GPU based parallel optimization of members of a truck floor

Abstract

Abstract Grillage - formally lattice - structures are made of longitudinal beams and cross members. Grillages can model vehicle frames, bus floor frames or an entire vehicle. The optimization of the cross members of this structure is shown in this article. The optimization method is a nature-inspired metaheuristic evolutionary method, the flower pollination algorithm. The target function is the total weight and cost of the optimized cross-member. Unknowns are typical cross-sectional dimensions of the cross-member. The design constraints considered are local buckling of web plate and flanges, and fatigue. The calculation was performed assuming the usage of aluminium alloys. In this paper, we propose a possible parallelization method, for computation of base algorithm and one group of fitness functions. The results show, that parallel computations can lead to significant reduction in computational time, if the population size is large and the number of variables are also large.