Aerodynamic Shape Optimization by means of a Hybrid Genetic Algorithm

This paper presents the comparison and coupling of two aerodynamic shape optimisation techniques based on Geneti Algorithm (GA) and Powell algorithm. The optimisation codes are coupled with an automatic grid generation and Navier-Stokes solver (FIDAP). The examples presented consider the optimisation for highest egiciency of a profile of given volume in laminar incompressible flow (Re=1000) in the case of a single point design. The results obtainef point out that a hybrid technique GA/Powell exhibits better performances than each singular algorithm but th computational cost is approximately doubled. In order to overcome the main drawback of GA, i.e. the high numbe_ of calls to the flow solver, the possibility of extrapolate data from the created database is investigated.