Artificial Neural Network (ANN) Based Formability Prediction Model for 22MnB5 Steel under Hot Stamping Conditions

Abstract

Production of high strength, light weight automotive components is one of the major thrust areas in the automotive sector, which is achieved by using advanced high strength steels. However, problems associated with these steels are lower formability and high springback effect at room temperature therefore hot stamping process is widely used to address those challenges. Hot stamping or press hardening process is a high temperature deformation process where specimen is heated above austenization temperature and held at this temperature for appropriate time for homogenization. Once temperature of specimen is uniform, specimen is transferred to forming press where forming and quenching take place simultaneously. Due to elevated temperature and quenching involved in the process, determination of formability is difficult. For determination and prediction of formability of 22MnB5 steel, experiments are performed using Gleeble-3800 at various temperatures and strain rates. In Gleeble-3800 uniaxial movement is possible only and therefore in this work notch and simple tensile test are performed to obtain Forming Limit Diagram (FLD) in drawing region. Thickness gradient-based necking criterion was used for FLD determination. ANN model was developed to predict flow behavior and FLD. It was found that in drawing region of FLD formability increases with increase in temperature and strain rate. The ANN model developed was able to accurately predict the constitutive behavior and limit strains in drawing region.