Safety risk assessment model of a subway

tunnel collapse system based on improved

DOW-FFTA method
..............................................................................................................................................................

Wen Li1,2 , Xuesong Lu1,2, *,† and Menglong Wu3

1

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

School of Architectural Engineering, Huanggang Normal University, Huanggang 438000,
China; 2 Huanggang Ecological Architecture and Renewable Resources Research Center,
Huanggang 438000, Hubei, China; 3 School of Resource and Environmental Engineering,
Wuhan University of Science and Technology, Wuhan 430081, China
.............................................................................................................................................
Abstract
Green construction considers factors such as quality, safety, efficiency, environmental protection and
ecology. Under the premise of balancing basic construction capacity and green construction, prioritizing
construction safety requirements and exploring the risk assessment system for the green construction of
subway projects is crucial. Traditional risk assessment methods in the study of local system risks have been
mostly applied to simple systems; thus, risk identification and assessment methods lack universality. The
questions of how to change the existing structure of the system safety risk assessment model and establish a
realistic modeling approach, as well as implement dynamic risk supervision, have become urgent problems.
Here, we investigated the event risk point and proposed an integrated evaluation method of inherent, initial
and real risks based on system attributes. The improved DOW chemical method was used to solve the
static inherent risk severity index, and fuzzy fault tree analysis (FFTA) was used to judge the probability
of out-of-control local state management. Considering the abnormal situation and the emergence of new
risk information, a dynamic correction model was proposed. Finally, a static risk assessment model was
established in line with the actual state of the system with local management and realistic risk assessment
models that could modify the initial risk over time.

Keywords: system security risk; green construction; inherent risks; risk assessment model; subway tunnel
collapse
*Corresponding author:
[email protected] Received 23 May 2023; revised 28 July 2023; accepted 14 August 2023
.................................................................................................................................................................................

1 INTRODUCTION Early accident control theories mainly highlighted the role of

‘unsafe human behavior’ and ‘unsafe state of objects’ in trigger-
The concept of building safety and green construction is increas- ing accidents [4, 5]. The traditional high-risk industry fits the
ingly valued by people as the development of society progresses model of accident causation theory because of its dense allocation
and low-carbon construction [1–3]. To comprehensively imple- of personnel on site, use of dangerous materials and complex
ment safe and green construction of building projects, the sig- technology. However, the traditional accident control model con-
nificance of safety and green construction has been analyzed siders the accident as the research object, which has inherent
based on risk factors. Owing to the complexity and uncertainty of ‘lag’ and ‘passivity’; therefore, the causes searched and measures
urban underground engineering, improper construction control formulated are not universal. In recent years, the investigation of
of subway tunnels can easily lead to collapse accidents. Therefore, many major accidents has focused on ‘people’s safety awareness,’
evaluating the safety risks of subway tunnel systems from the ‘management mode,’ ‘system implementation’ and other aspects
perspective of green construction may prevent their collapse. [6, 7]; however, the aspects of enterprise safety law, essential

† Wen Li, https://orcid.org/0000-0002-4504-0773

International Journal of Low-Carbon Technologies 2024, 19, 171–184
© The Author(s) 2024. Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which
permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
https://doi.org/10.1093/ijlct/ctad101 171
W. Li et al.

characteristics of accidents and production system have not been with the local system to modify the system property management
explored in depth. Therefore, although companies may invest status, form the initial risk. The new safety information affects the
substantial workforce and material resources to detect hidden safety of the whole system, re-combined with the new dynamic
dangers, they are unable to control the occurrence of accidents [8]. information to modify the properties of the system, forming the
Therefore, conducting safety risk research on the green construc- real risk [14]. Therefore, the concepts of inherent, initial and real
tion of subways is of utmost importance [9]. risks were interpreted based on the attributes of the characteristics
The early ‘4M1E’ method from the ISOquality management of the system and management state of tunneling works. The
system involved workforce, machines, materials, methods and emphasis of the analysis was different for the three types of risk,
environment, which are regarded as the five elements of on-site but there were some correlations, as discussed in Sections 2.2 to
management. Cha et al. [10] analyzed how to integrate the pro- 2.5. As shown in Fig. 1, the improved DOW chemical method was
duction resources of 4M1E and improved the management effi- used to assess the local static inherent risk of the system attributes.

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

ciency by integrating the management information and using the Fuzzy fault tree analysis (FFTA) method was used to analyze the
functions of standardized data and digital information. Wu et al. initial risk of dynamic changes caused by improper management.
[11] incorporated the 4M1E elements of field management into Finally, considering that the special period and the integration of
laboratory management to help improve management efficiency. new information affect the overall risk status of the system, the
Mao and Xu [12] investigated the relative importance of the five system was modified as a whole and the real risk assessment model
factors and whether the influence of 4M1E factors on engineering was constructed.
quality is significant, indicating that the factors of human, envi-
ronment and machinery are important for engineering quality.
Key factors determine the focus of the risk assessment. 2.2 Improved DOW method to determine the
Because the operation process is dynamic and continuous, new inherent-risk index
research methods based on the safety risk of the system have According to the inherent characteristics of the system attributes,
been proposed to reflect the changing trend of risk accumulation. the tunnel safety system was quantitatively analyzed. Several sim-
Li et al. [13–15] proposed an aggregation method based on risk ilarities (items a–e below) were observed between the inherent-
concept for event occurrence probability and consequence sever- risk assessment of the tunneling work safety system and DOW
ity, focusing on system attributes and management state factors, method when evaluating inherent attributes:
and they established a green construction risk assessment model a. The modes of influence of hazard sources in the tunnel
based on system attributes and governance state. Wang et al. [16] collapse system are similar to those of fire and explosion disasters.
proposed a learning probability modeling method to support the b. The risk characteristics of the evaluation indices are similar
inspection of pipeline reliability and risk. Zheng et al. [17] applied but differ because the substances affecting fire and explosion
fuzzy synthesis to green environmental protection evaluation of disasters mostly consist of chemical energy, whereas the inherent
water quality in water samples. Therefore, the evaluation of safety risk of tunneling collapse is majorly potential energy.
risk assessment should present the following: objective risks of c. Although exothermic and endothermic reactions and other
high-risk substances, processes and sites; effectiveness of man- process characteristics are present in the tunnel collapse sys-
agement level; dynamic risks subject to online monitoring and tem, system risk characteristics also exist, such as pressurization,
early warning results; earthquakes; and dynamic accident hidden decompression and instability.
dangers. d. Similar to the engineering background and comment basis of
Because of this discrepancy, the existing risk assessment meth- the DOW method, both are based on the objective risk attribute
ods should be re-examined, and an assessment method should of accident disasters and obtain the risk index of main factors.
be developed based on integrating system attributes and manage- e. Both have a severe impact on personnel, equipment and the
ment status from a new perspective, while retaining the advan- environment once accidents occur.
tages of traditional assessment methods. In this study, we explored Several differences (items a–d below) were also observed
the mechanisms behind establishing static, local dynamic and real between the improved and traditional DOW methods:
risk assessment models based on the actual state of the system, and a. Different methods of unit division. The improved method
we elucidated timely corrections to provide a theoretical and tech- is divided into evaluation units based on process conditions,
nical basis for safety risk assessment in green tunnel construction. whereas the traditional method is divided into evaluation units
based on major accident risk points.
b. Different quantitative methods of indicators.
2 METHODS c. The meanings represented by each indicator are different.
d. Different application fields. The improved method is mostly
2.1 Develop the security risk assessment index system used in the chemical industry, whereas the traditional method
The system security risk includes static and dynamic risks, as the provides a new evaluation path for the safety risk assessment of
well as dynamic risk of the whole system. Unlike the traditional tunnel engineering.
model of risk assessment, taking the risk source as the starting Based on these observations, the improved DOW method was
point, the system properties for identification assessment key, applied in tunnel engineering to measure the risk of tunnel system

172 International Journal of Low-Carbon Technologies 2024, 19, 171–184

 Subway tunnel collapse system

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

Figure 1. Realistic risk assessment model of system risk.

attributes and determine its inherent risk. Furthermore, based material coefficient, main factors affecting stability and weight
on the risk characteristics of system attributes, the indices of coefficient, and its expression is:
operation, site, equipment and facilities affecting accident risk
points were considered the coefficients of inherent-risk compen- 
n
∗
MF = MF + wi Ai (i = 1, 2, · · · n) (1)
sation. Finally, the severity index and grade of the inherent risk of
i=1
tunnel safety system were determined, and the final inherent-risk
assessment process is illustrated in Fig. 2. where MF is the unstable material coefficient of energy;MF ∗ is
the basic material coefficient, and depending on the situation, the
general engineering rock mass class value is 10; Ai represents the
3 INHERENT-RISK INDICATORS FOR GREEN main internal influencing factor of the hazardous substance, with
CONSTRUCTION values ranging from 0 to 10; and wi represents the weight of each
factor affecting the material coefficient.
Three indicators of risk were defined for green construction (2) Assessing the process risk coefficient of the event risk point.
related to determining the energy material and process risk The process risk coefficient F 3 of the risk point assessment
coefficients at event risk points, as well as determining the object of a tunnel collapse accident comprises the general and
compensation coefficient of the inherent-risk index of tunnel special process risk coefficients F 1 and F 2 . General process risk
collapse. These indicators are described below. is the main internal factor that affects the damage consequences
(1) Determination of energy material coefficient at event risk of dangerous process accidents, whereas special process risk is
point. the main factor that may lead to accidents at risk points [18].
Owing to the diversification of incident risk points of the tunnel Referring to the DOW method, the process risk coefficient in
safety system, hazardous substances were classified to determine event risk points is calculated as:
the material coefficient MF. MF is determined by the material 
m
 
potential energy, kinetic energy and inherent characteristics of the F1 = Bj ηj j = 1, 2, · · · m (2)
event risk point, and it is an internal factor that affects the tunnel j=1
stability of the safety system.
MF is calculated by the instability of the tunnel safety system, where Bj is the risk value of class j hazardous process factors, and
and its value comprises the comprehensive value of the basic the value ranges from 0 to 5; ηj represents the weight coefficient

International Journal of Low-Carbon Technologies 2024, 19, 171–184 173

W. Li et al.

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

Figure 2. Inherent-risk assessment process of improved DOW method.

Figure 3. Tunnel construction collapse risk accident tree.

of class j dangerous process factors, and: where Ca is the risk of the main factors that lead to accidents,
ranging from 0 to 5, and μa is the weight coefficient of each factor

b
of the risk coefficient for a special process.
F2 = Ca μa (a = 1, 2, · · · b) (3)
The process risk coefficient F 3 is calculated as:
a=1

F3 = F1 × F2 (4)

174 International Journal of Low-Carbon Technologies 2024, 19, 171–184

 Subway tunnel collapse system

Table 1. Correction coefficient of hazardous environment. 4 INDEX OF INHERENT RISK SEVERITY FOR
Number of people in danger area Correction coefficient k1 GREEN CONSTRUCTION
>100 2.0
60–99 1.5 Referring to the DOW calculation method, the severity index h
30–59 1.2 caused by the inherent-risk index of the tunnel collapse accident
1–29 1.0 safety system is estimated as:
0 0.5

h = M × F3 × k1 × k2 × k3 (6)

Effective risk assessment grading standards must be formulated

(3) Compensation coefficient of the inherent-risk index of tun- after several rounds of adjustment combined with practical appli-

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

nel collapse. cation. These standards correspond to the grading method of the
The correction coefficient of inherent risk of the event risk DOW chemical hazard index, which is continuously modified.
point can be divided into three categories: dangerous environ- After checking and adjusting the case data, the grading standard
ment, dangerous operation and dangerous equipment and facil- was finally determined, as shown in Table 4, and it could meet the
ities correction coefficients. requirements of tunnel engineering safety system evaluation.
1) Hazardous environment compensation coefficient (k1 ). The
main reason for the high inherent-risk environment is the labor-
intensive place prone to accidents, which may affect numerous
4.1 Initial risk assessment based on FFTA
people, resulting in an increase in casualties. Therefore, the num- 5 FFTA DETERMINES THE RISK EVENT
ber of people exposed represents the risk to the environment. PROBABILITY
In recent years, the safety of subway tunnel construction strictly
controls the number of people exposed to dangerous places and The FFTA method can deduce the event, its causes at various
strictly prohibits over-quota production. According to the stan- levels and predict the causality of the event [19]. Owing to the lack
dard of identification of major hazard sources of hazardous chem- of data or unclear information, the exact values of the basic events
icals (GB 18218-2018), a table of the correction coefficients of are difficult to determine. However, the FFTA method can effec-
exposed personnel was constructed, and the correction coeffi- tively predict the probability of risk occurrence quantitatively. The
cients of exposed personnel applicable to the hazardous environ- FFTA evaluation steps are as follows.
ment of the subway tunnel safety system are shown in Table 1. Step 1: Expert judgment determines probabilistic fuzzy num-
2) Compensation coefficient for hazardous operation (k2 ). bers. During expert judgment, it is reasonable to use natural
The correction coefficient for dangerous operation mainly language, such as ‘small’ and ‘large,’ to describe the probability
includes the risk of the current situation of safety measures of events because it is difficult to accurately estimate. However,
such as high risk, special and special equipment operations. because natural language has fuzziness, it cannot be processed
The safety achieved by control measures at different risk points by traditional methods. Therefore, fuzzy set theory is used to
varies. The operation compensation coefficient was determined process this uncertain information. The natural language adopted
comprehensively from control, as shown in Table 2. If there by experts is usually represented by trapezoidal or triangular fuzzy
are numerous types of operations, the compensation coefficient numbers [14]. Let the expression of the triangle fuzzy number be
corresponding to the maximum-risk operation is determined as f = (a, b, c), and the expression of the trapezoidal fuzzy number
the compensation coefficient corresponding to the inherent-risk be f = (a, b, c, d). The membership function expression of both is
operation according to the principle of maximum impact and the probability of event occurrence:
maximum risk. This factor can be expressed as follows:
⎧
(k2 )max ∈ [k21 , k22 · · · , k2c , · · · k2d ] (c = 1, 2, · · · d) (5) ⎪
⎪0, x < a
⎨ x−a , a < x ≤ b
f = (x, a, b, c) = b−a
c−x (7)
where c is the number of types of operations. ⎪
⎪ ,b < x ≤ c
⎩ c−b
3) Compensation coefficient of dangerous equipment (k3 ). The 0, c < x
reliability of hazardous facilities is represented by the essential ⎧
safety level of risk points. Therefore, the value of the compensa- ⎪
⎪ 0, x < a
⎪
⎪ x−a
tion coefficient of the inherent risk of dangerous equipment and ⎨ b−a , a < x ≤ b
facilities is shown in Table 3. As the safety and reliability of the f = (x, a, b, c, d) = 1, b < x ≤ c (8)
⎪
⎪ d−x
equipment gradually decrease with the increase of service life, ⎪ d−c , c < x ≤ d
⎪
⎩
there may be a risk of expansion. Therefore, the compensation 0, d < x
coefficient of hazardous equipment and facilities is > 1 and ranges
from 1 to 1.7. When there is no equipment or facilities at the risk The fuzzy number from λ cut set is shown in Table 5, in which
point, the compensation coefficient k3 = 1. VL, L, FL, M, FH, H and VH represent the fuzzy language for

International Journal of Low-Carbon Technologies 2024, 19, 171–184 175

W. Li et al.

Table 2. Compensation coefficient for hazardous operations.

Status of safety measures Very good Fairly good Good Medium Fairly poor Poor Very poor

k2 0.95–0.97 0.97–0.98 0.98–1.00 1.00–1.10 1.10–1.20 1.20–1.25 1.25–1.30

Table 3. Compensation coefficient for dangerous equipment and facilities.

Reliability of equipment Degree of intrinsic safety

and facilities Very good Fairly good Good Medium Fairly poor Poor Very poor

k3 1.1 1.2 1.3 1.4 1.5 1.6 1.7

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

Table 4. Classification of inherent-risk severity index. Step 4: The fuzzy number is transformed into a fuzzy pos-
h Severity level of inherent risk Level symbol sibility value. Because a fuzzy number may represent multiple
real numbers in different membership functions, it is difficult
h < 70 Negligible Five level
to compare the final fuzzy number results in FFTA; thus, the
70 ≤ h < 140 Mild Four level
140 ≤ h < 190 Moderate Three level comprehensive fuzzy probability was calculated mathematically.
190 ≤ h < 250 Serious Two level We adopted the fuzzy maximum minimum algorithm proposed
h ≥ 250 Catastrophic One level by Huang [24], where FPS represents the level of trust of experts
in the possibility of an event. This method fully uses the infor-
mation provided by the fuzzy number, and the X and Y axes
values corresponding to the maximum membership degrees on
very low, low, fairly low, medium, fairly high, high and very high, the left (pessimistic interval) and right (optimistic interval) of
respectively. the fuzzy set can be calculated with the same scale: the fuzzy
Step 2: Expert assessment was used to determine the nor- maximum FPSmax and the fuzzy minimum FPSmin . Then, the
malized weight. Preyssl [20] interpreted three types of effective fuzzy maximum and minimum are averaged to obtain the cor-
calculations of the probability of the occurrence of basic events: responding fuzzy maximum probability FPS. The fuzzy number
expert judgment, extrapolation and statistical analysis. Yazdi et al. is transformed into FPS based on the left and right fuzzy sorting
[21] suggested that expert judgment is more direct and practical methods, in which the maximum and minimum fuzzy sets are
through comparison. Lavasani et al. [22] suggested that the expert defined as:
judgment method applied to the study of rare events is a scientific x, 0 < x ≤ 1
consensus method. However, numerous researchers also indicate fmax (x) = (10)
0, others
that different educational and cultural backgrounds, work experi-
ence and familiarity of evaluation objects would lead to significant 1 − x, 0 < x ≤ 1
fmin (x) = (11)
differences in judgment results [23]. Therefore, to reduce subjec- 0, others
tive judgment errors of different experts, experts’ backgrounds
were scored according to Table 6 and evaluated using Eq. (9). Then, the left and right fuzzy possibility values of the fuzzy
number are:

n
 
wj = kj / kj j = 1, 2, · · · , n (9) FPSmax = fR (max) = SUP f z (x) ∧ fmax (x) (12)
x
j=1
FPSmin = 1 − fL (max) = 1 − SUP f min (x) ∧ fz (x) (13)
x
where wj represents the normalized weight of the jth expert
authority; kj represents the score of the jth expert; and n is the The average fuzzy probability value is:
number of experts.
Step 3: Base event occurrence probability calculation. Experts FPS = 1 2 (FPSmax + FPSmin ) (14)
in relevant fields are invited to form an assessment team to assess
the risk of an object, assessment of natural language by experts.
Step 5: The fuzzy probability value is transformed into the
Then, the total fuzzy number Z of expert evaluation opinions is:
Fuzzy Failure Probability (FFP), and the bottom event probability
of the FTA is solved. Calculating the FPS of each base event yields
f (z) = Zλ = [x1 , x2 ] (10) only the probability data of the bottom event, not the FFP of

176 International Journal of Low-Carbon Technologies 2024, 19, 171–184

 Subway tunnel collapse system

Table 5. Fuzzy numbers and λ cut set.

Fuzzy language Fuzzy number form λ cut set

Very low fVL = (0, 0, 0.1, 0.2) fVL λ = (0, 0.2 − 0.1λ)
Low fL = (0.1, 0.2, 0.3) fL λ = (0.1 + 0.1λ, 0.3 − 0.1λ)
Fairly low fFL = (0.2, 0.3, 0.4, 0.5) fFL λ = (0.2 + 0.1λ, 0.5 − 0.1λ)
Medium fM = (0.4, 0.5, 0.6) fMλ = (0.4 + 0.1λ, 0.6 − 0.1λ)

Fairly high fFH = (0.5, 0.6, 0.7, 0.8) λ = (0.5 + 0.1λ, 0.8 − 0.1λ)
fFH
High fH = (0.7, 0.8, 0.9) fHλ = (0.7 + 0.1λ, 0.9 − 0.1λ)
Very high fVH = (0.8, 0.9, 1, 1) λ = (0.8 + 0.1λ, 1)
fVH

Table 6. Expert weighted scoring criteria [ 13 , 21].

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

Position Score Job experience (year) Score Education Score Age Score

Senior academic 5 ≥30 5 PhD 5 ≤ 50 4

Junior academic 4 20–29 4 Master 4 40–49 3
Engineer 3 10–19 3 Bachelor 3 30–39 2
Technician 2 6–9 2 Higher national diploma 2 > 30 1
Worker 1 ≤5 1 School level 1

the base event [24]. For the occurrence probability of events at Table 7. Probability grading determination of risk points.
the bottom of the fault tree, part of the occurrence probability is
Level symbol Probability level P(T)
obtained by statistical analysis, whereas the other part is obtained
by combining expert judgment with the fuzzy set theory. To A Very likely 0.01 < P(T) ≤ 1
ensure consistency between the two, FPS must be transformed B Somewhat likely 0.001 < P(T) ≤ 0.01
C Unlikely 0.0001 < P(T) ≤ 0.001
into FFR; Onisawa used Eq. (15) to transform the possibility of D Very unlikely 0.00001 < P(T) ≤ 0.0001
fragile failure into the probability of failure, which has been widely E Practically impossible P(T) ≤ 0.00001
used in [25–27].

1  1
CFP = 0 1 − FPS 3
Step 8: Determine the risk control frequency index. In the
FFP = 10h h = 2.301 ×
0 CFP = 0 FPS tunnel engineering safety production system, the inherent risk
(15) mainly emphasizes the internal danger of the objective existence
of substances and the quality of management only represents the
Step 6: Solve the top event failure probability. According to the
high and low possibilities of risk events. However, the inherent
application characteristics of the gates and gates in the accident
risk does not disappear and the change of management state
tree, the structural function of the accident tree is listed and the
in risk management causes the inherent risk to fluctuate; thus,
simplest standard equation is extracted by Boolean algebra [25].
the probability of complement of the occurrence of uncontrolled
Then, the probability of the occurrence of the top event is obtained
management was used as the risk control to measure the fluc-
using Eq. (16):
tuation of the inherent risk. Furthermore, due to integrating the
⎧m safe production and operation mode of the enterprise, the non-
⎪ 
⎪
⎨ qi the“ and gate" of FTA intersecting and static physical objects and the enterprise have a
P(T) = i=1 (16) cross and dynamic correlation in the operation, which increases
m 
 
⎪
⎪ the original objective and independent physical risks. For exam-
⎩1 − 1 − qi the“ or gate" of FTA
i=1 ple, the risk of slope collapse in the uninhabited zone is small,
whereas the risk of slope collapse in the construction process
where qi = (FP)i represents the occurrence probability of the ith of open-pit mine operators is large. Therefore, the expression
bottom event. of the control frequency index of the event risk point due to
Step 7: Possibility grading judgment of event risk points. Based management failure is defined as follows:
on the FFTA model, the event probability is calculated at the top of
the accident risk point. The higher the probability, the greater the
G0 = 1 + P(T) (17)
intensity of risk control intensity, indicating a greater possibility
of the occurrence of risk events in the tunnel engineering safety
system. The probability grading judgment of the occurrence of where G0 represents the risk control frequency index of the risk
risk point events [18, 28] is shown in Table 7. point.

International Journal of Low-Carbon Technologies 2024, 19, 171–184 177

W. Li et al.

Table 8. Determination of safety risk levels. where a1 represents yellow alarm times, a2 represents orange
alarm times and a3 represents red alarm times.
Safety risk level Risk level Level symbol
If the online monitoring characteristic index warning signal is
R0 ≥ 320 Major risk I level normal, K 4 = 1; otherwise, the initial security risk increases by one
160 ≤ R0 < 320 Higher risk II level level.
80 ≤ R0 < 160 General risk III level
R0 < 80 Low risk IV level
(2) Modification of other dynamic indicators.
Special periods refer to statutory holidays and important
national or local activities. The natural environment refers to
meteorological, earthquake, geological and other disasters that
In conclusion, the risk control frequency index affecting the occur in a region. The major hidden danger index refers to the
fluctuation of inherent risk was established. The risk control fre- hidden danger of an accident that violates relevant laws and
quency classification standard can be determined with reference

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

regulations, departmental rules and national standards and may
to Table 7. lead to heavy personal injury or economic loss. If these three
disaster indicators are encountered, the initial level of safety risk
6 INITIAL SAFETY RISK AGGREGATION must be increased by one level, which is the final realistic risk
level. In contrast, if the enterprise implements the comprehensive
METHOD treatment of tunnel engineering, the initial safety risk level should
be reduced by one level.
According to the affiliation of the system attribute risk structure
and initial risk characteristics, the inherent-risk severity index of
risk events and its risk control frequency index are aggregated into 7.2 Realistic risk dynamic correction model
the initial safety risk, which is defined as: (1) Dynamic index correction method.
The correction coefficient of the alarm signal of the dynamic
R 0 = h × G0 (18) monitoring characteristic index dynamically modifies the inherent-
risk index of risk points:
where R0 represents the initial safety risk of the risk point.
Hd = HK4 (20)
7 INITIAL SAFETY RISK LEVEL
where Hd represents the revised value of the dynamic monitoring
From high to low, the safety risk is divided into four levels: major, index of the inherent-risk point.
higher, general and low risk. Among them, the main risk is the (2) Realistic risks.
highest (level I). The initial results of the risk classification were If dynamic correction is required, the high-risk control fre-
derived through data calculation, as shown in Table 8. quency and the inherent-risk index should be aggregated again.
Its model is expressed as:
7.1 Real risk dynamic correction index
(1) Correction coefficient of high-risk monitoring features. R = G0 Hd . (21)
The actual risk dynamic correction index modifies the initial
high risk (R0 ) or inherent-risk index (h) in real time. Includ- The actual risk before dynamic correction is equal to the initial
ing mainly the high-risk monitoring characteristic index (K4), safety risk, R = R0 . Actual risk results from the initial high security
the natural environment index, major hidden danger index and risk modified by the dynamic correction index of actual risk.
closed tunnel engineering. Dynamic correction is a timely correction for real risk, not at all
The inherent-risk index of the risk points is corrected for by the stages. Therefore, the actual risk classification standards can still
characteristic index alarm signal coefficient of high-risk dynamic be included in Table 8.
monitoring. The real-time alarm of the online monitoring project
is divided into levels I, II and III, corresponding to low, medium
and high alarms, respectively. When the online monitoring items
reach three level I alarm, it is recorded as one level II alarm; when
8 APPLICATION OF A SECURITY RISK
the monitoring items reach two level II alarms, it is recorded as ASSESSMENT OF A SUBWAY TUNNEL
one level III alarm. Therefore, the weights of the first, second and ENGINEERING SYSTEM
third levels of alarm are set as 1, 3 and 6, respectively, and the
normalized coefficients (correction coefficients of alarm signal) The main structure of a subway crossing a mountain tunnel was
are 0.1, 0.3 and 0.6, respectively. The equation is as follows: constructed. The rock mass itself was affected by different geo-
logical conditions, such as faults, bedding planes, fracture zones
K4 = 1 + 0.1a1 + 0.3a2 + 0.6a3 (19) and joint fissures, to form a rock mass structure with different

178 International Journal of Low-Carbon Technologies 2024, 19, 171–184

 Subway tunnel collapse system

Table 9. Determination of material coefficient of pressurized roof during tunnel excavation.

Each inherent material Evaluation criteria for each indicator

Weight coefficient
index I level II level III level IV level V level

Basic quality index of rock I type II type III type IV type V type 0.5
mass (A1 )
Deformation modulus >33 33–20 20–6 6–1.3 <1.3 0.5
index (A2 )
Risk value of each indicator 1–2 2–3 4–6 6–8 8–10 /

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

characteristics, which determines the deformation, strength and mass and disturbance influence factors. The general and special
overall stability of the rock mass. During the construction of the process risk factors F 1 and F 2 were evaluated from the inherent
tunnel, online monitoring facilities monitored events, such as the process attribute characteristics and disturbance factors. The risk
external and internal displacement of the rock mass. The daily value and the influence coefficient of each influence factor are
monitoring facilities worked normally. Twenty-two operators per shown in Table 10.
shift were involved in the tunnel construction and 28 residents Combined with the actual situation in this case, the risk index
were located within 1 km downstream of the tunnel. Emergency of the process at the risk point of the collapse event was obtained
plans and safety evaluation reports had been produced. The major from Table 10, and the risk coefficient of the general process was
hidden dangers were found through on-site investigation. Because obtained from Eqs. (2) to (4). F1 was determined to be 2.85, the
the localization deformation process of rock mass is complex, risk coefficient of the special process F 2 was 2.60 and process risk
quantifying the deformation evolution relationship within rock index F 3 was 7.41.
mass with specific mathematical relations is difficult. Therefore, (3) Determination of the risk compensation coefficient of the
to quantitatively assess the risk of typical collapse events, the risk point.
established safety risk assessment model was used to assess safety Considering that the maximum operator in a single shift is
risk of the subway tunnels. exposed to a dangerous environment, Table 1 shows that the risk
compensation coefficient k1 at the risk point of collapse acci-
dent was 1.50. According to Table 2, the maximum dangerous
operation compensation coefficient k2 was 1.20. Here, due to the
8.1 Inherent-risk severity of typical events disaster-causing factors of excavation equipment and facilities, the
(1) Determination of material coefficient with coping plate.
compensation coefficient of dangerous equipment k3 was 1.20,
According to the material class solution method of the pres-
selected from Table 3.
surized roof, the determination rules of the material coefficient
(4) Inherent risk severity.
of the tunnel were established from the dangerous characteristics
Using the improved DOW inherent risk assessment method,
that determine the possible energy burst of the material, as shown
the inherent risk severity index of the collapse event was com-
in Table 9.
prehensively solved using Eq. (6), where h was determined to
According to the classification standard for qualitative char-
be 224.
acteristics of the basic quality of the rock mass, the rock mass
According to Table 4, the consequences of inherent risks at risk
belonged to level III and the corresponding value was 4. The
points of collapse events were ‘serious,’ and the severity level was
static nature of the rock mass can be reflected in the deformation
‘level II’.
and elastic moduli. In addition, the deformation modulus reflects
the correlation between the elastic modulus, including elastic and
the plastic deformation, and elastic modulus, which only reflects
elastic deformation. Thus, to objectively characterize the initial 8.2 Typical event risk control frequency index
safety of the pressurized roof in the mining area, the risk value was From the root cause of the event risk, another main cause of
determined by the deformation modulus, which was calculated tunnel collapse event is improper management. The standardized
to be 4. Therefore, the comprehensive value was combined with site management system usually adopts the ‘3E’ safety manage-
the actual case in the evaluation, and the inherent material risk ment system method of enforcement, education and engineering.
coefficient of the pressurized roof was obtained from Eq. (1) as Potential risks can turn into accidents when poorly managed.
MF = 14. According to the characteristics of this case, a tunnel collapse fault
(2) Determination of process risk coefficient of collapse risk tree diagram caused by improper management was constructed to
point. solve the probability of a tunnel collapse event, as shown in Fig. 3.
Based on the rock mass basic quality classification standard, The probability of the top events can be obtained by estimating
literature and case data analysis, through the inherent risk identifi- the probability of the basic events using the FFTA method. To
cation mode, combined with the characteristics of the tunnel rock evaluate the events lacking information, fuzzy data were obtained.

International Journal of Low-Carbon Technologies 2024, 19, 171–184 179

W. Li et al.

Table 10. Risk values of the dangerous process parameters of the pressurized roof in mining space.
General process risk
Numerical
order Classification of parameter Numerical range Weight
coefficient

1 Joint Joint spacing (cm) <6 6–20 20–60 60–200 >200 0.25
conditions
Length of the joint (m) <1 1–3 3–10 10–20 >20
Joint roughness Friction Smooth Slightly Rough Very rough
mirror rough
B1 0–1 1–2 2–3 3–4 4–5

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

2 Physical and Level I II III IV V 0.30
mechanical
parameters of
rock mass
Density
  >2.65 >2.65 2.65–2.45 2.45–2.25 <2.25
ρ/g•cm3 ρ/
g•cm3
Poisson’s ratio u 0.2 0.2–0.25 0.25–0.3 0.3–0.35 <0.35
Shear strength Bond force on >60 60–50 50–39 39–27 <27
the structural
surface (◦ )
Angle of friction >2.1 2.1–1.5 1.5–0.7 0.7–0.2 <0.2
on the structural
plane (MPa)
B2 0–1 1–2 2–3 3–4 4–5

3 Stability of Span (s), Since s≤20 m, s = 10–20 m, basically stable s = 5–10 m, s > 5 m, Lack of 0.30
surrounding the steady ability long-term stable for usually not self-stability
rock in roof stability months self-stabilizing;
s≤5 m, stable
for several days
until January
B3 0–1 1–2 2–3 3–4 4–5

4 Orientation of the structural plane and its θ ≤ 60◦ ; θ ≤ 30◦ , Other range / / 0.15
combination with the axis of hole (θ ) θ >75◦ θ ∈ 30◦ ∼
75◦
B4 0–2 4–5 2–4 / /

Special process risk

1 Groundwater Long water inflow per 10 m holes >125 25–125 10–125 <10 / 0.18
state (L/min•10m)
Joint hydraulic (MPa) >0.5 0.2–0.5 0.1–0.2 <0.1 0
General situation Completely A little wet Wet Drop of water Water
dry flowing out
or overflows
C1 0–1 1–2 2–3 3–4 4–5

2 Initial stress BQ >550 550–450 450–350 350–250 <250 0.21

state
Extremely high stress region 1.0 1.0 1.0–1.5 1.0–1.5 1.0
The high stress area 0.5 0.5 0.5 0.5–1.0 0.5–1.0
C2 0–1 1–2 2–3 3–4 4–5

3 Velocity of particle vibration caused by blasting (cm/s) 3–10 10–20 20–30 30–40 >40 0.21
C3 0–1 1–2 2–3 3–4 4–5

4 Stability deformation rate of surrounding rock Slow down Constant Speed up / / 0.40
speed
C4 0–2 2–3 3–5 / /

180 International Journal of Low-Carbon Technologies 2024, 19, 171–184

 Subway tunnel collapse system

Table 11. Results of the bottom event questionnaire affecting top events.
Bottom event Expert fuzzy judgment language

E1E1 E2E2 E3 E4E4 E5E5

X1 Inadequate hazard identification M FL FL M FH

X2 Inadequate safety inspection H FH FH H H
X3 The effect of hidden danger treatment is poor M M M M M
X4 There are residents downstream VL L L M L
X5 Adjacent to the slope blasting area VL L L M M
X6 No training H VH H M VH
X7 Personnel safety awareness is weak M H H M M
X8 Employee’s illegal operation M H H M M

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

X9 Rock mass quality H H H FH VH
X10 Flood discharge facilities are blocked VL FL L VL M
X11 No flood discharge facilities H FH VH H H

Table 12. Normalized weight of experts. Based on left–right ordering and combining Eqs. (10) to (13),
the left–right fuzzy possibility value of the bottom event X1 could
Evaluation standard Proportion of assessment experts
be obtained; FPSmax was 0.5705 and FPSmin was 1.4172. Accord-
E1 E2 E3 E4 E5 ing to Eq. (14), FPSX1 was 0.5767. Eq. (15) was used to deblur the
Score 18 13 16 15 8 bottom event X1 , and FFP was 0.0084.
w 0.257 0.186 0.229 0.214 0.114 According to the above steps and expert judgment, the proba-
bility of other base events in the fuzzy fault tree was calculated as
shown in Table 13.
The failure probability of the top event was calculated according
According to Table 2, the bottom event possibility assessment to Table 13. To facilitate calculation and analysis, there were no
level was divided into seven levels. Five experts were invited to repeated events in the fault tree, and the number of ‘or gates’ far
make fuzzy evaluations for each event of the fault tree. The results exceeded that of ‘and gates’ (Fig. 2). The bottom events probability
of the expert judgment are shown in Table 11. in Table 13 were input in Eq. (16) to obtain the top event probabil-
The expert normalized weight was solved. To avoid the error of ity P(T), which was 0.0209. According to Table 2, the possibility
evaluation data caused by the expert weight, the basic information level was ‘A,’ indicating it was ‘very likely.’ Using Eq. (17), we
of experts was evaluated according to the scoring criteria in determined that the collapse risk control frequency index G0 of
Table 6. Expert weight results obtained from Eq. (9) are shown the tunnel engineering was 1.209. Because online monitoring
in Table 12. facilities were mentioned in the case and the online monitor-
The probability of failure of the bottom event was solved. ing equipment was normal; here, K 4 was 1, dynamic verifica-
The base event ‘X1 ’ was chosen as an example to calculate the tion was required, and Hd was 270.82, determined using Eq.
generated fuzzy number and convert natural language into fuzzy (20). Meanwhile, according to Eq. (21), the initial safety risk
numbers. The total fuzzy number Z evaluated by five experts is of collapse after dynamic correction was 270.82. Therefore, we
expressed as. judged that the initial safety risk level was ‘level II’ and belonged
 to ‘higher risk.’ As the case mentioned the presence of major
fM⊕FL⊕FL⊕M⊕FH (z) = w1 · fM (x) ⊕ w2 · fFL (x)
hidden dangers and risks, the initial risk should be raised by
 one level, and the final actual risk level was ‘level I,’ which is a
⊕ w3 · fFL (x) ⊕ w4 · fM (x) ⊕ w5 · fFH (x)
‘major risk.’
= [0.3284 + 0.1λ, 0.5813 − 0.1λ]
Zλ = [x1 , x2 ] = [0.3284 + 0.1λ, 0.5813 − 0.1λ]
x1 − 0.3284 0.5813 − x2
9 DISCUSSION
λ= ,λ =
0.1 0.1 The widely recognized risk matrix method was adopted to verify
the system security risk assessment method. According to the risk
Therefore, the relation function of the average fuzzy function matrix table, the risk level was divided into the following four
Z is:
⎧ x−0.3284 levels: red, level i, major risk; orange, class ii, high risk; yellow,
⎪
⎪ , 0.3284 < x ≤ 0.4284 grade iii, general risk; blue, grade iv, low risk. A risk matrix is
⎨ 0.1
1 , 0.4284 < x ≤ 0.4813 shown in Table 14.
fz (x) = 0.5813−x
⎪
⎪ , 0.4813 < x ≤ 0.5813 The consequence of the inherent-risk index obtained by the
⎩ 0.1
0 , others improved DOW method was ‘serious,’ and the probability of risk

International Journal of Low-Carbon Technologies 2024, 19, 171–184 181

W. Li et al.

Table 13. Probability of occurrence of bottom events.

Bottom events Fuzzy evaluation set Zλ FPS FFP

X1 [0.3284 + 0.1λ, 0.5813 − 0.1λ] 0.5767 8.402 × 10−3

X2 [0.5942 + 0.1λ, 0.8471 − 0.1λ] 0.4305 2.979 × 10−3
X3 [0.4342 + 0.1λ, 0.6342 − 0.1λ] 0.4545 3.588 × 10−3
X4 [0.1385 + 0.1λ, 0.3385 − 0.1λ] 0.4545 3.588 × 10−3
X5 [0.1727 + 0.1λ, 0.3727 − 0.1λ] 0.4545 3.588 × 10−3
X6 [0.6658 + 0.1λ, 0.8658 − 0.1λ] 0.4545 3.588 × 10−3
X7 [0.5245 + 0.1λ, 0.7245 − 0.1λ] 0.4545 3.588 × 10−3
X8 [0.5245 + 0.1λ, 0.7245 − 0.1λ] 0.4545 3.588 × 10−3
X9 [0.6686 + 0.1λ, 0.8258 − 0.1λ] 0.4740 4.147 × 10−3
X10 [0.1057 + 0.1λ, 0.3243 − 0.1λ] 0.4461 3.364 × 10−3

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

X11 [0.6857 + 0.1λ, 0.9043 − 0.1λ] 0.4461 3.364 × 10−3

Table 14. Risk matrix judgment table.

Consequence

Catastrophic Serious Moderate Mild Negligible

Possibility

Very likely

Somewhat likely

Unlikely

Very unlikely

Practically

impossible

point solved by the FFTA method was ‘very likely.’ According The risk matrix method, a tool for risk visualization, is a qual-
to the risk matrix in Table 8, the actual risk level was level i, itative risk assessment and analysis method that can compre-
which belongs to ‘major risk.’ Comparing the system security risk hensively evaluate the possibility of danger and the severity of
assessment method with the risk matrix method yielded the same injury. Its advantages include fast operation, ease of learning and
risk assessment results, further verifying the rationality of the ease of mastering. In addition, it can analyze the quantitative
model. Comparing their analysis results showed that they have characteristics, relations and changes of enterprise security status,
their own advantages and disadvantages, and their application and its risk value is convenient for researchers or risk decision
scope is as follows. makers to make effective judgments and interpretations through

182 International Journal of Low-Carbon Technologies 2024, 19, 171–184

 Subway tunnel collapse system

the comparison and analysis of these data. Its disadvantage is environment; therefore, these findings provide a theoretical basis
that the risk importance level determined using the risk matrix is for subsequent research.
determined by mutual comparison; thus, it is impossible to obtain
the risk importance level for the whole risk through mathematical
calculation. However, the results of the system security risk assess- AUTHOR CONTRIBUTIONS
ment method are usually expressed by obtaining index data from
statistical data or given certain data rules, and the risk value can Wen Li is the executor of the modeling design and theoretical
be obtained by processing and sorting mathematically. Accurate analysis of this study, and is responsible for manuscript writing.
identification and quantification of risks is still difficult, which Xuesong Lu completed data analysis and guided the writing and
largely depends on the experience of the evaluator. revision of the paper. Menglong Wu participates in the modeling
The analysis of system risk assessment results provided three process. All authors read and approved the final manuscript.

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

major findings:
(1) The evaluation model highlights key aspects of prevention
and control. Inherent risks from ‘high-risk materials, processes, ACKNOWLEDGEMENTS
environment, operation, equipment and facilities’ can highlight
This research was supported by the Science and Technology
the dangers of key groups, equipment, processes and places, and Research Program of the Education Department of Hubei
fully show the actual current risk. Province, China (grant Q20212906), Huanggang Ecological
(2) The material hazard index (M) and the site personnel expo- Architecture and Renewable Resources Research Center, China
sure index (E) account for the highest proportion in the assess- (grant 202316004), High-level Cultivation Program of Huang-
ment, the basic quality (A1 ) and deformation modulus indices gang Normal University, China (grant 204202112304) and Talent
(A2 ) of the rock mass M; with the heightening of A1 or A2 , M will Initiation Program of Huanggang Normal University, China
increase. E increases with a greater number of people near the tun- (grant 2042021014). I especially thank my tutor, Professor
nel construction site, resulting in a greater impact on the results Yicheng, and his team for their guidance and help in the writing
of the risk assessment. Therefore, comprehensive treatment of process.
tunnel engineering, reduction of personnel in exposed areas and
reduction of stockpile stock can significantly reduce inherent risk.
(3) The system security risk can present the dynamic change REFERENCES
process, considering the changes of online monitoring and other [1] He BJ, Wang W, Sharifi A et al. Progress, knowledge gap and future
potential risks. This method can better display the risk state of directions of urban heat mitigation and adaptation research through a bib-
each part of the system. liometric review of history and evolution. Energy Build 2023;287:112976.
[2] Liu Y, Lai J, Wang W et al. Low carbon optimization of deep foundation
pit support in undulating strata. Appl Sci 2023;13:6562.
[3] Rokooei S, Shojaei A, Alvanchi A et al. Virtual reality application for
10 CONCLUSIONS construction safety training. Saf Sci 2023;157:105925.
[4] Huang L, Wu C, Wang B et al. An unsafe behaviour formation mechanism
The traditional management model for distinguishing between based on risk perception. Hum Factors Ergon Manuf Ser Ind 2019;29:
key and non-key industries, as well as for identifying and address- 109–17.
ing hidden dangers, cannot satisfy the current practical require- [5] Wu C, Huang L. A new accident causation model based on informa-
tion flow and its application in Tianjin Port fire and explosion accident.
ments of safety production risk control. This study proposes Reliability Engineering & System Safety 2019;182:73–85.
improved inherent risk identification and evaluation techniques [6] Wang B, Wu C, Huang L et al. Prevention and control of major accidents
to systematically identify specific characterization methods for (MAs) and particularly serious accidents (PSAs) in the industrial domain
the inherent attributes of high-risk materials, processes, environ- in China: current status, recent efforts and future prospects. Process Saf
ment, operations, equipment and facilities in green construction Environ Prot 2018a;117:254–66.
[7] Wang B, Wu C, Kang L et al. Work safety in China’s thirteenth five-year
of tunnel engineering. Based on the DOW-FFTA model, the plan period (2016–2020): current status, new challenges and future tasks.
initial safety risks of green construction were determined, based Saf Sci 2018b;104:164–78.
on which a dynamic correction model was proposed. A green [8] Xu K, Chen XF. “High Five” risk management system based on major acci-
perspective subway construction safety risk assessment model dents prevention. Journal of Wuhan University of Technology (Information
was established, which accounted for the shortcomings of the and Management Engineering Edition) 2017;39:649–53. (In Chinese).
[9] Qin G, Huang Y, Wang Y et al. Pipeline condition assessment and finite
traditional subway construction evaluation system. Finally, the element modeling of mechano-electrochemical interaction between cor-
evaluation process of green construction in subway projects was rosion defects with varied orientations on pipelines. Tunn Undergr Space
validated through practical cases, to provide better quantitative Technol 2023;136:105101.
methods and a more comprehensive risk evaluation index sys- [10] Cha SK, Yoon JY, Lee SK et al. Development of production resources
tem for green construction evaluation in subway projects. The (4M1E) integration in real time and middleware for autonomous configu-
ration. J Korean Soc Precis Eng 2014;31:303–9.
research results encourage the improvement of the current green [11] Wu HP, Fu LR, Wang JC. Discussion on the application of site management
construction evaluation system of subway projects, as well as “4M1E” elements in the university laboratory management. Research and
promote the green degree of subway construction and its related Exploration in Laboratory 2013;32:230–3.

International Journal of Low-Carbon Technologies 2024, 19, 171–184 183

W. Li et al.

[12] Mao Y, Xu T. Research of 4M1E’s effect on engineering quality based [20] Preyssl C. Safety risk assessment and management the ESA approach.
on structural equation model. Systems Engineering Procedia 2011;1: Reliab Eng Syst Saf 1995;49:303–9.
213–20. [21] Yazdi M, Nikfar F, Nasrabadi M. Failure probability analysis by employ-
[13] Li W, Ye YC, Wang QH et al. Fuzzy risk prediction of roof fall and rib ing fuzzy fault tree analysis. International Journal of System Assurance
spalling: based on FFTA–DFCE and risk matrix methods. Environ Sci Engineering and Management 2017;8:1177–S1193.
Pollut Res 2019;27:8535–47. [22] Lavasani SM, Ramzali N, Sabzalipour F et al. Utilization of fuzzy fault tree
[14] Li W, Ye YC, Hu NY et al. Real-time warning and risk assessment of tailings analysis (FFTA) for quantified risk analysis of leakage in abandoned oil and
dam disaster status based on dynamic hierarchy-grey relation analysis. natural gas wells. Ocean Eng 2015;108:729–37.
Complexity 2020;1–14. [23] Li WJ, Liang W, Zhang LB et al. Performance assessment system of health,
[15] Li W, Lu XS, Dong X. Fire system safety risk cognition model and safety and environment based on experts’ weights and fuzzy comprehen-
evaluation of major public safety risks. Front Public Health 2022;10: sive evaluation. J Loss Prev Process Ind 2015;35:95–103.
1–14. [24] Huang C. 2007. Risk analysis with information described in natural lan-
[16] Wang Y, Xia A, Qin G. Probabilistic modeling for reliability analysis guage. In Computational Science-ICCS 2007, International Conference,

Downloaded from https://academic.oup.com/ijlct/article/doi/10.1093/ijlct/ctad101/7603306 by guest on 02 June 2024

of buried pipelines subjected to spatiotemporal earthquakes. Probabilis- Beijing, China, May 27–30, 2007, Proceedings, Vol. 5. DBLP. 1016–23.
tic Engineering Mechanics 2022;69:103315. https://doi.org/10.1016/j.probe [25] Lavasani SM, Wang J, Finlay J. Application of fuzzy fault tree analysis on
ngmech.2022.103315. oil and gas offshore pipelines. Int J Marine Sci Eng 2011;1:29–42.
[17] Zheng XX, Xu JL, Wang HX et al. Health risk assessment of a reservoir-type [26] Mottahedi A, Ataei M. Fuzzy fault tree analysis for coal burst occurrence
water source in the northeastern region of China. Human and Ecological probability in underground coal mining. Tunn Undergr Space Technol
Risk Assessment: An International Journal 2017;23:751–66. 2019;83:165–74.
[18] Li W. A dissertation submitted in partial fulfillment of the requirements [27] Wang D, Zhang P, Chen L. Fuzzy fault tree analysis for fire and explosion
for the degree of doctor of philosophy in engineering. Wuhan University of of crude oil tanks. J Loss Prev Process Ind 2013;26:1390–8.
Science and Technology, 2020. (In Chinese). [28] Morisset J, Johannson KA, Jones KD et al. Identification of diag-
[19] Deng QG, Gao JL, Liu MJ. Safety System Engineering (Chinese English nostic criteria for chronic hypersensitivity pneumonitis: an interna-
bilingual). Xuzhou: China University of Mining and Technology Press, tional modified Delphi survey. Am J Respir Crit Care Med 2018;197:
2011. 1036–44.

184 International Journal of Low-Carbon Technologies 2024, 19, 171–184