You are on page 1of 5

For the official version please visit the following DOI

http://dx.doi.org/10.1063/1.4850070
Incremental Analysis of Springback and Kinematic
Hardening by the Variation of Tension During Deep
Drawing
H. ul Hassan1, A. Gner1, N. Ben Khalifa1 and A. E. Tekkaya1
1
Institute of Forming Technology and Lightweight Construction, TU Dortmund
Baroper Strae 305, D-44227 Dortmund, Germany

Abstract. Springback is considered as one of the major problems in sheet metal forming. It leads to assembly defects and
cause a huge amount of cost for tool modifications. In this work a tool for incremental analysis of springback analysis has
been presented. Development of springback with punch travel has been analyzed for the simple U draw-bend geometry,
tunnel geometry with open base and modified tunnel geometry with closed base and variable flange height. The effect of
tension variation in the sheet with punch travel has been considered as the steering parameter for the springback and
various profiles of varying tension are studied, which would generate different tensile forces in sheet. It is found that the
tension in the part in the last quarter of punch travel has a profound effect on the springback reduction as compared to the
traditionally applied constant BHF. Two selected kinematic hardening models, namely Yoshida-Uemori(YU) model and
Armstrong-Frederick(AF) model are used to study the coupled effects of tension and material hardening.
Keywords: Incremental analysis, Springback, Tension, Kinematic hardening.
PACS: 07.05.Tp, 02.70.Dh, 81.40.Ef

INTRODUCTION
In the sheet metal formed components, springback is one of the main sources of geometrical and dimensional
inaccuracy. This phenomenon is being extensively investigated for the last four decades. The crux of the researches
comes out to be the fact that the final shape of the part depends on the amount of elastic energy stored in the part
during the process of sheet metal forming [1]. However, since the amount of stored elastic energy depends on a
number of parameters, prediction of springback proves to be a complicated task. Also the sensitivity of springback
prediction to the numerical parameters [2] of finite element (FE) simulation adds further challenge to this
complexity.
Reliable descriptions of the materials elastoplastic behavior should be used for obtaining accurate numerical
solutions. Strong strain-path changes, such as the typical bending/unbending due to die radius require the modeling
of kinematic hardening [3] and the springback cannot be accurately numerically predicted if only isotropic work-
hardening is considered, even for the case of low level of equivalent plastic strain.
In [4] Yoshida et al. proposed a new yield surface which does not change in size but its center translates with
deformation; the bounding surface changes both in size. Also, by comparing the influence of the work-hardening
models on springback, different trends can be expected depending on the selected sheet metal formed part as well as
the process conditions [5].
Different time varying process parameters like as blankholder force (BHF), friction and drawing velocity also
show a profound effect on the springback. Since the BHF and friction cause stretching in the sheet and are directly
related to the material flow in the die, to obtain the desired forming result, these time varying process parameters
should be controlled in an optimum manner. In [6] a hybrid optimization algorithm is presented to adjust the time-
course for the blank holder force in Hat-profile drawing so that no sheet cracks occur and the springback is as small
as possible. Liu et al. presented a method based on intermediate straining for the minimization of springback. In
their method they change the BHF after a certain punch travel and keep the BHF force at that value for the rest of
the process which reduces the springback [7]. Tommerup et al. found that by changing the strain paths during
forming, the residual stress distribution after forming was affected, and thus the stress relieving during springback
which would lead to smaller springback [8]. Oliveira et al. in their study concluded that strong variations in the
distribution of friction coefficient in the part being formed would affect the springback evolution [9]. Hence we can
conclude that by proper modeling of the material behavior and the proper variation of process parameters the
springback could be minimized.

FE MODEL AND NUMERICAL ANALYSIS


For the analysis of springback based on different material models, three different FEM models are prepared. In
all the cases the die, punch and blankholder are modeled as rigid bodies and the contact between blankholder and
sheet is set to be frictionless. The sheet is modeled using the shell elements with 7 integration points through the
thickness. In the current paper the blankholder force and the friction variation have been replaced by just a single
parameter namely tension. The primary function of the variation of BHF and friction coefficients is the variation of
tension in the sheet during the forming process; hence these technological parameters have been replaced by a rather
more physical parameter. The blankholder which is primarily responsible for the tension is replaced with an
equivalent axial force (Feq) which is calculated from the yield force (Fyield). The maximum tension is considered
being 1.1 times the flow stress and is varied in a number of ways to see the combined effect of material model and
the load application.
The tension has been applied just by straining from the end nodes as shown in the Figure 1(a) for case of U
geometry. Figure 1(b) shows different tension profiles that have been applied to the geometries for springback
analysis. The other geometries that are analyzed are shown in Figure 2(a) and 2(b).

FIGURE1: (a) Cross-sectional model for U-geometry (b) Different tension profiles used in analysis

FIGURE 2: (a) Tunnel geometry with open base (b) Tunnel geometry with closed base and variable flange height

The Figure 2a shows the open channel model with the open bottom. Figure 2b shows the open channel model
with the closed bottom and this model also has an S-curve in the die and the blankholder so that the height of flange
also changes to increase the geometrical complexity. The reason for selection of this geometry is that it could be
easily extended to the B-pillar of automotive.
The material is modeled using the two built-in material models from the LS_DYNA material library i.e. AF
model and YU model. The material used for the analysis is DP600 and the material parameters for these models
have been experimentally determined with the cyclic tests. The AF model has two back stress terms for kinematic
hardening combined with isotropic hardening. YU model is an elasto-plastic material law which combines
Yoshidas non-linear kinematic hardening rule with transverse anisotropy only. Yoshidas theory use two surfaces to
describe the hardening rule: the yield surface and the bounding surface.
For the incremental analysis of springback the complete punch travel is divided into a 100 increments and then
the springback is calculated at each interval as the average deviation of all the nodes in the loaded and unloaded
configuration. Hence by this procedure the evolution of springback is investigated over the punch travel. At the end
of the punch travel the surface comparison is also done for the measurement of geometrical deviations.
RESULTS AND DISCUSSION
The results of incremental analysis of springback for different geometry, tension profiles and material models
could be seen in Figure 3. Profile P1 shows the typical BHF which is used as a constant value, Profile P2 shows an
exponentially increasing BHF which has a higher value of tension in the last 14 % of the punch travel, Profile P3
shows a monotonic increase in the BHF and the profile P4 shows a mirror image of profile B but with an
exponential decrease to the minimum value. For all these force profiles incremental springback analysis have been
carried out and the variation of springback with the punch travel for all these profiles have been plotted in following
Figure 3.

FIGURE 3: Springback variation for different geometries over punch travel using two different material models

It could be seen that in all the different cases of varying tension, geometry and the hardening model, the tension
profile 2 lead to the smallest springback independent of the geometry. The profiles however show the springback
due to the unloading of the same part at that interval and the comparison is not made with the reference during the
punch travel. It is important to note that the springback estimated by the YU model is smaller than the AF model in
all the case which means that independent from the geometry the estimation of springback by YU model would give
smaller value of springback. The tension profile P4 shows a huge springback in all the case and hence confirms the
necessity of application of higher tension at the end of punch travel. The tension profile P3 also gives a smaller
springback then then constant tension profile P1 as it also stretches the part just at the end. For the case P2, a
geometrical comparison has also been done for the simulated tunnel geometry with the ideal CAD geometry for the
two material models and is shown in Figure 4.

FIGURE 4: Geometrical comparison of tunnel geometry (a) for YU model (b) for AF model

In the Figure 4 we can see that that the geometrical deviation of simulated tunnel geometry is smaller in case of
YU model than the AF model. It could also be seen that the maximum deviation for the YU model occurs in the
base while for AF model it occurs at the flange. A higher tension at the end of deep drawing is expected to give a
smaller springback in flange, which is truly captured by the YU model.

CONCLUSION
In this paper a comparison of the effect of different material models along with different tension profiles on
springback has been investigated. It is shown that the variable tension gives a smaller springback as compared to the
traditionally used constant BHF. This tension profile could be further used for the optimization of springback based
on some closed loop control system having springback as an objective function. It is also found that the YU model
gives smaller springback regardless of the investigated geometry which is primarily due to its ability to model non-
linear kinematic hardening properly. For the deeper understanding a stress analysis between the top surface and
bottom surface is proposed for the different geometries to find the difference of stress over the geometry which will
give an insight of the reason for different springback evolution.

ACKNOWLEDGMENTS
This work is done as part of the SFB 708, TP C3. The authors thank the Deutsche Forschungsgemeinschaft for
their financial support. The authors also thank Prof. Karl Roll for his valuable suggestions.

REFERENCES
1. N. Narasimhan. , M. Lovell, Finite Elem. Anal. Des. 33 (1) 1999. 2942.
2. W.L. Xu, C.H. Ma, C.H. Li, W.J. Feng, J. Mater. Process. Technol. 151 (2004) 217222
3. L. F. Menezes, S. Thuillier, P- Y. Manach, S. Bouvier, 2002. Maryland, Neat Press Fulton, pp. 331333.
4. Yoshida, F., Uemori, T., 2003. Int. J. Mech. Sci. 45 (10), 16871702.
5. M.C. Oliveira, J.L. Alves, B.M. Chaparro, L.F. Menezes , Int. J. Plasticity (2006), doi:10.1016/j.ijplas.2006.07.003
6. C. Jiang, X. Han, G. Liu, G. Li, J. Mater. Process. Technol. 182 (2007), 262-267
7. G. Liu, J. Zhongqin , Y. Bao, J. Mater. Process. Technol. 120 (2002) 259-264
8. S. Tommerup, B. Endelt, Int J Mater Form (2009) Vol. 2 Suppl 1:809812
9. M.C. Oliveira, J.L. Alves and L.F. Menezes Materials Science Forum Vols. 455-456 (2004) pp 737-741

You might also like