Experience with sequential stochastic design improvement methods

Publisher Summary Safety problems involve a high degree of uncertainty and variability, such as velocity of impact, mass of vehicle, angle of impact, dummy position, restraint system, and mass/stiffness of barrier. As a result, deterministic optimization methods may be insufficient for developing robust products. Sequential stochastic design improvement methods use sampling methods, such as Monte Carlo and Latin Hypercube, to select the best design and then iterate the process until a satisfactory design is found. This chapter presents a paper that evaluates the implementation in commercially available ST-ORM, marketed by EASi Engineering, for safety applications. The objective of the paper is to benchmark ST-ORM's Stochastic Design Improvement capabilities. Two typical safety application problems are used—side impact and inflatable knee bolster (1KB). The side impact model is chosen because of its large model size and the 1KB is chosen because of its highly nonlinear behavior. This paper provides the solution of these two problems using conventional optimization methods.