Safety part design optimisation based on the finite elements method and a genetic algorithm

摘要

This paper deals with a numerical approach for improving the mechanical properties of a safety belt anchor by optimizing its shape and the manufacturing process by using a multi-objective genetic algorithm (NSGA-2). This kind of automotive component is typically manufactured in three stages: blanking, rounding of the edges by punching and finally bending (90°). This study focuses only on the rounding and bending processes. The numerical model is linked to the genetic algorithm (GA) in order to optimize the shape of the part and the process parameters. This is implemented by using ABAQUS© script files developed in the Python programming language and CATIA© script files in VBScript. The algorithm modifies the part’s design parameters in the CAD system, imports the model in STEP format into ABAQUS CAE and starts the Finite Elements Analysis (FEA) automatically. The material behaviour is modelled using a specific Lemaitre material damage formulation implemented in ABAQUS© via a FORTRAN user subroutine. The influence of two process parameters (the die radius and the rounding punch radius) and five shape parameters on the component behaviour is investigated. The search for the optimum component design depends on three objective functions which are (i) the material damage state at the end of the forming process, (ii) the von Mises stress field and (iii) the maximum von Mises stress in the folded zone. A global optimisation is finally performed in order to improve the ultimate unbending load and the volume of the safety part. This work has two major areas of innovation: (a) the improvement of the genetic algorithm NSGA-2; and (b) the development of an integrated numerical procedure including “Computer aided design” and “mechanical finite element simulation” controlled by the genetic algorithm.