Modelling stochastic forming behaviour in deep drawing steels

26 Aug 2011

Eisso Atzema, Michael Abspoel, Marc Scholting; Tata Steel RD&T IJmuiden, The Netherlands

Structural car parts usually get stamped form sheet metal, most often steel. Material properties scatter and this results in scatter in the stamping process and consequently in the stamped parts. Tata Steel customers may require tighter material tolerances to reduce this scatter to acceptable proportions. However, an accurate prediction of the scatter might allow the design of a stamping process to be robust against material property scatter. And this could well be a more
sensible approach. The optimisation of a stamping process towards insensitivity is referred to as robust optimisation. For robust optimisation the material scatter needs to be modelled numerically and this requires complex data processing. In this paper we propose a more pragmatic approach. First the correlations between material properties have been examined.
Scatter in hardening and yield locus proved to be independent, so their modelling can also be done independently. Scatter in hardening was modelled by defining extreme properties in the collective of data gathered and then finding a dataset of the coil that represents these extremes the best. In earlier work the scatter of the yield locus was modelled using anisotropy coefficients r. The results of that approach were not trusted and an alternative approach based on texture
measurements will be presented here. For now the results are still based on analysis of all coils measured, rather than using an approach based on extremes as was employed for scatter in hardening.

Paper presented at Numisheet 2011, Seoul, Korea, August 21-26, 2011