Predictive modelling of hybrid phase change material fin-cooling for Lithium-Ion batteries using machine learning

Abstract

Effective thermal management is critical for lithium-ion battery safety and performance, especially in electric vehicles, where thermal runaway poses significant risks. While hybrid cooling systems combining phase change materials (PCMs) and fins show promise, predicting their thermal behaviour remains a key challenge. This study develops a data-driven predictive framework to optimize a hybrid PCM-fin cooling system, integrating numerical modelling with advanced machine learning (ML). A pressure-based finite volume method solves the governing equations, incorporating phase change via the enthalpy-porosity approach, and is rigorously validated against experimental data for PCM melting and battery discharge. The core novelty lies in the comparative development of five ML models—Artificial Neural Network (ANN), Support Vector Regression (SVR), Random Forest, XGBoost, and Gradient Boosting—to predict temperature distribution and system performance. The ANN model delivered RMSE <0.051 K, MAE < 0.04 K, and R2 > 0.94—markedly outperforming SVR and tree-based methods. The three-fin hybrid cooling system reduced peak surface temperature by 2.1 K, cut thermal gradients by 75.67 % during charging, and improved PCM melting uniformity by 9.5 % during discharge. This work provides a robust ML-assisted predictive framework for advanced battery thermal management. The validated ML-based temperature prediction framework offers a computationally efficient tool for the design, optimization, and real-time thermal management of lithium-ion battery systems, establishing a practical foundation for future multi-objective optimization and intelligent battery pack design.

More information

Identifiers

ORCID for Nader Karimi: orcid.org/0000-0002-4559-6245

Catalogue record

Export record

Altmetrics