Fire & explosion hazards due to thermal runaway propagation in lithium-ion battery systems

Archibald, Erik Johnson
Journal Title
Journal ISSN
Volume Title

Lithium-ion battery technologies are increasingly used in electric vehicles as well for electrical energy storage at residences, businesses and utilities. In a failure event, these cells may produce large quantities of gas that pose fire, explosion and toxicity hazards to building occupants and firefighters. To understand the hazard, it is important to understand how single cells fail and how that failure propagates to other cells. Once the quantity and composition of the gas released by a single cell is understood and the failure process to surrounding cells is understood then models can be applied to quantify the explosion or fire hazard of the system.

A series of experiments is conducted using single cell and linear 1D arrays of lithium-ion pouch cells to understand the thermal runaway process for a single cell and compare it to how thermal runaway propagates through an array of cells. Due to the difference in heating between single cells and cells failing in an array, there are differences in failure characteristics. Single cells vent gases faster and more violently than those failed in an array. This causes damaged cells to look very different. An idealized process for thermal runaway propagation in an array of cells is presented.

Many systems have complex geometry in which heat transfer via conduction, convection and radiation may cause thermal runaway propagation in 3 dimensions. Experiments are conducted with individual prismatic lithium-ion cells and modules comprising 14 cells to study the more complex runaway propagation behavior of commercially available energy storage modules. The quantity and species of gases released by a single cell is measured. The propagation of thermal runaway through the 14 cells in a module is observed for both a module in open air and modules in a rack. The impact on the temperature and heat flux within the rack and throughout the room is measured. A computational fluid dynamics model is developed and used to predict temperatures, heat fluxes and gas concentrations in the compartment.

Gas sensing, ignition and control systems are developed to perform explosion experiments with lithium-ion cells. Explosion experiments are performed at lab scale, intermediate scale and full-size closet scale. Full scale experiments are conducted in a closet which reveal that a single 94 Ah cell provides both the fuel and ignition to cause a partial volume deflagration which breaks the closet door.

Once gas release and propagation behavior are understood, a series of models can be used to estimate explosion consequences. Simple models are described to predict possible flammable gas mixing prior to an explosion. Battery vent gas compositions are used with models to calculate flammability limits, laminar flame speed, maximum adiabatic pressure and other properties. Gas mixture properties and compartment geometry are used in a 0D deflagration code to predict pressures and impulses. Single degree of freedom calculations are used to predict possible consequences from the calculated pressures and impulses. Models are validated against experimental literature and demonstrated in case studies. Flammability properties are calculated and summarized for a database of gas composition data. Gas properties are used to determine minimum amounts of energy storage required to cause an explosion.