Optimization of the Iteration Parameters of the Simulation of Incompressible Flow by a Genetic Algorithm

In optimizing the SIMPLE-like numerical procedure, a genetic algorithm (GA) which aims to minimize the calculation time in a space of iteration parameters was developed. Various iterative procedures for Poisson's pressure correction equations, which arise from application of this SIMPLE-like procedure, were employed: successive over-relaxation, modified incomplete factorization, a conjugate gradient method, and a conjugate gradient method with a modified incomplete factorization preconditioner. The temperature equation was solved by successive over-relaxation, biconjugate gradient-stabilized method, and the biconjugate gradient-stabilized and generalized conjugate residual methods with modified incomplete factorization preconditioner. The momentum equation was treated with the successive over-relaxation and the Gauss Seidel methods. A methodology is presented for the numerical solution of natural convection in a rectangular cavity.