Predictive hazard level assessment of Li-ion cell thermal runaway failure

Li-ion batteries (LIBs) are a widely adopted energy storage device that are increasingly used in transportation and stationary applications. However, LIBs come with the risk of thermal runaway (TR) which is known to be unpredictable. Previous computational work has assessed the sensitivity of TR to model inputs, while some experimental work has quantified the distribution of TR behaviour. However, no work is known to have used computational simulations of cell abuse to predict the probability of cell failure under typical abuse test standards. This work applies an abuse model to achieve this, as well as using key TR output variables to calculate the magnitude of cell failure according to the redefined EUCAR hazard level assessment. Abuse simulations of Underwriter Laboratory's oven test are simulated thousands of times considering parameter distributions with two different coefficient of variance sets. This work shows that it is possible to predict the change in the probability of failure against the change in oven temperature and the probability of different hazard levels. However, there is a need to better understand and refine the variance in cell parameters, specifically those related the kinetic behaviour, to allow for analysis that is more suitable for risk assessment purposes.