Response surface method (RSM) can be used to construct approximate functions to performance of structures. In this thesis a response surface method is utilized for structural optimization and implemented on a parametric CAD platform. Once an approximation is made, no derivative calculation is necessary for the design sensitivity analysis. The approximation gives the sensitivity information and intuition on the performance functions.
The scheme for the design of experiment chosen for the RSM has a large influence on the accuracy of converged solutions and the amount of computation. The D-optimal design criterion as implemented in this study is found efficient for the structural optimization. The number of experimental points can be selected arbitrary and the shape of design region irregular. After a set of designed experiments is performed, a response surface is obtained by a least square technique. The result is then used to get the next improved design. Inequality constraints are handled by a penalty approach.
The program developed is tested using several shape optimal design problems of such as a fillet, a torque arm and a belt clip. The results are compared with those by finite difference sensitivity analysis. It is observed that the RSM used provides a faster robust convergence than by the finite difference but the solutions are approximate rather than exact.