Predicting and Preventing Thermal Runaway and Navigating the Regulatory Environment of Lithium-ion Battery Safety

By Kent Griffith 

May 4, 2021 | Lithium-ion battery safety is once again headline news following the recent Tesla Model S crash in Texas and subsequent fire that took firefighters four hours and 30,000 gallons of water to extinguish. Although agencies such as the National Highway Traffic Safety Administration (NHTSA) and Insurance Institute for Highway Safety (IIHS) have determined that electric vehicle safety is comparable to that of combustion engine vehicles in the event of a crash, the batteries do present unique challenges to first responders. Officials in Texas noted that a vehicle fire is normally handled in minutes, rather than hours. In addition, there are a number of reported incidents of batteries reigniting from residual energy in the cells after suppression of the initial fire. It is clear that lithium-ion battery safety measures and the regulatory framework surrounding these devices is becoming increasingly important with the rising number of electric vehicles and residential, industrial, and utility-scale energy storage projects.

Leaders in the field came together virtually at the International Battery Seminar this spring to share updates and best practices on the evolving landscape of battery safety. Because safety is tied to all aspects of battery chemistry, manufacturing, and use-cases, the representation of stakeholders was particularly diverse. Representatives presented perspectives from safety testing and standards laboratories, regulatory committees, safety-oriented start-up companies, battery associations, national laboratories, and the university research community.

On the standards side, Cindy Millsaps from Energy Assurance LLC, formerly of Underwriters Laboratories, described the anticipated changes to IEC 62133-2—written by the International Electrotechnical Commission—which is most important international standard for exporting lithium-ion batteries. Initially released in 2002, IEC 62133 underwent major revisions in 2012 and 2017, including separating lithium systems into a distinct category from other rechargeable battery chemistries. Five years is a long period in the rapidly-expanding lithium-ion battery industry so the new version expected to come out in 2022 will contain many updates. In particular, changes in the test procedures of high-rate capable cells and systems are coming to reflect the growth in fast charging (see, Chemical and Engineering Solutions to Faster Charging Lithium-ion Batteries).

Another major change is the expansion of testing protocols related to “forced internal short circuit,” commonly thought of in the community as the nail penetration test. While tests like this will still be allowed, the updates will allow for less dramatic procedures such as cell teardown combined with X-ray and computed tomography examination of safety-related cell characteristics like negative electrode/separator/positive electrode overlap. The negative electrode should be larger than the positive to prevent dendrite formation at the electrode edges that could lead to short circuits. Similarly, the separator should extend beyond the negative or positive electrode to prevent short circuits.

George Kerchner, Executive Director of the Portable Rechargeable Battery Association (PRBA), noted that important changes are coming to the International Fire Code. Specifically, the changes include a new section on lithium battery collection and storage regulations that could have major consequences across the industry. The proposed regulations include obtaining permits, devising fire safety plans, developing technical reports based on hazard levels and protection measures, and rules regarding the quantity of containment of indoor and outdoor battery storage. There will be exemptions for small quantities, defined as about two 55-gallon drums; batteries stored at less than 30% state-of-charge; temporary storage during the manufacturing process; and small batteries packaged or installed with equipment. Furthermore, major changes are coming to powered micromobility devices such as electric scooters and bicycles, which will soon require mandatory Underwriters Laboratories listings and well-spaced direct electrical connection requirements during charging.

Thermal management is a critical factor in battery safety. Dr. Chuanbo Yang of the National Renewable Energy Laboratory (NREL) noted that factors to consider for thermal runaway at the battery materials level include the onset temperature, self-heating rate, and total enthalpy (i.e., heat generated). For cathode materials, their data indicate that lithium cobalt oxide (LCO) has the largest enthalpy per unit capacity in thermal runaway followed by the lithium nickel manganese cobalt aluminum oxide chemistries (NMC, NCA), lithium iron phosphate (LFP), and finally lithium manganese oxide (LMO). Lithium manganese oxide had less than one-fifth of the heat generation than lithium cobalt oxide, supporting its use in applications where safety is particularly critical.

Among the approaches for managing thermal runaway, Yang discussed heat dissipation with vapor chambers, heat absorption with aluminum heatsinks, and isolation with flame-retardant foams. Research from NREL has shown that flame-retardant additives or coatings can shift the onset temperature for thermal runaway by more than 100 °C, decreasing the likelihood for total battery loss from an individual cell failure. Dr. Judith Jeevarajan from Underwriters Laboratories is interested in similar thermal runaway suppression methods and her team has recently carried out a number of experiments on modules of a 5´5 array of 18650 cells where the central cell was triggered into thermal runaway. In these tests, performed with the NASA Johnson Space Center, materials including ceramic blanket, fiber insulation, and mica tubing were not able to suppress full thermal runaway and the modules reached peak temperatures of 1000–1200 °C. However, Jeevarajan noted that the tests were performed at 100% state-of-charge, a harsh and high-energy condition that is more extreme than the state in which the batteries would normally be stored or shipped.

Christiane Essl of Graz University of Technology also studied fully charged cells. She looked at differences in thermal runaway triggers, examining overtemperature conditions, overcharge, and nail penetration. Key takeaways from her research include that overcharge produces the most gas evolution, nearly 2.5 L per Ah, the gas evolution speciation depends primarily on the electrolyte used, and that overtemperature and overcharging show two gas venting events while nail penetration leads to such rapid failure that only one gas release is observed (J. Electrochem. Soc., DOI: 10.1149/1945-7111/abbe5a).

Part of avoiding thermal runaway is predicting it in time to take preventative action. Dr. Yuliya Preger of Sandia National Laboratories emphasized the importance of discovering recoverable faults such as individual misperforming cells or modules in large energy storage systems for grid-scale applications. Early detection could avoid responses such as firefighting measures that are dangerous and result in the total loss of the system. While cell failure and safety events are rare, often pegged at around 1 in 10,000,000 cells, the enormous scale of utility energy storage systems translates to billions of cells. Of note, Preger mentioned that South Korea was forced to suspend operations at 522 energy storage system units in January 2019 following 23 fires. She proposed moving toward a model of predictive maintenance but notes that there are business and technical challenges.

Among the former issues is downward cost pressure leading to less investment in sensors and infrastructure. A promising consideration is that predictive maintenance is common across the energy and utilities sector with examples from oil pipeline corrosion, power plant upkeep, PV partial shading faults, and wind turbine gearbox repair.

Dr. Peter Attia, currently at Tesla, presented work on battery lifetime and failure prediction from his previous role as a student at Stanford University. Attia applied a variety of machine learning methods to commercial LFP//graphite cells and identified subtle features in the voltage curves that enabled early prediction of cell degradation (Nature, DOI: 10.1038/s41560-019-0356-8 and Nature, DOI: 10.1038/s41586-020-1994-5). Both Attia and Dr. Shriram Santhanagopalan of NREL talked about the importance of data quality for battery management and, in particular, emerging big data approaches. Santhanagopalan noted that there has been more of an emphasis on data quantity than data quality, and it is important to ensure that trends picked up by machine learning methods are both real and actionable, and not artifacts of instrumental precision or parameterization. Large datasets are becoming available not just with electrochemical data such as cell voltage and capacity but also temperature, pressure, online gas monitoring, and tomographic data from within the cell. With appropriate care and considerations of statistical significance, these datasets can help us make decisions on battery second-life applications, predict maintenance requirements, and prevent catastrophic safety events.