Abstract: Uncertainties Quantification for Cup Forming process

Abstract: Uncertainties Quantification for Cup Forming process

The axisymmetric drawing of a circular blank is one of the simplest and most utilized processes for examining the anisotropic mechanical behavior of metallic sheets in metal forming. This straightforward test effectively simulates the key aspects of any deep drawing operation. The main objective of this study is to investigate how the uncertainties in the experimental characterization of the Lankford coefficients and the shape of the yield surface affect the cup forming process of a AA6061 Aluminum alloy. Due to the limited availability of experimental cup forming data and to minimize the number of required F.E. simulations, a machine learning-based approach is employed. Specifically, a limited number of high-fidelity F.E. simulations are conducted to train a surrogate model, which is then used to perform a Monte Carlo analysis. The surrogate model is initially trained based on randomly chosen sampling, which crudely explore the range of uncertainty. After the initial training, the surrogate quality is assessed. If the error is deemed too large, a supervised active learning method is performed. The study examines how experimental uncertainties and uncertainties in the yield surface shape influence the average height and earing profile’s amplitude of the fully drawn cup.