Trip Steels

Relations between microstructure and bendability on
TRIP-aided steels for automotive products

Delphine Reche
To cite this version:

Delphine Reche. Relations between microstructure and bendability on TRIP-aided steels for automo-
tive products. Engineering Sciences [physics]. École Nationale Supérieure des Mines de Paris, 2011.
English. �NNT : 2011ENMP0016�. �pastel-00973376�
HAL Id: pastel-00973376

https://pastel.hal.science/pastel-00973376
Submitted on 4 Apr 2014
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
École doctorale n° 432 : Science des Métiers de l’Ingénieur
Doctorat ParisTech
THÈSE
pour obtenir le grade de docteur délivré par
l’École nationale supérieure des mines de Paris

Spécialité “ Sciences et Génie des Matériaux ”
présentée et soutenue publiquement par
Delphine RECHE
Le 18 mars 2011
Relations entre microstructure et aptitude au pliage des aciers à

effet TRIP pour application automobile
Relations between microstructure and bendability on TRIP-aided

steels for automotive products
Thèse confidentielle jusqu’au 31 janvier 2014
Directeur de thèse : Anne-Françoise GOURGUES-LORENZON

Co-encadrement de la thèse : Thierry STUREL
Jury
T
Mme Sandrine THUILLIER, Professeur, Université de Bretagne Sud Président
M. Michel BRUNET, Professeur, INSA Lyon Rapporteur
H
M. Paul VAN HOUTTE, Professeur, Katholieke Universiteit Leuven, Belgique Rapporteur
Mme Véronique AUBIN, Professeur, Ecole Centrale Examinateur È
Mme Anne-Françoise GOURGUES-LORENZON, Professeur, Mines Paris Tech Examinateur
M. Thierry STUREL, Ingénieur de recherche, ArcelorMittal Global R&D Examinateur S
M. Jacques BESSON, Directeur de recherche CNRS, Mines Paris Tech Examinateur
E
MINES ParisTech
Centre des Matériaux
B.P. 87 - 91003 EVRY Cedex
Relations between microstructure and bendability on
TRIP-aided steels for automotive products
Abstract
In order to limit fuel consumption, automotive industries push the steelmakers to develop
thinner sheets with higher strength. Ultra High Strength Steels (UHSS) such as low alloy
TRIP (TRansformation Induced Plasticity)-aided steels enable to get complex shapes for
automotive parts. However, understanding the formability of these steels appears to be
complex and involves a detailed study of failure mechanisms during forming tests.
Therefore, links between microstructural features, in particularly banded structures, and
formability were investigated in order to better predict the bending ability of steels as a
function of their microstructure. In this study, four low alloy TRIP-aided steels exhibiting
different bending performances were characterized using both air and stretch bending tests.
From scanning electron and light microscopy observations, cracks initiate from the outer
surface or from just below in air-bending, but from the central segregation bands in stretch
bending. Fracture surfaces after bending tests are ductile and damage mainly appears by
ferrite/martensite interface decohesion and occasionally as microcracks within martensite
islands. An original procedure based on thickness-reduced specimens, with various
locations of the main hard band within the specimen thickness, was set up. It enabled to
propose a relationship between cracking, local thickness of hard band and local strain
during air-bending. In order to develop a local fracture criterion that could be used for any
bending test, stress and strain fields were computed by numerical simulation of the two
bending tests. For that purpose, an experimental database including tensile tests on smooth
and notched samples as well as shear tests was established. Material parameters of
constitutive equations, accounting for anisotropic elasto-plastic behaviour with mixed
hardening, were determined from this database. Finite element simulation of both bending
tests associated with a ductile damage criterion made it possible to satisfactorily predict
strain fields, bending load-displacement curves and fracture angle. The possibility of using
simplifying assumptions in the model (such as an isotropic flow criterion, pure isotropic
hardening, two-dimensional assumptions and simplified boundary conditions) is also
discussed.
Keywords: Ultra High Strength Steels, TRIP-aided steels, microstructure, damage

mechanisms, air-bending test, stretch bending test, finite-element simulation.
Relation entre microstructure et aptitude au pliage des
aciers à effet TRIP pour application automobile
Résumé
Pour limiter la consommation des véhicules, les industries automobiles demandent aux
aciéristes de développer des aciers de plus en plus fins avec des hautes résistances. Les
aciers très haute résistance (THR) comme les aciers TRIP (TRansformation Induced
Plasticity) permettent d’effectuer des pièces automobiles aux formes complexes.
Néanmoins, le compréhension des mécanismes de mise en forme de ces aciers n’est pas
simple et implique une étude détaillée des mécanismes de rupture apparaissant durant la
mise en forme. Ainsi, les liens entre les paramètres microstructuraux, en particulier les
structures en bandes, et la mise en forme ont été étudiés pour obtenir une meilleure
prédiction de la capacité en pliage des aciers en fonction de leur microstructure. Dans cette
étude, quatre aciers TRIP présentant des performances en pliage différentes ont été
caractérisés à la fois en pliage en V et en pliage sous traction. A partir d’observations au
microscope optique et à balayage, nous avons démontré que les fissures s’initiaient à partir
de la surface ou juste en dessous en pliage en V alors qu’elles s’initiaient au niveau de la
ségrégation centrale en pliage sous traction. Les surfaces de rupture après pliage sont
ductiles et l’endommagement apparait principalement par décohésion de l’interface
ferrite/martensite et occasionnellement par rupture des ilots martensitiques. Une procédure
originale basée sur des échantillons rectifiés présentant la bande de ségrégation à différents
endroits dans l’épaisseur a été établi. Ce travail a permis de proposer une relation entre
l’épaisseur d’une bande, son endommagement et sa déformation locale atteinte pendant un
test de pliage en V. Pour développer un critère de rupture pour ces tests de pliage, les
champs de contrainte et déformation ont été calculés par simulation numérique pour ces
deux tests. Pour ce faire, une base de données expérimentale incluant des tests de traction
sur éprouvettes lisses et entaillées et des tests de cisaillement a été établie. La loi de
comportement du matériau a été déterminée à partir de cette base expérimentale et présente
un comportement élasto-plastique anisotrope avec écrouissage mixte. La simulation
numérique des deux essais de pliage associée à un critère d’endommagement a permis de
prédire de manière satisfaisante les champs de déformation, les courbes force/déplacement
ainsi les angles de rupture. La possibilité d’utiliser des modèles simplifiés (tels qu’un
critère d’écoulement isotrope, un écrouissage purement isotrope, des calculs en 2D et des
conditions aux limites simplifiées) a également été discuté.
Mots clés: Aciers Très Haute Résistance, aciers TRIP, microstructure, mécanismes
d’endommagement, test de pliage en V, test de pliage sous traction, simulation numérique
par éléments-finis.
Remerciements
Au terme des ces trois années de thèse, le temps est venu de présenter mes remerciements à
toutes les personnes qui ont contribué à cette aventure scientifique et humaine. Sans leur
aide efficace et leur soutien précieux, ce manuscrit n’aurait pas abouti à ce qu’il est
aujourd’hui.
Je remercie vivement tous les membres de mon jury pour le temps consacré à la lecture de
mon manuscrit et pour leurs précieux conseils. Un grand merci à Paul Van Houtte et
Michel Brunet qui ont accepté d’être rapporteurs de ces travaux ainsi qu’à Sandrine
Thuillier et Véronique Aubin, les examinatrices de mon jury. Leurs remarques pertinentes
m’ont permis d’améliorer la qualité de ce travail.
Cette thèse a été menée en collaboration avec le Centre des Matériaux (CdM) de l’Ecole
des Mines de Paris et ArcelorMittal Global R&D. Ainsi, je tiens ainsi à remercier Esteban
Busso, directeur du CdM, de m’avoir accueillie dans cet environnement technique et
humain de qualité. Je remercie également Michel Babbit de m’avoir reçue au Centre Auto
ainsi que Messieurs Alain Carmet, Philippe Hein et Martin Munier qui m’ont permis
d’évoluer et de m’intégrer très rapidement au sein des équipes "MBIWS" puis "AUP²".
Pour reprendre un ordre plus chronologique, je suis très reconnaissante à Isabelle Boul
d’avoir lancé cette thèse à la suite de mon stage de fin d’études, de m’avoir fait confiance
et de m’avoir donné le goût de la recherche et plus particulièrement celui de la mise en
forme.
Je tiens maintenant à remercier sincèrement Anne-Françoise Gourgues-Lorenzon qui a

dirigé cette thèse et m’a si bien accompagnée durant ces trois années. Un grand merci pour
ta disponibilité, ton écoute et implication malgré la distance, la rigueur et les méthodes de
travail que tu m’as apportées mais aussi pour toutes ces relectures méticuleuse d’articles ou
de chapitres en temps imposé.
Je remercie mon encadrant chez ArcelorMittal, Thierry Sturel, de m’avoir accompagnée

pendant cette thèse. Merci pour ta confiance et ton soutien au cours de cette aventure. Je te
remercie également pour ta disponibilité, tes conseils et l’autonomie que tu m’as laissée
tout au long de cette thèse.
Un grand merci également à Jacques Besson. Vous avez hérité d’une énième thésarde alors
que vous en aviez déjà beaucoup mais grâce à vous, j’ai pu découvrir les terres inconnues
de la simulation numérique. Avant cette thèse, je n’avais jamais utilisé les éléments finis,
alors merci pour votre patience, pour les explications sur ces modèles complexes et pour
toutes les réponses aux problèmes que j’ai pu rencontrer sur ZéBuLon.
Ayant passé un an au Centre des matériaux et deux chez ArcelorMittal, je vais d’abord
remercier mes collègues d’Evry puis ceux de Maizières, en espérant n’oublier personnes
(si c’est le cas, je leur présente par avance mes excuses).
Je souhaiterai tout d’abord remercier les chercheurs et techniciens qui m’ont aidé au cours
de cette thèse. Merci à toutes les personnes de l’atelier, Christophe, Jean-Pierre, René,
Georges, Michel pour l’usinage des éprouvettes ainsi que pour celui du montage du pliage
vii
sous traction. Merci Odile de m’avoir aidée dans mes recherches bibliographiques. Merci
également aux personnes de l’informatique, Greg et Olivier mais aussi aux personnes du
service administratif. Merci Daniel pour les essais de RX. Merci Anne et Maria de m’avoir
initiée au MEB et un grand merci également à Nicole pour les manips de nanoindentations
qui étaient loin d’être une partie de plaisir…Merci également à Abdennour de m’avoir
aidée à dessiner sous Inventor mais aussi de m’avoir formée à l’utilisation des différentes
machines du bocal. Merci également à M. Pineau pour les discussions sur des problèmes
métallurgiques ou autres.
Cette aventure aurait sans doute été moins agréable sans la présence de l’ensemble des
thésards et autres personnes du CdM qui m’ont accueillie au cours de cette année à Evry et
où entraide et bonne humeur ont permis un travail efficace. Merci Djamel pour tous les
renseignements sur la mécanique, sur Linux ou encore sur Zébulon.
Pika…que dire de la morphomath, merci d’avoir été présent à chaque moment, pour ton
écoute, ton soutien, tes conseils, et pour tout le reste. Merci également pour le roller dans
les rues parisiennes...
Clémence, ma grande amie, sans toi, Evry n’aurait surement pas été pareil. Merci pour tout
ce que tu as fait pour moi, pour ton écoute, tes encouragements. Merci également pour le
prêt de ton clic-clac car sans toi mes aller-retour à Evry n’auraient sans doute pas été aussi
sympa.
Florence, merci également pour ton écoute et tes encouragements surtout en fin de thèse. Je
te souhaite bon courage pour la suite dans ton nouveau travail et félicitation à la première
"Docteur" de notre promo.
Angel, merci de m’avoir fait découvrir "le pantalon" et merci pour l’hébergement après les
thèses sur Paris. Bonne continuation.
Julie, merci de m’avoir initié à Zébulon et m’avoir appris à jouer avec les lignes de codes
car au début ce n’était vraiment pas simple. Bon courage pour la suite et j’espère que tu es
venue à bout de ces fameux aciers TWIP.
Je vais maintenant remercier tous les autres thésards ou autres personnes avec qui j’ai pu
partager du travail, des rires, des sorties, etc…merci Yohann pour ta terrasse, merci Julien
d’avoir été présent pendant cette année et d’avoir fait le petit bout de chemin entre le CdM
et le 407. Un grand merci également à Tony, Guillaume, Thomas, Cédric, Bahram, Sophie,
Vlad, Guilhem, la grande Julie, Jianqiang, Nikolai, Fabrice, Steeve, Clara, Laurent,
Florine, Greg, Aurélie, Thomas….à qui je souhaite une grande réussite.
Je souhaiterais maintenant remercier mes collègues d’ArcelorMittal. Je vais tout d’abord

remercier les ingénieurs et techniciens qui m’ont formée et aidée pour les différents essais
que j’ai eu à effectuer…parmi eux, Grégory pour les essais de RX, Aurélien pour les
manips de sigmamétrie, Stéphane pour les manips avec le logiciel Aramis et Nathalie et
Rémi pour les observations métallographiques de la fin de thèse. Je souhaiterai également
remercier tout le service métallo de m’avoir formée, aidée et conseillée quand j’avais des
problèmes d’attaques ou d’observations. Merci également à toutes les personnes d’AUP²
avec qui j’ai pu travailler et en particulier le service mise en forme.
Merci à Germain, Max et Flo (mon parrain) de m’avoir appris différents tests de mise en
forme et de m’avoir accueillie dans votre bureau. Merci également à Nico, Rémi, Cyril,
Benoît et Dominique pour l’aide et les conseils que vous m’avez apportés. Xavier, je te
remercie également pour tes conseils sur la simulation numérique et ton aide sur des
problèmes pas toujours simples à résoudre.
Merci également à toutes les autres personnes des deux autres services (soudage et fatigue)
pour votre soutien et encouragements tout au long de cette thèse…Bastien, Michel, Rémi
viii
(bon courage pour la fin de ta thèse), Marina, Bernard, Adrien, Mathieu, Daniel, Nora,
Irina, les deux Dom, Agatha, Nathalie, Tarik, Jérôme, Stéphanie, Florent (bon courage
également pour cette dernière année de thèse), etc. Je remercie aussi Rachel et Danielle, les
secrétaires de mon service ainsi que Sabine Fogel pour les recherches bibliographiques.
Merci beaucoup Jonas, Pape et Perrine de m’avoir incluse dans vos réunions de projet et
pour l’aide précieuse que vous m’avez apportée sur ces sacrés aciers TRIP !
Un grand merci également à toi, Olivier, pour ton implication, ton dynamisme, tes idées
nouvelles et pour tous les conseils au cours de ces trois années.
Alain, je ne trouverai sans doute pas tous les mots mais je souhaite te remercier
sincèrement pour ta grande disponibilité que ce soit pendant mon stage ou pendant ma
thèse. Merci pour tous les conseils avisés ainsi que pour ton implication qui m’ont été
d’une aide précieuse.
Cette aventure n’aurait pas été possible sans certaines rencontres et notamment les
personnes qui sont plus que de simples collègues…je pense naturellement à des personnes
de la team NCIS…Julie, Nico, Savine, Mike, etc merci pour tout ce que vous m’avez
apporté au cours ce ces années, merci d’avoir été là et de m’avoir remonté le moral dans
les moments de doute. Je vous souhaite bonne continuation (une meilleure santé) et une
belle réussite pour la suite.
Je souhaiterais maintenant remercier mes coéquipiers de volley-ball et amis…merci à vous
toutes et tous. Un grand merci à Sophie, Anne, Claire, Pat, Manu, Jean, DD, Sam, Julien, et
Rémi pour tout ce que vous avez fait. Un petit message pour toi Sophie, continue comme
ça, la vue est belle en haut de cette montagne. Grâce à vous, les séances de volley m’ont
permis de décompresser et même si j’avais des difficultés à dépasser le filet, j’ai toujours
beaucoup aimé jouer avec vous…que ce soit le mardi soir ou en matchs. Merci également à
vous tous pour les sorties, les rires, la danse et pour les encouragements tout au long de
cette thèse.
Merci également à tous mes amis de prépa, de Montpellier ou d’ailleurs pour ces
retrouvailles au cours de ces trois ans et pour votre soutien.
Enfin, je souhaiterais remercier toutes les personnes qui ont fait le déplacement à Paris le
jour de ma soutenance.
Mes dernières lignes iront à ma famille qui m’a toujours soutenue. Merci à mes parents et
ma sœur d’avoir toujours cru en moi et de m’avoir apporté un grand réconfort. Sabine,
crois en toi et en tes rêves…tu pourras faire des choses magnifiques alors garde ton cap et
continue comme ça. Bon courage pour ces nouvelles aventures qui t’attendent.
Enfin mes derniers mots seront pour toi, Stéphane. Je te remercie tendrement de m’épauler
et de m’aider tous les jours depuis six ans. Merci pour ton amour et ton soutien sans pareil.
Sans toi, je ne serai sûrement pas en train de terminer ce manuscrit. Je te souhaite bon
courage pour tes nouvelles aventures en terre inconnue…
ix
x
Table of content
INTRODUCTION ................................................................................................................ 1
CHAPTER 1 ....................................................................................................................... 11
I. Literature survey ........................................................................................................... 13

I.A. Bendability of sheet metals ...................................................................................... 13
I.B. TRIP steels................................................................................................................ 26
I.C. Banded structures formation..................................................................................... 30
II. Microstructural characterization of the four TRIP-aided steels studied................ 36

II.A. Chemical composition and phase morphology ....................................................... 36
II.B. Determination of the retained austenite content...................................................... 40
II.C. Nanohardness of ferrite matrix and secondary phases ............................................ 43
II.D. Summary ................................................................................................................. 44
III. Mechanical characterization of the four TRIP-aided steels studied ...................... 46

III.A. Tensile properties at room temperature ................................................................. 46
III.B. Air-bending properties ........................................................................................... 47
III.C. Stretch bending properties ..................................................................................... 49
III.D. Summary................................................................................................................ 51
IV. Summary of Chapter 1 ............................................................................................... 51
V. Choice of the materials and approach used in the present study............................. 52

V.A. Choice of the materials ........................................................................................... 52
V.B. Outline of the next three chapters ........................................................................... 52
CHAPTER 2 ....................................................................................................................... 55
I. Global analysis of the air-bending test......................................................................... 57
II. Complementary data for model predictions .............................................................. 81

II.A. Experimental database ............................................................................................ 81
II.B. Comparison with predictions by other models assuming either Von Mises isotropic
yield criterion or isotropic hardening rule ....................................................................... 83
III. Influence of simulation conditions of the air-bending test ...................................... 90

III.A. Influence of meshing conditions............................................................................ 90
III.B. Influence of the boundary conditions .................................................................... 92
III.C. Influence of the constitutive equations on the model predictions for the air-
bending test...................................................................................................................... 94
III.D. Summary: influence of modelling conditions on air-bending test predictions...... 96
IV. Summary of Chapter 2 ............................................................................................... 97
xii
CHAPTER 3 ..................................................................................................................... 101
I. Experimental characterization of damage mechanisms........................................... 103

I.A. Fractographic observations after air and stretch bending tests............................... 103
I.B. Damage development during air and stretch bending tests .................................... 106
I.C. Investigation of damage development in segregated bands using tensile specimens
....................................................................................................................................... 111
I.D. Summary of section I ............................................................................................. 116
II. Influence of segregated bands on bendability.......................................................... 117

II.A. Investigation of damage mechanisms in air-bending............................................ 117
II.B. Complementary observations................................................................................ 118
III. Summary of Chapter 3 ............................................................................................. 130
CHAPTER 4 ..................................................................................................................... 133
I. Mechanical analysis of the air and stretch bending tests ......................................... 135
II. Influence of simulation conditions on the stretch-bending test.............................. 157

II.A. Influence of meshing conditions ........................................................................... 157
II.B. Influence of boundary conditions.......................................................................... 159
II.C. Summary: influence of modelling conditions on stretch-bending test predictions162
III. Parametric study: effect of the location of the hard band on failure during air-
bending ............................................................................................................................. 162
IV. Summary of Chapter 4 ............................................................................................. 166
GENERAL CONCLUSIONS AND PROPOSAL FOR FUTURE WORK................ 169
BIBLIOGRAPHY............................................................................................................ 181
APPENDIX ...................................................................................................................... 195
Appendix 1: Microstructural characterization procedures ........................................ 196
Appendix 2: Mechanical characterization procedures ................................................ 200
xiii
Introduction
Ultra High Strength Steels (UHSS) for

automotive application
In the Kyoto protocol, it was decided to reduce

greenhouse gases emissions. Private cars are responsible
for more than 10% of CO2 emissions in the European
Union.
For that purpose, European, Japanese and Korean
automotive industries engage themselves to reduce CO2
emissions down to 120g/km before 2012.
Ultra High Strength Steels (UHSS) are thus developed
and used to decrease vehicle weight but formability
issues can be encountered with these new steels.
1
Introduction
Ultra High Strength Steels in automotive industry
Current environmental concerns require that greenhouse gases and CO2 emissions have to
be decreased. Therefore, European standards impose to the automotive industry a reduction
of vehicle fuel consumption and CO2 emissions. To achieve these requirements,
automotive industries have to use thinner sheets with higher strength and equivalent or
improved functional properties.
In this industrial context, steelworkers develop and propose nowadays thin sheets as Ultra
High Strength Steels (UHSS) which are very advantageous to reduce the vehicle weight by
keeping a high strength. They are commonly used in automotive body in white (BIW) to
make safety and structural parts such as bumper reinforcements and intrusion beams
(Figure 1).
Figure 1. Typical applications of high strength steels for crash resistance
UHSS make reference to three main steel families:
- Dual phase (DP) steels: composed of hard phases (martensite or martensite / bainite)
dispersed in a soft pure ferrite matrix. The combination of high strength and ductility and
the capacity for strain hardening lead to excellent fatigue properties and good energy
absorption characteristics, making these steels suitable for structural and reinforcement
components, such as longitudinal beams and cross-members.
- TRIP (TRansformation Induced Plasticity) steels: composed of a ductile ferrite matrix

containing small islands of hard bainite and retained austenite. Transformation of the
austenite into martensite during deformation causes significant strain hardening and delays
the onset of necking, providing high ductility.
- Multiphase (MP) steels: with extremely fine microstructures, consisting of ferrite / bainite
or precipitation-hardened bainite or mixture of bainite and martensite. They have excellent
combination of high strength and ductility, high YS (Yield Strength), good formability and
good resistance spot weldability. These grades are particularly suited for automobile safety
components requiring good impact resistance, such as bumper beams and door
reinforcements.
2
Introduction
The main quality of these steels is to offer a good compromise between strength and
ductility (Figure 2). In this graph, it is clearly seen that TRIP steels offer the best
compromise between the tensile strength and the fracture elongation. Thus, these steels are
potentially very attractive for the automobile industry.
Indeed, they differ from conventional steels by their remarkable combination of high
strength and ductility, resulting from their particular microstructure. The strain hardening
capacity of TRIP steels is considerable, ensuring efficient strain redistribution during
forming and hence good drawability, combined with very high yield strength in the
finished component.
Due to their high energy absorption capacity and good fatigue strength, TRIP steels are
particularly used for structural and safety components such as longitudinal beams, mid-
post reinforcements, etc.
Figure 2. Ultimate tensile strength and fracture elongation for main steel product families
Metallurgical route for TRIP steels production

Figure 3 illustrates the metallurgical route for TRIP steels processing. After the
steelmaking step, the first step corresponds to the continuous casting. Then, from the slab
state, there are 3 main steps: first the hot rolling, secondly the coiling and finally the cold
rolling. After cold rolling, the material is significantly strain hardened and the annealing
cycle (Figure 4) is designed to recrystallize the grain and produce a TRIP grade that meets
the specifications targeted by the final customer. This step is done so as to restore enough
ductility for the automotive part forming. This annealing cycle corresponds to the heat
treatment of the TRIP studied (with a galvanization).
During heating and soaking between Ac1 and Ac3, the microstructure recrystallizes from
initial ferrite+pearlite microstructure yielding at the end of the soaking a microstructure
made of approximately 60% of ferrite and 40% of austenite. Such distribution depends on
the soaking temperature. The higher the latter, the more initial austenite is formed.
Then, a slight cooling from 30 to 60°C/s is performed during which nucleation, growth and
coarsening of ferrite appears. This cooling shall be fast enough to avoid pearlite formation.
In order to stabilize austenite phase thanks to its carbon enrichment, a isothermal holding
in the bainitic region is carried out. In addition, austenite is transformed into bainite during
this holding. Note that this bainite formed is carbide-free due to the presence of Al and Si.
3
Introduction
After this step, the steel is hot dip galvanized (liquid Zn at 460°C) before going through a
furnace in order to form the galvannealed GA coating (Zn-Fe alloy).
The final cooling can lead to a transformation of the austenite into martensite especially in
the mid-thickness area which is enriched in C and Mn. At the end of annealing, retained
austenite is still present (between 8 and 20%).
Figure 3. Production of TRIP steels
Ac3 Soaking at intercritical temperature

Temperature (°C)
Ac1
Slight Galvanizing
cooling
Heating
Final
Holding in the cooling
bainitic region
Time (s)
Figure 4. TRIP steels heat treatment
Sheet formability by stamping

In automotive industry, most of parts are stamped to be given their final shape. Stamping
process differs according to the part shape and several stampings (less or more complex)
are sometimes necessary to obtain the final automotive part.
However, on new TRIP steel grades with increasing mechanical properties, cracking can
occur on radii during stamping (Figure 5). Connecting steel formability with standard
mechanical properties seems insufficient and the significance of many relationships is not
obvious. It is thus necessary to take the microstructure of these materials into account to
study in more detail the bendability of these steels.
4
Introduction
Figure 5. Bumper beam with a large crack along the radius after stamping
Scope of the research

To enlarge the use of UHSS in automotive body as structural parts, good knowledge and
experience must be developed regarding their formability.
During the development of these products, every limit in terms of in-use properties has to
be known and characterizations are necessary to help and lead metallurgical development
of new grades. Therefore, understanding the formability of UHSS for automotive
applications involves a detailed study of failure mechanisms during forming tests.
The aim of this work is to study the relationships between the microstructure and the
bendability of UHSS.
TRIP steels can generally be stamped more easily than other UHSS. In addition, these
steels are developed to be able to perform more complex automotive parts than it is
possible with DP steels for example. Therefore, it is necessary to get deeper information
and to study the bendability of TRIP steels in more detail. In this study, four TRIP-aided
steels exhibiting various bending performances and various microstructures were used. For
this work, two bending tests were considered: air-bending and stretch bending.
The main objectives of this project are:

- To investigate the link between microstructure and damage during bending.
- To quantitatively investigate, in the particular case of the low alloy TRIP-aided steels, the
main microstructural parameter controlling failure in bending conditions.
- To build a design tool enabling metallurgical engineers to rather easily test new ways of
metallurgical improvement of UHSS regarding bending performances.
This manuscript is divided into four chapters.
The first chapter is devoted to a literature survey on both bendability, TRIP steels and the
origin of the micro and macrosegregations. Then, using scanning electron and light
microscopy, a first microstructural characterization of the four studied materials is given.
Finally, based on tensile and bending tests, a mechanical analysis of the four steels is
proposed.
After this state-of-the-art survey and after a short presentation of the four TRIP-aided
steels studied, the approach of this study is explained in more detail. The outline of the
three other chapters will thus be given at the end of the first chapter.
5
Introduction
Aciers très haute résistance pour application automobile
Les exigences environnementales actuelles requièrent que les émissions de gaz à effet de
serre et de CO2 soient réduites. A cet effet, des normes européennes imposent aux
constructeurs automobiles une réduction de la consommation des véhicules ainsi qu’une
diminution des émissions de CO2. Pour atteindre ces exigences, les industries automobiles
doivent utiliser des tôles plus fines avec des résistances plus élevées et des propriétés
d’usage équivalentes ou améliorées. C’est dans ce contexte industriel que les aciéristes
développent et proposent des aciers Très Haute Résistance (THR) qui ont l’avantage de
réduire le poids des véhicules tout en gardant une haute résistance. Ces aciers sont
couramment utilisés dans la caisse en blanc (squelette structurel de la voiture) pour les
pièces de structure et de sécurité telles que les poutres de renfort ou les poutres de pare-
choc (Figure 1).
Figure 1. Application des aciers très haute résistance
Les aciers très haute résistance regroupent trois grandes familles :
- Dual Phase (DP): aciers composés de phases dures (martensite ou martensite/bainite)

dispersées dans une matrice ductile de ferrite. La combinaison haute résistance/ductilité et
la forte capacité d’écrouissage font de ces aciers des candidats très fiables pour des pièces
de structures ou de renforts tels que les longerons ou les traverses.
- TRIP (TRansformation Induced Plasticity): aciers composés d’une matrice ductile de

ferrite contenant des petits îlots de bainite et d’austénite résiduelle. Ces aciers offrent un
potentiel de durcissement supplémentaire par rapport aux DP grâce à la possibilité
d’activer la transformation martensitique de l’austénite au cours de la déformation. Leur
grande ductilité en font également des aciers de choix pour les pièces de structures ou de
sécurité.
- Multiphase (MP): ces aciers ont une structure ferrito-bainitique très fine ou une structure
bainitique durcie par des précipités ou alors une structure bainite-martensite. Ces aciers
ont un bon compromis résistance/ductilité, une haute limite élastique “Re“, de bonnes
caractéristiques de formage et une très bonne soudabilité par points. Ces aciers sont
utilisés pour des pièces automobiles de sécurité destinées à la résistance aux chocs tels que
longerons ou les renforts de portières.
6
Introduction
La principale qualité des ces aciers est d’offrir un bon compromis entre résistance et
ductilité (Figure 2). Dans ce graphique, nous pouvons constater que les aciers TRIP
offrent un meilleur compromis entre résistance en traction et allongement à rupture. C’est
pour cette raison que ces aciers sont très demandés par l’industrie automobile.
En effet, ils diffèrent des aciers conventionnels de part leur remarquable combinaison
résistance/ductilité, résultant de leurs microstructures. La capacité d’écrouissage des
TRIP est considérable, assurant une distribution de la déformation efficace pendant leur
mise en forme. En raison de leur forte capacité d’absorption des chocs et de leur bonne
résistance à la fatigue, les aciers TRIP sont particulièrement utilisés pour des pièces de
structures et de sécurité tels que les traverses, les longerons, les renforts, etc.
Figure 2. Résistance en traction en fonction de l’allongement à rupture pour différentes familles d’aciers
Route métallurgique des aciers TRIP

La figure 3 illustre la route métallurgique des aciers TRIP. Après la coulée continue, la
brame va suivre trois étapes : le laminage à chaud, le bobinage et enfin le laminage à
froid. Après le laminage à froid, le matériau s’est durci donc le recuit (Figure 4) va
permettre de recristalliser les grains et de produire des aciers TRIP ayant les propriétés
demandées par les clients. Cette étape permet de restaurer une certaine ductilité
nécessaire pour la mise en forme des tôles. Le cycle de recuit illustré à la Figure 4
correspond à celui des TRIP étudiés (avec un passage à la galvanisation). Pendant le
chauffage et le maintien entre Ac1 et Ac3, la microstructure initialement ferrito/perlitique
est recristallisée pour atteindre à la fin du maintien une microstructure composée
d’environ 60 % de ferrite et 40 % d’austénite. Cette distribution dépend de la température
de maintien et plus elle est élevée, plus la quantité d’austénite formée est importante.
Cette étape est suivi par un refroidissement entre 30 et 60°C/s assurant germination,
croissance et grossissement de la ferrite. Ce refroidissement doit être assez rapide pour
éviter la formation de perlite.
Pour stabiliser l’austénite grâce à son enrichissement en carbone, un maintien isotherme
dans la région bainitique est effectué. Lors de cette étape, une partie de l’austénite est
transformée en bainite. Il faut noter qu’en présence d’aluminium et de silicium, la bainite
formée est sans Fe3C.
7
Introduction
Après cette étape, l’acier est trempé dans un bain de Zn à 460°C avant d’entrer dans un
four pour former le revêtement GA (alliage Zn-Fe).
Le refroidissement final entraîne une transformation partielle de l’austénite en martensite,
surtout à mi-épaisseur qui est particulièrement enrichie en C et Mn. A la fin de ce recuit,
l’austénite restante est appelée austénite résiduelle et représente 8 à 20 % de l’acier.
Refroidissement
Coulée et bobinage Laminage à Recuit
froid continu Skin-pass
continue
Produit laminé à Produit laminé

chaud à froid
Figure 3. Production des aciers TRIP
Ac3 Maintien intercritique

Temperature (°C)
Ac1
Refroidissement
Galvanisation
Chauffage
Refroidissement
Maintien dans la final
région bainitique
Time (s)
Temps (s)
Figure 4. Traitement thermique des aciers TRIP
Mise en forme des tôles par emboutissage

Dans l’industrie automobile, la plupart des pièces sont embouties pour leur donner leur
forme finale. En fonction de la pièce désirée, le procédé d’emboutissage varie et il est
parfois nécessaire d’effectuer plusieurs emboutissages pour obtenir des pièces
automobiles avec des formes plus ou moins complexes.
Cependant, sur les nouveaux grades d’aciers TRIP ayant des propriétés mécaniques de
plus en plus élevées, des fissures apparaissent sur les rayons des pièces embouties (Figure
5). Relier les propriétés de mise en forme des aciers avec leurs propriétés mécaniques
classiques semblent maintenant insuffisant.
Il est donc nécessaire de prendre en compte la microstructure de ces aciers pour étudier
plus précisément leur comportement en pliage.
8
Introduction
Figure 5. Poutre de pare-choc fissurée au niveau d’un pli après emboutissage
Problématique et démarche de l’étude

Pour élargir l’utilisation des aciers THR pour les pièces de structure, les limites des
propriétés en mise en forme doivent être connues et une caractérisation est nécessaire
pour aider au développement de ces nouveaux grades d’aciers. Ainsi, la compréhension de
la mise en forme des aciers THR pour application automobile implique une étude détaillée
des mécanismes d’endommagement pendant les essais de mise en forme.
Le but de ce travail est donc d’étudier les relations entre microstructure et capacité au
pliage des aciers THR.
Les aciers TRIP peuvent généralement être emboutis plus facilement que les autres aciers.
De plus, ils sont développés dans le but d’effectuer des pièces aux formes très complexes.
Il est ainsi nécessaire d’étudier plus particulièrement ces aciers. Dans cette étude, quatre
aciers TRIP présentant des performances en pliage différentes et des microstructures
variées ont été utilisés. Deux tests de pliage ont particulièrement été étudiés : le pliage en
V et le pliage sous traction.
Les principaux objectifs de ce projet sont:

- Effectuer les liens entre microstructure et endommagement pendant les tests de
pliage.
- Etudier d’une manière quantitative, dans le cas particulier des aciers TRIP, le
principal paramètre microstructural qui contrôle la fissuration en pliage.
- Proposer un outil permettant aux ingénieurs métallurgistes de tester d’une manière
simple des voies d’amélioration métallurgique des aciers THR vis à vis de leur
performance en pliage.
Ce manuscrit est constitué de quatre chapitres:
Le premier chapitre est consacré à l’état de l’art concernant à la fois le pliage, les aciers
TRIP ainsi que l’origine des micro et macroségrégations. Dans une deuxième étape, une
caractérisation à la fois microstructurale et mécanique des quatre aciers étudiés est
effectuée.
Après cet état de l’art et une présentation rapide des quatre matériaux de l’étude, il sera
possible d’expliquer plus en détail l’approche de cette étude. La démarche des trois autres
chapitres sera donnée à la fin du premier chapitre.
9
Chapter 1. Characterization of the bendability of TRIP steels
Chapter 1
Characterization of the bendability of

TRIP steels
In the following part, a literature survey on bending tests

and a short presentation of TRIP steels was given.
Four TRIP steels were chosen in this study to investigate
and characterize their bendability.
No clear correlation between bendability and both usual
mechanical properties and microstructural features have
been observed.
Therefore, it is necessary to develop a more detailed
mechanical / microstructural analysis of failure in
bending.
11
Table of contents
I. LITERATURE SURVEY .................................................................................... 13
I.A. Bendability of sheet metals .......................................................................................... 13
I.B. TRIP steels.................................................................................................................... 26
I.C. Banded structures formation......................................................................................... 30
II. MICROSTRUCTURAL CHARACTERIZATION OF THE FOUR TRIP-AIDED

STEELS STUDIED............................................................................................... 36
II.A. Chemical composition and phase morphology ........................................................... 36
II.B. Determination of the retained austenite content.......................................................... 40
II.C. Nanohardness of ferrite matrix and secondary phases ................................................ 43
II.D. Summary ..................................................................................................................... 44
III. MECHANICAL CHARACTERIZATION OF THE FOUR TRIP-AIDED STEELS

STUDIED.............................................................................................................. 46
III.A. Tensile properties at room temperature ..................................................................... 46
III.B. V bending properties.................................................................................................. 47
III.C. Stretch bending properties ......................................................................................... 49
III.D. Summary.................................................................................................................... 51
IV. SUMMARY OF CHAPTER 1 .......................................................................... 51
V. CHOICE OF THE MATERIALS AND SCOPE OF THE STUDY...................... 52
V.A. Choice of the materials ............................................................................................... 52
V.B. Outline of three other chapters.................................................................................... 52
12
I. Literature survey
I.A. Bendability of sheet metals
The formability of sheet metals often refers to the limit of strain or stress states up to which
the forming process is accomplished successfully.
It is very difficult, however, to find suitable strain limit for sheet bending operations. The
bending limit could be defined as the amount of strain that initiates a crack at the outer
surface of the bending zone, from which a minimum bending radius is deduced. It is
affected by many testing and material parameters, such as the tool geometry, bending
procedure, material properties, sheet thickness and the surface or edge condition of the
sheet. Note that information provided in this part is the more precise as possible. When
information is missing, it is mean that all data were not mentioned in the various studies.
Bending tests can be classified in three categories: simple bending, stretch bending and
draw-bending. The two last categories correspond to bending under tension and differ from
the first test where bending is applied without tension. It is important to distinguish pure
bending by appliance of a torque on both sides from 3 points bending which induces both
pure bending and a little tension.
1. Impact of specimen dimension and mode of preparation
One of the most significant parameters for this type of test is the surface state of cut edges.
Indeed, the blanking process creates a section with three areas: a rollover (edge rounding),
a sheared zone and a fracture zone (including the burr), Figure I.1.
The last one is generally the most problematic as it may introduce lot of incipient damage
which will further develop under a stress field. Moreover, strain hardening due to cutting
will increase the hardness of the material over a significant depth, reducing locally
ductility.
Rollover
Sheared zone
Fracture zone
300µm Burr
Figure I.1 : Cut edge of a dual phase steel sheet (cross-section micro light optical micrograph) (Dalloz et al.,
2007).
To perform a bending test, it is necessary, before any other operation, to grind the surface
of the edges. Indeed, the bending test is considered as valid (no artefact due to specimen
preparation) only if edge effect is negligible. This means that a crack should start from the
middle of sample width and not from the edge.
In addition, the sample width has a non-negligible influence on the allowed minimal bend.
It is easier to bend samples with decreasing width because the necessary stress to bend the
13
sample will be lower. Moreover, a very wide sample leads to plane strain field whereas for
a narrow one, the stress state is closer to uniaxial tension.
A study done on a copper-alloyed material shows that the minimal bending radius can be
doubled if sample width increases from one to ten times the material thickness (Figure I.2).
This tendency becomes blurred for higher values of width. We can also imagine that this
dependency on specimen width should vary according to the ductility of material
(Determe, 1998).
Figure I.2: Minimum bending radius evolution according to sample width for copper alloys (CuFe2P) with
various values of thickness (Determe, 1998).
2. Impact of microstructure
The first cause of material damage is generally the presence of internal inhomogeneities.
For steels, we can distinguish various categories of defects, which are:
- Inclusions
- Microstructural heterogeneities.
Col (2010) showed that in presence of manganese sulphides (MnS), the bendability of a
steel sheet may be lowered due to the initiation of a crack at this type of inclusion (Figure
I.3). The material in this picture is a hot rolled mild steel of 1.5mm thickness.
Figure I.3. Crack due to a MnS on a wheel rim element (Col, 2010).
Such inhomogeneities can induce heterogeneous behaviour. Their location and orientation,
often elongated along the rolling direction, favour crack development along the transverse
direction. This can explain differences in bending behaviour noticed between rolling and
transverse directions (Figure I.4). Indeed, sheet bend along transverse direction has
generally a lower bendability. This effect was observed by Leu et al. (1997) where a poor
bendability was noticed on transverse bent specimens.
14
Figure I.4 : Bend orientation with respect to the coil (Determe, 1998).
Yamazaki et al. (1995) showed that the bendability of UHSS with a tensile strength higher
than 780MPa correlates better with microstructural homogeneity than with total
elongation. In their study, they evaluated the bendability of steels by placing a specimen
over a 90° V-block, forcing a punch of specified tip radius into the specimen, and
measuring the minimum bending radius at which the specimen did not crack. Before
bending, they performed hardness measurements on various sheets with various levels of
microstructural homogeneity. Rockwell hardness was measured at five points with 2mm
intervals from the surface of sheet specimens, and its standard deviation was taken as a
microstructural homogeneity index. The minimum bending radius does not correlate at all
with the total elongation in tension but fits quite well with the microstructural homogeneity
index (Figure I.5). In their study, they measured the strain distribution of the bend surface
and showed that higher strains were found for ferritic areas while lower strains were
measured for martensitic areas. In addition, cracks are initiated at the boundary between
high-strain portions and low-strain portions, i.e. between the hard and soft phases.
Figure I.5. Bending radius vs. total elongation and microstructural homogeneity index (Yamazaki et al.,
1995).
Nagataki et al. (1994) worked on bendability of UHSS with fully martensitic structure
having tensile strength ranging from 1100 to 1650 MPa and a thickness of 1.4mm. They
showed that the microstructure should be refined and homogenized for the improvement of
bendability.
From a few years, health and cleanliness of steels have been improved. This induces a
decreasing of inclusions into steels. Moreover with the increase use of UHSS, inclusions
still have an influence on the bendability but heterogeneities of microstructure (difference
between hard and soft phases in multiphase steels for example) seem to have the most
important role on the bendability of materials.
15
3. Relationships between bendability and other In Use Properties
Mechanical properties such as yield strength, tensile strength, strain hardening, reduction
of area in tensile test, etc can be tentatively used to characterize the formability of a sheet.
However, these values are not the most reliable parameters to predict the bendability of
sheets.
To investigate pure bending of anisotropic sheet metals, Leu et al. (1997) proposed a
simplified approach to identify the bending instability using the “maximum bending
moment” concept, with the incorporation of the normal anisotropic R-value (defined as the
ratio of the transverse plastic strain in the plane of the sheet to the plastic strain through the
thickness) and the strain hardening exponent “n”. The value of R is given by the following
averaging: R = (R0 + 2 R45 + R90) / 4 where R0, R45 and R90 are the Lankford coefficients
at 0°, 45° and 90° with respect to the rolling direction of the sheet.
They showed that the springback is almost proportional to the normal anisotropic value R
and decreases sharply with decreasing strain hardening exponent “n”. Concerning pure
bending, the authors have shown that the minimum bending radius is proportional to the
sheet thickness and decreases with the normal anisotropy “R” and the strain hardening
exponent “n”. However, this trend was not confirmed by Yamazaki et al. (1995) who
showed that the minimum bending radius was not correlated with the strain hardening
exponent. In this paper, sheet was bent on a 90° V-block contrary to Leu’s study where
pure bending was assumed.
Datsko et al. (1960) predicted the minimum bending radius for several materials
(magnesium, aluminium, 1018 steel, cast-iron and titanium), from the reduction of area
determined by a standard tensile test. They assumed that the fracture strain in the outer
fibre of a bending specimen equals the one in tensile test specimens. The material assumed
to be homogeneous, isotropic and the sheet was assumed to be bent in plane strain
conditions. In their paper, they determined several equations allowing linking the Rmin
(smallest radius that a material can withstand without cracking) to the Ra% (reduction of
area) with various location of the neutral fibre. In their study, three-point bending test was
used to determine the minimum bending angle as well as another bending method which
provided similar results.
In case where the neutral fibre coincides with the mid-thickness the following equation can
be used:
R 50
= −1 (1)
t Ra %
According to the authors, this equation is only valid for Ra% < 20.
When there is a shift in the neutral fibre, the equation (2) is chosen:
R 60
= −1 (2)
t Ra %
The Forming limit curve (FLC) can also be used to predict the bendability of a material.
For example, Sriram et al. (2003) performed stretch bending tests on various steels such as
skin panel steels (BH210, BH280, IF) and structural steels (HSLA350, DP600, TRIP600,
16
DP800, DP980) with thicknesses ranging from 0.6mm to 1.8mm. Strains were measured in
the failed samples both in the unsupported sidewall region and on the punch nose and
compared with the corresponding achievable limit strain determined by the empirical FLC.
These authors introduced a new element to define the stretch bending ability of UHSS
within a wide range of R/t ratios and introduced a new concept called “stretch bendability
index”. The purpose of the index is to isolate the stretching influence from coupled stretch-
bend condition. This index represents the ability of the material to withstand stretching in
the presence of severe bending.
The stretch bendability index is expressed as a ratio: A / B (Figure I.6). B is the limit strain
(from FLC) for a given strain path and A is the maximum major hoop strain in the
unsupported sidewall region for a given R/t ratio when the failure occurs at the punch nose.
The maximum achievable value of the stretch bendability index is 1.
Figure I.6. Strain path and stretch bendability index (SBI) for 0.93mm BH210 specimens (Sriram et al.,
2003).
In Figure I.7, it is noticed that the R/t value for a stretch bendability index (SBI) of 1
represents the critical R/t ratio (R/t)c for stretch bending for a given steel grade. For R/t
ratios below (R/t)c, bending severity limits achievable formability. For R/t ratios higher
than (R/t)c, the effect of bending on the achievable formability is minimal.
Figure I.7. SBI as a function of R/t ratio for some structural steels (Sriram et al., 2003).
17
4. Failure modes
a. Simple bending
In their work, Nagataki et al. (1994) observed fracture surfaces of specimens tempered at
400°C after bending with an inside radius close to 0 mm. They showed that fracture in
bending occurs by shear mode crack propagation followed by void nucleation. Steninger et
al. (1982) worked on steels with tensile strength ranging from 240 to 390 MPa and a
thickness of 5 mm. They performed V-bending tests on these specimens and observed
cracks on outer surfaces of the specimens. Fracture propagated along shear bands and
voids expanded preferentially along the shear direction.
Chien et al. (2004) reviewed two most commonly found failure modes: necking (which
involves void nucleation, growth and coalescence) and shear localization. The bendability
of aluminium sheets was investigated with the shear localization mode. For example, the
pictures of three samples bent with three different radii are shown in Figure I.8.
Note that there is a high indentation of the punch especially for specimen bent with lower
punches (radii of 0.3mm and 0.4mm).
Local necking
Figure I.8. Cross-sections of AA6111 specimens (aluminium sheet) with an initial thickness of 1mm. These
specimens were subjected to bending in a semi-guided wrap-bend test with bending radii of: (a) 0.6 mm, (b)
0.4 mm and (c) 0.3mm (Chien et al., 2004).
Vallellano et al. (2008) analysed the effect of strain gradient in stretch bending on the
failure of metal sheets. Depending on the severity on the gradient, two kinds of failure are
expected: a necking-controlled failure and a fracture-controlled failure. They showed that
the sheet failure is controlled by the ability of the less strained fibres (inner fibres) to neck.
This idea is consistent with the work reported by Tharrett and Stoughton (2003). In fact,
they observed necking on the sheet when the strain on the concave side of the sheet
reached the in-plane forming limit strain. A conservative failure criterion is to assume that
the sheet fails once the outer fibre fails. This criterion is obviously not new and has been
successfully checked in a variety of bending operations (Ragab and Saleh, 2005).
18
b. Stretch bending
Demeri (1981) performed stretch bending tests on several steels (AK steels with tensile
strength of 300MPa, HSLA-F50 and DP 80 with tensile strength of 680MPa) and on one
aluminium sheet (2036-T4) with thicknesses ranging from 0.5mm to 2.5mm in order to
determine the influence of material and geometrical variables on sheet metal formability.
Data obtained are curves relating the punch radius “R”, the sheet thickness “t” to the height
of the configuration at failure “H” (cf. Figure I.9 for AK steels).
For AK steel, fracture location depends on three factors: (i) punch radius, (ii) sheet
thickness and (iii) transverse constraints. The curves obtained for the various materials
could be divided into two regions:
i) predominantly stretching for values of (R/t) > 20

ii) predominantly bending for values of (R/t) < 20.
Indeed, for most of steel grades, failure moved from the punch contact region to the
unsupported sidewall with increasing the R/t ratio.
Figure I.9. (a) Stretch bent test, (b) Bendability curves for AK steel sheets (Demeri, 1981).
Damborg (1998) used another type of stretch bending test to determine the stretch
bendability of sheets. For that purpose, he carried out tests on two steels: HSLA (550 MPa,
1.5mm) and DQSK (Draw Quality Silicon Killed with 365MPa and 1.5mm thickness) and
one aluminium alloy (Al 6022-T4 with 330MPa and 0.9mm thickness) with various punch
radii and various tension loads (Figure I.10). Thanks to this test, he also noticed the two
possible fracture modes in stretch bending: bending or stretching fracture (Figure I.10).
19
Figure I.10. Schematics of a bending under tension test, illustrating the typical location of the bending and
stretching fracture (Damborg, 1998).
Sriram et al. (2003) found similar results in stretch bending (Figure I.11) except for some
DP steels where, even for for high R/t ratios, failure did not move from the punch nose to
the sidewall. In other words, DP steels were found to be more sensitive to bending fracture
(with respect to stretching fracture) than the other steels. According to the authors, the
different behaviour of DP steels is not clearly understood. They assumed that steel
processing, chemistry and distribution of martensite may influence its forming behaviour.
(a) (b)
Figure I.11. (a) Stretch bending device, (b) Height at failure as a function of R/t ratio for structural steels in
stretch bending (Sriram et al., 2003).
To summarize, in simple bending, cracks seem to initiate from the surface (on the bent
nose) and propagate along shear direction. This will be discussed in the following parts.
In addition, from the above mentioned results, the height at failure during a stretch bending
test can be expressed as a function of R/t. Moreover, several authors claimed that above a
critical ratio (R/t)c, failures move from the punch nose to the sidewall, which means that
above (R/t)c, stretching is predominant over bending.
20
5. Mechanical description of the bending tests
a. Analytical models
To simulate as well as possible what happens in material during bending, the stress and
strain distributions have to be known within the bent specimen.
During bending solicitation, strain goes from elastic to plastic modes until fracture. During
the elastic stage, stress repartition is homogeneous with:
- a zero value at neutral fibre
- a compression stress next to inner radius
- a tensile stress next to outer radius
Beranger et al. (1994) demonstrates that in pure bending, the neutral fibre is exactly at
mid-thickness (Figure I.12a). When tension appears in bending and in presence of friction,
this assumption is not true and thus the neutral fibre moves within the material due to
thickness reduction of sheet steel (Figure I.12b).
Figure I.12 : Stress distribution: a) pure bending b) bending + friction + tensile (Beranger et al., 1994).
In simple bending, without applied tension, and where the radius of curvature is more than
several times the sheet thickness, the neutral fibre approximately coincides with the mid-
thickness (Figure I.13) so that the axial strain ε can be given by the following equation:
y
ε ( y ) = ln(1 +
) (3)
R0
Where R0 corresponds to the radius at mid-thickness and y is the distance from the mid-
thickness.
Figure I.13: Strain and stress distribution for a material in pure bending that obeys the relation σ 1 = kε 1n
(Marciniak and Duncan, 1992).
21
In various studies (Chakrabarty et al., 2000; Chakrabarty et al., 2001, Livatyali et al., 2003;
Quach et al., 2004; Ragab et al., 2005), the true bending strain was defined as equation (3).
However, other authors developed more accurate equations to determine the bending
strain. For example, Zhang et al. (1998) used the following equation:
1 2 y −ε 0
ε = ε 0 + ln(1 + e ) (4)
2 R0
where ε0 is the strain at the neutral fibre. When the neutral fibre is undeformed, the value
ε0 becomes equal to zero. In this case, equations (3) and (4) can be approximated by the
same following value:
y
ε≈ (5)
R0
Wu et al. (2006) defined the bending strain at the outer fibre with the following equation:
⎛ t ⎞
⎜ ⎟
⎜ R0 ⎟
ε = ln⎜1 + ⎟ (6)
⎜ ⎛ t ⎞⎟
2⎜⎜1 + ⎟
⎜
⎝ ⎝ R0 ⎟⎠ ⎟⎠
Lemoine et al. (2009) compared the reliability of these formulas and showed that they are
all equivalent till a ratio thickness / bending radius equal to 0.1. For t/R > 0.1, significant
difference appear between these formulas (with y = t / 2, i.e. at the outer surface).
Analytical and semi-analytical models have been widely adopted in the study of bending
process (Leu, 1997; Zhang et al., 1998; Chakrabarty et al., 2000 and 2001; Quach et al.,
2004 and Ragab et al., 2005). Unfortunately, these models are generally used with
significant simplifications and simple constitutive equations laws (i.e. without damage) and
thus, the accuracy of the results are considerably affected. Therefore to avoid these
problems and to increase the accuracy of these models, finite-element simulation is needed.
b. Need of numerical simulation to predict the bendability of sheet
The understanding and development of bending mechanics are aimed at obtaining

important information for industrial applications: (i) springback prediction to determine
accurately the final dimensions of bent parts, (ii) bendability and fracture prediction, (iii)
estimation of the bending force reached during these tests and (iv) contact issues between
punch and sheet. As the fracture in bending is a localized phenomenon, it is intended to
model very accurately this test to access to strain and stress fields. Therefore, the numerical
simulation thanks to a finite-element method is generally used to model precisely bending
tests.
Only few numerical works can be found in literature concerning the modelling of bending
tests and prediction of fracture. Most finite-element simulations of bending tests focused
22
on the prediction of springback (Date et al., 1999; Gan and Wagoner, 2004). In order to
correctly predict bendability and fracture in bending tests, it is necessary to include a
damage criterion in the model and to select a suitable one, accurate knowledge of damage
mechanisms is essential. In the case of sheet-metal forming, a lot of authors studied the
mechanisms leading to final failure and proposed their own models.
The formability approaches can be classified in three main families: (i) Forming Limit
Diagrams (FLD) approaches which are based on strain fields and dedicated to sheet metal
forming (Keeler and Backofen, 1964; Goodwin, 1968) (ii) coupled damage approaches and
(iii) uncoupled damage approaches.
For many sheet forming processed, the FLD has been used for the evaluation of failure.
Laukonis and Ghosh (1978) and Graf and Hosford (1993) showed that FLD are only valid
with proportional strain path. However, for bending tests, the stress/strain state changes
continuously and so the strain path does not stay constant. In order to solve this issue, other
failure criterions were used.
Considering coupled damage approaches, damage evolution is considered within the
elastoplastic material law and fracture is then fulfilled when the damage variable reaches a
critical value. Gurson (1977) was one of those who develop a ductile damage macroscopic
constitutive law based on microvoid growth. Its model was largely used and extended in
the literature (Tvergaard, 1982; Tvergaard and Needleman, 1984). The Gurson model is
the first micromechanical model for ductile fracture which introduces a strong coupling
between deformation and damage. It is derived from an analysis similar to the one
performed by Rice & Tracey (1969) for an isolated void. Other scientists as Lemaitre
(1985) also developed damage formulation fully coupled with the behaviour laws of elasto-
plasticity.
Eventually, uncoupled damage approaches also named “fracture criterion” are used as soon
as the failure is fulfilled when the stress and/or the strain state reaches a critical value
To model forming tests including failure criterion, several finite-element analysis exist in
the literature (Table I.1). The main aim of these numerical simulations generally consisted
in the prediction of bendability, failure location and estimation of the stress and strain
fields. Note that most of the authors used an isotropic hardening law with either an
anisotropic yield criterion (Hill generally) or an isotropic criterion (Von Mises). According
to the authors who chose a mixed (isotropic+kinematic) hardening law, results were well
predicted and best results were obtained than with an isotropic hardening law.
23
Forming Constitutive Failure

Material Thick. 2D/3D Main results Reference
tests equation criterion
air
2024-O bending 2D: Isotropic No
Springback Wang et al.
aluminium 1.02mm in U or V- Plane hardening fracture
prediction 1993
alloy shape strain + Hill criterion
Extension
deep Isotropic Prediction of
of Gurson- Brunet et
mild steel 0.67mm drawing 3D hardening localized necking
Tvergaard al. 1996
test + Hill and damage
model
(i) isotropic
mild steel Cyclic 3- Best prediction
1.2mm (ii) kinematic No
& high point of the bendability Zhao and
and 3D (iii) mixed fracture
strength bending with mixed Lee 2002
1.4mm hardening criterion
steel test hardening
+ Von Mises
Prediction of
0.6% air Isotropic
Rice and damage Hambli et
carbon steel 3mm bending 3D hardening
Tracey evolution and F/d al. 2004
(780MPa) test + Von Mises
curve
wiping-
Isotropic
die Prediction of Bahloul et
HSLA 500 4mm 3D hardening Lemaitre
bending stress al. 2006
+ Von Mises
test
Saanouni
Mixed continuum
Orthotropic Hydrobul Prediction of and
3D hardening damage
thin sheet ging test fracture location Badreddine
+ Hill mechanics
2007
wiping- Mkaddem
Isotropic
die Prediction of and
HSLA 4mm 3D hardening Lemaitre
bending failure and load Bahloul
+ Von Mises
test 2007
Isotropic Prediction of Uthaisangs
galvanized Nakazima
1mm 3D hardening GTN local crack uk et al.
BH220 test
+ Von Mises initiation 2008
Isotropic No Prediction of
draw Kim et al.
DP980 3D hardening fracture failure and F/d
bend test 2009
+ Von Mises criterion curves
Prediction of
Isotropic
Nakazima Gologanu failure, F/d Falkinger
DP600 1.2mm 3D hardening
test model curves and strain et al. 2010
+ Barlat
fields
(i)3D,
Prediction of
(ii)plane
failure, F/d
stretch strain Isotropic MMC Luo et
curves and strain
DP780 1mm bending and hardening fracture Wierzbicki
and stress fields
test (iii)shell + Hill model 2010
(best agreement
element
with 3D)
s
pure 2D: Mixed extended
Prediction of Bettaieb et
DP600 4mm bending plane hardening Gurson
bendability al. 2010
test strain + Hill model
Table I.1. Studies dealing with numerical simulations of various forming tests.
24
6. Summary
To conclude on this part, it was first noticed that several parameters can influence the
bendability of steels (mechanical properties, microstructure, tools, etc…) making difficult
the understanding of formability issue. Among them, microstructural parameters, in
particular heterogeneities, seem to have a key role in bending operation.
Secondly, it was shown that, in bending tests, failure propagate by a shear mode and seem
to initiate from the outer surface in air-bending test and appear either at the punch nose or
in the sidewall in stretch bending tests depending on the bending conditions.
Finally, it is highlighted that finite-element numerical simulation is an efficient tool to
accurately relate global bending performance to mechanical behaviour and predict failure
mechanisms at microstructural scale. It was shown that a mixed hardening law coupled
with anisotropic criterion provides the best combination to accurately predict bendability.
Almost analyses were carried out with a coupled damage approach and most of them used
3D modelling except for simple bending test where 2D simulation with plane strain
condition was sufficient.
25
I.B. TRIP steels
1. Definition
Low alloy TRIP-aided steels (which will be termed “TRIP steels” throughout this
manuscript for the sake of simplicity) are composed of a ductile ferrite matrix containing
small islands of hard bainite, martensite and retained austenite. Transformation of retained
austenite into martensite during deformation causes significant strain hardening and delays
the onset of necking, providing high ductility. It is the “TRansformation Induced
Plasticity” (TRIP) effect that forms the basis of the exceptional properties of the TRIP
steels.
These steels differ from conventional steels by their remarkable combination of high
strength and ductility, resulting from their particular microstructures. The hardening
capacity of the TRIP steels is considerable, ensuring efficient strain redistribution during
forming and hence good drawability, combined with very high yield strength in the
finished component.
Due to their high energy absorption capacity, TRIP steels are principally used for structural
and safety automotive parts such as longitudinal beams, seat structures, etc.
2. Processing of TRIP steels
It was shown (Rigsbee and van der Arend, 1977; Speich and Miller, 1979) that retained
austenite contained in Dual Phase steel permits the improvement of their mechanical
properties.
If austenite is stable enough, the mechanical strength and the ductility are increased. The
stabilization is performed after the intercritical annealing. During the bainitic
transformation, austenite is supersaturated in carbon, ensuring its stability.
In order to obtain a multiphased structure with retained austenite, the thermal cycle used
for DP steel is modified by introducing holding in the bainitic domain (Matsumura et al.,
1987). Figure I.14 illustrates heat cycles for cold rolled DP steels (a) and TRIP steels (b).
Figure I.14. Heat treatment cycle and microstructures (a) DP steels (b) TRIP steels. F: ferrite; A: austenite;
B: bainite; M: martensite (Matsumura et al., 1987).
The evolution of the microstructure during the TRIP steel process is shown in Figure I.15
with:
(i) ferrite and austenite during the intercritical annealing
(ii) austenite decomposition and ferrite nucleation at grain boundary
(iii) ferrite, bainite and retained austenite: during and after the bainitic holding in a
range of temperature between 350 and 500°C.
26
Figure I.15. Microstructure formation during the thermal cycle. α = ferrite, αb = bainite, γ = austenite,
γr= retained austenite (Ohlert et al., 2002).
At room temperature, the microstructure of these TRIP steels is composed of ferrite,

bainite, retained austenite and sometimes martensite. A typical microstructure of a TRIP
steel by SEM is shown in Figure I.16.
Figure I.16. Typical microstructure of a TRIP steel by SEM. F = ferrite, A = retained austenite and B =
bainite (Jacques et al., 2002).
3. Role of retained austenite
In the case of DP steel, mechanical properties are mainly due to the interaction between a
hard phase (martensite) and a soft phase (ferrite). For TRIP steels, the progressive
transformation of retained austenite into martensite, coupled with the multiphased
microstructure are the factors responsible for their mechanical properties.
Destabilization of retained austenite can be caused by plastic deformation (Pickering,

1992). This transformation induced by mechanical loading improves the mechanical
properties of low-alloy TRIP steels with dispersed retained austenite islands
(Haidemenopoulos et al., 1989; Jacques, 1998).
Retained austenite influences the mechanical properties thanks to both its stability and its
distribution: too instable retained austenite is transformed into martensite too early, so that
necking cannot be efficiently delayed. On the contrary, if the retained austenite is too
stable, the martensitic transformation does not occur and the positive effect on delaying
necking is cancelled. Therefore, austenite should have the right good stability to ensure
progressive martensitic transformation during straining.
27
In addition, the repartition of retained austenite must be homogeneous in the part before
the deformation. Indeed, the mechanically induced martensitic transformation improves
strength and ductility by both composite strengthening resulting from the progressive
increase in the amount of hard martensite, and dislocation strengthening of the ferrite
matrix as a consequence of the transformation strain accompanying the martensitic
transformation (Jacques et al., 2001).
4. Influence of the alloying elements
The present study focuses on the low alloy TRIP-aided steel family, and more particularly
the Mn-Al-Si family. Effects of alloying elements have been largely studied. Some authors
speak about the role of carbon to stabilize austenite (Tsukatani et al., 1991; Sakuma et al.,
1991), other ones about the silicon or manganese contents (Pichler and Stiaszny, 1999;
Baik et al., 2000; Kim et al., 2001; Lee et al., 2002; Mahieu et al., 2002) to obtain a
multiphased microstructure and to improve the mechanical properties of the material.
Effects of various alloying elements can be summarized as:
Carbon stabilizes austenite by decreasing transformation temperatures Ms (martensite

start) and Mf (martensite finish).
Manganese is an austenite stabilizer. By delaying ferritic and pearlitic transformations, it

promotes the formation of retained austenite. It is also used to harden the material by solid
solution strengthening and to refine the ferritic structure. Additionally, pronounced
banding may occur in steels containing big amount of manganese (Kim et al., 2001).
Banding is attributed to the inter-dendritic segregation of solute manganese during
solidification of ingot/slab from liquid steel (Bastien, 1957).
Silicon inhibits the cementite formation during the bainitic transformation and causes the
stabilization of austenite thanks to the carbon rejected from the bainitic ferrite formed
during isothermal holding at low temperatures. As manganese, it can be used to harden the
material by solid solution strengthening.
However, one of its drawbacks is its tendency to segregate and therefore to lead to
heterogeneous properties. Moreover, it decreases the welding ability and has a deleterious
effect during the hot dip galvanizing of the sheet.
To avoid the drawbacks of high Si additions, lots of authors investigated the possibility to
partly replace Si by other elements such as P or Al (Gallagher et al., 2002; Chen et al.,
1989; Barbel et al., 2002; De Meyer et al., 1999, Girault et al., 2001). Addition of one of
these elements does not seem to affect final mechanical properties. For example, Traint et
al. (2002) showed that addition of P and more particularly of Al help to form more stable
retained austenite.
Aluminium is a strong ferrite stabilizer, as Si, it inhibits the formation of Fe3C as it is not
soluble in this element. Hence, the most suitable element which can readily provide the
same effects as Si is Aluminum. As a matter of fact, up to 1.5%, Aluminum shall not affect
coatability adversely (Mintz, 2001; Maki et al., 2003).
Molybdenum is a solid solution strengthener of ferrite and retards pearlite formation. It is

also known to increase hardenability and retained austenite fraction but has the drawback
to be expensive.
28
Finally, as for Dual Phase steels, final mechanical properties of TRIP steels can be
improved by adding micro-alloying elements such as Ti, Nb or V (Kim et al., 2002;
Bai et al., 1998; Pereloma et al., 1999). These elements are principally used to harden the
material by solid solution or by precipitation but they are also used to modify phase
transformations by acting both on thermodynamic equilibrium and on transformation
kinetics.
5. Martensitic transformation
It is a diffusionless transformation which implies the same chemical composition for the
austenite and martensite phases.
This transformation corresponds to an austenite shearing which leads to a shape change
and a volume increase. This highest volume leads to stress concentration around the newly
formed martensite islands. To approach 100% of martensite from retained austenite, it is
necessary to cool until a temperature lower than Mf. For cooling temperature between Ms
and Mf, martensitic transformation will be therefore only partial.
Several authors tried to give an empirical formula for Ms calculation as a function of the
chemical composition (Beranger et al., 1994). The use of such equations is often restricted
to a category of steels. For example, for low alloy TRIP steels, the following equation is
proposed (Andrews, 1965):
M s (°C ) = 539 − 423 ⋅ C − 30.4 ⋅ Mn − 7.5 ⋅ Si + 30 ⋅ Al (in wt%) (7)
In fact, the Ms temperature is also a function of other factors such as previous

transformations or strain state in retained austenite.
According to the stress or the strain state, various retained austenite transformation
mechanisms can be involved in a TRIP steel (Lee et al., 2003). Indeed, mechanical
properties that Lee et al. (2003) have obtained in uniaxial tension depend on the amount of
retained austenite. On the contrary, the formability in equibiaxial stress is controlled by the
stability of retained austenite. The strain rate is also an influent parameter.
6. Microstructure heterogeneities
As mentioned previously, microstructural parameters could have an influence on

bendability. TRIP steels have a very good combination of high strength and ductility due to
the progressive transformation of retained austenite into martensite. However, the
repartition of the phases in the material (especially retained austenite distribution) has an
influence on mechanical properties. When this repartition is heterogeneous, this could be
detrimental. For example, when retained austenite islands are linked by each others,
banded structures are formed and this alignment could have an influence on the
bendability. Therefore, these microstructure heterogeneities should have to be taken into
account more precisely. Hence, in order to understand the influence of banded structures,
the following part will deal with their origin.
29
I.C. Banded structures formation
Banded structures appear during the cooling after the hot rolling. They are not formed
during the solidification of the steel but it is a consequence of the micro and
macrosegregations.
1. Solidification of steel in continuous casting
Micro and macrosegregations are formed during the solidification of steels in continuous
casting (Figure I.17). Segregation can be defined as heterogeneity of the composition in the
as-cast product after complete solidification.
Two different kind of segregation must be distinguished:
- The microsegregation is a variation of composition at the dendritic scale, in the
order of 10 to 100µm.
- The macrosegregation is a variation of composition at the product scale, in the
order of few mm.
Ladle
Tundish
Mold
Secondary
cooling
Oxygen
cutting
Figure I.17: Continuous casting.
Figure I.18 summarizes the formation mechanisms of solidification structures and

segregation in continuous cast steel products (Didier and Bobadilla, 2008).
• At the first step of solidification, a solid shell is formed in the mold. The growth of
the solid shell is columnar.
• In a second step, the liquid in the center of the slab becomes undercooled and
equiaxed dendrites nucleate and grow in the liquid. These dendrites slowly settle
due to the density difference between the liquid and the solid.
• When the number of equiaxed dendrites ahead of the columnar front becomes too
important, the growth of the columnar zone is stopped. The width of the central
equiaxed zone and of the columnar zone is then fixed. The solidification continues
through the growth of the central equiaxed grains. The interdendritic liquid can
move between the dendrites arms.
30
• At the end of solidification, the deformations of the solid skeleton of the mushy
zone due to bulging and shrinkage lead to segregated interdendritic liquid motion
and to the formation of macrosegregated zones.
1- Columnar growth
2- Equiaxed growth
3- Columnar-Equiaxed Transition
4- Macrosegregation formation
Figure I.18: Solidification of continuously cast products (Didier and Bobadilla, 2008).
After the solidification, the slab has the structure shown in Figure I.19. Two areas can be
distinguished:
- Two columnar zones where dendrites are oriented in the perpendicular direction of
casting.
- An equiaxed zone with thinner dendrites, at mid-thickness of the product.
Columnar
zone
245 mm (slab
Equiaxed thickness)
zone
Axial
Columnar
segregation
zone
Casting direction
Figure I.19: Solidification structure (Bobadilla, 2008).
31
2. Microsegregation formation (Bobadilla et al., 2009)
The microsegregation that results from solute redistribution (C, Mn, Si, P, S…) during
dendritic solidification leads to variation in the solute concentration between the center and
the outside of a dendrite arm.
In as-cast product, because the dendritic microstructure size (primary and secondary arm
spacing) increases from the skin to the core of the product, the level of microsegregation
increases from the skin to the center part of the slab.
The elements with low diffusion coefficient (Mn, Si, Cr, Ni, Mo…) show high
microsegregation level at the end of solidification. This is due to the fact that very few
solute redistribution can occur in the solid phase during the solidification. Figure I.20
illustrates a Mn content map from the skin to the mid-thickness of the slab. Note that the
dendrites arms regions correspond to the lowest content of Mn (in blue) while the highest
concentrations of Mn are observed in the interdendritic region (in red). In addition, the
highest Mn content is observed at-mid-thickness, this means that microsegregations will
principally appear in this area.
2 mm 1.745
1.720
1.695
1.670
1.645
skin ¼ thickness mid- thickness
Figure I.20: Mn content for three locations on a slab through the thickness. TRIP800 steel (Bobadilla, 2003).
In as-cast product, the microsegregation level depends on the chemical composition of the
steels: it increases with carbon, manganese, phosphorus content and so on… In order to get
low microsegregation level it is necessary to refine the dendritic microstructure i.e. to
decrease the dendrite spacing.
3. Macrosegregation formation
The mechanisms of macrosegregation formation on flat products during continuous casting

are explained by segregated liquid flowing to the center of the slabs at the end of
solidification (Didier and Bobadilla, 2008). Two factors will lead to interdendritic liquid
motion during the continuous casting of flat products (Figure I.21):
• Bulging of the slabs between the rolls during casting lead to periodic expansion and
compression of the central mushy zone. This periodic deformation of the solid
skeleton of the mushy zone leads to segregated interdendritic liquid motion and
increases the formation of macrosegregation.
• Shrinkage will lead to a local depression at the end of solidification. This
depression will create movements of the segregated interdendritic liquid which will
be sucked up in the center of the slab.
32
The last interdendritic liquid, which is rich in alloying elements due to the
microsegregation, solidifies in the center of the slabs and forms the axial
macrosegregation.
Interdendritic liquid
movements due to
Liquid core
bulging
Solid shell
Bulging Shrinkage at the end
of solidification
cracks
Axial segregation
Figure I.21: Mechanisms of macrosegregation formation. Note that bulging is enhanced on the scheme
(Didier and Bobadilla, 2008).
4. Banded structures formation (Didier and Bobadilla, 2008)
The theory of the formation of banded structures is illustrated in Figure I.22 (Thompson
and Howell, 1992).
Banded structure formation depends on the spatial distribution of substitutionnal elements

(Mn, Si) resulting of the remaining segregation from solidification. During hot-rolling, the
microsegregated zones rich in Mn formed during the solidification are spread over the
rolling direction (Figure I.22a). The temperature at the end of rolling is generally around
800 to 900°C. The steel is fully austenitic in this temperature range. All the structure is
austenitic with higher amount of Mn in the segregated zones.
During cooling, the ferrite phase first appears in regions which are the poorest in Mn and
for which the transformation temperature for austenite into ferrite is higher (Figure I.22b).
As cooling progresses, the ferrite grains grow in the Mn-poor regions and carbon is
rejected into the remaining austenite phase, so that the carbon content in the segregated
zone increases (Figure I.22c). When the temperature reaches the eutectoid temperature,
pearlite is formed by simultaneous precipitation of ferrite and cementite in the segregated
zones which are rich in carbon. The final structure obtained is composed of ferrite grains
(in white) in the former intradendritic zones and of pearlite (in black) in the former
interdendritic regions (Figure I.22d).
This oriented structure called “banded structure” is responsible for an anisotropy of the
mechanical properties of the steel.
33
T1 ≈ 800°C – 900°C T 2 < T1

Mn-rich
areas
Austenitic
grains
b)
rolling direction
Formation of ferrite in the
Mn poor areas
T 3 < T2 T4 < Teutectoid
C
C
C
c) d)
C rejection in the Growth of ferrite grains Ferrite grains Precipitation of pearlite

remaining austenite
Figure I.22. Banded structure formation (Thompson and Howell, 1992).
5. Parameters allowing to control micro and macrosegregation
For a given steel grade, the main continuous casting parameters to control
macrosegregation and microsegregation are:
• Superheating
• Casting speed
• Secondary cooling
• Soft reduction
For metallurgists, the superheat is the temperature of the liquid steel measured in the mold
subtracted by the value of the liquidus. A high superheat is characterized by a thick central
segregation band and thinner segregation bands around it. A low superheat is characterized
by many thick bands in the center of the hot rolled sheet.
The major way used to control the macrosegregation during the continuous casting of flat
products is the soft reduction (Figure I.23). Soft reduction is applied at the end of
solidification of the continuous cast product. The principle is to reduce the thickness of the
product by a few mm near the end of solidification in order to avoid the interdendritic
liquid motion in the mushy zone responsible for axial macrosegregation.
rolls
th.1 th.2
Figure I.23: Soft reduction principle from thickness th.1 down to thickness th.2 at the end of the
solidification.
34
Steps following the continuous casting as the reheating of slabs or the rolling and coiling
also have an influence on microsegregation and macrosegregation.
- Slabs furnace: slab reheating reduces the intensity of microsegregation (especially

in the skin of the product) but does not reduce the intensity of axial
macrosegregation.
- Rolling: hot rolling does not affect the intensity of microsegregation and
macrosegregation but elongate them strongly.
- Coiling: the coiling temperature impacts on structural banding, is more important

than the superheat and must be as low as possible. Indeed, a low coiling
temperature means a higher cooling rate (run out table) before coiling and then less
time for carbon diffusion: less banded structure.
6. Summary
Figure I.24 illustrates the effects of (a) the superheat on the slab structure and (b) the
coiling temperature on the banded structures after hot rolling.
Low superheat
Distributed Low T Hot coil
macrosegregation spots coiling thickness
~ a few mm
Slab
thickness
~250mm
Hot coil
High T thickness
coiling ~ a few mm
Columnar Equiaxed
zone zone
Medium and high superheat

Low T Hot coil
coiling thickness
Slab ~ a few mm
thickness
~250mm
Hot coil
High T thickness
Aligned macrosegregation
spots coiling ~ a few mm
(a) (b)
Figure I.24: (a) Superheat effect on the slab structure before hot rolling and (b) temperature coiling effect on
the banded structures after hot rolling.
35
II. Microstructural characterization of the four TRIP-

aided steels studied
Four TRIP steels were chosen for this study. Steel 1, Steel 3 and Steel 4 have a thickness of
1.6mm while Steel 2 which come from the same casting as Steel 1 has a thickness of 2mm.
All four steels were provided in the galvannealed (GA) state. These materials provide
different microstructures and various bending performances. They were thus chosen to
check the effect of the microstructure on their bendabilities.
II.A. Chemical composition and phase morphology
Chemical compositions and microstructures at third-thickness (rolling direction) are

respectively reported in Table I.2 and Figure I.25. Figure I.26 shows mid-thickness
microstructures both along rolling and transverse directions for the four TRIP steels. As
Steel 1 and Steel 2 come from the same casting, they have the same chemical composition.
Differences with the two other materials are in fact very slight except for the Al+Si
content.
C Mn Si + Al P + Mo + Ti +Cr +Ni
Steel 1 216 1673 1480 194
Steel 2 Same composition as Steel 1
Steel 3 219 1694 1614 209
Steel 4 225 1663 1554 214
Table I.2. Chemical composition of the TRIP steels (x 0.001 wt%)
Steel 1 Steel 2
sheet
thickness
20 μm 20 μm
RD Steel 3 Steel 4
20 μm 20 μm
Figure I.25. Light optical micrographs of the four TRIP steels along rolling direction, LePera etching, third-
thickness microstructure.
36
Steel 1
Steel 2
Steel 3
Steel 4
(a) sheet (b) sheet

thickness thickness
RD TD
Figure I.26. Light optical micrographs of the four TRIP steel. Mid-thickness microstructures, LePera etching
(a) along rolling direction and (b) along transverse direction.
37
Microstructures of the four steels were observed by light microscopy after a LePera
etching (Appendix 1). It is noticed that:
- Martensite/austenite (M/A) islands appear in white.

- Ferrite appears in brown.
- M/A are present in micrographs as 2-5µm islands linked by thin bands of smaller
islands.
- All steels except Steel 4 have a central segregation band at mid-thickness. This
band has a thickness ranging between 2 and 20µm for the four steels studied. In
addition, from various cross-sections observations carried out on steels, this band is
found to be heterogeneous both with its presence and with its thickness along the
sheet.
- This central segregation is thicker for Steel 2 and Steel 3.
- In addition, this band differs along rolling and transverse direction. This band is
more pronounced along transverse direction than along rolling direction for Steel 3.
- Central segregation is composed mainly of martensite and occasionally on
austenite.
- Concerning banded structures, they are thinner (often < 2µm) and are mainly
present in the vicinity of the mid-thickness and occasionally close to the surface.
Note that these cross-sections observations are not appropriated to estimate the shape of
the central segregation band in 3D. From these observations, hard segregation bands seem
to be present as layer with variable thickness along the sheet. Similarly, it is difficult to
evaluate the shape of M/A islands in 3D. They could be connected with each other and
may play an important role on the cracking in bending.
In order to determine the M/A islands content in each steel, light microscopy observations
were performed at third-thickness. After a Klemm colorant etching (Appendix 1), the M/A
islands appear in white. Figure I.27 illustrates results obtained after etching with and
without threshold.
Table I.3 provides the amount of M/A islands for each steel. Note that their contents are
very similar and ranging between 17 and 20%. In order to determine if the islands are more
austenitic or more martensitic, retained austenite measurements are also carried out (see
next part).
Material % M/A islands

Steel 1 18%
Steel 2 17%
Steel 3 19%
Steel 4 20%
Table I.3. M/A islands quantification for the four steels studied.
38
Pictures without threshold Pictures with threshold of

of M/A islands M/A islands
Steel 1
20µm 20µm
Steel 2
20µm 20µm
Steel 3
20µm 20µm
Steel 4
20µm 20µm
Figure I.27. Quantification of M/A islands in the four TRIP-aided steels.
39
II.B. Determination of the retained austenite content
In one study (Jacques et al., 2009), measurements of retained austenite in several

multiphase TRIP steels (~1mm) were carried out by comparing six different techniques
(light microscopy, X-ray diffraction, EBSD, magnetic saturation, thermal diffusivity and
laser ultrasonics). According to these measurements, a large variability of the estimated
retained austenite contents has been observed. Among these techniques, light microscopy
and EBSD systematically brought larger and smaller amounts of retained austenite
respectively and X-ray diffraction and magnetic saturation are shown to be the most
common methods for retained austenite measurements.
The main drawback of X-ray diffraction (XRD) method is that the analysis is carried out
on the first few microns below the sample surface. Therefore, this technique cannot be
used to determine retained austenite content in steels which have a heterogeneous
repartition of phases through the specimen.
Magnetic measurements are also widely used and one of their advantages is that
measurements are conducted on the bulk. Austenite being paramagnetic contrary to other
steel’s constituents, the saturation magnetization of the sample can be linked on the amount
of retained austenite. Hence, the transformation of retained austenite into martensite
corresponds to a transformation of a paramagnetic phase into a ferromagnetic phase.
Although this method is easily usable, it has the drawback to be destructive compared to
XRD measurements.
Other techniques are also used to determine the amount of retained austenite. Among them,
the Mössbauer spectroscopy (Ladriere and He, 1986; Jacques, 1998) and neutron
diffraction (Huang et al., 2002; Reed et al., 1998) can be mentioned. However, these
methods require specific experimental means and their use was outside the scope of the
present study.
1. XRD measurements
In this study, XRD measurements were carried out to determine the retained austenite
content in each of the selected materials in the as-received conditions. Details of this
experiment are given in Appendix 1. The peaks used to measure respectively the amount of
ferrite and austenite are (200), (211), (220) and (200), (220), (311).
In order to investigate the transformation of retained austenite into martensite,
measurements were performed on as-received materials (flat samples containing austenite)
and on bent specimens for three steels (Figure I.26).
It is found that the amount of retained austenite in the as-received material is 8% for
Steel 1, 10% for Steel 2, 16% for Steel 3 and 15% for Steel 4. Steel 1 and Steel 2 have very
low amount of retained austenite. The difference between the M/A islands content obtained
after light microscopy and the retained austenite amount measured by XRD is the
martensite amount. Steel 1 and 2 have the highest martensite content with respectively
10% and 7%. Steel 3 and 4 have the lowest martensite content with respectively 3% and
5%. This result is in agreement with the mechanical properties because Steel 1 and 2 have
the highest tensile strengths.
40
35000
30000 α110 as received
Intensity (a.u)
25000 bent sample
20000 α211
15000
α200
10000 α220
γ111
5000 γ200 γ220 γ311
0
(a)
20 60 100 140
Angle (°)
35000
30000 as received
Intensity (a.u)
25000 bent sample

20000
15000
10000
5000
0
(b)
20 60 100 140
Angle (°)
35000
30000 as received
25000 bent sample

Intensity (a.u)
20000
15000
10000
5000
0
20 60 100 140
(c)
Angle (°)
Figure I.28. XRD measurements both on as-received and bent specimens along RD-TD plane. 2θ scale. (a)
Steel 1, (b) Steel 3, (c) Steel 4.
It is noticed from Figure I.28 that austenite peaks have disappeared with the bent sample.
This means that the austenite has been transformed into martensite during bending (just
below the surface – outer fibre – tensile area).
41
2. Magnetic measurements
Concerning magnetic measurements, sigmametry method was used. This technique

consists in measuring the saturation magnetization of specimens with and without
austenite. More details about the measurement are given in Appendix 1.
Although with this method, the whole thickness specimen is analysed, similar results to
XRD measurements are obtained for as-received materials except for Steel 1 which has a
higher value of retained austenite with this method. Indeed, Steel 1, Steel 3 and Steel 4
have respectively 12%, 16% and 13% of retained austenite. The difference between XRD
and magnetic measurements for Steel 1 is not well understood and could be due to an
heterogeneous repartition of retained austenite into the sample. For bent samples, retained
austenite was still detected (nearly 5% for all steels). This difference is due to the location
of measurements (over a few microns beyond the outer surface in the tension deformed
area for XRD versus all the thickness for magnetic measurements). So, it is possible to
conclude that a significant part of retained austenite is totally transformed into martensite
during bending. Indeed, the area which is less deformed has surely retained austenite
which was not transformed into martensite.
The carbon content in retained austenite was also determined for the steels investigated,
from the 2θ XRD plots.
Several authors proposed equations linking the austenite cell parameter, aγ (aγ in nm), to
alloy element content (in wt%):
a γ = 0.3555 + 0.0044.C (Roberts, 1953) (8)

a γ = 0.3573 + 0.0033.C (Ridley et al., 1969) (9)
a γ = 0.3553 + 0.000095.Mn + 0.00056. Al + 0.00006.Cr + 0.000474.C (Cheng et al., 1990) (10)
The last expression is the most suitable one for TRIP steels. It yields to the following
carbon content (%) in retained austenite:
- 1.15 % for Steel 1

As a conclusion, all steels have similar carbon content in retained austenite i.e. close to
1.15%. These values are in good agreement with the ones found in the literature. For
example, Sugimoto et al. (2002) calculated a C content in retained austenite ranging
between 1.10% and 1.20% for TRIP having about 10% of retained austenite and 0.2 C
(wt%). Srivastava et al. (2007) obtained a concentration of carbon in retained austenite of
0.82% for their TRIP steels. This lowest value can be due to the difference in tensile
strength: 600MPa for their steels and highest strength for our TRIP-aided steels. Girault et
al. (1998) found a C content in retained austenite of 1.05% and 1.2% for two TRIP steels
having respectively 0.11% and 0.27% C amount.
42
II.C. Nanohardness of ferrite matrix and secondary phases
Nanoindentation measurements were carried out on the three TRIP steels with a thickness
of 1.6mm. This technique consists in indenting the material with a slight depth and to
monitor the penetration evolution as a function of the applied load in order to determine
the local mechanical properties at the surface of the material. Additional details on this
technique are given in Appendix 1.
The nanohardness tests were coupled with an Atomic Force Microscope (AFM) allowing
locating the area to indent before testing.
An example of loading/unloading curves of Steel 1 both for M/A islands in the vicinity of
the mid-thickness and the ferritic matrix is shown in Figure I.29.
6000
ferritic matrix
5000 M/A islands
Load (microN)
4000
3000
2000
1000
0
0 100 200 300
Depth (nm)
Figure I.29. Loading/ unloading cycle for Steel 1.
Figure I.30 shows two AFM pictures of a M/A island before and after nanohardness
measurements.
2µm
indentation
(a) (b)
Figure I.30. AFM observations. M/A island is surrounded with yellow circle. (a) Island before the
indentation, (b) island after indentation with two prints surrounded with white squares.
Scanning electron microscopy was also used to have a better image resolution to measure
the size of indents. For example, Figure I.31 shows two observations with indentations on
the central segregation band (a) and in the ferritic matrix (b). As expected, the print on the
softer matrix is larger than the ones within the central segregation band and the size of
indents is much smaller than the size of phases, at least in the plane of view.
43
Central
segregation
ferrite
M/A island
(a) (b)
Figure I.31. SEM observations after nanohardness measurements on Steel 3. Prints are surrounding in white.
(a) on the central segregation band, (b) in the ferritic matrix.
Five nanohardness measurements carried out for the three steels both on ferritic matrix and
M/A islands close to the mid-thickness are reported in Figure I.32.
Similar ferritic matrix nanohardness is noticed for all TRIP steels, while the nanohardness
of M/A islands in the vicinity of the mid-thickness is slightly different for the three steels.
The ferrite nanohardness value is in very good agreement with the ones found by Jacques
et al. (2007) and Furnémont et al. (2002) where they measured a nanohardness of about
5GPa. In these studies, the austenite and the martensite were found to have a nanohardness
of respectively 11 and 17GPa. The values of the M/A islands close to the mid-thickness for
the three steels are in this range. The difference between the three steels could be due to the
nature of the island, i.e. mainly austenitic for Steel 1 and 4 and mainly martensitic for
Steel 3. In addition, this steel is also the most segregated one among the three materials.
20
M/A islands
Nanohardness (GPa)
16
ferritic matrix
12
0
Steel 1 Steel 3 Steel 4
Figure I.32. Nanohardness measurements for the three TRIP steels both for the M/A islands in the vicinity of
the mid-thickness and for the ferritic matrix.
II.D. Summary
Table I.4 provides a summary of the microstructural properties of the four studied
materials. Their microstructures are very different as well as the behaviour of each phase
(content, distribution, nanohardness, etc). These different behaviours could have a role
during bending tests.
44
Retained
Microstructure Central Martensite Nanohardness (GPa)
austenite
homogeneity segregation content Ferrite MA islands
(DRX)
Medium
Steel 1 Medium 8% 10% 4.1 11
(<5µm)
High
Steel 2 Low 10% 7% /
(<20µm)
High
Steel 3 Medium 16% 3% 4.3 15
(<20µm)
Low
Steel 4 High 15% 5% 5.5 10
(< 2µm)
Table I.4. Summary of the microstructural properties of the four TRIP studied.
45
III. Mechanical characterization of the four TRIP-aided

steels studied
A knowledge of the mechanical properties is essential for a practical application of any
steel. The aim of this section is to give a full analysis of the mechanical behaviour of the
studied TRIP steels both in tensile tests and in bending tests. All experimental procedures
and specimen dimensions are described in Appendix 2.
III.A. Tensile properties at room temperature
Tensile tests have been performed on ISO specimens 20x80 along each direction under
load line displacement control at an initial strain rate of 2.5.10-4 s-1. The longitudinal strain
was monitored with an extensometer of 80mm in gauge length. Engineering stress/strain
curves along transverse direction of each TRIP steel are shown in Figure I.33 and Table I.5
summarizes their mechanical properties.
According to their mechanical behaviours, the four chosen steel sheets can be separated in
two families. Steel 1 and Steel 2 have similar mechanical properties with high tensile
strength and medium fracture elongations and the two other ones Steel 3 and Steel 4 are
more ductile but present a yield point elongation (~0.4%) that might play a role at the
beginning of bending tests. The main difference between Steel 1 and Steel 2 is the lower
total elongation of Steel 2 compared to that of Steel 1.
1000
Engineering stress (MPa)
800
600
400 Steel 1
Steel 2
Steel 3
200
Steel 4
0
0 10 20 30
Elongation (%)
Figure I.33. Engineering stress / strain curves along TD for the four TRIP steels.
Note that fracture surface of tensile specimen are ductile with dimples and fractures appear
at 45° within the thickness as it is reported in Figure I.34.
Figure I.34. Broken tensile specimen with a 45° fracture through the thickness.
46
0.2% proof Tensile Uniform Total

Direction
stress (MPa) Strength (MPa) elongation (%) elongation (%)
Steel 1 RD 467 866 15.0 20.0
(1.6mm) TD 546 851 15.0 19.2
Steel 2 RD 464 816 16.6 22.9
(2mm) TD 521 834 13.6 17.0
Steel 3 RD 426 795 19.7 25.5
(1.6mm) TD 448 790 19.2 24.5
Steel 4 RD 433 764 20.0 27.2
(1.6mm) TD 451 782 18.2 22.8
Table I.5. Room temperature tensile properties of the four TRIP steels along rolling (RD) and transverse
(TD) directions (ISO 20x80mm tensile specimens).
III.B. Air-bending properties
This test is the most commonly used for bending because it is the most simple one and
easy one to set up. With this test, it is possible to determine the maximum bending angle
for a specified bending radius. It is important to underline that the inner radius (of the bent
sample) does not exactly correspond to the punch radius. This test is carried out on a
universal tensile machine (Figure I.35). Various punches are available (radius from 0.1 to
15 mm). Samples have to be ground before trials to avoid cracks starting from edges
(Dalloz et al., 2007). The dimensions of the samples are 50x100mm with full thickness.
This test is carried out without lubrication.
α: bending angle
(a) (b) β: angle of the bend (c)
Figure I.35 : (a) Air-bending device, (b) schematic and (c) bent specimens.
For this study, air-bending tests were performed on the four TRIP steels. Five samples
along each direction (rolling and transverse) were bent for Steel 1, 3 and 4 while Steel 2
was only bent along transverse direction as it is the most detrimental one (Table I.6). For
this test, the punch used has a radius of 0.1mm. Tests are interrupted as soon as the first
crack was considered to appear at the bend nose. As a criterion to stop the test, an
automatic method based on the load/displacement curve was used. As soon as the load
decreases, it is considered that a microcrack had appeared. The test is usually stopped after
5 to 10% decrease of the load. To check whether the specimen is cracked or not, a
magnifying glass is used.
47
The angle of the bend (β) of each sample was measured thanks to a protractor (after
unloading and thus specimen springback) in order to compare the steels with each other.
The bending angle (α) is then deduced. Finally, based on the five tests results, an average
and its standard deviation are calculated.
Steel 1 Steel 2 Steel 3 Steel 4

RD TD RD TD RD TD RD TD
Sample 1 51 63 62 Bending until
Bending until the
Sample 2 34 61 59 the limit of the
limit of the
Sample 3 34 66 58 machine
machine without
Sample 4 22 55 No 56 without
cracking (~15°)
Sample 5 22 64 data 58 cracking (~15°)
β angle (°) 33 62 59 <15 <15 <15 <15
α angle(°) 147 118 121 >165 >165 >165 >165
Stand. dev. 11.9 4.2 2.2 0 0 0 0
Table I.6. Air-bending results.
From Figure I.36, it is possible to note that:
- Due to the limit of the machine, it was not possible to bend Steels 3 and 4
specimens above 165°. As no crack was observed on them, it is not possible to use
this test to compare their bendability.
- Steels 3 and 4 have a better bendability than the two other ones.
- Steel 1 and Steel 2 which have similar mechanical properties, also present
equivalent air-bending behaviour along TD.
- There is a high anisotropy when comparing bending angles along rolling or

transverse directions (Steel 1). It is clearly noticed that Steel 1 has best bendability
along rolling direction. This confirms study of Leu et al. (1997) where a poor
bendability was noticed on transverse bent specimens.
180
170 Machine limit
Bending angles (°)
160
150 TD
140 RD
130
120
110
100
Steel 1 Steel 2 Steel 3 Steel 4
Figure I.36. Air-bending results for each TRIP steel.
48
III.C. Stretch bending properties
In the automotive industry, many parts are stamped, and some more complex forming
paths than simple bending have to be encountered. Stretch bending is one of them. As for
the Erichsen test, the stretch bending test consists in clamping blanks of a sample between
a die and a blank-holder. This test makes it possible to determine failure heights for various
values of punch radius, R, and then rank material formability for various R/t ratios (t being
the sheet thickness). It is also possible to evaluate the failure mode which seems to move
from punch region to sidewall for high punch radii, depending on the bendability of
material.
As for the air-bending test, five samples (size: 50x250mm) along each of the rolling and
transverse direction were bent for Steel 1, 3 and 4 and Steel 2 was only bent along
transverse direction. For the stretch bending test, three punch radii (R = 2, 4, 6mm) were
tested and there is a lubrication between the sheet and the punch. In this test, grease and a
Teflon film were laid on the punch radius to improve the friction between specimen and
tool.
As this test is very unstable, it is carried out with a hydraulic press and thus implies a lot of
stored energy, it is thus impossible to interrupt it as soon as a micro-crack appears in the
sample (Figure I.38). Therefore, the test was only stopped once the sample was broken in
two parts. To compare steels, the bending angle (α) was determined for each sample after
failure (equivalent to failure height). This angle was measured with a protractor by
reproaching the two broken parts of the specimens. Finally, an average and a standard
deviation are given for the five samples. Figure I.38 shows two stretch bent specimens (A:
with a punch radius of 6mm, B: with a punch radius of 2mm).
Die Sidewall Punch
Sample
Blank-holder
Figure I.37 : Stretch bending device
A
B
Figure I.38: Picture of two stretch bent samples (A: punch radius = 6mm, B: punch radius = 2mm).
Figure I.39 shows bending angles obtained after fracture for each steel and for the three
punch radii (2, 4 and 6mm) along rolling and transverse directions, except for Steel 2
which was only bent along the transverse direction.
49
90 90
Steel 1 Steel 1
80 80
Bending angle (°) Steel 3 Steel 2
Bending angle (°)

70 Steel 4 70 Steel 3
60 60 Steel 4
50 50
40 40
30 30
20 20
10 10
0 0
(a) R=2 R=4 R=6 (b) R=2 R=4 R=6
Figure I.39. Stretch bending results: (a) along rolling direction, (b) along transverse direction.
From Figure I.39, it is reported that for a punch radius of 6mm, Steel 1, Steel 3 and Steel 4
have similar bending behaviour and Steel 2 has a lower bending angle than the three other
steels along transverse direction. For lower punch radii (2 and 4mm), Steel 1 and 2 have
lowest bendability than the two other ones.
Regarding the fracture locations, all steels broke on the punch nose for punch radii of 2 and
4mm while for higher radius (6mm), Steels 3 and 4 have their cracks located in the
sidewall and not at the punch nose. For the 6mm punch radius, Steel 1 and 2 broke at the
punch nose. It is in agreement with studies discussed in part I.A. for which high values of
R/t, lead to shift of the crack from the punch nose to the sidewall.
Thus, to summarize, Steel 1 and 2 have lowest ability compared to Steel 3 and 4 in stretch
bending for all punch radii.
In order to overcome instability issues due to the previous machine, a new device was
designed (Appendix A2) to be able to stop the test at various steps to get non propagating
cracks. This device enables to investigate damage mechanisms more accurately and cross-
sectional observations of stretch bent specimens will be shown in Chapter 3. The device
used is a conventional servohydraulic testing machine Instron (load cells: 250 kN and 50
kN) which makes it possible to apply very low displacement rates (0.05mm/s instead of
0.5mm/s with the hydraulic press). The new device is shown in Figure I.40a and the bent
specimen is presented in Figure I.40b. With this tool, only a punch with a radius of 4mm
was used and the dimension of the samples was 40x180mm for further tests.
(a) (b)
Figure I.40. (a) Stretch bending tool after machining, (b) sheet bent under tension.
50
III.D. Summary
Firstly, Steel 1 and 2 have the lowest bendability compared to Steel 3 and 4 as well as in
air-bending than in stretch bending tests. However, the lowest bendability of the two first
steels compared with the two other ones is more visible in air-bending where bending
angles differences are more pronounced.
In addition, note that stretch bending angles are clearly lower compared to the air-bending
ones. This result was expected because the stretch bending test is more severe than the air-
bending one due to the stretching imposed. It is thus necessary to use these two bending
tests to obtain comprehensive database on both bending tests and represent the most
accurately the various forming modes appearing during parts stamping.
IV. Summary of Chapter 1

This chapter provided a literature survey on bendability and gives a short presentation on
TRIP-aided steels. In addition, the origin of the banded structures was explained in the last
part of this literature review. Finally, a microstructural and mechanical analysis was carried
out on the four TRIP-aided steels studied. This chapter intended both to fix the background
of this study and to highlight the most relevant findings related to this PhD work.
From the state of the art, it was noticed that several parameters can influence the
bendability of steels such as mechanical properties, microstructure, tools, etc…making
difficult the understanding of formability issues. It seems clear that higher tensile strength
decreases the bendability of materials. However, at this stage, it is not possible to affirm
what is the role of each microstructural parameter such as the central segregation, the
banded structures, the retained austenite content and the nanohardness of various phases.
The link between the microstructure and bendability is not well known and has to be
improved to investigate more accurately damage mechanisms.
Available published data on mechanical analysis of bending tests were presented in this
chapter. These data are summarized in section I.5. and more particularly in Table I.1.
Among these studies, most of them used an isotropic hardening law to simulate forming
tests and either a Von Mises or a Hill yield criterion was chosen to account for the
anisotropy. Studies with a mixed hardening rule exhibited more accurate results than with
purely isotropic one. Almost modelling was associated with a coupled damage approach to
predict fracture in bending and simulations were carried out mainly in 3D and occasionally
in 2D.
Finally, from first characterizations conducted on the four steels studied, it appears that
Steel 1 and 2 have the lowest bendability compared to Steel 3 and 4 but no clear
relationship with their microstructures was noticed. Therefore, complementary analyses are
necessary to describe more precisely the relationship between bendability and
microstructure.
51
V. Choice of the materials and approach used in the

present study
V.A. Choice of the materials
In order to investigate damage mechanisms and relationship between microstructure and

bendability, the four TRIP-aided steels were studied both with air and stretch bending tests.
As Steel 1 and Steel 2 have the lowest bendability, they were investigated in more detail.
The microstructure of Steel 2 is less homogeneous than that of Steel 1 and it exhibits more
hard segregated bands. In order to determine the influence of hard segregation bands as a
function of their locations through the thickness, Steel 2, which is the thickest material,
was used. Surface machining was carried out in order to artificially move the central
segregation from the mid-thickness and to obtain thickness-reduced specimens with
various locations of the main hard band.
In order to conduct a comprehensive mechanical analysis of both bending tests,

constitutive equations of Steel 1 was established and then the simulation of both air and
stretch bending tests was carried out.
In addition, air-bending test simulations were also performed on thickness-reduced Steel 2
in order to investigate more severe loading conditions (bending angle up to 180° and
thinner material). For that purpose, as Steel 2 has a mechanical behaviour very close to
Steel 1, constitutive equations of Steel 1 were used to model air-bending on Steel 2.
V.B. Outline of the next three chapters
A full mechanical analysis of the air-bending test is reported in the second chapter. For that
purpose, constitutive equations of Steel 1 were first determined and then modelling of the
air-bending test without damage was carried out thanks to finite element simulation. The
modelling of air-bending test of thickness-reduced Steel 2 specimens up to 180° is also
presented in this chapter.
The third chapter addresses investigation of damage mechanisms after both tensile and
bending tests. Observations and understanding of damage mechanisms linked to the
microstructure are reported. The influence of the segregated bands on air-bending ability is
investigated and a relationship between damage, band thickness and local strain is
proposed.
Finally, a comparison between stretch bending and air-bending tests is carried out in
chapter 4. Modelling of these two tests including damage development was performed
using finite element simulation.
52
Résumé du Chapitre 1
Ce chapitre établit à la fois une étude bibliographique ainsi qu’une présentation sommaire
des aciers TRIP étudiés. L’origine des structures en bandes a été détaillée dans la dernière
partie de cette étude bibliographique. Une analyse à la fois microstructurale et mécanique
a été réalisée sur les aciers de l’étude à la fin de ce chapitre. Les objectifs de ce chapitre
ont été d’une part de fixer le contexte de l’étude et d’autre part de mentionner les résultats
les plus significatifs disponibles dans la littérature en relation avec les travaux de cette
thèse.
A partir de cet état de l’art, il a été démontré que divers paramètres tels les propriétés
mécaniques, la microstructure, les outils, etc. peuvent avoir une influence sur la capacité
en pliage des aciers. Il semble clair qu’une résistance en traction élevée diminue l’aptitude
au pliage d’un acier. Néanmoins, cette étude n’a pas permis de dégager le rôle de chacun
des paramètres microstructuraux tels que la ségrégation centrale, les structures en bandes,
le taux d’austénite résiduelle et la dureté des phases. Les liens entre la microstructure et
l’aptitude au pliage des aciers ne sont pas vraiment connus et nécessitent d’être améliorés
pour étudier d’une manière précise les mécanismes d’endommagement.
Les données disponibles dans la littérature portant sur l’analyse mécanique des tests de
pliage ont été présentées dans ce chapitre. Elles sont synthétisées dans la partie I.5 et plus
particulièrement dans le tableau I.1.
Parmi ces études, la plupart des auteurs ont modélisé des essais de mise en forme avec soit
un critère d’écoulement isotrope de Von Mises, soit un critère anisotrope de Bron Besson.
De plus, les simulations numériques réalisées avec un écrouissage mixte ont permis
d’obtenir de meilleurs résultats que celles effectuées avec un écrouissage purement
isotrope. La plupart d’entre-elles ont inclus l’endommagement dans leurs modèles pour
prédire la rupture en pliage et ces modélisations ont principalement été conduites en 3D et
occasionnellement en 2D.
Enfin, à partir des premières caractérisations réalisées sur les quatre aciers de l’étude,
nous avons montré que les aciers nommées “acier 1“ et “acier 2“ ont la plus faible
aptitude au pliage comparé aux deux autres aciers mais nous n’avons pas relevé de
relation directe avec leurs microstructures. Des analyses complémentaires sont donc
nécessaires pour décrire de manière précise les relations entre microstructure et aptitude
au pliage.
53
Chapter 2. Mechanical analysis of Steel 1 and air-bending test modelling
Chapter 2
Mechanical analysis of Steel 1 and air-

bending test modelling
An experimental database including tensile tests on smooth

and notched samples as well as shear tests was established
to determine the constitutive equations of Steel 1. These
constitutive equations enabled to simulate both air and
stretch bending tests. In a journal paper, finite element
simulation of air-bending tests has been conducted first on
Steel 1 and then on thickness-reduced Steel 2 to investigate
other bending conditions. This modelling made it possible to
satisfactorily predict stress and strain fields and bending
load-displacement curves. The constitutive equations
accounted for anisotropic elasto-plastic behaviour with
mixed hardening in this article. The possibility of using
simplifying assumptions in the model (such as an isotropic
flow criterion, pure isotropic hardening, two-dimensional
assumptions and simplified boundary conditions) was also
discussed in complementary data of this article.
Regarding the model, three cases were investigated: (i) an
anisotropic Bron-Besson model with mixed
isotropic/kinematic hardening (article), (ii) a Bron Besson
yield criterion with a purely isotropic hardening (article +
complementary data) and (iii) a Von Mises yield criterion
with mixed hardening (complementary data).
55
Table of contents
I. GLOBAL ANALYSIS OF THE AIR-BENDING TEST............................................ 57
II. COMPLEMENTARY DATA FOR MODEL PREDICTIONS ............................... 81
II.A. Experimental database ................................................................................................ 81
II.B. Comparison with predictions by other models assuming either Von Mises isotropic
yield criterion or isotropic hardening rule ........................................................................... 83
III. INFLUENCE OF SIMULATION CONDITIONS OF THE AIR-BENDING TEST

............................................................................................................................................. 90
III.A. Influence of meshing conditions................................................................................ 90
III.B. Influence of the boundary conditions ........................................................................ 92
III.C. Influence of the constitutive equations on the model predictions for the air-bending
test........................................................................................................................................ 94
III.D. Summary: influence of modelling conditions on air-bending test predictions.......... 96
IV. SUMMARY OF CHAPTER 2 ................................................................................... 97
56
I. Global analysis of the air-bending test

This part is based on a journal paper which has been submitted for publication to
International Journal of Mechanical Sciences in May 2011. The text of this article is
reproduced below.
Analysis of the air-bending test using finite-element simulation:

Application to steel sheets
D. Rèche 1), 2), J. Besson 1), T. Sturel 2), X. Lemoine 2), A.F. Gourgues-Lorenzon 1)
1)
MINES ParisTech, Centre des Matériaux, CNRS UMR 7633, BP 87
91003 Evry Cedex, France
2)
ArcelorMittal Global R&D, Voie Romaine, BP 30320,
57283 Maizières-lès-Metz Cedex, France
Corresponding author: Delphine Rèche: [email protected]

Tel: 0033 3 87 70 48 01
Fax: 0033 3 87 70 47 14
Abstract
A full mechanical analysis of the air-bending test was performed in order to determine the
stress and strain fields involved in this test. Two low alloy TRIP-aided steels and two
bending conditions were considered: the first one is commonly used to test the bending
ability (sheet thickness 1.6mm, bending angle up to 150°) whereas the second one involves
fairly different loading conditions (sheet thickness 0.75mm, bending angle up to 180°).
Constitutive equations were determined from tensile and shearing tests to accurately
represent the flow behaviour of the sheet during both bending and unloading.
Predictions from the two-dimensional simulation of air-bending were in good agreement
with all experimental measurements (load vs. displacement curves, hardness and strain field
measurements of bent then unloaded specimens). While neither using three-dimensional
simulations nor representing the contact between punch and sheet were necessary, an
anisotropic yield criterion with both isotropic and kinematic contributions to hardening is
required. In particular, accounting for kinematic hardening is necessary for the correct
simulation of air-bending of a prestrained material. The strain path experienced by the
material both at the apex and close to the initial mid-thickness is then discussed. This
method can be applied to assess local mechanical loading, prestrain effects and possibility
of damage development to any high strength steel grade.
57
Keywords
Ultrahigh strength steels, TRIP steels, air-bending, constitutive equations, finite element
simulation.
1. Introduction
Advanced high strength steels (AHSSs) have been more and more intensively used in
recent years in the automotive industry for weight and cost reduction as well as for crash
performance improvement. However, a drop of formability might be experienced under
severe conditions, in particular forming modes such as bending.
Bending is one of the most frequently used sheet-forming operations in the automotive
industry to produce structural and safety parts such as bumper beams and lateral
reinforcement beams. Development and optimisation of bending operations for a given
steel sheet requires the knowledge of its strength and of its bendability, which is generally
defined as the ratio of the minimum bending radius that can be achieved without failure, to
the sheet thickness. In the particular case of AHSSs, due to their high strength, localized
fracture might occur and affect the bendability.
The minimum bending radius can be given in various ways. Datsko and Yang [1] predicted
the minimum bending radius for a specific material, from the reduction of area in uniaxial
tension. Leu [2] studied sheet metal and showed that the minimum bending radius is
proportional to the sheet thickness and decreases with increasing the Lankford coefficient
“r” ratio. In addition, a large strain hardening exponent was reported to favour low
minimum bending radius and to delay strain localisation and thus it was expected to
increase the fracture strain [2]. This trend was not in agreement with the study of
Yamazaki et al. [3] who worked on ultra high strength steels and showed that the minimum
radius was not correlated with the strain-hardening exponent.
The mechanical analysis of bending of metallic sheets focused on the prediction of

springback after removing the tooling. Among them, Nilsson et al. [4] used a finite element
method to predict the springback in V-die bending. Li et al. [5] showed that the hardening
model in the material constitutive equations directly affects the accuracy of springback
predictions. In their particular case, the best accuracy was found with an elasto-plastic
power-exponent hardening rule. Gan and Wagoner [6] developed a method for the design
of sheet metal forming dies to produce a specified part shape, by taking springback into
account. Eggertsen and Mattiasson [7] showed the importance of the choice of both
hardening constitutive equations and yield criterion to get a good accuracy of springback
prediction. To the authors’ knowledge, no finite element analysis focussed on the stress
and strain fields through the specimen thickness. Only analytical and semi-analytical
models were used to this aim. Although these models can save computation time, they
often require simplifying assumptions which can affect the accuracy of the results. For
example, some authors such as Chakrabarty et al. [8], Chakrabarty et al. [9] and Ragab and
Saleh [10] assumed that there is no variation of the specimen thickness during bending. As
will be shown in the present paper, such assumptions are not fulfilled in the case of air-
bending of AHSSs, so that a more accurate mechanical analysis is required to investigate
stress and strain fields in the whole air-bending specimen. That is why finite-element
method was used to study successfully bending tests. Little work has been done in the past
to give a full mechanical analysis of the air-bending test. Zhao and Lee [11] simulated
58
three-point bending of two types of sheet metals: mild and high strength steels with plane-
strain hypothesis by using three constitutive equations involving (i) isotropic hardening,
(ii) kinematic hardening or (iii) a combination of both. They showed that a combination of
isotropic and non linear kinematic hardening accurately predicts the load-displacement
curve. Wang et al. [12] showed that such numerical simulation can be carried out in plane-
strain conditions. In their modelling they assumed an isotropic hardening rule and also that
all bent fibres had the same curvature centre. In the following, it will be shown that this
assumption is not true. Hambli et al. [13] simulated the air-bending test assuming a 3D
elastic-plastic model with a Mises criterion, an isotropic hardening law coupled with a
damage criterion. Thuillier et al. [14] characterized the mechanical behaviour of an
aluminium thin sheet. They used the GTN model to predict the ductile rupture in bending.
In addition, they chose a purely isotropic hardening rule and a Von Mises yield criterion in
their work. To the authors’ knowledge, no study has reported yet simulation of the air-
bending test after tensile prestraining and with a kinematic contribution to hardening.
In order to predict the effect of microstructural non-homogeneity such as a central hard

band or banded structures on the bendability of AHSSs, a local fracture criterion must be
developed. This requires accurate assessment of the stress and strain fields that develop
during bending. The purpose of the present study was to build a methodology for the
mechanical analysis of the air-bending test, based on the finite element analysis technique,
and to apply it to the case of low alloy transformation-induced plasticity (TRIP)-aided
steels. The results are intended to be used to derive a fracture criterion in air-bending for
this family of steels.
The complexity of the model that has to be used was thoroughly analysed, by considering
the need of including (i) plastic yield anisotropy, (ii) accurate description of the contact
between the sheet and the tools, (iii) 3D formulation of the problem, (iv) both isotropic and
kinematic contributions to hardening. Following a preliminary mechanical analysis of the
test that gave the requirements for relevant constitutive equation determination, an
experimental database was built to identify constitutive equation parameters on a sound
and relevant basis. Finite element simulation of the air-bending tests was then performed
and validated against all available experimental measurements. The influence of tensile
prestraining before air-bending was also addressed.
2. Materials and experimental methods
2.1. Materials
The present study focused on two cold rolled TRIP-aided steels of different thickness (1.6
and 2mm) but from the same heat (C~0.2 wt%, Mn~1.6 wt%, Al+Si~1.5 wt%) and with
similar mechanical properties (Table 1), in particular along the transverse direction, which
was also the bending direction mainly used in the present study. In order to investigate
bending of a thin sheet and to be able to reach high fracture strain, modelling was also
performed with a bending up to 180° on thickness-reduced specimens of Steel 2. This steel
was previously used in Rèche et al. [15] to investigate the role of hard bands where the
central hardest band was artificially shifted to various locations in the sheet thickness in
order to obtain a “model material”. To keep a sufficient specimen thickness for air-bending
test, the thicker base material (Steel 2) was used. Surface machining followed by chemical
polishing made it possible to investigate the bending behaviour of various specimens (the
central hard band being shifted away from the mid-thickness) by reducing the thickness
59
from 2mm down to 0.75mm. The model was first validated with conventional bending
conditions (Steel 1) and then with more severe ones (thickness-reduced Steel 2). In the
following, only full-thickness Steel 1 and thickness-reduced Steel 2 are considered.
0.2% proof Tensile niform Fracture

stress (MPa) strength (MPa) longation (%) elongation (%)
Steel 1 (1.6mm)
RD 467 866 5.0 20.0
TD 546 851 5.0 19.2
Steel 2 (2mm)
RD 464 816 6.6 22.9
TD 521 834 3.6 17.0
Table 1: Room temperature tensile properties of Steel 1 and Steel 2 (all in full thickness) along rolling (RD)
and transverse (TD) directions (ISO 20x80mm tensile specimens)
2.2. Characterisation of the mechanical behaviour
2.2.1. Air-bending tests
Air-bending tests were carried out at room temperature using a universal tensile machine
and 50x100mm specimens. The load and punch displacement were continuously measured.
Two rotary rolls (in grey in Figure 1a) and various punches were used. The tests were
performed without lubrication. The spacing between either roll and the punch was equal to
the thickness of the considered sheet + a small gap of 0.1mm, i.e., 1.7mm for Steel 1 and
0.85mm for thickness-reduced Steel 2. Before the tests, specimen edges were ground to
avoid premature failure that could be due to cutting edge-induced damage effects
sometimes encountered in AHSSs [16]. For Steel 1, five specimens were tested along both
RD and TD with a punch radius of 0.1mm. Note that, in this case, the bending angle, α,
was limited to 150° due to the punch wedge (30°, see Figure 1b). A thinner punch
developed for that purpose was used to bend thickness-reduced Steel 2 along TD
(Figure 1c). This punch enabled to reach a bending angle, α, up to 180°. Therefore, with
thickness-reduced Steel 2 and this punch, higher strains can be reached at the end of the
test but provided that the hard bands were sufficiently far from the outer surface, no
cracking occurred even under these very severe bending conditions.
The evolution of strain fields during air-bending of Steel 1 and Steel 2 was analysed in
three dimensions with Aramis software [17] using a maximum allowable strain of 100%.
Two CCD cameras allowed monitoring surface deformation thanks to a speckle laid on the
surface of the specimen. Strain fields were then computed using digital image correlation.
The acquisition frequency was 2 images per mm of punch displacement, i.e. about 52
images for air-bending of Steel 1.
60
α 30°
10mm Punch tip
a b c
Figure 1: (a) schematics of the air-bending test with α = bending angle, (b) punch with radius = 0.1mm,
punch thickness = 10mm and (inlet) a bent sample, (c) thin punch with side stiffeners and radius = 0.4mm,
punch thickness = 0.8mm and (inlet) a 180° bent sample.
As a first approximation, the maximal strain reached at the outer surface after air-bending
was determined thanks to a simple analytical formula. Marciniak and Duncan [18]
proposed the following equation:
ε = ln(1 + y / R) (1)
where y is the distance between a bent fibre and mid-thickness and R is the bending radius
at mid-thickness. In the present case, to assess the strain at the outer surface, this equation
could be rewritten as:
ε = ln(1 + t /(2 Ri + t )) (2)
where t is the initial thickness and Ri is the inner radius of the bent specimen.
The initial thickness of the Steel 1 specimen was equal to 1.6mm and the inner radius
profile (curvature radius of inner fibre) was measured manually by determination of the
contour of the sheet in polished cross-sections to be equal to 2.05mm (average of three
specimens with very low experimental scatter) for a bending angle of 118° (TD sample).
Hence, for these bending conditions, the maximum hoop strain (at the outer surface) was
estimated to be 0.25. Due to the very low bend radius of thickness-reduced Steel 2 bent up
to 180°, this formula was not valid. No first estimation for this maximal strain could thus
be given. Finite element simulation was required to give a valid assessment of the hoop
strain in such bending conditions.
2.2.2. Experimental database dedicated to FE simulation of the bending test
To determine constitutive equations to be used in the FE simulations of bending, tensile

tests on smooth samples along RD, TD and 45° directions were performed at room
temperature to evaluate isotropic hardening as well as strength and strain anisotropy of the
material. The tensile specimens had a gauge area of W0 =12mm in width and 64mm (see
Fig. 2a) in length and the initial elongation rate was 10-4 s-1. Elongation was measured
using an axial extensometer with a L0 = 12mm gage length. A second extensometer was
used to measure the reduction in width of the specimen ( ΔW ) to evaluate the
corresponding Lankford coefficient (“r” ratio). In the following ( ΔL ) denotes the axial
displacement measured with the extensometer.
61
U-Notched tensile (NT) specimens (18x100mm with one full thickness notches of radius
2mm on either edge and width of the remaining ligament = 10mm, see Fig. 2c) cut along
TD, RD and 45° directions were also tested in tension at room temperature to reach higher
local strains (up to 0.40), necessary to accurately model the air-bending test (see section
2.2.1.). These specimens were also used to investigate the effect of stress multi-axiality on
plastic yielding. The prescribed notch opening rate was 0.5mm.min-1.
In-plane reverse shear tests (100x30mm specimens with a gauge part of 4.5mm in height)
were also carried out along RD under a prescribed loading rate of 0.05 kN.s-1 in order to
assess the Bauschinger effect and to determine the kinematic contribution (if any) to
hardening at high strains.
1000 0
800 -0.02
Wo )
ΔW //Wo
600 -0.04
400
RD (1+ΔW
ln (1+ -0.06 RD: r=0.65
TD
TD: r=0.85
45° 45°: r=0.68
200 -0.08
isotropic_case
a b
0 -0.1
0 0.1 0.2 0.3 0 0.05 0.1 0.15 0.2
ΔL/Lo ln (1+ ΔL //Lo
(1+ΔL Lo )
1200 600
γ = 0.1
400
1000 γ = 0.2
shear stress (MPa)
800 200 γ = 0.3

F/So (MPa)
600 0
RD
400 TD -200
45°
200 -400
c d
0 -600
0 0.5 1 1.5 -0.6 -0.4 -0.2 0 0.2 0.4
Notch opening displacement (mm) γ
Figure 2: (a) Engineering tensile stress-strain curves, (b) strain anisotropy (smooth tensile specimens), (c)
Tensile curves on NT specimens where F is the load and So is the initial cross section between notch roots,
(d) cyclic shear curves along RD. γ corresponds to the ratio between the horizontal displacement and the
specimen width = 4.5mm.
Figure 2a shows that the tensile response is almost isotropic. However, the “r” ratio (figure
2b) was significantly lower than 1 (between 0.65 and 0.85) yet with small differences
between loading directions. Slight stress anisotropy was observed for notched specimens
(Figure 2c). The three shearing tests evidenced a strong Bauschinger effect (Figure 2d), so
that constitutive equations should include both isotropic and kinematic contributions to
hardening to accurately describe the mechanical behaviour of the material, especially in the
case where bending after prestraining has to be modelled.
62
3. Determination of constitutive equations for Steel 1
For the low alloy TRIP-aided steels considered here, although microstructural
inhomogeneity and microstructural instability affect the local strain behaviour, the phase
transformation of retained austenite into martensite was not explicitly taken into account in
the model. The material was considered as a homogeneous medium and the effect of phase
transformation on the yield behaviour was implicitly taken into account in the model
parameters.
3.1. Constitutive equations
For rolled sheets, it has been proposed [19] to use the Hill quadratic yield criterion [20]
together with mixed isotropic/kinematic hardening. However this criterion can hardly
represent both stress and strain anisotropy so that more complex yield functions have been
proposed in literature. Barlat et al. [21] developed an anisotropic criterion that is more
appropriated for aluminium alloys. This yield function was further extended by Karafillis
and Boyce [22] and Bron and Besson [23]. In their model, they used 16 material
parameters to represent a complex yield surface and the convexity was proved. These
complex yield surfaces have principally been used assuming pure isotropic hardening.
Recently, some authors used anisotropic yield criteria coupled with a mixed
isotropic/kinematic hardening. For example, Eggertsen and Mattiasson [24] modelled the
bending / unbending behaviour with various anisotropic yield criteria including a
contribution of kinematic hardening. They demonstrated that the model with purely
isotropic hardening yielded to poorer prediction of the load-displacement curves than the
one with a model including mixed hardening. Williams et al. [25] used a non-linear
isotropic/kinematic hardening model associated with the Yld2000-2d anisotropic model
developed by Barlat et al. [26] and Yoon et al. [27] to predict the axial crush response of
hydroformed aluminium alloy tubes. Vladimirov et al. [28] used a non-linear isotropic and
kinematic hardening associated with an anisotropic Hill yield criterion to model sheet
metal forming such as deep drawing tests.
In the present study, according to the experimental results, an anisotropic yield criterion
coupled with kinematic+isotropic hardening was used as in [29]. In order to represent
kinematic hardening, a back stress X was introduced [30]. The yield surface was
expressed using the difference ( B ) (i.e. effective stress), between the Cauchy stress, σ ,
and the back stress: B = σ − X . In order to account for anisotropic plastic yielding, it was
necessary to use an anisotropic scalar measure of stress to define the yield surface. The
model proposed by Bron and Besson [23] was used in the following.
For any symmetric second order tensor, T , the anisotropic scalar measure ( T E ) of T is
defined as:
⎛K ⎞
a 1/ a
( )
TE (T) = ⎜⎜∑ α k TEk ⎟
⎟ (3)
⎝ k =1 ⎠
In practice, only two functions ( K = 2 ) were used. α k are positive coefficients with a sum
equal to 1, i.e. α 1 = α and α 2 = 1 − α in the present case. The function T E (T ) is positive and
homogeneous of degree 1. TEk are secondary anisotropic scalar measures. One first defines
two modified deviators:
63
T k = L : T , k = 1, 2 (equivalent to Tijk = Lkijkl Tkl ) (4)

k
where the fourth order tensor L k is expressed as follows using Voigt notations:
⎛1⎛ k 1 k 1 k ⎞
⎜ ⎜ LL + SS ⎞⎟
k
c c − c − c 0 0 0 ⎟
⎜3⎝ ⎠ 3 SS 3 LL ⎟
⎜ 1 k 1⎛ k 1 k
0 ⎟
k ⎞
− c ⎜ +
c c ⎟ − c 0 0
⎜ 3 SS 3 ⎝ SS TT ⎠ 3 TT ⎟
Lk = ⎜ 1 k 1 k 1⎛ k k ⎞ ⎟ (5)
⎜ − c − c ⎜ + c ⎟ c 0 0 0 ⎟
⎜ 3 LL 3 TT 3 ⎝ LL TT ⎠
⎟
⎜ 0 0 0
k
0 0 ⎟
⎜ c TL ⎟
⎜ 0 0 0 0
k
0 ⎟
⎜ c LS ⎟
⎜ 0 0 0 0 0
k
⎟
⎝ c
ST ⎠
The eigenvalues Tki of T k are then computed: Tk1 ≥ Tk2 ≥ Tk3 . T E1 and T E 2 are respectively
computed as:
1 / b1
⎡1 ⎛ b1 b1 b1 ⎞⎤
TE1 = ⎢ ⎜ T12 − T13 + T13 − T11 + T11 − T12 ⎟⎥ (6)
⎣2 ⎝ ⎠⎦
1 / b2
⎡ 3b2 ⎛ 1 b2 b2 b2 ⎞⎤
TE 2 =⎢ b ⎜ T2 + T22 + T23 ⎟⎥ (7)
⎢⎣ 2 + 2 ⎝
2 ⎠⎥⎦
a , b1 , b2 and α are four material parameters that influence the shape of the yield surface
and c1IJ, 2 ( I , J = L, T , S ) parameters control yield anisotropy. The model used in the present
study was simplified by assuming a = b1 = b2 .
As mechanical tests were performed along in-plane directions, coefficients corresponding
to shear along the thickness direction could not be adjusted. It was therefore assumed that
c 1ST,2 = c 1LS,2 = 1 . Following these simplifications, 10 out of a total of 16 parameters had to be
optimised to represent yield anisotropy.
The yield surface was then expressed by:
φ = B E − R ( p ) with BE = TE (B) (8)
In Eq. 8, p is the cumulative plastic strain and R is the current flow stress. Plastic flow
was then computed using the normality rule as:
∂BE (9)
ε& p = p&
∂σ
Where p& is the plastic multiplier such that ε& p : B = p& BE (equivalent toε& p ij Bij = p& B E )
Non-linear (Armstrong–Frederick type) kinematic hardening was included in the model
using two back-stresses [19] to accurately represent experimental data. Their evolutionary
equations are given by:
2 2
X = X 1 + X 2 with X& 1 = C1ε& p − p& D1 X 1 and X& 2 = C2 ε& p − p& D2 X 2 . (10)
3 3
64
where C1, D1, C2 and D2 are material parameters. Finally an additive form for
the R ( p ) function (isotropic hardening) was chosen as:
R ( p ) = R0 + Hp + Q (1 − exp(− bp )) (11)
Isotropic strain hardening is principally given by the exponential term at low strains and by
the linear term at high strains. This hardening law was also successfully used by Thuillier
et al. [31] to model bending tests.
The model thus included two non-linear kinematic hardening variables (involving four
material parameters: C1 , D1 , C 2 , D2 ), one non-linear isotropic hardening variable (involving
two material parameters: Q, b ) and one linear isotropic hardening variable (involving one
material parameter: H ). The use of a linear variable was found to be mandatory to account
for large strains experienced both during tension of notched specimens and during bending.
R0 corresponds to the initial value of the yield stress.
3.2. Finite element simulation procedure
The in-house finite element software Zset [32] was used to simulate all mechanical tests. A
fully implicit integration scheme was used to integrate the material constitutive equations.
All calculations were done using an updated Lagrangian formulation with linear brick
elements using selective integration (finite strain B-bar method, [33]). Tensile tests on
smooth specimens were simulated up to (experimental) necking using one volume element.
Notched (NT) and shearing specimens were analysed using 3D FE simulation;
corresponding meshes are shown in Figure 3. In addition to usual symmetry conditions,
displacements were prescribed over the AB segment along direction 2 for notched tensile
specimens (Figure 3a) and along direction 1 for the shear specimens (Figure 3b).
A B
A B
This notch is
used to
U2
attach the
C D
extensometer
U1
U3 a b
C D
Figure 3: two-dimensional view of (a) meshing of the NT specimen (1/8 of the specimen due to symmetry
with smallest element size = 130 x 150 x 800 µm close to the U-shaped notch) and (b) meshing of the
shearing specimen (1/2 specimen with element size = 500 x 450 x 800 µm)
3.3. Identification of model parameters
In a first step, the yield criterion parameters were determined. To this purpose, a first
approximation of strain hardening had to be made. In this first step, it was considered to be
purely isotropic (there were little changes in loading path in the tests considered here). The
65
value of R0 was set to 499 MPa. The Von Mises yield criterion was used to optimize
isotropic strain hardening parameters thanks to tensile tests on smooth and notched
samples along all directions. Then, the parameters of the Bron Besson yield criterion were
identified. The values of a and α influence the shape of the yield surface. c Ik,=J1=, 2L ,T
parameters were optimised from tensile tests on smooth and notched samples (RD, TD,
45°) by including experimental data on strain anisotropy i.e. measurements of “r” ratio.
In a second step, hardening parameters were determined. To this aim, isotropic hardening
parameters had to be modified and C i and Di (i=1, 2) coefficients were optimized from
shearing tests.
All model parameters were optimized using the optimizer integrated in the in-house finite
element software Zset with the simplex algorithm.
1000 1200
1000
800
800
F/So (MPa)
600
600
experiment
400
400 model prediction
experiment
200 model prediction 200
a b
0 0
0 0.05 0.1 0.15 0.2 0 0.5 1 1.5
ΔL/Lo Notch opening displacement (mm)
600
400 experiment
shear stress (MPa)
model prediction
200
-200
-400
c
-600
-0.5 -0.25 0 0.25
γ
Figure 4: Experiment vs. model predictions for Steel 1: tensile curves (a: smooth and b: NT specimens tested
along TD) and c: cyclic shearing curves along RD.
The model together with the optimized set of material parameters (Table 2) well
k =1, 2
reproduced tensile and reverse shearing curves (Figure 4). Note that cTT , LL , SS ,TL parameters
were close to 1, consistently with the slight anisotropy of the material. From numerical
simulations, the local amount of strain reached 0.40 during tension on NT specimens and
shear tests. The corresponding strain range was thus wide enough to encompass values that
were expected to be reached during the air-bending test (max. 0.25 for Steel 1, see Section
2.2.1). The same hypothesis remained to be validated for Steel 2 using finite element
calculations (see below).
66
Elastic properties E = 210 GPa , ν = 0.3

Isotropic hardening R = 499 MPa, H = 110 MPa
0
b = 8.4, Q = 382 MPa
Kinematic hardening C1 = 718 MPa, D1 = 9.4
C 2 = 9900 MPa, D2 = 66.7
Plastic anisotropy a = 6.1, α = 0.55
1
cTT = 0.99, c1LL = 0.96, c1SS = 0.92, cTL
1
= 1.14, c1LS = c1ST = 1
2 2 2 2 2 2
cTT = 0.99, cLL = 1.04, cSS = 1.17, cTL = 1.04, cLS = cST =1
Table 2: Optimized material parameters for constitutive equations (Steel 1)
4. Full mechanical analysis of the air-bending test
4.1. Mechanical analysis of the bending test in usual conditions (Steel 1)
4.1.1. Numerical representation of the test
As air-bending tests are more severe along TD because cracking usually occurs earlier than
along RD (at least in AHSSs), only simulation of air-bending along TD was carried out for
Steel 1.
As no variation in specimen width was experimentally detected during and after the air-
bending test, a plane strain assumption was considered to be reasonable in numerical
simulations of the test and was first used. Only one half of the specimen and punch and one
roll were meshed (Figure 5) due to symmetry conditions. The size of smallest elements
was about 25 x 50 μm just next to the punch (i.e. there were 32 elements along the sheet
thickness). For 3D simulations, 5 elements were considered along the specimen width.
Finite element simulations were performed for the same displacement range as for
experiments.
Tooling (punch + rolls) was assumed to behave elastically with a Young’s modulus of
210 GPa and a Poisson ratio of 0.3. Two contact zones were considered:
• Between the sheet and the rolls: as rolls rotated almost freely during the test, friction
could be neglected. Consequently, rolls were fixed and the corresponding friction
coefficient was set to 0.
• Between the sheet and the punch: the contact area was first fully represented in the
mesh. It was shown after modelling that only four nodes of the 2D mesh of the sheet
were in contact with the punch (half meshing). In 3D, this corresponds to four lines
of nodes along the direction of the sample width. Two conditions were then
compared:
(i) full representation of the punch (see Figure 5) assuming a Coulomb
condition with various values of the friction coefficient between the punch
and the sheet.
(ii) prescribed vertical displacement on the four nodes in 2D (four lines of nodes
in 3D) close to the symmetry axis.
67
punch
Transverse
direction
roll
sheet
Figure 5: Meshing of the air-bending test
In the following, the “reference” calculations conditions denote a 2D plane strain

assumption, a mesh with 32 elements in the sheet thickness and boundary conditions (i).
4.1.2. Modelling results
Very good agreement between experimental curves and 2D model predictions was
obtained except at the end of the curve where the decrease in load was slightly
underestimated by the model (Figure 6). In fact, crack initiation and stable crack
propagation occurred during this stage of the test [15] and the damage-free model not
expected to be able to describe this.
3.5
3 Steel 1
2.5
stable cracking
Load (kN)
2
during
1.5 experiment
experiment
1 simulation: 3D mixed hardening
0.5 simulation: 2D mixed hardening
simulation: 2D isotropic hardening
0
0 10 20 30
Punch displacement (mm)
Figure 6: Experimental and predicted load vs. displacement curves both in 2D and 3D with a mixed
isotropic/kinematic hardening rule (boundary condition (ii)) and in 2D with an isotropic hardening rule
(boundary condition (i)).
Hardness measurements were carried out through the thickness on cross-sections of bent
specimens of Steel 1. The softest area (i.e. that had experienced less work hardening) was
considered as “the neutral fibre” after bending and unloading. The location of the
minimum cumulated plastic strain predicted after bending and then unloading by
simulation well corresponded to the (experimentally measured) softest region, at about
850µm from the outer surface (Figure 7). Due to stretching of the outer part of the
specimen and compression of its inner part, the neutral fibre did not stay at mid-thickness
and shifted toward the inner surface. Furthermore, the material next to the mid-thickness,
68
initially in the compressive zone, was moved to the other side of the “neutral fibre” i.e. was
eventually pulled in tension during the bending test.
The minimum value of hardness measured on bent samples was very close to the hardness
of the undeformed material. This correlated well with the simulation results which showed
that the minimum value of the cumulated plastic strain was very low (~0.015) after the end
of the test.
0.35 350
Hardness on bent sample
cumulated plastic strain
0.3 300
0.25 250
Hv 0,5
0.2 Initial hardness profile 200
0.15 150
0.1 100
0.05
strain 50
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
distance from outer surface (mm)
Figure 7: Evolution of the cumulative plastic strain p predicted using the “reference” calculation conditions,
hardness through the thickness of bent + unloaded specimen and initial hardness profile. The “neutral fibre”
was shifted from the mid-thickness of the sheet toward the inner surface (initial compressive zone). The
dashed black line and the dashed grey line respectively represent “the neutral fibre” and the mid-thickness in
the updated geometry. The vertical grey line represents the inner surface after thickness reduction due to
bending.
The radius of curvature at the outer surface, at the end of the test (point A in Figure 8a) on
bent then unloaded specimens was experimentally determined from metallographic
observations to be 3.84mm. This value very well agreed with model predictions (also after
unloading: 3.92mm). Note that this radius did not exactly correspond to the sum of the
inner radius and the thickness of the sheet (i.e. 2.05 + 1.6mm). This difference was due to
the fact that the centres of curvature were not the same for inner and outer skin of the
specimen. This was mainly due to the high level of plastic strain that was locally reached
during the test. Thus, even for moderately severe bending conditions used for Steel 1, the
analytical formulae (Eq. 2) and the analytical approach from Ben Bettaieb et al. [34] could
only be used as a first approximation. However, the value predicted from Eq. 2 (0.25) was
in fair agreement with model prediction (∼0.27). In more detail, experimental measurement
of strain fields very well agreed with model predictions, both concerning the evolution of
maximum principal strain at the apex and the strain gradient perpendicularly to the apex
(Figure 8b). Consequently, the model together with reference calculation conditions
accurately predicted all experimental data concerning this air-bending test.
69
0.3 0.3
A
maximal principal strain

0.25 0.25
experiment 1
0.2 experiment 2 0.2
experiment 3
0.15 experiment 4 0,25 from 0.15 experiment 1
simulation Eq. 2 experiment 2
0.1 0.1 experiment 3
experiment 4
0.05 0.05
a simulation
b
0 0
0 10 20 30 0 1 2 3
displacement (mm) distance from apex (mm)
Figure 8: Evolution of strain during the air-bending test. Strain field measurements vs. model prediction
(reference calculation conditions).
(a) Evolution of a material point taken at the apex. The grey point corresponds to the value given by equation
(2) in section 2.2.1, (b) Strain along a line perpendicular to the apex (point A in (a)). Curvilinear abscissa in
mm.
4.2. Effects of meshing and boundary conditions
To investigate the effects of the numerical representation of the test, simulations were first
carried out by using boundary conditions (i) with two different friction coefficients
between the sheet and the punch. As no lubrication was used during the test between the
punch and the sheet specimen, a high friction coefficient was first assumed (0.25)
according to Gieck [35], Muller [36] and Stöcker et al. [37] who gave friction coefficients
values between 0.20 and 0.30 for two steels without lubrication. However, the actual value
of the friction coefficient was not known and the value of 0.25 was also tentatively
replaced by a lower one (0.1). Similar results were obtained by using both values so that
the value of 0.25 was kept for the rest of the 2D calculations with boundary condition (i).
In addition, similar results were obtained by replacing the punch by prescribing the
displacement nodes located just below the tool (boundary conditions (ii)).
Additional 3D calculations were then carried out, using boundary conditions (ii) to limit
the number of elements. The load-displacement curves predicted using 2D and 3D
calculations were almost identical (Figure 6), which indicated that a 2D calculation was
sufficient for modelling the air-bending test.
To summarize, this sensitivity study made it possible to perform accurate modelling of this
air-bending test using a 2D calculation with the plane strain hypothesis. The friction
coefficient between the sheet and the punch had little influence in the range investigated.
Moreover, the punch could be replaced by prescribed displacement of the four nodes
closest to the symmetry axis.
4.3. Role of kinematic hardening
4.3.1. Effect on the global response
To check whether it was actually necessary to take kinematic hardening into account in air-
bending test even without prestraining, purely isotropic hardening was alternatively
assumed. It was represented by the following equation:
70
R ' ( p ) = R0 + H ' p + Q ' (1 − exp (− b' p )) (13)
The initial value R0 (i.e. 499 MPa) was kept and the values of parameters H ' , Q' and b'
were optimized on a tensile test on smooth specimen along TD. The initial Bron Besson
parameters (given in Table 2) were also used in these constitutive equations. After
optimization, H ' , Q' and b' were respectively set to 800 MPa, 393 MPa and 23. While
these parameters were determined from a test along a specific direction, they can still be
used with acceptable accuracy for loading along other directions. From Figure 6, the load
vs. punch displacement curve was not as well predicted by assuming purely isotropic
hardening as by assuming mixed hardening. In fact, the isotropic hardening rule led to an
overestimation of the load compared to the mixed hardening rule.
4.3.2. Effect on springback
In the literature, Geng and Wagoner [38] and Eggertsen and Mattiasson [24] reported that
taking kinematic hardening into account improved the prediction of springback after
unloading compared to using a purely isotropic hardening rule. In the present study, one
test was interrupted by unloading after a prescribed punch displacement of 20mm
(corresponding to a bending angle α = 88°, to avoid any damage development in this
specimen). The load/displacement curve was well represented up to this value of punch
displacement assuming either mixed or pure isotropic hardening: both models predicted an
angle equal to 89°. After unloading, the bending angle was measured to 78°, to be
compared to 79° using mixed hardening and 81° using isotropic hardening. The agreement
was therefore slightly better with mixed hardening. However, the difference was rather
small, so that this air-bending test did not appear as an appropriate tool to study the
influence of the hardening rule on the prediction of springback effects.
4.3.3. Effect on the global response after prestraining
In industrial stamping, some regions of the sheet blanks may have already been deformed
before bending. It seems therefore necessary to investigate the influence of prior straining
on the subsequent behaviour in air-bending. For this purpose, uniaxial tensile tests were
carried out along TD using Steel 1 up to an elongation of 10%. Then, air-bending
specimens (using the same geometry as before) were cut from the gauge part of the
prestrained tensile specimens and tested in air-bending until microcracking, in the same
conditions as previously described. FE simulation of this test was performed in the
“reference” conditions but assuming either mixed or purely isotropic hardening. The
reduction in thickness which occurred during prestraining was taken into account in the
model. The corresponding curves are given in Figure 9. With a purely isotropic hardening,
the yield stress was largely overestimated, while the first half of the curve was very
satisfactorily described using the mixed hardening rule. At the end of the test, both curves
were almost superimposed and overestimated the experimental one. This was attributed to
the fact that the model did not account for damage development in the material [15].
Straining prior to bending affected the bendability of the sheet (i.e. damage developed
earlier than in the bending specimens that had not been previously strained). Cracking was
observed for lower values of punch displacement after 10% of prestraining than for the as-
received material. This was consistent with the work of Friedman and Luckey [39] on
aluminium alloys.
71
The true strain reached at the outer surface at microcracking for the as-received material
was evaluated to be 0.27 with the model prediction. If the amount of true strain introduced
by tensile prestraining is subtracted from this value, one gets a value of true strain equal to
0.175. The punch displacement yielding that value in the as-received material was found to
be equal to 20.5 mm. This value was in very good agreement with the punch displacement
at microcracking of the prestrained specimen (Figure 9). This means that, for the particular
case of tensile prestraining along the same direction as that of hoop strain applied during
bending, the effect of prestraining can be satisfactorily predicted by a critical cumulative
plastic strain criterion. However, due to the kinematic contribution to hardening, (i) the
load-displacement curve can only be predicted with the mixed hardening rule and (ii) this
simple result can be extended neither to prestraining along other directions nor to other
prestraining conditions.
3.5
2.5
Load (kN)
1.5
experiment
1
simulation mixed hardening
0.5 simulation isotropic hardening
0
0 5 10 15 20 25
Displacement (mm)
Figure 9: Experimental and predicted load / punch displacement curves of a prestrained and then air-bent
specimen. Model predictions assuming either purely isotropic or mixed hardening rules are respectively
represented with a dotted grey line and a continuous grey line.
4.3.4. Effect on local stress-strain paths
The evolution of hoop stress vs. hoop strain is plotted in Figure 10 for various nodes
located slightly below the mid-thickness (toward the inner surface) along the symmetry
axis. At the beginning of the test, all considered nodes were located in the compressive
zone but some of them were eventually loaded in tension (consistently with Figure 7) after
a slight compressive plastic strain. This loading path confirmed that using kinematic
hardening was necessary to accurately represent the local stress/strain path next to the mid-
thickness.
1200
800 0.2mm 0.1mm

Hoop stress (MPa)
400 0.3mm
0
0.4mm
-400
-800
-1200
-0.04 -0.02 0 0.02 0.04
Hoop strain
Figure 10: Loading paths for various nodes initially next to the mid-thickness, along the symmetry axis. The
nodes were initially located at distances from mid-thickness of, respectively, 0.1, 0.2, 0.3, and 0.4mm toward
the inner surface.
72
To summarize, integration of kinematic hardening in the constitutive equations slightly

improved the accuracy of the global response in terms of load vs. punch displacement
curves as well as the prediction of springback. In case of non-proportional loading paths
(as for prestraining before air-bending), kinematic hardening is needed to accurately
predict the global response. Finally, to accurately predict local stress and strain path next to
the mid-thickness of the sheet, kinematic hardening is also required.
4.4. Using the model to assess local stress and strain fields during air-bending
1500
stress strain predicted by 0.3
1000 the model “initial 0.2
Hoop stress (MPa)
coordinates”
500 strain 0.1
Hoop strain
predicted by
0 0
Eq. 2 “initial
-500 coordinates” -0.1
-1000 strain predicted by -0.2
-1500 the model “updated -0.3
geometry”
-2000 -0.4
0 0.4 0.8 1.2 1.6
Distance from outer surface (mm)
Figure 11: Hoop stress (black line) and hoop strain (grey line) predicted by the model in the updated
geometry, along the symmetry axis, for a punch displacement of 26mm (end of the test), before unloading.
For comparison purposes, the corresponding hoop strains predicted by Eq. 2 (grey diamonds) and predicted
by the model (grey crosses) are also represented in the initial coordinates (undeformed state), Steel 1. The
inner surface in the updated geometry is represented by the vertical dashed grey line.
Calculation of the strain using Eq. 2 yielded a maximal value close to 0.25, in close
agreement between the value given with both model prediction and Aramis measurements
(0.27). The difference between these values was due to an approximation made with the
analytical formula. The latter only applied when the neutral fibre stays at the mid-
thickness, which was not the case with the present bending conditions. Regarding the strain
gradient given by the model prediction and by Eq. 2 in the initial coordinates (i.e. no
decrease in thickness was taken into account), different curves were obtained and the
analytical formula clearly underestimated the strain with respect to the numerical
prediction (Figure 11). It is therefore necessary to account for the change in the sheet
thickness to well predict stress and even strain fields.
As tensile tests on notched specimens were performed up to a local strain of 0.4, it can be
concluded that the numerical simulation of the air-bending tests carried out on Steel 1 was
actually performed with constitutive equation parameters that were identified in a relevant
range of stress and strain states.
4.5. Application to the mechanical analysis of a more severe air-bending test on a thinner
sheet (Steel 2)
The model was then applied to thickness-reduced Steel 2 under severe bending conditions
(i.e. up to 180° bending). As Steel 1 and 2 had the same heat treatment, the same chemical
73
composition and similar mechanical behaviour, the same constitutive equations and
parameters were used as for Steel 1. The plane strain hypothesis was also used in this case
as no variation in the sample width was observed after bending. Continuous experimental
strain measurements were not made possible because rolls were too close to the specimen
to allow recording the strain evolution with the two cameras. The only way to obtain such
information was to take one picture of the specimen before the test and one of the
deformed specimen after bending and unloading. Thus, only the strain field after the end of
the test was accessed experimentally.
4.5.1. Mesh and boundary conditions
As previously, only one half of the specimen and punch and one roll were meshed. These
calculations were performed with 32 elements in the thickness and the size of smallest
elements was about 31 x 23μm just in front of the punch (close to the apex). The tools
(punch + rolls) were assumed as previously to behave elastically. Here, the punch radius
was equal to 0.4mm and the friction coefficient between sheet and punch was set to 0.25 as
for air-bending test on Steel 1. As the punch diameter was similar to the sheet thickness,
only boundary condition (i) was used, i.e. the punch was not replaced by prescribed node
displacement. As for Steel 1, the friction coefficient between sheet and rolls was set to
zero.
4.5.2. Mechanical analysis
Experimental and predicted load vs. punch displacement curves were in good agreement
(Figure 12a) in the first part of the test (punch displacement up to 15mm). The change in
slope (after a punch displacement of about 15mm) was qualitatively well predicted but still
quantitatively underestimated. In fact, the maximum strain achieved during this test is
higher than 0.4 so that constitutive equations are used here beyond the strain range used for
parameter identification. The discrepancy at the end of the test could also be due to the
difficulty to well represent the contact conditions for such severe bending test up to 180°.
Indeed, at the end of the test, friction between the sheet and the rolls might have occurred
as the contact area increased. In addition, the contact between the sheet and the punch
moved from its initial location (Figure 12b). At the end of the test, the inner radius of the
bent specimens was smaller than the punch radius. This phenomenon underlined the
difficulty to perform such calculations because of contact issues.
2
1.5 experiment
simulation
Load (kN)
0.5
0
a 0 10 20 30
b
Figure 12: (a) Load vs. punch displacement curves for experiment and 2D simulation for a 180° bending
angle, (b) The contact area between punch and sheet has moved during the test. Initial contact is illustrated
with white arrow and final contact is showed with a black arrow.
74
The strain gradient (section taken perpendicularly to the apex) after bending up to 180° and
then unloading is represented in Figure 13. The model prediction was in good agreement
with experiments in view of the experimental scatter. As for Steel 1, external and internal
radii measured on bent specimens were compared to model predictions, both after
unloading. Very good agreement was observed between the two values respectively of the
outer radius (1.14mm from experimental measurements vs. 1.17mm from model
predictions) and of the inner radius (0.32mm from experimental measurements vs. 0.33
mm from model predictions).
0.6
0.5
0.4
0.3
0.2 simulation
experiment 1
0.1 experiment 2
0
0 0.5 1 1.5
distance from apex (mm)
Figure 13: Evolution of the major strain at outer surface after bending up to 180° followed by unloading.
Experimental values (strain field measurements vs. model predictions). Section taken perpendicularly to the
apex at the end of the test. Curvilinear abscissa in mm along the outer surface.
Good correlation was also found between through-thickness hardness measurements and
model prediction of cumulative plastic strain (Figure 14), so that the location of the
“neutral fibre” was well predicted by the model. The slight difference in location of the
neutral fibre was probably due to the difficulty to perform hardness measurements through
such a thin material. Moreover, the minimum value of hardness measured on the bent
specimen was significantly higher (by about 40 HV0.3) than the hardness of the as-received
material. This correlated well with finite element results which indicated a minimum
cumulated plastic strain equal to 0.13 in this case (compared to 0,015 for Steel 1). The
evolution of hoop stress vs. hoop strain was plotted in Figure 14b for various nodes located
slightly below the mid-thickness (toward the inner surface) along the symmetry axis as for
bending of Steel 1. At the beginning of the test, all considered nodes were located in the
compressive zone but some of them were eventually loaded in tension after a slight
compressive plastic strain. Here, it was clearly necessary to take the kinematic contribution
to hardening into account to accurately represent the local stress/strain path next to the
mid-thickness. Although the prediction of the load/displacement curve was not in very
good agreement with experimental data at the end of the test, the numerical simulation was
considered to correctly predict the available database, making it possible to estimate strain
and stress fields.
75
1.2 350 1500

Hardness on bent 0.047mm
sample 300 0.094mm
cumulated plastic strain
1 1000
hoop stress (MPa)

0.14mm
250 500 0.19mm
0.8
Hv 0.3
200 0
0.6 Initial hardness profile
150 -500
0.4
100 -1000
0.2 strain
50 -1500
a b
0 0 -2000
0 0.15 0.3 0.45 0.6 0.75 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2
distance from outer surface (mm) hoop strain
Figure 14: (a) Evolution of cumulative plastic strain and hardness through the thickness on a specimen bent
up to 180° and initial hardness profile. The neutral fibre shifted from mid-thickness of the sheet toward the
inner surface, i.e., the initial compressive zone. The dashed black line and the dashed grey line respectively
represent the neutral fibre and the mid-thickness in the updated geometry. The continuous vertical grey line
represents the inner surface (thickness reduction).
(b) Loading paths for various nodes initially next to the mid-thickness along the symmetry axis. The nodes
were initially located at distances from mid-thickness of, respectively, 0.047, 0.094, 0.14, and 0.19mm
toward the inner surface.
4.5.3. Local stress and strain fields during air-bending of the thin specimens
Very high values of the hoop stress (1320MPa) and of the hoop strain (0.56) were
predicted at the end of the test by the model (Figure 15a). Note that unlike in air-bending
of Steel 1, thickness-reduced Steel 2 did not crack, except as soon as hard bands were
located close to the outer surface [15]. Internal damage was observed through the thickness
after air-bending of Steel 2 but no microcracking was observed at the outer surface. The
very low thickness of this material could possibly explain this result. This bending test
involved very high strain levels with a significant gradient through the thickness. In
particular, the node initially at mid-thickness underwent a total hoop strain close to 0.28 at
the end of the test (Figure 15b). Moreover, the increase in stress, in the compressive area,
with increasing distance from the outer surface was due to the shift of the contact region
between the sheet and the punch (Figure 12b).
1500 0.6 1200

1000 0.4
1000
Hoop stress (MPa)
500 0.2
Hoop stress (MPa)
Hoop strain
800
0 0
-500 strain -0.2 600
-1000 -0.4
400
-1500 -0.6
stress 200
-2000 -0.8
a b
-2500 -1 0
0 0.2 0.4 0.6 0.8 0 0.1 0.2 0.3
Distance from outer surface(mm) Hoop strain
Figure 15: (a) Hoop stress (black line) and hoop strain (grey line) along the symmetry axis for a 180° bending
(thickness-reduced Steel 2). The location of material initially at mid-thickness is represented by the vertical
dashed line (0.23mm from outer surface). The inner surface is represented by the vertical grey line (thickness
reduction), (b) hoop stress vs. hoop strain evolution for the node initially at mid-thickness.
76
5. Summary and concluding remarks
In order to carry out a realistic simulation of air-bending tests, a comprehensive

experimental database (including tensile tests on smooth and notched samples and shear
tests) was established for the considered low alloy TRIP steels. Using this database,
material constitutive equations accounting for anisotropic elasto-plastic behaviour with
mixed isotropic/kinematic hardening was determined. Using such material description
allowed good prediction of load vs. displacement curves obtained during air-bending tests.
Numerical simulation of the air-bending test showed the following results:

(i) The friction coefficient between the sheet and the punch has no influence (in
the investigated range: 0.1 vs. 0.25) in conventional bending conditions.
(ii) The test can be satisfactorily modelled by considering fixed rolls and perfect
sliding conditions between the sheet and the rolls.
(iii) Both two air-bending tests considered here involve plane strain conditions.
(iv) When using a very sharp punch (compared to the sheet thickness), there is no
need to explicitly model the contact between the sheet and the punch. It can be
replaced by prescribed node displacement of the sheet in the vicinity of the
symmetry axis.
(v) A slight shift of the neutral fibre (defined by the minimum cumulated plastic
strain) from the mid-thickness toward the compressive area was predicted, in
agreement with experimental observations. This implies that the whole sheet
thickness undergoes plasticity for sufficiently large bending angles.
(vi) Although kinematic hardening appears to have little effect on the
load/displacement curves and on the prediction of springback, the material
close to the sheet centre undergoes initial compression followed by reverse
plasticity in tension. Kinematic hardening should thus be taken into account to
predict the strain path next to the initial mid-thickness, especially for severe
bending.
(vii) Kinematic hardening should be taken into account in case of changes in loading
path e.g. by tensile prestraining prior to air-bending.
Acknowledgements
The authors are grateful for the financial support provided by ArcelorMittal Global R&D.
The authors also thank O. Bouaziz and A. Col for their valuable comments and S.
Douchamps for measurements performed with the Aramis facility.
References
[1] J. Datsko, C.T. Yang, Correlation of bendability of materials with their tensile
properties, J. Eng. Ind. 82 (1960) 309-314.
[2] D.K. Leu, A simplified approach for evaluating bendability and springback in plastic
bending of anisotropic sheet metals, J. Mater. Process. Technol. 66 (1997) 9-17.
[3] K. Yamazaki, Y. Mizuyama, M. Oka, Y. Tokunaga, Influence of microstructure on

bendability of ultrahigh-strength steel sheet, Journal of the Japan Society for Technology
of Plasticity 36 (1995) 973-978.
77
[4] A. Nilsson, L. Melin, C. Magnusson, Finite-element simulation of V-die bending: a

comparison with experimental results, J. Mater. Process. Technol. 65 (1997) 52-58.
[5] X. Li, Y. Yang, Y., Wang, J. Bao, S. Li, Effect of the material hardening mode on the
springback simulation accuracy of V-free bending, J. Mater. Process. Technol. 123 (2002)
209-211.
[6] W. Gan, R.H. Wagoner, Die design method for sheet springback, Int. J. Mech. Sci. 46
(2004) 1097-1113.
[7] P.A. Eggertsen, K. Mattiasson, On constitutive modelling for springback analysis, Int.
J. Mech. Sci. 52 (2010) 804-818.
[8] J. Chakrabarty, W.B. Lee, K.C. Chan, An analysis of the plane-strain bending of an
orthotropic sheet in the elastic/plastic range, J. Mater. Process. Technol. 104 (2000) 48-52.
[9] J. Chakrabarty, W.B. Lee, K.C. Chan, An exact solution for the elastic/plastic bending
of anisotropic sheet metal under conditions of plane strain, Int. J. Mech. Sci. 43 (2001)
1871-1880.
[10] A.R. Ragab, C.A. Saleh, Evaluation of bendability of sheet metals using void
coalescence models, Mater. Sci. Eng. A. 395 (2005) 102-109.
[11] K.M. Zhao, J.K. Lee, Finite element analysis of the three-point bending of sheet
metals, J. Mater. Process. Technol. 122 (2002) 6-11.
[12] C. Wang, G. Kinzel, T. Altan, Mathematical modeling of plane strain bending of sheet
and plate, J. Mater. Process. Technol. 39 (1993) 279-304.
[13] R. Hambli, A. Mkaddem, A. Potiron, Finite element damage modeling in bending
processes, J. Mater. Process. Technol. 147 (2004), 302-310.
[14] S. Thuillier, N. Le Maoût, P.Y. Manach, Bending limit prediction of an aluminium

thin sheet, Int. J. Mater. Form. 3 (2010) 223-226.
[15] D. Rèche, T. Sturel, O. Bouaziz, A. Col., A.F. Gourgues-Lorenzon, Damage

development in low allow TRIP-aided steels during air-bending, Mater. Sci. Eng. A 528
(2011) 5241-5250.
[16] A. Dalloz, J. Besson, A.F. Gourgues-Lorenzon, T. Sturel, A. Pineau, Effect of shear

cutting on ductility of a dual phase steel, Eng. Fract. Mech. 76 (2009) 1411-1424.
[17] Aramis, Optical 3D deformation analysis software developed by GOM GmbH,

http://www.gom.com/3d-software/aramis-software.html.
[18] Z. Marciniak, J.L. Duncan, Bending. The Mechanics of Sheet Metal Forming, first
edition, A.Arnold, 1992, pp. 68-99.
[19] J.L. Chaboche, A review of some plasticity and viscoplasticity constitutive theories,
Int. J. Plast. 24 (2008) 1642-1693.
78
[20] R. Hill, A theory of the yielding and plastic flow of anisotropic metals. Proc. Royal
Soc. London, A193, 1948, pp. 281-297.
[21] F. Barlat, D.J. Lege, J.C. Brem, A six component yield function for anisotropic
materials, Int. J. Plast. 7 (1991) 693-712.
[22] A.P. Karafillis, M.C. Boyce, A general anisotropic yield criterion using bounds and a
transformation weighting tensor, J. Mech. Phy. Solid. 41 (1993) 1859-1886.
[23] F. Bron, J. Besson, A yield function for anisotropy materials. Application to

aluminium alloys, Int. J. Plast. 20 (2004) 937-963.
[24] P.A. Eggertsen, K. Mattiasson, On the modelling of the bending-unbending behaviour

for accurate springback predictions, Int. J. Mech. Sci. 51 (2009) 547-563.
[25] B.W. Williams, C.H.M. Simha, N. Abedrabbo, R. Mayer, M.J. Worswick, Effect of
anisotropy, kinematic hardening and strain-rate sensitivity on the predicted axial crush
response of hydroformed aluminium alloy tubes, Int. J. Impact Eng. 37 (2010) 652-661.
[26] F. Barlat, J.C. Brem, J.W. Yoon, K. Chung, R.E. Dick, D.J. Lege, F. Pourboghrat,
S.H. Choi, E. Chu, Plane stress yield function for aluminum alloy sheets - Part 1: Theory,
Int. J. Plast. 19 (2003) 1297-1319.
[27] J. Yoon, F. Barlat, R.E. Dick, K. Chung, T.J. Kang, Plane stress yield function for
aluminum alloy sheets - Part II: FE formulation and its implementation, Int. J. Plast. 20
(2004) 495-522.
[28] I.N. Vladimirov, M.P. Pietryga, S. Reese, Anisotropic finite elastoplasticity with
nonlinear kinematic and isotropic hardening and application to sheet metal forming, Int. J.
Plast. 26 (2010) 659-687.
[29] Y. Shinohara, Y. Madi, J. Besson, A combined phenomenological model for the

representation of anisotropic hardening behaviour in high strength steel line pipes, Eur. J.
Mech. A Solids. 29 (2010) 917-927.
[30] J.L. Chaboche, Constitutive equations for cyclic plasticity and cyclic viscoplasticity.
Int. J. Plast. 5 (1989), 247-302.
[31] S. Thuillier, N. Le Maoût, P.Y. Manach, Influence of ductile damage on the bending
behaviour of aluminium alloy thin sheets, Materials and Design 32 (2011) 2049-2057.
[32] J. Besson, R. Foerch, Large scale object-oriented finite element code design, Comput.
Meth. Appl. Mech. Eng. 142 (1997) 165-187.
[33] B. Moran, M. Ortiz, C.F. Shih, Formulation of implicit finite-element methods for
multiplicative finite deformation plasticity, Int. J. Numer. Meth. Eng. 29 (1990) 483-514.
[34] M. Ben Bettaieb, X. Lemoine, L. Duchêne, A.M. Habraken, Simulation of the bending
process of hardening metallic sheets using damage model. Part I: Theoretical development
and numerical implementation, Mater. Sci. Eng. A. 528 (2010) 434-441.
79
[35] K. Gieck, Technical formulary, 10th Edition française, Dunod (In French), 1997.
[36] J. Muller, General mechanics formulary, 17th Edition, Paillart, 2000, p.147 (In
French).
[37] H. Stöcker, F. Jundt, G. Guillaume, Physics, 2nd Edition, Dunod, 2007, p.253 (In
French).
[38] L. Geng, R.H. Wagoner, Role of plastic anisotropy and its evolution on springback,
Int. J. Mech. Sci. 44 (2002) 123-148.
[39] P.A. Friedman, S.G. Luckey, Failure of Al-Mg-Si alloys in bending, Practical Failure
Analysis, ASM Inter. 1 (2002) 33-42.
80
II. Complementary data for model predictions

II.A. Experimental database
Constitutive equation parameters of Steel 1 have been determined by fitting model

predictions to experimental results, i.e. tensile tests on smooth and notched samples and
shear tests. In section I, only simulations along the transverse direction were given for
tensile tests. In the following, simulations along the three directions (rolling, transverse and
45°) are shown (Figure II.1).
For the following curves, a Bron Besson yield criterion coupled with a mixed
isotropic/kinematic hardening was used as in section I.
Note that the model is in good agreement with all experimental curves. The rolling
direction is slightly overestimated with the simulation compared to the experiment for
tensile tests on notched specimens while the simulation along 45° direction slightly
underestimates the experiment at the end of the test for tensile tests both on smooth and
notched samples. Finally, the prediction along transverse direction perfectly fit with
experiments. This is in this direction that air and stretch bending tests were carried out.
1000 1200
RD RD
1000
800
F/So (MPa)
800
600
600
400
experiment 400 experiment
model predictions model predictions
200 200
0 0
0 0.5 1 1.5
(a)
0 0.05 0.1 0.15 0.2 0.25 0.3 (b)
Elongation Notch opening displacement (mm)
1000 1200
TD TD
800 1000
800
F/So (MPa)
600
600
400
experiment 400 experiment
model predictions
200 model predictions
200
0 0
0 0.1 0.2 0.3 0 0.5 1 1.5
(c) Elongation (d) Notch opening displacement (mm)
81
1000 1200
45 45
800 1000
800
F/So (MPa)
600
600
400
experiment experiment
400
200
200
0 0
0 0.1 0.2 0.3 0 0.5 1 1.5
(e) Elongation
(f) Notch opening displacement (mm)
Figure II.1. Experiment vs. model predictions for Steel 1. Bron Besson yield criterion and mixed
isotropic/kinematic hardening. Tensile curves on smooth and notched specimens. Engineering tensile stress-
strain curves along (a) rolling, (c) transverse and (e) 45° directions. Tensile curves on notched specimens
along (b) rolling, (d) transverse and (f) 45° directions where F is the load and So is the initial cross section.
Shear tests were also carried out to determine constitutive equations of Steel 1. In section I,
only simulation curve for γmax = 0.2 was shown. Figure II.2 illustrates both experimental
curves and model predictions for γmax = 0.1, 0.2 and 0.3.
To summarize, there is very good agreement between all simulated and experimental
curves along the three directions for all tests (tensile and shear tests). This is due to the use
of both an anisotropic yield criterion (Bron-Besson equations) and mixed
isotropic/kinematic hardening rules, as will be explained in more detail in the following.
600 600
400 400
shear stress (MPa)
shear stress (MPa)
200 200
0 0
-200 -200
-400 -400
-600 -600
-800 -800
-0.4 -0.2 0 0.2 0.4 -0.4 -0.2 0 0.2 0.4
shear strain γ shear strain γ
(a) (b)
82
600
experiment
400
model predictions
shear stress (MPa)

200
-200
-400
-600
-800
-0.4 -0.2 0 0.2 0.4
(c)
shear strain γ
Figure II.2. Experimental vs. predicted curves for Steel 1. Cyclic shear curves (a) γmax = 0.1, (b) γmax = 0.2
and (c) γmax = 0.3.
II.B. Comparison with predictions by other models assuming either Von

Mises isotropic yield criterion or isotropic hardening rule
The model presented in section I included both a mixed isotropic and kinematic hardening
with an anisotropic yield criterion. In this section, a comparison with other models is
carried out. Simulations were first performed assuming a Von Mises yield criterion and a
mixed hardening rule. Then, model predictions with a Bron Besson yield criterion and a
purely isotropic hardening rule were quantified.
1. Constitutive equations accounting for a mixed hardening and a Von Mises yield
criterion
In this part, the model assumes an isotropic Von Mises yield criterion instead of an
anisotropic Bron Besson yield criterion and a mixed hardening rule. Figure II.3 and 4
illustrate both the predicted and the experimental curves along the three directions for
smooth and notched tensile specimens.
Although the difference is small, experiments are better simulated with the Bron Besson
yield criterion than with the Von Mises one and the curves simulated with the Von Mises
criterion systematically overestimate stresses compared to predictions from the Bron
Besson yield criterion. In fact, 45° direction simulations for tensile test on smooth
specimens are not well predicted with the Von Mises yield criterion. The two other
directions are closer to the experimental curves. Regarding tensile tests on notched
specimens, the rolling direction is the worse predicted with Von Mises yield criterion but
this is in agreement with predicted curves given by the Bron Besson yield criterion.
83
a. Tensile tests on smooth specimens along the three directions
900 900
800
RD 800
TD

700 700
600 600
500 500
400 400
300 300
Bron Besson + mixed hardening Bron Besson + mixed hardening
200 200
Von Mises + mixed hardening
100 Von Mises + mixed hardening 100
0 0
(a) 0 0.1 0.2 0.3 (b) 0 0.1 0.2 0.3
Elongation Elongation
900
45
800
700
600
500
400
experiment
300
Bron Besson + mixed hardening
200
100
0
(c) 0 0.1 0.2 0.3
Elongation
Figure II.3. Experiment vs. model predictions for Steel 1 both with a Von Mises (red curves) and a Bron
Besson (grey curves) yield criterion. Mixed isotropic/kinematic hardening rule. Engineering tensile stress-
strain curves along (a) rolling, (b) transverse and (c) 45° directions.
b. Tensile tests on notched specimens along the three directions

1200 1200
RD TD
1000 1000
800 800
F/So (MPa)
F/So (MPa)
600 600
experiment
experiment 400
400 Bron Besson + mixed hardening
Von Mises + mixed hardening 200 Von Mises + mixed hardening
200
0 0
0 0.5 1 1.5 0 0.5 1 1.5
(a) (b)
Notch opening (mm) Notch opening (mm)
84
1200
45
1000
800
F/So (MPa)
600
400 experiment
200 Von Mises + mixed hardening
0
0 0.5 1 1.5
(c) Notch opening (mm)
Besson (grey curves) yield criterion. Mixed isotropic/kinematic hardening rule. Tensile curves on notched
specimens along (a) rolling, (b) transverse and (c) 45° directions where F is the load and So is the initial cross
section.
c. Cyclic shear tests
In Figure II.5, it is clearly noticed that experiments are better modelled with a Bron Besson
yield criterion than with the Von Mises one. Note that during shearing, the Von Mises
yield criterion overestimates the stress for the three values of maximum shear strain. For
reverse shearing, this phenomenon is also noticed, despite the use of the same mixed
isotropic/kinematic hardening rules. In fact, only the yield stress in compression is
underestimated, experimental and predicted curves being parallel in the plastic part of the
reverse shear curve.
600 600
Bron-Besson + Bron-Besson +
400 mixed hardening 400 mixed hardening
Von Mises + Von Mises +
Shear stress (MPa)
shear stress (MPa)
200 200
mixed hardening mixed hardening
0 0
Experiment Experiment
-200 -200
-400 -400
-600 -600
(a)-800 (b)-800
-0.4 -0.2 0 0.2 0.4 -0.4 -0.2 0 0.2 0.4
85
600
Bron-Besson +
400 mixed hardening
Von Mises +
Shear stress (MPa)

200
mixed hardening
0
Experiment
-200
-400
-600
-800
-0.4 -0.2 0 0.2 0.4
(c) shear strain γ
Besson (grey curves) yield criterion. Mixed isotropic/kinematic hardening rule. Cyclic shearing curves (a)
γmax = 0.1 and (b) γmax = 0.2 and (c) γmax = 0.3.
d. Summary: effect of the yield criterion (with mixed hardening rule)
With simulations of tensile tests on both smooth and notched specimens, it is noticed that
similar curves are obtained with the Bron Besson or with the Von Mises yield criterion.
For the cyclic shear tests, the difference is more significant between the two model
predictions. Therefore, it seems clear that the Bron Besson yield criterion is the more
appropriated and the more accurate criterion for shear tests. For tensile tests, the choice of
the yield criterion is not discriminatory. In section III, model predictions with a Von Mises
criterion and a mixed hardening will be given for the air-bending test. However, based on
these results, it is expected to not have a real impact on the air bending test simulation.
2. Constitutive equations accounting for a purely isotropic hardening and a Bron-

Besson yield criterion
In order to determine the influence accounting for kinematic hardening on the accuracy of
model prediction, purely isotropic hardening combined with the Bron Besson yield
criterion was used. Note that Bron Besson yield criterion parameters defined in section I
are used in this model The isotropic hardening parameters were optimized on the tensile
curve on smooth specimen along the transverse direction (Figure II.6b). The new
hardening rule equation is described by the following equation:
R ( p ) = R0 '+ H ' p + Q' (1 − exp(− b' p )) (1)
where p is the cumulative plastic strain (calculated with the Bron Besson yield criterion)
and with R0’ = 499 MPa, H’ = 800MPa, Q’ = 393MPa and b’ = 23.
With these new constitutive equations, the entire database was modelled. Figure II.6 and 7
illustrate both the model predictions and the experimental curves along the three directions
for smooth and notched tensile specimens. It is noticed that both model predictions with
either the mixed or purely isotropic hardening rules are similar for all curves except for
86
smooth and notched specimens along the 45° direction for which the simulation is more
accurate with the Bron Besson model.
a. Tensile tests on smooth specimens along the three directions

900 900
800
RD TD
800
700 700
600 600
Stress (MPa)
Stress (MPa)
500 500
400 400
experiment
300 300 experiment
200
Bron Besson + isotropic hardening 200
Bron Besson + isotropic hardening
100 100
0 0
0 0.1 0.2 0.3
(a) (b) 0 0.1 0.2 0.3
Elongation Elongation
900
800
45
700
600
Stress (MPa)
500
400 experiment
200 Bron Besson + isotropic hardening
100
0
(c) 0 0.1 0.2 0.3
Elongation
Figure II.6. Experiment vs. model predictions for Steel 1 both with a purely isotropic hardening (blue curves)
and a mixed hardening (grey curves). Bron Besson yield criterion (same parameters for both models).
Engineering tensile stress-strain curves along (a) rolling, (b) transverse and (c) 45° directions.
b. Tensile tests on notched specimens along the three directions

1200 1200
RD TD
1000 1000
800 800
F/So (MPa)
F/So (MPa)
600 600
experiment
400 experiment
Bron Besson + mixed hardening 400
Bron Besson + isotropic hardening Bron Besson + mixed hardening
200 200 Bron Besson + isotropic hardening
0 0
(a) 0 0.5 1 1.5 (b) 0 0.5 1 1.5
Notch opening displacement (mm) Notch opening displacement (mm)
87
1200
45
1000
800
F/So (MPa)
600
experiment
Bron Besson + isotropic hardening
200
0
0 0.5 1 1.5
(c)
Notch opening displacement (mm)
Figure II.7. Experiment vs. model predictions for Steel 1 both with a purely isotropic hardening (blue curves)
and a mixed hardening (grey curves). Bron Besson yield criterion (same parameters for both models). Tensile
curves on notched specimens along (a) rolling, (b) transverse and (c) 45° directions where F is the load and
So is the initial cross section.
c. Cyclic shear tests
Shear tests were also modelled with a purely isotropic hardening law for γmax = 0.1, 0.2 and
0.3. In Figure II.8, it is clearly noticed that experiments are better modelled with a mixed
isotropic + kinematic hardening than with purely isotropic hardening. Indeed, for the
shearing, simulated blue curves (purely isotropic hardening) underestimate the experiments
(black curves). As expected, without kinematic hardening the reverse shearing is poorly
represented because the Bauschinger effect is not taken into account.
600 600
Bron-Besson + Bron-Besson +
400 mixed hardening 400 mixed hardening
Bron Besson +
shear stress (MPa)
Bron Besson +
shear stress (MPa)
200 200
isotropic hardening isotropic hardening
0 0
Experiment Experiment
-200 -200
-400 -400
-600 -600
-800 -800
-0.4 -0.2 0 0.2 0.4 -0.4 -0.2 0 0.2 0.4
(a) (b)
600
Bron-Besson +
400
mixed hardening
shear stress (MPa)
200 Bron Besson +

isotropic hardening
0 Figure II.8. Experiment vs. model predictions
-200 Experiment for Steel 1 both with purely isotropic hardening
(blue curves) and a mixed hardening (grey
-400 curves). Bron Besson yield criterion (same
parameters for both models). Cyclic shear
-600 curves (a) γmax = 0.1 and (b) γmax = 0.2 and (c)
γmax = 0.3.
-800
(c) -0.4 -0.2 0 0.2 0.4
shear strain γ
88
d. Summary: effect of the hardening rule by using the Bron Besson yield criterion
With simulations of tensile tests both on smooth and notched specimens, it is noticed that
similar curves are obtained both with a mixed hardening and with a purely isotropic
hardening. Regarding the cyclic shear tests, the difference is well visible between
predictions from the two models. As the material shows a Bauschinger effect, these tests
cannot be satisfactorily modelled without any kinematic contribution in the hardening rule.
Therefore, it seems clear that for these tests on this material, mixed hardening is more
appropriate and yields more accurate predictions than purely isotropic hardening rule. The
effect of a purely isotropic hardening on the air-bending test has been investigated on
section I and will be shown in section III with the model assuming a Von Mises yield
criterion.
3. Partial conclusion
In this section, three models were compared (Table 7):
Isotropic hardening law

YES NO
Von Mises + mixed hardening:
→ Bauschinger effect is taken into

account.
YES / → Stresses are overestimated.
→ Calculations are rather fast.
→ In air bending test, the load is
overestimated (see next section).
Yield Bron Besson + purely Bron Besson + mixed hardening

isotropic isotropic hardening
criterion
→ Tensile tests on notched or → Tensile tests on notched and
smooth specimens are in good smooth specimens are in very
agreement with experiments. good agreement with experiments.
NO
→ Bauschinger effect is not → Bauschinger effect is taken into
taken into account. account.
→ In air-bending test, the load → Air-bending test prediction is in
is overestimated. very good agreement with
experimental curve.
Table 7. Comparison of the three models.
Based on these results, it seems clear that a yield anisotropic criterion with mixed
isotropic/kinematic hardening rules have to be taken into account to well predict the whole
experimental database. In addition, to predict air-bending tests with accuracy, this model is
also required.
89
III. Influence of simulation conditions of the air-bending test

In this section, it is intended to study the influence of the mesh as well as the air-bending
test boundary conditions. In addition, the choice of the model for air-bending test
predictions is discussed at the end of this part.
III.A. Influence of meshing conditions
In order to determine the influence of the mesh, four calculations were performed with
various numbers of elements across the thickness close to the punch. The refined mesh in
the vicinity of the punch was chosen to be equal to 2mm. The size of the smallest element
along the transverse direction is constant and equal to 25µm for the four meshes. The four
cases studied were:
(i) 8 elements across the sheet thickness (i.e. 200µm elements size)
(ii) 16 elements across the sheet thickness (i.e. 100µm elements size)
(iii) 32 elements across the sheet thickness (i.e. 50µm elements size)
(iv) 64 elements across the sheet thickness (i.e. 25µm elements size)
In agreement with section I, this parametric study is carried out in 2D (plane strain
assumption), with the Bron Besson yield criterion and mixed isotropic/kinematic hardening
rules. In addition, calculations were carried out with the punch and a friction coefficient of
0.25. The corresponding meshes are illustrated in Figure II.9.
8 elements 16 elements
Thickness
TD
Figure II.9. Four meshes with various numbers of elements across the thickness. Total thickness = 1.6mm.
90
Figure II.10a illustrates load / displacement curves for four meshing (8 to 64 elements in
the thickness). Similar curves are obtained for the four simulations, meaning that the
number of elements across the thickness do not change the global response of the load /
displacement curve. Hoop stress and hoop strain distributions across the thickness at the
end of the test are illustrated in Figure II.10b. Strain and stress fields maps calculated using
each of the four meshes are respectively given in Figure II.11 and Figure II.12.
3.5 1500 0.3
3 1000 0.2
Hoop stress (MPa)

2.5 500 0.1
Hoop strain
Load (kN)
2 0 0
stress: 8 elem.
1.5 experiment stress: 16 elem.
-500 -0.1
stress: 32 elem.
8 elements
stress: 64 elem.
1 16 elements -1000 -0.2
strain: 8 elem.
32 elements strain: 16 elem.
0.5 64 elements -1500 -0.3
strain: 32 elem.
strain: 64 elem.
0 -2000 -0.4
0 10 20 30 0 0.4 0.8 1.2 1.6
(a) Displacement (mm) (b) Distance from outer surface (mm)
Figure II.10. (a) Influence of the mesh on the load / displacement curves. Calculations with the Bron Besson
yield criterion, mixed isotropic/kinematic hardening rules. (b) Influence of the mesh on the hoop stress and
hoop strain distributions across the thickness at the end of the test (i.e. for a punch displacement of 26mm).
x2
x1
x3
ε11
Figure II.11. Strain field after bending of Steel 1. ε11 map for the four simulations (8 to 64 elements across the
thickness) taken at the end of the test (i.e. for a punch displacement of 26mm).
91
16 elements
8 elements
x2
x1
x3
σ11
Figure II.12. Stress field after bending of Steel 1. σ11 map for the four simulations (8 to 64 elements across
the thickness) taken at the end of the test (i.e. for a punch displacement of 26mm).
Regarding stress and strain fields as a function of the local meshing size, it seems clear that
with 8 elements in the thickness, the stress and strain gradients are not well represented.
With 16 elements across the thickness, the accuracy is rather good. In addition, it is
reasonable to say that with 32 or 64 elements across the thickness, the stress and strain
distributions are very close. Thus, calculations could be performed with 16 or 32 elements
across the thickness to keep a good compromise between a good accuracy and a limited
elements number. Nevertheless, in Chapter IV, it is intended to include damage in the
model to simulate air-bending test. Therefore, it was decided to perform all simulations
with 32 elements across the thickness to get a better accuracy.
III.B. Influence of the boundary conditions
1. Influence of the friction coefficient
As mentioned in section I, air-bending tests were carried out without lubrication. Thus, a
high friction coefficient was first assumed (µ = 0.25). As the actual friction coefficient
between the sheet and the punch was not known, a lower value (µ = 0.1) was also tested.
Note that same conditions were kept (i.e. 32 elements across the thickness and Bron
Besson yield criterion with mixed hardening rules). The corresponding curves are
92
illustrated in Figure II.13a and b. The load / displacement simulated curves (Figure II.13a)
are similar, meaning that the friction coefficient between the sheet and the punch has no
influence in this configuration. Same results are obtained regarding the evolution of the
hoop stress and hoop strain (Figure II.13b) where no difference between µ=0.25 and µ=0.1
were noticed. Note that with a higher punch radius, the friction coefficient may have an
influence.
3.5 1500 0.3
3 1000 0.2
Hoop stress (MPa)

2.5 500 0.1
Hoop strain
Load (kN)
2 0 0
1.5 -500 -0.1

experiment stress for µ=0.25
1 -1000 -0.2
simulation: µ =0.25 stress for µ=0.1
0.5 simulation: µ = 0.1 -1500 strain for µ=0.25 -0.3
strain for µ=0.1
0 -2000 -0.4
0 10 20 30 0 0.4 0.8 1.2 1.6
(a) Punch displacement (mm) (b) Distance from outer surface (mm)
Figure II.13. Influence of the friction coefficient on the air-bending test with µ = 0.25 and µ = 0.1 (a)
experimental and predicted load / displacement curves, (b) hoop stress and hoop strain distributions for both
friction coefficients at the end of the test (i.e. a punch displacement of 26mm).
This result was expecting regarding more precisely the contact between the sheet and the
punch (Figure II.14). Indeed, due to its tight shape, the punch is in contact with only four
nodes, that is why the friction coefficient has no influence in this case.
Four nodes in
contact between the
punch and the sheet
Figure II.14. Zoom on the contact between the sheet and the punch. Only four nodes of the sheet are in
contact with the tool.
2. Influence of the punch
Based on previous observations, simulation was also conducted by replacing the punch
with prescribed displacement of the nodes located just under the punch (Figure II.14). Here
also, calculations were carried out in 2D with 32 elements across the thickness and with a
93
Bron Besson yield criterion and mixed hardening rules. The corresponding curves are
shown in Figure II.15a and b. Regarding the load / displacement curves (Figure II.15a), a
similar behaviour is noticed with simulations with or without the punch. Figure II.15b
illustrates same results because hoop stress and hoop strain have similar distributions
across the sheet thickness with or without the punch. These conclusions are only valid
because the punch radius is very small (R = 0.1mm) i.e. because the contact area between
the punch and the sheet is very small: the inner radius of the sheet after the end of the
bending (2.05mm) is much higher than the punch radius itself. This is due to both the
punch geometry, the sheet thickness and the constitutive behaviour of the sheet material. In
case of a larger punch or of much thinner sheet (such as thickness-reduced Steel 2), the
punch could not be replaced by prescribed nodes displacement.
3.5 1500 0.3
3 1000 0.2
Hoop stress (MPa)

2.5 500 0.1
Hoop strain
Load (kN)
2 0 0
1.5 -500 -0.1

experiment
1 -1000 stress: with punch -0.2
simulation: with punch
stress: without punch
0.5 simulation: without punch -1500 strain: with punch -0.3
strain: without punch
0 -2000 -0.4
0 10 20 30 0 0.4 0.8 1.2 1.6
(a) (b)
Punch displacement (mm) Distance from outer surface (mm)
Figure II.15. Influence of the punch on the air-bending test. (a) experimental and predicted load /
displacement curves with and without the punch, (b) hoop stress and hoop strain distributions across the
thickness both with and without the punch at the end of the test (i.e. a punch displacement of 26mm).
III.C. Influence of the constitutive equations on the model predictions for the
air-bending test
In section I, air-bending tests on Steel 1 were carried out in 2D with 32 elements across the
thickness and with a Bron Besson yield criterion associated with mixed isotropic/kinematic
hardening rules. In this part, it is intended to compare these results with the ones found
with (i) an isotropic yield Von Mises criterion and (ii) a purely isotropic hardening rule.
All these calculations were also performed assuming plane strain hypothesis (2D) with 32
elements across the thickness.
The air-bending test was first modelled assuming a Von Mises yield criterion instead of a
Bron Besson anisotropic criterion (Figure II.16). Then, the mixed hardening law was
replaced by a purely isotropic one (Figure II.17). As observed in part II.B.1, the Von Mises
yield criterion overestimates the load and is less accurate than the modelling with the Bron
Besson one (Figure II.16a). Regarding the hoop stress and strain distributions (Figure
II.16b), the yield criterion has no influence except for the points located close to the inner
surface. Indeed, hoop stress and hoop strain have lower values with the Von Mises yield
criterion in the compressive part than with the Bron Besson yield criterion.
94
3.5 1500 0.3
3 1000 0.2
2.5
Hoop stress (MPa)

500 0.1
Hoop strain
Load (kN)
2 0 0
1.5 -500 -0.1
1 Experiment -1000 -0.2

stress: Bron Besson
stress: Von Mises
0.5 Bron Besson + mixed hardening -1500 -0.3
strain: Bron Besson
strain: Von Mises
0 -2000 -0.4
0 10 20 30 0 0.4 0.8 1.2 1.6
Figure II.16. (a) Load / displacement curves for the air-bending test. Black, grey and red curves respectively
correspond to experiment, model predictions with a mixed hardening law and a Bron Besson yield criterion
(as in section I) and model predictions with a mixed hardening law and a Von Mises yield criterion. (b) Hoop
stress and hoop strain distributions across the thickness both with a Bron Besson and a Von Mises yield
criterions at the end of test (i.e. a punch displacement of 26mm).
Figure II.17a showed that similar load / displacement curves are obtained until a
displacement of about 10mm with both a mixed and a purely isotropic hardening rule.
After this displacement, the load is also overestimated with the isotropic hardening rule
compared to the mixed hardening one. Regarding the hoop stress and strain distributions
(Figure II.17b), the hardening rule has no influence except for the points located close to
the inner surface. Indeed, hoop stress and hoop strain have lower values with the purely
isotropic hardening rule in the compressive area than with the mixed hardening rule.
3.5 1500 0.3
3 1000 0.2
Hoop stress (MPa)
2.5 500 0.1

Hoop strain
Load (kN)
2 0 0
1.5 -500 -0.1
1 experiment -1000 -0.2

stress: mixed hard.
Bron Besson + mixed hardening stress: isotropic hard.
0.5 Bron Besson + isotropic hardening -1500 strain: mixed hard. -0.3
strain: isotropic hard.
0 -2000 -0.4
0 10 20 30 0 0.4 0.8 1.2 1.6
Figure II.17. (a) Load / displacement curves for the air-bending test. Black, grey and blue curves respectively
correspond to experiment, model predictions with a mixed hardening law and a Bron Besson yield criterion
(as in section I) and model predictions with an isotropic hardening law and a Bron Besson yield criterion. (b)
Hoop stress and hoop strain distributions across the thickness both with a mixed hardening rule and a purely
isotropic hardening rule at the end of the test (i.e. a punch displacement of 26mm).
95
III.D. Summary: influence of modelling conditions on air-bending test

predictions
In this part, the influence of the air-bending modelling conditions was discussed. It was
shown that the mesh has no influence on the global force / displacement curves but has an
impact on the stress and strain fields. Therefore to get the best compromise between the
accuracy of predictions and the number of elements, it was chosen to perform all air-
bending simulations with 32 elements across the thickness. Then, the influence of the
friction coefficient was considered. Regarding previous results, it was demonstrated that in
this case, the friction coefficient between the punch and the sheet has no influence. This is
in agreement with the fact that similar results were obtained with or without the punch.
Finally, the choice of the constitutive equations was discussed. A Von Mises yield criterion
combined with a mixed hardening rule as well as a Bron Besson yield criterion associated
with a purely isotropic hardening rule were used to simulate the air-bending test. In terms
of load / displacement curves, both results overestimate the experimental curve (especially
at the end of the test). More accurate results were obtained with the association of a mixed
kinematic + isotropic hardening and a Bron Besson yield criterion. Regarding the stress
and strain distributions, similar results were obtained except for the nodes close to the
inner surface which had lower values. A summary of all these results is given in Table 8.
Load /displacement curve Hoop stress and hoop strain

Slight influence
→ 8 elements: not enough accuracy
Meshing influence → 16 elements: correct accuracy
→ 8 elements → 64 elements: too many elements
→ 16 elements → 32 elements is the best
No influence
→ 32 elements compromise between the accuracy
→ 64 elements and the number of elements and this
across the thickness number will be used in the
following chapter when damage
will be taken into account.
Friction coefficient
→ µ = 0.25 No influence No influence
→ µ = 0.1
Effect of the punch
→ with the punch
No influence No influence
→ with prescribed
nodes displacement
2D / 3D
No influence No influence
calculations
Yield criterion Slight influence for nodes close to
Better predictions with the
→ Von Mises the inner surface both for hoop
Bron Besson yield criterion
→ Bron Besson stress and hoop strain
Hardening rule Better predictions with the Slight influence for nodes close to
→ Purely isotropic mixed isotropic / kinematic the inner surface both for hoop
→ Mixed hardening stress and hoop strain
Table 8. Summary of the influence of air-bending modelling conditions on the load / displacement curve and
the hoop stress and hoop strain distribution across the thickness.
96

In this chapter, a comprehensive experimental database including tensile tests on smooth
and notched specimens and shear tests was established. All these experiments enable to
determine the constitutive equations of Steel 1. The “reference” model chosen here account
for anisotropic elasto-plastic behaviour (Bron Besson yield criterion) with a mixed
kinematic + isotropic hardening rule. The possibility of using simplifying assumptions in
the model (such as an isotropic Von Mises yield criterion or a purely isotropic hardening
rule) was also discussed.
It was demonstrated that with a Von Mises yield criterion and a mixed hardening rule, the
load is always overestimated and the reverse shear tests were not well simulated. With a
Bron Besson yield criterion and a purely isotropic hardening rule, similar results were
obtained than with a mixed hardening law except for the cyclic shear tests. As expected,
without kinematic contribution in the hardening rule, the Bauschinger effect is not taken
into account and thus shear tests are poorly modelled, whereas satisfactory agreement was
obtained by using the “reference” model.
Using these “reference” constitutive equations allowed modelling the air-bending test. The
simulation of this test showed that for the punch geometry, sheet thickness and sheet
material considered here (Steel 1), the friction coefficient between the sheet and the punch
has no influence. In addition, when using a very sharp punch (R=0.1mm in this study), the
punch can be replaced by prescribing the displacement of the nodes located just below the
punch. The meshing has no influence on the load / displacement curve but has one on the
stress and strain distributions through the thickness. The final meshing chosen for the rest
of the work has 32 elements in the thickness. In addition, it is interesting to underline that
this modelling can be carried out in plane strain conditions. For thickness-reduced Steel 2,
same conditions are used, i.e. 32 elements across the thickness, plane strain calculations
and Bron Besson yield criterion associated with a mixed isotropic/kinematic hardening
rule. However, in this case, the punch was not replaced by prescribed nodes displacement.
Regarding the role of the yield criterion, simulations of the air-bending tests with a Von
Mises or with a Bron Besson criterion were carried out and simulation assuming a Bron
Besson yield criterion fitted better with the experiment than the modelling with a Von
Mises criterion. By using a Bron Besson yield criterion, a slight difference in
load / displacement curves with mixed and purely isotropic hardening rules was noticed.
Better predictions were obtained with a mixed hardening rule. The kinematic contribution
was also discussed by studying its influence on a prestrained and bent specimen. In this
case, it seems clear that when prestraining is performed before a air-bending test, a
kinematic contribution is needed to satisfactorily predict the load/displacement curve
during the bending.
Numerical simulation of the air-bending tests using the “reference” model revealed that the
thickness of the sheet decreases inhomogeneously and this leads to a shift of the neutral
fibre from the mid-thickness towards the compressive area. In addition, material points
close to the mid-thickness undergo initial compression followed by tension suggesting that
kinematic hardening is necessary to predict the strain path in the vicinity of the mid-
thickness especially for very severe bending such as that of thickness-reduced Steel 2 until
180°.
97
Dans ce chapitre, une base de données expérimentale détaillée a été établie incluant des
tests de traction sur éprouvettes lisses et entaillées ainsi que des tests de cisaillement. Ces
essais ont permis de déterminer la loi de comportement de l’acier 1. Le modèle de
“référence” établi précédemment inclut un comportement élasto-plastique anisotrope
(critère d’écoulement de Bron Besson) et un écrouissage mixte isotrope + cinématique. La
possibilité d’utiliser des modèles plus simples tels qu’un modèle d’écoulement isotrope de
Von Mises ou un écrouissage purement isotrope a également été envisagée.
Avec un critère de Von Mises et un écrouissage mixte, les efforts sont souvent surestimés.
Avec le critère anisotrope de Bron Besson et un écrouissage purement isotrope, nous
avons obtenu des résultats similaires à ceux obtenus avec un écrouissage mixte excepté
pour les essais de cisaillement qui étaient mal prédits. Sans écrouissage cinématique,
l’effet Bauschinger n’est pas pris en compte, et cela entraîne une mauvaise prédiction des
essais de cisaillement alors qu’ils étaient correctement simulés avec le modèle de
“référence”.
Cette loi de comportement a permis de modéliser l’essai de pliage en V. La simulation

numérique de ce test a montré que pour cette géométrie de poinçon, cette épaisseur de tôle
et l’acier de l’étude (acier 1), le coefficient de frottement entre la tôle et le poinçon n’a pas
d’influence. Lorsqu’un poinçon avec un rayon très fin est utilisé (R = 0.1mm dans cette
étude), ce dernier peut être remplacé par le déplacement imposé des nœuds situés sous le
poinçon. Le maillage n’a aucune influence sur les courbes force/déplacement mais en a
une sur la distribution des contraintes et des déformations à travers l’épaisseur. Le
maillage choisi pour toute l’étude a 32 éléments à travers l’épaisseur et les simulations
numériques du pliage en V peuvent être effectuées en déformation plane. Pour l’acier 2
rectifié en épaisseur, les mêmes conditions que précédemment ont été appliquées, à savoir,
32 éléments dans l’épaisseur, l’hypothèse d’un calcul en déformation plane, un critère
d’écoulement de Bron Besson et un écrouissage mixte. Néanmoins, dans ce cas là, le
poinçon n’a pas été remplacé par le déplacement imposé des nœuds sous le poinçon.
La modélisation du pliage en V a été réalisée à la fois avec un critère d’écoulement de Von

Mises, ainsi qu’avec Bron Besson. La simulation numérique utilisant le critère anisotrope
de Bron Besson a donné des résultats plus proches de l’expérience que celle avec le critère
de Von Mises. Utilisant un critère d’écoulement anisotrope de Bron Besson, nous avons
montré que les courbes force/déplacement étaient mieux prédites avec un écrouissage
mixte plutôt qu’avec un écrouissage purement isotrope. La contribution de l’écrouissage
cinématique a été discutée en étudiant son influence sur un échantillon plié préalablement
prédéformé en traction. Nous avons démontré que lorsqu’une prédéformation est
appliquée avant un test de pliage, il est nécessaire de prendre en compte l’écrouissage
cinématique pour prédire de manière satisfaisante le pliage.
La simulation numérique du pliage en V avec le modèle de “référence” a révélé que

l’épaisseur de l’acier diminuait de manière inhomogène et cela entraînait un déplacement
de la fibre neutre depuis la mi-épaisseur vers la zone en compression. De plus, nous avons
montré que les points situés dans le voisinage de la mi-épaisseur subissaient d’abord une
compression puis une tension, suggérant la nécessité d’inclure l’écrouissage cinématique
dans le modèle pour prédire de manière précise le trajet de chargement des points proches
de la mi-épaisseur.
98
Chapter 3. Investigation of damage mechanisms
Chapter 3
Investigation of damage mechanisms
In previous parts, it was shown that the microstructure

has a key role in the bendability of steels. Hence, it is
now necessary to precisely investigate damage
mechanisms involved both in air-bending and in stretch
bending tests.
In a first part, damage was characterized using scanning
electron microscope and light microscope on air-bent,
stretch bent and tensile specimens. Fractographic
observations showed that fracture modes are ductile both
in air and stretch bending tests. In addition, the influence
of both central segregation and banded structures were
investigated for the two bending tests.
Finally, in a second part, the role of the segregated
bands depending on their locations through the thickness
was investigated.
101
Table of contents
I. EXPERIMENTAL CHARACTERIZATION OF DAMAGE MECHANISMS .... 103
I.A. Fractographic observations after air and stretch bending tests……………...……….103
I.B. Damage development during air and stretch bending tests……………………….....106
I.C. Investigation of damage development in segregated bands using tensile specimens..111
I.D. Summary of section I………………………………..……………………………….116
II. INFLUENCE OF SEGREGATED BANDS ON BENDABILITY......................... 117
II.A. Investigation of damage mechanisms in air-bending……………………………….117
II.B. Complementary observations……………………………………………………….127
III. SUMMARY OF CHAPTER 3 ................................................................................. 130
102
In this chapter, damage mechanisms are investigated for the four TRIP-aided steels under
study. In stretch bending, first experiments were carried out with an hydraulic press. Then,
in order to be able to interrupt the test just before fracture, a new device was designed
enabling to perform tests with a lower displacement rate.
In order to investigate damage mechanisms in air-bending in more detail, Steel 1 and
thickness-reduced Steel 2 were chosen (section 2). The first objective of these
complementary investigations is to evidence the influence of the location of hard bands on
bendability. A quantitative relationship between damage, band thickness and local strain is
proposed at the end of this section.
Finally, in the second part of section 2, complementary observations of thickness-reduced
Steel 2 are given.
I. Experimental characterization of damage mechanisms

I.A. Fractographic observations after air and stretch bending tests
In order to determine fracture modes of air-bending and stretch bending specimens,

fractographic observations were carried out using scanning electron microscopy.
As air-bending specimens are not fully broken, it was very difficult to observe their
fracture surfaces (Figure III.1). Nevertheless, dimples can be observed in some areas of the
main microcrack (Figure III.2).
Figure III.1. Crack observed by SEM on Steel 1 after air-bending.
Figure III.2. Ductile failure of Steel 1 after air-bending.
In order to confirm the fracture mode, additional tests were carried out to open cracks
initiated during air-bending. After performing air-bending test on Steel 1 specimens until
103
cracking (classifications will be given in section 2), the partially cracked specimens were
then heat treated in a furnace at 300°C in air during 1 hour in order to oxidize the fracture
surface induced by air-bending. The specimens were then dipped into liquid nitrogen
during a few minutes and then quickly fractured in order to break them in a brittle manner
and therefore to easily distinguish the initial microcrack from final (cleavage) fracture
(Figure III.3). Figure III.4a shows the initial air-bending crack (in light grey) as well as the
final fracture surface (in dark grey). The difference between the ductile fracture surface
(due to air-bending) and the brittle fracture surface (cleavage after cooling) is clearly
observed (Figure III.4b).
Heat
Broken
treatment:
oxidization steel
of the crack
Figure III.3. Steps to mark the crack by oxidization.
(a) Thickness (b)
. TD
RD
(c) (d)
Figure III.4. SEM observations of a crack along RD (Steel 1), (a) general view of the crack induced by air-
bending, (b) transition between the two cracks, (c) brittle cleavage fracture at low temperature, (d) ductile
fracture with dimples occurred during air-bending.
104
In the particular case of Figure III.4, the crack due to bending has a width of about 1.6mm
and a maximum depth of 200µm. Note that the crack has a smooth propagation front
through the sheet thickness.
In order to check whether the fracture is also ductile during stretch bending tests,
fractographic observations were carried out on Steel 1, Steel 3 and Steel 4. As fracture is
unstable in the stretch bending tests, no particular sample preparation was carried out, the
specimens being completely broken into two parts after the stretch bending tests. For the
three steels, dimples were also observed, confirming the ductile nature of the fracture
mechanism with this forming mode (Figure III.5, Figure III.6).
It is important to underline that the mid-thickness is readily observable in the fracture

surfaces. In Figure III.5, two planes, which together form a V-shape, are distinguished. In
Figure III.6, lines oriented along the rolling direction are noticed at mid-thickness, this
could correspond to segregated bands or elongated inclusions. Thus, under stretch bending,
the role of segregated bands on damage development was further investigated.
Rolling direction
Thickness
(a) (b)
Figure III.5. Fractographic observations by SEM on Steel 3. (a) general view of the thickness, (b) close up
view of the mid-thickness.
Rolling direction
Thickness
(a) (b)
Figure III.6. Fractographic observations by SEM in the vicinity of the mid-thickness. (a) Steel 1, (b) Steel 4.
105
To summarize, the fracture modes are ductile both in air and stretch bending tests.
Particular features such as marked lines oriented along the rolling direction (for stretch
bending along TD) are probably linked to segregated bands.
I.B. Damage development during air and stretch bending tests
1. Air-bending test
To investigate damage mechanisms during air-bending, light and scanning electron

microscopy were used. After bending of various specimens up to “microcracking” (which
will be defined in section II), the cracks were first observed thanks to a binocular magnifier
(Figure III.7). Then, through-thickness metallographic observations perpendicular to the
crack were performed to investigate crack initiation.
The LePera colour etching reagent was used making it possible to reveal easily
martensite/austenite islands (in white) and the ferritic matrix (in brown).
(a) (b)
Figure III.7. (a) Bent sample with a crack on the bent apex surrounded with black line, (b) Crack observed
with binocular magnifier. Crack surrounded with black line.
Figure III.8 and Figure III.9 illustrate microcracks after an air-bending test before and after
LePera etching. Before etching, cavities (dark spots) are observed close to the microcrack,
highlighting damage all around it.
The microcrack is of nearly 180μm in depth and makes an angle of about 45° with the
outer surface. This confirms fracture mechanisms already observed in the past by
Chien et al. (2004). At first sight, the microcrack seems to initiate from the outer surface.
However, at about 50µm below the surface in tension, a slight discontinuity of the
microcrack is noticed (Figure III.8d). This could also suggest crack initiation from just
below the surface.
106
400 μm 400 μm
(a) (b)
discontinuity
40 μm 40 μm
(c) (d)
Figure III.8. Cross sectional observations using light microscopy on Steel 1 along the rolling direction. (a)
General view of the crack before etching, (b) general view of the crack after LePera etching, (c) close up
view of the crack before etching, (d) close-up view of the crack after LePera etching.
Figure III.9 shows another shape of microcrack for Steel 1 bent along the transverse
direction. The microcrack makes a kink with a kind of (vertical) flat area below the outer
surface suggesting two possibilities regarding crack initiation: either from the surface in
tension or, just below that surface, from the flat area where a change in the direction of the
microcrack is noticed. Here also, the microcrack develops with an angle of 45° with the
outer surface.
Possible
crack
initiation
400 μm 40 μm
(a) (b)
Figure III.9. Cross sectional observations using light microscopy on Steel 1 along the transverse direction,
LePera etching. (a) General view, (b) close up view.
Damage (dark spots) at the central segregation band and a very small microcrack at the
outer surface were also observed (Figure III.10) in another specimen. Moreover, there is a
kind of local necking at the central segregation between cavities. In “pure” bending, we
would not expect cavities at central segregation as this hard band would coincide with the
neutral fibre (no deformation). However, in air-bending, as this neutral fibre is slightly
shifted from the mid-thickness of the sheet, such events can be observed.
107
40 μm
400 μm
100 μm
Figure III.10. Metallographic observations on Steel 1 along rolling direction (LePera etching).
No metallographic observations of Steel 3 and Steel 4 are given here, as these steels did
not present any crack in air-bending.
In order to get more detail about the damage process in bending, scanning electron
microscopy was used. Figure III.11 illustrates two air-bent microcracks after a slight Nital
etching. The microcrack in Figure III.11a seems to have initiated from the outer surface. In
addition, damage is clearly noticed all around the crack and seems to nucleate either by
cleavage at martensite/austenite (M/A) islands, or by phase decohesion between M/A
islands and softer ferrite. Poruks et al. (2006) and Avramovic-Cingara et al. (2009) noticed
similar damage mechanisms after tensile tests, respectively on low-carbon bainitic steels
and DP600 steels. During the preparation stage (before the polishing), the microcracks
were filled with mounting resin. In the bent specimens, all microcracks were free from any
particles before metallographic preparation.
10µm 10µm
(a) (b)
Figure III.11. Two air-bent specimens of Steel 1 observed using scanning electron microscope.
(a) Microcrack which has surely initiated from the surface, (b) propagation of a microcrack through the
thickness. Damage is well noticed all around the microcrack.
108
2. Stretch bending test
Observations on stretch bent specimens were performed on both broken and unbroken
specimens. As fracture during this test is very unstable, it was therefore very difficult to
interrupt it before complete failure. Figure III.12 shows large cavities which have initiated
at central segregation but no significant crack propagation from this damage area did
occur. In fact, it seems very difficult to catch the stage between first damage development
at central segregation and full propagation across the thickness.
900μm 40μm
Figure III.12. Metallographic observations of stretch bent Steel 1 along the transverse direction.
Metallographic observations of broken samples are given in Figure III.13 to Figure III.15.
In each picture, rather large cavities located at the central segregation band are observed
whereas no crack is noticed close to the outer surface. Note that close to the mid-thickness,
a “V” shape of the fracture surface is also noticed. From these observations, the central
segregation seems to play a significant role during stretch bending as the crack apparently
initiates from it. Moreover, a similar orientation for the crack propagation (~ 45° from the
initial sheet plane) as for the air-bending test is observed.
900μm 100μm
Figure III.13. Metallographic observations of Steel 1 stretch bent along the transverse direction.
900μm
100μm
109
900μm 100μm
As the stretch bent specimens are completely broken, it is, however, difficult to prove that
cracks start from the central segregation region. Therefore, the device designed at Centre
des Matériaux was used to carry out interrupted tests and to stop the test just before failure,
as very low displacement rates can be reached with this device (0.05mm/s instead of
0.5mm/s with the hydraulic press).
As already mentioned, interrupting tests on this kind of device is still very difficult because
the crack remains very unstable. In one case, however, we successfully stopped the test
with an internal crack just before the failure (Figure III.16).
necking
Thickness
TD
100μm
400μm
Figure III.16. Interrupted stretch bending test on Steel 1, LePera etching. Crack initiation from the central
segregation region.
The shape of the crack clearly indicates that, in stretch bending, fracture initiates from the
central segregation while, in air-bending, the crack initiates from the outer surface or from
just underneath. Figure III.16 suggests the following mechanism: first, crack initiation
close to central segregation and then, crack propagation with a 45° angle, probably linked
to both mechanical fields and sensitivity of the material to strain localization.
In Figure III.16 we can also notice some necking of the sheet just above the crack, at the
outer surface. Indeed, the initial thickness of the sample was 1.6mm (before the stretch
bending) and after the stretch bending, just at the location of the crack, the thickness was
about 1.35mm. Out of the necking, the thickness was nearly 1.45mm.
Then, to study damage mechanisms in stretch bending at a finer scale, scanning electron
microscopy was used to observe metallographic sections after Nital etching (Figure III.17).
The main hard band is clearly seen in Figure III.17c with a lot of cavities within it. It is
also noticed that the thicker bands are more extensively damaged than the thinner ones.
Similar damage mechanisms are observed as for the air-bending test, i.e., cavities originate
either from decohesion between M/A islands and ferrite or from cleavage cracking of M/A
110
islands. In addition, at a macroscopic scale, as segregated bands are very hard (principally
composed of austenite and martensite), they fail earlier than the softer matrix, which can
sustain a higher amount of strain before to be damaged.
(a) (b)
Thickness
TD
(c)
Figure III.17. Stretch bent specimen of Steel 1. Observation of damage mechanisms after a slight Nital
etching using scanning electron microscopy. (a) General view of the crack, (b) close-up view of the crack at
mid-thickness, (c) close-up view of damage close to mid-thickness at the right of the crack.
I.C. Investigation of damage development in segregated bands using tensile

specimens
In order to determine the fracture strain of the segregation bands, several cross-section
observations were performed: first, on a broken tensile specimen, then on several
specimens deformed in uniaxial tension with various strain levels. SEM observations of the
broken tensile specimen are reported in Figure III.18. A decrease in the amount of damage
along the specimen axis (away from the fracture surface) is clearly observed.
111
fracture
Thickness
TD
fracture
Figure III.18. SEM observations along a broken tensile specimen of Steel 1 (Nital etching).
Interrupted tensile tests were performed on 12.5x50mm smooth tensile specimens for
various true strain levels (0.09 up to 0.21). These values of strain were measured with
Aramis software [Aramis system]. This facility allows monitoring the evolution of strain
during a test. For that purpose, a speckle has to be deposited on the surface of the studied
material and then, thanks to two CCD cameras, strain fields are derived from digital image
correlation. The values of strain are given for the regions where damage was actually
quantified thanks to SEM observations.
The following damage features are noticed (Figure III.20):

- No damage is observed for ε = 0.09 for Steel 1 and only a few cavities are observed for
Steel 4. Steel 3 is more severely damaged than the two other ones.
- For higher strains (0.11 up to 0.18), Steels 1 and 4 have similar amount of damage (cavity
of maximum 1 μm in size, similar number density) while Steel 3 exhibits much more
damage with larger cavities (up to 2 μm). The size of a cavity given here corresponds to its
opening (Figure III.19).
Band thickness
Size of cavities
Figure III.19. Schematics of damage quantification in a segregation band.
To summarize, for Steels 1 and 4, the first occurrence of damage in the microstructure is
observed for an amount of true strain ranging from 0.09 up to 0.11 while for Steel 3,
several larger cavities are noticed for a true strain of 0.09 meaning that they nucleated
earlier. Therefore, the fracture strain of the segregation bands for these steels is roughly
close to 0.10. In more detail, these results also show that the thicker segregated bands may
experience earlier damage nucleation during deformation.
112
Steel 1 Steel 3 Steel 4
ε = 0.09
ε = 0.11
ε = 0.18
Figure III.20. SEM observations after Nital etching of the three TRIP steels (1.6mm in thickness) after a
uniaxial tensile true strain of 0.09, 0.11 and 0.18.”
As Steel 3 is the most sensitive one to damage, the rest of the study concentrated on this
material. Figure III.21 shows damage at segregated bands for various tensile strain levels
(0.09 up to 0.21). As expected, the size of cavities increases with increasing tensile strain.
Tensile strain = 0.09 Tensile strain = 0.11 Tensile strain = 0.14
Tensile strain = 0.16 Tensile strain = 0.18 Tensile strain = 0.21
Figure III.21. Damage development during uniaxial tension of Steel 3 - various tensile true strain levels (from
0.09 up to 0.21) (Nital etching).
113
The spatial distribution of these cavities along the bands was also quantitatively studied.
Figure III.22 presents the average number of cavities contained over a 10-micron-long
segment along a segregated band. The standard deviation is very high, it is therefore very
difficult to provide a tendency in this plot. However, it can be suggested that the number of
cavities is relatively constant (~1cavity per 10μm in length) up to a strain of 0.16 and then,
a slight increase is noticed.
It suggests that the damage process may follow the following steps:
i) nucleation of first cavities just before a true strain of 0.09.
ii) growth of cavities up to a true strain of 0.16.
iii) nucleation of new cavities above this limit.
2
nb of cavities in 10µm
1.6
1.2
0.8
0.4
0
0 0.05 0.1 0.15 0.2 0.25
tensile strain
Figure III.22. Spacing of cavities along segregated bands as a function of the applied tensile strain.
To go further, the size distribution of cavities was determined for each strain level (Figure
III.23). The largest cavities are observed for the highest strain levels. For example, for a
true strain of 0.09, all cavities range between 0.2 and 0.8 μm in size. For higher strain
levels, cavities grow but there are also smaller ones. This means that there is nucleation of
new cavities during all steps of the tensile deformation. However, it is noticed that there
are a big amount of small cavities (size < 0.3 µm) for the material strained up to 0.18 and
0.21. Therefore, although nucleation seems to act at all strain levels, the main steps of
nucleation are for an amount of true strain close to 0.09 and from 0.18 and 0.21.
Tensile strain = 0.09 Tensile strain = 0.11

0.35 0.35
0.3 0.3
0.25 0.25
frequency
frequency
0.2 0.2
0.15 0.15
0.1 0.1
0.05 0.05
0 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.4 1.5 1.6 1.8 2 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.4 1.5 1.6 1.8 2
cavities size (µm) cavities size (µm)
114

0.35 0.35
0.3 0.3
0.25 0.25
frequency
frequency
0.2 0.2
0.15 0.15
0.1 0.1
0.05 0.05
0 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.4 1.5 1.6 1.8 2 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.4 1.5 1.6 1.8 2

0.35 0.35
0.3 0.3
0.25 0.25
frequency
frequency
0.2 0.2
0.15 0.15
0.1 0.1
0.05 0.05
0 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.4 1.5 1.6 1.8 2 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.4 1.5 1.6 1.8 2
Figure III.23. Cavities size distribution for the various true strain levels (0.09 up to 0.21).
Figure III.24 illustrates this assumption where larger cavities and smaller ones are noticed.
The big ones generally measure between 1 and 2 μm in width and are circled in red. The
cavities circled in yellow represent the “newer ones” (opening width < 0.3 μm).
(a) (b)
Figure III.24. SEM observation after Nital etching: spatial distribution of damage cavities along the band for
a specimen deformed under uniaxial tension up to a true strain of (a) 0.18 and (b) 0.21. Red circles: big
cavities (open at low strain), yellow circles: second generation of (i.e. newly nucleated) cavities.
115
I.D. Summary of section I
It was demonstrated that crack initiation mechanisms are different in air bending and in
stretch bending. In fact, the two bending procedures do not test the same microstructural
parameters. Stretch bending depends on the central segregation while air-bending is more
sensitive to banded structures i.e. to segregation bands closer to the surface of the sheet.
Regarding damage in the microstructure after uniaxial tension, it was demonstrated that the
amount of true strain needed to nucleate damage cavities at central segregation is about
0.10 for the three TRIP-aided steels studied. This value is strongly dependent on the band
thickness. Cavities were only observed for higher fracture strains for very thin hard bands.
To further investigate the influence of banded structures in air-bending, the main hard band
was shifted from the mid-thickness towards the outer surface in thickness-reduced Steel 2.
This is the aim of section II. In the following part, it is also intended to quantitatively
correlate band thickness and band sensitivity to damage nucleation during air-bending.
116
II. Influence of segregated bands on bendability

II.A. Investigation of damage mechanisms in air-bending
This part is based on a journal paper which has been accepted for publication to Materials
Science and Engineering A in March 2011. The text of this article is reproduced below.
117
118
119
120
121
122
123
124
125
126
II.B. Complementary observations
In this part, additional observations about bending of thickness-reduced Steel 2 with

various locations of hard bands are reported.
Figure III.25 shows specimens bent up to 180° for four main hard band configurations. As
expected, the severity of the test increases with the proximity of the central segregation
with the outer surface.
Segregation below the Segregation at mid- Segregation above Segregation just under
mid-thickness thickness the mid-thickness the outer surface
OK microcrack crack big crack
Segregation closer to the outer surface during bending
Increasing sensitivity to cracking

Figure III.25. Macrographic observations of four specimens of reduced-thickness Steel 2 bent along TD up to
180° with various locations of the main segregation band.
Figure III.26 illustrates cross-section optical microscopy observations for the four previous
cases after a 180° bending and a LePera etching.
400μm 100μm 400μm 100μm
(a) Segregation below the mid-thickness (b) Segregation at mid-thickness
400μm 100μm 400μm 100μm
(c) Segregation above the mid-thickness (d) Segregation just below the surface
Figure III.26. Optical micrographs of thickness-reduced Steel 2 for various locations of the main hard
segregation band with a general view and a close-up view, LePera etching. (a) Segregation below the mid-
thickness, (b) segregation at mid-thickness, (c) segregation above the mid-thickness, (d) segregation just
below the surface.
127
As observed in Figure III.25, cracking becomes more and more severe with a shift of the
segregation towards the outer surface. Both the opening and the depth of crack increase
with the proximity of the segregation with the surface.
In order to investigate crack initiation more precisely, additional air-bending tests were
carried out with specimens having the segregation just below the outer surface. Air-
bending tests were interrupted just before cracking by a trial and error procedure (bending
angles 110°, 142° and 160°) followed by metallographic observations using light optical
and scanning electron microscopy. Figure III.27 shows pictures obtained for two
specimens with the segregation band below the outer surface: one with the segregation
band close to the surface (~130µm) and the other one with the segregation band very close
to the surface (~50µm). Damage at hard band is observed for the two specimens without
any crack at the outer surface. As expected, damage at the segregation band is more intense
for the specimen with the segregation just below the surface.
80μm 30μm 80μm 30μm

(a) (b)
Figure III.27. Optical micrographs of thickness-reduced Steel 2 specimens bent without crack: (a)
Segregation 130µm below the surface with a 160° bending angle, (b) segregation 50µm below the surface
with a 110° bending angle. Air-bending along TD.
Then, additional tests were performed with a little more severe bending angle, in order to
stop the test just before cracking at the outer surface. Figure III.28 and Figure III.29 show
internal cracks at hard segregation bands with necking at the outer surface above each sub-
surface crack, meaning that in air-bending, cracks could initiate from just below the
surface, as already suggested in Part I of this chapter. It is also noticed in Figure III.29 that
the crack propagates with a 45° angle as already observed in part I.B.
80μm 100μm 80μm

(a) (b) (c)
Figure III.28. Optical micrographs of sub-surface cracks for three air-bent thickness-reduced Steel 2
specimens with a 142° angle (LePera etching).
128
45°
Figure III.29. SEM observations of damage at hard band located just under the outer surface (Nital etching).
Bending up to 110° for thickness-reduced Steel 2.
Finally, Figure III.30 illustrates three cracks due to air-bending. The central crack does not
give enough precise clues to determine the exact initiation point (maybe from the surface
or maybe from just below where there is a change in the direction of propagation). But
thanks to the two small cracks located at each side of the main microcrack, and being
initiated below the outer surface at the same hard band, we assume that all three cracks
initiated at that segregation band. There are also a lot of cavities or even of arrested
microcracks along that segregation band (e.g. in the bottom left picture of Figure III.30).
Figure III.30. SEM observations of sub-surface cracks of two thickness-reduced Steel 2 bent specimens
(Nital etching), 142° bending angle.
It can be concluded that in air-bending, cracks may actually nucleate from segregation
bands, provided that the combination of band thickness and local strain is favourable. A
quantitative relationship was given by Equation (3) in the previous part with (for this steel)
a = 0.1 and b = 0.11. However, this relationship is based on a very simplified analysis of
the bending test.
129
III. Summary of Chapter 3

In this chapter, damage mechanisms were studied both in air-bending and stretch bending
tests as well as in tensile tests with various true strain levels. In a second part, the influence
of segregated bands as a function of their locations through the thickness was more
particularly investigated using thickness-reduced specimens of Steel 2.
The investigation of damage mechanisms gave the following results:
- The fracture mode is ductile both in air-bending and in stretch bending. Using light
and scanning electron microscopy, cracks were shown to propagate with a 45°
angle with respect to the outer surface, indicating a strong sensitivity to strain
localization.
- Damage was noticed at hard segregated bands with more cavities for the thickest
bands for all bending tests. This suggests that segregation bands play a non-
negligible role in fracture during bending. In fact, in air-bending, the closer the
thicker segregation bands to the outer surface, the higher the severity of cracking.
- Cracks initiate from the outer surface or from just underneath (at a hard band) in
air-bending while they initiate from the central segregation band in stretch bending.
- Thanks to tensile tests, it was revealed that the fracture strain of the main hard band
is close to 0.10 for the three TRIP-aided steels studied. Based on this result and on
the previous observations, a simple quantitative relationship between cavity
nucleation, band thickness and local “mesoscopic” strain was proposed.
130
Dans ce chapitre, les mécanismes d’endommagement ont été étudiés à la fois en pliage en
V et sous traction, ainsi qu’en traction à différents taux de déformation. Dans une
deuxième partie, l’influence des bandes de ségrégation en fonction de leurs positions à
travers l’épaisseur a été plus particulièrement étudiée grâce aux échantillons de l’acier 2
rectifié.
L’étude des mécanismes d’endommagement a révélé les résultats suivants :
- Le mode de rupture est ductile à la fois en pliage en V et en pliage sous traction.

Grâce à la microscopie optique et à balayage, il a été observé que les fissures se
propagent à 45°, indiquant une forte sensibilité à la localisation en déformation.
- L’endommagement a principalement été observé sur les bandes de ségrégation et

nous avons montré que les bandes les plus épaisses étaient les plus endommagées.
Ce résultat suggère le rôle non négligeable des structures en bandes sur la
fissuration en pliage. De plus, en pliage en V, plus les bandes de ségrégation sont
proches de la surface, plus les éprouvettes fissurent tôt.
- Les fissurent s’amorcent en surface ou juste en dessous (sur les bandes de

ségrégation) en pliage en V, tandis qu’elles s’initient à partir de la ségrégation
centrale en pliage sous traction.
- Grâce aux tests de traction interrompus, nous avons montré que la déformation à
rupture des bandes de ségrégation était proche de 0.10 pour les trois aciers TRIP
étudiés. Utilisant ce résultat et les précédentes observations, une simple relation
quantitative entre la germination de cavités, l’épaisseur des bandes et la
déformation locale mésoscopique a été proposée.
131
Chapter 4. Mechanical analysis of fracture during air and stretch bending tests
Chapter 4
Mechanical analysis of fracture during

air and stretch bending tests
In chapter 2, air-bending tests have been modelled

thanks to finite-element simulation without taking
damage development into account. In chapter 3, it was
demonstrated that damage mechanisms are different in
air and stretch bending tests and the role of the hard
bands on damage development was experimentally
investigated.
In this section, constitutive equations were associated to
a Gurson-Tvergaard-Needleman yield potential for the
mechanical analysis of damage development and failure
during air and stretch bending tests. A comparison
between both air and stretch bending tests was first
carried out for Steel 1. A complementary parametric
study including the effect of meshing, friction coefficient
and 2D vs. 3D modelling is then presented for the stretch
bending test. In the last part of this chapter, the influence
of the location of the band within the sheet thickness is
numerically investigated on Steel 1 tested in air-bending.
133
Table of contents
I. MECHANICAL ANALYSIS OF THE AIR AND STRETCH BENDING TESTS135
II. INFLUENCE OF SIMULATION CONDITIONS ON THE STRETCH-BENDING

TEST ................................................................................................................................. 157
II.A. Influence of meshing conditions ............................................................................... 157
II.B. Influence of boundary conditions.............................................................................. 159
II.C. Summary: influence of modelling conditions on stretch-bending test predictions... 162
III. PARAMETRIC STUDY: EFFECT OF THE LOCATION OF THE HARD

BAND ON FAILURE DURING AIR-BENDING ........................................................ 162
IV. SUMMARY OF CHAPTER 4 ................................................................................. 166
134
I. Mechanical analysis of the air and stretch bending

tests
This part is based on a project journal paper which is to be submitted for publication to
Engineering Fracture Mechanics. The text of this article is reproduced below.
Flow and fracture analysis of a ultrahigh strength steel during stretch

bending and air-bending tests
D. Rèche 1), 2), J. Besson 1), T. Sturel 2), A.F. Gourgues-Lorenzon 1)

1)
MINES ParisTech, Centre des Matériaux, CNRS UMR 7633, BP 87
91003 Evry Cedex, France
2)
Abstract
In order to evaluate the bendability of a ultra high strength steel, air-bending and stretch
bending tests were performed along rolling and transverse directions. A full analysis of
these tests was given with both experimental and numerical data. To examine and
understand damage mechanisms involved, cross section observations were carried out both
after interrupted air-bent tests and after broken and interrupted stretch bent tests. Cracks
initiate from central segregation bands (if any) in stretch bending while cracks initiate from
the outer surface or from just below in air-bending. The two tests were then modelled
thanks to finite element simulation including a damage criterion, in 2D for the air-bending
test and in 3D for the stretch bending test. An extension of the Gurson-Tvergaard-
Needleman (GTN) model was used in this simulation with damage development
principally governed by nucleation of cavities. Good agreement was obtained between
experimental data and model predictions.
Key words
Air-bending, stretch bending, ultra high strength steel, finite element simulation, damage
mechanisms.
1. Introduction
Stamping is one of the most common methods in industry to give specified shapes to sheet
metal. Many different deformation modes are used to obtain the desired shape of sheet
steel during stamping, such as stretching, bending, deep drawing, embossing, and so on. In
recent years, the demand to reduce the weight of automotive vehicles led to increasing use
of Advanced High Strength Steels (AHSS). Thus, the forming behaviour of these steels
135
had to be quantified both experimentally and numerically. Therefore, the finite-element

simulation of various forming modes is required to characterize the formability of these
materials and to determine the limit to which the material can be formed without reaching
failure.
The forming limit depends on processing conditions and on the shape of the part. Hence, it
is necessary to find an adapted failure criterion to well predict the sheet formability. The
forming limit curve (FLC) is the most widely used criterion in sheet metal forming analysis
(Keeler and Backofen, 1964; Goodwin, 1968). However, Laukonis and Gosh (1978), Graf
and Hosford (1993) showed that FLCs are only valid along proportional strain paths.
During metal stamping, the stress/strain state changes continuously, so that the loading
path is generally not proportional. Consequently, traditional FLCs cannot be used for
accurate predictions of the behaviour during bending.
Only a few papers deal with the initiation of fracture in bending. For example, in bending
without tension, cracks seem to initiate from the surface (Chien et al., 1981; Steninger and
Melander, 1982; Sarkar et al., 2001) or from just below. In stretch bending, there are a lot
of papers addressed the location of the crack along the bent sheet, i.e., at the punch nose or
at a sidewall.
Among them, Demeri (1981), Damborg (1998), Sadagopan et al. (2003), Hudgins et al.
(2007) and Kim et al. (2009) performed stretch bending tests on several steels and
aluminium sheets to determine their stretch bendability. Typically, the height at failure as a
function of the R/t (punch radius over sheet thickness) ratio was reported as a measure of
formability under stretch bending conditions. The test was stopped once the first failure
appeared and they noticed that failure occurred at the punch nose radius of the specimen
for small R/t ratios. When R/t increased above some limit, (R/t)c, failure moved from the
punch nose to the sidewall. For example, Hudgins et al. (2007) found that TRIP steels have
a high value of (R/t)c compared to other UHSS. However, there was no information
concerning the location of crack initiation through the thickness in all these studies,
although the microstructure of the sheet may be not fully homogeneous through the
thickness. Therefore, a mechanical analysis across the thickness is necessary.
In order to predict fracture during forming processes, several fracture and damage criteria
have been proposed in the past (Cockroft and Latham, 1968; McClintock, 1968; Rice and
Tracey, 1969; Gosh, 1976; Gurson, 1977; Leroy et al., 1981; Lemaitre, 1985; etc.). The
family of continuum damage mechanics models was widely used to model forming tests.
In these models, damage evolution is coupled with elastoplastic constitutive equations and
fracture is predicted to occur as soon as the damage variable reaches a critical value.
Gurson (1977) developed a ductile damage macroscopic constitutive law based on
microvoid growth. Its model was largely used and extended in literature (Tvergaard, 1982;
Tvergaard and Needleman, 1984). Gurson’s model is the first micromechanical model for
ductile fracture introducing a strong coupling between deformation and damage.
Several models were used in the past to analysed bending tests. Among them, Date et al.
(1999) and Gan and Wagoner (2004) studied the springback of bent sheets. To accurately
predict both bendability and fracture, Brunet et al. (1996) used a modified Gurson model
with a Hill anisotropic yield criterion to simulate square-cup deep drawing. Hambli et al.
(2004) simulated the air-bending process in 3D by coupling a Von Mises yield criterion
with a Lemaitre damage equation. They showed that the simulation of air-bending,
respectively with and without damage in the model, led to similar load/punch displacement
136
curves up to the maximum in load. For higher values of punch displacement, the model
including damage development predicted a faster decrease in load than the damage-free
model. More recently, two studies reported numerical simulation of forming tests, both
using a Von Mises yield criterion coupled with a damage evolution equation, respectively,
the Gurson-Tvergaard-Needleman (GTN) model (Uthaisangsuk et al., 2008) and the
Lemaitre approach (Bahloul et al., 2006). Uthaisangsuk et al. (2008) used a galvanized
bake-hardening steel of 1mm in thickness to study the Nakazima test and showed that the
GTN model offers an acceptable sheet metal failure criterion. Bahloul et al. (2006)
obtained a good prediction of punch load and stress distribution during the wiping-die
bending process of an high strength low alloy steel using a 3D finite element simulation.
Ben Bettaieb et al. (2010) simulated a pure bending test on a dual phase steel using a
mixed hardening rule, an anisotropic yield criterion and an extended Gurson model. In
their study, Luo and Wierzbicki (2010) used Hill’s yield criterion coupled with a Modified
Mohr-Coulomb (MMC) ductile fracture criterion (Bai and Wierzbicki, 2010) to predict the
failure during stretch bending of DP780 steels. They demonstrated that the 3D finite
element simulations provided the best prediction in terms of fracture location,
load/displacement curves and failure wall stress compared to shell or plane strain finite
element simulations. Kim et al. (2009) simulated the draw bending test on a DP980 steel
with a coupled thermo-mechanical, damage-free finite-element model. They demonstrated
that their model accurately predicted failure location, maximum stress and displacement at
failure by taking the change in temperature during fracture into account. To summarize,
from these analyses, an anisotropic yield criterion is required to account for the anisotropic
behaviour of most of steels and to accurately predict curves, stress and strain fields. A
mixed hardening rule enables to account for the Bauschinger effect and is more reliable in
case of non proportional loading, as in the case of a prestraining before a bending test
(Rèche et al., 2011).
The aim of the present study being to predict failure conditions for two different bending
tests, the local approach to ductile fracture was chosen to be used by including damage in
the constitutive equations.
In this study, elastoplastic constitutive equations coupled with damage development were
chosen to describe damage mechanisms leading to fracture on one low alloy TRIP-aided
steel both in air and stretch bending. As stretch bending tests were carried out at a very low
displacement rate in the present study, it is reasonable not to take the change in
temperature into account in the model. Thus, no thermo-mechanical model as in Kim et al.
(2009) was used in this study. A model based on the GTN equations was used in the in-
house finite element software Zset (Besson and Foerch, 1997). First, air-bending tests,
stretch bending tests and metallographic observations were performed to determine
fracture initiation mechanisms. The damage criterion was then introduced into existing
constitutive equations (Rèche et al., 2011) for the modelling of both bending tests.
2. Materials and testing procedures
2.1. Material
A cold rolled low alloy TRIP-aided steel with a thickness of 1.6mm (C~0.2 wt%, Mn~1.6
wt%, Al+Si~1.5 wt%) was investigated in this study. Its mechanical properties are shown
in Table 1. This material has a multiphase microstructure with martensite/retained austenite
islands in a softer ferritic matrix. The third-thickness microstructure (Figure 1a) was
137
homogeneous while the mid-thickness microstructure presented bands of hard phases

principally composed of retained austenite and martensite (Figure 1b).
0.2% proof Tensile Uniform Fracture

stress (MPa) strength (MPa) elongation (%) elongation (%)
RD 467 866 15.0 20.0

TD 546 851 15.0 19.2
Table 1: Room temperature mechanical properties along rolling (RD) and transverse (TD) directions
(ISO 20x80mm tensile specimens).
sheet
thickness
RD
40µm 40µm
(a) (b)
Figure 1: Light optical micrographs of the as-received steel sheet, LePera etching. (a) third-thickness
microstructure, (b) mid-thickness microstructure. Hard phases (austenite + martensite) in white, ferrite in
grey.
2.2. Air-bending tests
Air-bending tests were carried out at room temperature using a universal tensile machine
on 50x100mm specimens (Figure 2). Air-bending tests were performed without
lubrication. The sheet was laid on two rotary rolls and a very sharp punch (tip radius of
0.1mm) went down to bend specimens. Before the tests, the specimen edges were ground
to avoid premature failure that could be due to cutting edge-induced damage effects
sometimes encountered in AHSSs (Dalloz et al., 2009). Five specimens were tested along
both rolling and transverse directions on as-received material until microcracking occurred
(corresponding to a load drop of 5%) and a very good reproducibility was observed.
Finally, the bending angle α (Figure 2b) was measured after unloading.
As in industrial stamping, bent areas are often already deformed when bending occurs, it
seemed necessary to investigate the influence of prestraining before an air-bending test. In
a previous study (Rèche et al., 2011), the influence of prestraining prior to air-bending and
more particularly, the influence of the kinematic contribution to hardening was discussed
but no damage was included in the model, leading to a good prediction of the
load/displacement curve except at the end of the test. Tensile tests along TD with a
prestraining of 10% were carried out on the as-received material and air-bending
specimens were cut from tensile specimens and tested along TD until microcracking
occurred.
138
α
(b)
(a)
Figure 2: (a) Air-bending device (b) schematics of the air-bending test with α = bending angle.
2.3. Stretch bending tests
In order to investigate the formability of this steel, stretch bending tests were performed at
room temperature. They were carried out on a conventional Instron servohydraulic testing
machine designed and developed for that purpose making it possible to apply very low
displacement rates (0.05mm/s) (Figure 3). As for an Erichsen test, the method consisted in
clamping a 180mm x 40mm blank between a die and a blank-holder. In order to limit
friction, lubrication was employed between the punch and the sheet. Grease was put on the
punch radius and then a Teflon film was added on it. Specimen edges were ground before
testing to avoid cracks starting from them. Five samples along both rolling and transverse
directions were bent with three different punches (tip radii = 2mm, 4mm and 6mm,
respectively) along the transverse direction. As expected, the higher the punch tip radius,
the better was the bendability. Only results with the punch radius of 4mm will be given in
the following. The test was stopped either once the sample was broken, or just before
fracture to investigate interrupted stretch bent tests. Finally, the bending angle at fracture
(α in Figure 3c) was given.
α
β
(a) (b) (c)
Figure 3: (a) stretch bending device, (b) stretch bending test in progress with specimen clamped between die
and blank-holder and a punch radius of 4mm, (c) schematics of the stretch bent sheet (thick line) with α =
bending angle.
3. Experimental results
3.1. Air-bending curves
Load vs. punch displacement curves are shown in Figure 4 both for as-received and
prestrained specimens. The prestrained specimen was bent up to a lower value of the punch
139
displacement to get the first crack. Therefore, prestraining prior to air-bending affects the
bendability of the sheet because cracking was observed earlier. This result is in agreement
with the work of Friedman and Luckey (2002) and Sarkar at al. (2001).
The hoop strain reached at the outer surface at microcracking for the as-received material
was evaluated to be 0.27 in Rèche et al. (2011). If the amount of elongation (10%) applied
during prestraining is removed from this value, the punch displacement corresponding to a
hoop strain of 0.175 at the outer surface was found to be equal to 20.5 mm. This value is in
very good agreement with the punch displacement at microcracking of the prestrained
specimen. Thus, as a first approximation, an additive rule on strain can be used to estimate
the bendability, for tensile strains applied along the same direction during both prestraining
and bending. However, this very simple rule may obviously not work for other prestraining
conditions.
Regarding the as-received steel bent along both rolling and transverse directions, slight
differences were observed between the two curves. Note that the maximum load reached
for the bent specimen along RD is lower than the one along TD. In addition, specimens
taken along RD were bent up to a higher punch displacement than for TD specimens,
meaning than the transverse direction is less ductile than the rolling one. These differences
are consistent with the slight anisotropy of this steel in uniaxial tension.
2.5 prestrained
TD
2 as-received
Load (kN)
TD
1.5
1 as-received
RD
0.5
0
0 5 10 15 20 25 30
Figure 4: Air-bending test. Load vs. punch displacement curves both on as-received and on prestrained
material.
3.2. Stretch bending curves
A stretch bending load vs. punch displacement curve is illustrated in Figure 5. In air and
stretch bending, curves look very different. While in air-bending, the load decreased rather
slowly after its maximum, in stretch bending, the load sharply decreased after its maximum
due to unstable fracture. Note that in stretch bending, the load reached just before fracture
was much higher than in air-bending test, this was due to the tension applied on the whole
specimen in these bending conditions.
140
60
50
40
Load (kN)
30
20
10
0
0 5 10 15 20 25
Displacement (mm)
Figure 5: Stretch bending test along TD. Load / displacement curve.
3.3. Crack initiation both in air-bending and stretch bending
In order to investigate damage mechanisms, cross-sectional observations were carried out

on one specimen after cracking in air-bending. As noticed in Figure 6, cracks generally
initiated from the outer surface. In a recent study (Rèche et al, 2010), it was also noticed
that cracks can initiate from just below the surface provided that there are thin hard bands
underneath the outer surface.
sheet
thickness
TD
Figure 6: Cross-section of an air-bent specimen. Optical micrograph, LePera etching.
As stretch bending specimens broke suddenly, it was very difficult to interrupt stretch
bending tests just before fracture even by using the device developed for this study. Stretch
bent specimens appeared to be either safe or broken into two parts so it was difficult to
stop the test with a crack inside the sheet thickness. In Figure 7, noticeable cavities located
at the central hard band (in white) are seen in a broken specimen. In addition, a V-shaped
kink was noticed in the cross-section of the fracture surface, close to the central hard band.
Based on these observations, the central hard band seemed to play a significant role during
stretch bending as a significant amount of damage initiated from it. In one case, however,
the test was successfully stopped with an internal crack just before specimen failure
(Figure 8). This figure suggests the following mechanism: first, cavity nucleation close to
the central hard band, then, microcrack propagation with a 45° angle probably linked to
both mechanical fields and sensitivity of the material to strain localization. In Figure 8a,
necking of the sheet just above the crack was observed. The initial thickness of the as-
received sample was 1.6mm and after stretch bending, just above the crack, the thickness
was about 1.35mm. Out of the crack, the thickness was 1.45mm.
141
sheet
thickness
(a) (b) TD
Figure 7: Optical micrographs after stretch bending. LePera etching. (a) General view of a broken sample, (b)
close-up view of damage close to the central hard band.
necking
sheet
thickness
TD
(a) (b)
Figure 8: Optical micrographs after an interrupted stretch bending test. LePera etching. (a) General view of
the cracked specimen, (b) close-up view of the crack which seems to have initiated from the central hard
band.
Fractographic observations (Figure 9a, 9b, 9c) showed that in stretch bending, cracks
propagated by a ductile mechanism, as dimples were observed on the fracture surface. In
Figure 9a, the mid-thickness of the sheet was well visible (horizontal dark area). Figure 9b
represents a close-up view of the mid-thickness part of the fracture surface. It seems that
cleavage-like facets are observed in this area, contrary to Figure 9c where dimples were
noticed everywhere out of this mid-thickness area. Damage was noticed with cavities
located either at the martensite/ferrite interface or by cleavage of martensite islands
(Figures 9d and 9e). In addition, damage was noticed to initiate first at hard bands and then
in the softer matrix.
cleavage
mid-thickness
50µm 20µm 20µm

(a) (b) (c)
142
100µm 10µm
(d) (e)
Figure 9: Fractographic observations by SEM of a broken sample after stretch bending. (a) General view with
mid-thickness surrounded by a white ellipse, (b) close-up view of the mid-thickness with some of cleavage
showed by white arrows, (c) close-up view of ductile failure with dimples far from the mid-thickness. (d, e)
SEM observations after Nital etching of the crack and damage close to the crack at hard bands. The crack in
the upper right corner of (e) is the one shown in (d).
Cross-sectional observations of cracking showed that crack initiation mechanisms are

different for the two bending tests (initiation at the outer surface in air-bending vs.
initiation from central hard band in stretch bending) so the key microstructural parameters
controlling crack initiation might not be the same for the two investigated bending tests.
3.4. Investigation of cavity nucleation both at and out of hard bands
Uniaxial tensile tests interrupted after various strain levels were performed in previous
work (Rèche et al., 2010) in order to investigate the beginning of damage cavity nucleation
at hard bands. For this steel, cavities were first observed for an average true axial strain of
0.11 in uniaxial tension (Figure 10a). Then, it was also interesting to determine the
beginning of nucleation in the rest of the steel (“softer matrix” composed of ferrite and
austenite/martensite islands). Based on SEM observations performed on air-bent specimens
(Figure 10b), cavities were observed in the tensile zone in the softer matrix. On the other
hand, by using finite element analysis of the air-bending test (Rèche et al., 2011), the
evolution of strain as a function of the thickness of the sheet was determined. Therefore, by
coupling SEM observations and results of mechanical analysis, the nucleation of first
damage cavities was observed in the softer matrix for a true strain close to 0.21.
(a) 5µm (b) 5µm
Figure 10: cross-section SEM observations of damage after a Nital etching. Damage cavities are surrounded
with white circles. (a) Damage at hard band after an interrupted uniaxial tensile test (ε = 0.11), (b) damage in
the softer matrix with either decohesion at martensite/ferrite interface or with cleavage at martensite islands
(air-bent specimen).
143
4. Modelling
4.1. Constitutive equations
In a previous work, (Rèche et al., 2011), constitutive equations of this material were
established thanks to an experimental database based on tensile tests on smooth and
notched specimens and on shear tests. These constitutive equations included both isotropic
and kinematic contributions to hardening with an anisotropic yield criterion (Bron and
Besson, 2004). Other authors such as Eggertsen and Mattiasson (2009), Williams et al.
(2010), Vladimirov et al. (2010) and Shinohara et al. (2010) also used a mixed hardening
rule coupled with an anisotropic yield criterion.
To simulate both air-bending and stretch bending tests, the damage model used was based
on extensions of the GTN model (Gurson, 1977; Tvergaard and Needleman, 1984).
Thuillier et al. (2010) used similar method by predicting bending test on an aluminium thin
sheet. In their study, they characterize the mechanical behaviour of the material using
tensile tests on smooth and notched specimens and equibiaxial tests. The GTN model was
used to predict the ductile rupture in bending but only a purely isotropic hardening rule and
a Von Mises yield criterion was used in their work.
As in Shinohara et al. (2010), kinematic hardening is represented by a back stress X

introduced by Chaboche (1989). The yield surface is defined by using the difference B
between the Cauchy stress σ and the back stress: B = σ − X . In order to account for
anisotropic plasticity, it was necessary to use an anisotropic stress measure to define the
yield surface. The model proposed in (Bron and Besson, 2004) was used in the following.
It consists in a generalization of previously published models (Barlat et al., 1991; Karafillis
and Boyce, 1993).
For any symmetric second order tensor, T , the anisotropic scalar measure ( TE ) of T is
defined as:
1/ a
⎛ K ⎞
∑ ( )
a
TE (T ) = ⎜⎜ α k T ⎟
⎟⎟ (1)
⎜ k =1 Ek
⎝ ⎠
The function T E (T ) is positive and homogeneous of degree 1. TEk are secondary

anisotropic scalar measures and α k are positive coefficients with a sum equal to 1. In
practice, only two functions ( K = 2 ) were used i.e. α 1 = α and α 2 = 1 − α . One first defines
two modified deviators:
T k = L :T (2)
k
where the fourth order tensor L k is expressed as follows using Voigt notations:
144
⎛1⎛ k 1 k 1 k ⎞
⎜ ⎜ LL + SS ⎞⎟
k
c c − c − c 0 0 0 ⎟
⎜3⎝ ⎠ 3 SS 3 LL ⎟
⎜ 1 k 1⎛ k 1 k
0 ⎟
k ⎞
− c ⎜ +c c ⎟ − c 0 0
⎜ 3 SS 3 ⎝ SS TT ⎠ 3 TT ⎟
Lk = ⎜ 1 k 1 k 1⎛ k ⎟ (3)
⎜ LL + TT ⎞⎟
k
⎜ − c − c c c 0 0 0 ⎟
⎜ 3 LL 3 TT 3 ⎝ ⎠ ⎟
⎜ 0 0 0
k
0 0 ⎟
⎜ c TL ⎟
⎜ 0 0 0 0
k
0 ⎟
⎜ c LS ⎟
⎜ 0 0 0 0 0
k
⎟
⎝ c
ST ⎠
The eigenvalues of T k are then computed: Tk1 ≥ Tk2 ≥ Tk3 . TE1 and T E 2 are respectively
computed as :
1 / b1
⎡1 ⎛ b1 b1 b1 ⎞⎤
TE1 = ⎢ ⎜ T12 − T13 + T13 − T11 + T11 − T12 ⎟⎥ (4)
⎣2 ⎝ ⎠⎦
1 / b2
⎡ 3b2 ⎛ 1 b2 b2 b2 ⎞⎤
TE 2 =⎢ b ⎜ T2 + T22 + T23 ⎟⎥ (5)
2 +2⎝ ⎠⎦⎥
⎣⎢ 2
a , b1 , b2 and α are four material parameters that influence the shape of the yield surface
1, 2
and the model was simplified assuming a = b1 = b2 . cTT ...ST parameters control yield
anisotropy. L, T and S respectively correspond to the longitudinal, the transverse and the
thickness directions of the sheet. As mechanical tests were performed along in-plane
directions, coefficients corresponding to shear in the thickness direction could not be
determined. It was therefore assumed that c ST
k
= c LS
k
= 1.
Experimental observations led to define the void fraction as a damage indicator. Among
the models using the void fraction as a damage parameter, the GTN model is the most
widely adopted. It couples the void volume fraction, f , with material constitutive
equations. For any symmetric second-order tensor, T , the corresponding effective
stress, T* (T , f ) , is implicitly defined by:
2
⎛T ⎞ ⎛q T ⎞
Ψ (T* , T E , Tkk , f ) = ⎜⎜ E ⎟⎟ + 2q1 f * cosh⎜⎜ 2 kk ⎟⎟ − 1 − q1 2 f * 2 = 0 (6)
⎝ T* ⎠ ⎝ 2 T* ⎠
In Equation (6), TE = TE (T ) is the anisotropic stress, Tkk is the trace of the stress tensor.
Parameters q1 and q 2 introduced by Tvergaard (1982) are assumed to be constant and
respectively equal to 1.5 and 1. These q parameters can be used to correct the effect of
void shape changes, of large strain hardening and of void interaction. f * is a function
introduced by Tvergaard and Needleman (1984) to represent the coalescence of damage
cavities:
fg + fn for f g < f c
f* (7)
fc + fn +δ ( f g − fc ) for f g > f c
145
In Equation (7) f c represents the critical growth void volume fraction for which
coalescence starts, f g represents the damage due to cavity growth and f n corresponds to
damage nucleation as in (Besson et al., 2000) . δ and f c are material parameters.
The yield function is then expressed by:
φ = B* ( B , f ) − R ( p ) (8)
Here, p is the cumulative plastic strain and R is the flow stress. The normality rule applies
and the strain rate tensor is given by:
∂B*
ε& p = (1 − f ) p& (9)
∂σ
Damage growth is both controlled by void nucleation and void growth. Therefore, the
evolution law for damage is expressed as (Chu and Needleman, 1980):
f& = (1 − f )tr (ε& p ) + An p& (10)
The first term corresponds to the evolution law for f g given by mass conservation and the
second term corresponds to the contribution of void nucleation. An is the nucleation rate
parameter which may depend on the material state variables such as plastic strain and
stress state.
Kinematic hardening is described using an intermediate tensor variable, α , for which the
evolution equation is given following the standard expression for non linear kinematic
hardening (Lemaitre and Chaboche, 1990) by:
α& = ε& p −
3D
p& X
(11)
2C
C and D are material parameters.
2
The equation ( X = Cα ) used in the damage-free version of the model (Rèche et al., 2011)
3
is not valid here. Therefore, an intermediate back stress, χ , is used. It represents the back
stress at the microscopic level and is linked to α by: χ = 2 Cα .
3
2 2
In practice, χ 1 = C1α1 and χ 2 = C2 α 2 and X = X 1 + X 2
3 3
According to Besson and Guillemer-Néel (2003), the actual back-stress, X , is defined
implicitly by the following equation:
2 ∂X i *
χi = X i* (12)
3 ∂Xi
To solve this equation, X * is defined as a solution of Ψ ( X * , X eq , X kk , f ) = 0 where X eq and

X kk are respectively the von Mises invariant of X and its trace.
146
Finally, the elastoplastic behaviour of the TRIP sheet has been identified thanks to tensile
tests on smooth and notched specimens and to shear tests (Rèche et al., 2011). The
evolution equation for isotropic hardening is given by the following:
R( p) = R0 + Hp + Q(1 − exp(− bp )) (13)
p is the cumulative plastic strain, R is the flow stress and H , Q and b are material
parameters.
In addition, note that while tensile and shearing tests were being carried out on the steel,
retained austenite was transformed into martensite during deformation, meaning that the
TRIP-effect was implicitly taken into account in the model. Hence, no micromechanical
model was used here because it was well beyond the scope of this study.
4.2. Adjustment of material parameters
Adjustment of all model parameters yielded the values reported in Table 2. As a relatively
high level for plastic strain has to be reached before damage occurs, it was therefore
possible to adjust parameters related to plastic hardening and plastic anisotropy
independently of damage parameters.
Consequently, plastic behaviour parameters were determined separately and their values
were given in (Rèche et al., 2011). As already mentioned, they were determined using an
experimental database based on tensile tests on smooth and notched specimens and shear
tests.
To include the effect of stress triaxiality on the damage cavity nucleation rate as evidenced
by (Kao et al., 1990), the void nucleation rate parameter An in Equation (10) was
expressed as a function of the local stress triaxiality ratio, Tr , and of the cumulative plastic
strain as:
0 ∀Tr ∈ ]−∞;−1 / 3] or p < p n

3 0 1
An = An (Tr + ) ∀Tr ∈ ]−1 / 3;1 / 3[ and p > p n (14)
2 3
∀Tr ∈ [1 / 3; ∞[ and p > p n
0
An
An0 is the maximum value of the nucleation rate parameter. It is assumed to be constant. p n
is the cumulative plastic strain at which nucleation can start. This choice for the expression
of An is dictated by the analysis of Bridgman’s results (Bridgman, 1952) made in (Bao and
Wierzbicki, 2005), which shows that failure does not occur in specimens where stress
triaxiality is lower than -1/3. The above formula provides a continuous transition between
−1 / 3 < Tr < 1 / 3 . Above this value, nucleation is assumed to be triaxiality independent.
SEM observations performed on bent and tensile specimens (Paragraph 3.4) showed that
cavity nucleation starts for a true strain close to 0.11 for the main hard band and close to
0.21 for the softer matrix. This implies that a given strain must be reached for nucleation to
start. In this case, the true strain (ε) was not exactly equal to the cumulative plastic strain
p . A true strain (ε) of respectively 0.11 and 0.21 corresponded to a cumulative plastic
strain at nucleation ( p n ) of respectively about 0.13 and 0.24. In the present model, the
sheet specimen was divided into three parts: (i) the material within the fold, with hard
bands (i.e. mixed ferrite + aligned second phases), with a value of p n = p b = 0.13 , (ii) still
147
within the fold, the “softer matrix” material with a higher value p n = p m = 0.24 and (iii) the
material far from the fold where no damage nucleation occurred. The corresponding
nucleation rate parameters were respectively evaluated to be equal to An = Ab = 2.3 for the
material with hard bands and An = Am = 1.15 for the “softer matrix” material. The value of
Am was mainly adjusted using air bending simulation and that of Ab using stretch bending
simulation.
Elastic properties E = 210GPa ,ν = 0.3

Isotropic hardening R0 = 499 MPa, H = 110 MPa
b = 8.4, Q = 382 MPa
Kinematic hardening C1 = 718MPa, D1 = 9.4
C 2 = 9900MPa, D 2 = 66.7
Plastic anisotropy a = 6.1, α = 0.55

1
cTT = 0.99, c1LL = 0.96, c1SS = 0.92, cTL
1
= 1.14, c1LS = c1ST = 1
2 2 2 2 2 2
cTT = 0.99, cLL = 1.04, cSS = 1.17, cTL = 1.04, cLS = cST =1
Damage q1 = 1.5, q2 = 1, fc = 0.05, δ = 4

p m = 0.24, pb = 0.13
Am = 1.15, Ab = 2.3
Size of finite elements Through-thickness: 50µm

Length: 50µm
Width (3D calculation): 500µm
Table 2: Optimized constitutive equation parameters.
4.3. Finite-element simulation procedure
The in-house finite-element software Zset (Besson and Foerch, 1997) with a fully implicit
integration scheme was used to simulate the two bending tests. All calculations were done
using an updated Lagrangian formulation with linear brick elements using selective
integration. The mesh used for the air-bending test is shown in Figure 11a and the one used
for the stretch bending test modelling is illustrated in Figure 11b. Tools (punch, roll and
die) were assumed to behave elastically with a Young’s modulus and a Poisson ratio equal
to 210 GPa and 0.3. To reduce the number of elements in the stretch bending calculations,
the punch was modelled as a layer of elements to reduce the number of degrees of
freedom. To ensure its stiffness, all displacements were prescribed inside the punch. The
same method was used for the die as well as for the roll in air-bending.
As the model included damage-induced softening, there was a strong dependence of
damage development on the mesh size (Besson et al., 2003). Therefore, it was decided to
fix the mesh size in order to compare the two bending tests in a rigorous manner. Hence,
smallest elements located close to the punch had a size of 50x50µm (corresponding to 32
elements across the thickness).
148
die
punch
roll
sheet
sheet punch
(a) (b)
Figure 11: (a) Meshing of the air-bending test, (b) meshing of the stretch bending test.
The modelling of the air-bending test had been performed using the damage-free model in
(Rèche et al., 2011). Identical numerical conditions were used here, i.e., plane strain
elements were used, only a half of the specimen and punch and one roll were meshed due
to symmetry conditions (Figure 11a). Two contact conditions were considered (Rèche et
al., 2011):
- Between the sheet and the punch, as the friction coefficient was shown to have little
influence on the results, it was fixed to 0.25.
- Between the sheet and the rolls, as rolls rotated freely during the test, friction was
neglected. Here, the rolls were fixed and the friction coefficient was set to 0.
The simulation of stretch bending was carried out in 3D, so only one quarter of the
specimen and punch and a half of one die were meshed due to symmetry conditions
(Figure 11b). The size of the smallest element located just above the punch was
50x50x500 µm (i.e. there were 32 elements across the sheet thickness and 4 elements
across the half-width). The punch and the die had respectively a radius of 4mm and 8mm.
The nodes located just below the die were considered to be fixed to well represent the
clamping of the sheet between the die and the blank-holder. Here, two contact conditions
were considered:
- Between the sheet and the punch, lubrication consisted in grease and a Teflon film
but the actual friction coefficient was not known. It was thus considered as a model
parameter.
- Between the die and the sheet: the friction coefficient between them had no
influence on stretch bending results due to its location far from the fold.
Consequently, it was arbitrarily set to 0.1.
According to cross-sectional observations illustrated in Figure 1, the investigated steel

exhibited at mid-thickness an area with hard bands and for simplification, it was
considered that the region containing the wider bands can be represented within a thickness
of 100µm in the meshing. The three parts of the sheet defined previously are illustrated in
Figure 12.
149
(a) (b) (c)
Figure 12: Closer view of the sheet thickness. Elements coloured in black are (a) material with hard bands
(100µm in thickness), (b) the softer matrix and (c) the material far from the fold.
5. Finite element analysis, results and discussion
5.1. Air-bending
Numerical simulations were performed both with and without damage in the model
(Figure13a). Calculation carried out without damage yielded rather good results but the
load did not decrease enough at the end of the test, i.e. after the onset of cracking. By
including damage in the model, the load decreased quicker and there was very good
agreement between the experimental curve and model predictions. A prestrained sample
was bent until cracking and Figure 13b illustrates both experimental and model predicted
curves. A good agreement was obtained between the two curves except at the end of the
test where the load decreased too early by including damage into the model. This could be
due to the change in strain path: from tension during prestraining to plane strain during
bending. The value of the plastic strain p is increased during prestraining so that damage
starts earlier during subsequent bending thus explaining the reduced ductility of the pre-
strained material. A better agreement with experiment could possibly be obtained by fitting
p m and Am on experimental curves shown in Figure 13b. With the parameters used here,
this model was very efficient to predict air-bending curves without prestraining, while
being a little bit less accurate to predict air-bending curves after prestraining.
3
3
2.5 2.5
2 2
Load (kN)
Load (kN)
1.5 1.5
experiment
1 experiment 1 simulation
simulation without damage
0.5 simulation with damage 0.5
(a) (b)
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Punch displacement (mm) Punch displacement (mm)
Figure 13: Air-bending tests (a) simple air-bending test, (b) air-bending test after 10% tensile prestraining.
The spatial distribution of damage is mapped in Figure 14b for the air-bent specimen
without prestraining. A similar damage map was observed for the prestrained sample.
There was very good agreement between simulation and experiments (Figure 14a) where
similar crack propagation at 45° was observed. Although the central hard band was taken
into account in the model, it had no effect on the global response because cracks initiated
from the outer surface and not from the central hard band. This result is in agreement with
experimental observations.
150
(a) (b) (c)
Figure 12: Closer view of the sheet thickness. Elements coloured in black are (a) material with hard bands
(100µm in thickness), (b) the softer matrix and (c) the material far from the fold.
5. Finite element analysis, results and discussion
5.1. Air-bending
Numerical simulations were performed both with and without damage in the model
(Figure13a). Calculation carried out without damage yielded rather good results but the
load did not decrease enough at the end of the test, i.e. after the onset of cracking. By
including damage in the model, the load decreased quicker and there was very good
agreement between the experimental curve and model predictions. A prestrained sample
was bent until cracking and Figure 13b illustrates both experimental and model predicted
curves. A good agreement was obtained between the two curves except at the end of the
test where the load decreased too early by including damage into the model. This could be
due to the change in strain path: from tension during prestraining to plane strain during
bending. The value of the plastic strain p is increased during prestraining so that damage
starts earlier during subsequent bending thus explaining the reduced ductility of the pre-
strained material. A better agreement with experiment could possibly be obtained by fitting
p m and Am on experimental curves shown in Figure 13b. With the parameters used here,
this model was very efficient to predict air-bending curves without prestraining, while
being a little bit less accurate to predict air-bending curves after prestraining.
3
3
2.5 2.5
2 2
Load (kN)
Load (kN)
1.5 1.5
experiment
1 experiment 1 simulation
0.5 simulation with damage 0.5
(a) (b)
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Punch displacement (mm) Punch displacement (mm)
Figure 13: Air-bending tests (a) simple air-bending test, (b) air-bending test after 10% tensile prestraining.
The spatial distribution of damage is mapped in Figure 14b for the air-bent specimen
without prestraining. A similar damage map was observed for the prestrained sample.
There was very good agreement between simulation and experiments (Figure 14a) where
similar crack propagation at 45° was observed. Although the central hard band was taken
into account in the model, it had no effect on the global response because cracks initiated
from the outer surface and not from the central hard band. This result is in agreement with
experimental observations.
150
0.3
(a) (b)
0
Figure 14: (a) crack on an air-bent specimen, (b) damage map at the end of the air-bending test.
5.2. Stretch bending
As the actual friction coefficient, µ, between the punch and the sheet was not known, the
variation in width was measured along stretch bent specimens to give additional
experimental information about this friction coefficient. Necking observed on samples
corresponded to a mean decrease in width by 5.6% +/- 0.2. The numerical simulation gave
a variation of 5.7% with μ = 0.1 and 5.3% with μ = 0.15. Therefore, for stretch bending
modelling along the transverse direction, the friction coefficient between the sheet and the
punch was set to 0.1.
Figure 15 illustrates the predicted curves obtained both without damage and with the set of
damage parameters given in the previous part. Excellent agreement between experimental
curve and 3D simulation was obtained. As seen in Figure 15, the load decreased very
quickly just after its maximum. This observation confirmed that very unstable failure
occurs in this test. Note that without damage in the model, the load did not decrease
proving the necessity to include a damage criterion to fit well with experiment.
60
experiment
50
simulation with damage
40
Load (kN)
30
20
10
0
0 5 10 15 20 25 30
Figure 15: Experimental and predicted stretch bending load vs. punch displacement curves with and without
damage in the model.
In addition, it was noticed that in this test, the neutral fibre was not at the mid-thickness.
Very quickly, it moved towards the inner skin. In fact, from a punch displacement of 2mm,
the compressive zone shrank toward the inner surface and reduced to a thickness of
0.5mm. For a punch displacement of 10mm, the compressive zone reduced to a thickness
151
of 0.15mm. Thus, the compression stress gradually decreased in that area and annihilated
for a punch displacement of about 15mm. This result confirms that, in stretch bending,
tension prevails over bending.
According to metallographic examinations of stretch bent samples (Figure 16a), cracks

initiated from the central hard band and propagated along a 45° direction with local
necking visible at the outer surface. Localization was clearly noticed within a 45° band
(Figure 16b) and damage was concentrated at the mid-thickness with development along
that band (Figure 16c). These results were in good agreement with experiment
(Figure 16a). The model predicted that damage extended along the hard band at mid-
thickness over a distance of about 2mm. From all metallographic cross-sectional
observations carried out on various stretch bent specimens, damage at mid-thickness was
observed to extend over a distance between 1 and 1.9mm. The thicker the hard band, the
more extended of the damage at mid-thickness. In the present model, the “hard bands”
containing material had a thickness of 100µm but in as-received materials, most of steels
had thinner hard bands at mid-thickness. Therefore, there was good agreement between
model predictions and experimental observations regarding the extent of damage at mid-
thickness.
Necking
0.7
Inactive
plasticity
zone
Active
plasticity zone 0
(a) (b) (c)
Figure 16: (a) cross-sectional observation of an internal crack during stretch bending (b) active and inactive
plasticity zones: localization within a 45° band and (c) damage map taken at mid-width for a punch
displacement of 22mm.
Conclusion
In this paper, analysis of rupture in air-bending and stretch bending tests was performed.
Thanks to an existing comprehensive database, constitutive equations without damage
were already determined for this material. In the present study, in order to predict fracture
more precisely, a model based on the GTN approach has been used introducing (i) mixed
isotropic and kinematic hardening, (ii) plastic anisotropy and (iii) damage evolution
essentially based on the nucleation and growth of voids. Using this model allowed good
predictions of both air and stretch bending behaviours. The key conclusions for this study
are:
1) Fracture modes and damage mechanisms are different for both bending tests. In air-
bending, cracks initiated from the surface or from just below while in stretch
bending tests, cracks inclined by 45° initiated from the main hard band located at
mid-thickness and propagated in an unstable manner leading to sudden fracture.
152
2) The Bron Besson yield criterion, together with mixed plastic hardening and ductile
damage development allowed the load vs. punch displacement curve to be very
well predicted as well as the propagation of a crack from the outer surface in air-
bending.
3) The numerical response of an air-bending test after prestraining along the same
direction also showed good agreement with experiment thanks to the kinematic
contribution to hardening being taken into account in the model.
4) Load vs. punch displacement curves as well as internal crack initiation and
propagation are well predicted for the stretch bending test.
5) Contrary to the air-bending test, a 3D simulation was required in stretch bending to
fit well with experiment. In this case, the friction conditions between the sheet and
the punch must be carefully modelled.
Acknowledgements
The authors would like to acknowledge ArcelorMittal Global R&D for financial support
and material supply. Thanks also are due to Alain Col, Olivier Bouaziz and Xavier
Lemoine for valuable discussions.
References
Bao, Y., Wierzbicki, T., 2005. On the cut-off value of negative triaxiality for fracture,
Engineering Fracture Mechanics 72, 1049-1069.
Bai, Y., Wierzbicki, T., 2010. Application of extended Mohr-Coulomb criterion to ductile
fracture, International Journal of Fracture 161, 1-20.
Besson, J., Foerch, R., 1997. Large scale object-oriented finite element code design,
Computer methods in applied mechanics and engineering 142, 165-187.
Besson, J., Devilliers-Guerville, L., Pineau, A., 2000. Modelling of scatter and size effect
in ductile fracture: Application to thermal embrittlement of duplex stainless steels,
Engineering Fracture Mechanics 67, 169-190.
Besson, J., Steglich, D., Brocks, W., 2003. Modeling of plane strain ductile rupture,
International Journal of Plasticity 19, 1517-1541.
Besson, J., Guillemer-Neel, C., 2003. An extension of the Green and Gurson models to
kinematic hardening, Mechanics of Materials 35, 1-18.
Ben Bettaieb, M., Lemoine, X., Duchêne, L., Habraken, A.M., 2010. Simulation of the
bending process of hardening metallic sheets using damage model. Part I: Theoretical
development and numerical implementation, Materials Science and Engineering A528,
434-441.
Ben Bettaieb, M., Lemoine, X., Duchêne, L., Habraken, A.M., 2010. Simulation of the
bending process of hardening metallic sheets using damage model. Part II: Numerical
investigations, Materials Science and Engineering A528, 442-448.
153
Bridgman, P.W., 1952. Studies in large plastic flow and fracture: With special emphasis on
the effects of hydrostatic pressure, McGraw-Hill.
Bron, F., Besson, J., 2004. A yield function for anisotropy materials. Application to
aluminium alloys, International Journal o f Plasticity 20, 937-963.
Chaboche, J.L., 1989. Constitutive equations for cyclic plasticity and cyclic viscoplasticity,
Chaboche, J.L., 2008. A review of some plasticity and viscoplasticity constitutive theories,
International Journal of Plasticity 24, 1642–1693.
Charpentier, P.L., 1975. Influence of punch curvature on the stretching limits of sheet
steel, Met. Trans. A, vol 6A, 1665-1669.
Chien, W.Y., Pan, J., Tang, S.C., 1981. A combined necking and shear localization
analysis for aluminium sheets under biaxial stretching conditions, International Journal o f
Plasticity 20, 1953-1981.
Chu, C., Needleman, A., 1980. Void nucleation effects in biaxially stretched sheets,
Journal of Engineering Materials and Technology 102, 249-256.
Cockroft, M.G., Latham, D.J., 1968. Ductility and workability of metals, Journal of the
Institute of Metals 96, 33-39.
Dalloz, A., Besson, J., Gourgues-Lorenzon, A.F., Pineau, A., 2009. Effect of shear cutting
on ductility of a dual phase steel, Engineering Fracture Mechanics 76, 1411-1424.
Damborg, D.D., 1998. Prediction of fracture in bending under tension, In: Proceeding of
the 19th Riso International Symposium on Material Science: Modelling of Structure and
Mechanics from Microscale to Product, Roskilde, Denmark, pp. 235-240.
Date, P.P., Narashima, K., Maiti, S.K., Singh, U.P., 1999. On the prediction of springback
in V-bending of metallic sheets under plane strain condition. Proceeding of the
International Conference on Sheet Metal, Earlangen, Germany, pp. 447-456
Demeri, M.Y., 1981. The stretch bend forming of sheet metal, J. Applied Metal Working 2,
3-10.
Eggertsen, P.A., Mattiasson, K., 2009. On the modelling of the bending-unbending

behaviour for accurate springback predictions, International Journal of Mechanical
Sciences 51, 547-563.
Faleskog, J., Gao, X., Shih, C., 1998. Cell model for nonlinear fracture analysis. I:
Micromechanics calibration, International Journal of Fracture 89, 355-373
Friedman, P.A., Luckey, S.G., 2002. Failure of Al-Mg-Si alloys in bending, ASM
International, Practical Failure Analysis 2, 33-42.
154
Gan, W., Wagoner, R.H., 2004. Die design method for sheet springback, International
Journal of Mechanical Sciences 46, 1097-1113.
Goodwin, G.M., 1968. Application of strain analysis to sheet metal forming in the press
shop, SAE paper 680093.
Ghosh, A.K., 1976. A criterion for ductile fracture in sheet under biaxial loading,
Metallurgical Transactions A 7A, 523-533.
Graf, A., Hosford, W., 1993. Effect of changing strain paths on forming limit diagrams of
AL2008-T4, Metallurgical Transactions 24A, 2503-2512.
Gurson, A.L., 1977. Continuum theory of ductile rupture by void nucleation and growth:
part I-yield criteria and flow rules for porous ductile media, Journal of Engineering
Materials and Technology 99, 2-15.
Hudgins, A., Matlock, D., Speer, J., Fekete, J., Walp, M., 2007. The susceptibility to shear
fracture in bending of advanced high strength sheet steels, In: Proceedings from Materials
Science and Technology (MS&T): Advanced High Strength and Other Specialty Sheet
Steel Products for the Automotive Industry, Detroit, Michigan, pp. 145-157.
Kao, A., Kuhn, H., Richemond, O., Spitzig, W., 1990. Tensile fracture and fractographic
analysis of 1045 spheroidized steel under hydrostatic pressure, Journal of Materials
Research 5, 83-91.
Kim, J.H., Sung, J.H., Wagoner, R.H., 2009. Thermo-mechanical modelling of draw-bend
formability tests, International Deep Drawing Research Group, International Conference,
Golden, CO, USA, pp. 503-512.
Keeler, S.P. and Backofen, W.A., 1963. Plastic instability and fracture in sheets stretched
over rigid punches. Transactions of the American Society for Metals 56, 25-48.
Laukonis, J.V., Ghosh, A.K., 1978. Effects of strain paths changes to the formability of
sheet metals, Metallurgical Transactions 9A, 1849-1856.
Lemaitre, J., 1985. A continuum damage mechanics model for ductile fracture, Journal of
Engineering Materials and Technology 107, 83-89.
Lemaitre, J., Chaboche, J.L., 1990. Mechanics of Solid Materials. Cambridge University
Press.
Luo, M., Wierzbicki, T., 2010. Numerical failure analysis of a stretch bending test on dual-
phase sheets using a phenomenological fracture model, International Journal of Solids and
Structures 47, 3084-3102.
LeRoy, G., Embury, J.D., Edwards, G., Ashby, M.F., 1981. A model of ductile fracture
between the nucleation and growth of voids, Acta Metallurgica 29, 1509-1522.
McClintock, F.A., 1968. A criterion of ductile fracture by growth of holes, Journal of

Applied Mechanics 35, 363-371.
155
Perrin, G. and Leblond, J.B., 1990. Analytical study of a hollow sphere made of plastic
porous material and subjected to hydrostatic tension-application to some problems in
ductile fracture of metals, International Journal of Plasticity 6, 677-699.
Rèche, D., Sturel, T., Bouaziz, O., Col., A., Gourgues-Lorenzon, A.F., 2010. Damage
development in low allow TRIP-aided steels during air-bending. Submitted to “Materials
Science and Engineering A”.
Rèche, D., Besson, J., Sturel, T., Lemoine, X., Gourgues-Lorenzon, A.F., 2011. Analysis
of the air-bending test using finite-element simulation: Application to steel sheets.
Submitted to “Journal of Materials Processing Technology”.
Rice, J.R. and Tracey, D.M., 1969. On the ductile enlargement of voids in triaxial stress
fields, Journal of the Mechanics and Physics of Solids 17, 201-217.
Sriram, S., Wong, C., Huang, M., Yan, B., 2003. Stretch bendability of advanced high
strength steels, SAE International World Congress, Detroit, Michigan, pp. 107-115.
Sarkar, J., Kutty, T.R.G., Conlon, K.T., Wilkinson, D.S., Embury, J.D., Lloyd, D.J., 2001.
Tensile and bending properties of AA5754 aluminium alloys, Materials Science and
Engineering A316, 52-59.
Shinohara, Y., Madi, Y., Besson, J., 2010. A combined phenomenological model for the
representation of anisotropic hardening behaviour in high strength steel line pipes,
European Journal of Mechanics / A Solids 29, 917-927.
Steninger, J., Melander, A., 1982. The relation between bendability, tensile properties and
particle structure of low-carbon steel, Scandinavian Journal of Metallurgy 11, no. 2, 55-71.
Thuillier, S., Le Maoût, N., Manach, P.Y., 2010. Bending limit prediction of an aluminium
thin sheet, International Journal of Material Forming 3, 223-226.
Tvergaard, V. and Needleman, A., 1984. Analysis of the cup-cone fracture in a round
tensile bar. Acta Metallurgica 32, 157-169.
Uthaisangsuk, V., Prahl, U., Münstermann, S., Bleck, W., 2008, Experimental and
numerical failure criterion for formability prediction in sheet metal forming,
Computational Materials Science 43, 43-50.
Vladimirov, I.N., Pietryga, M.P., Reese S., 2010. Anisotropic finite elastoplasticity with
nonlinear kinematic and isotropic hardening and application to sheet metal forming,
Williams, B.W., Simha, C.H.M., Abedrabbo, N., Mayer, R., Worswick, M.J., 2010. Effect
of anisotropy, kinematic hardening and strain-rate sensitivity on the predicted axial crush
response of hydroformed aluminium alloy tubes, International Journal of Impact
Engineering 37, 652-661.
156
II. Influence of simulation conditions on the stretch-

bending test
II.A. Influence of meshing conditions
In order to determine the influence of the mesh, four calculations in the same conditions as
for the air-bending test (Chapter 2) were performed, i.e. with various numbers of elements
across the thickness close to the punch, i.e. over 3mm (this value was set regarding the
stress and strain fields). As the GTN model is sensitive to the mesh size, all calculations
were performed without damage in the model and we chose to keep the same length
element size, i.e. 50µm. The four cases studied were:
(i) 8 elements across the sheet thickness (i.e. 200µm thick elements)
(ii) 16 elements across the sheet thickness (i.e. 100µm thick elements)
(iii) 32 elements across the sheet thickness (i.e. 50µm thick elements)
(iv) 64 elements across the sheet thickness (i.e. 25µm thick elements)
In agreement with section I, this parametric study was carried out in 3D, with the Bron
Besson yield criterion, mixed isotropic/kinematic hardening rules but no damage was
included in the model. These calculations were carried out with a friction coefficient of 0.1
between the punch and the sheet. The corresponding meshes are illustrated in Figure IV.1.
Thickness
TD
Figure IV.1. Four meshes with various numbers of elements across the thickness. Total thickness = 1.6mm.
Figure IV.2a illustrates load vs. punch displacement curves for the four meshing
conditions. Similar curves are obtained for the four simulations, meaning that the number
157
of elements across the thickness does not change the global response in the range
investigated here. Hoop stress and hoop strain distributions across the thickness at mid-
width are illustrated in Figure IV.2b. Here also, similar curves are obtained for the four
meshing conditions. Stress and strain fields calculated using each of the four meshes across
the sheet thickness are respectively given in Figure IV.3 and Figure IV.4.
60 1600 0.35
50 experiment
1400 0.3
8 elements 1200
hoop stress (MPa)

0.25
40 16 elements stress
Hoop strain
1000
Load (kN)
32 elements 0.2
30 64 elements 800
0.15
600 8 elem.
20
16 elem. strain 0.1
400 32 elem.
10 64 elem. 0.05
200 S
0 0 0
0 5 10 15 20 25 0 0.4 0.8 1.2 1.6
(a) (b)
Displacement (mm) Distance from outer surface (mm)
Figure IV.2. (a) Influence of the mesh on the load / displacement curves. Calculations with the Bron Besson
yield criterion, mixed isotropic/kinematic hardening rule and no damage. (b) Influence of the mesh on the
hoop stress and hoop strain distributions across the thickness for a punch displacement of 22mm at mid-
width.
16 elements
8 elements
x2
32 elements
64 elements x1
x3
σ11
Figure IV.3. σ11 field (in MPa) after stretch bending up to a punch displacement of 22mm (8 to 64 elements
across the thickness) at mid-width.
158
8elements 16 elements
x2
32 elements
64 elements
x1
x3
ε11
Figure IV.4. ε11 field after stretch bending up to a punch displacement of 22mm (8 to 64 elements across the
thickness) at mid-width.
Regarding stress and strain fields as a function of the local meshing size, similar maps are
obtained for the four calculations. However, as damage criterion has to be included into
constitutive equations, it is necessary to use a quite thin mesh. With 64 elements across the
thickness, calculations lasted a very long time. At the end, the mesh size is determined by
the best compromise between computation time and accuracy of damage prediction.
Therefore, all calculations were performed with 32 elements across the thickness i.e. for a
mesh size of 50 µm along the specimen thickness.
II.B. Influence of boundary conditions
1. Abnormal sensitivity to the friction coefficient in 2D
Stretch bending test simulations were first carried out in 2D with a plane strain state
hypothesis, as for the air-bending test. As the actual friction coefficient between the sheet
and the punch was not known, a parametric study was conducted. Several calculations
were launched with various values of the friction coefficient (µ = 0.2, 0.3, 0.4 and 0.5).
For these calculations, the Bron Besson yield criterion associated with mixed
isotropic/kinematic hardening was used as in section I but no damage was included in the
constitutive equations to only investigate the influence of the friction coefficient on the
flow curve.
159
A surprising result was obtained (Figure IV.5). The slope of the load vs. punch
displacement curves is similar for all calculations but the load drops at various values of
the punch displacement corresponding to the various values of the friction coefficient. In
fact, the higher the friction coefficient, the later the load drops. Necking in the thickness
was observed earlier with a lower friction coefficient (Figure IV.6). To be in accordance
with experiments, a very high friction coefficient should be used, i.e. close to 0.4.
However, this value of the friction coefficient is not realistic for this kind of test, and the
load does not drop as quickly as observed experimentally. This discrepancy could stem
from two phenomena:
(i) only strain localisation (and no damage) was involved in this model
(ii) the strong influence of the friction coefficient influences the ability for specimen to
shrink i.e. to be strained along the thickness direction.
Figure IV.6 illustrates ε11 field for two values of the friction coefficient (µ = 0.2 and µ =
0.4). Premature necking of the sheet thickness is clearly observed especially for the lowest
value of the friction coefficient, explaining the early drop of the load. These results
explicitly show that the 2D calculations are not correct (besides this abnormal influence of
the friction coefficient, stress and strain fields are not correct and the sheet does not shrink
as much experimentally). It is therefore necessary to perform all stretch bending
calculations in three dimensions to accurately predict the load vs. punch displacement
curves, stress and strain fields and the location of fracture.
60
50 experiment
2D: µ = 0.2
40 2D: µ = 0.3
2D: µ = 0.4
Load (kN)
2D: µ = 0.5
30
20
10
0
0 5 10 15 20 25 30
Displacement (mm)
Figure IV.5. Load /displacement curves in stretch bending. Influence of the friction coefficient in 2D.
x2 1.2
x1
x3
(a) (b) 0
Figure IV.6. ε11 field after stretch bending in 2D. Abnormal necking along the sheet thickness due to the
friction coefficient (a) µ = 0.2, (b) µ = 0.4.
160
2. Influence of the friction coefficient in 3D
Here it is checked that the abnormal effect reported in the previous section was actually an
artefact due to the plane strain assumption. As the actual friction coefficient of this sheet
with the bending tool was not known, various “reasonable” values of the friction
coefficient were tested: μ = 0.05, 0.1 and 0.15. In this case, the constitutive equations
included damage with parameters defined in section I. Corresponding load vs. punch
displacement curves are shown in Figure IV.7. The curves have similar slopes but different
load drops. As in 2D modelling, the lower the friction coefficient, the earlier the load
decreases. Here, the best fit was obtained with μ = 0.1, which is a reasonable value for the
lubrication conditions used experimentally. These curves confirm that the predicted load
drop is due to the introduction of damage in the model. In 2D, the load drop was a
calculation artefact due to abnormal necking along the sheet thickness.
60
50 experiment
µ = 0.05
40 µ = 0.1
Load (kN)
µ = 0.15
30
20
10
0
0 5 10 15 20 25
Figure IV.7. Load / displacement curves for various friction coefficients (μ = 0.05, μ = 0.1 and μ = 0.15) in
3D simulation with damage in the model.
The value of µ was further checked by measuring the variation in width along stretch bent
specimens (Figure IV.8). Necking observed along the specimen width corresponded to a
mean decrease in width by 5.6% +/- 0.2. The numerical simulation yielded a variation in
width by 5.7% with μ = 0.1 and by 5.3% with μ = 0.15. Thus, the value of μ = 0.1 leads to
global response and local strain that are consistent with experimental measurements,
therefore, this value was chosen for all stretch bending test simulations (see previous part).
Note that this observation was in agreement with the study of Luo and Wierzbicki (2010).
In fact, in their work, they showed that localized width necking was noticed in the vicinity
of the punch nose, that could not be simulated by a 2D calculation with a plane strain
assumption.
width variation width
(a) (b)
Figure IV.8. (a) Part of the stretch bent specimen geometry predicted by the model, with necking along the
width. (b) real stretch bent specimen with necking along the width.
161
II.C. Summary: influence of modelling conditions on stretch-bending test

predictions
In this part, the influence of the stretch-bending modelling conditions was discussed. It was
shown that until damage begins to develop, the 3D mesh size has an influence neither on
the global load vs. displacement curve nor on the hoop stress and hoop strain distributions
across the sheet thickness. In order to be able to simulate a realistic thickness of the
damaged hard band, it was chosen to perform all stretch-bending simulations with 32
elements across the thickness. Then, it was demonstrated that 3D element model provided
much better agreement with experiments than with a plane strain assumption. The
influence of the friction coefficient was also considered. It was demonstrated that a friction
coefficient of 0.1 should be used in this case between the sheet and the punch. A summary
of all these results is given in Table 9.
2D 3D
Influence of No influence on hoop stress and
meshing hoop strain distributions and on
→ 8 elements the load/displacement curve
→ 16 elements
→ 32 elements / → 32 elements across the
→ 64 elements thickness has been chosen to get
across the thickness the best compromise between
before damage time computation and accuracy
develops of damage prediction
Abnormal results: a very high
friction coefficient (µ=0.4) has
The best fit friction coefficient
to be used to fit reasonably with
in 3D is µ = 0.1. This value was
Friction coefficient experiment.
adjusted using the experimental
The load drop is a function of
variation in width along the bent
the necking along the sheet
specimen
thickness depending of the
friction coefficient.
Plane strain elements (2D calculations) induce artefacts and are
not adapted due to unrealistic premature necking of the specimen
Type of elements
along the thickness direction.
A three-dimensional calculation is required in stretch bending
Table 9. Summary of the influence of stretch-bending modelling conditions in 2D and 3D.
III. Parametric study: effect of the location of the hard

band on failure during air-bending
In section I, modelling was performed by assuming a hard band at mid-thickness which
had the highest sensitivity to damage. In chapter 3, we demonstrated that a hard band close
to the outer surface had a deleterious effect on the bendability because cracking appeared
earlier with the proximity of the band with the outer surface. Experiments investigating the
influence of the location of the band were carried out on thickness-reduced Steel 2 and the
162
simulation of air-bending of this material was performed in Chapter 2. A good agreement

was observed at the beginning of the test (i.e. good prediction of the load / displacement
curve) but not at the end, where the load was underestimated by the model. Therefore, it
seemed inappropriate to model the air-bending test on thickness-reduced Steel 2 with
various locations of the band because the model should first be improved.
One aim of the study was to quantitatively investigate the influence of banded structures
(see Figure IV.9) on bendability. In this case, the banded structure is assimilated to a
continuous hard band as for the central segregation. Therefore, air-bending simulations
with various locations of hard band across the thickness were carried out on Steel 1
(classical thickness for automotive products, i.e. 1.6mm) because calculations were much
simpler than for thickness-reduced Steel 2, leading to more reliable predictions. A row of
elements with the highest sensitivity to damage (representing a banded structure) was
located at various distances from the mid-thickness towards the outer surface. Each set has
a thickness of 50µm corresponding to the thickness of one element.
Sheet thickness
Figure IV.9. Schematics of a cross-section along the sheet thickness with the central segregation as a thick
black line at mid-thickness and banded structures across the thickness as thin grey lines.
Several cases were investigated:

(i) at 150 µm from the outer surface
(ii) at 200 µm from the outer surface
(iii) at 250 µm from the outer surface
(iv) at 300 µm from the outer surface
(v) at 350 µm from the outer surface
(vi) at 400 µm from the outer surface
Note that the distances between the outer surface and the band were measured in the
undeformed state. After air-bending, as the thickness of the sheet had decreased, all the
bands were located at a smaller distance from the outer surface.
As in section I, the Bron Besson yield criterion, a mixed isotropic / kinematic hardening
rule and damage were included in the model for all these calculations. The same damage
parameters as previously were kept both for the hard band and for the softer matrix.
Experimentally, in chapter 3, it was noticed that sub-surface cracks could appear with a
hard band located at 50µm from the outer surface after air-bending of thickness-reduced
Steel 2 (Figure IV.10).
Regarding damage maps given by the various calculations (Figure IV.11, air-bending test
of Steel 1), it is difficult to affirm whether cracks initiate from the surface or from just
below. However, it seems reasonable to say that as soon as bands are located as far as
250µm from the outer surface, cracks initiate from the surface even if hard bands are also
damaged. For the band located at 200µm from the surface, it is not possible to exactly
determine the exact initiation of the crack. For a band located between the outer surface
163
and 150µm below it, it seems that a sub-surface crack is noticed. Regarding this result, it
seems that sub-surface cracks can appear with bands farther from the outer surface than for
thickness-reduced Steel 2 (Figure IV.10) for which the critical distance was evaluated to
50µm. This difference can be due to various factors: (i) the sheet thickness (0.75mm for
thickness-reduced Steel 2 and 1.6mm for Steel 1), (ii) the hard band thickness (a few
microns for thickness reduced Steel 2 and 50µm for the modelling of Steel 1), etc.
50 µm
Figure IV.10. Sub-surface crack at 50µm from the outer surface after air-bending of thickness-reduced
Steel 2.
400µm from the 350µm from the

outer surface outer surface

164

Figure IV.11. Damage map for six initial locations of the band across the thickness. Damage initiates
internally as soon as the hard band is initially at 150µm or less from the outer surface.
The corresponding load / displacement curves are given in Figure IV.12. Note that the load
decreases earlier with the closer proximity of the band to the outer surface. This is in
agreement with the fact that cracks appeared earlier with bands located closer to the
surface.
3.5
2.5
Load (kN)
2
experiment
1.5 band at mid-thickness
band at 400µm from surface
1 band at 300µm from surface
0.5 band at 200µm from surface
0
0 5 10 15 20 25 30
Figure IV.12. Load / displacement curves for various locations of the band across the thickness.
As cracks initiate from the central hard band in stretch-bending tests, it was not necessary
to perform these same calculations with various locations of hard band for the stretch
bending test. It is expected that by shifting the hard band close to the outer surface in
stretch bending, cracks would also initiate from this band and fracture would appear
earlier.
165

In this chapter, the mechanical analysis of both air and stretch bending tests was conducted
by finite-element calculations taking damage into account. Damage-free constitutive
equations were first determined in chapter 2 thanks to an experimental database based on
tensile tests on smooth and notched specimens and to shear tests. In this chapter, the
“reference” constitutive equations included mixed isotropic/kinematic hardening, a Bron
Besson yield criterion and ductile damage development.
Using such a model made it possible to predict air bending as well as stretch bending tests
very accurately both as load vs. punch displacement curves and as fracture initiation
location. To represent the microstructure heterogeneity of the material along the sheet
thickness, a band with a higher sensitivity to damage was included in the model at mid-
thickness of the sheet. Using this meshing enabled to obtain a kind of “composite” model
of the sheet representative of the as-received material. This model is thus able to represent
the two bending tests and to help to predict the beginning of fracture.
In air-bending, cracks initiated from the outer surface when the hard band was located at
mid-thickness. Additional air-bending calculations carried out on Steel 1 to investigate the
influence of band location across the thickness demonstrated that cracks can initiate from
just below the surface when hard bands were initially located at 150µm or less from the
outer surface. In stretch bending, cracks initiated from the main hard band located at mid-
thickness and this was also well predicted by the model.
Regarding the stretch bending test, the 3D mesh size has no influence both of the load vs.
displacement curve and on the hoop stress and strain distributions through the thickness.
As a “composite” material had to be used to investigate the influence of hard band on both
bending tests, a thin mesh was required. Therefore, a mesh with 32 elements across the
thickness was chosen to get a good compromise between good accuracy and a limited
number of elements.
3D calculations are required to simulate stretch bending test while 2D modelling with
plane strain elements was satisfactory for air-bending. This was explained by the fact that
necking along the sample width occurs during the stretch bending test. The friction
coefficient required for the modelling of the stretch bending test was adjusted by width
reduction measurements on stretch bent specimens and was set to 0.1 between the punch
and the sheet.
166
Dans ce chapitre, une analyse mécanique à la fois du pliage en V et du pliage sous traction
a été effectuée par la méthode des éléments finis en prenant en compte l’endommagement
dans le modèle. La loi de comportement sans endommagement a été déterminée dans le
chapitre 2 grâce à une base de données expérimentale incluant des tests de traction et de
cisaillement. Dans cette partie, le nouveau modèle de “référence” inclut un écrouissage
mixte, un critère d’écoulement anisotrope Bron Besson ainsi qu’une prise en compte de
l’endommagement.
Grâce à cette loi de comportement, nous avons été capables de prédire aussi bien les
courbes force/déplacement que la localisation de la rupture à la fois pour le test de pliage
en V et pour le pliage sous traction. Pour représenter l’hétérogénéité de la microstructure
à travers l’épaisseur dans le matériau, une bande possédant une plus grande sensibilité à
l’endommagement a été incluse à mi-épaisseur dans le modèle. Ceci a permis d’obtenir un
matériau “composite” modèle, représentatif du matériau de l’étude. Ce modèle est donc
capable de représenter deux essais de pliage différents et de prédire l’apparition de la
rupture.
En pliage en V, les fissures s’amorcent en surface quand les bandes de ségrégation sont
situées à mi-épaisseur. Des calculs supplémentaires réalisés sur l’acier 1 pour déterminer
l’influence de la position des bandes ont permis de montrer que les fissures peuvent
s’initier sous la surface lorsque des bandes de ségrégation se trouvent à moins de 150µm
de la surface de l’acier. En pliage sous traction, les fissures s’amorcent à partir de la
ségrégation centrale et ce résultat a été correctement prédit par le modèle.
En pliage sous traction, le maillage 3D n’a aucune influence sur les courbes
force/déplacement ou sur la distribution des déformations et des contraintes à travers
l’épaisseur. Comme nous souhaitions modéliser un matériau “composite” pour déterminer
l’influence des bandes de ségrégation (structures en bandes, ségrégation centrale) sur les
deux tests de pliage, un maillage fin était nécessaire. Dans le but de garder un bon
compromis entre un nombre limité d’éléments et une bonne précision, un maillage avec 32
éléments dans l’épaisseur a été choisi.
Un calcul 3D est nécessaire pour modéliser le test de pliage sous traction alors qu’une
simulation numérique en 2D avec l’hypothèse des déformations planes est suffisante pour
simuler le test de pliage en V. Ceci provient du fait qu’une striction dans la largeur
apparait pendant le pliage sous traction. Le coefficient de frottement requis pour la
modélisation du pliage sous traction a été ajusté grâce aux mesures de réduction de
largeur sur les éprouvettes de pliage sous traction et a été évalué à 0.1 entre la tôle et le
poinçon.
167
General conclusions and proposals for future work
General conclusions and proposals for

future work
169
In order to reduce both vehicle weight and fuel consumption, steelworkers need to develop
and propose new steel grades such as Ultra High Strength Steels (UHSS). Among them,
TRIP-aided steels, composed of hard retained austenite/martensite and bainite islands
dispersed in a soft ferritic matrix, exhibit a very good compromise between strength and
formability. However, as TRIP steels are used to perform very complex parts, formability
issues can be encountered on these materials. On new TRIP steel grades with increasing
mechanical properties, cracking can occur on radii during part stamping and it is difficult
to directly connect steel formability with standard mechanical properties. It is therefore
necessary to take the complex microstructures of these steels into account to accurately
investigate their bendability.
This study aimed at understanding the relationships between microstructure and

bendability of Ultra High Strength Steels for automotive applications. The approach used
during this Ph.D. project was based on coupling experimental and numerical analyses.
Four TRIP-aided steels, exhibiting both various bending performances and
microstructures, were investigated through air and stretch bending tests.
Main results
The first chapter of this manuscript was devoted to a literature survey describing
bendability, TRIP-aided steels and the origin of micro and macrosegregations. It was
shown that several parameters such as mechanical properties, microstructure, tools,
etc…can influence steel bendability, making it difficult to understand formability issues.
Regarding simulation of bending tests, various models have already been used, including
(or not) an anisotropic yield criterion as well as a purely isotropic or mixed hardening rule.
However, no detailed mechanical analysis of what happens through the specimen thickness
was performed except with simplified models.
The aim of the present study was therefore to develop an experimental and numerical
approach to quantitatively assess the effect of microstructural inhomogeneities on the
bendability of ultrahigh strength steels. To introduce the study, this first chapter also
provided both microstructural and general mechanical characterization (including standard
bending tests) of the four selected TRIP-aided steels.
Among the four materials studied here, Steel 1 and Steel 2 exhibited the lowest bendability
compared to the two other ones and they were selected for the remaining of the study. This
result was in agreement with their tensile strength but no clear relationship with their
microstructures could be determined. From these preliminary tests, it was thus not possible
to determine the influence of microstructural inhomogeneities such as central segregation
and banded structures on both air and stretch bending tests.
In chapter 2, an experimental database, including tensile smooth and notched specimens, as

well as reverse shear tests was established. All these results made it possible to determine
constitutive equations of Steel 1 in order to model the two bending tests. These constitutive
equations included the anisotropic Bron Besson yield criterion associated with a mixed
isotropic/kinematic hardening rule. They were included in a finite element model. The
possibility to use “simplified models”, by assuming an isotropic von Mises yield criterion
and/or a purely isotropic hardening rule was also discussed. Using a von Mises yield
criterion, the experimental database was satisfactorily modelled even if the load was
systematically always overestimated. Using purely isotropic hardening, reverse shear tests
were not well predicted, as the Bauschinger effect was not taken into account.
170
Regarding the air-bending test, better predictions were obtained using a Bron Besson yield
criterion and a mixed isotropic/kinematic hardening rule than with simplified models. In
addition, when prestraining is carried out prior to air-bending, kinematic hardening is
required to obtain a good prediction of the load. Moreover, the reduction in thickness
during air-bending led to a shift of the neutral fibre toward the inner radius of the fold.
Indeed, a change in the strain path in the elements located in the vicinity of the mid-
thickness was observed as these nodes underwent initial compression followed by tension.
To investigate the effect of the central hard band on bendability, a specific air-bending
facility was set up to test specimens of Steel 2 that were reduced in thickness, to locate this
band at various distances from the outer radius of the fold. Model predictions of air-
bending of thickness-reduced Steel 2 were in very good agreement at the beginning of the
test but not at its end, possibly due to contact and/or very high strain issues. In addition,
although higher strains were reached after 180° bending of thickness-reduced Steel 2 than
with Steel 1, the former specimens did not crack with the central segregation at mid-
thickness. Damage developed inside the specimens but did not lead to microcracking at the
outer surface.
The third chapter dealt with the investigation of damage mechanisms in three tests:
interrupted tensile tests, air and stretch bending tests. Metallographic observations of
damage were performed using light and scanning electron microscopy. Ductile (dimple)
fracture modes were found in both air and stretch bending tests. Cracks initiated from the
outer surface or from just below in air-bending while cracks initiated from the central
segregation in stretch bending; in both cases, cracks propagated along a band at 45° from
the loading direction. Damage mainly nucleated by interfacial decohesion between
martensite/austenite (M/A) islands and softer ferrite and occasionally by cleavage of M/A
islands. Using tensile tests interrupted at various strain levels, the fracture strain of hard
bands was determined to be about 0.10 for the studied TRIP-aided steels. Thicker hard
bands promoted earlier damage nucleation than thinner ones and nucleation of new cavities
further occurred all along the tests.
Using thickness-reduced specimens of Steel 2 showed that the closer the hard band to the
outer surface, the higher the severity of cracking in air-bending. In fact, sub-surface cracks
initiated from these hard bands. A quantitative relationship between damage, band
thickness and local “mesoscopic” strain was eventually derived.
In chapter 4, a comparison between the two bending tests was presented. The constitutive
equations established in chapter 2 were used and associated with an extension of the
Gurson-Tvergaard-Needleman (GTN) model to well predict the fracture in both air and
stretch bending tests. Scanning electron microscopy observations coupled with numerical
results made it possible to determine the damage parameters for both “hard bands” material
and “softer matrix” material. Using a “composite” representation of the sheet specimens,
modelling of the two bending tests was carried out by assuming a hard band at mid-
thickness exhibiting higher sensitivity to damage development. Using such a model
allowed quantitative prediction of the load/displacement curves as well as of fracture
location for two different tests: air and stretch bending. In addition, because of necking
along the specimen width, 3D modelling was required for stretch bending simulations
while 2D calculations with a plane strain assumption were shown to be accurate enough in
the case of air-bending.
For the low alloy TRIP-aided steels studied here, hard bands do not have the same
influence in both air and stretch bending tests. Regarding air-bending, banded structures
171
have a potentially deleterious effect on bendability while the central segregation does not
lead to earlier cracking. Nevertheless, in one case, damage was observed at the central
segregation after air-bending. In that particular case, the central segregation band was
particularly thick but was still not responsible for microcracking at the outer surface. It was
demonstrated that sub-surface cracks can occur when banded structures are located in the
vicinity of the outer surface. Under the conditions described by the parametric numerical
study carried out at the end of chapter 4 (air-bending with various locations of the hard
band through the thickness for Steel 1), this critical distance was evaluated to about
150μm. It is worth noting that the potentially deleterious influence of a banded structure
will depend on a combination of several parameters such as its ability to deform without
damage, which will be a function of its average thickness, its location (distance from the
outer surface of the fold), the strain generated during bending and probably on other
parameters such as its hardness and so on. The higher the thickness and the amount of
strain, the lower this critical distance will be.
The most deleterious metallurgical feature in stretch bending tests is the central segregation
band, as fracture clearly initiates from this band. Therefore, when dealing with complex
forming paths involving a tensile contribution to loading, it may be important to control
and even to optimize the microstructure of those steels.
As shown in the first chapter, different steel processing parameters have to be used to act
on banded structures and central segregation, respectively. Usually, a low coiling
temperature is recommended to decrease the severity of the banded structure. In addition,
as banded structures too close to the outer surface are deleterious in air-bending, it will be
better to perform high superheat in the steelmaking process to concentrate bands towards
the mid-thickness. This has to be balanced, however, with stretch bending properties,
which strongly depend on the microstructure at mid-thickness. Regarding the central
segregation, one actuator to reduce it, is to use the soft reduction process.
Proposals for future work
To go further in the understanding of the link between microstructure and bendability, it

would be necessary to apply the methods developed in this study to other steel families
such as Dual Phase (DP) and Multiphase (MP) steels. In these materials, other parameters
such as inclusions, etc. can possibly play an important role on the bendability.
In this Ph.D. project, we only focused on two forming modes: air-bending and stretch
bending. It would be interesting to determine the influence of microstructure in other
formability tests such as stamping or expansion. For example, during stamping, some parts
of the specimen are in tension, and therefore, central segregation band could also have a
significant impact on formability.
Outputs of this work are mainly methodological and partly qualitative regarding
microstructural data. Regarding the air-bending test, banded structures are the most
important parameters to be controlled in order to improve the steel bendability.
It will be interesting to perform complementary experiments linking the band thickness,
the amount of strain reached during bending and the sensitivity to damage in air-bending,
to more precisely determine a critical thickness for banded structures and a minimum
distance from the sheet surface to avoid premature failure in bending. It would also be
important to include stress triaxiality effects in the phenomenological relationship
172
established here between the critical band thickness and the critical amount of strain, to
more accurately apply this equation to other forming cases.
Finally, in order to improve steel bendability, heat treatments could be used to homogenize
sheet microstructure and obtain “model” materials.
In this study, the damage initiation criterion was set at a macro or mesoscopic scale. A
more detailed analysis of the mechanical behaviour of each phase could be helpful to
determine the stress at the interface between harder and softer phases. The nanohardness of
each phase was given and it is now essential to provide information about the shape and
size of individual phases to perform a comprehensive micromechanical study of ductile
damage void nucleation during bending. In addition, to get more information about damage
mechanisms, it would be interesting to use tomography experiments. With all this data, it
would then be possible to use a micromechanical model to more accurately determine a
microstructure-based ductile damage initiation criterion.
The finite element method proved to be a good tool to investigate the bendability of a sheet
material. With accurate constitutive equations and by reliably taking damage into account,
the model can predict ductile fracture during bending. In view of the present results, it
seems necessary to continue the parametric study of the air-bending test with various
locations of the hard band across the specimen thickness. The present study was carried out
with a “numerical” band exhibiting a thickness of 50μm. In the four steels under study, the
central hard band was always thinner and the region containing such hard bands could only
occasionally reach this thickness. To use more realistic assumptions, it would be necessary
to use finer meshes and to again perform calculations with various locations of the band
along the specimen thickness. It would be also interesting to investigate interactions
between two hard bands (in particular, the effect of the distance between them) with
respect to damage initiation.
Similar parametric studies could also be performed in stretch bending to determine a
critical thickness that the central segregation should not exceed to avoid premature failure.
By using such a damage initiation map together with predicted load/displacement curves, it
will thus be possible to assess the bendability of a given steel sheet material.
173
Dans le but de réduire à la fois le poids des véhicules et leur consommation, les aciéristes
doivent développer et proposer des nouveaux grades d'aciers comme les aciers Très Haute
Résistance (THR). Parmi eux, les aciers TRIP, composés d'ilots durs d'austénite résiduelle,
de martensite et de bainite dans une matrice ductile ferritique, présentent un bon
compromis entre résistance et mise en forme. Cependant, comme les aciers TRIP sont
utilisés pour réaliser des pièces complexes, ils peuvent présenter des problèmes de mise en
forme. Sur les nouveaux grades d'aciers TRIP avec des propriétés mécaniques de plus en
plus élevées, des fissures peuvent apparaître sur les rayons pendant l'emboutissage et il est
difficile de directement relier la capacité de mise en forme d'un acier avec ses propriétés
mécaniques classiques. Il est donc nécessaire de prendre en compte la microstructure
complexe de ces aciers pour prédire d'une manière précise leur capacité à se plier.
Le but de cette étude était de comprendre les relations entre microstructure et capacité en
pliage des aciers THR pour application automobile. L'approche qui a été utilisée pendant
ce projet de thèse était basée à la fois sur des essais expérimentaux mais également sur
une analyse de simulation numérique. Quatre aciers TRIP possédant des capacités en
pliage différentes et des microstructures variées ont été étudiés à la fois en pliage en V et
en pliage sous traction.
Principaux résultats
Le premier chapitre de ce manuscrit a été consacré à une étude bibliographique décrivant

le pliage, les aciers TRIP et l'origine des micro et macroségrégations. Il a été montré que
divers paramètres tels que les propriétés mécaniques, la microstructure, les outils, etc.
peuvent avoir une influence sur la capacité en pliage, rendant difficile la compréhension
des problèmes de mise en forme.
D'après la simulation numérique du test de pliage en V, différents modèles ont été utilisés,
incluant (ou non) un critère d'écoulement anisotrope, un écrouissage purement isotrope ou
mixte. Cependant, aucune étude détaillée de ce qui se produit dans l'épaisseur n'a été
effectuée, mis à part sur des modèles simples.
Le but de cette étude était donc de développer une approche à la fois expérimentale et
numérique pour évaluer l'effet d'une microstructure inhomogène sur la capacité en pliage
des aciers THR. Pour introduire cette étude, une caractérisation à la fois mécanique et
microstructurale a été effectuée sur les quatre aciers TRIP dans le premier chapitre.
Parmi les quatre matériaux étudiés ici, l'acier 1 et l'acier 2 présentaient une capacité en
pliage faible comparé aux deux autres aciers et ils ont été choisis pour la suite de l'étude.
Ce résultat était en accord avec leur résistance en traction mais aucune relation claire
avec leurs microstructures n'a pu être déterminée. Ainsi, à partir de ces premiers essais, il
n'a pas été possible de déterminer l'influence des inhomogénéités microstructurales comme
la ségrégation centrale et les structures en bandes tant sur le test de pliage en V que sur le
test de pliage sous traction.
Dans le chapitre 2, une base de données expérimentale, incluant des tests de traction sur
éprouvettes lisses et entaillées, ainsi que des tests de cisaillement a été établie. Tous ces
résultats ont permis de déterminer la loi de comportement de l'acier 1, destinée à
modéliser les deux tests de pliage. Ce modèle inclut un critère d'écoulement anisotrope de
Bron Besson ainsi qu'un écrouissage mixte isotrope/cinématique. Cette loi a été incluse
dans un modèle de simulation numérique par éléments finis. La possibilité d'utiliser des
174
modèles simplifiés, tels qu'un critère d'écoulement isotrope de von Mises et/ou un
écrouissage purement isotrope a aussi été envisagée. Avec un critère d'écoulement de von
Mises, la base de données expérimentale a été prédite de manière satisfaisante même si les
efforts étaient toujours surestimés. Avec un écrouissage purement isotrope, les essais de
cisaillement réversibles n'ont pas été bien prédits, car l'effet Bauschinger n'était pas été
pris en compte.
Quant aux essais de pliage en V, de meilleures prédictions ont été obtenues en utilisant un
critère d'écoulement de Bron Besson et un écrouissage mixte isotrope/cinématique qu'avec
des modèles simplifiés. De plus, quand une prédéformation est effectuée avant un pliage en
V, un écrouissage cinématique est nécessaire pour prédire de manière satisfaisante les
courbes force/déplacement. La réduction de l'épaisseur apparaissant pendant le pliage en
V a mené à un déplacement de la fibre neutre vers le rayon intérieur du pli. En effet, les
éléments situés proches de la mi-épaisseur ont subi un changement de trajet (compression
initiale suivi par une tension).
Pour déterminer l'effet d'une bande de ségrégation sur le pliage, un essai spécifique du
pliage en V a été utilisé sur l'acier 2 rectifié en épaisseur, sur lequel cette bande dure a été
placée à différentes positions par rapport à la surface externe du pli. Les prédictions
données par le modèle du pliage en V de l'acier 2 rectifié étaient en accord au début du
test, mais pas à la fin. Ceci pourrait être dû à des problèmes de contact et/ou de
déformations atteintes très importante. De plus, bien que des déformations plus élevées
aient été atteintes après un pliage de l'acier 2 rectifié à 180° qu'avec l'acier 1, ces
échantillons n'ont pas fissuré avec la ségrégation centrale à mi-épaisseur.
L'endommagement s'est développé dans l'épaisseur mais n'a pas entraîné une
microfissuration sur la surface extérieure.
Le troisième chapitre a traité des mécanismes d'endommagement au cours de trois essais :

des tests de traction interrompus, du pliage en V et du pliage sous traction. Des
observations métallographiques de l'endommagement ont été effectuées en utilisant la
microcopie optique et celle à balayage. Les ruptures sont ductiles (avec cupules) aussi
bien en pliage en V qu'en pliage sous traction. Les fissures s'amorcent à partir de la
surface extérieure ou juste en-dessous en pliage en V alors qu'elles s'initient à partir de la
ségrégation centrale en pliage sous traction; dans les deux cas, les fissures se
propageaient à 45°. L'endommagement apparaît principalement par décohésion entre les
îlots martensite/austénite et la ferrite et de temps en temps par clivage des îlots M/A. En
utilisant des tests de traction interrompus à différents niveaux de déformation, la
déformation à rupture des bandes de ségrégation a été évaluée à environ 0.10 pour les
aciers TRIP étudiés. Les bandes de ségrégation les plus épaisses ont été endommagées
plus rapidement que les plus minces et la germination de nouvelles cavités est apparue
durant tout l'essai.
L'utilisation des échantillons de l'acier 2 rectifié en épaisseur a mis en évidence que plus
les bandes de ségrégation étaient proches de la surface extérieure, plus les éprouvettes
fissuraient tôt en pliage en V. De plus, des fissures en sous cutané se sont initiées sur ces
bandes. Une relation quantitative entre l'endommagement, l'épaisseur des bandes et la
déformation locale "mésoscopique" a été proposée.
Dans le chapitre 4, une comparaison entre les deux tests de pliage a été présentée. La loi
de comportement établie dans le chapitre 2 a été utilisée et associée à une extension du
modèle de Gurson-Tvergaard-Needleman (GTN) pour bien prédire la rupture à la fois en
pliage en V et en pliage sous traction. Des observations au microscope électronique à
balayage couplées à des résultats de simulation numérique ont permis de déterminer les
175
paramètres d'endommagement à la fois pour "les bandes de ségrégation" et pour "la

matrice ductile". En utilisant ce matériau "composite" représentatif de l'échantillon, une
modélisation des deux tests de pliage a été effectuée en mettant une bande dure à mi-
épaisseur possédant la plus grande sensibilité à l'endommagement. L'utilisation d'un tel
modèle a permis la prédiction des courbes de force/déplacement aussi bien que la
localisation de la rupture pour deux tests différents : le pliage en V et le pliage sous
traction. De plus, en raison de la striction apparaissant dans la largeur de l'échantillon,
une simulation en 3D a été nécessaire en pliage sous traction alors que des calculs 2D
étaient suffisamment précis pour la modélisation du pliage sous traction.
Pour les aciers TRIP étudiés, les bandes dures n'ont pas la même influence en pliage en V
ou en pliage sous traction. En pliage en V, les structures en bandes ont un effet néfaste sur
le pliage tandis que la ségrégation centrale n'entraîne pas une fissuration prématurée.
Néanmoins, dans un cas, nous avons observé de l'endommagement au niveau de la
ségrégation centrale en pliage en V. Dans ce cas particulier, la bande de ségrégation
centrale était particulièrement épaisse, mais n'était pas responsable de la microfissure de
la surface extérieure. Nous avons montré que des fissures sous-cutanées peuvent
apparaître quand les bandes de ségrégation sont placées proches de la surface extérieure.
Dans les conditions décrites par l'étude paramétrique effectuée à la fin du chapitre 4
(pliage en V de l'acier 1 avec différentes positions des bandes à travers l'épaisseur), nous
avons évalué cette distance critique à environ 150µm.
Il faut noter que l'influence néfaste d'une structure en bande dépendra d'une combinaison
de plusieurs paramètres comme sa capacité de se déformer sans endommagement, qui
dépendra de son épaisseur moyenne, son emplacement (distance par rapport à la surface
extérieure), la déformation atteinte en pliage et probablement d'autres paramètres comme
sa dureté, etc. Plus l'épaisseur de la bande et la déformation seront élevées, plus cette
distance critique sera faible.
Le paramètre métallurgique le plus néfaste en pliage sous traction est la ségrégation
centrale, étant donné que les fissures s'amorcent à partir de celle-ci. Lorsque des trajets de
chargement impliquant une contribution en traction sont appliqués, il est important de
contrôler et même d'optimiser la microstructure de ces aciers.
Comme indiqué dans le premier chapitre, différents procédés doivent être utilisés pour
agir sur les structures en bandes et sur la ségrégation centrale. Habituellement, une
température de bobinage basse est recommandée pour diminuer la sévérité de ces bandes.
De plus, comme des structures en bandes trop proches de la surface extérieure sont
néfastes en pliage en V, il sera préférable d'effectuer une surchauffe haute dans le
processus sidérurgique pour concentrer les structures en bandes vers la mi-épaisseur.
Cela doit être équilibré, cependant, avec les propriétés en pliage sous traction, qui
dépendent fortement de la microstructure à mi-épaisseur. Quant à la ségrégation centrale,
un remède pour la diminuer pourrait être d'utiliser le processus de réduction douce.
Perspectives
Pour aller plus loin dans la compréhension des liens entre microstructure et pliage, il
serait nécessaire d'appliquer les méthodes développées dans cette étude à d'autres familles
d'aciers comme les aciers Dual Phase (DP) ou Multiphasés (MP). Dans ces matériaux,
176
d'autres paramètres comme les inclusions, les éléments d'alliage, etc. pourraient peut être
jouer un rôle important sur le pliage.
Au cours de ce projet de thèse, nous nous sommes concentrés sur deux essais de mise en
forme : le pliage en V et le pliage sous traction. Il serait intéressant de déterminer
l'influence de la microstructure dans d'autres tests de mise en forme comme lors d'un
emboutissage ou d'un test d'expansion. Par exemple, pendant l'emboutissage, comme
certaines parties de la pièce sont en traction, la bande de ségrégation centrale pourrait
aussi avoir un impact pendant la mise en forme.
Les données de ce travail sont principalement méthodologiques et qualitatives quant aux

données microstructurales. En pliage en V, les structures en bandes sont les paramètres
les plus importants qui doivent être contrôlés pour améliorer le pliage des aciers.
Il sera intéressant d'effectuer des essais supplémentaires reliant l'épaisseur d'une bande,
sa déformation atteinte en pliage ainsi que sa sensibilité à l'endommagement, pour
déterminer précisément une épaisseur critique pour les structures en bandes ainsi que la
distance minimum par rapport à la surface à éviter pour limiter les fissures prématurées
en pliage. Il serait aussi important d'inclure les effets de la triaxialité des contraintes dans
la relation phénoménologique établie dans le chapitre 3 entre l'épaisseur critique d'une
bande et la déformation critique, pour appliquer de manière plus précise cette équation à
d'autres cas de mise en forme.
En pliage sous traction, la ségrégation centrale devrait être réduite et il sera essentiel
d'éviter des bandes dures continues.
Enfin, pour améliorer le pliage des aciers, des traitements thermiques pourraient être
utilisés pour homogénéiser la microstructure des tôles et ainsi obtenir des matériaux
"modèles".
Dans cette étude, le critère d'endommagement a été établi à une échelle macro ou
mésoscopique. Une analyse plus détaillée du comportement mécanique de chaque phase
pourrait être utile pour déterminer la contrainte à l'interface entre les phases dures et les
phases plus ductiles. La nanodureté de chaque phase a été donnée et il est maintenant
essentiel de fournir des informations sur la forme et la taille de ces phases pour effectuer
une étude micromécanique complète des mécanismes ductile d'endommagement pendant le
pliage. De plus, pour obtenir plus d'informations sur des mécanismes d'endommagement, il
serait intéressant d'utiliser la tomographie. Avec toutes ces données, il serait alors
possible d'utiliser un modèle micromécanique pour déterminer précisément un critère
d'endommagement.
La méthode des éléments finis s'est avérée être un bon outil pour étudier le pliage des
aciers. Avec une loi de comportement précise et en prenant en compte l'endommagement
de façon fiable, le modèle peut prévoir la rupture en pliage. Au vu des résultats, il semble
nécessaire de continuer l'étude paramétrique du test de pliage en V avec différentes
positions de la bande de ségrégation à travers l'épaisseur. L'étude actuelle a été effectuée
avec une bande de ségrégation "numérique" ayant une épaisseur de 50µm. Dans les quatre
aciers étudiés, la bande de ségrégation centrale était toujours plus mince et la zone
contenant de telles bandes dures pouvaient parfois atteindre cette épaisseur. Pour utiliser
des hypothèses plus réalistes, il serait nécessaire d'utiliser des maillages plus fins et de
refaire des calculs avec différentes positions de la bande à travers l'épaisseur. Il serait
également intéressant d'étudier les interactions entre deux bandes dures (particulièrement
l'effet de la distance entre elles) et leurs rôles sur l'endommagement.
177
Des études paramétriques semblables pourraient aussi être effectuées sur le pliage sous
traction afin de déterminer l'épaisseur critique que la ségrégation centrale ne devrait pas
excéder pour éviter des ruptures prématurées. En utilisant les cartes d'endommagement et
les courbes force/déplacement prédites par le modèle, il sera ainsi possible d'évaluer la
capacité en pliage d'un acier donné.
178
Bibliography
Bibliography
181
Bibliography
Bibliography
Andrews, K.W., 1965, Empirical formulae for the calculation of some transformation
temperatures, Journal of the Iron and Steel Institute, 203, p. 721-727.
Aramis system, Optical 3D deformation analysis software developed by GOM GmbH,

http://www.gom.com/3d-software/Aramis-software.html
Avramovic-Cingara, G., Ososkov, Y., Jain, M.K., Wilkinson, D.S., 2009, Effect of
martensite distribution on damage behaviour in DP600 dual phase steels, Materials Science
and Engineering A, 516, p. 7-16.
Bach, M., Broll, N., Cornet, A., Gaide, L., 1996, Diffraction X en traitements thermiques :
dosage de l'austénite résiduelle par diffraction des rayons X, Journal de physique IV, 6, p.
C4-887/C4-895.
Bahloul, R., Mkaddem, A., Dal Santo, Ph., Potiron, A., 2006, Sheet metal bending
optimisation using response surface method, numerical simulation and design of
experiments, International Journal of Mechanical Sciences, 48, p. 991-1003.
Bai, D.Q., Di Chiro, A., Yue, S., 1998, Stability of retained austenite in a Nb microalloyed
Mn-Si TRIP steel, Materials Science Forum, 284-286, p. 253-260.
Bai, Y., Wierzbicki, T., 2010, Application of extended Mohr-Coulomb criterion to ductile
fracture, International Journal of Fracture, 161, p. 1-20.
Baik, S.C., Kim, S., Jin, Y.S., Kwon, O., 2001, Effects of alloying elements on mechanical
properties and phase transformation of cold rolled TRIP steel sheets, ISIJ International, 41,
p. 290-297.
Bao, Y., Wierzbicki, T., 2005, On the cut-off value of negative triaxiality for fracture,
Engineering Fracture Mechanics, 72, p. 1049-1069.
Barbel, L., Tosal-Martinez, L., De Cooman, B., 2002, Effect of phosphorus on the
properties of a cold rolled and intercritically annealed TRIP-aided steel, in : Int. Conf. on
TRIP-Aided high strength ferrous alloys, Ghent, Belgium, p. 147-152.
Barlat, F., Lege, D.J., Brem, J.C., 1991, A six component yield function for anisotropic
materials, International Journal of Plasticity, 7, p. 693-712.
Barlat, F., Brem, J.C., Yoon, J.W., Chung, K., Dick, R.E., Lege, D.J., Pourboghrat, F.,
Choi, S.H., Chu, E., 2003, Plane stress yield function for aluminum alloy sheets - Part 1:
theory, International Journal of Plasticity, 19, p. 1297-1319.
Bastien, P.G., 1957, The Mechanism of formation of banded structure, ISIJ, p. 281-291.
182
Bibliography
Ben Bettaieb, M., Lemoine, X., Duchêne, L., Habraken, A.M., 2010, Simulation of the
bending process of hardening metallic sheets using damage model. Part I: Theoretical
development and numerical implementation, Material Science and Engineering A, 528, p.
434-441.
Ben Bettaieb, M., Lemoine, X., Duchêne, L., Habraken, A.M., 2010, Simulation of the
bending process of hardening metallic sheets using damage model. Part II: Numerical
investigations, Materials Science and Engineering A, 528, p. 442-448.
Béranger, G., Henry, G., Sanz, G., 1994, Le livre de l'acier, Lavoisier
Besson, J., Foerch, R., 1997, Large scale object-oriented finite element code design,
Computer methods in applied mechanics and engineering, 142, p. 165-187.
Besson, J., Devillers-Guerville, L., Pineau, A., 2000, Modeling of scatter and size effect in
ductile fracture: application to thermal embrittlement of duplex stainless steels,
Engineering Fracture Mechanics, 67, p. 169-190.
Besson, J., Steglich, D., Brocks, W., 2003, Modeling of plane strain ductile rupture,
International Journal of Plasticity, 19, p. 1517-1541.
Besson, J., Guillemer-Neel, C., 2003, An extension of the Green and Gurson models to
kinematic hardening, Mechanics of Materials, 35, p. 1-18.
Bleck, W., 2002, Using the TRIP effect –The Dawn of a promising group of cold formable
steels, in : Int. Conf. on TRIP-aided high strength ferrous alloys, Ghent, Belgium, p. 13.
Bobadilla, M., 2003, Microsegregation in steels, ArcelorMittal internal document.
Bobadilla, M., 2008, Solidification in continuous casting, ArcelorMittal internal document.
Bobadilla, M., Theyssier, M.C., Jolivet, J.M., Bonnet, F., Sturel, T., 2009,
Macrosegregations and microsegregations control in AHSS grades. Key actuators,
ArcelorMittal internal document.
Bouquerel, J., Verbeken, K., De Cooman, B.C., 2006, Microstructure-based model for the
static mechanical behaviour of multiphase steels, Acta Materialia, 54, p. 1443-1456.
Bridgman, P.W., 1952, Studies in large plastic flow and fracture: With special emphasis on
the effects of hydrostatic pressure, McGraw-Hill.
Bron, F., Besson, J., 2004, A yield function for anisotropic materials. Application to
aluminium alloys, International Journal o f Plasticity, 20, p. 937-963.
Brunet, M., Sabourin, F., Mguil-Touchal, S., 1996, The prediction of necking and failure in
3D sheet forming analysis using damage variable, Journal de Physique III, 6, p. C6-
473/C6-482.
Chaboche, J.L., 1989, Constitutive equations for cyclic plasticity and cyclic viscoplasticity.
183
Bibliography
Chaboche, J.L., 2008, A review of some plasticity and viscoplasticity constitutive theories,
Chakrabarty, J., Lee, W.B., Chan, K.C., 2000, An analysis of the plane-strain bending of
an orthotropic sheet in the elastic/plastic range, Journal of Materials Processing
Technology, 104, p. 48-52.
Chakrabarty, J., Lee, W.B., Chan, K.C., 2001, An exact solution for the elastic/plastic
bending of anisotropic sheet metal under conditions of plane strain, International Journal of
Mechanical Sciences, 43, p. 1871-1880.
Charpentier, P.L., 1975, Influence of punch curvature on the stretching limits of sheet
steel, Metallurgical Transactions A, 6, p. 1665-1669.
Chen, H.C., Era, H., Shimizu, M., 1989, Effect of phosphorus on the formation of retained
austenite and mechanical properties in Si-containing low-carbon steel sheet, Metallurgical
Transactions A, 20, p. 437-445.
Cheng, L., Böttger, A., De Keijser, T.H., Mittemeijer, E.J., 1990, Lattice parameters of
iron-carbon and iron-nitrogen martensites and austenites, Scripta Metallurgica, 24, p. 509-
514.
Chien, W.Y., Pan, J., Tang, S.C., 2004, A combined necking and shear localization
analysis for aluminum sheets under biaxial stretching conditions, International Journal of
Plasticity, 20, p. 1953-1981.
Chu, C.C., Needleman, A., 1980, Void nucleation effects in biaxially stretched sheets,
Journal of Engineering Materials and Technology, 102, p. 249-256.
Cockroft, M.G., Latham, D.J., 1968, Ductility and workability of metals, Journal of the
Institute of Metals, 96, p. 33-39.
Col, A., 2010, L’emboutissage des aciers, L’Usine Nouvelle, Dunod, p. 273.
Dalloz, A., 2007, Etude de l’endommagement par découpe des aciers Dual Phase pour
application automobile, thèse, Ecole des Mines de Paris.
Dalloz, A., Besson, J., Gourgues-Lorenzon, A.F., Sturel T., Pineau, A., 2009, Effect of
shear cutting on ductility of a dual phase steel, Engineering Fracture Mechanics, 76,
p.1411-1424.
Damborg, D.D., 1998, Prediction of fracture in bending under tension, in : Proceeding of

the 19th Riso International Symposium on Material Science: Modelling of Structure and
Mechanics from Microscale to Product, Roskilde, Denmark, p. 235-240.
Date, P.P., Narashima, K., Maiti, S.K., Singh, U.P., 1999, On the prediction of springback
in V-bending of metallic sheets under plane strain condition, in : Proceeding of the
International Conference on Sheet Metal, Erlangen, Germany, p. 447-456.
184
Bibliography
Datsko, J., Yang, C.T., 1960, Correlation of bendability of materials with their tensile
properties, Journal of Engineering for Industry, , p. 309-314.
Demeri, M.Y., 1981, The stretch bend forming of sheet metal, Journal of Applied
Metalworking, 2, p. 3-10.
De Meyer, M., Vanderschueren, D., De Cooman, B.C., 1999, The influence of the
substitution of Si by Al on the properties of cold rolled C-Mn-Si TRIP steels, ISIJ
International, 39, p. 813-822.
Determe, H., 1998, Cétim, Formation: Influence des caractéristiques de la tôle sur les
opérations de pliage, emboutissage et découpage.
Didier, M., Bobadilla, M., 2008, Solidification and casting. Microsegregation and
macrosegregation, ArcelorMittal internal document.
Eggertsen, P.A., Mattiasson, K., 2009, On the modelling of the bending-unbending

behaviour for accurate springback predictions, International Journal of Mechanical
Sciences, 51, p. 547-563.
Eggertsen, P.A., Mattiasson, K., 2010, On constitutive modelling for springback analysis,
International Journal of Mechanical Sciences, 52, p. 804-818.
Faleskog, J., Gao, X., Shih, C., 1998, Cell model for nonlinear fracture analysis. I:
Micromechanics calibration, International Journal of Fracture, 89, p. 355-373
Falkinger, G., Andrieux, F., Helm, D., Riedel, H., 2010, Finite-element modelling of
Nakajima tests in due consideration of anisotropic ductile damage, in : IDDRG, June 2010,
Graz, Autriche.
Friedman, P.A., Luckey, S.G., 2002, Failure of Al-Mg-Si alloys in bending, Journal of
Failure Analysis, 1, p. 33-42.
Furnémont, Q., Kempf, M., Jacques, P.J., Göcken, M., Delannay, F., 2002, On the
measurement of the nanohardness of the constitutive phases of TRIP-assisted multiphase
steels, Materials Science and Engineering A, 328, p. 26-32.
Gallagher, M.F., Speer, J.G., Matlock, D.K., Fonstein, N.M., 2002, Microstructure
development in TRIP-sheet steels containing Si, Al and P, in : Proceedings of the 44th
Mechanical Working and Steel Processing Conference and the 8th International Rolling
Conference and International Symposium on Zinc-Coated Steels, Orlando, 2002, p.153-
172.
Gan, W., Wagoner, R.H., 2004, Die design method for sheet springback, International
Journal of Mechanical Sciences, 46, p. 1097-1113.
Geng, L., Wagoner, R.H., 2002, Role of plastic anisotropy and its evolution on springback,
Gieck, K., 1997, Formulaire technique, 10th ed., Dunod.
185
Bibliography
Girault, E., Mertens, A., Jacques, P., Houbaert, Y., Verlinden, B., Van Humbeek, J., 2001,
Comparison of the effects of silicon and aluminium on the tensile behaviour of multiphase
TRIP-Assisted steels, Scripta Materialia, 44, p. 885-892.
Goodwin, G.M., 1968, Application of strain analysis to sheet metal forming in the press
shop, SAE paper 680093
Ghosh, A.K., 1976, A criterion for ductile fracture in sheets under biaxial loading,
Metallurgical Transactions A, 7, p. 523-533.
Graf, A., Hosford, W., 1993, Effect of changing strain paths on forming limit diagrams of
Al2008-T4, Metallurgical Transactions A, 24, p. 2503-2512.
Gurson, A.L., 1977, Continuum theory of ductile rupture by void nucleation and growth:
part I-yield criteria and flow rules for porous ductile media, Journal of Engineering
Materials and Technology, 99, p. 2-15.
Haidemenopoulos, G.N., Olson, G.B., Cohen, M., Tsuzaki, K., 1989, Transformation
plasticity of retained austenite in stage-I tempered martensitic steels, Scripta Metallurgica,
23, p. 207-212.
Hambli, R., Mkaddem, A., Potiron, A., 2004, Finite element damage modeling in bending
processes, Journal of Materials Processing Technology, 147, p. 302-310.
Hecker, S.S., Stout, M.G., Staudhammer, K.P., Smith, J.L., 1982, Effects of strain state and
strain rate on deformation-induced transformation in 304 stainless steel: Part I. Magnetic
measurements and mechanical behaviour, Metallurgical Transactions A, 13, p. 619-626.
Hill, R., 1948, A theory of the yielding and plastic flow of anisotropic metals, Proceedings
of the Royal Society of London, A193, p. 281-297.
Huang, J., Hammond, R.P., Conlon, K., Poole, W.J., 2002, An experimental study of the
ferrite austenite two phase regions in a Fe-C-Mn-Si steel using neutron diffraction, in :
Proc. Int. Conf. on TRIP aided high strength ferrous alloys, Ghent, Belgium, 2002, p.187.
Hudgins, A., Matlock, D., Speer, J., Fekete, J., Walp, M., 2007, The susceptibility to shear
fracture in bending of advanced high strength sheet steels, in: Proceedings from Materials
Science and Technology (MS&T): Advanced High Strength and Other Specialty Sheet
Steel Products for the Automotive Industry, Detroit, Michigan, 2007, p. 145-157.
Jacques, P., 1998, On the physics and mechanics of phase transformations in TRIP-assisted
multiphase steels, Thèse de doctorat, Université Catholique de Louvain, Belgium.
Jacques, P., Furnemont, Q., Pardoen, T., Delannay, F., 2001, On the role of martensitic
transformation on damage and cracking resistance in TRIP-assisted multiphase steels, Acta
Materialia, 49, p. 139-152.
Jacques, P.J., Petein, A., Harlet, P., 2002, Improvement of mechanical properties through
concurrent deformation and transformation : new steels for 21st century, in : Proc. Int.
Conf. on TRIP-Aided high strength ferrous alloys, Ghent, Belgium, 2002, p. 281.
186
Bibliography
Jacques, P.J., Furnémont, Q., Lani, F., Pardoen, T., Delannay, F., 2007, Multiscale
mechanics of TRIP-assisted multiphase steels: I. Characterization and mechanical testing,
Acta Materialia, 55, p. 3681-3693.
Jacques, P., Allain, S., Bouaziz, O., De, A., Gourgues, A.F., Hance, B.M., Houbaert, Y.,
Huang, J., Iza-Mendia, A., Kruger, S.E., Radu, M., Samek, L., Speer, J., Zhao, L., Van der
Zwaag, S., 2009, On measurement of retained austenite in multiphase TRIP steels – results
of blind round robin test involving six different techniques, Materials Science and
Kao, A.S., Kuhn, H.A., Richemond, O., Spitzig, W.A., 1990, Tensile fracture and
fractographic analysis of 1045 spheroidized steel under hydrostatic pressure, Journal of
Materials Research, 5, p. 83-91.
Karafillis, A.P., Boyce, M.C., 1993, A general anisotropic yield criterion using bounds and
a transformation weighting tensor, Journal of the Mechanics and Physics of Solids, 41,
p. 1859-1886.
Keeler, S.P. and Backofen, W.A., 1963, Plastic instability and fracture in sheets stretched
over rigid punches, Transactions of the American Society for Metals, 56, p. 25-48.
Kim, S.J., Lee, C.G., Choi, I., Lee, S., 2001, Effects of heat treatment and alloying
elements on the microstructures and mechanical properties of 0.15 wt pct C TRIP-Aided
cold rolled steel sheets, Metallurgical and Materials Transactions A, 32, p. 505-514.
Kim, S.J., Lee, C.G., Jeong, W.C., Park, I., 2002, Microstructure and mechanical
properties of the 0.15%C TRIP-Aided cold rolled steels containing Cu, Ni and Cr, in :
Proc. Int. Conf. on TRIP-Aided high strength ferrous alloys, Ghent, Belgium, 2002, p. 165.
Kim, D., Barlat, F., Bouvier, S., Rabahallah, M., Balan, T., Chung, K., 2007, Nonquadratic
anisotropic potentials based on linear transformation of plastic strain rate, International
Journal of Plasticity, 23, p. 1380-1399.
Kim, J.H., Sung, J.H., Wagoner, R.H., 2009, Thermo-mechanical modelling of draw-bend
formability tests, in : International conference, International Deep Drawing Research
Group, Golden, CO, USA, 2009, p. 503-512.
Ladriere, J.H., He, X.J., 1986, Mössbauer study on retained austenite in a Fe-Mn-C dual
phase steel, Materials Science and Engineering A, 77, p.133-138.
Lani, F., Furnémont, Q., Van Rompaey, T., Delannay, F., Jacques, P.J., Pardoen, T., 2007,
Multiscale mechanics of TRIP-assisted multiphase steels: II. Micromechanical modelling,
Acta Materialia, 55, p. 3695-3705.
Laukonis, J.V., Ghosh, A.K., 1978, Effects of strain paths changes on the formability of
sheet metals, Metallurgical Transactions A, 9, p. 1849-1856.
Lee, C.G., Kim, S.J., Oh, C.S., Lee, S., 2002, Effects of heat treatment and Si addition on
the mechanical properties of 0.1 wt% C TRIP-Aided cold rolled steel, ISIJ International,
42, p. 1162-1168.
187
Bibliography
Lee, C.G., Kim, S.J., Lee, T.H., Lee, S., 2004, Effects of volume fraction and stability of
retained austenite on formability in a 0.1C-1.5Si-1.5Mn-0.5Cu TRIP-aided cold rolled steel
sheet, Materials Science and Engineering A, 371, p.16-23.
Lemaitre, J., 1985, A continuous damage mechanics model for ductile fracture, Journal of
Engineering Materials and Technology, 107, p. 83-89.
Lemaitre, J., Chaboche, J.L., 1990, Mechanics of Solid Materials, Cambridge University
Press.
Lemoine, X., Bettaieb, M.B., Goncalves, J., 2009, Bibliography about bending process in
particular formability criteria, ArcelorMittal internal document, p. 1-31.
LeRoy, G., Embury, J.D., Edwards, G., Ashby, M.F., 1981, A model of ductile fracture
based on the nucleation and growth of voids, Acta Metallurgica, 29, p. 1509-1522.
Leu, D.K., 1997, A simplified approach for evaluating bendability and springback in
plastic bending of anisotropic sheet metals, Journal of Materials Processing Technology,
66, p. 9-17.
Li, X., Yang, Y., Wang, Y., Bao, J., Li, S., 2002, Effect of the material hardening mode on
the springback simulation accuracy of V-free bending, Journal of Materials Processing
Technology, 123, p. 209-211.
Lin, M., Olson, G.B., Cohen, M., 1992, Distributed-activation kinetics of heterogeneous
martensitic nucleation, Metallurgical Transactions 23A, p. 2987-2998.
Livatyali, H., Kinzel, G.L., Altan, T., 2003, Computer aided die design of straight flanging
using approximate numerical analysis, Journal of Materials Processing and Technology,
142, p. 532-543.
Luo, M., Wierzbicki, T., 2010, Numerical failure analysis of a stretch bending test on dual-
phase steel sheets using a phenomenological fracture model, International Journal of Solids
and Structures, 47, p. 3084-3102.
Mahieu, J., Van Dooren, D., Barbé, L., De Cooman, B.C., 2002, Influence of Al, Si and P
on the kinetics of intercritical annealing of TRIP-Aided steels: Thermodynamical
prediction and experimental verification, in : Proc. Int. Conf. on TRIP-Aided high strength
ferrous alloys, Ghent, Belgium, 2002, p. 159.
Maki, J. Mahieu, J., De Cooman, B.C., Claessens, S., 2003, Galvanisability of silicon free
CMnAl TRIP steels, Material Science and Technology, 19, p. 125-131.
Marciniak, Z., Duncan, J.L., 1992, Bending. The mechanics of sheet metal forming, E.
Arnold, p.68-99.
Matsumura, O., Sakuma, Y., Takechi, H., 1987, Enhancement of elongation by retained
austenite in intercritical annealed 0.4C-1.5Si-0.8Mn steel, Transactions of the Iron and
steel institute of Japan, 27, p. 570-579.
188
Bibliography
McClintock, F.A., 1968, A criterion of ductile fracture by growth of holes, Journal of

Applied Mechanics, 35, p. 363-371.
Mintz, B., 2001, Hot dip galvanising of transformation induced plasticity and other
intercritically annealed steels, International Materials Reviews, 46, 4, p. 169-197.
Moran, B., Ortiz, M., Shih, C.F., 1990, Formulation of implicit finite-element methods for
multiplicative finite deformation plasticity, International Journal for Numerical Methods in
Engineering, 29, p. 483-514.
Muller, J., 2000. Formulaire de mécanique générale, 17th ed, Paillart, p.147.
Nagataki, Y., Tsuyama, S., Hosoya, Y., Okita, T., 1994, Effect of microstructure on
bendability of ultra-high strength steel sheet, in : High-strength steels for automotive
symposium proceedings, p. 239-244.
Nilsson, A., Melin, L., Magnusson, C., 1997, Finite-element simulation of V-die bending:
a comparison with experimental results, Journal of Materials Processing Technology, 65,
p. 52-58.
Ohlert, J., Bleck, W., Hulka, K., 2002, Control of microstructure on TRIP steels by
niobium, in : Proc. Int. Conf. on TRIP-Aided High Strength Ferrous Alloys, Ghent,
Belgium, 2002, p. 199.
Patel, J.R., Cohen, M., 1953, Criterion for the action of applied stress in the martensitic
transformation, Acta Metallurgica, 1, p. 531-538.
Pereloma, E.V., Timokhina, I.B., Hodgson, P.D., 1999, Transformation behaviour in

thermomechanically processed C-Mn-Si TRIP steels with and without Nb, Materials
Science and Engineering A, 273-275, p. 448-452.
Perrin, G. and Leblond, J.B., 1990, Analytical study of a hollow sphere made of plastic
porous material and subjected to hydrostatic tension-application to some problems in
ductile fracture of metals, International Journal of Plasticity, 6, p. 677-699.
Pichler, A., Stiaszny, P., 1999, TRIP steel with reduced silicon content, Steel Research,
70, p. 459-465.
Pickering, F.B., 1992, Constitution and properties of steels, Materials science and
technology : a comprehensive treatment, 7, p. 272-284.
Poruks, P., Yakubtsov, I., Boyd, J.D., 2006, Martensite-ferrite interface strength in a low-
carbon bainitic steel, Scripta Materialia, 54, p. 41-45.
Quach, W.M., Teng, J.G., Chung, K.F., 2004, Residual stresses in steel sheets due to
coiling and uncoiling: a closed-form analytical solution, Engineering Structures, 26, p.
1249-1259.
189
Bibliography
Quach, W.M., Teng, J.G., Chung, K.F, 2009, Residual stresses in press-braked stainless
steel sections - I: Coiling and uncoiling of sheets, Journal of Constructional Steel Research,
65, p. 1803-1815.
Ragab, A.R., Saleh, C.A., 2005, Evaluation of bendability of sheet metals using void
coalescence models, Materials Science and Engineering A, 395, p.102-109.
Rèche, D., Sturel, T., Bouaziz, O., Gourgues-Lorenzon, A.F, 2011, Damage development
in low alloy TRIP-aided steels during air-bending, Materials Science and Engineering A,
528, p. 5241-5250.
Rèche, D., Besson, J., Sturel, T., Lemoine, X., Gourgues-Lorenzon, A.F., 2011, Analysis
of the air-bending tests using finite element simulation: Application to steel sheets,
submitted to International Journal of Mechanical Sciences, May 2011.
Reed, R.C., Akbay, T., Shen, Z., Robinson, J.M., Root, J.H., 1998, Determination of
reaustenitisation kinetics in a Fe-0.4C steels using dilatometry and neutron diffraction,
Material Science and Engineering A, 256, p. 152-165.
Rice, J.R. and Tracey, D.M., 1969, On the ductile enlargement of voids in triaxial stress
fields, Journal of the Mechanics and Physics of Solids, 17, p. 201-217.
Ridley, N., Stuart, H., Zwell, L., 1969, Lattice parameters of Fe-C austenites at room
temperature, Trans. Met. Soc. AIME, 245, p. 1834-1836.
Rigsbee, J.M., Van der Arend, P.J., 1977, Laboratory studies of microstructures and
structure-property relationships in dual phase steels, ed. A.T. Davenport, in : Proc. Conf.
Formable HSLA and Dual Phase steels, Chicago, AIME, p. 56-86.
Roberts, C.S., 1953, Effect of carbon on the volume fractions and lattice parameters of
retained austenite and martensite, Trans. AIME, 197, p. 203-204.
Saanouni, K., Badreddine, H., 2007, Damage prediction in sheet metal forming: theory,
numerics and application, in : Numiform’07, Proceedings of the 9th International
Conference on Numerical Methods in Industrial Forming Processes, Porto, Portugal, AIP
Conference proceedings, p. 105-110.
Sriram, S., Wong, C., Huang, M., Yan, B., 2003, Stretch bendability of advanced high
strength steels, SAE technical paper, n° 2003-01-1151.
Sakuma, Y., Matsumura, O., Takechi, H., 1991, Mechanical properties and retained
austenite in intercritically heat treated bainite transformed steel and their variation with Si
and Mn additions, Metallurgical Transactions A, 22, 1991, p. 489-498.
Sarkar, J., Kutty, T.R.G., Conlon, K.T., Wilkinson, D.S., Embury, J.D., Lloyd, D.J.,
Tensile and bending properties of AA5754 aluminum alloys, 2001, Materials Science and
Engineering A, 316, p. 52-59.
190
Bibliography
Shinohara, Y., Madi, Y., Besson, 2010, J., A combined phenomenological model for the
representation of anisotropic hardening behaviour in high strength steel line pipes,
European Journal of Mechanics / A Solids, 29, p. 917-927.
Speich, G.R., Miller, R.L., 1979, Mechanical properties of ferrite-martensite steels, in :

Proc. Conf. Structure and Properties of Dual Phase Steels, New Orleans, TMS AIME, p.
145-182.
Sriram, S., Wong, C., Huang, M., Yan, B., 2003, Stretch bendability of advanced high
strength steels, in : SAE International World Congress, Detroit, Michigan, p. 107-115.
Srivastava, A.K., Bhattacharjee, D., Jha, G., Gope, N., Singh, S.B., 2007, Microstructural
and mechanical characterization of C-Mn-Al-Si cold-rolled TRIP-aided steel, Materials
Science and Engineering A, 445-446, p. 549-557.
Steninger, J., Melander, A., 1982, The Relation between bendability, tensile properties and
particle structure of low-carbon steel, Scandinavian Journal of Metallurgy, 11, p. 55-71.
Stöcker, H., Jundt, F., Guillaume, G., 2007. Physique, 2nd ed., Dunod, p. 253.
Sugimoto, K.I., Kikuchi, R., Hashimoto, S.I., 2002, Development of high strength low
alloy TRIP-aided steels with annealed martensite matrix, Steel research, 73, p. 253-258.
Tharrett, M.R., Stoughton, T.B., 2003, Stretch-bend forming limits of 1008 Ak steel, in :
SAE paper, n° 2003-01-1157, p.1-7.
Thompson, S.W., Howell, P.R., 1992, Factors influencing ferrite/pearlite banding and
origin of large pearlite nodules in a hypoeutectoid plate steel, Materials Science and
Thuillier, S., Le Maoût, N., Manach, P.Y., 2010, Bending limit prediction of an aluminium
thin sheet, International Journal of Material Forming, 3, p. 223-226.
Thuillier, S., Le Maoût, N., Manach, P.Y., 2011, Influence of ductile damage on the
bending behaviour of aluminium alloy thin sheets, Materials and Design, 32, p. 2049-2057.
Traint, S., Pichler, A., Hauzenberger, K., Stiaszny, P., Werner, E., 2002, Influence of
silicon, aluminium, phosphorus and copper on the phase transformations of low alloyed
TRIP steels, in : Proc. Int. Conf. on TRIP-Aided high strength ferrous alloys, Ghent,
Belgium, 2002, p. 121.
Tsukatani, I., Hashimoto, S., Inoue, T., 1991, Effects of silicon and manganese addition on
mechanical properties of high-strength hot rolled sheet steel containing retained austenite,
ISIJ International, 31, p. 992-1000.
Tvergaard, V. and Needleman, A., 1984, Analysis of the cup-cone fracture in a round
tensile bar. Acta Metallurgica, 32, p. 157-169.
Tvergaard, V., 1989, Material failure by void growth to coalescence, Advances in Applied
Mechanics, 27, p. 83-151.
191
Bibliography
Uthaisangsuk, V., Prahl, U., Münstermann, S., Bleck, W., 2008, Experimental and
numerical failure criterion for formability prediction in sheet metal forming,
Computational Materials Science, 43, p. 43-50.
Vallellano, C., Morales, D., Garcia-Lomas, F.J., 2008, On the study of the effect of metal
sheets, in : Conference Numisheet, Interlaken, Switzerland, September 2008.
Vladimirov, I.N., Pietryga, M.P., Reese S., 2010, Anisotropic finite elastoplasticity with
nonlinear kinematic and isotropic hardening and application to sheet metal forming,
Wang, C., Kinzel, G., Altan, T., 1993, Mathematical modeling of plane strain bending of
sheet and plate, Journal of Materials Processing Technology, 39, p. 279-304.
Williams, B.W., Simha, C.H.M., Abedrabbo, N., Mayer, R., Worswick, M.J., 2010, Effect
of anisotropy, kinematic hardening and strain-rate sensitivity on the predicted axial crush
response of hydroformed aluminium alloy tubes, International Journal of Impact
Engineering, 37, p. 652-661.
Wu, J., Zhou, D., Zhang, L., Zhou, Y.J., Du, C.Q., Shi, M.F., 2006, A failure criterion for
stretch bendability of advanced high strength steels, SAE Technical Paper, n° 2006-01-
0349.
Yamazaki, K., Oka, M., Yasuda, H., Mizuyama, Y., Tsuchiya, H., 1995, Recent Advances
in UHSS for automotive structural use, Nippon steel technical report, n°64, p. 37-44.
Yamazaki, K., Mizuyama, Y., Oka, M., Tokunaga, Y., 1995, Influence of microstructure
on bendability of ultrahigh-strength steel sheet, Journal of the Japan Society for
Technology of Plasticity, 36, p. 973 – 978.
Yan, H.Y., Kai, G.Y., Jian, M.D., 2006, Transformation behaviour of retained austenite
under different deformation modes for low alloyed TRIP-assisted steels, Material Science
and Engineering A, 441, p. 331-335.
Yoon, J., Barlat, F., Dick, R.E., Chung, K., Kang, T.J., 2004, Plane stress yield function
for aluminum alloy sheets - Part II : FE formulation and its implementation, International
Journal of Plasticity, 20, p. 495-522.
Zackay, V.F., Parker, E.R., Fahr, D., Bush, R., 1967, The enhancement of ductility in high
strength steels, ASM Transactions Quarterly, 60, p. 252-259.
Zhang, Z.T., Hu, S.J., 1998, Stress and residual stress distributions in plane strain bending,
Zhao, K.M., Lee, J.K., 2002, Finite element analysis of the three-point bending of sheet
metals, Journal of Material Processing Technology, 122, p. 6-11.
192
Appendix
Appendix 1: Microstructural characterization

procedures
Appendix 2: Mechanical characterization procedures
195
Appendix 1: Microstructural characterization procedures
Appendix 1: Microstructural characterization

procedures
A1.1. Light microscopy
Light optical microscopy was performed on each TRIP steel to characterize their
microstructures and to conduct a quantitative analysis of phase components. Samples were
mounted in resin and polished down to 1µm.
For these steels, a colorant etching is used to reveal the microstructure. The LePera etching
makes it possible to reveal hard phases (martensite + retained austenite) in white and the
soft ferrite in brown. This etching is composed of:
- Sodium disulfite (Sodium metabisulfite Na2 S2 O5)
- Picric acid ( C6 H3 N3 O7 )
- Ethanol (C2 H6 O)
1) Dilute 1g of sodium disulfite in 100ml of distilled water → solution (A)

2) Dilute 4g of picric acid in 100ml ethanol → solution (B)
Just before using, mix these two solutions with identical proportions. Polished samples
have to be etched during 5 to 10s.
In order to quantify the M/A islands content in each steel, light microscopy was carried out
after a Klemm colorant etching. This etching enables to colour in white the M/A islands.
This etching is composed of:
- 2g of potassium disulfite in 100mL of solution (A)

- Solution (A): 1kg of sodium thiosulfate in 300mL of water (warm this solution to
help the dissolution).
A1.2. Scanning electron microscopy
The scanning electron microscope was used to observe damage more precisely and to
perform fractographic observations. To carry out fractographic observations, specimens
were directly put inside the microscope chamber and their fracture surfaces were observed
using secondary electron imaging.
Observations of damage mechanisms required polished and electrically conductive

specimens. In this case, samples were etched with 4% Nital.
Analysis conditions used for the study:

- acceleration tension: 15kV
- working distance close to 15mm
- secondary electron mode
196
A1.3. X-ray diffraction measurements
The amount of retained austenite contained in these TRIP steels was determined thanks to
X-Ray Diffraction (XRD). These measurements have been performed, for each steel, on
as-received material (without bending) and on bent samples which were first polished
down to 1µm using diamond pastes and then with OPS (colloidal silica suspension) to
avoid surface hardening and polishing-induced austenite transformation. Note that, with
XRD, the measurements are localized at the first few microns just below the prepared
sample surface. These measurements were carried out with a D500 Brucker diffractometer
and with a cobalt tube (λ = 1.8 Å), under 40kV and 25mA. The angle domain in 2θ is
between 30° and 130° with a step every 0.032° and 1second/step.
detector
tube
2θ
ψ
Goniometric
φ circle
Figure A1.1. X-ray diffraction scheme.
To determine the amount of retained austenite in each sample, it is necessary to use the
intensities of 6 peaks:
- austenite (γ): 200, 220, 311
- ferrite (α): 200, 211, 220
The amount of retained austenite is given by these formulas (Bach et al., 1996):
Cγ
=
∑ Kα . ∑ I γ and Cα + C γ = 1 where Kα and Kγ are correction coefficients and Iα, Iγ
Cα ∑ Kγ ∑ Iα
are the intensities of each peak.
A1.4. Magnetic measurements
The amount of retained austenite has also been measured by sigmametry for the three
1.6mm thickness steels, both on as-received material (without bending) and on bent
samples. This technique consists in measuring the saturation magnetization of specimens
with and without austenite. To perform this experiment, the magnetization is first measured
on as-received samples (i.e. with retained austenite) and then, heat treatment at 500°C for
two hours is carried out to decompose austenite contained in the material into ferrite and
carbides.
197
Therefore, the content of retained austenite can be deduced by the difference between the
two magnetizations by using the following equation:
J s 2 − J s1
Vγ =
J s2
Where Js1 is the saturation magnetization of the specimen with austenite and Js2 is the
saturation magnetization of the specimen without austenite and Vγ is the volume content of
retained austenite.
A1.5. Vickers microhardness measurements
Hardness measurements through the thickness of as-received and air-bent specimens were
performed. The device used is a Emco microhardness testing. In the present case, two load
values were tested: 300g for thin sheet specimens (0.75mm) and 500g for the other ones
(1.6mm). Measurements were carried out in staggered rows on each specimen to get more
points through the thickness. In addition, samples were polished before testing to have
good reliability of measurements.
A1.6. Nanohardness measurements
The nanoindentation is a technique making possible to determine the nanohardness of

various phases and to access at their Young modulus. This technique consists in indenting
the material with a slight depth and to follow the penetration evolution as a function of the
applied load. For these measurements, specimens were polished and a slight Nital etching
was performed on each specimen. Analysis of experimental curves allows deducing
mechanical properties of the surface of the material. The applied load (in µN) and the
depth are stored all along the test. From these curves, the hardness (H in GPa) is
determined by the ratio between the maximal applied load (P) and the indentation area (A).
P
Hence: H =
A
The unloading curve depends on the elastic behaviour of the material. The Young modulus
is then calculated by taking the tangent on the unloading curve. For these measurements, a
loading/unloading cycle was used with a maximum applied load of 5000µN following by
holding for about one second at this maximum load. The loading rate is about 1660µN/s
both for the loading and the unloading.
Nanohardness measurements were carried out in the Evry’s University with a

nanohardness testing (Triboscope®, Hysitron SA) coupled with an atomic force
microscope-AFM (Digital Instrument 3100). The AFM helps to locate the area to indent
before testing, even if the resolution is poor due to non-optimized observation conditions.
In the present study, five nanohardness measurements were carried out both on the ferritic
matrix and the M/A islands in the vicinity of the mid-thickness. Note that, due to the small
size of the M/A islands, it was very difficult to be sure to have made an indentation on
198
these islands. Hence, when an indentation was performed at the interface between an M/A
island and the ferritic matrix, another measurement was carried out.
199
A2.1. Tensile tests
First, ISO tensile specimens of 80 mm in gauge length and 20 mm in width were tested
according to the European standard EN 10002-1 (Figure A2.1). Load/line displacement of
0.5mm/min was prescribed with an initial strain rate of 2,5.10-4 s-1. The engineering stress-
strain curves for the four steels were determined both along rolling and transverse
directions. The mechanical properties are characterized by means of the proof stress
(Rp0.2%), the ultimate tensile strength (TS) and the uniform elongation (Ue%).
260 mm
100 mm
30 mm
20 mm
80 mm
Figure A2.1. Scheme of tensile specimen (ISO 20x80 mm).
Secondly, tensile tests on smooth specimens with another dimension were performed for
Steel 1 only (Figure A2.2). In the present case, tests were conducted with an initial
elongation rate of 10-4 s-1.
Figure A2.2. Additional tensile specimen used for Steel 1.
Tensile tests were also carried out on notched specimens. These specimens have a notch
radius of 2mm with a distance between them of 10mm (Figure A2.3). The prescribed notch
opening rate is 0.5 mm.min-1.
200
50. 00
18. 00
10. 00
2. 00
1. 25
1. 50
2. 00
Figure A2.3. Scheme of notch tensile (NT) specimen and close-up view of a notch.
A2.2. Shear tests
Shear tests were performed on a shearing device and σ12 vs γ curves were stored for
reverse shear tests. The cyclic shear tests were carried out along the rolling direction for
γ = 10, γ = 20 and γ = 30 with a prescribed loading rate of 0.05 kN.s-1. The scheme of
reverse shear specimen is given in Figure A2.4.
Upper clamping
zone
TD
100 mm b =3 4.5
b= mmmm
Sheared zone
RD
Lower clamping
zone
Immobilized
area
h = 30 mm
Figure A2.4. Reverse shear specimen.
A2.3. Air-bending tests
Air-bending specimens with a size of 50x100mm (Figure A2.6) were first edge ground to
avoid premature failure starting from the cut-edge side surfaces. Air-bending device is
shown in Figure A2.5.
The device was used on a Zwick universal tensile machine. The two rotary rolls have a
diameter of 60mm. Two punches are used in this study. The first one is used to bend sheet
with thickness of 1.6mm. This punch has a width of 10 mm and a radius close to 0mm
(Figure A2.7a). The second one is used to bent sheet with a thickness of 0.75mm. It has a
width of 0.8mm and a radius of 0.4mm (Figure A2.7b). All tests were conducted without
lubrication.
201
For this test, a load/line displacement rate of 1mm.s-1 is used. The test is automatically
stopped after a load drop between 5 and 10%. Finally the bending angle is given as a
measurement to compare steels.
l
(a) (b)
Figure A2.5. (a) Air-bending test scheme, (b) air-bending device used for the study.
50 mm
100 mm
Figure A2.6. Air-bending specimen
(a) (b)
Figure A2.7. Air-bending punches. (a) punch with radius close to 0mm, (b) thin punch with radius of 0.4mm
and side stiffeners.
A2.4. Stretch bending tests
a. With the hydraulic press
50x250mm specimens with ground edges (Figure A2.8) were stretch bent on a hydraulic
press (BUP1000) (Figure A2.9). This press has a die and a blank-holder with respectively a
202
diameter of 85mm and 76mm. In addition, the die has a radius of 8mm and the draw-beads
have a radius of 10mm. The blank-holder force on this press is set at 480 kN and the punch
velocity is about 0.5mm.s-1. Several punches with different radii are available (2, 4, 6, 8, 10
and 15 mm, Figure A2.10). In the present case, only three punch radii were used: 2, 4 and
6mm.
50 mm
250 mm
Figure A2.8. Stretch bending specimen (R = 2 to 15 mm).
(a) (b)
Figure A2.9. Hydraulic press. (a) die, (b) blank-holder with draw-beads and punch.
Figure A2.10. Various punches (R = 2 to 15 mm).
b. With the servohydraulic tensile machine
In order to be able to interrupt the test just before fracture, a new device was designed at
Centre des Matériaux making it possible to access at lower rate (0.05mm.s-1) (Figure
A2.11). Samples are a little bit smaller by using the new device. The dimensions of stretch
bent specimens are 40x180mm. Here also, cut-edges have to be ground to avoid fracture
starting from them. In the present case, only the punch with the radius of 4mm is used.
Detailed scheme of the entire device is given in Figure A2.12.
203
This device is composed of:

- Three main parts (upper, lower and middle).
- One punch with a radius of 4 mm.
- Four rods sliding through the eight bronze rings in the three main parts used for the
guiding during the test.
- Two blank-holders making it possible to clamp the sheet during the test thanks to 8
screws.
Figure A2.11. Designing of the tool with Inventor software.
Screwing in the Screw

load cell
Upper part
Blank-holder
Middle part
Punch Lower part
Rod
Bronze ring
Screwing in
the jack
Figure A2.12. General view of the stretch bending device.
204
A.2.5. Strain measurement
In order to get the evolution of the strain during both bending and interrupted tensile tests,
optical measuring system with Aramis software was used. This system used for optical 3D
deformation analysis is developed by GOM GmbH. It is available to measure a maximum
strain of 100%. This system is equipped with two CCD cameras allowing monitoring
surface deformation thanks to a speckle deposited on the surface of the specimen.
It is then possible to obtained strain fields of the sheet. For bending experiments, the
acquisition frequency was 2 images per mm of punch displacement.
205
Relations between microstructure and bendability on TRIP-aided steels for
automotive products
Abstract
In order to limit fuel consumption, automotive industries push the steelmakers to develop thinner sheets
with higher strength. Ultra High Strength Steels (UHSS) such as low alloy TRIP (TRansformation Induced
Plasticity)-aided steels enable to get complex shapes for automotive parts. However, understanding the
formability of these steels appears to be complex and involves a detailed study of failure mechanisms
during forming tests. Therefore, links between microstructural features, in particularly banded structures,
and formability were investigated in order to better predict the bending ability of steels as a function of their
microstructure. In this study, four low alloy TRIP-aided steels exhibiting different bending performances
were characterized using both air and stretch bending tests. From scanning electron and light microscopy
observations, cracks initiate from the outer surface or from just below in air-bending, but from the central
segregation bands in stretch bending. Fracture surfaces after bending tests are ductile and damage mainly
appears by ferrite/martensite interface decohesion and occasionally as microcracks within martensite
islands. An original procedure based on thickness-reduced specimens, with various locations of the main
hard band within the specimen thickness, was set up. It enabled to propose a relationship between
cracking, local thickness of hard band and local strain during air-bending. In order to develop a local
fracture criterion that could be used for any bending test, stress and strain fields were computed by
numerical simulation of the two bending tests. For that purpose, an experimental database including
tensile tests on smooth and notched samples as well as shear tests was established. Material parameters
of constitutive equations, accounting for anisotropic elasto-plastic behaviour with mixed hardening, were
determined from this database. Finite element simulation of both bending tests associated with a ductile
damage criterion made it possible to satisfactorily predict strain fields, bending load-displacement curves
and fracture angle. The possibility of using simplifying assumptions in the model (such as an isotropic flow
criterion, pure isotropic hardening, two-dimensional assumptions and simplified boundary conditions) is
also discussed.
Keywords: Ultra High Strength Steels, TRIP-aided steels, microstructure, damage mechanisms, air-
bending test, stretch bending test, finite-element simulation.
Relation entre microstructure et aptitude au pliage des aciers à effet TRIP pour
application automobile
Résumé
Pour limiter la consommation des véhicules, les industries automobiles demandent aux aciéristes de
développer des aciers de plus en plus fins avec des hautes résistances. Les aciers très haute résistance
(THR) comme les aciers TRIP (TRansformation Induced Plasticity) permettent d’effectuer des pièces
automobiles aux formes complexes. Néanmoins, le compréhension des mécanismes de mise en forme de
ces aciers n’est pas simple et implique une étude détaillée des mécanismes de rupture apparaissant
durant la mise en forme. Ainsi, les liens entre les paramètres microstructuraux, en particulier les structures
en bandes, et la mise en forme ont été étudiés pour obtenir une meilleure prédiction de la capacité en
pliage des aciers en fonction de leur microstructure. Dans cette étude, quatre aciers TRIP présentant des
performances en pliage différentes ont été caractérisés à la fois en pliage en V et en pliage sous traction.
A partir d’observations au microscope optique et à balayage, nous avons démontré que les fissures
s’initiaient à partir de la surface ou juste en dessous en pliage en V alors qu’elles s’initiaient au niveau de
la ségrégation centrale en pliage sous traction. Les surfaces de rupture après pliage sont ductiles et
l’endommagement apparait principalement par décohésion de l’interface ferrite/martensite et
occasionnellement par rupture des ilots martensitiques. Une procédure originale basée sur des
échantillons rectifiés présentant la bande de ségrégation à différents endroits dans l’épaisseur a été établi.
Ce travail a permis de proposer une relation entre l’épaisseur d’une bande, son endommagement et sa
déformation locale atteinte pendant un test de pliage en V. Pour développer un critère de rupture pour ces
tests de pliage, les champs de contrainte et déformation ont été calculés par simulation numérique pour
ces deux tests. Pour ce faire, une base de données expérimentale incluant des tests de traction sur
éprouvettes lisses et entaillées et des tests de cisaillement a été établie. La loi de comportement du
matériau a été déterminée à partir de cette base expérimentale et présente un comportement élasto-
plastique anisotrope avec écrouissage mixte. La simulation numérique des deux essais de pliage associée
à un critère d’endommagement a permis de prédire de manière satisfaisante les champs de déformation,
les courbes force/déplacement ainsi les angles de rupture. La possibilité d’utiliser des modèles simplifiés
(tels qu’ un critère d’écoulement isotrope, un écrouissage purement isotrope, des calculs en 2D et des
conditions aux limites simplifiées) a également été discuté.
Mots clés: Aciers Très Haute Résistance, aciers TRIP, microstructure, mécanismes d’endommagement,
test de pliage en V, test de pliage sous traction, simulation numérique par éléments-finis.

Trip Steels

Transféré par

Droits d'auteur :

Formats disponibles

Trip Steels

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Trip Steels

Transféré par

Droits d'auteur :

Formats disponibles

Relations between microstructure and bendability on

TRIP-aided steels for automotive products

To cite this version:

HAL Id: pastel-00973376

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

l’École nationale supérieure des mines de Paris

Relations entre microstructure et aptitude au pliage des aciers à

Relations between microstructure and bendability on TRIP-aided

Thèse confidentielle jusqu’au 31 janvier 2014

Directeur de thèse : Anne-Françoise GOURGUES-LORENZON

Keywords: Ultra High Strength Steels, TRIP-aided steels, microstructure, damage

Je tiens maintenant à remercier sincèrement Anne-Françoise Gourgues-Lorenzon qui a

Je remercie mon encadrant chez ArcelorMittal, Thierry Sturel, de m’avoir accompagnée

Je souhaiterais maintenant remercier mes collègues d’ArcelorMittal. Je vais tout d’abord

I. Literature survey ........................................................................................................... 13

II. Microstructural characterization of the four TRIP-aided steels studied................ 36

III. Mechanical characterization of the four TRIP-aided steels studied ...................... 46

IV. Summary of Chapter 1 ............................................................................................... 51

V. Choice of the materials and approach used in the present study............................. 52

I. Global analysis of the air-bending test......................................................................... 57

II. Complementary data for model predictions .............................................................. 81

III. Influence of simulation conditions of the air-bending test ...................................... 90

IV. Summary of Chapter 2 ............................................................................................... 97

I. Experimental characterization of damage mechanisms........................................... 103

II. Influence of segregated bands on bendability.......................................................... 117

III. Summary of Chapter 3 ............................................................................................. 130

CHAPTER 4 ..................................................................................................................... 133

II. Influence of simulation conditions on the stretch-bending test.............................. 157

IV. Summary of Chapter 4 ............................................................................................. 166

GENERAL CONCLUSIONS AND PROPOSAL FOR FUTURE WORK................ 169

APPENDIX ...................................................................................................................... 195

Appendix 1: Microstructural characterization procedures ........................................ 196

Appendix 2: Mechanical characterization procedures ................................................ 200

Ultra High Strength Steels (UHSS) for

In the Kyoto protocol, it was decided to reduce

Ultra High Strength Steels in automotive industry

Figure 1. Typical applications of high strength steels for crash resistance

UHSS make reference to three main steel families:

- TRIP (TRansformation Induced Plasticity) steels: composed of a ductile ferrite matrix

Metallurgical route for TRIP steels production

Figure 3. Production of TRIP steels

Ac3 Soaking at intercritical temperature

Figure 4. TRIP steels heat treatment

Sheet formability by stamping

Scope of the research

The main objectives of this project are:

This manuscript is divided into four chapters.

Aciers très haute résistance pour application automobile

Figure 1. Application des aciers très haute résistance

Les aciers très haute résistance regroupent trois grandes familles :

- Dual Phase (DP): aciers composés de phases dures (martensite ou martensite/bainite)

- TRIP (TRansformation Induced Plasticity): aciers composés d’une matrice ductile de

Route métallurgique des aciers TRIP

Produit laminé à Produit laminé

Figure 3. Production des aciers TRIP

Ac3 Maintien intercritique

Figure 4. Traitement thermique des aciers TRIP

Mise en forme des tôles par emboutissage

Figure 5. Poutre de pare-choc fissurée au niveau d’un pli après emboutissage

Problématique et démarche de l’étude