Proposal of a risk-factor-based analytical approach for integrating occupational health and safety into project risk evaluation

Accident Analysis and Prevention 48 (2012) 223–234
Contents lists available at ScienceDirect
Accident Analysis and Prevention
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / a a p
Proposal of a risk-factor-based analytical approach for integrating occupational
health and safety into project risk evaluation
Adel Badri a,∗, Sylvie Nadeau a , André Gbodossou b
a Mechanical Engineering Department, University of Quebec, École de technologie supérieure, 1100 Notre Dame West, Montreal (Quebec) H3C 1K3, Canada
b Unit of Education and Research in Management Sciences, University of Quebec Abitibi-Témiscamingue, Rouyn-Noranda (Quebec) J9X 5E4, Canada
a r t i c l e i n f o
Article history:
Received 18 October 2010
Received in revised form 9 April 2011
Accepted 8 May 2011
Keywords:
Occupational health and safety (OHS)
Project management
Analytical hierarchy process (AHP)
Risk factor concentration
Risk assessment
Industrial project
a b s t r a c t
Excluding occupational health and safety (OHS) from project management is no longer acceptable.
Numerous industrial accidents have exposed the ineffectiveness of conventional risk evaluation meth-
ods as well as negligence of risk factors having major impact on the health and safety of workers and
nearby residents. Lack of reliable and complete evaluations from the beginning of a project generates bad
decisions that could end up threatening the very existence of an organization.
This article supports a systematic approach to the evaluation of OHS risks and proposes a new proce-
dure based on the number of risk factors identified and their relative significance. A new concept called
risk factor concentration along with weighting of risk factor categories as contributors to undesirable
events are used in the analytical hierarchy process multi-criteria comparison model with Expert Choice ©
software.
A case study is used to illustrate the various steps of the risk evaluation approach and the quick and
simple integration of OHS at an early stage of a project. The approach allows continual reassessment of
criteria over the course of the project or when new data are acquired. It was thus possible to differentiate
the OHS risks from the risk of drop in quality in the case of the factory expansion project.
© 2011 Elsevier Ltd. All rights reserved.
1. Introduction
Industrial accidents continue to cause human suffering, capital
losses, environmental destruction and social problems (Duijm et al.,
2008; Kartam, 1997; Li et al., 2009; Shikdar and Sawaqed, 2003). In
recent years, accidents in construction and industry have occurred
in spite of rigorous management of projects and robust occupa-
tional health and safety (OHS) management systems (Makin and
Winder, 2008) in all phases of project lifecycle (Li et al., 2009).
The explosion of a power plant in the start-up phase while
testing a gas line in a populated region (43,000 inhabitants) of Con-
necticut (USA) on February 7, 2010 is reminiscent of a series of
similar industrial accidents over the decades in terms of gravity and
consequences. In most cases, inquiry into the causes of the accident
revealed failure in the identification and evaluation of the impend-
ing risks, placing at peril the health and safety of human beings
on site and in the surrounding areas. This was the case notably at
Bhopal (1984) and at Chernobyl (1986).
In general, risk is evaluated in terms of its consequences with
respect to project performance and rarely in terms of human suf-
fering. Smallwood (2004) confirmed that quality, planning and
∗ Corresponding author. Tel.: [phone]x7322.
E-mail addresses: [e-mail], [e-mail] (A. Badri).
costs are the parameters given the greatest consideration. This is
reflected in the decision to install many high-risk production plants
near or in densely populated areas (e.g. the AZF chemical plant in
Toulouse, France; the now closed Sigma-Lamaque mine in Val d’Or,
Quebec). In Quebec, high-risk installations still get the go-ahead in
spite of the efforts by the Environmental Public Hearings Office to
provide transparent information and to consult citizens.
The aim of this paper is to present a new systematic approach to
the evaluation of OHS risks and proposes a new procedure based on
the number of risk factors identified and their relative significance.
This approach is able to overcome the difficulties of current tools
in the manufacturing industry. The proposed approach is based on
known techniques and tools, such as multi-criteria analysis tech-
niques (e.g. analytic hierarchy process), expert judgment and the
analysis of accidents and incidents. The analytic hierarchy process is
selected to minimize the inconsistencies in expert judgments (Fera
and Macchiaroli, 2009) and to support approaches that use mixed
qualitative–quantitative assessment data (Chao et al., 2005).
This document is structured as follows. In Section 2, we begin
by discussing the relevant tools and approaches used to manage
project risk in different industrial sectors. We also give an overview
of the use of qualitative and quantitative tools in various indus-
tries. Section 3 presents the methodology, including the conceptual
model of the systematic approach to the evaluation of OHS risks.
Given its importance in the approach proposed, the AHP method
0001-4575/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aap.2011.05.009

224 A. Badri et al. / Accident Analysis and Prevention 48 (2012) 223–234
is outlined in Section 4. The proposed approach is then described
in detail in Section 5 and a case study of a factory extension is
presented to test the proposed approach. Section 6 follows with
discussion and suggests possible directions for future research and
a conclusion is provided in Section 7.
2. Literature review
Industrial work is risky in many economic sectors, in particu-
lar the construction industry (Fung et al., 2010), chemical plants
(Vernero and Montanari, 2010), nuclear power plants (Young,
2005) and the mining industry (Hermanus, 2007). Safety prob-
lems can result from any of several combinations of causes, which
vary from one industry to another. The high level of risk in the
construction industry is explained by the nature and characteris-
tics of construction work, low educational level of workers, lack
of safety culture and communication problems (e.g. Fung et al.,
2010; Gambatese, 2000b). In the mining sector, increasing num-
bers of subcontractors working in mines, the emergence of new
mining ventures and recognition of small-scale mining pose new
challenges to the practice of risk control (Hermanus, 2007).
The most effective way to improve OHS performance is to iden-
tify and eliminate hazards at the source (Glickman and White,
2007). Risk identification and assessment thus become primary
tasks that are part of hazard prevention (Manuelle, 2005). Risk anal-
ysis is the foundation of the risk management process (Fung et al.,
2010; Liu and Guo, 2009) and presents several challenges (Hagigi
and Sivakumar, 2009).
OHS has not always been a preoccupation of process engineers
(Hassim and Hurme, 2010). The motivations for integrating OHS
risk management into engineering have been discussed recently.
These include legislation (Gambatese, 2000b; Zachariassen and
Knudsen, 2002), awareness of the importance of protecting work-
ers (Gambatese, 2000a) and in some cases perceived potential to
increase profitability and remain competitive (Sonnemans et al.,
2002).
Industry has attempted to adapt engineering tools and methods
to the assessment of OHS risks. These include quality management
tools (e.g. failure methods and critical analysis (FMECA), “What
If” analysis and check lists) and other industrial safety approaches
(e.g. fault tree analysis (FTA), event tree (ET) and human reliability
analysis (HRA)). Several authors have developed OHS risk reduc-
tion tools and models used in conjunction with historical data and
shop floor know-how (e.g. Cameron and Hare, 2008; Ciribini and
Rigamonti, 1999; Fung et al., 2010; Gibb et al., 2006; Hare et al.,
2006; Kartam, 1997; Saurin et al., 2004; Suraji et al., 2001). It is
important to note that the abovementioned tools are used alone
rather than integrated into other types of risk management by an
organization.
Quantitative methods of risk management are widely used in
many industrial fields (Fera and Macchiaroli, 2009), for example
the aerospace and nuclear industries (e.g. Skelton, 2002). These
methods generally use equipment and software to analyze data.
Quantitative methods are generally expensive and require special-
ized analysts (Restrepo, 1995). One of the best-known methods is
that of the safety review and hazard and operability study (HAZOP)
(Calixto, 2007). This method allows assessment of complex situa-
tions based on knowledge of several key parameters of a system.
In many industrial fields, the data and information used to assess
risk are imprecise and incomplete (Ferdous et al., 2009). Quantita-
tive approaches do not give reliable results when data are lacking
(Pinto et al., 2010). Acquiring useful information using quantitative
risk assessment based on probabilistic models is not yet possible
(Jabbari Gharabagh et al., 2009). In the petrochemical industry,
Jabbari Gharabagh et al. (2009) attributed the current difficulties
in risk assessment to the complexity of the current quantitative
methods. These problems are more significant in the design stage
of industrial projects (e.g. Pinto et al., 2010).
Pinto et al. (2010) proposed a qualitative model for health
and safety risk assessment based on available data and using a
fuzzy logic approach. They concluded that qualitative approaches
for human-centered problems are flexible enough to assess risk.
Another method worth mentioning was developed by Hassim and
Hurme (2010) for assessing the health risks of a chemical pro-
cess during the design phase. The method takes into account both
the hazard associated with the presence of the chemicals and the
potential for the exposure of workers to them. An “Inherent Occu-
pational Health Index” has also been proposed to conduct the risk
evaluation early in the design phase. Jabbari Gharabagh et al. (2009)
concluded that the use of historical data is not only important in risk
management, but is also helpful in risk evaluation as an indicator
of acceptable risk criteria.
Neglecting the consideration of human factors in risk analysis
is due in part to the difficulty of quantifying many of them (e.g.
Human risk-taking behavior in Kotani et al., 2007). In addition,
human behavior cannot be predicted from analysis of accident and
incident histories alone. Evaluation based solely on historical infor-
mation always runs into difficulties in meeting the challenge of the
proactive treatment of risks.
It is always more effective and profitable to integrate risk evalu-
ation beginning at the project design phase (Charvolin and Duchet,
2006). Complete and accurate evaluation will contribute to reduc-
ing risks as well as justify monitoring of workers and residents of
the surrounding community in the event of damage to the instal-
lation, whether caused by an industrial accident or a natural event
(Pérusse and Bernier, 2009). Determining the risks and measures
for dealing with them before setting the project in motion is with-
out question the wisest course to follow (Gray and Larson, 2006).
Starting from the need to create an appropriate and effective
approach that integrates the management of all project risks in the
manufacturing sector, our paper explores the possibility of creating
such a model for industrial projects using an approach based on
mixed techniques.
The proposed approach allows quick prioritizing of identified
risks and allows evaluators to identify additional potential causes
of undesirable events without nullifying the previous risk element
compilation effort. The simplicity of the procedure should facilitate
its use in small and medium-sized businesses without requiring a
major investment.
3. Methodology
Based on the literature (Aubert and Bernard, 2004; Curaba et al.,
2009; Freivalds, 1987; Henderson and Dutta, 1992) and on con-
tinuous risk management standards (Dorofee, 1996), this paper
proposes a conceptual model for integrating occupational health
and safety into project risk evaluation based on multi-criteria com-
parison (AHP). We have considered a model of risk composed of
three elements detailed below and the conventional steps of risk
management.
In order to propose a conceptual framework for identifying and
assessing risks, we began by tracing the elements of risks that are
used for the identification steps. Once the elements of risk are iden-
tified, the causality links form the basis of the evaluation and the
control steps.
Our analysis is based on a model of risk composed of three prin-
cipal elements (Fig. 1), namely the risk factors, the undesirable
event, and the impact of the undesirable event. In order to control
risk, all of the elements must be identified and the various causal

A. Badri et al. / Accident Analysis and Prevention 48 (2012) 223–234 225
Fig. 1. Modelling of risk and its influence.
links likely to appear in a field or area of study must be clarified as
well as their mechanisms and the conditions that trigger them.
It should be noted that the project internal environment is made
up of controllable variables such as the effectiveness of health
and safety measures. The variables of the external environment
(e.g. weather-related) are always the most difficult to control or
modify.
The proposed approach is based on a risk factor approach (Fig. 2).
This is an original approach to risk evaluation, since it is based on
a novel parameter expressed as a fraction and representing the
presence or likely appearance of the risk factors that trigger an
undesirable event, or more specifically the direct influence of the
number of risk factors present on the probability of occurrence.
This new concept is called the “risk factor concentration”. When
this concentration increases, there is a greater chance of triggering
the associated undesirable event.
Aubert and Bernard (2004) present a similar approach without
specifying that the impact of an undesirable event may include sev-
eral types of loss. The causality links are identified by the evaluators
and determine how the potential impact of a risk will be evaluated.
Each link (i) between a factor, an event and an impact thus defines
a possible route of concretization of a risk as an event having a
negative impact.
4. The analytical hierarchy process (AHP)
The AHP (Saaty, 2000) method is a structured multi-attribute
decision method used in complex decision making and is the most
widely used of the multi-criteria comparison methods. Developed
in the USA by Saaty in the 1970s (Simei et al., 2009), this method
is based on three fundamental principles: decomposition of the
structure, comparison of judgments and hierarchical composition
(or synthesis) of priorities. AHP is applicable to decision situations
involving subjective expert judgments and uses both qualitative
and quantitative data (De Steiguer et al., 2003). This method cre-
ates a priority index for each expert decision or judgment. AHP
summarizes these judgments by ensuring their consistency.
The proposed approach involves the AHP method for the paired
comparison of the risk factors, which was carried out using the deci-
sion aid software Expert Choice© . The AHP method is used in project
management as a decision aid in order to choose a project on the
basis of company objectives. Al-Harbi (2001) discussed this method
in the context of the pre-qualification of construction contractors.
In the OHS field, attempts to use AHP began in the con-
text of ergonomic analysis done by Henderson and Dutta (1992)
and the comparison of ergonomic standards by Freivalds (1987).(i) (i)Category of factors F 1Category of factors F 2Category of factors F nF1.1F1.2F1.xF2.1F2.2F2.yFn.1Fn.2Fn.zEvent E 1Event E 2Event E pImpact I1Impact I2Impact Iq
Fig. 2. The links in a risk factors approach to risk analysis; example inspired from Aubert and Bernard (2004).

226 A. Badri et al. / Accident Analysis and Prevention 48 (2012) 223–234
Henderson and Dutta (1992) compared NIOSH recommendations
with those of the ECSC for the two-handed handling of loads in
the sagittal plane. In this study, 11 risk factors were compared
using the AHP model. These factors, namely frequency, distance,
height, dimensions, load shape, position of the load center of grav-
ity, anthropometric dimensions, gender and age of the individual
and limited biomechanical and physiological criteria, were pro-
posed in a previous study by Freivalds (1987). Using AHP, Freivalds
(1987) showed discrepancies between NIOSH and ECSC standards,
which were attributed to differences in the respective equations,
hypotheses and concepts.
Padma and Balasubramanie (2008) used AHP to develop a deci-
sion aid system that draws on a knowledge base in order to rank risk
factors associated with the occurrence of musculoskeletal problems
in the shoulder and neck. Another system using AHP to compare risk
factors associated with human error and with the causes of acci-
dents in the maritime transport sector was developed in a study
by Zhang et al. (2009). Topacan et al. (2009) used AHP to evaluate a
health information system with the aim of investigating the factors
that influence user preferences in the selection of health services.
Fera and Macchiaroli (2009) have selected AHP for their model of
industrial risk assessment to identify major events and validate the
actions taken.
In ergonomics research, AHP has been described as a reliable
method for comparing risk factors, evaluating risks, defining prior-
ities, allocating resources and measuring performance (Henderson
and Dutta, 1992). The use of AHP to analyze human factors should
make the hierarchical model more clear, simple and practical
(Zhang et al., 2009) and should also allow more structured dis-
cussion and easier examination of relevant information (Larson
and Forman, 2007). AHP reduces the inconsistency of expert judg-
ments and appears acceptable in terms of reliability (Fera and
Macchiaroli, 2009). This multi-criteria method allows incorporat-
ing both objective and subjective considerations into the decision
process (Forman and Selly, 2002).
In conclusion, the feature of combining both quantitative and
qualitative data and controlling the consistency of expert judg-
ments makes AHP the most applicable to the proposed approach.
We will provide objective judgments and reliable prioritization of
risks.
4.1. The theoretical background of AHP (Nguyen, 2009):
Given n alternatives {A1 , A2 , . . ., An} from which a selection is to
be made, the expert attributes a numerical scale aij from the scale
of binary combinations (Table A.1 in Appendix A) to each pair of
alternatives (Ai, Aj). The term aijk expresses the individual prefer-
ence of expert k regarding alternative Ai compared to alternative
Aj.
Once the overall expert judgments are created and computed
using the geometrical mean (1), they are inserted into the compar-
ison matrix D (2):
aij = n
√aij1, aij2, . . . , aijn (1)
D =
⎡
⎢
⎣
a11 a12 . . . a1n
a21 a22 . . . a2n
. . . . . . . . . . . .
a n1 an2 . . . ann
⎤
⎥
⎦ (2)
Matrix D is a comparison matrix with inconsistent judgments and
has the following properties:
aij > 0; aij = 1
aji
∀ i where j = 1, 2, . . . , n (3)
Matrix D is considered consistent when its elements meet condi-
tions (4) and (5):
aij · ajk = aik; ∀ i, j, k where i, j, k = 1, 2, . . . , n (4)
aij · aji = 1 where i, j = 1, 2, . . . , n (5)
The ordering of alternatives is taken as a result of the approximation
of comparison matrix D using matrix P:
P =
⎡
⎢
⎣
p11 p12 . . . p1n
p21 p22 . . . p2n
. . . . . . . . . . . .
p n1 pn2 . . . pnn
⎤
⎥
⎦ (6)
The elements of which are consistent judgments presented in the
form of weight ratios among alternatives:
pij = p i
pj
where i, j = 1, 2, . . . , n (7)
pi signifies the weights of the alternatives of the order vector p:
p = (p1, p2, . . . , pn)T (8)
We obtain the standardized order vector after the arithmetic nor-
malization:
p∗ = (p∗
1, p∗
2, . . . , p∗
n)T (9)
where
p∗
i = pi
∑n
i=0pi
(10)
Saaty (2000) uses the maximum eigenvalue method to approxi-
mate the judgment matrices:
D.p = maxp (11)
where max is the maximum eigenvalue of matrix D.
For reliable comparison, it is important to note that the incon-
sistency of the comparison matrix D must be less then 10%. This
condition means that the number of times that condition (4) is not
met must be below 10%.
5. Results and analysis
5.1. The proposed risk-factor-based analytical approach
The proposed approach is divided into three phases and each
phase is divided into steps. This approach outlines all phases of risk
management including: (1) risk identification; (2) risk assessment
and (3) actions.
The approach uses several methods and tools such as systematic
observations, interviews, multi-criteria analysis (AHP), analysis of
accidents and incidents and the new concept of risk factor concen-
tration. In Table 1, we report the tools and methods used for each
phase and step.
The model is based on teamwork and knowledge of multi-
criteria analysis techniques. The purpose of this model is to
integrate OHS risk with operational risk without creating a con-
flict and without complicating the process for the risk management
team. It should be noted that multi-criteria analysis is used partly
to compare the risk factors, not to compare the risks identified.
Like any approach to risk management, the model gives appro-
priate consideration to the phase of identifying risk elements (risk
factors, undesirable events and impact of undesirable events). The
risk assessment phase uses multi-criteria analysis, expert judgment
and the new concept of risk factor concentration. The analysis is
made according to the causal links between elements of identified
risks. The action phase is based on risk prioritization. This step can