Safety Science 94 (2017) 26–40

Contents lists available at ScienceDirect

Safety Science
journal homepage: www.elsevier.com/locate/ssci

Fire risk assessment for large-scale commercial buildings based on

structure entropy weight method
Fang Liu a,b,c,d,⇑, Shengzhong Zhao a, Miaocheng Weng a,b,c,d,⇑, Yongqiang Liu a
a
Faculty of Urban Construction and Environmental Engineering, Chongqing University, Chongqing 400045, PR China
b
Key Laboratory of Three Gorges Reservoir Region’s Eco-Environment, Ministry of Education, Chongqing 400045, PR China
c
National Centre for International Research of Low-carbon and Green Buildings, Chongqing 400045, PR China
d
Joint International Research Laboratory of Green Building and Built Environment, Ministry of Education, Chongqing 400045, PR China

a r t i c l e i n f o a b s t r a c t

Article history: Based on the fire characteristic and the maintenance of fire equipment in the large-scale commercial
Received 20 March 2016 buildings, this paper proposed a fire risk assessment system for large-scale commercial buildings. The
Received in revised form 25 November 2016 assessment system focus on evaluating the safety performance of the fire protection system in the build-
Accepted 11 December 2016
ings. Firstly, the index system of the assessment system was determined by on-site investigation and
experts’ suggestions. Secondly, the weight of each index was determined by the structure entropy weight
method. Thirdly, the score rules were determined according to the regulations from the relevant laws and
Keywords:
design codes. Finally, the score and the corresponding fire risk level could be obtained. Moreover, four
Fire
Risk assessment
large-scale commercial buildings in Chongqing, China have been taken as examples to calculate the val-
Structure entropy weight method ues of the fire risk level by the proposed assessment system. In addition, the fire safety performance of
Large-scale commercial buildings the four buildings were analyzed.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction Specifically, once a fire occurs in the commercial building, it may

With the continuous development of China’s economic and
acceleration of urbanization, people’s living standard is being
improved, the requirements on the level and environment of the
shopping mall is being raised, hence a lot of large-scale commercial
buildings are being constructed. As a comprehensive entertain-
ment center, the large-scale commercial building has the features
of large scale, atrium inside and multiple functions, it usually com-
bines restaurant, leisure, tourism, residence, finance, culture, art,
etc.
The large-scale commercial building is defined as any floor area
of the building is equal or greater than 5000 m2, or the total area is
equal or greater than 15,000 m2 (Committee, 2013a, 2013b). Based
on this definition, the large-scale commercial building demon-
strates the characteristics of large building area and scale, large
number of people and electric equipment inside, complex evacua-
tion route and high fire load, so it is in a high fire risk. According to
statistics of yearly top ten fire accidents in the recent 10 years,
commercial building fires accounted for 24%, and large-scale com-
mercial building fires accounted for 2% (Qian-li and Ting-lin, 2011).
⇑ Corresponding authors at: Faculty of Urban Construction and Environmental
Engineering, Chongqing University, Chongqing 400045, PR China.
E-mail addresses: drliufang@126.com (F. Liu), mcweng@outlook.com (M. Weng).
cause a lot of damage to property and casualties. Large-scale com-
mercial buildings belong to high-risk units. However, until to date,
there is no appropriate risk assessment system for large-scale com-
mercial buildings. Therefore, there is an urgent need to establish a
fire risk assessment system for large-scale commercial buildings
and determine the corresponding building fire risk level. What’s
more, it has great social and economic significance in reducing
the fire risk in large-scale commercial buildings as well as casu-
alties and property loss. Fire risk assessment was originated from
the American insurance industry. The Insurance company charged
a certain fee by underwriting all kinds of fire damage for the poli-
cyholders, and the charge was determined by the magnitude of the
fire risk, as a result, the problem of how to measure the level of fire
risk was extended. Domestic and international researchers have
studied and explored the methods and theories of fire risk assess-
ment, and the main methods contain: standard reference method,
logic analysis method, comprehensive evaluation method, and
deterministic analysis method.
There are many fire risk evaluation models and corresponding
software. Among them, the typical models include: FIERA system
(Fire Evaluation and Risk Assessment system) (Benichou, 2005),
CESARE-RISK (Centre for Environment Safety and Risk Engineering,
RISK) (Beck, 1998), Crisp II (Computation of Risk Indices by Simu-
lation Procedures) (Fraser-Mitchell, 1994), etc.

http://dx.doi.org/10.1016/j.ssci.2016.12.009
0925-7535/Ó 2016 Elsevier Ltd. All rights reserved.
 F. Liu et al. / Safety Science 94 (2017) 26–40
Generally speaking, each assessment model has its own charac-
teristics and applicable scope. The model adoption should depend
on the specific case during the assessment process, moreover, the
precondition and the restriction should be understood adequately.
In most cases, the models rely on a large number of input data and
particular fire scenario, in addition, the application condition is
very strict. According to the target of evaluation, previous
researchers have developed some assessment models based on
the system theory. For examples: Qian et al. built an urban fire risk
assessment index system according to the possibility and the
severity of urban fire (Qian, 2012). Ke-xuan et al. built a fire risk
assessment index system for large-scale shopping centers accord-
ing to the fire characteristics (Ke-xuan, 2013). Omidvari et al.
who comes from Iran built a fire risk assessment system for educa-
tion center laboratories based on RPN method (Omidvari et al.,
2015). Li et al. built a fire risk assessment system for coal mines,
based on TOPSIS method (Li et al., 2011). Jiang et al. built a fire risk
assessment system for super high-rise buildings (Jiang et al., 2015).
Zheng conducted a fire risk assessment on the Main Stadium of the
12th National Games of China by using the analytic hierarchy pro-
cess (AHP) method (Zheng, 2014). Sun and Luo, also built a fire risk
assessment system for super high-rise buildings (Sun and Luo,
2014). Yang et al. built a fire risk assessment system for under-
ground buildings by using grey relational analysis (Yang et al.,
2012).
Most of the mentioned fire risk assessment systems are based
on the features of target buildings and equipment of the fire pro-
tection system. Nevertheless, fire risk level in the two same build-
ings with same fire equipment will be different, if the maintenance
situation of the fire equipment and the fire safety management
capability are different. Besides, the existing large-scale commer-
cial buildings have relatively high fire safety requirements and var-
ious of fire equipment, therefore, only relying on the installed fire
equipment to evaluate the fire risk is not enough.
There is no special fire risk assessment system for the large-
scale commercial buildings, and the application of the existing fire
risk assessment system is lack of the consideration of the fire char-
acteristic in the large-scale commercial buildings. Therefore, to
establish a fire risk assessment system considering the mainte-
nance situation of the fire equipment for the large-scale commer-
cial buildings will provide significant theoretical and practical
values.
A fire risk assessment system for large-scale commercial build-
ings will be established in this paper, the comprehensive evalua-
tion method was adopted and the detailed scoring system for all
fire indexes were included in this system. Firstly, the investigation
and research on the management departments, maintenance
departments, insurance companies of the large-scale commercial
buildings were carried out. Secondly, the indexes in the fire risk
assessment system for large-scale commercial buildings was
established by using analytic hierarchy process (AHP) method,
and the weights of the indexes were obtained by questionnaire
and structure entropy weight method. Thirdly, the score of each
index was provided by on-site investigation. Finally, the fire risk
of the large-scale commercial buildings was evaluated by the total
score calculated from the proposed fire risk assessment system.
2. Methodology
The purpose of this research is to establish a fire risk assessment
system for large-scale commercial buildings, and take some real
large-scale commercial buildings as examples to calculate the total
score based on the system. Therefore, the fire risk level depends on
the total score. Based on the architectural characteristics of large-
scale commercial buildings and considering the maintenance situ-
27
ation of fire equipment and fire safety management, in order to
reduce the possibility of a fire and increase the possibility of fire
extinction, five main aspects including causing disaster factors,
passive fire protection method, active fire protection method, fire
protection management, and fire-fighting ability of fire brigade
have been covered. This research mainly includes four steps: 1.
On-site investigation and risk identification to determine the fire
risk assessment index system of large-scale commercial buildings.
2. Use the structure entropy weight method to determine the index
weight. 3. Determine the scoring principle, score for the real com-
mercial building and determine the fire risk level. 4. Case study.
Four steps were described as follows:
Step 1: Build a targeted fire risk assessment index system for
large-scale commercial buildings. First of all, the related fire
data, the status quo of fire risk assessment and the fire influence
factors of large-scale commercial buildings were achieved
through on-site investigation, reading relevant literatures and
visiting the insurance company and the third party fire risk
assessment institution. Under the comprehensive consideration
of reducing the possibility of a fire and increasing the possibility
of fire extinction, the assessment index system was preliminar-
ily built thereafter. Some fire experts from design companies,
consult organizations, management departments, research
institutions and fire brigades who engaged in fire engineering
were invited to evaluate the preliminary index system. Accord-
ing to experts’ suggestions, this system mainly focused on the
maintenance of fire equipment and fire safety management in
large-scale commercial buildings. Therefore, considering the
reliability of fire safety system in the large-scale commercial
buildings, the characteristic of commercial buildings needed a
further consideration in assessment system and the index sys-
tem needed to be refined to every part of the fire-fighting
equipment. The establishment of index system was similar to
the analytic hierarchy process (AHP); indexes were divided into
several layers. Five main aspects, including causing disaster fac-
tors, passive fire protection method, active fire protection
method, fire protection management, and fire-fighting ability
of fire brigade, have been covered. Finally, the fire risk assess-
ment index system used in the large-scale commercial build-
ings was established.
Step 2: After the index system was determined, the index
weight should be assigned to all indexes in each layer. If the tra-
ditional AHP method was used to assign the index weights, it
would greatly increase experts’ workload. Compared with the
AHP method, this paper introduced a new method, Structure
Entropy Weight Method, to assign the index weight. It com-
bined methods of subjective and objective assignments, as well
as the qualitative and quantitative analysis. The detailed princi-
ple and application method of the structure entropy weight
method will be explained in Section 2.3.
Step 3: To achieve the quantification of the assessment results,
it is necessary to assign the scores of the indexes in the bottom
layer. For different types of bottom indexes, Grade Security
Assessment and the Scoring Method were adopted respectively
according to the difference of the indexes. Combined with Chi-
na’s actual situation, the evaluation criteria have been
improved. The scoring methods were approved by the experts,
and the details will be explained in Section 2.4. Through the
previous steps, the index weights in each layer and the bottom
indexes scores could be acquired.
In this paper, in order to eliminate the difference between dif-
ferent layers used different methods, the bottom indexes were
processed uniformly by Hundred-mark score transferred. The
value and level of the fire risk of the large-scale commercial
buildings could be calculated and evaluated finally.
28 F. Liu et al. / Safety Science 94 (2017) 26–40
Step 4: Case study. Four large-scale commercial buildings were
chosen for the case study to verify the feasibility of the assess-
ment system.
Algorithm used in this study is illustrated in Fig. 1.
2.1. Principles
There were various fire indexes in the index system of the large-
scale commercial building. Actually, the contribution of each index
is different as the weight of each index is not the same. In order to
reflect the fire risk of the large-scale commercial buildings accu-
rately and objectively, it refers to the scientificity, guidance, oper-
ability, systematicness, comparability and comprehensiveness to
establish the index system in this paper.
2.2. Index system of fire risk assessment for large-scale commercial
buildings
2.2.1. On-site investigation and fire risk identification
The fire risk assessment usually begins from the internal fire
risk identification of the building. In order to build the targeted fire
risk assessment index system for large-scale commercial buildings,
five commercial buildings were selected for investigation. Mean-
while, the fire insurance companies, fire brigades and the third
party risk assessment institutions were visited and consulted.
The investigation steps are shown in Fig. 2.
Detailed on-site investigations were carried out on the fire pro-
tection system, fire equipment maintenance and fire safety man-
agement in the commercial buildings. Safety check list method
was used to systematically analyze the fire equipment of large-
scale commercial buildings.
Analyzing the characteristics of large-scale commercial
buildings, collecting historical fire data, reviewing the
relevant literatures
On-site investigation, visiting the
related companies and institutions
Identifying the fire influence factors
of large-scale commercial buildings
Building fire risk assessment index system
Step 1
Collecting experts’ comments and
forming typical order
Analyzing blind degree
(uncertainty) of indexes
Normalized treatment of indexes
Determining the index
weight of each layer
Scoring and calculating
the final score
Determining the risk level of building
Case study
Fig. 1. Algorithm for the present study.
Designing the plan of investigation
Visiting risk assessment
Investigating the large-scale
department of fire insurance
commercial buildings and
company and the third party
taking on-site test
risk assessment institution
Visiting the relevant experts and scholars of fire risk
assessment and Chongqing Fire Brigade
Summarizing research results
Fig. 2. Investigation steps.
A great effort was made to make a reliability analysis of the fire
equipment by consulting the staffs from the maintenance organi-
zation and checking the failure rate of the fire equipment in
large-scale shopping malls and various fire equipment repair
conditions.
Partial equipment and related data are shown in Figs. 3–11.
Existing fire risk assessment system of large-scale commercial
buildings, the principle of building fire risk classification and the
evaluation for the fire insurance premium were achieved through
visiting the insurance companies and the third party fire risk
assessment institutions. In addition, a fire insurance scheme for
some large-scale commercial buildings made by an insurance com-
pany was obtained.
The fire risk influence factors in large-scale commercial build-
ings were discovered after on-site investigation and visit. Mean-
while, a series of fire risk problems in the buildings were also
found, such as various fire equipment was installed, it can’t com-
pletely be covered by the existing fire risk assessment system for
residential buildings, public buildings and industrial buildings.
There were little communication and cooperation between the
safety protection department, operation department and mainte-
nance organization. A lot of fire equipment were lack of mainte-
nance, and management regulation was lack of execution. As a
result, the fire risk was in high danger. The on-site investigation
and visit provided a good foundation to establish the fire risk
assessment index system of large-scale commercial buildings.
Step 2
Step 3
Step 4
Fig. 3. Control room.
 F. Liu et al. / Safety Science 94 (2017) 26–40
Fig. 4. Smoke-exhaust system.
Fig. 5. Pump room.
Fig. 6. Indoor fire hydrant.
2.2.2. Initial establishment of index system
Based on the characteristics of large-scale commercial build-
ings, on-site investigation, domestic and international fire data,
the review of published literature, existing fire risk assessment sys-
tem of public buildings, and some codes including Chongqing fire
regulations (Committee, 2013a, 2013b), Code for fire protection
design of large-scale commercial buildings (DBJ-50-054-2013)
(Committee, 2013a, 2013b), Code for fire protection design of
buildings (GB50016-2014) (China, 2014), the index system was
divided into five main aspects: causing disaster factors A1 , passive
29
Fig. 7. Valve.
Fig. 8. Alarm device.
Fig. 9. The equipment maintenance record.
fire protection method A2 , active fire protection method A3 , fire
protection management A4 , and fire-fighting ability of fire brigade
A5 . Based on the result of on-site investigation and visit, several
sub-items of each aspect were constructed. For instance, the index
system of causing disaster factors involved distribution plan A11 ,
occupant density A12 , internal fire load A13 , internal decoration
A14 , construction height A15 , building service life A16 and external
environment A17 . Passive fire protection method involved fire
resistance rating A21 , fire compartment A22 , smoke zone A23 ,
30 F. Liu et al. / Safety Science 94 (2017) 26–40
Fig. 10. Fire doors and safety exit signs.
Fig. 11. Fire equipment test report.
emergency lighting and evacuation indicator A24 . Active fire pro-
tection method involved automatic sprinkler system A31 , fire
hydrant system A32 , smoke control system A33 , and automatic fire
alarm system A34 . Fire protection management involved fire safety
responsibility system A41 , fire emergency plan A42 , fire safety train-
ing A43 and fire safety inspection A44 . Fire-fighting ability of fire bri-
gade involved fire lane A51 and rescue capability of the local fire
brigade A52 .
2.2.3. The modification of index system
Some fire experts from design companies, consult organiza-
tions, management departments, research institutions and fire bri-
gades who engaged in fire engineering were invited to evaluate the
preliminary index system. The assessment index system was
revised according to the comments from the experts. According
to experts’ comments, this system mainly focused on the mainte-
nance of fire equipment and fire safety management in large-
scale commercial buildings. Compared the previous system, the
maintenance situation of fire equipment should be reflected in pro-
posed system. Therefore, considering the reliability of fire safety
system in the large-scale commercial buildings, the characteristic
of commercial buildings needed a further consideration in assess-
ment system and the index system needed to be refined to every
part of the fire-fighting equipment. Thereafter, the fire risk assess-
ment index system for the large-scale commercial buildings was
established.
Finally, the first grade indexes were determined as follows:
causing disaster factors, passive fire protection method, active fire
protection method, fire protection management, and fire-fighting
ability of fire brigade. The second or next grade indexes were rec-
ognized as the further supplement subsystem for the correspond-
ing first or previous grade indexes.
In order to satisfy the suggestion from the experts that mainte-
nance situation of fire equipment is very important, this index sys-
tem focused on the passive fire protection method and active fire
protection method. Consequently, to describe and calculate the
two main indexes, the passive fire protection method was
extended to the third grade index and the active fire protection
method was extended to the fourth grade index.
The multi-grade index of passive fire protection method and
active fire protection method were mainly referred to Code for
installation and commissioning of sprinkler systems (GB50261-
2005) (PRC, 2013), Code for design of automatic fire alarm system
(GB50116-2013) (Supervision, 2013), and Code of design on fire
protection water supply and hydrant systems (GB50974-2014)
(PRC, 2014). Take the active fire protection method as an example,
the third grade index nearly contained all components of the fire
protection system, and the fourth grade index was used to identify
the maintenance situation of the components. For example, the
third grade index of automatic sprinkling fire extinguishing system
included the fire pool/tank, sprinkler pump, sprinkler, valve, fire
alarm valve and water flow indicator. The detailed fire risk assess-
ment index system is shown in Fig. 12.
The bottom indexes of A2 are shown in Table 1 and the bottom
indexes of A3 are shown in Table 2.
2.3. Determination of index weight
The calculation method of the index weight can be divided into
two categories: subjective assignment and objective assignment. In
the subjective assignment, the weight of each index depends on
the experts’ knowledge structure and individual preference.
Although it reflects well the subjective intention, it still lacks the
scientific calculation. Considering its obvious disadvantages, it
generally applies only to the assessment under the condition of
data collection in difficulty or information inaccuracy. Unlike the
subjective assignment, in the objective assignment, the weight of
index achieved by means of relevant mathematical treatment
based on the original information, due to this reason, it avoids
the disadvantages of the subjective assignment to some extent.
In most cases, this will give a higher accuracy of the index weight,
likewise, the disadvantage is that the result will not agree the
actual situation and difficult to give a clear explanation.
The AHP, Delphi and the expert experience are the commonly
used methods to determine the index weight in fire risk assess-
ment, and the AHP is the most widely used method (Lau et al.,
2015). Because there were a lot of indexes in the proposed index
system, if the AHP was used to determine the index weight, it
would take a large amount of experts’ workload, which would
not conducive to acquire the judgment matrix. For this reason, this
paper introduced the structure entropy weight method to deter-
mine the index weight of the large commercial building’s fire risk
assessment system.
The structure entropy weight method combined methods of
subjective and objective assignments, as well as the qualitative
and quantitative analysis. It could reduce a lot of calculation work-
load but achieve higher accurate result, especially for there were
many indexes in the fire risk assessment system.
2.3.1. The principle of structure entropy weight method
The basic idea of the structure entropy weight method is to ana-
lyze the indexes of the assessment system and the interrelation-
ship between them, and then to classify the indexes into
 F. Liu et al. / Safety Science 94 (2017) 26–40 31

Fig. 12. The assessment index system for large-scale commercial buildings.

independent hierarchical grades. The execution steps are as fol- an experts’ comments collection table. Experts need to fill in the
lows (Qi-yue, 2010): form with the indexes’ order of importance.

(1) Experts’ comments collection (2) Blind degree (uncertainty) analysis

Questionnaire forms of indexes weights was designed for
experts to decide the qualitative ranking of indexes. Several
experts were invited to complete the questionnaire form in accor-
dance with the procedure and requirement of the Delphi method
(Dalkey, 1969). Moreover, the selection of the experts should be
representative, authoritative and fair. The experts ranked the
importance of each index independently according to their own
knowledge and experience. The indexes were ranked from high
to low according to their importance, for example, mark ‘‘1” repre-
sented ‘‘the most important”, mark ‘‘2” represented ‘‘more impor-
tant”, mark ‘‘3” represented ‘‘important” and so on. Some indexes
could be allowed to recognize as equally important, and the final
rank of the indexes could be discussed by the experts. Table 3 is
The potential deviation and uncertainty of the experts’ com-
ments on the indexes rank might be appeared due to the noisy
data. In order to eliminate noisy data and reduce uncertainty, the
qualitative judgment conclusion from the experts should be statis-
tical analyzed and dealt with. In order to reduce the uncertainty of
expert ranking, the entropy value was calculated by the entropy
theory. The execution steps are as follows:
Supposing that k experts were invited to take the questionnaire
survey, then k questionnaire forms would be returned, every form
was recognized as an index set, marked as U ¼ fu1 ; u2 ; . . . ; uk g, ui
refers to the expert ranking array denoted by
fai1 ; ai2 ; . . . ; ain gði ¼ 12 . . . kÞ, ai1 ; ai2 ; . . . ; ain can be any natural num-
ber from f1; 2; . . . ; ng. As previously mentioned, ‘‘1” represented
32 F. Liu et al. / Safety Science 94 (2017) 26–40

Table 1
Bottom indexes of A2 .

Second grade indexes Third grade indexes (bottom indexes)

Fire resistance rating A21 1. Whether the fire resistance is level 1 or not 0-1-
3-5
2. During the design and construction, the materials must satisfy the national laws and codes 0-1-
5-7
3. Whether the decoration materials satisfy the requirements or not, and whether them were non-combustible or 0-1-
fireproof 3-5
Fire compartment A22 1. Whether fire compartment satisfy the codes or not 0-1-
3-5
2. Inspect the fire door every day; Open test impact test on the normally closed fire door; For the automatic fire door
installed in the evacuation passageway, input the automatic or manual open signal, test the action and feedback of the
controller on the door
3. Inspect the fire shutter and whether goods are storied in the warning area every day, whether the controller of fire
shutter is energized and under automatic status
4. Inspect the functions of the alarm and manual control of the fire shutter’s controller every half a year
5. Inspect the functions of on-site startup, remote startup and automatic startup every half a year
Smoke zone A23 1. Whether smoke zone satisfy the codes or not 0-1-
3-5
2. Smoke zone is under the fire compartment
3. The smoke screen is in perfect condition
4. The sign of the fire screen is clear and the installation is firm
5. There are no barriers around the smoke screen
Emergency lighting and evacuation 1. Whether the minimum width of the escape route satisfy the codes or not 0-1-
indicator A24 3-5
2. Inspect the emergency lighting and evacuation indicator regularly, establish failure report and clearing system, and
replace the damaged parts in time

the highest level of importance. The index sort matrix obtained b1j þ b2j þ . . . þ bkj
bj ¼ ð5Þ
from the k table is shown as a matrix A. k
0 1 Blind understanding degree rj was defined as the uncertainty of
a11 a12 a13 . . . a1n
Ba a22 a23 . . . a2n C the evaluation on the index uj from k experts.
B 21 C
A¼B C ð1Þ
@ ... ... ... ... ... A rj ¼ jf½maxðb1j ; b2j ; . . . bkj Þ  bj  þ ½bj  minðb1j ; b2j ; . . . bkj Þg=2j
ak1 ak2 ak3 ... akn ð6Þ
In which aij represents the i th expert’s evaluation on the jth The global understanding degree xj was defined as the degree of
index. the evaluation on every index uj from all k experts who have been
The qualitative ranking result could be transformed into quan- invited.
titative result by a membership function, it can be defined as:
xj ¼ bj ð1  rj Þ ð7Þ
vðIÞ ¼ kpn ðIÞ ln pn ðIÞ ð2Þ
The evaluation vector of the index uj from k experts was express
In which, pn ðIÞ ¼ m1
mI
; k ¼ lnðm1Þ
1
, they can be put into Eq. (2): as X ¼ ðx1 ; x2 ; . . . xk Þ.
   
1 mI mI
vðIÞ ¼  ln ð3Þ (3) Normalized treatment
lnðm  1Þ m  1 m1
In order to obtain the weight of index uj , the Eq. (7) need further
ðmIÞ
Divided by ðm1Þ for both sides, assume 1  vðIÞ= ðm1Þ
ðmIÞ
¼ lðIÞ. And
normalized treatment
then, ,
Xk

lðIÞ ¼
lnðm  IÞ
ð4Þ
xj ¼ xj xj ð8Þ
lnðm  1Þ j¼1

I was defined as the qualitative ranking number of a certain
index evaluated by an expert. For example, a set of qualitative
ranking numbers 5, 2, 3, 4, 1 for five indexes A1 ; A2 ; A3 ; A4 ; A5 was
evaluated by one expert. Thus, it means the index A5 was the most
important, for the index A5 , I ¼ 1. m was the transformation
parameter, defined as m ¼ j þ 2, and j is the number of indexes.
The qualitative ranking number I was put into the Eq. (4), the
quantitative transformation value of bij would be obtained.
bij ¼ lðaij Þ was referred to as the membership degree of the quali-
tative ranking number I, and the matrix B ¼ ðbij Þkn was defined as
the membership degree matrix. A new parameter, average under-
standing degree bj , was introduced. It presented the consistency
degree of the evaluation on the index uj from k experts, its calcula-
tion is as follows:
P
Obviously, xj > 0, and kj¼1 xj ¼ 1. The x ¼ ðx1 ; x2 ; . . . ; xj Þ was
expressed as the weight vector of index set U ¼ ðu1 ; u2 ; . . . ; uj Þ.
2.3.2. Calculation of the index weight
In the proposed fire risk assessment system, the index weight
was determined from the first grade, it means that the indexes
A1 ; A2 ; A3 ; A4 and A5 would be determined first. Similarly, the
weight of the indexes on the second and third grade could be
determined in sequence. Because the proposed system focused
on the fire equipment maintenance, the reliability of their opera-
tion should be a key consideration in the qualitative ranking eval-
uated from the experts.
During the study, 4 groups of experts, and 8 experts in each
group, totally 32 experts from industries including architecture,

Table 2
Bottom indexes of A3 .
Third grade indexes
Fire pool/water tank A311
Sprinkler pump A312
Sprinkler A313
Valve A314
Alarm valve A315
Water flow indicator A316
Indoor fire hydrant A321
Outdoor fire hydrant A322
Fire pump A323
Vent A331
Fan A332
Fire damper A333
Distribution box for fan A334
Fire detector A341
Alarm button A342
Fire alarm controller A343
F. Liu et al. / Safety Science 94 (2017) 26–40 33
Fourth grade indexes (bottom indexes)
1. Inspect the water level once a month, and verify whether the water level is right or not
2. Water temperature is not lower than 5 °C, anti-freeze measures and supply system are in good condition
1. Clean and inspect the functions of automatic and manual operation every week
2. The pump will start in 30 s if it is started by automatic or manual method
3. Discharge by opening the drain valve for 3 min every week
4. The sign with corresponding system and No. should be installed
5. The inlet and outlet valves should be always kept open, and the sign are correct
6. Pressure gauge, test valve and overpressure protection device should be in good condition
1. Visual Inspection on all sprinklers once a month, and replace the damaged ones in time
2. Keep temperature-sensing element clean, and clear the dust, oil pollution, etc. in time
3. Some standby sprinklers should be prepared
1. All valves should be kept in the prescriptive status by seals
2. Signs for valves should be installed with No.
3. Inspect the valves in outdoor valve wells once every month, and ensure the valves are in open
1. Test the function of alarm valve once every quarter
2. All component of alarm valve are complete; water pressure gauge and barometer are be calibrated
3. The alarm valve disc will be opened automatically if the test valve is turned on, the continuous audible alerts will be caused by
the water-driven alarm in 5–90 s, the inspection should be carried out several times, please repair it in time if the action failure
Inspect the functioning of the water flow indicator once every two months
1. The hydrant box should be clean, dry and neat
2. Lubricating oil should be added to the scroll periodically
3. No leakage appeared in pipe junctions
4. All components in the box should be intact and in good condition
5. A comprehensive inspection and test should be carried out every half a year
1. Rotate the valve of the hydrant by a specialized wrench, and inspect the flexibility of operation, lubricating oil should be
added in necessary
2. The outlet blank cap should be sealed without defects
3. No spilling or rust oil paint appeared on the outside surface, or repair it in time
4. Test water supply for all hydrants on the ground one by one during the period after the spring and before the winter in every
year
5. Inspect the valve well before the hydrant regularly
6. Make sure the supported component are complete and effective
1. The sign with corresponding system and No. should be installed
2. The inlet and outlet valves should be always kept open, and the sign are correct
3. Pressure gauge, test valve and overpressure protection device should be in good condition
4. Turn on the electric-powered pump once a month or the diesel-engine-driven pump once a week. Turn on the pump
automatically once a month for the auto-controlled pump
5. On-site turn on and inspect the pump in the pump room every quarter
6. Remote turn on and inspect the pump in the fire control room every quarter
1. Smoke exhaust and air supply vents are in good condition, and no barriers are around them
2. Connection between the vent and duct and the screws are in good condition
3. Air velocity satisfy the design requirements
4. Air volume satisfy the design requirements
1. Whether the drive mechanism is distorted or damaged. Whether the impeller is touched the shell
2. Whether the wiring to the motor is loose. Whether the corrosion phenomena is appeared in the shell
3. Inspect the status of the lubricating oil in the scroll (dirt, mixed with mud, dust, etc.)
4. Inspect the liquid level of the lubricating oil in the scroll
5. Inspect the tightness of drive belt and linkage spindle
6. Inspect the abnormal vibration or noisy produced from the working motor
7. Turn on the fan without load through the fire control center or fan room every 1–2 weeks
8. Demarcate the air volume of fan every year
1. Whether the fire damper is worked according to the design requirements
2. Inspect the work condition and the reliability of the fire damper regularly, make sure its accuracy, or clear the failure in time
3. Repair the parts of the fire damper in deformation or corrosion in time and lubricating oil should be added to the scroll
periodically
4. Fuse satisfy the work requirements, make sure the function is in good condition by manual operation once or twice, check the
adjustment of blade angle, inspect the performance of mechanical spring and ensure it is efficient
1. No deformation, damage and corrosion is appeared in the box
2. The distribution diagram and operating instruction are complete
3. Pointers of voltmeter and ammeter are in the prescribed range
4. The switches are in good conditions
5. Inspect the performance of the switches and the status of the indicator lights
6. Inspect the work condition of relay and switcher
1. No obviously corrosion and mechanical damage is observed, and the sigh is clear
2. Action and indicator light inspection should be carried out by specialized instrument in batches every quarter
3. Inspect and test the installed detectors by specialized instrument every year
4. Clean and inspect the performance of all detectors every three years, only the detectors passed the test can continue to use
1. Good appearance, complete components and clear sign
2. Inspect the performance of the button, glint will be produced if press the button
3. The furthest distance from the fire compartment is less than or equal to 30 m
4. Fire alarm signal will output if press the button
1. Good appearance, complete components and clear sign
2. The ground wire is used by the RE wire, and the sign is obvious and lasting
3. The controller power is from special fire power or special battery, and it is connected to fire power directly without plugs
(continued on next page)
34 F. Liu et al. / Safety Science 94 (2017) 26–40

Table 2 (continued)

Third grade indexes Fourth grade indexes (bottom indexes)

Linkage control A344
Fire broadcast, communication, and
elevator A345
4. Inspect the function of the controller and fill in the record every day
5. Clean the controller periodically
6. Inspect the battery level by specialized instrument and replace the invalid battery in time
1. Inspect the performance of linkage and feedback of the relevant every month
2. Inspect the function of indicator and display of the linkage controlled equipment by manually or automatically control
1. The distance from the speaker to any point in the floor is less than or equal to 25 m, and rated power is less than or equal to
3W
2. The sound pressure level at the furthest point in the broadcast range is 15db higher than background noise
3. The fire broadcast can be switched mandatory from the regular broadcast for the multipurpose broadcast system in the fire
control room
4. Broadcast for certain floors (districts) can be controlled in fire control room
5. Conversation between the fire control room and equipment room is in good condition
6. Conversation between the fire control room and manual fire alarm button is in good condition
7. Direct alarm telephone of outside call to the local fire department should be installed in fire control room
8. Fire elevator can be descended to the ground floor manually or automatically if a fire is confirmed
9. The signal of linkage control to the elevator can be returned to the fire control room

Table 3
Experts’ comments collection table. (5) Based on the previous results and Eqs. (5) and (6), the blind
Index 1 Index 2 Index 3 Index 4 Index 5 ... understanding degree rj on the indexes from all experts
Expert 1 could be obtained. Then the evaluation vector X could be cal-
Expert 2 culated according to the blind understanding degree rj and
Expert 3 Eq. (7). Finally, the weight of each index could be achieved
Expert 4
by the normalized treatment method. The calculated result
...
of each parameter is shown in Table 5.
A2 , A3 belong to secondary indexes, and should be ranked indi-
Table 4 vidually to determine the weight index. Similarly, weight distribu-
The rank result from experts. tion of the third grade indexes can also be obtained by refining. The
index distribution is shown in Tables 6 and 7.
Index A1 Index A2 Index A3 Index A4 Index A5
Group 1 2 1 1 3 4
Group 2 3 2 1 4 5
Group 3 1 2 2 3 4 2.4. Scoring and calculation
Group 4 3 2 1 4 5
2.4.1. Scoring instructions
All indexes extended from the first grade should be scored, it
HVAC, electric and fire protection were invited to evaluate the means that the second grade indexes from causing disaster factor,
qualitative rank on all indexes except the indexes in the bottom fire protection management and fire-fighting ability of fire brigade,
grade. The weight determination of the indexes in the first grade the third grade indexes of passive fire prevention measures, and
was taken as an example: the fourth grade indexes of active fire prevention measures should
be scored.
(1) The rank result from the 4 groups of experts were collected The score of ‘‘1” to ‘‘5” points was given to each index in the sec-
in Table 4: ond grade of index A1 , A4 , A5 , and the scoring standards mainly
(2) The obtained rank matrix: based on the CIBW14 ‘‘work report” finished by the International
Architectural Research Union, Code for fire prevention design of
0 1 interior decoration of buildings (GB50222-2015) (Supervision,
2 1 1 3 4
B3 4 1 4 5C 2015) and Unified standard for reliability design of building struc-
B C
A¼B C ð9Þ tures (GB50068-2008) (PRC, 2008), some specific details referred
@1 2 2 3 4A
to national standards guide for fire risk assessment of building
3 2 1 4 5 property insurance(draft) (Yumin, 2014). Scoring rules of indexes
in the grade of index A1 , A4 , A5 are shown in Table 8.
(3) The calculated membership matrix B based on the Eqs. (4) For the bottom indexes of index A2 and A3 the scoring method of
‘‘0-1-3-5”and ‘‘0-1-5-7” was adopted based on the instruction in
0and (9), m was taken as 7. 1 the relevant codes (Hai-yun, 2011). This method of scoring was
0:8982 1 1 0:7737 0:6131
B 0:7737 0:6131 1 0:6131 0:3869 C
originated from the Grade Security Assessment, which is a semi
B C
B¼B C ð10Þ quantitative investigation method used for the preparation of
@ 1 0:8982 0:8982 0:7737 0:6131 A chemical industry proposed by Philli Petroleum Company (Zhang,
0:7737 0:8982 1 0:6131 0:3869 2011). According to the actual situation of China, the scoring
method has been improved, and the score is distinguished based
(4) The average understanding degree on a particular dimension on the specific words in the regulations, such as the positive words
from all experts: ‘‘should”, ‘‘must” and the negative words ‘‘should not”, ‘‘cannot”
b1j þ b2j þ b3j þ b4j and ‘‘must not”. For example, positive word ‘‘should” and negative
bj ¼ ¼ ð0:8614; 0:8524; 0:9746; 0:6934; 0:5000Þ words ‘‘should not” were used in the code regulations, the medium
4
ð11Þ risk score series ‘‘0-1-3-5” could be applied. Similarly, positive
word ‘‘must” and negative words ‘‘must not” were used in the code
 F. Liu et al. / Safety Science 94 (2017) 26–40 35

Table 5
Weight distribution of the first grade index.

Index A1 Index A2 Index A3 Index A4 Index A5

Group 1 2 1 1 3 4
Group 2 3 2 1 4 5
Group 3 1 2 2 3 4
Group 4 3 2 1 4 5
Average cognition degree bj 0.8614 0.8524 0.9746 0.6934 0.5000
Maximum 1 1 1 0.7737 0.6138
Maximum bj 0.1386 0.1476 0.0254 0.0803 0.1138
Minimum 0.7737 0.6138 0.8982 0.6138 0.3869
bj  Minimum 0.0877 0.2386 0.0764 0.0796 0.1131
rj 0.1132 0.1931 0.0509 0.0800 0.1135
1- rj 0.8869 0.8069 0.9491 0.9201 0.8866
Xj 0.76393 0.68780 0.92499 0.63796 0.44327
Weight 0.2209 0.1989 0.2675 0.1845 0.1282

Table 6 Firstly, the score of each bottom index could be acquired

The subordinate index weight distribution of A2 . through the score evaluation method introduced. Secondly, based
Index A21 A22 A23 A24 on the extended grades, the solution could be described as three
Weight 0.2183 0.2863 0.2235 0.2719
options. For the indexes A1 ; A4 and A5 , the scores Gi could be calcu-
lated by the principle shown as follows:
Pn
k¼1 Z k
Table 7
Gi ¼  100 ð12Þ
Si
The subordinate index weight distribution of A3 .
where Gi is the score of the index Ai on the first grade, i ¼ 1; 4; 5; Z k
Index A31 A32 A33 A34
is the bottom index score under the corresponding first grade index;
Weight 0.2592 0.2145 0.2531 0.2732 Si is the sum of the maximum score of each bottom index under the
Index A311 A312 A313 A314
corresponding first grade index; n is the number of the bottom
Weight 0.1716 0.1969 0.1732 0.1576
Index A315 A316 A321 A322 index under the corresponding first grade index.
Weight 0.1735 0.1273 0.3684 0.2795 For the index A2 , the score G2 could be calculated by the princi-
Index A323 A331 A332 A333 ple shown as follows:
Weight 0.3521 0.2456 0.2456 0.2867 Pn
Index A334 A331 A342 A343 k¼1 Z k
Weight 0.2221 0.2227 0.1574 0.2249 G2j ¼  100 ð13Þ
S2j
Index A344 A345
Weight 0.1723 0.2227 X
4
G2 ¼ x2j G2j ð14Þ
j¼1
regulations, the high risk score series ‘‘0-1-5-7” could be applied
(Zhang, 2011). where G2j is the score of the index A2j on the next grade of A2 ; Z k is
In the practical evaluation process, score ‘‘0” was given to the the bottom index score under the index A2j ; S2j is the sum of the
condition was totally unaccepted, score ‘‘1” was given to the con- maximum score of each bottom index under the index A2j ; n is
dition was much lower than the requirement in the code regula- the number of the bottom index under the index A2j ; x2j is the
tions, the higher score of ‘‘3” or ‘‘5” was given to the condition weight of the index A2j .
was slightly lower than the requirement in the code regulations, For the index A3 , the score G3 could be calculated by the princi-
and the highest score of ‘‘5” or ‘‘7” was given to the condition ple shown as follows:
was completely complied to the requirement in the code regula- Pn
k¼1 Z k
tions (Wu and Liu, 2011). The standard is mainly based on the G3ij ¼  100 ð15Þ
S3ij
Chongqing municipal engineering construction standards, Code
X
m
for the fire prevention design of large-scale commercial buildings G3i ¼ x3ij G3ij ð16Þ
(Committee, 2013a, 2013b), referred to as ‘‘commercial code”. j¼1
The indexes need to be scored by the method of Grade Security X
4
Assessment are shown in Table 9. G3 ¼ x3i G3i ð17Þ
Five scores including 5, 4, 3, 2 and 1 were used to evaluate the i¼1
score the index related to maintenance and management, and the
where G3ij is the score of the index A3ij on the next grade of A3i ; Z k is
score instructions is shown in Table 10.
the bottom index score under the index A3ij ; S3ij is the sum of the
maximum score of each bottom index under the index A3ij ; n is
2.4.2. Calculation results
the number of the bottom index under the index A3ij ; G3i is the score
If the value of each indexes in the first grade of the five main
of the second grade index; x3ij is the weight of the index A3ij ; m is
aspects can be calculated, the final result from the assessment sys-
the number of the third grade index under the index A3i ; x3i is the
tem will be obtained. Consequently, the solution of the value for
weight of the index A3i ;
each index in the first grade is the key to obtain the final result.
Thirdly, based on the results of G1 ; G2 ; G3 ; G4 and G5 , the final
Based on the indexes system established previous, single or multi
score could be calculated by the principle shown as follows:
grades indexes extended from each indexes in the first grade, for
example, one grade extended from indexes of A1 ; A4 and A5 , while X
5
two grades extended from index A2 , three grades extended from G¼ xi Gi ð18Þ
index A3 . The calculation method can be described as follows: i¼1
36
Table 8
Detailed scoring rules of A1 , A4 , A5 .
Second grade indexes
(bottom indexes)
Distribution plan A11
Occupant density A12
Internal fire load A13
Internal decoration A14
Construction height A15
Building service life A16
External environment
A17
Fire safety responsibility
system A41
F. Liu et al. / Safety Science 94 (2017) 26–40
Detailed scoring rules Score Remarks
1. Any aspect of the design and installation on distribution 1 r means the ratio of the maximum electrical load to the design
system, lightning protection system and ground system is not load
satisfied the code requirements, or the safety inspection on the
power is failed to pass, r > 1:2
2. The design and installation on distribution system, lightning 2
protection system and ground system are basically conformed
with the code requirements, the safety inspection on the power
is passed, and 1:0 < r 6 1:2
3. The design and installation on distribution system, lightning 3
protection system and ground system are basically conformed
with the code requirements, the safety inspection on the power
is passed, and 0:8 < r 6 1:0.
4. The design and installation on distribution system, lightning 4
protection system and ground system are basically conformed
with the code requirements, the safety inspection on the power
is passed, and 0:5 < r 6 0:8
5. The design and installation on distribution system, lightning 5
protection system and ground system are basically conformed
with the code requirements, the safety inspection on the power
is passed, and 0 < r 6 0:5
1. q>2 1 Determination method of the occupant density q:
2. 1<q62 2 (1) According to the statistical data. Leisure: 0.4–1.0, edu-
3. 0:5 < q 6 1 3 cation: 0.7–1.0, restaurant: 0.5–0.8, shopping mall and
4. 0:2 < q 6 0:5 4 office: 0.2–0.5
5. q 6 0:2 5 (2) The average occupant density was calculated by the
peak flow and low peak flow
1. d > 800 1 Referred to the average values of fire load density in different
2. 600 < d 6 800 2 buildings on the ‘‘work report” of CIBW14 reported by the
3. 400 < d 6 600 3 International Architectural Research Association
4. 200 < d 6 400 4
5. 0 < d 6 200 5
1. Combustible materials are used in the ceiling, and flammable 1 According to the relevant regulations in the code of GB50222-
materials are used in the wall or ground. 2001
2. Flame retardant materials are used in the ceiling, combustible 2
(flammable) materials are used in the wall or ground.
3. Flame retardant materials are used in the ceiling, wall and 3
ground.
4. Non-flammable materials are used in the ceiling, and flame 4
retardant materials are used in the wall and ground.
5. Non-flammable materials are used in the ceiling, wall and 5
ground.
1. H > 100 m 1 According to the regulation of 5.1.1 in the code of GB50016-
2. 50 m < H 6 100 m 2 2013
3. 24 m < H 6 50 m 3
4. H 6 24 m in the multi-story building 4
5. Single story building 5
1. Service life > 4n=5 1 According to the regulation of 1.0.5 in the code of GB50068-
2. 3n=5 < Servicer life 6 4n=5 2 2001
3. 2n=5 < Servicer life 6 3n=5 3
4. n=5 < Servicer life 6 2n=5 4
5. 0 < Servicer life 6 n=5 5
1. Closed to the dangerous buildings of producing or storing 1 Yearly dominant wind direction and the dangerous of the
combustible or explosive materials, and it is in the downstream adjacent building are the main consideration
direction of yearly dominant wind direction of the dangerous
buildings
2. Closed to the dangerous buildings of producing or storing 3
combustible or explosive materials, and it is in the upstream or
side direction of yearly dominant wind direction of the
dangerous buildings
3. No dangerous buildings of producing or storing combustible 5
or explosive materials around
1. All three aspects are not perfect 1 Including three aspects:
2. Any aspect is not perfect 3 1. Fire safety system is established
3. Three aspects are perfect 5 2. Clarify the responsibilities of departments and staff are
clear, and execute strictly
3. Fire safety management file is established
 F. Liu et al. / Safety Science 94 (2017) 26–40 37

Table 8 (continued)

Second grade indexes Detailed scoring rules Score Remarks

(bottom indexes)

Fire emergency plan A42 1. No emergency plan 1 Including four aspects:

2. Any three aspects are not satisfied 2 1. Effectiveness of emergency resources
2. Reasonability of command, adjust and response capability
3. Any two aspects are not satisfied 3 in fire emergency plan
4. Any one aspect is not satisfied 4 3. The fluent possibility of announcement and communication
5. All four aspects are satisfied 5 4. Organization of the fire drill per periodically

Fire safety training A43
Fire safety inspection A44
Fire lane A51
Rescue capability of the
local fire brigade A52
1. Neither of the two aspects cannot be executed 1 Including t two aspects:
2. Only the first aspect can be executed 2 1. Pre-job fire safety training and regular fire safety education
3. Only the second aspect can be executed 3 should be carried out for all the staff
4. Both two aspects can be executed without record 4 2. The special workers engaged in dangerous materials and
5. Both two aspects can be executed with clear record 5 processes and fire safety system should be certified by cor-
responding credentials
1. None of three aspects cannot be executed 1 Including three aspects:
2. Any one aspect cannot be executed 3 1. Determination of the key location and work position
3. All three aspects can be executed 5 2. Implementation of daily inspection, and establishment of
inspection record
3. Timely rectification of the potential fire risk
1. Not satisfy the code requirements or cannot be ensured 1 According to the regulation of 7.1 in the code of GB50016-2013
smooth
2. Satisfy the code requirements and can be ensured smooth 5
1. Any one aspect is not satisfied 1 Including three aspects:
2. Both two aspects are satisfied 5 1. Arrive at fire scene in 5 min after receive the fire alarm
2. Rescue equipment satisfy the requirements of the code
‘‘Construction Standards on Urban Fire Station”

Table 9
Security classification score sheet.

Second grade indexes Third grade indexes Source Score

Fire resistance rating A21 1. Whether the fire resistance is level 1 or not 3.0.2 of ‘‘commercial code” 0-1-
3-5
2. During the design and construction, the materials must satisfy the national laws Article 11 of ‘‘Fire Protection Law” 0-1-
and codes 5-7
3. Whether the decoration materials satisfy the requirements or not, and whether 4.5.2 of ‘‘commercial code” 0-1-
them were non-combustible or fireproof 3-5
Fire compartment A22 1. Whether fire compartment satisfy the codes or not 5.1.1 of ‘‘commercial code”,5.1.2 of 0-1-
‘‘commercial code” 3-5
Smoke zone A23 1. Whether smoke zone satisfy the codes or not 5.2.1 of ‘‘commercial code”,5.2.2 of 0-1-
‘‘commercial code” 3-5
Emergency lighting and 1. Whether the minimum width of the escape route satisfy the codes or not 6.3.2 of ‘‘commercial code” 0-1-
evacuation indicator A24 3-5

Table 10 experts who have significant impact on the fire risk evaluation
Scoring instructions. have been invited to score all the bottom indexes.
Score Instruction
Take BUILDING 1 for example, the result of all bottom indexes
scored from the expert group is shown in Fig. 14.
5 Maintenance management is perfect
The scores of G1 , G2 , G3 , G4 and G5 corresponding the five
4 Maintenance management is good
3 Maintenance management is mediocre indexes in the first grade could be calculated based on the pro-
2 Maintenance management is poor posed method and the Eqs. (12) to (18), the final score G and its fire
1 Maintenance management is very poor risk level of the BUILDING 1 could be achieved based on Table 11at
last. Similarly, the total scores and their fire risk levels of BUILDING
2, BUILDING 3 and BUILDING 4 could be obtained. The results of
where Gi is the score of the first grade index Ai ; xi is the weight of BUILDING 1, BUILDING 2, BUILDING 3, and BUILDING 4 are shown
the index Ai . in Table 12.
As a result, the corresponding fire risk level of large-scale com-
mercial buildings could be identified as shown Table 11. The algo-
rithm used in this study is illustrated in Fig. 13. 4. Discussions

Based on the result of the case study on four large-scale com-

3. Results (Case study)
Four large-scale commercial buildings marked from BUILDING 1
to BUILDING 4 were selected as case study to test the proposed fire
risk assessment system. According to the requirement of the fire
assessment proposed in this paper, an expert group with seven
mercial buildings, the total scores were more than 65 points. How-
ever, there were some differences in some aspects in them.
In the aspect of the causing disaster factors (G1 in Table 12),
BUILDING 2 was a large shopping center located in the CBD of
Chongqing, and it has been reconstructed for many times during
the past 20 years from opened. Besides, the fire load density is very
38 F. Liu et al. / Safety Science 94 (2017) 26–40

Table 11 high due to large population inside. As a result, the G1 score of

Fire risk level table. BUILDING 2 was lower than the other 3 buildings. In addition,
Number Risk Score Basic characteristics BUILDING 3 and BUILDING 4 were located in the new urban finan-
level range cial district of Chongqing, their planning and design were more
1 I 90–100 System operation condition is very good, fire reasonable than BUILDING 2.
risk is very low In the aspect of passive fire protection method (G2 in Table 12),
2 II 70–90 System operation condition is good, fire risk in the old shopping centers, lots of goods were normally stored in
is lower
3 III 50–70 System operation condition is mediocre, fire
the shop even in the passageways and near the fire door. Therefore,
risk is mediocre the evacuation signs were covered, the width of the evacuation
4 IV 30–50 System operation condition is poor, fire risk is was reduced and the security exits were blocked. The people inside
higher could not successfully evacuate from the building in case of a fire.
5 V 0–30 System operation condition is very poor, fire
During the on-site inspection, the previous scenarios were
risk is very high
appeared in BUILDING 2. Besides, some goods were stored under
the fire shutter and manual control system of fire shutter was

Determining the first grade indexes in fire risk

assessment of large-scale commercial buildings

Establishing the indexes Establishing the indexes Establishing the Establishing the Establishing the
system of causing system of passive measures. indexes system of indexes system of indexes system of fire
disaster factors. active measures. fire management. brigade fire ability.
Scoring the third
Scoring the second Scoring the fourth Scoring the second
grade indexes. Scoring the second
grade indexes. grade indexes. grade indexes.
grade indexes.
Calculating the score of
Calculating the Calculating the score of Calculating the Calculating
the second grade and
score of the first the third grade, second score of the first the score of the
the first grade indexes,
grade indexes. grade and the first grade grade indexes. first grade indexes.
in turn.
indexes, in turn.

Calculating the final score of building fire risk assessment.

Determining the fire risk level according to the final score

Fig. 13. Judgment of fire risk level.

Fig. 14. Bottom index scoring result of BUILDING 1.

 F. Liu et al. / Safety Science 94 (2017) 26–40
Table 12
Scores and risk levels of 4 buildings.
G1 G2 G3
BUILDING 1 68.57 80.91 89.07
BUILDING 2 60.00 71.26 77.53
BUILDING 3 91.42 80.91 79.65
BUILDING 4 88.57 88.97 94.46
out of work, hence the fire compartment could not be divided by
the fire shutter. So the G2 score of BUILDING 2 was lower than
the other 3 buildings.
In the aspect of active fire protection method (G3 in Table 12):
The G3 of BUILDING 2 and BUILDING 3 were lower than the other
two buildings. This was mainly due to the scheduled maintenance
and periodic inspection were lacked in BUILDING 2 and BUILDING
3. To be sure that the operation reliability of fire equipment would
be lower without scheduled maintenance and periodic inspection.
In these buildings, it was found that some devices were rusty or
even abandoned. During the on-site inspection, some truth was
found, such as some devices were rusty or even abandoned, some
functions of the on-line monitoring system were out of use, etc. In
general, the performance of active fire protection in the four build-
ing were good, and the average score of was over 77.
In the aspect of fire protection management capacity (G4 in
Table 12), on-site investigation indicated that only few staff from
the four buildings have participated the regular fire training. There
was no plan on the perambulation and inspection of fire hazard in
some high risk and important areas. But the overall result shown a
good behavior in fire safety responsibility, because all staff was
clear to his responsibility and the record and preservation work
of fire safety management documents were good.
In the aspect of fire-fighting ability of fire brigade (G5 in
Table 12), BUILDING 1 and BUILDING 2 are located in the CBD of
Chongqing, it means that the traffic around was congested. The
local fire brigade could not arrive at in 5 min after receiving fire
alarm. Nowadays, fire safety in large-scale commercial buildings
is strictly mandated by laws. As the fire codes are being rigorously
updated gradually, the maintenance of fire equipment in these
buildings becomes more and more important. To increase the suc-
cess rate of fire equipment operation can protect people’s lives and
properties effectively. Therefore, the daily maintenance of fire
equipment was also very important after the large-scale commer-
cial buildings construction completely. It is recommended that
government should increase the strength on supervision and
inspection in these buildings, and the improvement should be
enforced to execute by the management department of the build-
ing if the internal fire hazards were discovered.
5. Conclusions
In this paper, the indexes were divided into multi grade in the
proposed fire assessment system based on the AHP. This paper
focus on the maintenance of fire equipment, hence in order to
increase the assessment accurateness, the next grade indexes
extended from passive and active fire protection strategies were
proposed. Very rigorous methods were adopted in the basic data
collection, including on-side investigation and measurement,
maintenance record, insurance and third party evaluation compa-
nies visit and experts interview.
The structure entropy weight method was introduced, the rea-
sonable index weights were determined by experts’ comments col-
lection, blind degree analysis and normalized treatment. Detailed
score instructions have been stated in the proposed assessment
system based on the relevant code regulations, the influence
39
G4 G5 G Level
90.00 60.00 79.36 II
70.00 60.00 68.77 III
90.00 100.00 87.02 II
85.00 100.00 91.03 I
caused by subjective factors could be avoided. The proposed
assessment system is not only easy but also practical applied in
the engineering project. However, in order to widely use this sys-
tem in the engineering project, further optimization need to be car-
ried out.
The case study shows that the proposed system can be applied
feasibility and operability in the four large-scale commercial build-
ings. In the four buildings, the fire safety adjustments in BUILDING
2 need to be conducted to reduce its fire risk, because it was under
the highest fire risk.
Acknowledgements
This work was supported by key program foundation of Min-
istry of Public Security, China (Grant 2011ZDYJCQXF019), graduate
scientific research and innovation foundation of Chongqing, China
(Grant No. CYB16006) and the 111Project, No. B13041. We grate-
fully acknowledge the assistance of Mr. Shujiang Liao and Mr.
Xiangyang Zhang from Chongqing public security fire department
in the on-site investigation and data collection.
References
Beck, V., 1998. CESARE-RISK: A tool for performance-based fire engineering design.
In: Proceedings of the 2nd International Conference on Performance-based
Codes and Fire Safety Design Methods.
Benichou, N., 2005. FIERA system: a fire risk assessment tool to evaluate fire safety
in industrial buildings and large spaces. J. Fire. Prot. Eng. 15 (3), 145–172.
China, M. o. H. a. U.-R. C. o. t. P. s. R. o., 2014. Code for Fire Protection Design of
Buildings.
Committee, C.U.C., 2013. Chongqing Fire Regulations.
Committee, C.U.C., 2013. Code for Fire Protection Design of Large-scale Commercial
Buildings.
Dalkey, N., 1969. An experimental study of group opinion. Futures 1 (5), 408–426.
Fraser-Mitchell, J.N., 1994. An Object-Oriented Simulation (Crisp II) for Fire Risk
Assessment .
Hai-yun, H.A.N., 2011. Research on standardization method of risk assessment for
fire public liability insurance in assembly occupancies and underwriting
auditing. Proc. Eng. 11, 120–126.
Jiang, J., Chen, L., Jiang, S., Li, G.-Q., Usmani, A., 2015. Fire safety assessment of super
tall buildings: a case study on Shanghai Tower. Case Stud. Fire Saf. 4, 28–38.
Ke-xuan, L., 2013. A renovated method of fire risk evaluation of large emporium
based on ANN and grey correlation. J. Saf. Environ. 13 (1), 254–258.
Lau, C.K., Lai, K.K., Lee, Y.P., Du, J., 2015. Fire risk assessment with scoring system,
using the support vector machine approach. Fire Saf. J. 78, 188–195.
Li, X., Wang, K., Liu, L., Xin, J., Yang, H., Gao, C., 2011. Application of the entropy
weight and TOPSIS method in safety evaluation of coal mines. Proc. Eng. 26,
2085–2091.
Omidvari, M., Mansouri, N., Nouri, J., 2015. A pattern of fire risk assessment and
emergency management in educational center laboratories. Saf. Sci. 73, 34–42.
PRC, M. o. C. o. t., 2008. Unified Standard for Reliability Design of Building
Structures.
PRC, M. o. C. o. t., 2013. Code for Installation and Commissioning of Sprinkler
Systems.
PRC, M. o. C. o. t., 2014. Code of Design on Fire Protection Water supply and Hydrant
Systems.
Qi-yue, C., 2010. Structure entropy weight method to confirm the weight of
evaluating index. Syst. Eng. Theory Pract. 30 (7), 1225–1228.
Qian-li, M.A., Ting-lin, H., 2011. Analysis of and study on the difficulties in the fire
protection design of large commercial complex. Proc. Eng. 11, 302–307.
Qian, D., 2012. Risk evaluation of urban fire by regression model with latent
variables. China Saf. Sci. J. 7 (22), 159–163.
Sun, X.-Q., Luo, M.-C., 2014. Fire risk assessment for super high-rise buildings. Proc.
Eng. 71, 492–501.
Supervision, S. B. o. T., 2013. Code for Design of Automatic Fire Alarm System.
Supervision, S. B. o. T., 2015. Code for Fire Prevention Design of Interior Decoration
of Buildings.
40 F. Liu et al. / Safety Science 94 (2017) 26–40

Wu, J.-J., Liu, X.-C., 2011. Risk assessment of underground coal fire development at Yumin, T., 2014. Guide for Fire Risk Assessment of Building Property Insurance.
regional scale. Int. J. Coal Geol. 86 (1), 87–94. Zhang, D., 2011. The Inspection and Evaluation of Fire Safety System.
Yang, F., Qian, X., Huang, P., 2012. Fire safety assessment of underground buildings Zheng, W., 2014. Fire safety assessment of China’s twelfth national games stadiums.
based on grey relational analysis. Proc. Eng. 45, 89–95. Proc. Eng. 71, 95–100.