Model Reduction for Power Electronics Systems with Multiple Heat Sources
Résumé
Fast transient analysis of electronic systems thermal behavior is a vital tool for the examination of integrated circuits during the development. In an operational as well as in failure mode, this kind of analysis is needed to verify that operation conditions will remain safe for complex mixed signal electronic systems. Many of the questions, which have to be evaluated, are coming up in the last stages of the development, when the time to take a decision has to be short, and the designers need fast and accurate answers. A formal model reduction approach allows us to take a high dimensional finite element model and generate its low-dimensional approximation formally. As such, it is an ideal candidate for the goal above. Several research groups have already documented its successful application to a thermal problem with a single heat source. In the paper, we present the application of model reduction to a thermal problem with many independent heat sources. We demonstrate that the block Arnoldi process allows us to automatically build accurate compact dynamic thermal models while preserving independent heat sources in the reduced model. We discuss computational time necessary to perform model reduction. We show that a simple approach based on local error indicators allows us to choose automatically the dimension of the reduced system. We present the complete flow of using the method in engineering environment: Modeling in ANSYS the original problem, Model reduction with MOR for ANSYS, Evaluation of results using Mathematica.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...