Combining optimizer and metamodelling for railcar structural optimization

Abstract

Stress constraint is a hard issue for structural topology optimization, especially for large-scale structures, e.g. railcars. Another technique is proposed to combine a sizing optimizer with metamodelling for topology optimization. At the lower level, for each topology design sampled within the topology design space, a sizing optimizer finds feasible and optimal solutions in terms of sizing variables (plate thickness in continuum structures). All performance constraints such as stress, displacement, and stability, are handled only at this level. At the upper level, a metamodel is built to fit all the optimal solutions found at the lower level and is optimized for topology design. The only constraints left at the upper level are topological constraints and topological variable bounds. Only the objective function (e.g. weight) versus topological variables, and not the constraints, is approximated. The number of topology design variables is much smaller than those used in many other topology optimization approaches. Thus it may be able to handle large-scale structural systems. It was applied to two boxcar design projects, resulting in 18 per cent and 36 per cent weight savings and significant reductions in manufacturing cost and total cost.