Genetic algorithms for thermosciences research: application to the optimized cooling of electronic components

Abstract

Genetic algorithms are adaptive search procedures loosely based on the Darwinian notion of evolution. They use rules of natural selection to investigate highly complex, multidimensional problems and have been employed successfully in a variety of search, optimization and machine learning applications in science and engineering where other more traditional methods fail. In this study genetic algorithms are presented and discussed within the framework of an adaptive solution methodology for investigating otherwise intractable optimization problems in the thermosciences. The exposition focuses on their application to an electronics cooling problem where it is required to find optimal or nearly optimal arrangements of convectively cooled components placed in-line on the bottom wall of a ventilated two-dimensional channel. The present application is specific only for purposes of illustration. The power of the methodology rests on its generality of application and an indifference to the source of data (experimental, analytical or numerical) used in the optimization process. The study shows that genetic algorithms allow a cost-effective approach for investigating highly complex numerical and/or experimental thermosciences problems where it is desirable to obtain a family of acceptable problem solutions, as opposed to a single optimum solution.