Multi-Objective Optimisation of Aero-Engine Compressors

The design of a new aero-engine compressor is a complex task: design objectives are almost always conflicting, the design space is large, nonlinear and highly constrained, and the effects of some geometrical changes can be difficult to predict. Computational fluid dynamics (CFD) is now widely used in real-world applications and especially in the design of turbomachinery. However, the large design space and the time required for the numerical simulation of the whole turbomachine make the use of CFD in the early phases of the design process infeasible: preliminary design relies on a number of physical and empirical relations, still quite similar to those used in the early history of turbomachinery design. In this study, 87 independent parameters were used to define the geometry of a 7-stage compressor, the performance of which was evaluated using proprietary design codes for mean-line, multi-stage analysis. The effects on efficiency and surge margin of changing 44 design variables were analysed and their optimal values found by means of deterministic (gradient-based) and meta-heuristic (Tabu Search [TS]) optimisation methods. The results show clearly how the use of meta-heuristic optimisation tools can improve the preliminary design of turbomachinery, allowing a more thorough but still rapid exploration of the design space to identify the most promising regions that will then be verified and further analysed with higher fidelity tools. The results also reveal the impact of introducing various constraints into the design process, highlighting the effects of design decomposition.