Data-Efficient Machine Learning Surrogate Modelling for Battery Cooling Channel Design and Thermal Performance Prediction

This project aims to investigate whether data-efficient machine learning models can accurately predict the thermal performance of electric vehicle battery cooling channels without requiring computationally expensive simulations for every new design.

The research will involve developing a range of realistic cooling-channel geometries using Computer-Aided Design (CAD) and evaluating their heat transfer, temperature distribution, and pressure drop through Computational Fluid Dynamics (CFD) simulations. The resulting data will be used to train and compare machine learning surrogate models capable of predicting cooling performance under different geometric configurations and operating conditions.

A key focus will be evaluating how effectively these models generalize to previously unseen cooling-channel designs, particularly when trained on limited simulation data. The project will also investigate prediction accuracy, computational efficiency, and the limitations of applying machine learning to thermal engineering problems.

Expected outcomes include a structured simulation dataset, a validated machine learning benchmark, reproducible code, and a research manuscript presenting the findings. By combining computational modelling with data-driven analysis, this research aims to contribute to more efficient and accessible thermal management design methods for sustainable electric vehicle technologies.