Constitutive Modelling of Hot Deformation Behaviour of Nitrogen Alloyed Martensitic Stainless Steel

The relationship of the flow stress with the strain, strain rate, and temperature is complex and often described by constitutive equations. Constitutive equations are one of the principal inputs for the hot working simulation using finite element method (FEM). The current investigation focuses on predicting the high temperature flow behaviour of Fe-15.9Cr-1.7Mo-0.43C-0.14Nb-0.22N (wt%) nitrogen alloyed martensitic stainless steel using constitutive equations. The flow curves obtained from isothermal hot compression tests conducted in a temperature range of 1173–1423 K and strain rate range of 0.001–10 s−1 were used for modelling. Johnson Cook (JC), Modified Johnson Cook (m-JC) and artificial neural network (ANN) models were employed to formulate the hot deformation behaviour. Accuracy of the predictions was evaluated using parameters such as correlation coefficient (R) and average absolute relative error (AARE). The JC and m-JC models showed AARE of 19.6 and 1.2%, respectively. For developing the ANN model, some of the best training algorithms and transfer functions were explored. Bayesian-regularization employing hyperbolic tangent transfer function gave the best results with AARE of 0.3% and the correlation coefficient of 0.999.