Developing A Process To Evaluate Construction Project Safety Hazard Index Using The Possibility Approach in India

Developing a Process to Evaluate Construction
Project Safety Hazard Index Using the

Possibility Approach in India
D. A. Patel 1 and K. N. Jha 2
Abstract: Hazard evaluation is a critical task in safety management of construction projects because of their characteristics and dynamics.
Downloaded from ascelibrary.org by Budi Atmoko on 05/09/17. Copyright ASCE. For personal use only; all rights reserved.
This study provides a framework to evaluate the project hazard index (PHI), which represents the hazard level of a construction project by
using the possibility approach (fuzzy set theory). To achieve this objective, the study has been conducted in two phases. In the first phase, 10
hazardous trades and their physical attributes have been shortlisted and a three-level hierarchical structure has been constructed to evaluate the
PHI. In the second phase, the methodology of evaluation and determination of the weight of each hazardous trade with its attributes were
carried out using tools such as fuzzy measures, fuzzy integrals, fuzzy analytical hierarchy process (FAHP), and the weighted sum method
(WSM). Correlation between the PHI and safety budget of several project sites owned by the Delhi Metro rail project validates this study. The
PHI will be useful to differentiate the projects based on their project hazard level and accordingly, decisions can be made to improve safety
management in construction projects. Application of fuzzy measure and nonadditive fuzzy integral, nonprobability (possibility) approach,
and considerations of trades instead of activities are some of the key highlights and contributions of this study. DOI: 10.1061/(ASCE)CO
.1943-7862.0001205. © 2016 American Society of Civil Engineers.
Author keywords: Project hazard; Construction safety; Delphi method; Fuzzy logic; Fuzzy integrals; Fuzzy analytical hierarchy process
(FAHP); Weighted sum method (WSM); India; Labor and personnel issues.
Introduction measurement of the overall project hazard level is not an easy task.
The evaluation of a project hazard level can be considered a multi-
A hazard is a source of potential harm or a situation that can cause criteria decision-making problem in which multiple experts and
loss [Bureau of Indian Standards (BIS) 2007]. The loss may be respondents decide hazardous trades and evaluate the attributes
unintentional injuries or death to people, or damage to or loss of of each.
an item or belongings. Therefore, the estimation of the hazard level The present study uses the fuzzy set theory (possibility ap-
at construction sites should be considered for all on-site activities proach), fuzzy integral, and multicriteria decision making (MCDM)
that are deemed hazardous trades of the project. However, two main methods like the fuzzy analytical hierarchy process (FAHP) and
difficulties arise in doing so: (1) selection and measurement of haz- weighted sum method (WSM) to develop a project hazard level
ardous trades and their attributes, and (2) determination of an evalu- evaluation methodology. The fuzzy set theory conveniently and
ation method of the project hazard level of construction projects. effectively deals with subjective judgment. The fuzzy integral eval-
Because of their increasing complexity and dynamism, con- uates the attributes of hazardous trades and FAHP and WSM
struction projects raise significant uncertainties and subjectivities in structure hazardous trades associated with a project. The proposed
the evaluation of project hazard level. A typical construction project methodology incorporates knowledge and experience acquired
generally includes several activities and trades like excavation, from selected construction experts. The determination of hazardous
demolition, roof work, electrical work, road work, underground trades and attributes, framing of their structure, and sound subjec-
utilities trenching work, and use of different types of machinery and tive judgments are considered to assess the project hazard index
equipment, etc. Most of the time, construction professionals are (PHI), which reflects hazard level of a project.
able to express project hazard level in qualitative terms only such
as: high, very high, low, etc. They are also unaware of the evalu-
ation process for the project hazard level especially in the presence Objectives of the Study
of a large number of hazardous trades. In addition, due to the
In brief, the prime objectives of this study are to (1) determine the
number of physical attributes that define a hazardous trade, the
important hazardous trades existing in typical construction projects
1 and their measurement attributes; (2) prepare the framework to
Assistant Professor, Dept. of Civil Engineering, S.V. National Institute
of Technology, Surat 395007, Gujarat, India. E-mail: dapscholar@gmail
evaluate the project hazard level; (3) develop the evaluation process
.com to obtain the PHI using the proposed framework; (4) formulate and
2
Associate Professor, Dept. of Civil Engineering, Indian Institute of present the PHI to evaluate the project hazard level of a construc-
Technology Delhi, Hauz Khas, New Delhi 110016, India (corresponding tion project; and (5) validate the practical applicability of the PHI.
author). E-mail: knjha@civil.iitd.ac.in
Note. This manuscript was submitted on February 3, 2016; approved on
May 25, 2016; published online on July 18, 2016. Discussion period open Literature Review
until December 18, 2016; separate discussions must be submitted for in-
dividual papers. This paper is part of the Journal of Construction Engi- Each construction project has a unique nature and, therefore, the
neering and Management, © ASCE, ISSN 0733-9364. project hazard level depends on its characteristics and differs from
© ASCE 04016081-1 J. Constr. Eng. Manage.
J. Constr. Eng. Manage., 2017, 143(1): -1--1

site to site even in the same project. New hazards emerge on an function (Zadeh 1978; Sachs and Tiong 2009; Dubois and Prade
almost daily basis because the jobsite is continually changing as 2015). Zadeh (1995) argued that probability theory is not sufficient
construction proceeds (Laukkanen 1999). Therefore, identification with uncertainty and imprecision. However, probability theory and
of hazards and their assessment are major issues in a construction fuzzy logic are not competitive; rather, they complement each other.
project. Probability lacks the capability to handle perception-based informa-
In assessment of hazards, there are two main types of barriers tion, which is a product of human cognition. Possibility theory like
(Carter and Smith 2006): (1) knowledge and information barriers; fuzzy set theory provides solutions to problems that lack the math-
and (2) process and procedures barriers. Knowledge and informa- ematical rigor inherent to probability theory (Baloi and Price 2003).
tion barriers include lack of information sharing across projects, Therefore, possibility theory was successfully implemented in en-
lack of resources in smaller projects, subjective nature of hazard gineering and management areas including construction safety
identification and risk assessment, and reliance upon tacit knowl- management (Mohamed and McCowan 2001). Moreover, the prac-
edge; whereas process and procedure barriers often consist of a lack tice of compiling and publishing data on accidents and their relative
of standardized approaches and undefined structure for tasks and frequencies is rare, particularly in developing countries like India
hazards. (Patel and Jha 2015). Therefore, the possibility approach (fuzzy set
Several studies attempted to assess the risk of hazards existing theory) seems more appropriate for this study.
at the project site. Lott (2005) considered the workplace as one of The fuzzy set theory has been widely used in the construction
the hazards in the assessment of the workers’ risk-management pro- safety management field. Tam et al. (2002) developed the Non-
gram to predict incurred costs. According to his study, based on the structural Fuzzy Decision Support System (NSFDSS) to evaluate
project’s hazard level assessment, some strategic decisions may be construction safety management system using fuzzy set theory and
taken to improve the safety performance of the project. For exam- an analytical hierarchy process (AHP). Dağdeviren and Yüksel
ple, assessment of a project’s hazard level may be helpful to top (2008) presented a fuzzy AHP approach to evaluate the faulty
management in allocating resources. Liu and Tsai (2012) deter- behavior risk in a work system. Pinto (2014) developed a qualita-
mined and presented important types and causes of hazards in tive occupational safety risk-assessment model (QRAM) to assess
the construction industry and calculated the risk assessment value the aspects of safety climate and safety barrier effectiveness in con-
of each hazard’s cause, risk, and method of prevention and im- struction. Liu and Tsai (2012) presented a fuzzy risk-assessment
provement. In the United States, Dharmapalan et al. (2015) quan- method against occupational hazards in the construction industry.
tified the construction safety risk of each of the design elements In this study, an analytic network process (ANP) was used to iden-
present in typical multistory buildings and developed an online de- tify important hazard types and their causes. Failure mode and ef-
sign risk-assessment tool. However, the frequency ratings obtained fect analysis (FMEA) was used to assess the risk value of hazard
for this study are risk perceptions of construction personnel and are causes through the fuzzy interference approach. However, the study
not based on empirical data. was focused only on telecom projects and thus it is difficult to use
Gürcanli and Müngen (2009) presented a method to assess this method in other types of construction projects. In addition, Liu
the risks faced by workers at construction sites using a fuzzy- and Tsai (2012) believed that fuzzy numbers other than triangular
rule-based safety analysis using historical accident data, subjective fuzzy numbers (TFN) could be used for a better result. According
judgments of experts, and current safety level of a construction site. to Lee (2006), selection of a membership function depends on com-
They used three parameters, namely, accident likelihood, current putation efficiency, its simplicity, amount of data, and variance in
safety level, and accident severity. Similarly, Fung et al. (2010) de- the data set. The simplest types of membership functions are tri-
veloped the probability-based quantitative risk assessment model angular and trapezoidal because of their linear boundaries, simple
(RAM), which uses different levels of work trades. As a principle formulas, and computational efficiency. The trapezoidal function
of RAM, an historical accident record to calculate the probability is preferred to the triangular function when there is more variance
of occurrence of accidents is needed. The main issue with proba- in the data sets (Lee 2006).
bility analysis is the use of probability measure to evaluate uncer- Several researchers have developed hazard indices in their
tainty. Therefore, studies are needed to define and develop each specific studies. For example, Benardos and Kaliampakos (2004)
factor’s probability distribution using historical data in estimating developed a methodology to assess geotechnical hazards for tun-
relative frequencies. Because of the nonrepetitive nature of con- neling boring machines (TBM). They proposed a vulnerability in-
struction projects, the knowledge of relative frequencies cannot just dex to assess the hazards based only on the principles of a rock
be assumed and applied from one project to another. According to engineering system. Imriyas et al. (2008) proposed a new premium
Baradan and Usmen (2006), few construction firms have the quan- rating framework in the context of the Singaporean construction
tity and quality of data needed to perform meaningful risk analysis. industry based on 17 variables pertaining to the worker compensa-
In developing countries, practitioners are generally used to simply tion insurance (WCI) premium rating from an extensive literature
assuming the relative frequencies in every project without consid- review. They identified the most important and effective variables
ering the hazard level of the project. The hazard level of the project to decide the premium using the Pareto 80=20 statistical analysis.
depends on many factors like location of the project, scope and Imriyas et al. (2008) found that the project hazard level is the most
characteristics of the projects, use of equipment, etc. Thus, uncer- important variable to determine the premium of insurance for the
tainty associated with these projects is generally caused by the in- construction projects. To assess the project hazard level, they pre-
herent fuzziness of the parameter estimate rather than randomness. sented simple PHI using 11 hazardous trades and their respective
In multivariate data analysis, uncertainty is considered as prob- physical attributes in building projects, which were determined
ability phenomena, whereas possible data analysis is considered as based on literature review only.
possibility phenomena (Tanaka et al. 2000). Probability measures In brief, many hazardous activities and tasks exist in a construc-
the likelihood of occurrence, whereas possibility measures the tion project. To assess the project hazard level, it is quite difficult
degree of certainty. Uncertainty existing in real risk situations is to enlist and combine all activities and their measurement criteria.
related to the knowledge of things rather than depending on chance Construction projects engage a variety of trades for carrying out
(Mohamed and McCowan 2001). The possibility theory uses a pair a variety of tasks on project sites. There are varying degrees of haz-
of dual-set functions while the probability theory uses a one-set ards and risks associated with different types of work performed

by these trades (Baradan and Usmen 2006). Therefore, this study select the appropriate MADM method to develop a framework to
prefers to use trades instead of activities to differentiate the projects evaluate PHI of the construction project.
based on their hazard level. For example, the hazard in roofing trade
can be inferred by its attributes like the roof’s height, slope, and
materials used. The impact of hazardous trades and the level of their Research Methodology
physical attributes are different from one project to another. There-
fore, a list of general hazardous trades with their attributes is re- For the sake of convenience, the whole study was conducted in two
quired. However, because of the confounding nature of hazardous phases as shown in Fig. 2.
trades, it is almost impossible to reach a consensus about their
attributes among experts and practitioners. In addition, the measure- Steps in Phase 1 of Research
ment of some physical attributes is quite difficult due to their fuzzy
nature. Therefore, a framework for evaluation is required to develop In this phase, the first step enlists the typical hazardous trades and
the index, which can represent the hazard level of projects. their measurement attributes that represent the hazard level of dif-
ferent projects. For this, a questionnaire survey was designed and
analysis performed using the Delphi method.
Framework for Evaluation of Project Hazard Level The Delphi process generally consists of three rounds of
survey questionnaires. In the first round, experts respond to a broad
This study formulates a PHI that represents the project hazard level question, while each additional round builds upon the responses
of a construction project. The evaluation of a PHI is considered to collected from earlier rounds. The process is terminated when con-
be a multiattribute decision method (MADM) problem. Therefore, sensus is reached (Delbecq et al. 1975). Hallowell and Gambatese
a hierarchical framework to develop a PHI was constructed as (2010) presented a standard methodology for applying the Delphi
shown in Fig. 1. In this study, it is assumed that the hazard level method in rigorous studies and suggested the basic steps of the
of any construction project depends on the existing hazardous Delphi procedure. However, the Delphi method is sometimes criti-
trades. Of these, there are some critical and important hazardous cized because of the shortcuts and modifications to the prescribed
trades that can be considered for evaluating the project hazard research method. In this regard, as indicated in Table 1, the standard
level. Each of these hazardous trades is characterized by its attrib- guideline has also been presented by Hallowell and Gambatese
utes. Based on the research conducted by Imriyas et al. (2008), a (2010) to implement the Delphi method.
framework can be developed to evaluate the PHI for this study. The At the end of the first phase, the study determines hazardous
hazardous trades and their physical attributes are shown at Level 2 trades and their measurement attributes that represent the hazard
and Level 3, respectively, in Fig. 1, whereas the project hazard level level of different projects.
measured by the PHI is at Level 1 in Fig. 1.
The MADM approach has mainly four parts: (1) alternatives;
Steps in Phase II of Research
(2) attributes; (3) weight or relative importance of each attribute;
and (4) measure or performance of alternatives with reference to the Subsequently, in the second phase, the relative importance and
attributes. For illustration, consider one of the hazardous trades— weight of hazardous trades and their attributes are evaluated and a
roofing work. It is very difficult to measure the hazards involved in method to evaluate the project hazard level of a construction project
roofing work, and thus it needs some measurable physical attributes is presented. As the attributes are interdependent, fuzzy measures
that can reflect its hazard level. Quantum (volume) of roofing and fuzzy integrals are employed to evaluate and decide the relative
works, height of the roofing works, types of roofing materials, importance of physical attributes of each hazardous trade.
and inclination of the roof may be considered as some of the attrib- The concept of fuzzy measure was first presented by Sugeno
utes of roofing work. In addition, the relative importance and (1974). The fuzzy measure generalizes the usual definition of a
weight of each attribute needs to be known, even though the measure by replacing the usual additivity property with a monot-
MADM method specifies how to process and evaluate attribute onicity property with respect to set inclusion. Similar attempts
information. To avoid difficulties in calculation, Rao (2007) sug- regarding nonadditive measures were made by Shafer (1976) and
gests using no more than 10 alternatives (hazardous trades) in any Zadeh (1978). As suggested by Sugeno (1974), fuzzy measures are
MADM method. Therefore, this study needs to determine the 10 suitable to express grades of fuzziness, for example, the quantities
most important hazardous trades, with their physical attributes, and depending on human subjectivity. However, by the nature of the
Fig. 1. Framework for evaluation of project hazard level

Fig. 2. Research methodology
Table 1. Guidelines for the Delphi Procedure

Characteristic Minimum requirement by Hallowell and Gambatese (2010) Minimum requirement adopted in this study
Identifying potential • Membership in a nationally recognized committee in the • Membership of National Safety Council of India (NSCI)
experts focus area of the research (e.g., ASCE site safety committee)
• Primary writer of publications in ASCE journals • Known participation in similar expert-based studies
• Known participation in similar expert-based studies • Delegation and expert list of conference, seminar, training,
and other events
Qualifying panelists Experts must satisfy at least four of the following criteria in Experts must satisfy at least four of the following criteria in
as experts the topics related to the research: the topics related to the research:
• Primary or secondary writer of at least three peer-reviewed • Invited to present at a conference, training program
journal articles
• Invited to present at a conference • Member or chair of a nationally recognized committee,
society, or council
• Member or chair of a nationally recognized committee • At least 10 years of professional experience in safety in the
construction industry
• At least 5 years of professional experience in the • At least 2weeks of OHS related training
construction industry
• Faculty member at an accredited institution of higher • Primary or secondary writer of at least three peer-reviewed
learning journal articles
• Writer or editor of a book or book chapter on the topic of • Advanced degree in the field of civil engineering, CEM,
construction safety and health, or risk management or other related fields (minimum of a BE/B.Tech)
• Advanced degree in the field of civil engineering, CEM, • Professional registration such as Registered Engineer (PE),
or other related fields (minimum of a BS) Licensed Engineer
• Professional registration such as professional engineer (PE),
licensed
• Architect (AIA), certified safety professional (CSP),
associated risk manager (ARM)
Number of panelists 8–12 10
Number of rounds 3 3
Feedback for each round — —
Round 1 Data from preliminary research or archived data (if available) Data from preliminary research, existing literature,
interviews with experts or archived data (if available)
Round 2 Median response from Round 1 Median response from Round 1
Round 3 Median response from Round 2 and reasons for outlying Median response from Round 2 and reasons for outlier
responses responses
Measuring consensus Absolute deviation Absolute deviation (mean), coefficient of variation, range
of data
Note: BE = bachelor of civil engineering; CEM = construction engineering and management; OHS = occupational health and safety.
definition of a fuzzy measure g, the measure of the union of two and the λ-fuzzy measure for λ ¼ 0 is a probability measure (Banon
disjointed subsets cannot be directly computed from the component 1981). Appendix I provides further details on λ-fuzzy measure.
measures. In light of this, Sugeno (1974) introduced the so-called Using the notion of fuzzy measures, Sugeno (1974) presented
λ-fuzzy measures. A λ-fuzzy measure is indeed a fuzzy measure the concept of the fuzzy integral, which is an aggregator operator

Table 2. Linguistics Values of W ij and X ij
Linguistic variables for W ij Fuzzy linguistic values for W ij Linguistic variables for X ij Fuzzy linguistic values for X ij
Extremely unimportant (EUI) (0.00, 0.00, 0.10, 0.15) Extremely low (EL) (0.00, 0.00, 0.10, 0.15)
Very un-important (VUI) (0.10, 0.15, 0.25, 0.30) Very low (VL) (0.10, 0.15, 0.25, 0.30)
Little un-important (LUI) (0.25, 0.30, 0.40, 0.45) Little low (LL) (0.25, 0.30, 0.40, 0.45)
Neutral (N) (0.40, 0.45, 0.55, 0.60) Neutral (N) (0.40, 0.45, 0.55, 0.60)
Little important (LI) (0.55, 0.60, 0.70, 0.75) Little high (LH) (0.55, 0.60, 0.70, 0.75)
Very important (VI) (0.70, 0.75, 0.85, 0.90) Very high (VH) (0.70, 0.75, 0.85, 0.90)
Extremely important (EI) (0.85, 0.90, 1.00, 1.00) Extremely high (EH) (0.85, 0.90, 1.00, 1.00)
in multicriteria decision making. In brief, the λ-fuzzy measure and Y

n
fuzzy integral concept deal with the problems of weights; this is λþ1¼ ð1 þ λgi Þ ð4Þ
i¼1
nonadditive and therefore is known as a nonadditive fuzzy inte-
gral. Hence, this study uses a λ-fuzzy measure and fuzzy integral
to evaluate the attributes of hazrdous trades. Hence, if the fuzzy densities, gi , are known for i ¼ 1; 2; : : : ; n,
the λ-fuzzy measure can be constructed. Where gij ¼ SðW g ij Þ,
Evaluation of Attributes Using Fuzzy Measures and then
Integrals
l
The attributes of hazardous trades are evaluated in the following 1 Y
four steps: gλ ½ðX i1 ; X i2 ; : : : ; X il Þ ¼ ½1 þ λSðW ij Þ − 1
f f f g ð5Þ
λj¼1
Step 1: Define the Membership Functions of Fuzzy Linguistic
Sets. In this real evaluation process, respondents were asked to
evaluate the performance and the importance of the attribute on fij Þ, if SðX
Suppose hðX ij Þ ¼ SðX fil Þ > SðXg f
iðl−1Þ Þ > · · · > SðX i1 Þ
a seven-point scale. Accordingly, linguistic variable for X ij (lingus- then
tic evaluation value of jth attribute for ith hazardous trade for a
given construction project) and W ij (lingustic weight value of fi1 ; · · · ; X
hi ¼ hðX ij Þfgλ ðX fil Þ − gλ ½X~ i1; : : : ; Xg
iðl−1Þ g
jth attribute for the ith hazardous trade) and their membership func-
tions are defined consisting of seven elements as given in Table 2. þ · · · þhðX i1 Þgλ ½Xg f
iðl−1Þ ¼gλ ðX il ÞfhðX il Þ − h½X iðl−1Þ g
The symmetrical trapezoidal-shaped membership functions, which fi1 ÞhðX i1 Þ
þ · · · þgλ ðX ð6Þ
are used most often, are employed in this study.
Step 2: Aggregate Fuzzy Values of Both Attributes and Weights.
Saaty (1986) recommends the use of the geometric mean for the syn- After getting the value for the ith attribute, the weight of 10
thesis of judgments. According to the linguistic values defined in hazardous trades is determined by employing fuzzy analytical hi-
Step 1, after taking the geometric mean of the k (number of respond- erarchy process (FAHP) in the second level. Because 10 first-level
ents) values, fuzzy evaluation value of the jth attribute for the ith hazardous trades are independent, the additivity method, usually
hazardous trade X fij , and fuzzy weight value of the jth attribute FAHP, is appropriate for this kind of problem (Patel 2015).
for ith hazardous trade W g ij are computed following Eqs. (1) and (2):
Analytical hierarchy process (AHP) is easier to understand and
it can effectively handle both qualitative and quantitative data. The
application of AHP does not involve cumbersome mathematics.
fij ¼ ðX ij X ij · · · X ij Þ1=k
X ð1Þ However, the usual AHP cannot approximate the human thinking
1 2 k
style because of uncertainty and vagueness (Lee 2014). To handle
the uncertainty and vagueness of subjective perception and the ex-
g 1=k perience of the human decision-making process, many FAHP meth-
W ij ¼ ðW ij1 W ij2 · · · W ijk Þ ð2Þ
ods are proposed by several authors (Skibniewski and Chao 1992;
Hastak and Halpin 2000; Shapira and Goldenberg 2005).
Step 3: Defuzzification. Many methods of defuzzification are
The FAHP determines the weights of decision criteria for each
available (Chen and Hwang 1992). Delgado et al. (1998) suggested
relative interest group, including the owners, users, and experts.
that it is not proper to use a single method to transform from linguistic
The advantages of the FAHP method over Saaty’s (1980) classical
domain to numerical domain. They recommended using multiple
AHP method are: (1) decision makers have the liberty of estimation
transformation functions for defuzzification. In this study, three
regarding the overall goal of using fuzzy numbers—judgment can
methods are used to obtain defuzzification value: (1) distance meas-
g g go from optimistic to pessimistic; (2) fuzzy numbers are favorable
urement M 1 ðW ij Þ (Chen 2000); (2) central value M 2 ðW ij Þ (Delgado
g to extend the range of a distinct decision matrix in the classical
et al. 1998); and (3) center of gravity (center of area) M 3 ðW ij Þ.
AHP method; (3) the combination judgment from subcriteria to
The mean of M1 ðW g Þ, M 2 ðWg Þ, M 3 ð g
W Þ, is the final
ij ij ij major criteria is more suitable than the traditional single eigenvec-
g
defuzzification number MðW ij Þ as tor method, which forms a square comparison matrix for all the
criteria.
g
M 1 ðW g g
ij Þ þ M 2 ðW ij Þ þ M 3 ðW ij Þ
g
MðW ij Þ ¼ ð3Þ Determination of Weight of Hazardous Trades Using FAHP
3 The procedures for determining the weight of each hazardous trade
using FAHP can be summarized in the following five steps:
Step 4: Fuzzy Measure and Fuzzy Integral. According to • Step 1: Construct fuzzy pairwise comparison matrix by ques-
Eqs. (13) and (14) given in Appendix I and gðXÞ ¼ 1, the following tionnaire survey: The results of the questionnaire are converted
can be obtained: into a fuzzy pairwise comparison matrix by using Satty’s (1980)

Table 3. Scale for Comparison and Triangular Fuzzy Numbers
Fuzzy number Definition Explaination Scale of fuzzy number
1~ Equal importance Two criteria contribute equally to the objective (1,1,1)
3~ Weak importance Judgment slightly favors one criteria over another (2,3,4)
5~ Essential importance Judgment strongly favors one criteria over another (4,5,6)
7~ Very strong importance A criteria is favored very strongly over another (6,7,8)
9~ Absolute importance The evidence favoring one criteria over another is of the (9,9,9)
highest possible order
~ 4,
2, ~ 8~
~ 6, Intermediate values between x ¼ 2, 4, 6, and 8 (x − 1, x, x þ 1)
adjacent scale values
nine-point scale and their corresponding triangular fuzzy num- value of Pi is considered as the most hazardous project. Thus,
bers (TFN) as given in Table 3. at this level, using Eq. (11), overall hazardous value is computed
• Step 2: Group Aggregation: After constructing the fuzzy pair-
through an operation of WSM that integrates the score of each

wise comparison matrix, the geometric mean of each hazardous hazardous trade.
trade in the matrix can be calculated as follows (Buckley 1985): Thus, for this study, the final value of the PHI can be
calculated as
g
M fij 1 m
ij ¼ ðm fij k Þ1=k
fij 2 : : : : : : m ð7Þ
PHIi ¼ h1 · W 1 þ h2 · W 2 þ · · · þh10 · W 10 ð11Þ
where Mg ij = integrated triangular fuzzy number; mfij k = ith to
the jth attribute pair comparison value by the respondent’s k. where PHIi = numerical value of the hazard level for a construction
• Step 3: Construct the fuzzy positive reciprocal matrix: Obtain- project i.
ing the final calculated fuzzy numbers could form the fuzzy Delgado et al. (1998) also proposed a numerical linguistic trans-
positive reciprocal matrix after Step 2 as follows: formation function, which provides a representative label for a
given numerical value (Appendix II).
g
M ¼ ½M ij
where M = fuzzy positive reciprocal matrix and Mg g

ji ¼ 1=M ij . Data Collection
• Step 4: Estimate the attributes’ fuzzy weights: By using the
formula proposed by Buckley (1985), the fuzzy weight can be In the first phase of the study, a questionnaire survey was conducted
calculated as follows: to shortlist 10 typical hazardous trades based on experts’ opinions.
To apply the Delphi method, this study modifies selection criteria
Zei ¼ ðf
ai1 ⊗ af f
i2 ⊗ · · · ·⊗ a in Þ
1=n
; ∀ i ¼ 1; 2; : : : n; for panelists and experts (Table 1) in the context of the Indian con-
fi ¼ Zei ⊗ ðZ
W f1 Z
f2 · · · ·· Z
fn Þ −1
ð8Þ struction industry. To emphasize greater field experience, at least
10 years of professional experience in safety was fixed as one of
where af ij = relative importance between hazardous trade i and j;
the criteria for selection of experts. The experts belong to compa-
Zei = fuzzy geometric mean of hazardous trade i; and W fi = fuzzy nies that are the members of Construction Industry Council of India
weight corresponding to hazardous trade i. (CIDC) New Delhi. These companies are big and self-performing
• Step 5: Defuzzification and normalization: The defuzzification work organizations and have established safety management sys-
process can be conducted using the method of center of gravity tem in their building, industrial, and infrastructures projects.
(center of area) M 3 ðW gij Þ. Thus, for i ¼ 1; 2; : : : ; k, the weight
Based on existing literature reviews (NICMAR 1998; BIS 2001;
vector can be obtained using Eq. (9) Imriyas et al. 2008) and interviews with 10 experts in the first round
of survey under the Delphi process, 16 hazardous trades and their
f1 Þ; MðW
W ¼ ½MðW f2 Þ; : : : ::MðW
fk ÞT ð9Þ physical attributes of a typical construction project were selected as
listed in Table 4. In the second round of the survey, the findings of
where W = nonfuzzy number. the first round of interviews were presented to the experts. Experts
were asked to evaluate the importance of the hazardous trades and
Computing Project Hazard Index (PHI) Using Weighted physical attributes on a seven-point scale varying from extremely
Sum Method (WSM) unimportant (1) to extremely important (7). In the seven-point
Finally, to obtain the PHI, the WSM that integrates the score of scale, Point 4 stands for neutral, while Points 2, 3, 5, and 6 stand
each hazardous trade is used. The WSM is the simplest but most for intermediate values to reflect compromises as indicated in
widely used multiattribute decision-making method (Rao 2007; Table 4. The objective is to verify the consensus of the safety ex-
Chou et al. 2008). In this method, each hazardous trade is assigned perts and the reliability of the attributes for each hazardous trade as
a weight, and the sum of all weights must be 1. Each project is well as to limit the hazardous trades to a maximum of 10. Based on
assessed with reference to every hazardous trade. In this way, the the results of the second round, all required statistical parameters
overall or composite performance score of the project is calculated like absolute deviation (AD) (median), coefficient of variation
using Eq. (10) (CV), and range of the data were computed as presented in Table 5
to judge whether experts have reached consensus to a certain
X
h
Pi ¼ W j ðhij Þnormal ð10Þ extent. Jordan and Javernick-Will (2013) and Cao (2006) computed
j¼1 absolute deviation (AD) (median) and coefficient of variation (CV)
in their studies by using the Delphi method. Statistical parameters
where ðhij Þnormal = normalized value of ðhij Þ; and Pi = overall or and their cut-off values depend on the requirement of the study.
composite score of the alternatives. The project with the highest Therefore, it needs to fix the statistical parameters because of a lack

Table 4. List of Hazardous Trades and their Attributes
Hazardous trades and attributes Likert scalea
1. Demolition work 1 2 3 4 5 6 7
a. Volume/size of demolition 1 2 3 4 5 6 7
b. Type of structure and reasons of its demolition 1 2 3 4 5 6 7
c. Method and equipment used for demolition 1 2 3 4 5 6 7
d. Waste material (disposal, quantity, type of material, space for storage etc) 1 2 3 4 5 6 7
2. Excavation work 1 2 3 4 5 6 7
a. Excavation configuration (depth, width, length, slope) 1 2 3 4 5 6 7
b. Geological condition (soil type, water table, etc) 1 2 3 4 5 6 7
c. Underground utilities (electrical, water, and sewer lines, telephone line, gas line etc) 1 2 3 4 5 6 7
d. Nearby vehicular traffic (vibration and surcharge) 1 2 3 4 5 6 7
e. Nearby structures 1 2 3 4 5 6 7
3. Scaffolding and ladder usage 1 2 3 4 5 6 7
a. Volume of scaffolding and ladder usage 1 2 3 4 5 6 7
b. Height of the scaffold/ladder to be used 1 2 3 4 5 6 7
c. Adequacy of design (type of material, member size, bracing, guardrails, platform size, toe board) 1 2 3 4 5 6 7
4. Falsework (temporary structure) 1 2 3 4 5 6 7
a. Volume of falsework involved in the project 1 2 3 4 5 6 7
b Adequacy of design (material, member size, bracing, guardrails, platform size, toe board) 1 2 3 4 5 6 7
5. Roof work 1 2 3 4 5 6 7
a. Quantum (volume) of roofing involved 1 2 3 4 5 6 7
b. Height of the roof 1 2 3 4 5 6 7
c. Roofing material property such as slippery, brittleness, asbestos etc 1 2 3 4 5 6 7
d. Inclination of the roof 1 2 3 4 5 6 7
6. Erection of steel/precast concrete structures 1 2 3 4 5 6 7
a. Volume of erection work 1 2 3 4 5 6 7
b. Height of erection work 1 2 3 4 5 6 7
c. Erection method (partial/full erection at height, labor involvement level) 1 2 3 4 5 6 7
d. Adequacy of design (holding down bolts, anchor points, bracings, alignment, positioning, etc) 1 2 3 4 5 6 7
7. Lifting and hoisting machinery 1 2 3 4 5 6 7
a. Volume of lifting and hosting involved 1 2 3 4 5 6 7
b. Nature of materials lifted and hoisted 1 2 3 4 5 6 7
c. Operating platform 1 2 3 4 5 6 7
d. Nature of site vicinity (nearby structures, overhead cables, etc) 1 2 3 4 5 6 7
8. Construction tools and machinery use 1 2 3 4 5 6 7
a. Volume of plant and machinery used 1 2 3 4 5 6 7
b. Operating platform of plant and machinery (i.e., slope etc) 1 2 3 4 5 6 7
c. Site layout 1 2 3 4 5 6 7
d. Volume of tools used 1 2 3 4 5 6 7
e. Type of tools used 1 2 3 4 5 6 7
9. Works on contaminated sites 1 2 3 4 5 6 7
a. Type of contaminants on the site 1 2 3 4 5 6 7
b. Quantity of contaminants present 1 2 3 4 5 6 7
c. Duration of work on contaminated site 1 2 3 4 5 6 7
10. Welding, cutting and hot work 1 2 3 4 5 6 7
a. The volume of welding, cutting and hot works 1 2 3 4 5 6 7
b. Location of welding (confined space, underground, on ladders etc) 1 2 3 4 5 6 7
11. Work in confined spaces (including bore holes) 1 2 3 4 5 6 7
a. The volume of confined space works 1 2 3 4 5 6 7
b. Confined space configuration 1 2 3 4 5 6 7
c. Type of activity to be involved (e.g., welding, waterproofing, etc) 1 2 3 4 5 6 7
d. Current usage of the confined space (if any) 1 2 3 4 5 6 7
12. Electrical work 1 2 3 4 5 6 7
a. Power lines configuration 1 2 3 4 5 6 7
b. Extension codes (numbers) 1 2 3 4 5 6 7
c. Generator configuration 1 2 3 4 5 6 7
13. Deep foundation and piling 1 2 3 4 5 6 7
a. Foundation configuration (depth, width, length, slope) 1 2 3 4 5 6 7
c. Number of cranes used for lowering workers 1 2 3 4 5 6 7
d. Number of pile driven machines 1 2 3 4 5 6 7
14. Tunnelling 1 2 3 4 5 6 7
a. Tunneling configuration (depth, width, length, slope) 1 2 3 4 5 6 7
c. Underground utilities (electrical, water, and sewer lines, telephone line, gas line etc) 1 2 3 4 5 6 7
d. Nearby vehicular traffic (vibration and surcharge) 1 2 3 4 5 6 7
e. Tunneling method (e.g., Drilling, blasting, use of tunneling machine etc) 1 2 3 4 5 6 7
f. Dewatering (quantity and method) 1 2 3 4 5 6 7

Table 4. (Continued.)
Hazardous trades and attributes Likert scalea
15. Road work 1 2 3 4 5 6 7
a. Road configuration (length, lane, width, location, etc) 1 2 3 4 5 6 7
b. Traffic volume on the road 1 2 3 4 5 6 7
c. Use of machinery (excavators, graders, pavers, rollers, haulers etc) 1 2 3 4 5 6 7
d. Hot mix plant configuration 1 2 3 4 5 6 7
16. Concrete work 1 2 3 4 5 6 7
a. Concrete configuration (volume, location, etc) 1 2 3 4 5 6 7
b. Concrete mixing (in situ, ready mix, etc) 1 2 3 4 5 6 7
c. Concrete/material transportation and placing (length and methods) 1 2 3 4 5 6 7
d. Curing method 1 2 3 4 5 6 7
Note: A score of 4 stands for neutral while scores of 2, 3, 5, and 6 stand for intermediate values to reflect compromises.
a
Extremely unimportant (1) to extremely important (7).
Table 5. Statistical Parameters of Hazardous Trades

Hazardous trade Mean Median SD CV AD (median) Maximum Minimum Range
1. Erection of steel/precast concrete structure 5.80 6.00 0.92 0.16 0.60 7.00 4.00 3.00
2. Tunneling 6.40 6.50 0.70 0.11 0.60 7.00 5.00 2.00
3. Concrete work 3.60 3.00 0.84 0.23 0.60 5.00 3.00 2.00
4. Falsework (temporary structures) 5.10 5.00 0.99 0.19 0.70 7.00 4.00 3.00
5. Construction tools and machinery use 5.10 5.00 0.99 0.19 0.70 7.00 4.00 3.00
6. Excavation work 6.00 6.00 1.25 0.21 0.80 7.00 3.00 4.00
7. Scaffolding and ladder usage 5.80 5.50 0.92 0.16 0.80 7.00 5.00 2.00
8. Roof work 4.60 4.50 0.97 0.21 0.80 6.00 3.00 3.00
9. Lifting and hoisting machinery 6.20 7.00 1.14 0.18 0.80 7.00 4.00 3.00
10. Welding, cutting, and hot work 4.60 4.50 0.97 0.21 0.80 6.00 3.00 3.00
11. Works in confined spaces (including boreholes) 5.80 6.00 1.23 0.21 0.80 7.00 3.00 4.00
12. Electrical work 5.60 6.00 1.35 0.24 1.00 7.00 3.00 4.00
13. Deep foundation and piling 5.40 5.00 1.26 0.23 1.00 7.00 4.00 3.00
14. Road work 4.40 4.00 1.35 0.31 1.00 6.00 2.00 4.00
15. Demolition work 5.40 6.00 1.65 0.30 1.20 7.00 2.00 5.00
16. Work on contaminated sites 5.00 5.00 1.76 0.35 1.40 7.00 2.00 5.00
Note: AD = absolute deviation; CV = coefficient of variation; SD = standard deviation.
of a guideline to measure the consensus in the Delphi method (4) underground utilities and nearby vehicular traffic were deleted
(Hallowell and Gambatese 2010). Hence, this study adopts the cri- from tunneling; and (5) concrete mixing was deleted from concrete
teria that absolute deviation (AD) (median), coefficient of variation work. Thus the alpha value of almost all hazardous trades was im-
(CV), and range of the data, should be less than 1.00, 0.24, and 5, proved; these are shown in brackets in the second column of Table 6.
respectively. The criteria were decided based on variation in data It is observed that deleted attributes were difficult to measure in
and previous studies (Cao 2006; Hallowell and Gambatese 2010; order to evaluate their corresponding hazardous trade. However,
Jordan and Javernick-Will 2013; Patel 2015). As a result, five it appears that lifting and hoisting machinery (0.667), welding, cut-
hazardous trades—electrical work, deep foundation, road works, ting, and hot works (0.689), and works in confined spaces (0.420)
demolition works, and works on contaminated sites—were re- could not reach 0.70, although the alpha value of lifting and hoisting
moved from the list because either one or more of their statistical machinery (0.667) and welding, cutting, and hot works (0.689) is
parameters were higher than the permissible limit (Table 5). near 0.70. In this regard, Churchill (1979) advocates that a value of
Afterward, the Cronbach’s alpha was calculated for the remain-
0.6 can be considered as acceptable for newly developed scales.
ing 11 hazardous trades to test reliability as indicated in Table 6.
Therefore, this study has considered both of these hazardous
Reliability estimates of all hazardous trades were greater than 0.7,
trades—lifting and hoisting machinery (0.667) and welding, cut-
providing evidence of internal consistency within items (Hair et al.
1995). Referring to Table 6, scaffolding and ladder usage (0.381); ting, and hot works (0.689)—and dropped works in confined spaces
erection of steel (0.352); lifting and hoisting machinery (0.667); (0.420) because of its low value of alpha from the final list of haz-
welding, cutting, and hot works (0.689); works in confined spaces ardous trades to evaluate the project hazard level. Almost all haz-
(0.314); tunneling (0.604); and concrete work (0.580) have alpha ardous trades have an attribute related to volume or configuration of
values lower less than 0.7. However, alpha values could be im- work. Possibly this has led to the low value of alpha for the attribute
proved by dropping irrelevant attributes (Hair et al. 1995). There- “work in confined spaces”. The attribute “nearby vehicular traffic”
fore, as indicated by italics the first column of Table 6, the following obtains a low alpha score because this gets overshadowed by the
were deleted: (1) volume of scaffolding and ladder usage two attributes of tunneling, namely “tunneling configuration”
was deleted from scaffolding and ladder usage; (2) volume of and “geological condition”. Thus, at the end of the second round
erection work was deleted from erection of steel; (3) confined space of the Delphi method, 10 hazardous trades and their attributes were
configuration was deleted from works in confined spaces; finally shortlisted as presented in Table 7 to evaluate the PHI.

Table 6. Cronbach’s Alpha of Hazardous Trades (Measures)
Hazardous trades and their attributes Alpha
1. Scaffolding and ladder usage
a. Volume of scaffolding and ladder usage 0.381 (0.816)
b. Height of the scaffold/ladder to be used
c. Adequacy of design (type of material, member size, bracing, guardrails, platform size, and toe board)
2. Falsework (temporary structures)
a. Volume of falsework involved in the project 0.792
b. Adequacy of design (material, member size, bracing, guardrails, platform size, and toe board)
3. Roof work
a. Volume of roofing involved 0.782
b. Height of the roof
c. Roofing material property such as slippery, brittleness, asbestos, etc
d. Inclination of the roof
4. Erection of steel/precast concrete structure
a. Volume of erection work 0.352 (0.728)
b. Height of erection work

c. Erection method (partial/full erection at height, and labor involvement level)
d. Adequacy of design (holding down bolts, anchor points, bracings, alignment, positioning, etc.)
5. Lifting and hoisting machinery
a. Volume of lifting and hosting involved 0.667
b. Nature of materials lifted and hoisted
c. Operating platform
d. Nature of site vicinity (nearby structures, overhead cables, etc.)
6. Construction tools and machinery use
a. Volume of plant and machinery used 0.748
b. Operating platform of plant and machinery (i.e., slope, etc.)
c. Site layout
d. Volume of tools used
e. Type of tools used
7. Welding, cutting and hot work
a. The volume of welding, cutting and hot works 0.689
b. Location of welding (confined space, underground, on ladders, etc.)
8. Work in confined spaces (including bore holes)
a. The volume of confined space works 0.314 (0.420)
b. Confined space configuration
c. Type of activity to be involved (e.g., welding, waterproofing, etc)
d. Current usage of the confined space (if any)
9. Excavation work
a. Excavation configuration (depth, width, length, slope) 0.781
b. Geological condition (soil type, water table, etc.)
c. Underground utilities (electrical, water, and sewer lines, telephone lines, gas lines, etc,)
d. Nearby vehicular traffic (vibration and surcharge)
e. Nearby structures
10. Tunneling
a. Tunneling configuration (depth, width, length, and slope) 0.604 (0.721)
b. Geological condition (soil type, water table, etc.)
c. Tunneling method (e.g., drilling, blasting, use of tunneling machine, etc.)
d. Dewatering (quantity and method)
e. Underground utilities (electrical, water, and sewer lines, telephone lines, gas lines, etc.)
f. Nearby vehicular traffic (vibration and surcharge)
11. Concrete work
a. Concrete configuration (volume, location, etc.) 0.580 (0.734)
b. Concrete mixing (in situ, ready mix, etc.)
c. Concrete/material transportation and placing (length and methods)
d. Curing method
During the third and final round of the Delphi process, a panel of evaluate the project hazard level of a typical construction project
experts was asked to look at the results, analysis, and the final lists as indicated in Table 7.
of the hazardous trades obtained from round two, and rate the In the second phase of the study, a questionnaire survey was
statements accordingly again. Furthermore, whenever their individ- conducted to obtain necessary data to determine the weight of haz-
ual ratings were found to be different from the group rating, they ardous trades and their physical attributes. There are three sections
were asked to give an explanation. Hence, there was consensus in the questionnaire. The first section asks about the demographics
among all experts regarding the final list of hazardous trades with of the respondents. The second section is for the weight of hazard-
their physical attributes to evaluate the project hazard level. In brief, ous trade (W i ) in the fuzzy AHP process, and the third is for the
at the end of the first phase of the study, a list 10 typical hazardous weight of an attribute corresponding to its hazardous trade (W ij ) in
trades and their physical attributes was obtained and finalized to the fuzzy measure and fuzzy integral process.

Table 7. Final List of Hazardous Trades with their Attributes
Hazardous trade Attributes
Scaffolding and ladder usage Height of the scaffold/ladder to be used
Adequacy of design (type of material, member size, bracing, guardrails, platform size, toe board)
False work (temporary structures) Volume of false work involved in the project
Adequacy of design (material, member size, bracing, guardrails, platform size, and toe board)
Roof work Volume of roofing involved
Height of the roof
Roofing material property such as slippery, brittleness, asbestos, etc.
Inclination of the roof
Erection of steel/precast concrete structure Height of erection work
Erection method (partial/full erection at height, and labor involvement level)
Adequacy of design (holding down bolts, anchor points, bracings, alignment, positioning, etc.)
Lifting and hoisting machinery Volume of lifting and hosting involved
Nature of materials lifted and hoisted
Operating platform
Nature of site vicinity (nearby structures, overhead cables, etc.)

Construction tools and machinery use Volume of plant and machinery used
Operating platform of plant and machinery (i.e., slope, etc.)
Site layout
Volume of tools used
Type of tools used
Welding, cutting, and hot work The volume of welding, cutting and hot works
Location of welding (confined space, underground, on ladders, etc.)
Excavation work Excavation configuration (depth, width, length, and slope)
Geological condition (soil type, water table, etc.)
Underground utilities (electrical, water, and sewer lines, telephone lines, gas lines, etc.)
Nearby vehicular traffic (vibration and surcharge)
Nearby structures
Concrete work Concrete configuration (volume, location, etc.)
Concrete/material transportation and placing (length and methods)
Curing method
Tunneling Tunneling configuration (depth, width, length, and slope)
Geological condition (soil type, water table, etc.)
Tunneling method (e.g., drilling, blasting, use of tunneling machine, etc.)
Dewatering (quantity and method)
In the fuzzy AHP (second) section of the questionnaire, re- with all 10 hazardous trades selected in the previous section.
spondents were asked to assess the relative importance among Participants were approached at sites, conferences, training centers,
10 hazardous trades that were derived in the first phase of the workshops, and elsewhere. A total of 64 responses were collected
study using the Delphi method. A Likert scale of 1–9 was used from different kinds of construction projects like metro rail, high-
in the questionnaire, where AI (9) stands for absolute importance; way, and building construction projects across India.
DI (7) stands for demonstrated importance; SI (5) stands for strong
importance; WI (3) stands for weak importance; EI (1) stands for
equal importance; and Options 2, 4, 6, and 8 stand for intermediate Development of Project Hazard Index
values to reflect compromises. Similarly, in the third section, for
the fuzzy measure and fuzzy integral part, respondents were asked As shown in Fig. 1, there are three levels, namely top, middle, and
to evaluate the importance of each third level attribute on a 7-point bottom, in the framework. The process of evaluation is from bottom
scale from extremely unimportant (EUI) to extremely important to top, that is, from third level to first level as explained in the
(EI), where (1) stands for extremely unimportant; (7) stands for following subsections.
extremely important; (4) stands for neutral; and Options 2, 3, 5,
and 6 stand for intermediate values to reflect compromises.
A total of 40 undergraduate students in their final year and 5 Measurement of Attributes at the Third Level of the
safety professionals working and possessing at least 10 years of Hierarchy (Using Fuzzy Measure and Fuzzy Integral)
experience in construction projects were selected to interview as The 10th hazardous trade, tunneling, and its four attributes of
part of a pilot test to check the understanding of the questionnaire. (1) tunneling configuration, (2) geological condition, (3) tunneling
After their responses, the questionnaire was revised to make it more method, and (4) dewatering are taken to illustrate the results of the
comprehensible. Revisions of the questionnaire were mostly related four steps for fuzzy measure and fuzzy integral.
to its language, structure, understanding of the statements, etc. In Step 1, each respondent was asked to assess the importance
The questionnaire survey was conducted in the Indian construction of these four attributes on a seven-point scale, from EUI (extremely
industry to obtain the opinions of practitioners and executives (site unimportant) to EI (extremely important). For example, the re-
engineers, project executives, planning engineers, safety officers sponses of one respondent are neutral (N), very important (VI),
etc.) associated with the execution of construction projects and little important (LI), very important (VI), very unimportant (VUI),
who possessed at least 5 years of experience. Before the interviews, and extremely important (EI). The respective fuzzy values are
it was ensured that the respondents had enough experience to work (0.40, 0.45, 0.55, 0.60), (0.70, 0.75, 0.85, 0.90), (0.55, 0.60, 0.70,

Table 8. Fuzzy Linguistic Values of a Hazardous Trade: Tunneling Table 9. λ Fuzzy Measure of Hazardous Trades
Defuzzification λ fuzzy
Attribute Fuzzy linguistic value value Hazardous trade measure Weight
Tunneling configuration (0.601, 0.643, 0.743, 0.774) 0.505 Scaffolding and ladder usage −0.629 0.066
Geological condition (0.528, 0.570, 0.670, 0.705) 0.429 Construction tools and machinery use −0.962 0.070
Tunneling method (0.626, 0.669, 0.769, 0.794) 0.531 Lifting and hoisting machinery −0.956 0.153
Dewatering (0.531, 0.574, 0.674, 0.711) 0.435 Welding, cutting, and hot work 0.111 0.083
Erection of steel/precast concrete structure −0.915 0.092
False work (temporary structure) 0.059 0.053
Excavation work −0.988 0.096
0.75), (0.70, 0.75, 0.85, 0.90), (0.10, 0.15, 0.25, 0.30), and (0.85, Roof work −0.962 0.108
0.90, 1.00, 1.00) (Table 2). Concrete work −0.664 0.049
In Step 2, aggregating the 64 evaluation values yields the fuzzy Tunneling −0.890 0.230
linguistic values of the tunneling configuration attribute. Similarly,
in Step 3, their corresponding defuzzification values can also be
computed. The fuzzy linguistic values of the tunneling configura-

and ladder usage (Rubio-Romero et al. 2013), false work (Rubio-
tion attribute with their defuzzification values are given in Table 8.
Romero et al. 2013), and concrete work (El-Shahhat et al. 1993).
In Step 4, the value of λ should be first obtained using Eq. (5),
Most of the accidents are associated with tunneling (Wu et al.
which is λþ1¼ð1þ0.505λÞð1þ0.429λÞð1þ0.531λÞð1þ0.435λÞ.
2015), lifting and hoisting machinery (Neitzel et al. 2001), and roof
The value of λ is (−0.890).
work (Johnson et al. 1998). This shows that the results obtained are
For a given project, suppose the evaluation values of the four
reasonable.
attributes in descending order are hðX 101 Þ ¼ 0.650 > hðX 102 Þ ¼
0.650 > hðX 103 Þ ¼ 0.350 > hðX 104 Þ ¼ 0.22. Then, according to
Eq. (5), the fuzzy measure for the hazardous trade of tunneling Measurement of Project Hazard Index at the First Level
can be found as follows: of the Hierarchy (Using WSM)
gλ ðx1 Þ ¼ 0.505 The final index of the project hazard level can be calculated using
the following formula:
gλ ðx1 ; x2 Þ ¼ 0.505 þ 0.435 þ ð−0.890Þð0.505Þð0.435Þ ¼ 0.744 PHIi ¼ h1 · W 1 þ h2 · W 2 þ · · · þh10 · W 10
¼ h1 × 0.066 þ h2 × 0.070 þ h3 × 0.153 þ h4 × 0.083
gλ ðx1 ; x2 ; x3 Þ ¼ 0.744 þ 0.531 þ ð−0.890Þð0.744Þð0.531Þ ¼ 0.923
þ h5 × 0.092 þ h6 × 0.053 þ h7 × 0.096 þ h8 × 0.108
þ h9 × 0.049 þ h10 × 0.230 ð12Þ
gλ ðx1 ; x2 ; x3 ; x4 Þ ¼ 0.923 þ 0.429 þ ð−0.890Þð0.923Þð0.429Þ ¼ 1
The PHI is a function of 10 hazardous trades and their attributes.
Using Eq. (6), the value for the hazardous trade of tunneling can Its value varies from 0 to 1. However, a 0 or 1 value of the PHI is
be computed as follows: not realistic because PHI ¼ 0 means no hazard situation in the
project site whereas PHI ¼ 1 represents the highest level of project
h10 ¼ hðX 104 Þ × 1 þ ½ðhðX 104 Þ − hðX 103 Þ × 0.923 hazard in the project site, which does not seem practical as only an
þ ½ðhðX 103 Þ − hðX 102 Þ × 0.744 insensitive management will allow work in such situations. Higher
values of the PHI show higher project hazard levels and vice versa.
þ ½ðhðX 102 Þ − hðX 101 Þ × 0.505 ¼ ð0.200 × 1Þ
þ ð0.350 − 0.200Þ × 0.923 þ ð0.650 − 0.350Þ × 0.744
Practical Implications and Validation of the Study
þ ð0.650 − 0.650Þ × 0.505 ¼ 0.562
Based on the evaluation procedure of this framework, the value of
The other data can easily be computed using the MATLAB PHI can be easily evaluated. To determine the PHI of any project,
program and/or the MS Excel. The λs of the remaining hazardous attributes should be evaluated using linguistic terms on a seven-
trades are calculated and obtained as shown in Table 9. point Likert scale from extremely low (EL) to extremely high (EH),
where a value of 1 stands for EL; a value of 7 stands for EH;
a value of 4 stands for neutral (N); and values of 2, 3, 5, and 6
Measurement of Hazardous Trades at the Second
stand for intermediate values to reflect compromises as input, as
Level of the Hierarchy (Using Fuzzy AHP)
demonstrated in Table 2. For the sake of convenience, an online
As mentioned, five steps are employed to calculate the weights of web-based project-hazard-evaluation tool can easily be developed
the first-level hazardous trades. In Step 1, the fuzzy pairwise com- using computer software and programming. In this tool, a user has
parison matrix was created as given in Table 10. As stated earlier, to input evaluation value of the attributes of each hazardous trade
64 valid responses are available for this process. Subsequently, fol- taken on the linguistic scale only. Based this information, the tool
lowing Steps 3–5, the weight vector could be obtained as shown in will compute and display the PHI of that project. Thus construction
Table 9. The most important hazardous trade to measure the project industry can directly use the online system to measure the project
hazard level of the construction project is tunneling. This is fol- hazard level of the projects and improve them. It is also a practical
lowed by lifting and hoisting machinery, roof work, excavation tool to improve safety management of the projects. The PHI can be
work, erection of steel/precast concrete structures (El-Shahhat et al. used to compare the project hazard level among projects and it can
1993), welding, cutting, and hot works (Imriyas et al. 2007), con- also be employed to measure the hazard level of a project in differ-
struction tools and machinery use (Neitzel et al. 2001), scaffolding ent points of time.

(0.24,0.27,0.31)
(0.31,0.35,0.41)
(0.40,0.45,0.52)
(0.24,0.27,0.30)
(0.35,0.39,0.44)
(0.27,0.29,0.33)
(0.32,0.36,0.41)
(0.49,0.55,0.63)
(0.26,0.28,0.31)
(1.00,1.00,1.00)
Feng (2013) found a stronger positive effect of safety budget
(investment) on accident prevention under a higher safety culture
F10 level and project hazard level. This study has considered the differ-
ent project sites of Delhi metro construction project for the appli-
cation of the PHI and to examine the relationship between project
hazard level and safety budget. A rail route (from the source station,
(1.43,1.73,2.04)
(1.48,1.84,2.20)
(2.15,2.62,3.11)
(1.39,1.76,2.16)
(1.83,2.39,2.97)
(1.29,1.66,2.12)
(1.21,1.49,1.91)
(1.69,2.15,2.59)
(1.00,1.00,1.00)
(3.20,3.53,3.82)
Janak Puri West, to a destination station, Botanical Garden) of
Phase III of Delhi metro rail project is under construction, and it
F9
consists of 30 construction project sites. The authors visited and

computed the PHI of each project site, evaluating physical attrib-
utes of 10 hazardous trades and using Eq. (12). They also collected
(1.59,1.82,2.02) 2.02) safety budget information for each of these projects sites. Visiting
of sites and compilation of data took 50 days. After confirming
(0.62,0.70,0.81)
(0.50,0.60,0.72)
(0.88,1.06,1.31)
(0.63,0.73,0.86)
(0.79,0.94,1.11)
(0.54,0.63,0.74)
(0.71,0.86,1.06)
(1.00,1.00,1.00)
(0.38,0.47,0.59)
normality and linearity of data, Pearson correlation is computed at

the 0.01 significance level (two-tailed). It is found to be 0.624,
F8
which shows the correlation between safety budget and the PHI on
expected lines. The coefficient of determination (R2 ) is found to be
0.39. It means that 39% of the variability in the safety budget is
shared by the PHI alone. In a similar manner, the relationship be-
tween project hazard level and other determinants of safety perfor-
(0.45,0.52,0.63)
(0.61,0.68,0.79)
(1.22,1.51,1.83)
(0.64,0.77,0.95)
(0.61,0.73,0.88)
(0.59,0.72,0.89)
(1.00,1.00,1.00)
(0.94,1.16,1.41)
(0.52,0.67,0.83)
(2.46,2.81,3.10)
mance can be studied.

The proposed framework focuses on hazardous trades and their
F7
attributes to evaluate a project’s hazard level because in most

construction projects, various teams and contractors are working
tradewise. Therefore, it does not need to identify the important
hazardous trades every time, as in previous studies (Liu and Tsai
(1.69,2.06,2.46)
(1.38,1.73,2.13)
(2.01,2.46,2.96)
(1.57,1.96,2.35)
(2.03,2.58,3.19)
(1.00,1.00,1.00)
(1.12,1.38,1.69)
(1.35,1.59,1.84)
(0.47,0.60,0.78)
(3.05,3.39,3.65)
2012). Risk is the product of likelihood of an accident and its

severity. The probability approach believes in the closed-world
F6
model where every data point is known (Pinto 2014). However,

due to absence of reliable accident statistics and severity data, this
framework uses the possibility approach, which never assumes
that all data could have been known (Pinto 2014). Gürcanli and
(0.64,0.75,0.88)
(0.61,0.70,0.82)
(1.64,1.95,2.31)
(0.95,1.15,1.41)
(1.00,1.00,1.00)
(0.31,0.39,0.49)
(1.14,1.38,1.65)
(0.90,1.06,1.26)
(0.34,0.42,0.55)
(2.26,2.55,2.83)
Müngen (2009) used fuzzy logic, Mamdani fuzzy inference, and

If-Then rules in their model to transform a knowledge map into
F5
nonlinear mapping. However, it adds complexity into the model

while this proposed framework does not need to use If-Then rules
and fuzzy inference. Previous models (Dağdeviren and Yüksel
2008; Gürcanli and Müngen 2009; Pinto 2014) do not consider the
(0.45,0.55,0.67)
(0.75,0.90,1.08)
(1.99,2.35,2.75)
(1.00,1.00,1.00)
(0.71,0.87,1.05)
(0.43,0.51,0.64)
(1.06,1.29,1.55)
(1.16,1.37,1.58)
(0.46,1.76,0.72)
(3.39,3.76,4.11)
nonadditive nature of the subfactors or attributes lying at lower lev-

els in the framework of the studies, as in the proposed framework.
F4
On the other hand in this study, the 10 hazardous trades essential

in construction project sites for determining the hazard level of a
construction project were examined. These hazardous trades pro-
vide a thorough analysis of all their specific attributes involved
(0.27,0.31,0.33)
(0.28,0.31,0.36)
(1.00,1.00,1.00)
(0.36,0.43,0.50)
(0.43,0.51,0.61)
(0.30,0.41,0.50)
(0.55,0.66,0.82)
(0.76,0.94,1.14)
(0.32,0.38,0.46)
(1.92,2.23,2.50)
in the hazard level of construction sites. The use of the linguistic

variables and fuzzy integrals to assess attributes of hazardous trades
F3
proved to be a versatile and robust way of estimating and integrat-

ing them. Fuzzy set theory takes care of uncertainty, complexity of
human behavior, and the linguistic-scale measurement.
Table 10. Fuzzy Pair Wise Comparison Matrix
(0.94,1.09,1.27)
(1.00,1.00,1.00)
(2.74,3.22,3.62)
(0.92,1.11,1.33)
(1.22,1.43,1.65)
(0.47,0.58,0.73)
(1.27,1.46,1.65)
(1.40,1.67,2.00)
(0.45,0.54,0.67)
(2.42,2.85,3.25)
Summary and Conclusions

F2
This study aimed to develop a framework to process for evaluation

of the hazard level of a construction project. Therefore, 10 hazard-
ous trades with their corresponding physical attributes were deter-
(1.00,1.00,1.00)
(0.79,0.92,1.06)
(3.00,3.27,3.67)
(1.50,1.83,2.23)
(1.13,1.34,1.57)
(0.41,0.49,0.59)
(1.58,1.91,2.24)
(1.24,1.43,1.61)
(0.49,0.58,0.70)
(3.25,3.77,4.16)
mined using the Delphi method. The possibility approach was

preferred to probability approach because of the unavailability of
F1
past records and similar kinds of research in the existing literature.

FAHP was used to determine the weight of each hazardous trade,
and fuzzy measures and fuzzy integrals were adopted to evaluate
the attributes of each hazardous trade because of their nonadditive
Number
nature. To design the weight matrix in FAHP and evaluate

F10
F1
F2
F3
F4
F5
F6
F7
F8
F9
attributes, data were collected through questionnaire surveys from

several projects from different parts of India. By using WSM, the Sugeno (1974) defined fuzzy measure as follows:
PHI was developed and presented to evaluate the project hazard
level of a construction project. Thus, a framework was formulated g:ℵ → ½0; 1
to evaluate and formulate the index of project hazard level.
So far, most studies have focused on identification of hazards
and risk management using the probability approach. From a theo- 1. gðϕÞ ¼ 0, gðXÞ ¼ 1;
retical perspective, the possibility approach (non-probability) is a 2. gðϕÞ < gðXÞ if A ⊂ B; and
novel approach for evaluating the project hazard level in the con- 3. If ðAÞ∞
i¼1 is an increasing sequence of measurable sets, then
struction industry. Use of fuzzy logic in the evaluation process fur- limi→∞ gðAi Þ ¼ gðlimi→∞ Ai Þ.
ther extends the method of the AHP. In addition, fuzzy measures
and fuzzy integrals take care of attribute weights, which are non-
additive and in the same hierarchy as the evaluation framework. Properties of λ-Fuzzy
Based on this developed framework, a user-friendly web-based tool Properties of λ-Fuzzy measures are as below.
can easily be developed to compute the PHI. The computed PHI For all A, B ⊂ X and A ∩ B ¼ ϕ,
can be utilized for better resource management. Thus, this study
sets up a comprehensive map of quantification and integration of [

hazards at the project level including different trades and their gðA BÞ ¼ gðAÞ þ gðBÞ þ λgðAÞgðBÞ for some λ > −1 ð13Þ
attributes instead of considering the activities. Many subcontractors
and trades are associated with a construction project, and hazard
levels vary from one trade to another trade. Based on the weights Let X ¼ fx1 ; x1 ; : : : ; xn g be a finite set and let gi ¼ g½ðxi Þ. The
of different hazardous trades, they could be differentiated and thus mapping xi → gi is called a fuzzy density function. Suppose
the safety of crews involved in different trades can be managed A ¼ X ¼ fx1 ; x1 ; : : : ; xn g ⊆ X. A has m elements, xi1 is the first
effectively. Thus, a strategy to improve safety performance should element of the classifier i, and xim is the element m of the classi-
be developed from the whole-to-part approach. It means the safety fier i. Therefore
management system should focus on a crew or trade level to under-
stand and improve safety culture effectively. In the absence of X
m X
m−1 X
m
reliable accident data, the possibility approach can be used to evalu- gðAÞ ¼ gij þ λ gij gik þ · · · þλm−1 gi1 gi2 : : : gim
ate project hazard levels. j¼1 j¼1 k¼jþ1
In the first phase of the study, use of only 10 hazardous trades is m
1 Y
a main limitation of the study. This study does not include personal ¼ ð1 þ λgij Þ − 1 for − 1 < λ < ∞ ð14Þ
and psychological attributes of the hazardous trades in evaluation of λ j¼1
hazard level. Issues in fuzzy set theory, such as determining a mem-
bership function and selecting a fuzzy measure, still exist in the
second phase of the study. However, most construction personnel Eq. (14) is the generalization of the Eq. (13), which indicates the
do not have expertise or knowledge to carry out a hazard assess- union
P of only two subsets, A and B. The first section of Eq. (14)
ment of a project site. The purpose of this study is to fill this gap by ( m j¼1 gij ) is the summation of all the fuzzy densities from Element
developing a methodology to assess a project’s hazard level using 1 to m of the classifier i (m is the number
Pm−1 P of elements in subset A).
m
the possibility approach. In development of the PHI, the combined The second section of the Eq. (14) ( j¼1 k¼jþ1 gij gik ) means the
use of fuzzy integral, FAHP, and WSM is a novel attempt in the interinfluence of every two fuzzy densities, where gij is the jth
construction industry and safety science. fuzzy density of the classifier i and gik is the kth fuzzy density of
the classifier i. In the third section of Eq. (14), (λm−1 gi1 gi2 ; : : : ; gim )
means multiplying λ by λ for m−1 times (number of elements
Appendix I. λ Fuzzy Measure and Fuzzy Integral in subset A minus) and then multiplying the result with
(gi1 gi2 ; : : : ; gim ). These are the fuzzy density values of every
Fuzzy measures are the generalization of the classical measures. element of the subset A for classifier i. Eq. (14) calculates the
By a measurable space, it means a ðX; ℵÞ consisting of a set X λ-fuzzy measure for A, which is the subset of the finite set X. Thus,
and a ℵ-algebra of subset of X. A subset A of X is called measurable the value of λ can be found from the Eq. (14).
(or measurable with respect to ℵ) if A ∈ ℵ. A measure μ on a meas- Fuzzy integrals are nonlinear functional, analogous to Lebesque
urable space ðX; ℵÞ is a real nonnegative set function defined for integrals, where the integral is defined over measurable sets (Cao
all sets of ℵ such that μðϕÞ ¼ 0, and if ðAÞ∞ 2006). Let ðX; ℵÞ be a measurable space and let h:X → ½0,1 be a
i¼1 is a disjoint family of
sets with Ai ∈ ℵ, i ≥ 1, then ℵ-measurable function.
The fuzzy integral over A ⊂ X of the function h with respect to a
∞ X ∞ fuzzy measure is defined as follows:
μ ∪ Ai ¼ μðAi Þ
i¼1
i¼1 I
hðxÞ ∘ gð·Þ ¼ sup fmin½min hðxÞ; gðA ∩ EÞg
A measure μ has the following properties: A E⊆X x∈E
¼ sup min½α; gðA ∩ Fα Þ

μðAÞ ≤ μðBÞ if A ⊂ B α∈½0;1
1. If ðAÞ∞
i¼1 is an increasing sequence of measurable sets, then where Fα ¼ ½x:hðxÞ ≥ α.
limi→∞ μðAi Þ ¼ μðlimi→∞ Ai Þ; and A more general equation of fuzzy integral is defined by Sugeno
2. An important example of such a measure is the probability mea- (1974) as follows. Assuming that hðx1 Þ ≥ hðx2 Þ ≥; : : : ; hðxn Þ, then
sure, P, where PðXÞ ¼ 1. the fuzzy integral is

Z
where z = number of is dividing points; Gj ðst Þ = jth numerical
ðCÞ hdg ¼ hðxn ÞgðH n Þ þ ½hðxn−1 Þ − hðxn ÞgðHn−1 Þ
value of the linguistic variable si ; and Suppðst Þ = scope of the
numerical value of the linguistic variable si .
þ ··· þ½hðx2Þ − hðx1 ÞgðH1Þ¼hðxnÞ½gðHn Þ − gðHn−1Þ
þ hðxn−1 Þ½gðH n−1 Þ − gðH n−2 Þþ ··· þhðx1 ÞgðH1 Þ
ð15Þ References
where H1 ¼ fx1 g; H2 ¼ fx1 ; x2 g; : : : :Hn ¼ fx1 ; x2 ; : : : xn g ¼ X. Baloi, D., and Price, A. D. F. (2003). “Modelling global risk factors affect-
Moreover, if λ ¼ 0 and g1 ¼ g2 ¼; : : : ; ¼ gn , then hðx1 Þ ≥ ing construction cost performance.” Int. J. Project Manage., 21(4),
261–269.
hðx2 Þ ≥; : : : ; hðxn Þ is not a necessary condition.
Banon, G. (1981). “Distinction between several subsets of fuzzy measures.”
The following is the interpretation of the fuzzy integral. Suppose
Fuzzy Sets Syst., 5(3), 291–305.
that an object is calculated from the point of view of a set of sources Baradan, S., and Usmen, M. (2006). “Comparative injury and fatality risk
X. Let hðxÞ ∈ ½0; 1 denote the decision for the object when source analysis of building trades.” J. Constr. Eng. Manage., 10.1061/(ASCE)
x ∈ X is considered and let g½ðxÞ denote the degree of impor- 0733-9364(2006)132:5(533), 533–539.
tance of this source. At this time, suppose the object is evaluated Benardos, A. G., and Kaliampakos, D. C. (2004). “A methodology for
using sources from A ⊆ X. It is logical to consider a quantity assessing geotechnical hazards for TBM tunnelling-illustrated by the
WðAÞ ¼ min hðxÞ as the best security decision that the object pro- Athens Metro Greece.” Int. J. Rock Mech. Mining Sci., 41(6), 987–999.
x∈A
vides and gðAÞ expresses the grade of importance of this subset of BIS (Bureau of Indian Standards). (2001). Handbook on construction
safety practices, BIS SP70, New Delhi, India.
sources. The value obtained from comparing these two quantities in
BIS (Bureau of Indian Standards). (2007). “Occupational health and safety
terms of the min operator is interpreted as the grade of agreement
management system–requirements with guidance for use.” BIS 18001,
between real possibilities, hðxÞ, and the expectations, g. Therefore, New Delhi, India.
fuzzy integration is interpreted as searching for the maximal grade Buckley, J. J. (1985). “Fuzzy hierarchical analysis.” Fuzzy Sets Syst., 17(3),
of agreement between objective evidence and the expectation. 233–247.
The properties of fuzzy integral are (Cao 2006) H Cao, N. (2006). “Supply-chain performance measurement in textile and
1. If hðxÞ ¼ c, for all x ∈ X, 0 ≤ c ≤ 1, then X HhðxÞ ∘ gð·Þ ¼ c; apparel industries.” Ph.D. dissertation, Hong Kong Polytechnic Univ.,
H h1 ðxÞ ≤ h2 ðxÞ, for all x ∈ X, then X h1 ðxÞ ∘ gð·Þ ≤
2. If Hong Kong.
X h2 ðxÞ ∘ gð·Þ; H H Carter, G., and Smith, S. (2006). “Safety hazard identification on construc-
3. If A ⊂ B, then A hðxÞ ∘ gð·Þ ≤ A hðxÞ ∘ gð·Þ; and tion projects.” J. Constr. Eng. Manage., 10.1061/(ASCE)0733-9364
4. HLet fAi :i ¼ 1; 2; : : : ng be a portion of the set x, then (2006)132:2(197), 197–205.
Chen, C. T. (2000). “Extension of TOPSIS for group decision-making
X hðxÞ ∘ gð·Þ ≥ maxðe1 ; : : : ; en Þ, where ei is the fuzzy integral
under fuzzy environment.” Fuzzy Sets Syst., 114(1), 1–9.
of h with respect to g over Ai. The interpretation of all these
Chen, S. J., Hwang, C. L. (1992). Fuzzy multi-attribute decision-making:
properties related to the fuzzy integral as an information fusion
Methods and applications, Springer, Berlin.
technique should be obvious.
Chou, S. Y., Chang, Y. H., and Shen, C. Y. (2008). “A fuzzy simple additive
The calculation of the fuzzy integral when X is a finite set is weighting system under group decision-making for facility location se-
easily given. Let X ¼ fx1 ; x2 ; : : : ; xn g be a finite set and let h:X → lection with objective/subjective attributes.” Eur. J. Oper. Res., 189(1),
½0; 1 be a function. Suppose hðx1 Þ ≥ hðx2 Þ ≥; : : : ; hðxn Þ (if not, X 132–145.
is rearranged so that this relation holds). Then, a fuzzy integral e Churchill, G. A. (1979). “A paradigm for development of better measures of
with respect to a fuzzy measure g over X can be calculated by marketing constructs.” J. Marketing Res., 16(1), 64–73.
n
e ¼ maxfmin½hðxi Þ; gðAi Þg where Ai ¼ fx1 ; : : : ; xi g. Dağdeviren, M., and Yüksel, I. (2008). “Developing a fuzzy analytic hier-
i¼1 archy process (AHP) model for behavior-based safety management.”
When g is λ-fuzzy measure, the value of gðAi Þ can be deter- Inf. Sci., 178(6), 1717–1733.
mined recursively as gðA1 Þ ¼ g½ðx1 Þ ¼ g1 Delbecq, A. L., Van de Ven, A. H., and Gustafson, D. H. (1975). Group
techniques for program planning, Scott, Foresman and Company,
gðAi Þ ¼ g½ðx1 ; : : : ; xi−1 ; xi Þ ¼ gi þ gðAi−1 Þ þ λgi gðAi−1 Þ Glenview, IL.
for 1 < i ≤ n Delgado, M., Herrera, F., Herrera-Viedma, F., and Martinez, L. (1998).
“Combining numerical and linguistic information in group decision
In this way, the calculation of the fuzzy integral with respect to a making.” Inf. Sci., 107(1–4), 177–194.
λ-fuzzy measure would only require the knowledge of the density Dharmapalan, V., Gambatese, J., Fradella, J., and Moghaddam Vahed, A.
(2015). “Quantification and assessment of safety risk in the design of
function, where the ith density, gi , is interpreted as the degree of
multistory buildings.” J. Constr. Eng. Manage., 10.1061/(ASCE)CO
importance of the source xi , for i ¼ 1; 2; : : : ; n.
.1943-7862.0000952, 04014090.
Dubois, D., and Prade, H. (2015). “Possibility theory and its applications:
Where do we stand?” Springer handbook of computational intelligence,
Appendix II. Numerical Linguistic Transformation
Springer, Berlin, 31–60.
Function El-Shahhat, A. M., Rosowsky, D. V., and Chen, W. F. (1993). “Construc-
tion safety of multistory concrete buildings.” Struct. J., 90(4), 335–341.
Let r ∈ ½0,1 be a numerical value, and let si be a label verify-
Feng, Y. (2013). “Effect of safety investments on safety performance of
ing that building projects.” Ph.D. thesis, National Univ. of Singapore (NUS),
hðr; si Þ ¼ min½hðr; st Þj ∀ st ∈ S Singapore.
Fung, I. W. H., Tam, V. W. Y., and Lu, L. H. L. (2010). “Developing a risk
with assessment model for construction safety.” Int. J. Project Manage.,
( 28(6), 593–600.
z if r ∈= Suppðst Þ Gürcanli, G. E., and Müngen, U. (2009). “An occupational safety risk
hðr; si Þ ¼ Pz analysis method at construction sites using fuzzy sets.” Int. J. Ind.
j¼1 ½r − Gj ðst Þ2 if r ∈ Suppðst Þ Ergon., 39(2), 371–387.

Hair, J. F., Anderson, R. E., Tatham, R. L., and Black, W. C. (1995). Multi- Patel, D. A., and Jha, K. N. (2015). “Neural network model for the pre-
variate data analysis with readings, 4th Ed., Prentice Hall, Englewood diction of safe work behavior in construction projects.” J. Constr. Eng.
Cliffs, NJ. Manage., 10.1061/(ASCE)CO.1943-7862.0000922, 04014066.
Hallowell, M. R., and Gambatese, J. A. (2010). “Qualitative research: Patel, D. A. (2015). “Estimating the number of fatal accidents and inves-
Application of the Delphi method to CEM research.” J. Constr. Eng. tigating the determinants of safety performance in Indian construction.”
Manage., 10.1061/(ASCE)CO.1943-7862.0000137, 99–107. Ph.D. thesis, Indian Institute of Technology Delhi, New Delhi, India.
Hastak, M., and Halpin, D. W. (2000). “Assessment of life-cycle benefit Pinto, A. (2014). “QRAM a qualitative occupational safety risk assessment
cost of composites in construction.” J. Compos. Constr., 10.1061/ model for the construction industry that incorporates uncertainties by
(ASCE)1090-0268(2000)4:3(103), 103–111. the use of fuzzy sets.” Saf. Sci., 10.1016/j.ssci.2013.10.019, 57–76.
Imriyas, K., Low, S., Teo, A., and Chan, S. (2008). “Premium-rating model Rao, R. V. (2007). Decision making in the manufacturing environment:
for workers’ compensation insurance in construction.” J. Constr. Eng. Using Graph theory and fuzzy multiple attribute decision, Springer,
Manage., 10.1061/(ASCE)0733-9364(2008)134:8(601), 601–617. London.
Rubio-Romero, J., Rubio, M., and García-Hernández, C. (2013). “Analysis
Imriyas, K., Pheng, L. S., and Lin, T. A. (2007). “A decision support system
of construction equipment safety in temporary work at height.” J.
for predicting accident risks in building projects.” Archit. Sci. Rev.,
Constr. Eng. Manage., 10.1061/(ASCE)CO.1943-7862.0000567, 9–14.
50(2), 149–162.
Saaty, T. L. (1980). The analytic hierarchy process, McGraw Hill,
Johnson, H., Singh, A., and Young, R. (1998). “Fall protection analysis for
New York.
workers on residential roofs.” J. Constr. Eng. Manage., 10.1061/ Saaty, T. L. (1986). “Axiomatic foundation of the analytic hierarchy
(ASCE)0733-9364(1998)124:5(418), 418–428. process.” Manage. Sci., 32(7), 841–855.
Jordan, E., and Javernick-Will, A. (2013). “Indicators of community re- Sachs, T., and Tiong, R. L. (2009). “Quantifying qualitative information
covery: Content analysis and Delphi approach.” Nat. Hazards Rev., on risks: Development of the QQIR method.” J. Constr. Eng. Manage.,
10.1061/(ASCE)NH.1527-6996.0000087, 21–28. 10.1061/(ASCE)0733-9364(2009)135:1(56), 56–71.
Laukkanen, T. (1999). “Construction work and education: Occupational Shafer, G. (1976). A mathematical theory of evidence, Princeton University
health and safety reviewed.” Constr. Manage. Econ., 17(1), 53–62. Press, Princeton, NJ.
Lee, D. (2006). “A generalized approach for analyzing transportation user Shapira, A., and Goldenberg, M. (2005). “AHP-based equipment selection
perception using fuzzy sets.” Ph.D. thesis, Pennsylvania State Univ., model for construction projects.” J. Constr. Eng. Manage., 10.1061/
State College, PA. (ASCE)0733-9364(2005)131:12(1263), 1263–1273.
Lee, S. (2014). “Determination of priority weights under multi-attribute Skibniewski, M., and Chao, L. C. (1992). “Evaluation of advanced con-
decision-making situations: AHP versus fuzzy AHP.” J. Constr. Eng. struction technology with AHP method.” J. Constr. Eng. Manage.,
Manage., 10.1061/(ASCE)CO.1943-7862.0000897, 05014015. 10.1061/(ASCE)0733-9364(1992)118:3(577), 577–593.
Liu, H. T., and Tsai, Y. L. (2012). “A fuzzy risk assessment approach for Sugeno, M. (1974). “Theory of fuzzy integrals and its applications.” Ph.D.
occupational hazards in the construction industry.” Saf. Sci., 50(4), dissertation, Tokyo Institute of Technology, Tokyo.
1067–1078. Tam, C. M., Tong, T. K. L., Chiu, G. C., and Fung, I. W. H. (2002). “Non-
Lott, R. J. (2005). “Controlling workers’ compensation cost a risk man- structural fuzzy decision support system for evaluation of construction
safety management system.” Int. J. Project Manage., 20(4), 303–313.
agement program.”〈http://www.healthconsultantsusa.com/Risk_Mgmt
Tanaka, H., Guo, P., and Türksen, I. B. (2000). “Portfolio selection based
_Program_Controlling_Worker_Comp_Costs.pdf〉 (Oct. 31, 2011).
on fuzzy probabilities and possibility distributions.” Fuzzy Sets Syst.,
MATLAB R2010b [Computer software]. MathWorks, Natick, MA.
111(3), 387–397.
Mohamed, S., and McCowan, A. K. (2001). “Modelling project investment Wu, X., Liu, H., Zhang, L., Skibniewski, M. J., Deng, Q., and Teng, J.
decisions under uncertainties using possibility theory.” Int. J. Project (2015). “A dynamic Bayesian network based approach to safety deci-
Manage., 19(4), 231–241. sion support in tunnel construction.” Reliab. Eng. Syst. Saf., 134(0),
Neitzel, R. L., Seixas, N. S., and Ren, K. K. (2001). “A review of crane 157–168.
safety in the construction industry.” Appl. Occup. Environ. Hyg., Zadeh, L. A. (1978). “Fuzzy sets as a basis for a theory of possibility.”
16(12), 1106–1117. Fuzzy Sets Syst., 1(1), 3–28.
NICMAR (National Institute of Construction Management and Research). Zadeh, L. A. (1995). “Discussion: Probability theory and fuzzy logic
(1998). Safety management in the construction industry-A manual for are complementary rather than competitive.” Technometrics, 37(3),
project managers, 2nd Ed., Mumbai, India. 271–276.

Developing A Process To Evaluate Construction Project Safety Hazard Index Using The Possibility Approach in India

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Developing A Process To Evaluate Construction Project Safety Hazard Index Using The Possibility Approach in India

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Developing A Process To Evaluate Construction Project Safety Hazard Index Using The Possibility Approach in India

Uploaded by

Copyright:

Available Formats

Developing a Process to Evaluate Construction

Project Safety Hazard Index Using the

© ASCE 04016081-1 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

© ASCE 04016081-2 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

Fig. 1. Framework for evaluation of project hazard level

© ASCE 04016081-3 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

Fig. 2. Research methodology

Table 1. Guidelines for the Delphi Procedure

© ASCE 04016081-4 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

in multicriteria decision making. In brief, the λ-fuzzy measure and Y

© ASCE 04016081-5 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

through an operation of WSM that integrates the score of each

where M = fuzzy positive reciprocal matrix and Mg g

© ASCE 04016081-6 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

© ASCE 04016081-7 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

Table 5. Statistical Parameters of Hazardous Trades

© ASCE 04016081-8 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

b. Height of erection work

© ASCE 04016081-9 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

Nature of site vicinity (nearby structures, overhead cables, etc.)

© ASCE 04016081-10 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

computed. The fuzzy linguistic values of the tunneling configura-

© ASCE 04016081-11 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

consists of 30 construction project sites. The authors visited and

normality and linearity of data, Pearson correlation is computed at

mance can be studied.

attributes to evaluate a project’s hazard level because in most

2012). Risk is the product of likelihood of an accident and its

model where every data point is known (Pinto 2014). However,

Müngen (2009) used fuzzy logic, Mamdani fuzzy inference, and

nonlinear mapping. However, it adds complexity into the model

nonadditive nature of the subfactors or attributes lying at lower lev-

On the other hand in this study, the 10 hazardous trades essential

in the hazard level of construction sites. The use of the linguistic

proved to be a versatile and robust way of estimating and integrat-

Summary and Conclusions

This study aimed to develop a framework to process for evaluation

mined using the Delphi method. The possibility approach was

past records and similar kinds of research in the existing literature.

nature. To design the weight matrix in FAHP and evaluate

attributes, data were collected through questionnaire surveys from

© ASCE 04016081-12 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2017, 143(1): -1--1

sets up a comprehensive map of quantification and integration of [

¼ sup min½α; gðA ∩ Fα Þ

© ASCE 04016081-13 J. Constr. Eng. Manage.