Effective Move Limits for Approximate Structural Optimization

Abstract

In structural optimization problems, conventional approximations are often used. For this type of approximations, move limits are usually defined as allowable changes in design variables during optimization of the approximate problem. This paper shows that these common move limit constraints, which are suitable for first-order series approximations, might not be effective for various improved approximations. The improved approximations, developed in recent years and used extensively in various applications, require more effective constraints. Such alternative move limits, proposed in this paper, include angle move limits, which are explicit nonlinear constraints on the angle between the initial and the modified design lines and error move limits, which are implicit nonlinear constraints on the errors in the equilibrium equations. The effectiveness of the aforementioned types of move limit constraints is demonstrated by various numerical examples. It is shown that, for improved approximations, the proposed move limits provide more reasonable results than the common approach.