Cross-sectional shape optimization of thin-walled columns enduring oblique impact loads

This study aims to find the best cross-sectional shapes of thin-walled columns enduring an oblique impact loading for crashworthiness. For approximating to the shape, spline polynomials are used with four key-points benefiting from the double symmetry of the cross section. Crashworthiness is defined by using a multi-objective function. By using Latin hypercubes design of experiment methodology, the design space is sampled. Based on the finite elements analyses, the objective functions are approximated by adopting radial basis function network. The corresponding Pareto front is found by Non-dominated Sorting Genetic Algorithm II. It is found that plus-sign-like cross-sections have better performance than benchmarks for all objectives.