Real-Time FEM Car Body for Driving Simulator Assessments

The integration of flexible structural components in real-time driving simulators can improve the representation of vehicle dynamics, but it also introduces a significant computational challenge. This thesis investigates the reduction, integration, and validation of a finite element body-in-white model for real-time driving simulator applications, with particular focus on handling, steering response, structural interface behavior, and computational feasibility. Starting from a high-fidelity finite element model, an Adams-compatible flexible-body representation is generated through component mode synthesis and exported as a Modal Neutral File. The initial flexible model is reduced through interface-node optimization, event-based mode selection, and state-space-based Modal Dominancy. The event-based approach ranks modes from Adams vehicle simulations using strain energy, kinetic energy, and modal participation, while Modal Dominancy ranks modes from the component-level input-output response without requiring vehicle-level time-domain events. The reduced models are validated against a full flexible reference model using macroscopic vehicle-dynamics signals, transfer functions, reconstructed interface-node displacements, real-time solver performance, and subjective driver feedback. The results show that body-in-white flexibility alone has a limited influence on the selected low-frequency steering-response quantities, while suspension-related flexible components produce more visible changes in the macroscopic vehicle response. However, the interface-node displacement validation shows that local structural accuracy still depends strongly on the retained modal content. Overall, the study shows that reduced finite element car-body models can be used in real-time driving simulator applications, provided that interface definition, modal selection, and validation criteria are treated as coupled modeling decisions.