Coupling through move limits in multidisciplinary optimization

Abstract

Structural optimization has progressed significantly in the past twenty years and is now commonly used in the design of mechanical and aeronautical components. A multitude of optimization programs exist and many have been incorporated in finite element programs and are an integral part of that design tool. Problems consisting of a single discipline such as structural, vibrational, etc., have been dealt with successfully by researchers. Multidisciplinary problems such as the design of an aircraft which involves structural, aerodynamic, control and other disciplines were solved using sequentially generated optimized solutions. These sequentially generated optimized solutions need not necessarily lead to a global optimum solution for a coupled problem, as the interdisciplinary effects play a major role in reaching a globally optimized design.

This work is aimed at formulating an efficient procedure for coupling multiple disciplines, sensing the effect of one discipline over the other using approximations and move limit strategies. It involves using the information generated from the Two Point Exponential Approximation (TPEA) (which is a first-order approximation modified to include a parameter equivalent to the curvature of functions with respect to the design variables) in the coordination procedure for optimization of multidisciplinary problems.

An algorithm, the Best Design Selection Strategy (BDSS), for coupling the multiple disciplines is proposed. The main feature of this algorithm is that the coupling is done such that parallel analysis without any interdependency of one process over the other is still possible. BDSS uses the rule of “survival of the fittest”.

The BDSS algorithm is tested on three problems: a beam, a three-bar truss with lumped mass and a ten-bar truss with lumped mass, considering stress and displacement constraints as one discipline (static) and frequency constraints as a second discipline (dynamic).