Parametric model order reduction of thermal models using the bilinear interpolatory rational Krylov algorithm

The Bilinear Interpolatory Rational Krylov Algorithm (BIRKA; P. Benner and T. Breiten, Interpolation-based H-2-model reduction of bilinear control systems, SIAM J. Matrix Anal. Appl. 33 (2012), pp. 859-885. doi:10.1137/110836742) is a recently developed method for Model Order Reduction (MOR) of bilinear systems. Here, it is used and further developed for a certain class of parametric systems. As BIRKA does not preserve stability, two different approaches generating stable reduced models are presented. In addition, the convergence for a modified version of BIRKA for large systems is analysed and a method for detecting divergence possibly resulting from this modification is proposed. The behaviour of the algorithm is analysed using a finite element model for the thermal analysis of an electrical motor. The reduction of two different motor models, incorporating seven and thirteen different physical parameters, is performed.