STRUCTURAL OPTIMIZATION USING SPACE MAPPING AND SURROGATE MODELS

The aim of this paper is to determine if Space Mapping technique using Surrogate Models in combination with the Response Surfaces Methodology (RSM) is useful in optimization of crashworthiness applications. In addition, the efficiency of optimization using Space Mapping will be compared to conventional structural optimization using the Response Surface Methodology (RSM). To determine the response surfaces, several evaluations must be performed and each simulation can be computationally demanding. Space Mapping technique uses surrogate models, i.e. less costly models, to determine these surfaces and their associated gradients with respect to the object and constraint functions. The original full model is used to correct the gradients from the surrogate model for the next iteration. Thus, the Space Mapping technique makes it possible to reduce the total computing time, needed to find the optimal solution. Two application problems are used to illustrate the algorithm. All examples are constrained optimization problems with one or two design variables. In all applications, the algorithm converged to the optimum solution. For the crashworthiness design problems the total computing time for convergence was reduced with 53% using Space Mapping compared to the conventional RSM. The conclusions are that optimization using Space Mapping and Surrogate Models can be used for optimization in crashworthiness design with maintained accuracy but with a significant reduction in computing time compared to traditional RSM.