Collaborative Research Centre – TRR361/F90: CREATOR – Computational Electric Machine Laboratory
Subprojectt D03: Design Optimisation of Electric Machines under Uncertainty and Optimal Design of Experiments for Parameter Identification
In this project the design (shape and topology) of electric machines under uncertainties is optimised. Electric machines are described with sufficient accuracy by the magnetostatic approximation of Maxwell’s equations, which yields for a Permanent Magnet Synchronous Machine (PMSM) to an elliptic partial differential equation (PDE). Also production tolerances, material variations, usage scenarios and other influences have to be taken into account during the optimisation, because they lead to uncertainty in the shape or topology optimisation of electric machines.
The optimised design should be robust with respect to the considered uncertainties either in a worst case sense (robust optimisation) or in a stochastic sense while providing similar performance as the nominal optimum. The development of efficient methods for complex shape or topology optimisation under uncertainty is challenging. Such problems are a bilevel optimisation problem with PDE-contraints which are contained of an inner maximization problem and difficult to treat numerically.
To obtain tractability of the problem an approximation (linear, quadratic, reduced order model) of the problem is necessary.
Moreover, in order to determine material properties, e.g., local magnetic properties, accurately and to quantify their uncertainty, advanced methods for optimal experimental design governed by PDEs are required.
Kontakt: Theodor Komann, Stefan Ulbrich
