Analyzing cell-to-cell heterogeneities and cell configurations in parallel-connected battery modules using physics-based modeling

TL;DR

This work tackles cell-to-cell heterogeneity in parallel-connected battery modules by integrating experimentally validated physics-based ESPMs with a thermal-aging model and a data-driven multi-linear regression (MLR) analysis. It identifies electrode active material volume fraction variations ($\epsilon_{s,n},$ $\epsilon_{s,p}$) as primary drivers of capacity, energy, and thermal gradients, while interconnection resistance $R_{int}$ and SOC/temperature dependencies also shape outputs. The study demonstrates a simple yet effective cell-arrangement strategy—placing higher-capacity cells at the module entrance to reduce gradients and aging variability—yielding a 51.8% reduction in thermal gradients and a 5.2% reduction in long-term energy loss in aging simulations. These insights inform module design and BMS strategies to improve reliability and longevity of parallel-connected packs, especially for second-life applications. The approach combines high-fidelity modeling with scalable statistical analysis to illuminate how manufacturing tolerances propagate to system-level performance.

Abstract

In parallel-connected cells, cell-to-cell (CtC) heterogeneities can lead to current and thermal gradients that may adversely impact the battery performance and aging. Sources of CtC heterogeneity include manufacturing process tolerances, poor module configurations, and inadequate thermal management. Understanding which CtC heterogeneity sources most significantly impact battery performance is crucial, as it can provide valuable insights. In this study, we use an experimentally validated electrochemical battery model to simulate hundreds of battery configurations, each consisting of four cells in parallel. We conduct a statistical analysis to evaluate the relative importance of key cell-level parameters, interconnection resistance, cell spacing, and location on performance and aging. The analysis reveals that heterogeneities in electrode active material volume fractions primarily impact module capacity, energy, and cell current, leading to substantial thermal gradients. However, to fully capture the output behavior, interconnection resistance, state of charge gradients and the effect of the temperature on parameter values must also be considered. Additionally, module design configurations, particularly cell location, exacerbate thermal gradients, accelerating long-term module degradation. This study also offers insights into optimizing cell arrangement during module design to reduce thermal gradients and enhance overall battery performance and longevity. Simulation results with four cells indicate a reduction of 51.8% in thermal gradients, leading to a 5.2% decrease in long-term energy loss.

Figures (10)

• Figure 1: (a) Pseudo OCV curves of the fresh 19 cells under C/20 discharge test procedure at 23 $^{\circ} C$. (b) Boxplot of cells ohmic resistances at 10% SoC intervals at 23 $^{\circ} C$
• Figure 2: Schematic representation of the physics-based electrochemical-aging-thermal model for the battery module, where the module thermal model and the cell-level electrichemical model are highlighted.
• Figure 3: Single-cell parameter results for the 19 LG M50T cells. (a) and (c) voltage profile and SOC comparison between experiments and model simulations for cell P12 undergoing the validation cycle, respectively. (b) RMES between experimental voltage and model predictions for all cells. (d) RMSE between reference SOC and electrode SOC for all the cells.
• Figure 4: (a) Comparison of voltage profiles between experimental data and model simulations for $R_{int} = 0,1, \text{and } 3 \text{ m}\Omega$ at $25^{\circ} C$. (b) RMSE across all the considered experiment scenarios.
• Figure 5: Contour plots of the cumulative MSE for (a) cell currents and (b) temperatures. The cumulative MSE for cell currents, $\sum_{k=1}^{N_p} \text{MSE}(I_{\text{cell}}^{[k]})$, and for temperatures, $\sum_{k=1}^{N_p} \text{MSE}(T_{\text{cell}}^{[k]})$, are calculated based on the MSE values reported in Table \ref{['Tab:MSE']}.