Metamodel-Based Design Exploration and Optimization for Multiphysical Designs of Axial-Flux Synchronous Machines

Efficient electric drives tailored to their application play a crucial role in increasing the efficiency of the transportation sector and reducing global CO2 emissions. Depending on the specific application, a wide variety of requirements, solution variants, interactions with other systems, load profiles, and restrictions needs to be taken into account. The resulting modeling and evaluation complexity is becoming increasingly difficult to manage with conventional approaches, requiring more sophisticated methodologies for the development process of such technical systems. This paper presents a comprehensive methodological framework and its software implementation for design exploration and optimization based on metamodeling and the Functional Mock-up Interface Standard (FMI) for multiphysical system designs. The methodology is discussed and demonstrated using the multiphysics model of an axial flux synchronous machine (AFSM) in YASA topology with liquid-cooled hollow conductor coils, consisting of an electromagnetic and a thermal model. It is generally applicable to a wide range of multiphysics systems, for instance, the electrical and thermal model description of power electronics. The simulation of 10,000 designs in a broad design space takes only 92 minutes on a modern workstation, with the possibility of a subsequent application of a cost function. Changes in the optimization objectives do not require a recalculation of the entire multiphysics model.