Investigation and modeling of cyclic aging using a design of experiment with automotive grade lithium-ion cells

60 large format automotive grade lithium-ion pouch cells with graphite/NMC chemistry are tested following a design of experiment. Realistic driving profiles resembling a plugin-hybrid electric vehicle are used with variation of five aging factors: temperature, maximum and minimum state of charge, charging power, and the ratio of charge depleting vs. charge sustaining cycling. Capacity fade is cleaned from calendar aging and multivariate stepwise linear regression is used to parameterize an empirical model of cyclic capacity fade. Temperature and the ratio between charge depleting and charge sustaining cycling show the biggest impact on cyclic aging, whereas charging power has little effect in the chosen range of aging conditions. The importance of considering interdependencies between aging factors for modeling is pointed out, major interdependencies are found between the factors temperature and charging power and between minimum and maximum state of charge. Leave-one-out cross validation is used to show the capability of the comparatively simple model approach to predict cyclic aging within the tested range.