Shape optimization of notched mechanical members to reduce stress concentration

Abstract

Intensive competition in the market place today requires engineers to continuallysearch for better and more economic designs. Current interest especially in structural shapeoptimization is largely motivated by demands for more competitive designs throughout the industrial sector. Therefore, considerable effort has been devoted to developing efficienttechniques for shape optimization.The aim of this study is to optimize the shape of two-dimensional notchedmechanical members to reduce stress concentration. In order to determine the best design,a search algorithm, called Direct Search Simulated Annealing (DSA) technique which seeks the best configurations through randomly generated configurations, is used. Aboundary perturbation technique is used to generate random shapes. In this technique,initially, a set of design points is chosen and connected by spline curves, therebyconstituting the boundary of the model. In this optimization problem, the objective function to be reduced is the maximum equivalent stress. Besides, connectivity of thegeometric model, as a design constraint that should not be violated, is maintained throughthe optimization process. In this thesis, a general computer code is used, which was developed using ANSYSVersion 8.1 Parametric Design Language for shape optimization. This computer code wasapplied to three different models in order to check the effectiveness of optimizationtechnique. The results show that this technique can be applied for successfully twodimensional shape optimization problems.