Computationally Efficient Modelling of Electric Vehicle Thermal Systems

The increasing adoption of electric vehicles and stringent global sustainability targets highlighted the challenge of improving energy efficiency, particularly through an optimisation of the thermal management system. Considering this, the aim of this thesis is to develop in Matlab/Simulink a model of the Thermal Management System (TMS) that can serve as a foundation for the future implementation of a Model Predictive Control (MPC) strategy, designed to maximise energy efficiency while keeping the battery and cabin within the desired thermal limits. To achieve this, two complementary modelling approaches are developed: a Physics-based model and a data-driven model. The Physics-based model is based on fundamental thermodynamic laws and is implemented using simplified equations. Its main advantage lies in being usable immediately, in complete independence from a possible database. The model is validated through a comparison with a MathWorks Simscape reference model, demonstrating an optimal compromise between accuracy and computational efficiency, with ten times less simulation time. The data-driven model, on the other hand, uses neural networks trained on EFFEREST experimental data to capture the system's complex nonlinear relationships under different operating conditions, including battery heating/cooling and cabin climate control. The results of the study are analysed and discussed in the final chapter. The simplified Physics based model offers generalizability and fast execution times (about 0.7 seconds), showing decent alignment with the Simscape reference model, which takes 71 seconds per simulation. The data driven model provides a better estimation of the thermal management system and higher computational performance (0.2 seconds). However, this approach is highly dependent on the training dataset, thus requiring a large amount of data across different operating conditions. Through detailed benchmarking, this work provides not only a robust modelling foundation for the future development of predictive control strategies, but also operational guidelines for selecting the most suitable approach based on specific design requirements, paving the way for electric vehicles with greater range and optimized thermal performance.