Determination of suitable values for parameters governing B-spline based evolutionary structural optimisation using the boundary element method

Abstract

The basic evolutionary structural optimisation concept (ESO) has been developed for several years. Recently, the first ESO algorithm based on the boundary element method (BEM) has been presented. In this thesis, this algorithm is used for the 2D shape optimisation. The aim is to develop a greater understanding of the role of certain governing parameters that drive the optimisation using this algorithm, and to make recommendations as to appropriate values of these parameters that give rise to good optimal solutions most efficiently. Two problems, a short cantilever beam and a fillet, are selected as test cases in this work. By using a wide range of numerical tests, the performance of the optimisation has been evaluated using a variety of methods including mean performance analysis and multi-objective optimisation approaches using Pareto curves and weighted sums. Recommendations are made as to appropriate values of these parameters that give rise to good optimal solutions most efficiently. Sensitivity analysis is another important method in engineering design. In this work a new algorithm to undertake a sensitivity analysis has been developed and used in a small number of investigations for boundary element structural optimisation process. ESO is selected when computational efficiency is thought the most important consideration, since it can reach the optimum in fewer iterations and lower run-time compared with sensitivity analysis in structural optimisation.