Optimizing and Anticipating the Residual Stress in Al6061 Deep-Drawn Cups Employing RSM and ANN

Abstract

Abstract Residual stresses are inevitable phenomena in the deep drawing process due to the applied mechanical tools. However, they can be minimized as process parameters are optimized. This paper explores the use of response surface methodology (RSM), finite element method (FEM), and artificial neural networks (ANN) to optimize and predict the punch force, thickness distribution, and residual stresses in Al6061 deep drawn cups. The study aims to investigate eight key process parameters, including initial blank thickness, punch and die shoulder radii, blank-holder force, punch velocity, and coefficients of friction between die-blank, holder-blank, and punch-blank. RSM is employed to design experiments, identify the degree of importance of parameters, find optimal process parameters configuration regarding each output and provide the required dataset for training the network. Moreover, it is used as an extra comparison tool with ANN. To avoid time-consuming and costly residual stress assessment techniques, a verified FEM model is used. Finally, an ANN is trained to anticipate outputs considering different process parameters, most notably residual stress, which is difficult, time-consuming, and expensive to measure. The results show that the optimized process parameters lead to a significant reduction in residual stresses, required punch force and thickness distribution. Furthermore, the proposed methodology provides a reliable and efficient approach for anticipating the residual stresses in deep-drawn cups and improving the quality of the final product.