Evaluation of loading-path-dependent constitutive models for springback prediction in martensitic steel forming

Abstract

In this study, the performance of conventional and advanced constitutive models to the prediction of springback for ultra-high strength steel was assessed. The elasto-plastic behavior of a martensitic steel sheet sample was characterized to assess the conventional stress-strain behavior in uniaxial tension. More advanced tests such as uniaxial compression and tension-compression experiments were conducted to assess the strength-differential (SD) effect and hardening fluctuations during non-linear loading paths (NLLP effect). Three constitutive plasticity models, namely, the conventional isotropic hardening, Yoshida-Uemori (YU) two surface kinematic hardening and pure distortional hardening (HAH20) models were calibrated using an optimization procedure. Then, these models were employed for finite element predictions of springback after forming. The isotropic hardening model does not consider both SD and NLLP effects and the YU model captures only the influence of the NLLP effect. On the other hand, the HAH20 model describes both effects simultaneously. The performance of these models with respect to springback prediction was evaluated using three forming applications: wiper bending, C-rail drawing and Roof-Side-Rail crush forming. This work indicates that the HAH20 model is the best candidate for the prediction of forming and springback for such martensitic steel because of its ability to consider both SD and NLLP effects. However, the conventional isotropic hardening model led to results in good agreement with those obtained using the HAH20 model in specific cases. The reason, discussed in detail in this article, is likely due to the compensation of the permanent softening due to load reversal by the higher flow stress due to the SD effect.