Improving the Quality of Reshaped EoL Components by Means of Accurate Metamodels and Evolutionary Algorithms

The reshaping of End-of-Life (EoL) components by means of the sheet metal forming process has been considered largely attractive, even from the social and economic point of view. At the same time, EoL parts can often be characterized by non-uniform thicknesses or alternation of work-hardened/undeformed zones as the result of the manufacturing process. Such heterogeneity can hinder a proper reshaping of the EoL part, and residual marks on the reformed blanks can still be present at the end of the reshaping step. In a previous analysis, the authors evaluated the effectiveness of reshaping a blank with a deep-drawn feature by means of the Sheet Hydroforming (SHF) process: it was demonstrated that residual marks were still present if the deep-drawn feature was located in a region not enough strained during the reshaping step. Starting from this condition and adopting a numerical approach, additional investigations were carried out, changing the profile of the load applied by the blank holder and the maximum oil pressure. Numerical results were collected in terms of overall strain severity and residual height of the residual marks from the deep-drawn feature at the end of the reshaping step. Data were then fitted by accurate Response Surfaces trained by means of interpolant Radial Basis Functions and anisotropic Kriging algorithms, subsequently used to carry out a virtual optimization managed by multi-objective evolutionary algorithms (MOGA-II and NSGA-II). Optimization results, subsequently validated via experimental trials, provided the optimal working conditions to achieve a remarkable reduction of the marks from the deep-drawn feature without the occurrence of rupture.