Uncertainty quantification for simulating vehicle driving dynamics


The design of a vehicle undercarriage is crucial for the vehicle's safety, comfort and driving characteristics. Since these attributes are essential for a modern passenger car, extensive numerical simulation is already performed in the early stages of a new car model's design process.

Nevertheless, the value of those simulations could be increased considerably if it was possible to assess the results' credibility in a stringent manner. Uncertainty quantification (UQ) could be the key to this issue.  For this reason, a research cooperation between LNM and BMW AG has been initiated in order to explore the potential of UQ in the context of vehicle driving dynamics.

The scope of the research cooperation includes the investigation and enhancement of existing mathematical UQ-concepts as well as their application to BMW's driving dynamics simulation environment. The latter comprises simulation models of different fidelity levels. In this project, both a commercial multi-body simulation software package and a complex MATLAB/Simulink simulation tool are subjected to an extensive uncertainty quantification analysis. A major goal is to draw conclusions on the simulation tools’ validity and predictability in the presence of both aleatoric and epistemic uncertainty.