Case Studies in Thermal Engineering 61 (2024) 104764

Contents lists available at ScienceDirect

Case Studies in Thermal Engineering

journal homepage: [Link]/locate/csite

Experimental study on downward flame spread of thermal

insulation material inside the glass curtain wall configuration
under the ambient wind
Wei Duan a, Jinmo Wu b, *
a School of Architecture and Urban Planning, Anhui Jianzhu University, Hefei, Anhui, 230022, China
b Hefei Institute for Public Safety Research, Tsinghua University, Hefei, Anhui, 230601, China

ARTICLE INFO ABSTRACT

Keywords: Due to the chimney effect caused by curtain wall configuration, facade flame spread poses a seri-
Environment wind ous thermal hazard to upper floors. This work studied downward flame spread behavior of
Polyurethane foam polyurethane insulation material in narrow interlayer space of curtain wall. Flame morphology,
Narrow space mass loss rate and flame temperature under different lateral winds were investigated in detail.
Flame spread
Results indicated that downward flame spread was determined by the coupling effect of curtain
wall configuration, stack effect, and ambient wind. Firstly, the pyrolysis front had an inverted V-
shaped distribution under lateral wind effect. With increasing wind speed, flame inclination on
windward side increased first and then decreased, and flame front on leeward side disappeared
after a critical wind speed. Secondly, the confined space and ambient wind constructed a complex
turbulent flow, which led to the nonlinear relationship between flame height and mass loss with
wind speed. The peak temperature of gas flame near sample surface decreased due to cooling ef-
fect of wind. Thirdly, dripping mass, flame spread rate, and Froude number in downward flame
spread were positively correlated with wind speed. Ambient wind and gravity increase the possi-
bility of randomly melting-flame droplets from the original flame, thereby seriously increasing
the risk of secondary ignition.

1. Introduction
The glass curtain walls are widely used in high-rise building facade in recent years. However, the glass curtain wall is easy to
soften and break in fire [1–3], and then the indoor fire development may be accelerated by the fresh air coming in from outside (i.e.,
the ambient wind). In addition, a long and narrow interlayer space is formed between the building curtain wall and the external insu-
lation material of the wall [4], which leads to the chimney effect rised by fire smoke on the building facade. This effect greatly in-
creases the fire risk, and poses serious thermal threat to the adjacent rooms including direct flame contact (thermal convection) and
thermal radiation. The typical accidents are the fire in the Grenfell town in 2017 (Fig. 1(a)), CCTV town fire in 2009 (Fig. 1(b)) and
the fire in Wanxin Building of Shenyang in 2011. The fire spread quickly in the vertical cavity between the wall and the curtain wall
under the chimney effect and developed into a three-dimensional fire. Due to the limitation of air entrainment and buoyancy induced
by glass curtain wall, the induced airflow by the flame will be increased to a certain extent, and it will be further enhanced under the
influence of ambient wind, producing a thermal impact to the upper floor.

* Corresponding author.
E-mail address: wujinmo@[Link] (J. Wu).

[Link]
Received 30 January 2024; Received in revised form 12 June 2024; Accepted 28 June 2024
Available online 28 June 2024
2214-157X/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
([Link]
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

Fig. 1. Fire in Grenfell town and CCTV building ([Link]

Flexible polyurethane foam (FPU) is widely used as building facade insulation material [5], for its advantages of low thermal con-
ductivity, density, and price. As a common polymer insulation material, the FPU board is easy to burn when exposed to high tempera-
ture (due to the fire heat radiation) and produces molten liquid in the combustion process. When these molten liquids leave the FPU
board at the effect of gravity and ambient wind, it may ignite the unburned fuel in advance and cause a faster fire spread. Previous
studies show that opposed flame spread is not a simple one-dimensional spread along the material surface, but a narrow oil pool flame
spread along the width direction [6,7]. Under the effect of the curtain walls and ambient wind, the heat and mass transfer process of
downward flame spread is of complex mechanisms.
The ambient wind may leads to great effect on flame spread process [8–11], which can be reflected on changing the flame spread
direction and flame shape parameters. Due to the coupling effects of wind and flame buoyancy, the heat transfer process from flame
to the fuel and flame shape are quite different from still air condition [12,13]. The flame spread along thermal insulation materials
have been reported in many previous works [14–19], Ma et al. [14] revealed the influence of building curtain wall distance on down-
ward flame spread behavior over polyurethane. It was found that the flame spread behavior was determined by the stack effect, con-
straint effect and the heat transfer process induced by curtain wall. An et al. [4] studied the downward flame spread under the influ-
ence of parallel curtain walls. It can be concluded that the curtain wall space has great influence on air entrainment rate at pyrolysis
front and may lead to the irregular evolution of flame front surface. The flame stretching phenomenon has relatively large effect for
small spacing distance condition, and the flame height changes periodically. Hu [20] studied the physical properties and interrela-
tionships of pool fire in wind, including thermal feedback, combustion geometric characteristics, flame turbulence, smoke and radia-
tion intensity. Tang et al. [21] studied the temperature field distribution and air entrainment coefficient of tunnel fire by point extrac-
tion system, and found that the air entrainment coefficient increased with fire heat release rate.
This work presents experimental study on downward flame spread over polyurethane thermal insulation materials in architectural
glass curtain wall under various ambient wind speed and confined space. Key parameters including flame front shape, fuel burning
rate, melt dripping mass ratio, fire spread rate and heat transfer parameters were thoroughly measured. Theoretical analysis was con-
ducted on evolution of heat and mass transfer mechanisms, from which the effect of wind and confined space on flame spread behav-
iors could be revealed. The conclusion of this study not only help understand the controlling mechanisms of flame spread behaviors
under wind, but also provides consulting for building façade fire spread prevention design.

2. Experimental design
Fig. 2 gives the downward flame spread system over building facade insulation materials, which can adjust the ambient wind
speed and simulate the modern building curtain wall fire. The use of glass is firstly for the widely used glass curtain wall structure
at present, and secondly for the convenience of experimental observation when the glass structure is adopted, the heat transfer and
heed feedback must be different from ordinary walls. For example, the glass curtain wall would lose most of the radiant heat reflec-
tion due to its transmittance for radiation heat transfer but this difference would also the content of our research.
The experimental setup mainly consists of four parts: stable lateral air supply system, downward flame spread platform, curtain
wall, and data acquisition system.
The frequency conversion fan supplies stable lateral wind speed, and simulates six different external wind speeds Uw from 0 to
2.5 m/s with the interval of 0.5 m/s. The reason or basis for the setting is because of the main research object of this paper is small
scale experiment, we calibrate the wind speed according to the speed of the fan. When the wind speed from 0 to 2.5 m/s, the fire
spread phenomenon is different obviously. When the wind exceeds 2.5 m/s, the flame will tilt to leeside completely or even be blown
out. This configuration can limit the fluctuation range of turbulence intensity to less than 5 %. A four-probe anemometer with an ac-
curacy of 0.01 m/s was used for calibration. The result showed that the wind speed fluctuation was less than 0.1 m/s. The downward
flame spread platform consists of a high-precision electronic balance (Sartorius, Germany, with an accuracy of 0.01 g), an FPU
bracket, and a droplet trough. The electronic balance was used to record the mass loss rate during the downward flame spread
process. The dripping mass during the downward burning of FPU was collected by the droplet trough. The FPU board adapted in this
experiment is thermally thick solid which was 80 cm long, 20 cm wide, and 2 cm thick, and was fixed on the adiabatic gypsum board
with fine iron nails. Horizontal reference lines were drawn every 10 cm on the FPU plate surface to record the instantaneous morpho-
logical characteristics of the pyrolysis front during the fire spreading process. A propane igniter was used for linear ignition to ensure

2
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

Fig. 2. Experimental setup for downward flame spread of building facade insulation material in curtain wall with lateral ambient wind.

that the FPU board was ignited simultaneously in the width direction. K-type thermocouple sequences (which are widely used in fire
or combustion experiments) were employed in this work. An array of five thermocouples (T0–T4) was located along leeward side of
the FPU surface and another array was located along lateral side (T5–T9) along windward side. Reference lines were drawn at incre-
ments of 10 cm on board to allow visual analysis of the flame dispersion. The distance between two thermocouples in the same row is
approximately 18 cm. Thermocouple sequences installed 2 mm away from the sample surface and located at the middle of each inter-
val. The top thermocouple (T0, T5) is 15 cm from the top of the sheet, and the bottom thermocouple (T4, T9) is 25 cm from the bottom
of the sheet. Two water-cooled radiant heat flux gauges (Ami Instruments Co., LTD, with an accuracy of 0.01 kW/m2) recorded the ra-
diant heat flux above and below the surface of the FPU board during combustion. Two HD cameras 1# and 2# (SONY, FDR-AX100E)
were used on the front and side of the experimental system to monitor the flame shape and flame propagation speed.
The experiment was carried out in an experimental chamber with constant temperature, pressure, environmental temperature and
relative humidity (20 ± 2 K, 50 ± 3 %). Thus, the influence of the surrounding environment on the experimental results could be ig-
nored. All data such as temperature and mass loss were recorded at a frequency of 1Hz. Specifically, we conducted each experiment
under varying wind conditions at least three times to account for any variability and to ensure the robustness of our findings. This rep-
etition helped in identifying and minimizing random errors, thereby allowing for a more accurate representation of the trends ob-
served in our study. Additionally, we used high-precision instruments for our measurements, including a four-probe anemometer for
wind speed (with an accuracy of 0.01 m/s) and K-type thermocouples for temperature measurements (accuracy: 0.1 K), among oth-
ers. These tools have been calibrated carefully to further reduce systematic errors. To address the uncertainties directly, we calculated
the standard deviation for key experimental outcomes, such as flame height, mass loss rate, and temperature profiles. The physical pa-
rameters of FPU board and glass curtain wall are shown in Table 1.

3. Results and discussion

3.1. Flame front morphology
As shown in Fig. 3, during the downward flame spread of the FPU board in a confined space without wind, natural airflow driven
by the flame plume is constrained by the front wall. The airflow at the middle of the FPU board is larger than that at the two sides. The

Table 1
Physical parameters of FPU board and glass curtain wall.

FPU material: soft polyurethane foam board

Average molecular structure CH1.8O0.30N0.05

3
Density (kg/m ) 41.5
Specific heat (KJ/kg·K) 1.5
Heat conductivity (W/m·K) 0.037
Pyrolysis temperature (K) 440
Heat of combustion (MJ/kg) 30
Curtain wall material: Fire resistant glass
Dimension (height × width × thickness, cm) 100 × 40 × 0.3
3 3
Density (10 kg/m ) 2.7
Transmittance (%) 87–91

3
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

Fig. 3. Schematic diagram of heat transfer and air entrainment under (a) windless and (b) windy conditions.

air turbulence is much complex in this special case with sandwich structure and airflow. As shown in Fig. 4, laminar flow air is sucked
into the interlayer space of the curtain wall from the bottom, and gradually develop into turbulence. An “inverted V-shaped” pyrolysis
front is obtained due to the air entrainment at both sides and the bottom of the sample sheet. Under the lateral wind condition, as
shown in Fig. 5, the coupling effect of lateral wind and buoyancy of flame plume causes unstable airflow and impacts the “inverted V-
shape” of the flame front. Fig. 5 gives the flame front shapes of FPU board in a stable downward burning stage (Time = 150 s) at dif-
ferent wind speeds. The lateral wind changed the flame front shape, thereby affecting the preheating area in the unburned fuel. The
averaged heat flux between thermal penetration length was q˙ ′′ , When the curtain wall existed, heat balance becomes,
{ ′′
q ′′ = q w,rad + q f′′,conv + q ′′ ′′
FPU,cond − q loss,conv
(1)
q ′′ ′′ ′′
w,rad = q ref ,rad + q c,rad

The main heat transfer for vertical fire propagation includes four parts: radiation from the curtain wall q̇w,rad , convection in the in-
′′

terlayer space q̇f ,conv , thermal conduction inside FPU q̇FP U ,cond and heat loss q̇loss,conv influenced by air entrainment due to external
′′ ′′ ′′

wind. For the first term, the overall heat feedback of the curtain wall constituted two parts. The flame radiation reflection q̇ref ,rad by
′′

curtain wall which is inversely correlated with the light transmittance would further enhance the burning process. Additionally, the
direct thermal radiation heat feedback from flame were negligible for vertical flame propagation. Meanwhile, the radiation q̇c,rad will
′′

be emitted by curtain wall itself after heated by flame. It should be noted that the overall temperature field of the interlayer space will
increase as the spreading process, which means q̇f ,conv will increase at relatively low wind speeds, but on the other hand, rising en-
′′

trainment flow at the bottom (enhanced cooling effect) would also increase when the wind speed exceeds a certain value.
As shown in Fig. 5, the angle of the “inverted V-shape” flame front gradually decreased with the increase of wind speed, and the
pyrolysis front on the windward side became steeper. After the wind speed increased to a critical value of 2.0 m/s, the pyrolysis front
on the leeward side finally vanished because it is covered by the flame on the windward side. However, the inclination of the wind-
ward side will not increase monotonically with the wind speed, because when wind speed exceeded the critical value (μ∞ = 1.0 m/

Fig. 4. Analysis of lateral wind speed on convection state in curtain wall interlayer.

4
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

Fig. 5. Morphological of pyrolysis front under different lateral wind speeds.

s), the force of the lateral wind was much greater than that of the vertical flame plume, which led to a smaller variation of flame incli-
nation angle. Furthermore, with the increase of wind speed, the preheating area and heat feedback from the flame to the unburned
area were gradually enhanced. The flame on windward direction was stretched and nearly attached to the sample surface, which ac-
celerated the melting and pyrolysis of unburned materials. This phenomenon can be interpreted based on Gollner's [22] boundary
layer theory, whose extension to the Reynolds' analogy established a relationship between mass, momentum and heat transfer in a
boundary layer over the fuel surface. The equation is

𝜏s m t ′′
= (2)
u∞ v 2∕3 D 2∕3 ln (1 + B)

Where the air shear stress, τs, is the viscosity coefficient multiplied by the derivative of velocity. This term can be written as
( )
𝜕u 𝜕v
𝜏s = 𝜇 + (3)
𝜕y 𝜕x
′′
In addition, μ∞ is the free-stream velocity, v is the kinematic viscosity or momentum diffusivity, ṁt is the mass transfer caused by
shear flow, D is the species diffusivity, and B is the Spalding number is the mass transfer coefficient. This value is determined using the
equation

(4)
( ) ( )
B = ΔHc fY∞ + Cpg ΔT ∕ L + Cps ΔTs .

The mass transfer rate is positively correlated with τs, the effect of which is obvious during solid combustion (as opposed to during
a pool fire) due to an additional induced effect. The difference in flame behavior on the windward and leeward sides of the FPU board
under varying wind speeds is attributed to three effects. First, the flame size will be larger at the specimen sides than in the center of
the board. Second, shear entrainment due to wind from the lateral sides will accelerate the flame spreading velocity in these loca-
tions. Last, as the flame height and temperature reach their maximum values at the leeward side, the burning rate will be increased. In
the later stage of combustion, due to the enhanced burning at the leeward side resulting from the effect of τs, the spreading velocity at
the leeward sides will become significantly faster so as to form the different phenomenon.

3.2. Flame height

Flame height is a significant parameter, which reflects the hazard in a real fire scene. Fig. 6 gives the typical flame morphology ob-
tained by the HD camera in front of the FPU sample. In this work, using Zukoski [23] flame spot probability method to obtain the av-
erage flame height, double flame in the probability of 0.5 of the highest flame image of the bright spot is defined as the average flame
height. In this study, the MATLAB software is used for specific image analysis tailored on flame spread. A program is specialized for
calculating brightness statistics, helping to understand the flame's average shape and variations in luminosity. The flame shape devel-
opment during the downward flame spread of FPU under different lateral wind speeds is shown in Fig. 7.
The flame height is related to the Froude number because the Fr number represents the relative strength between the inertia force
of wind and buoyancy of flame plume. The dimensionless flame height has a negative power relationship with the Fr number, that is,
the dimensionless flame height decreases with the increase of wind speed.

5
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

Fig. 6. Flame shape under different wind speeds.

Fig. 7. Flame shape (a) U∞ = 0.5 m/s, (b) U∞ = 2.0 m/s.

U∞
Fr = √ ( ) (5)
ΔTZ ∕Ta gz

Where, U∞ is the ambient wind speed; z is the flame height; g is the acceleration of gravity; ΔTz is the local temperature difference be-
tween plume and ambient air; and Ta is the ambient temperature. In this experiment, the buoyancy is basically unchanged, and the in-
ertial force increases with the increase of wind speed. The lateral wind affects the thermal feedback, which will influence flame height
and length [24,25]. Fig. 8 shows that there is a nonlinear relationship between flame height and wind speed.
The flame height versus the lateral wind speed is shown in Fig. 8. The average flame height versus the lateral wind speed is shown
in Fig. 9. The result indicated that the flame height increased first and then decreased during the downward flame spread. When the

6
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

Fig. 8. Flame height versus time under different wind speeds.

Fig. 9. Average flame height at different wind speeds.

wind speed was relatively lower, the coupling effect of stack effect and wind in the interlayer of curtain wall led to the acceleration of
air convection, and the flame was stretched. When the lateral wind speed reached the critical value U∞ = 1.5 m/s, the external wind
provided enough air for combustion, thereby accelerating the flame spread and making the flame height reach its peak. If the lateral
wind force was far greater than that of the flame plume. The flame may closely tilt to the leeward side and flame height. The height
and length of the flame will decrease with the increasing wind speed until the flame was blown out.
For buoyant plume, the flame height can be determined by Ref. [26].

Uz2
Hf ∝ (6)
g

Where Hf is the flame height, Uz is induced airflow rate along the wall and g is acceleration of gravity. Equation (6) shows that the
flame height is proportional to 1/2 of the induced airflow velocity between the interlayers. Due to the stack effect, the outside air will
enter the vertical channel from the bottom and both sides, which will increase the airflow velocity induced by flame buoyancy. In this
experiment, the lateral wind would greatly accelerate the airflow velocity and play a leading role.

3.3. Mass loss and melt dripping

The mass loss during the downward flame spread is one of the important parameters to evaluate the burning rate. In this work, the
mass loss represents the total mass loss caused by combustion, including downward flame spread and combustion in falling droplets.
Fig. 10 shows the mass loss of FPU during downward burning at different wind speeds. There were obviously two stages in the mass
loss profile. The mass loss was relatively gentle in the first stage (0–200 s), and got fluctuation in the second stage (after 200 s). In the
early stage, the heat feedback from the flame to the preheating zone and the liquid accumulation at the flame front were relatively
small, and the melt dripping phenomenon was not obvious. As shown in Fig. 11, with the downward burning of FPU, more liquid fuel
was accumulated at the flame front, and drop randomly under the lateral wind condition. In addition, when the wind speed reaches

7
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

Fig. 10. Mass loss of FPU sheet at different wind speeds.

Fig. 11. Random droplet phenomenon in the downward burning.

Uw = 2.0 m/s, the mass loss rate decreased dramatically in the late burning stage. The wind speed promoted the air convection, sup-
plied sufficient air into the curtain wall for combustion. Additionally, the incomplete burning embers adhered to the gypsum board
will blow away from the experimental platform under a larger wind speed. Finally, the burning rate of the FPU board measured in this
work accelerated obviously under a larger wind speed.
As a typical thermosetting polymer, the FPU will get through a melting and dripping process when it is heated. During the down-
ward flame spread, the molten fuel generated at the pyrolysis front is higher than that evaporated and pyrolyzed. Thus, the liquid fuel
accumulates continuously at the flame front. The liquid fuel will drip randomly under gravity force and wind. The frequency and
mass of dripping are affected by the external wind speed. The analysis of the dripping and burning of molten liquid in the experiment
is helpful to understand the development of building fire.
The dripping ratio and burning ratio in the downward flame spread of the FPU plate can be respectively defined as.

8
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

md
ηd = (7a)
mi
mb
ηb = (7b)
mi

Where, mi is the initial mass of FPU plate before combustion, md refers to the molten dripping mass collected by the dripping trough
during the downward flame spread, and mb is the mass consumed by the sample combustion.
Table 2 shows the dripping ratio and burning ratio at different lateral wind speeds. The results illustrate that the droplet rate ηd
increases and the burning rate ηb decreases with the increase of lateral wind speed. The lateral airflow increases the possibility of the
molten liquid fuel to drip. Meanwhile, the melting dripping process and the lateral wind speed take away a great amount of heat,
which slows down the pyrolysis rate and finally lead to a lower burning rate. Obviously, molten dripping phenomenon occurs ran-
domly due to properties of the material and external wind. It is also a significant factor in the secondary fire. The dripping sub-flame
will ignite other unburned fuel in advance, and result in the three-dimensional fire including upwards fire with multi fire sources,
which increases the fire risk and the difficulty for firefighting. We need to take such factors into account of building facade fire pro-
tection design and set several fire partitions on the facade to reduce the probability of secondary fire.

3.4. Temperature and radiation in gas phase

Fig. 12 (a) and (b) show the temperature curves of the side and centerline of the FPU board measured by the thermocouple at an
external wind speed of Uw = 0 m/s and Uw = 2.0 m/s, respectively. Fig. 12 shows that the temperature measured by the side ther-
mocouples increases earlier than that by the center thermocouples. This result consists the “inverted V-shaped” flame as mentioned
above. Comparing the result in Fig. 12 (a) and (b), the flame temperature is relatively stable without wind, and fluctuate under the
lateral wind speed of 2.0 m/s.
The temperature at the side of the sample sheet is generally higher than that at the center during the downward flame spread in
previous studies. But the results in this experiment is different from the previous results. The reason can be related to the heat accu-
mulated between the FPU board and the glass curtain wall, and the thermal radiation fed back from the curtain wall play an impor-
tant role in the downward burning, which leads to the result that the centerline temperature is higher than the leeward side.
The peak flame temperature under wind speed of 2.0 m/s is generally lower than that under windless conditions. The result indi-
cates that the lateral wind may take away the combustion heat and reduce the peak flame temperature. Additionally, the results in the
previous section illustrates that the melting and dripping phenomena are more obvious under the windy condition. This may also be
an important factor in lowering the temperature of the flame, as the dripping fuel can take away heat. The flame radiation profile is
shown in Fig. 13. As the wind speed increases, the radiation response time measured by the radiometer is earlier. This is mainly be-
cause the flame spread rate gradually increases with wind speed. The peak value of radiant heat flux at the upper part of the vertical
plate is higher than that in the lower part. The peak radiant heat flux increases first and then decreases with the increase of wind
speed. Because the air entrainment is accelerated with the increase of wind speed in the downward flame spread, which promotes the
combustion in the gas phase. When the wind speed increases further (Uw = 2.5 m/s), the flame will incline and deviate from the FPU
plate. The radiation from the flame cannot reach the radiant flowmeter completely. Therefore, after the wind speed increases to a crit-
ical value, the radiant heat flux from the flame decreases. The results should be of significance for grasping the distance between the
escape and the fire scene.

3.5. Fire spread rate

The average flame spread rate Vi in this work is obtained by calculating the time interval between two lines drawn on the sample
sheet before the experiment.

ΔD Di+1 − Di
Vi = = (8)
Δt ti+1 − ti

Where, ti is the time when the flame front reaches the i-th mark line, Di is the position of the i-th mark line and ΔD is the distance be-
tween two adjacent mark lines.
The flame spread is mainly determined by the heat feedback from the combustion zone to the preheated fuel, which should in-
clude conduction, radiation, convection and radiant heat feedback from the glass curtain wall. The relationship among combustion,
pool fire size, and wind speed under cross airflow condition can be established as [27].

Table 2
The ratios of liquid drop and burning of FPU under different wind speeds.

Parameter Wind speed (m/s)

0 0.5 1 1.5 2.0 2.5

ηd 17.65 20.10 27.40 29.20 22.64 34.20

ηb 82.35 79.90 72.60 70.80 77.36 65.80

9
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

Fig. 12. Temperature profile on center line and leeward side of FPU sheet under lateral wind speed (a) Uw = 0.5 m/s and (b) Uw = 2.0 m/s.

 1∕2 
𝜌g kg
(9)
′′
  1∕2  
m (x) ∼ U∞ ∕d ln 1 + B∕𝜒 conv
1∕2
𝜇g c𝜌g

Where, U∞ is the wind speed, kg is the gas-phase thermal conductivity, cpg is the specific heat, B is the coefficient of the fuel mass
transfer.
( )
Yo2 ,∞ ΔHc ∕v0 MO − cp Tfuel,vaporization − T∞
B= ( ) (10)
hfg + cl Tfuel,vaporization − T0

(11)
( )
𝜒 conv = 1 − 𝜒 rad + 𝜒 cond

Where, T∞ , T0 are the temperature in ambient condition and initial temperature at liquid phase, respectively. χconv, χrad, χcond are the
convection, radiation and conduction feedback fractions, respectively. Equation (9) shows that the mass burning rate is directly pro-
portional to the wind speed. Namely, the flame spread rate is positively correlated with wind speed.
Fig. 14 shows the average flame spread rate on the windward side and leeward side at the lateral wind speed of Uw = 1.5 m/s.
The results showed that the flame spread rate on the windward side and leeward side gradually increase during the downward
burning. Under the condition of crosswind, there is a significant difference in flame spread rate between the two sides. In the early
stage of flame spread, the flame spread rate on the windward side is faster than that on the leeward side, while in the middle and
late stage, the flame spread rate on the leeward side increases and is higher than that on the windward side. Flame on the wind-
ward side inclines to the leeward side under lateral wind, and the leeward side receives more radiation, thereby accelerating flame
spread rate on leeward side. As is depicted in Fig. 5, the leeward side gradually vanishes during the downward burning. In the ac-
tual fire rescue, the fire firefighters need to evacuate people as soon as possible. The design of the building facade should meet the
needs of fire safety, and the technical means such as sandwich fire partition and filled fire materials can effectively improve the fire
rating of the external wall upon the experiment's findings. Horizontal fire isolation belts for using thermal insulation materials un-
der glass curtain walls. In the design and construction of curtain wall structure, a reasonable smoke discharge port and smoke ex-

10
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

Fig. 13. Radiant heat flux on the surface of FPU sheet under different wind speeds.

Fig. 14. Average flame spread rate on windward side and leeward side (Uw = 1.5 m/s).

haust system should be set up to ensure that the smoke can be removed in time when the fire occurs. location and scale of fire iso-
lation belts should be set reasonably. The curtain wall should be coated with fireproof paint.

4. Conclusion
In this work, the influence of ambient wind on the downward flame spread of thermal insulation material under the glass curtain
wall configuration was studied with a serial of small-scale experiments. The flame shape and temperature, burning rate, and droplet
mass ratio were determined. The conclusions are as follows:

11
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

(1) The flame was stretched and inclined to the preheating zone under the lateral wind. With the increase of wind speed, the
windward flame plume will cover the leeward side, which vanished the flame front at the leeward side. The inclination of the
flame front on the windward side increased first and then decreased with the wind speed.
(2) With the increase of ambient wind speed, the flame height increased first and then decreased. The stack effect formed by the
narrow interlayer of curtain wall accelerated air convection and stretched the flame structure. The mass loss rate of FPU
increased first and then decreased with wind speed. The mass percentage of molten dripping gradually increased under the
influence of wind speed, which increases the fire risk.
(3) The heat accumulation at the flame front during the downward flame spread will be strengthened by a curtain wall because
of its shielding effect. The temperature near the centerline is generally higher than that at the side of the FPU board. The
lateral wind, on the one hand, promoted the melting and pyrolysis rate, on the other hand, took away heat in combustion and
decreased the peak temperature.
(4) The technical means such as sandwich fire partition and filled fire materials can effectively improve the fire rating of the
external wall upon the experiment's findings. Horizontal fire isolation belts for using thermal insulation materials under glass
curtain walls. In the design and construction of curtain wall structure, a reasonable smoke discharge port and smoke exhaust
system should be set up to ensure that the smoke can be removed in time when the fire occurs. The curtain wall should be
coated with fireproof paint.

CRediT authorship contribution statement

Wei Duan: Writing – original draft, Software, Methodology, Data curation, Conceptualization. Jinmo Wu: Writing – review &
editing, Supervision, Software, Formal analysis.

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 Natural Science Foundation of Anhui Province under Grant no. 1908085QE250.

References
[1] Y. Wang, Q. Wang, G. Shao, H. Chen, Y. Su, J. Sun, L. He, K.M. Liew, Fracture behavior of a four-point fixed glass curtain wall under fire conditions, Fire Saf. J.
67 (2014) 24–34.
[2] Y. Wang, Q. Wang, Y. Su, J. Sun, L. He, K.M. Liew, Fracture behavior of framing coated glass curtain walls under fire conditions, Fire Saf. J. 75 (2015) 45–58.
[3] Y.K. Choi, J.T. Kim, M.G. Jeong, S.W. Cho, H.S. Ryou, The influence of fire size on breakage time for double glazed curtain wall system in enclosure fire, J. Mech.
Sci. Technol. 32 (2) (2018) 977–983.
[4] W. An, R. Pan, Q. Meng, H. Zhu, Experimental study on downward flame spread characteristics under the influence of parallel curtain wall, Appl. Therm. Eng.
128 (2018) 297–305.
[5] H. Singh, A.K. Jain, Ignition, combustion, toxicity, and fire retardancy of polyurethane foams: a comprehensive review, J. Appl. Polym. Sci. 111 (2) (2009)
1115–1143.
[6] R. Tu, Y. Zeng, J. Fang, Y. Zhang, The influence of low air pressure on horizontal flame spread over flexible polyurethane foam and correlative smoke
productions, Appl. Therm. Eng. 94 (2016) 133–140.
[7] X. Ma, R. Tu, C. Ding, Y. Zeng, Y. Wang, T. Fang, Thermal and fire risk analysis of low pressure on building energy conservation material flexible polyurethane
with various inclined facade constructions, Construct. Build. Mater. 167 (2018) 449–456.
[8] Q. Zhou, W. yang Gui, X. Ma, X. Cheng, F. Tang, Experimental study of the downward flame spreading characteristics of building facade flexible polyurethane
with a lateral environment wind effect, Case Stud. Therm. Eng. 26 (2021) 101090.
[9] E. Eftekharian, M. Rashidi, M. Ghodrat, Y. He, K.C. Kwok, LES simulation of terrain slope effects on wind enhancement by a point source fire, Case Stud. Therm.
Eng. 18 (2020) 100588.
[10] L. He, K. Liao, G. Zhu, Study on the influence of the longitudinal position of fire source and tunnel width on the maximum ceiling gas temperature, Case Stud.
Therm. Eng. 52 (2023) 103731.
[11] C.S. Lam, E.J. Weckman, Wind-blown pool fire, Part II: comparison of measured flame geometry with semi-empirical correlations, Fire Saf. J. 78 (2015)
130–141.
[12] L. Hu, S. Liu, J.L. De Ris, L. Wu, A new mathematical quantification of wind-blown flame tilt angle of hydrocarbon pool fires with a new global correlation
model, Fuel 106 (2013) 730–736.
[13] L. Hu, L. Wu, S. Liu, Flame length elongation behavior of medium hydrocarbon pool fires in cross air flow, Fuel 111 (2013) 613–620.
[14] X. Ma, R. Tu, W. An, L. Xu, S. Luo, J. Wang, F. Tang, Experimental study of interlayer effect induced by building facade curtain wall on downward flame spread
behavior of polyurethane, Appl. Therm. Eng. 167 (2020) 114694.
[15] W. An, J. Sun, K.M. Liew, G. Zhu, Effects of building concave structure on flame spread over extruded polystyrene thermal insulation material, Appl. Therm. Eng.
121 (2017) 802–809.
[16] Y. Kobayashi, Y. Konno, X. Huang, S. Nakaya, M. Tsue, N. Hashimoto, O. Fujita, A.C. Fernandez-Pello, Effect of insulation melting and dripping on opposed
flame spread over laboratory simulated electrical wires, Fire Saf. J. 95 (2018) 1–10.
[17] L. Zhou, A. Chen, L. Gao, Z. Pei, Effectiveness of vertical barriers in preventing lateral flame spread over exposed EPS insulation wall, Fire Saf. J. 91 (2017)
155–164.
[18] E.K. Asimakopoulou, D.I. Kolaitis, M.A. Founti, Thermal characteristics of externally venting flames and their effect on the exposed façade surface, Fire Saf. J. 91
(2017) 451–460.
[19] Y. Zhou, R. Bu, J. Gong, W. Yan, C. Fan, Experimental investigation on downward flame spread over rigid polyurethane and extruded polystyrene foams, Exp.

12
W. Duan and J. Wu Case Studies in Thermal Engineering 61 (2024) 104764

Therm. Fluid Sci. 92 (2018) 346–352.

[20] L. Hu, A review of physics and correlations of pool fire behaviour in wind and future challenges, Fire Saf. J. 91 (2017) 41–55.
[21] F. Tang, Q. He, F. Mei, Q. Shi, L. Chen, K. Lu, Fire-induced temperature distribution beneath ceiling and air entrainment coefficient characteristics in a tunnel
with point extraction system, Int. J. Therm. Sci. 134 (2018) 363–369.
[22] M.J. Gollner, F.A. Williams, A.S. Rangwala, Upward flame spread over corrugated cardboard, Combust. Flame 7 (2011) 1404–1412.
[23] E.E. Zukoski, B.M. Cetegen, T. Kubota, Visible structure of buoyant diffusion flames, Symp. (Int.) Combust. 20 (1) (1985) 361–366.
[24] Y. Oka, O. Sugawa, T. Imamura, Y. Matsubara, Effect of cross-winds to apparent flame height and tilt angle from several kinds of fire source, Fire Saf. Sci. 7
(2003) 915–926.
[25] J. Moorhouse, Scaling Criteria for Pool Fires Derived from Large Scale Experiments, 1982.
[26] J. Moorhouse, Scaling Criteria for Pool Fires Derived from Large Scale Experiments, British Gas Corporation,Solihull, England; United Kingdom, 1982.
[27] L. Hu, J. Hu, S. Liu, W. Tang, X. Zhang, Evolution of heat feedback in medium pool fires with cross air flow and scaling of mass burning flux by a stagnant layer
theory solution, Proc. Combust. Inst. 35 (3) (2015) 2511–2518.

13