Failure Assessment in Sheet Metal Forming Using A

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Available online

Available online at
at www.sciencedirect.com
www.sciencedirect.com

ScienceDirect
Available online at www.sciencedirect.com
Procedia
Procedia Engineering
Engineering 00
00 (2017)
(2017) 000–000
000–000
www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia
ScienceDirect
Procedia Engineering 207 (2017) 2066–2071

International Conference on the Technology of Plasticity, ICTP 2017, 17-22 September 2017,
Cambridge, United Kingdom

Failure assessment in sheet metal forming using a


phenomenological damage model and fracture criterion:
experiments, parameter identification and validation
Sebastian Heibelaa*, Winfried Nesteraa, Till Clausmeyerbb, A. Erman Tekkayabb
aa
Process Engineering
Process Engineering and
and Materials,
Materials, Mercedes-Benz
Mercedes-Benz Cars,
Cars, Benzstraße,
Benzstraße, 71059
71059 Sindelfingen,
Sindelfingen, Germany
Germany
Institute of Forming Technology
Institute of Forming Technology and
and Lightweight
Lightweight Design,
Design, TU
TU Dortmund
Dortmund University,
University, Baroper
Baroper Str.
Str. 303,
303, 44227
44227 Dortmund,
Dortmund, Germany
b
b
Germany

Abstract
Abstract

In
In this
this contribution
contribution microstructural
microstructural andand macroscopic
macroscopic experimental
experimental findings
findings are
are used
used to
to calibrate
calibrate the
the phenomenological
phenomenological damagedamage
model
model GISSMO (Generalized Incremental Stress State dependent Model) and the fracture criterion FFL/SFFL (Fracture
GISSMO (Generalized Incremental Stress State dependent Model) and the fracture criterion FFL/SFFL (Fracture Forming
Forming
Limit line/
Limit line/ Shear
Shear Fracture
Fracture Forming
Forming Limit
Limit line)
line) toto assess
assess failure
failure in
in sheet
sheet metal
metal forming
forming simulation.
simulation. It It is
is shown
shown that
that macroscopic
macroscopic
failure in
failure in aa commercial
commercial dual-phase
dual-phase steel
steel is
is initiated
initiated through
through aa combination
combination of of ductile
ductile damage
damage mechanisms
mechanisms and and local
local shear
shear banding
banding
at aa characteristic,
at characteristic, stress-state
stress-state dependent
dependent optical
optical or or tactile
tactile measureable
measureable fracture
fracture strain.
strain. The
The parameter
parameter identification
identification for
for both
both models
models
is based on ductile fracture experiments representing characteristic stress states. GISSMO is calibrated inversely
is based on ductile fracture experiments representing characteristic stress states. GISSMO is calibrated inversely using experimental using experimental
stress-strain curves,
stress-strain curves, optical
optical measured
measured fracture
fracture strains
strains and
and simulation
simulation data.
data. The
The FFL
FFL and
and SFFL
SFFL are
are constructed
constructed in in aa direct
direct manner
manner
with the
with the optically
optically and
and tactilely
tactilely measured
measured fracture
fracture strains.
strains. Both
Both models
models are are validated
validated comparatively
comparatively on on aa cross-die
cross-die cup
cup showing
showing
ductile fracture
ductile fracture with
with slight
slight necking.
necking. The
The use
use ofof the
the fracture
fracture criterion
criterion in
in combination
combination withwith aa direct
direct method
method of of determining
determining thethe fracture
fracture
lines based on tactile strain measurements leads to an overestimation of the instant of fracture initiation.
lines based on tactile strain measurements leads to an overestimation of the instant of fracture initiation. With the inversely With the inversely
identified parameters
identified parameters of of the
the phenomenological
phenomenological damage damage model
model the
the onset
onset of of fracture
fracture initiation
initiation can
can be
be accurately
accurately predicted.
predicted.
© 2017
© 2017 TheThe Authors.
Authors. Published
Published byby Elsevier
Elsevier Ltd.
Ltd.
© 2017 The Authors. Published by ofElsevier Ltd.
Peer-review
Peer-review
Peer-review under
under responsibility
under responsibility
responsibility ofof thescientific
the scientific committee
committee of theof the International
International Conference Conference
on the Technologyon theofTechnology
Plasticity.
of Plasticity..
Keywords:
Keywords: Phenomenological
Phenomenological damage
damage modelling;
modelling; fracture
fracture criterion;
criterion; damage
damage mechanisms;
mechanisms; failure
failure assessment;
assessment; sheet
sheet metal
metal forming;
forming; AHSS
AHSS

*
* Corresponding
Corresponding author.
author. Tel.:
Tel.: +49
+49 151
151 58627508;
58627508; fax:
fax: +49
+49 711
711 3052147161.
3052147161.
E-mail address:
E-mail address: sebastian.heibel@daimler.com
sebastian.heibel@daimler.com

1877-7058
1877-7058 ©© 2017
2017 The
The Authors.
Authors. Published
Published by
by Elsevier
Elsevier Ltd.
Ltd.
Peer-review under
Peer-review under responsibility of the scientific committee
responsibility of of the International Conference on the Technology of
Plasticity..

1877-7058 © 2017 The Authors. Published by Elsevier Ltd.


Peer-review under responsibility of the scientific committee of the International Conference on the Technology of Plasticity.
10.1016/j.proeng.2017.10.1065
Sebastian Heibel et al. / Procedia Engineering 207 (2017) 2066–2071 2067
2 Heibel et al./ Procedia Engineering 00 (2017) 000–000

1. Introduction

Advanced High Strength Steels (AHSS) are widely used in modern car body manufacture to achieve a reduction in
vehicle weight while increasing the crash performance. Higher strength is accompanied inevitably by a reduction of
ductility and formability. AHSS exhibit the classical stress-state depending overlapping stages of ductile damage
evolution: void nucleation, growth and coalescence. From a mechanical point of view the void growth is controlled by
hydrostatic stresses, represented by the stress triaxiality, whereas the shape is influenced by the deviatoric component
of the stress tensor, characterized by the Lode-parameter. From a materials point of view the respective damage
mechanisms are determined by the heterogeneity of the microstructure. The difference in hardness between the phases
leads to a heterogeneous strain distribution and thus to damage evolution [1]. During sheet metal forming this damage
evolution might lead to fractures with slight necking, fractures on tight radii or edge-fractures. To assess these damage
and failure mechanisms modelling approaches on different scales are presented in literature [1]. The sheet metal
forming simulation on the macroscopic component scale is usually based on an appropriate hardening law and yield
locus as well as on the prediction of the forming limits in terms of onset of localized necking with the FLC (Forming
Limit Curve) in the principal strain space. This failure criterion is a purely experimental approach and thus only valid
for proportional loading paths in the range between uniaxial and (equi-)biaxial tension. The FLC is not suitable to
predict shear fracture, fracture under bending or fracture with an underlying complex loading history or edge-fractures.
Consequently the FLC is often applied with high safety margins and thus the material capacity of AHSS is not fully
used. To overcome these drawbacks, special emphasis is placed on phenomenological damage models and advanced
fracture criteria in research and industry. These models incorporate the dependency of damage evolution and of ductile
fracture on the stress state by formulations depending of the fracture strain in the space of stress triaxiality and the
Lode-parameter. The loading-path-dependency is considered through the incremental accumulation of a damage
variable, which is controlled by evolution of the equivalent plastic strain. Advanced fracture criteria like the Xue-
Wierzbicki [2], modified Mohr-Coulomb [3], Lou-Huh [4] or the Hosford-Coulomb model [5] are able to represent a
fracture surface with a monotonic decreasing dependency of the fracture strain on stress triaxiality and a convex
relationship between fracture strain and Lode-parameter. Using a three-dimensional modelling approach in finite-
element analysis (FEA) they show good results in predicting ductile fracture. However the quality of failure assessment
is hereby strongly dependent on the ductile fracture experiments. Furthermore, the three-dimensional modelling
approach is not applicable in sheet metal forming simulation in an industrial context, where the use of shell elements
is standard. Assuming the plane-stress condition the fracture surface can be reduced to a fracture curve depending only
on the stress triaxiality. In this context the phenomenological damage model GISSMO (Generalized Incremental Stress
State dependent Model) was introduced by Neukamm et al. [6]. The model is based on an incremental damage
accumulation which depends on an arbitrary failure curve. Like the phenomenological Lemaitre model [7] GISSMO
takes up the idea of a reduction of the load bearing cross-section due to damage evolution in coupling the stress tensor
with the damage variable. For application in crash simulation GISSMO is calibrated inversely on the basis of stress-
strain curves for several stress-states [8]. Another approach assuming plane-stress condition is the use of fracture lines
in principal strain space. Following this approach Martins et al. merged the idea of a damage variable based on stress
triaxiality and fracture strains with the sheet metal forming observations of stress-state independency of fracture
thickness in the concept of FFL/SFFL (Fracture Forming Limit line/ Shear Fracture Forming Limit line) [9]. These
fracture lines can be constructed in a direct manner with a minimum of three ductile fracture experiments. Within this
work the damage and failure behavior of a commercial produced dual-phase steel with a tensile strength of 1.00 GPa
is investigated. On this basis two strategies for failure prediction in sheet metal forming are comparatively assessed,
GISSMO with an inverse parameter identification scheme and the concept of the FFL/SFFL, where the experimental
measured fracture strains are directly applied in simulation.

2. Failure modelling

Experimental investigations by Isik et al. show that for an aluminum alloy ductile tensile fracture (Fracture Mode
I) occurs at a constant through thickness strain at fracture  3 f , regardless of the loading history before necking [10].
Accordingly Martins et al. proposed the FFL in principal strain space falling from left to right with a slope of -1.
2068 Sebastian Heibel et al. / Procedia Engineering 207 (2017) 2066–2071
Heibel et al./ Procedia Engineering 00 (2017) 000–000 3

(a) ST 1 (b) f
UAT

ST UAT PST BAT


PST BAT

2 
0 13 1 3 23

Fig. 1. Ductile fracture specimens at the respective stress states in principle strain (a) and equivalent plastic strain to failure – triaxiality space (b)

Perpendicular to the FFL the shear fracture forming limit line can be constructed. Hereby the major  1 f and minor  2 f
in-plane strains and distortions  f reach critical values leading to in-plane shear fracture (Fracture Mode II) [9].

1 f   2 f   3 f (FFL) and 1 f   2 f   f (SFFL) (1)

Combined with damage mechanics and using Hill48’s anisotropic yield locus a fracture criterion can be formulated

1  rm   (FFL) and D s   12 d  1 1  rm     
f f
m
Dcrit  
0

d  
3
3f crit
0
 
2 1  2rm 
1f 2f
(SFFL) (2)

where Dcrit and Dcrits


are the critical damage variables,  is the equivalent plastic strain, rm is the normal anisotropy
coefficient,  m the mean stress and  the equivalent stress. The ratio  m  is defined as the stress triaxiality  .
Because of the independency of the initial loading history Dcrit can be interpreted as characteristic material value. [9]
Contrarily, GISSMO is based on an incremental accumulation of the damage variable D . The damage
accumulation depends at every stage of the simulation on the current triaxiality, the respective fracture strain and on
the damage exponent n . [8]

~   D  Dins  
m
n n 1

D  D n
d and    1     (3)
 f     1  Dins  
 

Failure is initiated for D  1 . The stress tensor is affected by damage when Dins is reached. Similar to the fracture
curve  f   , Din s is named instability curve. The coupling between damage and an arbitrary plasticity model through
~
introduction of effective stresses  can be controlled via the fading exponent m . For D  0 and m =1 the classical
principle of effective stress is invoked [7]. Detailed information about the models can be found in [8, 9].

3. Experiments

GISSMO as well as the FFL/SFFL are calibrated using experimental data of the DP steel (DP1000) with a sheet
thickness of t 0  0.98 mm from four specimens covering a triaxiality range between simple shear and biaxial tension:
a shear tensile specimen (ST), a uniaxial tensile specimen (UAT), a notched tensile specimen for plane-strain tension
(PST) and an (equi-)biaxial tensile specimen (BAT). The specimens are illustrated in Fig. 1. The geometries are
detailed in [8] and [11]. The tests are conducted with the digital image correlation system ARAMIS for optical strain
measurements. To determine the local failure strain a strain gauge length of 0.50 mm is adjusted and the images are
recorded with a frequency of 10 Hz.
4 Sebastian Heibel
Heibel et al./ et al. Engineering
Procedia / Procedia Engineering 207 (2017) 2066–2071
00 (2017) 000–000 2069

The DP steels microstructure is characterized via electron backscatter diffraction analysis (EBSD). Composed of
about 27 % ferrite, 60 % bainite/ tempered martensite, 10 % martensite and 2 % retained austenite with an average
ferrite grain size of 1.26 µm the multiphase microstructure leads to the mechanical properties in longitudinal direction
(LD) displayed in Table 1. These mechanical properties are determined after SEP1240 with specimen geometry
according to EN 10002-1 appendix B shape 2.

Table 1. Mechanical properties of DP1000 longitudinal to rolling direction (RD)


RD Rp0.2 [GPa] Rm [GPa] Rp0.2/ Rm AG [%] A80 [%] n2-A [-]
G
rm[-]
LD 0.708 1.060 0.67 9 14 0.126 0.957

The damage mechanisms of the DP1000 leading to ductile fracture are outlined exemplarily on the notched tensile
specimen. Therefore scanning electron microscopy (SEM) pictures are shown in Fig. 2. Voids are primarily elongated
in loading direction and can be mainly detected between fractured martensite particles (4) or at martensite-ferrite grain
boundaries (5). Both nucleation sites are strongly dependent on the local microstructural morphology. Void
coalescence can be identified along a martensite row (6). The damage evolution leads to void area fractions of about
1.00 %. As a result of the interplay between ductile damage mechanisms in the center of the specimen (1) and material
separation in the form of decohesion at phase boundaries or particle fracture due to local shear banding (2,3) a
A B C
(5)
(5)
45 (4) (4)
B C
(3) (6)
A (1)
(5)
(5)
(2) (4)
(4)
ND
200 µm 10 µm 5 µm (5) 5 µm
LD

Fig. 2. SEM pictures of the fracture zone of the nital-etched microstructure of the DP1000 cut out of a notched tensile specimen

combination of tensile fracture (Mode I) and out-of-plane shear fracture (Mode III) can be detected. The fracture
surface shows characteristic for shear bandingangles of 45° to longitudinal direction.
A considerable strain gradient in thickness direction develops during localized necking for stress states between
uniaxial tension and biaxial tension. This strain gradient depends on the material, the sheet thickness and on the
respective stress state and leads to differences in optical and tactile strain measurement. For the latter the reduction in
sheet thickness is measured on both parts of the fractured specimen (Fig. 3 (a)). Thus  3 f is averaged about the
thickness. Via optical strain measurements the strain field on the surface can be determined (Fig. 3 (b)). As such, the
respective values depend additionally on the chosen gauge length and on the image rate. The resulting stress-state
dependent fracture strains are summarized in (Fig. 3 (c)). In contrast to the UAT and PST specimen, the BAT specimen
exhibit minor localized necking. The strain in thickness direction is more homogenous and thus the fracture strain is
higher. The ST specimen shows an in-plane shear fracture (Mode II) with almost no reduction of sheet thickness
 3 f  0 . Hence, for this specimen the fracture strains are measured just optically on its surface.

(a) (b) Stage 233 f Stage 234 (c) optical tactile


Stress 3 f f 3 f f
State
ST - 0.73 - -
UAT -0.35 0.51 -0.51 0.69
1 mm PST -0.27 0.34 -0.50 0.60
BAT -0.54 0.54 -0.68 0.68

Fig. 3. Measurement of thickness reduction (a), optical strain measurement (b), and comparison of optically and tactilely measured strains (c)
2070 Sebastian Heibel et al. / Procedia Engineering 207 (2017) 2066–2071
Heibel et al./ Procedia Engineering 00 (2017) 000–000 5

4. Parameter identification

To compare the modelling approach with GISSMO and with the concept of FFL/SFFL the yield locus Hill48 with
isotropic hardening is chosen. The flow-curve is extrapolated inversely with a combined Swift-Hockett-Sherby law
on the basis of the ductile fracture experiments. All simulation are executed with fully-integrated shell elements with
an edge-length of 0.50 mm in the commercial FE-code LS-Dyna. As proposed in [10] the FFL is calibrated directly
on the basis of tactilely measured thickness strains  3 f under the assumption of volume constancy and a constant,
optically measured strain  2 f in direction of the localized neck. The SFFL is constructed by linking the optically
measured supporting point of ST with the FFL. The resulting fracture lines and the specimen dependent loadings paths
are illustrated in Fig. 4 (a). Especially in case of UAT and PST non-linear loading paths develop due to strain
localization. The failure onset is overestimated for every stress state when these fracture lines are used for failure
prediction in simulation of ductile fracture experiments (Fig. 4 (c-f)).
The supporting points of the fracture curve in GISSMO are inversely determined. Starting with the optically
measured equivalent plastic strains at fracture  f for the respective stress states while neglecting the non-proportional
loading paths the supporting points are optimized based on engineering stress-strain curves and the punch force-
displacement curve (Fig. 4 (b-f)). Non-linear damage accumulation is invoked by n =1. Due to the high fading
exponent m =100, there is barely an effect of coupling plasticity and damage model, which corresponds to the
experimentally measured low void volume fraction. Besides the good agreement between the simulation with

FFL/SFFL GISSMO Experiments


(a) 1 (b) f
1.00 1.00
Fracture curve
0.80 0.80
ST UAT Instability curve
SFFL ST BAT
0.60 PST 0.60
FFL BAT UAT PST
0.40 0.40

0.20 0.20

0.00 2 0 
-0.70 -0.50 -0.30 -0.10 0.10 0.30 0.50 0 0.33 0.66
(c) ST (d) UAT
1.20 1.20
Engineering Stress [GPa]

Engineering Stress [GPa]

1.00 1.00
0.80 0.80
0.60 0.60
0.40 0.40
0.20 0.20
0 0
0 0.02 0.04 0.06 0.08 0.10 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Engineering Strain [-] Engineering Strain [-]
(e) PST (f) BAT
1.20 160
Engineering Stress [GPa]

140
Punch Force [kN]

1.00
120
0.80
100
0.60 80
0.40 60
40
0.20
20
0 0
0 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0 10 20 30 40 50
Engineering Strain [-] Punch Displacement [mm]

Fig. 4. FFL/SFFL with loading paths leading to fracture (a), Fracture curve and instability curve of GISSMO with loading paths leading to
fracture (b), Engineering Stress-Strain curves: ST (c), UAT (d), PST (e) and Punch Force-Displacement curve: BAT (f)
Sebastian Heibel et al. / Procedia Engineering 207 (2017) 2066–2071 2071
6 Heibel et al./ Procedia Engineering 00 (2017) 000–000

(a) 24 mm (b) GISSMO 24 mm (c) FFL/SFFL 27 mm

Fig. 5. Fracture of cross-die cup with slight necking at a drawing depth of 24 mm (a), prediction of onset of fracture at a drawing depth of 24 mm
with GISSMO (b), prediction of onset of fracture at a drawing depth of 27 mm with FFL/SFFL (b)

GISSMO and the experimental curves in cases of UAT, PST and BAT the ST specimen shows a slight deviation and
oscillations due to a stepwise failure process because of element deletion.

5. Validation

Both failure modelling approaches are validated on a cross-die sample, which exhibits loading paths similar to a
B-pillar. As displayed in Fig. 5 (a), for the DP1000 cross-die cups show tensile fracture with slight necking. With
GISSMO the location as well as the instant of onset of failure are predicted correctly (Fig. 5 (b)). The concept of
FFL/SFFL overestimates the fracture onset corresponding to the failure prediction in the ductile fracture experiments.

6. Conclusion

Within this contribution the damage and failure behavior of a dual-phase steel is investigated. Damage evolution
due to plastic deformation leads to a small amount of void volume fraction. In context of the continuum damage theory
the effect of void volume fraction on the effective stress is small due to the small experimentally measured value. The
interplay between ductile damage mechanisms and local shear banding leads for stress states between uniaxial tension
and biaxial tension to a combination of ductile tensile fracture (Mode I) and out-of-plane shear fracture (Mode III).
The complex loading situation with a strain gradient in thickness direction and three-dimensional stress-states in the
zone of localized necking cannot be represented with shell elements in FEA. As result of this strain gradient the
experimentally measured strains depend on the respective material, the sheet thickness and the stress-state.
Consequently, calibration of fracture criteria in a direct manner might not be reasonable for predicting failure in sheet
metal forming as shown for the example of FFL/SFFL. In summary at least for modelling with shell elements an
inverse parameter identification scheme seems to be advantageous, where weaknesses in the modelling approach are
intrinsically considered.

References

[1] C.C. Tasan, M. Diehl, D. Yan, M. Bechtold, F. Roters, L. Schemmann, C. Zheng, N. Peranio, D. Ponge, M. Koyama, K. Tsuzaki, D. Raabe, An
Overview of Dual-Phase Steels: Advances in Microstructure-Oriented Processing and Micromechanically Guided Design, Annu. Rev. Mater.
Res. 45 (1) (2015) 391–431.
[2] T. Wierzbicki, Y. Bao, Y.-W. Lee, Y. Bai, Calibration and evaluation of seven fracture models, Int. J. Mech. Sci. 47 (4-5) (2005) 719–743.
[3] Y. Bai, T. Wierzbicki, Application of extended Mohr–Coulomb criterion to ductile fracture, Int. J. Fract. 161 (1) (2010) 1–20.
[4] Y. Lou, H. Huh, S. Lim, K. Pack, New ductile fracture criterion for prediction of fracture forming limit diagrams of sheet metals, Int. J. Solids.
Struct. 49 (25) (2012) 3605–3615.
[5] C.C. Roth, D. Mohr, Ductile fracture experiments with locally proportional loading histories, Int. J. Plast. 79 (2016) 328–354.
[6] F. Neukamm, M. Feucht, A. Haufe, K. Roll, On closing the constitutive gap between forming and crash simulation, Proceedings of the 10th
international LS-DYNA users conference, Detroit (2008)
[7] J. Lemaitre, A continuous damage mechanics model for ductile fracture, J. Eng. Mater. Technol. 107 (1985) 83–89
[8] F.X.C. Andrade, M. Feucht, A. Haufe, F. Neukamm, An incremental stress state dependent damage model for ductile failure prediction, Int. J.
Fract. 200 (1-2) (2016) 127–150.
[9] P. Martins, N. Bay, A.E. Tekkaya, A.G. Atkins, Characterization of fracture loci in metal forming, Int. J. Mech. Sci. 83 (2014) 112–123.
[10] K. Isik, M.B. Silva, A.E. Tekkaya, P. Martins, Formability limits by fracture in sheet metal forming, J. Mater. Process. Technol. 214 (8) (2014)
1557–1565.
[11] S. Heibel, W. Nester, T. Clausmeyer, A.E. Tekkaya, Damage characterization of high-strength multiphase steels, IOP Conf. Ser.: Mater. Sci.
Eng. 159 (2016) 12013.

You might also like