High-Fidelity Multi-criteria Aero-structural Optimisation using Hierarchical Parallel Evolutionary Algorithms

Abstract

An emerging technique for the solution of Multi-criteria and Multidisciplinary Optimisation problems are Evolutionary Algorithms (EAs). This paper describes a parallel multi-criteria (multi-objective) evolutionary algorithms (PEAs) for aero-structural problems. The foundations of the algorithm are based upon traditional evolution strategies and incorporate the concepts of a multi-objective optimisation, hierarchical topology, asynchronous evaluation and parallel computing. The algorithm works as a black-box optimiser and has been coupled to several aerodynamic and aircraft conceptual design solvers. The paper describes the features of the method and the application of the method for aero- structural design problems. The coupling of the algorithm with a higher order panel method, a commercial FEA solver and an automated aerodynamic and structural mesh generation program is described. In this automated process the optimiser defines design variables for the automatic mesh generation program which defines and produces the external geometry of the wing (Root chord, tip chord, sweep, etc) and the internal geometry (number of spars, number of ribs skin thickness, etc). This geometry is then analysed by the aerodynamic solver and maps the pressure forces into the structural model. Single or multiple objectives can be defined by the optimiser to find non-dominated -Pareto optimal solutions or the Nash equilibrium. The parallel computing capabilities in combination with hierarchical levels of aerodynamic and FEA solvers are exploited to reduce computational expense. Comparisons will be presented between a traditional analytical approach -weakly couple, and a high fidelity -strongly coupled- approach for the aero-structural analysis of a high aspect ratio aircraft/UAV wing.