Water Resources Management (2021) 35:3221–3233

https://doi.org/10.1007/s11269-021-02886-4

Risks Ranking in a Desalination Plant Construction Project

with a Hybrid AHP, Risk Matrix, and Simulation‑Based
Approach

Georgios K. Koulinas1   · Alexandros S. Xanthopoulos1 · Konstantinos A. Sidas2 ·

Dimitrios E. Koulouriotis1

Received: 29 September 2020 / Accepted: 23 June 2021 / Published online: 1 July 2021
© The Author(s), under exclusive licence to Springer Nature B.V. 2021

Abstract
This paper proposes a hybrid process based on the Analytical Hierarchy Process (AHP),
risk matrix, and Monte Carlo Simulation for prioritizing delay risks in construction pro-
jects and estimating the probability of timely project completion. The main contribution
of this study is the development of a novel framework for risk assessment that quantifies
the impact of more critical risks to each activity according to its nature and benefits from
the advantages of the AHP for constructing an enhanced risk matrix. In addition to the
above, the proposed framework exploits the computational potential of Monte Carlo Simu-
lation to obtain more accurate predictions of a project’s completion time. The proposed
approach was used for analyzing the duration, expected delay, and the probability of com-
pletion within the deadline, of a vital seawater desalination plant construction project, on
the island of Alonissos, Greece. The proposed framework predicted the project duration
more accurately than the classic PERT method did, and the results of the analysis docu-
mented that it could be a simple tool providing risk managers with helpful information
while making decisions under uncertainty.

Keywords  Delay risks · Project risk management · Uncertainty · Monte Carlo simulation ·
Analytical hierarchy process

* Georgios K. Koulinas
gkoulina@pme.duth.gr
Alexandros S. Xanthopoulos
axanthop@pme.duth.gr
Konstantinos A. Sidas
std112649@ac.eap.gr
Dimitrios E. Koulouriotis
jimk@pme.duth.gr
1
Department of Production and Management Engineering, Democritus University of Thrace, 12
Vas. Sofias st, 67100 Xanthi, Greece
2
Department of Engineering Project Management, Faculty of Science & Technology, Hellenic Open
University, Parodos Aristotelous 18 St, 26335 Patra, Greece

13
Vol.:(0123456789)
3222 G. K. Koulinas et al.

1 Introduction

The lack of drinking water is a problem experienced by 30% of the people worldwide,
while about 20% of the world’s population live in areas where water is a scarce resource
(World Health Organization 2011). Given these challenges, the desalination process is
widely used for providing drinking water in such cases. In addition, desalination can be
applied to the water of varying salinity, and thus, it is a complicated chemical process. All
the above denote the increasing importance of desalination plant construction projects.
In practice, the total project duration can be extended due to uncertainty that can cause
delays and costs overruns. Thus, a process for devising more accurate actual makespan
predictions could be beneficial while constructing baseline schedules. This research aims
to introduce a framework that combines a multi-criteria decision-making method to rank
partial risks which affect the duration of specific activities. Next, each risk’s probability
of appearance and impact is estimated, and the corresponding risk matrix is constructed.
Finally, triangular distribution with varying upper and lower bounds are assigned to activi-
ties’ durations for performing Monte Carlo Simulation (MCS) experiments and predicting
the possibilities of in-time project execution.
Risk analysis in projects is one of the most challenging processes impacting the pro-
ject’s cost, time, and scope. Numerous studies have been proposed to assist time and cost
predictions during the execution of activities.
Initially, probabilistic approaches were proposed for risks’ impact cost and duration esti-
mation. (Barraza 2011) used MCS to estimate time contingency of the project activities
while allocating project allowances, while (Elshaer 2013) merged an earned schedule man-
agement framework with simulation-predicted duration for activities. Recently, (Bianco
et  al. 2019) proposed a stochastic model of the project scheduling problem under uncer-
tainty and applied a process that considered the Beta probability density function for the
activities’ durations.
In addition, many papers used the Monte Carlo Simulation approach for the estima-
tion of risks. More specifically, (Kirytopoulos et al. 2008) and (Yi et al. 2019) underlined
the necessity of using historical data while performing duration estimations with MCS,
while (Vanhoucke 2011) introduced novel tracking methods and for projects with uncer-
tain durations while applying Monte-Carlo simulation. In addition, (Wang et  al. 2016)
proposed a new algorithm and formulation for operational cost minimization during con-
structing a building energy system and used MCS for performance validation. The study
of (Ntzeremes and Kirytopoulos 2018) proposed a novel method for assessing fire risks in
underground road tunnels, and (Rausch et al. 2019) applied Monte Carlo simulation on a
prefabricated construction assembly that needs less rework during assembly fit-up during
offsite construction. An excellent survey of the papers that propose methods for handling
uncertainty in projects can be found in (Aven 2016). The recent study of (Koulinas et al.
2020) proposed an approach that incorporates the manager’s expertise with Monte Carlo
Simulation for making project makespan delays estimations.
Additionally, few papers studied risks that appeared in desalination plants, such as
the study of (Attarzadeh et al. 2011), monitored the risks of a desalination project, and
performed a matrix model for studying uncertainty impact on construction. The study
of (Zhang et  al. 2020) introduced a fuzzy comprehensive evaluation and AHP-based
method to perform a quantitative risk evaluation in large-scale desalination projects.
Also, (Bouamri and Bouabdesselam 2018) used the MADS MOSAR methods for risk
analysis on a reverse osmosis seawater desalination plant, while (Fathy Awwad 2018)

13
Risks Ranking in a Desalination Plant Construction Project… 3223

assessed risks in several reverse osmosis desalination plants for investigating the main
risks that could have an impact on these plants.
Furthermore, many papers applied AHP and Monte Carlo Simulation to analyze
risks, especially for performing sensitivity analysis to the judgments that the decision-
maker assigns to the problem’s elements. More specifically, in the study of (Dong et al.
2013), the AHP has been used within a geographical information systems environment
to classify influencing factors and make predictions about the utilization and exploi-
tation of risks of the third aquifer in Tianjin City, China. In addition, (Bamakan and
Dehghanimohammadabadi 2015) used AHP and MCS to assess the risks of informa-
tion security management systems. In the study of (Li et al. 2015), the authors selected
dominant factors of groundwater development and utilization risk based on the various
groundwater function zones and built a risk assessment system for shallow and deep
groundwater. In their proposed system, the weight of each evaluation index was calcu-
lated using the AHP method.
Moreover, (Albogamy and Dawood 2015) proposed a process using AHP and MCS for
client-based risk management applied at the early design stage of construction processes,
while (Helbig et al. 2016) applied AHP along with MCS to estimate supply risk values of
materials for the construction of thin-film photovoltaics. In addition, (Fu et al. 2017) devel-
oped a Monte Carlo-based fuzzy AHP approach for environmental risk factors analysis
during ship operations in Arctic waters. The study of (Tapia et al. 2017) proposed a joint
Analytic Hierarchy Process Data Envelopment Analysis framework for carbon dioxide uti-
lization options selection via a combination of data and expert beliefs to extract efficiency
scores. Then, a Monte Carlo simulation was applied to determine the impact of uncertainty
on the selections.
Additionally, (Luthra et al. 2018) performed AHP and MCS for risks analysis in a sus-
tainable supply chain framework, and (Yu et al. 2018) developed an AHP-based model for
assessing risks during the maintenance of onshore oil and gas transmission pipelines and
tackled uncertain conditions using MCS. Additionally, (Dweiri et al. 2018) applied AHP
for location criteria ranking of a new desalination plant, and (Negahban 2018) proposed a
sampling-optimization-adjustment approach applied within an MCS AHP-based approach
to allow handling situations where inconsistent matrices constructed with sampling from
the pairwise comparison distributions. The developed approach is used to treat a real layout
design selection problem of an industrial facility.
Recently, (Kim et al. 2019) proposed a model for cost and schedule performance predic-
tions in offshore engineering, procurement, and construction mega-projects, which uses a
fuzzy inference system, the AHP, and MCS for investigating for possible cost overrun. In
another contemporary research (Zhong et al. 2019) combined the Cloud-based Information
Diffusion (CID) model and the AHP to develop an innovative approach for assessing flood
risks. This approach converts crisp-numbered observations to fuzzy values for improving
the information diffusion function.
The study of (Afzal et al. 2020) developed a framework based on fuzzy systems, ana-
lytical hierarchy process, and Monte Carlo simulation to identify and treat cost overrun
problems in transit projects. (Baghapour et al. 2020) proposed a novel methodology that
integrates a flexible water quality index and four multi-criteria decision-making meth-
ods (fuzzy AHP, TOPSIS, fuzzy TOPSIS, and fuzzy OWA) for dealing with uncertain-
ties regarding missing data while performing assessments. Also, (Spanidis et  al. 2021)
used a framework based on the Fuzzy Analytical Hierarchy Process, Expected Value (EV)
function, Monte Carlo simulation, and PERT analysis to assess hazards in mining projects
using the (FAHP) and predict the time and cost overruns.

13
3224 G. K. Koulinas et al.

The remainder of the paper consists of the following sections: (3) the proposed
approach, (4) the description of the application to a construction project, and (5) the dis-
cussion and conclusions section.

2 The Proposed Approach

The flowchart of the proposed methodology is shown in Fig. 1. Initially, the risks affect-
ing the execution of every activity are identified through brainstorming and search on the
comparative literature. Then, the pairwise comparison matrices are filled for each task, the
weights for each risk are extracted using the typical AHP process, and the ranking extracted
with descending risk weight order. After the most critical risk for every task is identified,
the risk manager estimates its possibility of appearance while executing the specific activ-
ity under study and its impact on the task’s duration. Then, a risk matrix is constructed, and
the risk manager estimates the optimistic, pessimistic, and most likely durations for every
task. These values are inputs to the Monte Carlo Simulation process, which is executed for
an adequate number of independent replications. The simulation scenarios are reviewed,
and the manager derives conclusions about whether the level of risk is tolerable. If the
total risk level is found to be acceptable, the process terminates; otherwise, risk mitiga-
tion measures are performed, and the process returns to the step of risk identification. It
is worth mentioning that the proposed process could be generalized and used in differ-
ent cases simply by inserting different judgments in the pairwise comparison matrices and
adjusting the duration deviation ranges, the distribution used, and the risks included in the
analysis.

2.1 Risks Identification

Initially, a scientific literature review was performed as well as brainstorming sessions with
the expert risk manager to identify the particular project’s risk factors. The first employed
for extracting an informed risk list, and the latter to further analyze and discuss with the
expert the list and add or delete risks according to his experience and knowledge. Given
that not all risk factors have a potential impact on all tasks, the manager considers a risk list

Idenfy risks for Apply AHP to rank risks

START Construct risk matrix
acvies for each acvity

Perform measures for Assign ranges to

reducing risks duraons’ distribuons

NO

YES
Is total risk Apply Monte Carlo
FINISH Extract conclusions
acceptable? Simulaon

Fig. 1  The flowchart of the proposed approach

13
Risks Ranking in a Desalination Plant Construction Project… 3225

for each task containing the factors affecting the correct completion of the specific activity.
This step’s output is used to construct the pairwise comparison matrices for each task to
extract its most important risk factor.

2.2 Ranking Risks with the Typical AHP

The typical AHP, initially proposed by (Saaty 1990), is a well-known and classic multi-cri-
teria decision-making method. The method organizes the decision-making problem into a
hierarchy, with at least two levels, and compares the problem’s elements one-to-one, using
a pairwise comparison matrix for each hierarchy level. The output is a weight value for
each element, one concerning the immediate upper-level element (local priority) and one
for the ultimate goal of the analysis (global priority). Such problems have a primary goal
­(1st level of the hierarchy), which is analyzed in criteria (­2nd level of the hierarchy), and
each criterion has many subcriteria ­(3rd level of the hierarchy). The input data of the AHP
method are the decision maker’s beliefs regarding the criteria, compared to one another,
which are expressed via a pairwise comparison matrix filled by the decision-maker. The
decision-maker can express the relative importance of the compared factors in a pairwise
comparison matrix using a standard scale imposed by the typical AHP (Saaty 1990). This
scale assigns linguistic variables to numerical values. More specifically:

• the value “1” is assigned when the task Ti and Tj are equally important,
• the value “3” when Ti is moderately more critical than Tj,
• the value “5” when Ti is strongly more important than Tj , the value “7” when Ti is very
strongly more important than Tj , and
• the value “9” when Ti is extremely more important than Tj.

Note that the intermediate values "2, 4, 6, 8" are used when compromise is needed. The
consistency of the decision maker’s judgments is checked by computing the Consistency
Ratio (CR) to validate the quality of the results. Note that the consistency index value for
random decision maker’s judgments, according to the order of a random pairwise compari-
son matrix (Saaty 1990), is 0 if the order is 2, 0.58 if the order is 3, 0.9 if the order is 4,
1.12 if the order is 5, 1.24 if the order is 6, and 1.32 if the order is 7. In this paper, we use
typical AHP for ranking risks of each activity, and then, for the most important risk, we
make estimations about the possibility of appearance and impact on the activity’s duration.

2.3 Risk Matrix Construction

After considering each activity’s most critical risk factor, the risk manager proceeds to
construct the probability-impact matrix for all top-rated risks. This step is undertaken to
estimate the possibility for a specific risk to be realized and to examine how the duration
of a task is affected due to the impact of this risk. As referred in (PMI 2017), such a matrix
provides combinations of occurrence probabilities and impact on the project (or on an
activity) for all (or even a subset of) the risk factors. Then, the risks are categorized accord-
ing to the value (Probability * Impact) that the risk manager gives to the most critical risk
of each task. The rating rules of the present case are described in the following Table 1.
More specifically, in the red area, high-risk factors are included; yellow represents a
medium risk, while the green area represents factors of the low-risk level.

13
3226 G. K. Koulinas et al.

Table 1  The rules for the construction of the risk matrix

Probability
Linguistic
of Value Probability * Impact ( ∗ )
variable
occurrence
Very High >90% 0.90 0.09 0.27 0.45 0.63 0.81
High 60-90% 0.70 0.07 0.21 0.35 0.49 0.63
Medium 30%-60% 0.50 0.05 0.15 0.25 0.35 0.45
Low 11-30% 0.30 0.03 0.09 0.15 0.21 0.27
Very Low <=10% 0.10 0.01 0.05 0.05 0.07 0.09
0.10 0.30 0.50 0.70 0.90
<10% 11-30% 31%-60% 61%-90% >90%
Very Low Low Medium High Very High
Risk Impact on time

2.4 Duration Deviation Range Estimation and Distribution Assignment

According to the categorization of each activity’s risk (and of its corresponding basic
risk), a range is assigned to express the delay estimation, and a triangular distribution
is considered for representing the activities’ duration. These are inputs to the Monte
Carlo Simulation process for total project makespan prediction. In this study, we
assigned a maximum of 10% delay if the risk belongs to the green area, a 20% maxi-
mum delay if the risk is considered in the yellow area, and a maximum of 40% delay if
the risk belongs to the red area. Note that this categorization and assignments are con-
sidered for the present case and that it could be modified in another project example.
Since there are no historical data about the actual durations for each activity of the
specific project, we selected the triangular distribution to represent activities’ dura-
tions. The rationale behind this choice is that this distribution can effectively dem-
onstrate the continuous uncertainty in such a project. The only required data is the
estimate of the most likely duration value, illustrating the base case scenario, and a
maximum and minimum duration value as the upper and lower bound, respectively. In
addition, the triangular distribution is frequently selected by risk analysts since it can
represent the uncertainty in a general way for such cases that there are no additional
data available, apart from the expertise and experience of the risk manager. (Rees
2015).

2.5 The Monte Carlo Simulation Process

The Monte Carlo Simulation is a classic sensitivity analysis technique. It comprises of

choosing a statistical distribution to represent each risk and running multiple iterations
using an equal number of random values for durations that belong to the prescribed
distribution. Every iteration refers to constructing a different schedule for the project
accompanied by the respective total duration estimate. After completing the total num-
ber of runs, the possibility of executing the project within a specific deadline is esti-
mated, and the descriptive statistic extracted can be utilized for estimating the total
risk level tolerance of a task.

13
Risks Ranking in a Desalination Plant Construction Project… 3227

3 Application in a Construction Project in Greece

The proposed process is used to analyze the duration risks of a desalination plant construc-
tion project. The desalination process will process the seawater to produce drinking water
daily and dispense it to Alonissos Municipality, which will be the project owner and the
contracting authority. All the necessary configurations of the surrounding area for support-
ing the installation and safe operation of water treatment plants are the responsibility of the
contractor construction company. The seawater treatment would be carried out through a
desalination unit with a nominal capacity of 600 ­m3 of drinking water per day. The mini-
mum amount of treated drinking water supplied to the island amounts to 160,000 m ­ 3 per
3
annum, for the cost of 1.44 € / ­m of desalinated water plus VAT.
The project aims to ensure water supply in Alonissos island, which is the responsibility
of Alonissos Municipality. It worth noting that the water supply, today, is quantitatively
marginal. The provision of a sufficient quantity of high-quality drinking water, with a con-
trolled rate of supply to the distribution system, is necessary to meet the peak demand dur-
ing the summer for the main settlements of Alonissos. Also, the unit’s function is critical
during the winter due to the degradation of the available water from the boreholes and
water sources of the network (salinization). The timely completion of the project is crucial
for the development of the island, the upgrade of the inhabitants’ quality of life, and the
environment’s protection. According to the contract, the unit must start operation no later
than 11/1/2021. A short description of the desalination plant, the sequence of works, and
the organization of facilities are presented in the remainder of this section.
For the seawater supply of the plant, an existing borehole is used. The produced drinking
water is channeled from the unit’s exit to a tank located near the installation site. The desal-
ination unit will be installed inside a container of indicative dimensions 12 × 2.5 × 2.5 m.
Among the currently commercially available desalination methods, reverse osmosis is pro-
posed due to the continuous technological improvements of the membranes, the unavaila-
bility of thermal energy, the simplicity of operation but mainly the cost of water production
for the considered supply (600 ­m3/day). The discarded water from the units (condensate)
will be disposed of at appropriate points indicated by Alonissos Municipality.
The contract considers the project start to be on 1/9/2020 and a total duration of
90 days. Based on the Critical Path Method, the project ends on 04/01/2021. Note that a
working week of five days has been considered, with eight working hours, and that October
­28th, December ­25th, January ­1st, and January ­6th are considered to be holidays. In addition,
the contract imposes the "hard constraint" of completing the project and commencing the
water desalination production at most by 10/1/21 (i.e., a maximum of 95 project days).
After that date, financial penalties were activated. Given that, without performing risk anal-
ysis, the manager (according to CPM) has a 7-day time buffer before meeting the “hard”
deadline. Table 2 illustrates the main data of the project that were input to the Monte Carlo
Simulation process.
According to the process illustrated in Fig.  1, firstly, we have identified the main risk
factors that may affect any activity of the current project, and then we assigned several
risks to each activity. The complete risk list is shown in Table 3.
Next, the AHP is applied for ranking risks according to their importance of impact. For
each activity’s risks, a hierarchy is constructed, containing the goal, the total risk of the
current activity, and the criteria represented by the risks affecting the given activity. As
identified by the expert, the risks coupled with the respective impact on each activity are
shown in Table 4. These risks are used to construct a pairwise comparison matrix in which

13
3228 G. K. Koulinas et al.

Table 2  The project data

ID Activity Duration Predecessors

A1 Start 0
A2 Project contract—insurance 5 1
A3 Construction site installation 5 2
A4 Configurations and access to the unit installation improvements 15 3
A5 Water intake drilling works 15 3
A6 Seawater supply network construction works 21 3
A7 The electricity supply network works 85 3
A8 Brine disposal and membranes leachate disposal network construction 6 4,5
A9 Drinking water to the water tanks transmission network construction 15 6,8
A10 Support base construction works 10 9
A11 Supply and assembly of equipment 15 10
A12 Transportation of unit and equipment (sea and land) 4 11
A13 Installation of the unit and connections to the rest networks 10 12
A14 Landscaping and completion works 10 12
A15 Finish 0 7,13,14

the decision-maker—risk manager makes judgments about the importance of each risk
(criterion) compared to all other risks of a given task. Columns 3–9 of Table 4 present the
judgments of the manager using the standard scale of the AHP method.
Using these pairwise comparisons, we compute the ranking of each task’s risks, and the
most important risk affecting each activity is determined. In addition, the consistency of
the manager’s judgments is monitored via the consistency ratio index (CR) computed for
every pairwise comparison matrix. Table 4 illustrates the CR, the corresponding rankings,
and the priorities (weights-scores) for each risk. Note that all matrices have a CR value less
than the thresholds described in Sect. 4.2, which means that the judgments were consistent.
After considering the most affecting risk for every task, the “Probability” and “Impact”
levels are determined according to the rules of Table  3 for each separated activity. The
result from the multiplication categorizes each risk (and each task) to the corresponding
area of the risk matrix. The red color means a risk/task of "High" risk, the yellow means a
risk/task of "Medium" risk, and the green color represents a "Low" level risk/task. In this
analysis, we have only tasks belonging to the red and yellow areas. According to a task’s
class of risk, the manager considers a duration range, namely, if the task is characterized as
of “High” risk, the optimistic value of the duration is considered equal to the initial (most
likely) duration, and the pessimistic value is equal to a 40% greater than the initial one. In
the case of the "Medium" risk level, the pessimistic value is equal to 120% of the initial,
and in "Low" risk tasks, it is equal to 110% of the CPM duration. The optimistic value
remains equal to 100% of the most likely duration. For example, activity A2 is classified to
the red area of the risk matrix, and its optimistic value is equal to the initial (5 days), while
its pessimistic value is 140% of the initial (7 days). In such a way, the columns "Optimis-
tic Duration" and "Pessimistic Duration" of Table 4 are filled in. These estimations, along
with the initial duration of each task, are input to the triangular distribution for perform-
ing 1000 iterations of the Monte Carlo Simulation process. The Monte Carlo Simulation
was implemented in the Visual Basic for Applications programming language within the
Microsoft Excel 2016 software environment.

13
Risks Ranking in a Desalination Plant Construction Project… 3229

Table 3  The risk registry for the specific project

Risk ID Description Risk ID Description

R1 Possible changes in laws and regula- R10 Possible poor coordination between Project
tions for the licensing of units and Owner, Supervisor, and Contractor (and
public procurement Technical Consultant, if any)
R2 Possible delay due to protest of R11 Delays in electricity supply
residents (due to location of the
unit, water costs, noise from an
operation, or other reasons)
R3 Lack of specialized materials R12 Possible inability to find suitable quality and
quantity of seawater (drilling)
R4 Inability to transport materials to an R13 Possible lack of technological knowledge,
island area skills, applied techniques for the implemen-
tation of work
R5 Unable to find a suitable brine R14 Delays in issuing the necessary licenses
disposal site
R6 Possible perceived environmental R15 Potential difficulty in finding suitable land
impacts from the project for the unit’s installation (accessibility,
closeness to the sea, drilling sites, municipal
reservoir sites, noise—no nearby residences
or tourist facilities, etc.)
R7 Degradation of the environment in R16 Lack of provisions in the contract (for delays,
the installation area water quality, controls, analyzes, etc.)
R8 Lack of state funding for this area R17 Delays during the licensing procedure
R9 Lack of experience and know- R18 Serious work accident
how in the licensing and auction
procedures

After performing the simulation iterations, the minimum (95.62), maximum (114.18),
and mean value (102.01) for the total project duration were computed, along with the stand-
ard deviation (4.09) of the makespan. Furthermore, the sensitivity analysis showed that
there is no possibility for the project to be completed within the initial CPM duration of
95 days (namely, end by 11/1/21). Moreover, there is only a 38.5% possibility that the pro-
ject will last 100 days, and it is predicted that the makespan will be less than 109.47 days,
with a 95% probability. Also, the percentiles (5%, 10%, 15%, 20%, 25%, 30%, 35%,40%,
45%, 50%, 55%, 60%, 70%, 75%, 80%, 85%, 90%) were found to be (96.71, 97.12, 97.61,
98.15, 98.60, 99.06, 99.66, 100.14, 100.83, 101.24, 101.81, 102.62, 103.16, 104.13, 104.86,
105.70, 106.78), respectively. From these results, it is concluded that the applied approach
makes clear the fact that the project is of high risk for not completing on time. Given the
importance of the desalination plant for the local economy of the island of Alonissos, this
could have serious negative consequences for the construction company. Note that we per-
formed the classic PERT method to the present case example for comparison purposes and
found a 33.57% probability for the project to end within 95 days and a probability of 57.21%
to end the project within 100 days. These results could easily mislead the risk manager to
act as if the project is exposed to less risk when, in fact, it is not. This difference between
the proposed method and PERT is attributed to the beta distribution used by the latter to
represent the duration of the activities. At the same time, the pessimistic and optimistic
value arises empirically without taking into account the risks related to the specific activity.
Therefore, it is considered that the proposed method helps the risk manager to make more
accurate assessments and ultimately more efficient risk management.

13
3230 G. K. Koulinas et al.

Table 4  The risk list, judgments, and results for each activity of the project
Optimistic Pessimistic
Task ID Risk ID Pairwise comparison matrix Score Ranking Probability Impact ∗
Duration Duration
A2 R8 R1 R9 R17 R15 R14 R16 5 7
R8 1 5 3 5 2 2 5 0.34 1 VH (0.9) H (0.7) 0.63
R1 1 1 1 1 1 1 0.09 4
R9 1 1 3 3 3 0.18 2
R17 1 1 1 1 0.09 4
R15 1 0.2 1 0.08 5
R14 1 3 0.15 3
R16 1 0.07 6
A3 R2 R10 5 7
VH
R2 1 7 0.88 1 VH (0.9) 0.81
(0.9)
R10 1 0.12 2
A4 R4 R6 R18 15 18
R4 1 5 3 0.66 1 M (0.5) M (0.5) 0.25
R6 1 1 0.15 3
R18 1 0.19 2
A5 R12 R13 R4 R6 15 18
R12 1 3 ½ 1 0.25 2
R13 1 ¼ 1 0.13 4
R4 1 1 0.39 1 H (0.7) M (0.5) 0.35
R6 1 0.23 3
A6 R4 R3 R6 21 25,2
R4 1 5 3 0.64 1 H (0.7) M (0.5) 0.35
R3 1 1/3 0.10 3
R6 1 0.26 2
A7 R5 R4 R3 R6 R7 85 102
R5 1 1/2 1/2 1 ½ 0.13 5
R4 1 1/2 1/2 ½ 0.14 4
R3 1 2 2 0.32 1 M (0.5) M (0.5) 0.25
R6 1 2 0.22 2
R7 1 0.19 3
A8 R4 R3 6 7,2
R4 1 2 0.67 1 H (0.7) M (0.5) 0.35
R3 1 0.33 2
A9 R4 R6 15 18
R4 1 5 0.83 1 H (0.7) M (0.5) 0.35
R6 1 0.17 2
A10 R11 1 VH (0.9) H (0.7) 0.63 10 14
A11 R3 R13 15 18
R3 1 ½ 0.33 2
R13 1 0.67 1 H (0.7) M (0.5) 0.35
A12 R4 1 H(0.7) H (0.5) 0.35 4 5.6
A13 R11 1 VH (0.9) H (0.7) 0.63 10 14
A14 R4 R6 R7 10 12
R4 1 5 5 0.71 1 M(0.5) M(0.5) 0.25
R6 1 1 0.145 2
R7 1 0.145 2

Apart from the calculations regarding the probability of timely project completion,
a separate analysis was conducted to determine the effect of each activity on the project
makespan. For the given mean baseline duration of 102.01 days, the task “Electricity sup-
ply network works” proved to be the most influential, causing the total duration to range
from 96.84 to 109.89 days, probably because of its initial large duration and of the amount
of uncertainty that it includes. The other tasks cause lighter deviations while the minimum
project duration (100.91) is relatively more significant than the one caused by the "Elec-
tricity supply network works," meaning that the risk manager must spend budget for miti-
gating risks in this task to achieve improved total project duration, and then manage risks
in the other activities.

4 Discussion and Conclusions

The experimental results led to several interesting conclusions. Initially, the expert easily
understood the proposed approach, which could favor the method’s efficiency, owing to
the AHP’s simplicity when inserting judgments and performing calculations for extract-
ing alternatives’ priorities, Also, the output of AHP, which is a ranking, could be consid-
ered as a list of priorities by risk managers when allocating budget since this is precisely
their target. Note that the reliability of the ranking is verified by the consistency ratio of

13
Risks Ranking in a Desalination Plant Construction Project… 3231

each pairwise comparison matrix. The proposed framework comprises a compact, fast, and
efficient tool for risk assessment that entails relatively low computational effort. Further-
more, it is observed that the riskiest activity can cause project duration variation of about
13.05 days, while the rest of the activities cause 2.41 days or less. This could be an indica-
tion of which risks should be addressed as a matter of priority.
This article proposed a hybrid approach that combines the AHP method, risk matrix, and
Monte Carlo Simulation, applied to estimate delay risks of a desalination plant construction
project. The AHP method successfully quantifies the expert’s knowledge, experience, and
value system while performing risk assessment and prioritization. The resulting rankings are
used to apply a comprehensive sensitivity analysis and make predictions about the expected
actual project completion time. The sensitivity analysis proved that there is no possibility
for the project to be completed within the initial CPM duration. The proposed approach was
compared to the classic PERT technique, and it was found to outperform it since it can pre-
dict duration deviations of activities more accurately. The current approach assists risk man-
agers in making informed decisions regarding the budget allocation and application of risk
mitigation actions after a specific risk arises and impacts the project. In addition, the present
framework could then be used as the basis for the development of the integrated platform for
joined time, cost, resources availabilities, and safety risks assessment.
The core contribution of this paper is the introduction of a process that combines
an effective multi-criteria method (AHP) for risk ranking with a classic risk classifica-
tion approach (risk matrix) and a robust sensitivity analysis tool (Monte Carlo Simula-
tion). Despite the promising results, this framework could be extended by incorporat-
ing fuzzy sets theory and provide support to a group of decision-makers. Furthermore,
future studies could include more intensive research on which risk factors impact spe-
cific activities, and extensions to risk factors impacting project costs would be interest-
ing enough to investigate.

Author Contributions Conceptualization, G.Koulinas, D.Koulouriotis; data curation, G.Koulinas,

A.Xanthopoulos, K.Sidas; formal analysis, G.Koulinas, K.Sidas, D.Koulouriotis; investigation, G.Koulinas,
K.Sidas; methodology, G.Koulinas, A.Xanthopoulos, D.Koulouriotis; writing—original draft, G.Koulinas;
writing—review and editing, G.Koulinas, A.Xanthopoulos, D.Koulouriotis.
All authors have read and agreed to the published version of the manuscript.

Funding  This research received no external funding.

Data Availability  Some or all data, models, or codes that support the findings of this study are available
from the corresponding author upon reasonable request.

Declarations 

Conflicts of Interest  We declare no conflict of interest.

References
Afzal F, Yunfei S, Junaid D, Hanif MS (2020) Cost-risk contingency framework for managing cost over-
run in metropolitan projects: using fuzzy-AHP and simulation. Int J Manag Proj Bus 13:1121–
1139. https://​doi.​org/​10.​1108/​IJMPB-​07-​2019-​0175

13
3232 G. K. Koulinas et al.

Albogamy A, Dawood N (2015) Development of a client-based risk management methodology for the
early design stage of construction processes. Eng Constr Archit Manag 22:493–515. https://​doi.​org/​
10.​1108/​ECAM-​07-​2014-​0096
Attarzadeh M, Kim Huat Chua D, Beer M (2011) Risk Management of Asalouye Desalination Pro-
ject. In: First International Symposium on Uncertainty Modeling and Analysis and Management
(ICVRAM 2011); and Fifth International Symposium on Uncertainty Modeling and Analysis
(ISUMA). Hyattsville, Maryland, United States, pp 805–812
Aven T (2016) Risk assessment and risk management: Review of recent advances on their foundation.
Eur J Oper Res 253:1–13. https://​doi.​org/​10.​1016/j.​ejor.​2015.​12.​023
Baghapour MA, Shooshtarian MR, Zarghami M (2020) Process Mining Approach of a New Water Qual-
ity Index for Long-Term Assessment under Uncertainty Using Consensus-Based Fuzzy Decision
Support System. Water Resour Manag 34:1155–1172. https://​doi.​org/​10.​1007/​s11269-​020-​02489-5
Bamakan SMH, Dehghanimohammadabadi M (2015) A Weighted monte carlo simulation approach to
risk assessment of information security management system. Int J Enterp Inf Syst 11:63–78. https://​
doi.​org/​10.​4018/​IJEIS.​20151​00103
Barraza GA (2011) Probabilistic Estimation and Allocation of Project Time Contingency. J Constr Eng
Manag 137:259–265. https://​doi.​org/​10.​1061/​(ASCE)​CO.​1943-​7862.​00002​80
Bianco L, Caramia M, Giordani S (2019) A chance constrained optimization approach for resource
unconstrained project scheduling with uncertainty in activity execution intensity. Comput Ind Eng
128:831–836. https://​doi.​org/​10.​1016/j.​cie.​2018.​11.​053
Bouamri M, Bouabdesselam H (2018) Risk analysis in seawater desalination sector: a case study of Beni
Saf Water Company “BWC. Adv Syst Sci Appl 18:93–106. https://​doi.​org/​10.​25728/​assa.​2018.​
18.2.​531
Dong D, Sun W, Zhu Z et  al (2013) Groundwater Risk Assessment of the Third Aquifer in Tianjin City,
China. Water Resour Manag 27:3179–3190. https://​doi.​org/​10.​1007/​s11269-​013-​0342-z
Dweiri F, Khan SA, Almulla A (2018) A multi-criteria decision support system to rank sustainable desalina-
tion plant location criteria. Desalination 444:26–34. https://​doi.​org/​10.​1016/j.​desal.​2018.​07.​007
Elshaer R (2013) Impact of sensitivity information on the prediction of project’s duration using earned
schedule method. Int J Proj Manag 31:579–588. https://​doi.​org/​10.​1016/j.​ijpro​man.​2012.​10.​006
Fathy Awwad A (2018) Risk Assessment and Control for Main Hazards in Reverse Osmosis Desalination
Plants. Ind Eng 2:1. https://​doi.​org/​10.​11648/j.​ie.​20180​201.​11
Fu S, Zhang D, Zhang M, Yan X (2017) Identification of environmental risk influencing factors for ship
operations in Arctic waters. Harbin Gongcheng Daxue Xuebao/Journal Harbin Eng Univ 38:1682–
1688. https://​doi.​org/​10.​11990/​jheu.​20160​6050
Helbig C, Bradshaw AM, Kolotzek C et al (2016) Supply risks associated with CdTe and CIGS thin-film
photovoltaics. Appl Energy 178:422–433. https://​doi.​org/​10.​1016/j.​apene​rgy.​2016.​06.​102
Kim MH, Lee EB, Choi HS (2019) A forecast and mitigation model of construction performance by assess-
ing detailed engineering maturity at key milestones for offshore EPC mega-projects. Sustain 11.
https://​doi.​org/​10.​3390/​su110​51256
Kirytopoulos KA, Leopoulos VN, Diamantas VK (2008) PERT vs. Monte Carlo Simulation along with
the suitable distribution effect. Int J Proj Organ Manag 1:24–46. https://​doi.​org/​10.​1504/​IJPOM.​2008.​
020027
Koulinas GK, Xanthopoulos AS, Tsilipiras TT, Koulouriotis DE (2020) Schedule Delay Risk Analysis in
Construction Projects with a Simulation-Based Expert System. Buildings 10. https://​doi.​org/​10.​3390/​
build​ings1​00801​34
Li F, Zhao Y, Feng P et al (2015) Risk Assessment of Groundwater and its Application. Part I: Risk Grading
Based on the Functional Zoning of Groundwater. Water Resour Manag 29:2697–2714. https://​doi.​org/​
10.​1007/​s11269-​015-​0964-4
Luthra S, Mangla SK, Venkatesh VG, Jakhar SK (2018) Management of risks in sustainable supply chain
using AHP and monte carlo simulation. Global Business Expansion: Concepts, Methodologies, Tools,
and Appl. Government Engineering College Nilokheri, India, pp 1633–1652
Negahban A (2018) Optimizing consistency improvement of positive reciprocal matrices with implications
for Monte Carlo Analytic Hierarchy Process. Comput Ind Eng 124:113–124. https://​doi.​org/​10.​1016/j.​
cie.​2018.​07.​018
Ntzeremes P, Kirytopoulos K (2018) Applying a stochastic-based approach for developing a quantitative
risk assessment method on the fire safety of underground road tunnels. Tunn Undergr Sp Technol
81:619–631. https://​doi.​org/​10.​1016/j.​tust.​2018.​08.​020
Organization WH (2011) Safe Drinking-water from Desalination. 1–34
PMI (2017) PMBOK Guide – Sixth Edition

13
Risks Ranking in a Desalination Plant Construction Project… 3233

Rausch C, Nahangi M, Haas C, Liang W (2019) Monte Carlo simulation for tolerance analysis in prefabri-
cation and offsite construction. Autom Constr 103:300–314. https://​doi.​org/​10.​1016/j.​autcon.​2019.​03.​
026
Rees M (2015) Business Risk and Simulation Modelling in Practice Using Excel, VBA and @RISK. John
Wiley & Sons Ltd
Saaty TL (1990) How to make a decision: The analytic hierarchy process. Eur J Oper Res 48:9–26. https://​
doi.​org/​10.​1016/​0377-​2217(90)​90057-I
Spanidis PM, Roumpos C, Pavloudakis F (2021) A fuzzy-ahp methodology for planning the risk management
of natural hazards in surface mining projects. Sustain 13:1–23. https://​doi.​org/​10.​3390/​su130​42369
Tapia JFD, Promentilla MAB, Tseng ML, Tan RR (2017) Screening of carbon dioxide utilization options
using hybrid Analytic Hierarchy Process-Data Envelopment Analysis method. J Clean Prod 165:1361–
1370. https://​doi.​org/​10.​1016/j.​jclep​ro.​2017.​07.​182
Vanhoucke M (2011) On the dynamic use of project performance and schedule risk information during pro-
ject tracking. Omega 39:416–426. https://​doi.​org/​10.​1016/j.​omega.​2010.​09.​006
Wang C, Jiao B, Guo L et al (2016) Robust scheduling of building energy system under uncertainty. Appl
Energy 167:366–376. https://​doi.​org/​10.​1016/j.​apene​rgy.​2015.​09.​070
Yi D, Lee EB, Ahn J (2019) Onshore Oil and Gas Design Schedule Management Process Through Time-
Impact Simulations Analyses. Sustain. 11
Yu X, Liang W, Zhang L et al (2018) Risk assessment of the maintenance process for onshore oil and gas
transmission pipelines under uncertainty. Reliab Eng Syst Saf 177:50–67. https://​doi.​org/​10.​1016/j.​
ress.​2018.​05.​001
Zhang Y, Wang R, Huang P et al. (2020) Risk evaluation of large-scale seawater desalination projects based
on an integrated fuzzy comprehensive evaluation and analytic hierarchy process method. Desalination
478:114286. https://​doi.​org/​10.​1016/j.​desal.​2019.​114286
Zhong M, Wang J, Gao L et  al (2019) Fuzzy Risk Assessment of Flash Floods Using a Cloud-Based
Information Diffusion Approach. Water Resour Manag 33:2537–2553. https://​doi.​org/​10.​1007/​
s11269-​019-​02266-z

Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

13