Data-driven optimization of nano-PCM arrangements for battery thermal management based on Lattice Boltzmann simulation

• Published in: Energy (Oxford) 2024-12, Vol.313, p.133670, Article 133670
• Main Authors: Zhang, Liwei, Shang, Bichen, Sun, Weijie, Tao, Yao, Li, Xueren, Tu, Jiyuan
• Format: Article
• Language: English
• Subjects: Back Propagation Neural Networks (BPNNs); Battery thermal management; Genetic Algorithm (GA); Lattice Boltzmann Method (LBM); Nano-PCM
• ISSN: 0360-5442
• DOI: 10.1016/j.energy.2024.133670

An efficient Battery Thermal Management System (BTMS) is vital for maximizing electric vehicle effectiveness and extending service life, essential for sustainable transportation. This study proposes a novel non-uniform nano-PCM distribution strategy within BTMS to tackle battery overheating challenges and further proposes a multi-objective optimization framework combining Back Propagation Neural Networks (BPNN) and Genetic Algorithm (GA) to achieve optimal design solutions for BTMS. Optimization data is derived from the well-validated Lattice Boltzmann Method (LBM) results across 343 cases. Initial evaluations show that a negative gradient distributed nano-PCM (Type 2) improves melting rate, heat dissipation power, and temperature uniformity by 4.67%, 4.87%, 19%, and 7.0%, respectively. The BPNN-GA optimization framework satisfactorily correlates nanoparticle distribution with four evaluation metrics, achieving R2 values from 0.9469 to 0.9987. Optimization improves melting rate, heat dissipation power, and regional and inter-regional temperature uniformity by 9.13%, 9.94%, 7.77%, and 29.73%, respectively. The BPNN-GA also demonstrates reasonable generalizability for the other two practical case scenarios with improvements in certain criteria up to 49.19%. This study highlights the potential of uneven nano-PCM configurations and the efficiency of the LBM-BPNN-GA framework in achieving superior thermal management for BTMS, which is expected to provide insights for future BTMS designs and implementations. [Display omitted] •A novel non-uniform distributed Nano-PCM configuration with BTMS is proposed.•Cooling performance of Nano-PCM is assessed via LBM studies across 343 cases.•A novel multi-objective optimization framework using BPNN and GA is proposed.•A negatively gradient-distributed Nano-PCM shows superior cooling performance.•BPNN-GA framework successfully improved multiple evaluation met-rics up to 29.73%.