Development of a consistent approach for design

and assessment of structural concrete members

using stress fields and strut-and-tie models

THÈSE NO 7911 (2017)

PRÉSENTÉE LE 25 AOÛT 2017
À LA FACULTÉ DE L'ENVIRONNEMENT NATUREL, ARCHITECTURAL ET CONSTRUIT
LABORATOIRE DE CONSTRUCTION EN BÉTON
PROGRAMME DOCTORAL EN GÉNIE CIVIL ET ENVIRONNEMENT

ÉCOLE POLYTECHNIQUE FÉDÉRALE DE LAUSANNE

POUR L'OBTENTION DU GRADE DE DOCTEUR ÈS SCIENCES

PAR

Filip NIKETIĆ

acceptée sur proposition du jury:

Prof. D. Lignos, président du jury

Prof. A. Muttoni, Dr M. Fernández Ruiz, directeurs de thèse
Prof. L. C. Hoang, rapporteur
Prof. D. Zastavni, rapporteur
Prof. K. Beyer, rapporteuse

Suisse
2017
Foreword
Since the early developments of structural concrete, designers have searched for
consistent and comprehensive methods for its design. One approach developed in this frame
was the stress field method, grounded on the theory of plasticity and allowing designers for an
enhanced freedom to arrange the reinforcement and deciding upon the main load-carrying
actions. Currently, other than tools suitable for design of new structures, the assessment of the
strength of existing structures is becoming a significant task to be performed by engineers. Such
assessment allows checking on the level of safety of existing structures and if they are suitable
for new or upgraded actions without the need of retrofitting them.
Within this frame, the thesis of Filip Niketić investigates on the application of stress
fields both for cases related to design and to assessment. The work presents first the conceptual
differences between design and assessment and possible approaches for both cases. Also, the
possibility of developing suitable stress fields in an automated manner valid both for design and
assessment purposes is discussed. This latter is approached by developing elastic-plastic stress
fields, which are justified to yield exact solutions according to limit analysis as, at failure, they
yield a licit stress field compatible with a licit failure mechanism. The research of Filip Niketić
on elastic-plastic stress fields is completed with a systematic comparison to test results, showing
excellent agreement, as well as a discussion on the safety format issues.
In addition, the thesis presents a detailed investigation on the strength of compression
fields developing in concrete accounting for the influence of transverse strains and its
interaction with the web reinforcement. This discussion, usually referred to as the strength
reduction factors or efficiency factors of concrete, is one of the basic inputs to calculate the
strength of stress fields. In the thesis, a new approach to the issue is presented, based on a
mechanical model describing the various potential failure modes. The results of this approach
are observed to be consistent to test observations and to explain the results of currently used
formulas with an empirical ground.

Lausanne, July 2017

Prof. Dr Aurelio Muttoni Dr Miguel Fernández Ruiz

 To my dear parents,
Snežana and Živorad

Grateful son
“Let the future tell the truth and evaluate each one according to his work and accomplishments.
The present is theirs; the future, for which I have really worked, is mine.”

Nikola Tesla, April 1927

Acknowledgements
I would like to express the deepest gratitude to my thesis supervisor Prof. Aurelio
Muttoni, and co-supervisor Dr. Miguel Fernández Ruiz at the Structural Concrete Laboratory
(IBETON) at Swiss Federal Institute of Technology in Lausanne (EPFL) for giving me the
opportunity to become an active member of their research team, and providing me with the
necessary support to carry out my work. Under your supervision I have grown both as a person
and as an engineer, and for that I am forever grateful.
I would also like to thank Dr. Eckart Hars for his generous help and encouragement.
Working with you has truly been an honour and a blessing. Thank you for each second of your
time.
I would like to acknowledge the members of my thesis jury, namely Prof. Linh Cao
Hoang from the Department of Civil Engineering at Technical University of Denmark (DTU),
Prof. Denis Zastavni from the Faculty of Architecture, Architectural Engineering and Urban
Planning at Catholic University of Louvain (UCL), Prof. Katrin Beyer, head of the Earthquake
Engineering and Structural Dynamics Laboratory at EPFL and Prof. Dimitrios Lignos
(president of the jury), head of the Resilient Steel Structures Laboratory at EPFL for their
engagement and useful comments that improved the quality of my thesis.
My work was funded by the Federal Roads Office (FEDRO) to whom I would like to
give many thanks.
The support of Dr. Peter Tanner and Dr. Miguel Prieto from the Eduardo Torroja
Institute of Construction Sciences in Spain is fully acknowledged.
Special thanks goes to Dr. Olivier Burdet for helping me create the webpage with my
database and providing me with the technical support for writing this thesis.
I would like to acknowledge Garbiela Jovičić for proofreading my thesis and correcting
my English in a very short period of time.
Ann, Fabio, Dr. Burdet, Raffaele, Michaele, Francesco, Max, Bastian, Marie-Rose and
Dražen are thanked for proofreading my abstract and translating it to French, Italian and
German.
I would like to thank João for many interesting discussions we had during the last two
years of my work, and for driving me to school almost each morning for the last two years!
Our secretary, Yvonne Bühl-Brauch, is thanked for the administrative and personal
support she generously provided during my stay at IBETON.

vii
Acknowledgements

Over the period of my PhD studies, I sheared my office with ten amazing people:
Stefano, Francesco (Donarini), Rinaldo, Hugo, Joël, Quentin, Simone, Carlos, Francesco
(Moccia), Luca and Francesco (Moccia again!). I would like to thank each and every one of
you for making my days at EPFL fun and my studies enjoyable. I am looking forward to joining
you on the “Wall of Fame”.
Special thanks goes to the officemate who managed to put up with me the longest:
Francesco Moccia. We spent a lot of great moment together, and I am lucky to have you as a
friend.
I would like to thank Fabio and his family (Riccardo, Claire and Sandro) for accepting
me from the very beginning, and making me feel at home in Switzerland. I am glad to have met
you, and I am looking forward to our future moments.
I am more than happy to know Marie-Rose, Qianhui, Angelica, Raluca, Sarah, Pia,
Maléna, Francisco, Frédéric, Dan, Jürgen, Darko, Michael, Patrick, Raffaele, Ioannis,
Francesco (Cavagnis), Ovidiu, (big) João, Shenghan, Alessandro, Hadi and many more
interesting, talented and fun people from EPFL. I will cherish the time we spent together for as
long as I live.
I would also like to thank my old friends Zoki, Ilija, Miloš, Luka, Vasja, Mile, Nemanja,
Dražen, Šeki, Paja, Buda, Knele, Isi and Urke for supporting me regardless of the physical
distance and the little time we spent together over the past four years.
To my family: Mira, Vasa, Vlada, Miša, Bane, Bojka, Buca, Stevan and Naca I say
thank you for giving me your love and attention, for nourishing and forming me as a person. I
love you all.
I am lucky to have a cousin who loves and supports me with all her hart. Ana you showed
me what it means to have a sister. I am glad that life brought us closer together.
Mici always had a special place in my heart, and the words cannot quantify the amount
of love and support she gave me from the very beginning. For this, I thank her greatly.
I could not ask for more loving and generous parents than Sneža and Žika. To you I owe
my life, my soul my everything. You have carried me swiftly through the darkest moments of
my life and have always made me feel safe and loved. Thank you for teaching me how to be
confident, persistent and strong.

Filip Niketić, July 2017

viii
Abstract
Designing new concrete members and assessing the strength of existing elements
subjected to in-plane stress state conditions is often conducted by means of Stress Fields (SF)
and Strut-and-Tie Models (STM). These methods are usually accounting for identical load-
carrying mechanisms for design and assessment, even if the goals for these two tasks are not
the same. When designing new elements, the aim is to obtain solutions that are in equilibrium
with external actions, with simple reinforcement layouts while ensuring satisfactory behaviour
at serviceability limit state. When assessing the ultimate strength of existing elements, the goal
is to avoid an unnecessary strengthening and limit the amount of retrofitting. The required
complexity of assessment models depends on the strength requirements, and need to be
gradually refined if the results from current models prove to be insufficient. This refinement
process ultimately leads to increasingly more exact solutions that eventually correspond to the
largest possible strength according to the limit analysis.
This thesis presents various strategies which can be employed to develop SF and STM
suitable for the design and assessment of structural concrete members. The idea of a gradual
model refinement (both for design and assessment) is introduced through practical examples,
while potential challenges related to each solution are indicated and discussed.
Moreover, the accuracy and the generality of exact solutions obtained using Elastic-
Plastic Stress Fields (EPSF) are investigated. To do this, ultimate loads estimated with EPSF
are compared to test results found in the literature. To facilitate further studies by other
researchers, they are available online. The analysis of structural members with insufficient
anchorage and indirectly supported concrete elements with EPSF are presented and discussed.
Furthermore, this thesis focuses on a sensitivity analysis of the EPSF, to investigate the
stability of the results as a function the size, shape and orientation of the finite elements. The
influence of the number of iteration steps on the accuracy of EPSF models is evaluated, and
clear recommendations are provided.
Stress fields based on exact solutions of the theory of plasticity simulate the physical
behaviour of structural concrete members more accurately than current code provisions. On this
basis, a procedure for tailoring partial safety factors for steel and concrete is presented and
discussed. Reduced PSF could potentially be used when assessing the strength of existing
reinforced and prestressed concrete elements, which would (could?) lead to significant cost
reductions in the field of structural maintenance.
To better understand the mechanical origins of concrete compression softening
(important for an accurate application of stress fields), a mechanical model for estimating the

ix
Abstract

effective concrete compressive strength is developed and discussed. Concrete cover spalling,
concrete crushing and crack sliding are taken into account. Special attention is given to the
dowel action of the reinforcement and its effects on the surrounding concrete matrix. The model
is validated using experimental results found in literature. Finally, the pertinence of existing
semi-empirical approaches for determining the effective concrete compressive strength is
evaluated and discussed.

Key-words: strut-and-tie models, stress fields, design, assessment, insufficient

anchorage length, partial safety factors, concrete compressive strength constitutive law,
mechanical model, efficiency factor

x
Résumé
Le dimensionnement d’éléments en béton armé et l’évaluation de la résistance de
structures existantes soumises à un état de contrainte plan sont souvent effectués au moyen de
Champs de Contraintes (CC) et de Modèles de Bielles-et-Tirants (MBT). Ces méthodes
complémentaires sont typiquement utilisées en prenant en compte des mécanismes de transfert
de charge similaires, même si les objectifs principaux du dimensionnement et de l’évaluation
soient fondamentalement différents. Dans le cadre du dimensionnement de nouveaux éléments,
le but est d’obtenir des solutions qui sont en équilibre avec les actions extérieures, ce qui permet
d’obtenir des schémas d’armature simples, tout en garantissant un comportement satisfaisant à
l’état limite de service. Dans le cadre de l’évaluation de la résistance d’un ouvrage existant, le
but est d’éviter tout renforcement inutile. La complexité requise pour les modèles d’évaluation
dépend donc des exigences de résistance de la structure et doit être progressivement raffinée si
les résultats s’avèrent insuffisants. Cette procédure itérative conduit finalement à la solution
exacte selon la théorie de la plasticité, qui est associée à la résistance la plus grande.
Cette thèse présente plusieurs stratégies qui peuvent être employées pour développer
des CC et MBT appropriés tant pour le dimensionnement que pour l’évaluation d’éléments
structuraux en béton armé. Un raffinement graduel de ces modèles est proposé au travers
d’exemples pratiques et les enjeux associés à chacune de ces solutions sont discutés.
Des investigations supplémentaires permettent d’appréhender la précision et
l’applicabilité des solutions exactes obtenues par la méthode des Champs de Contraintes
Élastiques-Plastiques (CCEP). Ceci a été réalisé au travers de comparaisons entre les valeurs
de charges ultimes estimées par cette méthode et celles provenant de la littérature spécifique.
Pour faciliter l’approfondissement de cette étude par d’autres chercheurs, les résultats sont
disponibles en ligne. L’analyse par CCEP traite des éléments structuraux avec ancrage
insuffisants, des éléments à supports indirects ainsi que des éléments précontraints.
Cette thèse présente également une analyse de la stabilité de la solution CCEP en
fonction de la taille, de la forme et de l’orientation des éléments finis utilisés. L’influence du
nombre d’itérations sur la précision des modèles CCEP est évaluée et des recommandations
particulières sont données.
Les champs de contraintes basés sur les solutions exactes de la théorie de la plasticité
simulent le comportement structurel d’éléments en béton armé de manière plus précise que les
normes actuelles. Cela pourrait permettre d’adapter les facteurs partiels de sécurité pour l’acier
et le béton. Une réduction de ces derniers peut être envisagée dans le cadre de l’évaluation

xi
Résumé

d’ouvrages existants en béton armé ou précontraint, ce qui peut conduire à une réduction
significative des coûts associés à la réparation dans le domaine des structures et ouvrages d’art.
Pour mieux comprendre l’origine des mécanismes liés au comportement adoucissant du
béton en compression (primordial pour une application optimale des champs de contraintes),
un modèle mécanique a été développé pour estimer la résistance effective du béton à la
compression. L’éclatement du béton d’enrobage, l’écrasement du béton et le glissement dans
les fissures sont pris en compte. Une attention particulière est donnée à l’effet goujon de
l’armature et ses effets sur la matrice de béton avoisinante. Le modèle est validé par des résultats
expérimentaux provenant de la littérature spécifique. Finalement, la pertinence des approches
semi-empiriques actuelles est évaluée vis-à-vis de la réduction de la résistance à la compression
provoquée par la présence de déformations transversales.

Mots-clés: modèles bielles-et-tirants, champs de contraintes, ancrage insuffisant,

dimensionnement, assainissement, éléments avec appuis indirects, facteurs de sécurité partiels,
résistance à la compression, lois constitutives, modèle mécanique, facteurs d’efficacité

xii
Zusammenfassung
Häufig werden Spannungsfelder und Fachwerkmodelle für die Bemessung und Prüfung
von neuen bzw. bestehenden Stahlbetonbauteilen verwendet. Üblicherweise berücksichtigen
diese Methoden bei der Bemessung neuer und dem Tragwiderstandsnachweis bestehender
Bauwerke dieselben Widerstandsmechanismen, obwohl die verfolgten Ziele unterschiedlich
sind. Bei der Bemessung von neuen Elementen möchte man eine sich mit den äusseren
Belastungen im Gleichgewicht befindende Lösung ermitteln, welche zu einer einfachen
Bewehrung führt und ein gutes Verhalten am Grenzzustand der Gebrauchstauglichkeit
vorweist. Beim Tragsicherheitsnachweis bestehender Bauwerke möchte man hingegen nicht
erforderliche Verstärkungsmassnahmen vermeiden und die Instandsetzungsarbeiten so gering
wie möglich halten. Hierfür müssen, abhängig von der erforderlichen Widerstandsfähigkeit,
unterschiedlich komplexe Modelle verwendet werden: sollten die Ergebnisse von einem Modell
keine ausreichenden Lastreserven vorweisen, dann muss dieses Modell schrittweise verfeinert
werden. Der Verfeinerungsprozess führt zu fortschreitend genaueren Ergebnissen, welche
letztlich dem Widerstandswert gemäss der Grenzwertanalyse entsprechen.
Diese Arbeit beschreibt verschiedene Strategien, die für die Erstellung von
Spannungsfelder und Fachwerkmodellen für den Entwurf und die Bewertung der Tragfähigkeit
von Stahlbetonbauteilen verwendet werden können. Das Prinzip der schrittweisen Verfeinerung
wird anhand praktischer Beispiele erläutert. Für jede Strategie werden möglicherweise
auftretende Schwierigkeiten und Probleme beschrieben und diskutiert.
Die Anwendbarkeit von Elastisch-Plastischen Spannungsfeldern (EPSF) sowie die
Genauigkeit der hiermit gefundenen Lösungen werden untersucht. Die ermittelten Grenzlasten
werden mit Versuchsergebnissen aus der Literatur verglichen. Um zukünftige Studien zu
erleichtern sind alle durchgeführten Simulationen online verfügbar. Die Analyse von
Stahlbetonträgern mit unzureichender Bewehrungsverankerung wurde ebenfalls durchgeführt
und diskutiert.
Die vorliegende Arbeit beinhaltet desweiteren eine Sensibilitätsstudie der EPSF, welche
die Auswirkung von Grösse, Form und Orientierung der finiten Elemente auf die Ergebnisse
untersucht. Auch der Einfluss der Anzahl der Iterationsschritte wird diskutiert, sodass klare
Empfehlungen formuliert werden können.
Spannungsfelder, welche auf den exakten Lösungen der Plastizitätstheorie basieren,
simulieren das physische Verhalten von Stahlbetonbauteilen genauer als derzeitige Norm-
vorgaben. Darauf aufbauend wird ein Verfahren zur Ermittlung der Teilsicherheitskoeffiziente
für Beton und Bewehrungsstahl präsentiert und diskutiert. Tiefere Teilsicherheitskoeffiziente

xiii
Zusammenfassung

könnten für die Auswertung von bestehenden Stahlbeton- und Spannbetonbauteile verwendet
werden, was zu bedeutend niedrigeren Kosten für die Instandsetzung und Instandhaltung von
Bauwerken führen kann.
Um die mechanische Ursache der Druckfestigkeitsabnahme im Nachbruchbereich
besser zu verstehen, wurde ein mechanisches Modell für die Schätzung der wirksamen
Betondruckfestigkeit entwickelt. Dieses ist für die Anwendung der Spannungsfelder
ausschlaggebend. Das Abplatzen der Betondeckung, der Betondruckbruch, sowie das Gleiten
von Rissen werden hierbei berücksichtigt. Besondere Aufmerksamkeit wird der Dübelwirkung
der Bewehrung geschenkt, sowie dessen Einfluss auf die umliegende Betonmatrix. Das Modell
wird anhand von Versuchsergebnissen aus der Literatur validiert. Schliesslich wird die Validität
existierender semi-empirischer Ansätze zur Ermittlung der wirksamen Betondruckfestigkeit
gewertet und diskutiert.

Stichwörter: Fachwerkmodelle, Spannungsfelder, Entwurf, mechanisches Modell,

unzureichende Bewehrungsverankerung, Bewertung, Teilsicherheitsbeiwerte, Effizienzfaktor,
Betondruckfestigkeits, konstitutives Stoffgesetz

xiv
Riassunto
La progettazione di nuove strutture e la stima della resistenza ultima per strutture
esistenti soggette ad uno stato di sforzi membranali sono spesso condotte mediante
modellazioni che utilizzano campi di tensione o sistemi puntone-tirante. Questi metodi sono
spesso utilizzati sia per il dimensionamento che per la verifica ipotizzando legami costituitivi e
meccanismi di trasmissione sostanzialmente simili malgrado gli scopi possano essere differenti.
Nel caso del dimensionamento di nuove strutture, l’obbiettivo del calcolo è l’ottenimento di
soluzioni che siano in equilibrio con le forze esterne, con disposizioni semplici delle armature
e assicurando allo stesso tempo un comportamento soddisfacente allo Stato Limite di Esercizio.
Nel caso della verifica della resistenza ultima di strutture esistenti, lo scopo è, invece, evitare
rinforzi strutturali non necessari.
La complessità richiesta nei modelli di verifica dipende dai requisiti di resistenza, e
necessita di un raffinamento graduale se i risultati ottenuti coi modelli attuali forniscono risultati
insufficienti. Questo processo di raffinamento porta a soluzioni sempre piu’ esatte che
corrispondono, infine, alla massima resistenza ammissibile calcolata aitraverso l’analisi limite.
Questa tesi presenta varie strategie che posso essere adottate in caso di modellizzazioni
di campi di tensione e modelli puntone-tirante nel caso di dimensionamento e verifica di
strutture in calcestruzzo armato. L’idea di un graduale raffinamento del modello (sia per il
dimensionamento che per la verifica) verrà introdotta mediante esempi pratici. Allo stesso
tempo, potenziali raffinamenti legati ad ogni soluzione saranno elencati e discussi.
Saranno inoltre investigate l’accuratezza e la generalità di soluzioni esatte ottenute con
Campi di Tensione Elasto-Plastici (CTEP). A tal fine, i carichi di rottura ottenuti mediante tali
campi saranno confrontati con risultati sperimentali presi dalla letteratura. Al fine di facilitare
futuri studi, tali risultati sono stati inseriti in un database disponibile online. Saranno presentati
e discussi, inoltre, analisi mediante CTEP di elementi strutturali con ancoraggi insufficienti o
indirettamente appoggiati.
Questa tesi si focalizza, anche, su analisi di sensitività dei campi di tensione elasto –
plastici, al fine di investigare la stabilità dei risultati in funzione della taglia, forma e
dell’orientamento degli elementi finiti. Infine, sarà analizzata l’influenza del numero di
incrementi di carico e chiare raccomandazioni saranno fornite a riguardo.
Campi di tensione basati su soluzioni esatte della teoria della plasticità simulano il
comportamento fisico di elementi strutturali in calcestruzzo in maniera piu’ accurata che gli
attuali codici. Per tale motivo, una procedura di calibrazione dei fattori parziali di sicurezza per
il calcestruzzo e l’acciaio sará presentata e discussa. I fattori di sicurezza parziali ridotti possono

xv
Riassunto

essere utilizzati nel caso di verifica della resistenza di strutture esistenti o in elementi strutturali
precompressi. Tali riduzioni potrebbero portare, a significativi abbassamenti dei costi nel
campo della manutenzione strutturale.
Al fine di comprendere al meglio gli aspetti constituivi legati al comportamento di
softening del calcestruzzo in compressione (importante per l’applicazione dei campi tensionali),
un modello meccanico per la stima della resistenza effettiva a compressione del calcestruzzo
sarà sviluppato e trattato in dettaglio. Lo “spalling” del copriferro, lo schiacciamento del
calcestruzzo per elevati sforzi di compressione e lo scorrimento della fessura saranno presi in
considerazione.
Un’attenzione speciale sarà data al contributo dell’effetto spinotto prodotto dalle
armature e ai suoi effetti sulla matrice di calcestruzzo circostante. Il modello sarà validato
mediante risultati sperimentali presenti in letteratura. Infine, sarà valutata la pertinenza degli
approcci semi-empirici esistenti per tenere conto della riduzione della resistenza a
compressione per softening, in presenza di deformazioni trasversali.

Parole-chiave: modelli puntone-tirante, campi tensionali, dimensionamento, verifica,

lunghezza di ancoraggio insufficiente, fattori parziali di sicurezza, legge costituitva della
resistenza a compressione, modello meccanico, fattore di efficienza

xvi
Сиже
Пројектoвање нових и процена носивости постојећих елемената у равном стању
напона, често се врши коришћењем модела Притиснутих и Затегнутих Штапова (ПЗШ),
односно развојем Поља Напона (ПН) читавог носача. Иако коначни циљ самог процеса
пројектовања и процене стања није идентичан, обе методе (ПЗШ и ПН) се често
употребљавају на исти начин током обављања ових принципијeлно другачијих задатака.
Када говоримо о пројектовању нових конструкција, задовољавајуће решење произилази
из модела који је у равнотежи са нанетим оптерећењем и уједно даје једноставне планове
арматуре који гарантују задовољавајуће понашање елемената у граничном стању
употребљивости. Током процене носивости, модели имају за циљ да избегну непотребна
ојачавања постојећих објеката или смање обим неопходних интервенција. Комплексност
ових модела зависи од захтеваног нивоа носивости. Почевши од једноставних решења,
комплексност модела треба постепено повећавати све док се гарантована отпорност не
нађе изнад захтеване. Ово усложњавање је могуће све до достизања егзактног решења,
које по принципима теорије пластичности даје максималну теоретску носивост једног
елемента.
У оквиру ове тезе представљени су различити процеси развоја ПН као и модела
ПЗШ оптималних за пројектовање нових или процену отпорности постојећих елемената.
Идеја постепеног усложњавања модела је показана кроз практичне примере. Уједно,
потенцијални проблеми који могу настати као последица неправилног моделирања су
назначени и објашњени.
Применљивост и прецизност егзактних решења која су добијена применом методе
Еласто-Пластичних Поља Напона (ЕППН) је истражена кроз поређење процењених и
експериментално измерених вредности лома носећих елемената. Како би се обезбедила
апсолутна транспарентност представљеног истраживања и уједно олакшао будући рад у
овом домену, сви модели су доступни онлајн. Посебна пажња посвећена је елементима
са неадекватно анкерисаном арматуром и индиректним ослонцима.
У циљу анализе стабилности решења добијених применом ЕППН (која су
заснована на Методи Коначних Елемената - МКЕ), одређени носачи су моделирани
помоћу КЕ различите величине, облика и орјентације. Утицај броја итерација на
прецизност коначног решења је испитана и препоруке за практичну примену ЕППН су
јасно назначене.
Чињеница да еласто-пластична поља напона приказују расподелу унутрашњих
сила далеко прецизније од препорука актуелних правилника за бетонске конструкције,

xvii
Сиже

омогућава редукцију Парцијалних Коефицијента Сигурности (ПКС) бетона и челика.

Посебна метода која дозвољава прерачунавање ПКС узимајући у обзир тачност ЕППН је
представљена и дискутована. У случају процене стања постојећих конструкција,
редукција ПКС би водила ка потенцијално значајним уштедама.
Ради бољег разумевања процеса омекшавања бетона под оптерећењем (веома
битном параметру од којег зависи прецизност поља напона) посебна пажња посвећена је
развоју маханичког модела за процену његове ефективне чврстоће, који директно узима
у обзир три различита механизма лома (дробљење, клизање и одвајање заштитног слоја
бетона). Поред тога, кривљење арматуре услед отварања пукотина које су управне на
правац шипки је узет у обзир. Механички модел је искоришћен како би се измоделирали
армирано бетонски панели који су оптерећени на смицање, и добијени резултати су
упоређени са експериментално измереним вредностима. Поред тога постојећи полу-
емпиријски модели за омекшавање бетона су упоређени са резултатима механичког
приступа ради даље валидације.

Кључне речи: модел притиснутих и затегнутих штапова, поље напона,

пројектовање, процена, недовољна дужина анкерисања арматуре, парцијални
коефицијенти сигурности, ефективна чврстоћа бетона на притисак, механички модел

xviii
Table of content
Foreword i
Acknowledgements vii
Abstract ix
Résumé xi
Zusammenfassung xiii
Riassunto xv
Сиже xviii
Notation xxiii
Upper Latin characters xxiii
Lower Latin characters xxiii
Upper Greek characters xxiv
Lower Greek characters xxiv
Subscript xxv
Upper Latin characters xxv
Lower Latin characters xxv
Greek characters xxvi
Numbers xxvi
Acronyms xxvi
Chapter 1: Introduction 1
1.1 Aims of the research 3
1.2 Structure of the thesis 4
1.3 Scientific contributions of the thesis 5
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment of
structural concrete members 7
2.1 Introduction 7
2.2 State of the art on strut-and-tie models and stress fields 13
2.3 Suitable approaches for designing new structural elements with stress fields 26
2.3.1 Design models inspired by the existing cases 26

xix
Table of content

2.3.2 Design models based on deviated thrust lines 28

2.3.3 Design models inspired by elastic uncracked stress fields 31
2.3.4 Design models inspired by cracked stress fields 32
2.4 Suitable approaches for assessing the ultimate strength of structural concrete elements with
stress fields 34
2.4.1 Assessing the ultimate strength using a rigid-plastic stress field approach 34
2.4.2 Assessing the ultimate strength using elastic-plastic stress field approach 40
2.5 Collected database and validation of the elastic-plastic stress fields 42
Chapter 3: Advanced modelling of structural concrete members with EPSF 47
3.1 Sensitivity analysis of EPSF method with respect to its finite element implementation 47
3.1.1 Investigating the influence of finite element size on EPSF analysis 48
3.1.1.1 Structural concrete members subjected to pure compression 49
3.1.1.2 Structural concrete members subjected to pure tension 50
3.1.1.3 Structural concrete members subjected to pure shear 51
3.1.1.4 Structural concrete beam under four-point-bending 52
3.1.2 Investigating the influence of finite element shape on EPSF analysis 55
3.1.3 Investigating the influence of finite element orientation on EPSF analysis 57
3.1.4 Convergence and required number of iterations for an EPSF analysis 59
3.2 Modelling elements with insufficient anchorage length using EPSF 61
3.3 Analysis of indirectly supported structural concrete members using EPSF 66
3.4 Tailoring partial safety factors suitable for EPSF method 72
Chapter 4: Effectiveness factor for concrete compressive strength accounting for the
presence of cracks and interaction with the reinforcement 81
4.1 Introduction and state of the art 81
4.1.1 Concrete compressive strength brittleness factor – ηfc 83
4.1.2 Concrete softening caused by imposed transversal strains – ηε 86
4.1.3 Concrete softening caused by the presence of post-tensioning ducts 94
4.1.4 Critique of the multiplicative approach for estimating effective concrete
compressive strength 99
4.2 Mechanical model for compression softening of reinforced concrete panels 102
4.2.1 Stages of mechanical behaviour 102
4.2.1.1 First stage: Linear elastic response 102
4.2.1.2 Second stage: Cracked behaviour 104
4.2.1.3 Third Stage: Behaviour after yielding 104
4.2.2 Plastic strength of rebars subjected to doweling and elongation 105

xx
 Table of content

4.2.3 Concrete strength accounting for the presence of initial cracks and rebars 111
4.2.3.1 Strength of the damaged struts 113
4.2.3.2 Strength of the undamaged struts 117
4.2.4 Parameters governing the compressive strength of concrete 118
4.3 Numerical solving procedure for the proposed compression softening model 120
4.3.1 Cracking of the panel 120
4.3.2 Onset of yielding or early concrete crushing 120
4.3.3 Failure of the panel 123
4.4 Experimental validation and comparison to available methods 124
Chapter 5: Conclusions and future research 131
5.1 Conclusions 131
5.1.1 Regarding the application of STM and SF as tools for design and assessment 131
5.1.2 Regarding the advanced modeling of structural concrete with EPSF method 133
5.1.3 Regarding the effectiveness factor for concrete compressive strength 134
5.2 Future research 135
References 137
Appendix
Curriculum vitae

xxi
Notation
Following notation is used in this thesis:

Upper Latin characters

A : area
D : aggregate diameter
E : young’s modulus of elasticity
F : applied force
M : bending moment
N : normal (axial) force
P : pole of rotation
Q : applied load
R : reaction force; resistance
S : strut
T : tie
V : shear force

Lower Latin characters

a : general dimension of a structure
b : thickness
c : concrete cover
d : differentially small section
f : material strength
h : size of a panel in z-direction
l : size of a panel in x-direction ; length
m : normalized moment
n : normalized normal force; number of layers; number of initial cracks in one half

xxiii
Notation

of an analysed panel
n’ : number of damaged struts in a critical section of analysed panel
q : applied distributed load; relative amount
s : spacing between two consecutive cracks; spacing between two neighbouring
rebars
u : displacement in horizontal direction
v : displacement in vertical direction
w : crack width

Upper Greek characters

 : one half of the total rebar elongation between the two plastic hinges
Ø : rebar diameter
Φ : normal distribution function
Ψ : rotation

Lower Greek characters

α : rebar tilting angle due to the doweling
β : reliability index
γ : shear strain; partial safety factor
δ : slip of a plastic hinge
δ’ : ratio between the external rebar diameter and the sum of the double effective
concrete cover thickness and the external rebar diameter
ε : strains
η : effectives factor
θ : direction of the principal compressive strains (stresses)
λ : integration factor
ν : applied stress on the panel; effectiveness factor
ξ : normalized slip of the plastic hinges in the doweling
ρ : reinforcement ratio
σ : stress
τ : shear/bond stress
φ : friction angle
ω : local stress field disturbance ratio

xxiv
 Notation

Subscript

Upper Latin characters

D : duct
DS : damaged struts
EPSF : according to elastic-plastic stress field
MODEL : according to proposed model
U : ultimate
UDS : undamaged struts
V : shear force

Lower Latin characters

a : width of the local disturbance zone
anch : anchorage
b : bond
c : concrete
cc : confined/confining concrete
ce : effective concrete
cr : crack
cte : effective tensile concrete strength
ctm : mean tensile concrete strength
conf,1 : direction perpendicular to the direction of confining concrete stress
conf,2 : direction parallel to the direction of confining concrete stress
d : design values
dow : doweling
eff : effective
ext : external
fc : accounting for the brittle behavior of the concrete
k : characteristic values
m : mean
max : maximal
min : minimal
model : model
nom : nominal values

xxv
Notation

p : inside a plastic hinge

r : reduced
s : steel; accounting for the presence of steel
s,1 : at the location of a crack
s,2 : at the mid-point between the two cracks
t : tensile; time
test : tested
ult : refers to stain (stress) state at ultimate limit state
x : in x-direction; height of the compressed zone
xz : in xz-plane
w : web
y : yielding
z : in z-direction

Greek characters
ε : accounting for induced transverse strains
τ : accounts for the presence of the initial cracks in concrete

Numbers
0 : reference value; refers to concrete compressive strength
1 : direction of the principal tensile strains (stresses)
3 : direction of the principal compressive strains (stresses)

Acronyms
ACI : American Concrete Institute
CoV : Coefficient of Variation
EC : Euro Code
EPSF : Elastic-Plastic Stress Field
CCEP : Champs de Contraintes Élastiques-Plastiques
CTEP : Campi di Tensione Elasto-Plastici
ЕППН : Еласто-Пластичнa Поља Напона
FE : Finite Element
GSP : Global Safety Format
LoA : Level of Approximation

xxvi
 Notation

MC : Model Code
MCFT : Modified Compression Field Theory
PSF : Partial Safety Format
RC : Reinforced Concrete
RPSF : Rigid-Plastic Stress Field
SF : Stress Field
SLS : Serviceability Limit State
std : Standard deviation
STM : Strut-and-Tie Model
ULS : Ultimate Limit State

xxvii
Chapter 1: Introduction
Application of reinforced and prestressed concrete in civil engineering has been
steadily increasing over the past century, ever since Hennebique [101] patented a
manufacturing process for ribbed reinforced concrete slabs, and used it in practice by the late
nineteenth century. Purely intuitive structural concrete analysis gradually evolved into
mechanically based models by the end of the fifties, when the development of exact solutions
according to the theory of plasticity took place. The increased demands for the residential,
infrastructural and industrial facilities caused by the global urbanization popularized the
application of concrete as reliable and accessible building material.
The end of the twentieth century marked the end of the designed life span for many
concrete structures, which now required assessment to reinsure safety of further usage.
Retrofitting the existing objects became a frequent engineering task, since the available
construction space was decreasing and the idea of sustainable building gained on importance.
At the same time, the general knowledge on mechanical behaviour of reinforced and
prestressed concrete members increased and the current codes of practice have been updated.
Compared to their previous versions, more conservative criteria for acceptable structural
behaviour have sometimes been imposed, which resulted in structural strengthening.
However, the fact that the design models usually govern conservative results compared to
reality (some load-carrying mechanisms are neglected), means that the strengthening can
often be avoided by increasing the accuracy of the applied methods.
Limit analysis provides a consistent frame suitable for the design and the assessment
of structural concrete members. Lower-bound theorem can be used to develop models that
give safe estimates of element’s strength, and as such they can be used for the structural
design. On the other hand, upper-bound theorem governs unsafe results by definition.
Nevertheless, in combination with the lower-bound theorem it can be used to obtain the exact
solution which, according to the theory of plasticity, gives the highest possible strength of an
analysed element and is therefore suitable for structural assessment.
This thesis focuses on the differences between the models that can be used when
designing new structural concrete members subjected to plane stress state and assessing the
strength of existing ones. It aims to show that the difference in principal goals (design or
assessment) dictates the form and accuracy of the final solution. Various procedures for
designing and assessing the strength of structural concrete members are presented and
discussed, using a deep beam with an opening as an example (originally introduced by
Schlaich et al. [107]).

1
Chapter 1: Introduction

Special attention is given to the Elastic-Plastic Stress Field (EPSF) approach, which
automatically accounts for all potential load-carrying mechanisms at ultimate limit state, thus
giving an exact solution according to the limit analysis. Results obtained using the EPSF are
compared to experimentally-measured values found in the literature in order to prove its
accuracy and generality. Based on this, a database containing 315 reinforced and prestressed
concrete members is created and placed online.
Robustness of the EPSF method has been evaluated by conducting a detailed
sensitivity analysis. Influence of finite element size, shape and orientation on the convergence
of EPSF are estimated and discussed.
Additional focus is placed on the development of EPSF suitable for investigating
members with insufficient anchorage length and elements with indirect supports (which is
particularly useful when analysing bridge diaphragms). In order to profit from increased
accuracy of the EPSF models, a procedure for tailoring partial safety format is presented and
discussed.
If the strength of an element is governed by the concrete, accuracy of the exact
solution depends on the pertinence of the approach used to estimate the effective concrete
compressive strength. This has been observed in the late seventies, and since then various
semi-empirical softening equations have been proposed by different authors (Vecchio and
Collins [120], Hsu [47], Pang [94], Kaufmann [52, 53], Muttoni [82, 89], Hars [41], etc.).
However, none of the existing models take directly into account the influence of the different
failure modes (crushing of the struts, spalling of the concrete cover and sliding of the struts
along the weakest planes). In order to overcome this, a mechanically based procedure for
estimating the effective concrete compressive strength has been developed and presented. The
model showed satisfactory results when compared to experimentally measured values.
Finally, the results of the presented mechanical approach were used to justify the accuracy
and the consistency of the existing semi-empirical approaches.

2
 1.1 Aims of the research

1.1 Aims of the research

The research presented in this thesis aims to:
 Show and compare various design procedures for reinforced and prestressed concrete
members with discontinuity regions;
 Show and compare various procedures for estimating the maximal load-bearing
capacity of reinforced and prestressed concrete members with discontinuity regions;
 Underline the main differences between the procedures suitable for design and
assessment of structural concrete members;
 Assemble a database of tested structural concrete members found in literature,
estimate their ultimate strength using models based on the exact solution of the theory
of plasticity (employing the EPSF method);
 Conduct a detailed sensitivity analysis of the EPSF method, with respect to its finite
element implementation;
 Propose a procedure for taking into account the insufficient reinforcement anchorage
in EPSF models;
 Propose a procedure for EPSF analysis of indirectly supported structural concrete
members;
 Propose a procedure for tailoring the partial safety factors for EPSF method, taking
into account its increased accuracy;
 Investigate the influence of concrete crushing, concrete cover spalling and sliding on
effective concrete compressive strength;
 Investigate the interaction between the concrete compressive strength and the
formation of cracks appearing at different load-stages;
 Investigate the interaction between the concrete compressive strength and doweled
rebars in structural panels failing in shear;
 Develop a mechanically based model for estimating the effective concrete
compressive strength and validate it using the results from experimental campaigns
found in literature.

3
Chapter 1: Introduction

1.2 Structure of the thesis

The thesis is organized in the following manner:
 Chapter 2 consists of results which were published in the 112th volume of ACI
Structural Journal (issue number 5) in September - October of 2015, titled “Design
versus Assessment of Concrete Structures Using Stress Fields and Strut-and-Tie
Models”, written by Aurelio Muttoni, Miguel Fernández Ruiz, and Filip Niketić. The
chapter gives a brief introduction and state-of-the-art on structural analysis using strut-
and-tie models and stress-fields. It also presents various strategies for design and
assessment of structural concrete members based on limit analysis. Finally, it
compares the results of the highest level of accuracy (exact solutions obtained with the
EPSF method) to an online database;
 Chapter 3 consists of results which were published in October 2016 as a part of the
FEDRO (Federal Roads Office) report number 680, titled “Assessment of Existing
Structures Based on Elastic-Plastic Stress Fields and Modelling of Critical Details and
Investigation of the In-Plane Shear Transverse Bending Interaction”, written by
Aurelio Muttoni, Miguel Fernández Ruiz, Filip Niketić and Marie-Rose Backes. The
report was a part of the research project AGB 2009/009, requested by the AGB bridge
research group. It presents the details of the finite elements sensitivity analysis
performed on the EPSF method. It also shows the modelling procedures for elements
with insufficient reinforcement anchorage length, as well as indirectly supported
structural concrete members. Finally, it presents a procedure for tailoring partial safety
factors for steel and concrete that could be used with the EPSF method;
 Chapter 4 consists of results which were accepted for publishing on the 7th of July
2017 by the Magazine of Concrete Research under the article number MACR-D-17-
00077R1. The work was titled: “Response of RC panel accounting for crack
development and its interaction with rebars”, and was written by Eckart Hars, Filip
Niketić and Miguel Fernández Ruiz. The chapter gives an introduction and state-of-
the-art on various factors for estimating the effective concrete compressive strength. It
presents a mechanically based model for the reduction of concrete compressive
strength. The model takes into account the effects of rebar doweling. Finally, it
compares the obtained results to the semi-empirical softening equation of
Vecchio and Collins [120], and validates the consistency of the proposed procedure
using a database of tested reinforced concrete panels found in literature.
 Chapter 5 gives conclusions based on the work presented in this thesis and proposes
recommendations for future research.

4
 1.2 Structure of the thesis

1.3 Scientific contributions of the thesis

The scientific contributions of this thesis are:
 Clarifying the difference between the development procedure of strut-and-tie models
and stress fields used for design and ultimate strength assessment of structural
concrete members;
 Assembling an open-source database containing 315 reinforced and prestressed
concrete members from 30 different experimental campaigns and the corresponding
EPSF models;
 Performing a detailed sensitivity analysis of the EPSF method and giving clear
recommendations on how the models should be conceived;
 Presenting and validating a strategy for taking into account the insufficient anchorage
length of the reinforcement in EPSF models;
 Presenting and validating a strategy for modelling crossed reinforced and prestressed
concrete members with EPSF method;
 Presenting a strategy for tailoring partial safety factors which can be applied on EPSF
models;
 Presenting and validating a mechanically based model for estimating the concrete
compressive strength effectiveness factor, which takes into account different failure
mechanisms occurring in the concrete;
 Presenting a procedure for taking into account dowelling for the reinforcement and
incorporating it into the concrete compressive strength effectiveness factor;
 Validating the accuracy of semi-empirical approach of Vecchio and Collins [120] by
comparing it to mechanically cased approach developed in this thesis.

5
Chapter 2: Strut-and-tie models and stress
fields as tools for design and
assessment of structural concrete
members
The work presented in the following chapter is based on a technical paper titled
“Design versus Assessment of Concrete Structures Using Stress Fields and Strut-and-Tie
Models”, written by Aurelio Muttoni, Miguel Fernández Ruiz, and Filip Niketić. The article
was published in ACI Structural Journal, volume 112, number 5 as a part of the September -
October 2015 issue of the paper.
The contributions of Filip Niketić to creation of this publication involved:
1. Developing the examples of stress fields and strut-and-tie models suitable for
designing a deep beam with an opening;
2. Developing the examples of stress fields and strut-and-tie models suitable for
assessing the ultimate strength of a deep beam with an opening;
3. Validating the generality and accuracy of the EPSF method by assembling a
database containing 150 structural elements (the database was later extended to
a total of 315 structural elements presented in this thesis);
4. Making a web page which contains the basic geometrical and mechanical
properties of analysed elements, as well as the corresponding EPSF models
which can be downloaded and recalculated in ICONC

2.1 Introduction
Structural analysis of concrete elements in modern civil engineering is widely done
using Stress Fields (SF) and Strut-and-Tie Models (STM). These approaches have been
incorporated in most codes of practice in the last decades. Given a fact that up to recently the
demand for assessing the ultimate strength of existing structural concrete members was
relatively low, means that most of the code recommendations were mainly directed towards
the design. However, this trend is gradually changing as retrofitting of the existing structures
becomes more and more common due to:
1. Limited space for conceiving new structures in heavily populated areas;

7
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

2. Application of more rigorous criteria for the acceptable behaviour of structural

elements by the valid codes of practice.
Consequently, some of the codes (such as MC 2010 [34] for example) have
incorporated the concepts related to life cycle design, or are providing recommendation for
the structural evaluation and rehabilitation (which is the case with ACI 364 [2], ACI 437 [3]
and SIA 269 [109]).
Today stress fields and strut-and-tie models are often used in the same manner when
designing new and assessing the strength of existing structures. Developing a suitable model
for these two conceptually different tasks should not necessarily be performed following the
same approach. The aim of the models used for designing is to derive simple solutions which
are in equilibrium with the external load, and are able to provide safe solutions at the Ultimate
Limit State (ULS). Out of multiple options, the ones showing acceptable behaviour at
Serviceability Limit State (SLS) should be selected. On the other hand, the aim of SF and
STM used for assessment of the existing structures is to avoid or minimize any element
strengthening. In this case it is usually not necessary to model structural behaviour at SLS,
since the structure itself has already been in usage and its current condition can be validated
through thorough in-situ inspection. Given the fact that the aim of the models used for design
and assessment is clearly different, indicates that the process of their development as well as
their final form should not always be the same. It is therefore necessary to clearly distinguish
between these two modelling techniques, and provide the engineers with the
recommendations for their development.

(a) σc V
Fan region bw·z·sinθ·cosθ

z·cosθ
V
σc z
c M
θ
σc

z/(2tanθ) z/(2tanθ) Constant stress inclination

V
(b)

V/sinθ V
M z

θ’ θ θ

cotθ’ c cotθ
+
2z 2
V

Figure 2.1: (a) Stress field of a simple beam; (b) corresponding strut-and-tie model
Stress field and strut-and-tie methods share many analogies. They are complimentary
tools which should be used in parallel weather designing new or assessing the strength of
existing structures. Stress field method [34] allows determining a licit stress state of a member

8
 2.1 Introduction

subjected to external actions (as presented in Fig.2.1a). Combining the internal forces with
the geometrical properties of the cross-section, allows one to determine the direction (θ) and
the intensity (σc) of the principal compressive stresses in the concrete. Combining them with
the number, position, and diameter of the rebars on the other hand determines the stress state
inside the reinforcement. Strut-and-tie model corresponding to a specific stress field,
represents nothing else but its resultant (refer to Fig.2.1b). It is easier to find an equilibrium
between the applied actions and the internal forces in this manner, and to determine which
regions of the analysed element are subjected to tension or compression. The inclination of
the struts and ties and the intensity of the forces inside them can be determined through
integration of the stress field over areas indicated in Fig.2.1. This becomes more obvious
when comparing the inclination of the strut which is closest to the support to the inclination of
the other struts in the beam. As it can be seen θ’ is not parallel to θ, since it corresponds to the
resultant of the fan region, and not the compressive field with constant inclination.
When searching for an optimal solution for design and assessment, the final goal is
formally the same: reduce the cost while ensuring a satisfactory structural behaviour. With
this in mind, the appropriate SF and STM for structural design leads towards the solutions that
minimize the amount of necessary material (both steel and concrete), and are easy to develop
and build (requiring little time and labour force). The two goals do not necessary have the
same direction, and sometimes more material is required for achieving simpler reinforcement
layouts. This means that the optimal solution for design will always depend on the current
market. The optimization process of the models used for strength assessment on the other
hand is more straightforward, since the cost are always reduced by minimizing or possibly
avoiding any structural strengthening. However, this does not mean that the model which
accounts for all potential load-carrying mechanisms of an element is the appropriate solution.
It is rather the simplest one which proves that the structure has sufficient resistance with
respect to the required design loads. Only in case a model does not provide sufficient
structural safety, it should be changed to account for the additional load-carrying mechanisms.
According to the limit analysis, this process can be repeated until exact solution has been
obtained. Only if the exact solution shows insufficient resistance compared to the design
loads, the structure should be strengthened accordingly.
This concept of gradual model refinement is known as the Levels-of-Approximation
(LoA) approach and was introduced in MC 2010 [34] by Muttoni et Fernández-Ruiz [85,86].
As presented in Fig.2.2a the lower LoA models take less time to develop but are more
conservative compared to the actual element resistance. As the LoA increases, so does the
models computational time and accuracy.
Fig.2.2b gives an example of the correct application of the LoA approach, by
calculating the shear resistance of a rectangular cross section according to section 7.3.3.3 of
MC 2010 [34]. Assuming that this is an existing element whose required resistance (Vreq) is
indicated in the graph using a dashed line (see Fig.2.2b), the modelling process should start
using the simplest solution – the 1st LoA where the minimal SF inclination is limited to 30°
and the concrete compressive strength effectiveness factor is 0.55. As presented it Fig.2.2b,
the obtained resistance does not satisfy the imposed requirements, which means that the 2nd

9
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

LoA model should be applied. According to MC 2010 [34], this implies more accurate
estimation of the minimal SF inclination angle as well as a re-evaluation of the concrete
compressive strength effectiveness factor. As presented in Fig.2.2b, the necessary
requirements have now been met, which means that there is no need to develop models of the
3rd and the 4th LoA.

(a) (b) 2.5

Actual resistance of an element
Ø8@200
2
Potential model th
4 LoA Vc,2nd
accuracy 2nd LoA d=400 mm
1.5

V/Vreq [-]
Accuracy

3rd LoA Vc,1st bw=100 mm

d fc=30 MPa
2nd LoA 1.0 Vreq fy=550 MPa
4Ø20
1st LoA 1st LoA Vs
0.5
bw
0
Time 20 25 30 35 40 45
θ [°]

Figure 2.2: Levels-of-Approximation (LoA) approach: (a) accuracy of a model in function of the
time necessary for it’s development; (b) shear strength of a given cross-section
estimated with two LoA according to MC 2010 [34]
Stress field and strut-and-tie methods are in agreement with limit analysis and provide
a consistent frame for the design and assessment of structural concrete. Since both methods
respect the yield condition at all points and are in equilibrium with applied loads, SF and
STM can be treated as lower-bound solutions. Drucker [27], described the application of the
lower-bound theorem to structural concrete in the following manner: “If an equilibrium
distribution of stress can be found in the concrete and the steel which is nowhere tensile in the
concrete and is everywhere at or below yield, the structure will not collapse or will just be at
the point of collapse”. It can therefore be seen that if a model respects the lower-bound
theorem of the limit analysis, the estimated strength of the corresponding element will be
conservative, and the members will have sufficient ductility. In other words, SF and STM
based on the lower bound theorem are suitable tools for the design of structural concrete.
A few examples of the lower-bound solutions are presented in Fig.2.3 b to d, where
stress fields (given in a form of moment diagrams) are developed for a double clamped beam
subjected to uniformly distributed load. Each of the models is in equilibrium with the applied
loads (refer to M-diagrams presented in red) and the ultimate strength of concrete and steel is
never exceeded along the beam (MR-diagrams always surpass the M-diagrams), which makes
them licit lower-bound solutions. SF which show a positive behaviour at SLS while assuring
sufficient safety at ULS should be selected when designing a structural concrete member (as
described by Muttoni et al. [89]). Given a fact that placing reinforcement only in the upper
part of the beam’s cross-section (see Fig.2.3b) or only in its lower part (see Fig.2.3c) leads to
unacceptably wide cracks in the middle of the beam or close to the supports, means that the
solution presented in Fig.2.3d is the most acceptable one (shows the best performance at
SLS). Whether it is an optimal solution or not depends on the amount of applied material,
simplicity of the final reinforcement layout, and the amount of robustness required from the
newly designed structure. Given a fact that in Fig.2.3d MR-diagram never touches the M-

10
 2.1 Introduction

diagram, indicates that there is some margin for the reduction of applied materials. Whether
this reduction is necessary to make or not is a question that should be answered by the
responsible designer.

(a) q

ULS SLS

(b) MR
2
q·l M
8

(c) M
q·l2
8
MR

(d) MR

q·l2 M
8

Figure 2.3: Lower-bound solutions (suitable for design): (a) Geometry and loading properties of a
double clamped beam; (b) to (d) moment diagrams and corresponding deformed shapes
resulting from reinforcement being placed:(b) on the top of the beam;(c) at the bottom
of the beam;(d) both on the top and at the bottom of the beam
Limit analysis has also an upper bound theorem which can be used to calculate the
strength of the existing structural concrete elements. According to Drucker [27]: “The
structure will collapse if there is any compatible pattern of plastic deformation for which the
rate of work of the external loads exceeds the rate of internal dissipation”. Once more, there
are multiple solutions which can satisfy the upper-bound theorem and they are non-
conservative by definition. Referring to the double clamped beam from Fig.2.4a, two
mechanisms (amongst infinite number of solutions) presented in Fig.2.4d and c can be
assumed to determine an upper bound of its strength. Both are kinematically admissible and
correspond to SF which is in equilibrium with the applied loads (refer to M1 and M2-
diagrams). However, estimated resistances of the same element using these two deferent
mechanisms are not equal. The solution presented in Fig.2.4b governs higher ultimate loads
compared to the one from Fig.2.4c. In addition to this SF from Fig.2.4b violates the yield
condition in the middle of the beam. Therefore, such mechanism cannot be accepted as a licit
solution and is therefore disregarded. On the other hand, the mechanism of Fig.2.4c fully
respects the yield condition along the beam yet allows the development of the assumed
kinematics. In addition to this it is in equilibrium with applied loads, which means that it
respects both lower and upper-bound theorems of the limit analysis. Such solution is called
the exact solution. According to the limit analysis, it is considered to be the best potential SF
which can develop in an existing structure, since it provides the largest strength of all
potential safe (lower bound) solutions. When assessing the strength of an existing element,
exact solutions present the most refined (highest LoA) models, which minimize the cost of

11
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

structural strengthening either by avoiding it or making the minimal necessary interventions

in the critical regions.

(a)

Mechanisms ULS
qult,1
(b)
MR qult,1·l2 M1
8

qult,2
(c)
MR qult,2·l2 M2
8

Figure 2.4: Lower-bound solutions (suitable for the ultimate strength assessment) (a) Geometry
properties and the reinforcement layout of a double clamped beam (b) and (c) assumed
mechanisms and corresponing moment diagrams resulting from a: (b) four-hinge
mechanism; (c) three-hinge mechanism
It can be concluded that the lower-bound solutions are suitable for design of structural
concrete members, but can also be used as tools for safe assessment of the strength of existing
structures. The upper-bound models on the other hand, generally lead towards unsafe
solutions, but they can be optimized in order to find an exact solution. Such solution can be
used as a powerful tool when estimating the strength of existing structural concrete element,
in which case it requires the analysis of the highest LoA. Exact solution can also be used for
designing, in case it is required to determine the minimal amount of building materials
necessary to support the applied loads. However, such level of accuracy is not necessarily
positive in designing, since it leaves little space to compensating for the unexpected actions
during the construction and over the lifetime of a structure.
This chapter will focus of presenting various strategies which can be used for
developing suitable stress fields and strut-and-tie models for design and assessment of
structural concrete members, through practical examples and references. The concepts are
explained from a general point of view even though structural design and strength assessment
are performed on elements subjected to in-plain loading. The accuracy of limit analysis is
finally compared to available test data showing the consistency of the approach and its
generality.

12
 2.2 State of the art

2.2 State of the art on strut-and-tie models and stress fields

Initial design of structural concrete elements was purely intuitive. First patents and
guidelines which came out by the late nineteenth century, were based on simple mechanisms
and engineering experience. One of such systems (presented in Fig.2.5) was introduced by
Hennebique [101]. Recommendations on how to place the longitudinal reinforcement
according to different boundary conditions were given (refer to Fig.2.5a), along with proper
rebar anchorage details. In addition to this, guidelines for shaping and placing the transverse
reinforcement (which was also used to keep the longitudinal reinforcement in place) were
included (refer to Fig.2.5b and c).

(a) (b) Section A-A (c)

Anchorage A Slab

A Stirrup
Rib

Figure 2.5: Reinforcement layout patented by Hennebique:(a) shape of the longitudinal rebars and
disposition of the transverse reinforcement; (b) correct placing the reinforcement in the
cross-section of the rib;(c) detail of the stirrup (adopted from [101])
Inspired by the patent of Hennebique, Ritter [101] introduced a simple approach for
designing structural concrete beams, presented in Fig.2.6.

(a) (b) q
n
Neutral axis Concrete
stress state
Strain state
h
z
θ
Steel
Reinforcement stress state e

Figure 2.6: Design approach proposed by Ritter:(a) stress and strain state of an analysed cross-
section; (b) truss-model (adopted from [101])
Determining the necessary area of the longitudinal steel and concrete, was based on
sectional analysis presented in Fig.2.6a. After obtaining the position of the neutral axis, a
linear strain distribution across the section is assumed. This determines the strain state of
every fibre, and consequently its stress state. According to Ritter, concrete can take
compression stresses (following a nonlinear distribution) as well as the tensile stresses
(following a liner distribution). However, since the tensile strength of concrete is quite small
(approximately 1/10 of its compressive strength), it was proposed that the entire tensile force
acting across a section should be supported by the reinforcement.
In addition to this, Ritter introduced a truss model which should be used to calculate
the position and the amount of the stirrups in a beam (refer to Fig.2.6b). The model was based
on an observation that a reinforced concrete beam subjected to external loads develops cracks
which form at approximately 45°. This means that a truss-like mechanism is formed inside

13
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

such an element, allowing the compressive forces that pass between the cracks to be taken by
the concrete, before being suspended by the vertical stirrups (which cross the cracks), until
finally reaching the supports. This implies that the principal direction of concrete struts is
parallel to the direction of the cracks (θ=45°), and that the vertical reinforcement should be
sufficient to take the applied shear force. According to the presented truss model,
recommended stirrup spacing (e in Fig.2.6c) is equal to the beam’s lever arm (z). Finally,
based on the equilibrium conditions, Ritter concluded that the optimal stirrup inclination is
45° (perpendicular to the compressed struts). However, since vertical reinforcement can be
disposed on a construction site with greater accuracy, 90° inclinations should be used in
practice.

Q Q
(a) (b) Smeared truss

z 45°

l Q

Figure 2.7: (a) RC beam with vertical stirrups tested by Mörsch; (b) smeared truss model
Truss model of Ritter was later investigated and improved by Mörsch [80], who
conducted a large experimental campaign containing reinforced concrete beams with vertical
stirrups (refer to Fig. 2.7a), reinforced concrete beams with inclined rebars (refer to Fig. 2.8a
and c), and reinforced concrete beams with no transverse reinforcement (refer to Fig.2.9a).

Q
(a) Q (b)

z 45° 2·z

(c) Q Q Q
(d) Smeared truss

z z

l
Q

Figure 2.8: (a) RC beam with inclined rebars tested by Mörsch; (b) truss model of a beam with
inclined rebars;(c) RC beam with closely spaced inclined rebars tested by Mörsch;
(d) smeared truss model for beams with inclined rebars
The spacing of the transverse bars in the investigated elements was reduced compared
to recommendations given by Ritter [101]. Consequently, the compression field of the beam
was able to spread over the entire web of the element, leading to more efficient usage of the
concrete (as presented in Fig. 2.7b). The same conclusion was obtained after conducting a
similar experimental campaign involving beams with inclined rebars (notice the difference
between the two trusses in Fig. 2.8b and Fig. 2.8d).
Investigation of elements with no transfers reinforcement led to a conclusion that a
significant shear force can be transferred by a beam even though none of the truss load-

14
 2.2 State of the art

carrying mechanisms presented in Fig. 2.7 and Fig. 2.8 can be formed. Mörsch explained this
by introducing a direct strut action (refer to Fig.2.9b), where shear force is carried through the
web by means of a single element in compression. Aside from the material properties of steel
and concrete, strength of such elements mainly depends on their slenderness (ratio between l
and z in Fig.2.9b, later investigated by Kani [50], followed by many others [84]).

Q (b) Q
(a)

Direct strut
z

l Q

Figure 2.9: (a) RC beam without transverse reinforcement tested by Mörsch; (b) direct strut action
Mörsch also combined the direct strut action with the smeared truss model, obtaining a
more realistic picture of the load-carrying mechanisms in a structural concrete beam (refer to
Fig.2.10a). All this finally gave birth to a new truss model (presented in Fig.2.10b). As it can
be seen, it implied a variable inclination of the struts as well as smaller inclination of the
compression field (θ<45°). Nevertheless, the truss of Mörsch was not used in practice at the
time, since he himself recommended that the inclination of the concrete struts should be equal
to 45° for the design purposes.

Q (b) Q
(a) Direct strut Smeared truss Variable inclination

z z θ

l Q l Q

Figure 2.10: Truss models of Mörsch with:(a) smeared struts and direct strut action;(b) variable
strut inclination
The practical application of variable stress fields with compressive strut direction
lower that 45° was not used until the sixties when Kupfer [64] reintroduced this idea in order
to reduce the material costs of newly designed reinforced concrete structures. He showed that
the reduction of the stress field inclination increases the number of stirrups which are actively
participating in caring shear force, thus allowing to reduce the total area of transverse
reinforcement. He also distinguished the fan region (presented in Fig.2.1a) from the region
having a constant stress field inclination in a reinforced concrete beam.
Truss models were later developed so that they could be used for analysing various
structural concrete elements (such as deep beams, members with openings or dapped-ends,
pile caps, bridge diaphragms etc.). Deriving a consistent approach for analysis of such
elements was initially led by Leonhardt and Walther [69] who conducted an experimental
campaign on reinforced concrete deep beams and intersecting elements. It was shown that
there is no need for a load-carrying model to be an actual truss (statically determinate or
indeterminate system). Instead funicular models which are in equilibrium with the applied
actions are also suitable for analysis or structural concrete elements (refer to Fig.2.11c). This

15
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

discovery lead towards to the development of modern strut-and-tie models which are still used
today. The work of Leonhardt was later continued and generalized by Schlaich et al. [107],
who focused on defining methods which can be used for defining the suitable geometry of
STM. His approach was based on elastic uncracked stress field of a member (presented in
Fig.2.11b), which can be used to determine the location of the strut and the ties (the resultant
of compressive stresses corresponds to location of struts, and the resultant of tensile stresses
defines the geometry of the ties, as shown in Fig.2.11c).
Such approach was very convenient at that time since elastic uncracked internal forces
could be calculated using the photoelasticity or linear finite element method. Schlaich et al.
[107] also proposed a simplification when analysing a structure with STM. Instead of
developing a STM for an entire member, it is sufficient to analyse just the regions where
Bernoulli-Naviers hypothesis is not respected. The rest of the member can be easily calculated
using the beam theory. Hence, an element would be divided into Beam regions (B-regions)
and Discontinuity regions (D-regions), as presented Fig.2.11a, and a STM would only be
required to verify the detail close to the supports.

(a) q (b) q (c)

Q Q

D - region Compression
Strut

B - region h h h

h h Tension Tie

l > 2·h l<h Q l<h Q

Figure 2.11: Strut-and-tie models proposed by Schlaich et al.:(a) B-regions and D-regions of a deep
beam (adopted from Schlaich et al. [107]);(b) elastic uncracked stress field of a deep
beam;(c) corresponding strut-and-tie model of a deep beam
The stress fields originated from a completely different basis compared to STM. It
involved the direct application on the theory of plasticity based on a crude assumption that the
materials do not deform until they have reached the point of yielding (refer to adopted
constitutive laws in Fig.2.12a and b).

(a) σs Actual (b) σc

fct
fy
Adopted
εs

- fy -fce
Adopted -fc
Actual

Figure 2.12: Rigid-plastic constitutive laws: (a) actual and adopted stress-stain diagram for steel;
(b) actual and adopted stress-stain diagram for concrete
This assumption is acceptable in case a constitutive law shows significant deformation
capacity of the material once it reached its elastic limit, such is the case with steel (refer to

16
 2.2 State of the art

actual stress-strain diagram from Fig.2.12a). The actual stress-stain diagram of concrete on
the other hand, does not show a similar behaviour (see Fig.2.12b). In order to insure sufficient
ductility of concrete, the actual compressive strength (fc) has to be reduced down to effective
compressive strength (fce) while the tensile strength of concrete has to be neglected (as can be
seen in Fig.2.12b). The reduction of the concrete strength was first introduced in Denmark
1969 [91], where it was argued that the lower bound solution of theory of plasticity can give
unconservative results in case unreduced concrete compressive strength is used. In 1979
Exner [30] gave the first theoretical bases for estimating the concrete compressive strength
effectiveness factor. Based on the test results involving RC beams with no transverse
reinforcement he introduced the following empirical expression:
3 .2
 (2.1)
fc

The effective concrete compressive strength would then be calculated as:

f ce    f c (2.2)

where fc represents the concrete compressive strength obtained from the standard cylinder
test.
The effective concrete compressive strength was later investigated by many authors
(Muttoni [82, 89], Vecchio and Collins [120], Hsu [47], Pang [94], Kaufmann [52, 53], Hars
[41], etc.), as it will be presented in Chapter 4 of this thesis. However, in the beginning of the
20th century when the theory of plasticity was first applied on structural concrete the ductile
behaviour of elements was ensured by imposing steel yielding as the principal failure mode.
Even though the origins of the theory of plasticity can first be found in the works of
Ingerlsev [48] and Johansen [49], Gvozdev [40] was the first to consistently formulate the
concepts of the yield surface, upper- and lower-bound solutions as well as the flow rule (the
initial publication of Gvozdev was made in 1936 and was translated in English in 1960).

(a) (b) l
½l QR qR

½h ½h
¾h ¾h
h h
b·fc ½h b·fc ½h

External External
anchorage ½lqR anchorage
b·fc ½QR b·fc

Figure 2.13: Lower-bound stress field of a simple beam: (a) subjected to point load; (b) subjected to
distributed load (adopted from Drucker [27])
The theory of plasticity was later used by Drucker [27] to develop a stress field for a
simply supported beam subjected to a point load (refer to Fig.2.13a) and distributed loading
(refer to Fig.2.13b). Both models are in equilibrium with external action and are not
surpassing the strength of applied materials (steel and concrete), meaning that they are licit
lower-bound solutions. As indicated in Fig.2.13, light grey zones of the SF are subjected to

17
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

uniaxial compression equal to concrete compressive strength (fc). This maximises the lever-
arm of a beam, and reduces the total amount of reinforcement. Due to the fact that the tensile
strength of concrete is neglected (as presented in Fig.2.12b), these zones have a constant
width. Dark grey regions are subjected to equal stresses in all directions (quasi-hydrostatic
stress state equal to fce), while the rest of the beam is stress free.
In addition to the lower-bound solution presented in Fig.2.13, Drucker also focused on
the upper-bound solutions using a simple beam subjected to a point load (refer to Fig.2.14).
Two potential mechanisms were assumed. One having completely rigid reinforcement and
failing due to the concrete crushing in the mid-span and next to the external anchorage plates
(as presented in Fig.2.14a), and the other assuming yielding of the reinforcement along with
the concrete crushing in the mid-section of the element (as indicated in Fig.2.14c).

(a)
l QR (b) ½QR
Δ
Rotation
point ½h bfc
Crusing Crusing Crusing
½h bfc ½h

½QR Δh/l ½QR

(c)
l QR (d) ½QR
Δ
Rotation
point bfc
½h
Crusing
½h Asfy

Yielding
½QR 2Δh/l ½QR ½QR

Figure 2.14: Upper-bound solution: (a) “mechanism 1” – assuming concrete crushing in the middle
of the beam and concrete crushing close to the supports; (b) free-body corresponding to
the “mechanism 1”; (c) “mechanism 2” – assuming concrete crushing and steel
yielding in the middle of the beam; (d) free-body corresponding to the “mechanism 2”
(adopted from Drucker [27])
Free-bodies corresponding to the mechanisms in Fig.2.14a and c are given on their
right (Fig.2.14b and d). After expressing all the necessary displacements of the two
mechanism as a function of Δ (see Fig.2.14b and d), the expression for internal and external
work can be written. Drucker then showed that both mechanisms governed the same ultimate
load (QR) which is identical to the one in equilibrium with the lower-bound stress field from
Fig.2.14a, thus obtaining the exact solution according to the theory of plasticity. Even though
ultimate loads derived from Druckers exact solution were sometimes unconservative
compared to experimentally obtained values (due to the fact that the concrete strength
effectiveness factor was not applied, as indicated in Fig.2.12b), he established a clear
theoretical bases which is still used today.
The stress fields were later developed particularly in Denmark and Switzerland. An
alternative approach for obtaining the exact solution using stress fields was proposed by

18
 2.2 State of the art

Nielsen and Hoang [92]. A free body of a deep beam indicated in Fig.2.15a was divided into
triangles having a constant stress state (regions 1 to 3 in Fig.2.15b).

(a) q (b) qR (c)

3 τ
b·fce
N 3 2
Free-body D B
E
τR σ
h
A qR 1
2 C

b·Φ·fce b·τR Φ·fce

l ½·l fce
R R

Figure 2.15: Stress field of a simply supported deep beam: (a) geometry and loading properties;
(b) stress field and border stresses inside the analysed free-body; (c) stress state of each
region represented with a Mohr’s circle (adopted from Nielsen et Hoang [92])
As it can be seen, the analysed disk (deep beam in this case) is assumed to have plastic
tensile strength, which can be determined using the degree of reinforcement (noted as Φ in
Fig.2.15b and c). In order to maximize the lever arm, point N in Fig.2.15b should be placed
high enough so that the border stress acting on the vertical surface of the area 3 is equal to fce.
Once the location of this point is determined, the geometry of all three triangles is known.
This can then be used to obtain the stress state of all the regions using the Mohr’s circle
(presented in Fig.2.15c). Point C represents the pole of circle 1 and point A the pole of circle
3. Lines A-B and B-C, which are parallel to the lines separating zones 2-3 and 1-2, are used to
determine the shear and normal stress acting along the borders between the zones (points D
and E in Fig.2.15c). Once all the points are determined Mohrs circles can be drawn, and the
ultimate load qR is obtained.

Q Q Q Q
(a) (b) (c) (d)

Q Q Q Q

Wide crack Wide crack

Wide crack

Figure 2.16: Developing a suitable stress field for a panel with an opening: (a) to (d) potential strut-
and-tie models at ULS and corresponding behaviour at SLS (adopted from Muttoni [89])
Some of the most significant works in the field of stress fields and strut-and-tie models
can be found in Thürlimann et al. [116], Müller [81], Marti [73] and Muttoni et al. [89] who

19
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

gave clear recommendations for the design of reinforced and prestressed concrete elements
using SF method and STM, most of which can be found in codes of practice. A general
procedure for the development of SF for nonstandard structural concrete members (such as
elements with opening, dapped-ends or complex loading) was introduced by
Muttoni et al. [89] (see Fig.2.16).
The procedure takes into account the behaviour of a member at ULS (refer to the top
row in Fig.2.16) and in a qualitative manner at SLS (refer to bottom row in Fig.2.16). When
talking about the ULS, all the models indicated in Fig.2.16a to d seem suitable (due to the fact
that they are in equilibrium with the external loads). However, when looking at SLS it
becomes clear that only the last STM shows appropriate behaviour, since some of the cracks
indicated in Fig.2.16a to c are unacceptably wide. It is important to indicate that the
procedure proposed by Muttoni et al. is iterative. In this sense Fig.2.16 does not show four
independent models that can be used to design the same structural concrete element, but rather
an evolution (refinement) of a single solution. The stress transfer mechanism should always
be kept as simple as possible, and the model should be modified until the structural behaviour
at ULS and SLS becomes acceptable.
In order to overcome the fact that modelling of elements with D-regions using SF and
STM is time consuming and requires good engineering intuition and experience, various
procedures for their automatic development have been proposed. Inspired by the approach of
Dorn et al. [26] who worked on the automatic optimization of steel trusses, Kumar [62]
introduced a procedure which could generate STM starting from an initial truss using the
energy deformation criterion. Similar procedures then followed using different optimization
criteria. Xie and Steven [125] used an algorithm which favorited elements in tension based on
the amount of force they carried. Biondini et al. [12] used the material volume criterion
whereas Ali and White [4] focused on determining the amount of virtual work or each
potential tie in the model, favouring those having minimal work as well as the ties which were
placed in horizontal and vertical direction.
Starting from the work of Ali and White [4] Kostić [59] developed a procedure which
transforms a STM (presented in Fig.2.17a) into a SF (see Fig.2.17b) using an algorithm that
optimized the geometry of the nodal regions by imposing a pseudo-hydrostatic state of stress.

(a) Q Q Q Q Q Q Q Q
(b)

4Q 4Q b·fce b·fce

Figure 2.17: (a) Generated strut-and-tie model; (b) corresponding stress field with pseudo
hydrostatic nodes (adopted from Kostic [59])
The process of automatic stress field generation was taken one step further by
Fernández Ruiz and Muttoni [32] who developed the Elastic-Plastic Stress Field method
(EPSF) and implemented it in a finite element program (JCONC – which has been integrated

20
 2.2 State of the art

as an applet into an educational web site http://i-concrete.epfl.ch/). According to this approach

stress fields are obtained iteratively for each load step by assessing the stresses in concrete
and reinforcement based on constitutive laws (elastic-plastic material behaviour) and imposed
deformations.
Reinforcing steel is assumed to behave as an elastic-plastic material in tension and
compression with the possibility to account for the strain hardening as presented in Fig.2.18a
It is important to mention that a small amount of strain hardening Eh=0.33 is automatically
introduced in every model, in order to assure the numerical stability of the results. Its response
is governed by the yield strength of the material (fy), its elastic modulus (Es) and its hardening
modulus (Eh). Both can be estimated based on the actual stress-strain diagram, as indicated in
Fig.2.18a. The strain hardening part can also be neglected in case the EPSF are used for
design of structural concrete elements. However, when the ultimate strength of the existing
members needs to be assessed, introduction of strain hardening modulus can significantly
increase the ultimate resistance, especially when hot rolled reinforcement has been applied
(since the difference between the yield and the ultimate strength of the material can be
noticeable).

(a) σs Actual (b) σc (c) σc,2

fy fct -fce,1
Ec = tanφc σc,1
Eh = tanφh Adopted εc
Es = tanφs φs
φc
εs ε·

-fce -fce,2
φh - fy Adopted -fc ε·
Actual

Figure 2.18: Elastic-plastic constitutive laws: (a) actual and adopted stress-stain diagram for steel;
(b) actual and adopted stress-stain diagram for concrete; (c) concrete failure surface
The concrete constitutive law is presented in Fig.2.18b. Its behaviour under
compression is assumed to be elastic-perfectly plastic, whereas its tensile strength is
completely neglected. Concrete modulus of elasticity (Ec) is adopted as secant modulus of the
material, and the effective concrete strength (fce) is calculated in the following manner:
f ce  f c  fc  (2.3)

where fc represents the uniaxial concrete compressive strength measured in cylinders;

ηfc represents the concrete compressive strength reduction factor, which accounts for its brittle
behaviour and can be calculated according to Muttoni [82,89] in the following manner:

f co
η fc  3 1 (2.4)
fc

where fco is equal to 30 MPa in case of normal strength concrete [34].

ηε represents the concrete compressive strength reduction factor which takes into account the
presence of transverse strains in the concrete struts, which can be evaluated on the basis of the
compression-softening law proposed by Vecchio and Collins [122]:

21
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

1
ηε  1 (2.5)
0.8  170  ε1

where ε1 represents the transverse strains in concrete struts.

One of the most significant advantages of the EPSF compared to RPSF, is related to
the estimation of the effective concrete compressive strength (refer to Eq.2.3). Unlike the
compression-softening law proposed by Vecchio and Collins (ηε), rigid-plastic stress field
approach uses fixed reduction values (ν), depending on the angle between the struts and ties
(see Fig.2.19). The coefficients were proposed by Muttoni et al. [89] and later adopted by
MC2010 [34] and SIA262:2013 [108]. As it can be seen, compressive strength of a plain
concrete which is not subjected to any imposed strains (Fig.2.19a) assumes a ν value equal to
1.0, in which case the effective compressive strength is equal to the plastic concrete strength
(fce=fc∙ηfc). On the other hand, imposing the transverse strains means reducing the concrete
compressive strength by the factor 0.75-0.80 in case the strains are perpendicular to concrete
struts or 0.55-0.60 in case they are inclined (angle between the stress field and the
reinforcement is smaller than 65 [34]).

fce·bw fce·bw fce·bw

(a) (b) (c)

ν=1.0 ν=0.75-0.8 ν=0.55-0.6

fce·bw fce·bw fce·bw

Figure 2.19: Reduction factors for concrete compressive strength in RPSF: (a) no imposed
transverse strains; (b) imposed perpendicular strains; (c) imposed inclined strains
In-plane concrete strength can be represented using a Mohr-Coulomb yield surface
with a tension cut-off (refer to Fig.2.18c). The effect which transverse strains have on
material strength can be understood as shrinkage in the yield surface with respect to the
positive transverse strain increase. It should be noted that effective concrete strength (fce)
cannot be increased due to application of negative transverse strains (concrete confinement is
neglected).

(a) (b) (c) (d)

εs σs Fs,i
lFE
2 (u2,v2) 2 2 2

Bar FE εs σs Fs,i
1 (u1, v1) 1 1 1

Figure 2.20: Bar finite element: (a) geometry properties and imposed nodal displacements; (b)strain
state; (c) stress strate; (d) nodal forces

22
 2.2 State of the art

Finally, bond between the reinforcement and surrounding concrete is assumed to be

perfect, meaning that these two materials have equal strains in the direction of reinforcement
(εs = εc) resulting in zero slip.
Implementation of the previous constitutive laws into a nonlinear finite element
method analysis [32] can be performed by means of two different types of element: link
elements and constant strain triangles.
Reinforcement bars and prestressing tendons are modelled using link (1D) finite
elements with uniaxial behaviour (neglecting the dowel effect) as presented in Fig.2.20.
Based on a given displacement and the length of a FE (refer to Fig.2.20a), strain state of the
structural steel member can be obtained directly (see Fig.2.20b). This value is then used as an
input parameter to determine the stress state (refer to Fig.2.20c) based on the constitutive law
presented in Fig.2.18a. The stresses can then be used to determine the nodal (Fs,i) by means of
simple integration over the cross-section surface of the analysed element (refer to Fig.2.20d).
σc,2
(a) (b) (c) (d) Fc,2 (e)
σc,2 σc,2
conc3N FE 2 (u2,v2) 2 2
2
lFE, 1-2 Fcj,i
εc,2 ½ lFE, 1-2 ½lFE,2-3
lFE, 2-3 βj
εc,1 θi
(u1, v1) 1 1 1 1 j
lFE, 1-3 βj+1
θi ½ lFE,1-3 3 Fc,1 3
3 (u3,v3) 3
Fc,3 θi
σc,2
σc,2

Figure 2.21: Concrete finite element: (a) geometry properties and imposed nodal displacements;
(b) strain state; (c ) stress state; (d) nodal forces; (e) definition of the characterisitc
angle β in the node j
Concrete is modelled using constant strain triangles. Once again, a displacement field
is imposed on a concrete FE (refer to Fig.2.21a) which is then used to obtain a strain state (εx,
εy and γxy). Using the Mohr’s transformations the principle strains (ε1 and ε2) as well as their
principle direction θi are obtained (refer to Fig.2.21b).
Assuming that the principal stresses are parallel to the principal strains, concrete
stresses can be directly calculated (refer to Fig.2.21c) using the constitutive law defined in
Fig.2.18b. Finally as presented in Fig.2.21d and e, nodal forces can be derived as following:

lj l j 1 
Fcj ,i  σ i    cosβ j    cosβ j 1  i  1,2& j  1,2,3 (2.6)
2 2 
where Fcj,i represents the nodal force in j-node and i-principle stress direction determined
from a concrete FE;
σc,i represents the concrete stress in i-principle direction;
lj represents the length of j-side of the FE triangular;
βj represents the characteristic angle in j-node presented in Fig.2.21e.

23
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

The iterative process for obtaining an EPSF for a given load, geometry and
reinforcement layout starts with a linear elastic FEM calculation, in which both materials
(concrete and steel) have infinite strength in tension and compression. This is done in order to
obtain an initial displacement field, which is then imposed back to the model. This time
however, the materials (concrete and steel) are assuming with an elastic-plastic behaviour (as
presented Fig.2.18). Strains, stresses and nodal forces coming from the bar and concrete FE
are obtained (as described above), after which equilibrium of each node is checked. This
usually results in having some residual (un-equilibrated) forces. The intensity and the
direction of these forces are then used to correct the initial displacement field, by moving the
nodes at a certain rate using the Newton-Raphson algorithm. This defines a second
displacement field, which is then reused in the same manner. The entire process is repeated
for a predefined number of steps until a convergence is reached (a point in which residual
forces become insignificant).
q q
(a) (b)

6×Øst5
1’600

160 160
8×Ø8

1’600 7×stØ5

Figure 2.22: Elastic-plastic stress field method: (a) geometry properties and reinforcement layout of
a RC wall tested by Leonhardt [59]; (b) corresponding FE mesh and EPSF
As described above, the method verifies the equilibrium of the system and the yield
condition everywhere (lower bound theorem is respected). It also respects compatibility
conditions and eventually becomes a failure mechanism (the upper bound theorem is
respected). According to the theory of plasticity, this means that the final solution is the exact
solution. One major advantage of the EPSF is the fact that the failure kinematics of an
analysed member is clearly visible, and can be used to develop an upper bound solution
according the RPSF to verify the results. An example of an EPSF is given in Fig.2.22.
Geometry properties and the reinforcement layout of a deep beam tested by Leonhardt [69]
are presented in Fig.2.22a. The corresponding EPSF at its ULS is given in Fig.2.22b. Blue
lines indicate the principal concrete compressive stress direction and intensity (with their
inclination and length) and the thickness of the red lines indicated the stress level in the
reinforcement (dark red represents yielding).
An alternative approach to EPSF was introduced by Kuchma and Tjhin [61]. Their
Computer Aided Strut-and-Tie design tool (CAST) has been used to generate the STM of
elements with D-region and estimate the minimal necessary space to pass the concrete struts
and place the critical nodes.
Hoogenboom [46] introduced an iterative procedure that uses linear and nonlinear
stringer-panel models to optimize both geometry and reinforcement layout of a structural

24
 2.2 State of the art

concrete element, based on the maximal acceptable crack width at SLS and required design
load at ULS.
Another computer aided design tool was developed by Lourenço and Almeida [71] as
a part of their Adaptive Stress Field Approach, which offers a different SF for different load
level. As a part of their research they conducted an advanced analysis or RC deep beams
(tested by Leonhardt [69]) during which they focused on elements ductility, deformations and
crack width. The method proved to be very useful when analysing structural concrete
members at SLS.
Considering the practical application of stress fields and strut-and-tie models, it can be
concluded that each method has its advantages. Using them in a combined manner is perhaps
the most suitable approach, as both are expressing the same physical behaviour in a different
manner. Stress fields can be used to determine the minimal necessary width of the struts in
order to satisfy the compatibility condition, or estimate the stress state in a specific region of
the structure in order to compare it to concrete strength. In addition to this they allow
understanding and identifying the location where smeared reinforcement is to be arranged.
The development of stress fields is thus particularly useful for detailing (required space for
the struts, reinforcement bents, stresses in the nodal regions) and to account for variable angle
of the compression fields (fan or constant-angle compression fields). Nevertheless, a complete
development of the stress field for a structural concrete element may be too time-consuming.
This becomes even more evident during the design, when an iterative procedure involving a
gradual model refinement is applied. In order to overcome this, stress fields can be combined
with strut-and-tie models, which in fact represent nothing else but a simplification accounting
only for the resultants (forces) of the stress fields. However, a single STM can be interpreted
in different manners leading to different reinforcement layout (refer to [42]). This is why a
local stress field analysis of the critical concrete regions based on the final reinforcement
layout needs to be conducted in order to assure a satisfactory behaviour of the structural
members.

25
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

2.3 Suitable approaches for designing new structural elements with stress fields
Design of new elements should be performed in order to obtain safe structures at ULS
with satisfactory behaviour at SLS while keeping the reinforcement layout as simple as
possible (thus accommodating a precise and fast execution at the construction site). There are
various strategies which can be used for this purpose, four of which will be presented and
compared in the following section.
The methods can be used separately or be combined, depending on the given case. The
selection of the most suitable approach should be decided by the designer based on his/her
experience, the complexity of the problem and the required level of accuracy of the analysis.

2.3.1 Design models inspired by the existing cases

Design of new structural elements is often performed based on their loading and
geometrical analogies to some well-known solutions found in literature. This will be
presented using a practical example given in Fig.2.23a, where a deep beam with an opening
(first used by Schlaich et al. [107]), is loaded with 2’000 kN concentrated force.
In a very simplistic manner, this element can be considered as a dapped-end beam
(neglecting the presence of vertical and horizontal member around the opening), which allows
the application of a STM as indicated in Fig.2.23b. Consequently, the tension and
compression forces can be estimated (refer to Tab.2.1), which governs the amount of
necessary reinforcement. In case further refinement of the model is required, the load-carrying
mechanism can be improved by taking into account the contribution of the column and the
coupling beam around the opening (see Fig.2.23b). This results in strut-and-tie model
presented in Fig.2.23c, which once again allows a simple design of the main reinforcement
(refer to Tab.2.1).
Comparing the STM models from Fig.2.23b and c one can see that the amount of the
main reinforcement (the ties around the opening and the tie on the bottom of the deep beam)
is quite similar. However, the second model does provide some reduction in the amount of the
reinforcement in the upper point of the wall while in the bottom point, reinforcement has to be
provided to respect SLS requirements. As already indicated, whether this difference is
sufficient enough to invest the additional time for its development or not depends on the
responsible designer. The resulting reinforcement layout (using the mixed STM from
Fig.2.23c) is given in Fig.2.23d. The assumed design yield strength of the reinforcement is
equal to 435 MPa.
Table 2.1: Forces corresponding to the strut-and-tie models from Fig.2.23b and c
Struts Force Struts Force Struts Force Ties Force Ties Force
“S“ [kN] “S“ [kN] “S“ [kN] “T” [kN] “T” [kN]
S1 -946 S7 -638 S13 -591 T1 691 T7 311
S2 -632 S8 -844 S14 -837 T2 620 T8 461
S3 -911 S9 -576 S15 -606 T3 593 T9 83
S4 -597 S10 -322 S16 -1460 T4 99 T10 433
S5 -1460 S11 -207 T5 289 T11 509
- -
S6 -665 S12 -248 T6 691 - -

26
 2.3 Suitable approaches for designing

2’000 kN

4’750 mm 2’750 mm 2’000 kN

(a) (b)

bw=250 mm
fc=20 MPa S2
4’750 mm
2’750 mm

fy=435 MPa
S1 T3 S4 S5

T2
1’500 mm

S3

T1

500 mm 1’500 mm 5’000 mm 500 mm

7’500 mm 714 kN 1’286 kN

2’000 kN
(c) (d)

S9 2Ø8@200

S8 T11 S15 S16

2Ø20
Ø20
T8 2Ø20

S14
S7 S12 2Ø20
Ø20 Ø20
T7 T9 S11 S13 2×2Ø20
T6
S6
T4 S10 T5 T10 stØ8@200
714 kN 1’286 kN

Figure 2.23: Developing a design model based on analogies with the existing solutions: (a) geometry
and load properties of the analysed element; (b) strut-and-tie model of a dapped-end
detail combined with the contribution of the column and beam around the opening;
(c) merged model; (d) reinforcement layout
Some of the nodes (S6-S10-T4, S6-S7-T7, S7-S8-T8, S13-T5-T6-T10 and S16-R-T6)
are placed in the proximity of the elements surface, meaning that the anchorage of the
reinforcement needs to be sufficient enough to assure the required force transfer [13]. In order
to do that, the U-shaped re-bars having the same diameter as the once from the main
reinforcement (dimeter Ø20 in this case) are placed in the nodal regions (see Fig.2.23d). The
reinforcement is usually completed with a minimum smeared reinforcement in order to
control the cracking and to allow spreading of the concentrated loads (refer to the ground
mesh of Fig.2.23d). Applying the minimal reinforcement increases the load-bearing capacity
of the element. Even though this reinforcement can be taken into account in the STM (making
it more accurate), having some reserve in the ultimate strength is considered as positive when
designing new elements. Over its expected lifetime, a structure might change its purpose
(which imposes different load conditions), or experience unexpected events that could
jeopardise its integrity. Both situations would benefit from unaccounted load-carrying
mechanisms. When developing STM suitable for design, it is important to keep in mind that
from a practical point of view exact solutions do not necessarily give the most satisfying

27
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

results since they leave very little space for human errors that can occur during the
construction phase and can be time consuming to develop and validate.

2.3.2 Design models based on deviated thrust lines

Another manner to design the element from Fig.2.23a consists of drawing the “ideal”
STM which would carry the forces to the supports in the most direct manner, regardless of the
actual geometry of the member. Such model is presented in Fig.2.24b.
This clearly indicates parts of STM that are violating the boundary condition of the
element (strut S1 in this case, since it is passing through the opening). After extracting the
free-body (as indicated in Fig.2.24a), and making sure that it is in equilibrium, suitable
reinforcement should be placed in order to deviated the struts while making sure that they
remain within the available concrete.

2’000 kN

(a) S1 (b)

S5

S4 T3 S8 S1 S2

T2

S3 S7
S6
T1 T1

714 kN 714 kN 1’286 kN

Figure 2.24: Developing a design model for a deep beam with an opening using the deviated thrust
lines: (a) STM spreading the strut around an opening; (b) funicular STM
The internal forces corresponding to the STM from Fig.2.24 are given in Tab.2.2 and
can be used to determine the reinforcement layout, similar to the one presented in Fig.2.23d.

Table 2.2: Forces corresponding to the strut-and-tie models from Fig.2.24

Struts Force Struts Force Ties Force
“S“ [kN] “S“ [kN] “T” [kN]
S1 -944 S5 -575 T1 691
S2 -1460 S6 -644 T2 510
S3 -634 S7 -871 T3 510
S4 -845 S8 -606 - -
Deviating thrust line method has been acknowledged as a general tool for developing
stress fields, especially for the cases where direct analogy with a known solution is not so
evident. However, not every funicular shape which is in equilibrium with external loads is
automatically acceptable. The development process is iterative, and in order to obtain a model
with acceptable behaviour at ULS and SLS, four following cases should absolutely be
avoided:

28
 2.3 Suitable approaches for designing

1. Having zones without reinforcement where tensile stresses are expected;

2. Having unacceptably low angles between the struts and the ties;
3. Having STM with TTT nodes (see Fig. Fig.2.26d);
4. Having uncontrolled cracking related to anchorage issues.
The first problem is presented in Fig.2.25b. It shows a model where the primary
tension tie located above the opening (disregarding the lower part of the structure). The
proposed solution is statically admissible (the internal forces are in equilibrium). However, in
order to activate the assumed STM, large flexural cracks may develop in an unsuitable
manner within the unreinforced region (refer to Crack 1 in Fig.2.25b). Even in case the
reinforcement for crack control is placed below the tie, Crack 1 might still be unacceptably
wide. In order to avoid such cases, Muttoni et al. [89] proposed their iterative approach
described in Fig.2.16. For the case presented in Fig.2.25a, this means qualitative analysis of
the crack pattern (given in Fig.2.25b) and adapting the reinforcement layout until more
suitable structural response in achieved. In other words, another tie needs to be added below
the opening (as presented in Fig.2.23d or Fig.2.24).

(a) (b)

Crack 2 Crack 2

Crack 1

Figure 2.25: Potential problems of a STM neglecting the presence of lower part of the wall:
(a) statically admissible solution; (b) corresponding deformed shape
The second issue regards the limit of the angle between the struts and the ties in STM,
and is presented in Fig.2.25 (refer to Crack 2). When no compatibility conditions are
accounted (assuming for instance a rigid-plastic material behaviour) and if no transverse
reinforcement is available, the angle between the struts and the ties should usually be
considered larger or equal than 45° (refer to Muttoni et al. [89]). In case the transverse
reinforcement is present in a model, the angle between the struts and the ties can be reduced.
However, the angle between struts and ties (named θ in Fig.2.25a) should not be lower than
20°-25°. This limit is grounded by Grob [39] on the fact that otherwise the strain state of the
member (tension in the reinforcement and compression in the concrete) can otherwise become
incompatible, requiring large tensile strains as well as crack widths to develop (refer to
Crack 2 in Fig.2.25a and b). This potentially reduces the effective concrete compressive

29
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

strength (ν reduction factor to be exact, refer to Fig.2.19) below the conventional thresholds
(0.55-0.6 in this case), resulting in an un-conservative estimation of elements load capacity.

(a) SA SB (b) S (c) (d) T

SB A S B
SB TA TA TA Cracks
SA fce S TA TC
SC T fce TB
fce lB,anch
fcp fce lanch lB,anch
SC T TB TB
lC,anch
fce Cracks Cracks
SA fcp Cracks lA,anch Cracks lA,anch
fce fcp Cracks
TC
SB

Figure 2.26: Nodal regions of RPSF: (a) Compression-Compression-Compression (CCC);

(b) Compression-Compression-Tension (CCT); (c) Compression-Tension-Tension
(CTT); (d) Tension-Tension-Tension (TTT)
Application of RPSF implies using four different types of nodes presented in Fig.2.26:
CCC, CCT, CTT and TTT (where C refers to compression and T to tension). CCC nodes (see
Fig.2.26a) can be used in design without any restrictions. As shown by Tepfers [115], each tie
induces cracking of the surrounding concrete matrix, caused by steel-to-concrete stress
transfer mechanisms. This occurs along tie’s anchorage length (refer to red ties in Fig.2.26b to
d), as well as it’s body (refer to black ties in Fig.2.26b to d), which consequently reduces the
effective concrete compressive strength (see Fig.2.19b and c). Providing that each tie has
sufficient anchorage length, and that the minimum angles between the struts and the ties are
respected (as previously discussed), CCT and CTT nodes can be used. On the contrary, TTT
(Tension-Tension-Tension) nodes (presented in Fig.2.26d) should absolutely be avoided. This
is justified by the fact that large crack widths can potentially develop inside the nodal region,
leading to very low values of the strength reduction factor ν. Each time a TTT node is present
in a load-carrying model, the topology has to be modified in order to avoid such kind of node.
Prestressing of at least one of the ties could also be an acceptable solution.

(a) (b)

Cracks
Crack

Figure 2.27: Potential problems of a STM resulting in a reinforcement layout without any crack
contorl reinforcement: (a) staticaly admisible solution; (b) corresponding deformed
shape

30
 2.3 Suitable approaches for designing

As discussed by Schlaich et al. [107] or Muttoni et al. [89], the strut-and-tie model has
to consider that compression in concrete tends to occupy all available space following the St-
Venant’s principle. This is particularly significant with respect to the introduction of
concentrated loads [14], and should be also considered for the anchorage of tension ties. For
instance, Fig.2.27 presents a load-carrying model where the vertical tension tie is anchored
within the element but not at its edge.
As experimentally shown by Maxwell [77], this may lead to development of wide
cracks on top of the anchorage region, since the strut tends to occupy all available concrete
(unreinforced anchorage region of the vertical tie, indicated in Fig.2.27a). These cracks
ultimately join with the crack that originates from the top left corner of the opening (see
Fig.2.27b) forming a kinematically admissible mechanism.

2.3.3 Design models inspired by elastic uncracked stress fields

A classical approach to development of strut-and-tie models was early investigated by
Leonhardt [69] and later refined by Schlaich [107]. The approach considers the elastic
uncracked stress field analysis of a member (refer to Fig.2.28a) and arranges the
reinforcement following the location, the direction, and the intensity of the tensile stresses (as
presented in refer to Fig.2.28b).
This method generally gives a conservative solution with respect to its ULS as well as
a satisfactory behaviour at SLS due to the fact that the deformation of the tensile zones is
controlled by the reinforcement.

2’000 kN 2’000 kN
(a) (b)

714 kN 1’286 kN 714 kN 1’286 kN

Figure 2.28: Element design inspired by an elastic uncracked behaviour: (a) elastic uncracked stress
field of a deep beam with an opening; (b) corresponding STM
Nevertheless, this approach does have a few weak points:
1. The location and arrangement of the reinforcement is not decided by the designer.
Following the distribution of elastic stresses inside an uncracked member leads to
application of inclined bars which may be difficult to put in place and inspect at the
construction site (see Fig.2.28b). Nonetheless such reinforcement layout assures a
positive behaviour.

31
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

2. The method leads potentially to TTT nodes (for instance for corner frames with
opening moments), which need to be avoided in design.
3. Finally, the amount of required reinforcement is not necessarily the minimum required
for equilibrium reasons. This is for instance the case when imposed strains are
considered as actions (imposed strains can be dissipated provided that the member has
sufficient ductility). Also, this results from the location of the tension ties, which are
generally more efficient when placed at the outermost fiber of the member respecting
concrete cover to maximize the lever arms (as presented in Fig.2.29c), rather than at
the resultant of the uncracked stress field (see Fig.2.29a and b). In a similar manner,
placing the concrete strut in the resultant of compressive stresses leads to reduction of
the lever arm, and thus increases the required amount of reinforcement for the design.

(a) q (b) (c)

Q Q Q Q

zUNCR zCR

Q Q Q Q

Figure 2.29: (a) Uncracked elastic SF of a deep beam subjected to uniformly distributed load;
(b) STM resulting from the uncracked elastic SF; (c) STM resulting from a plastic SF

2.3.4 Design models inspired by cracked stress fields

More suitable stress fields that can be used to develop consistent load-carrying models
accounting for the cracked behaviour of concrete using advanced constitutive material
models. Fernández Ruiz and Muttoni [32] proposed a simple elastic-plastic stress field
approach, accounting for the cracked behaviour of concrete. This can be seen as a
simplification of a more general constitutive model that leads to suitable results when
sufficient transverse reinforcement is available to avoid crack localization. Fig.2.30a shows
an EPSF of previously investigated deep beam with an opening (see Fig.2.23a). The EPSF
clearly indicates the location of the compression field (its intensity is presented using blue
lines with different length) and intensity of tensile forces carried by the reinforcement (dark
red indicates yielding of the steel).
As shown in [32], cracked SF can be used to develop strut-and-tie models (or rigid-
plastic stress fields) the same way elastic uncracked SF were used by Schlaich et al. [107].
However, the results of the EPSF can also be directly applied for the design and optimization
of the reinforcement given a fact that they already represent a licit stress field. An approach to
optimize the reinforcement layout and amount has been also presented by Fernández Ruiz and
Muttoni [32]. Starting from a preliminary analysis where a minimum reinforcement required
for the crack control (Asi,min) is arranged according to the designers preferences (for example

32
 2.3 Suitable approaches for designing

the orthogonal mesh from Fig.2.30a), the stress state of the reinforcement can be calculated
for each bar FE (σsi). In the next step, the area of the reinforcement (Asi) can then be updated
so that the steel stress does not violate the yield strength of the material:
 si
Asi , j  Asi , j 1  Asi ,min (2.7)
f yi

where fyi represents the yield strength of the reinforcement;

i represents the bar FE number;
j represents the iteration number (j = 2 is the first iteration, while j = 1 is the initial step
corresponding to Asi,min).

2’000 kN
(a) σc / fce
(b)
1.0
2Ø8@200

2Ø16
Ø16
0.5 2Ø16

2Ø16
Ø16 Ø16
2×2Ø16
0.0

714 kN 1’286 kN

Figure 2.30: Design of a deep beam with an opening using the EPSF method: (a) EPSF of fully
cracked element; (b) resulting reinforcement layout
With the updated reinforcement area, the EPSF analyses can be repeated until the
solution converges to a final reinforcement layout, such as the one presented in Fig.2.30b
(assuming design yield strength of the reinforcement equal to 435 MPa). The final rebar
diameter is slightly increased to round it to available commercial value (Ø8 and Ø16 in this
cases), which introduces some reserve of the ultimate strength. The required number of
iterations to attain the final solution is quite low, and the method is robust for its practical
application (investigated by Kostić [59]).
Unlike previous approaches, where minimum crack control reinforcement (As,min) was
added in the end of the calculations to assure a positive behaviour of the model at SLS and
ULS, the EPSF considers it from the beginning. In this manner, its contribution to resist the
applied design loads is taken into account, which leads to potential savings of the
reinforcement amount. Finally, taking advantage of the compatibility conditions of the EPSF,
the final reinforcement layout can be also analysed at serviceability limit state, both for
deflections and cracking.

33
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

2.4 Suitable approaches for assessing the ultimate strength of structural concrete
elements with stress fields
In case an existing structure is subjected to increased loads, requires retrofitting or is
deteriorated after several decades of usage, its structural assessment is usually performed. In
addition to this, estimation of a load-carrying capacity is also required when new code
provisions become more severe compared to the existing ones, making the structural
assessment more and more common in modern engineering practice.
Nonetheless structural retrofitting is an expensive and complicated procedure that
involves a number of constrains during its physical execution. Therefore, estimating the load-
carrying capacity of existing structures is directed towards avoiding or minimizing the
necessary interventions. As already stated in the beginning of this chapter, the modelling
process can be performed following a Levels-of-Approximation approach [85], starting with
simple load-carrying models and refining them whenever the obtained structural resistance
proves to be insufficient.
During the process of structural design, even when using rational approaches that were
previously presented, some elements are not considered within the load-carrying model. This
usually refers to the minimal reinforcement for crack control. Nevertheless, this additional
reinforcement should be considered as part of the load-carrying mechanism for an assessment,
since it increases the strength of the member in potentially non-negligible manner (especially
when analysing RC walls with significant quantity of crack control reinforcement). In
addition to this, design based on lower-bound solutions according to limit analysis (such are
stress fields or strut-and-tie models) implies that more efficient load-carrying mechanisms
may develop within the member for the available geometry and reinforcement (which was
graphically presented in Fig.2.4h).
In case conventional analyses do not allow ensuring sufficient strength with respect to
the design actions, refined estimates of the strength are required. In this case, it means
selecting a lower-bound solution that gives the highest possible resistance according to the
limit analysis, in other words exact solution (refer to Fig.2.4h). Different techniques can be
used to obtain it, and in the following section two approaches will be examined:
1. The one based on rigid-plastic stress field approach;
2. The one based on elastic-plastic stress field approach.
As for design, strength assessment will be presented using a practical example.
Therefore, a deep beam with an opening from Fig.2.23a will be analysed assuming the
reinforcement layout presented in Fig.2.23d.

2.4.1 Assessing the ultimate strength using a rigid-plastic stress field approach
The use of rigid-plastic stress fields combined with mechanisms for the search of an
exact solution has been discussed by Muttoni et al. [89]. This can be performed by selecting a
licit collapse mechanism (upper-bound solution) whose free bodies are separated with discrete

34
 2.4 Suitable approaches for assessing the ultimate strength

cracks and concrete hinges as presented in Fig.2.31a. According to the upper-bound theorem
of the theory of plasticity, all rebars crossing the cracks have already reached their yield
strength, which means that the intensity of the forces in the ties at ultimate limit state can be
directly calculated (refer to Fig.2.32 and Tab.2.3). The contact zones between the free bodies
are considered as compression zones, which accommodate the development of the struts or
the nodal regions. There are two methods to calculate the load-carrying capacity of a given
mechanism:
1. Using the work equation, where the work of external loads (sum of scalar product of
external forces and related displacements) has to be equal to the internal plastic
dissipation (sum of all reinforcement yielding forces and concrete forces multiplied by
their elongation or shortenings);
2. Finding the equilibrium of every free-body.
With respect to the second method, solving the equilibrium equations of every free
body can be performed by:
1. Following an iterative procedure based on optimizing the thrust -lines inside the free-
bodies;
2. Determining the contact forces between the free-bodies and then solving a system of
equations relating all the implied variables.
In order to qualitatively define the crack length and the location of the compressed
zones of the assumed failure mechanism, detailed analysis of its kinematic behaviour is
conducted and presented in Fig.2.31c.

(a) (c) PA
ψA
A PAB B

PAC

C
PBD

PCD D

(b)

PD
ψD

PBC PC ψc ψB
PB

Figure 2.31: Kinematical analysis of an assumed failure mechanism: (a) rigid plates A to D;
(b) corresponding deformed shape; (c) plan of rotation poles

35
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

The deep beam is separated in four rigid-plates (marked as A, B, C and D in

Fig.2.31a), whose intermediate poles of rotation (PAB, PBD, PCD and PAC) are indicated using
white circles. Red circles in Fig.2.31c present the rotation poles of each rigid plate which
were determined using the Arnold-Kennedy’s Theorem. According to it, three intermediate
rotation poles of the rigid plates have to lie on a straight line, and have to be between the
poles of the corresponding plates. Ones the rotations ΨA, ΨB, ΨC and ΨD are known, the crack
opening can be qualitatively defined (as presented in Fig.2.31b). The internal forces
associated to reinforcing bars can then be calculated directly (refer to Fig.2.32 and the ties of
the free-bodies A, B, C and D in Tab.2.3). However, since the exact depth of the compression
zones between the free-bodies is not known, nor is the precise location and the direction of the
compression forces (see the struts SAB, SAC, SCD and SBD in Fig.2.32), the system has to be
solved through iterations.

SAB
A y
QU x
TAB1
SAB
B
TAB2 TAB2
TAB1
TAB3
SAC TAB4
TAB4
SAC TAB3

C
TCD TCD SBD SBD

SCD SCD
D
TBD TBD
RL RR

Figure 2.32: Free-body equilibrium of parts A to D

Looking back at the free-body A, presented in Fig.2.33a, the intensity of the internal
forces SAB and SAC as well as the location of points A and B can be calculated on the basis of
the concrete compressive strength (refer to Fig.2.33b).
Using the kinematic considerations from Fig.2.31c, it can be concluded that at the
ULS, strut SAB has to touch the upper surface of the deep beam and that the strut SAC has to
touch the top left corner of the deep beam opening (as is presented in Fig.2.33a).
As a first approximation in this case, it can be assumed that the force SAC is vertical,
which allows determining the moment equilibrium for the point A, whose location is also
known since the intensity of SAB can be obtained from an equilibrium of forces in the
horizontal direction. This allows calculating the shear strength of the free body A, which can
also be determined by using the force diagram shown in Fig.2.33b (the represented diagram
relates to the final result where SAC has a horizontal component). The same approach can be
followed for the other free bodies. In the case of free body D, the steel forces TBD and TCD are
known (reinforcement at yielding). It can be assumed (as a first approximation) the axial force
of the coupling beam is negligible (which is consistent with the previous assumption that SAC

36
 2.4 Suitable approaches for assessing the ultimate strength

is vertical). Thus, the horizontal component of concrete forces SBD and SCD can be calculated
(equal to TCD and TBD) and vertical component is assumed equal to 0. The next step could be
the investigation of body C which allows calculating the force at the left support (RL) and the
horizontal component SAC. The latter is however not compatible with previous assumptions so
that a second iteration step is needed. To do so, the horizontal component of the force SAC
increases the shear strength of the free body A as the shear span is reduced (as presented in
Fig.2.33a). However, it also reduces the shear strength of the free body D as a tensile normal
force acts now on the coupling beam. The iterative process is finished when the calculated
horizontal component of SAC from equilibrium of the free-body C equals the assumed value
for investigating body A. Once the iterations are completed, equilibrium can be checked
investigating the equilibrium of body B, and therefore the load-carrying capacity of the
member (QU) is obtained (refer to Tab.2.3 to see the intensity of analysed struts and ties after
the 1st and final iteration). This check can also be done by developing the force diagrams
(refer to Fig.2.33b) which is also called the Cremona diagram.
final iteration 2408 kN
(a) fce (c)
SAB
N1
1st iteration Point A
N2

N3 TAB1

TAB2

N4 TAB3

fce Point B TAB4

SAC 860 kN SAC 1548 kN
(b) SAC
TAB3 TAB1 bAC=
fce·bw
TAB4 SAB
bAB fce
SAC Point A
TAB2
SAB fce
SAB bAB= Point B
fce·bw
bAC

SAC

Figure 2.33: Verifying equilibrium and boundary conditions of an assumed mechanism: (a) analysis
of the A free-body using RPSF approach; (b) cremona diagram of the thrust lines
inside the free-body and RPSF details of points A and B; (c) resulting STM
Alternatively, the contact forces between the free-bodies can be determined directly,
without the need of performing any iterations. This holds true as the 12 unknown force
components (two components of each of the four contact forces Sij plus the load-carrying
capacity QU and the three reactions of the statically determinate system, refer to Fig.2.32) can
be calculated using the 12 equilibrium conditions (3 for each of the four free bodies). In
addition to this, the location of the points A and B can be calculated on the basis of the width
of the struts (bAB and bAC in Fig.2.33b) in the following manner:

37
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

Sij
bij  (2.8)
bw  f ce
where bw represents the width of the wall;
Sij represents the compressive force in a strut between the free-bodies i and j;
fce represents the effective concrete compressive strength (as defined in Eq.2.3 using ν
reduction factor presented in Fig.2.19 instead of ηε).

Table 2.3: Forces corresponding to the strut-and-tie models from Fig.2.32 and 2.33
Free Struts Force 1st Force final Ties Force 1st Force final
body “S“ iteration [kN] iteration [kN] “T” iteration [kN] iteration [kN]
SAB,x -754 -897 TAB1 481 481
SAB,y 71 -28 TAB2 -437 -437
“A”
SAC,x 0 143 TAB3 273 273
SAC,y 639 738 TAB4 -273 -273
SAB,x 754 897 TAB1 -481 -481
SAB,y -71 28 TAB2 437 437
SBD,x 590 447 TAB3 -273 -273
“B”
SBD,y 0 122 TAB4 273 273
RR 1150 1548 TBC -590 -590
QU -1789 -2408 - - -
SAC,x 0 -143 TCD 590 590
SAC,y -639 -738
“C” SCD,x -590 -447
- - -
SCD,y -275 -122
RL 639 860
SCD,x 590 447 TCD -590 -590
SCD,y 275 122 TBD 590 590
“D”
SBD,x -590 -447
- - -
SBD,y -275 -122
The load carrying capacity QU of the selected mechanism (presented in Fig.2.31a),
calculated by any of the previous procedures, is still not proved to be corresponding to that of
the exact solution according to limit analysis, as other potential mechanisms (such as the one
shown in Fig.2.34b) can develop. As this procedure is based upon the upper-bound theorem
of limit analysis, the exact load carrying capacity (QU,exact) could be lower than the calculated
value QU. An interesting approach to verify if the selected mechanism corresponds to the
exact solution can be performed by combining both theorems of limit analysis. As stated by
Drucker [27]: “agreement of upper and lower bounds proves that the load carrying capacity
is exactly halved”. This procedure can be seen as an optimization where the criterion is related
to the load-carrying capacity.
In our example, this can be done by developing a stress fields (STM) inside the
investigated element in order to verify if the boundary conditions, as well as the strengths of
steel and concrete elements will be respected. This is shown in Fig.2.33a and c. The location
of the node N4 can be determined by using the direction of the forces SAC and TAB3. After
solving the nodal equilibrium, a similar procedure can be used to obtain the location of the

38
 2.4 Suitable approaches for assessing the ultimate strength

node N3. The strut deviates at every intersection with the reinforcement, until it reaches the
internal force SAB on the opposite side of the free-body A. Provided that the compression
fields remain within concrete (as is the case in Fig.2.33c), a licit solution will have been
obtained.

(a)
A PAB B

PB
ψB

(b) (c)

PA ψA

Figure 2.34: Kinematical analysis of an assumed failure mechanism: (a) rigid plates A and B;
(b) corresponding deformed shape; (c) plan of rotation poles
On the contrary, when the calculated stress field leads to compression forces
developing where no concrete is available (as it is presented in Fig.2.35b), the selected
mechanism is not corresponding to the exact solution and the ultimate strength is
overestimated (QU = 3’201 kN).
3201 kN
(a) (b)
SAB1 SAB2

TAB1 TAB1

TAB2 TAB2

TAB3 TAB3
Violation of the
TAB4 TAB4 boundary conditions

TAB5 TAB5
RL RR 1144 kN
2057 kN

Figure 2.35: (a) Free-body equilibrium of parts A and B; (b) corresponding STM
In this cases, other mechanisms have to be investigated. The kinematics of the
corresponding mechanism is presented in Fig.2.34, which shows the discontinuity lines (in
Fig.2.34a), deformed shape of the deep beam in Fig.2.34b and plan of the rotation poles for
the rigid plates A and B in Fig.2.34c. As it was the case with the mechanism analyzed in

39
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

detail from Fig.2.31a, the ultimate force corresponding to the kinematics from Fig.2.34a, can
be determined through iterations or by solving a system of equations (having 8 unknown in
this case). The intensity of the contact forces presented in Fig.2.35a can be found in Tab.2.4.
The obtained result (from Fig.2.33c) for the reinforcement layout of Fig.2.23d, leads
to a failure load of 2’408 kN. This result is above the original design load of 2’000 kN and
shows a strength reserve of 20%. This is due to the fact that:
1. The selected strut-and-tie model for design of the main reinforcement is a lower
bound;
2. Necessary reinforcement was rounded (increased) to the next available commercial
rebar diameter;
3. The minimum reinforcement amount was not considered to contribute to the beam’s
strength (contrary to the EPSF approach for design shown in Fig.2.30b).

Table 2.4: Forces corresponding to the strut-and-tie models from Fig.2.35 and 2.34
Free Struts Ties
Force [kN] Force [kN]
body “S“ “T”
SAB1,x -1700 TAB1 435
SAB1,y -1144 TAB2 370
“A” RL 1144 TAB3 305
TAB4 295
- -
TAB5 295
SAB2,x 1700 TAB1 -435
SAB2,y -2057 TAB2 -370
“B” RR 2057 TAB3 -305
TAB4 -295
- -
TAB5 -295

2.4.2 Assessing the ultimate strength using elastic-plastic stress field approach
Despite the generality of the approach explained in the previous section, obtaining a
solution following a rigid-plastic approach might be time-consuming. It requires iterations to
solve as well as the analysis of more than a single failure mechanism. In addition to this,
sometimes it is not so evident how a potential failure mechanism might develop due to
complex geometry of the element and its loading conditions. In case of over-reinforced
structures, where elements can potentially fail due to the crushing of the concrete, the analysis
of an upper-bound solution becomes even more challenging.
A suitable alternatively overcoming this shortcoming is the development of elastic-
plastic stress fields, that can be used to obtain exact solutions according to limit analysis in an
automated and time-efficient manner. This type of stress field considers an elastic behaviour
for concrete and steel (following their elastic stiffness) until the plastic plateau is reached. No
tensile stresses are also considered for concrete. This type of stress field, since it accounts for
the conditions of a lower-bound solution (equilibrium and yield conditions) but also for the
compatibility of displacements (stress field calculated on the basis of a displacement field),
allows obtaining exact solutions in an automated manner. In addition to this, the effective

40
 2.4 Suitable approaches for assessing the ultimate strength

concrete compressive strength (refer to Eq.2.3) is estimated in a much more consistent

manner, using the Vecchio and Collins softening equation [120], unlike the ν reduction factors
proposed for the rigid-plastic stress field approach (see Fig.2.19).

2’678 kN
(b) 2’678 kN
(a)
σc / fce ηε
1.0 0.0

0.5 0.5

0.0 1.0

956 kN 1’722 kN 956 kN 1’722 kN

Figure 2.36: Exact solution obtained using the EPSF approach: (a) stress fields; (b) corresponding
kinematics
This is for instance shown in Fig.2.36 where the elastic-plastic stress field (Fig.2.36a)
and its corresponding displacement field (Fig.2.36b) are presented for the investigated
member using the reinforcement layout of Fig.2.23d. The blue lines indicate the intensity and
the direction of the concrete stresses whereas the thickness of the red lines represents the label
of stress in the reinforcement, with dark red lines indicating yielding. The dark surfaces in
Fig.2.36b show the concrete compressive strength reduction according to Vecchio and
Collins [120], marking more significant reduction (which translates into increased transverse
strains of the concrete and therefore more damage) with darker shades of grey.
The failure load according to the EPSF method is 2’678 kN. The difference with
respect to the mechanism calculated using the rigid-plastic approach can mostly be explained
by the vertical reinforcement of the column at the left-hand side of the opening (which was
neglected for the rigid-plastic analysis) and by the role of the compression reinforcement (also
neglected for the rigid-plastic analysis shown previously). It can also be noted that the
computed failure mechanism (from Fig.2.36b) nicely agrees with the selected mechanism at
failure for the rigid-plastic analysis.

41
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

2.5 Collected database and validation of the elastic-plastic stress fields

In order to investigate the accuracy of the EPSF method and explore the limits of its
applicability for analysis of structural concrete elements, a database containing 315 reinforced
and prestressed concrete elements (such are beams, corner frames, dapped-ends, diaphragms
and walls with and without openings) was assembled and placed online at http://i-
concrete.epfl.ch/epsf/epsf.html.
The database contains finite element models of all 315 members, which are available
for download, and can be used to further explore the structural behaviour of the members by
focusing on deformations, failures mechanism, inclination of the compression field in the
elements, etc. The investigations have shown that the kinematics of the EPSF models were
always in agreement with the test results, and so was the ultimate strength. The average ratio
of the actual failure load to predicted one is equal to 1.04 with a fairly low value of the
coefficient of variation (CoV = 0.10 considering all the tests), as it can be seen in Fig.2.37.
Tab.2.5 gives a summary of the online database.

Table 2.5: Summary of the Online Database

N° Ref. Type : N°of el. Cross-section Loading
1 Vecchio et Shim, 2004 [122] RC 9 Rec. CS 3PB
2 Yoon, Cooc et Mitchell, 1996 [126] RC 9 Rec. CS 3PB
3 Sagaseta et Vollum, 2011 [104] RC 7 Rec. CS 3PB/CLCB
4 Mansur et al., 1991 [72] RC 8 Rec. CS CLCB
5 Hong et all., 2002 [45] RC 7 Rec. CS 4PB
6 Sørensen., 1974 [110] RC 10 T CS 4PB
7 Leonhardt et Walter, 1963 [68] RC 16 T CS 4PB
8 Kaufmann et Marti, 1996 [54] RC 4 TT CS PB
9 Nagrodzka-Godycka et Piotrkowski, 2012 [90] RC 12 Dapped-end 4PB
10 Mata Falcón, 2015 [76] RC 50 Dapped-end 3PB
11 Chan, 1979 [19] RC 8 Dapped-end 3PB
12 Khan, 1981 [55] RC 9 Dapped-end 3PB
13 Cook, 1987 [23] RC 3 Dapped-end 3PB
14 Zhu et all., 2003 [128] RC 6 Dapped-end 3PB
15 Herzinger, 2007 [43] RC 18 Dapped-end 4PB
16 Campana et Muttoni, 2011 [17] RC 13 Corner el. PB
17 Placas, 1969 [96] RC 23 Rec. CS + T CS 3PB
18 Bach et al., 1980 [8] RC 12 T CS 4PB
19 Leonhardt et Walter, 1966 [69] RC 2 Rec. CS SBCL
19 Leonhardt et Walter, 1966 [69] RC 2 Cross-beam 3PB
20 Leonhardt et all., 1968 [70] RC 5 Cross-beam 4PB
21 Baumann, Rüsch, 1970 [9] RC 7 Cross-beam 4PB
22 Saqan et Frosch, 2009 [105] PC 9 Rec. CS 3PB
23 Kaufman et Ramirez, 1988 [51] PC 6 TT CS 4PB

42
 2.5 Collected database and validation

24 Kuchma, D. et al., 2008 [60] PC 19 TT CS SBCL

25 Rupf et Muttoni, 2012 [103] PC 13 TT CS CLCB
26 Fernández Ruiz et Muttoni, 2008 [31] PC 6 TT CS 4PB
27 Moore, 2014 [79] PC 11 TT CS 3PB
28 De Wilder K. et al., 2015 [25] PC 6 TT CS 4PB
29 Leonhardt at all., 1973 [66] PC 3 Cross-beam 3PB
30 Büeler et Thoma, 2010 [15] PC 2 Cross-beam 3PB
RC: Reinforced concrete PC: Prestressed concrete
Rec. CS: Rectangular cross section Dapped-end: Dapped end beam
T CS: T cross section Cross-beam : Crossed beams elements
TT CS: Double T cross section
3PB: Three-point bending PB: Pure bending
4PB: Four-point bending SBCL: Simple beam continuous load
CLCB: Concentrated load on a continuous beams

The results from Fig.2.37 are presented in function of four physical parameters
(concrete compressive strength, average prestress, longitudinal and transversal reinforcement
ration).

(a) (b)
2.0

1.5
QTEST/QEPSF

Moving average
1.0

0.5

0
0 20 40 60 80 100 120 140 0 5 10 15 20 25
fc [MPa] σp,inf = Fp,inf /A [MPa]

(c) (d)
2.0
Aver: 1.04
CoV: 0.10
1.5
QTEST/QEPSF

1.0

0.5

0
0 1 2 3 4 5 0 0.5 1 1.5 2 2.5 3
ρ [%] ρw [%]

Figure 2.37: Ratio of measured and estimated strength of the elements from the online database
using the EPSF method in function of: (a) concrete compressive strength; (b) initial
prestress; (c) longitudinal reinforcement ratio; (d) transversal reinforcement ratio

43
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

Each of the graphs contains a red line, which was obtained by computing an average
value of the five nearest points in each graph (averaging values on both axis). This “moving
average”, as it was named, facilitates observing any trends in the results. As it can be seen no
trends have been noted in the results despite the wide range of physical parameters,
investigated by numerous researchers from different institutes (which makes the results even
more satisfying due to the fact that certain level of scatter caused by human error or various
types of measurement equipment is already present).
Tab.2.6 gives the results of EPSF analysis of each experimental campaign (average
and coefficient of variation). Once again it can be seen that the EPSF prediction of the
ultimate resistance is sufficiently accurate for all investigated cases. Elements which failed in
compression due to the spalling of the concrete cover (refer to dapped-end beams in the online
database), required application of link finite elements that introduce the fracture energy of
concrete back into the system (refer to Mata Falcón 2015 [76] for further details).

Table 2.6: Synthesis of the EPSF results

N° Ref. Qtest/QEPSF CoV

1 Vecchio et Shim, 2004 [122] 1.03 0.05

2 Yoon, Cooc et Mitchell, 1996 [126] 0.95 0.07
3 Sagaseta et Vollum, 2011 [104] 1.03 0.09
4 Mansur et al., 1991 [72] 0.99 0.06
5 Hong et all., 2002 [45] 1.06 0.05
6 Sørensen., 1974 [110] 1.16 0.06
7 Leonhardt et Walter, 1963 [68] 1.11 0.06
8 Kaufmann et Marti, 1996 [54] 1.03 0.03
9 Nagrodzka-Godycka et Piotrkowski, 2012 [90] 0.98 0.07
10 Mata Falcón, 2015 [76] 0.99 0.07
11 Chan, 1979 [19] 1.02 0.05
12 Khan, 1981 [55] 1.03 0.07
13 Cook, 1987 [23] 0.99 0.04
14 Zhu et all., 2003 [128] 1.06 0.07
15 Herzinger, 2007 [43] 0.99 0.06
16 Campana et Muttoni, 2011 [17] 0.97 0.06
17 Placas, 1969 [96] 1.04 0.14
18 Bach et al., 1980 [8] 1.14 0.12
19 Leonhardt et Walter, 1966 [69] 1.02 0.04
19 Leonhardt et Walter, 1966 [69] 1.02 0.04
20 Leonhardt et all., 1968 [70] 1.01 0.04
21 Baumann, Rüsch, 1970 [9] 1.06 0.06
22 Saqan et Frosch, 2009 [105] 1.23 0.10
23 Kaufman et Ramirez, 1988 [51] 1.07 0.07
24 Kuchma, D. et al., 2008 [60] 1.09 0.07

44
 2.5 Collected database and validation

25 Rupf et Muttoni, 2012 [103] 1.06 0.05

26 Fernández Ruiz et Muttoni, 2008 [31] 0.98 0.05
27 Moore, 2014 [79] 1.00 0.07
28 De Wilder K. et al., 2015 [25] 0.98 0.04
29 Leonhardt at all., 1973 [66] 1.05 0.03
30 Büeler et Thoma, 2010 [15] 1.01 0.01
Total: 1.04 0.10

However, it can be seen that the EPSF method gives conservative results for a few
elements (Qtest/QEPSF > 1.25). These were the members, which had no transverse
reinforcement (Saqan and Frosh [105]), or had very low concrete compressive strength (less
than 15 MPa). Limit analysis is not applicable in the first case, since members experienced
crack concentration (refer to Campana [16]), and the concrete quality of element in the second
case was not well documented.

(a) (b)
2.0

1.5
QTEST/QEPSF

Moving average
1.0

0.5

0
0 20 40 60 80 100 120 140 0 5 10 15 20 25
fc [MPa] σp,inf = Fp,inf /A [MPa]

(c) (d)
2.0
Aver: 1.02
CoV: 0.08
1.5
QTEST/QEPSF

1.0

0.5

0
0 1 2 3 4 5 0 0.5 1 1.5 2 2.5 3
ρ [%] ρw [%]

Figure 2.38: Selected EPSF results from the Database presented in function of: (a) concrete
compressive strength; (b) inital prestress; (c) longitudinal reinforcement ratio;
(d) transversal reinforcement ratio
After taking such elements out from Fig.2.37, the results become even better (average
Qtest/QEPSF is equal to 1.02 with 0.08 CoV, as presented in Fig.2.38). The two remaining
models with slightly conservative results are beams with significant doweling action

45
Chapter 2: Strut-and-tie models and stress fields as tools for design and assessment

(Sørensen [110] and Placas [96]). Once again, no trends in the results have been observed,
thus confirming the generality and accuracy of the approach.
It is important to emphasise that the current database is the results of the collective work
of multiple PhD students who applied the EPSF during their research at IBETON (Structural
Concrete Laboratory, at École Polytechnique Fédérale de Lausanne in Switzerland):
Kostić [59], Campana [16], Rupf [103], Mata-Falcon [76], Argirova [6, 7], Moccia [79] and
myself (all the models have been revised by the author). The main reason for assembling a
database containing work of multiple authors is to assure the generality of the approach before
introducing it into every-day engineering practice. In this manner additional user related
uncertainties can be taken into account, which would not be the case if the entire modelling
has been done by a single author. The mesh size, shape and refinement, number of iterations,
introduction of concentrated loads in ICONC (FE implementation of EPSF available for free
download at http://i-concrete.epfl.ch in the “applets” menu) was slightly varied by each user,
which had some influence on the final result (more on this will be presented in following
chapter of the thesis). In the future, as more EPSF models are provided the Database is
expected to grow, potentially opening new areas of research.

46
Chapter 3: Advanced modelling of structural
concrete members with EPSF
This chapter is based on the FEDRO (Federal Roads Office) report number 680, titled
“Assessment of Existing Structures Based on Elastic-Plastic Stress Fields and Modelling of
Critical Details and Investigation of the In-Plane Shear Transverse Bending Interaction”,
written by Prof. Aurelio Muttoni, Dr. Miguel Fernández Ruiz, Filip Niketić and Marie-Rose
Backes. This was part of the research project AGB 2009/009, requested by the AGB bridge
research group. The report was published by the OFROU in October 2016.
Contributions of Filip Niketić involved:
1. Conducting a sensitivity analysis of the EPSF method with respect to FE size,
shape, orientation and number of iterations until the convergence;
2. Developing and validating a procedure for modelling rebars with insufficient
anchorage length using the EPSF method;
3. Developing and validating a procedure for modelling indirectly loaded and/or
supported structural concrete members using the EPSF method;
4. Tailoring partial safety factors which can be used with the EPSF method
following the works of P. Tanner et al. [112, 113, 114].

3.1 Sensitivity analysis of EPSF method with respect to its finite element
implementation
Considering the fact that EPSF are developed using finite element analysis, it is
important to investigate the sensitivity of the method with respect to mesh properties. In other
words, it is necessary to show how size, shape and finite element orientation affect the form of
ultimate stress fields and if this has any influence on the estimated strength of a member.
Consequently, it is important to determine what are the limits in which satisfactory accuracy is
ensured, and how much error is introduced in case optimal meshes cannot be applied.
When developing an EPSF, the number of iterations steps until the convergence needs
to be predefined, which could also play an important role in the accuracy of the results. This is
why special attention is given to this parametric study, with a goal to determine the smallest
number of iterations required to obtain satisfactory results.

47
Chapter 3: Advanced modelling of structural concrete members with EPSF

3.1.1 Investigating the influence of finite element size on EPSF analysis

The total number of applied FE can significantly increase the computation time of an
elastic-plastic stress field, meaning that it is beneficial to use as little FE as possible for each
simulation. One of the most effective strategies involves targeted mesh refinement (FE size is
reduced in the critical regions of a structure). Even though this approach is effective, sometimes
the location of the critical regions is not obvious, and the prediction of the licit failure
mechanism is challenging. In such cases FE meshes are best kept uniform, so that the ultimate
stress field can be derived with equal precision in each segment of the structure.
However, the size of applied FE can have an influence on the final results in nonlinear
calculations. Therefore, clear recommendations on how to choose a suitable mesh needs to be
provided. In addition to this it is important to investigate potential problems which might occur
when some of the given recommendations cannot be respected due to various reasons
(geometrical or load constraints for example). The effect which finite element size has on the
development of EPSF at the ULS was investigated using two groups of numerical models:
1. Members with uniform stress state (presented in Fig.3.1a)
2. Members with non-uniform stress state (presented in Fig.3.1b)
Elements subjected to uniform stress state consisted of three groups of reinforced
concrete panels subjected to pure compression, pure tension and pure shear (as can be seen in
Fig.3.1a).

Q Central
Surface

(a) (b)
800
bw bw bw

510 mm2
720 720 720
bw 1’200

720 720 720

3’840
13’50 mm2

Figure 3.1: Geometry properties of structural concrete members used to investigate the sensitivity
analysis of EPSF method: (a) RC Panels subjected to uniform compression, tension and
shear; (b) RC beam subjected to 4-point bending
This allows to investigate the behaviour and stability of EPSF models used to simulate
the basic stress states which can occur in structural concrete elements separately. The sensitivity
analysis also involved a reinforced concrete beam subjected to four-point bending (refer to
Fig.3.1b), which was used to perform the same sensitivity analysis of a non-uniform stress state.
The idea behind it was to investigate a realistic model which combines all three stress states in
a structure with potential redistributions amongst the elements.

48
 3.1 Sensitivity analysis

Four types of meshes (M1 to M4, presented in Fig.3.2) were applied when modelling
each of the structural concrete elements presented in Fig.3.1. Meshes were kept uniform (all
FE had identical shape and orientation), while the area of applied FE triangles was varied
(surface of the FE triangle from mesh M4 is four times the size of that from mesh M1). This
allowed a direct comparison of the results and avoided the influence of any undesired mesh
parameters (such are presence of skewed FE and local mesh refinement).

(a) (b) (c) (d)

Bar FE M4 M3 M2 M1
Node

720
Concrete FE

720 720 720 720

Figure 3.2: Investigation of different finite element mesh size:(a) geometry properties of mesh
M4 – disposition of the concrete FE, bar FE and corresponding nodes;(b) geometry
properties of mesh M3;(c) geometry properties of mesh M2;(d) geometry properties of
mesh M1
As it is presented in Fig.3.2a, each mesh consists of blocks of four concrete FE triangles
which are surrounded with four bar FE. Given the fact that the concrete cannot carry any tensile
stresses (one of the main assumptions of the EPSF method) means that it is important to avoid
models where concrete FE are not connected to at least a single bar FE. In extreme cases where
the spacing of the rebars is such that the corresponding mesh would be unacceptably crude (too
big FE), one could refine it. However, in such cases it is important to pay attention to the
concrete compressive strength reduction factor (ηε) that can be unrealistically low, which would
lead to conservative estimates of failure loads. This can be explained by the fact that assumed
deformations of concrete FE are exaggerated (due to the fact that nothing can resist the tension
forces). In such cases, it is better to smear the reinforcement (place smaller fictitious bars closer
to each other) so that the deformation of concrete FE would be more realistic. It is important to
emphasise that the actual rebar spacing was always sufficient to assure satisfactory behaviour
of EPSF models within the scope of presented investigation, and the reinforcement was never
smeared.
When modelling structural concrete panels (refer to Fig.3.1a), each concrete FE in the
model shared nodes with bar FE in both directions (as it is presented in Fig.3.2), which is an
exaggeration (a more realistic case will be used when analysing the I cross-section beam
presented in Fig.3.1b). However, such disposition was selected in order to avoid any anisotropy
in the RC panels.

3.1.1.1 Structural concrete members subjected to pure compression

Sensitivity of EPSF models simulating the behaviour of structural elements subjected to
pure compression was investigated on a RC panel presented in Fig.3.3c (with reinforcement
ratio equal to 0.1% in both directions). Plastic concrete compressive strength was 38 MPa, and

49
Chapter 3: Advanced modelling of structural concrete members with EPSF

steel yielding strength was 550 MPa. The element was subjected to an uniformly distributed
load over the two facing edges, as presented in Fig.3.3b and c.
The element was modelled using four meshes presented in Fig.3.2 (M1 to M4), and the
corresponding ultimate loads QM1 to QM4 were compared (all models are available for download
at http://i-concrete.epfl.ch/epsf/epsf.html).

(a) (b) (c)

QM4=15.4 MN/m’ QMi
1.50

1.25 bw
QMi / QM1

M1 M3 AMi
1.00 ρx = ρz = 0.1%
M2 M4
0.75 bw = 400 mm

0.50
0 1 2 3 4 5 QM4=15.4 MN/m’ QMi
AMi / AM1

Figure 3.3: Sensitivity of EPSF results obtained after analysing structural concrete elements
subjected to pure compression:(a) ultimate strength estimated using four meshes with
different FE size;(b) ultimate stress field of a model using M4 mesh;(c) geometry
properties and load conditions of analysed member
The results are presented in Fig.3.3a. As it can be observed there is no mesh
dependency, and all models failed at exactly the same load level. The horizontal axis shows the
normalized area of applied FE (area of applied FE triangles was divided with the area of the FE
applied in M1-mesh), and vertical axis shows the normalized failure loads (Qult of each
simulation was divided with the Qult obtained after applying the M1-mesh). The stress field of
the panel at ULS is presented in Fig.3.3b, black FE triangles indicate concrete crushing,
whereas the blue lines mark the direction of the principal compressive stresses.

3.1.1.2 Structural concrete members subjected to pure tension

The sensitivity analysis of EPSF method was further investigated on elements subjected
to pure tension. For this purpose, an element presented in Fig.3.4c was used.

(a) (b) QM4=220 kN/m’ (c) QMi

1.50

1.25 bw
QMi / QM1

M1 M3 AMi
1.00 ρx = ρz = 0.1%
M2 M4
bw = 400 mm
0.75

0.50
0 1 2 3 4 5 QM4=220 kN/m’ QMi
AMi / AM1

Figure 3.4: Sensitivity of EPSF results obtained after analysing structural concrete elements
subjected to pure tension:(a) ultimate strength estimated using four meshes with
different FE size;(b) ultimate stress field of a model using M4 mesh;(c) geometry
properties and load conditions of analysed member

50
 3.1 Sensitivity analysis

The material and geometry properties of a panel subjected to pure tension are the same
as the one subjected to pure compression (fcp=38MPa and fy=550MPa). After applying the
meshes given in Fig.3.2 (M1 to M4) no difference in the governing ultimate load was observed
(as it can be seen in Fig.3.4a). Once more the two axis show normalized values of the FE area
and failure load and the models can be downloaded at http://i-concrete.epfl.ch/epsf/epsf.html.
Ultimate stress field of the panel using the M4-mesh is presented in Fig.3.4b. Dark red
colour of the reinforcement indicates yielding of the steel, and white concrete triangles suggest
that they are not subjected to any stresses.

3.1.1.3 Structural concrete members subjected to pure shear

The method proved to be quite stable in case of elements subjected to uniform shear
stress state as well. This was investigated using a panel PV4 (refer to Fig.3.5c), tested by
Vecchio and Collins [120].

(a) (b) (c)

1.50 QM4=179 kN/m’
QMi
1.25 QTEST / QM4 bw
QM4=179 kN/m’

QM4=179 kN/m’
QMi / QM1

M1 M3 AMi
1.00 QMi ρx = ρz = 1.05 %
M2 M4
QMi
0.75 bw = 70 mm

0.50 QM4=179 kN/m’ QMi

0 1 2 3 4 5
AMi / AM1

Figure 3.5: Sensitivity of EPSF results obtained after analysing structural concrete elements
subjected to pure shear: (a) ultimate strength estimated using four meshes with different
FE size; (b) ultimate stress field of a model using M4 mesh; (c) geometry properties and
load conditions of analysed member
Once again four different meshes, presented in Fig.3.2 (M1 to M4) were applied. No
difference in the ultimate shear stress was observed between the models, and all of them failed
when 2.56 MPa shear force was applied (compared to the measured 2.86 MPa). Element’s
geometry properties and the results of the EPSF analysis are presented in Fig.3.5a. Concrete
compressive strength was 26.6MPa, and structural steel yielded at 242 MPa. Same as before,
horizontal and vertical axis are normalized and the applied models can be found at http://i-
concrete.epfl.ch/epsf/epsf.html.
The ultimate stress field of the panel using the M4 mesh is presented in Fig.3.5b. Dark
red colour of the reinforcement indicates yielding of the steel, whereas the dark grey triangles
indicate the solicitation of the concrete. In this case concrete compressive strength was
uniformly reduced over the entire surface of the panel (ηε=0.278) and the solicitation of the
concrete was equal to 69% (σc3/fce=0.69).
Previous example showed that the FE size does not have any influence on EPSF result
in case of symmetrically reinforced structural concrete panels subjected to pure shear (ρx = ρz).
In order to push the limits of the sensitivity analysis even further, the same panel was

51
Chapter 3: Advanced modelling of structural concrete members with EPSF

investigated assuming different reinforcement ration in horizontal and vertical direction

(ρx = 0.5×ρz). The final results are given in Fig. 3.6.

(a) 1.50 (b) (c)

QM4=153 kN/m’
QMi
1.25 bw

QM4=153 kN/m’

QM4=153 kN/m’
QMi / QM1

M1 M3 AMi
1.00 QMi ρx = 0.53%
M2 M4 ρz = 1.05 %
QMi
0.75 bw = 70 mm

0.50 QM4=153 kN/m’ QMi

0 1 2 3 4 5
AMi / AM1

Figure 3.6: Sensitivity of EPSF results obtained after analysing structural concrete elements
subjected to pure shear: (a) ultimate strength estimated using four meshes with different
FE size; (b) ultimate stress field of a model using M4 mesh; (c) geometry properties and
load conditions of analysed member
Ultimate strength of an unsymmetrically reinforced concrete panel estimated with the
EPSF method did not depend on the applied FE size (all four model failed at exactly the same
load, as presented in Fig. 3.6a and b). Reinforcement yield in both directions and concrete
compressive strength was uniformly reduced to approximately 15% of its initial value
(ηε=0.145). The solicitation of compressive struts was at 100% (σc3/fce=1.00), which is very
important to underline. Therefore, it can be concluded that the EPSF method governs identical
ultimate loads of members subjected to uniform shear stresses regardless of applied FE size.

3.1.1.4 Structural concrete beam under four-point-bending

In order to investigate the sensitivity of EPSF models that can experience stress
redistributions (as is usually the case in real structures), meshes presented in Fig.3.2 were used
to model a web of a reinforced concrete beam with I cross-section subjected to four-point
bending.
2’640
Q
5×100mm2 Area: 810mm2
CL
uniform mesh M1, M2, M3 and M4 3×170mm2
Bar FE
Node
240
1’200

Concrete FE
720

bw = variable
240

bf = 800 mm
3×4500mm2
720 2’400 720
3’840

Figure 3.7: Geometry properties of an I-cross section beam used to investigate the influence of
finite element size on EPSF results – disposition of concrete FE, bar FE and
corresponding nodes

52
 3.1 Sensitivity analysis

Geometry and mesh properties of analysed beam (modelled using M4 mesh) can be seen
in Fig.3.7. The beam was designed to experience shear failure of its central (web) region. Web
provided a space with uniform geometrical properties, which allowed direct application of
meshes M1 to M4. Total amount of transverse steel in the web (810 mm2) was kept constant for
all investigated models. Vertical reinforcement was smeared between concrete finite elements
as presented in Fig.3.7 (4 concrete FE triangles were placed between 2 bar FE). It is important
to emphasize that in this case (contrary to previous investigations of elements subjected to pure
compression, tension and shear), applied meshes could not be entirely uniform, due to the
presence of flanges where slightly elongated finite elements were applied. However, this
influence became negligible by forcing the beam to fail inside the web.
1.50

1.25

M3
QMi / QM1

M1
1.00
M2 M4

0.75
ρw=1.13% ρw=0.34%
ρw= 0.68% ρw=0.17%
0.50
0 1 2 3 4 5
AMi / AM1

Figure 3.8: Sensitivity of EPSF results obtained after analysing structural concrete elements
subjected to non-uniform stress state with ultimate strength estimated using four meshes
with different FE size on four beams with different transverse reinforcement ratio
In order to cover a wide range of shear failure modes (starting from yielding of the
stirrups, to crushing of the concrete) four beams having four different web thicknesses were
analysed:
1. bw = 200 mm corresponding to ρw = 0.17%
2. bw = 100mm corresponding to ρw = 0.34%
3. bw = 50mm corresponding to ρw = 0.68%
4. bw = 30mm corresponding to ρw = 1.13%
where ρw presents the transverse shear reinforcement ratio.
Plastic concrete compressive strength was equal to 38 MPa and steel yield strength was
equal to 550 MPa.
As it can be seen in Fig.3.8, EPSF models showed some mesh dependency. Application
of smaller FE governed lower failure loads for all four transverse reinforcement ratios.
However, it can be observed that the difference between the ultimate loads obtained using
meshes M1 and M4 was almost 9% in case RC beam had ρw = 1.13%, whereas this difference
dropped down to around 5% and 1% when relative amount of transverse reinforcement was

53
Chapter 3: Advanced modelling of structural concrete members with EPSF

equal to ρw = 0.34% and ρw = 0.17%. This means that the stability of the results depends on the
type of failure. Structural members which experienced concrete crushing without yielding of
the transverse reinforcement (refer to ρw = 1.13%) are more sensitive to FE size compared to
elements that failed in concrete crushing along with yielding of the steel (refer to ρw = 0.17%).
ρw = 0.17%

ρw = 1.13%
Legend:

ρw = 0.34%

ρw = 0.68%
SF

SF

SF

SF
η

ηε

η
ε

ε
σc / fce

M4: t4 ~ 30 s
0.0

1’039 kN

776 kN

552 kN

400 kN
0.5

CL

CL

CL

CL
1.0

SF

SF

SF

SF
η
η

ηε

ε
ε

M3: t3 ~ 1 min
ηε
1.0

1’027 kN

764 kN

386 kN
536 kN
0.5

CL
CL

CL

CL
SF

SF

SF

SF
0.0

η
η

ηε

ε
ε

M2: t2 ~ 2 min
Compression

1’022 kN

754 kN

518 kN

373 kN
CL
CL

CL

CL
Traction

SF

SF

SF

SF
η
η

ηε

ε
ε

M1: t1 ~ 3 min 30 s
Yielding

concentration
Strain

concentration
Strain

concentration
Strain

concentration
Strain
1’020 kN

740 kN

513 kN

368 kN
CL
CL

CL

CL

Figure 3.9: Stress fields and concrete compressive strength reduction factor for I-cross section
beams having four different transverse reinforcement ratios analysed with four finite
elements mesh sizes

54
 3.1 Sensitivity analysis

The reason for this can be observed in Fig.3.9, which shows the stress field (SF), the
concrete compressive strength efficiency factor (ηε) as well as the calculation time for all sixteen
analysed beams (four different web thicknesses modelled using four meshes M1 to M4, available
to download at http://i-concrete.epfl.ch/epsf/epsf.html). All models failed locally due to
crushing of the concrete close to one of the flanges (top flange in case of models with higher
transverse reinforcement ratio, and bottom flange in case of the beams with lower transverse
reinforcement ratio). As the FE area decreased, so did the size of the damaged zone. Since
models did not have any longitudinal reinforcement in the web, there was nothing to evenly
spread the deformations (especially in case of ρw = 1.13%). Therefore, the concrete compressive
strength was significantly reduced in a single FE row – in other words, strain concertation was
observed. Since the reinforced concrete beam was only numerically analysed (it was not a part
of an experimental campaign) it is not sure whether or not the strain localization close to the
flange would actually appear in real life of not.
The reason why the beams with lower transverse reinforcement ratio (ρw = 0.34% and
ρw = 0.17%) were less affected by this phenomenon lays in the fact that the stains were much
better averaged over the web once the stirrups yielded. Since concrete compressive strength
reduction factor (ηε) is directly depended on the principal tensile strains, the ultimate strength
of the reinforced concrete beams with higher transverse reinforcement ratio was much more
affected by the reduction of FE size.
Considering the fact that the area of applied FE in M1 models is four times smaller than
the one in M4 proves that the results are reasonably stable even in case of non-uniform stress
fields. Higher drop in ultimate load was observed on a series of beams which had significantly
high amount of transverse reinforcement (more than 1%), which is usually not the case in
engineering practice. When this ratio dropped down to 0.4% or less, the results were much more
stable.
It can be concluded that the EPSF method gives stable results for various sizes of applied
FE. A general recommendation when selecting an optimal FE size is to use the stirrup spacing
as main grid for meshing, and place two concrete FE triangles in between (as presented in
Fig.3.7). Using this criterion along with the recommendations concerning the FE shape and FE
orientation (that will be presented in following sections) ensured satisfactory behaviour of
EPSF models. Online database was generally modelled in this manner, and final results proved
to be quite accurate when compared to experimentally measured values (average QTEST/QEPSF
equal to 1.04 with coefficient of variation equal to 0.10).

3.1.2 Investigating the influence of finite element shape on EPSF analysis

The shape of applied finite elements is another parameter which affects EPSF analysis.
In order to investigate its significance and provide practical recommendations for meshing, a
reinforced concrete beam with 100 mm thick web (corresponding to ρw = 0.34%), presented in
Fig.3.7 was investigated using 5 additional meshes:
1. MDR that served as a benchmark;
2. MD1 to MD4 whose geometry properties are given in given in Fig.3.10.

55
Chapter 3: Advanced modelling of structural concrete members with EPSF

(a) (b) (c) (d) (e)

a:h=1:1 MDR a:h=1:2 MD1 a:h=1:3 MD2 a:h=1:4 MD3 a:h=1:5 MD4
720 720 720 720 720

720
Concrete FE h h h h h

Bar FE a a a
a FE Node a

Figure 3.10: Investigation of the finite element distortion on EPSF results: (a) geometry properties of
the reference mesh MDR – disposition of concrete FE, bar FE and corresponding nodes;
(b) geometry properties of mesh MD1; (c) geometry properties of mesh MD2;(d) geometry
properties of mesh MD3; (e) geometry properties of mesh MD4
Finite element distortion was defined as ratio between the height of a constant strain
triangle and length of its corresponding side (a:h ratio presented in Fig.3.10). As it can be seen
this ratio varied from 1:1 to 1:5.
Meshes MDR to MD4 (presented in Fig.3.10) were used to form beam webs of all five FE
models. The results of EPSF analyses are given in Fig.3.11, which shows that the ultimate load
was quite stable regardless of applied mesh. Maximal difference between the reference model
(MDR) and the most distorted one (MD4) was approximately 2%, proving that the presence of FE
distortion has little influence on the final results.

1.50

1.25
QMDi / QMDR

MDR MD2 MD4

1.00
MD1 MD3

0.75

0.50
1 2 3 4 5
Models

Figure 3.11: Sensitivity of EPSF results obtained after analysing I cross-section beam using five
meshes with different FE shape
Based on the general recommendations for FE analysis found in literature [35], maximal
triangle height-to-side ratio used in practice should not be greater than 1:3. A fact that the
difference in results is not too big even if this ratio is larger than the recommended values is
definitely reassuring, but should be avoided as much as possible. Meshes containing zones with
distorted finite elements due to the complex geometry of the element for example can be locally
accepted. All 5 models can be downloaded from http://i-concrete.epfl.ch/epsf/epsf.html.

56
 3.1 Sensitivity analysis

3.1.3 Investigating the influence of finite element orientation on EPSF analysis

Orientation of finite element triangles forming a mesh has an effect on the final results.
The fact that the hypotenuses of the constant strain triangles are parallel throughout a model
affects the shape of the ultimate EPSF as well as the value of ultimate strength.
In order to investigate this phenomenon, the reinforced concrete beam presented in
Fig.3.7 (with web thickness of bw = 100mm, corresponding to ρw = 0.34%) was modelled once
more, using a regular mesh, formed out of FE triangles with 1:1 height-to-side ratio. Triangles
were first orientated in a way which allowed their hypotenuses to descend from top left to
bottom right corner (refer to Fig.3.12a), and then vice versa (refer to Fig.3.12d). Zig-zag FE
orientation was also analysed (refer to Fig.3.12c) and finally, a mesh with random FE triangle
orientation was applied (refer to Fig.3.12b). Fig.3.12 shows all 4 meshes, along with
corresponding stress fields at ULS and concrete compressive strength effectiveness factors (ηε).
CL CL CL
(a) M QO1 822 kN
O1

Strain
concentration

(b) M QO2 798 kN

O2

Strain
concentration

QO3
(c) M 768 kN
O3

Strain
concentration

QO4 764 kN
(d) M
O4

Strain
concentration

Figure 3.12: Investigation of the finite element orientation on EPSF results - geometry properties,
corresponding stress field and concrete compressive strength reduction factor - ηε of:
(a) mesh MO1; (b) MO2; (c) MO3; (d) MO4
As it can be seen, the difference between the ultimate loads is not negligible and can be
as high as 8% (Qult=822 kN in case of MO1 compared to Qult=764 kN in case of MO4). In addition
to this, difference in shape between the four stress fields can be also observed. If an angle
between the hypotenuses of the concrete FE triangles and the direction of the principle
compressive stresses is small, compressive stress field tends to be more concentrated governing
a lower ultimate load (refer to MO4).

57
Chapter 3: Advanced modelling of structural concrete members with EPSF

On the other hand, if the finite element hypotenuses are almost perpendicular to the
principle compressive stresses, the stress field tends to spread over a wider web area, governing
a higher ultimate load (refer to MO1).

(a) (b) σc,2

F2=√2/4F F2=√2/4F
2 σc,2 2
σc,2
½l √2l √2l

σc,2 ½l F3=√2/4F ½l ½l
F3=0
1 3 1 3
F1=√2/2F F1=√2/4F
σc,2 σc,2 σc,2
σc,2

Figure 3.13: Derivation of the nodal forces from concrete stresses for: (a) FE with descending
hypotenuses; (b) FE with ascending hypotenuses
The explanation behind this behaviour lies in the process of nodal force derivation,
which depends on the angle between the principle stress direction and each side of the constant
strain triangles (refer to Fig.3.13). Assuming that the direction of the principal concrete
compressive stress is equal to 45°, nodal forces have been derived for two cases FE with
descending hypotenuses from top-left to bottom-right and vice-versa. It can be clearly seen that
application of the first mesh (presented in Fig.3.13a) activates all three nodes, contrary to the
second mesh, which activates only two (as shown in Fig.3.13b). In case of a uniform stress state
in which principal compressive stress direction does not change over a certain region of FE (for
example in panels presented in Fig.3.5), the two meshes will govern identical resistances.
However, when the direction changes, application of the first over the second FE results in
larger ultimate strength, due to the fact that a single concrete FE generates 3 instead of 2 nodal
forces and distributes them more evenly across the mesh (as it was observed in Fig.3.12).
1.50

1.25
QMorien,i / QMorien, 3

MO1
MO3
1.00
MO2 MO4

0.75

0.50
1 2 3 4
Models

Figure 3.14: Sensitivity of EPSF results obtained after analysing I cross-section beam using four
meshes with different FE orientation
In order to overcome this, suitable FE mesh should not have any preferential
hypotenuses inclination. Instead, zig-zag FE (refer to MO3 in Fig.3.12) or random FE inclination
(refer to MO2 in Fig.3.12) are preferred. However, it is important to mention that in case of
random FE inclination it is necessary to perform more than one simulation in order to obtain
acceptable results.

58
 3.1 Sensitivity analysis

Fig.3.14 summarises failure loads obtained with meshes presented in Fig.3.12. Ultimate
resistances are normalized using the value acquired with a zig-zag FE disposition
(corresponding to Qult=768 kN governed by MO3 mesh from Fig.3.12). It can be concluded that
the application of meshes which favour certain FE orientation should be avoided as much as
possible. Meshes with zig-zag inclination of hypotenuses or the ones with horizontal and
vertical FE hypotenuses (presented in Fig.3.2) should be selected instead. In case a random FE
orientation is applied, average value of at least 5 different simulations should be taken as
representative (unfortunately this can be time consuming depending on the size of analysed
models). All 8 FE models that were investigated in this chapter can be found at http://i-
concrete.epfl.ch/epsf/epsf.html.

3.1.4 Convergence and required number of iterations for an EPSF analysis

EPSF are developed based on assumed displacement fields, which are validated through
iterations. This process can lead to the convergence of the model (thus obtaining exact solutions
according to the theory of plasticity) or its divergence (failure of the element). Convergence
process is expressed through an error which is displayed at the end of each iteration, and
represents the ratio between the vector sum of residual (un-equilibrated) nodal forces and the
vector sum of applied loads.
During a calculation process, this error initially increases (refer to the two peaks in
Fig.3.15), and then gradually decreases ultimately reaching a stable asymptotic value. If this is
the case, the analysed model converged (refer to red line in Fig.3.15). Otherwise the model
diverged indicating that the applied loads are too high (presented with black line in Fig.3.15).
It can be seen that the difference in the final error between the two simulations presented in
Fig.3.15 is quite significant, making it quite clear if a simulation in question converged or not.
50
Divergence
Q = 755 kN
40

30
Error

20

10

0
Convergence
Qult = 754 kN
0
50 100 150 200 250 300
Iteration steps

Figure 3.15: Convergence and divergence of an EPSF simulation

Element in question was the reinforced concrete beam from Fig.3.7 (web thickness
bw = 100mm, corresponding to ρw = 0.34%) applying the M2 mesh (refer to Fig.3.9). The model
converged under 754 kN load but diverged under 755 kN (as presented in Fig.3.15), thus

59
Chapter 3: Advanced modelling of structural concrete members with EPSF

indicating that its ultimate strength is equal to 754 kN. Aside from looking at the error and
visual validation of the obtained stress field, two more checks should be performed in order to
validate if the EPSF is exact or not. The sum of the reaction forces should correspond to the
applied loads, and the deformation of the analysed element should be the biggest at ULS. As
the applied load approaches its maximal value, sometimes even if the simulation converged
(error is acceptably small), element deformation is smaller for higher loads, which is physically
impossible. In this case the biggest load that governs the largest deformations at the same time
should be selected as governing.
Graph presented in Fig.3.15 rises a bit of concern, since there are no indications that any
of the two models will diverge until the 150th iteration. This could lead to a conclusion that if
the predefined number of iterations was 150, instead of 250 steps, both simulations would
converge and therefore overestimate the element strength. This is why a sensitivity analysis of
this parameter (number of necessary iteration steps) was investigated.

1.50

1.25
Qult,i / Qult, 250

Qult, 250=754 kN
1.00

0.75

0.50
0 200 400 600 800 1000
Total number of iterations

Figure 3.16: Estimated ultimate strength of an I cross-section reinforced concrete beam using EPSF
analysis with different number of iterations
The ultimate strength of the same reinforced concrete beam from Fig.3.15 was estimated
using different number or iterations. The obtained results are presented in Fig.3.16. It can be
seen that the difference between estimated ultimate loads obtained with 250 and 1000 iterations
is negligible (in this case there is no difference at all). Moreover, the value of maximal load
became quite stabile after performing simulations with 100 iteration steps. Even though, the
given example indicates that 100 iteration steps are sufficient for achieving sufficient accuracy,
after modelling 315 structural concrete elements from the online database, general
recommendation for the minimal number of required iterations is 250 steps.

60
 3.2 Modelling elements with insufficient anchorage length

3.2 Modelling elements with insufficient anchorage length using EPSF

Elastic-plastic stress field method assumes perfect bond behaviour, meaning that there
is no displacement between the nodes of concrete and bar FE. Consequently, yielding of the
reinforcement is possible regardless of the provided anchorage length, and a rebar can be fully
activated already within the first FE of the model.
Previous assumption differs from reality, where complex stress-transfer mechanism
between the two materials exists. In case re-bars are not properly anchored, effects of bond need
to be indirectly introduced in EPSF, otherwise the obtained solutions might be unconservative.

(a) (c) lanch

li

Øi
Øeff,i Øeff,max
Øj

(d) 30
Øeff,max
25
Øeff,max
(b)
20
Øeff [mm]

τb0
15 Øeff,i
Rebar yielding

10
τb1
5 2 bars Ø19
fc=23.5MPa
0
δ1 0 50 100 150 200 250 300 350 400 450
l [mm]

Figure 3.17: Modelling rebars with insufficient reinforcement anchorage using EPSF approach:
(a) actual and effective rebars introduced to EPSF model; (b) concrete-to-steel bond
stress; (c) varying bar FE area over rebar’s anchorage length; (d) effective bar
diameter representing two Ø19 rebars imbedded in 23.5 MPa concrete accounting for
their actual anchorage length
Fig.3.17 shows the modelling process of two rebars Ø19 with insufficient anchorage
length (yield strength of the steel is equal to 390 MPa), imbedded in 23.5 MPa concrete. Actual
disposition of the bars is presented on the top of Fig.3.17a. However, the two rebars cannot be
modelled separately according to EPSF method, and instead need to be replaced with a single
(effective) bar (refer to the bottom Fig.3.17a), whose diameter is calculated in the following
manner:

 f y ,i 

 eff ,max     2
i  
f y ,eff 
(3.1)

where Øi represents the diameter of the rebars which are placed at the same cross-section depth

61
Chapter 3: Advanced modelling of structural concrete members with EPSF

fy,i represents the yield strength of corresponding rebar;

fy,eff represents the yield strength of effective rebar.
Concrete-to-steel bond law can be assumed as rigid-plastic, according to the Tension
Chord Model (TCM), introduced by Marti et al. [74]. The TCM adopts a simplified bond stress-
slip relationship (refer to Fig.3.17b) that assumes a constant transition of bond stress between
reinforcement and concrete equal to:
τ b 0  2  f ctm , up to yielding point of ordinary ribbed bars, and (3.2)

τ b1  f ctm , once the bars have reached yielding. (3.3)

Mean tensile strength of concrete is represented with fctm in previous equations (Eq.3.2
and Eq.3.3), and according to MC2010 [34] it can be estimated in the following manner:

0.3   f c 2 / 3 , for f c  50MPa

f ctm   (3.4)
2.12  ln(1  0.1  f c ), for f c  50MPa
where fc represents the uniaxial concrete compressive strength measured in cylinder.
The anchorage length necessary to fully yield a rebar can therefore be estimated
assuming bond stress equal to τb0 (refer to Eq.3.2) in the following manner:

  fy
lanch  (3.5)
4  b 0

where Ø represents the actual re-bar diameter;

fy represents the yielding strength of the reinforcing steel.
It is important to emphasise that in case the actual anchorage length of the rebar is equal
or larger than the one calculated in Eq.3.5, reinforcement can be modelled using the equivalent
re-bar diameter throughout its entire length, thus keeping the model as simple as possible.
Otherwise, the applied diameter needs to be reduced over the calculated anchorage length (refer
to Eq.3.5), as presented in Fig.3.17c. In this manner, the amount of tensile force that can be
taken by a single bar FE is physically limited, which is an indirect way of accounting for the
bond stress.
Since the geometry of the rebar is discretized using bar FE, the reduction of the effective
diameter is done in steps. Each FE assumes the average diameter of the bar over its length using
the following equation:

li

 eff ,i  
 eff ,max  (3.6)
lanch

where Øeff,i represents the diameter of a rebar in ith FE;

Øeff,max represents the effective re-bar diameter (refer to Eq.3.1);
lanch presents the minimal necessary anchorage length (refer to Eq.3.5);

62
 3.2 Modelling elements with insufficient anchorage length

li presents the distance between the beginning of the bar and the mid-point of the ith FE (refer
to Fig.3.17c).
The reduction of the effective rebar diameter over an anchorage length of two Ø19
rebars (yield strength of the steel equal to 390 MPa), imbedded in concrete that has a
compressive strength of 23.5 MPa is presented in Fig.3.17d. In order to limit the maximum
force that can be taken by the rebars the effective diameter was varied from 4.9 mm to 26.9 mm,
over the anchorage length of 375 mm. Each point in the graph represents the effective dimeter
of a rebar corresponding to a FE from Fig.3.17c.

Table 3.1: Geometry and material properties of RC beams tested by Hong et al. [45]
N° Spec. a/d sw [mm] lanch [mm] N° Spec. a/d sw [mm] lanch [mm]
1 SS-1 0.75 250 78 5 LBS-1 1.00 250 0
2 SS-2 1.00 250 78 6 LBS-2 1.00 250 156
3 SS-3 1.25 250 78 7 VSR-1 1.00 200 78
4 SS-4 1.5 250 78 8 VSR-2 1.00 150 78
d = 520 mm; fc = 23.5 MPa; fy = 392 MPa
The presented modelling strategy was used to investigate a series of reinforced concrete
beams subjected to four-point bending tested by Hong et al. [45]. The elements were designed
to fail in shear due to an insufficient anchorage length. Their geometrical and material
properties are given in Tab.3.1, and the results of the EPSF analysis are presented in Tab.3.2.
All the FE models can be downloaded from http://i-concrete.epfl.ch/epsf/epsf.html.

Table 3.2: Results of EPSF analysis of RC beams tested by Hong et al. [45]
Qtest QEPSF Qtest Qtest QEPSF Qtest
N° Spec. N° Spec. Aver. CoV
[kN] [kN] / QEPSF [kN] [kN] / QEPSF
1 SS-1 2×331 2×305 1.09 5 LBS-1 2×199 2×185 1.08
2 SS-2 2×305 2×269 1.13 6 LBS-2 2×290 2×288 1.01
1.06 0.05
3 SS-3 2×280 2×252 1.11 7 VSR-1 2×297 2×282 1.05
4 SS-4 2×240 2×247 0.97 8 VSR-2 2×329 2×304 1.08
As it can be seen, the presented procedure gives satisfactory results when compared to
measured values (average Qtest/QEPSF = 1.06), with very small dispersion (coefficient of
variation – CoV equal to 5%). Detailed results of the EPSF analysis are presented in Fig.3.18
and Fig.3.19. Each figure shows a stress field of a specimen (corresponding to its ULS), the
reduction of concrete compressive strength (ηε) along with its deformed shape, and a sketch of
the crack patterns observed by Hong et al. [45] at failure. Fig.3.18 summarises the EPSF results
corresponding to the beam LBS-1, which had longitudinal reinforcement without any
anchorage length (refer to Tab.3.1 for more details). Looking at the ultimate stress field (see
Fig.3.18a) it can be seen that the strength of the model is governed by local failure of the
concrete above the support plate (induced by the reduction of concrete compressive
strength - ηε) along with the local yielding of the rebars (refer to dark red sections of the
longitudinal bars in Fig.3.18b). Even though the concrete was locally crushed in this case, it is
actually the reinforcement which is limiting the amount of compression which can be deviated
by the between the loading plate and the supports. Looking at the ultimate crack pattern

63
Chapter 3: Advanced modelling of structural concrete members with EPSF

(presented in Fig.3.18c), it can be observed that a shear crack which originated from the support
plate caused the failure of the system. Such failure was characterized as anchorage failure by
Hong et al [45]. When comparing the results from Fig.3.18c to disposition of concrete
compressive strength reduction factors from Fig.3.18b, a clear analogy can be observed. The
EPSF results correspond quite well to the behaviour observed in experimental campaign. Both
EPSF model and the tested beam experienced the same anchorage failure.

185 kN C 185 kN 199 kN C

(a) L (b) C
L (c) L

Anchorage
failure LBS 1

185 kN 185 kN 199 kN

Figure 3.18: Results of EPSF analysis of a beam LBS 1 investigated by Hong [45]: (a) stress field at
ULS; (b) deformed shape of the beam with concrete compressive strength reduction
factor - ηε; (c) observed crack pattern and measured strength of the element
Similar conclusion can be drawn after analysing Fig.3.19 that gives detail of the EPSF
simulation of beam VSR-1 (refer to Tab.3.1 for more details). Crack pattern at ULS (refer to
Fig.3.19c) correspond quite well to the disposition of the concrete compressive strength
reduction factor (refer to Fig.3.19b). The EPSF failed due to concrete crushing induced by the
local yielding of the rebar close to the support. Such behaviour corresponds well to observed
anchorage problems indicated by Hong et al. [45].

282 kN C C 297 kN
L 282 kN L CL
(a) (b) (c)

Anchorage
failure VSR 1

282 kN 282 kN 297 kN

Figure 3.19: Results of EPSF analysis on a beam VSR 1 investigated by Hong [45]: (a) stress field at
ULS; (b) deformed shape of the beam with concrete compressive strength reduction
factor - ηε; (c) observed crack pattern and measured strength of the element
An alternative approach to what was described in this chapter involves reduction of the
yield strength of the steel over the anchorage length of a rebar. An argument for doing so would

64
 3.2 Modelling elements with insufficient anchorage length

be to maintain the axial stiffness of the element. However, this could result in underestimating
the deformations of the analysed member (due to the fact that the slip is neglected and the
reinforcement maintained its stiffness). On the other hand, the amount of tensile force which
can be taken by each rebar would be the same compared to what was obtained after reducing
the effective rebar diameter. Another approach would be to actually model the bond stress
transfer mechanism (presented in Fig.3.17b) by introducing additional spring FE. This would
significantly increase the computational time of the elastic-plastic stress fields, and would still
represent an approximation of the reality. At the same time, the amount of tensile force which
can be taken by a single rebar at ULS would be identical to what is obtained by reducing the
effective bar diameter. In order to keep the EPSF models as simple as possible without
decreasing their accuracy at ULS, a general recommendation for taking into account the
insufficient anchorage length of the rebars is to simply reduce their effective diameter.
It can be concluded that the presented modelling procedure gives satisfactory results when
compared to test results. Ultimate resistance and failure modes are accurately predicted with
EPSF, and a clear view inside the stress distribution of analysed elements is given. Finally, it
should be noted that the presented approach governs slightly conservative solutions by
definition, since the bond properties are enhanced by confining concrete stresses, which was
here not taken into account.

65
Chapter 3: Advanced modelling of structural concrete members with EPSF

3.3 Analysis of indirectly supported structural concrete members using EPSF

Elastic-plastic stress field method proved to be a powerful tool when analysing
individual structural concrete members (simple beams, deep beams, dapped ends, wall with
openings etc. which can be modelled as plane members - 2D members). It was able to
accurately predict structural resistance along with the governing failure mode and give a clear
view of the distribution of internal forces at ULS of tested specimens (as can be seen at http://i-
concrete.epfl.ch/epsf/epsf.html). However, in reality individual elements are often connected
into statically indeterminate systems, which are able to facilitate additional internal stress
redistribution. Even though such systems can be analysed element-per-element by developing
each stress field separately and imposing boundary and loading conditions in between, actual
structural behaviour might be overlooked due to such crude assumptions. This becomes
especially important when assessing the ultimate resistance of existing structures, when all
potential load-carrying mechanisms should be accounted for (so that any unnecessary structural
strengthening can be avoided). In order to investigate this complex phenomena, the EPSF
method has been used to analyse behaviour of crossed structural concrete members (reinforced
and prestressed concrete beams that have been indirectly loaded and/or supported) found in
literature.
Crossed members as load carrying mechanisms in civil engineering are most commonly
used in bridge diaphragms, where they allow bearings to be placed outside of bridge-web planes
(as presented in Fig.3.20). Bridge diaphragms of box-girder bridges also allow development of
uniform torsion conditions in order to resist asymmetrical cross-section loading in more
efficient manner.

Suspension reinf. Diaphragm B

A
Deck

Figure 3.20: Bridge diaphragm – principle function and strut-and-tie model

Looking back at Fig.3.20, it can be seen how the internal forces are deviated from the
longitudinal webs, through suspension reinforcement to bridge diaphragms in order to reach the
support. In the perspective of EPSF method, this means that two simultaneous simulations are
required in order to analyse the stress states of members in different planes in order to correctly
distribute shear force transfer from one element to another. Analysing elements separately and
assuming that the entire shear force is suspended from point A (see Fig.3.20) can be too
conservative, while assuming that it is entirely applied in point B (Fig.3.20) might be unsafe.

66
 3.3 Analysis of indirectly supported structural concrete members

Leonhardt F., Walther R.

fc: 28 MPa
Fp,inf/A: 0 MPa
a/d: 0.51 - 1.02
ρw: 0.38 - 0.71%

Baumann T., Rüsch H. Leonhardt F., Walther R., Dilger W.

fc: 48 - 59 MPa
Fp,inf / A: 0 MPa fc: 25 - 30 MPa a / d: 3.05 - 5.17
a / d: 1.64 - 3.97 Fp,inf / A: 0 MPa ρw: 0.16- 0.86%
ρw: 0.37 - 0.80%
Leonhardt F., Kroch R., Rostásy F.

fc: 30 - 40 MPa a / d: 1.82 - 3.94

Fp,inf / A: 4.30 MPa ρw: 0.70- 1.01%
Büeler Ch., Thoma K.

fc: 34 MPa a / d: 1.45 - 4.36

Fp,inf / A: 2.70 MPa ρw: 0.73%

Figure 3.21: Mechanical and geometrical properties of crossed structural concrete members
investigated with EPSF method
Placing the suspension reinforcement exactly in the crossing zone of the longitudinal
and transversal elements can be physically impossible (due to limed space or construction site

67
Chapter 3: Advanced modelling of structural concrete members with EPSF

constraints). In such cases suspension reinforcement is distributed over a wider zone which
brings up a question whether such arrangement can still effectively transfer the shear force, and
if this can affect the failure mechanism of the system. However, the most important question
that will be addressed in this chapter is: can the EPSF method be used to correctly predict the
ultimate capacity and failure mechanisms of such systems and with what accuracy?
For the purpose of this research a database containing 19 results form 5 different
experimental campaigns was compiled:
1. Leonhardt F. and Walther R. (1966) - 2 tests [69]
2. Leonhardt F., Walther R. and Dilger W. (1968) - 5 tests [70]
3. Baumann T. and Rüsch H. (1970) - 7 tests [9]
4. Leonhardt F., Koch R. and Rostasy F. (1973) - 3 tests [66]
5. Büeler Ch. and Thoma K. (2010) - 2 tests [15]

(a) (b)
Crossing Crossing
area area
FE mesh
FE mesh

Crossing
area FE mesh Rigid bars
Rigid bars
1
Crossing Crossing
area area 2 1
3 FE mesh
2
FE mesh 4
5
6
7
Rigid bar 1-1
Rigid bar 2-2

1
2 1
3
Crossing 4 2
area 5
6
7
FE mesh

Figure 3.22: Modelling technique for crossed-members using EPSF method: (a) decomposition of
the analysed specimen; (b) connecting crossed-members with rigid bar FE
The first three experimental campaigns contain only reinforced concrete specimens,
whereas the last two applied prestressed concrete beams. All five experimental campaigns were
conducted in order to better understand the mechanism of beam to beam shear transfer action,
and as such are adequate to simulate behaviour of bridge diaphragms (reinforcement details and
main parameter are given in Fig.3.21).

68
 3.3 Analysis of indirectly supported structural concrete members

The modelling technique that was applied in order to analyse crossed members by means of
EPSF method is presented in Fig.3.22.
In this figure, a crossed system consists of 4 elements: 3 transverse walls connected by
a longitudinal wall (as can be seen in the middle of Fig.3.22a). The 2 exterior transverse
elements act as supports while the load is applied indirectly through the central transverse wall.
In order to analyse the entire system using EPSF method all specimens need to be placed in a
single plane (as presented in Fig.3.22b). In a given example this means that the transverse
members need to be rotated for 90° (refer to Fig.3.22b) and placed above the longitudinal
element in case they serve as supporting members, or below it in case they are used to introduce
the loads. Finally, finite elements in the crossing zone need to be linked node-by-node using
rigid bars in order to impose equal nodal displacement between the members (refer to
Fig.3.22b). In order to keep Fig.3.22b more comprehensible, FE mesh was not shown for the
entire specimen, but only for its crossing zones. Node 1 placed both in longitudinal and
transversal walls are connected using the Rigid bar FE 1-1. The same is valid for the Node 2,
and so on until Node 7. It was observed that the best results are obtained in case the linking in
done in a single line where the symmetry axes of the two elements cross each other. Following
the general meshing recommendations given in section “3.1 Sensitivity analysis” of this thesis,
it is important to form such a mesh that allows each concrete FE triangle to have at least one
connection with a bar FE (which is able to take tensile forces and control the nodal
displacement). This is especially important for FE in crossing areas (see Fig.3.22b). All this
implies that the meshes of the crossing regions need to have equal number of FE over their
height (as presented in refer to Fig.3.22b) in order to equally distribute the points that are used
to impose equal deformations of the specimens.
CL CL
Crushing
SF ηε

Crushing
Observed

46 kN 46 kN

SF SF

46 kN 46 kN
ηε ηε

Figure 3.23: Results of EPSF analyses performed on a 65/2A specimen tested by Baumann and
Rüsch – governing flexural failure of the member
EPSF obtained after applying the described modelling procedure is presented in
Fig.3.23. The element in question is 65/2A, which was a part of the experimental campaign
conducted by Baumann and Rüsch [9]. General outline as well as the loading and supporting

69
Chapter 3: Advanced modelling of structural concrete members with EPSF

conditions of the specimen can be seen in the centre of the Fig.3.23. The stress field of the
longitudinal beam at ULS, as well as the corresponding concrete compressive strength
reduction factor (ηε) along with the crack pattern observed during the experimental campaign
are presented above the 3D sketch of the specimen. The behaviour of the transverse elements
(both stress field and distribution if the ηε factor) are shown on the sides.
It can be seen that the behaviour of the specimen observed during the test corresponds
well to the stress field resulting from the EPSF analysis. Distribution of the concrete
compressive strength reduction factor (ηε) is well correlated with the observed crack pattern at
ULS. The specimen failed due to crushing of the concrete in the upper zone close to the
symmetry line (as indicated in Fig.3.23), which was also the case in experimental campaign.

(a) 2.0 (b)

1.5
QTEST /QEPSF

1.0

0.5

0
20 30 40 50 60 1 2 3 4 5
fc [MPa] σp,inf = Fp,inf /A [MPa]

(c) 2.0 (d)

1.5
QTEST /QEPSF

1.0

0.5

0
0 1 2 3 4 0 0.25 0.50 0.75 1
ρ [%] ρw [%]

Baumann T. and Rüsch H. Leonhardt F. Koch R. and Rostasy F.

Leonhardt F. and Walther R. Büeler Ch. and Thoma K.
Leonhardt F. Walther R. and Dilger

Figure 3.24: Ratio of measured and estimated strength of indirectly supported structural concrete
members using EPSF method in function of: (a) concrete compressive strength;
(b) initial prestress; (c) longitudinal reinforcement ratio; (d) transverse reinforcement
ratio

70
 3.3 Analysis of indirectly supported structural concrete members

EPSF accurately predicted the ultimate strength of all analysed crossed structural
specimens, and each time correctly indicated the failure mode (the models can be downloaded
from http://i-concrete.epfl.ch/epsf/epsf.html). In some cases, the failures occurred in one of the
crossed-beams (as was the case presented in Fig.3.23 for example), while the others failed close
to the connecting regions (as was the case with specimen IWT2 tested by Leonhardt and
Walther [69]).
Satisfactory results have been obtained after applying the described modelling technique
on all 19 specimens found in literature (see Tab.3.3). As it can be seen, each of the five analysed
series gave satisfactory results with respect to the ultimate strength prediction as well as the
governing failure mode (which is very important when it comes to assessing the ultimate
strength of existing structural members).
On average EPSF method yielded 4% lower values of specimen’s ultimate strength
when compared to what was measured during the tests (average Qtest/QEPSF=1.04). In addition
to this the results showed very little scatter (CoV=0.05) and no trends presented as a function
of 4 basic mechanical parameters (concrete strength, prestress level, longitudinal and transverse
reinforcement ratio, as can be seen in Fig.3.24).

Table 3.3: The results of the EPSF for crossed structural concrete members
Qtest QEPSF Qtest Qtest QEPSF Qtest
N° Spec. Aver. CoV N° Spec. Aver. CoV
[kN] [kN] /QEPSF [kN] [kN] /QEPSF
Leonhardt et Walter [69] Baumann, Rüsch [9]
1 IWT1 1152 1130 1.02 1 64/1 102 102 1.00
1.04 -
2 IWT2 1177 1114 1.06 2 65/1A 140 130 1.08
Leonhardt et all. [70] 3 65/1B 104 104 1.00
1 ETI1 273 276 0.99 4 65/2A 93 92 1.01 1.06 0.06
2 ETI2 257 250 1.03 5 65/2B 103 96 1.07
3 ETI3 240 222 1.08 1.01 0.04 6 65/3A 92 80 1.15
4 ETI4 245 250 0.98 7 65/3B 112 98 1.14
5 ETI5 240 246 0.98 Leonhardt at all. [66]
Büeler et Thoma [15] 1 ILT1 1810 1690 1.07
1 LT1 635 630 1.01 2 ILT2 1565 1540 1.02 1.05 0.03
1.01 -
2 LT2 863 860 1.00 3 IILT1 1667 1552 1.07

Based on the presented results the EPSF method has proved to be efficient and accurate
tool for simulating behaviour of crossed-beam members capable of predicting the correct
ultimate load and failure kinematics both on local and global level.

71
Chapter 3: Advanced modelling of structural concrete members with EPSF

3.4 Tailoring partial safety factors suitable for EPSF method

The level of risk associated with the design and assessment of reinforced and prestressed
concrete elements varies depending on the amount of available information regarding their
usage and specifications. Analysis of new elements involves more uncertainty concerning
material properties, load level, load history, geometry and reinforcement layout compared to
existing structures. In addition to this, models used for design often employ more conservative
assumptions compared to approaches that are used for assessment of existing elements. In order
to ensure the same probability of failure for each member over its expected lifetime (regardless
of the fact that it is being designed or assessed) certain type of safety format is applied for each
simulation. Due to increased accuracy of models used for assessment, it seems natural that
corresponding safety factors should be decreased compared to the ones which are applied for
structural design.
This chapter investigates different strategies that could be used to ensure sufficient
structural safety of EPSF models (global and partial safety format), and recommends which one
is best suited for stress field analysis. In addition to this, it the discusses pertinence of partial
safety format being applied with global nonlinear structural analysis. Finally, it gives tailored
values of partial safety factors that can be used when estimating ultimate strength of existing
structural concrete members with EPSF, based on work of Tanner et al. [112, 113, 114].
When dealing with global nonlinear finite element analysis in practice, there is always
a question of what is the correct implementation of structural safety. There are two main
approaches:
1. Partial Safety Format (PSF)
2. Global Safety Format (GSF)
PSF implies reducing the material characteristic resistance values (fck for concrete and
fsk for steel that correspond to 5% fractile on a normal resistance distribution) in order to obtain
the design values (fcd and fsd), which are then used to determine the design resistance of analysed
members. In other words:
f ck
f cd  (3.7)
γc

f sk
f sd  (3.8)
γs

Rd  f ( f cd ; f sd ; anom ) (3.9)

where γc represents the partial safety factor for concrete equal to 1.50 [29,34];
γs represents the partial safety factor for steel equal to 1.15 [29,34];
Rd represents the design resistance of an element;
anom represents the nominal geometry values.

72
 3.4 Tailoring partial safety factors

On the other hand, GSF uses mean values of material properties and nominal values of
geometry properties in order to determine the mean elements resistance, after which a single
(global) safety factor is applied in order to determine design resistance in the following manner:
Rm  f ( f cm ; f sm ; anom ) (3.10)

Rm
Rd  (3.11)
γd

where Rm represents the mean resistance of an element;

Rd represents the design resistance of an element;
γd represents the global safety factor.
Since the PSF uses different factors to reduce the strength of steel and concrete, this
means that in case of global nonlinear FE analyses, internal stress redistribution that occurs
between a structural element is not necessarily the most probable one. Instead it favours a brittle
failure mode, which is the main reason why some modern codes of practice (for example EC2
[29]) recommend that only GSF should be applied when conducting a global nonlinear FE
analysis. This recommendation is debatable (which is clearly indicated in the EC2[29] itself),
since it is not in agreement with the general design format of the code (partial safety format).
Nevertheless, GSF gives more accurate estimates of the absolute magnitude of deformations
which is important when taking into account the second order effects in the nonlinear
calculations. In addition to this EC2 [29] does suggest that PSF is suitable for hand-calculations
of STM, which presents a global nonlinear calculation method, and as such takes into account
the internal stress redistributions between steel and concrete. This raises a question of why PSF
should not be used with nonlinear FE models, since the fact that a calculation is done by hand
or by means of nonlinear FE does not change the physical nature of the approach. Finally, EC2
[29] concludes that a non-linear analysis with PSF is acceptable but conservative since the
potential benefits of using a refined calculation method are lost.
According to subsection 7.11.3 of MC2010 [34] PSF can be applied for global non-
linear analyses giving a safe estimate of the design resistance. This approach is consistent since
it employs the same safety strategy for hand and finite element based calculation methods (in
other words RPSF and EPSF). Thus, the classical partial safety factors (γc=1.5 and γs=1.15)
should be applied when designing new structures with EPSF method. When it comes to
assessment of the existing structures it is pertinent to keep the same safety format but one can
argue that such approach gives slightly conservative results. This is why MC2010 [34] also
allows application of the probabilistic design methods when conducting the nonlinear FE
simulations. Aside from performing a full Monte-Carlo simulation to determine the design
resistance, MC2010 [34] introduces an approach that estimates the coefficient of variation of
the ultimate resistance and allows one to calculate the global safety factor for each particular
case (depending on the sensitivity of the element’s resistance towards the variation in material
strength). In addition to this, fixed values of the global safety factor are also provided as a
further simplification.

73
Chapter 3: Advanced modelling of structural concrete members with EPSF

In order to take advantage of enhanced accuracy of nonlinear FEM analysis while

keeping the PSF when assessing the strength of structural concrete members, a method for
tailoring partial safety factors was developed by Tanner et al. [112, 113, 114] . So far this
methodology showed that it is in agreement with code recommendations, but is sensitive to
scatter of concrete compressive strength (systematic assessment of actual material properties
on site is required for this approach). Therefore, the reduction of the partial safety factors should
be taken with caution. Based on this work, custom partial safety factors which are meant to be
used with EPSF when assessing the strength of existing structural concrete members were
calculated.
When talking about structural uncertainties, there are two major groups that influence
the final value of partial safety format for steel and concrete:
1. Uncertainties related to material properties (γC for concrete and γS for steel);
2. Uncertainties related to model and geometrical dimensions (γRd,c and γRd,s).
The two factors are multiplied in order to obtain the final value of partial safety
coefficient for concrete and steel in the following manner:
γc  γC  γ Rd ,c for the concrete partial safety factor (3.12)

γs  γS  γ Rd ,s for the steel partial safety factor (3.13)

In the previous equations (Eq.3.12 and 3.13), partial safety factor taking into account
material uncertainty can be estimated as following [58]:
f
C  ck
  ( COV ) fc
 1.00 for the concrete (3.14)
f cm  e

f sk
S   1.00 for the steel (3.15)
f sm  e   (COV ) fs

where fck and fcm respectively represent the characteristic and the mean value of concrete
compressive strength (assuming a log-normal distribution of the variable);
(COV)fc represents the coefficient of variation of concrete compressive strength;
α represents the sensitivity factor;
β is the reliability index.
The notation for the steel safety factors is the same except the index letter “S” indicating
steel instead of “C” indicating concrete.
The reliability index depends on the acceptable probability of failure (pf) and the
corresponding reference period (n) in years. For a standard design case acceptable probability
of failure is equal to 10-6 [34] and the design reference period is equal to 50 years [34]. The
reliability index can be calculated in the following manner [95]:

β1  Φ 1( 1  p f ) (3.16)

74
 3.4 Tailoring partial safety factors

β  Φ1 (Φ( β1 )n ) (3.17)

where Φ is the normal distribution function;

β1 represents the reliability index for one year reference period;
n represents the design reference period (in years).
After introducing the recommended values to Eq.3.16 the reliability index for one year
is equal to 4.75. Based on Eq.3.17 the same index for 50-year reference period is equal to 3.89.
MC2010 [34] recommends a value of 3.80 for the reliability index (50-year reference period),
which was assumed for the purpose of this research.
The sensitivity factor is separating dominating uncertainties from non-dominating
uncertainties. The final value of a partial safety factor (refer to Eq.3.12 and Eq.3.13) is obtained
assuming that one of the two sub-factors is dominating (due to the material uncertainties or
modelling uncertainties) and the other one is not. Out of the two combinations the one
governing the maximum value of the total partial safety factor is selected. According to
MC2010 [34] in case the variable uncertainty is of key importance (dominant variable) it should
be assumed a sensitivity factor equal to α=0.80; otherwise α=0.24.
As it was already indicated, the most significant parameter (the one having the strongest
impact on the final value of the partial safety factor) is the coefficient of variation of the
materials. For steel, this scatter is not very high and a value of 0.055 is assumed as fixed
(according of MC2010 [34]). The CoV was back-calculated assuming lognormal distributions
of the variable with fsk=500 MPa and fsm=550 MPa. Variation of the concrete compressive
strength on the other hand is something which is much less uniform. In addition to this, quality
control systems have changed a lot over the past century, so it is very difficult to give general
recommendations. This is why it is very important to conduct a thorough statistical analysis of
concrete compressive strength for each particular structure which is being assessed. However,
in order to have an idea of the scatter of concrete compressive strength, a literature review has
been conducted based on information for concrete cast in the 1960 [44, 28, 97] and presented
in Fig.3.25, which compares the results from three different authors (Himsworth [44], Erntroy
[28] and Rüsch [97]) to modern code recommendations (MC2010 [34]).
0.5 10

0.4 8

0.3 6
std [MPa]
CoV

0.2 4
Erntroy (1960)
0.1 2 MC 2010
Rüsch (1964)
Himsworth (1954)
0 0
15 25 35 45 55 15 25 35 45 55
fc [MPa] fc [MPa]

Figure 3.25: Coefficient of variation and standard deviation of concrete compressive strength
according to various authors

75
Chapter 3: Advanced modelling of structural concrete members with EPSF

Geometry and model uncertainties governing the value of partial safety coefficient can
be expressed in the following manner [58]:
QTEST / QMODEL k
 Rd ,c   1.00 (3.18)
QTEST / QMODEL m  e    (COV
2
MODEL )  ( COV geometry , c ) 2

QTEST / QMODEL k
 Rd , s   1.00 (3.19)
QTEST / QMODEL m  e    (COV
2
MODEL )  ( COV geometry , s ) 2

where (QTEST/QMODEL)m represents the average ratio between the actual (tested) and estimated
resistance for a specific calculation method;
COVMODEL represents the corresponding coefficient of variation;
(QTEST/QMODEL)k represents the characteristic value of the model uncertainty (corresponding to
5% fractile on a log-normal distribution);
COVgeometry,c represents the coefficient of variation of the geometry properties which are relevant
for the resistance strength of concrete (external cross section dimensions and effective depth);
COVgeometry,s represents the coefficient of variation of the geometry properties which are relevant
for the resistance strength of steel (effective depth and reinforcement diameter).
The first three parameters (QTEST/QMODEL)m, COVMODEL and (QTEST/QMODEL)k can be
estimated using the elements from the online database (http://i-concrete.epfl.ch/epsf/epsf.html).
In addition to this the analysed members can be divided into two groups:
1. Members failing in flexure
2. Members failing in shear
For each of these groups the average QTEST/QMODEL ratio and COVMODEL can be estimated
using the EPSF method. Reinforced concrete elements from the same database have been
assessed using the recommendations from MC2010 [34] in order to compare the final values of
partial safety coefficient. It is expected that the PSF for EPSF will be lower (due to the higher
accuracy of the method) compared to code recommendations, and that the PSF for code
recommendations should be equal to γc=1.5 and γs=1.15. Tab.3.4 shows the results of EPSF
analysis and code recommendations of the database.

Table 3.4: Ultimate strength of structural concrete elements from the online database estimated using
EPSF method and MC2010 code provisions
Model EPSF MC2010
Flexural Shear Flexural Shear– Lo1 Shear– Lo2
QTEST/QMODEL 1.075 1.048 1.059 1.439 1.223
COVMODEL 0.060 0.083 0.080 0.212 0.168

It is important to emphasise that the following test campaigns were selected from the
database for the purpose of this research, in order to account only for the specimens that had
clear flexural and shear failures (Vecchio and Shim [122], Yoon et all. [126], Sagaseta and

76
 3.4 Tailoring partial safety factors

Vollum [104]; Sørensen [110], Leonhardt and Walter [68], Kaufmann and Marti [54];
Kaufman and Ramirez [51], Kuchma et al. [60] without specimen G10, Rupf and Muttoni [103]
without specimen SR31B, Moore [79] and Fernández Ruiz and Muttoni [31]). Specimens
which experienced some local failures during the experiments were therefore discarded.
Selected EPSF results are presented in Fig.3.26.

(a) (b)
2.0

1.5
QTEST/QEPSF

Moving average
1.0

0.5

0
0 20 40 60 80 100 120 140 0 3 6 9 12 15
fc [MPa] σp,inf = Fp,inf /A [MPa]

(c) (d)
2.0
Aver: 1.05
CoV: 0.08
1.5
QTEST/QEPSF

1.0

0.5

0
0 1 2 3 4 5 0 0.5 1 1.5 2
ρ [%] ρw [%]

Figure 3.26: Selected member from the online database used to estimate the tailored partial safety
coefficients for EPSF analysis - predicted and estimated strength in function of:
(a) concrete compressive strength; (b) initial prestress; (c) longitudinal reinforcement
ratio; (d) transverse reinforcement ratio
When it comes to variation of geometry properties applied CoV are given in Tab.3.5
(taken from Tanner et al. [112, 113, 114]). All variables assume normal distribution.

Table 3.5: Main parameters estimating geometrical uncertainties in structural concrete

Variable Rk/Rm* CoV
External cross-section 1.000 0.030
Effective depth 1.000 0.040
Reinforcement diameter 1.000 0.020
*Bias value – ratio between characteristic and mean value of a variable

77
Chapter 3: Advanced modelling of structural concrete members with EPSF

Final values of PSF for concrete are presented in Fig.3.27 to Fig.3.28 (where MC2010
[34] equations were used to assess the ultimate strength of the elements for the database) and
Fig.3.29 to Fig.3.30 (where the same was done using the EPSF method).
3.0
Rusch (1964) Erntroy (1960) MC2010 Himsworth (1954)
2.5

2.0
Dom. material Dom. model
γc 1.5 Dom. material Dom. model Dom. material Dom. model Dom. material Dom. model

1.0 γc,max=1.28 γc,max=1.36 γc,max=1.40 γc,max=1.48

0.5
MC2010 Flex. fail.
0
20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70
fc [MPa] fc [MPa] fc [MPa] fc [MPa]

Figure 3.27: Tailored partial safety factors for concrete (flexural failure estimated with MC2010)
3.0
Rusch (1964) Erntroy (1960) MC2010 Himsworth (1954)
2.5

2.0
Dom. model Dom. material Dom. model Dom. material Dom. material Dom. model Dom. material Dom. model
γc 1.5

1.0 γc,max=1.39 γc,max=1.39 γc,max=1.40 γc,max=1.48

0.5
MC2010 Shear fail. LoA 1

3.0
Rusch (1964) Erntroy (1960) MC2010 Himsworth (1954)
2.5

2.0
Dom. model Dom. material Dom. material Dom. model Dom. material Dom. model Dom. material Dom. model
γc 1.5

1.0 γc,max=1.31 γc,max=1.36 γc,max=1.40 γc,max=1.48

0.5
MC2010 Shear fail. LoA 2
0
20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70
fc [MPa] fc [MPa] fc [MPa] fc [MPa]

Figure 3.28: Tailored partial safety factors for concrete (shear failure estimated with MC2010)
3.0
Rusch (1964) Erntroy (1960) MC2010 Himsworth (1954)
2.5

2.0
Dom. material Dom. model
γc 1.5 Dom. material Dom. model Dom. material Dom. model Dom. material Dom. model

1.0 γc,max=1.28 γc,max=1.36 γc,max=1.40 γc,max=1.48

0.5
EPSF Flex. fail.
0
20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70
fc [MPa] fc [MPa] fc [MPa] fc [MPa]

Figure 3.29: Tailored partial safety factors for concrete (flexural failure estimated with EPSF)

78
 3.4 Tailoring partial safety factors

3.0
Rusch (1964) Erntroy (1960) MC2010 Himsworth (1954)
2.5

2.0
Dom. material Dom. model
γc 1.5 Dom. material Dom. model Dom. material Dom. model Dom. material Dom. model

1.0 γc,max=1.28 γc,max=1.36 γc,max=1.40 γc,max=1.48

0.5
EPSF Shear fail.
0
20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70 20 30 40 50 60 70
fc [MPa] fc [MPa] fc [MPa] fc [MPa]

Figure 3.30: Tailored partial safety factors for concrete (shear failure estimated with EPSF)
Shear resistance of the elements has been estimated using both first and second Level
of Approximation (LoA) from MC2010 [34]. In each of the graphs, the red line gives the value
of the partial safety factor for concrete assuming that the material uncertainties are dominant.
On the other hand, the black line gives the same values assuming that model and geometry
uncertainties are dominant. The PSF are given in a function of concrete compressive strength.
Each of the four graphs in Fig.3.27 to Fig.3.30 assumes different scatter of concrete
compressive strength (indicated each time in the upper right corner of the graph).
In case the structural analysis is based on equations presented in MC2010 [34] highest
partial safety factor for the concrete is equal to 1.48 (which is close to 1.50 indicated by the
codes) for elements failing in flexure and shear. This value is governing each time for the low
strength concrete (around 20 MPa) if the dispersion of the concrete compressive strength is
assumed according to Himsworth [44]. In case of elements failing in flexure, it can be observed
that the PSF for concrete can be reduced as the concrete compressive strength increases, due to
the low CoV of concrete compressive strength and high accuracy of the applied model.
Elements failing in shear, on the other hand, do not experience the same significant reduction
of partial safety factors. This is due to the fact that the applied model is much less accurate
compared to the one for flexure (refer to Tab.3.4). In addition to this, it can be observed that
the first level of approximation for shear according to MC2010 [34] governs higher partial
safety factor for concrete (γc=1.39) compared to the second level of approximation γc=1.31 for
higher concrete compressive strength, which is to be expected. In order to keep a uniform value
of the PSF for concrete, the maximal obtained should be selected (referring to 1.48 in this case),
which indicates that the tailoring procedure is in agreement with the code regulations.
For the EPSF method partial safety factors are similar for both failure modes (refer to
Fig.3.29 and Fig.3.30), due to high model accuracy for both types of failure (refer to Tab.3.4).
Similarly to elements failing in flexure estimated with MC2010 [34], the material uncertainties
are dominant in case of lower strength concrete (around 20 MPa), and govern the maximal
value of PSF (γc=1.48). It can be observed that this value is highly dependent on CoV of concrete
compressive strength (in case concrete compressive strength scatter is assumed according to
Rüsch [97] for example the maximal PSF for concrete is γc=1.28). This is why it is crucial to
accurately estimate the level of concrete compressive strength dispersion. PSF tailored for
EPSF method can be reduced as the concrete compressive strength increases (refer to Fig.3.29
and Fig.3.30). According to the proposed procedure, it can be as low as 1.17 in case of 70 MPa

79
Chapter 3: Advanced modelling of structural concrete members with EPSF

concrete, at which point it becomes almost equal to PSF for steel, indicating that the partial
safety factor practically becomes a global safety factor.
The final value of partial safety factor for concrete that should be used with EPSF
method is not explicitly given in this chapter. Similarly, to LoA approach, one should first apply
γc = 1.5. In case this simulation does not give satisfying accuracy, the value of PSF for concrete
could potentially be reduced according to the proposed procedure. Thorough statistical analysis
needs to be conducted in order to verify the scatter of concrete compressive strength for a given
element. In addition to this, the value of sensitivity coefficients needs to be revised in order to
make sure that the 0.24 and 0.8 are pertinent values for that particular case.
The partial safety factor for steel has also been calculated, but since the dispersion of
the yield strength is much smaller, the results are not presented in the form of a graph. Tab. 3.6
summarizes the obtained results.

Table 3.6: Tailored partial safety factors for steel

Failure mode Flexural Failure Shear Failure
MC2010 1.171 -
EPSF 1.146 1.162

It can be observed that the PSF for steel are much more uniform and are not that different
from 1.15 indicated by the codes. The reason why the PSF for steel was not calculated in case
of elements failing in shear for the MC2010 [34] is due to the fact that in reality, failure of the
tested specimens was always governed by the simultaneous failure of concrete and
reinforcement. Even though the codes have predicted a few simulations failures of elements in
shear, the available number of such elements was not sufficient for a pertinent statistical
analysis.

80
Chapter 4: Effectiveness factor for concrete
compressive strength accounting
for the presence of cracks and
interaction with the reinforcement
Technical paper titled “Response of RC panel accounting for crack development and its
interaction with rebars”, written by Dr. Eckart Hars, Filip Niketić and Dr. Miguel Fernández
Ruiz was based on this chapter. The article was accepted for publishing by the Magazine of
Concrete Research on the 7th of July 2017 under an article number MACR-D-17-00077R1.
Contributions of Filip Niketić to creation of this publication involved:
1. Developing expression for estimating the plastic strength of rebars subjected to
doweling;
2. Deriving the expression for estimating plastic strength of damaged and
undamaged concrete struts;
3. Quantifying damaged and undamaged concrete struts in a panel;
4. Developing and implementing a solving procedure for estimating the shear
strength of the RC panels using the proposed model;
5. Assembling a database with 77 structural panels and validating the proposed
procedure;
6. Producing figures for the article;
7. Writing the manuscript of the article.

4.1 Introduction and state of the art

Limit analysis (through application of stress fields and strut-and-tie models) provides a
consistent framework for design and assessment of structural concrete members. It can be
applied to a wide range of cases, involving elements with various geometry, and material
properties subjected to in-plane as well as out-of-plane loading [92]. However, the application
of the theory of plasticity on structural concrete elements was challenged, due to the brittleness
of concrete. Unlike the stress-strain diagram of steel, which shows ductile behaviour in tension
and compression after reaching its yielding point (see fy in Fig.4.1b), the strength of concrete
decreases once its ultimate resistance (fc) has been achieved (see Fig.4.1a). Such behaviour

81
Chapter 4: Effectiveness factor for concrete compressive strength

compromises the development of the plastic zones in concrete and therefore the application of
limit analyses, especially for cases where the ultimate resistance of structural elements is
governed by crushing of the compressed zones. In order to overcome this, early elements were
designed in a way that avoid reaching the concrete compressive strength at ULS. In other words,
ultimate load-bearing capacity of members was governed by yielding of the reinforcement.
(a) σs Actual (b) σc

fct
fy
Adopted
εs

- fy -fce
Adopted -fc
Actual

Figure 4.1: Actual and adopted stress-strain diagram for: (a) steel; (b) concrete according to full
plastic approach
When Drucker [27] used the theory of plasticity to obtain the exact solution of a simple
beam subjected to a point load in 1961, the direct application of the limit analysis on structural
concrete members was theoretically proven. Experimental campaign of Leonhardt and Walther,
conducted in 1962 [67], showed that reinforced concrete beams which failed in flexure did
experience plasticisation of the compressed zones. All this lead to further development of stress
fields and strut-and-tie models as general tools for design and assessment of structural concrete
elements.
In 1979 Exner [30] proved that the application of the theory of plasticity in combination
with unreduced concrete compressive strength (fc in Fig.4.1a) can lead to unconservative
solutions. He argued that the concrete compressive strength needs to be reduced as a
consequence of material softening. In other words, an effective concrete compressive strength
fce should be applied instead of fc (see Fig.4.1a). Using the experimental results of RC beams
with no transversal reinforcement, he defined an empirical expression for concrete compressive
strength effectiveness factor:
3 .2
  (4.1)
fc

According to Exner [30], the effective concrete compressive strength can be estimated
in the following manner:

f ce    f c (4.2)

Current codes of practice (such as MC2010 [34], EC2 [29] or ACI 318 [1]) include
reduction of concrete compressive strength measured on a cylinder when estimating structural
capacity of load-bearing members. According to MC2010 [34], the effective concrete
compressive strength is obtained in the following manner:

fce  η fc  kε  αcc  fc (4.3)

where ηfc represents the concrete compressive strength brittleness factor;

82
 4.1 Introduction and state of the art

kε represents the concrete compressive strength softening due to imposed transversal strains;
αcc takes into account the effects of loading rate and loading level on concrete compressive
strength.
According to MC2010 [34], the effects of the post-tensioning ducts on effective
concrete compressive strength are taking into account indirectly, by reducing the thickness of
the analysed web in the following manner:

bw, nom  bw  k D   
D (4.4)

where bw,nom represents the nominal web thickness;

bw represents the initial web thickness;
ØD represents the diameter of the post-tensioning ducts.
This chapter analyses all concrete compressive strength reduction factors from Eq. 4.3
except the last one (αcc). The most significant models in this field are presented and discussed.
A critique of the MC2010 [34] multiplicative approach (refer to Eq. 4.3) is introduced, and
finally, a mechanically based model for assessing the concrete compressive strength
effectiveness factor that takes into account different failure mechanisms of structural elements
(concrete crushing, spalling and sliding) is presented and discussed. Its consistency is validated
by means of extensive comparison to available test data found in literature.

4.1.1 Concrete compressive strength brittleness factor - ηfc

Response of a standard concrete cylinder subjected to uniaxial compression, consist of
an ascending branch that reaches up to its ultimate strength, followed by a descending softening
branch. Constitutive law of concrete has been investigated by many authors in numerous
experimental campaigns, resulting in simplified empirical expressions such as the one proposed
by Fernández et al. [33]:
1
c   Ec   c (4.5)
1   c  0 


  fc
0  (4.6)
Ec    1
11  

  0.5  f c 20  f c2 1'500 (4.7)

where σc represents the concrete compression stress;

εc represents concrete compressive strain;
Ec represents Young’s modulus of elasticity for concrete;
fc represents the concrete compressive strength.
Dashed lines in Fig.4.2a represent the actual behaviour of concrete subjected to uniaxial
compression (described in Eq.4.5). As it can be seen, brittleness of concrete increases in

83
Chapter 4: Effectiveness factor for concrete compressive strength

function of the uniaxial compressive strength. Descending branch of a 100 MPa concrete for
example is much steeper (hence, more brittle) compared to the one corresponding to a 40 MPa
concrete, while the stress-strain diagram of 20 MPa concrete leaves an impression of an almost
ductile material behaviour.
(a) (b)
120 1.2
Actual behaviour
Plastic behaviour fc5
100 1.0

fc4
80 0.8
fcp5
σ [MPa]

fc3 fcp4
60 ηfc 0.6
fcp3
fc2
40 fcp2 0.4

fc1 fcp1
20 0.2
fco=30MPa - acc to MC2010
fco=20MPa - acc to Muttoni 1990
0 0
0 1 2 3 4 0 20 40 60 80 100
ε [×10-3] fc [MPa]

Figure 4.2: Concrete brittleness factor: (a) actual and plastic constitutive law of concrete; (b) the
brittleness factor according to Muttoni [82]
If the limit analysis is to be applied on structural concrete, actual stress distribution at
ULS should correspond to the one assumed by the theory of plasticity (constant stress blocks).
Fig.4.3 illustrates this problem, using an element subjected to pure bending. Assuming that the
maximal compressive strains are approximately equal to 3.5‰, and that the sections remain
perfectly plane at ULS (refer to Fig.4.3a), allows one to estimate the concrete stress
distributions at the critical cross-section. Applying the concrete constitutive law presented in
Eq.4.5 results in stress distribution presented in Fig.4.3b, which does not correspond to constant
stress block assumed by the theory of plasticity. In order to overcome this, a concrete
compressive strength brittleness factor (ηfc), which smears the actual stress distribution down
to a block with constant stress intensity needs to be applied (refer to fc∙ηfc in Fig.4.3c).
(a) (b) (c) (d)
Fc ~3.5‰
x xp
~2.0‰
fc fc·η
εc,ult fc·ηfc
Mult

Fs εs,ult Fs,ult Fs,ult Fs,ult

Figure 4.3: Analysis of a critical section subjected to pure bending at ULS: (a) strain distribution;
(b) stress distribution assuming actual concrete constitutive law; (c) stress distribution
assuming plastic concrete constitutive law with brittleness factor; (d) stress distribution
assuming plastic concrete constitutive law without the brittleness factor
Applying the strength reduction factor while smearing the stresses over the entire
compression zone (marked as x in Fig.4.3c) is only one approach for dealing with brittleness of
concrete. An alternative method is presented in Fig.4.3d, where the intensity of the stresses

84
 4.1 Introduction and state of the art

block is equal to fc∙η while the height of the compression zone is reduced down to xp. According
to MC2010 [34]:

x p  0.8  x for f c  50MPa (4.8)

x p  0.8   f c  50 400 for 50  f c  90MPa (4.9)

η  1.0 for f c  50MPa (4.10)

η  1.0   f c  50  200 for 50  f c  90MPa (4.11)

Even though the second approach can be applied for bending, in other cases (such as
shear for example), the reduction of the area upon which compressive stresses are acting is not
so evident. Hence, the application of a brittleness factor provides more consistent solutions for
various stress states. It now becomes evident why the resistance of structural concrete elements
can be overestimated in case the intensity of stress block remains equal to fc while the height of
the compressed zone (x) stays the same (as argued by Exner [30]).
In order to compensate for the brittle behaviour of concrete Muttoni [82] proposed a
following expression:

f 
 fc  3  c 0   1.0 (4.12)
 fc 
where fco represents the concrete compressive strength up to which the actual constitutive law
for concrete exhibits a ductile behaviour.
(a)
F F /A·fc 1 Q / ∑ Ai·fc
ηfc=1 ηfc

Q 0
F 0 (b) δult
δult

EI=∞
δ
A2
N2
N2 /A2·fc N1 /A1·fc
1 1
ηfc A1 A1 ηfc
N1 N1

0 0
δult δult

Figure 4.4: Redistribution mechanisms: (c) response of a standard concrete cylinder; (b) response of
a system consisting of three cylinders and an infinitely rigid beam (adopted from Muttoni
et al. [89])
Fig.4.2b shows the concrete compressive strength brittleness factor according to
Muttoni [82]. The initial limit which separates ductile from brittle behaviour of concrete
cylinders was set equal to fco=20 MPa (refer to the dashed line in Fig.4.2b), but was later

85
Chapter 4: Effectiveness factor for concrete compressive strength

modified to 30 MPa (refer to the full line in Fig.4.2b). Applying the brittleness factor on actual
stress-strain diagrams for concrete, results in the elastic-perfectly plastic constitutive laws for
cylinders, presented in Fig.4.2a with full lines. The reduction of concrete compressive strength
becomes more significant as the ultimate resistance increases (refer to Fig.4.2a and b), and the
actual behaviour becomes more brittle.
Brittleness factors should also be applied when estimating ultimate resistance of
structural concrete members subjected to constant compressive strains; not just when there is a
distribution of strains over the compressed zone (such is the case with bending, presented in
Fig.4.3 for example). The reason for this is schematically presented in Fig.4.4, which compares
the behaviour of a standard cylinder (see Fig.4.4a) to a system of three colons connected with
an infinitely stiff beam (see Fig.4.4b). Obtaining the failure load of a cylinder does not require
the application of a brittleness reduction factor, since the maximal material stress is reached at
the same point across the entire cross-section, and the element does not need to facilitate any
additional redistribution (in other words ηfc=1 as presented in Fig.4.4a). On the other hand,
applying the same displacement (δ) over three short columns made out of identical concrete,
results in different strain states in each one of them. The shortest (interior) column is already
plasticized by the time the system fails (see the bottom left graph in Fig.4.4b). On the other
hand, the longer (exterior) columns have just reached the peak of the stress-strain diagram (refer
to bottom right graph in Fig.4.4b). Consequently, summing the individual resistances of the
three members while using the actual concrete compressive strength (fc), results in
overestimation of the system’s strength. In order to compensate for this, a concrete compressive
strength brittleness factor (ηfc) should be applied to all three columns (as indicated in Fig.4.4b).
The same effect can be observed within a single structural concrete element which has
reinforcement in compression. Strain-wise, the yielding point of the two materials (concrete
and steel) does not coincide, thus leading to softening of concrete prior to reaching the yielding
of the reinforcement (similarly to what is described in Fig.4.4b).
It can be concluded that the stress redistribution in structural concrete elements should
be anticipated in case of isostatic as well as hyperstatic systems. Therefore, the concrete
compressive strength brittleness factors (ηfc) should always be applied in order to facilitate safe
usage of the limit analysis.

4.1.2 Concrete softening caused by imposed transversal strains - ηε

Being a quasi-brittle material which cracks at relatively low loads, concrete relies on
reinforcement in order to obtain the necessary strength and ductility when subjected to various
actions. As the reinforcement starts to deform, it induces strains by means of concrete-to-steel
bond, causing the formation of smeared cracks (as presented in Fig.4.5a). This reduces the
compressive strength of concrete and limits the ultimate resistance of analysed elements. In
addition to this, the physical presence of the rebars in the concrete matrix introduces
discontinuities which deviate the compressive stress path. Depending on the geometry
properties of analysed members, this might lead to structural failures.

86
 4.1 Introduction and state of the art

σc

(a)
τ
σc τ
(b)
σs

τ
RC panel τ

Figure 4.5: Physical behaviour of concrete: (a) in a reinforced concrete beam; (b) in a panel
Many experimental programmes have been conducted in the past in order to better
understand this complex problem ([5], [10], [47], [56], [57], [75], [94], [106], [118], [119],
[121], [124], [127]), and their results are today incorporated in modern codes of practice
(MC2010 [34], EC2 [29], ACI 318 [1]), as well as nonlinear numerical methods [35]. Most of
the studies have been performed on reinforced concrete panels (such as the one presented in
Fig.4.5b), since the specimens can be tested until failure while being subjected to uniform stress
states (pure shear, shear combined with tension or compression, pure compression, compression
combined with tension etc.). However, some of the experimental campaigns focusing on the
investigation of effective concrete compressive strength were performed on reinforced concrete
beams failing in shear (Muttoni [82] and Zwicky [129] for example). Such elements were
subjected to a non-uniform stress state, meaning that a part of shear was taken by the
compressed flange, as well as the direct strut action and doweling of the longitudinal
reinforcement. Consequently, estimating the effective concrete compressive strength becomes
much more challenging, which is why the application of reinforced concrete panels was
preferred. However, it is important to emphasise that experimental investigation of beams gives
crucial information on the realistic behaviour of structural elements, and as such, it is very
important from a practical point of view.
Fig. 4.6 shows the influence of transversal strains on concrete compressive strength.
The black line represents the stress-(transversal) strain diagram of a standard cylinder test,
showing a clear softening of the material once the compressive strength (fc) has been reached.
Given a fact that similar curves can be obtained for a structural concrete panel (refer to the red
curve in Fig.4.6) Muttoni [82] proposed that the softening branch of the stress-strain diagram
can be used as a failure criterion for the element, allowing one to determine the effective
concrete compressive strength (fce). The average transversal strain (ε1) of a panel can be
obtained based on the crack opening in the direction of the reinforcement (refer to Fig.4.6).
However, the shape of the softening branch in a stress-(transversal) strain diagram is quite
challenging to predict due to the large scatter in the experimental results. In addition to this, the
proposed procedure does not account for all potential interactions between the concrete and the
reinforcement (as it will be later presented).

87
Chapter 4: Effectiveness factor for concrete compressive strength

σ3 fc

σ3
fce
ε1 ε1
σ3

ε1 σ3

σ3
ε1

Figure 4.6: Descending branch of a standard cylinder test used as a failure criterion for reinforced
concrete panels (adopted from Muttoni et al. [89])
There are three basic failure modes which have been observed in the experimental
campaigns involving reinforced concrete panels (observed by Vecchio and Collins [120]):
1. Crushing of the concrete struts (presented in Fig.4.7a);
2. Spalling of the concrete cover (presented in Fig.4.7b);
3. Sliding along an initial crack (presented in Fig.4.7c).
(a) (b) (c)

A
Sliding
Crushing
Out-of-plane C
crack
B
In-plane Out-of-plane
Spalling crack crack

A
C
Confining
In-plane
B cracks

Out-of-plane crack Out-of-plane crack

Figure 4.7: Failure mechanisms of reinforced concrete panels: (a) concrete crushing; (b) concrete
cover spalling; (c) concrete sliding
First two failure modes are a result of the physical interaction between the rebars and
the surrounding matrix. In case of concrete crushing, the resistance of the element is reached
when the out-of-plane cracks, imposed through bond by the elongation of the rebars (see
Fig.4.7a) soften the surrounding concrete matrix up to its failure. Looking at the horizontal cut,
made at the level of the rebars (refer to the bottom of Fig.4.7a) one can observe the formation
of the out-of-plane cracks according to the model of Tepfers [115] based on the experimental
observations of Goto [38] (point A marks the location of an out-of-plane crack). The described

88
 4.1 Introduction and state of the art

behaviour is mechanically consistent with the softening of concrete cylinders (presented in

Fig.4.6), meaning that the approach of Muttoni [82,89] gives satisfying results when structural
elements are failing due to crushing of the concrete.
Concrete cover spalling is governed by the in-plane cracks forming above and below
the rebars. Principal concrete compressive stresses are deviated towards the rebar due to the
change in stiffness (a similar phenomenon occurs when post-tensioning ducts are present in the
concrete webs as will be shown later on). Consequently, tensile stresses are imposed in the cross
section, which ultimately leads to cracking. As the in-plane cracks separate the concrete cover
from the core, the difference between the average strains of the cover and the core becomes
greater. Gradually, the concrete cover begins to bend and form the longitudinal cracks (as
presented in Fig.4.7b). Vertical cross-section of the panel, showing the disposition of cracks
leading to concrete cover spalling at point B is presented in Fig.4.7b.
Finally, sliding failures occur when maximal contact stress that can be transferred
through an initial crack (by means of the aggregate interlock combined with the dowel action
of the reinforcement) is surpassed, resulting in a relative slip between the two segments of the
panel (as presented in Fig.4.7c). The slip can also occur along the plane which contains the
points of the rebars weakened by the doweling (this mechanism will be analysed later on in
detail). Even though the two hypothesis for the origin of the slip are different, they lead to the
same physical results at ULS (sliding of the panel along with the yielding of the reinforcement).
A section made through the mid-plane of the panel shows doweling of a horizontal rebar located
at the point C, as well as its interaction with the surrounding concrete matrix (refer to confining
concrete regions in Fig.4.7c).
Failure of panels reported in the experimental campaigns usually combines two or even
all three presented mechanisms (crushing, spalling and sliding), which limit the concrete
compressive strength simultaneously. Depending on material and geometrical properties of
analysed members, one of them is governing at ULS.
One of the first models to account for steel-to-concrete interaction was introduced by
Collins [22] in 1978, as a part of the Compression Field Theory:
3. 6
   1 .0 (4.13)
2   1   3 
1
0
where ε1 represents the principal tensile stain (introduced as a positive value);
ε3 represents the principal compressive strain (introduced as a negative value):
ε0 represents the strains corresponding to cylinder compressive strength (assumed equal to 2‰).
The reduction factor is presented in form of a surface diagram in Fig.4.8a, according to
which effective concrete compressive strength depends both on tensile and compression strains.

89
Chapter 4: Effectiveness factor for concrete compressive strength

(a) (b)
1.2 1.2
1.0 1.0
0.8 0.8
ηε 0.6 ηε 0.6
0.4 0.4
0.2 0.2
0 0
0 0
3 0 3 0
6 3 6 3
9 6 9 6
ε1 / ε0 9 ε1 / ε0 12 9
12 12 ε3 / ε0 12 ε3 / ε0
15 15

Figure 4.8: Concrete compressive strength softening law taking into account the concrete interaction
with reinforcement: (a) according to Collins 1978 [22]; (b) according to Vecchio and
Collins 1986 [120]
The proposed empirical expression has two important issues:
1. Reduction of the concrete compressive strength as a function of principal compressive
strains is not physically consistent;
2. The fact that the tensile and compressive strains are biased means that the concrete
compressive strength depends on the total amount of strains in a system, rather than the
amount of imposed tensile strains.
γ/2 γ/2

Case 1 Case 2

ε3 ε1 ε ε3 ε1 ε

Figure 4.9: Two different strain states of a structural concrete member governing the same concrete
compressive strength according to Collins [22]
In other words, the same effective concrete compressive strength is obtained when
imposing higher transversal strains into a system, in case this increase is compensated with
lower compressive strains. For example, effective concrete compressive strength for Case A is
identical to the one from Case B in Fig.4.9.
The two inconsistences were later corrected in 1986, when Vecchio and Collins [120]
introduced the Modified Compression Field Theory (MCFT), which is one of the most
commonly used expressions for the reduction of concrete compressive strength accounting for
the interaction with the reinforcement:
1
   1.0 (4.14)
0.8  170  1

Eq.4.14 is presented in a form of a surface graph in Fig.4.8b. When comparing the

failure surface from Fig.4.8a to that from Fig.4.8b, it can be seen the reduction of the concrete
compressive strength is much smaller for the same amount of transversal tensile strain. In

90
 4.1 Introduction and state of the art

addition to this, Eq. 4.14 is completely independent from the principal concrete compressive
strains. Compared to experimental results, MCFT proved to be quite accurate and easy to
implement (as shown in [120]). Nonetheless, the model of Vecchio and Collins is semi-
empirical, and as such it does not distinguish between the various failure modes presented in
Fig.4.7. Without a proper prediction of the governing failure mechanism (concrete crushing,
spalling or sliding), it is difficult to enhance the strength of an analysed element in the most
effective manner.
After the introduction of MCFT, multiple authors have been working on improving the
compression softening equation, by conducting additional experimental campaigns in order to
invoke all three failure modes presented in Fig.4.7. The range of mechanical properties of tested
RC panels was widened, as well as type of loading, the number of reinforcement layers, size of
the elements etc. Some of the most significant woks in this field includes the approach of
Belarbi and Hsu [10]:
0.9
  (4.15)
1  250  1

The proposed expression governs greater strength reduction for elements subjected to
lower transversal tensile strains, compared to Eq.4.14. This can be explained by the fact the RC
panels that fail in concrete cover spalling or concrete sliding, usually reach smaller principal
tensile strains at failure. The expression of Belarbi and Hsu tried to account of such behaviour.
Nevertheless, the softening of concrete is still estimated in a semi-empirical manner, depending
only on the amount of transversal tensile strains in a panel.
Another semi-empirical softening expression was introduced by Kaufmann [52], who
analysed elements which have been subjected to significant transversal strains (ε1/ε0 up to 15,
assuming ε0 equal to equal to 2‰). According to him the effective concrete compressive
strength should be accounted for in the following manner:
1
  (4.16)
1.08  81 1

It is important to emphasise that according to Kaufmann [52], Eq.4.16 should be

multiplied with the brittleness concrete compressive strength factor proposed by Muttoni (refer
to Eq.4.12, fco=20 MPa), in order to obtain the effective concrete strength.
Zwicky [129] also proposed a simplified expression for reducing the concrete
compressive strength:

  0.67 14 1 (4.17)

Previous equation proved to be accurate when compared against a series of structural

concrete beams tested by the author, in which case the imposed transversal tensile strains at
failure varied between 5‰ and 25‰.

91
Chapter 4: Effectiveness factor for concrete compressive strength

In order to improve the results of MCFT by taking into account the sliding which occurs
along the initial cracks of RC panels, Vecchio [117] introduced another expression for concrete
compressive strength reduction factor:
1
  (4.18)
1
0.95  0.15 
0
Eq. 4.18 was implemented in the Disturbed Stress Field Method [117], which was able
to capture the behaviour of panels with small rotation of the principal stress field direction at
failure more accurately than MCFT (refer to [120]).
A direct comparison between the different approaches for estimating the softening of
concrete compressive strength due to imposed transversal strains (refer to Eq.4.14 - Eq.4.18),
is presented in Fig.4.10a. It can be seen that the reduction proposed by Belarbi and Hsu [10],
Kaufmann [52] and Zwicky [129] govern similar results, which is slightly different compared
to that of Vecchio and Collins [120] and Vecchio [117].
(a) (b)
1.2 1.2
Vecchio and Collins ‘86 Hars ‘06
Belarbi and Hsu ‘95 MC2010 (LoA I)
1.0 1.0 MC2010 (LoA II)
Kaufmann ‘98
Vecchio ‘00
0.8 Zwicky ‘02 0.8

ηε 0.6 ηε 0.6

0.4 0.4

0.2 0.2

0 0
0 3 6 9 12 15 15 20 25 30 35 40 45
ε1 / ε0 θ [°]

Figure 4.10: Concrete compressive strength softening laws taking into account the presence of
transversal strains: (a) according to various authors; (b) simplified approach of
Hars [41] and recommendations provided by MC2010 [34]
In general, the first three approaches, indicate more significant concrete compressive
strength reduction for elements with less transversal strain, whereas the former two do the same
in case of members subjected to higher transversal tensile strains. All proposed expressions
showed satisfying behaviour when compared to corresponding databases (refer to the original
publications). However, each approach lacks a clear physical background, which would allow
it to predict one of the three failure mechanisms presented in Fig.4.7 as governing (concrete
crushing, spalling and sliding). Therefore, the development of a mechanically based model is
necessary in order to truly understand the actual behaviour of structural elements. This is
important when designing new members, since a variation in the rebar diameter and spacing
may significantly influence the concrete compressive strength (see Fig.4.7). The same can
happen when changing the thickness of the concrete cover (refer to Fig.4.7b). Using current

92
 4.1 Introduction and state of the art

semi-empirical expressions (refer to Eq.4.14 – Eq.4.18) can only suggest if the weakest part of
a structure is the concrete or the reinforcement.
Aside from estimating the effective concrete compressive strength in the most accurate
manner, it is also important to provide simplified expressions for the design purposes (presented
in Fig.4.10b). According to Hars [41], the effective concrete compressive strength can be
determined based on the direction of the principal concrete compressive stress. Such approach
is very straightforward to use, especially in case of reinforced and prestressed concrete beams
which is very important in practice:

  0.5  1  sin2   (4.19)

where θ represents the direction of the principal compression stresses.

MC2010 [34] on the other hand introduces the same reduction following the LoA
approach in a slightly different manner. In case of the 1st LoA, recommended concrete
compressive strength efficiency factor for reinforced and prestressed concrete beams is 0.55.
At the same time, the minimal inclination of the principal concrete compressive stresses
direction is equal to 30° in case of RC elements, and 25° in case of prestressed concrete elements
(refer to Fig.4.10b). The 2nd LoA introduces an expression, which estimates the effective
concrete compressive strength as a function of the principal tensile strains in the following
manner:
1
   0.65 (4.20)
1.2  55  1

1   x  ( x  0.002)  cot 2  (4.21)

where εx represents the longitudinal strain at the mid-depth of the effective shear depth.
The minimal principal concrete compressive stress direction for reinforced and prestressed
concrete elements is limited to:

min  20  10'000  x (4.22)

Fig.4.10b compares all three simplified approaches for estimating the concrete
compressive strength efficiency factor. Even though the approach of Hars [41] governs slightly
higher values compared to 2nd LoA of MC2010 [34], the two curves are almost parallel. This
indicates that the behaviour of an element does not change significantly when applying the two
different methods. It should be noted that even though the approach of Hars [41] is very
straightforward to use, it implies that the inclination of the principal compressive field in a
reinforced concrete beam can be lower than 20°, which is not necessarily consistent with
Eq.4.22. The longitudinal strain at the mid-depth (εx) is usually positive, meaning that θmin
should not be smaller than 20° (refer to Eq.4.22). Finally, according to the MC2010 [34], the
effective concrete compressive strength determined using the 1st LoA represents a safe estimate
compared to the 2nd LoA in case of RC elements. In case of prestressed concrete however,
ηε=0.55 slightly exceeds the limit imposed by the 2nd LoA for angles between 25° and 27°.

93
Chapter 4: Effectiveness factor for concrete compressive strength

From a practical point of view this is not a problem as long as the strength of an analysed
element is not governed by the crushing of the concrete. This is due to the fact that the 2nd LoA,
allows lowering the direction of the principal concrete compressive stress (thus activating more
stirrups), meaning that even though the concrete compressive strength is lower, the ultimate
strength of an element is higher. However, if according to the 1st LoA element fails due to
crushing of the concrete while the direction of the principal concrete compressive stress is
smaller than 27°, switching to the 2nd LoA will slightly decrease its resistance.

4.1.3 Concrete softening caused by the presence of post-tensioning ducts

Placing steel or plastic post-tensioning duct in webs of structural concrete elements
influences the flow of the principal compressive stresses throughout the thickness of analysed
members. After conducting compressive tests on structural concrete panels (such are the ones
presented in Fig.4.11a and b), it has been observed that the presence of post-tensioning ducts
invokes out-of-plane strains into the section, which leads to failure of investigated specimens
([37,65,21,18,100,36]). The right-hand side of Fig.4.11a shows the principal compressive stress
flow (refer to dashed lines) inside a panel with an empty post-tensioning duct.
½bw-bI bI
(a) (c)
σc
A B
W7 σc

σc
½bI
C D E
F G P x
σc

Model
(b)
a
W5
φ
σcII σcI

H I J σcII σcI K
σt
φ
½ØD
L M N O

σcII σcI

½ØD
sinφ ½(bw-ØD)

Figure 4.11: Interaction of structural concrete with prestress ducts: (a) crack pattern and stress flow
in a panel with an ungrouted duct; (b) crack pattern and stress flow in a panel with a
grouted duct; (c) mechanical model proposed by Hars [41]
Given a fact that an empty duct represents a void in the concrete matrix, means that the
principal concrete compressive stresses need to deviate around this discontinuity region.
Consequently, the tensile strength of concrete is utilized in order to facilitate the necessary

94
 4.1 Introduction and state of the art

deviation (see full lines in Fig.4.11a). At one point, the tensile strength of concrete is surpassed,
which leads to formation of in-plane cracks and failure of the panel. Such behaviour can be
seen on one of the panels tested by Hars [41] (refer to the left-hand side of Fig.4.11a).
Grouted post-tensioning ducts on the other hand present the stiffest zones in the concrete
matrix, causing the compressive stresses to deviate towards it (see Fig.4.11b). Similar to
previous situation, this deviation is accommodated by the tensile strength of concrete (full lines
in Fig.4.11b). As the applied load increases so does the amount of tensile stress necessary to
deviate the stress field in the proximity of the duct, which ultimately leads to cracking and
failure (as presented in Fig.4.11b).
On average elements containing grouted plastic post-tensioning ducts have reached
lower ultimate strength compared to the ones that used grouted steel ducts. This can be
explained by the surface roughness of the two materials (plastic and steel, as indicated in
Fig.4.11b). When the internal compressive force reaches the duct, its tangential projection on
the circular surface needs to be equilibrated by the friction occurring between the duct and the
concrete in order for the compressive force to pass through. Since plastic is a smoother and less
stiff than steel, the amount of compressive force which can be transferred through the duct is
smaller. Consequently, more force flows around the duct inducing higher transversal tensile
stress in the section, which lowers the element’s strength.
Hars [41] developed a mechanically based model, predicting out-of-plane failures of
structural panels containing ungrouted or grouted (plastic and steel) post-tensioning ducts. The
equilibrium-based model is given in a form of a stress field and presented in Fig.4.11c. It shows
the internal stress distribution for one quarter of the entire cross section (as indicated in
Fig.4.11b). A segment of the post-tensioning ducts is located in the bottom left part of
Fig.4.11c. The panel is subjected to uniaxial compression from the top (resulting in σc
compressive stress), leaving its side surfaces completely stress-free.
Area ABCDE of the stress field (shaded using light-grey in Fig.4.11c) is subjected to
uniaxial compression (equal to σc). Zone FGEHIJK, presented in dark grey separates the stress
field in two branches – one passing through the duct (branch FGHI) and the other going around
it (branch GEJK). Both branches are subjected to simultaneous tension and compression, and
their widths are governed by the surface roughness of the duct, in the following manner:
  tan  (4.23)

where μ is the friction coefficient

φ is the angle defined in Fig.4.11c.
The top and the middle part of the stress field are separated with a quasi-hydrostatic
node that spans over the entire width of the panel (area CDEFG). The reason why the node is
referred to as quasi-hydrostatic is due to the fact that the Fig.4.11c describes a plane stress state,
meaning that stress in perpendicular direction is equal to zero. The line GJ is defined as a 2nd
order parabola, and is not completely perpendicular to the line GE. It is assumed that the failure
of the element occurs at point J once the maximal tensile stress (σt) reaches the effective tensile

95
Chapter 4: Effectiveness factor for concrete compressive strength

strength of concrete (fcte). The bottom part of the stress field (zones HILM and JKNO) is once
more subjected to uniform stress state (σcII and σcI, respectively), which is higher compared to
σc.
Equilibrium of the free-body DGJKE gives (refer to Hars [41]):
t
H : x  a (4.24)
 c  t
2  b1
V :  cI   c  (4.25)
bw   D

 b1 bw  
D
M P :  cI  bw   D        t  a  a  x  (4.26)
2 4 
After combining previous equations, one can link the maximal compressive stress (σc)
with maximal tensile stress (σt) in the following manner:

 2 a2 
c t   1 (4.27)
 b1  2  b1  bw  
 D  

Vertical equilibrium of free-body CFHIGD gives:

bw  2  b1
V :  cII   c  (4.28)
 D  sin 


Horizontal position of points D and G can be obtained by assuming that σcI (Eq.4.25) is
equal to σcII (Eq.4.28):
bw 1 
b1   (4.29)
2 1    1  sin 
D
 (4.30
bw

According to Hars [41], effective concrete tensile strength (fcte) can be estimated in the
following manner:

f cte 750   
 5/3  1  3c  (4.31)
f ct fc  fc 
where σ3c represents the applied compressive stress;
fct represents the maximal concrete tensile strength equal to:

f ct  0.3 f c2 / 3 (4.32)

As it can be seen fcte<fct, due to the fact that the concrete is not only subjected to pure
tension, but tension combined with compression (σ3c). This effect was observed and
investigated by Kupfer [63] and Curbach et al. [24], and Eq.4.31 was used as a failure criterion

96
 4.1 Introduction and state of the art

for the model of Hars [41]. Ultimate strength of a panel is reached at point J, once the effective
tensile strength of concrete is surpassed. Maximal compressive stress is obtained after
combining Eq.4.25 and Eq.4.29:
1
 3c   cI   c  (4.33)
1    1  sin 
Equations 4.32 and 4.33 can then be introduced to Eq.4.31, and finally, after assuming
that fcte=σt (from Eq.4.27) as well as normalizing the entire expression with fc, concrete
compressive strength softening coefficient taking into account presence of post-tensioning
ducts can be obtained:
1
 f2  2 1
4  a bw   1    1  sin    
2 
 c 1 
D    1       (4.34)
 225    1  sin    1    1    1  sin  

Previous expression is valid for ungrouted as well as grouted (steel or plastic) post-
tensioning ducts. The only difference between them is in the amount of compressive stresses
which can be transferred through the duct. In other words, the width of FGHI branch changes
(see Fig.4.11c) as a function of the friction coefficient, which translates into angle φ (refer to
Eq.4.23 and Fig.4.11c). In case of ungrouted ducts, the proposed friction coefficient is equal to
zero. Grouted plastic ducts on the other hand assume μ=0.2 (φ=0.2 rad), and finally the friction
coefficient of grouted steel ducts is equal to μ=0.6 (φ=0.54 rad). According to Hars [41] the
local stress field disturbance ratio a/bw (see Eq.4.34) should be equal to two.
Model for estimating the effective concrete compressive strength of structural concrete
members containing a post-tensioning duct proposed by Hars [41] has been presented in form
of a graph in Fig.4.12a to c. Equation 4.34 assuming μ=0 and varying fc between 20 MPa and
100 MPa can be seen in Fig.4.12a as a function of δ . Similar curves are given in Fig.4.12b and
c, assuming μ=0.2 and μ=0.6, which corresponds to grouted plastic and steel ducts. According
to Hars [41] concrete compressive strength reduction caused by the presence of post-tensioning
ducts is quite significant, especially in case when empty ducts are combined with high strength
concrete. However, in practice grouted steel ducts are used most of the time. Fortunately Hars’s
model predicts that the effective concrete compressive strength remains relatively high in such
cases (see Fig.4.12c).
MC2010 [34] proposes a simplified expression for estimating the effective concrete
compressive strength of prestress concrete, which is especially suitable for design purposes.
The reduction can be applied on material or geometrical level, by reducing the concrete
compressive strength or thickness of the web in the following manner:
D  1  kD   (4.35)

bw,eff  bw  k D  
D (4.36)

where bw,eff represented the effective (reduced) web thickness;

bw represents the actual web thickness;

97
Chapter 4: Effectiveness factor for concrete compressive strength

kD represents the post-tensioning duct coefficient, which is equal to 0.5 in case of grouted steel
ducts, 0.8 in case of grouted plastic ducts and 1.2 in case of ungrouted duct.
(a) (b)
1.2 1.2
Ungrouted ducts Plastic grouted ducts
1.0 1.0

0.8 0.8

ηD 0.6 ηD 0.6

0.4 0.4
20 MPa 20 MPa
40 MPa 40 MPa
0.2 60 MPa 0.2 60 MPa
80 MPa 80 MPa
100 MPa 100 MPa
0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
δ δ
(c) (d)
1.2 1.2
Steel drouted ducts MC2010
1.0 1.0

0.8 0.8

ηD 0.6 ηD 0.6
δ=0.125

0.4 0.4
20 MPa
40 MPa
0.2 60 MPa 0.2 Ungrouted ducts
80 MPa Plastic grouted ducts
100 MPa Steel grouted ducts
0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
δ δ

Figure 4.12: Concrete compressive strength reduction factor taking into account the presence of
prestress ducts (assuming a/bw=2):(a) model of Hars [41] applied to ungrouted ducts;
(b) model of Hars [41] applied to grouted plastic ducts: (c) (a) model of Hars [41]
applied to grouted steel ducts; (d) recommendations of MC2010 [34]
Equation 4.35 can be seen in Fig.4.12d, where ηD factor is presented in function of δ
(refer to Eq.4.30). Compared to curves proposed by Hars [41], MC2010 [34] does not directly
depend on the concrete compressive strength, nor does it take into account the influence of
imposed tensile stress on strength of a member. In addition to this, it governs less conservative
values of concrete compressive strength (especially in case of high-strength concrete). This is
why the multiplicative approach of MC2010 [34] for estimating the effective concrete
compressive strength has been investigated and criticized by Hars [41].

98
 4.1 Introduction and state of the art

4.1.4 Critique of the multiplicative approach for estimating effective concrete compressive
strength
When looking at a prestressed concrete panel at ULS (presented in Fig.4.13a), it can be seen
that there are two distinctive families of cracks:
1. Out-of-plane cracks, accounted for through ηε effectiveness factor;
2. In-plane cracks, accounted for through ηD effectiveness factor
The multiplicative approach of MC2010 [34] (presented in Eq.4.3) implies that the two
reduction factors (ηε and ηD) should be combined in order to obtain the effective compressive
strength. However, when looking at the physical origins of η –coefficients it can be seen that
even though they act simultaneously, concrete compressive strength is reduced in two
perpendicular planes (as presented in Fig.4.13a). Hence, the multiplicative approach of
MC2010 [34] does not entirely capture the mechanical behaviour of prestressed concrete
panels.
In addition to this, Hars [41] showed that despite the fact that multiplicative approach
seems to be on the safe side, this is not the case when ungrouted or grouted plastic post-
tensioning ducts are applied, due to the fact that the presence of deviating tensile stress is
completely neglected in the cross-section. In order to be more mechanically consistent,
Hars [41] proposed following expression for obtaining the effective concrete compressive
strength:


f ce  f c  fc  min  D (4.37)
   1  k D   
where ηD represent the concrete compressive strength factor taking into account the presence of
the duct, which should be obtained according to Eq.4.34 ;
ηε represents the concrete compressive strength effectiveness factor taking into account the
presence of transversal strains in the concrete, which should be assumed according to
MC2010 [34];
kD and δ are respectively the duct type coefficient and relative post-tensioning ducts thickness,
and should be assumed according to MC2010 [34].
Expression 4.37, clearly distinguishes between the in-plane and out-of-plane failure of
panels, while at the same time acknowledges that the presence of the post-tensioning ducts
causes additional in-place stress concentration by multiplying the ηε coefficient with (1-kD·δ).
Fig.4.13b to d gives the results of concrete compressive strength criterion from Eq.4.37 applied
on different types of post-tensioning ducts and various concrete compressive strengths. The
effect of ηD was neglected for all the cases when δ ≤ 0.125 (as recommended by MC2010 [34]).
Each of the graphs clearly indicates whether the minimal concrete compressive strength was
governed by ηD or enhanced ηε reduction factor. Fig.4.13b and c shows that when high-strength
concrete (fc above 80 MPa) is combined with ungrouted and grouted plastic ducts, it is ηD not
enhanced ηε that governs the strength of a member. This means that the MC2010 [34] code
provisions are unconservative in such situations. However, in case of grouted steel post-

99
Chapter 4: Effectiveness factor for concrete compressive strength

tensioning ducts current code provisions are on the safe side (as can be seen in Fig.4.13d). It is
important to mention that the three curves from Fig.4.13d (corresponding to concrete
compressive strength of 60 MPa, 80 MPa and 100 MPa) are overlapped, which explains why
only the red line remains visible.
(a) (b)
1.2

A-A Ungrouted ducts

A
σc 1.0 60 MPa
80 MPa
Out-of-plane 100 MPa
crack
0.8
δ=0.125

ηε
ηD
ηε
A 0.6
ηε ηD η ε
(1-kD
δ)
η ε
(1-kD
δ)
In-plane 0.4
crack ηD

σcI σcI 0.2

0
0 0.1 0.2 0.3 0.4 0.5
δ
(c) (d)
1.2 1.2
Plastic grouted ducts Steel drouted ducts
1.0 60 MPa 1.0 60 MPa
80 MPa 80 MPa
100 MPa 100 MPa
0.8 0.8
δ=0.125 δ=0.125
η ε
η D

η ε
η D

ηε ηε
0.6 0.6
η ε
(1-kD
δ) η ε
(1-kD
δ)
η ε
(1-kD
δ)
0.4 ηD 0.4

0.2 0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
δ δ

Figure 4.13: Critique of the multiplicative approach for estimating effective concrete compressive
strength: (a) crack distribution in an prestressed concrete panel; (b) to (c) parallel
approach of Hars [41] compared to multiplicative approach of MC2010 [34] applied on
members with ungrouted and grouted plastic/steel ducts
After presenting, comparing and analysing current models used to estimate the concrete
compressive strength effectiveness factor (refer to Eq.4.3 - Eq.4.37), it becomes clear that the
mechanisms limiting in-plane and out-of-plane resistance of structural members need to be
clearly distinguished and accounted for separately from each other. Going forward even further,
it is necessary to derive a mechanically based model which is able to differentiate between the
three failure modes presented in Fig.4.7:
1. Crushing of the concrete strut;

100
 4.1 Introduction and state of the art

2. Sliding of the concrete;

3. Spalling of the concrete cover.
This will allow designers not only to suitably understand the mechanisms governing the
structural failures, but to comprehend their interaction as well as.
This chapter presents a mechanical approach for assessing the concrete compressive
strength efficiency factor that takes into account actual behaviour of structural panels as well
as interactions between steel and concrete. The consistency of the approach was validated by
means of extensive comparison to available test data. Finally, the model was used to better
understand the relationship between various strength-governing mechanisms.

101
Chapter 4: Effectiveness factor for concrete compressive strength

4.2 Mechanical model for compression softening of reinforced concrete panels

The response of a reinforced concrete panel subjected to shear and/or to bi-axial
tension/compression can be described by analysing three physical stages presented in Fig.4.14:
1. The first stage corresponds to a load transfer action valid up to first cracking of the panel
(refer to Fig.4.14a);
2. The second stage (refer to Fig.4.14b) describes behaviour which is applicable between
first cracking of the panel and yielding of the reinforcement in one direction (this phase
is also valid in case a panel fails prior to yielding of any steel);
3. The third and final stage (refer to Fig.4.14c) covers the loading history between yielding
of the reinforcement in one direction and failure of the panel (due to the crushing of the
concrete or yielding of the reinforcement in both directions).
τxz,y ,σz,y τxz,ult ,σz,ult
(a) z τxz,cr ,σz,cr (b) z (c) z
θcr θcr θult θcr

τzx,cr τzx,y τxz,ult

σx,cr σx,y σx,ult

x θy x x

(d)
2·c+Øz 2·c+Øx
Øz
Undamaged struts Ultimate cracks
bw
Ultimate cracks Damaged struts Confining region - primary doweling
Unconfined rebar Confining region - secondary doweling
Øx Confined rebar Plastic hinges - primary doweling
Initial cracks Plastic hinges - secondary doweling
Initial crack (ητ ) Spalling cracks (ηs and ηdow)

Figure 4.14: Modelling the physical behaviour of a reinforced concrete panel subjected to shear and
axial loads: (a) cracking point; (b) onset of yielding; (c) failure of the panel and (d) in-
plane and out-of-plane reduction mechanism of concrete compressive strength

4.2.1 Stages of mechanical behaviour

The behaviour of a panel will be characterized for an element as shown in Fig.4.15a. Also the
notations indicated in Fig.4.15b and Fig.4.15c will be used to refer to the stress and strain state
respectively.

4.2.1.1 First stage : Linear elastic response

During the early phases of loading, a reinforced concrete panel can be analysed
assuming a linear elastic behaviour. Once the tensile strength of concrete is reached, cracks start
to develop (refer to Fig.4.14a). As the direction of the principal tensile strains can be assumed

102
 4.2 Mechanical model for compression softening

parallel to principal tensile stresses for the elastic response, this means that the stress state,
strain state and initial cracking angle (θcr) are unambiguously defined (refer to Fig.4.14a).
z
(a) τxz ,σz bw (b) τ (c) γ/2
τxz σcx γxz/2 εx Pole

lz σc3 σc1 σ ε3 θ ε1 ε
τxz ,σx θ

y
Pole τxz σcz γxz/2 εz
lx x

Figure 4.15: (a) Geometry and loading properties of an analysed panel; (b) stress state of a concrete
strut and (c) average strain state of the panel
The amount of stress acting along the edge of a panel at any given point in load-history,
can be defined using the applied force as follows (refer to Fig.4.14a):
Fx
σx  (4.38)
h  bw

Fz
σz  (4.39)
l  bw

V xz
τ xz  (4.40)
l h
where Fx represents the applied force in x-direction;
Fz represents the applied force in z-direction;
Vxz represents the applied shear force;
l represents the size of the panel in the x-direction;
h represents the size of the panel in the z-direction;
bw represents the thickness of the panel.
Assuming that the panel behaves as a uniform elastic continuum up to the cracking, one
can calculate the principle tensile stress of the panel as:
2
σ  σz  σ  σz 
σ c1  x   x   τ xz
2
(4.41)
2  2 
The limit case of the elastic phase is reached when the principle tensile stress becomes
equal to the mean tensile strength of concrete (σc1 = fctm). According to MC2010 [34], this value
can be estimated as:

0.3   f c  , for f c  50MPa

2/ 3

f ctm   (4.42)
 2.12  ln(1  0.1  f c ), for f c  50MPa

103
Chapter 4: Effectiveness factor for concrete compressive strength

where fc represents the uniaxial concrete compressive strength measured in cylinder.

The initial cracking angle can now be determined using Mohr’s transformations (presented in
Fig.4.15b) as following:
σ cx  σ c 3
tan θcr  (4.43)
τ xz

4.2.1.2 Second stage: Cracked behaviour

The initial cracking angle remains unchanged throughout the second stage (as presented
in Fig.4.14b). However, the direction of the principal compressive stresses starts to rotate in
order to remain parallel to the principal compressive strains (θy in Fig.4.14a). This means that
a portion of the concrete compressive struts, named hereafter the damaged struts, crosses the
initial cracks (refer to Fig.4.14b). The strength of such struts is governed by aggregate
interlocking occurring along the initial cracks. The remaining portion of the struts are
considered as undamaged struts. Even though they are characterized by the absence of in-plane
cracks, their strength is not equal to the concrete compressive strength determined from the
standard cylinder test. This is justified as the presence of the reinforcement induces the
formation of spalling cracks, leading to out of plane failures (refer to Fig.4.14d). These spalling
cracks limit the compressive strength of the concrete over a given portion of the panel’s
thickness (shown in red in Fig.4.14d). This effect has been taken into account based on the
work of Hars [41]. The spalling reduction applies not only to the undamaged, but to the
damaged struts as well. In this manner, the strength of each strut is governed by the presence
of in-plane and/or out-of-plane cracks.
With respect to the yield strength of the reinforcement, it is limited by the doweling of
the bars in the region near the cracks. As plastic hinges start to develop (indicated using white
circles in Fig.4.14b), the concrete located between the surfaces of the cracks and the plastic
hinges are locally equilibrating the doweling forces (see the red zones along the reinforcement
in Fig.4.14b). This region, shown using dashed lines in Fig.4.14b, which is partly developing
in the undamaged and damaged struts, introduces a reduction of the concrete strength due to
the out-of-plane tensile stresses, potentially leading to spalling failures (presented in Fig.4.7b).

4.2.1.3 Third Stage: Behaviour after yielding

The third and final stage is presented in Fig.4.14c. At this point, a second family of
cracks develops. Their inclination is assumed to be parallel to the direction of the principal
concrete compressive stress (and strain) at failure (θult). The presence of a second family of
cracks causes additional doweling of the reinforcement (plastic hinges activated at this phase
are presented in Fig.4.14c using red circles). Given the fact that the direction of principal
compressive stress continues to rotate compared to the previous stage means that the percentage
of surface corresponding to damaged struts increases, and that of the undamaged struts
decreases. The strength reduction mechanisms governing the strength of each type of the
concrete struts are methodologically the same compared to previous stage. The stress field
continues to rotate until the reinforcement in the other direction reaches eventually yielding or

104
 4.2 Mechanical model for compression softening

until the principal compressive stresses become equal to the effective concrete compressive
strength.

4.2.2 Plastic strength of rebars subjected to doweling and elongation

As the initial and the ultimate cracks start to open, the horizontal and vertical
reinforcement of the panel is subjected to local bending. Consequently, plastic hinges start to
develop in the vicinity of the crack faces (refer to Fig.4.14b and Fig.4.14c), originating
doweling action. As the bar is dowelled, this reduces its capacity to carry tensile forces and
determines the location at which plastic hinges will originate. Physical evidence of the doweling
action in the reinforced concrete panels has been reported for instance by Vecchio et al. [121]
in their experimental campaign. In addition to this, panels which have experienced sliding
failures at ultimate limit state (refer to Fig.4.7c) also show significant doweling of the
reinforcement.
(a) Stress: (b)
fy,r< fy Ndow 1.2

fy Ndow 1.0 4·Ndow

fy Mdow 0.8 fy·Ø2·
Strain:
ε 0.6
Δ
α 0.4
6·Mdow
0.2
fy·Ø3
0
Ø 0 15 30 45 60 75 90
θ [°]

Figure 4.16: (a) Distribution of strains, stresses and internal forced inside a plastic hinge of a rebar;
(b) axial strength reduction of a horizontal rebar as a function of the crack opening angle
The plastic stress distribution within a hinge at the ultimate limit state is presented in
Fig.4.16a. The central part of the cross-section carries tension, while the rest carries the bending
moment. In order to account for such interaction, the apparent yield strength of the steel in
tension needs to be reduced (fy,r). It must be noted, however, that this detrimental effects of the
doweling in one direction (fy,r), brings some beneficial effects to the opposite direction (Vdow),
as presented in Fig.4.17d and Fig.4.17e This is justified as the bent rebar can resist some shear
forces, reducing the amount of the force that has to be carried by the bars in the perpendicular
direction. Depending on the inclination of the crack, the number of rebars which are actively
contributing in carrying applied loads through the doweling action can vary. In case this angle
is higher than 45° (refer to Fig.4.17d) the number of rebars is governed by their spacing (sx and
sz) and the height of a panel (h), whereas in case the angle is lower than 45° (see Fig.4.17e) the
number is determined based on the length of a panel (l) and the spacing between the rebars (sx
and sz).
The development of tensile and dowel forces in a plastic hinge for a given crack
direction and kinematics is investigated in Fig.4.17. The inclination of the initial crack is
defined by θcr, and the direction of its opening by θ. These two angles are considered to account

105
Chapter 4: Effectiveness factor for concrete compressive strength

that the crack is not necessarily (at all cases) parallel to the principal compressive stress and
strain direction, and is therefore subjected to simultaneous opening and sliding.
wini (c) wini
(a) (b)
w ldow w
ldow ldow ldow ½ζ w·sinθ
τb θ θ
w θcr Δ θcr
τb τb α Sliping of the
plastic hinge
Plastic hinge w·cosθ
Future crack w·cosθ w·sinθ τb (1-ζ)w·sinθ Δ
ldow
ldow ½ζ· w·sinθ
Confining struts

τxz ,σz τxz ,σz

σcc
(d) (e)
Mdow
Mdow Vdow
nx·fyx nx·fyx
σcc
sx θ Vx,dow sz
Vx,dow
τzx sx Vz,dow
nx·fyx σx nx·fyx nz·fyz
τzx Vx,dow
σx h Vx,dow h Mdow
θ Vz,dow
nx·fyx nz·fyz
Vz,dowVx,dow Vdow
nz·fyz
nx·fyx
Vx,dow Ndow
l l

Figure 4.17: Doweling mechanism of the reinforcement: (a) location of the future plastic hinges and
the initial crack; (b) local bending of a horizontal rebar in the proximity of an initial
crack – no slip along the rebar; (c) local bending of a horizontal rebar in the proximity
of an initial crack – accounting for the slip along the rebar; (d) contributing rebars for
crack angles bigger than 45° and (e) contributing rebars for crack angles smaller than
45°
The plastic hinges are located at a distance to the crack named the doweling length (ldow,
see Fig.4.17a). When the crack opens, the plastic hinges are horizontally and vertically moved
from each other (refer to Fig.4.17b). Since the rebar is not perfectly bonded to concrete, it slips
for a value which can be expressed as function of the corresponding crack width. This is defined
by coefficient ξ in Fig.4.17c. Consequently, the rebar tilts (at a specific angle α) and elongates
for 2·Δ. Along the doweling length, the rebar pushes against the surrounding concrete
transferring the doweling forces to the concrete (σcc). According to Rasmussen [99] the
doweling length for the case of doweling with no eccentricity and no normal load can be
estimated as:

fy
l dow    (4.44)
3  f cc

where ldow represents the doweling length;

Ø represents the dimeter of the rebar;
fy represents the initial yield strength of the rebar;
fcc represents the concrete strength in confining state.
Sørensen [111] states that in case of an increased eccentricity and additional normal
load, the doweling length actually decreases, which reduces the moment contribution as well

106
 4.2 Mechanical model for compression softening

as the percentage of confining (i.e. damaged) struts within a panel (refer to Fig.4.14b and
Fig.4.14c). Therefore, the presented approach is slightly conservative.
The enhanced concrete strength equilibrating the doweling forces can be estimated
accounting for the influence of the concrete cover. Based on the work of Vintzeleou and Tassios
[123], the confined concrete strength can be estimated as:
2
 cconf ,1  cconf , 2 
f cc  f c  1    0.02   0.15    5  f c (4.45)
    
where cconf,1 represents the concrete cover perpendicular to the confining concrete stress (cannot
be larger than 4Ø) and cconf,2 represents the concrete cover parallel to the confining concrete
stress (cannot be larger than 8Ø).
(a) z (b) z (c)
Øz , ρz srm srm
Effective tie
σs ldow
sz Øeff , ρeff F F
seff
Plastic hinge
sx seff
Øx
ρx
ε εs2 εs1 ξ=2δ/w
srm

δ εscr w
θcr θcr x
x x

Figure 4.18: (a) Geometry properties of the reinforcement in horizontal and vertical direction;
(b) geometry properties of the effective reinforcement and effective tie and (c) strain state
along the effective tie
On average, the corresponding values for the confined concrete strength are
approximately equal to 3.7·fc. The actual concrete strength (σcc), is however determined at a
later stage accounting for the actual doweling moment in the rebar (Mdow), and is always lower
than fcc.
Relative rebar slip is estimated using an effective tie presented in Fig.4.18b and
Fig.4.18c. Starting from the geometry properties of an analysed panel (given in Fig.4.18a),
crack spacing is estimated using the expression resulting for a constant bond strength according
to [120]:
1
s rm  (4.46)
sin θ cr cos θ cr

s rmx s rmz

3 1ρx 

srmx    x (4.47)
2 ρx 8

3 1ρz 

srmz    z (4.48)
2 ρz 8
where ρx represents the reinforcement ratio in x-direction;

107
Chapter 4: Effectiveness factor for concrete compressive strength

Øx represents the dimeter of the rebar in x-direction;

ρz represents the reinforcement ratio in z-direction;
Øz represents the dimeter of the rebar in z-direction.
and where the effective reinforcement ratio in the direction perpendicular to the direction of
cracks (refer to Fig.4.18b) is calculated in the following manner:
ρeff  ρ x sin 2 θ cr  ρ z cos 2 θ cr (4.49)

The effective rebar spacing is assumed equal to the average spacing of the horizontal
and vertical reinforcement projected to the direction of the cracks:

1  sx sz 
s eff      (4.50)
2  sin  cr cos  cr 
where sx represent the spacing between the rebars in x-direction;
sz represent the spacing between the rebars in z-direction.
Fig.4.18c represents the reinforcement strain distribution inside the effective tie at its
cracking point, and for a higher stress level. As it can be seen, both strains are linearly varying
between the two consecutive cracks (constant bond strength). Maximal steel strains – εs1 are
expected at the location of the crack, while the minimal steel strains – εs2 are expected in the
middle of the two consecutive cracks (concrete strains are neglected):
s
 s1  (4.51)
Es

s 2   b  s rm
 s2   (4.52)
Es E s   eff

where σs represents the stress in the reinforcement;

Es represented the Young’s modulus of elasticity;
Øeff represents the effective rebar dimeter (presented in Fig.4.18b);
τb represents the average bond stress, which can be estimated according to Marti et al. [74] (for
deformed rebars):
τ b  2  f ctm (4.53)

where fctm represents the mean tensile concrete strength given in Eq.4.42
In case or smooth rebars the bond law is calculated according to MC2010 [34]:

τ b  0.15  fc (4.54)

Plastic hinges are assumed to form at the cracking point of the panel in the vicinity of
the initial cracks (as presented in Fig.4.18c). As the stress level increases in the rebar, the plastic
hinges move towards the crack. The slip (δ) can be estimated by integrating the area marked in

108
 4.2 Mechanical model for compression softening

red in Fig.4.18c. As presented in Fig.4.18c, this value depends on the amount of steel stress
(σs). A mean value between the steel stress at the point of cracking (σs,cr) and the yield strength
of the steel (fy) shall be assumed, in other words:
 s ,cr  f y
s  (4.55)
2
f ctm
 s ,cr  (4.56)
 eff

The amount of slip itself can be estimated as:

 s   s ,cr  s rm 
   l dow  (4.57)
Es  2 
In order to do express the slip of a hinge in a more general manner (in function of the
corresponding crack width), the ξ parameter is defined as:
2
 (4.58)
w
 s1   s 2
w  s rm  (4.59)
2
Based on the mechanism given in Fig.4.17a, the rebar tilting angle can be expressed as:

 w  cos  
  arctan  (4.60)
 2  ldow  1     w  sin  
Going back on the same mechanism, the rebar elongation can be determined in the following
manner (refer to Fig.4.17c):

2ldow  22  2ldow  1     w  sin 2  w  cos 2 (4.61)


1
2
 2

 4l dow  4l dow  w  sin   1     w 2  1  2  sin      sin    l dow
2 2
(4.62)

where w represents the total displacement, which can be estimated according to mechanism
presented in Fig.4.17b as:
wini
w (4.63)
cos( cr   )

where wini represents the initial crack width.

In order to estimate the amount of rebar doweling based on the crack kinematics, it is
necessary to define the incremental difference of the rebar tilting angle and its elongation in
function of the crack opening. In other words, one needs to find the first derivatives of Eq.4.60
and Eq.4.62:

109
Chapter 4: Effectiveness factor for concrete compressive strength

 2l dow  cos
  

w 4l dow  4l dow  w  sin   1     w 2  1  2  sin  2    sin  2
2
 (4.64)

 



2ldow  sin   1     w  1  2  sin      sin  
2 2
 (4.65)
w 2 4l 2 2

dow  4l dow  w  sin   1     w  1  2  sin      sin  
2 2

The yield condition of the circular cross section subjected to simultaneous tension and
bending is similar to the one derived for a rectangular cross section but is far more complex
[111], which is why the doweling of the rebar will be derived using the simpler expression [92]:

f  N, M   m  n 2  1  0 (4.66)

M 6M
m  3 (4.67)
Mp 
  fy

N 4 N
n  2 (4.68)
   f y
Np 

where M represents the bending moment acting on a rectangular cross-section;

N represents the axial force acting on the rectangular cross-section.
Application of the normality condition on Eq.4.66 gives:
f M , N  f M , N  m 1
       (4.69)
M m M Mp

f M , N  f M , N  n 1
        2n (4.70)
N n N Np

where λ represents an integration factor.

After combining Eq.4.64, Eq.4.65, Eq.4.69 and Eq.4.70, it is possible to estimate the
relative amount of axial tensile force in a rebar for a given crack kinematics:

ndow  

 3   2l dow  sin   1     w  1  2  sin  2    sin  2  
 4  
 4  l dow  cos   (4.71)
2

 4l dow  4l dow  w  sin   1     w 2  1  2  sin      sin  
2 2

Once the amount of the axial tensile force in a rebar is determined, the amount of
moment can be expressed as:

mdow  1  ndow
2
(4.72)

The distribution of these two parameters (ndow and mdow) can be seen in Fig.4.16b as a
function of the crack opening angle. As it can be seen, the maximal amount of moment is
obtained for θ=0°. At this angle the crack opens perpendicularly to the rebar. For a general case
(refer to point A in Fig.4.19a and Fig.4.19b), the neutral axis is inside the analysed rebar, which

110
 4.2 Mechanical model for compression softening

means that a part of the cross section is subjected to uniform compression while the rest is
subjected to uniform tension. When the bar is subjected to pure normal force, the stress state of
the point B can be achieved through various strain profiles (refer to Fig.4.19b), since the neutral
axis is always outside of the rebars cross section. According to the rigid plastic approach, in
point B it is therefore possible to have no moment but some level of curvature.

(a) (b) Point A Point B

. n <1 n =1 n =1 n =1
1.0 Δ m>0 m=0 m=0 m=0
.
α fy fy fy fy
0.8
A
0.6 ε ε ε ε
m [-]

. . . .
Δ . Δ . Δ . Δ
α α α
0.4

0.2 Potential re-bar

stress state
B
0
Ø Ø Ø Ø
0 0.2 0.4 0.6 0.8 1.0
n [-]

Figure 4.19: (a) Moment - normal force interaction diagram; (b) distribution of strains and stresses
in a rebar
The maximal shear carried by doweling can be calculated accounting for the pressure
developed in the concrete (Fig.4.17c):
V dow    l dow   cc (4.73)

where σcc represents the effective concrete pressure acting along the dowel length, which may
be estimated from a simple free-body equilibrium (Fig.4.17c):

2  M dow mdow   2  fy
 cc  (4.74)
  l dow
2
3  l dow
2

The total force carried by the doweling can thus be calculated on the basis of Fig.4.17d and
Fig.4.17e, by adding the contribution of each bar (Vdow) for the total number of bars intersected
by the cracks.

4.2.3 Concrete strength accounting for the presence of initial cracks and rebars
The effective concrete compressive strength of the analysed panel can be expressed by
combining the resistances of the two types of struts (damaged and undamaged) presented in
Fig.4.20a in the following manner:
f ce  f c  η fc  ηDS  qDS  ηUDS  1  qDS  (4.75)

where fc represents the uniaxial concrete compressive strength measured on a cylinder;

ηfc represents the effectiveness factor that accounts for the brittle behaviour of concrete in
compression according to Muttoni [82] (refer to Eq.4.12 assuming fc0=30 MPa);

111
Chapter 4: Effectiveness factor for concrete compressive strength

qDS represents the relative amount of the damaged struts in the panel;
ηDS represents the effectiveness factor for the compressive strength of the damaged struts;
ηUDS represents the effectiveness factor for the compressive strength of the undamaged struts.
Damaged strut
(a) (b) Initial crack (c)
Undamaged strut lcr,aver srm
lcr,i θcr

B θcr
h h h
srm

A Analysed
section
θcr θ θcr θ θ
l l l

(d) Detail B
Undamaged struts not affected by the confinement
Detail A δUDS,dow
δUDS,dow srm / cos(θcr- θ) Undamaged struts affected by the confinement
δUDS,dow Damaged struts (affected by the confinement)
ldow,z·cosθ θ
ldow,z
Plastic hinges
ldow,x·sinθ ldow,x θ Initial Real cracks
θcr crack
δUDS,dow Fictitious cracks

Figure 4.20: (a) Surface quantity and distribution of damaged and undamaged struts in a panel;
(b) number of initial cracks in a panel; (c) effective damaged strut length and width;
(d) details determining the confining concrete strut width
The relative amount of damaged (and consequently undamaged) struts in the analysed
panel depends on the length of the initial cracks (which can be seen in Fig.4.20a and Fig.4.20b).
In order to account for their uneven length, an effective (average) crack length is calculated
based on geometrical considerations from Fig.4.20b:
l  0  s rmx l  1  s rmx l  2  s rmx l  n  s rmx
 2  2    2 
cos θ cr  cos θ cr  cos θ cr  cos θ cr 
l cr , aver  (4.76)
2n 1
l s rm n  n  1
l cr , aver   
cos θ cr  sin θ cr   cos θ cr  2  n  1

where n can be calculated as:

l  sin  cr  (4.77)
n
s rm

After combining Eq.4.76 and Eq.4.77, it becomes:

l  n  (4.78)
l cr , aver   
cos θ cr   2  n  1 

Finally, the relative amount of damaged struts can be estimated on the analysed section
(indicated in Fig.4.20c) in the following manner:

112
 4.2 Mechanical model for compression softening

n '  l cr , aver  sin θ cr  θ   cos θ

q DS  (4.79)
h
h  cos  cr    (4.80)
n'  1
cos   s rm

where n’ represents the effective number of damaged struts;

srm represent the average initial crack spacing;
l represents the size of an analysed panel in x-direction;
h represents the size of an analysed panel in z-direction.

4.2.3.1 Strength of the damaged struts

The compressive strength of the damaged struts is limited by the parameter ηDS, which
takes into account:
1. Presence of the rebars in the concrete struts, governing the out-of-plane failures (ηs, in
analogy to Hars [41]);
2. Presence of the initial cracks crossing the damaged struts, governing the in-plane
element failures (ητ, which will be defined further in this paper);
3. Interaction between the in-plane and out-of-plane element failures (adopted from
Hars [41]);
4. Presence of doweled rebars (ηdow, which will be defined further in this paper).
The value of the effectiveness factor for the compressive strength of the damaged struts
can thus be calculated as:

ηs
  0   bw 

ηDS  min  η  max  η  min  (4.81)
 bw  4c  2ο ext   4c  2ο ext   1    δ '
 bw bw
dow

where ηs represents the effectiveness factor that accounts for the presence of the reinforcement
in concrete (out-of-plane failure);
ητ represents the effectiveness factor that accounts for the presence of the initial cracks in
concrete (in-plane-failure);
ηdow represents the factor which takes into account the presence of rebars pressing against the
concrete due to the doweling (see Fig.4.17c);
δ’ represents the ratio between the external rebar diameter and the sum of two times the concrete
cover thickness and the external rebar diameter;
bw represents the thickness of the panel;
c represents the concrete cover thickness.

113
Chapter 4: Effectiveness factor for concrete compressive strength

With respect to Eq.4.81 it can be seen that the resistance of a damaged strut is limited
either by the presence of the out-of-plane cracks (considered by means of the ηs coefficient), or
by the presence of the in-plane cracks (considered by means of the ητ coefficient) combined
with the doweling of the reinforcement (ηdow).
The effectiveness factor that accounts for the presence of reinforcement in concrete (ηs)
is developed on the basis of the work of Hars [41]. Reinforcement bars introduce local defects
into the concrete cover region which may cause it to spall off (see the red strips in Fig.4.14d),
while the rest of the section remains undisturbed. This effect is physically governed by the out-
of-plane cracks (named spalling cracks in Fig.4.14d), and can be quite pronounced in case of
heavily reinforced concrete panels. On the basis of [41] the effectiveness factor that accounts
for the presence of the reinforcement in concrete (ηs) is evaluated as:

 0   bw 
max   min  
bw  4c  2ο ext   4c  2ο ext 
ηs   
bw bw (4.82)
1
 f2  1

 1  δ '1  sin  s   
2
4 s
2
 c  1  1 
   
 225  δ '1  sin  s   1  δ'   1  δ '1  sin  s  
 

 1.5 
 1 
 s  min  sext  ο ext  (4.83)
 2 cos θ 2c  ο ext 

ο ext
δ'  (4.84)
2c  ο ext

arctan 0.6 for plain reinforcement bars

s   (4.85)
arctan 1.0 for ribbed reinforcement bars
where c represents the concrete cover;
sext represents the distance between the external reinforcement bars;
Øext represents the external rebar diameter;
ωs represents the local stress field disturbance ratio.
It should be noted that the Eq.4.83 gives the expression for the ωs coefficient assuming
that the external reinforcement is in the vertical direction. In case a horizontal rebar is closer to
the panel’s surface, cos(θ) should be replaced by sin(θ) in Eq.4.83. As already mentioned, the
effectiveness factor that accounts for the presence of the initial cracks in concrete (ητ) limits the
strength of the damaged concrete struts. This coefficient is estimated based on the amount of
shear stress that can be transferred through the initial crack.

114
 4.2 Mechanical model for compression softening

(a) (b) (c) 1.5

θ
σc1
Initial crack σc1 σc3 τcr 1.2
σc3 σc1 θcr-θ τcr,max
τcr 0.9
θcr dl τcr,45°
Free-body 0.6 τcr,45°
σc3 σn,cr
θ 0.3
θcr τcr,MIN
0
0 15 30 45 60 75 90
θcr- θ [°]

Figure 4.21: (a) Stress state of a damaged strut and the inclination of an initial crack; (b) stresses
acting along the sides of the analysed free-body and (c) shear stresses acting along the
crack surface in function of the stress field rotation angle
By investigating a free-body indicated in Fig.4.21a and Fig.4.21b, one can find
equilibrium of the forces acting along each side of the free-body in the direction of the initial
crack (θcr) as follows:
σ c 3  sin  cr     cos  cr     σ c 1  cos  cr     sin  cr     τ cr (4.86)

Expressing the principal concrete compressive strength from the previous equation
gives:
2  τ cr
σ c3   σ c1 (4.87)
sin2   cr   

Since the concrete strength in this free-body is limited by the amount of shear stress that
can be transferred through a crack and the brittleness of the concrete, this means that:
σ c3  f c  fc  (4.88)

The effectiveness factor that accounts for the presence of the initial cracks in concrete
(in-plane-failure) is estimated as:
2  τ cr σ c1
ητ   1 (4.89)
f c  η fc  sin 2   cr    f c  η fc

where τcr represents the necessary amount of shear stress that can be transferred through a crack;
σc1 represent the average tensile stress in a concrete strut.
Average tensile stress of a strut can be estimated based on the crack spacing, which is
assumed equal to 1.5 times the bond length [74]:
6
σ c1   f ctm  0.38  f ctm (4.90)
16
When the contact forces develop at 45° from the crack surface, this value can be
evaluated on the basis of the proposal by Vecchio and Collins [120] (refer to Fig.4.21c):

115
Chapter 4: Effectiveness factor for concrete compressive strength

fc
τ cr , 45  (4.91)
24  wini
0.31 
Dmax  16
where wini represents the initial crack width;
Dmax represents the maximal aggregate diameter [mm] (Dmax = 0 for fc > 60 MPa).
The minimal shear stress which can be transferred through a crack corresponds to a case
at which the direction of the principal compressive stress is parallel to the face of a crack (see
Fig.4.21c). Based on the work of Randl [98] this value can be estimated as:
0.8 if f c  35MPa 
τ cr ,min  0.2  3 f c   
1.0 if f c  35MPa  (4.92)
 
  f y ,dow, x   sx   x  sin  cr    f y ,dow, z   sz   z  cos cr 

where ρx represents the horizontal reinforcement ratio;

ρz represents the vertical reinforcement ratio;
σsx represents the steel stresses in x-direction;
fy,dow,x represents the doweled steel strength x-direction;
σsz represents the steel stresses in z-direction;
fy,dow,z represents the doweled steel strength z-direction.
For other values of the angle (θcr-θ) the value can be calculated as (refer to Fig.4.21c):

τ cr τ cr ,min 2 sin cr     τ cr , 45 τ cr ,min  (4.93)

This assumption is based on the fact that the amount of shear that can be transferred
through a crack increases quite fast as the stress field starts to rotate [120], and then stabilizes
as the θcr and θ become close to perpendicular. After analysis of 77 reinforced concrete panels,
the proposed law showed a good agreement with the test results, although the authors
acknowledge that future work is required to verify its general consistency.
The last remaining effectiveness factor from Eq.4.81 is the ηdow, which takes into
account the presence of the confining concrete struts. As it was previously mentioned, the
enhanced concrete compressive strength along the region influenced by the doweling is already
used to equilibrate the shear force of the rebar (see Fig.4.17c). This means that this region of
the concrete cannot be included in carrying any compressive stresses in the principal direction.
This effect can be taken into account in the following manner:
f cc
 dow   1 .0 (4.94)
fc

It is important to emphasize that even in case where the yield strength of the
reinforcement is not reduced due to the doweling (point B from Fig.4.19), the rebars still move

116
 4.2 Mechanical model for compression softening

towards the concrete inducing the formation of the cracks and activating the confinement effects
of the concrete, meaning that Eq.4.94 needs to be applied in these cases as well.

4.2.3.2 Strength of the undamaged struts

The compressive strength of the undamaged struts is limited by the parameter ηUDS,
which takes into account:
1. The presence of the rebars in the concrete struts, governing the spalling failures (as
defined in Eq.4.82) and limiting the in-plane strength of the concrete (adopted from
Hars [41]);
2. The presence of the doweled reinforcement (as defined in Eq.4.94).
The parameter itself is proposed to be calculated as:

ηs
  0   bw 

ηUDS  qUDS ,dow  min max  min 
 bw  4c  2ο ext   4c  2ο ext   1   δ '
 bw bw
dow

(4.95)
ηs
  0   bw 

 1  qUDS ,dow  min max  min 
 bw  4c  2ο ext   4c  2ο ext   1  1  sin    δ '
 bw bw
s

φs represents the angle of the friction between the rebar surface and the surrounding concrete;
qUDS,dow represents the relative amount of the undamaged struts affected by the doweling.
As it can be seen the only difference between the Eq.4.95 and the Eq.4.81 is the fact that
the factor ητ is now equal to one (since there are no cracks crossing the undamaged struts), and
that unlike the damaged struts, the undamaged struts are only partly affected by the doweling.
This can be seen in Fig.4.20a, where the confining concrete region (marked in dashed lines)
crosses the undamaged struts only in their top-right and the bottom-left. The relative width of
concrete affected by the doweling (sdow) can be expressed in function of the total width of the
undamaged struts.
Based on the geometrical consideration presented in Fig.4.20d, this factor can be
estimated in the following manner:

cos  cr     l dow , x  sin 

 UDS , dow   max  (4.96)
s rm l dow , z  cos 
2 UDS , dow
q UDS , dow  (4.97)
1  q DS

117
Chapter 4: Effectiveness factor for concrete compressive strength

4.2.4 Parameters governing the compressive strength of concrete

As it can be seen, the mechanical model for the effective concrete compressive strength
is quite complex and depends on a significant number of parameters. However, the two
parameters with the largest significance in the results are the width of the crack at the point of
first yielding (wini), and the ultimate angle of the principal concrete compressive stresses.
Fig.4.22 gives a clear view on how these two parameters influence the concrete compressive
strength effectiveness factor.
Governed by ηs and ηdow

1.0
Governed by ηs , ηdow and ητ
0.8
fce 0.6
fc 0.4
Governed by ητ and ηdow
0.2
0 45
40
0.2 35
0.5 30 θ [°]
0.8
1.1 25
1.4 20
w [mm] 1.7
2.0

Figure 4.22: Representation of the failure criterion for the concrete compressive strength as a function
of the stress-field rotation and the crack opening at first yielding
As it can be seen, there are three different regions in the failure surface. The first
corresponds to the situation in which the rotation of the stress field is limited (see Fig.4.23a).
For these cases, the strength of the damaged and the undamaged struts is governed by the
presence of the spalling cracks (ηs) and the doweling action (ηdow). It can be seen that the width
of the initial cracks does not influence the failure criterion. The stress field rotations, on the
other hand, reduce the concrete compressive strength, since the amount of the undamaged struts
which are affected by the dowel action of the reinforcement increases.
(a) θ ≈ θcr (b) θ < θcr (c) θ < θcr
Sliding Sliding

Spalling Spalling
Crushing Crushing

θcr θcr
θ θcr
θ
θ

Figure 4.23: Potential failure mechanism of structural panels: (a) spalling of the concrete cover in
case of small stress field rotation; (b) spalling and combined crushing with sliding failure
in case of moderate stress field rotation and (c) crushing and sliding of the concrete struts
in case of significant stress field rotations
As the stress field continues to rotate, the strength of the damaged struts becomes
governed by the presence of in-plane cracks (ητ), while the strength of the undamaged struts
remains dependent on the out-of-plane cracks (ηs), as presented in Fig.4.23b. This corresponds

118
 4.2 Mechanical model for compression softening

to the second region of the failure criterion, which is depended on the initial crack width (the
wider the cracks, the larger the reduction of the concrete compressive strength). The physical
failure of the concrete struts in this region can occur due to the spalling or the crushing of the
material.
As the stress field increases still its rotation, the amount of the undamaged struts in the
panel reduces, and eventually becomes zero. This leads to the third stage of the failure criterion,
when the strength of the entire panel is dependent only on the characteristics of the in-plane
cracks (ητ) (see Fig.4.23c). Once more, the influence of the initial crack width is quite
pronounced, and the failure mechanism of the concrete struts corresponds to crushing.
Doweling of the reinforcement affects the concrete strength of the second as well as the third
region of the failure surface. It has to be noted that when failure occurs by yielding of the
reinforcement in both directions, the principal concrete compressive stresses does not reach fce.

119
Chapter 4: Effectiveness factor for concrete compressive strength

4.3 Numerical solving procedure for the proposed compression softening model
In order to estimate the stress and the strain state of a panel, it is first necessary to define
the points that correspond to the limit cases of the three stages of behaviour (the elastic
uncracked phase, the onset of yielding and the failure). Once these points are known, full stress-
strain curves can be produced by assuming that the stress field rotates linearly in between.

4.3.1 Cracking of the panel

First cracking can be directly calculated on the basis of Eq.4.38 to Eq.4.43.

4.3.2 Onset of yielding or early concrete crushing

Finding the second characterising point in the panels load history requires the use of an
iterative procedure (as presented in Fig.4.24a). There are three potential cases which can
determine the behaviour of a panel:
1. Case 1 (C1): which results in yielding of the reinforcement in x-direction;
2. Case 2 (C2): which results in yielding of the reinforcement in z-direction;
3. Case 3 (C3): which results in concrete crushing prior to yielding of the reinforcement.
After assuming the direction of the principal concrete compressive stresses (θσ) and the
final width of the initial cracks (wass), it is possible to estimate the dowel strength of the
reinforcement.
The average cracks spacing is assumed equal to be 1.5 times the bond length [74], which
means that the average concrete tensile stress within the cracks is equal to:
3
 c1   b (4.98)
16
The principal concrete compressive stresses inclination may be calculated by using a Mohr’s
circle (shown in Fig.4.15b):
τ xz
tan θ  (4.99)
σ c1  σ cx
Assuming that the x reinforcement yields the first (σsx=fy,dow,x), equilibrium conditions
give:
σ cx  ρx  f y ,dow, x  vx (4.100)

σ cz  ρz  σ sz  vz (4.101)

τ xz  vxz (4.102)

Eq.4.98 to Eq.4.102 present a system of 5 equation with 5 unknowns, which can be

solved for τxz, which can then be used to determine all required stresses in concrete and steel.

120
 4.3 Numerical solving procedure

(a) START (b) START

C1
Assume C2 Assume C3 Assume Assume Assume Assume
θσ ;wass θσ ;wass θσ ;wass θσ= θε θ σ= θ ε θ σ= θ ε

fy,dow,x fy,dow,x Assume fy,dow,x fy,dow,x fy,dow,x

fy,dow,z fy,dow,z σc3 = fce fy,dow,z fy,dow,z fy,dow,z

σc1 σc1 C2 σc1

Assume Assume fy,dow,x=min(C1;C2 )
σsx = fy,dow,x σsz = fy,dow,z fy,dow,z=min(C1;C2 ) C1
σsz ;σcx ;σcz σsx ;σcx ;σcz
σc1 σc1 σc1 τ ;σc3= σ3 σsz ;σcx ;σcz σsx ;σcx ;σcz τ ;σc3= σ3
τ ;σc3= σ3 τ ;σc3= σ3
σsz ;σcx ;σcz σsx ;σcx ;σcz τ εz εx
τ ;σc3= σ3 τ ;σc3= σ3 ε3 ;ε1 εz εx ε3 ;ε1
σsx ;σsz ε3 ;ε1 ε3 ;ε1
εx ;εz εx ;εz σcx ;σcz w w
ε3 ;ε1 ε3 ;ε1 w w
εx ;εz
θε ;w θε ;w ε3 ;ε1
NO
σc3 = fce σc3 = fce NO
NO NO
σsz = fsz,dow σsx = fsx,dow
NO YES YES
θ σ= θ ε θ σ= θ ε θε ;w
NO NO YES YES
wass= w wass= w
σsz ≤ fy,dow,z σsx ≤ fy,dow,x
YES YES σsz ≤ fy,dow,z NO
σc3 ≤ fce σc3 ≤ fce NO

YES YES

NO NO YES YES YES

σsz ≤ fy,dow,z σsx ≤ fy,dow,x END END
END END
YES YES σsx ≤ fy,dow,x Concrete Concrete
crushing Steel yielding Steel yielding crushing
YES in both directions in both directions
NO NO
σc3 ≤ fce σc3 ≤ fce
END
YES YES
Concrete
END END crushing

Steel yielding Steel yielding

in x-direction in z-direction

Figure 4.24: (a) Solving procedure at the onset of yielding; (b) solving procedure at failure
Applied stress rate in each direction is usually known, meaning that νx and νz can be
expressed using νxz. The results are checked determining the strain state of the panel and making
sure that the assumed principal stress angle (θσ) is equal to the principal the stain angle (θε) and
by confirming that the assumed initial crack width (wass) is equal to the calculated one (w).
(a) 1.2 (b) γ/2

1.0 γxz/2 εx Pole

0.8
σc ε3 θ ε1
Undamaged Damaged ε
0.6 struts
fc struts

0.4

0.2 γxz/2 εz

0
0 0.5 1.0 1.5 2.0
εc /εco = ε3 /εco

Figure 4.25: (a) Stress-strain parabola for concrete compressive strength; (b) average strain state of
the panel
The average strains of the panel in the x-direction and z-direction are equal to:

121
Chapter 4: Effectiveness factor for concrete compressive strength

f yx ,dow τ b  srmx
εx   (4.103)
Es  x  Es

 sz τ b  srmz
εz   (4.104)
Es  z  Es
srmx represents the average crack spacing in x-direction;
srmz represents the average crack spacing in z-direction.
The principal concrete compressive strains can be determined by using the uniaxial
stress-strain parabola (refer to Fig.4.25), assuming that the undamaged struts follow its
ascending branch and that the damaged struts follow its descending branch:
ε3  ε3c,DS  qDS  ε3c,UDS  1  qDS  (4.105)


ε3c, DS  ε0  1  1   c3 f c  (4.106)


ε3c,UDS  ε0  1  1   c3 f c  (4.107)

ε0  2  f c Ec  0.002 (4.108)

Eq.4.103, Eq.4.104 and Eq.4.105 define 3 points in a Mohr’s circle, which means that
the Fig.4.25b can be established, and the principal compressive stain direction can be
calculated:
 x 3
tan θ  (4.109)
1   x
Finally, it is possible to determine the initial crack opening:
w ini  s rm  ε1   c 1 E c  (4.110)

As indicated in Fig.4.24a, it is now necessary to verify that the assumed stress state is
respected, or try another case if the answer is not complying with this condition.
Case C2 is almost identical to C1, whereas C3 is a bit simpler. The only difference
between the solving procedure C1 and C2 is in the fact that the vertical instead of the horizontal
reinforcement is assumed to have reached yielding by the end of the second phase of the panel's
load-history. Apart from this, all the necessary validations remain identical (refer to Fig.4.24a).
The doweled steel strength of x-reinforcement according to case C1 is lower than that of
case C3, since the angle θ is higher for case C1 than for C3. In the same way, the doweled steel
strength of z-reinforcement according to case C2 is lower than that of case C3. These two values
for both doweled steel strengths (of C1 and C2) are adopted for case C3, which is a slightly
conservative simplification.

122
 4.3 Numerical solving procedure

4.3.3 Failure of the panel

In order to determine the third (and final) characteristic point of the panel’s load history,
another iterative procedure needs to be applied. Looking at the scheme from Fig.4.24b, it can
be seen that depending on the determining case from Fig.4.24a, it is now possible to estimate
the average strain in one direction (εx in case of C2 or εz in case of C1).
The direction of the principal compressive stresses is assumed to be parallel to the
direction of the principal compressive strains. The secondary doweling should only be applied
on the reinforcement which did not yield during the previous phase (x-direction in case of C2
and z-direction in case C1). The additional doweling should only be applied on the remaining
steel stresses in the reinforcement that yields second, from the onset of the yielding of the
direction yielded first reinforcement.
The stress field rotates until the failure of the concrete or steel is reached. The stress and
strain transformation which were applied at the onset of yielding should be applied during this
phase as well. In the end of the solving procedure it is verified that the ultimate stress direction
is lower than the one at the onset of yielding.

123
Chapter 4: Effectiveness factor for concrete compressive strength

4.4 Experimental validation and comparison to available methods

In order to validate the assumptions of the proposed model, a database comprising 77
structural panels has been collected and presented in Tab. 4.1. It contains results of 12 separate
experimental campaigns performed by various authors (the corresponding references as well as
the basic geometrical and mechanical properties are provided in Tab. 4.1).

Table 4.1: Database of the reinforced concrete panels failing in shear

bw Øx Øz fc fyx fyz
N° Ref. Name ρx [%] ρz [%]
[mm] [mm] [mm] [MPa] [MPa] [MPa
1 PV4 70 3.43 3.43 1.06 1.06 27 242 242
2 PV6 70 6.31 6.31 1.79 1.79 30 266 266
3 PV10 70 6.31 4.72 1.79 1.00 15 276 276
4 PV11 70 6.31 5.39 1.79 1.31 16 235 235
5 PV12 70 6.31 3.15 1.79 0.45 16 469 269
6 PV16 70 4.06 4.06 0.74 0.74 22 255 255
7 PV19 70 6.31 3.99 1.79 0.71 19 458 299
[119]
8 PV20 70 6.31 4.44 1.79 0.89 20 460 297
9 PV21 70 6.31 5.37 1.79 1.30 20 458 302
10 PV22 70 6.31 5.83 1.79 1.52 20 458 420
11 PV23 70 6.31 6.31 1.79 1.79 21 518 518
12 PV25 70 6.31 6.31 1.79 1.79 19 466 466
13 PV27 70 6.31 6.31 1.79 1.79 21 442 442
14 PV28 70 6.31 6.31 1.79 1.79 19 483 483
15 PHS2 70 8.05 5.72 3.23 0.41 66 606 521
16 PHS3 70 8.05 5.72 3.23 0.82 58 606 521
17 PHS5 70 8.05 5.72 3.23 0.41 52 606 521
18 [121] PHS6 70 8.05 5.72 3.23 0.41 50 606 521
19 PHS8 70 8.05 5.72 3.23 1.22 56 606 521
20 PHS9 70 8.05 5.72 3.23 0.41 56 606 521
21 PHS10 70 8.05 5.72 3.23 1.22 51 606 521
22 [75] PP1 287 19.54 11.28 1.94 0.65 27 479 480
23 PC1A 70 5.72 5.72 1.65 0.82 28 500 500
[118]
24 PC4 70 5.72 5.72 1.65 0.82 25 260 260
25 A2 178 16.00 16.00 1.20 1.20 41 462 462
26 A3 178 19.50 19.50 1.78 1.78 42 446 446
27 A4 178 25.20 25.20 2.97 2.97 42 469 469
28 B1 178 16.00 11.30 1.20 0.60 45 462 444
29 [94] B2 178 19.50 16.00 1.78 1.20 44 446 462
30 B3 178 19.50 11.30 1.78 0.60 45 446 444
31 B4 178 25.20 11.30 2.97 0.60 45 469 444
32 B5 178 25.20 16.00 2.97 1.20 43 469 462
33 B6 178 25.20 19.50 2.97 1.78 43 469 446
34 VA1 178 11.30 11.30 1.20 1.20 95 445 445
35 VA2 178 16.00 16.00 2.40 2.40 98 409 409
36 VA3 178 19.50 19.50 3.57 3.57 95 455 455
[127]
37 VA4 203 25.20 25.20 5.23 5.23 103 470 470
38 VB1 178 16.00 11.30 2.40 1.20 98 409 445
39 VB2 178 19.50 11.30 3.57 1.20 98 455 445

124
 4.4 Experimental validation

40 VB3 178 25.20 11.30 5.96 1.20 102 470 445

41 VB4 178 19.50 11.30 1.78 0.60 97 455 445
42 HB1 178 16.00 11.30 1.20 0.60 67 409 445
43 [47] HB3 178 19.50 11.30 1.78 0.60 67 447 445
44 HB4 178 25.20 11.30 2.98 0.60 63 470 445
45 SE1 285 19.50 11.30 2.91 0.98 43 492 479
[56]
46 SE6 285 19.50 11.30 2.91 0.33 40 492 479
47 00R 60 6.00 6.00 0.86 0.86 28 310 310
48 15R 60 6.00 6.00 0.86 0.86 28 310 310
49 30R 60 6.00 6.00 0.86 0.86 28 310 310
50 45R 60 6.00 6.00 0.86 0.86 28 310 310
51 00D 60 7.60 7.60 0.87 0.87 28 318 318
52 30D 60 7.60 7.60 0.86 0.87 28 318 318
53 45D 60 7.60 7.60 0.87 0.87 28 318 318
54 [124] 00DI 60 7.00 7.00 1.39 1.39 31 294 294
55 22.5DI 60 7.00 7.00 1.39 1.39 31 294 294
56 45DI 60 7.60 7.60 1.30 1.30 31 318 318
57 45DII 60 7.60 7.60 2.61 2.61 31 318 318
58 45PCI 60 5.44 5.44 0.77 0.77 30 1187 1187
59 45PCII 60 5.44 5.44 1.55 1.55 30 1187 1187
60 45PCIII 60 5.44 5.44 1.55 0.77 30 1187 1187
61 45PCIV 60 5.44 5.44 1.55 0.77 45 1187 1187
62 PK02 70 6.50 6.50 1.07 1.07 19 660 660
63 PK04 70 6.50 6.50 1.07 1.07 20 660 660
64 PK07 70 6.50 6.50 1.07 1.07 21 660 660
65 EGE6F1 100 6.50 6.50 0.66 0.66 16 465 465
66 EGE6F2 100 6.50 6.50 0.66 0.66 16 465 465
67 EGE6F3 100 6.50 6.50 0.66 0.66 15 465 465
[57]
68 EGE6F4 100 6.50 6.50 0.66 0.66 17 465 465
69 EGE6F7 100 6.50 6.50 0.66 0.66 19 465 465
70 EGE6F8 100 6.50 6.50 0.66 0.66 13 465 465
71 EGE7F1 100 6.50 6.50 0.66 0.66 16 660 660
72 EGE7F2 100 6.50 6.50 0.66 0.66 15 660 660
73 EGE7F3 100 6.50 6.50 0.66 0.66 17 660 660
74 #2 100 10.00 10.00 3.14 3.14 26 582 582
[106]
75 #6 100 10.00 10.00 1.57 1.57 26 582 582
76 KP1 140 12.70 12.70 2.03 1.02 25 430 430
[5]
77 TP4A 70 6.35 6.35 2.03 2.03 25 450 450
The selected elements vary significantly in concrete compressive strength (from 13 MPa
to 103 MPa), reinforcement ratios (from 0.33 % to 5.96 %), steel yield strength (from 235 MPa
to 1187 MPa) as well as concerning the ratio of the reinforcement placed in two perpendicular
directions (from 0.11 to 1.00). The size of the specimens varied from 500 mm to 2510 mm,
while the thickness varied from 60 mm to 287 mm. With respect to the loading conditions, most
of the panels have been subjected to pure shear (55 out of 77 specimens), some of them
combined shear with compression (19 out of 77 specimens), and a few were loaded to shear
and bi-axial traction (3 out of 77 panels). The panels experienced different failure modes
(concrete spalling, crushing or sliding) and while some failed due to weakening of the concrete,
others yielded reinforcement in both directions.

125
Chapter 4: Effectiveness factor for concrete compressive strength

In order to assess the accuracy of the presented approach and to compare it to some of
the existing strain-based approaches, the database from Tab. 4.1 has been assessed using the
Elastic-Plastic Stress Field Method (EPSF) [32] implementing the Vecchio and Collins
softening equation [120]. Tab.4.2 summarizes the results of both analyses, and gives the
ultimate strength assessment for each approach. It can be seen that both approaches provide
satisfactory accuracy. However, the proposed mechanical procedure shows some higher level
of accuracy (1.01 instead of 1.11) and lower scatter of the results (0.12 instead of 0.15)
compared to EPSF approach. Additionally, it provides with information on the governing
failure mode, which matched for the 77 specimens to the observed one.

Table 4.2: Results of the EPSF analysis and the mechanical model proposed in this paper
N° Ref. Name τTEST [MPa] τEPSF [kN] τMODEL [kN] τTEST/τEPSF τTEST/τMODEL
1 PV4 2.84 2.56 2.62 1.11 1.08
2 PV6 4.47 4.76 4.81 0.94 0.93
3 PV10 3.97 3.69 3.73 1.08 1.06
4 PV11 3.56 3.59 3.63 0.99 0.98
5 PV12 3.13 2.55 3.05 1.23 1.03
6 PV16 2.14 1.89 1.95 1.13 1.10
7 PV19 3.96 3.73 4.01 1.06 0.99
[119]
8 PV20 4.26 4.22 4.46 1.01 0.96
9 PV21 5.03 5.18 5.46 0.97 0.92
10 PV22 6.07 6.39 7.11 0.95 0.85
11 PV23 8.88 7.69 8.09 1.15 1.10
12 PV25 9.13 8.39 7.58 1.09 1.20
13 PV27 6.35 6.74 7.94 0.94 0.80
14 PV28 5.61 5.89 6.56 0.95 0.86
15 PHS2 6.66 5.18 5.21 1.29 1.28
16 PHS3 8.10 7.67 7.92 1.06 1.02
17 PHS5 4.81 3.54 3.96 1.36 1.21
18 [121] PHS6 7.62 7.30 7.80 1.04 0.98
19 PHS8 10.84 9.53 9.91 1.14 1.09
20 PHS9 9.16 7.52 7.87 1.22 1.16
21 PHS10 8.25 7.57 7.82 1.09 1.05
22 [75] PP1 4.95 5.11 4.79 0.97 1.03
23 PC1A 5.61 5.84 6.05 0.96 0.93
[118]
24 PC4 4.84 5.10 5.75 0.95 0.84
25 A2 5.37 5.54 6.13 0.97 0.88
26 A3 7.65 7.94 8.49 0.96 0.90
27 A4 11.31 11.91 14.08 0.95 0.80
28 B1 3.96 3.82 4.54 1.04 0.87
29 [94] B2 6.13 6.63 7.25 0.92 0.85
30 B3 4.35 4.59 4.89 0.95 0.89
31 B4 5.06 5.50 5.60 0.92 0.90
32 B5 7.15 8.27 8.03 0.86 0.89
33 B6 9.14 9.96 10.11 0.92 0.90
34 VA1 6.16 5.34 6.19 1.15 0.99
35 VA2 9.73 9.84 10.62 0.99 0.92
[127]
36 VA3 15.08 16.25 16.91 0.93 0.89
37 VA4 21.42 21.05 21.36 1.02 1.00

126
 4.4 Experimental validation

38 VB1 7.50 7.25 7.82 1.03 0.96

39 VB2 9.14 9.32 8.94 0.98 1.02
40 VB3 9.71 10.55 10.2 0.92 0.95
41 VB4 4.86 4.66 6.28 1.04 0.77
42 HB1 4.32 3.63 4.41 1.19 0.98
43 [47] HB3 4.89 4.62 5.06 1.06 0.97
44 HB4 5.33 5.84 5.85 0.91 0.91
45 SE1 6.77 7.52 7.32 0.90 0.92
[56]
46 SE6 3.76 3.92 3.96 0.96 0.95
47 00R 3.14 2.66 2.72 1.18 1.15
48 15R 3.15 2.66 2.72 1.18 1.16
49 30R 3.13 2.66 2.72 1.18 1.15
50 45R 3.42 2.66 2.72 1.29 1.26
51 00D 2.97 2.70 2.83 1.10 1.05
52 30D 2.56 2.70 2.82 0.95 0.91
53 45D 2.84 2.70 2.83 1.05 1.00
54 [124] 00DI 4.96 4.10 4.17 1.21 1.19
55 22.5DI 5.06 4.10 4.17 1.23 1.21
56 45DI 3.97 4.15 4.21 0.96 0.94
57 45DII 7.61 8.29 8.35 0.92 0.91
58 45PCI 7.78 6.55 9.26 1.19 0.84
59 45PCII 11.72 8.39 11.46 1.40 1.02
60 45PCIII 9.44 7.35 9.51 1.29 0.99
61 45PCIV 10.63 8.65 11.12 1.23 0.96
62 PK02 9.12 7.4 8.39 1.23 1.09
63 PK04 8.91 6.64 8.75 1.34 1.02
64 PK07 9.04 6.79 9.06 1.33 1.00
65 EGE6F1 8.00 6.01 7.28 1.33 1.10
66 EGE6F2 8.1 5.68 7.37 1.43 1.10
67 EGE6F3 6.90 5.17 6.72 1.34 1.03
[57]
68 EGE6F4 7.20 5.71 7.69 1.26 0.94
69 EGE6F7 8.50 7.32 8.67 1.16 0.98
70 EGE6F8 7.35 5.23 6.03 1.40 1.22
71 EGE7F1 8.20 5.89 7.28 1.39 1.13
72 EGE7F2 7.70 5.32 7.00 1.45 1.10
73 EGE7F3 8.70 6.47 7.93 1.35 1.10
74 #2 11.78 10.94 9.94 1.08 1.18
[106]
75 #6 14.07 9.3 10.93 1.51 1.29
76 KP1 5.62 6.08 5.79 0.93 0.97
[5]
77 TP4A 8.72 8.00 9.2 1.09 0.95
Aver : 1.11 1.01
COV : 0.15 0.12
The ratio between the measured and the estimated strength is also presented in Fig.4.26
as a function of four basic parameters (concrete compressive strength-fc, steel strength in z-
direction (which was the weaker direction) (fyz), reinforcement ratio in x-direction (ρx) and
reinforcement ratio in z-direction (ρz). Each of the plots contains a red line which gives an
average value of the 5 nearest τTEST/τMODEL points (with respect to a given parameter), along the
entire database. As it can be seen, the proposed mechanical approach gives consistent results

127
Chapter 4: Effectiveness factor for concrete compressive strength

over the investigated domain, and shows no clear trends with respect to the four parameters,
which is satisfactory.
(a) (b)
2.0

1.5

τTEST
τMODEL 1.0

0.5

0
10 35 60 85 110 200 450 700 950 1200
fc [MPa] fy [MPa]
(c) 2.0 (d)

1.5

τTEST
1.0
τMODEL

0.5

0
0 1.5 3.0 4.5 6.0 0 1.5 3.0 4.5 6.0
ρx [%] ρz [%]

Figure 4.26: Shear strength prediction of the database as a function of: (a) concrete compressive
strength; (b) steel yield strength; (c) horizontal reinforcement ratio; (d) vertical
reinforcement ratio
Fig.4.27 compares the accuracy of the presented approach to EPSF [32] with respect to
the stress field rotation.
(a) (b)
2.0
Proposed approach Vecchio and Collins ‘86

1.5

τTEST
1.0
τMODEL

0.5

0
0 7.5 15 22.5 30 0 7.5 15 22.5 30
θcr- θ [°] θcr- θ [°]

Figure 4.27: (a) Shear strength prediction of the database using the mechanical model presented in
this paper; (b) using the softening equation of Vecchio and Collins [120] as a function
of stress field rotation angle
As it can be seen the mechanically based model shows better accuracy for panel with
small rotations when compared to the Vecchio and Collins softening equation [120] approach,

128
 4.4 Experimental validation

but most importantly it shows greater precision for panels with large rotations, which is relevant
for analysis of prestressed concrete beams. However, it should be noted that the number of
panels which experienced significant stress field rotations is not sufficiently large, and more
specimens are required in order to consolidate these results.
In order to compare the proposed mechanical procedure to existing strain-based
approaches for estimating the effective concrete compressive strength, the ratio between the
effective and cylinder concrete compressive strength (fce/fc) of each panel has been presented in
function of its average principal tensile stain (ε1) in Fig.4.28. The red line in the graph
corresponds to the concrete compressive strength reduction factor according to Vecchio and
Collins [120].
1.2

1.0

0.8
fce
0.6
fc
0.4 1
0.8 + 170·ε1
0.2

0
0 5 10 15 20
ε1 / 0.002

Figure 4.28: Calculated effective concrete compressive strength in function of normalized principal
tensile strains
It can be seen that the results of the mechanical approach are consistent with the semi-
empirical equation established within the Modified Compression Field Theory [120]. Panels
with smaller average tensile strains experience a smaller reduction in strength, while vice-versa.
However, unlike the MCFT equations, the mechanical efficiency factor can never be equal to
one. Even in panels with relatively low strength concrete (fc < 30 MPa), the presence of the
reinforcement will always induce in-plane cracks, and therefore reduce its effective
compressive strength. This theoretical result is in agreement to other semi-empirical formulas
of Hsu [47] or Kaufmann [52].

129
Chapter 5: Conclusions and future research
Chapter 2 focused on the differences between the strut-and-tie models and stress fields
suitable for the design and assessment of structural concrete members. Elastic-plastic stress
fields were used for this purpose, and the obtained results were compared to those acquired
using rigid-plastic stress fields. Accuracy and generality of the approach was finally validated,
by comparing the estimated failure loads to experimentally measured values from an online
database (which can be found at: http://i-concrete.epfl.ch/epsf/epsf.html).
In addition to this, advanced modelling techniques for EPSF, which can be applied when
analysing structural concrete members with complex loading and geometry was presented in
Chapter 3. Special focus was directed towards numerical stability of the approach by
performing detailed finite element sensitivity analysis. Cases where structural concrete
elements had insufficient anchorage length were investigated, as well as members composed of
crossed structural concrete beams (which is particularly interesting when analysing bridge
diaphragms for example). In order to profit from increased accuracy of the EPSF method
(compared to current code provisions), a procedure for tailoring the partial safety factors was
presented.
In order to justify the accuracy and generality of the concrete compression strength
softening equation implemented in EPSF method (originally proposed by Vecchio and Collins
in 1986 [120]), a mechanically based model that takes into account different failure modes of
concrete (crushing, sliding and spalling) was developed and validated in Chapter 4.
Final conclusions resulting from the work presented in this thesis are summarized in
following chapter. In addition to this, some recommendations for future work in the field of
limit analysis, structural reliability and concrete compressive strength effectiveness factors are
provided.

5.1 Conclusions
The main conclusions resulting from the research presented in this thesis, are
summarized in the following section. They are divided into three groups, each of them referring
to the work described in previous chapters.

5.1.1 Regarding the application of STM and stress fields as tools for design and assessment
This chapter focused on comparing suitable approaches for the design and assessment
of structural concrete members using stress fields and strut-and-tie models. Its denouements are
synthesised as following.

131
Chapter 5: Conclusions and future research

1. Design and assessment of structural concrete elements can be based on limit analysis,
providing that its main hypotheses are respected (having smeared cracks in the
elements and sufficient anchorage of the reinforcement). Nevertheless, the two
processes should not necessarily be performed following the same strategy.
2. Simple load-carrying models, which are in equilibrium with the external loads, provide
sufficient element strength at ULS and ensure a suitable behaviour at SLS are suitable
for design of structural concrete members. Solutions leading to simple reinforcement
layouts that can be easily inspected and put in place are preferred.
3. Strength assessment of existing structural concrete members should be performed in
steps, beginning with simple solutions which are gradually refined until they provide
sufficient element strength compared to imposed requirements. According to the limit
analysis this refinement can be performed until an exact solution has been reached,
which provides the highest possible strength of all the safe estimates (lower-bound
solutions). In case imposed strength requirements are still not met the structure needs
to be reinforced. However, the quantity of required interventions is minimized.
Serviceability issues of such elements can be directly checked in-situ.
4. Stress fields and strut-and-tie models are complementary approaches based on the
lower-bound theorem of limit analysis. While strut-and-tie models are simpler to
develop for an entire element and particularly useful to determine the amount of
required reinforcement, stress fields are most suited for detailing. They are useful to
verify the behaviour of concrete in critical regions (if the struts respect the boundary
conditions of the element) and indicate nodes which require additional considerations
when it comes to anchorage.
5. Exact solutions according to the theory of plasticity can be developed by hand using
an iterative procedure, starting from a failure mechanism. In case a corresponding
rigid-plastic stress field can be developed inside such a mechanism (respecting the
boundary conditions) the solution can be accepted. Otherwise, the failure mechanism
needs to be adapted accordingly.
6. Elastic-plastic stress field method is suitable for design and assessment of structural
concrete members. The method respects the equilibrium and yield conditions (lower-
bound theorem), which are derived from an imposed deformation field. At ULS, this
deformation field becomes a licit failure mechanism, meaning that the upper-bound
theorem of the limit analysis is also respected. Consequently, the EPSF can be used to
obtain exact solutions according to limit analysis in an automated manner.
7. The EPSF automatically accounts for all potential load-carrying mechanism,
particularly the ones resulting from the presence of crack control reinforcement. In
addition to this, it allows a refined estimation of the effective concrete compressive
strength by imposing a softening equation of Vecchio and Collins [120].
8. The exact solutions (according to limit analysis) obtained using the EPSF method show
satisfactory behaviour when compared to a database containing 315 elements

132
 5.1 Conclusions

(available online at http://i-concrete.epfl.ch/epsf/epsf.html). Wide number of different

geometries has been analysed (dapped-end beams, deep beams, prestressed girders,
corner frames, wall, crossed beams and walls), failing in various types of mechanisms
(shear, bending, local failures). The method provided accurate estimates of their
strength and showed low scatter of the results, thus confirming the consistency and
generality of the approach.

5.1.2 Regarding the advanced modeling of structural concrete with EPSF method
This chapter focused on modelling aspect of elastic-plastic stress field method by
analysing its mesh related problems, simulation of particular structural details and application
of suitable safety format. Its denouements are synthesised as following.
1. The EPSF method gives stable results for various size of applied finite elements. A
general recommendation when selecting an optimal FE size is to use the stirrup spacing
as the main grid for meshing, and place two concrete FE triangles in between.
2. The EPSF method gives stable results for very distorted FE triangles (a:h ration equal
to 1:5). A general recommendation is to keep this ratio up to 1:3.
3. Orientation of the FE triangles on the other hand has a significant influence of the
results. Hence, using the meshes which favour certain FE orientation should be
avoided and FE with random or zig-zag inclination of hypotenuses should be selected.
4. Number of iterations until the convergence of the EPSF model at ULS is 250 steps.
However, when estimating the strength of a structural concrete elements, it is advised
to use less iterations in the beginning in order to reduce the time required for a single
simulation. The results obtained with 100 iterations are already quite close to final
solution.
5. Structural concrete members with insufficient anchorage length can be analysed using
EPSF method with satisfying accuracy. The diameter of a rebar which does not have
sufficient anchorage length should be reduced in order to limit the amount of tensile
force which can be taken by a single bar FE along the required anchorage length. For
this purpose, plastic bond law introduced by Tension Chord Model [74] proved to give
satisfying results.
6. Structural concrete elements which are connected and placed in two perpendicular
planes can be accurately analysed using the EPSF method, by linking the nodes of the
crossing areas with rigid FE bars in order to impose equal deformations between the
crossing elements. Providing that the basic meshing recommendations are respected,
this procedure gives satisfying results when compared to test results.
7. The Partial Safety Factors can be tailored in order to combine them with EPSF method.
Depending on the applied concrete compressive strength and type of failure (shear of
flexural), PSF can assume higher or lower values. The method proved to be sensitive
to scatter in material properties, meaning that average material resistance and the
coefficient of variation need to be properly estimated for each case.

133
Chapter 5: Conclusions and future research

8. For concrete with high compressive strength (over 50 MPa), the partial safety format
becomes the global safety format, since both concrete and steel partial safety factors
become equal to 1.16.
9. In case of lower concrete compressive strength (around 20 MPa), the tailored partial
safety factors assume standard values (1.48 for concrete and 1.16 for steel).
10. In case accurate estimates of concrete compressive strength are not available, standard
partial safety factors for concrete and steel should be used (γc=1.50 and γs=1.15)

5.1.3 Regarding the effectiveness factor for concrete compressive strength

This chapter investigated the behaviour of panels subjected to in-plane forces and
proposed a model to describe their behaviour and potential failure modes. Its denouements are
synthesised as following.
1. The traditional description of the phenomenon of compression-softening due to the
transverse cracking, by means of semi-empirical equations based on transverse strains,
is simple to use but does not provide information on the actual failure mode or
parameters governing failure.
2. A consistent approach to this issue shall account for different stages of behaviour,
namely: uncracked, cracked state before yielding and cracked state after yielding with
the potential development of a secondary set of cracks.
3. Each stage of behaviour may be governed by different failure modes. These are out-of-
plane (spalling) failures, concrete crushing and crack sliding (after rotation of the
principal stress direction). These failure modes can be consistently calculated.
4. Doweling of the bars has an influence on the effective concrete compressive strength. It
allows carrying shear through the cracks but may originate spalling of the concrete
cover. In general, the various shear-carrying actions are thus not independent.
5. A mechanical approach based on the equilibrium and compatibility conditions can be
developed accounting for each stage of behaviour and its associated failure modes. A
possible manner to do so is proposed in this chapter.
6. The results of the mechanical approach show nice agreement to test results (better than
available semi-empirical formulas) and allows for a complete description of the load-
deformation response. This is relevant not only for accuracy, but to provide designers
with a clear physical understanding of the failure mechanisms and how to enhance the
behaviour of the member.

134
 5.2 Future research

5.2 Future research

The following section provides some suggestions for future researches, which can be
conducted with respect to the scope of research presented in this thesis. Once again, the
recommendations are divided into three groups, each of them resulting from the work shown in
Chapter 2, Chapter 3 and Chapter 4, respectively.
One of the principal ideas behind the work conducted in Chapter 2 of this thesis, was to
assemble an open source database containing experimental results of reinforced and prestressed
concrete elements found in literature, model them using EPSF method and place the database
online (at http://i-concrete.epfl.ch/epsf/epsf.html). This accommodates complete transparency
of the research, and provides examples for the correct application of the EPSF method. With
this in mind:
1. Structural concrete elements from additional experimental campaigns along with
corresponding EPSF models could be introduced into the existing database;
2. Accent could be placed on very old experimental investigations (from the beginning of
the XX century) in order to validate the applicability of the EPSF approach for such
cases (some work has already been done in this direction at IBETON);
3. Improve the solver of ICONC in order to reduce the required computational time, which
would allow to efficiently analyze large-scale members without heavy finite element
meshes.
Continuing the work from Chapter 3 of this thesis could involve investigating:
1. Validation of the results of EPSF analyses in cases of imposed displacements as a load
case;
2. Analyzing the cases with insufficient anchorage length using plane (smooth) rebars;
3. Further investigation of the effect of reliability index in the process of tailoring the PSF
for EPSF method;
4. Comparing the ultimate resistance of structural concrete members, when tailored global
and partial safety factors have been combined with EPSF method (some work has
already been done in this direction at IBETON).
Following the results presented in Chapter 4 of this thesis, further research is required
in order to:
1. Propose a simplified expression for the concrete compressive strength efficiency factor,
which could be easily implemented in finite element programs or even used in hand
calculations;
2. Investigate the accuracy of full and simplified expressions for effective concrete
compressive strength on reinforced and prestressed beams failing in shear (some work
has already been done in this direction at IBETON);

135
References
1) ACI 318-08 (2008): 'Building Code Requirements for Structural Concrete', ACI
Committee 318, 467 p., USA

2) ACI Committee 364.1 R-94 (1994): 'Guide for Evaluation of Concrete Structures Prior
to Rehabilitation', ACI, 22 p., USA

3) ACI Committee 437R-03 (2003): 'Strength Evaluation of Existing Concrete

Buildings', ACI, 28 p., USA

4) Ali M.A. and White R.N. (2001): 'Automatic Generation of Truss Model for Optimal
Design of Reinforced Concrete Structures', ACI Structural Journal, Vol. 98, N° 4,
pp. 431-442, USA

5) André H.M.O. (1987): 'Toronto Kajima Study on Scale Effects in Reinforced Concrete
Elements', University of Toronto, Department of Civil Engineering, PhD Thesis,
267 p., Toronto, Canada

6) Argirova G., Fernández Ruiz M. and Muttoni A. (2014): 'How simple can nonlinear
finite element modelling be for structural concrete?', Informes de la Construcción,
Vol. 66, 8 p., Spain

7) Argirova G. and Muttoni A. (2012): 'Verification of details of existing structures with

the elastic-plastic stress field method', 9th fib International PhD Symposium in Civil
Engineering, Vol. 1, pp. 153-158, Karlsruhe, Germany

8) Bach F., Nielsen M.P. and Braestrup M.W. (1980): 'Shear Tests on Reinforced
Concrete T-Beams - Series V, U, X, B and S', Structural Research Laboratory,
Technical University of Denmark, Report N° 120, 87 p., Copenhagen, Denmark

9) Baumann T. and Rüsch H. (1970): 'Schubversuche mit indirekter Krafteinleitung:

Versuche zum Studium der Verdübelungswirkung der Biegezugbewehrung eines
Stahlbetonbalkens', Deutscher Ausschuss für Stahlbeton, Heft 210, 83 p., Berlin,
Germany

10) Belarbi A. and Hsu C.-T.T. (1995): 'Constitutive laws of softened concrete in biaxial
tension-compression', ACI Structural Journal, Vol. 92, N° 5, pp. 562-573, USA

11) Bentz E.C., Vecchio F.J. and Collins M.P. (2006): 'Simplified Modified Compression
Field Theory for Calculating Shear Strength of Reinforced Concrete Elements', ACI
Structural Journal, Vol. 103, N° 4, pp. 614-624, USA

137
References

12) Biondini F., Bontempi F. and Malerba P. (1996): 'Ricerca di Modelli Strut-and-Tie
Mediante Programazione Lineare', Politecnico di Milano, Studi e Ricerche, Vol. 17,
pp. 121-156, Italy

13) Brena S.F. and Morrison M.C. (2007): 'Factors Affecting Strength of Elements
Designed Using Strut-and-Tie Models', ACI Structural Journal, Vol. 104, N° 3,
pp. 267-277, USA

14) Brown D.B. and Bayrak O. (2006): 'Minimum transverse reinforcement for bottle-
shaped struts', ACI Structural Journal, Vol.103, N° 6, pp. 813-821, USA

15) Büeler C. and Thoma K. (2010): 'Indirekt gelagerter Spannbetonträger', Lucerne

University of Applied Sciences and Arts, 61 p., Lucerne, Switzerland

16) Campana S. (2013): 'Éléments en béton armé soumis à une combinaison de flexion,
effort tranchant et forces de déviation', EPFL, IBETON, Thèse, N° 5574, 162 p.,
Lausanne, Switzerland

17) Campana S. and Muttoni A. (2011): 'Essais sur angles de cadre d'une tranchée couverte
à section polygonale (rapport de synthèse)', EPFL, IBETON, Test Report , 184 p.,
Lausanne, Switzerland

18) Campbell T.I., Batchelor B. and Chitnuyanondh L. (1979): 'Web crushing in concrete
girders with prestressing ducts in the web', PCI Journal, Vol. 24, N° 5, pp. 71-87, USA

19) Chan T.C-K (1979): 'A Study of the Behavior of Reinforced Concrete Dapped End
Beams ', University of Washington, Master Thesis, 167 p., Washington, USA

20) Chen W.F. (1982): 'Plasticity in reinforced concrete ', McGraw-Hill Education, 474 p.

21) Clarke J.L. and Taylor H.P.J. (1975): 'Web crushing - a review of research', Cement
and concrete association, Technical report, Vol. 42, N° 509, 16 p., London, UK

22) Collins M.P. (1978): 'Towards a Rational Theory for RC Members in Shear', ASCE,
Journal of the Structural Division, Vol. 104, N° 4, pp. 219-231, USA

23) Cook W.D. (1987): 'Studies of Disturbed Region Near Discontinuities in Reinforced
Concrete Members', Mc Gill University, Department of Civil Engineering and Applied
Mechanics, Master Thesis, pp. 153, Montreal, Canada

24) Curbach M., Hampel T., Scheerer S. and Speck K. (2002): 'Hochleistungsbeton unter
zwei-und dreiaxialer Beanspruchung', Beton-und Stahlbetonbau, Ernst & Sohn,
Vol.97, N° 6, pp. 275-280, Berlin, Germany

25) De Wilder K., Lava P., Debruyne D., Wang Y., De Roeck G. and Vandewalle L.
(2015): 'Stress Field Based Truss Model for Shear-Critical Prestressed Concrete
Beams', The Institution of Structural Engineers, Vol. 3, pp. 28–42, London, UK

26) Dorn W.S., Gomory R.E. and Greenberg H.J. (1964): 'Automatic Design of Optimal
Structures', Journal de Mécanique, Vol. 3, pp. 25-52

138
 References

27) Drucker D.C. (1961): 'On Structural Concrete and the Theorems of Limit Analysis',
IABSE International Association for Bridge and Structural Engineering, Report, N°21,
Zürich, Switzerland

28) Erntroy H.C. (1960): 'The Variation of Works Test Cubes', Cement and Concrete
Association, Report, N°10, London, UK

29) Eurocode 2 (2004): 'Design of concrete structures-Part 1-1: General rules and rules for
buildings', European Committee for Standardization (CEN), Brussels

30) Exner H. (1979): 'On the Effectiveness Factor in Plastic Analysis of Concrete', IABSE
reports of the working commissions, Vol. 29, pp. 35-42, Zürich, Switzerland

31) Fernández Ruiz M. and Muttoni A. (2008): 'Shear strength of thin-webbed post-
tensioned beams', ACI Structural Journal, Vol. 105, N° 3, pp. 308-317, USA

32) Fernández Ruiz M. and Muttoni A. (2007): 'On Development of Suitable Stress Fields
for Structural Concrete', ACI Structural Journal, Vol.104, N° 4, pp. 495-502, USA

33) Fernández Ruiz M., Muttoni A. and Gambarova P. (2007): 'Relationship between
nonlinear creep and cracking of concrete under uniaxial compression', Journal of
Advanced Concrete Technology, Vol. 5, N° 3, pp. 383-393, Japan

34) fib (2011): 'Model Code 2010', Special Activity Group 5, 653 p., Lausanne,
Switzerland

35) Frey F. and Jirousek J. (2001): 'Méthode des éléments finis (TGC volume 6) Analyse
des structures et milieux continues', Presses Polytechnique et Universitaires
Romandes, 298 p., Lausanne, Suisse

36) Ganz H.R., Ahmad A. and Hitz H. (1992): 'Load Transfer through Concrete Sections
with Grouted Ducts', VSL, Technical report, N° 242, 32 p., Bern, Switzerland

37) Gaynor R.D. (1965): 'Effect of Horizontal Reinforcing Steel on the Strength of Molded
Cylinders', ACI Structural Journal, Vol. 62, N° 7, pp. 837-840, USA

38) Goto Y. (1971): 'Cracks Formed in Concrete around Deformed Tension Bars', ACI
Journal, N° 68-26, pp. 244-251, Japan

39) Grob J. and Thürlimann B. (1976): 'Ultimate Strength and Design of Reinforced
Concrete Beams Under Bending and Shear', IABSE, N° 36, pp. 105-120

40) Gvozdev A.A. (1960): 'The Determination of the Value of the Collapse Load for
Statically Indeterminate Systems undergoing Plastic Deformation', (English
translation of the Russian original published in Proceedings of the Conference on
Plastic Deformations, Akademiia Nauk SSSR, Vol. 1, pp. 19-38, Moskau-Leningrad,
December 1936), International Journal of Mechanical Sciences, pp. 322-335,
Amsterdam, Netherlands

139
References

41) Hars E. (2006): 'Zum Querkraftwiderstand von Stahl-und Spannbetonträgern mit

dünnen Stegen', EPFL, IBETON, Thèse, N° 3551, 275 p., Lausanne, Switzerland

42) Heinzmann D. (2012): 'Stringer-Tafelmodelle Für Stahlbeton', ETHZ, Institut für

Baustatik und Konstruktion, Dissertation, N° 20303, 197 p., Zürich, Switzerland

43) Herzinger R. and Elbadry M. (2007): 'Alternative Reinforcing Details in Dapped Ends
of Precast Concrete Bridge Girders', Journal of the Transportation Research Board,
N° 2028, pp. 111-121

44) Himsworth F.R. (1954): 'The variability of concrete and its effect on mix design',
Proceedings of the Institution of Civil Engineers, Vol. 3, N° 2

45) Hong S.G., Kim D.-J., Kim S.-Y. and Hong N.K. (2002): 'Shear Strength of
Reinforced Concrete Deep Beams with End Anchorage Failure', ACI, Vol. 99, N° 1,
pp. 12-22, USA

46) Hoogenboom C.J. (1998): 'Discrete Elements and Nonlinearity in Design of Structural
Concrete Walls', Technical University Delft, Civil Engineering and Geosciences,
171 p., Delft, Netherlands

47) Hsu C.-T.T. and Zhang L.-X. (1997): 'Nonlinear Analysis of Membrane Elements by
Fixed-Angle Softened-Truss Model', ACI Structural Journal, Vol.94, N° 5, pp. 483-
491, USA

48) Ingerslev A. (1922): 'The Strength of Rectangular Slabs', The Institution of Structural
Engineers, pp. 3-19

49) Johansen K.W. (1931): 'Beregning af Krydsarmerede Jernbetonpladers Brudmoment',

BSM 3‐1 Bygningsstatiske Meddelelser, pp. 1-18

50) Kani G.N.J. (1964): 'The riddle of shear failure and its solution', ACI Journal, Vol. 61,
N° 4, pp. 441-467, USA

51) Kaufman M.K. and Ramirez J.A. (1988): 'Re-evaluation of the Ultimate Shear
Behavior of High-Strength Concrete Prestressed I-Beams', ACI Structural Journal,
Vol. 85, N° 3, pp. 295-303, USA

52) Kaufmann W. (1998): 'Strength and Deformations of Structural Concrete subjected to

In-Plane Shear and Normal Forces', IBK report, N° 234, 147 p., Zürich, Switzerland

53) Kaufmann W. and Marti P. (1998): 'Structural concrete: cracked membrane model',
Journal of Structural Engineering, Vol.124, N° 12, pp. 1467-1475, USA

54) Kaufmann W. and Marti P. (1996): 'Versuche an Stahlbetonträgern unter Normal- und
Querkraft', ETHZ, Institut für Baustatik und Konstruktion, Dissertation, N° 226,
141 p., Zürich, Switzerland

55) Khan M.A. (1981): 'A Study of the Behavior of Reinforced Concrete Dapped End
Beams ', University of Washington, Master Thesis, 145 p., Washington, USA

140
 References

56) Kirschner U. (1986): 'Investigating the behaviour of reinforced concrete shell

elements', University of Toronto, Department of Civil Engineering, PhD Thesis, 83 p.,
Toronto, Canada

57) Kollegger J. and Mehlhorn G. (1990): 'Experimentelle Untersuchungen zur

Bestimmung der Druckfestigkeit des gerissenen Stahlbetons bei einer
Querzugbeanspruchung', Deutscher Ausschuss für Stahlbeton, Wilhelm Ernst & Sohn,
Vol.413, 132 p., Berlin, Germany

58) König G. and Hosser D. (1981): 'The simplified level II method and its application on
the derivation of safety elements for level I', Comité Euro-International du Béton
(CEB), Bulletin N° 147, pp. 147-224, Lausanne, Switzerland

59) Kostic N. (2009): 'Topologie des champs de contraintes pour le dimensionnement des
structures en béton armé', EPFL, IBETON, Thèse, N° 4414, 235 p., Lausanne,
Switzerland

60) Kuchma D., Kim K.S., Nagle T.J., Sun S. and Hawkins N.M. (2008): 'Shear Tests on
High-Strength Prestressed Bulb-Tee Girders: Strengths and Key Observations', ACI
Structural Journal, Vol. 105, N° 3, pp. 358-367, USA

61) Kuchma D. and Tjhin T.N. (2001): 'CAST (Computer Aided Strut-and-Tie) Design
Tool', ASCE Journal of Structural Engineering, Vol.109, 7 p., USA

62) Kumar P. (1977): 'Optimal Force Transmission in Reinforced Concrete Deep Beams',
Computers & Structures, Vol. 8, pp. 223-229, UK

63) Kupfer H. (1973): 'Das Verhalten des Betons unter mehrachsiger Kurzzeitbelastung
unter besonderer Berücksichtigung der zweiachsigen Beanspruchung', Deutscher
Ausschuss für Stahlbeton, Vol.229, 131 p., Berlin, Germany

64) Kupfer H. (1964): 'Erweiterung der Mörsch'schen Fachwerkanalogie mit Hilfe des
Prinzips vom Minimum der Formänderungsarbeit', Comité Euro-International du
Béton (CEB), Bulletin d'Information, N° 40, pp. 44-57, Paris, France

65) Leonhardt F. (1969): 'Abminderung der Tragfähigkeit des Betons infolge

stabförmiger, rechtwinklig zur Druckrichtung angeordnete Einlagen Festschrift
Rüsch', pp. 71-78, Berlin, Germany

66) Leonhardt F., Koch R. and Rostasy F.S. (1973): 'Schubversuche an

Spannbetonträgern', Deutscher Ausschuss für Stahlbeton, Vol. 227, 179 p., Berlin,
Germany

67) Leonhardt F. and Walther R. (1962): 'Schubversuche an einfeldrigen Stahlbetonbalken

mit und ohne Schubbewehrung zur Ermittlung der Schubtragfähigkeit und der oberen
Schubspannungsgrenze', Deutscher Ausschuss für Stahlbeton, Wilhelm Ernst & Sohn,
Vol. 151, 83 p., Berlin, Germany

141
References

68) Leonhardt F. and Walther R. (1963): 'Schubversuche an Plattenbalken mit

unterschiedlicher Schubbewehrung', Deutscher Ausschuss für Stahlbeton, Wilhelm
Ernst & Sohn, Vol. 156, 84 p., Berlin, Germany

69) Leonhardt F. and Walther R. (1966): 'Wandartige Träger', Deutscher Ausschuss für
Stahlbeton, Wilhelm Ernst & Sohn, Vol. 178, 159 p., Berlin, Germany

70) Leonhardt F., Waltner R. and Dilger W. (1968): 'Schubversuche an indirekt

gelagerten, einfelbrigen und durchlaufenden Stahlbetonbalken', Deutscher Ausschuss
für Stahlbeton, Wilhelm Ernst & Sohn, Vol. 201, 69 p., Berlin, Germany

71) Lourenco M. and Almeida J. (2010): 'Adaptive stress field models for structural
concrete', 3rd fib International Congress, N° 262, 19 p., Washington D.C., USA

72) Mansur M.A., Lee Y.F., Tan K.-H. and Lee S.L. (1991): 'Tests on RC Continues
Beams with Openings', Journal of Structural Engineering, Vol. 117, N° 6, pp. 1593-
1606

73) Marti P. (1980): 'Zur plastischen Bemessung von Stahlbetonbauten', ETHZ, Institut
für Baustatik und Konstruktion, Dissertation, N° 6602, 176 p., Zürich, Switzerland

74) Marti P., Alvarez M., Kaufmann W. and Sigrist V. (1998): 'Tension chord model for
structural concrete', Structural Engineering International, IABSE, Vol. 8, N° 4,
pp. 287-298, USA

75) Marti P. and Meyboom J. (1992): 'Response of Prestressed Concrete Elements to In-
Plane Shear Forces', ACI Structural Journal, Vol.89, N° 5, pp. 503-514, USA

76) Mata Falcón J (2015): 'Serviceability and Ultimate Behaviour of Dapped-end Beams
(In Spanish: Estudio del comportamiento en servicio y rotura de los apoyos a media
madera)', Universitat Politècnica de València, Valencia, PhD Thesis, 747 p., Valencia,
Spain

77) Maxwell B.S. and Breen J.E. (2000): 'Experimental Evaluation of Strut-and-Tie Model
Applied to Deep Beam with Opening', ACI Structural Journal, Vol.97, N° 1, pp. 142-
149, USA

78) Moccia F. (2015): 'Investigation of Shear Strength of Post-Tensioned Girders', EPFL,

IBETON, Master Thesis, 225 p., Lausanne, Switzerland

79) Moore A.M. (2014): 'Shear Behavior of Spliced Post-Tensioned Girders', Faculty of
the Graduate School, The University of Texas at Austin, PhD Thesis, 250 p., Austin,
USA

80) Mörsch E. (1908): 'Der Eisenbetonbau - Seine Theorie und Anwendung', 3. Auflage,
376 p., Stuttgart, Germany

81) Muller J. (1978): 'Enseignement tiré de l'exécution des ouvrages pour une meilleure
conception', Journées de l'A.P.C., pp. 99-109, France

142
 References

82) Muttoni A. (1990): 'Die Anwendbarkeit der Plastizitätstheorie in der Bemessung von
Stahlbeton', Birkhäuser Verlag, Institut für Baustatik und Konstruktion ETH Zürich,
N° 176, 164 p., Basel, Switzerland

83) Muttoni A., Burdet O. and Hars E. (2006): 'Effect of Duct Type on the Shear Strength
of Thin Webs', ACI Structural Journal, Vol.103, N° 5, pp. 729-735, USA

84) Muttoni A. and Fernández Ruiz M. (2007): 'Shear strength predictions according to
the critical shear crack theory and the Swiss code SIA 262 (2003)', Workshop on
assessment methods for determining the shear strength of existing structures, 14 p.,
Rotterdam, Netherlands

85) Muttoni A. and Fernández Ruiz M. (2012): 'The levels-of-approximation approach in

MC 2010: application to punching shear provisions', Structural Concrete, Vol. 13,
N° 1, pp. 32-41

86) Muttoni A. and Fernández Ruiz M. (2012): 'Levels-of-approximation approach in

codes of practice', Structural Engineering International, Vol. 2, pp. 190-194, Zurich,
Switzerland

87) Muttoni A. and Fernández Ruiz M. (2008): 'Shear strength of members without
transverse reinforcement as function of critical shear crack width', ACI Structural
Journal, Vol. 105, N° 2, pp. 163-172, USA

88) Muttoni A., Fernández Ruiz M. Niketic F. (2015): 'Design versus Assessment of
Concrete Structures Using Stress Fields and Strut-and-Tie Models', ACI Structural
Journal, Vol.112, N° 5, pp. 605-616, USA

89) Muttoni A., Schwartz J. and Thürlimann B. (1996): 'Design of Concrete Structures
with Stress Fields', Birkhäuser Verlag, 143 p., Basel, Switzerland

90) Nagrodzka-Godycka K. and Piotrkowski P (2012): 'Experimental and Analytical

Investigation of Classic Concrete Beam Tests', ACI Structural Journal, Vol.109, N° 1,
pp. 11-20

91) Nielsen M.P. (1969): 'Om forskydningsarmering i jernbetonbjælker (Shear

reinforcement in beams)', Proceedings of the Sessions Related to Structural Design,
Analysis and Testing, ASCE Structural Congress, Vol. 38, N° 2, pp. 33-58

92) Nielsen M.P. and Hoang L.C. (2011): 'Limit Analysis and Concrete Plasticity', CRC
Press, 3rd edition, 788 p., Boca Raton, USA

93) Okamura H. and Maekawa K. (1991): 'Nonlinear analysis and constitutive methods of
reinforced concrete', University of Tokyo, 850 p., Tokyo

94) Pang X.-B.D. and Hsu C.-T.T. (1995): 'Behavior of Reinforced Concrete Membrane
Elements in Shear', ACI Structural Journal, Vol.92, N° 6, pp. 665-667, USA

143
References

95) Pimentel M., Brühwiler E. and Figueiras J. (2014): 'Safety examination of existing
concrete structures using the global resistance safety factor concept', Engineering
Structures, Vol. 70, pp. 130-143

96) Placas A. (1969): 'Shear Strength of Reinforced Concrete Beams', Imperial College of
Science and Technology, PhD Thesis, 581 p., London, England

97) Popovics S. (1998): 'Strength and Related Properties of Concrete: A Quantitative

Approach', John Wiley & Sons Inc., New York, USA

98) Randl N. (2013): 'Design recommendations for interface shear transfer in fib Model
Code 2010', Structural Concrete, N° 3, pp. 230-241, Berlin, Germany

99) Rasmussen B.H. (1962): 'Strength of transversely loaded bolts and dowels cast into
concrete', Laboratoriet for Bugningastatik, Denmark Technical University,
Meddelelse, Vol. 34, N°2

100) Rezai-Jorabi H. and Regan P.E. (1986): 'Shear resistance of prestressed concrete
beams with inclined tendons', The Structural Engineer, Vol. 64B, N° 3, pp. 63-75,
London, UK

101) Ritter W. (1899): 'Die Bauweise Hennebique', Schweizerische Bauzeitung, pp. 41-
149, Zürich, Switzerland

102) Rupf M. (2014): 'Querkraftwiderstand von Stahlbeton- und Spannbetonträgern mittels

Spannungsfeldern', EPFL, IBETON, Thèse, N° 6004, 132 p., Lausanne, Switzerland

103) Rupf M. and Muttoni A. (2012): 'Schubversuche an vorgespannten Stahlbetonträgern

mit ungenügender Schubbewehrung', EPFL, IBETON, Test Report, 159 p., Lausanne,
Switzerland

104) Sagaseta J. and Vollum R.L. (2011): 'Influence of beam cross-section, loading
arrangement and aggregate type on shear strength', Magazine of Concrete Research,
Vol.53, N° 2, pp. 139-155, London, UK

105) Saqan E.I. and Frosch R.J. (2009): 'Influence of flexural reinforcement on shear
strength of prestressed concrete beams', ACI Structural Journal, Vol. 106, N° 1,
pp. 60-68, USA

106) Schäfer K., Schelling G. and Kuchler T. (1990): 'Druck und querzug in bewehrten
betonelementen', Deutscher Ausschuss für Stahlbeton, N° 408, pp. 5-85, Berlin,
Germany

107) Schlaich J., Schäfer K. and Jennewein M. (1987): 'Toward a Consistent Design of
Structural Concrete ', ACI Journal, Vol.32, N° 3, pp. 75-150, USA

108) SIA 262:2013 (2013): 'Structures en béton', Société suisse des ingénieurs et des
architectes, 102 p., Zurich, Switzerland

144
 References

109) SIA 269:2011 (2011): 'Existing Structures - Basis for examination and interventions',
Société suisse des ingénieurs et des architectes, 30 p., Zurich, Switzerland

110) Sørensen H.C. (1974): 'Shear Tests on 12 Reinforced Concrete T-Beams', Technical
University of Denmark, N° R60, 52 p., Lyngby, Denmark

111) Sørensen J.H., Hoang L.H., Olesen J.F. and Fischer G. (2016): 'Catenary action in
rebars crossing a casting joint loaded in shear', 11th fib International PhD Symposium
in Civil Engineering, pp. 735-742, Tokyo

112) Tanner P., Lara C. and Hingorani R. (2007): 'Structural Safety - a struggle with
uncertainties', Hormigón y Acero, N° 245, pp. 59-78, Madrid, Spain

113) Tanner P., Lara C. and Prieto M. (2011): 'Semi-probabilistic models for the assessment
of existing concrete structures', Applications of Statistics and Probability in Civil
Engineering, Köhler and Nishijima, Taylor & Francis Group, pp. 1039-1047, London,
UK

114) Tanner P., Lara C. and Prieto M. (2014): 'Analysis of the present condition and
deterioration in the main dome over La Laguna cathedral', Proceedings of the IABSE
Conference Engineering for Progress, Nature and People, 10 p., Madrid, Spain

115) Tepfers R. (1979): 'Cracking of concrete cover along anchored deformed reinforcing
bars', Magazine of Concrete Research, Vol.31, N° 106, pp. 3-12, UK

116) Thürlimann B., Marti P., Pralong J., Ritz P. and Zimmerli B. (1983): 'Anwendung der
Plastizitätstheorie auf Stahlbeton', Institut für Baustatik und Konstruktion ETH
Zürich, 252 p., Zurich, Switzerland

117) Vecchio F.J. (2000): 'Disturbed Stress Field Model for Reinforced Concrete:
Formulation', ASCE Journal of Structural Engineering, Vol.126, Issue 8, pp. 1070-
1077, USA

118) Vecchio F.J. and Chan C.C.L. (1990): 'Reinforced concrete membrane elements with
perforations', ASCE Journal of Structural Engineering, Vol. 116, Issue 9, pp. 2344-
2360

119) Vecchio F.J. and Collins M.P. (1982): 'Response of reinforced concrete to in-plane
shear and normal stresses', Department of civil engineering, University of Toronto,
N° 82-03, 332 p.

120) Vecchio F.J. and Collins M.P. (1986): 'The modified compression-field theory for
reinforced concrete elements subjected to shear', ACI Structural Journal, Vol.83, N° 2,
pp. 219-231, USA

121) Vecchio F.J., Collins M.P. and Aspiotis J. (1994): 'High-Strength Concrete Elements
Subjected to Shear', ACI Structural Journal, Vol.91, N° 4, pp. 423-433, USA

145
References

122) Vecchio F.J. and Shim W. (2004): 'Experimental and Analytical Investigation of
Classic Concrete Beam Tests', ASCE Journal of Structural Engineering, Vol.130,
N° 3, pp. 460-469, USA

123) Vintzeleou E. and Tassios T.P. (1990): 'Eccentric dowels loaded against core of
concrete sections ', ASCE Journal of Structural Engineering, Vol. 116, N° 10,
pp. 2621-2633

124) Watanabe F. and Muguruma H. (1989): 'Ultimate Strength and Deformations of RC

Panel', Proceedings of the Sessions Related to Structural Design, Analysis and Testing,
ASCE Structural Congress, pp. 31-38

125) Xie Y.M. and Steven G.P. (1994): 'Optimal Design of Multiple Load Case Structures
Using an Evolutionary Procedure', Engineering Computation, Vol. 11, pp. 295-302,
UK

126) Yoon Y.-S., Cook W.D. and Mitchell D. (1996): 'Minimum Shear Reinforcement in
Normal, Medium and High-Strength Concrete Beams', ACI Structural Journal, Vol.93,
N° 5, pp. 576-584, USA

127) Zhang L.-X. and Hsu T.T.C. (1998): 'Behavior and Analysis of 100 MPa Concrete
Membrane Elements', Journal of Structural Engineering, Vol. 124 N° 1, pp. 24-34,
USA

128) Zhu R.R.H., Wanichakorn W, Hsu C.-T.T. and Vogel J (2003): 'Crack Width
Prediction Using Compatibility-Aided Strut-and-Tie Model', ACI Structural Journal,
Vol.100, N° 4, pp. 413-421, USA

129) Zwicky D. (2000): 'Bruchversuche an ausgebauten Brückenträgern', Institut für

Baustatik und Konstruktion, N° 258, 167 p., Zürich, Switzerland

146
Appendix
1) Appendix 1 – Summary of the online database
The following appendix gives the mechanical properties of 315 reinforced and prestressed
structural elements, which were used to investigate the accuracy of the EPSF method in
Chapter 2 of this thesis. Both measured and estimated strength of each member are indicated
along with the references to the tests. Basic statistical analysis of the EPSF results has been
conducted and presented (the average prediction of the ultimate load and the coefficient of
variation for each series).
All the values correspond to the ones available at: http://i-concrete.epfl.ch/epsf/epsf.html.

1.1) Reinforced concrete members

fc ρ ρw σp,inf Failure Failure Qtest QEPSF Qtest
N° Name
[MPa] [%] [%] [MPa] mode Subtype [kN] [kN] QEPSF
Vecchio F.J. and Shim W. (2004): 'Experimental and Analytical Investigation of Classic Concrete Beam
Tests', ASCE Journal of Structural Engineering, Vol.130, N° 3, pp. 460-469, USA
1 A1 22.6 1.94 0.10 0 V CR 459 450 1.02
2 A2 25.9 2.44 0.10 0 V CR 439 452 0.97
3 A3 43.5 2.94 0.10 0 F CR 420 426 0.99
4 B1 22.6 2.58 0.15 0 V CR 434 416 1.04
5 B2 25.9 2.58 0.15 0 V CR 365 331 1.10
6 B3 43.5 3.25 0.15 0 F CR 342 344 0.99
7 C1 22.6 2.45 0.20 0 V CR 282 247 1.14
8 C2 25.9 3.89 0.20 0 V CR 290 290 1.00
9 C3 43.5 3.89 0.20 0 F CR 265 260 1.02
Average: 1.03
CoV: 0.05
Yoon Y.-S., Cook W.D. and Mitchell D. (1996): 'Minimum Shear Reinforcement in Normal, Medium and
High-Strength Concrete Beams', ACI Structural Journal, Vol.93, N° 5, pp. 576-584, USA
1 N1_N 36 2.8 0.08 0 V DT 914 840 1.09
2 N2_S 36 2.8 0.08 0 V DT 726 800 0.91
3 N2_N 36 2.8 0.12 0 V CR 966 1030 0.94
4 M1_N 67 2.8 0.08 0 V DT 810 962 0.84
5 M2_S 67 2.8 0.12 0 V CR 1104 1120 0.99
6 M2_N 67 2.8 0.16 0 V CR 1378 1370 1.01
7 H1_N 87 2.8 0.08 0 V DT 966 1012 0.95
8 H2_S 87 2.8 0.14 0 V DT 1196 1302 0.92
9 H2_N 87 2.8 0.24 0 V CR 1442 1656 0.87
Average: 0.95
CoV: 0.07
Sagaseta J. and Vollum R.L. (2011): 'Influence of beam cross-section, loading arrangement and aggregate
type on shear strength', Magazine of Concrete Research, Vol.53, N° 2, pp. 139-155, London, UK

A1
Appendix

1 BG1 31.7 3.32 0.50 0 V CR 950 990 0.96

2 BG2 31.7 3.32 0.83 0 V CR 1074 1250 0.86
3 BL1 53.1 3.32 0.50 0 V SY 1169 1126 1.04
4 BL2 53.1 3.32 0.83 0 V SP 1594 1476 1.08
5 CB1 49.4 2.80 0.36 0 V SY 1029 1020 1.01
6 CB2 49.4 2.80 0.53 0 V SY 1429 1257 1.14
7 DB1 49.4 2.80 0.36 0 V SY 597 540 1.11
Average: 1.03
CoV: 0.09
Mansur M.A., Lee Y.F., Tan K.-H. and Lee S.L. (1991): 'Tests on RC Continuos Beams with Openings',
Journal of Structural Engineering, Vol. 117, N° 6, pp. 1593-1606
1 B1 38.4 1.54 0.28 0 F SP 135 127 1.06
2 B2 40.5 1.54 0.28 0 F SP 155 145 1.07
3 B3 43.8 1.54 0.28 0 F SP 140 144 0.97
4 C1 43.8 1.54 1.01 0 F SP 260 261 1.00
5 C2 38.4 1.54 1.01 0 F SP 230 244 0.94
6 C3 40.5 1.54 1.01 0 F SP 230 232 0.99
7 C4 28.8 1.54 1.01 0 F SP 240 253 0.95
8 C5 28.8 1.54 1.01 0 F SP 180 200 0.90
Average: 0.99
CoV: 0.06
Hong S.G., Kim D.-J., Kim S.-Y. and Hong N.K. (2002): 'Shear Strength of Reinforced Concrete Deep
Beams with End Anchorage Failure', ACI, Vol. 99, N° 1, pp. 12-22, USA
1 SS_1 23.5 1.66 0.42 0 V DT 662.28 610 1.09
2 SS_2 23.5 1.66 0.42 0 V DT 610.34 538 1.13
3 SS_3 23.5 1.66 0.42 0 L A 560.66 507 1.11
4 SS_4 23.5 1.66 0.42 0 L A 479.22 494 0.97
5 LBS_2 23.5 1.66 0.42 0 L A 579.96 576 1.01
6 VSR_1 23.5 1.66 0.52 0 L A 593.29 566 1.05
7 VSR_2 23.5 1.66 0.70 0 V DT 658.07 608 1.08
Average: 1.06
CoV: 0.05
Sørensen H.C. (1974): 'Shear Tests on 12 Reinforced Concrete T-Beams', Technical University of Denmark,
N° R60, 52 p., Lyngby, Denmark
1 T23 34.2 1.06 0.34 0 V DT 139 139 1
2 T1a 22.9 1.06 0.59 0 F CR 133 115 1.16
3 T2a 24.6 1.06 0.41 0 F CR 137 122 1.12
4 T3a 24.6 1.06 0.49 0 V AS 126 105 1.20
5 T4a 25.2 1.06 0.34 0 V AS 131 114 1.15
6 T1b 23.1 1.06 0.44 0 V CR 118 102 1.16
7 T2b 24.9 1.06 0.30 0 V DT 129 108 1.19
8 T3b 24.6 1.06 0.29 0 V DT 116 89 1.30
9 T4b 24.7 1.06 0.20 0 V DT 106 94 1.13
10 T5 25.5 1.06 0.20 0 V DT 110 95 1.16
Average: 1.16
CoV: 0.06
Leonhardt F. and Walther R. (1963): 'Schubversuche an Plattenbalken mit unterschiedlicher Schubbewehrung',
Deutscher Ausschuss für Stahlbeton, Wilhelm Ernst & Sohn, Vol. 156, 84 p., Berlin, Germany
1 TA1 15.2 0.84 1.29 0 V CR 670 582 1.15
2 TA2 15.2 0.84 0.86 0 V SY 638 530 1.20
3 TA3 15.1 0.84 0.59 0 V SY 544 476 1.14
4 TA4 15.1 0.84 0.34 0 V SY 458 382 1.20
5 TA13 17.9 0.84 1.29 0 F CR 700 635 1.10
6 TA14 17.9 0.84 0.86 0 V SY 666 597 1.12
7 TA15 17.1 0.84 0.59 0 V SY 584 514 1.14
8 TA9 24.8 0.84 1.29 0 F Y 700 700 1.00

A2
 Summary of the online database

9 TA10 24.8 0.84 0.86 0 F Y 714 690 1.03

10 TA11 24.4 0.84 0.59 0 V SY 670 586 1.14
11 TA12 24.4 0.84 0.34 0 V SY 530 468 1.13
12 TA5 15.1 0.84 1.30 0 L SP 453 450 1.01
13 TA17 20.3 0.84 1.30 0 V CR 677 614 1.10
14 TA18 26.8 0.84 1.30 0 F Y 709 628 1.13
15 TA6 15.1 0.84 0.59 0 V SY 465 480 0.97
16 TA16 17.1 0.84 0.59 0 V SY 587 506 1.16
Average: 1.11
CoV: 0.06
Kaufmann W. and Marti P. (1996): 'Versuche an Stahlbetonträgern unter Normal- und Querkraft', ETHZ,
Institut für Baustatik und Konstruktion, Dissertation, N° 226, 141 p., Zürich, Switzerland
1 VN1 53.9 4.23 0.34 0 V SY 542 546 0.99
2 VN2 52.6 4.23 0.34 0 V SY 548 522 1.05
3 VN3 60.2 4.23 0.34 0 V SY 540 510 1.06
4 VN4 61.9 4.23 0.34 0 V SY 564 555 1.02
Average: 1.03
CoV: 0.03
Nagrodzka-Godycka K. and Piotrkowski P (2012): 'Experimental and Analytical Investigation of Classic
Concrete Beam Tests', ACI Structural Journal, Vol.109, N° 1, pp. 11-20
1 WB_1_L 36.4 0.63 0.76 0 L SY 130 126 1.03
2 WB_1_P 36.4 0.63 0.76 0 L SY 140 126 1.11
3 WB_2_L 36.4 0.63 0.76 0 L SP 180 188 0.96
4 WB_2_P 36.4 0.63 0.76 0 L SP 180 180 1.00
5 WB_3_L 36.4 1.26 1.52 0 L SY 206 206 1.00
6 WB_3_P 36.4 1.26 1.52 0 L SY 206 206 1.00
7 WB_4_L 36.4 1.26 1.52 0 L SP 270 272 0.99
8 WB_4_P 36.4 1.26 1.52 0 L SP 280 272 1.03
9 WB_5_L 36.4 0.56 0.73 0 L SY 172 180 0.96
10 WB_5_P 36.4 0.84 0.98 0 L SP 200 248 0.81
11 WB_6_L 36.4 0.56 0.73 0 L SP 238 256 0.93
12 WB_6_P 36.4 0.84 0.98 0 L SP 320 354 0.90
Average: 0.98
CoV: 0.07
Mata Falcón J (2015): 'Serviceability and Ultimate Behaviour of Dapped-end Beams (In Spanish: Estudio del
comportamiento en servicio y rotura de los apoyos a media madera)', Universitat Politècnica de València,
Valencia, PhD Thesis, 747 p., Valencia, Spain
1 DEB1.1_T1 41.1 1.1 0.27 0 L SY+SP 484 471 1.03
2 DEB1.2_T1 39.3 0.99 0.27 0 L SY+SP 365 339 1.08
3 DEB1.2_T2 39.3 0.99 0.27 0 L SY+SP 332 339 0.98
4 DEB1.3_T1 39.9 1.10 0.23 0 L SY 303 338 0.90
5 DEB1.3_T2 39.9 1.10 0.23 0 L SY 332 338 0.98
6 DEB1.4_T1 40.4 1.10 0.27 0 L SY+SP 457 442 1.04
7 DEB1.4_T2 40.4 1.10 0.27 0 L SY 426 442 0.97
8 DEB1.5_T1 40.8 0.99 0.27 0 L SY 313 303 1.03
9 DEB1.6_T1 31.1 2.38 0.55 0 L SY+SP 773 634 1.22
10 DEB1.6_T2 31.1 2.38 0.55 0 L SY+SP 627 634 0.99
11 DEB1.7_T1 30.0 2.15 0.55 0 L SY 486 506 0.96
12 DEB1.7_T2 30.0 2.15 0.55 0 L SY 472 506 0.93
13 DEB1.8_T1 32.2 1.69 0.41 0 L SY 488 575 0.85
14 DEB1.8_T2 32.2 1.69 0.41 0 L SY 498 575 0.87
15 DEB1.9_T1 31.9 1.54 0.41 0 L SY 354 421 0.84
16 DEB1.9_T2 31.9 1.54 0.41 0 L SY 364 421 0.86
17 DEB2.1_T1 40.2 1.07 0.27 0 L SY+SP 487 471 1.03
18 DEB2.1_T2 40.2 1.07 0.27 0 L SY 499 471 1.06
19 DEB2.2_T1 33.3 2.29 0.54 0 L SY+SP 805 792 1.02

A3
Appendix

20 DEB2.2_T2 33.3 2.29 0.54 0 L SY 824 792 1.04

21 DEB2.3_T1 33.3 1.64 0.42 0 L SY+SP 601 619 0.97
22 DEB2.4_T1 36.9 2.25 0.54 0 L SY+SP 780 814 0.96
23 DEB2.4_T2 36.9 2.25 0.54 0 L SY+SP 774 814 0.95
24 DEB2.5_T1 37.1 2.18 0.53 0 L SY+SP 663 706 0.94
25 DEB2.5_T2 37.1 2.18 0.53 0 L SY+SP 737 706 1.04
26 DEB2.6_T1 38.3 2.62 0.53 0 L SY+SP 820 788 1.04
27 DEB3.1_T1 33.7 2.25 0.54 0 L SY+SP 795 814 0.98
28 DEB3.1_T2 33.7 2.25 0.54 0 L SY+SP 851 814 1.04
29 DEB3.2_T1 37.2 2.18 0.53 0 L SY+SP 780 743 1.05
30 DEB3.2_T2 37.2 2.18 0.53 0 L SY+SP 796 743 1.07
31 DEB3.3_T1 38.8 2.62 0.53 1.24 L SY+SP 876 847 1.03
32 DEB3.3_T2 38.8 2.62 0.53 1.06 L SY+SP 841 842 1.00
33 DEB3.4_T1 34.5 2.38 0.55 0 L SY+SP 654 669 0.98
34 DEB3.4_T2 34.5 2.38 0.55 0 L SY+SP 665 669 0.99
35 DEB3.5_T1 33.0 2.29 0.54 0 L SY 849 838 1.01
36 DEB3.5_T2 33.0 2.29 0.54 0 L SY 856 838 1.02
37 DEB3.6_T1 36.7 1.69 0.41 0 L SY 567 591 0.96
38 DEB3.6_T2 36.7 1.69 0.41 0 L SY+SP 553 591 0.94
39 DEB3.7_T1 45.5 2.38 0.55 0 L SY 832 796 1.04
40 DEB3.7_T2 45.5 2.38 0.55 0 L SY 821 796 1.03
41 DEB3.8_T1 48.8 2.38 0.55 0 L SY 908 882 1.03
42 DEB3.8_T2 48.8 2.38 0.55 0 L SY 904 882 1.02
43 DEB3.9_T1 48.4 2.38 0.55 0 L SY 886 929 0.95
44 DEB3.9_T2 48.4 2.38 0.55 0 L SY 925 929 1.00
45 DEB3.10_T1 41.8 0 0.53 1.75 L SY+SP 887 888 1.00
46 DEB-3.1_T2 41.8 0 0.53 1.39 L SY+SP 925 886 1.04
47 DEB3.11_T1 45.5 0 0.53 2.23 L SY+SP 1028 1001 1.03
48 DEB3.11_T2 45.5 0 0.53 2.29 L SY+SP 988 1001 0.99
49 DEB3.12_T1 48.4 0 0.53 3.41 L SY+SP 1010 1089 0.93
50 DEB3.12_T2 48.4 0 0.53 3.04 L SY+SP 1033 1085 0.95
Average: 0.97
CoV: 0.07
Chan T.C-K (1979): 'A Study of the Behavior of Reinforced Concrete Dapped End Beams', University of
Washingon , Master Thesis, 167 p., Washington, USA
1 1A 33.6 1.71 0.47 0 L - 144 136 1.06
2 1B 30.5 2.33 0.47 0 L - 191 184 1.04
3 2A 33.0 1.88 0.44 0 L - 178 182 0.98
4 2B 30.9 2.33 0.44 0 L - 169 184 0.92
5 3A 37.0 1.88 0.47 0 L - 216 194 1.11
6 3B 30.3 2.33 0.47 0 L - 176 181 0.98
7 4A 30.3 1.88 0.43 0 L - 189 187 1.01
8 4B 29.4 2.33 0.43 0 L - 177 169 1.05
Average: 1.02
CoV: 0.05
Khan M.A. (1981): 'A Study of the Behavior of Reinforced Concrete Dapped End Beams ', University of
Washingon , Master Thesis, 145 p., Washington, USA
1 1A 28.8 2.39 0.53 0 L - 215 198 1.09
2 1B 29.8 2.71 0.50 0 L - 188 190 0.99
3 2A 29.7 2.71 0.54 0 L - 208 184 1.13
4 2B 31.0 2.97 0.53 0 L - 189 168 1.13
5 3A 33.7 2.41 0.61 0 L - 197 210 0.94
6 3B 37.1 2.67 0.59 0 L - 189 205 0.92
7 4A 29.0 2.16 0.59 0 L - 176 165 1.06
8 5B1 33.7 2.43 0.60 0 L - 164 152 1.07
9 5B2 34.5 2.43 0.60 0 L - 143 148 0.96

A4
 Summary of the online database

Average: 1.03
CoV: 0.07
Cook W.D. (1987): 'Studies of Disturbed Region Near Discontinuities in Reinforced Concrete Members', Mc
Gill University, Department of Civil Engineering and Applied Mechanics, Master Thesis, pp. 153, Montreal,
Canada
1 D-1 29.8 2.61 0.4 0 L - 307 324 0.95
2 D-3 36.3 2.51 0.47 0 L - 372 382 0.97
3 D-4 36.3 2.60 0.46 0 L - 340 324 1.05
Average: 0.99
CoV: 0.04
Zhu R.R.H., Wanichakorn W, Hsu C.-T.T. and Vogel J (2003): 'Crack Width Prediction Using Compatibility-
Aided Strut-and-Tie Model', ACI Structural Journal, Vol.100, N° 4, pp. 413-421, USA
1 T2 41.75 0.52 0.25 0 L - 563 544 1.03
2 T3 33.55 0.64 0.36 0 L - 538 513 1.05
3 T4 41.46 0.52 0.25 0 L - 572 618 0.92
4 T5 38.96 0.52 0.36 0 L - 921 789 1.17
5 T6 43.33 0.52 0.36 0 L - 467 445 1.05
6 T7 47.08 0.61 0.44 0 L - 1196 1062 1.13
Average: 1.06
CoV: 0.07
Herzinger R. and Elbadry M. (2007): 'Altermative Reinforcing Details in Dapped Ends of Precast Concrete
Brigde Girders', Journal of the Transportation Research Board, N° 2028, pp. 111-121
1 DEA1.0 T1 38.1 2.31 0.42 0 L - 216 204 1.06
2 DEA1.0 T2 48.4 2.31 0.42 0 L - 255 222 1.15
3 DEA0.5 T1 38 2.31 0.45 0 L - 231 231 1.00
4 DEB1.0 T1 38.6 2.28 0.42 0 L - 203 219 0.93
5 DEB1.0 T2 40.4 2.28 0.42 0 L - 226 219 1.03
6 DEB0.5 T1 36.9 2.28 0.45 0 L - 205 228 0.9
7 DEB0.5 T2 36.9 2.28 0.45 0 L - 222 228 0.97
8 DEC1.0 T1 39.1 2.3 0.38 0 L - 181 203 0.89
9 DEC1.0 T2 41.6 2.3 0.38 0 L - 212 203 1.05
10 DEC*1.0T1 42.2 2.3 0.49 0 L - 260 275 0.94
11 DEC*u1.0T1 41.9 2.3 0.54 0 L - 270 269 1.01
12 DED1.0T1 38.8 2.23 0.40 0 L - 220 222 0.99
13 DEDu1.0T1 36.8 2.23 0.40 0 L - 213 224 0.95
14 DEDu1.0T2 37.4 2.23 0.40 0 L - 222 224 0.99
15 DED*1.0T1 39.9 2.21 0.41 0 L - 214 222 0.96
16 DED*1.0T2 40.5 2.21 0.41 0 L - 203 222 0.91
17 DEDu*1.0T1 39.2 2.21 0.41 0 L - 212 220 0.96
18 DEDu*1.0T2 40.3 2.21 0.41 0 L - 227 220 1.03
Average: 0.99
CoV: 0.06
Campana S. (2013): 'Éléments en béton armé soumis à une combinaison de flexion, effort tranchant et forces
de déviation', EPFL, IBETON, Thèse, N° 5574, 162 p., Lausanne, Switzerland
1 SC26 41.9 0.71 0 0 L DT 108 117.5 0.92
2 SC27 41.6 0.71 0 0 L DT 124 127.5 0.97
3 SC31 41.7 0.71 0 0 L DT 119 127.5 0.93
4 SC34 41.4 0.72 0 0 L DT 114 107.5 1.06
5 SC35 42.1 0.72 0 0 L CR 134 127.5 1.05
6 SC38 31.3 0.70 0.17 0 L DT 110 112.5 0.98
7 SC39 31.1 0.71 0.19 0 L DT 109 122.5 0.89
8 SC40 30.9 0.7 0.19 0 L DT 106 125.0 0.85
9 SC41 30.9 0.7 0.22 0 L CR 132 127.5 1.03
10 SC42 31.0 0.71 0.22 0 L CR 127 127.5 1.00
11 SC43 31.1 0.7 0.26 0 L CR 129 127.5 1.01
12 SC44 30.9 0.7 0.19 0 L DT 118 122.5 0.97

A5
Appendix

13 SC45 30.8 0.7 0.22 0 L CR 123 125.0 0.99

Average: 0.97
CoV: 0.06
Placas A. (1969): 'Shear Strength of Reinforced Concrete Beams', Imperial College of Science and
Technology, PhD Thesis, 581 p., London, England
1 R10 30 0.97 0.21 0 V CR 76 73 1.03
2 R11 26 1.95 0.21 0 V CR 90 83 1.08
3 R12 34 4.17 0.21 0 V CR 110 98 1.12
4 R14 29 1.46 0.14 0 V DT 90 67 1.34
5 R17 13 1.46 0.21 0 V CR 70 54 1.30
6 R20 43 1.46 0.21 0 V CR 90 92 0.98
7 R22 30 1.46 0.21 0 V CR 80 82 0.98
8 R24 31 1.46 0.21 0 V DT 93 84 1.10
9 R25 31 4.17 0.21 0 V DT 105 95 1.11
10 T1 28 0.31 0.21 0 V DT 111 95 1.16
11 T3 28 0.36 0.21 0 V DT 105 97 1.08
12 T4 33 0.48 0.21 0 V DT 110 114 0.96
13 T7 27 0.75 0.21 0 V DT 110 118 0.93
14 T8 31 1.04 0.21 0 V DT 125 130 0.96
15 T10 28 0.36 0.14 0 V DT 87 90 0.97
16 T13 13 0.36 0.21 0 V DT 90 64 1.41
17 T15 33 1.04 0.21 0 V SY 105 115 0.91
18 T16 33 1.04 0.14 0 V SY 90 117 0.77
19 T19 30 1.04 0.21 0 V CR 113 112 1.00
20 T25 54 0.36 0.21 0 V SY 115 125 0.92
21 T31 31 0.36 0.21 0 V SY 95 103 0.92
22 T34 34 2.08 0.21 0 V SY 113 117 0.96
23 T35 34 0.59 0.21 0 V SY 115 119 0.97
Average: 1.04
CoV: 0.14
Bach F., Nielsen M.P. and Braestrup M.W. (1980): 'Shear Tests on Reinforced Concrete T-Beams - Series V,
U, X, B and S', Structural Research Laboratory, Technical University of Denmark, Report N° 120, 87 p.,
Copenhagen, Denmark
1 V6002W 35.7 0.72 0.27 0 V DT 245 233 1.05
2 V6002E 35.7 0.72 0.27 0 V DT 253 233 1.09
3 V6004W 36.4 0.72 0.43 0 V DT 306 292 1.05
4 V6004E 36.4 0.72 0.43 0 V DT 347 292 1.19
5 U6002W 19.5 0.72 0.13 0 V DT 194 144 1.35
6 U6002E 19.5 0.72 0.13 0 V DT 200 144 1.39
7 U6004W 21.1 0.72 0.27 0 V DT 224 193 1.16
8 U6004E 21.1 0.72 0.27 0 V DT 237 193 1.23
9 X6009W 7.3 0.32 0.27 0 V DT 133 145 0.92
10 X6009E 7.3 0.32 0.27 0 V DT 143 145 0.99
11 B6009W 10.7 0.57 0.23 0 V DT 286 237 1.21
12 B6009E 10.7 0.57 0.23 0 V DT 245 230 1.07
Average: 1.14
CoV: 0.12
Leonhardt F. and Walther R. (1966): 'Wandartige Träger', Deutscher Ausschuss für Stahlbeton, Wilhelm
Ernst & Sohn, Vol. 178, 159 p., Berlin, Germany
1 WT4 28 0.40 0.16 0 F SY 1526 1590 0.96
2 WT7 30 0.40 2.51 0 F SY 1119 1130 0.99
3 IWT1 28 0.98 0.71 0 L CR 1152 1130 1.02
4 IWT2 28 0.98 0.38 0 V CR 1177 1114 1.06
Average: 1.02
CoV: 0.04

A6
 Summary of the online database

Leonhardt F., Waltner R. and Dilger W. (1968): 'Schubversuche an indirekt gelagerten, einfelbrigen und
durchlaufenden Stahlbetonbalken', Deutscher Ausschuss für Stahlbeton, Wilhelm Ernst & Sohn, Vol. 201, 69
p., Berlin, Germany
1 ETI1 30 1.32 0.16 0 V SY 273 276 0.99
2 ETI2 26 1.40 0.28 0 F CR 257 250 1.03
3 ETI3 25 1.40 0.76 0 F CR 240 222 1.08
4 ETI4 27 1.40 0.86 0 F CR 245 250 0.98
5 ETI5 28 1.42 0.27 0 V SY 240 246 0.98
Average: 1.01
CoV: 0.04
Baumann T. and Rüsch H. (1970): 'Schubversuche mit indirekter Krafteinleitung: Versuche zum Studium der
Verdübelungswirkung der Biegezugbewehrung eines Stahlbetonbalkens', Deutscher Ausschuss für
Stahlbeton, Heft 210, 83 p., Berlin, Germany
1 64/1 59.3 3.48 0.37 0 F CR 102 102 1.00
2 65/1A 50.5 3.48 0.37 0 L SY 140 130 1.08
3 65/1B 50.5 3.48 0.37 0 F CR 104 104 1.00
4 65/2A 56.3 3.48 0.37 0 F CR 93 92 1.01
5 65/2B 56.3 3.48 0.80 0 F CR 103 96 1.07
6 65/3A 48.2 3.48 0.37 0 F CR 92 80 1.15
7 65/3B 48.2 3.48 0.80 0 F CR 112 98 1.14
Average: 1.06
CoV: 0.06

Notation:
fc: concrete compressive strength measured on a cylinder;
ρ: longitudinal reinforcement ration;
ρw: transversal reinforcement ration;
Qtest: measured ultimate load;
QEPSF: calculated ultimate load.

Failure modes:
F: Flexural failure;
V: Shear failure;
L: Local failure.
Failure subtype:
CR: Concrete crushing
SP: Concrete spalling
DT: Diagonal tension
SY: Reinforcement yielding
AS: Arch stability
A: Anchorage failure

A7
Appendix

1.2) Prestressed concrete members

fc ρ ρw σp,inf Failure Failure Qtest QEPSF Qtest
N° Name
[MPa] [%] [%] [MPa] mode Subtype [kN] [kN] QEPSF
Saqan E.I. and Frosch R.J. (2009): 'Influence of flexural reinforcement on shear strength of prestressed concrete
beams', ACI Structural Journal, Vol. 106, N° 1, pp. 60-68, USA
1 V-4-0 52.1 0 0 1.9 V DT 488 408 1.2
2 V-4-0.93 52.7 0 0 1.9 V DT 668 600 1.11
3 V-4-2.37 53.4 0 0 1.9 V DT 734 734 1.00
4 V-7-0 54.5 0 0 1.94 V DT 740 552 1.34
5 V-7-1.84 53.1 0 0 1.94 V DT 968 708 1.37
6 V-7-2.37 53.1 0 0 1.94 V DT 856 726 1.18
7 V-10-0 51.7 0 0 1.95 V DT 812 584 1.39
8 V-10-1.51 51.7 0 0 1.95 V DT 880 702 1.25
9 V-10-2.37 51.7 0 0 1.95 V DT 880 738 1.19
Average: 1.23
CoV: 0.10
Kaufman M.K. and Ramirez J.A. (1988): 'Re-evaluation of the Ultimate Shear Behavior of High-Strength
Concrete Prestressed I-Beams', ACI Structural Journal, Vol. 85, N° 3, pp. 295-303, USA
1 I-1 57.5 0 0.29 7.45 F SY 1094 942 1.16
2 I-2 57.5 0 0.24 7.45 V CR 1288 1130 1.14
3 I-3 57.7 0 0.33 7.66 V CR 890 890 1.00
4 I-4 57.7 0 0.24 7.76 V CR 978 908 1.08
5 II-1 62.7 0 0.33 7.94 V SY 1246 1328 0.94
6 II-2 62.7 0 0.33 7.79 F SY 1788 1604 1.11
Average: 1.07
CoV: 0.07
Kuchma D., Kim K.S., Nagle T.J., Sun S. and Hawkins N.M. (2008): 'Shear Tests on High-Strength
Prestressed Bulb-Tee Girders: Strengths and Key Observations', ACI Structural Journal, Vol. 105, N° 3, pp.
358-367, USA
1 G1E 83.4 0 0.55 7.7 V SY 4438 4228 1.05
2 G1W 83.4 0 0.55 7.7 V SY 5102 4954 1.03
3 G2E 86.9 0 0.93 8.5 V SY 5916 5100 1.16
4 G2W 86.9 0 0.93 8.5 V SY 6856 6014 1.14
5 G3E 109.6 0 0.82 9.6 V SY 6098 5594 1.09
6 G3W 109.6 0 0.82 9.6 V SY 6634 5870 1.13
7 G4E 112.4 0 1.70 9.6 V SY 7780 7554 1.03
8 G4W 112.4 0 1.70 9.6 V SY 7780 7780 1.00
9 G5E 122.7 0 0.18 6.4 V SY 3626 3454 1.05
10 G5W 122.7 0 0.18 6.4 V SY 2980 2638 1.13
11 G6E 87.6 0 0.85 10.4 V SY 5550 5138 1.08
12 G6W 87.6 0 0.85 9.1 V SY 4842 4484 1.08
13 G7E 86.2 0 0.82 10.4 V SY 5786 5076 1.14
14 G7W 86.2 0 0.82 10.4 V SY 6400 6336 1.01
15 G8E 91.7 0 0.82 9.9 V SY 5508 5738 0.96
16 G9E 66.2 0 1.57 8.5 V CR 5998 5504 1.09
17 G9W 66.2 0 1.57 8.5 V CR 5982 5808 1.03
18 G10E 73.1 0 1.14 8.9 V SY 6116 4932 1.24
19 G10W 73.1 0 1.14 8.9 V SY 7302 5618 1.30
Average: 1.09
CoV: 0.07
Rupf M. and Muttoni A. (2012): 'Schubversuche an vorgespannten Stahlbetonträgern mit ungenügender
Schubbewehrung', EPFL, IBETON, Test Report , 159 p., Lausanne, Switzerland
1 SR21 30.8 0 0.09 2.5 V SY 1197 1110 1.08
2 SR22 33.7 0 0.13 2.5 V CR 1377 1290 1.07
3 SR23 35.3 0 0.06 2.5 V SY 1092 1065 1.03
4 SR24 31.3 0 0.25 2.5 V CR 1737 1680 1.03

A8
 Summary of the online database

5 SR25 33.1 0 0.09 5 V CR 1452 1410 1.03

6 SR26 36.9 0 0.06 5 V SY 1371 1335 1.03
7 SR27 28.3 0 0.19 5 V CR 1818 1740 1.04
8 SR28 37.8 0 0.09 0 V SY 666 660 1.01
9 SR29 29.8 0 0.25 2.5 V CR 1755 1680 1.04
10 SR30 31.4 0 0.25 2.5 V CR 1743 1620 1.08
11 SR31 31.3 0 0.09 3 V DT 927 795 1.17
12 SR31B 31.3 0 0.09 3 V DT 909 795 1.14
13 SR32 35.2 0 0.09 0 V DT 519 525 0.99
Average: 1.06
CoV: 0.05
Fernández Ruiz M. and Muttoni A. (2008): 'Shear strength of thin-webbed post-tensioned beams', ACI
Structural Journal, Vol. 105, N° 3, pp. 308-317, USA
1 SH1 53.4 0 0.63 4.2 V CR 2980 3136 0.95
2 SH2 52.3 0 0.63 4.2 V CR 2520 2400 1.05
3 SH3 55.8 0 0.63 4.2 V CR 3060 3188 0.96
4 SH4a 49.5 0 0.63 4.2 V CR 2240 2434 0.92
5 SH4b 60.0 0 0.63 4.2 V CR 3340 3478 0.96
6 SH5 47.2 0 0.63 4.2 V CR 3320 3254 1.02
Average: 0.98
CoV: 0.05
Moore A.M. (2014): 'Shear Behavior of Spliced Post-Tensioned Girders', Faculty of the Graduate School, The
University of Texas at Austin, PhD Thesis, 250 p., Austin, USA
1 Tx62-1S 73.1 1.51 0.93 11.8 V CR 3056 3217 0.95
2 Tx62-2S 82.7 1.51 0.93 12.3 V CR 3332 3744 0.89
3 Tx62-2N 82.7 1.51 0.93 12.3 V CR 3630 3742 0.97
4 Tx62-3 80.7 1.51 0.93 0 V CR 4386 4024 1.09
5 Tx62-4S 95.8 1.51 1.40 12.5 V CR 3696 4062 0.91
6 Tx62-4N 93.8 1.51 1.40 12.5 V CR 3701 3738 0.99
7 Tx62-5S 86.2 1.51 0.31 12.5 V CR 3127 3191 0.98
8 Tx62-5N 86.2 1.51 0.31 12.5 V CR 3269 3302 0.99
9 Tx62-6S 85.5 1.52 1.14 13 V CR 4137 3831 1.08
10 Tx62-6N 91.0 1.52 1.14 13 V CR 4888 4887 1.00
11 Tx62-7S 84.1 1.52 1.14 13 V CR 5186 4630 1.12
Average: 1.00
CoV: 0.07
De Wilder K., Lava P., Debruyne D., Wang Y., De Roeck G. and Vandewalle L. (2015): 'Stress Field Based
Truss Model for Shear-Critical Prestressed Concrete Beams', The Institution of Structural Engineers, Vol. 3,
pp. 28–42, London, UK
1 B101 77.5 2.08 0.27 19.3 V DT 378 367 1.03
2 B102 77.5 2.08 0.27 19.3 V DT 322 309 1.04
3 B104 88.9 2.08 0.27 9.6 V DT 282 303 0.93
4 B105 88.9 2.08 0.27 9.6 V DT 251 255 0.98
5 B107 89.3 0.97 0.27 10.7 F CR 271 284 0.95
6 B108 89.3 0.97 0.27 10.7 F CR 214 221 0.97
Average: 0.98
CoV: 0.04
Leonhardt F., Koch R. and Rostasy F.S. (1973): 'Schubversuche an Spannbetonträgern', Deutscher Ausschuss
für Stahlbeton, Vol. 227, 179 p., Berlin, Germany
1 ILT1 30.4 0.35 1.01 4.3 V SY 1809 1690 1.07
2 ILT2 30.4 0.35 0.70 4.3 V SY 1565 1540 1.02
3 IILT1 33.6 0.35 1.01 4.3 v SY 1667 1552 1.07
Average: 1.05
CoV: 0.03
Büeler C. and Thoma K. (2010): 'Indirekt gelagerter Spannbetonträger', Lucerne University of Applied
Sciences and Arts, 61 p., Lucerne, Switzerland

A9
Appendix

1 LT1 34 0.17 0.25 2.7 F CR 635 630 1.01

2 LT2 34 0.4 0.25 2.7 F CR 863 860 1.00
Average: 1.01
CoV: 0.01

Notation:
fc: concrete compressive strength measured on a cylinder;
ρ: longitudinal reinforcement ration;
ρw: transversal reinforcement ration;
σp,inf : prestress after the initial losses due to wedge slippage;
Qtest: measured ultimate load;
QEPSF: calculated ultimate load.

Failure modes:
F: Flexural failure;
V: Shear failure;
L: Local failure.
Failure subtype:
CR: Concrete crushing
SP: Concrete spalling
DT: Diagonal tension
SY: Reinforcement yielding
AS: Arch stability
A: Anchorage failure

A10
 Reinforced concrete panels subjected to shear

2) Appendix 2 – Reinforced concrete panels subjected to shear

The following appendix gives geometrical and mechanical properties of 77 reinforced
concrete panels that were used to validate the mechanical model for estimating the effective
concrete compressive strength in Chapter 4 of this thesis. Both measured and estimated failure
loads of the panels are indicated, along with the basic statistical analysis of the results (the
average prediction and the coefficient of variation).
fc fyx fyz ρx ρz Failure τtest τmodel τtest
N° Name
[MPa] [MPa] [MPa] [%] [%] mode [MPa] [MPa] τmodel
Vecchio F.J. and Collins M.P. (1982): 'Response of reinforced concrete to in-plane shear and normal stresses',
Department of civil engineering, University of Toronto, N° 82-03, 332 p.
1 PV4 26.6 242 242 1.06 1.06 Y-xz 2.838 2.62 1.08
2 PV6 29.8 266 266 1.79 1.79 Y-xz 4.472 4.81 0.93
3 PV10 14.5 276 276 1.79 1.00 Y-z+C 3.968 3.731 1.06
4 PV11 15.6 235 235 1.79 1.31 Y-xz 3.561 3.633 0.98
5 PV12 16.0 469 269 1.79 0.45 Y-z+C 3.134 3.049 1.03
6 PV16 21.7 255 255 0.74 0.74 Y-xz 2.144 1.947 1.1
7 PV19 19.0 458 299 1.79 0.71 Y-z+C 3.955 4.012 0.99
8 PV20 19.6 460 297 1.79 0.89 Y-z+C 4.261 4.457 0.96
9 PV21 19.5 458 302 1.79 1.30 Y-z+C 5.029 5.462 0.92
10 PV22 19.6 458 420 1.79 1.52 C 6.07 7.114 0.85
11 PV23 20.5 518 518 1.79 1.79 C 8.876 8.086 1.1
12 PV25 19.2 466 466 1.79 1.79 C 9.125 7.583 1.2
13 PV27 20.5 442 442 1.79 1.79 C 6.35 7.935 0.8
14 PV28 19.0 483 483 1.79 1.79 C 5.611 6.56 0.86
1 PV4 26.6 242 242 1.06 1.06 Y-xz 2.838 2.62 1.08
2 PV6 29.8 266 266 1.79 1.79 Y-xz 4.472 4.81 0.93
3 PV10 14.5 276 276 1.79 1.00 Y-z+C 3.968 3.731 1.06
4 PV11 15.6 235 235 1.79 1.31 Y-xz 3.561 3.633 0.98
5 PV12 16.0 469 269 1.79 0.45 Y-z+C 3.134 3.049 1.03
Average: 0.99
CoV: 0.11
Vecchio F.J., Collins M.P. and Aspiotis J. (1994): 'High-Strength Concrete Elements Subjected to Shear', ACI
Structural Journal, Vol.91, N° 4, pp. 423-433, USA
1 PHS2 66.1 606 521 3.23 0.41 Y-z+C 6.662 5.21 1.28
2 PHS3 58.4 606 521 3.23 0.82 Y-z+C 8.098 7.917 1.02
3 PHS5 52.1 606 521 3.23 0.41 Y-z+C 4.805 3.955 1.21
4 PHS6 49.7 606 521 3.23 0.41 Y-z+C 7.617 7.795 0.98
5 PHS8 55.9 606 521 3.23 1.22 Y-z+C 10.837 9.908 1.09
6 PHS9 56 606 521 3.23 0.41 Y-z+C 9.162 7.868 1.16
7 PHS10 51.4 606 521 3.23 1.22 Y-z+C 8.248 7.822 1.05
Average: 1.12
CoV: 0.09
Marti P. and Meyboom J. (1992): 'Response of Prestressed Concrete Elements to In-Plane Shear Forces', ACI
Structural Journal, Vol.89, N° 5, pp. 503-514, USA
1 PP1 27 479 480 1.94 0.65 Y-z+C 4.95 4.794 1.03
Average: -
CoV: -
Vecchio F.J. and Chan C.C.L. (1990): 'Reinforced concrete membrane elements with perforations', ASCE
Journal of Structural Engineering, Vol. 116, Issue 9, pp. 2344-2360
1 PC1A 27.9 500 500 1.65 0.82 Y-z+C 5.61 6.045 0.93
2 PC4 24.9 260 260 1.65 0.82 Y-z+C 4.84 5.752 0.84
Average: 0.88

A11
Appendix

CoV: 0.05
Pang X.-B.D. and Hsu C.-T.T. (1995): 'Behavior of Reinforced Concrete Membrane Elements in Shear', ACI
Structural Journal, Vol.92, N° 6, pp. 665-667, USA
1 A2 41 462 462 1.2 1.20 Y-z+C 5.367 6.13 0.88
2 A3 42 446 446 1.78 1.78 Y-z+C 7.655 8.492 0.90
3 A4 42 469 469 2.97 2.97 C 11.31 14.079 0.80
4 B1 45 462 444 1.20 0.60 Y-xz 3.962 4.544 0.87
5 B2 44 446 462 1.78 1.20 Y-xz 6.125 7.246 0.85
6 B3 45 446 444 1.78 0.60 Y-xz 4.354 4.888 0.89
7 B4 45 469 444 2.97 0.600 Y-z+C 5.064 5.597 0.90
8 B5 43 469 462 2.97 1.2 Y-z+C 7.152 8.027 0.89
9 B6 43 469 446 2.97 1.78 Y-z+C 9.143 10.106 0.90
Average: 0.88
CoV: 0.04
Zhang L.-X. and Hsu T.T.C. (1998): 'Behavior and Analysis of 100 MPa Concrete Membrane Elements',
Journal of Structural Engineering, vol. 124 n° 1, pp. 24-34, USA
1 VA1 95.1 445 445 1.20 1.20 Y-z+C 6.156 6.19 0.99
2 VA2 98.2 409 409 2.40 2.4 Y-z+C 9.73 10.615 0.92
3 VA3 94.6 455 455 3.57 3.57 Y-z+C 15.08 16.914 0.89
4 VA4 103.1 470 470 5.23 5.23 C 21.42 21.357 1.00
5 VB1 98.2 409 445 2.40 1.20 Y-z+C 7.497 7.819 0.96
6 VB2 97.6 455 445 3.57 1.20 Y-z+C 9.137 8.94 1.02
7 VB3 102.3 470 445 5.96 1.20 Y-z+C 9.709 10.199 0.95
8 VB4 96.9 455 445 1.78 0.60 Y-xz 4.858 6.276 0.77
Average: 0.94
CoV: 0.08
Hsu C.-T.T. and Zhang L.-X. (1997): 'Nonlinear Analysis of Membrane Elements by Fixed-Angle Softened-
Truss Model', ACI Structural Journal, Vol.94, N° 5, pp. 483-491, USA
1 HB1 66.5 409 445 1.2 0.6 - 4.322 4.41 0.98
2 HB3 66.8 447 445 1.78 0.6 - 4.889 5.064 0.97
3 HB4 62.9 470 445 2.98 0.6 - 5.334 5.85 0.91
Average: 0.94
CoV: 0.01
Kirschner U. (1986): 'Investigating the behaviour of reinforced concrete shell elements', University of
Toronto, Department of Civil Engineering, PhD Thesis, 83 p., Toronto, Canada
1 SE1 42.5 492 479 2.91 0.98 Y-z+C 6.77 7.323 0.92
2 SE6 40 492 479 2.91 0.33 Y-z+C 3.755 3.963 0.95
Average: 0.94
CoV: 0.01
Watanabe F. and Muguruma H. (1989): 'Ultimate Strength and Deformations of RC Panel', Proceedings of the
Sessions Related to Structural Design, Analysis and Testing, ASCE Structural Congress, pp. 31-38
1 00R 28.15 310 310 0.86 0.86 Y-x+C 3.14 2.724 1.15
2 15R 28.15 310 310 0.86 0.86 Y-xz 3.15 2.724 1.16
3 30R 28.15 310 310 0.86 0.86 Y-xz 3.13 2.724 1.15
4 45R 28.15 310 310 0.86 0.86 Y-xz 3.42 2.724 1.26
5 00D 28.15 318 318 0.87 0.87 Y-x+C 2.97 2.831 1.05
6 30D 28.15 318 318 0.86 0.87 Y-xz 2.56 2.823 0.91
7 45D 28.15 318 318 0.87 0.87 Y-xz 2.84 2.831 1.00
8 00DI 31 294 294 1.39 1.39 Y-x+C 4.96 4.167 1.19
9 22.5DI 31 294 294 1.39 1.39 Y-xz 5.06 4.167 1.21
10 45DI 31 318 318 1.3 1.3 Y-xz 3.97 4.213 0.94
11 45DII 31 318 318 2.61 2.61 Y-xz 7.61 8.347 0.91
12 45PCI 30.4 1187 1187 0.77 0.77 Y-xz 7.78 9.264 0.84
13 45PCII 30.4 1187 1187 1.55 1.55 C 11.72 11.456 1.02
14 45PCIII 30.4 1187 1187 1.55 0.77 Y-z+C 9.44 9.512 0.99
15 45PCIV 45 1187 1187 1.55 0.77 Y-z+C 10.63 11.121 0.96

A12
 Reinforced concrete panels subjected to shear

Average: 1.05
CoV: 0.12
Kollegger J. and Mehlhorn G. (1990): 'Experimentelle Untersuchungen zur Bestimmung der Druckfestigkeit
des gerissenen Stahlbetons bei einer Querzugbeanspruchung', Deutscher Ausschuss für Stahlbeton, Wilhelm
Ernst & Sohn, Vol.413, 132 p., Berlin, Germany
1 PK02 19.4 660 660 1.07 1.07 C 9.12 8.393 1.09
2 PK04 20.2 660 660 1.07 1.07 C 8.91 8.747 1.02
3 PK07 20.9 660 660 1.07 1.07 C 9.04 9.058 1.00
4 EGE6F1 15.8 465 465 0.66 0.66 C 8.00 7.279 1.10
5 EGE6F2 16 465 465 0.66 0.66 C 8.10 7.372 1.10
6 EGE6F3 14.6 465 465 0.66 0.66 C 6.90 6.722 1.03
7 EGE6F4 17.1 465 465 0.66 0.66 C 7.20 7.686 0.94
8 EGE6F7 18.8 465 465 0.66 0.66 C 8.50 8.674 0.98
9 EGE6F8 13.1 465 465 0.66 0.66 C 7.35 6.025 1.22
10 EGE7F1 15.9 660 660 0.66 0.66 C 8.20 7.282 1.13
11 EGE7F2 15.3 660 660 0.66 0.66 C 7.70 7.004 1.10
12 EGE7F3 17.3 660 660 0.66 0.66 C 8.70 7.931 1.10
Average: 1.07
CoV: 0.07
Schäfer K., Schelling G. and Kuchler T. (1990): 'Druck und querzug in bewehrten betonelementen',
Deutscher Ausschuss für Stahlbeton, N° 408, pp. 5-85, Berlin, Germany
1 #2 25.8 582 582 3.14 3.14 C 11.775 9.943 1.18
2 #6 25.8 582 582 1.57 1.57 C 14.065 10.93 1.29
Average: 1.24
CoV: 0.04
André H.M.O. (1987): 'Toronto Kajima Study on Scale Effects in Reinforced Concrete Elements', University
of Toronto, Department of Civil Engineering, PhD Thesis, 267 p., Toronto, Canada
1 KP1 25.2 430 430 2.03 1.02 C 5.62 5.794 0.97
2 TP4A 24.9 450 450 2.03 2.03 C 8.72 9.199 0.95
Average: 0.96
CoV: 0.01

Notation:
fc: concrete compressive strength measured on a cylinder;
fyx: yield strength of the rebars in x (horizontal) direction;
fyz: yield strength of the rebars in z (vertical) direction;
ρx: reinforcement ration in x (horizontal) direction;
ρz: reinforcement ration in z (vertical) direction;
τtest: measured ultimate shear stress;
τmodel: calculated ultimate shear stress;

Failure modes:
C: Concrete crushing;
Y-xz: Yielding of the rebars in both directions;
Y-x: Yielding of the rebars in x (horizontal) direction;
Y-z: Yielding of the rebars in z (vertical) direction;

A13
 Curriculum vitae

Filip NIKETIĆ
Structural Engineer with experience in the advanced design and
assessment of reinforced, prestressed and post-tensioned
concrete members of complex geometry

INFO: Date of birth: 15.08.1987.

Nationality: Serbian
CONTACT: E-mail: [email protected]
PROFESSIONAL EXPERIENCE:
École Polytechnique Fédérale de Lausanne (EPFL)
2012-2017
Faculté Environnement Naturel, Architectural et Construit (ENAC)
Laboratoire de construction en béton - IBETON
- Assessing the strength of existing structural concrete members by means of rigid-plastic
and elastic-plastic stress field analysis, focusing on bridge structures
- Development of mechanical models for concrete compression softening
- Teaching assistant of the following courses:
Structures en béton, Structure et architecture, Structures I et II,
Dimensionnement et analyse des éléments à l'aide de champs de contraintes
University of Novi Sad
2011-2012
Faculty of Technical Sciences
- Teaching assistant of following courses:
Structural analysis II and Theory of plates and shells
PUBLICATIONS:
2015 Aurelio Muttoni, Miguel Fernández Ruiz, and Filip Niketić, “Design versus Assessment of
Concrete Structures Using Stress Fields and Strut-and-Tie Models”, ACI Structural Journal,
Vol. 112, N°5
2016 Aurelio Muttoni, Miguel Fernández Ruiz, Filip Niketić and Marie-Rose Backes,“Assessment
of Existing Structures Based on Elastic-Plastic Stress Fields and Modelling of Critical Details
and Investigation of the In-Plane Shear Transverse Bending Interaction”, FEDRO report
N°680
2017 Eckart Hars, Filip Niketić and Miguel Fernández Ruiz, “Response of RC panel accounting for
crack development and its interaction with rebars”, Magazine of Concrete Research, article
number MACR-D-17-00077R1
EDUCATION:
2006-2010 HONORS BACHELOR DEGREE IN CIVIL ENGINEERING
University of Novi Sad
Faculty of Technical Sciences
2010-2011 HONORS MASTER DEGREE IN STRUCTURE DESIGN
University of Novi Sad
Faculty of Technical Sciences
ACHIEVENTS:
- The Best Graduate Student of the Faculty of Technical Sciences in 2011
- The Best Graduate Student of the Civil Engineering Department in 2011
- Dositeja Certificate for the academic 2010/11
- Fund for Progress Encouragement of Gifted Students and Young Scientists and Artists Certificate in
the academic 2008/09 and 2009/10
- Reward for Outstanding Accomplishments in the academic 2006/07, 2007/08, 2008/09 and 2009/10
- Civil Engineering Department Plaque for the best student in the academic 2006/07, 2007/08, 2008/09
and 2009/10
LANGUAGES:
Serbian Native speaker
English C2 Level
French B1 Level