Non-linearities modeling for high-fidelity vehicle models

It is crucial for car manufacturers to be able to simulate the non-linearities (NL) present in the vehicles to allow a greater insight avoiding time-consuming and expensive testing during development and following phases. This thesis has the aim to extract, analyze and reproduce the non-linearities detected from test results, in both 3D Multi Body Simulation (MBS) at suspension level and on a more general 1D Full Vehicle (FV) level. To model non-linearities, typically test data is acquired providing both the input force or moment and the output displacement, velocity, or acceleration. It is to be considered that this is possible only with a component on a test bench but not entirely feasible when a full vehicle that is driving on the road is considered. There exist, in fact, technologies to estimate loads in correspondence of FV suspension joints but can be intrusive and incredibly expensive. Hence, operational observations (metrics) are taken as references to obtain equivalent results in simulation by implementing and updating non-linearities in the models. The described approach can lead to a significant improvement on the knowledge of the vehicle behavior through localization of NL and estimation of their contribution to the vehicle responses.