Accepted Manuscript

Title: Experimental study on fire smoke control using water

mist curtain in channel

Authors: Zhigang Wang, Xishi Wang, Yanqing Huang,

Changfa Tao, Heping Zhang

PII: S0304-3894(17)30618-0
DOI: http://dx.doi.org/doi:10.1016/j.jhazmat.2017.08.026
Reference: HAZMAT 18791

To appear in: Journal of Hazardous Materials

Received date: 19-4-2017

Revised date: 9-8-2017
Accepted date: 10-8-2017

Please cite this article as: Zhigang Wang, Xishi Wang, Yanqing
Huang, Changfa Tao, Heping Zhang, Experimental study on fire smoke
control using water mist curtain in channel, Journal of Hazardous
Materialshttp://dx.doi.org/10.1016/j.jhazmat.2017.08.026

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Experimental study on fire smoke control using water mist curtain
in channel

Zhigang Wanga†, Xishi Wanga†,*, Yanqing Huanga, Changfa Taob, Heping Zhanga

a
State Key Lab. of Fire Science, University of Science & Technology of China, Hefei, 230026, China
b
School of Automotive and Transportation Engineering, Hefei University of Technology, Hefei, 230009, China

†
The authors contributed equally to this work and should be considered co-first authors.

Corresponding author: Tel.: 86-551-63606437, Fax: 86-551-63601669, Email: wxs@ustc.edu.cn

Highlights
 A new effective method for fire smoke control in a channel was proposed and tested.

 Each nozzle type had an optimal working pressure for smoke control.

 The flow field of the induced smoke was visualized by means of sheet illumination and

FDS.

 A mathematical model for the stability of the smoke layer was developed.

ABSTRACT

The hazards of the spread of fire smoke in a channel have been recognized. This paper

relates to the potential use of a water mist curtain (WMC) for preventing the spread of fire

smoke, focusing particularly on smoke control at the early stage of a fire, with the aim of

reducing the harm of fire smoke and allowing time for people to escape. Fatal factors for

occupant evacuation in a fire, such as carbon monoxide concentration, smoke temperature,

and visibility, were measured in the section controlled by the WMC. The results indicate that

the WMC can be effective in preventing fire smoke from spreading at the early stage, and

may provide a useful reference for developing a novel method of smoke control. Furthermore,

the effects of nozzles with different spray characteristics were investigated and an optimal
working pressure was suggested. In addition, a mathematical model was simplified and used

to analyze the interaction between the fire-induced smoke layer and WMC spray.

KEYWORDS: Carbon monoxide; Fire smoke; Risk mitigation; Smoke control; Water mist

curtain

NOMENCLATURE
a Upper width of Zone Ⅲ (m) n Number of WMC nozzles

Width of slit (m) Contact length of ceiling and spray generated via single
B N
nozzle (m)

c Bottom width of Zone Ⅲ (m) P Water mist working pressure (MPa)

Optical density of smoke per unit length (dB/m)
D Q
A drop diameter such that the cumulative volume, from zero
DVf diameter to this respective diameter, is the fraction, f , of the Qc
corresponding sum of the total distribution
Original flow rate of one nozzle (L/min)
Flow rate of one nozzle passing through slit and
forming curtain (L/min)

FD Total drag force (N) Smoke temperature in spray region surrounded by

Ti ,a
ambient air (K)

FDZ Drag force on single water droplet in z-direction (N) Smoke temperature in spray region surrounded by
Ti , s smoke (K)

g Gravity acceleration (m/s2) v Velocity of water droplet (m/s)

Smoke layer height before discharging (m) Vertical velocity of water droplet (m/s)
h vz
Smoke layer height after Nozzle 1 discharging (m) Initial vertical velocity of water droplet (m/s)
h1 vz ,in
h2 Smoke layer height after Nozzle 2 discharging (m)
 z ,out Final vertical velocity of water droplet (m/s)

Combustion heat under complete combustion (kJ/mol) Volume flow of WMC above channel ceiling with one
hc V
nozzle (m3)

h Downward smoke displacement (m)

V1 Volume flow of WMC passing through slit (m3)

Vertical height of thermocouple (m) Smoke volume flow from smoke vent (m3)
H V'
Heat release rate (kW) Volume of spray region surrounded by ambient air (m3)
HRR Vi ,a
Width of channel (m) Volume of spray region surrounded by smoke (m3)
S Vi , s
S' Center line distance between fire source and smoke vent (m)
t Time needed for droplets to travel through Zone III

Contacting area between WMC and channel ceiling of one nozzle W Fuel pan size (cm)
S1
(without slit) (m2)
Contacting area between WMC and slit of one nozzle (m2)
S '1 Greek

T Ambient temperature
 Thickness of WMC (cm)

T0 Transmitting terminal signals of obscuration-meter (cd/m2)  Combustion yield

T1 Receiving terminal signals of obscuration-meter (cd/m2)
is Smoke density inside spray region surrounded by
smoke (kg/m3)

mw Original mass flow of one nozzle (kg)

ia Smoke density inside spray region surrounded by
ambient air (kg/m3)

mw Original mass flow rate of one nozzle (kg/s)

s Smoke density outside spray region (kg/m3)

m'w
Mass flow of WMC passing through slit and forming curtain (kg)
a Ambient air density (kg/m3)

m w Mass flow rate of WMC passing through slit and forming curtain
d Water density (kg/m3)
(kg/s)


m comb Consumed fuel flow rate (kg/s)
 Spray angle (degree)

1. Introduction

Toxic gas from fire smoke, known as carbon monoxide (CO), is the most fatal factor in

fire safety [1, 2]. Although lower visibility due to smoke particles makes it more difficult for

people to find a way out, toxic gases may directly harm or kill the evacuee [3]. Therefore, the

spread of smoke and CO concentration levels are two major concerns in fire safety risk

assessment. Channels such as aisles or corridors commonly exist in subways, high-rise

buildings, underground shopping streets, and other large-space buildings, where the personal

load is usually also very large. When a fire occurs in such areas, the channel is one of the

main means of smoke spread; therefore, sufficient attention should be paid to prevent smoke

or toxic gases, such as CO, being released by the fire, from spreading along the channels [3].

Solid obstructions, such as firewalls, fire-resisting shutters, fire-resistant glass, and fire

doors, have traditionally been used for fire and smoke compartmentation in buildings.

However, these measures may not always be available or practical in channels, due to

available clear height limitations or their special functions [3]. In particular, these obstructions

may hamper not only evacuation at the early stage of a fire, but also emergency rescue by

firefighters. Therefore, a more appropriate measure should be adopted that does not influence

normal channel use, by occupying less space, and does not prevent evacuation of people.

Water mist has been widely recognized as a clean and efficient fire-extinguishing method in

practical fire suppression applications [4, 5], and is preferable for fire compartmentation due

to its improved performance in attenuating flame thermal radiation [6, 7]. Therefore, a method

of fire smoke control using a water mist curtain (WMC) in a channel is proposed and tested in
this study. The primary intention is to restrict the smoke and toxic gases that are released

during the initial fire stage.

Pan et al. [8] conducted an experimental study of smoke control using a water mist system

located at the front of subway shield doors, and their results indicated that water mist could

aid in reducing the smoke temperature and toxic species concentration, improving survival

conditions for passenger evacuation. Amano et al. [9] and Murakami et al. [10] proposed

water screen and water spray systems for tunnels. The water screen system of special nozzles

with a 200 μm diameter arranged in rows was used to form a fire compartment to prevent

smoke spread. Morlon et al. [11] studied de-stratification and optical effects during

smoke/mist interactions. Bettelini et al. [12] evaluated all possible applications of flexible

devices for smoke control in road tunnels, including massive doors, water and air curtains,

and inflatable plugs, which could be used in similar manners. The latest works on fire smoke

control using a WMC with a special nozzle were performed by Zhu et al. [6] and Li et al.

[13]. Many other studies indicated that water mist could reduce visibility due to its special

characteristics [14-16]. On the whole, fundamental studies on the confinement of fire-induced

buoyancy-driven smoke flow, particularly the transportation of toxic CO and the interaction

between the smoke layer and WMC, are still very scarce.

Therefore, experimental tests were performed in a 8.5  1.4  2.4 m3 channel model in

order to study the control of smoke and CO transportation at the initial stage of a fire, by

means of a specially arranged WMC system. Two types of water mist nozzles arranged along

a pipeline were installed above the channel in order to generate a relatively homogeneous

WMC. The water mist characteristics of the nozzles were measured by means of

shadowgraphy technology. Both the experimental phenomena and principal analysis of the

fire smoke’s downward displacement are described in detail in the following sections.

2. Experimental apparatus and methodology

2.1 Experimental apparatus

Figure 1 provides a schematic diagram of the experimental apparatus. All the channel
sections were constructed out of steel board, except for the front section, which was made of

toughened glass to enable visualization. A light sheet was used for illumination during flow

field visualization. The wind flow was conditioned by two rectifying meshes at a distance of

150 cm from the pool center for minimizing the fan disturbance. In this work, the airflow fan

was not activated. In order to reduce the influence of the rectifying mesh on smoke

movement, we opened a smoke vent of approximately 30 cm wide and 140 cm long in the

ceiling. The center distance between the vent and fire source was approximately 135 cm. We

used an FDS model to simulate the smoke vent in different positions and measure the smoke

discharge quantity. As illustrated in Fig. 2, it was found that if the center distance between the

smoke vent and fire source exceeds 50 cm, the smoke vent hardly influences the natural fire

smoke, as is the case with the smoke layer height and WMC. A laser light sheet was used for

illumination during flow field visualization.

Two typical square pans with edge lengths of 20 and 25 cm, filled with 150 and 250 g of

diesel fuel, respectively, were tested. The depth of the stainless steel pan was 40 mm, while

the wall thickness was 4 mm. The burning pool was placed on a 4 cm-thick asbestos board

and two 12 cm-thick refractory bricks. The fuel was ignited with 10 and 15 g heptane at the

beginning of each test. An electric balance with sampling intervals of 0.1 s and a resolution of

0.01 g was positioned below the refractory bricks to record the fuel’s mass loss history.

Smoke layer images were recorded with a 3.3 MP SONY HDR-PJ10 digital camera. The CO

concentration, gas temperature, and obscuration value in the protected zone were compared

with and without application of the WMC.

Smoke logging occurs when the drag force of water droplets is larger than the thermal

buoyancy of the smoke layer; otherwise, the layer remains stable [11, 14-15]. Therefore, in

order to ensure that the smoke layer remains as stable as possible during the smoke control

process, the redundant drag force of dispersed water droplets should be reduced or the thermal

buoyancy of the smoke layer increased. It is very difficult to increase the thermal buoyancy of
the smoke layer when the heat release rate (HRR) is kept constant. The evaporating droplets

dispersing in the air extract energy from the flame and combustion products and reduce the

thermal buoyancy of the smoke layer [16]; therefore, we installed the nozzles above the

channel and opened a 4 cm slit along the channel ceiling. This maintained the velocity of the

water mist droplets in the vertical axis downward direction when the water mist moved across

the slit, and transformed the conical water mist into an approximate rectangular curtain. The

arrangement of the nozzles is shown in Fig. 3 (a), where I is the height of the nozzles above

the channel ceiling. In order to ensure that the smoke cannot pass through the slit among the

nozzles, I must satisfy Eq. (1), as follows:

2nI tan( / 2)  S , (1)

where n is the number of nozzles,  is the spray angle, S is the channel width, B is the slit

width, and N is the contact length between the ceiling and spray generated via a single

nozzle. However, I cannot be so large that it does not supply sufficient momentum to confine

the smoke. In this work, the values I =0.3 m and n=7 were used.

2.2 Methodology

2.2.1 Experimental conditions

The experimental tests consisted of two parts, namely characterization of the WMC and

controlling of smoke via the WMC. Tables 1 and 2 provide information on all the cases that

were investigated. During the smoke control test, the doors and windows of the laboratory

room were closed in order to avoid the effects of ambient wind, while two smoke exhaust fans

on the roof were operating during the tests. The WMC was activated 10 s after igniting the

flame. All the thermocouples were calibrated and their accuracy was estimated to be less than

3%. All test cases were carried out two or three times to obtain an average value, and began

with a similar ambient temperature (25±5 ℃).

2.2.2 Water mist characterization

 In this study, we used shadowgraphy technology to measure the velocity and size of the

WMC droplets by means of a LaVision Particle Master system. This system (as depicted in

Fig. 4) is based on high-resolution imaging with pulsed backlight illumination, and is

independent of the particle shape and material (transparent or opaque). However, because of

the difficulties of optical measurement due to the extensive field of a practical WMC, the

nozzles (with one orifice) along the central axis of the WMC system were considered during

the characterization tests. The details of the optical arrangements can be found in the literature

[17]. The droplet quantities were calculated using 100 recorded images.

The nozzle spray characteristics measured at a height of 0.5 m are listed in Tables 3 and 4.

From the preliminary test, we determined that the WMC generated via nozzle 1 was more

efficient for smoke control, even under low working pressures. Therefore, in the tests, the

measurement of nozzle 1 was conducted with pressures from 0.2 to 1.2 MPa, while that of

nozzle 2 was conducted with pressures from 0.6 to 3.0 MPa. The measured spray

characteristics of these two nozzles were proven to meet the water mist definition conditions

[18]. It can be seen that the cumulative volume diameter, Dvf (f = 50, 90), and the Sauter mean

diameter (D32) decreased with an increase in working pressure. Furthermore, changes in the

nozzles’ spray angle were relatively small as the working pressure was increased.

Another key factor of the nozzle is the flow coefficient, which is calculated by the

following equation:

K Q 10P (2)

where K is the nozzle flow coefficient (L/min/MPa1/2) and P is the operating pressure (MPa).

The average K factors of the two injectors were 0.85 and 0.22 L/min/MPa1/2, respectively.

2.2.3 HRR evaluation

The fuel mass loss was monitored with an accuracy evaluated as 1% in the range of

interest, which is directly related to the HRR through the standard relationship as in [19]:
HRR  m
 combhc , (3)

where η is the combustion yield for which a value of 0.9 was set following sufficient burning.

The steady state combustion heat of the diesel oil used in the current application was 44400

kJ/kg. The diesel fuel properties are displayed in Table 5.

In comparison with other liquid fuel, diesel produces more smoke due to its nature

characteristic. Furthermore, given that the channel is located in the area next to the

compartment entrance, only a small fire is considered in Fong’s experiments [20]. Weng et al.

[21] outlined the smoke hazards and experimental basis that used to derive the tenability

limits for occupants to evacuate safely in the event of a building fire. From the height and

obscuration value of the smoke layer without the WMC for the two fire types in the

experiments, we can conclude that the fires used in our experiment exceed the tenability limits

for occupants. The typical temporal evolution of the fuel mass loss and mass loss rate, with

and without the WMC, was modeled in three stages, as shown in Fig. 5, which corresponds to

a pool size of 0.2 m width and a WMC pressure of 1.2 MPa. The maximum mass loss rate of

the steady state was approximately 0.42 g/s and the corresponding HRR was calculated to be

approximately 16.8 kW. The WMC discharge resulted in a slightly lower level and greater

fluctuation of HRR, where the fluctuation was probably due to the air disturbance caused by

small dispersed water droplets; however, the change is not significant in terms of the average

mass burning rate. It was confirmed that smoke-mist interactions have no influence on the

combustion process, and nor does the quantity of the fire-induced smoke.

3. Results and discussions

3.1 Interaction between smoke layer and WMC

The instability that occurs at the surface of the WMC and bottom of the smoke layer for

different velocities is due to Kelvin–Helmholtz (K-H) instability [22, 23]. As illustrated in

Fig. 6, the smoke layer as well as the interaction between the two WMCs at a 1.0 MPa

pressure are visualized at the fire’s early stage with the aid of a laser light sheet. The WMC
without fire is also visualized in Fig. 6. It should be noted that in order to attempt to clearly

present the information in Fig. 6, we increased the contrast and brightness of the images by 20

and 40%, respectively. The curtain surface deformation is rolled up and leads to the formation

of a vortex, which is a typical observational signature of K-H instability. On account of the

droplet’s initial velocity having only a large axial velocity, a very small vortex formation and

less radical dispersed droplets were observed, which contributed to keeping the smoke layer

stable [16].

Furthermore, on account of the vertical downward velocity of the WMC and horizontal

velocity of the smoke layer, as well as the blocking function of the ground and K-H

instability, two large vortexes were formed on both sides of the WMC. As shown in Fig. 7, to

display the large vortex and smoke spread process clearly, a complementary computational

fluid dynamics simulation was further performed to illustrate the flow field affected by the

WMC with 1.0 MPa pressure in a same-scale channel, while the other model conditions were

kept as consistent as possible with the experimental tests. The internal boundary material of

the simulated channel was set to be the same as that of the experiments. The materials of the

ceiling and side wall were set to be metal, while the floor was set to be concrete. A mixture

fraction combustion model was used for LES simulation [24]. The pool fire with an HRR of

16.8 kW was located 1.35 m from the WMC, and the ambient temperature was approximately

25 ℃. Multi-mesh FDS technology was applied, with 0.01 m (x) × 0.01 m (y) × 0.01 m (z) for

every control volume [25]. We used the equivalent block area to replace the influence of the

rectifying mesh and airflow fan on the smoke. A more detailed design method of the nozzle is

referenced in our previous report [6], with strong agreement shown between the predicted and

measured values in the full-scale tests.

The formation of the two large vortexes in Fig. 7 (a) enhances the entrainment of the

surrounding smoke into the WMC. The smoke in the protected zone is entrained even though

it has passed through the WMC before it discharges, as illustrated in Fig. 6 (c) and (d). The
smoke spread is restrained and the smoke layer in the confined zone maintained, which can

benefit occupant evacuation in the channel. The WMC in Fig. 6 (b) is brighter than that in (d),

allowing us to infer that more fire smoke is mixed in the WMC in Fig. 6 (d). Moreover, from

the smoke spread process shown in Fig. 7 (b), we can conclude that the smoke layer moves

down to the floor.

3.2 Smoke temperature

From Fig. 6, it can be seen that the WMC activation alters thermal stratification,

longitudinal flow, and smoke plume behavior significantly, where the hot smoke plume

entrained under the ceiling is finally pushed down and strongly cooled, and the vertical

temperature gradient in the protected zone is very low. The average temperature distributions

along the centerline of the thermocouple tree, under different working pressures of the two

nozzles, are presented in Fig. 8. According to Mie’s theory [26], the spectral characteristics of

spherical water droplet absorption and scattering may also be significant in the attenuation of

thermal radiation and temperature. Therefore, it can be concluded that the WMC acting as a

shield to gas flow provides a very effective function for the separation of fire smoke

prevention. The effectiveness of the WMC has also been proven in a full-scale experiment

[6]. However, we found that the temperature of the lowest position was higher than the

adjacent point in Fig. 8, which is contrary to our common sense. This may be caused by the

vortex shown in Fig. 7, which induces the smoke downwards, and the smoke density in the

lowest position is higher than the adjacent point.

3.3 Visibility

As illustrated in Figs. 6 and 7, it can be considered that most of the smoke particles being

entrained by the vortex and within the surrounding smoke are finally decontaminated and

sedimentated in the WMC. Furthermore, a significantly greater amount of smoke in the

unprotected zone was controlled by the vortex and aerodynamic action, which were affected

by the WMC; therefore, smoke can rarely pass through the WMC at the early stage. Visibility
is an important parameter in fire safety assessment, and was measured by an AML

obscuration-meter, based on the attenuation of a laser beam at a 635 nm wavelength. Fig. 8

illustrates the visibility at 2 m above the channel floor in the protected zone under different

working pressures.

The obscuration value is negatively correlated with visibility. Based on the Lambert–Beer

law, the optical density of smoke can be defined as [13]:

D  10log T1 T0  / Z (4)

where T1 and T0 are the transmitting and receiving terminal signals of the obscuration-

meter, respectively, and Z is the length between the transmitting and receiving terminals of

the obscuration-meter.

From Fig. 9, it can be seen that the visibility in the protected zone was obviously

improved when the WMC was discharged. The lowest obscuration value with an HRR of 16.8

kW was identified when the pressure was 0.8 MPa for nozzle 1 and 1.6 MPa for nozzle 2,

which means that the visibility reached the maximum value under the corresponding pressures

of the two nozzles, and the corresponding minimum total obscuration values of the two

nozzles were reduced to 82.3 and 66.2%, respectively. However, the visibility could not be

greatly improved in the cases with a relatively large HRR of 32.8 kW, particularly for nozzle

2.

3.4 CO concentration

The WMC may prevent fire smoke by decreasing its horizontal velocity, which can be

analytically attributed to a reduction in CO concentration. The horizontal velocity of the

smoke strengthens the flow vortex intensity, increasing the recirculation of toxic gas and

preventing smoke from passing through the WMC, which may cause a major decrease in CO

concentration as measured via the gas analyzer at point “G.” Furthermore, the aerodynamic

action affected by the WMC provides a function for blocking CO movement. Fig. 10 displays

the concentration–time curve, which indicates that the CO concentration in the first stage was
low, and a low concentration was retained for a period after the WMC was applied. However,

as shown in Fig. 10 (b) and (d), for the cases of P=0.2 MPa for nozzle 1 and P=0.2-0.6 MPa

for nozzle 2 with HRR=32.8 kW, there was hardly any effect on decaying of the CO

concentration.

The time average was between 250 and 350 s, which is considered as the steady state, as

shown in Fig. 11. The lowest CO concentrations were determined in the cases of P=0.8 MPa

for nozzle 1 and P=2.0 MPa for nozzle 2. However, with a further increase in the WMC

working pressure, an improvement in visibility and decay of CO concentration is not obvious,

and the engineering cost increases. Considering the abovementioned several key factors

affecting human lives in a channel fire, we can conclude that there exists an optimal nozzle

working pressure. For example,, the optimal working pressure of nozzle 1 is approximately

0.8 MPa, and that of nozzle 2 is 2.0 MPa when the HRR is 16.8 kW. We used a 32.8 kW fire

to simulate the fire that develops to a certain extent, and a 16.8 kW fire to simulate the fire at

the early stage. The smoke control effectiveness was found to be reduced for the 32.8 kW fire,

as much more smoke moved across from the space between the WMC droplets. Nevertheless,

the WMC also affects smoke control, although the control effect is reduced to a certain extent

for larger fires. In terms of the smaller droplets of nozzle 2, the velocity attenuates quickly

after the discharging, disperses in the channel, and evaporates quickly. The evaporating

droplets dispersed in the air extract energy from the flame and combustion products, which

reduces the smoke layer’s thermal buoyancy; thus, the smoke layer stability is broken and it is

induced downwards. Therefore, we can conclude that nozzle 1 is more efficient in smoke

control based on our experimental tests.

3.5 Downward displacement of fire smoke induced by WMC

The downward displacement of the fire smoke induced by the WMC should be analyzed

because it may directly affect the smoke control efficiency. Li et al. [27, 28] discussed the

drag component of sprinkler spray adjacent to the horizontal smoke vent by means of a
mathematical model. Tang et al. [14, 16] developed a new model that proved to be more

pronounced for smaller droplets; therefore, Tang’s model was simplified and used to calculate

the downward displacement in this study.

As indicated in Fig. 12, the z-direction is vertical and r refers to the radial direction, while

vz and vr are the vertical and radial components, respectively, of the droplet velocity. The final

result of a single water droplet will determine the volume of Zone III and total drag force. It

can be assumed that the initial radial momentum of the water entering the channel is

destroyed by the ceiling slit, and we neglect the interaction between the water droplets and

smoke plume movement, so that the drop’s initial radial velocity component is zero. This

assumption is believed to be reasonable, and is similar to that of Bullen’s study [29].

Therefore, the velocity in the vertical direction is the droplet velocity, which can be

determined by

v vz2  vr2  vz  (5)

In the experimental tests, the height of the smoke layer was approximately 0.6 m when

HRR=16.8 kW. The nozzle was installed 0.3 m above the channel ceiling, so vz,in is equal to

the velocity of the spray measured at 0.3 m, while vz,out is equal to the velocity of the spray

measured at 0.9 m from the nozzle orifice. The corresponding velocities measured by means

of shadowgraphy are listed in Table 6.

For the downward forces resulting from the cooling effect (net downward buoyancy in the

smoke region) and drag force from water droplets, Δh can be calculated by means of Eq. (6),

while FD can be calculated by Eq. (7) [16].

 
mw  vz ,in  vz ,out   mw g  i ,s   s Vi ,s g   a  i ,a Vi ,a g (6)
total drag force

FD  mw  vz ,in  vz ,out   mw g  (7)

In this study, the droplet velocity is adopted by the average velocity, so the time for an

individual droplet to pass through the smoke can be determined by:

 2(h  h)
t  (8)
vz ,in  vz ,out

The water mass flow rate of a nozzle can be calculated by the density and original flow

rate of the nozzle ( Q ), as follows:

mw  Q (9)

The water mass flow of a nozzle is

mw  m
 w t (10)

The actual water droplet mass flow rate, which is the mass flow rate passing through the

slit with a width of B, and the total water droplet mass can be calculated by

1 '
S 1h
V1 3 7 NB

mw  n mw  7 mw  mw (11)
V 1 
S1h  ( I tan )2
3 2

NB
m' w  7 mw  (12)

 ( I tan ) 2

Therefore, the true drag force can be calculated as

FD  m
 w (vz ,in  vz ,out )  mw g (13)

Furthermore, Vi,s and Vi,a as shown in Fig. 12 can be determined from the water droplet

trajectories, as follows:

 1
Vi , s  B   a  b  h
 2
  (14)
V  1 B   b  c  h


i ,a
2

Consequently, Δh can be determined: in Eq. (15), F’D can be calculated from Eq. (13),

while the other quantities (ρi,s, ρs, ρa,ρi,a, and h) are input variables. Note that in Eq. (13),

mW , mw' , vz,out, and vz,in are also input variables.

FD' 
1
2
 i ,s  s  gB   a  b  h   a  i ,a  gB   b  c  h 
1
2
(15)
 Tables 7 and 8 provide the results of the depth of downward displacement for the model

and experimental results. The sums of Δh and h in Tables 7 and 8 of the two nozzles are

sufficiently larger than the channel height, and are consistent with the experimental

observation, as illustrated in Figs. 6 and 7. This means that the smoke entrained in the WMC

becomes unstable and falls to floor. The smoke is suppressed in order to pass through the

WMC, and the entrained smoke is decontaminated and sedimentated. From the two nozzles’

true drag force, we determined that each nozzle with a lower Δh has a smaller true drag force,

which is consistent with Tang’s results [16].

5. Conclusion

An experimental study was carried out in order to investigate the possibility of using a

WMC for fire smoke containment in a channel. The results indicate that the smoke particles

and CO gas released by the fire can be reduced to a great extent. The obscuration in the

protected zone was also demonstrated to exhibit an attenuating effect when the WMC was

discharged, and each of the two nozzle types had optimal pressure in smoke control.

Furthermore, the WMC system could effectively decrease the temperature of the protected

zone, even under a low working pressure.

Moreover, an analytical model for the effect of the WMC on a fire-induced smoke layer

was developed based on the current model presented by Tang et al. [16]. It was proven that

the smoke entrained in the WMC may become unstable and fall to the floor during the early

stage after discharging of the WMC, which strongly agrees with the experimental results.

Acknowledgements

The authors appreciate the support of the Key national R & D program (Grant No.

2016YFC0800603), Natural Science Foundation of China (Grant No. 51323010 and

51408181), and the Fundamental Research Funds for the Central Universities (Grant No.

WK2320000035).
 References

[1] Hu, L. H, Fong N K, Yang L Z, et al. Modeling fire-induced smoke spread and carbon monoxide
transportation in a long channel: Fire Dynamics Simulator comparisons with measured data, J.
Hazard. Mater.140 (2007) 293-298.
[2] Besserre, R., Recent studies prove that the main cause of death during urban fires is poisoning by
smoke, Urgences Médicales.16 (1997) 77-80.
[3] Hu, L.H, Zhou, J.W, Huo, R., et al, Confinement of fire-induced smoke and carbon monoxide
transportation by air curtain in channels, J. Hazard. Mater.156 (2008) 327-334.
[4] Grant, G., Brenton, J. and Drysdale D. Fire suppression by water sprays, Prog. Energ.
Combust.26 (2000) 79–130.
[5] Huang X., Wang X. S., Liao G. X. Characterization of an effervescent atomization water mist
nozzle and its fire suppression tests, Proceedings of the Combustion Institute. 33(2011): 2573-
2579.
[6] Zhu, P., Wang, X. S., Wang, Z. G., et al. Experimental and numerical study on attenuation of
thermal radiation from large-scale pool fires by water mist curtain, J. Fire Sci. 33 (2015) 398-403.
[7] Buchlin, J.M., Thermal shielding by water spray curtain, J. of Loss Prevention in the Process
Industries. 18 (2005) 423-432.
[8] Pan, L.W., Lo, S.M., Liao, G.X., et al, Experimental study of smoke control in subway station for
tunnel area fire by water mist system, Procedia Engineering. 11 (2011):335-342.
[9] Amano, R., Kurioka, H., Kuwana, H., Murakami, M., Tsuruda, T., Suzuki, T., Ogawa, Y.
Applicability of water screen fire disaster prevention system to road tunnels in Japan. In: 3rd
International Conference ‘Tunnel Safety and Ventilation’, Graz, (2006) pp. 162–173.
[10] Murakami, M., Kurioka, H., Imazeki, O., Kuwana, H., Amano, R. Numerical simulation in effect
of compartmentalization with water screen in a tunnel fire. In: Barták, Hrdina, Romancov Zlámal
(Eds.), Underground Space- the 4th Dimension of Metropolises. (2007) pp. 1705–1710.
[11] Morlon, R., Boulet, P., Parent, G., et al, Study of de-stratification and optical effects observed
during smoke/mist interactions, Fire Technol. 51 (2015)1231-1248.
[12] Bettelini, M., Rigert, S., Seifert, N. Flexible devices for smoke control in road tunnels. In: 6th
International Conference ‘Tunnel Safety and Ventilation’, Graz, (2012) pp. 265–272.
[13] Li, Y., Wang, X. S., Lan, M. J., Zhu, P., Fire smoke control by water mist curtain, Journal of
Combustion Science and Technology. 20(2014):263-269 (in Chinese).
[14] Tang, Z.J., Vierendeels, Study on the dynamics of smoke movement under some typical
sprinklers, Ph.D.Thesis, University of Science and Technology of China. 2006.
[15] Tseng, C.C., Viskanta, R., Absorptance and transmittance of water spray/mist curtains, Fire Saf.
J. 42 (2007) 106-114.
[16] Tang, Z., Vierendeels, J., Fang, Z., et al, Description and application of an analytical model to
quantify downward smoke displacement caused by a water spray, Fire Saf. J. 55(2013) 50-60.
[17] Wang, X.S., Zhu, P., Li, Y., Ni, X.M., Fan, M.H., Effect of low ambient air pressure on spray
characteristics of water mist. Exp. Therm. Fluid Sci. 66 (2015) 7–12.
[18] NFPA 750. Standard on Water Mist Fire Protection Systems, Edn., National Fire Protection
Association, Quincy, MA 02169, USA (2015).
[19] Babrauskas, V., Heat release rates (Chap. 3). In: SFPE Handbook of fire protection engineering,
3rd edn. SFPE, New England, (2002) 3–37.
[20] Fong, N.K., Chow, W.K., Chan, W.T., Preliminary studies on smoke spreading prevention and
thermal radiation blockage by water curtain, International Journal on Engineering Performance-
Based Fire Codes, Volume 3, Number 3, (2001) 98-103.
[21] Weng, P., Tenability criteria for design of smoke hazard management systems, Ecolibrium 8,
(2011) 3-37.
[22] Johnson, J.R., Wing, S., Delamere, P.A., Kelvin Helmholtz instability in planetary
magnetospheres, Space Science Reviews. 184 (2014) 1-31.
[23] Rothe, P.H, Block, J. A, Aerodynamic behavior of liquid sprays, International Journal of
Multiphase Flow. 3(1977)263-272.
[24] McGrattan, K.B., Forney, G.P., Fire Dynamics Simulator (Version 6.4.0)—User’s Guide. NIST
Special Publication 1019, National Institute of Standards and Technology, Gaithersburg, MD,
(2016).
[25] Petterson, N. M., 2002. Assessing the feasibility of reducing the grid resolution in FDS field
modelling. In: Fire Engineering Research Report, School of Engineering, University of
Canterbury, Christchurch, New Zealand, (2002).
[26] Hergert, W., Wriedt, T., The Mie Theory: Basics and Applications, Springer, Berlin (2012).
[27] Li, K.Y, Spearpoint, M, J., Simplified calculation method for determining smoke downdrag due
to a sprinkler spray, Fire Technol. 47 (2011) 781-800.
[28] Tang, Z., Fang, Z., et al, Yuan, J.P., Merci, B., Experimental study of the downward
displacement of fire-induced smoke by water sprays, Fire Saf. J. 55(2013) 35-49.
[29] Bullen, M. L, The effect of a sprinkler on stability of a smoke layer beneath a ceiling, Fire
Technol.13 (1977).
Fig. 1. Schematic of experimental apparatus
 0.14
S

0.12

0.10
V' (m3/min)

0.08

0.06

0.04

-0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

'
S (m)

Fig. 2. Smoke volume flow varies with center line distance between fire and smoke vent
 Slit

(a) Arrangement of nozzles above ceiling (b) Spray pattern of single nozzle

Fig. 3. Arrangement of WMC nozzles

Fig. 4. Schematic of water mist characterization by means of Particle Master Shadow system
 mwithout m'with
mwith m'without
168 Fluctuation range (P=1.2MPa) 0.6

Mass buring rate(g/s)

147
0.5
126
Mass(g)

105 0.4
84 0.3
63 Steady stage 0.2
42
0.1
21 Fluctuation range (P=0MPa)
0 0.0
0 50 100 150 200 250 300 350 400
T(s)

Fig. 5. Typical temporal evolution of mass loss and mass burning rate of pool fire
with and without effect of WMC (P=1.2 MPa, L=0.2 m, nozzle 1)
 (a) 20 s after ignition, (b) 20 s without fire,
without water mist P=1.0 MPa, nozzle 2

(c) 20 s after ignition, (d) 20 s after ignition,

P=1.0 MPa, nozzle 1 P=1.0 MPa, nozzle 2

Fig. 6. Visualization of smoke decay via laser light sheet

 WMC
Nozzle Smoke vent

Pool fire

Rectifying mesh

(a) 20 s after ignition, (b) 20 s after ignition,

P=1.0 MPa, nozzle 1 P=1.0 MPa, nozzle 1
Fig. 7. Visualization of flow field and smoke spread process by FDS model simulation
 2.4
2.2 E

2.0
1
1.8
1.6 2
H(m)

1.4 No water mist

3 0.2 MPa
1.2 0.4 MPa
1.0 4 0.6 MPa
0.8 MPa
0.8 5 1.0 MPa
1.2 MPa
0.6
25 30 35 40 45 50 55 60
Temperature(℃ )

(a) Temperature at measurement points of thermocouple tree and position E (for the cases of nozzle 1, T∞=23 ℃)

2.4
2.2 E

2.0
1
1.8
1.6 2
H(m)

1.4 No water mist

3 0.6 MPa
1.2 1.0 MPa
4 1.6 MPa
1.0 lowest position
2.0 MPa
0.8 3.0 MPa
5 4.0 MPa
0.6
25 30 35 40 45 50 55 60
Temperature(℃ )

(b) Temperature at measurement points of thermocouple tree and position E (for the cases of nozzle 2, T∞=28 ℃)

Fig. 8. Smoke temperature measured at different vertical positions in the protected zone
 70 70
HRR 32.8 kW HRR 32.8 kW

Obscuration value（ %/m）

Obscuration value（ %/m）
60 HRR 16.8 kW HRR 16.8 kW
60
50
50
40
40
30
30
20

10 20

0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 1 2 3 4

Water mist curtain pressure (MPa) Water mist curtain pressure (MPa)

(a) Nozzle 1 (b) Nozzle 2

Fig. 9. Obscuration measured in the protected zone by WMC under different operating pressures
 120
No water mist No water mist
60
0.2MPa 0.2MPa
0.4MPa
100 0.4MPa
50
0.6MPa 0.6MPa
0.8MPa 80 0.8MPa
40

CO (ppm)
CO (ppm)

1.0MPa 1.0MPa
1.2MPa 60 1.2MPa
30
Water mist Water mist
20 discharged 40 discharged

10 20

0 retain low concentration

0

0 100 200 300 400 0 100 200 300 400

Time (s) Time (s)
(a) Nozzle 1 (HRR=16.8 kW) (b) Nozzle 1 (HRR=32.8 kW)

120 No water mist

60 No water mist
0.6MPa 0.6MPa
100 1.0MPa
50 1.0MPa
1.6MPa 1.6MPa
2.0MPa 80 2.0MPa
40
CO (ppm)

3.0MPa
CO (ppm)

3.0MPa
4.0MPa 60
30 4.0MPa
Water mist
20 discharged 40 Water mist
discharged
10 20

0 0

0 100 200 300 400 0 100 200 300 400

Time (s) Time (s)
(c) Nozzle 2 (HRR=16.8 kW) (d) Nozzle 2 (HRR=32.8 kW)

Fig. 10. CO concentration measured on side protected by WMC

 HRR 16.8kW 100 HRR 16.8kW
100
HRR 32.8kW HRR 32.8kW

80 80

CO (ppm)
CO（ ppm）

60 60

40 40

20
20

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 1 2 3 4
Water mist curtain pressure (MPa) Water mist curtain pressure (MPa)

(a) Nozzle 1 (b) Nozzle 2

Fig. 11. Time-averaged CO concentration deduction under different WMC working pressures
Fig. 12. Schematic of instability criterion from Bullen theory
Table 1. Test cases for smoke control with WMC

Nozzle type Initial fuel mass (g) P (MPa) W (cm) HRR (kW)
Nozzle 1 150 0.2/0.4/0.6/0.8/1.0/1.2 20 16.8
250 25
Nozzle 2 150 0.6/1.0/1.6/2.0/3.0/4.0 20 32.8
250 25
Table 2. WMC characterization cases
Nozzle type P (MPa) Distance between nozzle and measuring point (m)
Nozzle 1 0.2/0.4/0.6/0.8/1.0/1.2
0.1/0.3/0.5/0.7/0.9
Nozzle 2 0.6/1.0/1.6/2.0/3.0/4.0
Table 3. Characteristics of single injector of WMC nozzle 1 (H=0.5 m)
P DV90 DV50 D32 Spray Q (L/min) Qc (L/min)
(MPa) (m) (m) (m) angle (θ)  (cm)
(Calc.*)
0.2 192 162 157 62 1.20 0.07 10.3
0.4 180 149 149 64 1.70 0.11 12.5
0.6 171 126 134 65 2.08 0.13 13.4
0.8 155 121 128 68 2.40 0.15 14.6
1.0 150 114 123 69 2.68 0.17 16.1
1.2 132 110 121 69 2.94 0.18 17.3
‘Calc.*’means that the values were calculated.
Table 4. Characteristics of single injector of WMC nozzle 2 (H=0.5 m)
P DV90 DV50 D32 Spray Q (L/min) Qc (L/min)
(MPa) (m) (m) (m) angle (θ)  (cm)
(Calc.*)
0.6 76 54 50 49 0.53 0.07 10.1
1.0 68 49 45 50 0.71 0.09 11.9
1.6 57 43 40 53 0.89 0.10 13.2
2.0 53 38 36 53 0.97 0.11 14.7
3.0 56 40 40 55 1.20 0.13 15.3
4.0 55 38 39 55 1.39 0.15 20.5
*‘Calc.’ means that the values were calculated.
Table 5. Fuel properties
Fuel type Specific heat at Flash point Heat of Density Boiling
1 atm combustion point
0# diesel oil 2.1 kJ/kg ℃ 38 ℃ 44,400 kJ/kg 0.86 g/cm3 170-390

℃
Table 6. Measured velocities of water mist droplets under different working pressures

Nozzle 1 Nozzle 2
P (MPa) v(H=0.3 m) v(H=0.9 m) P (MPa) v(H=0.3 m) v(H=0.9 m)
0.2 2.65 1.34 0.6 2.13 2.1
0.4 4.37 2.23 1.0 3.64 2.9
0.6 5.77 3.24 1.6 4.42 2.64
0.8 5.87 4.19 2 4.93 3.66
1.0 6.77 4.23 3 7.07 4.36
1.2 10.27 5.19 4 7.76 4.65
 Table 7. Summary of tests (h=0.6 m, nozzle 1, HRR=16.8 kW)
P (MPa) ṁ'w (kg/s) Δh (m) (experiment) Δh (m)
(Calc.*) (current model)
0.2 0.0082 1.80
0.4 0.0128 Greater than 1.80 m 1.90
0.6 0.0152 (distance between floor and 2.03
0.8 0.0175 bottom of hot smoke layer) 2.09
1.0 0.0198 (see Fig. 6 (c)) 2.17
1.2 0.0210 2.37
* ‘Calc.’ means that the values were calculated.
 Table 8. Summary of tests (h=0.6 m, nozzle 2, HRR=16.8 kW)
P (MPa) ṁ'w (kg/s) Δh (m) (experiment) Δh (m)
(Calc.*) (current model)
0.6 0.0082 2.13
1.0 0.0105 Greater than 1.80 m 2.21
1.6 0.0117 (distance between floor and 2.28
2.0 0.0128 bottom of hot smoke layer) 2.34
3.0 0.0152 (see Fig. 6 (d)) 2.43
4.0 0.0175 2.57
* ‘Calc.’ means that the values were calculated.