Research data - Physics-informed aging-sensitive equivalent circuit model for predicting the knee in lithium-ion batteries

This dataset contains the research data (Matlab code, parameter sets, measurement data, figure files) of the journal article: Patricia O. Mmeka, Matthieu Dubarry, and Wolfgang G. Bessler, “Physics-informed aging-sensitive equivalent circuit model for predicting the knee in lithium-ion batteries,” DOI: 10.1149/1945-7111/adf9cb Abstract: Predicting the lifetime of lithium-ion batteries is a complex task due to the nonlinear and strongly coupled interactions of various aging mechanisms, resulting in nonlinear aging dynamics, referred as "knee" phenomenon. While physicochemical models capture these dynamics, they are often computationally intensive and involve complex parameter derivation. In this study, a physics-informed equivalent circuit model is proposed. The model integrates key aging modes (loss of active material at positive and negative electrodes and loss of lithium inventory), and describes their kinetics using physics-informed rate laws. Rate expressions for each aging mode include a calendric component (dependent on half-cell potential) and a cycle component (dependent on current and relative volume changes in active materials). The model was calibrated using measured degradation modes of a 3.35 Ah nickel cobalt aluminium oxide and graphite lithium-ion cell, which was aged calendarically and cyclically using a dynamic stress test. The equivalent circuit model demonstrates good agreement with experimental aging data (capacity loss, internal resistance increase) and successfully predicts the knee in the capacity loss curve, where the knee formation could be primarily attributed to the onset of lithium plating. This occurs when increased resistance at the negative electrode causes its potential to drop below the plating potential. Copyright and IP information: All supplementary material (c) 2025 Patricia Mmeka and Wolfgang G. Bessler. The MATLAB code and research data provided are licensed under the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license. Quick start: Copy all files expcept the Figure1-8.zip into one folder. Open and run "Figure7Simulation.m" with Matlab. Adjust the number of cycles and acceleration factor accordingly to obtain the desired cyclic aging results. Description of the files: Matlab code (tested using versions R2019a): Figure7Simulation.m: this script simulates cycling aging during the DST (dynamic stress test), hence produces the capacity loss and resistance increase shown in Figure 7 LIBquivAging.m: this script contains the Matlab class which represents a virtual lithium-ion battery PanasonicNCR18650B.m: this script imports all battery parameters of the investigated NCA/graphite cylindrical cell Cell parameter data: ResistancesAlawa.mat: MATLAB data file containing resistance values as function of current and SOC GraphiteAlawa.dat: text file containing the half-cell data of the NE (i.e. tabular data with Stoichiometry x [], dH [J/mol], dS [J/mol/K]) NCAAlawa.dat: text file containing the half-cell data of the PE (i.e. tabular data with Stoichiometry x [], dH [J/mol], dS [J/mol/K]) Experimental data: Dubarry 2019 Batteries Capacity Fig. 3a.csv: the excel sheet contains the capacity fade data from Baure and Dubarry 2019, Batteries 5:42; doi:10.3390/batteries5020042 Dubarry 2019 Batteries Degradation Modes DST Fig. 9.csv: the excel sheet contains the degradation modes (LLI, LAM_NE and LAM_PE) data from Baure and Dubarry 2019, Batteries 5:42; doi:10.3390/batteries5020042 Dubarry 2019 Batteries IR Fig. 3b.csv: the excel sheet contains the internal resistance increase data from Baure and Dubarry 2019, Batteries 5:42; doi:10.3390/batteries5020042 Figure files: Figures1-8.zip: contains .TIFF (Tagged Image File Format) and .fig (Matlab format) versions of Figures 1-8 of the paper.