Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Contents lists available at ScienceDirect

Journal of Loss Prevention in the Process Industries

journal homepage: [Link]

The integration of HAZOP study with risk-matrix and the

analytical-hierarchy process for identifying critical control-points and
prioritizing risks in industry – A case study
Panagiotis K. Marhavilas *, Michail Filippidis, Georgios K. Koulinas, Dimitrios E. Koulouriotis
Dep. of Production & Management Engineering, Democritus Univ. of Thrace, Vas. Sofias 12 St., 67132 Xanthi, Greece

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

Keywords: The integration of Hazard and Operability (HAZOP) study with the Decision-Matrix Risk Assessment (DMRA)
Sour crude-oil technique and the Analytical Hierarchy Process (AHP) as well, is proposed as a new framework for identifying
Hazard identification critical points and potential hazards and also prioritizing risks, in industry. Therefore, the objective of this article
Risk analysis
is the development and application of the extended HAZOP-DMRA-AHP approach (E-HAZOP) in process in­
HAZOP study
Operability study
dustries. Initially, the application of the conventional HAZOP study is performed (as a case study) in a sour crude-
Risk-matrix oil processing-plant in order to identify the possible fault causes of abnormal conditions (deviations) in the plant.
DMRA With the results of the HAZOP study on the one hand, and the usage of the DMRA risk assessment technique for
Process safety assessing the risks, on the other side, the typical AHP process is then introduced, in order to extend HAZOP
Analytical hierarchy process analysis by prioritizing the risks in the worksite of the plant. The proposed E-HAZOP framework can provide a
AHP trustworthy basis to boost process safety and improve occupational health and safety. Moreover, it could be a
Multicriteria decision making useful tool for the decision makers to estimate the emergency of investing constrained budget in actions pre­
venting specific kind of deviations.

1. Introduction the requirements are as follows: (i) to identify, as complete as possible,

the hazards that arise from all the activities of the facility (ii) to make the
1.1. Basics of risk, safety, hazard identification and risk ranking hazard identification tailored to the equipment, systems and operations
of the facility in which the operation will take place (Vinnem, 1998).
Tolerable risk is succeeded by the procedures of risk assessment Besides, hazard identification techniques can be divided into four cat­
(consisted of risk-analysis and risk-evaluation subprocesses) and risk egories depending on the area in which they are predominantly applied:
reduction (ISO/IEC, 1999; 2009), while risk management can be (i) process hazards identification, (ii) hardware hazards identification,
considered as the entire scheme that includes both qualitative and (iii) control hazards identification and (iv) human hazards
quantitative analysis techniques, and can be separated into these two identification.
procedures. The final subphase is characterized by taking appropriate There has been a development of numerous hazard identification
measures in order to be controlled and/or reduced the risk. A significant techniques, of which some are more established than others, and the
part of risk analysis subphase, and as a consequence of the total process appropriateness of these techniques to identify hazards in specific pha­
of risk management, is the hazard identification process (IEC, 1995; ses of a project is not universal. Furthermore, as it is clear from the
Suokas, 1988; Olsson, 1999; Lee, 2006; Marhavilas et al., 2011a, 2011b; scientific literature, there is a great choice when it comes to hazard
Marhavilas and Koulouriotis, 2012; Marhavilas, 2015; Jain et al., 2018a, identification techniques, however the chemical and process industry
2018b; Aziz et al., 2019). has, in practice, adopted and used extensively a few of them, with the
The purpose of hazard identification is to identify, at least in theory, remarkable example of HAZOP (Andersen and Mostue, 2012; Marhavi­
all possible hazards that may occur during normal operations, as well as, las et al., 2011a; Othman et al., 2016).
during specific operations carried out on the installation. Such activities There are so different types of hazards existed in various installations
include start up and shut down of the installation, or maintenance. Thus, that a HAZOP-study may definitely take into account (for example toxic

* Corresponding author.
E-mail address: marhavil@[Link] (P.K. Marhavilas).

[Link]
Received 17 May 2019; Received in revised form 20 August 2019; Accepted 8 October 2019
Available online 15 October 2019
0950-4230/© 2019 Elsevier Ltd. All rights reserved.
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

release, fire, explosion, chemical explosions, physical explosions, reac­ processing plant, which is particularly very significant because it com­
tivity, etc). Some of the possible hazards, which need to be identified bines many hazards, such as flammable liquids and gases and also toxic
especially in process installations, are associated with the management gases. Finally, our main aim is to make public that the suggested E-
of highly hazardous materials and are presented by CCPS with a HAZOP technique could be a handy tool for the decision makers to
guidelines-list (CCPS, 2001, 2008). The HAZOP study is a structured improve occupational and process safety.
method of identifying potential hazards and problematic operation of a
process by examining the effects of various deviations from the design 1.3. Structure of the article
conditions. Chemical processes involve a wide array of interconnected
equipment, instrumentation, utilities and other devices to achieve the The paper is organized by the following sections:
installation design intent, for example the stabilization of crude-oil.
Equipment grouping into a node can be difficult because of the inter­ (1) introduction, which focuses on (i) elements of risk, safety, hazard
connectivity that is inherent in the process industry. identification and risk ranking, (ii) the objectives of the work
HAZOP cannot be conducted at the conceptual phase of a project (2) literature review
because of the lacking of any detailed Piping and Instrumentation Dia­ (3) technical presentation of a sour crude-oil process industry, which
grams (P&IDs) at that point. A hazard review must be performed to is used for the conventional and the extended HAZOP-study
identify potential hazards in the process. An evaluation of the severity of application,
those hazards may follow. This review should be carried out as early as (4) presentation of the methodological framework of an extended-
possible, so that there will be an adequate time-frame to implement HAZOP study (E-HAZOP), implemented by the combination of
fundamental modifications of the design concept, following principles of HAZOP, DMRA and AHP,
inherent safe design. HAZOP originally was considered suitable for new (5) application of the new proposed E-HAZOP method (as a case
installations only, but in a short period of time it came to be recognized study) in a sour crude-oil processing-plant,
as a useful tool for also existing facilities. This was due to the intro­ (6) discussion of main points, and
duction of new sets of regulations, like the SEVESO directive, that (7) conclusions
required the periodic hazard analysis of an installation, and partly
because of the increasing safety awareness of the industry. The potential 2. Literature review
hazards in an existing plant cannot be calculated or identified at the
beginning of its operations because the processed involved can change HAZOP studies were introduced by the “Critical Examination”
dramatically over time. Also some of the modifications that are being technique of the Imperial Chemical Industries (ICI) in the mid-1960s. A
made in older facilities may not have been analyzed for the possible decade later, HAZOP was formally published as a disciplined procedure
hazards that may entail for the process. These modifications could to identify deviations from the design intent (Lawley, 1974, 1976). In
compromise the safety of the original design but as process installations 1977, the Chemical Industries Association (CIA) in the UK published the
are a dynamic system, they are necessary. There many different kinds of first guideline to HAZOP (CIA, 1977).
changes that take place in the course of a lifetime of an installation. It was proposed as a technique to be used in the process industries to
There are vessels or pumps added, alterations of the process parameters identify hazard and plan safety measures. The objectives should always
to increase productivity, reduce cost or conserve energy. It would not be include the identification of hazards in the installation and the potential
cost effective to perform a HAZOP study for each and every change that operating problems. The hazards can be environmental or health and
may be implemented in an installation. safety hazards that could originate or having an impact both on and off
To continue, risk ranking is a useful tool in HAZOP analysis which site (Qureshi, 1988).
allows a prioritization of the recommendations made at the end of the The concept of a Hazard and Operability (HAZOP) study was
process. Establishments often develop corporate risk-ranking schemes to developed with the aim of identifying possible hazards which are pre­
make sure that there is a consistency in the results of the HAZOP studies. sent in installations that operate using or producing highly hazardous
Risk ranking is achieved by performing qualitative estimations of the materials. The main concern was to eliminate any source that could
severity and likelihoods of hazardous scenarios and combining them to possibly lead to a major accident, like a toxic release, explosions and
create risk estimates in matrix, table or grid form. fires (Swann and Preston, 1995). Nonetheless, over the years, HAZOP’s
Furthermore, Analytic hierarchy process (AHP), introduced by Saaty application has been extended to numerous other types of facilities
(1990), is one of the most popular and efficient multi-criteria decision because of its ability not only to identify hazards, but also operational
making (MCDM) methods. Its hierarchical and systematic methodology deviations from the desired state. So, HAZOP has been adopted for
gives it a popular nature to solve MCDM problems, and so that it has medical diagnostic systems (Chudleigh, 1994), road-safety measures
been successfully incorporated in various fields, from education, to (Jagtman et al., 2005) and hazard analysis in photovoltaic facilities
military purposes. In safety, AHP has been applied in various circum­ (Ftenakis and Trammell, 2003) among others. The diversity of fields that
stances, but its assimilation to HAZOP studies is rare (Othman et al., HAZOP has been implemented shows that it has come to be considered a
2016). powerful technique to improve many kinds of systems.
The process of HAZOP requires the concerted efforts of a team of
1.2. Objectives people in management system and is divided into three phases: (i)
Definition and Preparation, (ii) Organization, and (iii) Execution and
This work concentrates on the expansion of the HAZOP method, Documentation. The first phase requires the definition of the purpose,
which is one of the most widely used in industry, with other two valu­ scope and objectives of the study. Also a team leader and the team that
able techniques, the DMRA and AHP, taking into account that their will conduct the study should be assembled, after taking into consider­
simultaneous incorporation to HAZOP, is nonexistent in the scientific ation the skills and knowledge required for the specific installation and
literature. Consequently, the objectives of this article are the compre­ the phase in which the survey will take place. In the second phase two
hensive presentation of (i) the conventional HAZOP study, the DMRA key commitments should be addressed: (i) to divide the process into
technique, and the AHP process, and (ii) an especial HAZOP-study, manageable parts for immediate reviewing, (ii) to plan the study
extended with Risk-Matrix (DMRA) and the Analytical Hierarchy Pro­ schedule and to arrange for the meetings. Finally the third phase will
cess (AHP), i.e. the E-HAZOP approach. In addition, the applications of consist of the identification and documentation of the hazardous sce­
the combined HAZOP-DMRA technique and also the new proposed E- narios that could materialize from the process design intent and oper­
HAZOP framework are performed as case studies, in a sour crude-oil ation. Also, the recommendations of the HAZOP team should be

2
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

documented in order to be considered (Kletz, 1997, 1999; Baladeh et al., process is based on empirical judgement and this inherently mean a lack
2019). of precision.
Although it has such a structured methodology, the HAZOP study can To proceeded, the integration of multicriteria methods with HAZOP
be tedious, complex and time consuming. The way of organizing the has been attracted significant research interest. Mainly, AHP used for
study into manageable sections to be analyzed, is critical for the success occupational health and safety problems, initially by Henderson and
of the endeavor. The proper division into suitable sized nodes is a Dutta (1992) and especially for ergonomic factors assessment. Also,
difficult task. The practice can be considered as something of an art, Badri et al. (2012) used multicriteria analysis methods for occupational
because there is no assisted node management, so it is based entirely on health-safety (OHS) risks evaluation, while Aminbakhsh et al. (2013)
the leader’s judgement (Dunjo � et al., 2010). used AHP for safety risks prioritization in construction projects. More­
An excellent literature survey for HAZOP-analysis, with the purpose over, Gul and Guneri (2016) presented a fuzzy multi criteria
of classifying the research conducted over the years, and defining the risk-assessment methodology based on a risk-matrix technique, in order
HAZOP state-of-the-art, can be found in the study of Dunjo � et al. (2010). to help aluminum industry stakeholders to struggle with hazards more
The person responsible for deciding why and when to conduct a efficiently. Besides, Othman et al. (2016) used AHP for ranking risk
HAZOP study, in both existing and new process facilities is the project- factors identified by a HAZOP study, and applied to a reactor unit and a
manager. When the study will take place in a new facility, it can be complex system of dividing wall column pilot plant. Recently, Cheraghi
considered only after the basic engineering documentation has been et al. (2019) used HAZOP with fuzzy AHP and fuzzy TOPSIS to deter­
produced. The configuration of the HAZOP study may need to be mine the weight of risk factors and to prioritize the hazards on a gas
updated in the case of existing facilities. It is a critical decision whether wellhead facility.
to conduct a new hazard identification analysis when a change in the
process is introduced. Whatever the case, the validation procedure will 3. Presentation of the sour crude-oil processing plant
have to be performed. Consequently, the decision on why and where to
conduct a HAZOP study depends directly on the process lifetime. The 3.1. Description of the installation and its purpose
purpose, scope and objectives should be defined according to that
(Herbert, 2011). A crude stabilization unit processes sour crude-oil (i.e. crude-oil that
Besides, many accidents in the process industry have occurred due to contains hydrogen sulfide) from the oil production platform and pro­
the unforeseen consequences of minor modifications. So, it is up to the duces marketable crude-oil. The processes that take place are:
management of the facility to determine whether a modification con­
tains risks to health, safety or the environment. Furthermore, whenever � Desalinization and stabilization of crude-oil. The unit has been
the conclusion of this process is that a hazard may be present then the designed to desalinate and stabilize 100 m3/h incoming crude-oil
existing HAZOP study has to be updated (Baybutt, 2013). from the production platform. During this process sour gas is pro­
What is more, HAZOP studies are performed for a variety of reasons. duced, approximately 70.000 Nm3/day that are treated in a different
To validate good engineering practices, to comply with regulatory re­ industrial unit.
quirements, to investigate accidents, to meet the recommendations of � Gas compression. Sour gas that is being produced from the top of
internal audits or to guide in the creation of a safety management system the crude stabilizer tower has to be compressed in order to be merged
(Mentzer et al., 2014). Additionally, HAZOP is considered as an effective with the sour gas stream originating from the three-phase separator
method for recognizing hazards, which additionally could increase the vessel. The sum of the two streams is cooled by air cooled heat ex­
safety levels, prevent accidents and enhance the reliability of systems via changers and then is passed through a three-phase separator where
the reduction of operational problems (Alaei et al., 2014; O Herrera water and Natural Gas Liquids (NGL) are separated.
et al., 2018). � NGL stripping. During this process, the NGL produced during the
Nevertheless, conventional HAZOP has several drawbacks (Cheraghi processes mentioned above is being stripped of hydrogen sulfide
et al., 2019) and so that, the development of HAZOP-analysis is in (H2S) and some volatile hydrocarbons before being mixed with the
progress, in such a way, that other processes are being embodied, day stabilized crude-oil and routed to storage.
after day, in HAZOP. Risk-matrix or decision-matrix risk assessment
(DMRA) is a characteristic case. For instance, Alaei et al. (2014) eval­ There are one input and three outputs in the unit. The input is sour
uated the level of safety in the Claus reaction furnace crude-oil that is being transferred to the unit via a submerged 800 pipeline
package/blower/heat-exchanger equipments of a sulfur recovery unit from the oil production platform. The first output is stabilized crude-oil
(SRU), using a HAZOP-study joint with DMRA. Moreover, Filippidis that is driven to storage after it is mixed with H2S-free NGL. The second
(2014) merged the DMRA with HAZOP for achieving hazard identifi­ output is sour gas that is produced in the unit and is driven to another
cation and risk analysis in the first stage separation of a crude-oil pro­ unit for further treatment. The third output is the sour water that is the
cessing plant. sum of the water content that arrives in the facility with the crude-oil
DMRA is a systematic and commonly used approach for estimating and the injected water that has been mixed with it, before being
risks and consists of measuring and categorizing risks on an informed recovered. There are also utilities flows in the unit.
judgment basis as to both probability and consequence and as to relative
importance (Reniers et al. 2005; Marhavilas et al., 2011a, 2011b; Gul � 20.7 bar steam is being provided to heat exchanger to provide the
et al., 2018). Once the hazards have been identified, the question of energy required for the crude stabilization and NGL stripping.
assigning severity and probability ratings must be addressed. It is very � Potable water is being mixed with the sour crude-oil to aid in the
important to note that frequency estimates and consequence estimates desalinization process.
are very well considered and performed by experienced risk managers. � Compressed air is being provided for the operation of the control
The current consensus is that risk ranking in HAZOP studies should be valves.
used for qualitative reasons, in order to prioritize the recommendations � Nitrogen is used occasionally for vessel or pipeline for insertion.
that would emerge. It is beyond the point of HAZOP to perform a
quantification of the risks involved in the process as there are more The input stream of the sour crude-oil arrives at the unit through an
suitable analysis methods to that purpose (Montague, 1990). There are 800 pipe. It has a pressure of 20 bars and the temperature is that of the sea,
some guidelines involving rink ranking. The more levels, either for so it fluctuated seasonally from 14 � C in the winter to 25 � C in the
severity or frequency, the more precise will the assessment be. However summer. It is a mixture of saturated volatile hydrocarbons, various
in the case of risk assessment in the context of a HAZOP study the volatile compounds (Nitrogen, Hydrogen Sulfide, Carbon-dioxide),

3
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

medium volatility hydrocarbons and heavier compounds. Table 1

The untreated sour crude-oil that enters the unit also contains brine, A concise description of the installation equipment.
dissolved sodium, calcium and magnesium chlorides in small quantities Nr Type of the part of specifications
of water. The brine is dispersed in the crude-oil in the form of tiny installation equipment
suspended droplets and depending on the diameter of these droplets, 1. Vessels V-101: It is a three-phase separator vessel in
they can be separated by gravity in calm conditions or by using electrical which sour gas is being released from the sour
methods. crude-oil due to the drop of pressure that is
The unit is designed to remove water, salts, suspended particles, controlled by PV-101. Also any water content in
the sour crude-oil is being caught in the boot of
hydrogen sulfide and a large part of the volatile hydrocarbons from the the vessel and moved to the sour water
sour crude-oil and transforming it into stabilized crude-oil. Stabilized processing unit before it is released in the sea. As
crude-oil is safer for storage, transportation and further processing in a sour crude-oil passes through PV-101, pressure
refinery. The final stabilized crude-oil must have specific characteristics drops to 10 bars and is maintained at that level
with PV-102 that releases sour gas to E 104. At
such as:
the exit of the vessel there is a demister to trap
some hydrocarbon droplets. Oil is exiting the
� Vapor pressure reid must be below 12 PSI vessel through flow control and FV-107 to V-
� H2S content must be below 15 ppm 102. The LV-103 on the vessel triggers a low
� Water content must be as close to zero as possible alarm if the level of the sour crude in the vessel
drops below 25% and a high alarm if the level
� Salt content must be below 28 mg/lt rises above 75%. It also triggers a shut-down
function if the level reaches 80%.
The removal of the dissolved hydrogen sulfide is necessary because it ME-101: It is a desalinization vessel that uses an
is a volatile, toxic and flammable gas that would render storage and electrical field to achieve separation of the salts
in the sour crude-oil and their dissolution in the
transportation significantly more dangerous. Although non-toxic, vapor
water at the bottom of the vessel. The mixture of
pressure in the final product must be kept under a maximum value by crude-oil and water enters the vessel through an
removing volatile hydrocarbons to avoid creating an explosive envi­ internal distributor into the high voltage
ronment during storage and transportation. electrical field. Part of the water is separated by
The amount of salts in the final product is important for all phases of gravity and the rest is passed through the
electrical field with the crude-oil. The high
its processing (from stabilization to refining), even in small quantities. It
voltage field aids in the aggregation of the small
can leave deposits in process equipment that would cause reduction of water droplets to larger that can be separated by
their efficiency and shut down for maintenance. Also it can increase the gravity. The low voltage electrical field inside
corrosion in the transportation pipelines and in storage facilities. the vessel further aids in the separation of the
water droplets by achieving additional
Furthermore, it can increase the cost of moving the stabilized crude-oil
aggregation of the water droplets. Water and
to the refineries, as it is not a value product and finally it can cause a various impurities dissolved in it (oil droplets,
significant damage to fraction distillation columns in the refineries. asphaltenes, iron oxides, iron sulfide and others)
There are chloride salts (and especially magnesium chloride) which are are gathered in the bottom of the vessel and form
broken down and form the extremely corrosive hydrochloric acid. of water layer. This sour water is constantly
being removed and guided to the sour water
As far as the produced sour gas is concerned, the gas produced in the
processing unit for treatment. Water level is
unit must be separated from its liquefiable contents before being driven controlled by the LV-101 and the pressure inside
to the gas treatment unit. This is pursued to avoid problematic operation the vessel is controlled by PV-101. The electrical
in the sour gas treatment unit. Also they contain valuable hydrocarbons transformer of the vessel ME-101 has a
selectable voltage output in five steps, from
that must be recovered and reinjected in the final crude-oil product to
15 KV to 22 KV. The voltage applied is chosen
increase its value. This has to be achieved without violating the based on the samples taken at the output of the
maximum permissible value for the vapor pressure. The NGL that is vessel. If the water and salts level is above the
recovered from the gas produced in the unit contains H2S and so, it must specified level then the voltage can be increased
be stripped before the spiking of the crude-oil can occur. to better aggregate the water droplets. Inside the
desalter there are two parallel grids. Voltage is
applied to the bottom one whereas the top one is
grounded, as is the vessel itself. In reality inside
3.2. Installation equipment the vessel there are two electrical fields. The one
is strong and is formed between the charged and
the grounded grid and the other is weak and is
In Table 1 we a present a concise description of the installation
formed between the charged grid and the water
equipment, which is illustrated by the process flow diagram (PFD) of layer that is in contact with the grounded vessel
Fig. 1 and divided into vessels, heat exchangers, pumps and compres­ wall.
sors. In this table, it is also included a brief description of their func­ V-102: It is the crude stabilizer column. It has 30
tioning, not extended to all their details (more information exists in the valve tray disks and operates at 2.6 bars and the
bottom temperature is 160 � C. Sour crude-oil is
work of M. Filippidis, 2017). inserted through a distributor above the top
The installation includes, on the one hand, two cooling heat ex­ tray. Bellow the bottom valve tray there is
changers using sea water and another one using fans, and on the other chimney tray, where crude-oil is being collected
side, two heating heat exchangers that use 20.7 bar steam as a source of and is passed through the reboilers E 101 to be
heated and partly gasified. The returning gases
energy. In addition, there are existed a heat-recovery heat exchanger,
and liquid crude-oil is returned to V-102 below
three pumps and one compressor. Moreover, there are three types of the chimney tray where the gases can ascent
control valves in the facility, i.e. pressure, flow and level valves that through the chimney and come into contact with
control the respective element. They are pneumatic valves that draw the sour crude-oil descending.
their power from compressed air supplied from a utilities-unit. If the air As the gases move upwards through the valves of
the disks of the column they come into contact
supply is lost due to any reason the control valves will assume a pre­ and heat the incoming crude-oil releasing the
determined position, either closed or open, to ensure the safety of the more volatile components such as light
installation. There are also several pressure relief valves and alarm (continued on next page)
switches to ensure the safe operation of the facility.

4
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Table 1 (continued ) Table 1 (continued )

Nr Type of the part of specifications Nr Type of the part of specifications
installation equipment installation equipment

hydrocarbons and hydrogen sulfide. The result E-101: It is a heat exchanger that uses 20.7 bar
of this process is that the crude-oil collected in steam to heat the sour crude-oil in V-102. The
the bottom of V-102 is virtually free of hydrogen steam provided to E 101 is controlled through
sulfide and with low enough vapor pressure to FV-110.
meet the specifications of the final crude-oil to E-102: It is a heat recovery heat exchanger that
storage requirements. Stabilized crude-oil from utilizes the energy from the stabilized crude-oil
the bottom of the column is fed in the P-101 leaving V-102 to heat the incoming sour crude-
pumps to be transferred to storage after heat is oil entering the unit. There are additional
recovered in E 102 and is cooled in E 103. The advantages to this process, as the increased
level in the bottom is controlled via LV-106 temperature of the sour crude-oil is increasing
which is located after the heat exchanger its mobility by lowering its viscosity.
E 103. The pressure inside the vessel is kept Furthermore there are advantages in the
constant with pressure control and PV-103. desalinization process in ME-101 as the
V-103: It is scrubber vessel that is intended to increased temperature favors the separation of
retain water that is contained in the gas output the water and salts from the crude-oil. Only one
from the top of V-102. The sour water that is of the two heat exchangers is in operation at any
collected in V-103 is guided through level given moment.
control and LV-109 to the sour water processing E-103: It is a cooling heat exchanger that uses
unit for treatment before it is released in the sea. sea water to achieve its purpose. It a situated
At the top of the vessel there is a demister to after E 202 and the stabilized crude-oil after
increase the collection of water droplets from being cooled in E 203 is shifted to storage. The
the gas. reason for cooling the crude-oil before moving
V-106: It is a vessel that acts as a buffer between the stabilized crude-oil to storage is to avoid
the potable water intake and the crude-oil creating an explosive environment in the tanks
pipeline injection point. It has attached in its and avoid losing valuable volatile hydrocarbons
bridle a level transmitter that controls the LV- due to evaporation in the tanks.
102 keeping the water level inside the vessel E-104: It is an air cooled heat exchanger that is
constant. Blocking the valve and measuring the used to cool the stream of sour gas originating
decrease in the water level aids in the from C-101 and V-101. It has 3 fans and
calculation of the water injection rate. At the regulated shutters to control the gas output
bottom of the vessel there is a connection to P- temperature. The gas, after exiting E 104, is
102. passed through V-110 where it deposits its
V-107: It is NGL separator and is intended to liquefiable content.
collect the NGL from the cooled gas stream from E-105: It is a heat exchanger that uses 20,7 bar
E 104 that also contains the NGL separated at steam to heat the sour crude-oil in V-108. The
V-110. It has a side intake for the gas and a steam provided to E 105 is controlled through
demister at the top bellow the gas exit. The NGL FV-127.
collected at the bottom is fed to P-103 to be E-106: It is water cooled heat exchanger using
stripped at V-108. The gas that exits from the top sea water as a cooling medium. NGL exiting V-
is guided to the sour gas treatment unit as it is 108 passes through E 106 and is cooled to 38 � C
mostly hydrogen sulfide. and afterwards it joins the stream of stabilized
V-108: It is a condensate stripper column. The crude-oil to storage.
sour condensate that is separated in V-107 is fed 3. Pumps P-101: It is a centrifugal pump to move the
to the condensate stripper V-108 through flow stabilized crude-oil from the bottom of V-102
control valve FV-124. The stripper column through E 102 and E 103 to storage. It has a
consists of 3 layers of filler material. A liquid design flow rate of 216 m3/hr. The rated suction
outtake is situated above a chimney tray disk at pressure is 3.1 barg and the rated discharge
the bottom of the vessel. The liquid is moved pressure is 8.1 bar, so it has a dP of 5 bar. The
through there to E 105 where it is heated with design operating temperature of the liquid is
20,7 bar steam to gasify it and returned at a 160 � C. It is rotating at 2900 rpm and has a rated
point below the chimney tray. As the hot gases hydraulic power of 30.2 kW. The efficiency is
ascent in the column they come into contact 72.6% and the motor power is 75 kW.
with the sour NGL that is descending forcing P-102: It is a reciprocating pump to inject
hydrogen sulfide and some more volatile potable water in the sour crude-oil stream. The
components to be released and exit through the suction pressure is 1 barg and the discharge
top of the column. The gas from the top joins the pressure is 20 barg in order to be able to be
stream from V-107 and is fed to the sour gas connected with the crude-oil pipeline. The water
processing unit. The H2S free NGL at the bottom supply is from V-106 and the flow rate is
is cooled in E 106 using sea water and it joins controlled through a recirculating valve which
the stream of crude-oil to storage. defines the percentage of the total water flow
V-110: It is a three-phase separator. Cooled gas rate will be mixed with the crude-oil. The
from E 104 enters the vessel and separates into maximum flow rate is 3 m3/h. It has a rated
liquid and gas. The gas exits the vessel from the power of 4 kW.
opposing side to the entry. The liquids are P-103: It is a centrifugal pump to move the sour
further separated into water which is collected NGL from the bottom of V-107 to V-108 for
by gravity in the boot and NGL between the stripping. It has a design flow rate of 3.4 m3/hr.
water and the gas. There is a demister situated The rated suction pressure is 9.3 barg and the
before the boot of the vessel to retain more rated discharge pressure is 12.3 bars, so it has a
liquid droplets. Sour water from the boot is dP of 3 bar. The design operating temperature of
guided to the sour water processing unit and the liquid is 16 � C. It is rotating at 2900 rpm and
NGL is reinjected in the gas stream after the PV- has a rated hydraulic power of 0.29 kW. The
118. This is done because there is minimum flow efficiency is 18% and the motor power is
line before PV-118 that recirculates gas that 1.64 KW.
exits the vessel back to C-101. 4. Compressors C-101: It is a two-stage double action
2. Heat Exchangers reciprocating gas compressor that is in place to
(continued on next page)

5
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Table 1 (continued ) follows the risk-quantification step, and third is applied the cοmbinatiοn
Nr Type of the part of specifications and jοintly evaluatiοn using the АНР and DМRА, as a part οf the
installation equipment quantified risk evaluatiοn. Τhe АНР method is used fοr priοritizing the
pressurize gas originating from the top of V-102
risk factοrs. Finally, in case that the resulting amount of risk is not
to be able to merge with the gas released from V- tolerable, the process incοrpοrates safety related decisiοn making in
101. The first stage of the compressor has a order to apply policies to reduce risk. We nοte that the new alternative
suction pressure of 2.2 barg at 71 � C from V-103 risk-assessment framewοrk and the develοpment οf this figure were
and a discharge pressure of 5 barg at 108 � C. In
based οn the safety aspects–guidelines οf ІSΟ/ІЕC (1999, 2009),
the second stage the pressure rises to 10 barg at
145 � C. The rated flow rate for the compressor is Нøj and Κro €ger (2002) and Мarhavilas (2015; chapter 10).
4900 Nm3/h. It is rotating at 490 rpm and has an Below, we present some fundamental elements of the HAZOP-study,
electric motor with a rated power of 400 kW. To DMRA technique and the typical AHP in order to be more comprehen­
achieve the minimum flow rate required by the sible, the application of E-HAZOP (as case study) in a sour crude-oil
compressor there is a recirculating scheme
where some of the gas, after being cooled is
industry.
returned through pressure control to V-103 to be
fed to C-101 once more. There is a PD vessel on
the suction and discharge side of each stage to 4.1. HAZOP study
damp the pressure fluctuations on the
compressor. Because of the fact that the There are two aspects that a HAZOP team leader must take into
compressor contains sour gas which is lethal to
consideration when deciding on the node selection issue: (i) The
humans and highly flammable if released in the
atmosphere, there is a sealing system in place on equipment in the node must share the same design intention, (ii) The
each stage. Beside the mechanical barriers there sections should be equal in size. There are specific steps that a team
is space containing pressurized nitrogen that leader must take in order to ensure a well selected set of nodes,
continuously flows to the flare system. If the regardless of the process in which it is being applied. The data that are
mechanical seals fail then the flow rate would
required to assist in this process are the following (Kletz, 1997, 1999): (i)
increase, thus increasing the flow rate and
alarming the operators to stop the compressor. Process description, (ii) Process Flow Diagrams (PFDs), (iii) Piping &
Besides this, there are numerous switches that Instrumentation Diagrams (P&IDs).
would either alarm the operators of a deviation The execution subphase of HAZOP analysis, concerning a part of the
from the normal operation or in some cases
system with specific function and interconnectivity, follows the orga­
would cause the compressor the shut-down.
nization of this system into nodes. The method is based on using
guidewords, such as “no”, “more” or “less”, in combination with various
4. The methodological framework of E-HAZOP parameters, such as “temperature”, “flow” or “pressure” in order to
create deviations from normal operation. This procedure is applied on
In Fig. 2, we illustrate, by “amalgamating” HAZOP with Risk-Matrix every node. The HAZOP team, after the exploration of all possible de­
(DMRA) and the typical Analytical Hierarchy Process (AHP), the flow­ viations of a node, proceeds to exploration of practical causes and
chart of the new-proposed E-HAZOP alternative risk-assessment possible consequences. For every “cause/consequence” combination,
framework, as a part of the risk-management process. which constitutes a scenario, existing safeguards must be identified (or
This methοdοgical framewοrk, cοnsists οf four separate steps: the newly proposed), that could prevent, detect, control or mitigate the
first includes hazard-sοurces’ identificatiοn using НАΖΟР-analysis, next hazardous situation. Finally, if the safeguards are insufficient to prop­
erly handle the problem, then the HAZOP-team will have to provide its

Fig. 1. The unit process flow diagram (PFD).

6
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Fig. 2. The flowchart of the proposed E-HAZOP alternative risk assessment framework by including HAZOP analysis, Risk Matrix (DMRA) technique and the typical
Analytical Hierarchy Process (AHP), as a part of the risk management process.

recommendations (Macdonald and Mackay, 2004; CCPS, 2008). Furthermore, Table 3 illustrates according to Alaei et al. (2014) and
Filippidis (2014, 2017) the risk-matrix parameters for the severity (S)
and likelihood (L), whereas Table 4, the produced risk matrix. There­
4.2. The DMRA technique fore, the assigned values for likelihood and severity, concerning each of
the hazards identified in the HAZOP process, have to be placed in the
Marhavilas and Koulouriotis (2008) presented a DMRA technique, risk-matrix in order to be acquired a single risk value (that means the
which was based on the works of Johnson (1973), Woodruff (2005) and risk assessment of that specific risk).
Reniers et al. (2005). In particular, they developed two tables, one for
showing the criticality rating (i.e. the severity S of consequences rat­
ings), and a second one for the hazard probability ratings L (i.e. the 4.3. The Analytical Hierarchy Process
frequency rating). Finally, the risk (R) was resulted by the development
of the risk matrix according to relation R ¼ SxL. The AHP method supports organizing the multicriteria problem into
Table 2 depicts, according to IEC (2003), the basic risk-ranking levels a hierarchical structure and then, via pairwise comparisons and
by combining the severity levels with the cause frequency levels. weightings, defining the local and global priorities for the criteria and
subcriteria of the problem. AHP uses as input the judgments of the de­
cision maker on how important is a criterion over another, during
Table 2
Basic risk ranking criteria (IEC, 2003). conducting the pairwise comparisons, and outputs a ranking according
the importance of each criterion and/or subcriterion of the analysis.
Risk Matrix (R) Consequence Severity Levels (S)
Assignments of the decision maker’s qualitative estimations of impor­
SM SS SE tance to numerical values are realized using a standard scale (Table 5).
Cause Frequency Levels (L) LL R6 R5 R3 A very important characteristic of the method is that it checks the
LM R5 R4 R2 input judgments of the decision maker, for possible inconsistency. The
LH R3 R2 R1 latter improves the quality of the results. In the present study, we use
Annotations: typical AHP for ranking hazards identified by the HAZOP method.
Minor (SM): Impact initially limited to local area of event with potential for
broader consequence, if corrective action not taken. 5. Case study
Serious (SS): Impact event could cause serious injury or fatality on site or off site.
Extensive (SE): Impact event that is five or more times severe than a serious
5.1. Application of HAZOP with DMRA on a sour crude-oil processing
event.
plant
Low (LL): A failure or series of failure with a very low probability of occurrence
within the expected lifetime of the plant.
Medium (LM): A failure or series of failure with a low probability of occurrence The descriptive/analytical study was carried out in a sour crude-oil
within the expected lifetime of the plant. processing plant, located in Greece, by a three-members team. The
High (LH) : A failure can reasonably be expected to occur within the expected HAZOP team was consisted, on the one hand, of two unit-operators in
lifetime of the plant. the process facility under examination, i.e. the installation’s operations

7
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Table 3 MDS was a mechanical engineer. The unit operators had long experience
Ranking of the risk-matrix parameters for the severity and likelihood [the table in operating the facility and acquired valuable knowledge on previous
development was based on the works of Alaei et al. (2014) and Filippidis (2014, incidents and possible hazards. The HAZOP sessions took place during a
2017). period of six months, and were completed totally in twelve sessions. At
Ranking Likelihood Severity (S) the first two sessions the process was explained to the team and the
(L)
Consequences Consequences nodes of the unit were selected. During the compilation of the HAZOP
regarding regarding tables, any guidewords that didn’t apply to the particular node were
personnel environmental omitted. Also any deviation that didn’t cause any effect was also left out.
injury damage Finally, in the last session, the three-member team of experts extracted
1 Very likely Catastrophic/ single fatality or Multiple the judgments for the hazards-ranking of the typical AHP method.
(occurs Fatal multiple environments
frequently): fatalities) affected
� Step 1 - Node selection and purpose: The HAZOP meeting is
Once per
month started with node selection. A node represents a section of the pro­
2 Likely Severe/ fracture, Major localized cess in which the condition has significant change in term of pres­
(recurrent Serious hospitalisation effect sure, temperature, chemical composition. During the compilation of
but not >24 hrs, the HAZOP tables, any guidewords that didn’t apply to the particular
frequent): incapacitation
Once per >4 weeks
node were omitted. Also any deviation that didn’t cause any effect
year was also left out. Taking into account the function of the equipment
3 Possible Major strain/twist, Localized effect involved and the connections between the vessels, totally nine (9)
(could occur, sprain/ cramp/ nodes were selected as follows:
but dislocation,
� Node 1: Water Injection System
uncommon): incapacitation >
Once per 10 3 days � Node 2: Preheating and Cooling Exchangers
years � Node 3: Desalter Vessel ME-101
4 Unlikely Small/Minor small cut, Minor effect � Node 4: Low Pressure Separator
(occurs abrasion/ � Node 5: Crude stabilization column
rarely): scratch/scrape,
Once per 25 basic first-aid
� Node 6: Gas Compressor station and its accompanying scrubber
years need, no- vessel
hospitalization � Node 7: Compressor gas cooling & liquid recovery
5 Remote Trivial/No discomfort, slight Slight/no effect � Node 8: Condensate stripper column feeds vessel and pump P-103
(almost impact, only bruising, self-
never): Minor help recovery
Once per
100 years Table 5
The basic scale used for assigning numerical values to linguistic variables (Saaty,
1990).
superintendent (IOS) and the maintenance department superintendent Definition Intensity of importance
(MDS), and on the other side, of the risk manager (RM), who directed
Two activities are equally important to the objective 1
this team, and who was also the expert and decision maker (DM) for the An activity is moderately more important than another 3
judgments of Saaty’s scale in AHP technique. The RM/DM was a qual­ An activity is strongly more important than another 5
ified mechanical engineer with significant experience in hazard identi­ An activity is very strongly more important than another 7
fication in process industries. The IOS was a qualified chemical engineer An activity is extremely more important than another 9
Intermediate values, used when compromise is needed 2,4,6, and 8
with considerable expertise in the chemical process industry, while the

Table 4
Risk Matrix.

8
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

� Node 9: Condensate stripper and cooling unfolded in Fig. 3, we proceeded to the application of HAZOP analysis
� Step 2 - Process guideword/safe limits: The basic idea of HAZOP is concerning the elements “Level”, “Pressure”, “Flow” and “Containment”
identifying the deviation from design or safe process conditions. So, with the depicted results of Tables 7–10, respectively. We clarify, that
the process parameter which will be discussed should be identified. the conventional HAZOP-DMRA technique was also performed for the
The general process parameters are: “Flow”, “Pressure”, “Tempera­ rest nodes (i.e Nodes 2–9) in order to achieve, in a second phase, the
ture”, “Level”, “Composition”, “Containment” and “Phase”. And the application of E-HAZOP-technique in the crude-oil processing-plant.
guidewords for deviation from the safe operating limits are: “More”, In the produced HAZOP tables, concerning the established nodes
“Less”, “No”, “Reverse”, and “Other than”. The guidewords are 1–9, all the probable sources and hazards originating from a specific
combined with process parameters to identify the deviation from deviation have been identified. Many of them are common for multiple
safe process condition. Table 6 illustrates the HAZOP study elements deviations and some further analysis was necessary to identify unique
associated with the above selected nodes. hazards. The analysis of the data created from the HAZOP method will
� Step 3 - Identification of hazards and their causes: Once the nodes be presented and also the key role which this data analysis can perform
are described and the guidewords and process parameters are in organizing maintenance and inspection operations of the facility.
defined, the hazards are determined. A hazard is a deviation from the It is worth noting that the unique hazards identified by the HAZOP
safe operating limit which is determined by using the particular study have been grouped into three categories: (i) the first one includes
guidewords. The HAZOP team identified the root causes of the “hazards to human life and the environment” (HLE hazards) origi­
hazards associated with the specific node. nating from the installation, (ii) the second one presents the “opera­
� Step 4 - Consequences: At the next step the HAZOP team had bility hazards” (OP hazards) i.e. the hazards reacting on the product
determined the consequence of hazards in term of safety, environ­ quality and economic operation of the installation, while (iii) the third
mental, and economic. These consequences are determined with and group contains the “hazards with both economic and health/envi­
without safeguards in place. ronmental impact” (EHE hazards) i.e. the hazards that affect the
� Step 5 - Identification of safeguards: The relevant available safe­ economic performance of the installation and simultaneously pose a
guards to detect the identified deviations (hazards) and prevent the danger to human life and/or the environment.
identified consequences are considered and documented in HAZOP Taking into account the enormous size of the crude-oil processing-
worksheet. plant, the plethora of the established HAZOP nodes, and also their
� Step 6 - Recommendations: The recommendations to decrease the functionality, we concentrate through this paper, on the first category of
level of risk are discussed by HAZOP team and recorded in the hazards i.e. the HLE hazards, while in a future paper we will focus on the
HAZOP worksheet. OP hazards. Therefore, Table 11 illustrates the data of HLE hazards
created from the above referred HAZOP method, and more explicitly,
The risk-matrix format that was used in the risk assessment will have the data coming from the column “Consequences” of Tables 7–10. For
the form illustrated in Table 4. Our method was based on significant example the hazard “Explosive environment in ME-101” (rows 6 and 7,
historical data to analyze occurrences per number of operations for Table 11) is presented as a consequence in Table 7 (4 times), Table 9 (2
specific hazards. The likelihood of the identified hazards has been times).
assigned by using the information base of previous incidents in the unit The most significant hazards according to Table 11 are H2S-
with similar equipment. Moreover, the failure rate of common control releasing and hydrocarbons-releasing as well, taking into account the
equipment, such as level or pressure transmitters, has been calculated by implications of inhalation that may occur to people, causing consider­
the usage of previous incidents as well. On the other hand, the estima­ able harms to their health. Furthermore where hydrocarbons are
tion of the hazards severity was based on the consensus of the team and mentioned, these may be either in liquid or gas form. The distinction can
personal judgment. We assigned values for the hazard severity (S) and be difficult because of the volatility of these components. NGL can be
likelihood of occurrence (L) (taking into account the frequency and liquid inside a vessel under pressure and be evaporated almost instan­
duration of exposure, and according to Table 3) on a scale of 1–5, then taneously, when it is exposed to atmospheric conditions. The pipeline
multiply them together to give the rating band. The overall aim is to and vessel fracture (or rupture) may be a combination of elevated
reduce or remove the risk to an acceptable (as close to 1 as possible) pressure and poor condition of the equipment. Rust is an especially
level. critical problem, when considering the age of the facility under
The study was based on the unit’s PFD diagram illustrated in Fig. 1. examination.
Taking into account the enormous size of the crude-oil processing-plant, The HAZOP study can be a useful guide to provide maintenance and
the plethora of the established HAZOP nodes, and also their function­ inspection staff a list of priorities. By far, the most frequent causal issue
ality, we concentrate through this section, on the more significant node for the hazards listed above, was identified in the HAZOP study as the
of the processing-plant as far as the risk is concerned. In other words, we malfunction of level and pressure transmitters and controllers. These can
choose Node 1, for performing (as an example) the conventional often fail because they come into contact with impure substances with
HAZOP-DMRA technique, in order to identify the possible fault causes high deposits content. This work suggests that due to the fact that the
due to abnormal conditions (deviations). So, for the first node, which is cause of high percentage of hazards identified in the HAZOP study is

Table 6
HAZOP study elements associated with the selected nodes.
Nodes Elements

Level Flow Pressure Temperature Composition Phase Containment

1 ✓ ✓ ✓ ✓
2 ✓ ✓ ✓ ✓ ✓
3 ✓ ✓ ✓ ✓ ✓ ✓
4 ✓ ✓ ✓ ✓ ✓
5 ✓ ✓ ✓ ✓ ✓ ✓
6 ✓ ✓ ✓ ✓ ✓ ✓
7 ✓ ✓ ✓ ✓ ✓
8 ✓ ✓ ✓ ✓
9 ✓ ✓ ✓ ✓ ✓

9
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Fig. 3. Node 1: Water injection system.

Table 7
Node 1, “Level” element.

attributed to level and pressure transmitter failure, they should be with high pressure content, and especially those with H2S content.
checked thoroughly. Also, pressure and level transmitters that have an Generally speaking, the equipment in oil facilities is very expensive,
age of more than ten years should be replaced preemptively. due to increased quality specifications are required and due to the fact,
Another very common issue that could have catastrophic conse­ such equipment is usually custom-engineered to specific applications.
quences is pipeline and vessel rapture. Because of the facility’s long Thus, any action of replacing the whole or a part of such equipment can
operations span and the highly corrosive chemicals it contains, extensive have a severe economic impact for the company operates the facility.
pipeline and vessel wall thickness inspections must be performed
regularly. Priority should be given to vessels and pipelines operating

10
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Table 8
Node 1, “Pressure” element.

Table 9
Node 1, “Flow” element.

Table 10
Node 1, “Containment” element.

5.2. Application of E-HAZOP on a sour crude-oil processing plant implementation of E-HAZOP method, in this industry.
In the beginning, we note that we have subgrouped the “hazards to
Taking into account the results of the previous HAZOP-DMRA study, human life or the environment” presented in Table 11, in order to create
which are depicted in Table 11, we proceed (as a case study) to the next such a compact hierarchical structure that will help the decision maker
step i.e. the application of AHP process on the sour crude-oil processing- to define the local and global priorities and the input judgments as well.
plant (by using the hazards of Table 11) in order to complete the Therefore, Table 11 illustrates the “hazards to human life or the

11
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Table 11
List of hazards only related to “human life or the environment”, after subgrouping the hazards.

environment” after the essential subgrouping (intermediate table of depicted in Table 12 by the row (or new hazard number/code) of “[4]”.
hazards), while Table 12 the final ten transformed hazards, necessary for It is worth mentioning, that the different colors don’t express any
the pairwise comparisons (transformed table of hazards). More specif­ meaning concerning each hazard, but on the other side, we have used
ically, for attaining the necessary hazard-subgrouping, we have taken them in order to present more clearly (or to help the reader to under­
into account the presented similarities, as far as the nature/origin/ stand easily) the way of subgrouping the “hazards to human life or the
properties of the hazards are concerned, and in association with the environment” of Table 11, as far as the creation of the hierarchical
nodes which incorporate a specific hazard. For example the hazard “H2S structure is concerned.
release due to pipeline rapture” undergoes almost in the entire number of Consequently, the hierarchy of the present case consists of ten hazard
nodes (i.e. node 1 to 9) as it is illustrated by the rows 14–21 in Table 11. factors belonging to the same level, as illustrated in Fig. 4. It contains the
So, we have grouped them according to Table 11, and the result is analysis’ goal which is preserving safety in the workplace, and ten

12
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Table 12
Unique hazards to human life or the environment, necessary for the pairwise comparisons of AHP process
(transformed table of hazards).

criteria at the same hierarchy level which means that they are directly 5.3. Results of the E-HAZOP application
comparable to their contribution to the overall risk. In this analysis, we
considered as criteria the ten groups of hazards (Table 12) resulted from Table 14, summarizes the results of E-HAZOP after its final step (i.e.
grouping the unique hazards to the human life or the environment ac­ the application of AHP) and is constituted with the priorities of hazards
cording to their source, included in Table 11. These unique hazard calculated by the typical AHP (column a) and also the ranking of hazards
factors considered after making sub-groups of the hazards which in­ with decreasing importance (column b).
cludes all the hazards irrespective of their category. According to the typical AHP calculations, and based on the input
The expert asked to be the decision maker, made judgments using the judgments of the expert, the factors “H2S release to atmosphere (due to
classic Saaty’s scale (Table 5) and filled in the pairwise comparisons human error/opened valve) in a specific node” (03) and “Hydrocarbons
matrix of the hazard factors (Table 13), in which he expressed his release to atmosphere (due to human error/opened valve) in a specific
preference of importance of each hazard in comparison with all the node” (07) are more important than the other hazards, since they cause
other hazards of the present case. The Typical AHP is used to transform the 50.33% of the total risk.
expert judgments into weights and then to rank the risks in descending Next, the factors “H2S release due to pipeline rapture in a specific
order of significance, according to the decision maker’s values. It is node” (04), and “H2S release due to Vessel rapture (or failure) in a
crucial for the resulting ranking that the judgments proved to be specific node” (05) consist the second more important hazards group,
consistent since the Consistency Ratio (CR) (Saaty, 1990) found to be responsible for the 27% of the total risk.
less than 10% for the pairwise comparison matrix. Also, the hazards “Hydrocarbons release due to pipeline rapture in a
In the present analysis, there is one person considered as the decision specific node” (08), “Hydrocarbons release due to Vessel rapture (or
maker. The specific expert chosen to assign preferences in the pairwise failure) in specific node” (09), and “H2S release due to other machinery-
comparison matrix serves as the leading engineer-production manager rapture (or failure) in a specific node” (06) have slightly different
(i.e. the RM/DM) of the sour crude-oil processing-plant, with several resulting weights but clearly higher than the impact of the last three
years of experience in managing risks in industry. Thus, due to his hazards, namely “Hydrocarbons release due to other machinery-rapture
experience and expertise, his opinion considered of being very impor­ (or failure) in specific node” (10), “Explosive environment in ME-101
tant to the proposed approach application in the specific case. However, (due to air or hydrocarbons)” (02), and “C-101 failure (due to pres­
there is a merit in extending the present approach to group decision sure level or liquids presentation)” (01) which are responsible, cumu­
making in order to include the opinions of a larger number of executive latively, for the 7.04% of the total risk.
managers/engineers and increase the reliability of the final total risk
estimations and rankings. 6. Discussion

The chemical process industry (CPI) is currently one of the safest

workplaces in the world (Lees, 1980; Filippidis, 2017). This is due to the

Fig. 4. The hierarchical structure for the hazards of the case.

13
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Table 13
The judgments of the expert for extracting ranking using the typical AHP.
Safety [01] [02] [03] [04] [05] [06] [07] [08] [09] [10]

[01] 1 1/2 1/8 1/5 1/4 1/4 1/7 1/5 1/3 1/3
[02] 1 1/7 1/5 1/4 1/3 1/6 1/4 1/3 1/2
[03] 1 3 3 5 2 5 6 5
[04] 1 4 6 1/4 4 5 6
[05] 1 3 1/4 4 3 5
[06] 1 1/5 1/3 1/2 3
[07] 1 5 6 7
[08] 1 2 4
[09] 1 2
[10] 1

et al., 2014). In addition, HAZOP studies have the main purpose of

Table 14
hazards identification and are considered as the best PHA techniques for
Results of E-HAZOP application: (i) the priorities of hazards calculated by the
producing data for subsequent quantitative analysis techniques such as
typical AHP (column a), and (ii) the ranking of hazards (column b).
LOPA, FTA or QRA.
Weights for the hazards Ranking of hazards The main interest for implementing the HAZOP methodology in the
(a) (b) sour-COPI, was to validate its safety and identify some safety issues that
[01] 1,85% [03] 25,57% may have been arisen during its operation for many years. For example,
[02] 2,32% [07] 24,76% pipeline rapture may not have been considered as a probable occurrence
[03] 25,57% [04] 16,75% when the installation was constructed. However, after many years of
[04] 16,75% [05] 10,25%
being in operation, issues like this become more possible to occur. In
[05] 10,25% [08] 6,80%
[06] 4,33% [09] 4,50%
other words, the installation under examination is an aged facility but
[07] 24,76% [06] 4,33% with high safety standards for the age it was constructed. The goal of the
[08] 6,80% [10] 2,87% authors was to validate that those safety standards stand true today. As
[09] 4,50% [02] 2,32% the safety regulations and technology evolve, there are more strict safety
[10] 2,87% [01] 1,85%
standards that must be adhered in the chemical process industry.
However, there are a lot of legacy installations that may not be up to
strict procedures and extreme caution that the industry exhibits towards standards and furthermore there is the added issue of the material and
hazards. However the public perception towards CPIs is that, they are equipment aging that complicates the issue.
highly dangerous and people often protest the installation of chemical Consequently, the objective of this article is dual, first the presen­
process facilities near their place of residence, fearing for their lives and tation of the HAZOP method, and secondly the implementation of the
properties. extended HAZOP-DMRA-AHP approach (E-HAZOP) in the previous
During the last thirty years, the risk management and process safety referred COPI. The selection of the specific facility for the application of
have been developed to an amazing level and cases like Bopal and HAZOP is highly advantageous for the demonstration of the method
Seveso are beginning to fade from memory. The goal of CPI is to change because of the numerous hazards involved in the process (such as
the hazardous public perception into what it actually is today, an in­ flammable liquids and gases and also toxic gases) and also due to its
dustrial and economic powerhouse that is beneficial for the society as a sophisticated design. The identification of hazards by HAZOP indicates
whole. Especially the crude-oil process industry (COPI) can be highly that they could, systemically, assess and criticize the process. These
beneficial for local economies, either for the job opportunities it pro­ techniques can be, therefore, considered as an effective method for
vides or due to the general economic benefits. Nonetheless, like all CPIs, recognition and prediction of hazards and it may increase the safety
there is always present a potential for catastrophe. The experience over levels, prevent accidents and enhance the reliability of systems via the
the past 30 years is that with proper risk management, catastrophe reduction of operational problems (Alaei et al., 2014). The
would be impossible. processing-plant that came under examination proved to be safe and
One of the most important aspects of risk management is hazard reliable, in the condition that proper installation-inspections and
identification and a highly effective tool to that point, is the HAZOP maintenance are enforced. To this point, it is suggested that hazard
methodology. The HAZOP study execution may be tedious and laborious identifications should be used as a priority setting mean for inspection
but it has proven over the years to be one of the most important tool of and maintenance operations. The structured approach and thorough
chemical process industry to assure the safety of the people working in a analysis that HAZOP provide, is an excellent tool on which to judge the
facility, the environment and of course the integrity of the facility itself. focus of those departments.
In addition, HAZOP is one of the best and highly disciplined and ver­ Although being effortless and uncomplicated, HAZOP has the
satile techniques for hazard identification in the chemical engineering weakness of a deficient systematic-approach to clarify different con­
industry, and also widely used in the process industries for identifying clusions into an integrated result, thus it is vulnerable to imprecise and
potential hazards and operability problems (Mayes and Kilsby, 1989; baseless decisions (Othman et al., 2016). On the other side, risk priori­
Baladeh et al., 2019; Danko et al., 2019). As the motive for all industrial tization can surely help the manager to act in order to reduce or better
operations is the maximization of profit, HAZOP analysis has the great eliminate the most urgent risk factors and protect workers’ health by
additional advantage that it aims in identifying not only hazards to implementing the most important safety measures (Koulinas et al.,
health and safety but also operational problems of the installation. 2019). Better allocation of the constrained budget can reduce support
Operational problems may cause increased cost for the facility’s man­ and mortgage costs, and generally allow managers to have budget
agement or unsatisfactory products. Experience has shown that HAZOP available to further reduce the risk in projects.
technique does work, identifying hazards and critical control points with Therefore, by this work, an extended-HAZOP (E-HAZOP) study (i.e.
priorities for control (Mayes and Kilsby, 1989; Baladeh et al., 2019). By the conventional HAZOP combined with other two precious techniques,
using the HAZOP analysis, the operators can quickly seek out the loca­ both the DMRA and the AHP), is proposed as a new approach for (i)
tions of necessary documents to address abnormal situations (Alaei incorporating prioritization in HAZOP analysis through the usage of

14
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

AHP, and (ii) identifying and prioritizing critical points and potential listing of hazards. Nevertheless, weights-assignment throughout the
hazards in a sour crude-oil process industry (as a case study). In the pairwise comparison-phase is subjected to individual choice (appraiser)
beginning, the application of the conventional HAZOP study is per­ and consequently, should be bound by the judgement of a capable team
formed, in order to identify the possible causes and consequences, which of executive managers/engineers in order to include the opinions of a
correspond to abnormal conditions or deviations. With the results of the number of managers (i.e. group decision-making) and increase the
HAZOP study, on the one side and also the application of the DMRA reliability of the final total risk estimations and rankings.
technique for risk-Assessment, on the other side, the typical AHP process Overall, the findings indicate that the extended-HAZOP (E-HAZOP)
is then used successively, to extend HAZOP analysis by prioritizing the framework, proposed in this paper: (i) is an efficient safety assessment
risks in the plant. method for the complex chemical industry, (ii) can provide a trust­
Even though, conventional HAZOP is a popular technique for hazard worthy basis to boost process safety and improve occupational health
identification and risk ranking in hazardous facilities (according to and safety, and (iii) could be a useful tool for the decision makers to
Cheraghi et al., 2019) it has various handicaps: (i) it considers a limited estimate the emergency of investing constrained budget in actions pre­
number of risk factors (i.e. only the frequency and the severity of hazards), venting specific kind of deviations. Accordingly, by using this technique
(ii) it assumes equal weights for the risk factors (so that, “low-­ as a decision-making tool, the project-team will be capable of priori­
probability/high-consequence” and “high-probability/low-consequence” tizing any action to industry modification, retrofitting or construction
hazards are considered equally important as far as their ranking is con­ within the existing resources constraints.
cerned), and (iii) it uses precise data, which are rarely available (or highly As a final point, the examined COPI plant, despite its long time span,
uncertain), especially in the case of complex COPI facilities. has acceptable risk in its operation, due to proper maintenance and
The present work is an attempt to improve the foregoing negative inspection.
aspects of conventional HAZOP via the extended E-HAZOP approach
(HAZOP-DMRA-AHP). In the scientific literature the growth of HAZOP- References
analysis is in progress, so that other processes have been embodied in
conventional-HAZOP by the following combinations: (i) HAZOP-DMRA Ahn, J., Chang, D., 2016. Fuzzy-based HAZOP study for process industry. J. Hazard.
Mater. 317, 303–311. [Link]
(Alaei et al., 2014; Filippidis, 2014), (ii) HAZOP-dynamic FTA (Guo and Alaei, R., Mansoori, S.A.A., Moghaddam, A.H., Mansoori, S.M., Mansoori, N., 2014.
Kang, 2015), (iii) HAZOP-AHP (Othman et al., 2016), (iv) Fuzzy-based Safety assessment approach of hazard and operability (HAZOP) for sulfur recovery
HAZOP (Ahn and Chang, 2016), (v) Fuzzy-DMRA (Gul and Guneri, unit Claus reaction furnace package; blower; heat exchanger equipment in South
Pars gas processing plant. J. Nat. Gas Sci. Eng. 20, 271–284.
2016), (vi) Fuzzy multi-attribute HAZOP technique (FMA-HAZOP), and Aminbakhsh, S., Gunduz, M., Sonmez, R., 2013. Safety risk assessment using analytic
in particular HAZOP-fuzzy AHP and HAZOP-fuzzy TOPSIS (Cheraghi hierarchy process (AHP) during planning and budgeting of construction projects.
et al., 2019). J. Saf. Res. 46, 99–105. Retrieved from. [Link]
/S0022437513000479.
It is worth noting that for first time, through this work, we determine Andersen, S., Mostue, B.A., 2012. Risk analysis and risk management approaches applied
the weight of risk factors and prioritize the hazards by using together to the petroluem industy and their applicability to IO concepts. Saf. Sci. 50,
DMRA and AHP in the environment of HAZOP. Moreover, the applica­ 2010–2019.
Aziz, A., Ahmed, S., Khan, F.I., 2019. An ontology-based methodology for hazard
tion of the E-HAZOP scheme (HAZOP-DMRA-AHP), on a sour crude-oil
identification and causation analysis. Process Saf. Environ. Prot. 123, 87–98. https://
plant, shows that E-HAZOP presents a more transparent environment as [Link]/10.1016/[Link].2018.12.008.
far the rank of hazards is concerned, compared to conventional HAZOP. Badri, A., Nadeau, S., Gbodossou, A., 2012. Proposal of a risk-factor-based analytical
We have the opinion, that the extension of HAZOP-DMRA with AHP approach for integrating occupational health and safety into project risk evaluation.
Accid. Anal. Prev. 48, 223–234. [Link]
(i.e. E-HAZOP approach) could create a useful tool for the decision Baladeh, A.E., Cheraghi, M., Khakzad, N., 2019. A multi-objective model to optimal
makers to estimate the emergency of investing constrained budget in selection of safety measures in oil and gas facilities. Process Saf. Environ. Prot. 125,
actions preventing specific kind of deviations. In addition, the proposed 71–82. [Link]
Baybutt, P., 2013. The role of people and human factors in performing process hazard
E-HAZOP framework can provide a trustworthy basis to boost process analysis and layers of protection analysis. J. Loss Prev. Process. Ind. 26, 1352–1365.
safety and improve occupational health and safety. CCPS, Center for Chemical Process Safety, 2001. Layer of Protection Analysis- Simplified
According to Guo and Kang (2015), the improvement of the Process Risk Assessment. American Institute of Chemical Engineers, New York, ISBN
978-0-8169-0811-0.
completeness of HAZOP analysis is in progress, in such a way that CCPS (Center for Chemical Process Safety), 2008. Guidelines for Hazard Evaluation
various procedures (like the DyPASI) are incorporated in HAZOP sce­ Procedures. American Institute of Chemical Engineers, New York.
nario analysis and provide effective contribution to the Cheraghi, M., Baladeh, A.E., Khakzad, N., 2019. A fuzzy multi-attribute HAZOP
technique (FMA-HAZOP): application to gas wellhead facilities. Saf. Sci. 114, 12–22.
extended-HAZOP analysis. Taking into account this scientific tendency, [Link]
in a future research, we will introduce a new extended-HAZOP scheme, Chudleigh, M.F., 1994. Hazard analysis of a computer based medical diagnostic system.
i.e. the Fuzzy-E-HAZOP procedure for prioritizing the risks in the above Comput. Methods Progr. Biomed. 1 (44), 45–54.
CIA, 1977. A Guide to Hazard and Operability Studies. Imperial Chemical Industries and
referred COPI, and extend the present approach to group decision
Chemical Industries Associations Ltd., London, UK.
making in order to include the opinions of a larger number of executive Danko, M., Jano�sovský, J., Labovský, J., Jelemenský, L., � 2019. Integration of process
managers/engineers and increase the reliability of the final total risk control protection layer into a simulation-based HAZOP tool. J. Loss Prev. Process.
estimations and rankings. Ind. 57 [Link] ISSN 0950-4230.
Dunj� o, J., Fthenakis, V., Vílchez, J.A., Arnaldos, J., 2010. Hazard and operability
(HAZOP) analysis. A literature review. J. Hazard. Mater. 173, 19–32.
7. Conclusions Filippidis, M., 2014. Hazard Identification and Risk Analysis in First Stage Separation of
a Crude-Oil Processing Plant. MSc Diploma Thesis. Depertment of Petroleum and
Natural Gas Technology, Faculty of Engineering, Eastern Macedonia and Thrace
By this work, an alternative approach in prioritizing hazards in in­ Institute of Technology, Kavala, Greece.
dustry, by using the combination of HAZOP-study with the DMRA- Filippidis, M., 2017. Hazard and Operability Study of a Sour Crude-Oil Processing Plant,
technique and the AHP-process, is introduced. Our approach, not only Diploma Thesis, Depertment of Production and Management Engineering.
Democritus University of Thrace (DUTH), Xanthi, Greece.
supports the cooperation of HAZOP and AHP for ranking risks, but in Ftenakis, V.M., Trammell, S.R., 2003. Reference Guide for Hazard Analysis in PV
addition, it incorporates the combination of HAZOP with DMRA in order Facilities. Brookhaven National Laboratory, Upton NY.
to construct a process for identifying, assessing and prioritizing risks in Gul, M., Guneri, A.F., 2016. A fuzzy multi criteria risk assessment based on decision
matrix technique: a case study for aluminum industry. J. Loss Prev. Process. Ind. 40
industry. The proposed method (called as E-HAZOP) has been applied to [Link] ISSN 0950-4230.
a sour crude-oil processing-plant in order to identify health and safety Gul, M., Guven, B., Guneri, A.F., 2018. A new Fine-Kinney-based risk assessment
risks in the plant. framework using FAHP-FVIKOR incorporation. J. Loss Prev. Process. Ind. 53, 3–16.
[Link] ISSN 0950-4230.
The results prove that, the new method is appropriate for identifying
and ranking the most considerable hazards amid the identified lengthy

15
P.K. Marhavilas et al. Journal of Loss Prevention in the Process Industries 62 (2019) 103981

Guo, L., Kang, J., 2015. An extended HAZOP analysis approach with dynamic fault tree. Marhavilas, P.K., Koulouriotis, D.E., Mitrakas, C., 2011. On the development of a new
J. Loss Prev. Process. Ind. 38, 224–232. [Link] hybrid risk assessment process using occupational accidents’ data: application on the
Henderson, R.D., Dutta, S.P., 1992. Use of the analytic hierarchy process in ergonomic Greek Public Electric Power Provider. J. Loss Prev. Process Ind. 24 (5), 671–687.
analysis. Int. J. Ind. Ergon. 9 (4), 275–282. [Link] [Link] Elsevier.
90061-4. Mayes, T., Kilsby, D.C., 1989. The use of HAZOP hazard analysis to identify critical
Herbert, I.L., 2011. Learning the lessons - retrospective HAZOPs. In: SPE Offshore Europe control points for the microbiological safety of food. Food Qual. Prefer. 1 (2), 53–57.
Oil and Gas Conference and Exhibition. Aberdeen. Longman Group UK Ltd.
Høj, N.P., Kr€oger, W., 2002. Risk analyses of transportation on road and railway from a Mentzer, R.A., Zhang, J., Xu, W., Mannan, S.M., 2014. What does “safe” look and feel
European Perspective. Saf. Sci. 40 (1–4), 337–357. like? J. Loss Prev. Process. Ind. 32, 265–275.
IEC (International Electrotechnical Commission), 1995. International Standard 60300-3- Montague, D.F., 1990. Process risk evaluation - what method to use? Reliab. Eng. Syst.
9, Dependability Management – Part 3: Application Guide – Section 9. Risk analysis Saf. 29, 27–53.
of technological systems, Geneva, 1995. O Herrera, M.A.d.l., Luna, A.S., Costa, A.C.A.d., Lemes, E.M.B., 2018. Risk Analysis: A
IEC 61511, 2003. Functional Safety: Safety Instrumented Systems for the Process Sector. Generalized Hazop Methodology State-Of-The-Art, Applications, and Perspective in
International Electrotechnical Commission, Geneva, Switzerland. the Process Industry. [Link]
ISO/IEC Guide 51, 1999. Safety Aspects – Guidelines for Their Inclusion in Standards, nalysis_A_generalized_Hazop_methodology_stat.pdf. (Accessed 26 June 2019).
second ed. ISO/IEC, Geneva. 1999. Olsson, Fr., 1999. Tolerable Fire Risk Criteria for Hospitals. Report 3101, ISSN: 1402-
ISO/IEC Guide 73, 2009. Risk Management-Vocabulary. 3504, ISRN: LUTVDG/TVBB–3101—SE. Department of Fire Safety Engineering,
Jagtman, H.M., Hale, A.R., Heijer, T., 2005. A support tool for identifying evaluation Lund University, Lund, Sweden.
issues of road safety measures. Reliab. Eng. Syst. Saf. 90, 206–216. Othman, M.R., Idris, R., Hassim, M.H., Ibrahim, W.H.W., 2016. Prioritizing HAZOP
Jain, P., Rogers, W.J., Pasman, H.J., Keim, K.K., Mannan, M.S., 2018. A resilience-based analysis using analytic hierarchy process (AHP). Clean Technol. Environ. Policy 18
integrated process systems hazard analysis (RIPSHA) approach: Part I plant system (5), 1345–1360. [Link]
layer. Process Saf. Environ. Prot. 116, 92–105. [Link] Qureshi, A.R., 1988. The role of hazard and operability study in risk analysis of major
psep.2018.01.016. hazard plant. J. Loss Prev. Process. Ind. 1, 104–109.
Jain, P., Rogers, W.J., Pasman, H.J., Keim, K.K., Mannan, M.S., 2018. A resilience-based Reniers, G.L.L., Dullaert, W., Ale, B.J.M., Soudan, K., 2005. The use of current risk
integrated process systems hazard analysis (RIPSHA) approach: Part II management analysis tools evaluated towards preventing external domino accidents. J. Loss Prev.
system layer. Process Saf. Environ. Prot. 116, 115–124. [Link] Process. Ind. 18, 119–126.
psep.2018.06.037. Saaty, T.L., 1990. How to make a decision: the analytic hierarchy process. Eur. J. Oper.
Johnson, W.G., 1973. The Management Oversight and Risk Tree. Prepared for the U.S. Res. 48 (1), 9–26. [Link]
Atomic Energy Commission. Suokas, J., 1988. The role of safety analysis in accident prevention. Accid. Anal. Prev. 1
Kletz, T.A., 1997. HAZOP - past and future. Reliab. Eng. Syst. Saf. 55, 263–266. (20), 67–85.
Kletz, T., 1999. HAZOP and HAZAN: Identifying and Assessing Process Industy Hazards. Swann, C.D., Preston, M.L., 1995. Twenty five years of HAZOPs. J. Loss Prev. Process.
Institution of Chemical Engineers, Rugby. Ind. 8 (8), 349–354.
Koulinas, G.K., Marhavilas, P.K., Demesouka, O.E., Vavatsikos, A.P., Koulouriotis, D.E., Vinnem, J.E., 1998. Evaluation of methodology for QRA in offshore operations. Reliab.
2019. Risk analysis and assessment in the worksites using the fuzzy-analytical Eng. Syst. Saf. 61, 39–52.
hierarchy process and a quantitative technique – A case study for the greek Woodruff, J.M., 2005. Consequence and likelihood in risk estimation: a matter of balance
construction sector. Saf. Sci. 112, 96–104. [Link] in UK health and safety risk assessment practice. Saf. Sci. 43 (5–6), 345–353.
ssci.2018.10.017. Elsevier.
Lawley, H.G., 1974. Operability studies and hazard analysis. Chem. Eng. Prog. 4 (70),
105–116. Аbbreviations and αcronyms
Lawley, H.G., 1976. Size up plant hazards this way. Hydrocarb. Process. 55 (4), 247–261.
Lee, M., 2006. How Does Climate Change Affect the Assessment of Landslide Risk?
BDV: Blowdown Valve
[Link]
CCPS: Center for Chemical Process Safety
Lees, F.P., 1980. Loss Prevention in the Process Industries: Hazard Identification,
CPI: Chemical Process Industry
Assessment and Control. Reed, Oxford.
COPI: Crude-Oil Process Industry
Macdonald, D., Mackay, S., 2004. Practical Hazops, Trips and Alarms. Newnes, Oxford.
ESD: Emergency Shutdown
Marhavilas, P.K., 2015. Risk Assessment Techniques in the Worksites of Occupational
ESDV: Emergency Shutdown Valve
Health-Safety Systems with Emphasis on Industries and Constructions. PhD Thesis.
FTA: Fault Tree Analysis
Department of Production and Management Engineering, Democritus University of
LAHH: Level Alarm High High
Thrace, Xanthi. [Link] (Accessed 23 March
LALL: Level Alarm Low Low
2019).
P&IDs: Piping and Instrumentation Diagrams
Marhavilas, P.K., Koulouriotis, D.E., 2008. A risk estimation methodological framework
PFD’s: Process Flow Diagrams
using quantitative assessment techniques and real accidents’ data: application in an
PSV: Pressure Safety Valve
aluminum extrusion industry. J. Loss Prev. Process Ind. 21 (6), 596–603. [Link]
PHA: Preliminary Hazard Analysis
org/10.1016/[Link].2008.04.009. Elsevier.
PIC: Pressure Indicator Controller
Marhavilas, P.K., Koulouriotis, D.E., 2012. Developing a new alternative risk assessment
PPE: Personal Protective Equipment
framework in the work sites by including a stochastic and a deterministic process: a
PT: Pressure Transmitter
case study for the Greek Public electric power provider. Saf. Sci. 50 (3), 448–462.
PV: Pressure Valve
[Link] Elsevier.
SIL: Safety Integrity Level
Marhavilas, P.K., Koulouriotis, D.E., Gemeni, V., 2011. Risk analysis and assessment
SMS: Safety Management System
methodologies in the work sites: on a review, classification and comparative study of
the scientific literature of the period 2000-2009. J. Loss Prev. Process. Ind. 24 (5),
477–523. [Link] ISSN 0950-4230.

16