Case Studies in Thermal Engineering 38 (2022) 102342

Contents lists available at ScienceDirect

Case Studies in Thermal Engineering

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

Analyzing the reflash feature of large-area liquid fire in

channel-like structures: A case study
Jiaming Zhao , Zhisheng Xu , Lei Fang , Yulun Zhang *, Chengbo Yu
School of Civil Engineering, Central South University, Changsha, 410075, PR China

H I G H L I G H T S

• A study was conducted to explore the reflash of channel-like structure fires.

• Tunnel fire reflash is mainly affected by temperature level and fuel vapor amount.
• Improper portal-opening moment will induce the reflash in tunnel fire scenarios.
• In this work, 280 oC ceiling temperature and 0.35 m2 pool area are one of the necessary conditions for redraft.
• Currently, maximum radiant heat flux near portals induced by reflash reaches 505.5 W/m2.

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

Keywords: Several fire tests were performed in a 1/7 model tunnel to explore the potentially reflash issue
Large-area liquid fire after blocking tunnel fire, and the temperature right above fire source, temperature distribution
Blocking inside tunnel and radiant heat flux near tunnel portal were acquired. Results show that the
Tunnel-like structure occurrence conditions of tunnel fire reflash is mainly related to the amount of combustible vapor
Reflash accumulated in the internal space and temperature level near fire source. Improper portal-
opening moment will induce the reflash in such channel-like structure fire scenarios. In current
experiment, when the both portals were suddenly opened, the ceiling temperature right above fire
source of not less than 280 oC and liquid pool area of not less than 0.35 m2 are one of the
necessary conditions for the reflash occurrence. For the cases with large liquid fuel pool, the
obvious thermal plume and flame will overflow out portals when reflash occurred. Within the
scope of current test, maximum radiant heat flux near portals induced by reflash reaches 505.5
W/m2. This work can provide some reference for the understanding of reflash feature of large-
area fire and the corresponding emergency treatment of inside channel-like structure.

Abbreviations and nomenclatures

Af Surface area of isothermal flame plate [m2]

Ff→1 Geometrical view factor
ΔH The heat of combustion [kJ/kg]
HGV Heavy Goods Vehicles
ṁ Time-averaged mass loss rate [kg/s]
MLR Mass loss rate of liquid fuel [kg/s]

* Corresponding author. School of Civil Engineering, Central South University, Changsha, Hunan, 410075, PR China.
E-mail address: [email protected] (Y. Zhang).

https://doi.org/10.1016/j.csite.2022.102342
Received 21 April 2022; Received in revised form 15 July 2022; Accepted 2 August 2022
Available online 7 August 2022
2214-157X/© 2022 The Author(s). 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/).
J. Zhao et al. Case Studies in Thermal Engineering 38 (2022) 102342

Q̇ Heat release rate [kW]

q̇′′ flame,rad Radiative heat flux obtained at target location from flame [W/m2]
r Distance between isothermal flame plate and target object [m]
S Surface area of target object [m2]
Tf Flame temperature [K]
t Burning duration time [s]

Greek symbol
ε Flame emissivity
σ Stephen-Boltzmann constant [W/m2/K4]
θ1 θ2 Plane angle [◦ ]
χ Combustion efficiency

1. Introduction
The reflash phenomenon can be triggered in fire situations where the ventilation is limited and there is a fuel-rich region in a
confined space [1], which has typical nonlinear characteristics. Tunnel, as a typical long-narrow structure, has the large length-width
ratio and length-height ratio, which is obviously different from the structural characteristics of general compartment buildings [2].
Due to this special structural characteristics, the ventilation conditions in the channel-like structure are relatively poor, under the
premise of the failure of forced ventilation facilities. Furthermore, when a large area fire occurs in tunnel structure, blocking the portal
is one of the alternative treatment methods, which can effectively limit the inflow of fresh air, and then achieve the purpose of
extinguishing the fire. In this scenario, the ventilation conditions inside the tunnel are further deteriorated [3–5]. If the opening of
tunnel portal is not properly handled, there is a large amount of flammable vapor and high temperature in the internal space of tunnel.
That is a possibility of triggering the reflash of a large area fire, which is a great security threat to firefighters performing the fire­
fighting and rescue tasks [6,7]. Meanwhile, on the premise of ensuring fire safety, opening the tunnel to resume traffic as early as
possible can reduce indirect economic losses as much as possible. However, there are few researches on the potential reflash of
large-area fire in long-narrow channel-like structures.
At the present stage, the reflash problem involved in the fire of compartment buildings has been studied by many scholars with
focus attention [8–19]. Weng et al. [8,9] studied reflash of compartment fire under ventilation limitation by means of small-scale
experiments and theoretical analysis. It is found that reflash occurs when unburned fuel vapor reaches a critical value. Based on
the nonlinear dynamics theory, a mathematical model is established to describe the reflash phenomenon of compartment fire. Horvat
et al. [10] used numerical simulation method to simulate and reproduce the reflash phenomenon in the compartment, and found that
the reflash phenomenon was induced at the moment when fresh air entered the compartment with the form of gravity flow. Ferraris
et al. [11] also studied reflash by means of numerical simulation. The concept of "flame index" based on fuel and oxygen gradients was
proposed, and the backflash phenomenon is classified as different characteristic stages. Wu et al. [12] studied reflash in compartment
fire by means of model-scale experiment. The influencing mechanisms of gas leakage flow rate, ventilation conditions and ignition
position were identified. Chen et al. [13], based on the results of small-scale fire tests and combined heat transfer analysis and fire
dynamics analysis, established a theoretical model to describe the occurrence conditions of backflash. Tsai et al. [14] conducted a
full-scale room fire tests to study reflash characteristics. It is found that one reflash can cause two temperature peaks. Ashok et al. [15]
used numerical simulation to simulate the reflash process in different gravity environments, and the influence mechanism of gravity on
reflash of compartment fire was explored. Król et al. [16] numerically studied the reflash phenomenon in the process of fire devel­
opment through computational fluid dynamics method. The characteristic phases of the phenomenon and the occurrence of the gravity
current were revealed. Wu et al. [17] conducted a series of model-scaled compartment fire tests to study the critical factors governing
the relash onset, using polypropylene and high density polyethylene samples as fuel. Zhao et al. [18] carried out a series of small-scaled
fire tests using woodblocks as fuel to study the reflash issues. The effect mechanism of compartment window woodblock areas on
reflash features were revealed and identified. Myilsamy et al. [19] used the large eddy simulation with mixing-controlled fast
chemistry combustion model to study reflash behavior in the reduced-scale compartments. The reasonable results for reflash features
and gravity current were provided by numerical simulation.
Although previous scholars have conducted a lot of interesting studies on the reflash issue in compartment-like buildings fire, there
are few reports on the potential reflash problem in the large-area liquid fire of narrow-long channel-like structures [13,20]. Physically,
due to the large length-width ratio and length-height ratio of tunnel structure and the frequent occurrence of large-area fuel it is
possible that the reflash accidents could be triggered if the ventilation conditions near tunnel portal are not properly controlled.
However, the further identification for the occurrence conditions of reflash and thermal characteristics induced and driven by reflash is
particularly rare in such scenario.
Considering this, in this work, a special study was performed in a reduced-scale model tunnel (1/7) to further explore the potential
reflash issue of large-area fires in tunnel-like structures. The ceiling temperature, temperature distribution inside the tunnel, radiant
heat flux near tunnel portal are measured and identified, and the reflash features of the large-area fuel fire inside the tunnel are
explored. The study could give some experimental information on when to open the blocked tunnel portals and the caused reflash
characteristics, which is expected to provide some reference for the understanding of reflash feature of large-area fire and the

2
J. Zhao et al. Case Studies in Thermal Engineering 38 (2022) 102342

corresponding emergency treatment of inside channel-like structure.

2. Experimental configurations
The experimental tests are carried out in a 1/7 model tunnel (7 m (length) × 1.0 m (width) × 0.8 m (height)), as shown in Fig. 1.
The main material used for channel walls and sealing portal is 3 cm thick fire-proof board (heat capacity: 1130 J/kg K, density: 870 kg/
m3 and heat conduction: 0.1736 W/m/K) with white color. Methanol pool fire is used as fire source and the dimension of single
rectangular steel pan is 0.7 m × 0.25 m × 0.10 m. Fuel pans are located at the central position of tunnel. In fire test, each fuel pan was
injected with 3.8 kg methanol, which ensured that fuel combustion could reach the quasi-steady state and maintain for a certain
period, without the influence of sealing operation. Fuel mass loss rate is measured instantaneously by an electronic balance with 0.1g
accuracy. The response time is 1 s (i.e., to obtain a data within 1s). The temperature distribution inside the tunnel is measured with a
total of 95 temperature sensors (the diameter of K-type sheathed thermocouple is 1.0 mm, and the response time is also 1 s), which the
error tolerance of ±1.7%*T (T is measured value of gas temperature). The location of specific measurement points is shown in Fig. 2.
Thereinto, considering the higher temperature and temperature gradient near the fire source region and tunnel ceiling, a certain
encryption arrangement was hence carried out. Furthermore, it should be pointed out that the ceiling temperature was obtained by the
measuring point 5 cm below celling. Usually, due to the heat transfer from fire smoke to cold ceiling, the location of maximum ceiling
temperature was not tightly close to tuunel ceiling surface, but at a certain distance, which is related to specific experimental con­
ditions, especially the thermal characteristics of ceiling materials. In order to estimate the location of maximum ceiling temperature
under current experimental condition, a set of preliminary tests were conducted in this 1/7 model tunnel. Finally, combined with the
overall layout of points, the location were determined to be installed 5 cm below the ceiling surface under current test. In addition,
three radiation heat flow meters (GTT-25 series Gardon heat flow sensor) are enabled to measure the radiant heat outside tunnel
portal, which are mounted 1.0 m away from the tunnel portal with a 0.3 m vertical spacing. The absorptivity of radiation sensor is 0.92
and the measurement error is about 3.2%.
During the implementing process of blocking strategy, both tunnel portals were completely blocked by fire-proof asbestos board.
The execution time of blocking is judged by whether the internal burning of fuel pool reaches quasi steady state. Based on the judgment
of the pre-experiment, the internal burning approaches quasi-steady state about 5 min after ignition, so the blocking time in this work
is set as t = 5 min. Similarly, the open-blocking time of tunnel portal is set as t = 30 min. Experimental program is summarized in
Table 1. Under the experimental conditions, the area (or heat release rate) of liquid fire was chosen as the primary variable. On the one
hand, the emergence of reflash is closely related to heat release rate and ventilation conditions. The heat release rate of pool fire and
relative ventilation capacity inside tunnel can be adjusted by changing the area of fire source under the condition that the cross section
size of both tunnel portals is fixed, which is beneficial to the current study of reflash features of large pool fire inside the channel-like
structure. On the other hand, in real tunnel fire accidents, fire sources are often different types of vehicles, which usually correspond to
different fire source area and heat release rate. Hence, in current experimental work, the three area of fire sources (0.175 m2, 0.35 m2,
0.525 m2) was considered, and the heat release rate in quasi-stable burning stage was 66.1 kW, 202.3 kW and 397.6 kW, respectively.
Based on Froude’s scale law, the heat release rates of full-size configuration corresponding to the burning size in the current scaled
model tunnel were 8.6 MW, 26.2 MW and 51.5 MW, which respectively correspond to the multiple-passenger-car fire, HGV fire and
tanker fire in the real scenario [21]. In addition, the scaling relationship of temperature parameters is TM = TF . The subscript M and F
represent the model-scaled and full-scaled parameters, respectively. The current similarity ratio LM /LF is 1/7. Through these scaling
relations, the fire experiment results of current model-scaled test can be extrapolated to the real full-scale fire scenario. Specific

Fig. 1. Sectional layout of the model tunnel system (Unit: m).

3
J. Zhao et al. Case Studies in Thermal Engineering 38 (2022) 102342

Fig. 2. Specific measuring points of temperature and radiant heat.

Table 1
Experimental procedures and operating steps.

Experimental steps Execute operations

Step 1 Adjust the number and location of liquid pool

Step 2 Activate the experimental measurement equipment in advance
Step 3 At t = 0 min, start the experiment by igniting the liquid pools
Step 4 At t = 5 min, use fireproof boards to completely block both portals
Step 5 At t = 30 min, open portals at both ends of tunnel
Step 6 When flame dies out naturally, save data, end this case, and prepare for next case

experimental cases and parameters are summarized in Table 2, and all experimental cases are carried out under natural ventilation. It
should be noted that the given heat release rate refers to the quasi-stable experimental value before the portal blocking is performed.
Experimental heat release rate was obtained from Eq. (1):

Q̇ = χ ṁΔH (1)

where ṁ is time-averaged mass lost rate of fuel pan in relatively steady stage of burning (g/s), which is obtained from experimental
measurements; χ is the combustion efficiency, for clean liquid fuel such as methanol, χ = 1 is acceptable value [22]; ΔH is the heat of
combustion (kJ/kg).

3. Results and discussion

3.1. Temperature curve above fire source
Fig. 3 shows the variation trend of temperature right above the center location of fire source under different fuel pool areas. It can
be seen that after 5 min of ignition, the liquid fuel burning in the tunnel is fully developed and the maximum ceiling temperature is
between 800 and 1000 oC. When t = 5 min, the portals at both tunnel ends are completely closed. For the case with 0.175 m2 fuel pool,
after the portal is completely blocked, the temperature above the fire source does not immediately decay, but still maintains a high
temperature for a period of time. This can be attributed to the fact that the current case with 0.175 m2 fuel pool area requires a
relatively small amount of air for internal burning, compared with the other two cases with the larger fuel pool area. When the tunnel
portal is completely blocked, the amount of air remaining inside the tunnel still supports burning for a period of time (for 0.175 m2 fuel
pool), although the outside air cannot flow into the tunnel space. Subsequently, as the amount of air inside the tunnel is consumed, the
fuel pool burning rapidly decays due to lack of oxygen until the flame goes out. For the cases with 0.35 m2 and 0.525 m2 pool area, the
amount of air required for burning is relatively large. After the portals at both tunnel ends are completely blocked, the internal burning
begins to decay immediately, and the temperature above the fire source decreases rapidly. It should be noted that although the
temperature inside the tunnel decreases gradually due to the attenuation of burning intensity of fire source, the temperature above the
fire source inside the tunnel is still relatively high compared with the boiling point temperature of liquid fuel. In other words, the liquid
fuel inside the tunnel is still rapidly vaporizing, and the concentration of fuel vapor in the tunnel gradually increases. When the portals
at both tunnel ends are opened at t = 30 min, a large amount of thermal smoke gathered inside the tunnel rapidly flows out, while the
fresh air outside flows into the tunnel rapidly and mixes with a large amount of fuel vapor gathered above the fire source. At this time,

Table 2
Experimental cases and parameters.

Case no. Fire source area (m2) a

Heat release rate (kW) Sealing time (min) Opening time (min)

A-1 0.175 66.1 5 30

A-2 0.35 202.3 5 30
A-3 0.525 397.6 5 30
a
The Heat release rate is average value acquired based on the measured mass loss rate and Eq. (1).

4
J. Zhao et al. Case Studies in Thermal Engineering 38 (2022) 102342

Fig. 3. Temperature variation trend right above the fire source under different fuel pool area.

Fig. 4. Temperature distribution contourin tunnel under different fuel pool area.

5
J. Zhao et al. Case Studies in Thermal Engineering 38 (2022) 102342

if the temperature above the fire source is relatively high (compared to the methanol ignition point), the tunnel fire reflash will occur.
It can be seen from the current data that the temperature variation trend above the fire source in the cases of 0.35 m2 and 0.525 m2 fuel
pool area increases rapidly after t = 30 min, indicating that the flame appears again above the fire source, that is, reflash phenomenon.
In addition, the temperature data measured by Chen et al. work [13] are also shown in Fig. 3. It should be pointed out that the fire
source of Chen et al. used in their work [13] is wood fire, which is different from the current liquid pool fire. It can be seen that in the
initial rising stage, the temperature of liquid pool fire rises up rapidly, while that of wood fire is in the smoldering state, and its heating
process is slow. Meanwhile, for the burning of liquid pool fire with the reflash occurrence, the temperature variation on the whole
shows a rapid increase and then attenuation, and then accompanied by reflash, the temperature increases rapidly again, and finally
decreases with the fuel consumption. However, for solid fuel fires, the temperature increase is slow due to the existence of early
smoldering stage [13]. After the reflash, the temperature increases rapidly, again.
Fig. 4 shows the variation trend of temperature inside the tunnel at typical moment under different fuel pool areas. It can be seen
that as the area of fuel pool increases, the area of high-temperature region inside the tunnel increases significantly (before blocking the
portal). After the portals at both tunnel ends are completely blocked (t = 5 min), the tunnel temperature decreases rapidly and the area
of high-temperature region decreases significantly (t = 15 min). The complete closure of tunnel portals will prevent fresh air from
flowing into the tunnel to support the burning of liquid pool fire. The burning maintenance of liquid fuel depends only on the residual
air inside the tunnel. When the remaining air inside tunnel is completely consumed, the flame inside the tunnel gradually disappears.
Based on experimental results, the residual air inside the tunnel could not support the burning of liquid fuel for more than 10 min in all
cases, resulting in almost no flame inside the tunnel at t = 15 min (the blocking operation was performed at t = 5 min). When t = 30
min, the portals at both tunnel ends are opened, and the fresh air from the outside rapidly pours into the internal space of tunnel and
mixes with a large amount of fuel vapor gathered inside the tunnel. In the cases of 0.35 m2 and 0.525 m2 fuel pool area, the tem­
perature above the fire source is still relatively high when the portal is opened (t = 30 min), compared with the ignition point of liquid
fuel. After the tunnel portal was opened, the obvious high temperature region appeared again above the fire source. In current study,
the temperature data is used to determine whether reflash occurs, which the temperature standard is selected as 650 oC. It can be
clearly seen from Fig. 4(b) and (c) that obvious reflash occurred in these two cases, while the case shown in Fig. 4(a) did not occur. In
addition, it can also be found that the reflash intensity in the case with 0.525 m2 liquid pool area is significantly stronger than that of
the case with 0.35 m2 liquid pool area. This is mainly because when the liquid pool area is 0.525 m2, after the tunnel portals at both
ends are blocked, more combustible vapor is generated and accumulated inside the tunnel space under the action of high temperature.
At the moment that tunnel portals are opened, a stronger reflash occurred under the condition of high temperature and high con­
centration of fuel vapor. In addition, it should be pionted out that in curent work, the layer burning phenomenon has not been clearly
observed. Zhang et al. [23] carried out a series of experimental tests to investigate the layer burning phenomenon. This special fire
behavior is mainly the result of ignition of excess fuel gases at the interface between the upper hot gas and lower air layers. The absence
of layer burning phenomenon in this work is caused by the comprehensive influence of the differences in geometric structure, the
configuration characteristics of fire source, the obstructed state of openings.

3.2. The occurrence of backflash

Fig. 5 shows the variation trend of ceiling temperature right above the fire source under different fuel pool areas. It can be seen that
the ceiling temperature increases gradually with the increase of fuel pool area, and the maximum ceiling temperature is 683 oC, 850.3
o
C and 930 oC for the case with 0.175 m2, 0.35 m2 and 0.525 m2, respectively. As the tunnel portal is closed at t = 5 min, the fresh air
cannot be sucked into the internal space of tunnel in time to support the burning of liquid fuel pool, and the ceiling temperature
gradually decreases. However, when t = 30 min, the portals at both tunnel ends are opened, and the reflash occurs in both cases with
0.35 m2 and 0.525 m2 fuel pool area. The ceiling temperature of tunnel increases rapidly, and the maximum temperature is 588.5 oC
and 783 oC, respectively. In addition, based on this, it can be seen that within the current research scope, at the moment of opening the

Fig. 5. Ceiling temperature level at the center location of tunnel when the reflash can occurred.

6
J. Zhao et al. Case Studies in Thermal Engineering 38 (2022) 102342

both tunnel portals, the tunnel fire reflash can occur only when the ceiling temperature directly above the fire source reaches more
than 280 oC and the fuel pool area is not less than 0.35 m2. When these condition is not met (for example, the case with the 0.18 m2 fuel
pool area), the tunnel fire reflash does not occur. The possible reasons are analyzed that, on the one hand, the case with 0.18 m2 fuel
pool area is relatively small compared with other cases. Even if the both tunnel portals are blocked, the amount of air inside the tunnel
can still support the burning of liquid fuel pool for a period of time. During this duration, the liquid fuel was heavily consumed, and the
reflash did not occur due to insufficient fuel vapor after the tunnel portals were opened. Another possible reason is that the temperature
attenuation range inside the tunnel is too large after the blocking of both portals, leading to the relatively low temperature inside the
tunnel at the tunnel portals were opened again. At this time, there is enough fuel vapor, but due to the low temperature, there is no
reflash phenomenon.
The reflash issue of tunnel fire threatens the safety of emergency personnel and tunnel structure. It is of certain value to clarify the
fire reflash characteristics and critical conditions in the narrow-long tunnel space structures. Combined with the current experimental
work, Fig. 6 shows the schematic diagram of fire reflash under different fuel pool areas. Theoretically, there is a certain difference
between the reflash of narrow-long tunnel structure fire and that of conventional building fire. Channel-like structure has the char­
acteristics of large length-width ratio and length-height ratio. In this situation, tunnel fire smoke needs to pass a longer distance to
spread out of tunnel space. At the same time, the fresh air from the outside has to travel a long distance to be sucked into internal space
of tunnel to support the burning of liquid fuel pool. During this process, the thermal smoke transfers heat with the ceiling and sidewalls
of the tunnel over a longer distance, and the thermal buoyancy of smoke gradually decreases, resulting in the disorder and settlement
of thermal smoke, which is not conducive to the inflow of fresh air outside to a certain extent.
For the tunnel fire accident with large liquid fuel area, blocking the tunnel portal is an alternative treatment method. The blocking
of both tunnel portals more obviously restricts the flow of hot smoke and cold air. At the same time, the larger the fire source area is, the
more smoke volume generated during combustion is, and the greater the demand for external fresh air is. In addition, the main
material of tunnel structure walls and blocking portals is fire-proof board, which has certain heat insulation ability. Meanwhile, the
surface color of fire-proof board is white, which is conducive to re-radiation to a certain extent, and the flame radiation is feedback to
the liquid fire source inside the tunnel, which is potentially conducive to the occurrence of reflash to a certain extent. In current
experimental work, for the cases with large fuel pool area (0.35 m2 and 0.525 m2), the burning inside the tunnel produces a large
number of thermal products. When the tunnel portal is blocked, these thermal products cannot be discharged in time and are accu­
mulated inside the tunnel space. Hence, in the case of large fuel pool area, the temperature attenuation amplitude inside the tunnel
after blocking is smaller than that in the case of small fuel pool area (as shown in Fig. 5). The high temperature inside the tunnel can
continue to support the vaporization of liquid fuel to a certain extent, which increases the concentration of fuel vapor inside tunnel
space. In these cases, the high concentration of fuel vapor and high temperature have hidden hidden dangers of tunnel fire reflash.
Theoretically, the larger the fuel area is, the higher the possibility of fire reflash inside the tunnel, under the premise that the other
conditions remain the same.

3.3. Radiant heat flux near tunnel portal

Fig. 7 shows the variation trend of external radiant heat flux at tunnel portal under typical cases (the specific locations of mea­
surement points R1, R2 and R3 are shown in Fig. 2). It can be seen that for the case with 0.35 m2 fuel pool area, the maximum radiant
heat flux appears in R2 location before the tunnel portal is blocked, and its peak value is 458.9 W/m2. After the portal is blocked, the
radiant heat flux at the measurement point decreases rapidly because the flame is shielded. At t = 30 min, the tunnel portal is opened, a
large amount of air pours into the internal space of the tunnel, and the tunnel fire reflash occurs, and the thermal plume and flame
overflow from the tunnel portal again. The radiant heat flux at the measurement point increases again, with the maximum value
reaching 191.1 W/m2. For the case with 0.525 m2 fuel pool area, the radiant heat flux outside the tunnel entrance shows a similar
variation trend. However, with the increase of fuel pool area, the maximum radiant heat flux before and after the closure of tunnel

Fig. 6. Schematic diagram of the occurrence of reflash of tunnel fire under different fuel pool area.

7
J. Zhao et al. Case Studies in Thermal Engineering 38 (2022) 102342

Fig. 7. Variation trend of radiant heat flux outside tunnel portal under different fuel pool area.

portal increases obviously, and reach 1501.9 W/m2 and 505.5 W/m2, respectively.
In addition, it can be found that the maximum heat radiation data of R2 point is significantly larger than that of other measurement
points (R2>R1>R3, as shown in Fig. 8), which may be related to geometrical view factor during radiative heat transfer process. For the
case with 0.525 m2 fuel pool area, the continuous flame region can continuously impinge tunnel ceiling, and the part flame expands
along the ceiling, thus previous scholars usually viewed flame shape as the combination of a cylinder and a disk [24]. It is noted that
the extended flame temperature decreases gradually in contact with the tunnel ceiling and sidewalls, as well as the facing direction of
radiant heat flow meter, hence most of the contribution to heat radiation data at R1-R3 comes from the higher-temperature vertical
flame part. In addition, considering that only qualitative interpretation of the variation trend of maximum heat radiation at present
and that the vertical flame is mainly continuous flame region with a more stable flame temperature, the vertical flame is further
simplified as an isothermal plate [25], as shown in Fig. 9. Based on the above assumptions, taking the measuring point R1 as an
example, the receiving radiation heat flow can be simply expressed as [26]:

q̇′′ flame,rad = ε · Ff →1 · σTf4 (2)

where ε, σ Tf , are flame emissivity, Stephen-Boltzmann constant and flame temperature, respectively. For a specific fire scenario, these
parameters are determined, and only the geometrical view factor Ff →1 varies with the measuring point position. Starting from the
definition of geometrical view factor, Ff→1 can be simply calculated by the following formula [27]:
∫
1 cos θ1 cos θ2
Ff →1 = ​ dS (3)
π S r2

The parameters in the above formula are shown in Fig. 9. Obviously, in R2 point, the integrated function cos θ1r2cos θ2 is bigger. Hence,
the heat radiation data of R2 point is obviously larger than that of other measurement points. In addition, the flame overflow near
tunnel portal is closer to R1 point, which results in a higher heat radiation intensity at R1 point than at the lower position of R3. On the
whole, the maximum heat radiation data near tunnel portal shows the trend of R2>R1>R3.

4. Conclusions
This work provides an experimental-case study on the reflash issue of large-area liquid fire inside channel-like structure. A series of
fire tests were performed on a 1/7 model tunnel, and the temperature right above fire source, ceiling temperature, temperature
distribution inside tunnel, and radiant heat flux outside tunnel portal were acquired. The occurrence conditions of reflash and thermal
characteristics induced and driven by reflash were identified. Major finds are summarized as follows:
(1) Blocking the portals at both tunnel ends can significantly suppress the burning intensity of large area liquid fuel, and the
temperature above fire source decreases significantly. Improper portal-opening moment does induce reflash in such fire sce­
narios. In current experiment work, the reflash occurs in the case that pool area of fire source is not less than 0.35 m2.
(2) The occurrence conditions of tunnel fire reflash is mainly related to the amount of combustible vapor accumulated in the in­
ternal space and temperature level near fire source. In the current experiment, when the both portals were suddenly opened, the
ceiling temperature right above fire source of not less than 280 oC is one of the necessary conditions for the reflash occurrence of
current tunnel fire test.
(3) For the cases with large liquid fuel pool, the obvious thermal plume and flame will overflow out portals again when reflash
occurred, which potentially pose the threat to firefighters. Outside the tunnel portal, the radiant heat flux intensity of the middle
part of tunnel portal section is the highest. And the maximum radiant heat flux induced by backflash reaches 505.5 W/m2.
Due to the limitation of experimental conditions, only three liquid fuel area is considered currently. Experimental data and in­
formation is limited to the reflash issue of large-area fire in narrow-long channel-like structures, and can only support the current
qualitative analysis, which is expected to provide some references for the discussion and study on the reflash issue of large-area liquid
fire in such channel-like structures. It should be pointed out that the reflash issue of large-area fire in narrow-long tunnel space is worth

8
J. Zhao et al. Case Studies in Thermal Engineering 38 (2022) 102342

Fig. 8. Maximum radiant heat flux near tunnel portal in typical reflash cases.

Fig. 9. Simple diagram for geometrical view factor.

studying. In the next stage of work, the first problem to be considered is the monitoring of combustible vapor concentration inside
tunnel interior space under high temperature conditions and the effect of the portal-opening moment, which is related to the quan­
titative disclosure of the critical occurrence conditions for reflash. In addition, the location of fire source also has an effect on reflash
characteristics. When the fire source distribution is asymmetrical, the thermal characteristics induced by reflash will also change due
to the difference of air filling distance between the portals and fire source. These issues would be gradually revealed in the future work.

CRediT authorship contribution statement

Jiaming Zhao: Conceptualization, Formal analysis, Methodology, Investigation, Writing – original draft, Writing – review &
editing. Zhisheng Xu: Methodology, Investigation, Writing – review & editing, Funding acquisition, Project administration. Lei Fang:
Investigation, Data curation, Writing – review & editing. Yulun Zhang: Conceptualization, Investigation, Project administration.
Chengbo Yu: Data curation, Investigation, Software.

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.

Data availability

Data will be made available on request.

Acknowledgements
This work was supported by the Fundamental Research Funds for the Central Universities of Central South University (No.

9
J. Zhao et al. Case Studies in Thermal Engineering 38 (2022) 102342

2020zzts166). The authors appreciate the supports deeply.

References
[1] G. Xu, G.Q. Zhu, R.L. Pan, X. Liu, Investigation on temperature distribution under the coupling action of transverse position and fire sealing of linear fire in
tunnel, Case Stud. Therm. Eng. 26 (2021), 101032.
[2] A. Radwan, T. Katsura, L. Ding, A.A. Serageldin, A.I. El-Seesy, K. Nagano, Design and thermal analysis of a new multi-segmented mini channel based radiant
ceiling cooling panel, J. Build. Eng. 40 (2021), 102330.
[3] Z.Y. Wang, X.P. Jiang, H. Park, L.J. Wang, J. Wang, Numerical Investigation on the length of the near-field region of smoke flow in tunnel fires, Case Stud.
Therm. Eng. 28 (2021), 101584.
[4] L.J. Li, D.Q. Zhu, Z.H. Gao, P. Xu, W.C. Zhang, A study on longitudinal distribution of temperature rise and carbon monoxide concentration in tunnel fires with
one opening portal, Case Stud. Therm. Eng. 28 (2021), 101535.
[5] Z. Wang, G.Q. Zhu, J.J. Zhou, G.Q. Chai, P.F. Wang, Experimental study on fire propagation and temperature distribution of passenger car under different
opening conditions, Case Stud. Therm. Eng. 14 (2019), 100472.
[6] N. Hua, N. Elhami-Khorasani, A. Tessari, Review of tunnel fire damage assessment methods and techniques, Transport. Res. Rec. 2675 (2021) 279–290.
[7] M.S. Tomar, S. Khurana, Impact of passive fire protection on heat release rates in road tunnel fire: a review, Tunn. Undergr. Space Technol. 85 (2019) 149–159.
[8] W.G. Weng, W.C. Fan, Critical condition of backdraft in compartment fires: a reduced-scale experimental study, J. Loss Prev. Process. Ind. 16 (2003) 19–26.
[9] W.G. Weng, W.C. Fan, Nonlinear analysis of the backdraft phenomenon in room fires, Fire Saf. J. 39 (2004) 447–464.
[10] A. Horvat, Y. Sinai, Numerical simulation of backdraft phenomena, Fire Saf. J. 42 (2007) 200–209.
[11] S.A. Ferraris, J.X. Wen, S. Dembele, Large eddy simulation of the backdraft phenomenon, Fire Saf. J. 43 (2008) 205–225.
[12] J.X. Wu, Y.T. Zhang, X. Gou, M.J. Yan, E.Y. Wang, L.S. Liu, Experimental research on gas fire backdraft phenomenon, Proc. Environ. Sci. 11 (2011) 1542–1549.
[13] A.P. Chen, L. Zhou, B. Liu, W.H. Chen, Theoretical analysis and experimental study on critical conditions of backdraft, J. Loss Prev. Process. Ind. 24 (2011)
632–637.
[14] L.C. Tsai, C.W. Chiu, Full-scale experimental studies for backdraft using solid materials, Process Saf. Environ. Protect. 91 (2013) 202–212.
[15] S.G. Ashok, T. Echekki, A numerical study of backdraft phenomena under normal and reduced gravity, Fire Saf. J. 121 (2021), 103270.
[16] A. Król, M. Król, S. Krawiec, A numerical study on fire development in a confined space leading to backdraft phenomenon, Energies 13 (2020), 13071854.
[17] C.L. Wu, S. Santamaria, R. Carvel, Critical factors determining the onset of backdraft using solid fuels, Fire Technol. 56 (2020) 937–957.
[18] J.L. Zhao, Y.F. Li, J.M. Li, Y.B. Huang, J.X. Wu, Experimental study on the backdraft phenomenon of solid fuel, PLoS One 16 (2021), e0255572.
[19] D. Myilsamy, C.B. Oh, B.I. Choi, Large eddy simulation of the backdraft dynamics in compartments with different opening geometries, J. Mech. Sci. Technol. 33
(2019) 2189–2201.
[20] J. Mao, Y.H. Xi, G. Bai, H.M. Fan, H.Z. Ji, A model experimental study on backdraught in tunnel fires, Fire Saf. J. 46 (2011) 164–177.
[21] Y.Z. Li, H. Ingason, Overview of research on fire safety in underground road and railway tunnels, Tunn. Undergr. Space Technol. 81 (2018) 568–589.
[22] B. Karlsson, J. Quintiere, Enclosure Fire Dynamics, CRC press, 2000.
[23] X.C. Zhang, Z.J. Zhang, Z.Y. Zhang, W.B. Xu, Q.B. Luo, H.W. Tao, X. Li, Experimental investigation of compartment fires with circular opening: from the aspects
of internal temperature and facade flame, Combust. Flame 213 (2020) 107–116.
[24] Y. Shintani, T. Nagaoka, Y. Deguchi, K. Ido, K. Harada, Simple method to predict downward heat flux from flame to floor, Fire Sci. Technol. 33 (2014) 17–34.
[25] N.A. Liu, J.M. Wu, H.X. Chen, X.D. Xie, L.H. Zhang, B. Yao, J.P. Zhu, Y.L. Shan, Effect of slope on spread of a linear flame front over a pine needle fuel bed:
experiments and modelling, Int. J. Wildland Fire 23 (2014) 1087–1096.
[26] J. Ji, H.X. Wan, Z.H. Gao, Y.Y. Fu, J.H. Sun, Y.M. Zhang, K.Y. Li, S. Hostikka, Experimental study on flame merging behaviors from two pool fires along the
longitudinal centerline of model tunnel with natural ventilation, Combust. Flame 173 (2016) 307–318.
[27] R. Siegel, J. Howell, Thermal Radiation Heat Transfer, Taylor and Francis, 2000.

10