Experimental and Computational Analyses of Thermal Runaway in Lithium Iron Phosphate Cells and Battery Packs

Thermal runaway (TR) is a significant safety concern of Li-ion Batteries (LIBs) that can lead to hazardous events, such as fire and explosion of the battery, that presents a great risk to persons/assets in the vicinity of the battery system. However, through experimental investigation and computational modelling this phenomenon can be better understood, and hence, allow for improved LIB design that leads to safer systems. The aim of this work was to further the understanding of TR in LFP LIBs, in conjunction with developing improved theory and greater understanding of LIB TR models. The work is split into two main studies. First, experimental studies, in which the TR behaviour of LFP cells by accelerated rate calorimetry (ARC) and oven testing is investigated to carry out a novel assessment of LFP cells at various state of charge (SOC) and under rapid heating scenarios. Second, computational studies, in which a novel advanced abuse model (AAM) is developed to model the TR of LIBs, which is then parameterised for the study of LFP cells and extended to investigate the thermal runaway propagation (TRP) resilience of LFP LIB packs. Within the development of the AAM a novel representation of the cell pressurisation is considered, viz. assuming that the electrolyte/gas mixture within the cell is at bubble point. From the experimental work, results show, at SOC of 100% and 110%, the negative and positive electrode reactions are the main contributors to TR, while at lower SOC it is the negative electrode reaction that dominates. Cells at 100% SOC exposed to high temperatures during oven tests show, along with the ARC analysis, that the presence of the cathode and electrolyte reactions leads to an increase in the severity of a TR event for oven temperatures above 200°C. By comparing the heat generated in ARC and oven testing, it is shown that ARC does not fully capture the self-heating and TR safety hazard of a cell, unlike oven testing. This work gives new insight into the nature of the decomposition reactions and also provides an essential data set useful for model validation which is of importance to those studying LFP cells computationally. In developing the AAM, the novel bubble pressure assumption is validated against experimental data, and it is shown that the AAM significantly improves the predictions of time to TR and of temperatures after TR. Further, it is shown that there is significant uncertainty in the parameters defining the decomposition reactions for LFP cells. Importantly, cell pressurisation is most dependent on the gases released by the solid electrolyte interphase reaction, and venting is dependent on cell burst pressure and reaction activation energies. The AAM is essential for accurate abuse modelling, due to its improved temperature predictions, and considerably enhances the LIB TR field of study. In studying the TRP potential of LFP packs it is shown that TRP does not occur when an initiation cell undergoes total internal short circuit. This occurs, as, even though the abused cell undergoes complete TR, the amount of energy released by the abused cell does not raise the temperature of adjacent cells to the point that the energetic NE reaction develops significant decomposition rates. It is also shown that given different reaction parameters that lead to similar TR events of a single cell do not lead to significant variation in pack results. Finally, lower cell surface emissivities are shown to reduce the overall cell-to-cell heat transfer, and hence can enable safer lighter packs by simple cell surface alterations.