Driver and vehicle dynamics modelling for powertrain testing through virtual on-track session

Nowadays in the automotive industry design virtualization is gaining importance due shorter time to market, higher differentiation and increased complexity. Models allows powertrain engineers to move the design and tuning work much earlier in the development process reducing needs of expensive and time-consuming prototypes and experimental testing. One of the main potentials of virtual design is the possibility to explore any possible vehicle mission profile with any environmental condition, but often the generation of realistic duty cycles for the product under development may be challenging. This MSc thesis project, carried out in collaboration with Maserati and Powertech Engineering, aims to develop a tool in MATLAB-Simulink for generating realistic powertrain mission profiles for vehicle running on driving tracks at the limit of handling. This has been carried out by developing a virtual high-performance driver to drive a vehicle model following a given track trajectory, pursuing the best lap time possible according to maximum vehicle performances and tarmac (dry, wet) conditions. Vehicle model has been modelled as a 3 DoF Non-Linear Single-Track model, also known as bicycle model, and tuned according to provided experimental data and meaningful assumptions. Pacejka tire model has been also implemented in order to simulate tire behaviour according to vertical and longitudinal loads applied to wheels. Moreover, both a simplified map-based engine and a more accurate GT-Power engine model have been implemented with the possibility to select between different simulation modalities. At first, Vehicle model validation has been carried out with open-loop tests by imposing experimental vehicle commands actuation (pedals, steering wheel) and comparing model results with experimental telemetry. Then, virtual Driver model has been developed introducing steering wheel, accelerator pedal, brake pedal and gear shift controllers making the integrated model independent from experimental vehicle command actuation. Eventually, the developed model is able to closely represent the behaviour of a vehicle driven by a professional driver on a driving track on the base of X,Y trajectory only: model’s results match well the provided telemetry data as vehicle longitudinal and lateral accelerations, speed and it is able to predict lap times with a lower than 1 second accuracy.