Artificial neural networks-based multi-objective optimization of immersion cooling battery thermal management system using Hammersley sampling method

Abstract

This research optimizes lithium-ion battery module cooling through immersion cooling, addressing pressure drop and after discharge average cell temperature. Using the Hammersley method, various module designs are generated. Multi-objective optimization, using ANN-based multi objective genetic algorithms, is conducted on a 16S1P configuration at 4C discharge and 0.008 kg/s. The optimized design achieved an 83 % average cell temperature reduction at a 4C discharge rate and 0.008 kg/s compared to an uncooled battery cell, while also reducing the pressure drop by 88.6 % relative to the base design. The pressure drop is approximately 12 Pa at a mass flow rate of 0.02 kg/s, with an average cell temperature of 3.13°C in the optimized design. This represents a 68.4 % reduction in pressure drop compared to the base design, which experiences approximately 40 Pa at a lower mass flow rate of 0.008 kg/s. Additionally, the optimized design achieves a 20.8 % reduction in average cell temperature, lowering it from 3.95°C in the base design to 3.13°C. These findings highlight improved pressure and thermal performance in lithium-ion battery modules, with implications for enhanced design and operation. Future work could extend these optimizations to various battery chemistries and conditions.

Keywords

• Optimization
• Battery thermal management
• Immersion cooling
• Artificial neural network
• Hammersley sampling method