Development and Evaluation of two Hybrid Shape Design Algorithms in Fluid Mechanics

Shape design problems، in general، and inverse design problems، in particular، are often solved via optimization techniques. Evolutionary algorithms provide robust and efficient solution methods for such problems. This paper focuses on the application of genetic algorithms (GA)، particle swarm optimization (PSO)، and two hybrid variants of GA and PSO. Performance of these optimization methods in the solution of inverse design problems is examined and it is shown that hybridization of GA and PSO can be used to improve the convergence rate of the iterative design procedure. Global Minimums of a number of well known optimization test functions are found by the proposed hybrid algorithms and the solutions of both internal and external flow inverse design problems are discussed. Up to 30% speed up is observed in the numerical test cases when the hybrid methods are employed and it is also shown that hybrid methods can get closer to the optimum solution as compared to either GA or PSO.