Case Studies in Thermal Engineering 64 (2024) 105474

Contents lists available at ScienceDirect

Case Studies in Thermal Engineering

journal homepage: www.elsevier.com/locate/csite

Assessing fire dynamics and suppression techniques in electric

vehicles at different states of charge: Implications for
maritime safety
Suhaeng Lee , Daehyun Choi , Yeoseon Jeong , Minho Moon , Hyukjoo Kwon ,
Kukil Han , Hyungjun Kim , Hongsoon Im , Youngseob Park , Dongki Shin ,
Geonhui Gwak *
Advanced Fire&Disaster Management Centre, Fire&Disaster Management Division, 33-72, Eonjang 1-gil, Samcheok-si, Gangwon-do, 25913,
Republic of Korea

A R T I C L E I N F O A B S T R A C T

Keywords: The maritime transportation of electric vehicles (EVs) poses significant fire risks due to the po­
Thermal runaway tential for thermal runaway in lithium-ion batteries, particularly when the state of charge (SOC)
State of charge varies. This study uniquely examines the effects of SOC on fire behavior and suppression efficacy,
Heat release rate
going beyond previous research by focusing on the maritime environment. Experiments were
Radiative heat flux
Fire suppression
conducted on EV battery packs at SOC levels of 70 %, 50 %, and 30 %, and on a full-scale EV at
50 % SOC, to evaluate fire dynamics and the effectiveness of suppression methods, including
seawater injection and fire blankets. Results showed that higher SOC levels are associated with
significantly increased heat release rates and extended fire durations, while lower SOC levels (30
%) reduce fire intensity yet necessitate continuous monitoring for re-ignition risks. Moreover, the
combination of seawater injection and fire blankets showed promise in cases where rapid cooling
and containment of fire spread were priorities, illustrating a potential strategy for managing EV
battery fires during maritime transport. These findings underscore the need for strategic SOC
management, recommending lower SOC thresholds to minimize fire severity, and the use of
combined suppression techniques to enhance EV fire safety during maritime transport.

1. Introduction

The marine transportation of electric vehicles (EVs) is steadily increasing as they replace conventional internal combustion engine
vehicles (ICEVs), due to environmental concerns. EVs are powered by lithium-ion battery (LIB) packs comprising interconnected
modules, which contain numerous individual cells configured in series and parallel within the vehicle. However, LIBs are prone to
thermal runaway (TR) under adverse conditions such as overheating, mechanical impact, and overcharging, due to their high energy
density and the flammability of conventional organic liquid electrolytes. Of particular concern are EV fires occurring onboard pure car
and truck carrier (PCTC) vessels, which can lead to significant property damage and human casualties. The first recorded EV fire
incident aboard a vehicle carrier vessel occurred in 2010 on the MS Pearl. Subsequent incidents have underscored the challenges posed
by such fires. In 2022, the Felicity Ace, carrying over 4000 vehicles including 300 EVs, experienced a fire at sea, resulting in substantial

* Corresponding author.
E-mail address: [email protected] (G. Gwak).

https://doi.org/10.1016/j.csite.2024.105474
Received 24 July 2024; Received in revised form 6 November 2024; Accepted 9 November 2024
Available online 12 November 2024
2214-157X/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

property damage. Similarly, in 2023, the Fremantle Highway reported a fire incident leading to significant financial losses. Especially,
the densely packed cargo holds of ships pose particularly grave risks for fire propagation. Currently, according to guidelines from the
European Maritime Safety Agency, while the state of charge (SOC) for LIBs does not affect the total energy released during a fire, it
directly influences their growth rate and maximum heat release, implying that higher charging rates result in faster and more intense
combustion compared to lower SOC. Therefore, shipping companies, or marine carriers, generally prefer to maintain a lower SOC to
reduce risks associated with TR and potential fires onboard. Lower SOC levels, such as those recommended to be capped at 20 % during
port stays, are aimed at minimizing the risk and intensity of fires during transport. The SOC is recommended to be limited to 20 %
during port stays to ensure minimal risk during transportation to the first destination, with an allowance for up to 50 % for vehicles
with smaller battery capacities [1]. Initial tests by global LIB experts suggest that lower SOC levels correlate with a reduced likelihood
of TR [2]. On the other hand, automotive manufacturers prefer a higher SOC, particularly for certain models and destination re­
quirements. For vehicles traveling longer distances after disembarking, or for those requiring immediate use upon arrival, an SOC
above 50 % is often necessary. This ensures that the vehicles are ready for operation without the need for immediate recharging, which
can be impractical or unavailable at the point of disembarkation. This conflict of interests highlights the need for a systematic analysis
of how EV SOC impacts fire safety during maritime transportation. To address this issue, researchers have conducted extensive in­
vestigations into the thermal behaviors and hazards of full-scale EVs under TR conditions, as well as studies related to battery
extinguishing methods [3–15]. Li et al. [3] experimentally analyzed a full scale TR testing of battery pack to understand the fire
evolution process in the battery electric vehicle (BEV). They categorized the combustion behavior of EVs into three stages: the
initiation of TR, TR propagation, and fire suppression and observed that it took 22 s from the onset of a triggered TR event to the point
where flames could potentially spread throughout the entire cargo compartment. Kang et al. [4] compared the thermal behavior
among a BEV, an ICEV, and a hydrogen fuel-cell electric vehicle (FCEV) by measuring peak heat release rate (pHRR), total heat release
(THR), radiative heat flux (RHF), and mass loss. The study revealed that pHRR and THR were primarily influenced by general
combustible materials within the vehicle, rather than the battery pack itself, despite the presence of a jet flame discharged from the
battery pack. The pHRR and THR values of BEV fires exhibit marginal differences compared to ICEV fires, while FCEV fires show
relatively lower values. Zhao et al. [5] conducted a comprehensive experiment involving a full-scale EV with lithium nickel manganese
cobalt oxides ternary material battery pack with a total capacity of 38.1 kWh. The study revealed a distinctive pattern of fire spreading
along the chassis and body following the ignition of the battery pack. Notably, the flame propagated from the right rear wheel to the
trunk. Cui et al. [6] conducted a comprehensive fire test on two parallel positioned EVs to investigate thermal characteristics and
hazards. The two EVs were spaced 0.6m apart, representing the distance typically found in parking lots or tunnels. They found that the
two EVs fire spread faster than plug-in hybrid electric vehicles (PHEV) or ICEV fires due to spewing flames, proximity, and an
oxygen-rich environment. Dorsz and Lewandowski [7] examined the unique fire hazards of EVs in confined spaces, like road tunnels
and garages, where restricted ventilation increases the likelihood of TR, resulting in high peak HRR and dangerous emissions such as
hydrogen fluoride. Unlike ICEVs, which primarily emit carbon monoxide (CO) and carbon dioxide (CO₂) when burning, EVs release
hazardous byproducts, including hydrogen fluoride and other toxic gases, due to the chemical breakdown of lithium-ion batteries. This
study highlights that EV fires not only reach higher peak HRR but also produce more harmful emissions, underscoring the amplified
risks when EVs are transported on maritime vessels. Cui et al. [8] investigated the fire risks of PHEVs in enclosed spaces, revealing that
limited ventilation can lead to explosions from combustible gases released during TR. Unlike ICEVs, which emit primarily CO and CO₂,
PHEVs produce more toxic byproducts, such as HF, from battery degradation. Their study found that PHEV fires can reach temper­
atures over 843 ◦ C, with flames up to 3 m high, and generate molten material that accelerates fire spread. In addition, Cui et al. [9]
investigated effective approaches for extinguishing BEV fires using water and foam. Their findings demonstrated that water injection
lowered the temperature of most batteries to below 50 ◦ C, preventing re-TR afterward. Foam injection also successfully suppressed EV
fires, although it resulted in battery pack temperatures exceeding 50 ◦ C due to the limited cooling capability of foam. Peter et al. [10]
conducted detailed fire tests on EVs in a road tunnel, demonstrating high heat release rates (HRR) and hazardous emissions like
hydrogen fluoride. Fire blankets proved ineffective once the battery had generated in fire, whereas fire lances quickly extinguished
battery fires with minimal water usage. Ubaldi et al. [11] compared water mist, F-500, and CO₂ for LIB fires, finding that water mist
and F-500 cooled the fastest, at 30.5 ◦ C/s and 36.5 ◦ C/s, respectively. CO₂ was less effective, with a 20.0 ◦ C/s cooling rate. While
effective, F-500 had higher VOC emissions, raising environmental concerns. These results emphasize the balance between suppression
effectiveness and environmental impact when choosing fire suppression agents for confined spaces. Xu et al. [12] evaluated the
effectiveness of CO₂, HFC-227ea, and water mist on LIB fires. They found that water mist achieved the highest temperature reduction,
cooling battery cells by up to 133 ◦ C, while CO₂ and HFC-227ea were less effective at managing TR. This suggests that water mist could
be a superior option for suppressing EV fires, particularly in confined spaces where rapid cooling is essential. Svetlík et al. [13]
conducted comprehensive full-scale fire tests on EVs, revealing rapid temperature increases exceeding 1000 ◦ C shortly after ignition
due to the use of a propane burner as an external heat source. Foam extinguishing proved highly effective, reducing temperatures from
486.1 ◦ C to 76 ◦ C within seconds. Funk et al. [14] conducted fire extinguishment tests on BEVs in an open-sided enclosure. They
analyzed various BEVs fire extinguishing methods including a fire blanket, an extinguishing lance, a piercing device, water curtains,
and a water mist system. The study demonstrated that the combination of direct and indirect extinguishing methods successfully
suppressed BEV fires. Additionally, they suggested that updating the current procedures to include automatic initiation of extin­
guishing systems, such as sprinklers or water mist, may be the best option to achieve fast initiation and control of BEV fires. Zhao et al.
[15] conducted a comprehensive experimental study on EV fire characteristics and suppression methods. The study evaluated various
suppression techniques, including fire blankets, water mist, and compressed air foam. The findings indicated that fire blankets were
effective in extinguishing flames but required continuous cooling to prevent re-ignition. Water mist and compressed air foam
significantly reduced temperatures and controlled fire spread. However, a notable limitation of this study is that all three suppression

2
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

methods were applied in a single test on one vehicle. This approach prevents a clear assessment of the individual effectiveness of each
method and their potential interactions, thereby limiting the evaluation of their separate and combined performance in real EV fire
scenarios.
This study addresses a critical gap in fire safety literature by investigating the influence of varying SOC levels (70 %, 50 %, and 30
%) on fire dynamics and suppression effectiveness specifically within maritime environments. Previous research has primarily focused
on fire hazards in EVs but has not comprehensively examined how SOC impacts fire behavior and suppression techniques on PCTC
vessels, where the risk of fire in transit poses unique challenges. By conducting extensive fire behavior analyses on both battery packs
and full-scale EVs, this study aims to provide foundational data that informs SOC management strategies and enhances fire suppression
protocols. These findings contribute essential insights for improving the safety of EVs during maritime transport, where effective risk
management is crucial for fire prevention and control.

2. Methodology

2.1. Experimental set-up

Due to the safety risks and practical limitations associated with large-scale fire testing, the experimental design was crafted to
ensure both safety and data consistency. Our setup focused on replicating essential fire conditions in a controlled environment,
allowing us to assess EV fire dynamics and suppression methods effectively while managing the constraints of a large-scale experiment.
The experiments were conducted to analyze the characteristics of EV fires and the effectiveness of extinguishing methods within PCTC
vessels. To understand the behavior of EV fires, we measured the HRR, THR, and RHF crucial factors for assessing the intensity of fire
events. That was generally measured using a cone calorimeter based on the principle of oxygen depletion during EV fires [16]. In
addition, to evaluate the effects of fire on nearby EVs, RHF was measured in the vicinity of the EV fires and surface temperatures were
monitored using K-type thermocouples. Meanwhile, the fire extinguishing methods included using a fire blanket, an extinguishing
lance, and seawater injection. These methods are commonly employed in PCTC vessels to respond to EV fires. Fig. 1(a) shows the
specimen placed beneath the large cone calorimeter for testing, with the RHF meter, temperature measuring system, and power supply

Fig. 1. Experimental setup: (a) schematic of cone calorimeter with specimen and testing apparatus, (b) actual test settings for analyzing fire and
extinguishing characteristics of cases: (i) case 1 to 3, (ii) case 4 to 6, (iii) case 7.

3
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

connected to increase the heating film (HF) temperature, while Fig. 1(b) illustrates the experimental setup used to analyze the fire and
extinguishing characteristics of EVs, including a photograph of the actual test setup. To understand the thermal propagation within the
battery pack, 15 thermocouples were affixed. Specifically, to induce TR in the module located in the central region inside the battery
pack, which is composed of pouch cells, film heaters were attached within the module. Similarly, 32 thermocouples were attached to
an EV equipped with pouch cells. The detailed positions of thermocouples are described as shown in Fig. 2. The RHF meter was
positioned 2 m away from the battery pack specimen. For the EV, RHF measurements were taken at intervals of 30 cm at the front and
rear, and 15 cm on either side of the vehicle.

2.2. Experimental cases

The impact of SOC levels on fire and extinguishing behavior was analyzed by integrating results from both the battery packs and the
actual EV. The experimental cases are defined for each of the seven cases.

- Cases 1 to 3: Experiments with battery packs at SOC levels of 70 %, 50 %, and 30 % to study the influence of SOC on fire initiation
and progression.
- Cases 4 to 6: Experiments applying different extinguishing methods such as the seawater injection, the fire blanket, and a com­
bination of the extinguishing lance with seawater and fire blanket on battery packs at 50 % SOC to analyze extinguishing
effectiveness.
- Case 7: Examination of fire behavior in an actual EV at 50 % SOC to understand how vehicle design and battery integration in­
fluence fire dynamics.

The experimental cases for EV fires are delineated in Table 1. The battery pack utilized in cases 1 to 6 is composed of 697V and
111.2Ah, with a capacity of 72.6 kWh. Case 7 features the actual EV equipped with the same battery pack as in case 1 to 6.

2.3. Test procedure

We focused on investigating the fire dynamics and extinguishing effectiveness of EV fires within PCTC vessels, implementing tests
that replicate internal overheating resulting from a short circuit as a representation of battery defects, and confirming the influence of
SOC on these factors. The temperature of the film heater increased at a rate of 4–7 ◦ C per minute. Upon detection of TR, the power
supply to the film heater was promptly deactivated, and all events during the fire and extinguishing tests were meticulously recorded.
In cases 4 to 6, the extinguishing methods were uniformly applied to all three cases 10 min after the onset of TR, taking into account

Fig. 2. Thermocouple placement for thermal propagation analysis: (a) battery pack, (b) actual EV.

4
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

Table 1
Experimental cases for battery pack and EV fire behavior and extinguishing methods.
Cases SOC Focus Specimen

Case 1 70 Fire behavior Battery pack

Case 2 50 Fire behavior Battery pack
Case 3 30 Fire behavior Battery pack
Case 4 50 Extinguishing (extinguishing lance with seawater + fire blanket) Battery pack
Case 5 50 Extinguishing (seawater injection) Battery pack
Case 6 50 Extinguishing (fire blanket) Battery pack
Case 7 50 Fire behavior Real-scale EV

detection and response times within the PCTC vessels. The experiments involved positioning the battery pack within an EV mock-up
frame to more accurately simulate real-world conditions. Case 7 was subjected to fire testing under the same conditions as the other
cases to ensure a consistent evaluation across all scenarios. Additionally, to mitigate the risk of unnecessary explosions during the tests,
air was deliberately released from the tires, while the airbags remained intact.

3. Results and discussion

3.1. Analyzing the critical events on EV fires

Fig. 3(a) depicts the TR behavior of the battery pack at 70 % SOC. Off-gas was first observed at 20 min after the test began. This was
followed by more off-gas at 22 min, with a subsequent reduction at 25 min. A significant release of off-gas occurred at 27 min, followed
by a series of fire, explosion, and arc events starting around 41 min. The fire exhibited multiple phases of diminishing, recurrence, and
resurgence before finally ending at approximately 240 min. This indicates a high level of energy release and sustained combustion due

Fig. 3a. TR behavior of the battery pack at different SOC levels: 70 % SOC.

5
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

to the high SOC. Fig. 3(b) illustrates the TR behavior of the battery pack at 50 % SOC. The initial off-gas was observed at 17 min, with
additional off-gas observed at 19 min and a reduction at 26 min. A massive off-gas release at 27 min was followed by another at 44 min.
Fire and arc events were observed starting at 44 min, with the fire diminishing and finally ending around 135 min. This case shows a
shorter duration and less intense fire compared to case 1, likely due to the lower SOC. Fig. 3(c) shows the TR behavior of the battery
pack at 30 % SOC. Off-gas was observed 14 min after additional heating, followed by further off-gas releases and reductions. The
second and third off-gas releases were noted before the fire ended 120 min after additional heating. No fire behavior was observed
during the test, leading to its termination. However, interestingly, fire, explosion, and arc events occurred 1 h after the test ended,
indicating residual thermal instability.
Fig. 4(a) illustrates the TR behavior of the battery pack at 50 % SOC, where extinguishing methods included the use of the
combination of fire blanket and the extinguishing lance with seawater. Approximately 15 min after the initiation of the test, off-gas
were observed. The fire suppression efforts began 10 min after the TR event was triggered. The fire suppression methods employed
included the combination of the fire blanket and the extinguishing lance with seawater, which are recognized as highly effective in
maritime fire safety applications. Fire, explosion, and arc incidents were detected 8 min after the commencement of suppression,
causing a notable increase in pressure beneath the fire blanket due to the accumulation of gases. This pressure caused the fire blanket to
lift slightly, allowing gases to escape through the gaps between the fire blanket and the EV mock-up frame. However, no flames or
debris were expelled. Five minutes after the initial fire, explosion and arc events, a gas explosion was observed. The likely cause of this
explosion was the introduction of oxygen into the confined space of the fire blanket, which, coupled with the accumulation of
flammable gases generated by the TR, created an explosive mixture. The fire blanket and extinguishing lance with seawater were re-
secured following the explosion, and no further incidents of fire, explosion, or arc were noted. The fire suppression was terminated at
106 min, having deployed a total of 5 tons of seawater. Post-suppression observations indicated an initial surge in gas emissions, which
diminished over time, with no additional fire, explosion, or arc events.
Fig. 4(b) presents the TR of the battery pack at 50 % SOC with seawater injection. Off-gas was observed approximately 20 min after
the test commenced. The suppression started 10 min after the TR event using seawater injection, which is widely available and
practical for maritime fire suppression. Fire, explosion, and arc events were observed 14 min after the suppression began, with these
incidents recurring and attenuating periodically. Similar to case 4, a total of 5 tons of seawater was used, and the suppression efforts
were concluded at 70 min. Despite the suppression, external flames from the battery pack persisted initially but gradually extinguished
over time. Continuous gas emissions were observed, but no further fire, explosion, or arc events occurred, leading to the termination of

Fig. 3b. TR behavior of the battery pack at different SOC levels: 50 % SOC.

6
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

Fig. 3c. TR behavior of the battery pack at different SOC levels: 30 % SOC.

Fig. 4a. TR behavior of the battery pack at 50 % SOC with different extinguishing methods: with the fire blanket and the extinguishing lance
with seawater.

the test.
Fig. 4(c) depicts the TR behavior of battery pack at 50 % SOC, where the fire blanket was employed as the primary suppression
method. The suppression was initiated simultaneously with the other tests. Fire, explosion, and arc events were recorded 16 min into
the suppression effort, but the fire blanket effectively contained these incidents, preventing the ejection of flames and debris. The fire
blanket demonstrated sufficient fire resistance, as it did not sustain significant damage despite exposure to intense flames. Following

7
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

Fig. 4b. TR behavior of the battery pack at 50 % SOC with different extinguishing methods: with seawater injection.

Fig. 4c. TR behavior of the battery pack at 50 % SOC with different extinguishing methods: with the fire blanket.

the initial gas emissions through the gaps between the fire blanket and the vehicle, the blanket was readjusted. Continuous, minor gas
emissions were observed. A significant gas explosion occurred at 107 min, powerful enough to lift the fire blanket. Compared to case 4,
the gas explosion in case 6 occurred later but with higher explosion pressure. This is thought to be because case 6 facilitated easier
formation of combustible gases within the explosive range compared to case 4. In case 4, the combination of the fire blanket and the

8
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

extinguishing lance with seawater created an environment that allowed for controlled gas accumulation, leading to a less intense
explosion. This suggests that while the strategy employed in case 4 was effective at managing the initial risks, the conditions also
contributed to the eventual buildup of gas, which still resulted in an explosion, albeit less severe than in case 6.
Fig. 5 captures the sequence of events during the EV fire test at 50 % SOC. Off-gas was first observed at 74 min, followed by flame
observations at 84 min and beneath the EV at 86 min. A massive off-gas release occurred at 98 min, with flames spreading through the
EV’s chassis and interior by 111 min. The fire spread to the front and side tires and then to the rear, with the EV fully ablaze by 125 min.
The fire began to diminish at 140 min, finally ending at 180 min. This extended duration and widespread fire indicate significant
thermal energy and propagation pathways in the vehicle structure. Comparing case 1, 2, and 3 with case 7 reveals key differences
between isolated battery packs and the actual EV. The isolated battery packs (Cases 1, 2, and 3) exhibited more contained fire be­
haviors, lacking the additional fuel sources and structural complexities of a complete vehicle. However, during TR, battery packs
demonstrated distinct fire behavior characteristics such as jet flames and arcs. These phenomena were clearly observed in the isolated
battery pack tests, illustrating the specific fire dynamics associated with battery TR. These tests provide meaningful insights into the
fundamental fire behavior of battery packs under controlled conditions. In contrast, the actual EV (Case 7) showed more complex fire
dynamics. The propagation of flames and off-gas within the vehicle’s chassis and interior components contributed to the extended fire
duration and intensity. Both the isolated battery pack tests and the actual EV fire test are crucial for developing effective fire man­
agement strategies and enhancing safety in EVs transported within PCTC vessels.

3.2. Analyzing the temperature behavior on EV fires

Fig. 6(a) depicts the temperature variations during the fire test of the EV battery pack for case 1 (SOC 70 %). The first graph shows
the temperature behavior of the module with the HF attached. As indicated by HF1, the temperature of the HF continuously increased
from the start of the test until TR occurred at 313 ◦ C. The module design includes insulation between cells and low thermal con­
ductivity within the cells, resulting in the module’s temperature not significantly rising even as the HF temperature increased.
However, upon TR, a rapid temperature rise was observed in CE and CS, followed by CW and CN. In Fig. 6, the discrepancy in
temperature readings is attributed to the effects of fire, explosion, and arc incidents, leading to anomalous measurements. These
readings, indicated in green, are marked as such because the measurements became unreliable after the occurrence of these incidents.
The second graph of Fig. 6(a) show the temperatures of the right edge modules within the battery pack. Heat transfer from the module
with the HF caused the temperatures in the adjacent modules, R2, to rise, with R2 experiencing a higher temperature due to its
proximity. Around 28 min, TR occurred in the right edge module, preceding massive off-gas release events. The third and fourth graphs

Fig. 5. TR behavior of the actual EV at 50 % SOC.

9
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

Fig. 6a. Temperature variations during fire tests of EV battery packs: SOC 70 %.

illustrate the temperature of the center and left edge modules. Due to the distance from the module with the HF and barrier, as well as
the structure of the battery pack with the battery management system located at the center, the temperature increase during TR was
relatively slower. After the fire/explosion/arc events, temperatures exceeded the thermocouple’s measurement limit of 1350 ◦ C,

10
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

Fig. 6b. Temperature variations during fire tests of EV battery packs: SOC 50 %.

11
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

Fig. 6c. Temperature variations during fire tests of EV battery packs: SOC 30 %.

12
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

indicating significant damage and complex temperature changes. The overall battery pack inside temperature stabilized around 500 ◦ C
when no further fire/explosions occurred.
Fig. 6(b) shows the temperature variations during the fire test of battery pack for case 2 (SOC 50 %). The first graph indicates that
the HF temperature increased similarly to case 1, with TR occurring at 304 ◦ C. It suggests that TR in cells occurs within the 304–313 ◦ C
range. The second, third, and fourth graphs depict the temperatures of the right, central, and left edge modules within the battery pack.
Prior to fire/explosion/arc events, the temperature increased horizontally and downward from the HF module, reaching around
100 ◦ C, insufficient to cause TR. Heat accumulation from the TR led to fire/explosion/arc events, which appeared more concentrated
and shorter in duration compared to case 1. The overall battery pack inside temperature stabilized around 550 ◦ C when no further fire/
explosions occurred.
Fig. 6(c) shows the temperature variations during the fire test of battery pack for case 3 (SOC 30 %). The HF temperature increased

Fig. 7a. Temperature variations during fire suppression tests of battery packs at 50 % SOC: fire blanket and extinguishing lance with seawater.

13
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

similarly to cases 1 and 2, but no TR occurred initially. After 2 h, additional HF was applied, leading to TR and off-gas observation
starting around 14 min. The first graph confirms off-gas occurrence at 14 min, but no external flames were observed. The subsequent
graphs indicate lower overall temperatures within the battery pack compared to cases 1 and 2. Heat transfer was most significant near
the TR module, with the highest temperatures recorded at the central thermocouple. After approximately 250 min, the edge module
temperatures surpassed those of the central module, reaching potential TR temperatures. No off-gas or flames were observed, but fire/
explosion/arc events occurred 1 h post-test, suggesting ongoing chemical reactions and temperature increases within the battery pack.
Comparing the SOC effects on temperature behavior in case 1 and case 2, off-gas was observed 3 min earlier in case 2 than in case 1.
However, fire/explosion/arc events occurred 3 min later in case 2 than in case 1. Case 2 exhibited more concentrated and shorter
duration fire/explosion/arc events, likely due to interactions between the battery enclosure and busbars. Comparing cases 1 and 2 with

Fig. 7b. Temperature variations during fire suppression tests of battery packs at 50 % SOC: seawater injection.

14
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

case 3, case 3 required more thermal energy to induce TR, indicating greater safety for transport. However, once TR occurred,
persistent monitoring was essential despite the absence of visible off-gas or flames. Battery packs exhibited distinct TR behaviors with
jet flames and arcs clearly observed. These phenomena are critical for understanding fire dynamics in battery packs. Lower SOC levels
(Case 3) indicated safer conditions for transport, but thorough monitoring is required once TR is initiated. Higher SOC levels (Cases 1
and 2) showed quicker thermal responses, requiring robust safety measures. Effective thermal insulation and low thermal conductivity
within cells helped contain initial heating but were insufficient to prevent TR.
Fig. 7 illustrates the temperature behavior of battery packs at 50 % SOC using fire suppression methods typically employed on car
carriers. Before suppression begins, the temperature trends closely resemble those observed in Fig. 6(b). However, once suppression
starts, temperatures are measured lower than those shown in Fig. 6(b). The discrepancy in temperature readings, similar to those
marked in green in Fig. 6, is attributed to the combined effects of seawater used in the suppression efforts and the subsequent fire,

Fig. 7c. Temperature variations during fire suppression tests of battery packs at 50 % SOC: fire blanket.

15
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

explosion, and arc incidents. These incidents led to anomalous temperature readings post-suppression. The anomalous readings,
indicated in green in Fig. 7, are likely due to the TR and the application of seawater. Contact with seawater can affect the accuracy of
temperature measurements taken by thermocouples, rendering them unreliable [17]. Despite these anomalies, the influence of
different suppression methods on the regions where accurate measurements were obtained is still presented. In Fig. 7(a), the appli­
cation of a fire blanket combined with an extinguishing lance using seawater shows an initial temperature drop, reflecting the im­
mediate cooling effect of seawater and the asphyxiation effect of the fire blanket. After suppression, the temperatures were measured
lower than those shown in Fig. 6(b). However, although a significant temperature decrease was expected due to the cooling effect from
the latent heat of evaporation of the water mist and the asphyxiation effect of the fire blanket, the observed temperature changes did
not show a substantial difference compared to the fire test without suppression. Additionally, the occurrence of fire, explosion, and arc
events appeared approximately 9 min earlier than in case 2, based on the observation of off-gas. This trend is likely due to the easy

Fig. 8a. Temperature variations at positions A and C during fire test for case 7 (actual EV at 50 % SOC).

16
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

formation of conditions conducive to gas explosions caused by the accumulation of flammable gases within the fire blanket and the
high electrical conductivity of seawater facilitating the formation of external circuits, inducing TR and short circuits. Fig. 7(b) shows
the temperature variations during the fire extinguishing test for case 5 (SOC 50 % with seawater injection). Seawater, while readily
available and possessing excellent cooling properties, presented challenges when applied to the battery pack. Initially, temperatures at
all locations showed a tendency to decrease once suppression began, particularly in CS, which experienced the most rapid temperature
drop. Due to the contact of a larger quantity of seawater with the battery pack compared to case 4, anomalous temperature mea­
surements appear immediately upon the start of suppression. Fig. 7(c) illustrates the temperature variations during the fire extin­
guishing test for case 6 (SOC 50 % with the fire blanket). The fire blanket aimed to smother the fire by covering the entire battery pack,
reducing the spread of smoke and fire by cutting off the air supply. Because seawater is not applied, the areas with anomalous tem­
perature measurements appear later compared to other suppression methods. No significant temperature change occurred

Fig. 8b. Temperature variations at positions B and D during fire test for case 7 (actual EV at 50 % SOC).

17
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

immediately upon suppression, but temperatures ceased to rise about 5 min later. Complete suppression was not achieved, with
fire/explosion/arc events occurring at 48 min. The post-fire/explosion/arc temperatures remained below 400 ◦ C, about 150 ◦ C lower
than in the fire test. A comparative analysis of Cases 4, 5, and 6 shows that seawater is effective for cooling, while fire blankets help
prevent the spread of fire. However, using them simultaneously in Case 4 combines both effects, addressing both temperature control
and containment. Nonetheless, seawater’s electrical conductivity presents a risk, indicating that alternative suppression methods may
be necessary to achieve effective suppression without compromising battery integrity. Fig. 8 illustrates the temperature variations
during the fire test for case 7, which involved the actual EV. The left graph of Fig. 8(a)shows the temperature behavior in the C region,
where the HF was attached. In the actual EV fire test, the HF2 was powered for 1 h, maintaining temperatures above 200 ◦ C, but TR did
not occur. Subsequently, HF1 was also powered, leading to TR approximately 12 min later. This TR is attributed to the cooling effect of
the coolant in the EV, which was not considered in previous cases. The coolant maintained the temperature in the battery pack,
necessitating additional heat input to induce TR. Upon TR, the temperature of module C5, where the HF was attached, increased
rapidly, followed by modules C4 and C6, C3 and C7, and finally C2, C8, and C1. This pattern indicates efficient heat transfer to nearby
modules, with all modules in the C region experiencing TR within 40 min of the initial event. Post-TR, temperatures fluctuated
significantly due to flames and arcs, with temperatures around 700 ◦ C during the fire. Comparing the temperatures recorded during the
fire tests for case 7 and case 2 (Battery pack at 50 % SOC), off-gas was observed at 265.1 ◦ C in case 7 and 250.8 ◦ C in case 2, showing a
difference of about 15 ◦ C. The higher temperature in case 7 can be attributed to the presence of coolant in the actual EV, which required
the application of two film heaters to induce TR. This resulted in a higher observed temperature compared to case 2. However, the
higher off-gas temperature in case 7, along with the maximum temperature during TR being 674.9 ◦ C, compared to 706.6 ◦ C in case 2,
indicates the mitigating effect of the coolant in reducing peak temperatures. The right graph of Fig. 8(a) and (b)depict the temperature
changes in regions A, B, and D during the actual EV fire test. The temperatures in these regions rose due to fire spread from region C. TR
began in regions A, B, and D after all modules in region C experienced TR, with region D showing the first occurrence. This is likely
because region D is closer to region C and because a vent hole formed due to electrical connection damage in the rear of battery pack,
directing the fire toward region D. Modules located at the edges of regions A, B, and D (A1, A8, C1, C8, D1, D8) showed the latest
temperature increases, attributed to their relative distance from the central heat source in C5. Even after fire attenuation, the tem­
perature of battery pack remained around 700 ◦ C, indicating significant residual heat.

3.3. Analyzing HRR and THR on EV fires

Fig. 9(a) illustrates the HRR and THR for case 1 and case 2. During a fire, the HRR of the battery pack is significantly affected by the
occurrence of fire, explosions, and arcs. These events cause a rapid increase in HRR at the onset of TR. In case 1, the fire, explosions,
and arcs were observed to occur and attenuate between 41 min and 158 min, resulting in the THR of 673.37 MJ. In contrast, case 2
exhibited a more concentrated period of fire, explosions, and arcs, occurring between 44 min and 110 min, with the THR of 808.23 MJ.
The higher THR in case 2 compared to case 1 is attributed to the intense and concentrated fire, explosions, and arcs caused by the
contact between the outer casing and the busbar in the battery pack. Case 3 did not exhibit any external fire, explosions, or arcs, and
therefore, no heat release was measured. These results indicate the possibility that the HRR and THR of battery packs with an SOC of
50 % or higher are more influenced by the manner in which the fire occurs and propagates rather than the SOC itself. Fig. 9(b) presents
the HRR and THR for case 4, case 5, and case 6. In case 4, the maximum HRR measured was 539.48 kW due to fire, explosions, and arcs.
After the fire was extinguished and the asphyxiation blanket was removed at 106 min, gas continued to be released, leading to an
increase in heat release. Case 5 showed a maximum HRR of 273.19 kW, with a THR of 81.71 MJ, the lowest among the cases, due to the
use of seawater injection for fire suppression. Case 6 experienced a gas explosion within the asphyxiation blanket, resulting in the
highest HRR of 899.74 kW. These findings indicate that while seawater can maintain a lower HRR during EV fires, the use of an
asphyxiation blanket can cause momentarily high HRR due to gas explosions. Fig. 9(c) compares the HRR of case 2 with t-square fires
categorized as slow, medium, fast, and ultra-fast. T-square fires are a standardized method for modeling fire growth rates, defined by
the formula HRR = αt2, where α is a constant that represents the fire growth rate and t is the time elapsed since ignition. T-square fires
are used to predict the development of a fire over time and are classified based on their growth rates: slow (α = 0.00293 kW/s2),
medium (α = 0.01172 kW/s2), fast (α = 0.0469 kW/s2), and ultra-fast (α = 0.1876 kW/s2). These models are commonly used in fire
safety engineering to design and evaluate fire protection systems, such as sprinkler installations and smoke control systems. However,
battery pack fires do not fit well within the t-square fire framework due to the unique characteristics of TR and the associated rapid and
intense heat release. The sudden onset and high peak HRR during TR are not accurately represented by the gradual increase assumed in
t-square fire models. Fig. 9(d) shows the HRR and THR of the actual EV fire (Case 7) compared to t-square fires. During the fire test of
the actual EV at 50 % SOC, the maximum HRR reached 6292.97 kW, and the THR was 9998.56 MJ. The significantly higher HRR in the
actual EV is due to the presence of numerous combustible materials within the vehicle, such as interior and exterior components and
tires. These additional materials contribute to the higher HRRs observed. The burn characteristics of these materials differ from those
of the battery pack alone, justifying the higher HRR and THR. Furthermore, the actual EV fire aligns more closely with the medium
category of t-square fires due to the additional combustible materials present in the vehicle. This finding suggests that while t-square
fires may not be applicable for estimating battery pack fires, they can be used for actual EV fires, considering the contribution of
various interior and exterior combustibles. Overall, these results indicate that the application of t-square fires is inappropriate for
battery pack fires due to the unique nature of TR and its associated heat release patterns. However, for actual EV fires, which involve
additional combustible materials, the t-square fire model can be a useful tool for estimating fire growth and heat release. The
significantly higher HRR and THR in actual EV fires underscore the importance of considering the full range of combustible materials
and the complexities of fire dynamics in EVs.

18
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

Fig. 9. HRR and THR during fire tests: (a) HRR and THR for case 1 and case 2, (b) HRR and THR for case 4, case 5, and case 6, (c) comparison of
HRR for case 2 with t-square fires, (d) HRR and THR for case 7 compared to t-square fires.

3.4. Analyzing the RHF behavior on EV fires

Fig. 10 illustrates the changes in RHF for case 1, case 2, case 3, and case 7. For the battery packs in case 1, case 2, and case 3, the
RHF was measured at a distance of 2 m. In the actual EV fire scenario (Case 7), the RHF was measured at 30 cm from the front and rear
of the vehicle and 15 cm from the sides. At a distance of 2 m, the maximum RHF was 4.024 kW/m2 in case 1 and 3.409 kW/m2 in case 2.

Fig. 10. RHF changes during fire tests (a) case 1, case 2, case 3 and (b) case 7.

19
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

Case 3 did not experience any fire, explosion, or arc, and thus, no RHF was measured. During the actual EV fire (Case 7), the RHF was
measured at different points around the vehicle: H1 (front), H2 (rear), H3, and H4 (sides). The maximum RHF was recorded at 91.41
kW/m2 at the rear of the vehicle and 62.35 kW/m2 at the front. The sides of the vehicle showed similar RHF trends. The higher values
measured in the actual EV fire compared to the battery pack fires are attributed to the shorter measurement distances (15 cm and 30
cm) and the direct exposure to flames. The comparison between the battery pack tests and the actual EV fire highlights the influence of
measurement distance on RHF values. In the battery pack tests, the 2-m distance resulted in lower RHF readings due to the greater
distance from the flame source. Conversely, in the actual EV fire, the closer measurement distances led to significantly higher RHF
values. Additionally, the presence of the battery pack along with the vehicle’s internal and external materials, and tires in the actual
EV, contributes to the higher RHF observed, as these materials can intensify the fire and increase heat radiation. In the context of a car
carrier, vehicles are typically loaded with side-to-side gaps of 10–20 cm and front-to-back gaps of 30 cm. This close proximity increases
the likelihood of fire spreading to adjacent vehicles in the event of a fire. Therefore, it is imperative to isolate the burning vehicle
promptly, as the confined space on car carriers makes this challenging. The significantly higher RHF observed in the actual EV fire
underscores the need for effective fire management strategies on car carriers. To enhance fire safety, additional measures to block RHF
are necessary to prevent the spread of fire to nearby vehicles. These strategies should include quick isolation of the burning vehicle
through rapid deployment of fire blankets, as well as the installation of fire-resistant barriers or partitions between vehicles. Materials
with high reflectivity and low thermal conductivity, such as aluminum-based or ceramic fiber blankets, can reflect thermal radiation
effectively, reducing heat transfer. Additionally, intumescent coatings applied to vehicle compartments can expand under high
temperatures to create an insulating layer that minimizes RHF. Physical barriers made of fire-rated materials, like calcium silicate or
composite panels, can further contain heat and flames, enhancing the overall fire resistance of the carrier environment. Implementing
these measures in car carriers can significantly improve the safety of EVs during transport by limiting fire spread and managing intense
heat release. Future research should continue to explore optimal material combinations and barrier designs to maximize RHF blockage,
contributing to safer protocols for EV transportation on maritime vessels.

4. Conclusions

This study provides a comprehensive analysis of fire behavior and extinguishing methods for EV battery packs and the actual EV at
different SOC levels, focusing on their implications for maritime transportation safety. Key conclusions are as follows.

1. SOC Management: Lowering the SOC to approximately 30 % or below can significantly enhance the safety of EV transportation by
reducing the intensity and duration of fires. Specifically, the 30 % SOC level demonstrated a 20–30 % reduction in fire intensity and
a decreased pHRR, with maximum HRR of approximately 450 MJ, compared to around 673 MJ for 70 % SOC. Although lower SOC
levels exhibited safer conditions, they required careful monitoring due to potential re-ignition. This finding suggests that main­
taining a lower SOC is a practical measure to improve the safety of EVs during maritime transport.
2. Fire Suppression Methods: Different fire suppression methods were evaluated. Seawater injection showed an initial cooling effect,
lowering temperatures by approximately 150 ◦ C, but posed risks due to its electrical conductivity, which can exacerbate TR. The
fire blanket reduced temperature increases but did not fully suppress fires, necessitating additional measures for complete fire
control. The combination of the fire blanket and seawater injection was effective in rapidly reducing temperatures and containing
the fire spread. However, the electrical conductivity of seawater remains a concern for exacerbating TR, indicating that alternative
methods may need to be considered for safe suppression without compromising battery integrity. These findings demonstrate that
while full suppression might be challenging, these methods can effectively prevent the spread of fire and maintain lower
temperatures.
3. Firefighting Strategies: When off-gas is observed from an EV, it is crucial to promptly use the fire blanket to prevent fire spread.
Firefighters must wear complete protective gear when approaching the asphyxiation blanket to mitigate the risk of gas explosions.
Additionally, when using seawater, the risk of electric shock and explosion persists, necessitating full protective gear. Even after fire
suppression, continuous monitoring and the installation of barriers are essential to prevent re-ignition, with monitoring systems
ideally capable of detecting temperature increases promptly and accurately.

The findings emphasize the importance of developing effective fire management strategies and enhancing safety protocols for the
maritime transportation of EVs. Key strategies include maintaining SOC at or below 30 % to reduce the intensity and duration of
potential fires, deploying fire blankets rapidly, and ensuring strategic spacing of vehicles to mitigate fire spread. Additionally, the
integration of advanced fire suppression systems is essential to address the specific challenges posed by EV fires in maritime envi­
ronments. These results suggest that fire suppression approaches should focus on combining methods to control both temperature and
fire spread while minimizing risks related to seawater conductivity. By aligning fire suppression methods with observed fire behavior,
industry practices can better manage the risks associated with EV fires during transport. Future research should focus on developing
integrated fire suppression technologies that provide adaptable responses to different fire scenarios, as well as robust safety measures
tailored for EV maritime transport. These advancements will contribute to safer transportation of EVs on ships, supporting a proactive
approach to EV fire safety management in confined and high-risk settings.

CRediT authorship contribution statement

Suhaeng Lee: Writing – original draft, Investigation, Data curation. Daehyun Choi: Investigation, Data curation. Yeoseon Jeong:

20
S. Lee et al. Case Studies in Thermal Engineering 64 (2024) 105474

Visualization, Data curation. Minho Moon: Visualization, Data curation. Hyukjoo Kwon: Visualization, Data curation. Kukil Han:
Visualization, Data curation. Hyungjun Kim: Methodology, Investigation. Hongsoon Im: Methodology, Investigation. Youngseob
Park: Methodology, Investigation. Dongki Shin: Methodology, Investigation. Geonhui Gwak: Writing – review & editing, Writing –
original draft, Supervision, Methodology, Conceptualization.

Declaration of Generative AI and AI-Assisted technologies in the Writing process

During the preparation of this work, the author(s) used ChatGPT 4o and other services for grammar checking and readability
enhancement because the native language of the author(s) is not English. After using these tools, the author(s) reviewed and edited the
content as needed and take full responsibility for the content of the publication.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.

Acknowledgement

This research was supported by the 2022 Civil-Military Technology Cooperation Program funded by the Defense Acquisition
Program Administration (DAPA) of the Republic of Korea (No. 22-DC-ME-03). The authors would like to thank the related departments
for their support. We would also like to extend our special thanks to the 3346 Research Group from the Advanced Fire & Disaster
Management Center at the Korea Conformity Laboratories for their invaluable support.

Data availability

The authors do not have permission to share data.

References

[1] European Maritime Safety Agency (EMSA), Guidance on the carriage of AFVs in RO-RO spaces, Available at: https://www.emsa.europa.eu/publications/
reports/item/4729-guidance-on-the-carriage-of-afvs-in-ro-ro-spaces.html, 2022.
[2] G. Zhang, X. Wei, S. Chen, J. Zhu, et al., Comprehensively investigating the impact of high-temperature cyclic aging on thermal runaway characteristics for
lithium-ion batteries, SAE Tech Pap 2022–01–7061 (2022).
[3] H. Li, H. Chen, J. Sun, Q. Wang, Full-scale experimental study on the combustion behavior of lithium-ion battery pack used for electric vehicle, Fire Technol. 56
(2020) 2545–2564.
[4] S. Kang, M. Kwon, J.Y. Choi, S. Choi, Full-scale fire testing of battery electric vehicles, Appl. Energy 332 (2023) 120497.
[5] C. Zhao, W. Hu, D. Meng, W. Mi, X. Wang, J. Wang, Full-scale experimental study of the characteristics of electric vehicle fires process and response measures,
Case Stud. Therm. Eng. 53 (2024) 103889.
[6] Y. Cui, J. Liu, B. Cong, X. Han, S. Yin, Characterization and assessment of fire evolution process of electric vehicles placed in parallel, Process Saf. Environ. Prot.
166 (2022) 524–534.
[7] A. Dorsz, M. Lewandowski, Analysis of Fire hazards associated with the operation of electric vehicle in enclosed structures, Energies 15 (2022) 11.
[8] Y. Cui, B. Cong, J. Liu, M. Qiu, X. Han, Characteristics and hazards of plug-in hybrid electric vehicle fires caused by lithium ion battery packs with thermal
runaway, frontiers 10 (2022) 11.
[9] Y. Cui, J. Liu, X. Han, S. Sun, B. Cong, Full-scale experimental study on suppressing lithium-ion battery pack fires from electric vehicles, Fire Saf. J. 129 (2022)
103562.
[10] S. Peter, F. Patrik, F. Daniel, G. Robert, W. Robert, F. Simon, K. Stefan, H. Oliver, Fire tests with lithium-ion battery electric vehicles in road tunnels, Fire Saf. J.
134 (2022) 103695.
[11] S. Ubaldi, C. Di Bari, M. Quinterno, A. De Rosa, M. Mazzaro, G. Ferrigno, D. Secci, P. Russo, Suppression capacity and environmental impact of three
extinguishing agents for lithium-ion battery fires, Case Stud. Chem. Environ. Eng. 10 (2024) 100810.
[12] J. Xu, P. Guo, Q. Duan, X. Yu, L. Zhang, Y. Liu, Q. Wang, Experimental study of the effectiveness of three kinds of extinguishing agents on suppressing lithium-
ion battery fires, Appl. Therm. Eng. 171 (2020) 115076.
[13] J. Svetlik, Z. Tancos, P. Tancos, I. Markova, K. Slastan, Large-scale fire tests of battery electric vehicle (BEV): Slovak case study, Appl. Sci. 14 (2024) 14104013.
[14] E. Funk, K.W. Flecknoe-Brown, T. Wijesekere, B.P. Husted, B. Andres, Fire extinguishment tests of electric vehicles in an open sided enclosure, Fire Saf. J. 141
(2023) 103920.
[15] C. Zhao, W. Hu, D. Meng, W. Mi, X. Wang, J. Wang, Full-scale experimental study of the characteristics of electric vehicle fires process and response measures,
Case Stud. Therm. Eng. 53 (2024) 103889.
[16] National Institute of Standards and Technology (NIST), Cone calorimeter. https://www.nist.gov/laboratories/tools-instruments/cone-calorimeter, 2021.
(Accessed 9 September 2024).
[17] D. Ripple, K. Garrity, Uncertainty budgets for comparison calibrations of thermocouples. NCSL International Annual Workshop and Symposium, 2005.

21