Model order reduction (MOR) techniques seem a natural tool for including physics-based battery models in system-level simulation. Such an approach can make it computationally feasible to assess candidate battery designs by, for the example of electric vehicles, "driving the electrical system around until it overheats". But the use of MOR in this setting is far from straight-forward, and we use simulating an electric vehicle's thermal management system to demonstrate some of the challenges. In particular, we show that the battery nonlinearities, particularly in deep discharge, are not easily handled with existing reduction approaches. And we show that decomposing the system in to reducible subsystems, given the presence of volume fields (e.g. temperature), requires a level of insight that is not easily automated. A. Hochman, S. Asgari, M. Tsuk, J. White
Copyright. All Rights Reserved.