Practical approach to multi-objective optimization in ANSYS softwares

Abstract

Optimization of a complex system, such as an electric traction motor, often requires a multiphysics approach. It is a coupled case, where the outputs of one system are also the inputs of another system. If we talk about the design of the electric motor, then the output of the electromagnetic design is losses, which are then used as a heat source for thermal calculation. Hence the need for coupling, which is often a source of problems. This paper aims to present a practical procedure for optimizing the cooling of an air-cooled traction motor. Serialization of data of electromagnetic, thermal, and ventilation system will be also presented here. Subsequent optimization thus represents an independent task that works with already prepared input and output parameters. The optimization itself is based on a surrogate model, namely the Response Surface. The advantage of this approach is the reduction of computational complexity. Based on the mapped design space of input and output parameters, a reduced model is created, from which we can interpolate in the definition range of selected parameters. The paper focuses on electromagnetic, thermal, and ventilation systems, while the presented optimization procedure can be applied to any system, for example mechanical. All software used is from ANSYS. Specifically, Ansys Fluent for thermal and ventilation calculations, Ansys Maxwell for electromagnetic calculations, and optimization Ansys DesignXplorer is used.