Battery Thermal Runaway in Surprise, Arizona

What can we do to improve the safety of Battery Energy Storage Systems (BESS) facilities? In this article, James Close, CEng, MIChemE, examines a thermal runaway incident involving battery deflagration and fires, as well as the main contributing factors. Next, mitigation strategies are presented, along with recommended testing.

Surprise, Arizona, US Case Study

On April 19, 2019, a thermal runaway event occurred at a Battery Energy Storage System (BESS) in Surprise, Arizona, injuring four firefighters. We can see in the photos below that there was definitely a meltdown in rack 15. There was also a lot of debris. A lot of the BESS structure was crumpled and ejected outwards. This occurred because there was a secondary explosion that happened when firefighters arrived on the scene.


Source: Arizona Public Service McMicken Battery Energy Storage System Technical Analysis and Recommendations Report

Arizona Incident Timeline Start

16:54:30
The battery voltage dropped off in the 7th cell, module 2 rack 15 (406V to 3.82V).

16:54:38
The total voltage in rack 15 dropped from 799.9 to 796. 1V BSM loses module level data.

16:54:40
Temperature readings in the back of rack 15 begin to increase.

16:55:20
Smoke alarms 1 and 2 in the BESS are detected and the fire protection system is triggered and causes several circuit breakers to open.

16:55:45
A ground fault was detected.

Looking at the timeline, this is essentially how the incident played out. In rack 15, there was identified an internal short circuit within the pouch cell. This thermal runaway caused bursting within that cell, releasing toxic flammable gases. Then the cell caught fire and the fire spread throughout the module and then also rack 15. The fire was detected and the BESS did have a Novak 1230 dispenser, a fire suppression system, to suppress fires.

Arizona Incident Timeline Midpoint

16:57:00
APS contacted Fluence to verify the fire suppression system discharged.

16:55:50
The fire suppression system discharges Novec 1230 suppression agent and shuts down the ventilation system.

17:07:00
Fluence advises APS that its field service engineer is going to provide visual confirmation of the fire.

17:12:00
APS dispatches trouble man to the site to investigate the issue.

The problem was that even though the Novak 1230 gas was released and the fire was contained, unfortunately, the thermal runaway was still occurring, spreading through to rack 15. While the fires were well contained, the flammable gases being produced rose and accumulated on top of the BESS.

Arizona Incident Timeline Ending

17:40:00
Fluence field service engineer calls 911.

17:48:00
Fire department arrives.

20:02:00
Front door opened by emergency responders.

20:04:00
Explosion occurs.

Unfortunately, when the firefighters arrived on the scene and opened the door to the BESS, the Novak 1230 gases came out of the system, and then air rushed in. Then the flammable gases came, back down, and mixed with the air. And then, as mentioned, the thermal runaway in rack 15 was still continuing. The thermal runaway caused an ignition point and then a deflagration occurred.

Cause of the Incident

Although we're moving in a more sustainable direction, there are still a lot of chemical hazards we need to look out for. The main factors that contributed to this incident were:

• Internal failures in cells initiated thermal runaways;
• The fire suppression systems were not capable of stopping thermal runaway;
• There was also a lack of thermal barriers not only between cells but also between modules within the rack;
• The cooling system for maintaining the temperatures of the cells during operation, in my opinion, is what made things worse. The cells were air-cooled, so this spread the flammable gases throughout the energy storage system quicker, therefore it became a lot easier to collect the flammable gases into an accumulation at the top of the energy storage system.

All these factors, as shown below, unfortunately contributed to the explosion and fires of this Battery Energy Storage Systems (BESS) incident.

What factors contributed to the incident

The first step is determining what happened. There was some dendrite formation and some deformation internally within the cell. That caused internal short circuits and then a huge thermal runaway reaction. There was also some ejection of material in the cell, as mentioned earlier, some debris and flammable gases that accumulated and caused the deflagration. The thermal runaway spread was not being controlled. There was also the potential of secondary thermal runaways that occurred within the rack. Then the fire suppression systems just didn't do their job.

Mitigation Solutions

My method is to examine first what the issues were, explore different methodologies, and then take an optimistic approach to improve safety. The current approaches that we have available to us include thermal runaway modeling, similar to performing thermal runaway dynamic modeling of a storage tank. You can do a detailed kinetic model or an ISO conversion model, which is just as helpful.

The next method is propagation modeling and design which essentially looks at the thermal runaway spread from cell to cell. But this method also has to be backed up by data, some validation.

At ioKinetic, we have the capability to do single cell module tests using an EVx Calorimeter instrument. But one of the big issues around 2020-2021 was that there was no discussion about any gas ejection or any material ejection, analyzing the flammable limits of these gases being produced during that thermal runaway. That is becoming crucial now.

In terms of approved methodologies, we can perform computational fluid dynamics (CFD) modeling. We have done 1D explosion dynamics to model BESS deflagration and can size blast panels using Process Safety Office^® SuperChems^® software, which looks at the Layers of Protection (LOPA) aspect. This is very helpful for these sorts of installations.

If we look at the module, pack, or cell level, we really need to understand the energy distribution because this will help us design better modules, and also different thermal-resistant materials and thermal barriers to help prevent the propagation to the rack itself.

We have lots of tools at our disposal at the ioKinetic lab. We do more component testing to look at the individual chemistry of electrodes and electrolyte mixtures, things like that. We also do a lot of heat capacity and conductivity testing. This is very important if you wish to understand the heat of reaction of your battery and what your heat capacity is. For heat transfer calculations, you need to understand the thermal diffusivity conductivity. It is absolutely vital to have those accurate measurements to simulate and better improve thermal barriers and double resistances at a module level.

We do a lot of battery testing using an EVx calorimeter to test different chemistries and different gas collection methodologies to characterize things like temperature rises and pressure rises. We can store the gas and run actual testing in terms of composition, and then also use Process Safety Office^® SuperChems^® software to see what type of combustion reactions we would get, flammability limits, etc.

We can take all these experimental data, all these pressure measurements, voltage measurements, and then the collected gas to put it through GC mass spectrometry and really identify what kind of compositions we have, and then really nail down the flammability limits.

This really helps from a standards point of view and a regulatory point of view. For example, after this incident occurred, regulation UL9540A changed its testing structure. If you wish to follow the UL9540A testing structure now, you are required to do a single cell level test, module test, rack test, and then an installation level test.

Throughout all of those levels, you need to understand the gas ejection, what the gas composition is, what flammability limits it has, and also the diesel migration potential. Battery testing is really good for meeting compliance and also design purposes as well, to make safer products, safer batteries.

But we are not just stopping at the GC mass spectrometer, we're taking it a step further. I mentioned that the spread and the understanding of the distribution of energy is really important when dealing with single cells, single cells to modules, and things like that. The EVx Calorimeter instrument has its benefits, but what we really need to start thinking about is the distribution of energy during a thermal runaway when cells go off.

There's a huge amount of ejected material coming from either the top or bottom of the cell. It can go in either direction. Then also the currents can get very, very hot and melt away. We need to understand what the distribution is between the vented material and then also the EVx Calorimeter itself, because currently this equipment is great for a sort of lumped look at the thermal runaway, but do need to take it a step further and understand geometrically where the energy distributions are.

This brings me to the Fractional Thermal Runaway Calorimetry (FTRC) instrument. This system can quantify the distribution of energy based on the location of the cell. When we enter a cell into this chamber in the middle, it will trigger a thermal runaway either by overheating, or nail penetration, or another form of overcharge or mechanical deflagration. The FTRC captures and quantifies the energy at the actual cell itself, and then also the vented gas and the potential solids coming out.

The FTRC allows us to quantify the energy distribution down to every component. The results are presented as a pie chart that will be incredibly handy for any future design improvements and modeling purposes. We are able to quantify what portion of the energy is through the actual casing and then what portion of the energy is distributed through the ejector. A good majority of that energy goes through the ejected material, 74.7%, and then the remainder of the proportion, 21.7%, typically goes through the main cell body itself.

The FTRC data is very important because when we understand the distribution of the cell, we can put this into our battery modeling and design and start to answer two questions. We can see the fire suppression system's effectiveness and look at the energy carryover from fireball materials. We can also start to look at how effective the barriers would be in terms of stopping thermal runaway. For example, one of the current big challenges is that a lot of people are working on thermal barriers, but they place cells with no barriers in between. Sometimes they are shocked to find out that, depending on the thermal barriers they use, a thermal runaway still occurs further down the cell module.

What I suspect, and what I've also pursued in my doctoral research, is that a significant amount of energy is being carried and distributed further across the module by the temperatures, combustion processes, and energy that are happening. We need to understand this distribution of energy a lot better to solve this problem and make these systems a lot more safe.

Below is an example of thermal propagation modeling. We need to understand how these energy distributions work. The graph below shows that if I didn't change the efficiency, it would cause these problems.

Example of energy distribution data modeling

Data from the FTRC shows the energy released via the gas and the ejected flames. This equipment will help battery designers, module designers, and battery safety system designers optimize their thermal runaway mitigation processes and improve overall safety.

We Can Help

Lithium-ion batteries are great in terms of sustainability for capturing energy and discharging it, but they are highly thermally unstable. Thermal runaways can result in fire and spread from module to module and vertically in a pack. During thermal runaway, lithium-ion batteries release flammable and toxic gases. These gases can accumulate in a BESS if ignited and also cause deflagrations.

Thermal runaway models, safeguard and mitigation designs, the usage of firefighting equipment, thermal barrier designs, and other applications can all benefit from battery testing from ioKinetic.

Our experts can work with you to determine the level of analysis that is required for your particular needs. Call us at 1-844-ioKinetic or send us a note. We'll be glad to help.