- Email: [email protected]
A study of the probability distribution of pool fire extinguishing times using water mist
Accepted Manuscript Title: A Study of the Probability Distribution of Pool Fire Extinguishing Times using Water Mist Author: Tianshui Liang Mengjie Liu Zhonglin Liu Wei Zhong Xiukun Xiao Siuming Lo PII: DOI: Reference:
S0957-5820(14)00079-2 http://dx.doi.org/doi:10.1016/j.psep.2014.05.009 PSEP 448
To appear in:
Process Safety and Environment Protection
Received date: Revised date: Accepted date:
18-7-2013 7-5-2014 20-5-2014
Please cite this article as: Liang, T., Liu, M., Liu, Z., Zhong, W., Xiao, X., Lo, S.,A Study of the Probability Distribution of Pool Fire Extinguishing Times using Water Mist, Process Safety and Environment Protection (2014), http://dx.doi.org/10.1016/j.psep.2014.05.009 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.
Research Highlights
Ac
ce pt
ed
M
an
us
cr
ip t
Two kinds of extinguishing situation of pool fire by water mist were observed. In one, the flames had not yet been suppressed when the fire was extinguished via a blowoff process. The primary mechanism for fire extinguishment is flame cooling in the first situation In the other one, the flame was first suppressed and then gradually reduced in size. The primary mechanism for fire extinguishment is fuel surface cooling in the second. .
Page 1 of 30
*Manuscript
A Study of the Probability Distribution of Pool Fire Extinguishing Times using Water Mist Tianshui Liang1, Mengjie Liu1, Zhonglin Liu1, Wei Zhong1*, Xiukun Xiao 2, Siuming Lo3 1
us
cr
ip t
School of Chemical Engineering and Energy, Zhengzhou University, Zhengzhou 450001, China 2 Institute of Industrial Technology of Guangzhou and Chinese Academy of Science, Guangzhou, 511458, China 3 Department of Civil and Architectural Engineering, City University of Hong Kong, Kowloon, Hong Kong, SAR China
Ac ce pt e
d
M
an
ABSTRACT: Water mist, a replacement for Halon gaseous agents in fire fighting, has been studied for decades. However, the fire-extinguishing reliability of water mist is debated. For example, there are significant differences in extinguishing times between tests conducted under the same conditions, and water mists have difficulty extinguishing small fires. To date, no study of the probability distribution of extinguishing times has been reported. In this study a statistical analysis of the extinguishing time distribution of pool fires extinguished using water mist is presented. The fire sources were circular/square stainless steel pans with gasoline, diesel, ethanol or Daqing RP-3 as fuel. Two types of extinguishing scenarios were observed. In one situation, the fire was extinguished via a blow off process, when the flames had not yet been suppressed. Flame cooling is the primary fire extinguishing mechanism; the mass loss rate and combustion heat of the fuel are two key factors. In the other situation, the fire was initially suppressed and subsequently extinguished after a long suppression stage. Surface cooling is the primary fire extinguishing mechanism; the flash point of the fuel is the key factor. KEYWORDS: Pool Fires; Water Mist; Fire Extinguishment; Flame cooling; Surface cooling; Variance Analysis
*
Corresponding author, [email protected], TEL: 86-371-67739005, FAX:
86-371-67781801
1 Page 2 of 30
1. Introduction Fire extinguishment and suppression using fixed water mist systems have been studied for 60 years [1, 2]. Interest in this subject renewed
environmental
after
halogen-based
reasons.
Notable
agents
qualitative
were
banned
for
ip t
was
and
quantitative
cr
investigations of the fire extinguishing mechanisms of water mist have
us
been conducted [3-11]. It has been found that the extinguishing mechanism changes according to the type of fire encountered [12].
an
Heskestad [3], Kim [11] and Liu [12] focused on the critical water
M
flow, critical spray flux and water mist characteristics required to extinguish a fire, respectively. These critical factors provide a
d
foundation for the design of water mist extinguishing systems. While
Ac ce pt e
these elements are now well-understood, no previous study has established criteria that could be used to evaluate the time required under different scenarios to extinguish a fire with a water mist system. Many full-scale experiments on the performance of water mist in
fire extinguishment and control have also been conducted [13-16]. It has been demonstrated that water mist fire-fighting systems are able to suppress a wide variety of fires and that large fires are easier to extinguish using water mist when contained in a compartment [16]. Many countries and organizations have promulgated water mist fire protection system standards or test protocols, for the individual 2 Page 3 of 30
elements of a system or for entire systems. These standards are listed in Table 1. Table 1. Water mist fire protection system standards Date (First edition)
Country or organization
IMO Res. A800(19)
1995
IMO
NFPA750
1996
NFPA
AS 4587
1999
Australia
UL 2167
2002
Underwriters' Laboratories
EN 14972
2004
ANSI/FM 5560
2007
DS/CEN/TS 14972
2008
cr
ip t
Standard
UK
us
USA
Europe
an
However, several limitations of water mist have also been reported. The extinguishing time of water mist is long compared to that of
M
gaseous agents (for example halon 1301 and heptafluoropropane), and
d
water mist systems are sometimes unable to extinguish a fire at all
Ac ce pt e
[13-17]. Specifically, water mist systems cannot extinguish small, obstructed or shielded fires [16]. Previous research has shown that large fires are easier to extinguish in an enclosed space. It may result from self-extinguishing as Back III [16] pointed out that fires could still be extinguished without any agents reaching them if the fire size is above a critical value dictated by the conditions in a confined space. According to these previous studies, it may be concluded that water mist fire extinguishing systems have variable extinguishing times that are affected by fire type, fire size, fire location, and obstruction. Consequently,
studies
on
improving
the
fire
extinguishing
performance of water mist have been conducted. Liu [17] discovered 3 Page 4 of 30
that cycling the discharge of water mist could considerably enhance the efficiency of water mist in fire suppression. The efficacy of water mist systems with inorganic additives [18, 19, 20] and AFFF
ip t
(Aqueous Film-Forming Foam) additives [21, 22] have also been examined.
cr
Statistical methods have proven to be useful and powerful tools in
us
fire safety science and are widely used in fire research. Soares [23] developed probabilistic models for offshore fires. Kong [24, 25]
an
employed Monte Carlo simulation methods to study the uncertainty of
of several variables.
M
heat fire detectors and ASET (available safe egress time) as a function
d
A water mist system is the epitome of a performance-based design.
Ac ce pt e
The design of a water mist system must be based on comprehensive fire tests conducted by an internationally recognized laboratory [26], as recommended in NFPA 750. The fire protection objective of water mist fire protection systems is fire control, fire extinguishment or fire suppression. In fact, the extinguishing characteristics of water mist in two tests using the same scenario can differ. To date, no study of the probability distribution of extinguishing times has been reported, while the uncertainty of the extinguishing time affects the reliability of water mist fire-fighting systems.
4 Page 5 of 30
In this research the extinguishing time distribution of pool fires extinguished with water mist is investigated and the extinguishing mechanisms involved in the different situations is discussed.
ip t
2. Experiment Setup
cr
2.1 Fire Sources
us
Liquid combustibles are the main fire hazards at industrial sites. Pool fires are likely the simplest form of fire hazards that threaten a
an
wide range of industries. Therefore, pool fires were selected as the
M
focus of this study.
If a fire cannot be extinguished at the beginning of applying water
d
mist, the flame in a square fuel pan will always burn in one or more
Ac ce pt e
corners of this pan. To avoid the formation of corners in the fuel pan, a circular fuel pan with a slightly lower height was also used. Previous studies[6] have demonstrated that the extinguishing time increases with increasing fire size in an open space using water mist. Therefore, the chosen circular fuel pan was larger in area than the square fuel pan. A square stainless steel pan with a side length of 250 mm and 50 mm depth was used to replicate the behavior of a fire in a rectangular open-top tank or in an isolated combustible liquid spill. A circular stainless steel pan with a diameter of 320 mm and 20 mm depth was assumed to represent a fire in a circular open-top tank or in a liquid 5 Page 6 of 30
combustible spill on the ground. The fuel pan was placed on a steel pedestal 400 mm above the ground to reduce the effects of this surrounding ground surface on the fire behavior.
ip t
The fire was allowed to burn for 30 s in case of gasoline and 40 s for the other liquid combustibles diesel, ethanol and Daqing RP-3 to
cr
ensure a quasi-steady flame before the water mist was applied.
us
2.2Water Mist
an
The pressurised spray head used in the experiments had 7 nozzles, each with an orifice diameter of 1.8 mm. The water mist was applied
M
downwards, as shown in Figure 1. The water mist nozzle was placed
d
2.9 m above the floor, which is a typical sprinkler height in a building.
Ac ce pt e
Experiments were repeated 10 to 30 times for each scenario, as shown in Table 2. After a fire was extinguished, a lighter was used to reignite the residual contents of the pan. If the contents could be ignited, the pool fire was considered to have been successfully extinguished by water mist. The pool fire extinguishing process with water mist was recorded with a digital video at a rate of 25 fps. The extinguishing times were obtained from the video records. The spray angle of a single nozzle was approximately 60°for the pressure range of 1.8–2.4 MPa. The relevant parameters of the water mist, such as droplet size and velocity, were measured using an 6 Page 7 of 30
LDV/APV system at a cross-section below the nozzle. The
us
cr
ip t
cross-section was 1 m away from the nozzle exit.
an
Figure 1. Schematic of the experimental setup Typical results of the water mist characteristics in the pressure
M
range of 1.8 to 2.4 MPa are shown in Figures 2 and 3. The water flux distribution was obtained by collecting water with
Ac ce pt e
d
cups placed in a plane 2.5 m underneath the nozzle exit, as shown in Figure 4. A total of 25 cups with a diameter of 25 mm were placed in the plane radially outwards from the center of the spray. The distance between the heart lines of two adjacent cups was 50 mm. The collection time was approximately 2 min. The collected water from the cups was measured using a balance, and the mean water flux was calculated for the experiments performed for each scenario. Results of the water flux along the radial distance are shown in Figure 5.
7 Page 8 of 30
145
1.8Mpa 2.4Mpa
135
130
125
ip t
Axial Mean Diameter (m)
140
120
110 0
50
100
150
200
cr
115
250
300
us
Radial distance from centerline (mm)
Figure 2. Radial distribution of the water mist mean diameter
an
The mean diameter of droplets generated by a nozzle increases with
M
the increase of the radial distance of the mist from the exit of the nozzle [27]. As can be seen from Figure 3, the fire extinguishing
Ac ce pt e
d
system used for this research is comprised of 7 nozzles; one central nozzle which is directed straight downward, whereas the other six are placed radially and under an angle compared to the central nozzle. In this way the measured particle size distribution as function of the radial direction has two maximum values. The first maximum is found in the centerline (0 mm), which is due to the central nozzle and the second maximum at around 150 mm which is due to the superimposition of the largest droplets generated with the 6 radial nozzles and the finer particles of the central nozzle.
8 Page 9 of 30
2.0
1.8Mpa 2.4Mpa
1.6
1.4
1.2
1.0
ip t
Axial Mean Velocity (m/s)
1.8
0.6 0
50
100
150
200
cr
0.8
250
300
us
Radial distance from centerline (mm)
Ac ce pt e
d
M
an
Figure 3. Radial distribution of the water mist axial mean velocity
Figure 4. Schematic of the water mist flux measurement
9 Page 10 of 30
5.5
2.4Mpa 1.8Mpa
4.5
4.0
ip t
Spray Density(L/(Min*m^2))
5.0
3.5
0
50
100
150
200
cr
3.0
250
300
us
Radial distance from centerline(mm)
an
Figure 5. Radial distribution of the mean water flux
2.3 Data Analysis
M
A one-sample Kolmogorov-Smirnov test is a nonparametric test
d
that can be used to compare a sample with a reference probability
Ac ce pt e
distribution. The Kolmogorov–Smirnov statistic quantifies the difference between the distribution function of the sample and the cumulative distribution function of the reference distribution. A one-sample Kolmogorov-Smirnov test was conducted to compare the sets of extinguishing times with a normal and a lognormal distribution. A so-called p-value of less than 0.05 means that the sample data have a significant deviation from the reference distribution. The Q–Q plot ("Q" stands for quantile) is a plot of the quantiles of two distributions against each other. The pattern of points in the plot is used to compare the two distributions by providing a 10 Page 11 of 30
graphical view of how properties such as location, scale, and skewness are similar or different in the two distributions.
3. Results and Discussion
ip t
3.1 Two Types of Extinguishing:
cr
Two different fire-extinguishing scenarios were observed. In one,
us
which is referred to as situation 1, the fire was suddenly extinguished at an early stage when the water mist was applied. In the other, which
an
is referred to as situation 2, the fire was eventually extinguished after
M
a long period of applying the water mist after the fire had been suppressed. Figure 6 highlights the differences between these two
d
scenarios at the stage of the pool fire-water mist interaction. The fire
Ac ce pt e
extinguishing times in both scenarios fit binomial distributions. The characteristics of situation 1 are as follows: 1) The flames had not yet been suppressed (large fluctuations in the size and spatial structure of flames), when 2) The fire was extinguished via a blow-off process, as shown in Figure 7.
The characteristics of situation 2 are as follows: 1) The flames were first suppressed (the size of the flame was reduced more than 50%, and the flame only appeared near the edge of the fuel pan), then 11 Page 12 of 30
2) The flames gradually reduced in size and were ultimately extinguished, as shown in Figure 8. After the fire was extinguished, the contents of the pans could be
ip t
re-ignited in both situations. Situation 2 actually represents a failure in extinguishing the fire. In this situation most of the oil in the fuel pan
cr
was consumed, and the remainder could not sustain combustion in
us
water. However, if the pan would have been 5 times deeper, the fire would not have been extinguished and would have continued burning.
an
Table 2 shows the distribution of extinguishing times for the two
M
scenarios. From this table, one can see that only situation 1 was observed when the pool fire occurred in the circular pan, while both
d
situations were observed for all pool fires in the square pan, with the
Ac ce pt e
exception of ethanol fires. In a diffusion flame, the fuel-to-oxidizer ratio is not constant. There is a fuel rich region near the base of pool fires. As is known, concentrations higher than the UEL (upper explosive limit) are “too rich” to burn, and the UEL of ethanol, diesel, RP-3 kerosene and gasoline are about 19%, 7.5%, 5% and 7.6%, respectively. In addition, the evaporation rate of ethanol is the lowest among these fuels (diesel, RP-3 kerosene and gasoline) [30]due to the lowest radiation feedback to fuel surface from flame. After the application of water mist, it is easier to maintain a small corner flame for an ethanol pool fire because of the higher UEL and the lowest 12 Page 13 of 30
evaporation rate. This could explain why situation 1 was not observed for square ethanol fires. From Table 2 it can be seen that type of fuel, the pressure of the
ip t
water mist system, and even the shape of the fuel pan change the probability of each of these two situations occurring. A real fire
cr
scenario is more complex than an experimental one on laboratory
us
scale. First, the fuel type involved in a fire may vary. Second, the obstructions which would shield fires are different in width, height,
an
shape and/or location. Third, the uneven ground surface which would
M
affect the shape of spilled fuel pool may make the situation more complex. All of these factors affect the outcome of the effectiveness
d
of the water mist system. It is consequently difficult to guarantee the
Ac ce pt e
fire extinguishing efficiency of water mist extinguishing systems and this poses a problem in the design of such systems. Figures 9 and 10 show the mean value and standard deviation of
the extinguishing times for each scenario. The mean value for the extinguishing times of situation 1 is dramatically smaller than that of situation 2, which implies that in the latter case the fires will be hard to extinguish after suppression, i.e., the fire cannot be extinguished at the early stages of the water mist fire-fighting processes. Previous studies have discovered that small fires are more difficult to extinguish than large fires in confined spaces[16]. This study 13 Page 14 of 30
further demonstrates that small, shielded fires are more difficult to extinguish even by directly covering the fire with water mist. The walls of the fuel pan can shield the flames after suppression. The
ip t
water mist stream and the combustible vapor stream travel in opposite directions. As such the combustible vapor will be blown away from
cr
the fuel pan by the water mist. However, some vapor was obstructed
us
by the walls and remained at high concentrations in one or more corners of the square pan. Moreover, the high temperature of the wall
an
of the pan, heated by the fire, sometimes was sustaining the
M
combustion process and hence the flames were only suppressed and not extinguished in the corners (as shown in Figure 8). This explains
d
as to why the fires in the square pan were more difficult to extinguish
Ac ce pt e
than those in the circular pan.
The fundamental reason for the significant difference in
extinguishing times between the two scenarios is that the primary mechanism involved in extinguishing the fires is different. The primary mechanism for suppressing/extinguishing flames by spraying water mist are fuel vapor dilution and flame cooling. If a local area sustains a sufficient concentration of fuel vapor after the mist has been activated, the flame will be very difficult to extinguish (situation 2). For a square fuel pan, it is difficult for water mist to reach the corners. Therefore, the fuel vapor concentration in these corners is difficult to 14 Page 15 of 30
dilute using water mist. After being suppressed, the flame is reduced in size but is sustained in and near the corners. The circular pan has no corners, so all extinguishing processes of the circular pan fire obey the
ip t
conditions as described in situation 1. It should be noted that there was a difference in the depth of the pans, as well as that the circular pan
cr
was thinner than the square pan. The deep wall can shield the small
us
suppressed flame, whereas the shallow wall is less capable in doing so. This is the reason why situation 2 was not observed in the cases of
an
extinguishing the shallow circular pool fire. It is more difficult for
M
water mist to cool a thick fuel pan than a thin one. So the thick fuel pan was more capable of maintaining its temperature and sustaining
d
the evaporation rate of the fuel while under the application of water
Ac ce pt e
mist.
Situation1
Situation2
Water mist Free Burning activated
Extinguished
Water mist activated
Suppressed
Free Burning
Extinguished
A period: 2-8 times simulation1 0
10
20
60
120
180
Figure 6. Two scenarios of water mist interaction with pool fires
15 Page 16 of 30
Free burning
10.40 s
10.48 s
10.56 s
10.80s
10.64 s
Figure 7. The blow-off phenomenon in a pool fire extinguished with water mist
Free burning
us
cr
ip t
(situation 1)
10.5 s
9s
20 s
55 s
32 s
an
Figure 8. The interaction of a diesel pool fire with water mist (situation 2) Table 2. Extinguishing/suppressing success as a function of pressure of water mist
Situation 1 ( Extingui shment ) 2 (Suppressi on) 1 ( Extingui shment ) 2 (Suppressi on)
□
2.4
Ethanol ○
10/30
12/12
4/32
20/30
28/32
30/30
○
Diesel
□
Ac ce pt e
1.8
Gasoline
d
Water mist System pressure [MPa]
M
system, fuel type and shape of fuel pan.
16/16
10/10
26/30
4/30
□
○
□
○
12/12
8/30
16/16
20/20
10/10
7/30
23/30
Kerosene (Daqing RP-3)
22/30
10/10
25/30
10/10
5/30
□ Square pan, ○ Circular pan
3.2 Variance Analysis of Fuel Types In Extinguishing Times Figure 9 shows that the order of the extinguishing times from long to short are gasoline, daqing RP-3, diesel and ethanol respectively. Flame cooling is the primary fire extinguishing mechanism of situation 1; 1600 K is often selected as the threshold temperature to 16 Page 17 of 30
predict fire extinguishment[28,29]. The energy balance of the flame at the fire extinguishing point is[12] f'' H c Q Air Q Mist S m
(1)
ip t
where m f'' is the evaporation rate of the fuel (kg/m2-sec), H c is the heat of combustion of that fuel (kJ/kg), Q Air the heat dissipated to
cr
the surrounding air, Q Mist the heat loss to the water droplets. In case
us
S<0 this means fire extinguishment, i.e. the heat generation will not be
an
sufficient to keep the flame temperature above the threshold value. From Equation (1) it can be seen that the value of m f'' H c is the
M
primary factor affecting fire extinguishment. Table 3 gives the
d
evaporation rate and heat of combustion of the fuels, the product of
Ac ce pt e
which yields the amount of energy released per unit time. The order of the product m f'' H c from high to low is gasoline, daqing RP-3, diesel and ethanol, which coincides with the same order in extinguishing times as given in Figure 9.
The reason for this consistency is that the boiling point of gasoline
is lower than that of the other fuels, as well as that the evaporation rate per unit area is higher than that of the other fuels [30]. Consequently, the possibility that a combustible vapor remains in its flammable range in one or more corners in case of gasoline is therefore highest. Based on the energy balance of the fuel surface, the evaporation rate can be given as follows: 17 Page 18 of 30
m f''
Q Q f L Lvf
(2)
is the heat transferred to Lvf is the latent heat of evaporation of the fuel and Q f the fuel surface from the flame. The latter term is comprised of conductive heat
ip t
transfer from the fuel pan, and convective and radiative heat transfer from the flame:
cr
Q Q f cond Qconv Qradi
us
The radiation attenuation is another main extinguishing mechanism of water mist. Q radi will decrease significantly after the application of water mist, which
an
leads to the decrease of m f'' . As is known, radiation feedback from the ethanol
M
flame to the liquid ethanol surface is low compared to gasoline, diesel and kerosene, because the soot generation rate of the ethanol flame is smaller[30].
Ac ce pt e
not so pronounced
d
Hence the reduction of the evaporation rate of the ethanol by the water mist was
Table 3 Fire Evaporation Rate Data[30] Items
Gasoline
Daqing RP-3*
Diesel
Ethanol
m f'' (kg/m2-sec)
0.055
0.051
0.044
0.015
H c (kJ/kg)
43,700
43,500
44,400
26,800
*The data of JP-4.
18 Page 19 of 30
Mean extinguishing time and standard deviation (s)
60
50
40
30
20
10
50
Gasoline Gasoline Diesel Diesel Ethanol Ethanol Daqing RP-3 Daqing RP-3
45 40 35 30 25 20 15 10 5
ip t
Mean extinguishing time and standard deviation (s)
Gasoline Gasoline Diesel Diesel Ethanol Daqing RP-3 Daqing RP-3
70
Figure 9 The mean extinguishing time and standard deviation of different fuels
cr
and shape of pan for the case of extinguishment (situation 1,Left: 1.8Mpa, right:
us
450
an
Gasoline 1.8Mpa Diesel 1.8MPa Diesel 2.4MPa Ethanol 1.8MPa Ethanol 2.4MPa Daqing RP-3 1.8Mpa Daqing RP-3 2.4Mpa
400 350
200 150 100
d
250
M
300
Ac ce pt e
Mean extinguishing time and standard deviation (s)
2.4Mpa, □: Square pan, ○: Circular pan)
50
Figure 10 The mean extinguishing time and standard deviation of different fuels for the square pan leading to suppression (situation 2).
19 Page 20 of 30
10
(a) 1.8Mpa water mist 6
Normal distribution Log-normal distribution
(b) 2.4Mpa water mist
Log-normal distribution Normal distribution
9 8
5
7 6
Count
Count
4
3
2
5 4 3 2
1
1 0 20
40
60
80
0
100
10
20
30
40
50
60
ip t
0
70
Extinguishing times (s)
Extinguishing times (s)
cr
Figure 11. Frequency of extinguishing times with a log-normal and normal distribution curve fit for gasoline in a square pan with full extinguishment
130
110
(a) Log-normal Q-Q Plot 1.8Mpa
120
80 70 60 50 40 30 20
0 20
30
40
50
60
70
80
90
70 60 50 40 30 20 10
0 -10
d
10 10
Expected Value Reference Line
80
90
0
(b) Normal Q-Q Plot 1.8Mpa
an
90
Expected Normal Value
100
M
Expected Lognormal Value
100
Expected Value Reference Line
110
us
(situation 1).
100 110 120 130
-10
0
10
20
Ac ce pt e
Extinguishing times (s)
30
40
50
60
70
80
90
100
110
Extinguishing times (s)
90
(c) Log-normal Q-Q Plot 2.4Mpa
(d) Normal Q-Q Plot 2.4Mpa
70
Expected Value Reference Line
70
60
Expected Normal Value
Expected Lognormal Value
80
60 50 40 30 20
0
10
20
30
40
50
60
40 30 20 10 0
10
0
Expected Value Reference Line
50
-10 70
80
90
-10
0
10
20
30
40
50
60
70
Extinguishing times (s)
Extinguishing times (s)
Figure 12. Log-normal and normal Q-Q Plot of extinguishing times of gasoline in a square pan.
20 Page 21 of 30
Table 4. The P-value of one-sample Kolmogorov-Smirnov test results with a normal and lognormal distribution (in parentheses) to determine the success rate of extinguishing a fire.
system pressure
Gasoline Situation
□
Diesel ○
□
Ethanol ○
□
[MPa]
1 2.4 2
0.866
(0.958)
(0.898)
(0.911)
(0.981)
0.94 (0.92)
0.995
----
(0.935)
----
---0.999
(0.981)
□
○
0.995
0.949
0.975
(0.976)
(0.799)
(0.712)
cr
0.832
us
2
0.947
----
0.700
-----
(0.661)
0.999
0.996
0.980
0.960
0.530
0.999
0.999
0.999
(0.960)
(0.889)
(0.910)
(0.988)
(0.856)
(0.981)
(0.879)
(0.788)
----
----
□ Square pan, ○ Circular pan
an
1.8
0.999
○
0.410
(0.585)
----
0.992 (0.973)
----
0.616 (0.532)
-----
M
1
Kerosene
ip t
Water mist
Figure 10 shows that there is a significant difference between the
Ac ce pt e
d
different fuels with respect to pool fire extinguishing times according to the second scheme (situation 2). After the fire was extinguished, the contents of the fuel pan could be re-ignited in some cases. The fire was quenched as the rate of supply of fuel vapor was reduced due to fuel surface cooling and was not sufficient to sustain the flame. Consequently, the flash point of the fuel was the primary factor that determines the extinguishing time. As is known, the flash points of gasoline, ethanol, daqing RP-3 and diesel are about -43。C, 6。C, 38。C and 52。C, respectively. Figure 10 shows that the long-short sequence of extinguishing times of the fuels is gasoline, ethanol, daqing RP-3 21 Page 22 of 30
and diesel, which is in agreement with the same low to high order of the flash points of these fuels. Other factors, such as the latent heat of the fuel, which affect the evaporation of fuel, also played a role. The
ip t
primary fire extinguishing mechanism was fuel surface cooling after the fire was suppressed, which explains why the ethanol fire, easiest to
cr
extinguish in situation 1, became more difficult to extinguish than the
us
fires of diesel and daqing RP-3 in situation 2. When the ambient
an
temperature was about 20。C larger than the flash point of the fuel (gasoline and ethanol), the extinguishment of the fire was not
M
achieved by surface cooling but by fuel depletion. This may explain
d
why there is a significant difference between the fuels with respect to
Ac ce pt e
the pool fire extinguishing times according to the second scheme (situation 2). Figure 9 also shows that there was a significant difference between gasoline on the one hand and other fuel types (diesel, ethanol, Daqing RP-3) on the other hand with respect to pool fire extinguishing times according to the first scheme (situation 1). The reason is not only the high evaporation rate of the gasoline but also its low flash point. The low boiling point helps sustain the high evaporation rate of the fuel after the application of the water mist. Figures 9 and 10 show that increasing the pressure in the water mist system decreased the mean value and standard deviation of the extinguishing times. An increase in the pressure of the water mist 22 Page 23 of 30
system increases the droplet velocity and water flux, and decreases the droplet diameter, hereby enhancing the cooling effect of the water mist of both the flame as the liquid fuel surface. From Table 2 we also
ip t
can see that with increasing water mist pressure extinguishment of situation 1 improves, whereas for situation 2 is gets worse. This is
cr
because the high spray density can diluted fuel vapor concentration
us
more effectively, and small corner flame are more difficult to sustain
an
for high spray density condition.
3.3 Probability Distribution of Extinguishing Times
M
A probability density function can be used to predict the probability
d
of a random variable falling within a certain range. So the probability
Ac ce pt e
distribution of the extinguishing times was examined to evaluate the probability of a water mist system extinguishing a fire within a given time frame. Table 4 shows the one-sample Kolmogorov-Smirnov test results for each scenario. From this table, it can be observed that the distributions of the extinguishing times of the same scenario had no significant difference with the normal distribution or the lognormal distribution. A further comparison was conducted by using frequency and Q-Q plot of extinguishing times, as shown in Figure 11 and Figure 12. From Figure 11 it can be noted that the log-normal curve fits the distribution of the extinguishing times well when extinguishing times 23 Page 24 of 30
are smaller than the mean value, while the normal curve fits the distribution well when the extinguishing times are larger than the mean value. Figure 12 shows log-normal and normal Q-Q plots of the
ip t
extinguishing times. The deciles of the distributions are shown in red. From Figure 12a and Figure 12c, it can be seen that five outliers are
cr
evident at the high end of the range; otherwise, the data fit the
us
log-normal distribution well. From Figure 12b and Figure 12d, it can be observed that some outliers are evident at both the low and high
an
end of the range. When fire safety engineers try to predict the
M
probability of a water mist system extinguishing a fire within the given time frame. Based on this, fire safety engineers may judge
d
which of the distributions can meet requirements in predicting the
Ac ce pt e
probability of a water mist system extinguishing a fire within certain time range. Log-normal distribution would be better when the range include small value and exclude larger value. While normal distribution would be better when the range include larger value and exclude small value.
4. Conclusion Two types of extinguishing situations were observed when attempting to extinguish pool fires with water mist. In one situation, the fire was abruptly extinguished via a blow off process at an early stage of water mist application, when the flames had not yet been 24 Page 25 of 30
suppressed. Here, flame cooling is the primary mechanism for fire extinguishing. The mass loss rate and combustion heat of the fuel are two key factors that affect the extinguishing time.
ip t
In the other situation, the fire was suppressed effectively at first, and extinguished after a long suppression stage; in fact the fire
cr
extinguishment failed. Here, the cooling of the fuel surface is the
us
primary mechanism for fire extinguishing. The flash point of the fuel is the key factor that affects the extinguishing time.
an
Fuel type and even fuel pan shape affected the probabilities with
M
which the two situations occurred. Pool fires in a square pan are difficult to extinguish using water mist even with complete coverage
d
of the pan by the mist. Consequently, if the fire protection objective is
Ac ce pt e
fire extinguishment, multiple tests should be performed in each scenario to guarantee the extinguishing efficiency of the water mist system. In addition, many factors affect the extinguishing efficiency of a water mist system, so the test scenario should match the application precisely.
For the fuels with a low flash point, water mist fire extinguishing
systems must be capable of extinguishing the fire through the first situation. For the fuels with a high flash point, if the water mist fire extinguishing system is incapable of extinguishing the fire through the first situation, the duration of water supply must be sufficient to cool 25 Page 26 of 30
the fuel. In addition, the nozzle installation will have to be directed towards these possible suppressed small flames, which often occur at corners or near obstructions.
ip t
For a water mist fire suppression systems, an appropriate duration of water supply for each project needs to be selected according to the
cr
required time for firefighters to arrive at the scene as fires can be
us
reignited by suppressed small flames if the water mist system depletes
an
its water storage.
M
ACKNOWLEDGEMENTS
d
The authors appreciate the support of China Postdoctoral Science
Ac ce pt e
Foundation funded project(2014M550386) and the Opening Fund of the State Key Laboratory of Fire Science at University of Science and Technology of China under grant no. HZ2012-KF01.
REFERENCES
[1] M.M. Braidech, J.A. Neale, A.F. Matson, R.E. Dufour, The mechanism of extinguishment of fire by finely divided water, Underwriters Laboratories Inc., National Board of Fire Underwriters, NY, 1955. [2] D.J. Rasbash, Z.W. Rogowski, G.W.V. Stark, Mechanism of extinction of liquid fuel fires with water sprays, Combustion and Flame. 1960, 4(3):223–234. [3] G. Heskestad, Extinction of gas and liquid pool fires with water sprays, Fire Safety Journal. 2003, 38(4):301–317.
26 Page 27 of 30
[4] X. Huang, X.S. Wang, X. Jin, G.X. Liao, J. Qin. Fire protection of heritage structures: Use of a portable water mist system under high-altitude conditions. Journal of Fire Sciences. 2007, 25(3):217-239. [5] Z.G. Liu, C. Don, A.K. Kim, Cooling characteristics of hot oil pool by water mist during fire suppression, Fire Safety Journal. 2008, 43(4):269–281. [6] X.S. Wang, G.X. Liao, J. Qin, W.C. Fan. Experimental study on effectiveness
ip t
of extinction of a pool fire with water mist. Journal of Fire Sciences. 2002, 20(4):279-295.
cr
[7] X. Huang, X.S. Wang, G.X. Liao. Characterization of an Effervescent Atomization Water Mist Nozzle and Its Fire Suppression Tests. Proceedings of
us
the Combustion Institute. 2010, 33(2):2573–2579.
[8] F. Takahashi, V.R. Katta, Extinguishment of diffusion flames around a
an
cylinder in a coaxial air stream with dilution or water mist, Proceedings of the Combustion Institute. 2009, 32(2):2615–2623.
[9] A.R. Awtry, B.T. Fisher, R.A. Moffatt, V. Ebert, J.W. Fleming. Simultaneous
M
diode laser based in situ quantification of oxygen, carbon monoxide, water vapor, and liquid water in a dense water mist environment, Proceedings of the
d
Combustion Institute. 2007, 31(1):799–806.
Ac ce pt e
[10] H.K. Chelliah, Flame inhibition/suppression by water mist: Droplet size/surface area, flame structure, and flow residence time effects, Proceedings of the Combustion Institute. 2007, 31(2):2711–2719. [11] M.B. Kim, Y.J. Jang, M.O. Yoon, Extinction Limit of a Pool Fire with a Water Mist, Fire Safety Journal. 1997, 28(4):295-306. [12] Z.G. Liu, K.K. Andrew, C. Don, A study of portable water mist fire extinguishers used for extinguishment of multiple fire types, Fire Safety Journal. 2007, 42(1):25–42.
[13] R.G. Bill, R.L. Hansen, K. Richards, Fine-Spray (Water Mist) Protection of Shipboard Engine Rooms, Fire Safety Journal. 1997, 29(4):317-336. [14] G.G. Back, Full Scale Tests of Water Mist Fire Suppression Systems for Navy Shipboard Machinery Spaces, Fire Safety Science - Proceedings of Fourth International Symposium. 1996, pp. 435-444 [15] J.H. Dyer, Water Mist Fire Suppression Systems: Application Assessment Tests on Full Scale Enclosure, Fire Engineers Journal. 1997, 57(191):35-42.
27 Page 28 of 30
[16] G.G. Back III, C.L. Beylel, R. Hansen, The Capabilities and Limitations of Total Flooding, Water Mist Fire Suppression Systems in Machinery Space Applications, Fire Technology. 2000, 36(1):8-23. [17] Z. Liu, A.K. Kim, and Z.S. Joseph, Examination of the Extinguishment Performance of a Water Mist System Using Continuous and Cycling Discharges, Fire Technology. 1999, 35(4):336-361.
ip t
[18] R. Zheng, K.N.C. Bray, B. Rogg, Effect of Sprays of Water and NaCl-Water solution on the Extinction of Laminar Premixed Methane-Air counterflow flames,
cr
Combustion Science and Technology. 1997, 126(1/6):389-401.
[19] H.K. Chelliah, A.K. Lazzarini, P.C. Wanigarathne, G.T. Linteris, Inhibition
us
of premixed and non-premixed flames with fine droplets of water and solutions, Proceedings of the Combustion Institute. 2002, 29(1):369-376.
an
[20] X.M. Zhou, G.X. Liao, B. Cai, Improvement of water mist’s fire-extinguishing efficiency with MC additive, Fire Safety Journal. 2006, 41(1):39–45.
M
[21] G. LeFort, A. Marshall, M. Pabon, Evaluation of Surfactant Enhanced Water Mist Performance, Fire Technology. 2009, 45(3):341–354.
d
[22] B.H. Cong, G.X. Liao, Experimental Studies on Water Mist Suppression of
Ac ce pt e
Liquid Fires with and without Additives, Journal of Fire Sciences. 2009, 27(2):101-123.
[23] C.G. Soares, A.P. Teixeira, Probabilistic modelling of offshore fires, Fire Safety Journal. 2000, 34(1):25-45. [24] D.P. Kong, S.X. Lu, L. Feng, Q.M. Xie. Monte Carlo analysis effect
of
heat
release
rate
uncertainty
on
of the
available safe egress time.
Journal of Fire Protection Engineering, 2012, 23(1):5-29 [25] D.P. Kong, S.X. Lu, L. Feng, Q.M. Xie. Uncertainty and sensitivity analyses of heat fire detector model based on Monte Carlo simulation. Journal of Fire Sciences, 2011, 29(4):317-337. [26] NFPA 750, Standard on Water Mist Fire Protection Systems. 2000 Edition. [27] J. Zhao, L. Yang. Simulation and Experimental Study on the Atomization Character of the Pressure-Swirl Nozzle, Open Journal of Fluid Dynamics, 2012, 2(4A): 271-277.
28 Page 29 of 30
[28] T.S. Liang, S.M. Lo, G.X. Liao, X.S. Wang. Study on Fire Extinguishing Performance of Ultra Fine Water Mist in a Cup Burner. SCIENCE CHINA Technological Sciences, 2012, 55(7):1982-1987 [29] J.G. Quintiere, A.S. Rangwala. A theory for flame extinction based upon flame temperature. Fire and Materials, 2004, 28(5):387–402
ip t
[30] I. Naeem, W. Sunil. Fire Dynamics Tools：Quantitative Fire Hazard Analysis Methods for the U.S. Nuclear Regulatory Commission Fire Protection Inspection
Ac ce pt e
d
M
an
us
Nuclear Reactor Regulation, 2004:Chapter 3,4-6
cr
Program, Washington, DC: U.S. Nuclear Regulatory Commission Office of
29 Page 30 of 30
S0957-5820(14)00079-2 http://dx.doi.org/doi:10.1016/j.psep.2014.05.009 PSEP 448
To appear in:
Process Safety and Environment Protection
Received date: Revised date: Accepted date:
18-7-2013 7-5-2014 20-5-2014
Please cite this article as: Liang, T., Liu, M., Liu, Z., Zhong, W., Xiao, X., Lo, S.,A Study of the Probability Distribution of Pool Fire Extinguishing Times using Water Mist, Process Safety and Environment Protection (2014), http://dx.doi.org/10.1016/j.psep.2014.05.009 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.
Research Highlights
Ac
ce pt
ed
M
an
us
cr
ip t
Two kinds of extinguishing situation of pool fire by water mist were observed. In one, the flames had not yet been suppressed when the fire was extinguished via a blowoff process. The primary mechanism for fire extinguishment is flame cooling in the first situation In the other one, the flame was first suppressed and then gradually reduced in size. The primary mechanism for fire extinguishment is fuel surface cooling in the second. .
Page 1 of 30
*Manuscript
A Study of the Probability Distribution of Pool Fire Extinguishing Times using Water Mist Tianshui Liang1, Mengjie Liu1, Zhonglin Liu1, Wei Zhong1*, Xiukun Xiao 2, Siuming Lo3 1
us
cr
ip t
School of Chemical Engineering and Energy, Zhengzhou University, Zhengzhou 450001, China 2 Institute of Industrial Technology of Guangzhou and Chinese Academy of Science, Guangzhou, 511458, China 3 Department of Civil and Architectural Engineering, City University of Hong Kong, Kowloon, Hong Kong, SAR China
Ac ce pt e
d
M
an
ABSTRACT: Water mist, a replacement for Halon gaseous agents in fire fighting, has been studied for decades. However, the fire-extinguishing reliability of water mist is debated. For example, there are significant differences in extinguishing times between tests conducted under the same conditions, and water mists have difficulty extinguishing small fires. To date, no study of the probability distribution of extinguishing times has been reported. In this study a statistical analysis of the extinguishing time distribution of pool fires extinguished using water mist is presented. The fire sources were circular/square stainless steel pans with gasoline, diesel, ethanol or Daqing RP-3 as fuel. Two types of extinguishing scenarios were observed. In one situation, the fire was extinguished via a blow off process, when the flames had not yet been suppressed. Flame cooling is the primary fire extinguishing mechanism; the mass loss rate and combustion heat of the fuel are two key factors. In the other situation, the fire was initially suppressed and subsequently extinguished after a long suppression stage. Surface cooling is the primary fire extinguishing mechanism; the flash point of the fuel is the key factor. KEYWORDS: Pool Fires; Water Mist; Fire Extinguishment; Flame cooling; Surface cooling; Variance Analysis
*
Corresponding author, [email protected], TEL: 86-371-67739005, FAX:
86-371-67781801
1 Page 2 of 30
1. Introduction Fire extinguishment and suppression using fixed water mist systems have been studied for 60 years [1, 2]. Interest in this subject renewed
environmental
after
halogen-based
reasons.
Notable
agents
qualitative
were
banned
for
ip t
was
and
quantitative
cr
investigations of the fire extinguishing mechanisms of water mist have
us
been conducted [3-11]. It has been found that the extinguishing mechanism changes according to the type of fire encountered [12].
an
Heskestad [3], Kim [11] and Liu [12] focused on the critical water
M
flow, critical spray flux and water mist characteristics required to extinguish a fire, respectively. These critical factors provide a
d
foundation for the design of water mist extinguishing systems. While
Ac ce pt e
these elements are now well-understood, no previous study has established criteria that could be used to evaluate the time required under different scenarios to extinguish a fire with a water mist system. Many full-scale experiments on the performance of water mist in
fire extinguishment and control have also been conducted [13-16]. It has been demonstrated that water mist fire-fighting systems are able to suppress a wide variety of fires and that large fires are easier to extinguish using water mist when contained in a compartment [16]. Many countries and organizations have promulgated water mist fire protection system standards or test protocols, for the individual 2 Page 3 of 30
elements of a system or for entire systems. These standards are listed in Table 1. Table 1. Water mist fire protection system standards Date (First edition)
Country or organization
IMO Res. A800(19)
1995
IMO
NFPA750
1996
NFPA
AS 4587
1999
Australia
UL 2167
2002
Underwriters' Laboratories
EN 14972
2004
ANSI/FM 5560
2007
DS/CEN/TS 14972
2008
cr
ip t
Standard
UK
us
USA
Europe
an
However, several limitations of water mist have also been reported. The extinguishing time of water mist is long compared to that of
M
gaseous agents (for example halon 1301 and heptafluoropropane), and
d
water mist systems are sometimes unable to extinguish a fire at all
Ac ce pt e
[13-17]. Specifically, water mist systems cannot extinguish small, obstructed or shielded fires [16]. Previous research has shown that large fires are easier to extinguish in an enclosed space. It may result from self-extinguishing as Back III [16] pointed out that fires could still be extinguished without any agents reaching them if the fire size is above a critical value dictated by the conditions in a confined space. According to these previous studies, it may be concluded that water mist fire extinguishing systems have variable extinguishing times that are affected by fire type, fire size, fire location, and obstruction. Consequently,
studies
on
improving
the
fire
extinguishing
performance of water mist have been conducted. Liu [17] discovered 3 Page 4 of 30
that cycling the discharge of water mist could considerably enhance the efficiency of water mist in fire suppression. The efficacy of water mist systems with inorganic additives [18, 19, 20] and AFFF
ip t
(Aqueous Film-Forming Foam) additives [21, 22] have also been examined.
cr
Statistical methods have proven to be useful and powerful tools in
us
fire safety science and are widely used in fire research. Soares [23] developed probabilistic models for offshore fires. Kong [24, 25]
an
employed Monte Carlo simulation methods to study the uncertainty of
of several variables.
M
heat fire detectors and ASET (available safe egress time) as a function
d
A water mist system is the epitome of a performance-based design.
Ac ce pt e
The design of a water mist system must be based on comprehensive fire tests conducted by an internationally recognized laboratory [26], as recommended in NFPA 750. The fire protection objective of water mist fire protection systems is fire control, fire extinguishment or fire suppression. In fact, the extinguishing characteristics of water mist in two tests using the same scenario can differ. To date, no study of the probability distribution of extinguishing times has been reported, while the uncertainty of the extinguishing time affects the reliability of water mist fire-fighting systems.
4 Page 5 of 30
In this research the extinguishing time distribution of pool fires extinguished with water mist is investigated and the extinguishing mechanisms involved in the different situations is discussed.
ip t
2. Experiment Setup
cr
2.1 Fire Sources
us
Liquid combustibles are the main fire hazards at industrial sites. Pool fires are likely the simplest form of fire hazards that threaten a
an
wide range of industries. Therefore, pool fires were selected as the
M
focus of this study.
If a fire cannot be extinguished at the beginning of applying water
d
mist, the flame in a square fuel pan will always burn in one or more
Ac ce pt e
corners of this pan. To avoid the formation of corners in the fuel pan, a circular fuel pan with a slightly lower height was also used. Previous studies[6] have demonstrated that the extinguishing time increases with increasing fire size in an open space using water mist. Therefore, the chosen circular fuel pan was larger in area than the square fuel pan. A square stainless steel pan with a side length of 250 mm and 50 mm depth was used to replicate the behavior of a fire in a rectangular open-top tank or in an isolated combustible liquid spill. A circular stainless steel pan with a diameter of 320 mm and 20 mm depth was assumed to represent a fire in a circular open-top tank or in a liquid 5 Page 6 of 30
combustible spill on the ground. The fuel pan was placed on a steel pedestal 400 mm above the ground to reduce the effects of this surrounding ground surface on the fire behavior.
ip t
The fire was allowed to burn for 30 s in case of gasoline and 40 s for the other liquid combustibles diesel, ethanol and Daqing RP-3 to
cr
ensure a quasi-steady flame before the water mist was applied.
us
2.2Water Mist
an
The pressurised spray head used in the experiments had 7 nozzles, each with an orifice diameter of 1.8 mm. The water mist was applied
M
downwards, as shown in Figure 1. The water mist nozzle was placed
d
2.9 m above the floor, which is a typical sprinkler height in a building.
Ac ce pt e
Experiments were repeated 10 to 30 times for each scenario, as shown in Table 2. After a fire was extinguished, a lighter was used to reignite the residual contents of the pan. If the contents could be ignited, the pool fire was considered to have been successfully extinguished by water mist. The pool fire extinguishing process with water mist was recorded with a digital video at a rate of 25 fps. The extinguishing times were obtained from the video records. The spray angle of a single nozzle was approximately 60°for the pressure range of 1.8–2.4 MPa. The relevant parameters of the water mist, such as droplet size and velocity, were measured using an 6 Page 7 of 30
LDV/APV system at a cross-section below the nozzle. The
us
cr
ip t
cross-section was 1 m away from the nozzle exit.
an
Figure 1. Schematic of the experimental setup Typical results of the water mist characteristics in the pressure
M
range of 1.8 to 2.4 MPa are shown in Figures 2 and 3. The water flux distribution was obtained by collecting water with
Ac ce pt e
d
cups placed in a plane 2.5 m underneath the nozzle exit, as shown in Figure 4. A total of 25 cups with a diameter of 25 mm were placed in the plane radially outwards from the center of the spray. The distance between the heart lines of two adjacent cups was 50 mm. The collection time was approximately 2 min. The collected water from the cups was measured using a balance, and the mean water flux was calculated for the experiments performed for each scenario. Results of the water flux along the radial distance are shown in Figure 5.
7 Page 8 of 30
145
1.8Mpa 2.4Mpa
135
130
125
ip t
Axial Mean Diameter (m)
140
120
110 0
50
100
150
200
cr
115
250
300
us
Radial distance from centerline (mm)
Figure 2. Radial distribution of the water mist mean diameter
an
The mean diameter of droplets generated by a nozzle increases with
M
the increase of the radial distance of the mist from the exit of the nozzle [27]. As can be seen from Figure 3, the fire extinguishing
Ac ce pt e
d
system used for this research is comprised of 7 nozzles; one central nozzle which is directed straight downward, whereas the other six are placed radially and under an angle compared to the central nozzle. In this way the measured particle size distribution as function of the radial direction has two maximum values. The first maximum is found in the centerline (0 mm), which is due to the central nozzle and the second maximum at around 150 mm which is due to the superimposition of the largest droplets generated with the 6 radial nozzles and the finer particles of the central nozzle.
8 Page 9 of 30
2.0
1.8Mpa 2.4Mpa
1.6
1.4
1.2
1.0
ip t
Axial Mean Velocity (m/s)
1.8
0.6 0
50
100
150
200
cr
0.8
250
300
us
Radial distance from centerline (mm)
Ac ce pt e
d
M
an
Figure 3. Radial distribution of the water mist axial mean velocity
Figure 4. Schematic of the water mist flux measurement
9 Page 10 of 30
5.5
2.4Mpa 1.8Mpa
4.5
4.0
ip t
Spray Density(L/(Min*m^2))
5.0
3.5
0
50
100
150
200
cr
3.0
250
300
us
Radial distance from centerline(mm)
an
Figure 5. Radial distribution of the mean water flux
2.3 Data Analysis
M
A one-sample Kolmogorov-Smirnov test is a nonparametric test
d
that can be used to compare a sample with a reference probability
Ac ce pt e
distribution. The Kolmogorov–Smirnov statistic quantifies the difference between the distribution function of the sample and the cumulative distribution function of the reference distribution. A one-sample Kolmogorov-Smirnov test was conducted to compare the sets of extinguishing times with a normal and a lognormal distribution. A so-called p-value of less than 0.05 means that the sample data have a significant deviation from the reference distribution. The Q–Q plot ("Q" stands for quantile) is a plot of the quantiles of two distributions against each other. The pattern of points in the plot is used to compare the two distributions by providing a 10 Page 11 of 30
graphical view of how properties such as location, scale, and skewness are similar or different in the two distributions.
3. Results and Discussion
ip t
3.1 Two Types of Extinguishing:
cr
Two different fire-extinguishing scenarios were observed. In one,
us
which is referred to as situation 1, the fire was suddenly extinguished at an early stage when the water mist was applied. In the other, which
an
is referred to as situation 2, the fire was eventually extinguished after
M
a long period of applying the water mist after the fire had been suppressed. Figure 6 highlights the differences between these two
d
scenarios at the stage of the pool fire-water mist interaction. The fire
Ac ce pt e
extinguishing times in both scenarios fit binomial distributions. The characteristics of situation 1 are as follows: 1) The flames had not yet been suppressed (large fluctuations in the size and spatial structure of flames), when 2) The fire was extinguished via a blow-off process, as shown in Figure 7.
The characteristics of situation 2 are as follows: 1) The flames were first suppressed (the size of the flame was reduced more than 50%, and the flame only appeared near the edge of the fuel pan), then 11 Page 12 of 30
2) The flames gradually reduced in size and were ultimately extinguished, as shown in Figure 8. After the fire was extinguished, the contents of the pans could be
ip t
re-ignited in both situations. Situation 2 actually represents a failure in extinguishing the fire. In this situation most of the oil in the fuel pan
cr
was consumed, and the remainder could not sustain combustion in
us
water. However, if the pan would have been 5 times deeper, the fire would not have been extinguished and would have continued burning.
an
Table 2 shows the distribution of extinguishing times for the two
M
scenarios. From this table, one can see that only situation 1 was observed when the pool fire occurred in the circular pan, while both
d
situations were observed for all pool fires in the square pan, with the
Ac ce pt e
exception of ethanol fires. In a diffusion flame, the fuel-to-oxidizer ratio is not constant. There is a fuel rich region near the base of pool fires. As is known, concentrations higher than the UEL (upper explosive limit) are “too rich” to burn, and the UEL of ethanol, diesel, RP-3 kerosene and gasoline are about 19%, 7.5%, 5% and 7.6%, respectively. In addition, the evaporation rate of ethanol is the lowest among these fuels (diesel, RP-3 kerosene and gasoline) [30]due to the lowest radiation feedback to fuel surface from flame. After the application of water mist, it is easier to maintain a small corner flame for an ethanol pool fire because of the higher UEL and the lowest 12 Page 13 of 30
evaporation rate. This could explain why situation 1 was not observed for square ethanol fires. From Table 2 it can be seen that type of fuel, the pressure of the
ip t
water mist system, and even the shape of the fuel pan change the probability of each of these two situations occurring. A real fire
cr
scenario is more complex than an experimental one on laboratory
us
scale. First, the fuel type involved in a fire may vary. Second, the obstructions which would shield fires are different in width, height,
an
shape and/or location. Third, the uneven ground surface which would
M
affect the shape of spilled fuel pool may make the situation more complex. All of these factors affect the outcome of the effectiveness
d
of the water mist system. It is consequently difficult to guarantee the
Ac ce pt e
fire extinguishing efficiency of water mist extinguishing systems and this poses a problem in the design of such systems. Figures 9 and 10 show the mean value and standard deviation of
the extinguishing times for each scenario. The mean value for the extinguishing times of situation 1 is dramatically smaller than that of situation 2, which implies that in the latter case the fires will be hard to extinguish after suppression, i.e., the fire cannot be extinguished at the early stages of the water mist fire-fighting processes. Previous studies have discovered that small fires are more difficult to extinguish than large fires in confined spaces[16]. This study 13 Page 14 of 30
further demonstrates that small, shielded fires are more difficult to extinguish even by directly covering the fire with water mist. The walls of the fuel pan can shield the flames after suppression. The
ip t
water mist stream and the combustible vapor stream travel in opposite directions. As such the combustible vapor will be blown away from
cr
the fuel pan by the water mist. However, some vapor was obstructed
us
by the walls and remained at high concentrations in one or more corners of the square pan. Moreover, the high temperature of the wall
an
of the pan, heated by the fire, sometimes was sustaining the
M
combustion process and hence the flames were only suppressed and not extinguished in the corners (as shown in Figure 8). This explains
d
as to why the fires in the square pan were more difficult to extinguish
Ac ce pt e
than those in the circular pan.
The fundamental reason for the significant difference in
extinguishing times between the two scenarios is that the primary mechanism involved in extinguishing the fires is different. The primary mechanism for suppressing/extinguishing flames by spraying water mist are fuel vapor dilution and flame cooling. If a local area sustains a sufficient concentration of fuel vapor after the mist has been activated, the flame will be very difficult to extinguish (situation 2). For a square fuel pan, it is difficult for water mist to reach the corners. Therefore, the fuel vapor concentration in these corners is difficult to 14 Page 15 of 30
dilute using water mist. After being suppressed, the flame is reduced in size but is sustained in and near the corners. The circular pan has no corners, so all extinguishing processes of the circular pan fire obey the
ip t
conditions as described in situation 1. It should be noted that there was a difference in the depth of the pans, as well as that the circular pan
cr
was thinner than the square pan. The deep wall can shield the small
us
suppressed flame, whereas the shallow wall is less capable in doing so. This is the reason why situation 2 was not observed in the cases of
an
extinguishing the shallow circular pool fire. It is more difficult for
M
water mist to cool a thick fuel pan than a thin one. So the thick fuel pan was more capable of maintaining its temperature and sustaining
d
the evaporation rate of the fuel while under the application of water
Ac ce pt e
mist.
Situation1
Situation2
Water mist Free Burning activated
Extinguished
Water mist activated
Suppressed
Free Burning
Extinguished
A period: 2-8 times simulation1 0
10
20
60
120
180
Figure 6. Two scenarios of water mist interaction with pool fires
15 Page 16 of 30
Free burning
10.40 s
10.48 s
10.56 s
10.80s
10.64 s
Figure 7. The blow-off phenomenon in a pool fire extinguished with water mist
Free burning
us
cr
ip t
(situation 1)
10.5 s
9s
20 s
55 s
32 s
an
Figure 8. The interaction of a diesel pool fire with water mist (situation 2) Table 2. Extinguishing/suppressing success as a function of pressure of water mist
Situation 1 ( Extingui shment ) 2 (Suppressi on) 1 ( Extingui shment ) 2 (Suppressi on)
□
2.4
Ethanol ○
10/30
12/12
4/32
20/30
28/32
30/30
○
Diesel
□
Ac ce pt e
1.8
Gasoline
d
Water mist System pressure [MPa]
M
system, fuel type and shape of fuel pan.
16/16
10/10
26/30
4/30
□
○
□
○
12/12
8/30
16/16
20/20
10/10
7/30
23/30
Kerosene (Daqing RP-3)
22/30
10/10
25/30
10/10
5/30
□ Square pan, ○ Circular pan
3.2 Variance Analysis of Fuel Types In Extinguishing Times Figure 9 shows that the order of the extinguishing times from long to short are gasoline, daqing RP-3, diesel and ethanol respectively. Flame cooling is the primary fire extinguishing mechanism of situation 1; 1600 K is often selected as the threshold temperature to 16 Page 17 of 30
predict fire extinguishment[28,29]. The energy balance of the flame at the fire extinguishing point is[12] f'' H c Q Air Q Mist S m
(1)
ip t
where m f'' is the evaporation rate of the fuel (kg/m2-sec), H c is the heat of combustion of that fuel (kJ/kg), Q Air the heat dissipated to
cr
the surrounding air, Q Mist the heat loss to the water droplets. In case
us
S<0 this means fire extinguishment, i.e. the heat generation will not be
an
sufficient to keep the flame temperature above the threshold value. From Equation (1) it can be seen that the value of m f'' H c is the
M
primary factor affecting fire extinguishment. Table 3 gives the
d
evaporation rate and heat of combustion of the fuels, the product of
Ac ce pt e
which yields the amount of energy released per unit time. The order of the product m f'' H c from high to low is gasoline, daqing RP-3, diesel and ethanol, which coincides with the same order in extinguishing times as given in Figure 9.
The reason for this consistency is that the boiling point of gasoline
is lower than that of the other fuels, as well as that the evaporation rate per unit area is higher than that of the other fuels [30]. Consequently, the possibility that a combustible vapor remains in its flammable range in one or more corners in case of gasoline is therefore highest. Based on the energy balance of the fuel surface, the evaporation rate can be given as follows: 17 Page 18 of 30
m f''
Q Q f L Lvf
(2)
is the heat transferred to Lvf is the latent heat of evaporation of the fuel and Q f the fuel surface from the flame. The latter term is comprised of conductive heat
ip t
transfer from the fuel pan, and convective and radiative heat transfer from the flame:
cr
Q Q f cond Qconv Qradi
us
The radiation attenuation is another main extinguishing mechanism of water mist. Q radi will decrease significantly after the application of water mist, which
an
leads to the decrease of m f'' . As is known, radiation feedback from the ethanol
M
flame to the liquid ethanol surface is low compared to gasoline, diesel and kerosene, because the soot generation rate of the ethanol flame is smaller[30].
Ac ce pt e
not so pronounced
d
Hence the reduction of the evaporation rate of the ethanol by the water mist was
Table 3 Fire Evaporation Rate Data[30] Items
Gasoline
Daqing RP-3*
Diesel
Ethanol
m f'' (kg/m2-sec)
0.055
0.051
0.044
0.015
H c (kJ/kg)
43,700
43,500
44,400
26,800
*The data of JP-4.
18 Page 19 of 30
Mean extinguishing time and standard deviation (s)
60
50
40
30
20
10
50
Gasoline Gasoline Diesel Diesel Ethanol Ethanol Daqing RP-3 Daqing RP-3
45 40 35 30 25 20 15 10 5
ip t
Mean extinguishing time and standard deviation (s)
Gasoline Gasoline Diesel Diesel Ethanol Daqing RP-3 Daqing RP-3
70
Figure 9 The mean extinguishing time and standard deviation of different fuels
cr
and shape of pan for the case of extinguishment (situation 1,Left: 1.8Mpa, right:
us
450
an
Gasoline 1.8Mpa Diesel 1.8MPa Diesel 2.4MPa Ethanol 1.8MPa Ethanol 2.4MPa Daqing RP-3 1.8Mpa Daqing RP-3 2.4Mpa
400 350
200 150 100
d
250
M
300
Ac ce pt e
Mean extinguishing time and standard deviation (s)
2.4Mpa, □: Square pan, ○: Circular pan)
50
Figure 10 The mean extinguishing time and standard deviation of different fuels for the square pan leading to suppression (situation 2).
19 Page 20 of 30
10
(a) 1.8Mpa water mist 6
Normal distribution Log-normal distribution
(b) 2.4Mpa water mist
Log-normal distribution Normal distribution
9 8
5
7 6
Count
Count
4
3
2
5 4 3 2
1
1 0 20
40
60
80
0
100
10
20
30
40
50
60
ip t
0
70
Extinguishing times (s)
Extinguishing times (s)
cr
Figure 11. Frequency of extinguishing times with a log-normal and normal distribution curve fit for gasoline in a square pan with full extinguishment
130
110
(a) Log-normal Q-Q Plot 1.8Mpa
120
80 70 60 50 40 30 20
0 20
30
40
50
60
70
80
90
70 60 50 40 30 20 10
0 -10
d
10 10
Expected Value Reference Line
80
90
0
(b) Normal Q-Q Plot 1.8Mpa
an
90
Expected Normal Value
100
M
Expected Lognormal Value
100
Expected Value Reference Line
110
us
(situation 1).
100 110 120 130
-10
0
10
20
Ac ce pt e
Extinguishing times (s)
30
40
50
60
70
80
90
100
110
Extinguishing times (s)
90
(c) Log-normal Q-Q Plot 2.4Mpa
(d) Normal Q-Q Plot 2.4Mpa
70
Expected Value Reference Line
70
60
Expected Normal Value
Expected Lognormal Value
80
60 50 40 30 20
0
10
20
30
40
50
60
40 30 20 10 0
10
0
Expected Value Reference Line
50
-10 70
80
90
-10
0
10
20
30
40
50
60
70
Extinguishing times (s)
Extinguishing times (s)
Figure 12. Log-normal and normal Q-Q Plot of extinguishing times of gasoline in a square pan.
20 Page 21 of 30
Table 4. The P-value of one-sample Kolmogorov-Smirnov test results with a normal and lognormal distribution (in parentheses) to determine the success rate of extinguishing a fire.
system pressure
Gasoline Situation
□
Diesel ○
□
Ethanol ○
□
[MPa]
1 2.4 2
0.866
(0.958)
(0.898)
(0.911)
(0.981)
0.94 (0.92)
0.995
----
(0.935)
----
---0.999
(0.981)
□
○
0.995
0.949
0.975
(0.976)
(0.799)
(0.712)
cr
0.832
us
2
0.947
----
0.700
-----
(0.661)
0.999
0.996
0.980
0.960
0.530
0.999
0.999
0.999
(0.960)
(0.889)
(0.910)
(0.988)
(0.856)
(0.981)
(0.879)
(0.788)
----
----
□ Square pan, ○ Circular pan
an
1.8
0.999
○
0.410
(0.585)
----
0.992 (0.973)
----
0.616 (0.532)
-----
M
1
Kerosene
ip t
Water mist
Figure 10 shows that there is a significant difference between the
Ac ce pt e
d
different fuels with respect to pool fire extinguishing times according to the second scheme (situation 2). After the fire was extinguished, the contents of the fuel pan could be re-ignited in some cases. The fire was quenched as the rate of supply of fuel vapor was reduced due to fuel surface cooling and was not sufficient to sustain the flame. Consequently, the flash point of the fuel was the primary factor that determines the extinguishing time. As is known, the flash points of gasoline, ethanol, daqing RP-3 and diesel are about -43。C, 6。C, 38。C and 52。C, respectively. Figure 10 shows that the long-short sequence of extinguishing times of the fuels is gasoline, ethanol, daqing RP-3 21 Page 22 of 30
and diesel, which is in agreement with the same low to high order of the flash points of these fuels. Other factors, such as the latent heat of the fuel, which affect the evaporation of fuel, also played a role. The
ip t
primary fire extinguishing mechanism was fuel surface cooling after the fire was suppressed, which explains why the ethanol fire, easiest to
cr
extinguish in situation 1, became more difficult to extinguish than the
us
fires of diesel and daqing RP-3 in situation 2. When the ambient
an
temperature was about 20。C larger than the flash point of the fuel (gasoline and ethanol), the extinguishment of the fire was not
M
achieved by surface cooling but by fuel depletion. This may explain
d
why there is a significant difference between the fuels with respect to
Ac ce pt e
the pool fire extinguishing times according to the second scheme (situation 2). Figure 9 also shows that there was a significant difference between gasoline on the one hand and other fuel types (diesel, ethanol, Daqing RP-3) on the other hand with respect to pool fire extinguishing times according to the first scheme (situation 1). The reason is not only the high evaporation rate of the gasoline but also its low flash point. The low boiling point helps sustain the high evaporation rate of the fuel after the application of the water mist. Figures 9 and 10 show that increasing the pressure in the water mist system decreased the mean value and standard deviation of the extinguishing times. An increase in the pressure of the water mist 22 Page 23 of 30
system increases the droplet velocity and water flux, and decreases the droplet diameter, hereby enhancing the cooling effect of the water mist of both the flame as the liquid fuel surface. From Table 2 we also
ip t
can see that with increasing water mist pressure extinguishment of situation 1 improves, whereas for situation 2 is gets worse. This is
cr
because the high spray density can diluted fuel vapor concentration
us
more effectively, and small corner flame are more difficult to sustain
an
for high spray density condition.
3.3 Probability Distribution of Extinguishing Times
M
A probability density function can be used to predict the probability
d
of a random variable falling within a certain range. So the probability
Ac ce pt e
distribution of the extinguishing times was examined to evaluate the probability of a water mist system extinguishing a fire within a given time frame. Table 4 shows the one-sample Kolmogorov-Smirnov test results for each scenario. From this table, it can be observed that the distributions of the extinguishing times of the same scenario had no significant difference with the normal distribution or the lognormal distribution. A further comparison was conducted by using frequency and Q-Q plot of extinguishing times, as shown in Figure 11 and Figure 12. From Figure 11 it can be noted that the log-normal curve fits the distribution of the extinguishing times well when extinguishing times 23 Page 24 of 30
are smaller than the mean value, while the normal curve fits the distribution well when the extinguishing times are larger than the mean value. Figure 12 shows log-normal and normal Q-Q plots of the
ip t
extinguishing times. The deciles of the distributions are shown in red. From Figure 12a and Figure 12c, it can be seen that five outliers are
cr
evident at the high end of the range; otherwise, the data fit the
us
log-normal distribution well. From Figure 12b and Figure 12d, it can be observed that some outliers are evident at both the low and high
an
end of the range. When fire safety engineers try to predict the
M
probability of a water mist system extinguishing a fire within the given time frame. Based on this, fire safety engineers may judge
d
which of the distributions can meet requirements in predicting the
Ac ce pt e
probability of a water mist system extinguishing a fire within certain time range. Log-normal distribution would be better when the range include small value and exclude larger value. While normal distribution would be better when the range include larger value and exclude small value.
4. Conclusion Two types of extinguishing situations were observed when attempting to extinguish pool fires with water mist. In one situation, the fire was abruptly extinguished via a blow off process at an early stage of water mist application, when the flames had not yet been 24 Page 25 of 30
suppressed. Here, flame cooling is the primary mechanism for fire extinguishing. The mass loss rate and combustion heat of the fuel are two key factors that affect the extinguishing time.
ip t
In the other situation, the fire was suppressed effectively at first, and extinguished after a long suppression stage; in fact the fire
cr
extinguishment failed. Here, the cooling of the fuel surface is the
us
primary mechanism for fire extinguishing. The flash point of the fuel is the key factor that affects the extinguishing time.
an
Fuel type and even fuel pan shape affected the probabilities with
M
which the two situations occurred. Pool fires in a square pan are difficult to extinguish using water mist even with complete coverage
d
of the pan by the mist. Consequently, if the fire protection objective is
Ac ce pt e
fire extinguishment, multiple tests should be performed in each scenario to guarantee the extinguishing efficiency of the water mist system. In addition, many factors affect the extinguishing efficiency of a water mist system, so the test scenario should match the application precisely.
For the fuels with a low flash point, water mist fire extinguishing
systems must be capable of extinguishing the fire through the first situation. For the fuels with a high flash point, if the water mist fire extinguishing system is incapable of extinguishing the fire through the first situation, the duration of water supply must be sufficient to cool 25 Page 26 of 30
the fuel. In addition, the nozzle installation will have to be directed towards these possible suppressed small flames, which often occur at corners or near obstructions.
ip t
For a water mist fire suppression systems, an appropriate duration of water supply for each project needs to be selected according to the
cr
required time for firefighters to arrive at the scene as fires can be
us
reignited by suppressed small flames if the water mist system depletes
an
its water storage.
M
ACKNOWLEDGEMENTS
d
The authors appreciate the support of China Postdoctoral Science
Ac ce pt e
Foundation funded project(2014M550386) and the Opening Fund of the State Key Laboratory of Fire Science at University of Science and Technology of China under grant no. HZ2012-KF01.
REFERENCES
[1] M.M. Braidech, J.A. Neale, A.F. Matson, R.E. Dufour, The mechanism of extinguishment of fire by finely divided water, Underwriters Laboratories Inc., National Board of Fire Underwriters, NY, 1955. [2] D.J. Rasbash, Z.W. Rogowski, G.W.V. Stark, Mechanism of extinction of liquid fuel fires with water sprays, Combustion and Flame. 1960, 4(3):223–234. [3] G. Heskestad, Extinction of gas and liquid pool fires with water sprays, Fire Safety Journal. 2003, 38(4):301–317.
26 Page 27 of 30
[4] X. Huang, X.S. Wang, X. Jin, G.X. Liao, J. Qin. Fire protection of heritage structures: Use of a portable water mist system under high-altitude conditions. Journal of Fire Sciences. 2007, 25(3):217-239. [5] Z.G. Liu, C. Don, A.K. Kim, Cooling characteristics of hot oil pool by water mist during fire suppression, Fire Safety Journal. 2008, 43(4):269–281. [6] X.S. Wang, G.X. Liao, J. Qin, W.C. Fan. Experimental study on effectiveness
ip t
of extinction of a pool fire with water mist. Journal of Fire Sciences. 2002, 20(4):279-295.
cr
[7] X. Huang, X.S. Wang, G.X. Liao. Characterization of an Effervescent Atomization Water Mist Nozzle and Its Fire Suppression Tests. Proceedings of
us
the Combustion Institute. 2010, 33(2):2573–2579.
[8] F. Takahashi, V.R. Katta, Extinguishment of diffusion flames around a
an
cylinder in a coaxial air stream with dilution or water mist, Proceedings of the Combustion Institute. 2009, 32(2):2615–2623.
[9] A.R. Awtry, B.T. Fisher, R.A. Moffatt, V. Ebert, J.W. Fleming. Simultaneous
M
diode laser based in situ quantification of oxygen, carbon monoxide, water vapor, and liquid water in a dense water mist environment, Proceedings of the
d
Combustion Institute. 2007, 31(1):799–806.
Ac ce pt e
[10] H.K. Chelliah, Flame inhibition/suppression by water mist: Droplet size/surface area, flame structure, and flow residence time effects, Proceedings of the Combustion Institute. 2007, 31(2):2711–2719. [11] M.B. Kim, Y.J. Jang, M.O. Yoon, Extinction Limit of a Pool Fire with a Water Mist, Fire Safety Journal. 1997, 28(4):295-306. [12] Z.G. Liu, K.K. Andrew, C. Don, A study of portable water mist fire extinguishers used for extinguishment of multiple fire types, Fire Safety Journal. 2007, 42(1):25–42.
[13] R.G. Bill, R.L. Hansen, K. Richards, Fine-Spray (Water Mist) Protection of Shipboard Engine Rooms, Fire Safety Journal. 1997, 29(4):317-336. [14] G.G. Back, Full Scale Tests of Water Mist Fire Suppression Systems for Navy Shipboard Machinery Spaces, Fire Safety Science - Proceedings of Fourth International Symposium. 1996, pp. 435-444 [15] J.H. Dyer, Water Mist Fire Suppression Systems: Application Assessment Tests on Full Scale Enclosure, Fire Engineers Journal. 1997, 57(191):35-42.
27 Page 28 of 30
[16] G.G. Back III, C.L. Beylel, R. Hansen, The Capabilities and Limitations of Total Flooding, Water Mist Fire Suppression Systems in Machinery Space Applications, Fire Technology. 2000, 36(1):8-23. [17] Z. Liu, A.K. Kim, and Z.S. Joseph, Examination of the Extinguishment Performance of a Water Mist System Using Continuous and Cycling Discharges, Fire Technology. 1999, 35(4):336-361.
ip t
[18] R. Zheng, K.N.C. Bray, B. Rogg, Effect of Sprays of Water and NaCl-Water solution on the Extinction of Laminar Premixed Methane-Air counterflow flames,
cr
Combustion Science and Technology. 1997, 126(1/6):389-401.
[19] H.K. Chelliah, A.K. Lazzarini, P.C. Wanigarathne, G.T. Linteris, Inhibition
us
of premixed and non-premixed flames with fine droplets of water and solutions, Proceedings of the Combustion Institute. 2002, 29(1):369-376.
an
[20] X.M. Zhou, G.X. Liao, B. Cai, Improvement of water mist’s fire-extinguishing efficiency with MC additive, Fire Safety Journal. 2006, 41(1):39–45.
M
[21] G. LeFort, A. Marshall, M. Pabon, Evaluation of Surfactant Enhanced Water Mist Performance, Fire Technology. 2009, 45(3):341–354.
d
[22] B.H. Cong, G.X. Liao, Experimental Studies on Water Mist Suppression of
Ac ce pt e
Liquid Fires with and without Additives, Journal of Fire Sciences. 2009, 27(2):101-123.
[23] C.G. Soares, A.P. Teixeira, Probabilistic modelling of offshore fires, Fire Safety Journal. 2000, 34(1):25-45. [24] D.P. Kong, S.X. Lu, L. Feng, Q.M. Xie. Monte Carlo analysis effect
of
heat
release
rate
uncertainty
on
of the
available safe egress time.
Journal of Fire Protection Engineering, 2012, 23(1):5-29 [25] D.P. Kong, S.X. Lu, L. Feng, Q.M. Xie. Uncertainty and sensitivity analyses of heat fire detector model based on Monte Carlo simulation. Journal of Fire Sciences, 2011, 29(4):317-337. [26] NFPA 750, Standard on Water Mist Fire Protection Systems. 2000 Edition. [27] J. Zhao, L. Yang. Simulation and Experimental Study on the Atomization Character of the Pressure-Swirl Nozzle, Open Journal of Fluid Dynamics, 2012, 2(4A): 271-277.
28 Page 29 of 30
[28] T.S. Liang, S.M. Lo, G.X. Liao, X.S. Wang. Study on Fire Extinguishing Performance of Ultra Fine Water Mist in a Cup Burner. SCIENCE CHINA Technological Sciences, 2012, 55(7):1982-1987 [29] J.G. Quintiere, A.S. Rangwala. A theory for flame extinction based upon flame temperature. Fire and Materials, 2004, 28(5):387–402
ip t
[30] I. Naeem, W. Sunil. Fire Dynamics Tools：Quantitative Fire Hazard Analysis Methods for the U.S. Nuclear Regulatory Commission Fire Protection Inspection
Ac ce pt e
d
M
an
us
Nuclear Reactor Regulation, 2004:Chapter 3,4-6
cr
Program, Washington, DC: U.S. Nuclear Regulatory Commission Office of
29 Page 30 of 30