Evolutionary optimization techniques as versatile solvers for hard‐to‐converge problems in computational fluid dynamics

Evolutionary algorithms mimic the process of natural evolution governed by the ‘survival of the fittest’ principle. In this work, a genetic algorithm (GA) is successfully used to solve problems in potential flow in a gradual contraction, viscous flow over a backward facing step, and non‐Newtonian flow using the power law model. Specifically, the GA heuristically searches the domain for potential solutions, precluding many convergence difficulties associated with the stiffness of a problem. The GA was able to solve problems that the gradient‐based method could not mainly because of its relative indifference to regions of high gradients when performing the search and that systems of discretized equations are never actually solved. The GA exhibited excellent scalability properties for solving problems with a large number of nodes. These results show evolutionary techniques to be of great utility in solving stiff problems in fluid flow. Copyright © 2006 John Wiley & Sons, Ltd.