Bernhard Thiele
German Aerospace Center (DLR), Institute for System Dynamics and Control, Wessling, Germany
Martin Otter
German Aerospace Center (DLR), Institute for System Dynamics and Control, Wessling, Germany
Sven Erik Matsson
Dassault Systèmes AB, Ideon Science Park, Lund, Sweden
Download article
http://dx.doi.org/10.3384/ecp14096381Published in: Proceedings of the 10th International Modelica Conference; March 10-12; 2014; Lund; Sweden
Linköping Electronic Conference Proceedings 57:40, p. 381-393
Published: 2014-03-10
ISBN: 978-91-7519-380-9
ISSN: 1650-3686 (print), 1650-3740 (online)
The demand to ever increase realism and scope of models routinely exceeds the currently available computing power and thus requires thoughts on improving simulation efficiency. This is especially true for real-time simulations; where fixed timing constraints do not allow to just “wait a bit longer”.
This paper presents a new approach in Modelica that allows a modeler to separate a model into different partitions for which individual solvers can be assigned. In effect; this allows to use multi-rate and multi-method time integration schemes that can contribute to improve the efficiency of a (real-time) simulation.
The first part of the paper discusses basic consideration relating to modular (real-)time integration. Afterwards; the implementation of a convenient Modelica library for the partitioning of physical models is briefly described. Finally; the presented library is used to partition a detailed six degree of freedom robot model for modular simulation. The simulation performance of that partitioned model is compared to the simulation performance achieved by using “conventional” global solvers.
Multi-rate / multi-method time integration; simulation; clocked discretized continuous-time partitions.
