A UQ-Based Methodology for Rational Upscaling of Battery Failure Results for Maximum Credible Event Analysis
Abstract
For many reasons, the United States Department of Defense has made electrification of vehicles and systems a priority. A critical component for electrifying their systems is energy stored in lithium-ion cells. Army vehicles will use lithium-ion batteries for starting engines, hybrid-electric propulsion, and, eventually, fully electric propulsion. The Navy will use energy stored in lithium-ion cells on surface ships to improve efficiency and meet mission goals and transport Army vehicles containing these cells. While lithium-ion batteries are generally safe, under some conditions they can fail in a catastrophic event called thermal runaway. There are challenges in extinguishing incipient battery thermal runaway failures and there are concerns that such fires could lead to widespread damage to the vessel. For commercial vessels, there are already examples and concerns that transport of lithium-ion powered vehicles and technologies could lead to unacceptable losses.While some limitedexperimental data exist for large-scale failures involving lithium-ion cells, there is very little in the way of model framework that allows an analyst conducting a hazard analysis to quantitatively assess the maximum credible event that might result when a given cell chemistry is packaged in a specific module type, within a particular vehicle, in a container or compartment. While the Navy#s existing safety certification process has been successful, there are opportunities to improve it by reducing the time and cost required to assess new systems. A modern approach to safety certification includes an element known as certification by analysis. Applying the certification by analysis framework to lithium-ion safety certification would require a set of tools that can upscale results of cell and module experimental test data in a rational manner to make predictions that a decision maker might use in either requesting additional test data or increasing the level of mitigation technologies desired to meet a particular level of safety.The objective of this project is to develop a modeling and experimental testing framework that can be used to improve decision making on the risks from failures of lithium-ion cells in various formfactors and packages. To meet this objective, an uncertainty quantification and propagation framework will be developed to determine and validate maximum credible events for a given cell failure event. We will use existing and new data from Navy labs, our lab, and the literature at the cell scale, module scale, vehicle scale, and compartment scale to parameterize models from various sources. We will identify what level of models and sets of parameters are most important for upscaling cell level failure to site level consequences. We will develop and validate a workflow for coupling models at different scales for assessing site scale failure characteristics.We will leverage our experience in experimental and modeling characterization of lithium-ion battery failures at various scales for this project. At the largest scales, we have conducted house scale tests of battery failures with a variety of lithium-ionpowered devices and ordinary combustible fuels. We have also conducted tests at the compartment scale (e.g., shipping container scale) where we have designed tests to result in either an explosion or a fire. Models have been developed and used for these scenarios. At smaller scales, we will leverage data and models for single-cell failure and cell-to-cell thermal runaway propagation. There is a need to couple the range of models that we have used and that exist in the literature in a rational manner to generate statistically relevant results. It is our hypothesis that by statistically parameterizing existing models, it will be possible to couple these models in a way to generate statistical predictions useful for a risk analyst in making decisions about the likelihood of complex, site-scale, failure events.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 29, 2023
- Source ID
- N000142312552
Entities
People
- Ofodike A Ezekoye
Organizations
- Office of Naval Research
- United States Navy
- University of Texas at Austin