Multipoint shape optimisation of an automotive radial compressor using a coupled computational fluid dynamics and genetic algorithm approach

Automotive turbochargers operate over a wide range and require high efficiencies and pressure ratios. These conflicting requirements and a myriad of design parameters render iterative design techniques unfeasible. However, over the last decades the combination of numerical flow solvers and evolutionary algorithms has established itself as a viable option in the pursuit of reaching desired performance characteristics. This study seeks to perform a three-dimensional, multipoint and multiobjective optimisation of an automotive radial compressor by modifying blade shape as well as the meridional contour of the flow path. The method couples steady-state computational fluid dynamics (CFD) with a genetic algorithm (GA) to maximise isentropic efficiency in the region close to surge, while ensuring no significant reduction in choke margin. The results of two optimisation studies are presented and a flow-field analysis based on entropy generation rate is carried out revealing regions of flow improvement. The results are further compared against experimental data, indicating good agreement between the numerical and test data. The experiments however imply a detrimental impact on the surge margin for larger impeller speeds, which is attributed to unfavourable blade loading. Two additional optimisation runs are presented mitigating the effect of loading unbalance between main blade and splitter. (C) 2018 Elsevier Ltd. All rights reserved.