Multi-Objective Numerical Optimization of Radial Turbines

Abstract

Abstract The growing demand of high flexibility and wide operating ranges of radial turbines in turbocharger applications necessitates new methods in the turbomachinery design process. Often, design criteria such as high performance at certain operating conditions or low inertia contradict the requirement for high durability. This article demonstrates a newly developed optimization approach for radial turbines that allows to optimize for several design objectives. The presented approach is based on a parametric model of the turbine wheel geometry. On the one hand, the model is designed to capture the most important geometry and design features, and on the other hand, it is flexible for use on various machines. A surrogate model-based genetic algorithm is used to optimize the geometries with respect to several objectives, including efficiency, durability (HCF), low-cycle fatigue (LCF), inertia, and mass. Certain operating points or criteria can be particularly emphasized and specified constraints throughout the process allow for customized optimization. The simulations underlying the optimization are state-of-the-art computational fluid dynamics (CFD) and finite element (FE) analyses, involving the respective components. The newly developed and fully automated approach includes tasks of different disciplines. In the end, a selection of several promising geometries is examined more intimately to finally find a most suitable geometry for the given application. For the current study, this geometry has been manufactured and tested on a hot gas test facility to successfully validate the design process.