Structural Optimization Using Ansys Classic And Radial Basis Function Based Response Surface Models

Abstract

This thesis describes a methodology to develop an effective and simple structural optimization method using Response Surfaces. It also further demonstrates how these Response Surfaces can then be utilized for further Multi Objective Optimization. The true or exact responses were calculated using Ansys Classic while the codes for the training and simulation of Response Surface is pro-grammed in Matlab. Initially we used a Halton sequence to generate a DOE of the design variables. Then we used Ansys Classic to generate the exact responses. Once the responses were imported the Response Surface are generated using a RBF approximation model. This RBF model is then optimized, then another de-sign point and the optimized points are then added to the DOE. This process is carried out until the results converge. The center points and the coefficient points of each loading condition are then used for multiobjective analysis.