Multi-objective shape optimization of aircraft cabin ventilation components using adjoint CFD

Abstract

The design of the air distribution systems in aircraft, trains and cars still relies on the experience of the engineers and on extensive testing. However, the performance of these air distribution systems can be improved by applying numerical optimization techniques. In this paper we use adjoint Computational Fluid Mechanics to optimize a cabin outlet of the air distribution system of the Dornier 728 aircraft with the aim to develop a procedure based on which a multi-objective optimization can be performed for the complete air distribution system. With state of the art techniques for multi-objective optimization, like the weighted-sum method, the computational costs are beyond the capabillity when it comes to complex geometries. To avoid these high costs we present an engineering approach which is based on optimized combinations of sensitivity fields obtained with single-objective optimizations in a post-processing step. It is shown that the variation of the scaling factors allows to minimize the objective functions and to approach the utopia point. To be able to systematically vary the scaling factors a multi-objective genetic algorithm is proposed. In the last section the multi-objective problem is reformulated with an epsilon-constraints method and the variation of the scaling factors is optimized with a conjugated gradients optimization method.