Optimisation of aerodynamic and coupled aerodynamic-structural design using parallel Genetic Algorithms
Denis Doorly, Joaquim Peiró, Jonathan Oesterle · 6th Symposium on Multidisciplinary Analysis and Optimization · 1996
This paper describes a parallel genetic algorithm (GA) which is linked to aerodynamic and structural analysis for multidisciplinary optimisation of wings. The method has been applied to 2D aerodynamic section design, to the aerodynamic optimisation of wings and to combined aerodynamic and structural optimisation. The parallel genetic algorithm commences with the subdivision of the entire population of trial solutions into semi-isolated sub-populations, (so-called `demes'). The paper describes a model in which one or more demes may be mapped to a given processor; each processor concurrently executes the CFD and structural analysis for its deme(s). The algorithm is applied to various test problems; it is shown that this form of GA implementation provides advantages both in terms of search efficiency, and for parallel implementation. 1 Introduction The development of computational methods which attempt automatically to optimise the design of airfoils and wings is comparatively recent; ne...