A novel comprehensive approach for multi-objective optimization of the aluminum extrusion process

Barbara Reggiani, Lorenzo Donati, Luca Tomesani - University of Bologna

CHALLENGE Optimization of aluminum extrusion profiles involves conflicting objectives and only a limited amount of works tackle this problem from a multi-objective perspective. The aim is to perform a multi-objective optimization of the extrusion process by using the most significant geometric input variables to create optimal profile quality, good die strength and high performance.

SOLUTION - Seven input variables and eight objective functions were used in the optimization methodology for two selected test case of industrial complexity. The first case involves a round tube profile manufactured by means of a three-bridges porthole die. Input variable selection with six input variables, and CAD parametrization is done. The DOE was chosen based on the merging of experimental and numerical designs properly distributed in the variable space. Optimization was run with the NSGA-II genetic algorithm in modeFRONTIER. The second case investigated the optimization of a more complex, rectangular profile.

BENEFITS - A significant improvement has been achieved for the two case studies both in terms of welding quality and die life. The proposed approach could provide an effective automated guidance for practical productions to be used at the die and process design stage in order to maximize process performances both for die makers (die strength) and extruders (profile quality). Welding quality in the first test case increased by 13.8% and by 48.9% in the second case.