Modelling and correlation of a high performance electric vehicle motor thermal management system

Nowadays Battery Electric Vehicles are a fundamental field of research in science and technology, because of their contribution to sustainable development and decarbonization. Moreover, they enable flexible strategies concerning Electric Drive Unit and auxiliaries system control. Beside such clear opportunities, there is the challenge of complying with the available on-board energy sources, which need to be handled in a smart way. In this direction, building a model describing how the Thermal Management System behaves can help OEMs to seek for an improvement of the driving range, not only according to the homologating cycles, but for real life vehicle usage. Therefore the goal of this activity was to provide a tool allowing for the verification of such a system performance, whose reliability is crucial to be assessed also out of testing facilities, in order to keep the components safe and durable, and to achieve desirable system efficiency. In this work, after developing an understanding of the architecture of the “Pininfarina Battista”, the rear motor cooling circuit was modelled by following a lumped approach, taking into account namely the left motor, the right motor and the mechanical transmission for an estimation of the heat rejection contribution; as well as the radiator, which represents the coupling element between the water-glycol loop and the air duct, and the pump and the fan, acting respectively on such two fluids. Before connecting all the subsystems together, the blocks of transfer functions realized using “Simulink” were tested and functionally validated independently. The trends of temperature characterizing the system gave the basis for a comparison of the aforementioned model simulated signals with the log signals sampled during real driving tests, making possible to find out suitable correlation parameters. This phase required the establishment within the tool of the possibility of switching between simulated information and log information when running the entire model, enabling the user to select iteratively how each coefficient could have been better tuned.