J. Sherman, A. Bombik
University of North Carolina at Charlotte,
United States
Keywords: Battery, Safety, State of Health
Summary:
Lithium ion batteries are commonly used in hybrid and battery electric vehicles which are often involved in car crashes. Previous nondestructive abusive tests on cells with varying chemistries and states of charge have shown reduction in performance in the near and long term. These cells often are either continued to be used after impact which, leads to reduced life cycles and increased cost to the consumer, or used in second life scenarios which adds complications to the implementation alongside other uncompromised packs. To investigate this response, fully discharged commercial LCO pouch cells were cycled and damaged with a drop weight at 3J to minimize short risk and allow for future cycling. This post abuse cycling showed a reduction of capacity when compared to similarly undamaged cells demonstrating underlying electrochemical changes from the abuse. Equivalent resistances and capacitances from HPPC also demonstrated accelerated capacity decay instigated by mechanical damage. Disassembly of cells immediately after damage showed severe delamination of the anode electrode with minimal structural damage to the cathode, showing the primary cause for reduced performance is from the anode material. By performing ICA analysis on cells prior and after impact it was shown that such damage could be reliably detected by an on board BMS. Further investigation on cells from the same batch involved performing EIS measurements prior and after impact when the cells were discharged to critical voltages determined by ICA analysis. The resulting changes in EIS demonstrated underlying changes in charge transfer, SEI and diffusion kinetics from mechanical impact, which was then used as a prognostic tool to predict future reduction in performance from mechanical impact. This work allows for detection of nondestructive mechanical damage in batteries, which then can be extrapolated for future performance predictions in continuing use scenarios or for second life applications.