign.
In the first section of Chapter 6, the test set-up and procedure, the test specimens and the main
results of an experimental campaign with twelve large-scale tests on reinforced concrete slab strips
carried out at the Swiss Federal Institute of Technology are described.
In the second section of Chapter 6, the models introduced in the second part of the thesis are
adapted to the static system and loading conditions of the test set-up and selected results are present-
ed.
The model results are compared to the experimental data in the third section of Chapter 6. The
comparison starts with the material, proceeds to the section behaviour and finally to the system
behaviour. Good agreement of the results is highlighted and significant differences are discussed.

7.2 Conclusions

7.2.1 General Conclusions

In the first part of this thesis, simplified analytical solutions for typical slab strip sections that yield
moment-axial strength envelopes and moment-curvature relations of sufficient accuracy for many
practical problems are presented. It is shown that (i) moment-curvature relations generally contain
multiple peaks; (ii) a shift in the governing peak with changing axial force can lead to non-
convexities in the moment-axial force strength envelope; (iii) the maximum moment for a certain
axial force depends on the load history; and (iv) computational routines concentrating on the first
peak do not necessarily yield the strength envelope, but a conservative approximation.
In the second part of the thesis, it is shown that in the presence of axial forces the equilibrium
conditions have to be formulated for the deformed system in order to gain meaningful results. Rigid-
perfectly plastic models are the simplest models, but they only yield a limited amount of information
on the structural behaviour of the structure. When closed form solutions for the differential equations
describing a given system are found, they are well suited for the conceptual design phase, where a
wide range of parameters needs to be investigated efficiently. Numerical integration lends itself more
to the detailed design phase, because it allows the consideration of stiffnesses varying with the load
and over the length of the beam. A work-around in form of a virtual hinge length allows the use of

132
Conclusions

the models in cases with non-perfectly plastic hinge behaviour. For the propped cantilever with a
uniform transverse load used as an example, softening after the moment peak reduces the ultimate
load in the absence of axial tension. In the presence of axial tension, perfectly plastic behaviour of
the hinge in the span results in the expansion of the hinge. The behaviour in this hinge sector can be
interpreted as the sum of a funicular polygon contribution and a bending contribution. The main
factors influencing the behaviour of a given structural system under axial tension are (i) the tensile
force itself; (ii) the plastic moment(s); and (iii) the stiffness. For a given beam section the plastic
moment and the stiffness depend on the axial force and the variability of the results can be expressed
as a function of the axial force only. Generally, the influence of parameter variations and of a more
detailed modelling vanishes with increasing transverse load and increasing axial tension, as the
funicular polygon contribution becomes dominant. For a given beam section, independently of the
span, a hinge forms in the span at a uniform transverse load q ≥ Tχpl, provided the moments over the
supports are not positive.
Because of the interconnection of moment and shear, it is essential to model the flexural
behaviour of a structure adequately when the shear capacity is a concern. Under the assumption that
flexural cracks prevent axial tensile stresses but not the transfer of shear stresses, the shear stress
distribution prior to shear cracking can be calculated based on the elastic shear flow theory for a
transformed section. The resulting cracking shear force for a given section depends on the moment
and axial force applied, the longitudinal reinforcing ratio and the bond shear capacity as well as on
the load history. It can be sensitive to residual strains. For a symmetric section under axial tension
and with full bond shear capacity, the cracking shear force is constant at a value around Vcr = 2ebfct
for moments, where flexure is resisted by the reinforcing steel only. The model shows that in many
cases, the first shear crack occurs at a distance from the support with the largest shear force and that
for a given system the load corresponding to shear cracking increases with the axial tension. The
formation of a shear crack is not equal to shear failure as different mechanisms enable a force trans-
fer in and around a crack, even in the absence of shear reinforcement. When failure occurs, the crack
has expanded and the static system has changed. The structural response of a system after shear
failure can be conservatively estimated by replacing the beam over the length of the shear failure
with a funicular polygon. For a propped cantilever without axial force, shear failure is equal to
system failure. In the presence of axial tension, the deflections increase and the system continues to
resist increasing transverse loads.
The application of the model to a wide range of slab strip section geometries and axial tensile
forces and the comparison of the model results with experimental data in the third part of this thesis
show that differential equation models allow an excellent prediction of the load-deformation behav-
iour prior to a shear failure and a reasonable prediction of the behaviour after shear failure. In sys-
tems under axial tension, rigid-perfectly plastic models are sufficient to describe the load-deflection
behaviour at higher transverse loads. The significant uncertainties regarding the actual section geom-
etry and material properties should be kept in mind not only when designing a new structure but also
when verifying an existing structure. Considering the sensitivity of the load corresponding to shear
failure to changes in the flexural behaviour, exact shear failure load predictions are illusory and a
cautious assessment in design and verification is advisable. While the cracking shear force model
presented allows an insight into shear behaviour, it is not sufficient to predict the shear failure load.
One reason is the large number of factors influencing the shear resistance and another is that shear
failure occurs over a certain length of a member and therefore, methods taking into account a certain
length are more adequate than section based models. For the example, reasonable predictions of the
shear failure load were found by combining an upper bound solution for the shear resistance with a
differential equation model to determine the shear forces corresponding to a certain load.

133
Summary and Conclusions

7.2.2 Influence of Axial Tension on the Structural Response of RC Slab Strips

The main conclusions with regard to the influence of axial tension are: (i) differential equation mod-
els provide a realistic description of the structural response of reinforced concrete slab strips under
axial tension and transverse load; (ii) axial tension increases the ultimate resistance of the system;
(iii) the system stiffness increases with increasing axial tension; (iv) shear failure does not cause
system failure in systems under axial tension (provided that the reinforcement is detailed such that
the tension can be sustained in the shear-failed zone); and (v) the accuracy of a shear failure pre-
diction is considerably smaller than the accuracy of the prediction of the flexural behaviour.

7.3 Recommendations for Future Research


The influence of axial tension has been studied for systems where the axial force remains constant. In
many real structures, the tensile force will increase with increasing deflection because of the support
conditions. The influence of such support conditions can be studied by means of the models provided,
but requires an extra iteration on the compatibility of the horizontal movement at the support(s).
The influence of axial tension has been studied based on the assumption of plane sections
remaining plane. This assumption greatly simplifies the calculations. However, the local force flow at
the supports and the loading points is not represented adequately by such models. The assumption
also leads to the reinforcement yielding over a range of moments (and the corresponding cracking
shear force dropping to a very low level). According to the tension cord model, the reinforcement
yields only in a single point and is either elastic or in the hardening phase in the other parts as long as
there is a minimal bond shear stress. Furthermore, slipping of the reinforcement following a splitting
failure cannot be modelled by pure section based models. Developing a stress field model for the
deformed geometry could serve as basis to investigate these issues. The stress field model could be
complemented by a modified tension cord model taking into account splitting failure and by a model
of the stress transfer capacity over a crack. Discrete cracks could be taken into account in such a
model.
The section based model does not take into account that crushing of a part of the compression
zone results in an unloading of the adjacent areas of the compression zone. This unloading and corre-
sponding local force flow around the crushed zone would be a valuable subject of further investiga-
tions.
The longitudinal reinforcement was continuous in the test specimens and continuity was also
implied in the models. In reality, the longitudinal reinforcement is often spliced. In new structures,
the splices can usually be positioned outside of the critical zones without difficulties. In existing
structures, however, the splices can be encountered in critical zones. An investigation of the influ-
ence of such splices would be a valuable aid for practicing engineers.

134
Recommendations for Future Research

Appendix A – Coefficients for Differential Equation Solutions

Differential Equation Solution for Case with Axial Force, Softening at Plastic Hinges
The coefficients of the differential equation solutions for the two beam sectors can be computed as:

 0 
 
 c11   qL22 2T 
c   
 12   q Tκ 2

 c13   q κ2 + M θ 
   
 c14  −1  2 
=  A= R with R  qL 1 2T 
 c21   0 
c22   
   q  1 − L1L  + M θ 1 − L1  
c23   T  κ 2 2  T  L  
c   
 24    1 ζkL1  
 q  2 + T  − ξM θ 
 κ  

 1 0 0 1 0 0 0 0 (A.1)
 
 0 0 0 0 cosh ( λ 2 ) sinh ( λ 2 ) L2 1
 0 0 0 0 cosh ( λ 2 ) sinh ( λ 2 ) 0 0
 
−k κ −k
A = 
T 0 0 0 0 0
cosh ( λ1 ) sinh ( λ1 ) L1 1 −1 0 0 −1
 
cosh ( λ1 ) sinh ( λ1 ) 0 0 −1 0 0 0
 0 a27 a37 −1 0 0 0 0
 
 a18 a28 ζk 0 0 −ζk κ −ζk 0 

k κ  L1  k  L1 
where a= − 1 a=  L − 1 − L1
T  L
27 37
 T  
a18 T cosh ( λ1 ) + ζk κ sinh ( λ1 ) =
= a28 T sinh ( λ1 ) + ζk κ cosh ( λ1 )

135
Appendix B – Cracking Shear Force Relations

Appendix B – Cracking Shear Force Relations

Fig. B1 illustrates the variation of the cracking shear force Vcr with increasing curvature for the
example section defined in Fig. 3.1 for no axial force and for an axial tension of N = 2867 kN. The
figure complements Fig. 5.7 where the cracking shear force relations are shown without the details.

Fig. B1 – Details of moment-curvature and cracking shear force-curvature relations for selected axial forces:
(a) N = 0; (b) N = 2867 kN.

The awkward peak of the cracking shear force at χ = 49 mrad/m for N = 0 is a result of the
(arbitrary) distribution of the shear flow between reinforcing steel and concrete in the calculations
and is not real.

136
Recommendations for Future Research

Literature

ACI Committee 408, 2003, Bond and Development of Straight Reinforcing Bars in Tension,
ACI 408R-03, American Concrete Institute, Farmington Hills, MI, USA, 49 pp.

Adebar, P., and He, W., 1994, “Influence of Membrane Forces on Transverse-Shear Reinforcement
Design,” Journal of Structural Engineering, V. 120, No. 4, pp 1347–1366.

Alvarez, M., 1998, Einfluss des Verbundverhaltens auf das Verformungsvermögen von Stahlbeton,
IBK Bericht, No. 236, Institute of Structural Engineering (IBK), Swiss Federal Institute of
Technology, Zürich, Switzerland, 182 pp.

Ammann, J.W., 1983, Stahlbeton- und Spannbetontragwerke unter stossartiger Belastung,


IBK Bericht, No. 142, Swiss Federal Institute of Technology, Zürich, Switzerland, 285 pp.

Bischoff, P.H., and Perry, S.H., 1991, “Compressive Behaviour of Concrete at High Strain Rates,”
Materials and Structures, V. 24, No. 6, pp 425–450.

Burns, C., 2012, Serviceability Analysis of Reinforced Concrete Based on the Tension Chord Model,
IBK Bericht, No. 342, Institute of Structural Engineering (IBK), Swiss Federal Institute of
Technology, Zürich, Switzerland, 147 pp.

Carrasquilio, R.L., Nilson, A.H., and Slate, F.O., 1981, “Properties of High Strength Concrete Sub-
ject to Short-Term Loads,” ACI Journal, V. 78, No. 3, pp 171–178.

CEB, 1988, Concrete Structures under Impact and Impulsive Loading, Bulletin d’Information,
Comité Euro-International du Béton (CEB), Lausanne, Switzerland, 166 pp.

Chang, G.A., and Mander, J.B., 1994, Seismic Energy Based Fatigue Damage Analysis of Bridge
Columns: Part I - Evaluation of Seismic Capacity, Technical Report, No. NCEER-94-006, Na-
tional Center for Earthquake Engineering Research, State University of New York at Buffalo,
Buffalo, New York, USA, 208 pp.

Collins, M.P., and Mitchell, D., 1997, Prestressed Concrete Structures, Response Publications,
Canada, 766 pp.

Cosenza, E., Greco, C., and Manfredi, G., 1993, “The Concept of ‘Equivalent Steel,’” CEB Bulletin
d’information, No. 218, pp 163–183.

Dodd, L., and Restrepo-Posada, J., 1995, “Model for Predicting Cyclic Behavior of Reinforcing
Steel,” Journal of Structural Engineering, V. 121, No. 3, pp 433–445.

Ehmann, J., 2003, Querkrafttragfähigkeit zugbeanspruchter Stahlbetonplatten in Verbundbrücken,


PhD Thesis, Universität Stuttgart, Stuttgart, Germany, 164 pp.

Eligehausen, R., Popov, E.P., and Bertero, V.V., 1982, “Local Bond Stress-Slip Relationships of
Deformed Bars under Generalized Excitations,” Proceedings of the 7th European Conference
on Earthquake Engineering, V. 4, pp 69–80.

137
Literature

Eligehausen, R., Popov, E.P., and Bertero, V.V., 1983, Local Bond Stress-Slip Relationships of
Deformed Bars Under Generalized Excitations, No. UCB/EERC-83/23, EERC, University of
California, Berkeley, USA, 169 pp.

Ernst, G., and Marti, P., 1998, Versuche an Betonplatten mit integrierten Schalungselementen aus
Stahlbeton, IBK Bericht, No. 235, Institute of Structural Engineering (IBK), Swiss Federal
Institute of Technology, Zürich, Switzerland, 109 pp.

Eskola, L., 1996, Zur Ermüdung teilweise vorgespannter Betontragwerke, IBK Bericht, No. 223,
Institute of Structural Engineering (IBK), Swiss Federal Institute of Technology, Zürich,
Switzerland, 141 pp.

Etter, S., 2012, Zum Tragverhalten von Verbundbauteilen aus altem und neuem Stahlbeton, IBK
Bericht, No. 348, Institute of Structural Engineering (IBK), Swiss Federal Institute of Tech-
nology, Zürich, Switzerland, 146 pp.

Etter, S., Heinzmann, D., Jäger, T., and Marti, P., 2009, Versuche zum Durchstanzverhalten von
Stahlbetonplatten, IBK Bericht, No. 324, Institute of Structural Engineering (IBK), Swiss
Federal Institute of Technology, Zürich, Switzerland, 64 pp.

Etter, S., Villiger, S., and Marti, P., 2012, Bruchversuche an Plattenstreifen aus altem und neuem
Stahlbeton unter exzentrischem Längsdruck, IBK Bericht, No. 336, Institute of Structural
Engineering (IBK), Swiss Federal Institute of Technology, Zürich, Switzerland, 76 pp.

Fehlmann, P., Wolf, T., and Vogel, T., 2011, Versuche zum Ermüdungsverhalten von Stahlbet-
onbrücken, IBK Bericht, No. 332, Institute of Structural Engineering (IBK), Swiss Federal
Institute of Technology, Zürich, Switzerland, 95 pp.

Fenwick, R.C., and Paulay, T., 1968, “Mechanisms of Shear Resistance of Concrete Beams,” Journal
of the Structural Division, V. 94, No. ST10, pp 2325–2350.

fib, 2000, Bond of Reinforcement in Concrete, fib Bulletin, International Federation for Structural
Concrete (fib), Lausanne, Switzerland, 427 pp.

fib, 2010, Model Code 2010, First Complete Draft - Volume 1, fib Bulletin, International Federation
for Structural Concrete (fib), Lausanne, Switzerland, 292 pp.

Fürst, A., and Marti, P., 1999, Versuche an Trägern mit Unterspannung aus vorfabrizierten,
vorgespannten Betonzuggliedern, IBK Bericht, No. 243, Institute of Structural Engineering
(IBK), Swiss Federal Institute of Technology, Zürich, Switzerland, 108 pp.

Galmarini, A., Locher, D., and Marti, P., 2012, Response of RC Slab Strips Subjected to Axial Ten-
sion and Transverse Load - Prediction Competition, Institute of Structural Engineering (IBK),
Swiss Federal Institute of Technology (ETH), Zürich, Switzerland, 7 pp.

Galmarini, A., Locher, D., Wyss, J., and Marti, P., 2013, Versuche an Plattenstreifen aus Stahlbeton
unter Längszug und Querbelastung, IBK Bericht, No. 349, Institute of Structural Engineering
(IBK), Swiss Federal Institute of Technology, Zürich, Switzerland, 131 pp.

Goto, Y., 1971, “Cracks Formed in Concrete around Deformed Tension Bars,” ACI Journal Proceed-
ings, V. 68, No. 4, pp 244–251.

Grote, D.L., Park, S.W., and Zhou, M., 2001, “Dynamic Behavior of Concrete at High Strain Rates
and Pressures: I. Experimental Characterization,” International Journal of Impact Engineering,
V. 25, No. 9, pp 869–886.

138
Recommendations for Future Research

Hagsten, L.G., 2010, “Test on Moment Redistribution in Continuous Reinforced Concrete Beams,”
Bygningsstatiske Meddelelser, V. 81, No. 3-4, pp 79–149.

Hjorth, O., 1976, Ein Beitrag zur Frage der Festigkeiten und des Verbundverhaltens von Stahl und
Beton bei hohen Beanspruchungsgeschwindigkeiten, Schriftenreihe des Institutes für
Baustoffkunde und Stahlbetonbau der Technischen Universität Braunschweig, No. Heft 32,
Amtliche Materialprüfanstalt für das Bauwesen, Braunschweig, Germany, 188 pp.

Hoang, L.C., and Nielsen, M.P., 1996, “Continuous Reinforced Concrete Beams - Stress and Stiff-
ness Estimations in the Serviceability Limit State,” Bygningsstatiske Meddelelser, V. 67,
No. 3-4, pp 45–89.

Hognestad, E., 1951, A Study of Combined Bending and Axial Load in Reinforced Concrete Members,
No. Bulletin No. 399, University of Illinois Engineering Experiment Station, Urbana-
Champaign, USA, 128 pp.

Hsu, T.T.C., Slate, F.O., Sturman, G.M., and Winter, G., 1963, “Microcracking of Plain Concrete
and the Shape of the Stress-Strain Curve,” ACI Journal Proceedings, V. 60, No. 2, pp 209–224.

Hussein, A., and Marzouk, H., 2000, “Behavior of High-Strength Concrete under Biaxial Stresses,”
ACI Materials Journal, V. 97, No. 1, pp 27–36.

Jäger, T., and Marti, P., 2006, Versuche zum Querkraftwiderstand und zum Verformungsvermögen
von Stahlbetonplatten, IBK Bericht, No. 294, Institute of Structural Engineering (IBK), Swiss
Federal Institute of Technology, Zürich, Switzerland, 358 pp.

Jiang, L., Huang, D., and Xie, N., 1991, “Behavior of Concrete Under Triaxial Compressive-
Compressive-Tensile Stresses,” ACI Materials Journal, V. 88, No. 2, pp 181–185.

Jourawski, D.I., 1856, “Remarques sur la résistance d’un corps prismatique et d’une pièce composée
en bois ou en tôle de fer à une force perpendiculaire à leur longueur,” in Mémoires et docu-
ments, relatifs à l’art des constructions et au service de l’ingénieur, Annales des ponts et
chaussées, 3. Series, V. 12, Dalmont, Paris, France, pp 328–351.

Kaufmann, W., 1998, Strength and Deformations of Structural Concrete Subjected to In-Plane Shear
and Normal Forces, IBK Bericht, No. 234, Institute of Structural Engineering (IBK), Swiss
Federal Institute of Technology, Zürich, Switzerland, 147 pp.

Kaufmann, W., and Marti, P., 1996, Versuche an Stahlbetonträgern unter Normal- und Querkraft,
IBK Bericht, No. 226, Institute of Structural Engineering (IBK), Swiss Federal Institute of
Technology, Zürich, Switzerland, 141 pp.

Kenel, A., and Marti, P., 1996, Faseroptische Dehnungsmessungen an einbetonierten


Bewehrungsstäben, IBK Bericht, No. 271, Institute of Structural Engineering (IBK), Swiss
Federal Institute of Technology, Zürich, Switzerland, 93 pp.

Kupfer, H., 1973, “Das Verhalten des Betons unter mehrachsiger Kurzzeitbelastung unter besonderer
Berücksichtigung der zweiachsigen Beanspruchung,” Deutscher Ausschuss für Stahlbeton,
V. 229, pp 1–105.

Locher, D. et al., 2014, Versuche an Plattenstreifen aus Stahlbeton unter Längszug und Querbelas-
tung, 2. Serie, IBK Bericht, No. 353, Institute of Structural Engineering (IBK), Swiss Federal
Institute of Technology, Zürich, Switzerland, 133 pp.

Lowes, L.N., 1999, Finite Element Modeling of Reinforced Concrete Beam-Column Bridge Connec-
tions, PhD Thesis, University of California, Berkeley, Berkeley, USA, 416 pp.

139
Literature

Lutz, L.A., and Gergely, P., 1967, “Mechanics of Bond and Slip of Deformed Bars in Concrete,”
ACI Journal Proceedings, V. 64, No. 11, pp 711–721.

Malvar, L.J., 1998, “Review of Static and Dynamic Properties of Steel Reinforcing Bars,”
ACI Materials Journal, V. 95, No. 5, pp 609–616.

Mander, J.B., 1983, Seismic Design of Bridge Piers, PhD Thesis, Department of Civil Engineering,
University of Canterbury, Canterbury, New Zealand, 442 pp.

Manjoine, M.J., 1944, “Influence of Rate of Strain and Temperature on Yield Stresses of Mild Steel,”
Journal of Applied Mechanics, pp 211–218.

Marti, P., Alvarez, M., Kaufmann, W., and Sigrist, V., 1998, “Tension Chord Model for Structural
Concrete,” Structural Engineering International, V. 8, No. 4, pp 287–298.

Mattock, A.H., Kriz, L.B., and Hognestad, E., 1961, “Rectangular Concrete Stress Distribution in
Ultimate Strength Design,” ACI Journal Proceedings, V. 57, No. 2, pp 875–928.

Meyboom, J., and Marti, P., 2001, Experimental Investigation of Shear Diaphragms in Reinforced
Concrete Slabs, IBK Bericht, No. 263, Institute of Structural Engineering (IBK), Swiss Federal
Institute of Technology, Zürich, Switzerland, 165 pp.

Van Mier, J.G.M., 1984, Strain-Softening of Concrete under Multiaxial Loading Conditions,
PhD Thesis, Eindhoven University of Technology, Eindhoven, Netherlands, 349 pp.

Van Mier, J.G.M., 1986, “Fracture of Concrete under Complex Stress,” Heron, V. 31, No. 3, pp 1–90.

Mörsch, E., 1902, Der Eisenbetonbau, 1st ed., Konrad Wittwer, Stuttgart, Germany, 118 pp.

Mörsch, E., 1923, Der Eisenbetonbau, V. 1, 6th ed., Konrad Wittwer, Stuttgart, Germany, 490 pp.

Mörsch, E., 1929, Der Eisenbetonbau, V. 2, 6th ed., Konrad Wittwer, Stuttgart, Germany, 541 pp.

Nelissen, L.J.M., 1972, “Biaxial Testing of Normal Concrete,” Heron, V. 18, No. 1, pp 1–90.

Newman, K., and Newman, J.B., 1971, “Failure Theories and Design Criteria for Plain Concrete,” in
Structure, Solid Mechanics and Engineering Design, Wiley-Interscience, London, pp 963–995.

Nielsen, M.P., and Hoang, L.C., 2011, Limit Analysis and Concrete Plasticity, 3rd ed., CRC Press,
Boca Raton, USA, 788 pp.

Niwa, Y., Kobayashi, S., and Koyanagi, W., 1967, “Failure Criterion of Lightweight Aggregate
Concrete Subjected to Triaxial Compression,” Memoires of the faculty of engineering, Kyoto
University, V. 29, No. 2, pp 119–131.

Nowak, A.S., and Szerszen, M.M., 2003, “Calibration of Design Code for Buildings (ACI 318):
Part 1 - Statistical Models for Resistance,” ACI Structural Journal, V. 100, No. 3, pp 377–382.

Okubo, S., and Nishimatsu, Y., 1985, “Uniaxial Compression Testing Using a Linear Combination of
Stress and Strain as the Control Variable,” International Journal of Rock Mechanics and Min-
ing Sciences & Geomechanics Abstracts, V. 22, No. 5, pp 323–330.

Pajak, M., 2011, “The Influence of the Strain Rate on the Strength of Concrete Taking into Account
the Experimental Techniques,” Architecture Civil Engineering Environment, No. 3, pp 77–86.

140
Recommendations for Future Research

Pfyl, T., and Marti, P., 2001, Versuche an stahlfaserverstärkten Stahlbetonelementen, IBK Bericht,
No. 268, Institute of Structural Engineering (IBK), Swiss Federal Institute of Technology,
Zürich, Switzerland, 137 pp.

Popovics, S., 1970, “A Review of Stress-Strain Relationships for Concrete,” ACI Journal Proceed-
ings, V. 67, No. 3, pp 243–248.

Popovics, S., 1971, “Factors Affecting the Elastic Deformations of Concrete,” in Mechanical Behav-
ior of Materials, International Conference on Mechanical Behavior of Materials. The Society
of Materials Science, Kyoto, Japan, pp 172–183.

Rehm, G., 1957a, “The Fundamental Law of Bond,” in Proceedings of the Symposium on Bond and
Crack Formation in Reinforced Concrete, RILEM Symposium on Bond and Crack Formation
in Reinforced Concrete. Stockholm, Sweden, pp 491–498.

Rehm, G., 1957b, “Stress Distribution in Reinforcing Bars Embedded in Concrete,” in Proceedings
of the Symposium on Bond and Crack Formation in Reinforced Concrete, RILEM Symposium
on Bond and Crack Formation in Reinforced Concrete. Stockholm, Sweden, pp 499–507.

Rokugo, K., and Koyanagi, W., 1992, “Role of Compressive Fracture Energy of Concrete on the
Failure Behavior of Reinforced Concrete Beams,” in A. Carpenteri (ed.), Applications of Frac-
ture Mechanics to Reinforced Concrete, Taylor & Francis, London, pp 437–464.

Rüsch, H., 1960, “Researches Toward a General Flexural Theory for Structural Concrete,”
ACI Journal Proceedings, V. 57, No. 7, pp 1–25.

Sargin, M., 1971, Stress-Strain Relationships for Concrete and the Analysis of Structural Concrete
Sections, Studies Series, Solid Mechanics Division, University of Waterloo, Waterloo, Ontario,
Canada, 167 pp.

Seefeld-Ebert, B., Ott, C., and Marti, P., in press, Versuche zur Querkraftverstärkung einseitig
zugänglicher Stahlbetonplatten mit eingemörtelten Bewehrungsstäben, IBK Bericht, Institute
of Structural Engineering (IBK), Swiss Federal Institute of Technology, Zürich, Switzerland.

Seelhofer, H., 2009, Ebener Spannungszustand im Betonbau: Grundlagen und Anwendungen, IBK
Bericht, No. 320, Institute of Structural Engineering (IBK), Swiss Federal Institute of Tech-
nology, Zürich, Switzerland, 236 pp.

Shima, H., Chou, L.-L., and Okamura, H., 1987, “Micro and Macro Models for Bond in Reinforced
Concrete,” Journal of the Faculty of Engineering, The University of Tokyo, V. 39, No. 2,
pp 133–194.

Sigrist, V., 1995, Zum Verformungsvermögen von Stahlbetonträgern, IBK Bericht, No. 210, Institute
of Structural Engineering (IBK), Swiss Federal Institute of Technology, Zürich, Switzerland,
159 pp.

Sigrist, V., and Marti, P., 1993, Versuche zum Verformungsvermögen von Stahlbetonträgern,
IBK Bericht, No. 202, Institute of Structural Engineering (IBK), Swiss Federal Institute of
Technology, Zürich, Switzerland, 112 pp.

Stoffel, P., and Marti, P., 1997, Modellversuche Europabrücke, IBK Bericht, No. 227, Institute of
Structural Engineering (IBK), Swiss Federal Institute of Technology, Zürich, Switzerland,
118 pp.

141
Suprenant, B.A., 1995, “Core Strength Variation of In-Place Concrete,” The Concrete Producer, 3 pp.

Tepfers, R., 1979, “Cracking of Concrete Cover along anchored Deformed Reinforcing Bars,”
Magazine of Concrete Research, V. 31, No. 106, pp 3–12.

Thorenfeldt, E., Tomaszewicz, A., and Jensen, J.J., 1987, “Mechanical Properties of High-Strength
Concrete and Application to Design,” in I. Holand (ed.), Proceedings of the Symposium: Utili-
sation of High Strength Concrete, Tapir, Trondheim, Stavanger, Norway, pp 149–159.

Trümpi-Althaus, S., 2006, Tragverhalten und Bemessung von Stahlbetonvortriebsrohren,


IBK Bericht, No. 292, Institute of Structural Engineering (IBK), Swiss Federal Institute of
Technology, Zürich, Switzerland, 141 pp.

Vecchio, F.J., and Collins, M.P., 1986, “The Modified Compression Field Theory for Reinforced
Concrete Elements Subjected to Shear,” ACI Structural Journal, V. 83, No. 2, pp 219–231.

Wight, J.K., and MacGregor, J.G., 2011, Reinforced Concrete: Mechanics and Design, 6th ed.,
Pearson, 1157 pp.

Xie, J., Elwi, E., and MacGregor, J.G., 1995, “Mechanical Properties of Three High-Strength Con-
cretes Containing Silica Fume,” ACI Materials Journal, V. 92, No. 2, pp 135–145.

142
Recommendations for Future Research

Notation

Roman capital letters


A cross-sectional area, auxiliary parameter
A matrix used to compute differential equation solution coefficients
Agt strain at peak stress in steel tests
B auxiliary parameter
C vector of coefficients used to compute differential equation solution coefficients
E modulus of elasticity / Young’s modulus
EcF strain softening modulus in stress-strain relation of concrete
Esh strain-hardening modulus of steel
EI flexural stiffness
I
EI stiffness of elastic-uncracked section
EIII stiffness of elastic-cracked section
EIs stiffness of elastic section where only steel is active
G weight
GF specific fracture energy (per unit area)
I moment of inertia
L span length, length of beam sector
L0 length of beam sector that failed in shear
M moment
M0 moment referring to the undeformed reference system
Mcr cracking moment
Mpl plastic moment
Mu flexural strength
Mθ offset of linear M-θ relation at θ = 0
N axial force
P point load
Q point load
QSF load at the occurrence of shear failure
R vector used to compute differential equation solution coefficients
S first moment of area
T tensile force
UcF specific fracture energy (per unit volume)

143
Notation

V shear force
V0 shear force referring to the undeformed reference system
Vcr cracking shear force
Vu shear resistance
W passive work

Roman small letters


a clearance between loading plates used to determine the shear resistance
as cross-sectional area of reinforcing steel (in slab strips: per unit slab width)
b width
c cohesion, coefficient
d diameter
dD dissipated energy per unit volume
dH hysteretic dissipation per unit volume
dU strain energy per unit volume
e eccentricity
f function
fc compressive strength of concrete, nominal compressive strength of concrete
fc,cube compressive strength of concrete determined by cube test
fcc compressive strength of concrete determined by cylinder test
fcc3 compressive strength of concrete in multiaxial loading states
fcR3 residual strength of concrete in multiaxial loading states
fct tensile strength of concrete
fcts splitting tensile strength of concrete
fs0 nominal yield strength of steel
fs0.2 proof stress taken as yield strength of steel for steel without yield plateau
fsy yield strength of steel
fsu tensile strength of steel
g uniformly distributed load, self-weight
h height
l length
lb embedment length of reinforcing bar in pull-out tests
lF length of fracture zone
lp length of plastic hinge
k auxiliary parameter, rotational spring stiffness
k3 auxiliary parameter in modified Coulomb failure criterion
k3ε auxiliary parameter in the model of the crushing strain in multiaxial loading states

144
Recommendations for Future Research

kc auxiliary parameter for stress-strain relation according to Collins and Mitchell


kd auxiliary parameter for bond shear stress-slip relation according to Model Code
kE coefficient accounting for the influence of the aggregate type on the elastic modulus of
concrete
kp auxiliary parameter for bond shear stress-slip relation according to Model Code
ks auxiliary parameter for bond shear stress-slip relation according to Model Code
kσ auxiliary parameter for stress-strain relation according to Sargin
nc auxiliary parameter for stress-strain relation according to Collins and Mitchell
nbar number of reinforcing bars in a layer
ns ratio Es / Ec
q line load
s coordinate along deformed reference system, spacing of reinforcing bars
sR spacing of ribs on reinforcing bar
t time
w crack width, deflection
w′ slope
x coordinate, distance
y coordinate
z coordinate
zc location of centroid in z-direction

Greek letters
Δ increment, difference
∑ sum
α stress block factor, auxiliary parameter
β auxiliary parameter
δ displacement
δb bond slip
ε strain
ε̇ strain-rate
ε0 axial strain
εc1 strain of concrete when the compressive strength is reached
εcR residual strain of concrete crushing zone
εct tensile strain of concrete
εcu crushing strain of concrete
εsg plastic strain of reinforcing steel at rupture
εsh strain of reinforcing steel at the beginning of strain hardening

145
Notation

εsy yield strain of reinforcing steel


εsu rupture strain of reinforcing steel
ζ auxiliary parameter for stress-strain relation according to Sargin, ratio of rotational
spring stiffnesses
θ rotation
κ auxiliary parameter for differential equation solution
λ auxiliary parameter for differential equation solution
ν Poisson’s ratio
ξ auxiliary parameter, ratio of plastic moments
ρ geometric reinforcement ratio
σ stress
σ1,2,3 principal stresses
τ shear stress
τb bond shear stress
τbf residual bond shear strength
τbu bond shear strength
τ̇b bond shear stress rate
φ angle, angle of internal friction
χ curvature
χcF softening stability coefficient used for concrete stress-strain relation
ω mechanical ratio of reinforcing steel

Subscripts
0 initial, reference
c concrete, centroid
cr cracking
dyn dynamic
eff effective
eq equivalent
i pointer
k characteristic
l left
m mean
n transformed section
max maximum
min minimum
r right

146
Recommendations for Future Research

ref reference
s steel, spacing
spec specified
stat static
t tensile
tan tangent
trans transition
unl unloading
vol volume, volumetric
u ultimate, crushing
y yielding

Special Symbols
@ spacing of reinforcing bars
Ø diameter of reinforcing bar
° degrees

147

You might also like