1 s2.0 S0957582022007455 Main

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Process Safety and Environmental Protection 166 (2022) 524–534

Contents lists available at ScienceDirect

Process Safety and Environmental Protection


journal homepage: www.journals.elsevier.com/process-safety-and-environmental-protection

Characterization and assessment of fire evolution process of electric


vehicles placed in parallel
Yan Cui a, Jianghong Liu a, Beihua Cong b, Xin Han b, *, Sumiao Yin a
a
Ocean Science and Engineering College, Shanghai Maritime University, Shanghai 201306, China
b
Shanghai Institute of Disaster Prevention and Relief, Tongji University, Shanghai 200092, China

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

Keywords: Electric vehicle (EV) fire accidents induced by the thermal runaway of li-ion batteries are increasingly frequent,
Full-scale fire experiment and the fire dynamics of EVs parked in rows are still unclear. Hence the fire evolution process and characteristics
Electric vehicle of two parallel placed EVs were studied by analyzing the burning behaviors and thermal fields. The fire source
Lithium-ion battery
was the battery pack of a battery electric vehicle (BEV). Results showed that the precursor to the BEV fire was the
Thermal runaway
Fire characteristic
emission of white smoke from the chassis. Flames did not appear until the accumulated smoke exploded, and
they engulfed both the EVs. The flame switched between momentum and buoyancy control on the vehicle scale.
The maximums of the length and duration of the jet fires were determined. The maximum temperature of the EV
fire was consistent with that of internal combustion engine vehicle (ICEV) fires, but the flame spread faster
between the EVs compared to ICEVs. Pragmatic methods were proposed to quantify the fire evolution rate. The
heat effect of the EV fire on humans was also quantitatively evaluated. These results provide a fundamental
understanding of fire rules in a high EV concentration scenario.

1. Introduction capacity of battery packs is constantly increasing, which causes the cars
to release lots of energy when they catch fire. Sun et al. (2020) collected
Lithium-ion batteries (LIBs) have been used in energy storage and combustion data from 9 BEV fire tests and found that their peak heat
portable electronics since they were commercialized in 1991 (Mao et al., release rate (PHRR) ranges from 4.2 to 7 MW, and their total heat release
2020; Leuthner, 2013; Lisbona and Snee, 2011). With the rise of green (THR) is between 4.7 and 8.5 GJ. These issues raise concerns about the
transportation in the past two decades, LIBs are now the core part of fire safety of EVs and onboard LIB energy storage systems.
electric transportation due to their lightweight and high energy density EV fires have hazards of high temperature, jet fire, and explosion.
compared to lead-acid batteries. Traction LIB packs with large capacity Boe (2017) studied EV fire behaviors using cars converted from internal
and high voltage enter the automobile industry, making hybrid electric combustion engine vehicles (ICEVs). The flame temperature measured
vehicles (HEVs), plug-in hybrid electric vehicles (PHEVs), and battery during the experiment was above 900 ◦ C. Andersson et al. (2016)
electric vehicles (BEVs) more popular. The global electric vehicle market analyzed the temperature distribution inside a burning electric hybrid
is estimated to rise to $93.1 billion by 2025 (Diaz et al., 2020). bus and found that its maximal flame temperature is about 1000 ℃. A jet
However, the automotive battery packs may suffer from thermal flame is one of the fire hazards of burning EVs. It can injure occupants
runaway (TR) when they encounter overheat, overcharge, or a crash. and ignite nearby inflammable goods, triggering a more severe fire. A
The flame from TR battery packs can spread throughout the EV and LIB under TR produces plenty of gas inside its shell, and the accumu­
result in an EV fire. For instance, the average annual number of EV fire lation of the gas not only creates high pressure but when it is released
accidents from 2011 to 2018 in China is 31, and even in the epidemic- and encounters a heat source, a jet flame may be formed (Chen et al.,
affected year 2020, there are 124 reported EV fire incidents, according 2019; Cui and Liu, 2021). The maximum height of jet flame of the
to incomplete statistics (Diaz et al., 2020; Qiu et al., 2022; United Na­ 18650-type LIB was 0.31 m (Mao et al., 2021). The peak jet velocity was
tions Economic Commission for Europe, 2018). Other countries are 42.1 m/s and appeared at the moment when the battery safety valve
facing the same problem. Besides, to improve the range of EVs, the opened (Zhou et al., 2021). A 50 Ah LIB module has a jet flame height of

* Corresponding author.
E-mail address: hanxin@tongji.edu.cn (X. Han).

https://doi.org/10.1016/j.psep.2022.08.055
Received 28 May 2022; Received in revised form 21 August 2022; Accepted 22 August 2022
Available online 25 August 2022
0957-5820/© 2022 Institution of Chemical Engineers. Published by Elsevier Ltd. All rights reserved.
Y. Cui et al. Process Safety and Environmental Protection 166 (2022) 524–534

0.96 m (Ping et al., 2015). In the full-size PHEV fire experiment con­ the battery pack. The LIB pack of PHEV A was a cuboid and was located
ducted by Cui et al. (2022), the length of the jet flame from the vehicle within the chassis directly under the rear seats. Due to the small capacity
chassis was approximately 1 m. Part of the jet fire may be obscured by of the battery pack, the PHEV in pure electric mode could only travel 63
the fabric. Moreover, a PHEV fire is likely accompanied by explosions. km. Its fuel tank was empty. Both the battery packs were original and
This phenomenon originates from the bursting of burning tires and the fully charged. Prismatic cells were the primary component of the battery
rapid oxidation and expansion of combustible gases. The TR gases packs. Steel armor plates that were also original products protected the
include electrolytes, carbon monoxide, hydrogen, and alkanes (Wang LIB pack in each car.
and Huang, 2020; Ouyang et al., 2022; Yuan et al., 2020). When the
gaseous mixture accumulates in the outside space of the batteries, there
2.2. Instrumentation
is a risk of an explosion once it meets a spark of sufficient strength.
Nevertheless, the intensity of the jet flame and whether the gas phase
Two electric furnaces (radiant type) triggered TR of the battery pack
explosion occurs are unknown in BEV fires (Feng et al., 2020; Bisschop
of BEV A to simulate an EV fire caused by external fire sources. The
et al., 2020). Most of the EVs used in the existing fire tests are rebuilt or
heating power of the furnace was constant at 3000 W, and they were
old models, making it challenging to explore fire patterns of modern
immediately turned off when there was a visible fire around the vehicle.
EVs.
These electric furnaces were only present underneath BEV A and were
Massive EV fires in parking lots, tunnels, and roll-on/roll-off ships
right below its battery pack, and PHEV A was not directly heated. Baffles
are catastrophic. Two ro-ro vehicle carrier fires, which may have been
were placed around the BEV to maintain the heating performance of the
induced by automotive batteries (National Transportation Safety Board,
furnaces by blocking part of the ambient airflow. PHEV A was ignited by
2021), occurred in the last two years, namely the Höegh Xiamen and
the flames escaping from BEV A. Cars’ tires were partially deflated to
Felicity Ace fires. The flame spread among vehicles was an essential
prevent burning tires from exploding and damaging the measurement
contributor to the destruction of the two vessels. The characteristics of
equipment. Fig. 1 shows the schematic of the experimental rigs used in
fire propagation from a burning ICEV to the surrounding cars are well
this study. The distance between BEV A and PHEV A was 0.6 m. This
established. Fire data is the basis for fire safety design (Dorsz and
value represented the space between vehicles in a realistic scenario, such
Lewandowski, 2022; Jiang et al., 2020; Brzezinska and Bryant, 2022).
as a parking lot or tunnel. Moreover, the left or right direction
However, there is no available data on behaviors and features of fire
mentioned in this study was based on the driver’s angle.
propagation among multiple modern EVs (Sun et al., 2020; Lesiak et al.,
A total of 40 K-type thermocouples were employed to record the
2021), and ICEV fire parameters are not suitable for evaluating this fire
temperature variation. The K-type thermocouple had a diameter of
risk.
3 mm and a thermal response time of 0.6 s. The recording error was
This paper designs a full-scale fire experiment to explore the fire
± 0.8 ◦ C. The thermocouples were divided into four groups. Three
evolution process and characteristics of two parallel placed EVs. One
bunches of thermocouples (T1~T7, T8~T14, and T15~T21), the first
vehicle is a BEV, and the other is a PHEV, both of which are brand new
group, were used to measure the temperature of the flames in the
products with good market. This configuration of vehicles is represen­
exterior space of car compartments of BEV A, including the trunk and
tative of a typical scenario that BEVs and PHEVs are parked together.
rear and front passenger compartments. Thermocouples T3–4, T10–11,
The initial ignition position is the battery pack of the BEV. The fire be­
and T17–18 were mounted next to the rear and side windows of BEV A
haviors, hot plume spread process, and spatial distributions of temper­
and about 3 cm away from the windows. These thermocouples were
ature and heat radiation of the two-EV fire are analyzed in detail.
directly below thermocouples T1–2, T8–9, and T15–16, and the distance
Methods are developed to quantify the propagation speed of the hot
between two vertically adjacent thermocouples was 0.5 m. Thermo­
smoke and flame based on the temperature field. And the thermal effect
couples T5–7, T12–14, and T19–21 were fixed near the chassis and had a
of the BEV fire on the human body is quantitatively assessed to illustrate
ground level of about 0.15 m. Thermocouples T22–33 were the second
EV fire hazards. In addition, requirements and optimization directions
group and tracked the temperature of flames on the left side of BEV A.
for EV fire safety are provided. These results enhance the understanding
The height of thermocouples T22, T24, T26, T28, T30, and T32 was the same
of EV fire dynamics and afford guidance on safety aspects for the
as that of the windows, while thermocouples T23, T25, T27, T29, T31, and
application of LIB energy storage technology to transportation.
T33 were at the level of 0.15 m. The gap between thermocouples T28–33
and T22–27 and BEV A was 0.5 m and 1 m, respectively. The temperature
2. Experimental setup
of the flames that reached the position of PHEV A was monitored by
thermocouples T34–36, which were fastened near the chassis at the height
2.1. Description of the cars
of 0.15 m. Thermocouples T37–40 logged the temperature inside the
compartments of BEV A. To illustrate the burning process of plastic parts
The full-scale fire experiment of LIB-powered road vehicles was
and seats, thermocouple T37 was mounted inside the trunk, and ther­
conducted in an open field. A total of two EVs were used. Table 1
mocouples T38–40 were placed above the seats.
summarizes the types and properties of the vehicles. BEV A and PHEV A
A total of five radiant heat flux gauges were placed on the rear and
were sedan passenger vehicles, and they were both brand new vehicles
left of BEV A to measure the thermal radiation effect of the BEV fire on
produced by one of the global leading car companies. During the
humans. The recording error was ± 0.4%. Radiative heat flux meters H1,
experiment, all windows of the two cars were closed. Traction LIBs in­
H2, and H5 were 1.5 m away from the car body, while H3 and H4 were
side BEV A were distributed in an H-shape in the chassis, which could
1 m away. All the heat flux gauges were 1.5 m from the ground.
operate the BEV for 278 km. A master safety valve was located on top of
Furthermore, a thermal imager recorded the fire evolution, and its

Table 1
Vehicle types and properties.
Code name Vehicle type Battery type Battery capacity (kWh) Fuel tank volume (L) Length (m) Width (m) Height (m)
a
BEV A BEV LiNixCoyMnzO2 / 38.1 — 4.670 1.806 1.474
graphite
PHEV A PHEV —b 13.0 50 4.948 1.836 1.469
a
The BEV did not have a fuel tank.
b
Official data was not available.

525
Y. Cui et al. Process Safety and Environmental Protection 166 (2022) 524–534

Fig. 1. Schematic of the experimental apparatus.

measuring range was from 0 to 275 ℃. induced this severe explosion. This kind of explosion has been observed
in the preceding fire experiments of PHEVs (Cui et al., 2022). Electrolyte
3. Results and discussion vapors, CxHy, CO, and H2 from the TR batteries were the ingredients of
the white smoke. To prevent confusion, the experiments conducted by
3.1. Combustion behavior of BEV Cui et al. (2022) are referred to as single-vehicle tests. It is worth noting
that in this two-car test, the duration that the white smoke was released
The fire evolution process of BEV A is shown in Fig. 2. The time at before visible flames appeared was shorter. In the single-vehicle tests,
which the electric furnaces started heating the chassis of BEV A was the battery pack of a PHEV emitted smoke periodically for 51 min 19 s,
defined as the experiment onset time. BEV A began to release white during which no flame was created. This distinction may be caused by
smoke at 91 min 58 s after the test started, as illustrated in Fig. 2a. the difference in abuse modes and the existence of the baffles. In this fire
Because of the shielding of the baffles, most of the white smoke was not test, the baffles allowed the combustible smoke to accumulate, whereas
dispersed by the wind but gathered at the bottom of the vehicle. Fig. 3a the electric furnaces timely ignited the smoke after it reached an
shows the extent of the accumulation. The smoke was released inter­ explosive concentration. This process is much more challenging for the
mittently for about 7 s, and then an explosion occurred at 92 min 5 s. PHEVs of the single-vehicle tests, which did not employ baffles and
Fig. 4 describes the explosion process. Flames first appeared around the where an external short circuit triggered TR. The PHEV has to wait for
rear wheel before rushing toward the front of the vehicle (Fig. 4b~d). the smoke to fail to be dispersed by the wind, and its battery pack needs
The shock wave damaged the fixed structure of the car hood, which time to produce a heat source that can ignite the cloud of the flammable
caused the hood to rise (Fig. 4e). Subsequently, the flames quickly smoke. Ultimately, the relatively confined space increases the explosion
contracted to the vehicle chassis (Fig. 4f). risk of EV fires.
The white smoke was released from the battery pack of BEV A and After the explosion, flames surrounded the chassis, and the baffles

Fig. 2. Burning process of BEV A between 91 min 58 s and 98 min 33 s.

526
Y. Cui et al. Process Safety and Environmental Protection 166 (2022) 524–534

Fig. 3. Smoke released from BEV A before (a) and after (b) the explosion.

Fig. 4. Explosion process during the test (92 min 5 s).

were removed (Fig. 2b). The burning battery pack, lacquered plate, 3.2. Fire propagation behaviors between BEV and PHEV
electrical wiring, and other polymers started generating brown and
black smoke (Fig. 3b). As shown in Fig. 2c, a jet flame appeared on the Thermal energy was transferred from the burning BEV to the PHEV
side of the BEV at 92 min 45 s. This discharge of the jet fire lasted about via heat convection and radiation. Fig. 5 presents the optical images and
3 s. Subsequently, jet flames erupted from the chassis two times, but thermal imageries of the fire spread process. In the early stage of the BEV
their durations were less than 11 s, and the flame length did not exceed fire, flames did not come into contact with the body of PHEV A, and the
2 m. After 94 min 23 s of the onset of the experiment, flames spread to PHEV absorbed heat through thermal radiation from flames as well as
the rear passenger and trunk compartments (Fig. 2d). Meanwhile, a jet thermal convection and radiation from hot smoke. At 94 min 16 s,
fire escaped from the rear of BEV A and lasted for approximately 26 s flames suddenly touched the car body, and the contact lasted for 2 s,
(Fig. 2e). This jet fire was much more intense than the previous jet then smoke began emitting from the left of PHEV A (Fig. 5a). The
flames, and it reached its maximum length of about 2.5 m at 94 min temperature of the smoke was about 80 ℃ (Fig. 5e). At 95 min 16 s,
37 s, which was also the maximal length of jet fire of the BEV. This fierce flames appeared on the left car doors after another contact between the
jet flame ended at 94 min 49 s. Then flames reached the front passenger car doors and BEV flames (Fig. 5b). The temperature of the alight area
and engine compartments at 95 min 53 s and 96 min 18 s, respectively. reached over 275 ℃ (Fig. 5f). At 96 min 10 s, the flames hit PHEV A
During this period, jet flames appeared again around the bodywork. again, which expanded the combustion area of the PHEV (Fig. 5c). The
Eventually, it took 98 min 33 s for flames to surround the entire vehicle. entire side of the vehicle was gradually engulfed in flames. It wasn’t
The battery pack under TR was the source of the jet flames. until 104 min 45 s after the experiment began that BEV flames and
Combustible gases and organic solvents were emitted from the battery PHEV flames blended (Fig. 5d). Furthermore, when flames were in
pack due to the internal pressure of the batteries, and they were ignited contact with the PHEV, heat convection of the flames was incorporated
to form jet fires when they met flames of other burning parts. Because of into the heat transfer process and accelerated the fire development.
the unstable gas production of the batteries, the jet flames could not be The movement direction of flames in PHEV A was different from that
maintained all the time, so the flame state of the BEV fire constantly in BEV A. In the BEV, flames propagated upward along the vehicle body
switched between momentum and buoyancy control. Furthermore, the and forward along the chassis, with the heated battery pack as the
injection direction of jet flames was influenced by the position of the origin. But for the PHEV, flames covered its left side first, then gradually
master safety valves and openings of the battery pack. The jet fire was eroded the longitudinal sections of the vehicle and advanced to the right
emitted from the side once and from the rear four times during the test. side. Flames poured out of the sunroof at 108 min 31 s. At the same time,
Flames that escaped through the safety valve emerged from the side the rear bumper was already half burned. The time for flames to reach
after being deflected by the vehicle fabric. The interface between the the right side was 110 min 38 s. It took another 45 s for the flame to
circuit system and the battery pack was located at the posterior of the engulf the PHEV completely. During the burning process, the battery
battery pack, and after the high temperature melted the interface, the jet pack of PHEV A was in an environment with flames, and the flame
flames were released through this opening. consumed the TR gas that was released from it. Therefore, the PHEV did

527
Y. Cui et al. Process Safety and Environmental Protection 166 (2022) 524–534

Fig. 5. Flame propagation between BEV A and PHEV A.

not undergo a gas-phase explosion. As presented in Fig. 6, a flowchart The space surrounding the BEV was divided into twenty-one areas
reviews the fire evolution of the two adjacent EVs. At the end of the with a thermocouple in the center of each area, and the thermocouple
experiment (111 min 48 s), professional firefighters extinguished the EV reading reflected the temperature evolution of the area. For ease of
fire. distinction, these areas were coded with numbers that corresponded to
the numbers of the thermocouples. Moreover, timings, when combus­
tibles were surrounded by hot smoke or a flame as well as when the
3.3. Evolution of temperature field of EV fire combustibles were ignited, were defined to quantitatively describe the
evolution of a car fire. When thermocouple readings exceeded 50 ℃ for
Fig. 7 records temperature curves of thermocouples T1~T21 from more than 3 s, it was assumed that the hot plume, including hot smoke
91 min 19 s to 120 min 29 s. The gray dashed lines in Fig. 7 mark the and flames, had enclosed the combustibles near the thermocouple, and
time at which BEV A released the smoke. At 101 min 13 s, influenced by the area where the heated combustibles were located was termed the
burning tires, the BEV fire achieved its maximum temperature at the heated zone. The temperature of the fire plume (Heskestad, 2016) and
chassis (T6) with a value of 1126 ◦ C. The temperature of the fire flame at the ambient temperature (30–40 ℃) determined the threshold. The
about 0.4 m from the roof of BEV A reached 1047.1 ◦ C. The maximum location of 50 ℃ is marked with orange dotted lines in Fig. 7. Fig. 8
temperature around the rear passenger compartment (T11) was presents the variation of the heated zones during the test. The ignition
1054.6 ℃, and the seat cushion, car mat, and interior panel were fire temperature was set to 300 ℃, taking into account that most solid ma­
sources. Thermocouple T16 obtained the maximum temperature of the terials have an ignition temperature above 300 ℃ (Ji and Cheng, 2018;
cockpit fire at 107 min 46 s, which was 965.8 ◦ C. The above values are Khan et al., 2016; Torero, 2016). Similarly, temperature curves needed
consistent with the temperature characteristics of ICEV and electric to have three consecutive values greater than or equal to 300 ℃. The
hybrid bus fires (SP Technical Research Institute of Sweden, 2016; region that met these conditions was referred to as the flaming zone.
Weisenpacher et al., 2016; Zhu et al., 2020; Krüger et al., 2016) since the After a zone entered the flaming state, combustibles in this area began
materials of seat sponge, mat, and tires are similar across models.

Fig. 6. Flowchart of significant fire-spreading events in the two-EV fire.

528
Y. Cui et al. Process Safety and Environmental Protection 166 (2022) 524–534

Fig. 7. Curves of temperature around the BEV.

Fig. 8. Evolution of the heated zone of the BEV.

burning. In Fig. 7, the location of 300 ◦ C is indicated by the crimson completed, and the yellow squares indicated that the evolution was in
dashed line. The propagation process of the flaming zone is shown in progress.
Fig. 9. Furthermore, Figs. 8 and 9 consider the spatial distribution of The development of the heated zones revealed the origin and tra­
these areas. The red squares were the areas where the evolution was jectory of the hot plume. The hot plume released from the burning parts

529
Y. Cui et al. Process Safety and Environmental Protection 166 (2022) 524–534

Fig. 9. Evolution of the flaming zone of the BEV.

first moved upwards, and then the expansion of the burning area
Table 2
increased its coverage. This process lasted 456 s before the twenty-one
Flame spread rates of PHEV fires.
areas were completely enveloped. The formula for calculating the
propagation speed of the hot plume is as follows: Vehicle Battery Production Number Propagation Flame
type capacity date of flaming duration (s) spread
v1 = n1 /Δt1 (1) (kWh) zones rate
(s− 1)
where n1 is the number of the heated zones; Δt1 is the duration (s) of SUV- 13 2021 14 268 0.0522
propagation of the hot plume. The propagation duration is the difference type
between the time of appearance of the last heated zone and the time of PHEV
Sedan- 13 2021 8 554 0.0144
appearance of the first heated zone.
type
The diffusion rate of the hot plume was 0.0439 s− 1, which reflected PHEV
the rate of change in the thermal field of the space where the burning
BEV was situated. But it did not consistently maintain this value. Most of
the intervals between occurrences of the heated zones were within 7 s. rate of the PHEV fires (Cui et al., 2022) following the same calculation
The intervals between heated zones 15, 16, 18, 7, 9, and 4 (listed in method. The sedan-type BEV fire evolved the fastest among these three
order of appearance) were in the range of 11–24 s. The wind may in­ vehicles, followed by the SUV-type PHEV fire, and finally by the
fluence the accumulation of heat in these regions, but the distance of sedan-type PHEV fire. Both the energy contained in the battery pack and
these zones from the battery pack is essential in determining these in­ the car structure affected the fire performance of the EVs, and the former
tervals. It took more time for the hot plume to cover the outlying areas. had more of an impact. The battery pack of BEV A, with a capacity of
Finally, at 99 min 54 s, the temperature in all zones reached over 50 ℃. 38.1 kWh, provided more heat for the development of the EV fire than
The propagation path of flame on the combustible surface was the battery pack of 13 kWh. Although the frequent jet flames from the
different from the motion trail of the hot plume. The first flaming zone battery pack dissipated partial heat of combustion to the surroundings,
appeared at 92 min 34 s after ignition. Flames expanded first in the the remaining energy was enough to accelerate the flame spread.
chassis and then moved to the sides of the BEV after 94 min 15 s. It took Flames from the chassis of BEV A were considered to have the ability
296 s from the appearance of the first flaming zone to that of the last to transfer heat to PHEV A. Thus, the propagation of flames from the
flaming zone. Eventually, a total of 20 areas entered the combustion BEV to the PHEV started after zone 14 transformed into a flaming zone.
state. The temperature of zone 20 was maintained at around 50 ℃, Fig. 10 shows the temperature data of T34–36 from 91 min 19 s to
which may be attributed to the lack of oxygen and combustibles in the 120 min 29 s. As presented in Fig. 10, the temperature of thermocouple
area. And flames near zone 20 were tilted toward the vehicle sides where T35 achieved 300 ℃ at 96 min 56 s. After 532 s, the temperature at the
there was sufficient oxygen. The speed of flame propagation among chassis of the PHEV hit 988 ℃. Moreover, the appearance of flames on
these flaming zones is quantified by ν2, which is defined as follows: the left side of PHEV A was 130 s apart from the moment when flaming
zone 14 appeared. Given that the two cars were 0.6 m apart, the fire
v2 = n2 /Δt2 (2) propagation rate can be expressed by the following equation:

where n2 is the number of the flaming zones; Δt2 is the duration (s) of the v3 = d/Δt3 (3)
flame propagation. The continuance of the process is the time gap be­
tween the occurrence of the last flaming zone and the occurrence of the where d is the lateral distance (m) between two adjacent vehicles; Δt3 is
first flaming zone. the interval (s) between the time when the flaming zone appears on the
The v2 of the BEV fire was 0.0676 s− 1. A comparison of this value side of the vehicle, which is the fire source, and the time when com­
with the flame propagation speed of PHEV fires was used to illustrate bustion occurs on the side of the other vehicle. The two sides mentioned
further the burning rate of the BEV. Table 2 provides the flame spread above are opposite.

530
Y. Cui et al. Process Safety and Environmental Protection 166 (2022) 524–534

rubber, panel, and carpet (Cui et al., 2022).

3.4. Assessment of heat hazards from EV fire

Both convective and radiant heat from flames can incapacitate oc­
cupants. Therefore, it is necessary to assess the thermal risk of human
exposure to EV fires. The profile for radiative heat flux from the left and
rear of the BEV is presented in Fig. 12. This graph shows the data of H1–5
from 91 min 19 s to 120 min 29 s. The maximum measurement of the
radiometers was obtained at 94 min 57 s and was 2.347 kW/m2,
measured at point H3. At the same time, flames with a temperature of
694.6 ℃ appeared on the left. Moreover, jet fires from the rear and
flames pouring out of the trunk affected the radiant heat fluxmeter H5.
Between 93 min 34 s and 96 min 32 s, there were multiple peaks on the
data curve of H5. The peak radiation heat flux at this point was
0.555 kW/m2. The combustion intensity of the BEV fire determined the
heat flux. However, as the thermal radiation source is difficult to gather
Fig. 10. Curves of temperature at 0.6 m from the BEV. in open space, the radiometers 1.5 m away from the vehicle were
slightly affected by the flame heat radiation, and the maximum flux of
Under these experimental conditions, v3 was 0.0046 m/s. In the radiative heat at this location was 0.434 kW/m2. After 96 min 32 s,
contrast, the flame propagation speed between two neighboring ICEVs the radiative heat from the BEV fire to each gauge was below 0.25 kW/
placed outside (Terziev, 2019) was 0.0030 m/s. The distance between m2.
the two ICEVs was 0.8 m, and their Δt3 was around 270 s. Moreover, the The tolerance limit of human skin to radiation heat is about 2.5 kW/
indoor vehicle fire spreads more slowly to adjacent vehicles than out­ m2, below which occupants can withstand exposure for at least a few
door car fires. The spreading speed of flame between two ICEVs (Park minutes (Purser and McAllister, 2016). Once the thermal radiation
et al., 2019) placed 0.5 m apart was 0.0013 m/s, and this process took equals or exceeds this level, the skin is burned within several seconds.
place in a spacious indoor space and cost about 380 s. Overall, the The thermal hazards are divided into multiple levels, including severe
spewing flames from the BEV, the closer proximity of the two EVs, and skin pain, second-degree burns, and third-degree burns. The time
the oxygen-rich environment accelerated the fire propagation. consumed by radiant heat to cause the above damage to the exposed skin
The temperature evolution in the compartment of BEV A is illus­ reflects the fire severity. To calculate the time (min) to the different
trated in Fig. 11. This graph shows the data of T37–40 from 91 min 19 s to endpoints, the following formula (Purser and McAllister, 2016; Digges
120 min 29 s. After 300 s of the smoke emission, the trunk’s tempera­ et al., 2008) was used:
ture reached 304.7 ◦ C, and its temperature eventually peaked at /
tIrad = r q1.33 (4)
1104.3 ℃. This measured value was the highest temperature in the
burning BEV cockpit, and the peak fire temperature in the compartment
where r is the threshold for thermal radiation exposure dose [(kW/m2)4/
of ICEVs was also around 1000 ℃. As flames gradually spread from the 3
min], 1.33–1.67 for severe skin pain, 4–12.2 for second-degree burns,
rear of the BEV to the front, the temperature around the seats began to
and 16.7 for third-degree burns; q is the radiative heat flux (kW/m2).
rise at about 300 s. Before the seats ignited, the vehicle interior was
The value of r in this assessment was taken from the bottom limit of
filled with black smoke. 338 s after the trunk ignited, the upholstery of
the threshold in consideration of the elderly and children, as well as
the car seats began burning, and their maximum flame temperature
people with mobility problems. In the case of the maximal radiant heat
amounted to 1015.1 ◦ C. There was a period of 293 s from the appear­
flux, the exposed skin of occupants or firefighters would feel pain after
ance of flames on the undercarriage until the trunk started burning. The
25.7 s at a distance of 1 m from the burning BEV. If they remained
fire-resistant car frame prevented direct upward propagation of the
trapped after feeling pain, they would be burned after 51.5 s. Moreover,
flames. Chassis flames entered the vehicle compartment through the gap
the radiant heat of a fire is directional, and the above thermal damages
between the trunk lid and the car frame and ignited the trunk’s sealing
describe the heat exposure of the skin at a specific location. These

Fig. 11. Curves of temperature in the compartment of BEV A. Fig. 12. Curves of heat flux around BEV A.

531
Y. Cui et al. Process Safety and Environmental Protection 166 (2022) 524–534

radiative heat hazards are worse in a closed environment with poor source, respectively.
ventilation. More radiant heat is released by soot and gases that collect Fig. 14a shows the results of FED for the thermal hazard of the BEV
in the confined space (Chen et al., 2018; Chatterjee et al., 2011). fire to a group of people 1 m away from it. 50% of the victims were
The fire flame temperature at 0.5–1 m from BEV A is exhibited in incapacitated after 4.93 min of exposure. This period is long enough to
Fig. 13. This graph shows the data of T22–33 from 91 min 19 s to 120 min allow most of the trapped persons to be evacuated, but vulnerable per­
29 s. The BEV fire affected the temperature of the space on its left side sons may be injured or immobilized, especially if they are subjected to
after approximately 93 min 23 s. And the hot plume entering this area the direct impact of flames. When the flame temperature peaked, the
came from the undercarriage flame and the left rear tire fire. Due to the FED grew at the fastest rate. The peaks marked by points A and B
proximity to these fire sources, the maximum flame temperature at accelerated the deterioration of the thermal environment of the fire.
0.5 m on the left was 737.6 ◦ C. In comparison, the temperature at the They corresponded to the sudden increase in radiant heat at 94 min 57 s
left 1 m was below 230 ℃. Furthermore, thermocouples T22, T24, and and the rapid rise in convective heat at 96 min 41 s, respectively. The
T26 at the same level as the car window were more susceptible to flames increment of thermal convection also influenced the value at point C. In
than those (T23, T25, and T27) at the height of 0.15 m. The flames fluc­ addition, when 0.5 m from the BEV, the thermal situation was worse
tuated during the fire. As the flame gradually moved towards the engine because the convective heat of the flame decreased with distance
compartment and the combustion at the rear was reduced, the temper­ (Fig. 14b). The time required to immobilize half the population was
ature in the left area tended to decrease after 97 min 33 s. reduced to 2.28 min. And most individuals may be at risk of burns. The
Serious injury caused by convective heat derives from the direct hazard is even conservative since the tIrad used the output from radiant
contact between hot smoke or flame and skin. For individuals exposed to heat fluxmeter H3. Eventually, the FED attained its maximal growth rate
thermal convection, the time (min) to incapacitation is calculated by: at point D.

tIconv = 5 × 107 T-3.4 (5)


4. Fire safety requirements for EVs
where T is the temperature (℃) of flames and hot smoke in the fire
Existing studies concentrate on improving the electrochemical per­
environment (ISO 13571, 2012; Peng et al., 2020).
formance and thermal resistance of LIBs under abuse modes but lack
Under the impact of the fire flame of 230 ℃, humans were inca­
knowledge of fire safety optimization of EVs and the onboard LIB energy
pacitated after standing 1 m from the burning BEV for 28 s. Unprotected
systems. In the future, their fire performance can be improved in the
skin can only tolerate convective heat below 120 ℃, otherwise people
product development phase from the following aspects.
will feel considerable pain or even be burned (Purser and McAllister,
The gas production of LIBs with TR should be controlled to eliminate
2016). 150 ℃ was the temperature that the BEV fire maintained at a
the gas phase explosion of EVs. Under a certain capacity, the type and
distance of 1 m for a long time, making personnel incapable of escaping
formula of the positive electrode, negative electrode, and electrolyte in a
after 119.8 s. The distance was shortened from 1 m to 0.5 m, and the
LIB determine the direction and process of the TR reactions (Zhang et al.,
thermal threat increased exponentially. The body was only able to
2020; Yang et al., 2020; Feng et al., 2015). Thus, choosing lithium salts
maintain its functions for 0.53 s in an environment with a heat flow of
and electrolytes that produce less flammable smoke is advisable. The
737.6 ℃. At 0.5 m, the temperature of the hot plume was above 350 ℃,
avoidance of gas accumulation is also warranted based on the explosion
leaving only 6.7 s of escape time for trapped people. If the radiant heat is
limit. Ameliorating the design of the safety valve of the battery pack is a
taken into account, the safe time is further limited. Purser’s fractional
potentially effective measure to disperse explosive gases, such as the
effective dose (FED) model (ISO 13571, 2012; Purser, 2012) focuses on
activation threshold, number, and location of the safety valve(s).
evaluating the combined thermal hazard of convective and radiant heat,
Furthermore, traction battery packs should be tested for explosion risks
as shown in Eq. (6). The FED model expresses the dose of heat received
before they are put into service to ensure the holistic fire safety of EVs.
by a fire victim when exposed to a fire scenario for a spell, and the heat
The length and duration of the jet fire from the EVs should be
incapacitates 50% of the trapped population when the FED exceeds 1.
shortened. The TR propagation tends to induce the flame injection at the
∫ t2 ( )
1 1 pack level. Although there are some patterns in the behavior of the jet
FED = + Δt (6)
t1 tIrad tIconv fire, it speeds up the spread of the EV fire to surrounding combustibles.
Its intensity can be weakened by using non-flammable or flame-
where t1 and t2 are the start time and end time of exposure to the heat retardant materials to produce LIBs. Moreover, restricting the jet

Fig. 13. Temperature variation curves of the thermocouples on the left side of BEV A.

532
Y. Cui et al. Process Safety and Environmental Protection 166 (2022) 524–534

Fig. 14. Evolution of FED of thermal hazards from BEV fire between 91 min 19 s and 120 min 29 s.

direction of flames by adjusting the structure of the battery pack and the burning BEV safely. Under the coupling effect of the thermal radiation
car body is another option. In some fire scenarios, the downward jet fire and convection of the fire, a high percentage of the victims were inca­
of EVs may be acceptable. Nevertheless, this solution imposes additional pacitated in minutes.
demands on the fire resistance of the concrete floor of parking buildings This work reveals the fire dynamics of parallel parked modern EVs.
and the steel deck of roll-on/roll-off ships. The results are instructive for the assessment of EV fire risk, the design of
EVs, especially large battery-powered buses, are expected to have fire detection systems in tunnels and underground car parks, the pro­
fire detection and alarm systems to protect people inside or outside the tection of firefighters and occupants from the thermal hazard, and the
vehicles. Modern and well-equipped battery packs are installed with fire safety optimization of automotive LIB energy systems.
built-in temperature, current, and voltage monitoring systems. Howev­
er, these devices may not provide advance warning signals and lack Declaration of Competing Interest
appropriate feedback to passengers. When the battery pack is over­
charged, the temperature sensors on the battery surface cannot instantly The authors declare that they have no known competing financial
report the thermal state in the battery because TR is generated insidi­ interests or personal relationships that could have appeared to influence
ously. Fluctuations in battery voltage and current are also difficult to the work reported in this paper.
identify under overheating abuse conditions. The battery pack releases
smoke for a while before flames appear, so installing smoke detectors in Acknowledgments
the battery pack or on the vehicle frame may effectively prolong the
evacuation time. After detecting a fire, the vehicle alarm equipment The authors appreciate the support of this work by Project
warns the driver or nearby people to prevent casualties. 20dz1200900 supported by the Science and Technology Commission of
Shanghai Municipality and Project 2021YBR015 supported by the Top-
5. Conclusions notch Innovative Talent Training Program for Graduate Students of
Shanghai Maritime University.
A full-scale fire experiment on two parallel placed modern electric
vehicles (EVs) was conducted to explore their burning behaviors, fire
References
spread process, and thermal characteristics and hazards.
White smoke released from the vehicle chassis was a precursor to the Terziev, A., 2019. Study of the fire dynamics in a burning car and analysis of the
battery electric vehicle (BEV) fire when the traction lithium-ion battery possibilities for transfer of fire to a nearby vehicle. E3S Web Conf. 112, 01015.
(LIB) pack encountered external heat sources. Flames appeared only Bisschop, R., Willstrand, O., Rosengren, M., 2020. Handling lithium-ion batteries in
electric vehicles: preventing and recovering from hazardous events. Fire Technol. 56,
after the white smoke accumulated and exploded. Buoyant diffusion 2671–2694.
flames dominated the BEV fire, and multiple appearances of jet fires Boe, A.S., 2017. Full scale electric vehicle fire test. (https://www.fireproductsearch.
from the thermal runaway (TR) battery pack made it more dangerous. com/full-scale-electric-vehicle-fire-test/).
Brzezinska, D., Bryant, P., 2022. Performance-based analysis in evaluation of safety in
The total capacity of the LIBs and the position of the master safety valve car parks under electric vehicle fire conditions. Energies 15, 649.
affected the duration, length, and injection direction of the jet flames on Chatterjee, P., de Ris, J.L., Wang, Y., et al., 2011. A model for soot radiation in buoyant
the vehicle scale. The peak temperatures of the external and internal diffusion flames. Proc. Combust. Inst. 33, 2665–2671.
Chen, M., Ouyang, D., Weng, J., et al., 2019. Environmental pressure effects on thermal
flames of the BEV compartment are consistent with that of internal runaway and fire behaviors of lithium-ion battery with different cathodes and state
combustion engine vehicles (ICEVs) since the two models were similar of charge. Process Saf. Environ. Prot. 130, 250–256.
with respect to the material type and quantity of the combustible Chen, T.B.Y., Yuen, A.C.Y., Wang, C., et al., 2018. Predicting the fire spread rate of a
sloped pine needle board utilizing pyrolysis modelling with detailed gas-phase
polymers. The presence of the traction battery pack and the absence of combustion. Int. J. Heat Mass Transf. 125, 310–322.
the automotive fuel tank did not influence the maximum temperature Cui, Y., Liu, J., 2021. Research progress of water mist fire extinguishing technology and
characteristics of the vehicle fire. Methods were established to quanti­ its application in battery fires. Process Saf. Environ. Prot. 149, 559–574.
Cui, Y., Cong, B., Liu, J., et al., 2022. Characteristics and hazards of plug-in hybrid
tatively describe the average spread rate of the hot plume around the
electric vehicle fires caused by lithium-ion battery packs with thermal runaway.
BEV and the flame propagation rate between the BEV and plug-in hybrid Front. Energy Res. 10, 878035.
electric vehicle (PHEV). The flames of the two-EV fire spread faster than Diaz, L.B., He, X., Hu, Z., et al., 2020. Meta-review of fire safety of lithium-ion batteries:
that of PHEV or ICEV fires, which was attributed to the spewing flames Industry challenges and research contributions. J. Electrochem Soc. 167, 090559.
Digges, K.H., Gann, R.G., Grayson, S.J., et al., 2008. Human survivability in motor
from the BEV, the closer proximity of the two EVs, and the oxygen-rich vehicle fires. Fire Mater. 32, 249–258.
environment. Occupants had extremely limited time to escape from the Dorsz, A., Lewandowski, M., 2022. Analysis of fire hazards associated with the operation
of electric vehicles in enclosed structures. Energies 15, 11.

533
Y. Cui et al. Process Safety and Environmental Protection 166 (2022) 524–534

Feng, X., Sun, J., Ouyang, M., et al., 2015. Characterization of penetration induced Peng, Y., Yang, L., Ju, X., et al., 2020. A comprehensive investigation on the thermal and
thermal runaway propagation process within a large format lithium ion battery toxic hazards of large format lithium-ion batteries with LiFePO4 cathode. J. Hazard.
module. J. Power Sources 275, 261–273. Mater. 381, 120916.
Feng, X., Ren, D., He, X., et al., 2020. Mitigating thermal runaway of lithium-ion Ping, P., Wang, Q., Huang, P., et al., 2015. Study of the fire behavior of high-energy
batteries. Joule 4, 743–770. lithium-ion batteries with full-scale burning test. J. Power Sources 285, 80–89.
Heskestad, G., 2016. Fire plumes, flame height, and air entrainment. In: Hurley, M.J., Purser, D., 2012. Validation of additive models for lethal toxicity of fire effluent
Gottuk, D., Hall, J.R., et al. (Eds.), SFPE Handbook of Fire Protection Engineering. mixtures. Polym. Degrad. Stab. 97, 2552–2561.
Springer, New York, pp. 396–428. Purser, D.A., McAllister, J.L., 2016. Assessment of hazards to occupants from smoke,
ISO 13571, 2012. Life-threatening components of fire – guidelines for the estimation of toxic gases, and heat. In: Hurley, M.J., Gottuk, D., Hall, J.R., et al. (Eds.), SFPE
time to compromised tenability in fires. Handbook of Fire Protection Engineering. Springer, New York, pp. 2308–2428.
Ji, J., Cheng, Y., 2018. Fire dynamics. China University of Mining and Technology Press, Qiu, M., Cui, Y., Niu, S., et al., 2022. Experimental study on the effectiveness of water
Xuzhou. spray systems in suppressing electric vehicle fires. Fire Sci. Technol. 41, 82–86.
Jiang, Z.Y., Qu, Z.G., Zhang, J.F., et al., 2020. Rapid prediction method for thermal SP Technical Research Institute of Sweden, 2016. Full scale fire-test of an electric hybrid
runaway propagation in battery pack based on lumped thermal resistance network bus. Sweden.
and electric circuit analogy. Appl. Energy 268, 115007. Sun, P., Huang, X., Bisschop, R., et al., 2020. A review of battery fires in electric vehicles.
Khan, M.M., Tewarson, A., Chaos, M., 2016. Combustion characteristics of materials and Fire Technol. 56, 1361–1410.
generation of fire products. In: Hurley, M.J., Gottuk, D., Hall, J.R., et al. (Eds.), SFPE Torero, J., 2016. Flaming ignition of solid fuels. In: Hurley, M.J., Gottuk, D., Hall, J.R.,
Handbook of Fire Protection Engineering. Springer, New York, pp. 1143–1232. et al. (Eds.), SFPE Handbook of Fire Protection Engineering. Springer, New York,
Krüger, S., Hofmann, A., Berger, A., et al., 2016. Investigation of smoke gases and pp. 633–661.
temperatures during car fire - large-scale and small-scale tests and numerical United Nations Economic Commission for Europe, 2018. Statistics and analysis on fire
investigations. Fire Mater. 40, 785–799. accidents for EVs-China. UNECE.
Lesiak, P., Pietrzela, D., Mortka, P., 2021. Methods used to extinguish fires in electric Wang, Y., Huang, C., 2020. Thermal explosion energy evaluated by thermokinetic
vehicles. Saf. Fire Technol. 58, 38–57. analysis for series- and parallel-circuit NMC lithium battery modules. Process Saf.
Leuthner, S., 2013. Übersicht zu lithium-ionen-batterien. In: Korthauer, R. (Ed.), Environ. Prot. 142, 295–307.
Handbuch Lithium-Ionen-Batterien. Heidelberg. Springer. Weisenpacher, P., Glasa, J., Halada, L., 2016. Automobile interior fire and its spread to
Lisbona, D., Snee, T., 2011. A review of hazards associated with primary lithium and an adjacent vehicle: parallel simulation. J. Fire Sci. 34, 305–322.
lithium-ion batteries. Process Saf. Environ. Prot. 89, 434–442. Yang, X., Duan, Y., Feng, X., et al., 2020. An experimental study on preventing thermal
Mao, B., Chen, H., Jiang, L., et al., 2020. Refined study on lithium ion battery combustion runaway propagation in lithium-ion battery module using aerogel and liquid cooling
in open space and a combustion chamber. Process Saf. Environ. Prot. 139, 133–146. plate together. Fire Technol. 56, 2579–2602.
Mao, B., Zhao, C., Chen, H., et al., 2021. Experimental and modeling analysis of jet flow Yuan, L., Dubaniewicz, T., Zlochower, I., et al., 2020. Experimental study on thermal
and fire dynamics of 18650-type lithium-ion battery. Appl. Energy 281, 116054. runaway and vented gases of lithium-ion cells. Process Saf. Environ. Prot. 144,
National Transportation Safety Board, 2021. Fire aboard roll-on/roll-off vehicle carrier 186–192.
Höegh Xiamen. USA. Zhang, L., Li, Y., Duan, Q., et al., 2020. Experimental study on the synergistic effect of gas
Ouyang, D., Weng, J., Chen, M., et al., 2022. What a role does the safety vent play in the extinguishing agents and water mist on suppressing lithium-ion battery fires.
safety of 18650-size lithium-ion batteries? Process Saf. Environ. Prot. 159, 433–441. J. Energy Storage 32, 101801.
Park, Y., Ryu, J., Ryou, H.S., 2019. Experimental study on the fire-spreading Zhou, Z., Zhou, X., Wang, D., et al., 2021. Experimental analysis of lengthwise/
characteristics and heat release rates of burning vehicles using a large-scale transversal thermal characteristics and jet flow of large-format prismatic lithium-ion
calorimeter. Energies 12, 1465. battery. Appl. Therm. Eng. 195, 117244.
Zhu, H., Gao, Y., Guo, H., 2020. Experimental investigation of burning behavior of a
running vehicle. Case Stud. Therm. Eng. 22, 100795.

534

You might also like