Evaluation of Fire Spread and Suppression Techniques in Micro-Mobility Battery Packs

Lithium-ion battery packs used in micro-mobility devices, such as e-bikes and e-scooters can lead to substantial safety hazards should a single cell go into thermal runaway. In this paper we explore the extent and severity of e-mobility battery fires resulting from a single cell thermal runaway failure and evaluate various suppression techniques a user may attempt to implement if they experience a battery fire at home. We tested a household water hose as well as different fire blankets deployed both before the forced thermal runaway event and after initiation. The water hose was unable to supply a sufficient amount of water to extinguish the thermal event, however, the average pack temperature was decreased and the cell-to-cell propagation rate was slowed. Neither fire blanket tested was able to contain the flames or debris ejected from the battery packs and both acted to hold in the heat from the event, increasing the temperature, rather than allowing it to dissipate. In addition, we also demonstrated how various design approaches, such as added thermal insulation between cells, can help prevent cell-to-cell propagation and reduce the severity of a battery pack failure.

In the past decade, there has been a rapid growth of electric micro-mobility transportation modes as citizens, businesses, and governments look to adopt more environmentally benign and efficient transportation options.In 2022, the global micro-mobility market share was estimated to be $89.4 billion USD and is projected to increase to at least $210 billion USD by 2030. 1,2D'Acierno et al. recently compared the environmental impact of different transportation systems in urban settings and showed the advantages of alternative micro-mobility systems. 3The increasing number of battery-powered micro-mobility devices available on the market places a growing importance on battery cell, pack, and device manufacturers, as well as on governments, to design, build, maintain, and regulate the production of safe, reliable battery packs.As a result of the high energy density of micro-mobility battery packs, lack of high-quality design and manufacturing control, and the unfamiliarity of general consumers with large lithium-ion batteries, numerous safety incidents have been associated with micro-mobility battery pack failures.Over the past couple of years, substantial media attention has been given to fire safety concerns surrounding electronic micro-mobility devices, such as e-bikes and e-scooters, following several high-profile incidents. 4,5Thus, understanding and improving the safe use of these battery packs and battery-powered devices is critical for the safety of the users, protection of property, and the continued use and growth of the micro-mobility industry.
Lithium-ion (Li-ion) battery packs, such as those predominantly used in electric micro-mobility devices, can suffer thermal failures from a variety of root causes, including poor design or manufacturing, external abuse, and misuse by consumers.While many methods exist to properly operate these battery-powered products, there is often poor information, instructions, and/or training to ensure that the batteries are being handled appropriately.Particularly, there is a lack of general awareness surrounding the best practices and procedures for responding to battery thermal events, even among first responders.This lack of understanding and appropriate guidance has led to ineffective and hazardous actions being taken during the early stages of thermal runaway events on micro-mobility battery packs, leading to more severe and hazardous outcomes.
7][8] The possibility of creating stranded energy with the application of effective fire extinguishing techniques complicates matters even further. 9If a Li-ion battery fire is effectively extinguished, there is a high likelihood that the battery pack will contain stranded electrical energy without an effective means to remove it.
To further enhance the understanding of micro-mobility battery thermal runaway events and to evaluate some of the potential responses, forced thermal runaway abuse tests were performed on multiple e-bike battery packs to study their behaviour and potential mitigation strategies.Several applications of fire blankets were tested, including covering a battery pack in a fire blanket both prior to and after initiating thermal runaway.Additionally, a battery pack experiencing thermal runaway was sprayed with water to extinguish (or slow down) the thermal event.This work also demonstrates how design aspects of the battery pack, such as adding thermal insulation between cells and increasing the separation distance between cells, can help to prevent cell-to-cell propagation and reduce the severity of a thermal event when a failure occurs.

Experimental
Exponent performed forced thermal abuse testing on several lithium-ion battery packs which were designed for use in e-mobility devices such as e-bikes and e-scooters.Thermal runaway was induced by external heating of a single cell to failure.Thermal testing was used to characterize the duration, severity, and extent of the battery thermal events.Specifically, the factors that were investigated include the rate of cell-to-cell propagation, the effective secondary ignition radius, and the spread of debris, as well as the effectiveness of fire suppression methods (two types that would likely be readily available in a residential property) for reducing the severity of a thermal event or aiding in evacuation.Data was collected on the cell voltages and temperatures, as well as the distance sparks, embers, flames, or other competent ignition sources travelled.
Materials and methods.-Thebattery packs used in this investigation were commercially available, had a nominal voltage of 48 V, and a capacity of 20 Ah.Each pack contained 104 18650 format lithium nickel manganese cobalt oxide (NMC) cells in a 13S4P configuration.The packs were disassembled to assess the build quality, analyse the cell voltages, and to allow for the installation of K-type thermocouples (temperature range of −200 to 1270 °C), voltage sense wires, and external heating source(s) (Fig. 1) within the pack.
z E-mail: dtorelli@exponent.comECS Advances, 2024 3 010501 Fire suppression testing.-AFirechief SVB2/K100-P fire blanket (1.2 m × 1.2 m, double-sided, K100 silicone-coated, glass fibre) was used for the purpose of investigating the effect of fire suppression during charging/before initiation of thermal runaway.The fire blanket was secured over the pack being tested with a weight at each of the four corners prior to initiation of thermal runaway.This configuration is akin to wrapping an e-mobility device and/or battery pack in a fire blanket during charging, transportation, or when it is not in use and the pack then experiences a thermal event.
A Vigil VP-LBFB32 lithium battery fire blanket (3 m × 2 m, double-sided, 130 g silicon-coated, 430 gsm fibreglass fabric) was used for the purpose of investigating the effectiveness of a blanket to contain the thermal energy when it is deployed after thermal runaway initiation.The fire blanket was concertinaed and placed two meters away from the incident pack.Two steel cables were attached to the fire blanket handles to allow for safe deployment during the thermal runaway event.This configuration is akin to an incident where a battery pack goes into thermal runaway and it is necessary for a user or bystander to attempt to contain or extinguish the fire.
The results of fire suppression were observed visually (via video recording and photo documentation) and by monitoring the voltages and thermocouples attached to the pack.
Cell-to-cell propagation testing.-SamsungICR18650-26 F 2.6 Ah lithium cobalt oxide cells were added to the pack for the purpose of testing the likelihood of cell-to-cell propagation.Two cells were added to a single pack, each encased with different thermal insulations, and a third cell original to the pack was used to test the effects of no insulation.One cell was thermally insulated with a two-part polyurethane potting compound (Epic Resins S7253) with a thickness of 4 mm around the diameter of the cell (Cell 1).The second cell was insulated with Nomex fabric with a total thickness of 0.75 mm (fabric was 0.25 mm thick and was wrapped three times around the cell) (Cell 2).Prior to the addition of insulation, for the purpose of forcing thermal runaway, an electrical heater (17 cm of nickel-chromium wire (Ni:Cr 80:20, 0.23 mm diameter) wrapped around the circumference of the cell three times) was attached to the can wall of each cell and fixed in place using Kapton-insulated wire.Both Cell 1 and Cell 2 were electrically insulated with a layer of Kapton tape to prevent shorting of the heating coil to the cell can.As a control, the third cell, with no thermal insulation and in its "asmanufactured" state (Cell 3), was instrumented with an external cartridge heater (175 W, 38.1 mm long, 12.7 mm diameter).All three cells were also fitted with a thermocouple in the region of the heating element (T1, T2, and T3) to monitor the temperature of the heating element.Finally, neighbouring cells original to the pack were fitted with thermocouples to monitor the extent of heat transfer during the initial heating process and thermal runaway (T4, T8, and T9).Additional thermocouples were added along the length of the pack (T5, T6, and T7) (Fig. 1).The rate of thermal propagation was measured by the progression of temperature increase throughout the pack and observing the venting of cells during thermal runaway.
The heating elements were supplied with external power (up to 240 V) using a single phase 1920VA Variac variable transformer to adjust the mains voltage for power (and temperature) control.
Prior to each test, the pack under investigation was centred on gypsum boards (1.8 m × 0.9 m pieces); to monitor the spread of flames and ejecta, two circles 2 m and 4 m in radiuswith the pack in the centre, were marked using yellow acrylic paint.
Water as an extinguisher testing.-Ahousehold water hose was used to evaluate the effectiveness of using a residential water supply to suppress a Li-ion battery fire to aid escape.The hose was fixed to a scaffold aimed at the incident Li-ion pack, approximately 8 m away.A target flow rate similar to household water hoses was used (10-15 litres per minute).

Results
To evaluate the severity and likelihood of thermal runaway propagation, six forced total thermal runaway abuse tests were performed on four exemplar e-bike battery packs (three tests were conducted on one pack and one test each on the remaining three packs); thermal runaway events were initiated by an external heater and video, voltage, and temperature data were recorded.Furthermore, the efficacy of the use of fire blankets and water suppression by potential consumers on the propagating battery packs was recreated and assessed.Table I provides an overview of the tests performed, which are discussed in more detail in the subsequent sections.
Baseline thermal runaway test.-Acontrol experiment was performed by initiating thermal runaway in a single cell at one end of the battery pack and allowing it to fully propagate throughout the pack without any attempts to extinguish or contain the heat or ejecta (Test 1).The first plume of smoke was observed outside of the pack almost two seconds after the first cell vented (Fig. 2, 00:01:57) and flames were observed exiting the pack after almost seven seconds (Fig. 2, 00:06:96).Once thermal runaway had begun to propagate between cells, flames were observed extending approximately 1-2 m away from the pack (Fig. 2) and test combustibles (bundles of dry straw) were ignited as far as 2 m away (red arrows, Fig. 2, 4:15:10).
The final cell(s) appeared to go into thermal runaway just over seven minutes after initiation of the first cell (Fig. 2, 07:14:68) however, the combustible materials within the pack continued to burn for approximately eight additional minutes.The total duration for which flames were produced was approximately fifteen minutes, when the thermal event self-extinguished.

Fire Blanket Test-Pre-Deployed
Two thermal abuse tests were conducted to evaluate fire blankets as a potential means of containing the thermal event.The first test utilized a pre-deployed fire blanket fully covering the pack in the moments up to and after thermal runaway had been initiated (Test 2).
The first indication of audible cell venting was recorded and considered to be the start of the thermal runaway event (Fig. 3, 00:00:00).White smoke was observed exiting the fire blanket from 00:01:29 until 00:11:35, at which point the smoke transitioned to black, indicating that particulate matter from the cells had begun to  become entrained in the vent gases.As the cells in thermal runaway were venting and the thermal event began to propagate to additional cells, the heat released from the pack created a hole in the fire blanket directly above the pack (Fig. 3, 00:13:95, orange arrow).Soon after, a spark was visible through a second hole in the blanket (Fig. 3, 00:15:09, red arrow); the first flames were observed exiting from underneath the blanket almost immediately after that (Fig. 3, 00:15:11).After nearly twenty-two seconds from the initiation of the thermal event, flames were produced from the pack which rapidly reached close to 4 m in length ejected from both sides of the pack (Fig. 3, 00:21:90).Within fifteen seconds of the initiation of thermal runaway of Test 2, the fire blanket failed to contain the thermal damage.Despite having anchored the blanket to the test surface before the test, the flames, sparks, and debris were not contained by the blanket and ejecta was found as far as 15 m from the pack.The fast and furious nature of the event meant all cells experienced thermal runaway within the first fifty seconds, after which only remaining combustible materials were left burning.
In addition to the lack of containment provided by the fire blanket, a more violent, short-lived thermal runaway event was produced when the pack was covered compared to the control test (Test 1).Flames were observed to have travelled a greater distance in Test 2 compared to an uncovered pack (Test 1), and the temperatures of the exterior of the covered pack were measured above 600 °C for over four minutes (see Fig. 4), indicating that the fire blanket did not extinguish or significantly reduce the severity of the thermal event.These effects are likely, at least in part, due to the presence of the fire blanket which prevented the convection of hot gasses away from the pack rather than allowing them to dissipate more effectively as in the unconstrained pack test (Test 1) and the test with a water extinguishant (Test 6).

Fire Blanket Test-Deployed After Thermal Runaway Initiation
A second test was performed with a fire blanket deployed approximately one minute after the initiation of thermal runaway (Test 3) (Fig. 5, 00:55:78).Within one second of covering the pack with the fire blanket, flames and hot material exiting the pack produced a hole in the blanket (Fig. 5, 00:56:79, red arrow).Approximately ten seconds after the initial breach of the fire blanket, flames were observed exiting through the portion of the blanket in direct contact with the pack, as well as from the edge of the blanket, extending over 4 m away from the centre of the pack (Fig. 5, 01:05:02, yellow arrows).Flames continued to exit through or underneath the blanket for over three minutes, as the pack fully propagated and ultimately self-extinguished.Temperatures inside the pack were measured over 550 °C for approximately ten minutes after the start of the thermal event and remained until the test was terminated (Fig. 6).
Although differences between the two fire blanket tests (predeployed vs. deployed after thermal runaway) were observed both in terms of speed of the reaction and length of time at temperatures above 550 °C, both tests had more violent, short-lived thermal runaway events than the Test 1 (Baseline Thermal Runaway Test).This was evident by the distance flames travelled from the pack, as well as the total length of time of the thermal runaway event.It should however be noted that there is a fair amount of stochasticity between thermal runaway events in batteries failing under nominally the same conditions, so repeating these tests would help confirm the extent to which the fire blanket actually increased the speed of propagation.

Cell-to-Cell Propagation Testing
The propensity for cell-to-cell thermal runaway propagation within a pack was evaluated through separately forcing three initiator cells into thermal runaway and monitoring the resulting effects on the adjacent cells.The first initiator cell (Cell 1, Test 4) was thermally insulated with a urethane potting compound, the second was wrapped with a Nomex fabric (Cell 2, Test 5), and the third was unaltered and only had air as a separation barrier between the adjacent cells, as designed (Cell 3, Test 6).All initiator cells were instrumented in similar positions within the same pack and were forced into thermal runaway; the two insulated cells used identical external heaters (wrapped around the cells) and the uninsulated cell used an external cartridge heater.
During the thermal event, the exterior can wall of the potted initiator cell (Cell 1) reached a maximum temperature of 367 °C (T1), while the highest surface temperature on a neighbouring cell was measured at only 64 °C (T8) (Table II and Fig. 7).No cell-tocell propagation, ejection of flames or sparks, or severe thermal damage to the exterior of the battery pack was observed.After the thermal event, Cell 1's surface temperature dropped below 100 °C after approximately eleven minutes.
Once the battery pack returned to ambient temperatures, the Nomex-wrapped cell (Cell 2), which was instrumented on the opposite side of the pack, was then forced into thermal runaway.The exterior can wall of the wrapped cell reached a maximum temperature of 558 °C (T2), while the highest surface temperature of a neighbouring cell reached 132 °C (T8) (Table III and Fig. 8).While under controlled adiabatic heating conditions, such as in an accelerating rate calorimeter, the thermal runaway onset temperature for NMC-based (nickel, manganese, cobalt) lithium-ion batteries has been measured at 140 °C-145 °C; the onset temperature is typically much higher for rapid-heating events such as this test. 10No cell-tocell propagation, ejection of flames or sparks, or severe thermal damage to the exterior of the battery pack was observed.After the thermal event, Cell 2's temperature dropped to near 100 °C after approximately seventeen minutes.
The third initiator cell, which had no thermal insulation added (Cell 3), was then forced into thermal runaway; the exterior can wall of the initiating cell rapidly increased to 395 °C (T3) and about 30 s later, a nearby cell (T9) went into thermal runaway (Fig. 9).Several distinct temperature spikes were observed after the initiating event, including at approximately 30, 60, and 85 s, indicative of thermal runaway propagation to other cells.The rate of thermal runaway propagation throughout the pack is clearly shown in Fig. 10, which    During the propagating event, temperatures were measured above 1270 °C (T4), although this could have been a result of exposure of a thermocouple to a direct flame.Cell-to-cell propagation continued until all cells had gone into thermal runaway.Flames were observed exiting the pack approximately one minute and 56 s after initiating thermal runaway in Cell 3. At their peak intensity, the flames were observed to extend nearly 2 m away from the pack, which occurred after approximately two minutes (see Fig. 12).Combusting materials, including cell remains, were observed to have been ejected  ECS Advances, 2024 3 010501 from the battery pack at distances up to 10 m, indicating that the area exposed to a potential fire hazard was much than the immediate surroundings.Several thermocouples became damaged during the thermal event and reliable temperature data at these locations was no longer available.

Water as an Extinguishing Agent Test
As thermal runaway propagated throughout the pack, water was sprayed onto the pack using a household water hose approximately three minutes and eight seconds after initiating thermal runaway in Cell 3 (uninsulated, Test 6).The water lowered the overall temperatures in the pack (Fig. 13) and extinguished some of the flames exiting the pack (Fig. 14), but cell-to-cell propagation continued within the pack and flames reappeared approximately ten seconds after the continued application of water and sustained for the rest of the thermal event.
Videos and photographs taken during the thermal runaway event were used to identify the initiation of individual cell thermal runaway events.The time at which discrete visual cues and/or flames exiting the pack were observed after the initial cell entered thermal runaway, are shown in Fig. 15.While each plume of smoke or flame may not correspond to a single cell and may instead correspond to several cells entering thermal runaway at the same time, this data gives an approximate measure of the speed of cell-tocell propagation within the pack.After the first six thermal events were documented, a roughly linear region of cell-to-cell propagation was observed with approximately one cell (or cell group) entering thermal runaway about every five seconds (Fig. 15, red dots).After  ECS Advances, 2024 3 010501 the addition of water, the cell-to-cell propagation appeared to slow from approximately eleven to six cells minute, but maintained a roughly linear trend with approximately one cell (or cell group) entering thermal runaway about every ten seconds (Fig. 15, blue dots).

Discussion
Fire blankets are readily available and are an effective means of fire protection and containment for many household fires.These blankets act to smother fires, cutting off the supply of atmospheric oxygen, and protecting nearby bystanders from heat or flames produced from a combusting object.However, as demonstrated in both tests utilizing a fire blanket here (Tests 2 and 3), the fire blankets were insufficient to prevent flames, ejecta, or sparks from exiting the pack, affecting nearby areas.Both tests had flames extending over 4 m away from the battery pack, which was approximately 2 m further than any of the uncovered packs.In addition, an increase of hot ejecta was produced and often travelled further (∼10-15 m) than the unmitigated pack test (Test 1),  particularly on the pack which had the fire blanket deployed after thermal runaway initiation (Test 3).The diagram in Fig. 16 shows approximate location and quantity of ejecta (primarily cell cans, windings, and flaming debris) which could act as ignition sources if combustible objects were placed in the general vicinity of the pack.While some contributions of wind speed and direction were observed (approximate average wind speed was 16 kph with gusts up to 26 kph), even in the absence of wind it is likely that ignition sources would travel multiple metres away from the pack and could act to spread the fire to the surrounding environment despite the use of the fire blanket.It should be noted that there is a fair amount of stochasticity between thermal runaway events in batteries failing under nominally the same conditions, so repeating these tests would help confirm the extent to which the fire blanket actually increased the speed of propagation.
The ineffectiveness of fire blankets to contain lithium-ion thermal runaway events is likely due to several factors.First, many fire blankets are made of woven fibreglass fabrics with working temperatures in the range of 50 °C-550 °C. 11While these materials and operating temperature range are acceptable for small household fires (e.g., kitchen fires involving cooking oil which burn at around 230 °C-360 °C12 ), they were insufficient to provide acceptable protection from a lithium-ion battery fire of this size and intensity where peak temperatures inside the pack exceeded 1270 °C and exterior pack temperatures exceeded 660 °C.Secondly, the fire blanket acts to restrict heat loss from the pack to the surroundings, thus increasing the rate of heating of neighbouring cells within the pack and increasing the rate of cell-to-cell propagation.This mechanism contrasts with the use of water as an extinguishant, which acts to reduce the external temperature of the pack.As the  external pack temperature is reduced, the rate of cell-to-cell propagation decreases, and the severity of the can be reduced (see Test 6).Thirdly, fire blankets are most useful for combustion events which require atmospheric oxygen to propagate.In the case of lithium-ion battery fires, heat is produced from internal cell reactions and decomposition of materials which are either anaerobic or can sustain solely from the oxygen produced from active material and electrolyte decomposition. 13,14While atmospheric oxygen often helps accelerate thermal runaway propagation within a pack and is needed for the sustained combustion of many pack components (e.g., plastics, adhesives, labels), the lithium-ion batteries can still undergo thermal runaway and propagate in an environment with reduced availability of atmospheric oxygen-.
Although the addition of water from a household water hose to the thermal runaway event did slow down the rate of cell-to-cell propagation and lower the overall pack temperature, it was insufficient to stop thermal runaway and the event proceeded until all cells were consumed.As such, it is clear that lithium-ion battery fire extinguishing and containment is a difficult task, and that there are typically significant ignition and safety hazards from a battery pack during a thermal event.Thus, improved battery pack designs are required to prevent the initial thermal runaway events and to slow (or stop) the spread of thermal runaway from one cell to another during a failure.An illustration of a viable method was shown in Tests 4 and 5 which utilized a urethane potting compound and Nomex thermal insulation on initiator cells, respectively; these tests were able to demonstrate the effectiveness of added insulation between cells on the prevention of cell-to-cell propagation.Both materials served to prevent cell-to-cell propagation at the earliest stages, which resulted in relatively benign thermal events with no external flames or high-temperature debris produced from the pack.Thermal events like those in Tests 4 and 5 present relatively low risk compared to other tests, specifically Tests 1, 2, 3, and 6, where cellto-cell propagation and complete failure of the pack was observed.
Numerous other failure mitigation strategies can be implemented in the design at the cell-, pack-, and device-level; further, adjustments to the surrounding environment can also be made.External fuses, contactors, etc. can help ensure that, in the event of an external short circuit event or an intrinsic failure mode, the positive and negative terminals of the battery pack or cell are electrically disconnected, thus preventing further heating by the electrical short circuit.In addition, the use of certain chemistries or cell form factors may add further layers of safety depending on the pack design.At the pack-level, manufacturers may design packs with added thermal insulation, such as those utilized here, or other insulators like endothermic barriers, aerogels, and ceramic fibres.Cell-to-cell distances can be increased, and battery management systems  (BMSs) can be designed in such a way to prevent thermal runaway events by balancing cells, providing under or overvoltage protection, overcurrent temperature control, and cell health monitoring.In more complex designs, a BMS may even detect a thermal runaway event during its early stages through cell voltage, current, or impedance measurements, or by more sophisticated techniques such as gas detection.Furthermore, at the device-level, manufacturers may design for thermal runaway mitigation by increasing the pack's resistance to moisture and debris ingress, adding advanced thermal and electrical systems, and by designing for failure in a manner that limits the consequences.Of course it is also important that high quality battery cells are used in packs, where failure rates for catastrophic events are often quoted to be on the order of 1 in 10 million to 1 in 40 million cells. 15ven with sophisticated cell, pack, and device design mitigation measures in place, environmental and operative controls should be taken as well to prevent thermal runaway or reduce the severity if a thermal event does occur.These controls include practices such as providing adequate spacing between charging or stored battery packs and nearby combustibles, having appropriate fire suppression methods on hand (i.e., appropriate fire extinguishers or sprinklers), and monitoring the condition of all battery packs and devices for signs of damage to the pack or device housing.
The most robust approaches to preventing and mitigating battery thermal events in electric micro-mobility battery packs, involves incorporating redundant layers of protection from the cell to systemlevel.Pack manufacturers and designers can include both passive and active safety systems to prevent the initial failures and to mitigate the thermal event from propagating at the earliest stages.With proper handling and operation, many common battery failure modes can be avoided, or their likelihood significantly reduced, however, it is often unclear what the appropriate guidance from pack manufacturers and regulators is.The optimal strategies for failure mitigation and suitable response procedures to failed micro-mobility batteries requires additional research, however, the work demonstrated here provides valuable experimental results of current approaches and their strengths and limitations.

Conclusions
Forced thermal runaway of an uninsulated single cell in a widely available micro-mobility battery pack consistently resulted in cellto-cell propagation and complete thermal consumption of the packs.The severe thermal events, which released flames and ejected burning materials meters away from the pack, can present a significant safety hazard to nearby beings, property, and the surrounding environment.The pack-level testing performed in this work demonstrated that once a single cell thermal failure within the pack occurs, it can propagate rapidly throughout the pack and greatly increase the hazards to the surroundings.It was observed that once a propagating thermal event started, attempting to extinguish the flaming battery pack with fire blankets or moderate water dousing was ineffective at stopping the thermal event.Depending on the circumstances, attempting to suppress the battery thermal event with a fire blanket may cause an increase in severity of the thermal event, as the heat produced by the event is not efficiently dissipated, which can result in an acceleration of the cell-to-cell propagation rate and increased pack surface temperatures.Additional research of current suppression strategies or the development of new strategies is needed to safely extinguish thermal runaway events of micromobility sized battery packs.
The difficulty of containing or extinguishing a lithium-ion battery fire is well known and highlighted here.As such, the importance of taking appropriate design measures and using products according to their intended designed purpose is essential to reduce the risk of hazardous thermal runaway events.Design steps such as adding sufficient thermal insulation between cells or directing hot vent gases and ejecta away from other cells, can help to reduce the risk of cellto-cell propagation.Other mitigation strategies have been identified and this remains an active field of research and development.While these designs can help make packs safer, they often come with tradeoffs such as pack size, weight, or energy density.Manufacturers, regulators, and consumers must evaluate these compromises in performance with the potential hazards these battery packs present to establish a tolerable level of risk.Furthermore, once appropriate standards and regulations are set, it is imperative that manufacturers and pack designers properly test their battery pack designs in order to fully assess the potential hazards and optimize for pack safety.

Figure 1 .
Figure 1.Photograph showing the cell-to-cell propagation test pack wiring with the added cells, two with thermal insulation (Cell 1 and 2) and one "as manufactured" (Cell 3) on the left side of the pack, as well as the locations of the thermocouples placed for temperature monitoring (T1-9).

Figure 2 .
Figure 2. Test 1 photographs at various times during the thermal runaway event (mm:ss:ms) with no suppression or containment of the thermal event attempted.The arrows above show combustible location markers at 1 m (orange) and 2 m (red) away from the centre of the pack; the timing of combustible ignition is noted at 00:56:28 (1 m) and 4:15:10 (2 m).

Figure 3 .
Figure 3. Test 2 photographs (pre-deployed fire blanket) at various times during the thermal runaway event (mm:ss:ms).The orange arrow shows the location of the first hole, the red arrow shows the location of the second hole in which a spark could be seen (00:15:09).

Figure 4 .
Figure 4. Test 2 temperature data of the pack which had a fire blanket deployed prior to thermal runaway after thermal runaway was initiated in the first cell (0 s). T3 became damaged during the test and data from this thermocouple was not used in this analysis.

Figure 5 .
Figure 5. Test 3 photographs of the battery pack and fire blanket at various times during the thermal runaway event and blanket deployment (mm:ss:ms).

Figure 6 .
Figure 6.Test 3 temperature data of the pack which had a fire blanket deployed during thermal runaway after initiation of the first cell (0 s).

Figure 7 .
Figure 7. Test 4 temperature data during the thermal runaway event of Cell 1 (insulated with urethane potting compound).

Figure 8 .
Figure 8. Test 5 temperature data during the thermal runaway event of Cell 2 (insulated with Nomex fabric).

Figure 9 .
Figure 9. Test 6 temperature data during the thermal runaway event of Cell 3 (uninsulated).

Figure 10 .
Figure 10.Test 6 pack voltage (black line) and individual cell group (CG) voltages for the entire battery pack after the onset of thermal runaway of Cell 3.

Figure 12 .
Figure 12.Test 6 photograph of the cell-to-cell propagation test pack approximately 2 minutes and 14 s after initiating thermal runaway in Cell 3 (uninsulated).Inner yellow circle is 2 m away from pack and exterior yellow circle is 4 m away from pack.

Figure 13 .
Figure 13.Test 6 temperature data during thermal runaway after Cell 3 (uninsulated) was initiated.Water was sprayed onto the pack using a household water hose starting approximately three minutes after the initiating cell went into thermal runaway.

Figure 14 .
Figure 14.Test 6 photographs right before water was applied (left) to the pack and immediately after (right).Flames were initially extinguished but later reignited despite continued application of water.

Figure 15 .
Figure 15.Temporal plot of cell thermal runaway during Test 6 based on visible puffs of smoke or flame generation exiting the pack of initial cells (green dots), cells where no extinguishing was attempted and thermal runaway was allowed to propagate uninhibited (red dots), and cells which entered thermal runaway after water was added to the pack (blue dots).

Figure 16 .
Figure 16.Diagram of cell ejecta mapped out showing distance travelled from the end of the pack during Test 3. The furthest ejecta was measured approximately 16 m from the end of the pack.An average wind direction is indicated on the diagram.Average wind speeds of 16 kph, and gusts up to 26 kph, were measured at a nearby weather station in Waddington, England.The wind speed and direction likely contributed to the distance and direction the ejecta travelled.

Table I .
Test result summary table.

Table II .
Maximum temperature during thermal runaway of Cell 1 (insulated with urethane potting compound) measured at various locations throughout the pack (Test 4).