Testing Large Scale System Simulation using Linear Implicit Equilibrium Dynamics

Abstract

The concept of flattening where all model equations are collected in a single set is deeply hardwired into the Modelica language. While flattening enables effective symbolic manipulation such as the reduction of systems with a higher-index, it also imposes limitations. Two of these limitations are that the code generation for very large systems may not scale very well and that a statement on the regularity of the system often cannot be made before the flattening took place. Whereas it is difficult to overcome these limits in the general case, there is a subclass within Modelica models that is comparatively easy to precompile while still enabling the modeling of complex systems. This paper explores this path and presents first experiments on the scalability for larger systems.