Journal of Loss Prevention in the Process Industries 51 (2018) 137–149

Contents lists available at ScienceDirect

Journal of Loss Prevention in the Process Industries

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

Integrated self-assessment module for fire rescue safety in a chemical plant – T

A case study
Shih-Fang Tsaia,b, An-Chi Huanga, Chi-Min Shuc,∗
a
Graduate School of Engineering Science and Technology, National Yunlin University of Science and Technology (YunTech), Taiwan, ROC
b
Fire Bureau of Taichung City Government, Taiwan, ROC
c
Center for Process Safety and Industrial Disaster Prevention, School of Engineering, YunTech, Taiwan, ROC

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

Keywords: An integrated self-assessment module using disaster simulation software (suite for the assessment of flammable,
Integrated self-assessment module explosive, and toxic impacts, SAFETI) was employed to analyze the consequences of chain fire disasters in
Chain fire disasters chemical plants. Fire incident rescue procedures were successfully simulated and quantified. This study was
Fire incident rescue procedures implemented based on environmental data, risk frequency, and the physical and chemical characteristics of the
Risk frequency
chemical materials involved. Factors such as thermal radiation and the high pressure induced by an explosion
Domino effect
were evaluated, and a second-round SAFETI simulation was then designed. The results of the study can assist
supervisors in determining the influence of domino effects during related incidents and also help determine the
appropriate deployment of rescue personnel and vehicles to avoid casualties and fatalities.

1. Introduction Response Guidebook (ERG) to determine safety distances at which they

The chemical industry provides a wide variety of high-tech mate-
rials used in daily life. However, whether they are raw, semi-finished,
or finished, these materials often possess toxic, corrosive, flammable, or
even explosive properties. Government authorities have established
various chemical safety laws for each sector of industry according to
each country's distinct management styles. Nevertheless, numerous
accidents have still sporadically occurred in Taiwan and worldwide
during the years 2005−2015, as demonstrated in Tables 1 and 2, re-
spectively. Firefighters are often injured during rescue operations, and
failure to properly assess safety distances affects the extent of fire da-
mage and explosions and the number of resultant casualties (Liu et al.,
2017). Therefore, safety distance guidelines for chemical plant disaster
rescue operations merit further investigation.
Firefighters currently rely on a disaster incident response system
developed by the California Specialized Training Institute (http://
www.caloes.ca.gov) and a toxic chemical disaster emergency proce-
dure card. In preliminary rescue procedures, firefighters reference
safety data sheets, the Globally Harmonized System of Classification
and Labelling of Chemicals, and the 2016 edition of the Emergency
can operate. However, considerable discrepancies often exist between
the safety distances stated in the ERG and those that would prevent
harm from occurring to rescue workers, which results in inefficient
rescues or firefighter casualties (Mishra et al., 2014).
Over the past two decades, research on risk analysis and safety as-
sessment of domino effect has been paid more and more attention. In
terms of risk analysis, the main focus is on the assessment of the risk
factor of the domino effect (Abdolhamidzadeh et al., 2010; Cozzani and
Salzano, 2004a; Cozzani et al., 2006; Khakzad et al., 2013; Khakzad,
2015; Khakzad and Reniers, 2015; Landucci et al., 2009, 2015; Zhou
et al., 2016). Furthermore, safety assessment involves various types of
consequence analysis software to simulate the impact of the domino
effect. Consequence analysis software—such as ChemPlus, areal loca-
tions of hazardous atmospheres (ALOHA), atmospheric dispersion
model for denser-than-air release (SLAB), suite for the assessment of
flammable, explosive, and toxic impact (SAFETI), the fire dynamics
simulator (FDS), and the flame acceleration simulator (FLACS)—is
commonly used to simulate disasters including fires, pipeline leaks, and
chemical storage tank explosions (Cheng et al., 2009; Dadashzadeh
et al., 2013; I and Cheng, 2008; Santos and Landesmann, 2014; Zhou

∗
Corresponding author.
E-mail address: shucm@yuntech.edu.tw (C.-M. Shu).

https://doi.org/10.1016/j.jlp.2017.12.011
Received 17 January 2017; Received in revised form 29 July 2017; Accepted 14 December 2017
Available online 15 December 2017
0950-4230/ © 2017 Elsevier Ltd. All rights reserved.
S.-F. Tsai et al. Journal of Loss Prevention in the Process Industries 51 (2018) 137–149

Table 1
Major accidents with casualties in Taiwanese chemical plants, 2005–2015.

Items Plants location Date Chemicals Type Injured/death Remark

1 Tuku, Yunlin, Taiwan 04.02.2005 Trinitrotoluene Explosion 3/2 Firecracker plant

2 Nantun, Taichung, Taiwan 07.03.2005 Sodium nitrite Fire 22/0 Three firefighters injured
3 Hsitun, Taichung, Taiwan 02.20.2006 Hydrofluoric acid Splash 2/0
4 Mailiao, Yunlin, Taiwan 03.06.2006 Fire 5/1 Improper operation
5 Shulin, New Taipei City 05.21.2007 Nickel sulfate Leakage 4/2
6 Luchu, Taoyuan, Taiwan 06.25.2007 Propylene glycol methyl ether Explosion 5/1
7 Xinwu, Taoyuan, Taiwan 08.16.2007 Nitrite Explosion 6/0 Improper operation
8 Guanyin, Taoyuan, Taiwan 09.22.2009 Calcium fluoride Explosion 3/1
9 Mailiao, Yunlin, Taiwan 11.18.2009 Phosgene Leakage 12/0
10 Nantun, Taichung, Taiwan 06.11.2010 Xylene Explosion 5/2
11 Nantun, Taichung, Taiwan 06.18.2011 Foam Fire 1/1 Grinder sparks
12 Mailiao, Yunlin, Taiwan 09.14.2011 Styrene Explosion 2/0
13 Lukang, Changhua, Taiwan 05.17.2012 Toluene Explosion 13/1
14 Nantun, Taichung, Taiwan 08.01.2012 Methyl isobutyl ketone Explosion 3/0 Static spark
15 Dalin, Chiayi, Taiwan 08.10.2012 Hydrogen Fire 3/0
16 Minsyong, Chiayi, Taiwan 01.10.2014 Dodecylbenzene sulfonic acid Explosion 1/1 Welding spark
17 Mailiao, Yunlin, Taiwan 03.05.2014 Hydrogen Fire 2/0
18 Guishan, Taoyuan, Taiwan 03.27.2014 Hydrofluoric acid Leakage 1/0 One firefighter injured
19 Cianjhen, Kaohsiung, Taiwan 03.31.2014 Propylene Explosion 308/32 Twenty-four firefighters injured, six firefighters dead

Sources: Compiled by the authors.

Table 2
Major accidents with casualties in chemical plants globally, 2005–2015.

Items Plants location Date Chemicals Type Injured/death Remark

1 Texas, USA 03.23.2005 Light oil Explosion 180/15

2 Jilin, China 11.13.2005 Nitrobenzene Explosion 60/8
3 Buncefield, UK 12.11.2005 Gasoline Explosion 43/0
4 Osaka, Japan 12.19.2006 Aluminium Explosion 2/0
5 Hebei, China 05.11.2007 Fire 80/4 Improper operation
6 Florida, USA 12.19.2007 Trinitrotoluene Explosion 14/4
7 Istanbul, Turkey 01.31.2008 Trinitrotoluene Explosion 68/17
8 Markazi Province, Iran 05.25.2008 Explosion 38/30 Welding spark
9 Georgia, USA 07.02.2008 Dust Explosion 42/13
10 Liaoning, China 09.14.2008 Fire 2/3
11 Penang, Malaysia 04.24.2009 Gas Explosion 5/1
12 Ahmedabad, India 07.06.2009 Trinitrotoluene Explosion 100/30
13 Kharg Island, Iran 07.24.2010 Explosion 0/4 Boiler pressure was too high
14 Liaoning, China 01.19.2011 Heavy oil Explosion 30/0
15 Guangxi, China 11.23.2011 Trinitrotoluene Explosion 11/4 Firecracker plant
16 Amuay, Venezuela 08.25.2012 Propane Explosion 130/50
17 Hyogo, Japan 09.29.2012 Acrylic acid Explosion 30/1 One firefighter deaded
18 Jeollanam, Korea 03.14.2013 High density polyethylene Explosion 11/6 Welding spark
19 Jiangsu, China 04.16.2014 Dust Explosion 9/8
20 Moerdijk, Netherlands 06.03.2014 Ethylbenzene Explosion 2/0

Sources: Compiled by the authors.

Fig. 1. Relationship tree of the domino effect.

138
S.-F. Tsai et al. Journal of Loss Prevention in the Process Industries 51 (2018) 137–149

Fig. 2. The general flow chart for this study.

Fig. 3. Schematic plot of the plant for this study.

139
S.-F. Tsai et al. Journal of Loss Prevention in the Process Industries 51 (2018) 137–149

Table 3
Data on relevant buildings.

Company name Plant type Number of floors Floor area (m2) Structure Application

Sin Hun A raw material warehouse One 1300 Iron sheet Warehouse
Sin Hun B raw material warehouse One 1300 Iron sheet Warehouse
Everspring Old plant Tree 4300 Reinforced concrete, Iron roof Plant, Office
Chun Yuan An iron sheet plant One 1650 Iron sheet Plant

Sources: Compiled by the authors.

Table 4
Types and amounts of hazardous material in storage tanks.

Number Category Chemical Location Capacity (kg)

1 Flammable liquid Benzene 1st plant 19,000

2 Flammable liquid Methanol 2nd plant 12,000
3 Flammable liquid Isopropanol 2nd plant 3800
4 Flammable liquid Acetonitrile 2nd plant 1700
5 Flammable liquid N-heptane 2nd plant 3400

Sources: Compiled by the authors.

Table 5 (Zhou and Liu, 2012). FDS also lacks map integration and cannot es-
Hazardous material in storage. timate overall spatial concepts or the extent of the strain placed on staff,
especially firefighters. FLACS runs 3D simulations of fires or explosions
Number Category Chemical Capacity (kg)
stored
caused by combustible gas leaks (I et al., 2009). In summary, these
software assessment models are incomplete, simulate single accidents
1 Pyrophoric substances and substances Ammonia 4380 only without expanding or changing disaster types or patterns and did
which in contact with water emit not evolve into the safety distance for rescue.
flammable gas
2 Flammable liquids Hydrazine 5800
Catastrophes at chemical plants are usually the result of a disastrous
domino effect caused by the spread of a relatively minor accident, such
Sources: Compiled by the authors. as a chemical leakage, fire, or explosion. Firefighters are sometimes
subject to tremendous danger as such accidents escalate due to their
Table 6 failure to keep a safer operating distance. Therefore, this study proposes
Stock objects in the temporary outdoor storage area. an integrated assessment of safety disaster rescue operating distance
using SAFETI to simulate the catastrophic domino effect and simulta-
Chemical stored Raw material/ Quantity Public dangerous
recycle (drum) goods
neously visualize fires and explosions on maps (Cozzani et al., 2007,
2009; Santos and Landesmann, 2014).
Acetophenone Raw material 27 Class Ⅲ petroleum
Benzoate Raw material 48 Class Ⅲ petroleum
Butanol Raw material 6 Class Ⅱ petroleum 2. Research methods
Butanol Recycled 43 Class Ⅱ petroleum
Cyclohexane Recycled 8 Class Ⅰ petroleum 2.1. Environmental data analysis
Dimethyl sebacate Raw material 400 Class Ⅲ petroleum
Dipropylamine Recycled 58 Class Ⅰ petroleum
Monitoring data from the Taiwan Central Weather Bureau regarding
Dipropylamine Raw material 93 Class Ⅰ petroleum
Formic acid Raw material 8 Class Ⅱ petroleum average wind speed, temperature, and relative humidity were used
N-heptane Recycled 112 Class Ⅰ petroleum (http://www.cwb.gov.tw/). Atmospheric stability was selected ac-
Isopropanol Raw material 13 Alcoholic cording to the intensity of sunlight and other parameters related to
Isopropanol Recycled 23 Alcoholic
wind speed. The surface roughness parameter was selected on the basis
Methanol Recycled 27 Alcoholic
Octylamine Raw material 234 Class Ⅱ petroleum
of the geography of the chemical plant and its adjacent areas (Patra,
Xylene Raw material 46 Class Ⅰ petroleum 2006).
Xylene Recycled 82 Class Ⅰ petroleum

2.2. SAFETI
Sources: Compiled by the authors.

SAFETI is an impact analysis program for simulating various types

and Liu, 2012). Such simulations provide information on effective
preventive measures and treatment for stress and casualties among
rescue responders. However, ChemPlus, ALOHA, and SLAB lack map
integration and risk analysis models and are unable to calculate barrier
and wind effects, restricting their scope of use and contributing to large
discrepancies between simulation results and actual accident conditions
of accidents. The relationship between distance and the thermal ra-
diation pressure waves caused by explosions can be considered in either
the physical or effect modes (Reniers et al., 2015).
2.2.1. Physical mode
Casualty assessments are implemented after accidents, including
hazardous properties and areas. This study defined the mode as the fire

140
S.-F. Tsai et al. Journal of Loss Prevention in the Process Industries 51 (2018) 137–149

Fig. 4. Accident scenario at case study.

Table 7
Weather in the disaster zone in July 2005 (http://www.cwb.gov.tw/).

Item Parameter Numerical value

Temperature (°C) Maximum temperature 35.4

Lowest temperature 22.4
Average temperature 28.8
Wind velocity (m/s) Maximum wind velocity in ten 10.4
minutes
Maximum instantaneous wind 34
velocity
Average wind velocity 1.5
Wind direction (360°) Maximum wind in ten minutes 360
Maximum instantaneous wind 350
Average wind 220
Relative humidity (%) Minimum humidity 49
Average humidity 76
Rainfall (mm) 378.1
Precipitation days (day) ≧0.1 mm 11

Sources: Compiled by the authors.

explosion mode, wherein thermal radiation and pressure waves are Q = Hc MS (1)
measured after a flammable substance has been discharged and exposed
to a source of ignition (Cozzani et al., 2006).
Q
R=
12πqf (2)
2.2.2. Effect mode
This involves evaluation of the extent of casualties or building da- Explosion hazard is a function of pressure (Cozzani and Salzano,
mage caused by fires or explosions of the hazardous materials of in- 2004a, b; Cozzani et al., 2007; Piccinini, 2008).
terest. This study also investigated thermal radiation and pressure
1 1 1
waves within this mode. The extent of damage is a function of thermal ΔP = 1.1 R1
+ 4.3 2
+ 14 3
radiation (Cozzani et al., 2007, 2009; Du et al., 2017; Hiroumi et al., 3ϖ ( )
R1
3ϖ ( )
R1
3ϖ (3)
2015; Kyunghyun et al., 2014; Rachid et al., 2017).

141
S.-F. Tsai et al. Journal of Loss Prevention in the Process Industries 51 (2018) 137–149

Fig. 5. Wind rose plot for the statistics of wind direction.

Table 8 determined by using the previous 20 years' data from the Major
Physical and chemical properties of sodium nitrite (Abdollahi and Khaksar, 2014; Feng Accident Reporting System, Major Hazard Incident Data Service, Major
et al., 2016).
Accident Database, Failure and Accident Technical Information System,
Object Physical and chemical properties and Major Accident Hazards Bureau Database. This frequency includes
disasters caused by a domino effect (Fig. 1) (Darbra et al., 2010; Gómez-
Color White or light yellow Mares et al., 2008; Zhang and Zheng, 2012).
Exterior Deliquescent powder or granules
Odor Tasteless
Solubility Extremely soluble in water 2.3.2. Integrated impact analysis
Molar mass 69 g/mol A SAFETI simulation of an initial fire or explosion was performed
Melting point 271 °C that measured the increments of thermal radiation and pressure shock-
pH value About 9.0
wave strength that could trigger a second or even third accident
Flash point 537 °C
Density (water = 1.0) 2.17
(Gómez-Mares et al., 2008; Spoelstra et al., 2015), and the general flow
Normal boiling point 320 °C (decomposes) chart is demonstrated in Fig. 2.
Solubility value 82 g/100 mL (20 °C)
Auto–ignition temperature 489 °C 3. Case study
Vapour pressure (25 °C) 9.9E–17 hPa
LD50 180 mg/kg
LD50 5500 μg/m3/4 h (Rats, inhale) 3.1. Site location
LC50 ≦0.1 mg/L/96 h
The company used as the case study is located in Taichung
Sources: Compiled by the authors.
Industrial Park, Taichung, Taiwan. Accommodating more than 800
companies, it is currently one of the largest industrial parks in Taiwan;
its layout is presented in Fig. 3. The first and second plants were three-
story buildings, and the third plant was a five-story building. Ware-
2.3. Integrated assessment houses A and B, which stored raw materials, were metal buildings with
floor areas of 1300 m2 each. They were adjacent to the old plant and
2.3.1. Integrated risk frequency analysis had reinforced concrete with tin roofs above the third floor. The plant
The frequency of secondary accidents in fires and explosions caused had a total area of 4300 m2 (Table 3). Warehouse A was approximately
by thermal radiation, pressure waves, or explosion fragmentation was 2.4 m from warehouse B and approximately 4.8 m from the new plants.

142
S.-F. Tsai et al. Journal of Loss Prevention in the Process Industries 51 (2018) 137–149

Fig. 6. Thermal radiation of hydrazine tank release after the explosion.

Fig. 7. Thermal radiation and distance of hydrazine tank release.

143
S.-F. Tsai et al.
Fig. 8. Pressure wave radii of hydrazine tank release mapping with the overview.
3.2. Public storage of hazardous materials
The chemical company stored its hazardous goods in three loca-
tions: sump area, warehouses, and outdoor storage area. Types and
amounts of hazarding materials stored in tanks are listed in Table 4. The
reservoir mainly stored flammable liquids and was located next to the
company's plants 1 (north side) and 2 (east side). The warehouses were
located on the east side of the finished product warehouse. They stored
raw materials (including hazardous materials; Table 5) and were
burned out during the fire. The outdoor storage area stored materials
including toxic substances (Table 6) and was located next to the drying
chamber on the west side of the plant; fortunately, this area was un-
affected by the fire.
3.3. Fire rescue unit case study
A fire accident report was submitted at 12:10 on July 03, 2005, for
an incident that had occurred at 22:33 the previous day. In the period
between those two events, explosions occurred once every 2–3 min on
average. A 53-gallon drum exploded inside the plant, with the resulting
fire flowing around an iron sheet. At the time of the accident, there was
68% relative humidity and a prevailing westerly wind with a speed of
1.5 m/s, which accelerated the spread of the fire. Even after the fire was
144
Journal of Loss Prevention in the Process Industries 51 (2018) 137–149
controlled, numerous large-scale resurgences occurred. The first,
second, and third floors of warehouse A were thoroughly destroyed by
the fire, which spread to 300 m2 of the old plant. Fire rescue operations
were performed for 10 h 20 min (Fig. 4). A total of 22 people was in-
jured, including two professional firefighters and one volunteer fire-
fighter (Chen et al., 2010).
3.4. Environmental data analysis
Statistical data from the Taiwan Central Weather Bureau for July
2005 are listed in Table 7, including monthly average temperature,
maximum temperature, relative humidity, and wind speed. A wind rose
plot revealing the wind direction statistics is presented in Fig. 5.
3.5. Integrated impact analysis
3.5.1. Integrated risk frequency analysis
The selected databases revealed that the cases of domino effect
disasters with fires or explosions as the initial events had occurred over
the previous 20 years. The risk data of the plants of similar type and
level was adopted and used failure rate and event data within risk as-
sessments. The risk frequency was obtained using the value 1 × 10−3
(times per year). This frequency was used in the consequence analysis.
S.-F. Tsai et al.
Fig. 9. Pressure wave radii of hydrazine tank release.
3.5.2. Integrated impact analysis
Sodium nitrite (Table 8) was improperly stored and ignited the
chemicals it was stored adjacent to in the warehouse, including the
flammable liquids hydrazine (approx. 5800 kg) and ammonia water
(approx. 4380 kg). Environmental data, such as the inherent risks in the
physical and chemical properties of lawful hazardous materials, were
numerically entered SAFETI to simulate disasters involving a domino
effect caused by both thermal radiation and pressure shock waves.
4. Results and discussion
4.1. Initial accident
When firefighters arrived at the scene, the warehouse has sprung
out of the smoke and fire. After the rescue, identified by the fire in-
vestigators due to improper storage of sodium nitrite in the warehouse,
resulting in fire and igniting nearest-neighbor storage of the hydrazine.
Fig. 6 illustrates that sodium nitrite ignited inside a warehouse con-
taining hydrazine, releasing 28.5 kW of thermal radiation within a ra-
dius of 13.0 m (Fig. 7). At 14.0 m from the source, the thermal radiation
had decreased to 18.0 kW/m2. The human body begins to feel pain after
being exposed to this environment after 2.0 s at the distance of 32.0 m,
where the thermal radiation density is 4.0 kW/m2; rescue teams must
thus wear appropriate personal protective equipment to be able to
properly perform their duties. When under prolonged exposure to
145
Journal of Loss Prevention in the Process Industries 51 (2018) 137–149
9.5 kW/m2 of thermal radiation at the distance of 22.5 m, the com-
bustible matters were ignited, and caused the domino effect.
As illustrated in Fig. 8, the ignition of hydrazine caused a blast
overpressure of 0.1379 bar within a radius of 23.0 m, and a blast
overpressure of 0.2068 bar within a radius of 16.0 m. Fig. 9 shows the
pressure wave radii of hydrazine tank release. The pressure from the
explosion could have readily caused structural damage to the building
and generated a chain explosion. At a maximum distance of 87.0 m, at
which the explosion pressure had decreased to 0.0207 bar, there was
only 5% probability of lethal effects. Rescue teams must wear appro-
priate personal protective equipment to be able to properly perform
their duties under a blast overpressure of 0.0207 bar within a radius of
87.0 m.
The 2012 ERG classifies hydrazine as a flammable and corrosive
liquid, which, in the event of a leak, should be isolated in a radius of at
least 50.0 m as a precaution. The initial evacuation and isolation zone
should have been set at 800.0 m.
According to thermal radiation formulae, the possible thermal ra-
diation levels were determined to be 28.5, 18.0, 4.0, and 9.5 kW/m2 at
radii of influence 24.1, 30.3, 32.0, and 41.76 m, respectively. The ex-
plosion pressure wave formulae indicated possible pressures of 0.2068,
0.1379, and 0.0207 bar at radii of influence 15.0, 17.5, and 31.0 m,
respectively.
To keep rescue personnel at a safe distance in the event of a created
pressure wave from the thermal radiation after a hydrazine accident,
S.-F. Tsai et al.
Fig. 10. Thermal radiation of hydrazine and ammonia tanks release.
the following aspects should have been considered:
A. Domino effect
(1) Effects of thermal radiation: At the distance of 22.5 m from the
ignition sources, the thermal radiation was calculated as
9.5 kW/m2.
(2) Pressure wave caused by an explosion: At the distance of 16.0 m
from the ignition sources, the overpressure decreased to less
than 0.2068 bar.
B. Safety distance for rescue personnel
(1) Thermal radiation effect: At the distance of 32.0 m from the
ignition sources, the thermal radiation value decreased to
4.0 kW/m2 or less.
(2) Effects of explosion overpressure: At the distance of 87.0 m from
the ignition sources, the overpressure decreased to less than
0.0207 bar.
4.2. Second accident
According to Fig. 10, the release of radiation from the hydrazine
combustion ignited the ammonia in the warehouse, resulting in a
146
Journal of Loss Prevention in the Process Industries 51 (2018) 137–149
thermal radiation level of 8.3 kW/m2 within a radius of 10.0 m
(Fig. 11). At 23.0 m, the thermal radiation was 6.3 kW/m2. The human
body could have withstood exposure to this environment for 1.0 min at
the distance of 37.0 m, at which point the thermal radiation had de-
creased to 4.0 kW/m2. Therefore, rescue personnel with appropriate
personal equipment could have kicked off to properly perform their
duties under these conditions, but prolonged exposure to a thermal
radiation of 8.3 kW/m2 could have possibly resulted in damage, as
could the domino effect if it had occurred when proper protection was
not provided. Fig. 12 indicates that the ammonia combustion resulted
in an explosion of 0.2068 bar within a radius of 29.0 m, a blast over-
pressure of 0.1379 bar within a radius of 35.0 m. This secondary ex-
plosion facilitated the ignition of surrounding combustible materials. At
the distance of 145.0 m, the explosion overpressure decreased to less
than 0.0207 bar to reach 5% probability of lethal effects. Rescue per-
sonnel must wear appropriate personal protective equipment to be able
to properly perform their duties under blast overpressure of 0.0207 bar
within a radius of 145.0 m.
The ERG classifies ammonia as a corrosive gas. When leaks occur,
the area should be isolated in a radius of at least 150.0 m (800.0 m at
night) as an immediate preventive action. According to thermal
S.-F. Tsai et al.
Fig. 11. Thermal radiation and distance of ammonia tank release.
radiation formulae, the thermal radiation levels were estimated to be
8.3, 6.3, and 4.0 kW/m2 at radii of influence of 21.82, 25.05, and
31.43 m, respectively. The blast overpressure formula determined
pressures of 0.2068, 0.1379, and 0.0207 bar at corresponding radii of
influence of 16.8, 20.0, and 58.0 m, respectively.
In a SAFETI simulation of the secondary fire event—in which am-
monia water ignited and caused an explosion with a chain reaction
from the thermal, radiation, and overpressure effects—the following
should be considered:
A. Domino effect: A pressure wave was caused by an explosion 29.0 m
away from the ignition sources, with an overpressure of 0.2068 bar.
B. Safety distance for rescue personnel:
(1) Thermal radiation effect: At the distance of 37.0 m from the
ignition sources, the thermal radiation decreased to less than
4.0 kW/m2.
(2) Effects of explosion overpressure: At the distance of 145.0 m
from the ignition sources, the overpressure decreased to
0.0207 bar or less.
5. Conclusions
The ERG is currently the prevailing reference available for fire-
fighters to determine the distances at which they can safely, as well as
adequately, perform their duties. This study constructed an integrated
self-assessment module to design a chemical plant disaster rescue as-
sessment that could determine safety distances with corresponding
147
Journal of Loss Prevention in the Process Industries 51 (2018) 137–149
scenarios. According to the simulation results, the safety distance of the
firefighters was 87.0 m, and if the rescue personnel had worn the proper
equipment, the level of risk acceptable would have been between 32.0
and 87.0 m at the time of the initial accident. And if the combustibles in
the range of 22.5 m could not be removed or the thermal radiation
values in the range of 22.5 m were reduced to 22.5 kW/m2, the domino
effect would have occurred and led to a second accident. When the
second accident occurred, the safety distance of the firefighters in-
creased to 145.0 m and if the rescue personnel had worn the proper
equipment, the level of acceptable risk would have been between 37.0
and 145.0 m at the time of the second accident. In addition, if there are
combustibles within 29.0 m in a vicinity of the fire, there is still a
chance for a third accident to happen.
The initially completed integrated assessment module used SAFETI
to simulate domino effects caused by fires and explosions. For those in
charge of the plant, this study provides a visual impact of the domino
effect, which can be pre-simulated with a way to plan all kinds of
preventive measures, and help improve their own safety awareness. On
the other hand, for firefighters and rescuers, to more precisely de-
termine safety distances within emergency rescue guidelines, improve
rescue efficiency, and reduce casualties, the following should be im-
plemented: First, the ability of a domino effect to occur should be
minimized by removing flammable material from the area. Water-jet
cooling, firefighting, foam covering, and other measures should be used
to counteract ignition hazards. Second, vehicles and staff should be kept
at an appropriate safety distance to alleviate the likelihood of casualties
and fatalities.
S.-F. Tsai et al. Journal of Loss Prevention in the Process Industries 51 (2018) 137–149

Fig. 12. Pressure wave radii of hydrazine and ammonia tanks release.

Nomenclature Cozzani, V., Salzano, E., 2004b. The quantitative assessment of domino effects caused by
overpressure. Part I. Probit models. J. Hazard Mater. 107 (3), 67–80.
Cozzani, V., Tugnoli, A., Salzano, E., 2007. Prevention of domino effect: from active and
Hc Heat of combustion [kJ/kg] passive strategies to inherently safer design. J. Hazard Mater. 139 (2), 209–219.
M Mass burning rate [kg/s·m2] Cozzani, V., Tugnoli, A., Salzano, E., 2009. The development of an inherent safety ap-
ΔP Explosion pressure [MPa] proach to the prevention of domino accidents. Accid. Anal. Prev. 41 (6), 1216–1227.
Dadashzadeh, M., Khan, F., Hawboldt, K., Amyotte, P., 2013. An integrated approach for
Q Heat release rate [kW] fire and explosion consequence modelling. Fire Saf. J. 61, 324–337.
qf Critical radiant flux [kW/m2] Darbra, R.M., Palacios, A., Casal, J., 2010. Domino effect in chemical accidents: main
R Minimum safety distance [m] features and accident sequences. J. Hazard Mater. 183 (1–3), 565–573.
Du, S., Wang, R.Z., Chen, X., 2017. Development and experimental study of an ammonia
R1 Distance from the center of the explosion [m]
water absorption refrigeration prototype driven by diesel engine exhaust heat.
S Burning area [m2] Energy 130, 420–432.
T Time [s] Feng, X., Li, C., Jia, X., Guo, Y., Lei, N., Hackman, R.M., Chen, L., Zhou, G., 2016.
Influence of sodium nitrite on protein oxidation and nitrosation of subjected to
Ω Object mass [kg]
processing and storage. Meat Sci. 116, 260–267.
Gómez-Mares, M., Zárate, L., Casal, J., 2008. Jet fires and the domino effect. Fire Saf. J.
References 43 (8), 583–588.
Hiroumi, T., Hiroshi, T., Mitsuo, K., Yu, D., 2015. Hypergolic ignition and flame struc-
tures of hydrazine/nitrogen tetroxide co-flowing plant jets. Proc. Combust. Inst. 35,
Abdolhamidzadeh, B., Abbasi, T., Rashtchian, D., Abbasi, S.A., 2010. A new method for 2199–2206.
assessing domino effect in chemical process industry. J. Hazard Mater. 182 (1–3), I, Y.P., Cheng, T.L., 2008. The development of a 3D risk analysis method. J. Hazard Mater.
416–426. 153 (1–2), 600–608.
Abdollahi, M., Khaksar, M.R., 2014. third ed. Sodium Nitrite, Encyclopedia of Toxicology, I, Y.P., Shu, C.M., Chong, C.H., 2009. Applications of 3D QRA technique to the fire/
vol. 4. Elsevier Inc., Academic press, Radarweg, NX Amsterdam, The Netherlands, pp. explosion simulation and hazard mitigation within a naphtha-cracking plant. J. Loss
334–337. Prev. Process. Ind. 22 (4), 506–515.
Chen, C.C., Wang, T.C., Chen, L.Y., Dai, J.H., Shu, C.M., 2010. Loss prevention in the Khakzad, N., 2015. Application of dynamic Bayesian network to risk analysis of domino
petrochemical and chemical–process high–tech industries in Taiwan. J. Loss Prev. effects in chemical infrastructures. Reliab. Eng. Syst. Saf. 138, 263–272.
Process. Ind. 23 (4), 531–538. Khakzad, N., Khan, F., Amyotte, P., Cozzani, V., 2013. Domino effect analysis using
Cheng, S., Chen, G., Chen, Q., Xiao, X., 2009. Research on 3D dynamic visualization Bayesian networks. Risk Anal. 33 (2), 292–306.
simulation system of toxic gas diffusion based on virtual reality technology. Process Khakzad, N., Reniers, G., 2015. Using graph theory to analyza the vulnerability of process
Saf. Environ. Protect. 87 (3), 175–183. plants in the context of cascading effects. Reliab. Eng. Syst. Saf. 143, 63–73.
Cozzani, V., Gubinelli, G., Salzano, E., 2006. Escalation thresholds in the assessment of Kyunghyun, R., George, E., Zacharakis, J., Kong, S.K., 2014. Performance enhancement of
domino accidental events. J. Hazard Mater. 129 (1–3), 1–21. ammonia–fueled engine by using dissociation catalyst for hydrogen generation. Int. J.
Cozzani, V., Salzano, E., 2004a. The quantitative assessment of domino effect caused by Hydrogen Energy 39, 2390–2398.
overpressure. Part II. Case studies. J. Hazard Mater. 107 (3), 81–94.

148
S.-F. Tsai et al.
Landucci, G., Argenti, F., Tugnoli, A., Cozzani, V., 2015. Quantitative assessment of safety
barrier performance in the prevention of domino scenarios triggered by fire. Reliab.
Eng. Syst. Saf. 143, 30–43.
Landucci, G., Gubinelli, G., Antonioni, G., Cozzani, V., 2009. The assessment of the da-
mage probability of storage tanks in domino events. Accid. Anal. Prev. 41,
1206–1215.
Liu, X., Zhang, L., Guo, S., Fu, M., 2017. A simplified method to evaluate the fire risk of
liquid dangerous chemical transport vehicles passing a highway bridge. J. Loss Prev.
Process. Ind. 48, 111–117.
Mishra, K.B., Wehrstedt, K.D., Krebs, H., 2014. Amuay refinery disaster: the aftermaths
and challenges ahead. Fuel Process. Technol. 119, 198–203.
Patra, A.K., 2006. Influence of wind speed profile and roughness parameters on the
downwind extension of vulnerable zones during dispersion of toxic dense gases. J.
Loss Prev. Process. Ind. 19 (5), 478–480.
Piccinini, N., 2008. Dust explosion in a wool factory: origin, dynamics and consequences.
Fire Saf. J. 43 (3), 189–204.
149
Journal of Loss Prevention in the Process Industries 51 (2018) 137–149
Rachid, A., Toshiyuki, K., Nobuyuki, A., Keiichi, H., 2017. Hydroxylammonium nitrate
(HAN)–based green propellant as alternative energy resource for potential hydrazine
substitution: from lab scale to pilot plant scale–up. Combust. Flame 176, 334–348.
Reniers, G., Van Lerberghe, P., Van Gulijk, C., 2015. Security risk assessment and pro-
tection in the chemical and process industry. Process Saf. Prog. 34 (1), 72–83.
Santos, F.d.S., Landesmann, A., 2014. Thermal performance–based analysis of minimum
safety distances between fuel storage tanks exposed to fire. Fire Saf. J. 69, 57–68.
Spoelstra, M., Mahesh, S., Kooi, E., Heezen, P., 2015. Domino effects at LPG and propane
storage sites in The Netherlands. Reliab. Eng. Syst. Saf. 143, 85–90.
Zhang, H.D., Zheng, X.P., 2012. Characteristics of hazardous chemical accidents in China:
a statistical investigation. J. Loss Prev. Process. Ind. 25 (4), 686–693.
Zhou, J., Reniers, G., Khzkzad, N., 2016. Application of event sequence diagram to
evaluate emergency response actions during fire–induced domino effects. Reliab.
Eng. Syst. Saf. 150, 202–209.
Zhou, Y., Liu, M., 2012. Risk assessment of major hazards and its application in urban
planning: a case study. Risk Anal. 32 (3), 566–577.