Multi-objective optimization of the extrusion process

Barbara Reggiani, Lorenzo Donati, Luca Tomesani (University of Bologna)

CHALLENGE -  Extruders and die makers require a proper optimization process to be performed at the design stage to concurrently account for all manufacturing requirements. The aim of the present work was to perform a multi-objective optimization of an industrial three-bridges porthole die used to manufacture a thick round tube profile made of AA7003.

SOLUTION - A three-bridges porthole die, used to produce thick round tube profiles, was selected as test case. Eight competitive objective functions were selected aimed at increasing the seam weld quality, the die lifetime, the production rate and at restraining the die yielding and the peak process load with respect to an initial already optimized solution. Six geometric input variables were included in the optimization procedure (welding chamber and bridges height, ports width, undercut on ports, die entry angle, mandrel-bridges fillet radius) together with the ram speed. Multi-objective optimization was performed with modeFRONTIER using meta-models generated over a selected set of experimental and numerical training designs. modeFRONTIER integrated CAD software, used for geometry parametrization, with the selected FE code used for the numerical simulations. 







BENEFITS - The average pressure in the welding chamber was increased by 2.6% and the quality index by 13.8% with respect to V1 experimental tested design. The evolutionary algorithm used in the optimization procedure based on meta-models was the Non-dominated Sorting Genetic Algorithm (NSGA-II) implemented in modeFRONTIER with the initial population of the training 144 designs