Effect Analysis on the Maximum Nondimensional Temperature in the Cold Plate in Battery Thermal Management System-Based Artificial Neural Network

Abstract

A serpentine channel cold plate is a unique configuration of cold plate used extensively in battery thermal management systems due to its low-pressure drop and high heat transfer performance. Generalized analysis on serpentine channel cold plate for battery thermal management is very limited, especially using the finite element method (FEM). Through this study, we seek to obtain the maximum temperature on the cold plate subjected to uniform heat flux conditions from the Li-ion battery pack. The governing equations for the heat transfer through the cold plate under steady-state conditions are nondimensionalized to reduce the number of operating parameters from 12 to 4. The artificial neural network (ANN) is used to develop a correlation between nondimensionalized maximum temperature and the four nondimensional operating parameters. The ANN prediction has obtained a mean squared error (MSE) loss of the order of 10−6 and R2 value equal to 1 on the validation and test datasets. The temperature surface plots of the cold plate have been obtained for multiple channel configurations. The present study helps in reducing the overall computational time (59.13 s for 1296 simulations) and provides a generalized ANN-based correlation to predict the maximum temperature, which is vital to operate the battery under safe temperature limits.