PSI - Issue 82

Andreas Taucher et al. / Procedia Structural Integrity 82 (2026) 295–301 A. Taucher et al. / Structural Integrity Procedia 00 (2026) 000–000

299

5

3.4. Yield surface criteria The material model creators do not use the Hill48 and Hill90 yield surface criteria, since both criteria are not able to describe the yield surface of aluminum alloys properly. The OEM interview partners and one of the engineering service providers use the BBC05 criterion because of its good prediction of biaxial yield stresses, as confirmed by the simulation software developer. This criterion can be considered as standard in deep drawing simulations of aluminum components for the automotive industry. However, two OEM interview partners occasionally use the older Barlat89 criterion if experimental material data required for the BBC05 criterion are lacking. Furthermore, one interview partner stated that the more recent Vegter17 criterion (Abspoel et al., 2017) needs further verification to demonstrate its potential and reliability. Figure 4b shows the difference between three common yield criteria. All material model creators use uniaxial tensile tests and bulge tests to determine the parameters for defining the yield surface; one material model creator additionally uses biaxial tensile tests and shear tests to better adjust the yield surface. 3.5. Strain rate and kinematic hardening parameters Additional effects such as strain rate dependency and kinematic hardening can be considered in the material models. Nevertheless, only two interview partners account for the influence of the strain rate, which is assumed to change the flow stress up to about 10 %. One interview partner performs uniaxial tensile tests and bulge tests at speeds similar to industrial forming speeds. Nevertheless, all interview partners supposed that the strain rate dependency of aluminum alloys is more significant in crash simulations, but it has just minor influence in forming simulations. Since elaborate experiments are required for determining parameters that describe kinematic hardening, only two interview partners consider this influence in their calculations, if materials are known as sensitive to kinematic hardening. 3.6. Forming limit curve Three of the material model creators use the Nakajima test to determine the forming limit curve, one of them even integrates the Marciniak test. One material model creator additionally uses the fracture strain from the uniaxial tensile test to adjust the data point at the corresponding strain path in the FLC. Two material model creators also rely on synthetic FLC models such as the Arcelor (Cayssials et al., 2005) or the Abspoel-Scholting (Abspoel et al., 2013) models. The simulation software developer confirmed that synthetic FLC models are still underrated, although they provide a good approximation with little effort. As schematically illustrated in Figure 5, the interview partners usually shift the original FLC downward to consider a safety margin. In addition, one OEM uses the fracture strain determined by uniaxial tensile testing to adjust the left side of the FLC.

Fig. 5. FLC based on Nakajima tests with safety margin and additional point from uniaxial tensile test.

3.7. Crack detection at sheet edges During the interviews, the participants mentioned that cracks at the edges of the sheets or deep-drawn components, respectively, can appear even at lower strains than predicted by the simulations. Therefore, eight interview partners