Introduction

Over the years, various risk factors of construction delays in the construction industry have been reviewed from many articles [1-11]. An assessment of risk in conservation refurbishment projects in a deterministic model was conducted [12] . The refurbishment of the oil tank project activities contains complicated activities such as dismantling the existing structures then constructing the new ones which are performed in a small narrow area.

Limited articles have simulated the risk assessment for the crude oil tank refurbishment projects in the light of qualitative and stochastic approaches using risk analyses application packages [12,13]. Hence, it is a challenge to identify, assess and simulate risks of delays in the oil tank refurbishment projects based on the stochastic methods for better understanding managing of risks in the area of the oil industry around the world. The simulation of stochastic models was developed for estimating and predicting a range of output values under certain operational conditions suits to industrial purposes under uncertainty [14].

These article objectives were to; (i) identify risk factors causing project delay, (ii) establish a list of risk indexes, and (iii) simulate the risks using Monte Carlo simulations as well as the sensitivity risks.

The refurbishment of the existing crude oil tank project or Free Water Knock out (FWKO) project in Duri oil gathering station, Sumatra Island, Indonesia was taken as a case study in this article. There were more than 50 similar crude oil tanks in need to be repaired and replaced shortly in Indonesia [15]. This article is expected may assist the project owner and constructors in developing a strategic decision (during project planning phase) for analyses risks in the similar FWKO projects in the future.

This study performed risk analyses procedures as follow; (i) risk identification for establishing a risks list, (ii) risk assessment for quantifying risks, and (iii) risk analyses Figure 1 [16].

<bold id="bold-cdeaaf97206a9f6c44c4e0567a676058">Figure 1.</bold> Three Stages of Risk Analyses Procedure

The risk identification was performed by distributing a list of questioners to the FWKO stakeholders. The lists of questionnaires were then analyzed utilizing Likert scales for measuring the probability magnitude of; (i) the risk frequencies, and (ii) risk impacts [17]. It is not necessary the frequency directly proportional to the impact, for example, the impact of lack of work safety procedures during construction was relatively high, but the frequency to occur was relatively so-so. The risk frequency and impact scales were also summarized in Table 1.

<bold id="bold-3bdb575526a1601dc7eef8e3c62f0505">Table 1</bold>. Risk Frequency and Impact

Scale Value Frequency (A) Impact (B)
1 0.00 Never occur Very low
2 0.25 Rarely Low
3 0.50 Frequent Moderate
4 0.75 Very frequent High
5 1.00 Absolutely occur Very high

Then, risk assessment were performed to classified risk according to their index. The risk index was classified into three simple main categories [18], as follow:

· The low-risk category was calculated as a very low to low relative risk importance index (Figure 8). The low risk is considered insignificant to cause project delay. This is because the frequency of the risk occurrences was relatively low and so does the impact.

· The medium-risk category is considered as either its frequency or impact is relatively moderate or high. It is considered to take any necessary plans for controlling the risks (as long as there are adequate resources provided for mitigating the risks); and

· The high-risk category is calculated as the frequency of the risk occurrence and the impacts are relatively high or very high. It is recommended to perform risk mitigation procedures for minimizing the potential risk occurrences [16].

A sensitivity risk analysis was also performed to describe uncertainties in the output model which were affected by the different uncertainties in the input model in the form of tornado diagram [19,20]. The sensitivity analyses are commonly performed for estimating a range of probable project schedules or revenue generations, and other feasibility study parameters [21].

The application of @risk is an add-in to Microsoft Excel was utilized for simulating this case study. This application could simulate and analyze risk using Monte Carlo simulation which presents graphically all possible outcomes of the designated results (for example project durations, and costs) [22,23].

Monte Carlo simulations are one of the risk analysis methods relying on the repeated random sampling using statistical analysis to yield probable results [24]. A Monte Carlo simulation is also well known as “what-if analysis” [25,26] .

The advantages of using Monte Caro simulations compared to the deterministic model were that the input variables for the models depend on the variation of the external and internal conditions [25,27] . The probability risk analyses may also dynamic and may yield different results for the different inputs model, once the model is simulated [25,28].

<bold id="bold-9aabc6fca599e5e4c1a2880bc7d6d0ba">Figure 2.</bold> Case-based models (Samik Raychaudhuri, 2008)

Based on literature reviews the utilization of this Monte Carlo simulation has improved quantitative project analysis for developing decisions [25,26,29] .

 Methods

The methods applied in this study are following these 3 steps ; (i) Establishing questioner surveys to identify a list of risk factors that may occur in this project, (ii) assessing for quantifying the risk frequency and its impact causing the project delays by conducting qualitative risk analyses procedures, and (iii) performing a sensitivity risk analysis ( Figure 1 ).

Conductiong Survey

In order to identify a list of risk factors, this study conducted a field questioner survey from several project stakeholders [30]. This study conducted 1st questioner survey by interviewing 25 respondents consisted of 1 respondent from the project owner, 12 respondents from the supervisory consultant, and 12 respondents from the project contractor. This research applied Disproportional Stratified Random Sampling (DSRS). The Disproportional Stratified Random Sampling divided the population into strata and selecting a simple random sample from each stratum. The sample represents the characteristics of the total population [24].

Performing Risk Assessment

A Qualitative Risk Assessment was performed by conducting a questioner survey. The questioners were designed to quantify risk frequency and its impact causing project delays (by conducting 2nd Questioner Survey), and the collected data were analyzed using qualitative analyses [31-34] . This 2nd questioner survey was also conducted to validate a list of risk factors based on a Lickert scale using 1 to 5 scales.

Performing Risk Analyses

Risk analyses were performed by simulating the existing input data using “what it is scenario”. The risk simulation results were presented in the form of Probability Distribution Function (PDF) and Cumulative of Distribution Function (CDF) graphs. The results would be conducted after running 1000 times iterations.

In this article, a spider graph diagram was used to illustrate the sensitivity of the project delay over the project activities. The final results were drawn in the form of probability distributions graphs using Monte Carlo simulation utilizing @risk software [22,23].

Refurbishment of Oil Tank Projects

A refurbishment of the oil tank project is defined as a partial or an overall replacement activity package related to the tank structures and facilities [35]. The refurbishment project may encompass various activities, with a high degree of complexity, and uncertainty including integrative demolition of the existing tank, and construction of the new one within a small and narrow working area [15,36,37] It was reported that the existing tanks within the Duri oil field were operated for more than 40 years and in urgent need to be repaired as there has been corrosion and thinning undergone inside and outside the tank plates. In this case study, the tank’s diameter dimension was 22 m in height was 9.6 m with a tank capacity of 21.9 thousand barrels of crude oil ( Figure 3 (b, c)). Again the results of this case study may be a reference for the oil industry stakeholders in managing the risk of delay in this tank oil refurbishment project soon. The location of the FWKO project and the condition of the existing tank were presented in Figure 3 (a).

Figure 3. (a) project location, (b) crude oil tank refurbishment in progress, (c) crude oil final construction

Results and Discussions Free Water Knock Out (FWKO) Refurbishment Project

The refurbishment FWKO activities in Duri consisted of at least 4 main packages encompass; (i) tank cleaning activities, (ii) dismantling of ring wall and tank bottom plat, (iii) installation of tank bottom plate and shell, and (iv) installation of the center column and tank roof. The total number of FWKO sub-activities (level 3) may include 54 activities [15].

Stage 1. Risks identification and Brainstorming of the Project Schedule

It was identified, 16 main activities existed in the critical path of the FWKO project.

The initial schedule for accomplishing this project was designated for 180 days. During the construction phase, it was calculated that the actual duration became 437 days (delayed for 257 days or almost 250% from the initial project duration) In this paper, the terminology of working days, later on, will be defined as “days” only.

Based on the literature review, it was identified that there were approximately 30 to 49 risk factors inherent in the type of oil project [6,16,18,38,39].

This study conducted deep questioner surveys to the oil stakeholders in the restricted oil industry areas for identification of a range of risk input for conducting risk analyses using the Monte Carlo simulations [26,36] .

To obtain a list of risk factors, the identified risk frequencies and risk impacts were then multiplied. The results of the multiplication of risk frequencies and impacts = risk index. The risk index can be seen in Table 3 and Figure 7. The risk analysis was then listed based on the Risk Index [16,28,33,39-41]. This index is presented as follow [16] :

\begin{equation} R=I \times P \tag{1} \end{equation}

where, R = Risk index, I = Risk probability (frequency), and P = Risk impact.

The 2nd questioner survey data were analyzed using the SPSS application package to perform the validation test. Based on the phase II questionnaire survey data, then the multiplication between the frequency of risk events and the impact of the risks that occur for each single risk index for the delay was calculated and validated.

This article illustrated 9 the highest list of the risk indexes causing the delays (based on their total correlations). These risk indexes encompass; (i) poor coordination among the project parties (X8), (ii) Lack of experienced human resources (X15), (iii) Lack of supervisors quality (X16), (iv) inadequate work safety procedures during construction (X17), (v) Lack of construction equipment (X20), (vi) limited and narrow working area (X21), (vii) design changes during construction (X24), (viii) the difficulty in obtaining daily work permits (X26), and (ix) A complicated inter-dependencies critical path activities (X27).

The risks were then indexed into 3 simple levels (encompassing minimum, medium, and maximum). The maximum risk index level was 0.568, and the minimum average value was 0.178, and the range value = 0.390.

The average distance of the value is 0.390/3=0.130 (as there was assumed that the risk index would be classified into 3 categories; low, medium, and high). This low risk range value = 0.178+0.130= 0.308, medium=0.308+0.130=0.438, and high = 0.438 + 0.130 = 0.568 ( Figure 4 ).

<bold id="bold-5a702f8382405879a28f859668d8886a">Figure 4</bold>. Level of Risk

Stage 3. Simulating Risks by Conducting Risk Analyses and Sensitivity Analyses

In this article the risk simulations were implemented by conducting; (i) risks analyses, and (ii) sensitivity analyses.

Risk Analyses (What it is Scenario)

Risk analyses were performed using the distribution of the input values of the project risks which were obtained from the 2nd survey data.

A Monte Carlo simulation using Pert Distribution was applied for calculating the duration of project activities (with a minimum duration of 165 days, most likely to be 180 days and a maximum of 437 days).

For example, historically based on the initial project planning the duration for installation of water column activities was estimated 24 days (most likely duration), in practice, these activities were conducted in 38 days (maximum one) and the minimum duration for performing these activities would be 24 days x (100%- a base value of the minimum risk of 17.8%) = 20 days (Table 2). Based on 3rd surveys the delay was identified as the cause of various risk factors such as; poor coordination among the project parties in the construction of water column (X8), lack of supervisor quality in managing the activities (X16), inadequate construction equipment within the project location (X20), narrow working area (X21).

<bold id="bold-3f3ef94ef09ecb854b2486bf6232b403">Table 2.</bold> Three estimated durations before conducting mitigation process

Activities Initial Plan (Duration)
Minimum Most Likely Maximum Minimum Most Likely Maximum

Table 2 shows that there were 3 main columns encompass, initial planning duration (days), a 3-point estimate before conducting risk mitigation procedures (in percentage), and a 3-point estimate before conducting risk mitigation procedures (days).

After performing simulation for scenario 1 using Palisade Decision Tool @risk application package utilizing Table 2 data, with 1000 iterations, it was identified that there was a 90% probability the total duration of the project would be accomplished in the range of 227 days to 260 days. There was also an 80% probability the project could be accomplished in 250 days (Figure 5). However, the total maximum duration would be 437 days ( Table 2 ).

<bold id="bold-cecaf3936204d864070aa53e9a6a3571">Figure 5.</bold> Probability Distribution Function Graph (PDF) and Cumulative of Distribution Function (CDF) Graph of the total duration activities Before Mitigation

The initial project duration was 180 days. There was the possibility of this project would experience delays from 47 days to 80 days (126%-144%). And there was a 0% probability for accomplishing the project duration (at 180 days) under those identified project risk and uncertainty conditions. Hence, based on these findings the project owner and managers may evaluate their schedule and working process for performing tank refurbishment projects in the future systematically achieving the project objectives.

<bold id="bold-4d0056d55234dbfd49113612908972d6">Table 3</bold>. Head to head of estimated project durations in 3 conditions (initial plan schedule, actual schedule and simulation schedule)

No Conditions Completion Duration Notes
1 Initial plan 180 days Determined the estimated project duration without predicting and identification of any risks inherent in the project.
2 Actual 437 days Implementing the project without conducting any simulation of the probability of the project delay and risks.
3 Simulation (90% probability) at the range of 227 days to 260 days Simulating the project duration using Monte Carlo and considering the uncertainty as well as project risks.

Table 3 shows that there were 3 different results for completing project schedule. At the initial plan, there was estimated that the project would be accomplished in 180 days using deterministic analyses without considering any risks inherent in the FWKO project. In fact, the project had been finished in 473 days. Hence, it is necessary to calculate the project duration using simulations in order to yield a comprehensive result. Based on the implementation of Monte Carlo simulations running on @risk application package it was calculated that the project duration would be at the range of 227 days to 260 days.

Developing Sensitivity Risk Analyses

The important roles of sensitivity analysis in the technical and economic perspectives was highlighted [9]. After conducting simulation utilizing table 2 data using @risk application, it was identified 8 main activities contributing significantly to the project delays such as; installation of the existing spools activity, installing water leg, installing roof platform, etc. (Figure 6). The more sensitive the activity, the more its effect to change output results (causing project delay). For example, the change in the duration for installing all the existing spools activity at 35% may affect to increase the total project delay up to 10%, and the additional duration in removing and installing rib rafter 25% will contribute to 9% of the total project delay.

<bold id="bold-dedc15c05f8e40d5dd9dbfc62bd55e14">Figure 6.</bold> Spider graph before conducting risk mitigation

The list of sensitive activities in this article highlights the most sensitive activities that affect project delay. The list of activities was as the following order; install all existing spools, remove and install rib rafter, install water leg, install roof platform, install the center column and rafter, install water leg box, install (safety in design) SID equipment, install the roof, and internal tank (Figure 6). Thus, it is important to pay attention for evaluating and mitigating these 8 sensitive activities for controlling and reducing the risk impacts in the future.

 Conclusion

Based on the Monte Carlo simulations, there was a 90% probability total duration for completing the project was at the range of 124%-144% from the total initial project schedule. It stated that there was no chance for completing the project schedule within 180 days (the initial project planning). Hence, it proves that by implementing the project simulation using @risk application package using the Monte Carlo risk analyses yielded a range of estimated output values (of the project delay) under the uncertainties. These results become a basis of the project owner and managers prior to developing a comprehensive strategic planning for the refurbishment of oil tanks in the future.

 Acknowledgment<bold id="bold-2"/>

The author would like to thank the PT Chevron Pacific Indonesia, the Civil Engineering Department University of Riau, Rizki Ramadhan Husain, and the team for assisting the author in supporting this research study.

2nd international congress of structure, architecture, and urban development Tabriz, Iran 2014 Babaei H. Hosseini M. The causes of delay in the implementation of 170 projects in Iran and the strategies for delay reduction 5 321-326 21 2003 10.1016/S0263-7863(02)00055-8 Frimpong Yaw Oluwoye Jacob Crawford Lynn International Journal of Project Management Causes of delay and cost overruns in construction of groundwater projects in a developing countries; Ghana as a case study 4 05 349-357 24 2006 10.1016/j.ijproman.2005.11.010 Assaf Sadi A. Al-Hejji Sadiq International Journal of Project Management Causes of delay in large construction projects 2 11 1995 10.1061/(ASCE)0742-597X(1995)11:2(45) Assaf Sadi A. Al-Khalil Mohammed Al-Hazmi Muhammad Journal of Management in Engineering Causes of Delay in Large Building Construction Projects 1 11-16 4 2018 10.24178/ijare.2018.4.1.11 Kazemi Aliyeh Katebi Ali Kazemi Mohammad-Hossein International Journal of Advanced Research in Engineering Causes of delay in construction projects: The case of oil and gas projects 1 51-59 18 2000 10.1016/S0263-7863(98)00060-X Al-Momani Ayman H. International Journal of Project Management Construction delay: a quantitative analysis 1 17-29 16 1998 10.1080/014461998372556 Kumaraswamy Mohan M. Chan Daniel W. M. Construction Management and Economics Contributors to construction delays 1 55-63 15 1997 10.1016/S0263-7863(96)00039-7 Chan Daniel WM Kumaraswamy Mohan M International Journal of Project Management A comparative study of causes of time overruns in Hong Kong construction projects 2008 10.2139/ssrn.1268242 Hermeling Claudia Mennel Tim ZEW - Centre for European Economic Research Discussion Paper No. 08-068 Sensitivity Analysis in Economic Simulations - A Systematic Approach 3 31 2015 10.1061/(ASCE)ME.1943-5479.0000248 Ruqaishi Mohammed Bashir Hamdi A. Journal of Management in Engineering Causes of Delay in Construction Projects in the Oil and Gas Industry in the Gulf Cooperation Council Countries: A Case Study 5 450-461 20 2018 10.1080/15623599.2018.1484991 Muneeswaran G. Manoharan P. Awoyera P. O. Adesina A. International Journal of Construction Management A statistical approach to assess the schedule delays and risks in Indian construction industry 5 238-244 19 2001 10.1108/02630800110412480 Reyers John Mansfield John Structural Survey The assessment of risk in conservation refurbishment projects 555-561 56 2015 10.1016/j.engfailanal.2014.10.003 Liu Hongyong He Weitao Guo Ji Huang Qiushuang Engineering Failure Analysis Risk propagation mechanism: Qingdao Crude Oil Leaking and Explosion case study 4 622-633 8 2017 10.14716/ijtech.v8i4.9480 Bouayoune Karima Selmani Boudi El Mostapha Bachir Aziz International Journal of Technology A Stochastic Method based on the Markov Model of Unit Jump for Analyzing Crack Jump in a Material Duri, Indonesia 2018 Arvin A Free Water Knock-Out (FWKO) Tank 4th edition Project Management Journal 40 2009 10.1002/pmj.20125 Rose Kenneth H. Indelicato Greg Book Review: A Guide to the Project Management Body of Knowledge 4 396-403 7 2015 Joshi Ankur Kale Saket Chandel Satish Pal D. K. British Journal of Applied Science & Technology Likert Scale: Explored and Explained 2019 10.1080/15623599.2019.1644762 Sandhyavitri Ari International Journal of Construction Management Stochastic analyses for managing risk of delay in Duri oil construction projects, Indonesia 1st edition Wiley 2004 978-0-470-87093-8 Saltelli Andrea Tarantola Stefano Campolongo Francesca Ratto Marco Sensitivity Analysis in Practice: A Guide to Assessing Scientific Models 1st Edition London Routledge 1993 9780203474556 10.4324/9780203474556 Raftery J. Risk Analysis in Project Management 3 579-590 22 2002 10.1111/0272-4332.00040 Saltelli Andrea Risk Analysis Sensitivity Analysis for Importance Assessment New York USA Palisade Corporation 2010 Palisade Guide to Using @RISK - Risk Analysis and Simulation Add-In for Microsoft Excel Version 5.5 New York, USA Palisade Corporation 2007 Palisade @RISK for Six Sigma 9 1-10 8 2017 10.1051/ijmqe/2016030 Belouafa Soumia Habti Fatima Benhar Saïd Belafkih Bouchaïb Tayane Souad Hamdouch Souad Bennamara Ahmed Abourriche Abdelmjid International Journal of Metrology and Quality Engineering (IJMQE) Statistical tools and approaches to validate analytical methods: methodology and practical examples 2008 Winter Simulation Conference, Miami, USA 91-100 2008 10.1109/WSC.2008.4736059 Raychaudhuri Samik Introduction to Monte Carlo simulation 29-60 John von Neumann Institute for Computing 2004 3-00-012641-4 Frenkel Daan Attig Norbert Binder Kurt Grubmuller Helmut Kremer Kurt Computational Soft Matter: From Synthetic Polymers to Proteins, Lecture Notes Introduction to Monte Carlo Methods 1 44-57 9 2007 Kwak Young Hoon Ingall Lisa Risk Management Exploring Monte Carlo Simulation Applications for Project Management 4 586–594 16 2010 10.3846/jcem.2010.65 Kim Seon-Gyoo Journal of Civil Engineering and Management Risk Performance Indexes and Measurement Systems For Mega Construction Projects 6 2018-2026 7 2017 10.18517/IJASEIT.7.6.2338 Razali Nor Shahida Dahlan Nofri Yenita Abbas Wan Faezah Mohamad Hasmaini International Journal on Advanced Science, Engineering and Information Technology Visual Basic GUI for an Improved of M&V Framework Considering Risk Assessment 5 245-252 19 2001 10.1108/02630800110412615 Mills Anthony Structural Survey A systematic approach to risk management for construction 3 320-340 35 2017 10.1016/j.ijproman.2017.01.010 Muriana Cinzia Vizzini Giovanni International Journal of Project Management Project risk management: A deterministic quantitative technique for assessment and mitigation 4 369-381 21 2018 10.1080/15623599.2018.1541703 Vacanas Yiannis Danezis Chris International Journal of Construction Management An overview of the risk of delay in Cyprus construction industry 1 33-46 16 2010 10.3846/jcem.2010.03 Zavadskas Edmundas Kazimieras Turskis Zenonas Tamošaitiene Jolanta Journal of Civil Engineering and Management Risk assessment of construction projects 1 90-100 12 2021 10.14716/ijtech.v12i1.4156 Hatmoko Jati Utomo Dwi Astuti Pertiwi Kusuma Farania Sani Nur Insuring Project Risks: Contractor Expectations versus Insurance Company Policies Surabaya Institut Teknologi Sepuluh Nopember 2014 Renaldhi Muhammad Revi Analisis Risiko Keterlambatan Proyek Pembangunan Tangki X Di TTU-Tuban (Studi Kasus : PT Pertamina UPMS V) China South China Institute of Environmental Sciences, Ministry of Environmental Protection of the People’s Republic of China 2008 Jianming Zhang Gas Tank Project Environmental Impact Assessment Report 14th Annual ARCOM Conference 1998 Rahmat Ismail Torrance Victor B. Young Barbara A. University of Reading. Association of Researchers in Construction Management The planning and control process of refurbishment projects 2 85-102 5 2014 10.22059/imj.2013.50271 Attafar Ali Eghbali Mohammad Industrial Management Journal Analyzing the Factors Influencing Delay of Projects in Zone 3 of Iranian Gas Transmission Company 24th Annual ARCOM Cardiff, UK 819-827 2008 Salama Mohamed El Hamid Moustafa Abd Keogh Bill Investigating the causes of delay within Oil and Gas projects in the U.A.E. 5 439-451 21 2019 10.1080/15623599.2018.1557997 Sarkar Debasis Singh Manvinder International Journal of Construction Management Development of risk index for mass rapid transit system project in Western India through application of fuzzy analytical hierarchy process (FAHP) Buckinghamshire, UK The Association for Project Management 2000 Norris Catriona Perry John Simon Peter Project Risk Analysis and Management