Project 5: Coupling of Model Order Reduction and Multirate Techniques for coupled heterogeneous time-dependent systems in an industrial optimization flow

 

Project Partners:

Bergische Universität Wuppertal, Germany
STMicroelectronics, Italy

 

Project Scientists:

Marcus Bannenberg
Early Stage Researcher (ESR05)

 

Michael Günther (University Wuppertal)

 

Jan ter Maten (University Wuppertal)

 

Andreas Bartel (University Wuppertal)

 

Angelo Ciccazzo (STM)

 

Project Description

In industrial circuit and device simulation, e.g. for estimating failure probabilities due to aging, simulation problems have to be run many times in the loop of an optimization flow. This can only be done by drastically reducing simulation costs via model order reduction (MOR). This is particularly challenging for coupled systems of various simulation packages for the different subcomponents and physical domains. For efficiency, MOR and multirate error estimates have to be linked to define overall error estimates, balanced to the accuracy requirements of the iteration level of the optimization flow.

Objectives of this research project are the combination of advanced concepts of model order reduction and multirating on hierarchies of submodels, the preservation of overall properties of the system (such as passivity, energy conservation etc.) and stability of the dynamic iteration process, as well as the incorporation of manifold mapping techniques.