New findings suggest a method to foresee and mitigate temperature surges and fires in lithium-ion batteries, which are often used to power electric vehicles.
Keeping electric vehicle batteries cool is vital due to the safety risks associated with temperature increases, which can have severe outcomes.
A recent study conducted by a doctoral student at the University of Arizona introduces a technique to predict and prevent overheating in the lithium-ion batteries used in these vehicles.
The research article titled “Advancing Battery Safety,” spearheaded by Basab Goswami, a doctoral student in the College of Engineering, has been published in the Journal of Power Sources.
Backed by a grant of $599,808 from the Department of Defense’s Defense Established Program to Stimulate Competitive Research, Goswami along with his advisor, Professor Vitaliy Yurkiv of aerospace and mechanical engineering, created a system employing multiphysics and machine learning models. This system detects, forecasts, and identifies overheating of lithium-ion batteries, known as thermal runaway.
According to Goswami, this system could eventually be incorporated into the battery management systems of electric vehicles to prevent overheating, ultimately ensuring the safety of drivers and passengers.
“The transition to green energy is necessary,” stated Goswami, “but it comes with safety risks related to lithium-ion batteries.”
Learning from the past to anticipate the future
Thermal runaway poses significant dangers and is notoriously hard to foresee.
“When a battery’s temperature rises, it escalates rapidly, potentially igniting a fire,” Goswami explained.
An electric vehicle battery is made up of multiple interconnected battery “cells.” Modern electric vehicles can contain over 1,000 cells in each battery pack.
When thermal runaway affects one cell, the adjacent cells are highly likely to heat up as well, triggering a catastrophic chain reaction. If unchecked, this could lead to the entire battery pack exploding, according to Goswami.
To avert such outcomes, the researchers recommend installing thermal sensors around the battery cells. These sensors would feed historical temperature data into a machine learning algorithm, which would then forecast future temperatures, effectively predicting where and when a thermal runaway incident could begin.
“Understanding where the hotspot originates (the start of thermal runaway) enables us to devise solutions to stop the battery before it reaches that critical point,” stated Goswami.
Yurkiv expressed his admiration for the accuracy of Goswami’s algorithm, noting that prior to this research, machine learning models had not been employed for predicting thermal runaway.
“We were surprised by how much more effective machine learning is in predicting thermocouple temperature and pinpointing hotspot locations with such precision,” added Yurkiv. “No human analyst could achieve that level of accuracy.”
This research builds on previous work by Goswami and Yurkiv published in January, which explored using thermal imaging to predict thermal runaway—an approach that required bulky imaging equipment for ongoing monitoring.
In contrast, the method proposed in their recent paper is lighter and more cost-efficient.
Addressing global demand
Goswami’s findings come at a significant time in the history of automobile manufacturing in the U.S. In July, coinciding with the publication of the research, the Biden administration unveiled a $1.7 billion investment in electric vehicle manufacturing in eight states. In 2023, sales of electric vehicles surged by 35% compared to 2022.
As demand for electric vehicles grows, implementing safety measures becomes increasingly important, as noted by Goswami.
“Many consumers remain wary of adopting batteries due to safety concerns,” he mentioned. “For the public to accept electric vehicles broadly, it’s vital to demonstrate that active research is addressing these essential safety challenges.”
