Modeling degradation of lithium-ion batteries considering cell-to-cell variations. (15th December 2021)

• Record Type: Journal Article
• Authors: Galatro, Daniela; Romero, David A.; Freitez, Juan A.; Da Silva, Carlos; Trescases, Olivier; Amon, Cristina H.
• Is Part Of: Journal of energy storage. Volume 44(2021)Part B
• Publication Date: 2021-12-15
• Subjects: Lithium-ion batteries -- Degradation -- Cell-to-cell variation -- Non-homogeneous gamma process -- Sampling strategy; AF Acceleration factor -- BMS Battery management system -- BOL Beginning-of-life -- BTMS Battery thermal management system -- CDF Cumulative distribution function -- DOD Depth-of-discharge -- ECM Equivalent circuit model -- EOL End-of-life -- EV Electric vehicle -- GOF Goodness of fit -- KS Kolmogorov-Smirnov -- LCO Lithium cobalt oxide -- LIB Lithium-ion battery -- MAPE Mean absolute percentage error -- MLE Maximum likelihood estimation -- MM Method of moments -- NHGP Non-homogeneous gamma process -- NMC Nickel manganese cobalt oxide -- SEI Solid electrolyte interface -- SOC State-of-charge -- SOH State-of-health -- SSPE Sample size percentage error -- TG Temperature gradient -- TR Thermal runaway; Energy storage -- Periodicals; Energy storage -- Research -- Periodicals; 621.3126
• DOI: 10.1016/j.est.2021.103478
• Languages: English
• ISSNs: 2352-152X

Abstract

Highlights: A 3 –parameter non-homogeneous gamma process is used to account for the cell spreading in lithium-ion batteries. This approach allows predicting capacity-fade and time-to-failure for any battery architecture. The distributions of the fitted degradation data are adjusted with acceleration factors. Abstract: Battery packs exhibit both intrinsic cell-to-cell variations and spatio-temporal cell-to-cell differences in temperature and other stress factors, shaping the evolution of the degradation paths of the cells. To account for these variations and differences in degradation or cell spreading, we propose a statistical approach for modeling the degradation of lithium-ion batteries that utilizes a 3-parameter non-homogeneous Gamma process. This approach predicts the capacity fade or time-to-failure for any battery architecture and adjusts the distributions of the fitted degradation data of the cells with acceleration factors. At the pack level, cells are modeled with compositions of Gamma-distributed variables for configurations in parallel and series. Actual values of the capacity fade or time-to-failure at different thermal conditions are compared with predicted values, showing relative errors in the range 1 – 12%. We also present a methodology for estimating the minimum number of cells required for modeling the evolution of the spreading and degradation paths by analyzing the effect of the sample size on estimating the degradation for different sets of cells. This sampling strategy is particularly useful for reducing the computational cost of running simulations required for designing battery packs, battery management systems and battery thermal management systems.