Improving Structural Design by Genetic Search

Design optimization has long been a goal eagerly sought by engineers. The established classic methods of optimization demanded large‐scale computing facilities met only by the mainframe computers of the 1960s through 1980s. Two important developments of the last decade have changed the course of events. First, the microcomputer has arrived as a large‐capacity, rapid‐processing computing capability well up to the needs of optimization. Second, the methods of optimization have been joined by new approaches based on concepts of evolutionary or “genetic” progression in which an initial population (collection) of individuals (designs) is changed progressively in the direction of improved “fitness.” The concept of searching a large design space, notionally created by combining all possible values of the design variables and tracing a path through increasingly “better” designs, has become a realistic prospect. The purpose of this paper is to describe the ideas behind the concept of genetic search, to outline the basic principles of the genetic algorithm, to illustrate the genetic algorithm with a simple example of structural optimization, and to consider further developments in this area.