Você está na página 1de 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.


Comparison between Gurson and Lemaitre damage models in wiping die

bending processes

Article  in  International Journal of Advanced Manufacturing Technology · January 2004

DOI: 10.1007/s00170-003-1701-3


16 647

3 authors, including:

Ridha Hambli Alain Potiron

Université d'Orléans Ecole Nationale Supérieure d'Arts et Métiers


Some of the authors of this publication are also working on these related projects:

anisotropic multi-scale modeling of the dynamic human bone fracture View project

material parameters identification using new coupled methods View project

All content following this page was uploaded by Ridha Hambli on 03 October 2014.

The user has requested enhancement of the downloaded file.

Int J Adv Manuf Technol (2004) 23: 451–461
DOI 10.1007/s00170-003-1701-3


Ali Mkaddem Æ Ridha Hambli Æ Alain Potiron

Comparison between Gurson and Lemaitre damage models

in wiping die bending processes

Received: 15 November 2002 / Accepted: 13 March 2003 / Published online: 19 December 2003
Ó Springer-Verlag London Limited 2003

Abstract This paper is devoted to a finite element pre- Numerical simulations are widely used in industries
diction of material damage distribution within the to optimise sheet metal forming processes such as
workpiece during wiping die bending processes. The stamping, bending, and hydro-forming. However, in
damage mechanics approach has been used in this order to have confidence in the results of such simula-
investigation in order to describe the progressive damage tions, an accurate material model is required. The
evolution within the sheet. A comparative study between accuracy of a material model is affected by the constit-
the results obtained by the simulations using the utive equations and the values of the material parame-
Lemaitre and Gurson damage models is presented and ters.
discussed. The elastoplastic constitutive laws are inte- To describe the sheet behaviour in a realistic way as
grated by means of an incremental formulation which the operation is carried out, various parameters must be
has been implemented in the finite element code used, such as the material state of hardening and dam-
ABAQUS. The punch load, influenced by the friction age, the metallurgical morphology (which concerns the
coefficient, is investigated for different cases of die ra- shape and size of the grains), the crystallographic tex-
dius. The springback angle, which depends on the elastic ture, and the intergranular structures as they evolve
properties of the material, is computed for all cases. during the operation.
Both models give similar results in the modelling of All the aforementioned physical parameters affect the
bending operations. The Gurson one is shown to have geometrical quality and the mechanical state of the
more flexibility for applications. workpiece. Especially, the states of hardening and
damage conditions of the components have a measur-
Keywords Damage Æ Bending Æ Springback Æ Gurson Æ able effect upon the fatigue life of its future use.
Lemaitre The plastic deformations are localized in the fold and,
generally, more important at surfaces, which leads to
strengthening the part in this zone and may be the cause
of initiation and propagation of cracks as illustrated by
1 Introduction Fig. 1.
In order to accurately simulate sheet metal bending
Sheet metal wiping die bending processes are widely processes, the numerical prediction of the damage evo-
used for mass production. The design of bending pro- lution, crack initiation, and propagation can be de-
cesses is connected with time-consuming and costly scribed by means of a continuum damage approach such
experiments. Therefore, the finite element simulation of as those described by the Lemaitre and Gurson damage
the process could be a helpful tool for the designer and models.
for the quality assurance of the products. In general, the deformation localization and ductile
fracture in metals involve a considerable amount of
texture evolution [1]. The distribution of the crystallo-
graphic texture of a sheet is not homogeneous and the
A. Mkaddem (&) Æ A. Potiron
Ensam/CER, Angers-LPMI-ERT,
plastic deformation and mechanical properties of the
2 Boulevard du Ronceray, BP3525–49035 Angers, France material depend on the evolution of the preferred grain
E-mail: ali.mkaddem@angers.ensam.fr orientation.
R. Hambli
Spherical (Fig. 2a) or elliptical (Fig. 2b) holes, having
Istia-Lasquo, 62 Avenue Notre Dame du Lac, an arbitrary orientational distribution as observed by
49000 Angers, France Tsukrov et al. [2], are one of the main factors governing

Fig. 1 Cracking phenomenon

close to outer fold edges

knowledge of the failure processes is required for the

choice of suitable damage parameters.
In the field of continuum damage mechanics, many
models have been developed by Wang [7, 8],
Chandrakanth et al. [9], and Bonora [10, 11, 12]. These
models have been derived from the concept proposed by
Lemaitre [13] and Chaboche [14, 15] which allows the
prediction of damage evolution under different stress
state conditions.
A conclusion drawn from the above papers is that the
description of the progressive damage evolution leading
to crack initiation and propagation must take into ac-
count the finite element simulation of bending processes.
In such a situation, the concept of continuum damage
mechanics can be applied to describe the damage process
of the material before the final fracture.
In this paper, Gurson [9] and Lemaitres [10] behav-
iour laws coupled with damage have been retained to
simulate the process. To compare the performances of
both theories, a comparative study has been carried out
between the results obtained by the simulation.
After bending, springback occurs upon the removal
of the tool [1, 2, 3, 4]. The amount of elastic recovery
depends on the stress level and the modulus of the
elasticity of the material. Most finite element simulations
of sheet metal bending have focussed on the prediction
of the part springback during air bending processes.
Such efforts typically involve the use of 2D models [2, 5,
6, 7]. However, little work has been done in the past to
Fig. 2 Micro-defects shape observed in a Longitudinal and b
transversal sections of 0.09% sheet metal carbon steel
predict springback during wiping die bending processes.
In order to improve the accuracy of the wiping die
bending process, the material properties variation dur-
the formability of the roll sheets and their behaviour. As ing and after bending must be predicted, especially the
suggested by Gurson [3], void nucleation and growth are damage contour within the sheet.
commonly observed as processes which are character- The accuracy of the springback prediction depends
ized by a large local plastic flow such as ductile fracture. on the accuracy of the constitutive equations and their
As mentioned by Sevostianov et al. [4], porosity corresponding material parameters. Springback is also
enhances plasticity in elastic plastic materials; this phe- sensitive to a range of material and process parameters,
nomenon is observed in aspects of the growth and such as strain hardening [11, 12, 13, 14], evolution of
coalescence of ductile materials and the resulting chan- elastic properties [15, 16], elastic and plastic anisotropy
ges in the overall porosity. Examining the fractured [16], and the presence of a Bauschinger effect [13, 17, 18,
edge, Kalpakjian [5] shows that failure depends on the 19].
micro-defects density, the micro-defects shape, and the The elastic properties variation is the consequence of
strain hardening ratio. As originally suggested by Kra- the damage evolution generated by the plastic defor-
jcinovic [6], the area density of voids and other material mation as stated by Lemaitre [20].
decohesions in a given cross-section provides an intuitive In order to investigate the sensitivity of sheet
characterization of the material damaging. Accurate springback simulations, the Lemaitre and Gurson

damage models have been used in this work, which

allows the study of the influence of the properties
variations on the final quality of the parts.

2 Modelling of the sheet behaviour

Industrial applications like deep drawing, blanking

processes, and bending are all characterized by the
presence of small zones where damage accumulates
further as plastic deformations localize. The correct
description of the mechanical behaviour of the material
in these small zones is important.
In order to predict damage evolution within the sheet, Fig. 3 Hardening Law
the continuum damage mechanics approach has been
applied as a means to describe the workpiece behaviour The derivative form of / gives the rising gradient of
during its forming process. The conditions for initiation the damage noted HG:
and growth of micro-cracks in metals have been studied @/
extensively using micromechanics analysis. It has been HG ¼ ð4Þ
observed from metallurgical test results that ductile
fracture occurs due to void nucleation, growth, and As defined by Crisfield [24], the normal to the frontier
coalescence into micro-cracks. of the yield function is deduced from the derivative form
The application of damage theories allows for the of f by r. it is generally noted by a and written in the
identification of the cause of defects, the prediction of vectorial form:
ductile fracture during forming processes, and the eval-
@f @f 3
uation of the material properties variation generated by a¼ ¼ ¼ Ls ð5Þ
the damage. Such variations have significant effects on @r @s 2req
the fatigue life of the workpieces. where s is the deviator stress and L is a diagonal
In this investigation, Gurson [9] and Lemaitre [10] matrix.
behaviour laws coupled with damage have been retained The normality rule implicates that the plastic strain
to simulate the wiping die bending process. To study the increment is calculated normally to the yield function
performance of both theories, a comparative study has frontier. Moreover, it can be written:
been carried out between the numerical results.
dePl ¼ dePl  a ð6Þ
where dePl is the Lagrange multiplier defined by:
3 Lemaitre model
Pl 2 PlT 1 Pl
The von Mises criterion are taken as the basis of the de ¼ e_ L e_ ð7Þ
theoretical development of the constitutive equations.
h i T
The damage evolution is based on the isotropic and dePl ¼ dePlT ¼ dePl Pl Pl
x ; dey ; dez ; dcxy ; dcxz ; dcyz is
assumption as suggested by different authors [6, 14, 15, the vectorial form of the plastic strain rate.
18, 19, 20, 21, 22, 23]. The yield function, depending on When the material flow occurs, the expression of the
the equivalent stress, the hardening law, the variable D, yield function becomes:
and also the plastic strain, is given by:    
  req   f ¼ G 2~e  3dePl  ry þ R ~ ð1  D Þ ¼ 0 ð8Þ
f req ; ePl ; D ¼  ry  R~ ePl ; D ð1Þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1D T
where G is the shear modulus, ~e ¼ 32 ^e L1^e, and
As illustrated by Fig. 3, the hardening law R~ is expressed Total
^e ¼ eEl
n þ de as they are defined by Hibbit et al. [25].
The nonlinear Eq. 8 can be resolved using the
R ~ ePl n ð2Þ Newton-Raphson iterative method. This approach leads
to the final expression of the plastic corrector as written
where K~ ¼ K ð1  DÞ is the hardening modulus of the in the following relation:
puissance hardening law and n is the hardening expo-
nent. CnPlNEWT
The damage evolution is taken to be linear. It de- G0 ð1  DÞ 2~e  3dePl ~ ePl
n1  ð1  DÞ ry þ R  n1
pends on the equivalent plastic strain and is written as: ¼
3G0 ð1  DÞ þ HD þ ðHG G0 ð2~e  3dePl ÞÞ
D ¼ / ePl ð3Þ ð9Þ

where ~ is an elastic parameter of material and depends on

  the damage effect, as well as on Youngs modulus and
~ ePl
HR ¼ ð10Þ the shear modulus. Then:
~ ð DÞ ¼ K0 ð1  DÞ
K ð24Þ
~ HG
HD ¼ Pl
¼ ð1  DÞHR  ry þ R ð11Þ E0
@e K0 ¼ ð25Þ
3ð1  2mÞ
The theory of isotropic plasticity leads to the rela-
tionship between the deviatoric stress and the deviatoric and
strain such as:
~ ¼ K0 @D ¼ K0 HG @ePl
@K ð26Þ
@s ¼ AD @^e ð15Þ
Taking this last expression into account, Eq. 19 and
where s is the deviatoric form of stress and AD is the Eq. 23 lead to:
matrix defined as presented below:
" ~ m  3K0 HG @ePl em
@rm ¼ 3K@e ð27Þ
req 1
AD ¼ L  <D and then:
! # ~ T @e  K0 HG @ePl jjT e
@r ¼ AD @^e þ Kjj ð28Þ
þ   ssT
2G~e ð1 þ BD Þð1  DÞ þ H3G ð2~e  3dePl Þ It can be demonstrated that the equivalent plastic
strain can be written as:
where BD ¼ H3GD , G ¼ G0 ð1  DÞ is the shear modulus, and 3G^e L1 @^e
@ePl ¼ ð29Þ
<D is the expression given in the next relation: ~e½HD þ 3G þ G0 HG ð2~e þ 3dePl Þ
3 HD Pl ð1DÞ Substituting @ePl with its expression in the relationship
<D ¼ 1 de   of Eq. 28:
2req req ~e ð1þBD Þð1DÞþ H3G ð2~e 3dePl Þ

ð17Þ 1 T ~ T @e
@r ¼ AD I  jj @e þ Kjj
Also, the relationship leading to the calculation of the ^eT L1 @^eð1  DÞ
stress increment from the strain increment can be found.  K0 HG jem  
It is already well known that: ~e ð1 þ BD Þð1  DÞ þ H3G ð2~e  3dePl Þ
e ¼ e þ jem ; e ¼ ^e ð18Þ
where After some modifications, the last term has to be
modified so that de will appear. Then a new expression is
1 1  found in such a manner that:
em ¼ j T e ¼ ex þ ey þ ez ð19Þ
3 3  T  
1 1
and jT ¼ ð111000Þ. jem eT L1 @e ¼ jjT eeT I  jjT L1 I  jjT @e ð31Þ
3 3
It can also be written as follows:
1 The substitution of Eq. 31 in Eq. 30 transforms it
e ¼ e þ jjT e ð20Þ into:
1 T ~ T @e
The differentiation of the next relation gives: @r ¼ AD I  jj @e þ Kjj
1 T
@^e ¼ I  jj @e ð21Þ K0 ð1  DÞHG
~e ð1 þ BD Þð1  DÞ þ H3G ð2~e  3dePl Þ
where I is the unit matrix.
On the other hand, we have:    
T T 1 T T 1 1 T
 jj ee I  jj L I  jj @e ð32Þ
r ¼ s þ jrm ð22Þ 3 3
where The expression above relayed the stress increment to
1 the strain increment thanks to the consistent tangent
rm ¼ jT r ¼ 3KðDÞe m ð23Þ operator KtgD given in the following expression:
~ G
KH plastic strain. q1, q2, q3 are material parameters. The
KtgD ¼ AD    ranges of values reported in the literature [27] and [29],
~e ð1 þ BD Þð1  DÞ þ H3G ð2~e  3dePl Þ as specified for typical metals, are q1=1.0–1.5, q2=1.0,
  !  #
1 T T 1 1 T and q3=q12=1.0–2.25. The original Gurson model is
jj ee I  jj L I  jj þ Kjj ~ T
recovered when q1=q2=q3=1.0. The plastic strain is
3 3 assumed to be normal to the yield function:
ð33Þ @U
e_ Pl ¼ e_ Pl ð36Þ
The hardening of the fully dense matrix material is
4 Gurson model described by r0 ¼ r0 e_ Pl . The evolution of the equiva-
lent plastic strain in the matrix material is obtained from
The concept of continuum damage mechanics as ob- the following equivalent plastic work expression:
served by Kachanov [26] is based on the isotropic dis-
tribution of the spherical voids existing in the matrix ð1  f Þr0e_ Pl ¼ r : e_ Pl ð37Þ
material. Under the critical conditions of loading, the
growth of voids is inevitable when material flow occurs. The total change in void volume fraction is given as:
Later, the increase of micro-defects density leads to a
cracking phenomenon and consequently to failure. The f_ ¼ f_gr þ f_nucl ð38Þ
relationships defining the model are expressed in terms where f_gr expressed the growth of existing voids and
of the void volume fraction, named f [27]. It is defined as f_nucl expressed the nucleation of new voids. Growth of
the ratio of the volume of voids to the total volume of the existing voids is based on the law of conservation of
the material. It follows that f=1)r, where r is the rela- mass and is expressed in terms of the voids fraction:
tive density of a material defined as the ratio of the
volume of solid material to the total volume of material. f_gr ¼ ð1  f Þe_Pl : I ð39Þ
f=0 implies that material is fully dense and f=1 implies
that the material is completely void and has no stress- The nucleation of voids is given by a strain-controlled
carrying capacity. Gurson [3] proposed a yield condition relationship:
as a function of the void volume fraction. Later,
Tvergaard [28] modified this yield condition and trans- f_nucl ¼ Ae_ Pl ð40Þ
formed it into the following form: where A is an exponential function of ePl and the volume
 2   fraction of the nucleated voids is as given in the fol-
req 3rH  
U¼ þ2q1 f cosh q2  1 þ q3 f 2 ¼ 0; lowing expression:
ry 2ry 
ð34Þ Pl
1 e eN
fN sN
A ¼ pffiffiffiffiffiffi e ð41Þ
where sN 2p
1 The normal distribution of the nucleation strain has a
rH ¼  r : I ð35Þ
3 mean value eN and a standard deviation sN. fN is the
 Pl   Pl 
and r0 e ¼ ry þ R~ e is the yield stress of the fully volume fraction of the nucleated voids during the ten-
dense matrix material as a function of the equivalent sion case. The complete list of parameters used for the
Gurson model simulation is recapitulated in Table 1.
Table 1 Gurson parameters used for simulation Table 2 gives a comparison between the Lemaitre and
Gurson damage models. The total number of parame-
q1 q2 q3 r0 eN sN fN
ters is computed for each model.
1.50 1.10 2.25 1 0.10 0.05 0.04 Gurson flow stress (Eq. 34) depends on a scalar
damage variable f describing the void volume fraction.
Table 2 Comparison between Lemaitre and Gurson damage models

Lemaitre model Gurson model

Damage D ¼ A ATotal f_ The growth rate of the micro-void volume fraction f_ ¼ f_gr þ f_nucl
D ¼ / ePl  
Pl rH
Variables e ; req Tr e_Pl
: First
of the plastic
 strain rate tensor
  n  req 2 3rH 2

Yield function f ¼ req  ð1  DÞ ry þ K ePl ¼0 U ¼ ry þ2q1 f cosh q2 2ry  1 þ q3 f ¼ 0
Youngs Modulus E=E0(1)D) (variable) E=E0 (fixed)
Application Shearing processes Tensile and compressive processes
Damage parameters 5 parameters 7 parameters

The model allows for the damage prediction of porous element (F.E.) mesh, including element splitting, sepa-
materials. The normality assumption ensures that plastic rating of nodes, element deletion, and stiffness F.E.
dilation can occur if both the mean stress and damage decreasing. This last technique has been used in this
values are non-zero. The first invariant of the plastic work in order to simulate crack propagation. During the
strain rate tensor Tr e_Pl is the main variable of the analysis, the initiation of a crack is assumed to occur at
damage evolution (Eq. 39). any point in the structure where the damage reaches its
In the past few years several different approaches critical value Dc. The crack propagation is simulated by
were developed and applied to the analysis of ductile the propagation of a completely damaged area. From a
fracture, commonly known as the coupled damage numerical point of view, the damage value is abruptly set
approach, incorporating the accumulation of damage in to the value DR (DR1). This method leads to a decrease
the constitutive equations. The first attempts in this field in the stiffness of the concerned elements.
of research were made by Gurson who developed a
ductile damage macroscopic constitutive law based on
micro-void growth and by Tvergaard [28] who improved 5 Simulation of wiping die bending operation
Gursons model by including some coalescence effects.
These initial investigations and other subsequent devel- The bent part characteristics are affected by several
opments enabled the development of a continuous factors. In this study, only the influence of the die shape
damage mechanics model for ductile fracture by is treated. The material parameters obtained by tensile
Lemaı̂tre. The main advantage of the coupled damage tests are recapitulated in Table 3.
approach is that it allows the yield surface of the The meshing of the part is carried out by using tri-
material to be modified by the density changes and void angular and quadrangular node continuum elements.
growth induced by the accumulation of damage. 1192 elements are used for the mesh. The tool (punch,
The finite element model, using Gursons yield crite- die, and blank holder) is modelled by adopting analyti-
rion, allows for the prediction of the damage evolution cal rigid bodies. The mesh region and the bending
during the plastic loading of structures. The model geometry are shown in Fig. 4.
assumes that throughout the process, the matrix mate- The data used in the numerical modelling are
rial work hardens according to the power law given in reported in Table 4. All parameters remain constant
Eq. 34. The Gurson model assumes that the total vol- except for the die radius.
ume of voids grows, consequently weakening the mate- Rp and Rd are the punch and the die radius, respec-
rial. However, one should be aware of the fact that the tively. J is the clearance between die and punch, C is the
work hardening rate may change at larger strains punch stroke according to lower springback angles in
beyond the start of necking. The model simulates a industrial press machines, a is the bending angle, and t is
weakening of the material due to the propagation of the sheet thickness.
damage but does not include a further failure (cracking)
of the blank material.
The literature values for the model parameters (q1, q2, Table 3 Elastic and hardening parameters
q3, eN, sN, fN, r0) for typical metals were used. These
E0 ry ~k n m0 HG
values showed sufficient similarity between the simulated
GPa MPa MPa – – –
and the actual results from the tensile test [31].
It has to be remarked that in the constitutive frame- 200 560 800 0.745 0.28 0.4
work proposed by Lemaitre, damage affects internal
associated variables, such as strain, the yield condition,
and Youngs modulus. According to this approach,
when damage reaches its critical value, the material
stiffness is reduced to zero. Contrary to the Gurson
approach, the Lemaitre approach allows for the simu-
lation of crack initiation and propagation within the
When the critical Lemaitre damage value is satisfied
within an element, the element fractures and cracks
occur. The direction of crack propagation and the crack
tip position are then determined by the value contour of
the fracture or damage value at each element of the
For the simulation of processes where rupture is to be
considered, it is necessary to develop and apply methods
using mechanical fracture models where a material sep-
aration must be predicted. There are at least four pos-
sible methods to simulate a crack propagation in a finite Fig. 4 Mesh used for the fold zone

Table 4 Process parameters for numerical model The bending force curve versus punch penetration is
affected by the damage of the sheet. The results show
RP Rd t J C a
mm mm mm mm mm (°) that the curves are superimposed until a punch pene-
tration of about 7 mm, and there is a significant differ-
4 2-3-4-5-6 4 4 29 90 ence between both evolutions for higher punch
In Fig. 6a it can be observed that the punch load
Table 5 Friction coefficient for the contact modelling [32] increases sharply, especially for low values of the die
radius, reaching a maximum value, and then the force
Surfaces Punch Die Blank holder
decreases rapidly and similarly. The curves fit well to-
Blank top 0.09 – 0.15 gether with a higher maximum force without damage
Blank bot – 0.15 – influence simulations.
Figure 7 shows the evolution of maximum bending
force versus the die corner radius obtained by the dif-
The contact model is simulated by the coulomb fric- ferent simulations; without damage influence and, also,
tion law. In industrial processes, lubrication is mostly using Gurson and Lemaitre models. The punch pene-
used in order to reduce the tools wear and friction. For tration corresponding to the maximum force values for
this reason, relative sliding between punch and blank is each configuration is also reported in Fig. 7.
modelled by adopting a low friction coefficient [32]. The It can be observed that the predicted maximum
l-coefficients used for numerical modelling are reported bending force decreases with an increasing corner die
in Table 5. radius. The curves have the same trend; nevertheless,
The computation results, corresponding to different there is a significant difference between curves with and
displacement steps of the punch penetration are pre- without damage, especially for low values of the die
sented in Fig. 5. It can be seen that the springback radii.
takes place when the punch is removed from the sim- Figure 7 shows that the damage affects the maximum
ulation. bending force especially for a small value of the die
radius. As observed, the difference between Lemaitre
and Gurson models does not reach 2.5%.
6 Simulation results

6.1 Punch load 6.2 Springback

The punch load is investigated during the bending Corrections for springback are essential during die de-
operation. The numerical calculation using the Lemaitre sign in order to obtain specified final shapes.
and Gurson models are superimposed as shown in Figure 8 shows the effects of the die radius on the
Fig. 6. springback angle obtained by simulation with and
Figure 6a describes the load evolution of the bending without damage influence. When damage is accounted
process with various die radii obtained by finite element for, it can be observed that the springback is lower as a
simulation. At the initial stage of the bending process, consequence of the material parameters variation,
material first flows at the fold zone. With the punch especially the elastic modulus and the strain within the
force, the blank comes into contact with the punch and fibres. The difference between the curves decreases with
is drawn into the die. Although a large displacement the increase of the die radius. The decrease is attributed
occurs in the corner, the large deformation is located in to the damage reduction within the sheet for higher die
this zone. A severe gradient of stress exists in this part radius values.
when tensile strain outside the fold is excessive [33]. The springback angle increases rapidly against the die
Excessive tensile strain will causes a high damage dis- radius. The simulation results computed by both models
tribution due to the growth and nucleation of micro give similar results with a deviation less then 4%.
defects. However, the nucleation of voids is rather large
in comparison to the growth of voids as can be seen in
Fig. 6b. 6.3 Damage evolution

Damage evolution versus arc width curves for the wiping

die bending process are shown in Fig. 9. The damage is
computed at the outer surface of the fold zone. Both
damage curves, plotted for a die radius of 2 mm, evolve
similarly, reaching their maximum in the central zone of
the fold and then fall to zero.
The Lemaitre formulation gives a larger value than
Fig. 5 Deformed configuration at different steps the Gurson one. The difference noticed especially at the

Fig. 6 a Punch force evolution

and b Damage distribution
using the Gurson model and the
Lemaitre one when the die
radius changes from 2 mm to
6 mm during the bending

Fig. 8 Springback angle computed with Lemaitre and Gurson


can be noted that for low values of the die radius, the
Fig. 7 Maximum load versus punch penetration for different springback error between the Lemaitre model and the
values of the die radius Gurson one increases sharply, whereas it decreases for
bending force. For high values of the die radius, the
maximum value is essentially due to the choice of bending force evolves similarly for both models. The
damage parameters for modelling. difference in springback angle becomes considerable,
In order to evaluate the reliability of each method, reaches a maximum, and falls slightly as can be observed
Lemaitre and Gurson, error is quantified using the fol- in Fig. 10.
lowing relations: In spite of the difference, which is essentially due to
the choice of problem parameters particularly for the
jhLem  hGur j Gurson simulation, it can be concluded that the two
vh ð%Þ ¼
hLem models give similar results. The maximum error reached
Punch is 3.7%.
F Punch
Lem  FGur
vF punch ð%Þ ¼ Punch
7 Discussion and comparison
vh is the error associated with the spring back angle
obtained by means of the two models and vF Punch is the The comparative study of numerical computation using
error associated with the punch load during bending. It Gurson and Lemaitre models shows that finite element

Fig. 9 Damage evolution at the

outer fold zone for Rd=2 mm

From a practical point of view, the simulation can

provide useful information about the influence of the
tools design, especially the die shape, on the quality of
the final product.

8 Conclusion

In this work, a numerical simulation of wiping die

bending processes using the continuum damage
mechanics approach has been applied in the modelling
of the operation. The Lemaitre damage model was
implemented in a finite element code allowing for the
description of the damage evolution within the sheet.
The comparative study between the results obtained
by the simulations using Gurson and Lemaitre models
Fig. 10 Springback and punch load errors between the Lemaitre
and Gurson models showed that the material behaviour has to be accurately
known, especially in accounting for damage evolution
and rupture simulation. Both models give similar results
on springback prediction with a deviation less than 4%.
Table 6 Increment number required for convergence The 2D simulation provides a good modelling tool
Rd (mm) 2 3 4 5 6 for the sheet metal forming problems where it is not
necessary to go into more detail. Stretching phenomena
Lemaı̂tre model 247 223 218 219 233 and the boards effects are well known in a bent product,
Gurson model 243 210 216 211 227 and they cannot be observed by adopting a 2D simula-
tion. In fact, a 3D simulation of wiping die bending
operations becomes necessary in order to predict real-
modelling using the Lemaitre damage law in describing istically the final state of damage, bending part aptitude,
the operation of wiping die bending is more costly. and the geometry of the final product.
Incidentally, the total increment number computed
when the Lemaitre model is adopted is higher than the Acknowledgments The authors would like to thank Deville S.A
one required for the convergence using Gurson model. Industry for its technical support.
In fact, the CPUS(Computer Processor Units)-time,
which is tied to the increment number, is more consid-
erable when the Lemaitre model is used for numerical References
modelling. Numerical results are reported for two
models in Table 6. 1. Tang CY, Tai WH (2000) Material damage and forming limits
However, Lemaitre assumes that damage evolves in of textured sheet metals. J Mater Process Technol 99:185–140
the same way under a state of compression as well as 2. Tsukrov I, Kachanov M (2000) Effective moduli of an aniso-
tropic material with elliptical holes of arbitrary orientational
under a state of tension. Thus, to apply Lemaitre law to distribution. Int J Solids Struct 37:5919–5941
describe sufficiently and realistically material damage 3. Gurson AL (1977) Continuum theory of ductile rupture by
and response during bending, damage must be deacti- void nucleation and growth, part I, yield criteria and flow rules
vated for a negative value of hydrostatic stress (Eq. 23). for porous ductile media. Trans ASME, J Eng Mater Technol
Otherwise, simulation can lead to nonrealistic predic- 4. Sevostianov I, Kachanov M (2001) On the yield condition for
tions in wiping die bending processes as shown in [34]. anisotropic porous materials. Mater Sci Eng A313:1–15
In addition, damage parameters are very difficult to 5. Kalpakjian S (1991) Manufacturing process for engineering
describe by experimental methods, which are the only materials, 2nd edn, Addison-Wesley, USA
way to do so for Lemaitre damage characterisation. 6. Krajcinovic D (1983) Constitutive equations for damaging
materials. J Appl Mech 50:355–360
Gurson ones, on the other hand, are adjusted theoreti- 7. Wang TJ (1992) Unified CDM model and local criterion for
cally and the choice of parameters is more flexible ductile fracture-I, unified CDM model for ductile fracture. Eng
compared to Lemaitre ones. Fract Mech 42:177–183
As found by Hambli [30], the Gurson model is not 8. Wang TJ (1994) Further investigation of a new continuum
damage mechanics criterion for ductile fracture: experimental
able to predict the fracture propagation in a realistic way verification and applications. Eng Fract Mech 48:217–230
during shearing processes; he showed that only the Le- 9. Chandrakanth S, Pandey PC (1995) An isotropic damage
maitre damage law can give good results. Throughout model for ductile material. Eng Fract Mech 50:457–465
the shearing processes, the Lemaitre model can also be 10. Bonora N (1997) A linear CDM model for ductile failure. Eng
Fract Mech 58:11–28
used in wiping die bending processes when carefully 11. Bonora N (1988) Low cycle fatigue life estimation for ductile
applied. Crack initiation and propagation can be accu- metals using a nonlinear continuum damage mechanics model.
rately predicted by means of the Gurson damage model. Int J Struct 35:1881–1894

12. Bonora N, Salvini P, Lecoviello F (1996) Experimental iden- 25. Hibbit K, Sorensen Inc (1995) Mechanical constitutive theories.
tification of damage evolution law in Al-Li2091 alloy. XXV In: ABAQUS theory manual, version 5.5, pp–
AIAS National meeting, Galipoli, pp 345–355 26. Kachanov LM (1986) Introduction to continuum damage
13. Lemaitre J, Chaboche JL (1988) Endommagement, mécanique mechanics, mechanics of elastic stability. Kluwer, USA
des matériau solides. Dunod, Paris 27. Hibbit K (2001) Porous metal plasticity. In: ABAQUS/Stan-
14. Chaboche JL (1988) Continuum damage mechanics. Part I: dard users manual, vol II, version 6.2, Sorensen Inc, pp 11.2.7–
general concepts. J Appl Mech 55:59–64 1–11.2.7–6
15. Chaboche JL (1988) Continuum damage mechanics. Part II: 28. Tvergaart V (1991) Mechanical modelling of ductile fracture.
damage growth, crack initiation, and crack growth. J Appl Mech 26:11–16
Mech 55:65–72 29. Hambli R, Potiron A (2001) Damage and fracture prediction in
16. Ye D, Wang Y (2001) An approach to investigate pre-nucle- bending process using 3D finite element model. In: Proceedings
ation fatigue damage of cyclically loaded metals using Vickers of the 9th international conference on sheet metal, Leuven,
micro hardness tests. Int J Fatigue 23:85–91 Belgium, pp 229–236
17. Widmark M, Melander A, Meurling F (2000) Low cycle con- 30. Hambli R, Potiron A (2001) Comparison between Lemaitre
stant amplitude fully reversed strain controlled testing of low and Gurson damage models in crack growth simulation during
carbon and stainless sheet steels for simulation of straightening blanking process. In: Proceedings of the 9th international
operations. Int J Fatigue 22:307–317 conference on sheet metal, Leuven, Belgium, pp 555–562
18. Simo JC, Taylor RL (1985) Consistent tangent operator for 31. Klingenberg W, Singh UP (2003) Finite element simulation of
rate independent elastoplasticity. J Comput Meth Appl Mech the punching/blanking process using in-process characterisa-
Eng 48:101–118 tion of mild steel. J Mater Process Technol 134(March):296–
19. Simo JC, Ju JW (1987) Strain and stress based continuum 302
damage models I: formulation. Int J Solids Struct 7:821–840 32. Magny C (2002) Lois de frottement évolutives destinées à la
20. Simo JC, Ju JW (1987) Strain and stress based continuum simulation numérique de lemboutissage. La Revue de Métal-
damage models II: computational aspects. Int J Solids Struct lurgie – CIT/Science et Génie des Matériaux, pp145–156
23:841–869 33. LP Lei, SM Hwang, BS Kang (2000) Finite element analysis
21. Doghri I (1993) Fully implicit integration and consistent tan- and design in stainless steel sheet forming and its experimental
gent modulus in elastoplasticity. Int J Numerical Methods Eng comparison. J Mater Process Technol 110:70–77
36:3915–3932 34. Mkaddem A, Huneau B, Lebrun JL, Boude S (2003) Experi-
22. Doghri I (1995) Numerical implementation and analysis of a mental quantification of damage in sheet metal bending pro-
class of metal plasticity models coupled with ductile damage. cess: comparison with numerical prediction. In: Proceedings of
Int J Numerical Methods Eng 38:3403–3431 the 10th international conference on sheet metal, Belfast, North
23. Murakami S (1998) Mechanical modelling of material damage. Ireland
Trans ASME, J Appl Mech 55:280–286
24. Crisfield MA (1991) Basic plasticity, non-linear finite element
analysis of solids and structures, Mc-Graw Hill, Great Britain,
pp 152–199

View publication stats