ICID & IRRIGATION AUSTRALIA 22 – 29 June 2012, ADELAIDE, AUSTRALIA ICID: 63rd IEC MEETING AND 7th ASIAN REGIONAL CONFERENCE IRRIGATION AUSTRALIA 2012 CONFERENCE & TRADE SHOW Evaluating Schedule Uncertainty: The Case Of Gönen Irrigation And Drainage Project In Turkey Dr. Önder ÖKMEN Civil Engineer Ministry of Forestry and Water Affairs General Directorate of State Hydraulic Works (DSİ) Ankara, TURKEY E-mail: [email protected] Contents Introduction Correlated Schedule Risk Analysis Model (CSRAM) Example Application General description (Gönen Irrigation Project in Turkey) Critical Path Method (CPM) application Monte Carlo Simulation (MCS)-based-CPM application Correlated Schedule Risk Analysis Model application Conclusions Introduction Targets of Project Management Time Cost Quality Scope Project management process cycle Controlling Planning Scope Quality Coordinating Organizing Time Cost Directing Introduction Insufficiency of the Critical Path Method (CPM) in modeling the real situations in spite of its wide usage in planning, scheduling and management of construction projects. Risks, uncertainties and uncertain activity durations. Correlation between activities and between risk-factors that create uncertainty on activity durations. Concurrently modeling the uncertain activity durations, the uncertainty effect of risks and the correlation. Introduction Implicating CPM with using Monte Carlo Simulation (MCS) and integrating to the model Variable activity criticalities, variable activity path criticalities, variable project completion time, risk – network sensitivity analysis Correlated Schedule Risk Analysis Model (CSRAM) Based on CPM. Utilizes MCS. Enables evaluating the construction activity networks under uncertainty through the probabilistic quantitative output produced by qualitative and quantitative input. Developed as a risk analysis model that can be used in risk management and decision making processes of construction projects. Computerized by using MS Excel and @Risk software. Detailed description available in “Ökmen Ö., Öztaş A. (2008). Construction project network evaluation with Correlated Schedule Risk Analysis Model. Journal Of Construction Engineering And Management, ASCE 134(1) pp 49–63”. Correlated Schedule Risk Analysis Model (CSRAM) CSRAM flowchart CSRAM input-output Example Application/General description Gönen Irrigation and Drainage Project in Turkey Turkey’s total land area is 78 million ha. Almost one third of this, 28 million ha, can be classified as cultivable land. Recent studies indicate that an area of about 8.5 million ha is economically irrigable under the available technology. Until now, an area of about 2.8 million ha has been equipped with irrigation infrastructures by the General Directorate of State Hydraulic Works (DSİ in Turkish acronym) which is the primary executive state agency of Turkey for water resources planning, managing, execution and operation. If the irrigation projects completed by the other state agencies are also taken into account, an area of 4.9 million ha in total has been irrigated so far. However, rehabilitation is required for some of the old projects. These records reveal the importance of schedule risk analysis in irrigation and drainage projects that will be implemented in Turkey, as a country aiming at irrigating all the economically irrigable land of 8.5 million ha in the near future. In this regard, this study tries to draw attention to this issue. It presents an example CSRAM application to an irrigation project called “Gönen Project”, which was realized in Turkey between the years 2006 and 2007. The study also includes CPM and Monte Carlo Simulation (MCS) applications on the case of Gönen Project for the purpose of comparison. Example Application/General description Gönen Irrigation and Drainage Project in Turkey The Gönen Project was undertaken by a company on behalf of DSİ. The official name of the project is «Gönen Project, Gönen Plain Pumping Irrigation Project Design Job». The participants signed the contract on December 22, 2005 and the Owner handed over the job to the Contractor on January 2, 2006. This was also the official start date of the project. The aim of Gönen Project is to supply irrigation water with pressurized pipes to an area of 3750 ha of farm land in Balıkesir City/Gönen Region of Turkey. The main water resource of the project is the reservoir of the Gönen Dam. However, water will be delivered by four pumping stations to the project area from the main open channels of another project called «Gönen Plain Open Channel Irrigation Project» which takes water from the Gönen Dam. The irrigation system including main, secondary and tertiary channels will be constructed with pressurized pipes. The Contractor carries the responsibility of preparing the design drawings and reports in compliance with the contract clauses and the technical criteria. In other words, the case of Gönen Project includes only the pre-construction design stage. After the completion of Gönen Project, the construction stage of the project was awarded through a public tender performed in 23.12.2008. The construction still continues. Example Application/General description Gönen Irrigation and Drainage Project in Turkey Example Application/General description Gönen Irrigation and Drainage Project in Turkey The schedule submitted to the Owner by the Contractor at the beginning of the job was a simple bar chart. There were 19 different activities in the bar chart schedule. The Owner approved this schedule as a contract document. Thus, the job had to be completed within 300 calendar days between January 2, 2006 and October 28, 2006 according to the bar chart schedule. However, actual situation occurred differently. The Contractor requested time extensions from the Owner three times before June 2007 and the Owner had to accept all of these requests in compliance with the contract clauses. Another time extension request was made towards the end of June 2007. In other words, the 300 calendar days of contractual (planned) project duration was increased to 514 calendar days by June 2007 through three time extensions. The reasons of this time overrun which occurred early in the project are the insufficiency of the bar chart schedule and the inaccuracy of the predicted activity durations. The dependencies between the activities and the critical activities which directly affect the project duration could not be provided by a bar chart. After June 2007, two more time extensions occurred and the project was completed 683 days after the start date - January 2, 2006. In other words, the actual duration of the Gönen Project has become higher than two times of the planned duration, which is 300 days in the contract documents. Example Application/CPM application Gönen Irrigation and Drainage Project in Turkey In this regard, the officially approved bar chart was converted to an activity network and CPM was applied by using the same activity durations used in the bar chart in order to compare these two methods. However, 32 activities have been used in CPM application instead of 18 in order to build predecessor relationships. CPM application was implemented by Primavera Project Planner software. The completion date of the project was found as November 13, 2006 (project duration: 316 days) by CPM. However, it was October 28, 2006 (project duration: 300 days) according to the bar chart schedule although the same activity durations have been used for both the CPM and bar chart schedules. After investigating the official documents of the Gönen Project, it has been detected that the time extension requests were due to the late approval of the “Preliminary Report, general layout and upper plain plans”, and “pre-application network plan/profile design drawings” compared to the finish dates of these activities in the bar chart schedule. Example Application/CPM application Gönen Irrigation and Drainage Project in Turkey These activities were determined as critical or near-critical in CPM application according to the total float times. Since the bar chart schedule could not show the criticalness of these activities, the necessity of revision was overlooked. Consequently, the time extensions occurred. Let’s assume that the Contractor applied CPM, detected the critical activities, noticed the insufficiency of the critical activities’ durations and eventually, decided to revise the schedule. What would be the extent of this revision? Unfortunately, CPM cannot model the uncertainty in activity durations and accordingly, it cannot evaluate the uncertainty in project completion time. Therefore, simulation-based risk analysis techniques should be utilized to overcome this limitation. For this reason, MCS-based-CPM and CSRAM have been applied in this study. Example Application/CPM application Gönen Irrigation and Drainage Project in Turkey CPM network diagram of Gönen Project Example Application/MCS based CPM application Activity durations were modeled by triangle distributions formed by maximum-minimum-most likely durations. In order to apply MCS-basedCPM, CPM was computerized on MS Excel and @Risk, which is a simulation add-in program, was integrated into the MS Excel. After conducting 1000 MCS iterations, it was found that the probability of completing the job in 300 days determined in the approved bar chart schedule is 0 percent, the probability of completing in 316 days determined by CPM application is 0 percent, the duration of completion with 0 percent probability is 355 days, and the duration of completion with 100 percent probability is 630 days. These results indicate the unreliability of the findings of CPM and bar chart schedules due to the uncertainty on activity durations and the project completion time. MCS-based-CPM models the uncertainty effect but ignores the correlation effect unless the correlation coefficients are provided directly before the simulation. Furthermore, it is not capable of considering the risk-factor effect, correlation effect (unless the correlation coefficients are directly provided), and the sensitivity of the schedule to risk-factors. For this reason, the results of MCS-based-CPM application are misleading. Example Application/CSRAM application The purpose of application of CSRAM is to evaluate the uncertainty effect on the schedule of Gönen Project without ignoring the correlation effect between activity durations and between risk-factors. First, the risk-factors, activity/risk-factor influence degrees, risk-factor situation probability boundary values, and correlated risk-factors were determined. The same activities, maximum-minimum-most likely activity durations, predecessor relationships and lag times used in CPM and MCS-based-CPM were used in CSRAM application. In order to execute CSRAM, the model was computerized on MS Excel and @Risk, which is a simulation add-in program, was integrated into the MS Excel. Risk-factor-3 (Design changes by the General Directorate) and risk-factor4 (Owner's delay in approving the design drawings and reports) were introduced to the model as correlated. After conducting 1000 MCS iterations, CSRAM found the duration of completion with 0 percent probability as 281 days, and the duration of completion with 100 percent probability as 735 days. Example Application/CSRAM application Input data of CSRAM application Example Application/CSRAM application When the results of the CSRAM and MCS-based-CPM are compared, the maximum possible project duration found by CSRAM is greater and the minimum possible project duration found by CSRAM is lower. In other words, CSRAM overestimated with respect to MCS in case the maximum possible project duration is concerned and it underestimated in case the minimum possible project duration is concerned. However, CSRAM discloses the uncertainty in project duration more realistically with respect to MCS. In other words, the difference between the maximum and minimum possible project durations estimated by the CSRAM application is greater, i.e. the project duration is more uncertain. Furthermore, the probability of completing the project in 300 days (the project completion time determined by the bar chart and written in the contract) is 0.02 and the probability of completing the project in 316 days (the project completion time determined by the CPM application in this study) is 0.10, which means that CSRAM underestimated with respect to MCS-basedCPM. However, CSRAM’s results related to the uncertainty in project duration are more reliable and realistic due to the correlation incorporated and risk-factor based simulation algorithm of the model. As previously mentioned, the actual duration of the Gönen Project had risen up to 683 days after several time extensions. “683 days” corresponds to a cumulative probability of 0.95 according to CSRAM. In other words, CSRAM estimated the actual project duration within the min.- max. probability range. However, MCS could not estimate the actual duration because the maximum possible duration found by MCS-based-CPM was only 630 days which was lower than the actual duration, 683 days. Example Application/CSRAM application Project duration vs. cumulative probability graph Example Application/CSRAM application Implementation of the project – risk sensitivity analysis by CSRAM has shown that risk-factor-3 (Design changes by the General Directorate) and risk-factor-4 (Owner's delay in approving the design drawings and reports) are the most effective risks on the uncertainty of project duration. Riskfactor-5 (Labor productivity inside the Contractor Firm) and risk-factor-9 (Disputes between the project participants) have been determined as the next most effective risks. When the official documents of the project were examined, the risk-factor-4 (Owner's delay in approving the design drawings and reports) had been shown as the cause of the three time extension decisions approved by June 2007. This shows the success of CSRAM in detecting the most effective risk-factor on the uncertainty of project completion time. Such an information is important from the managerial point of view for the sake of taking necessary risk mitigation and risk controlling precautions in advance. However, it is not possible to obtain such information either by bar charts and CPM schedules or by MCS-based-CPM analysis. It should be mentioned that CSRAM is also capable of conducting activityrisk and path-risk sensitivity analyses besides project-risk sensitivity analysis. Through these analyses, it becomes possible to explore which risk-factors are more effective on activity and path criticalities. Example Application/CSRAM application Input data and results of MCS-based-CPM and CSRAM applications Example Application/CSRAM application Results of risk sensitivity analysis of CSRAM application Conclusions In this study, the Correlated Schedule Risk Analysis Model (CSRAM) has been applied to a real project called “Gönen Project”, which started in Turkey at the beginning of 2006 and ended at the end of 2007. CSRAM enables evaluating the uncertainty effect on construction schedules in case of correlated activity durations and correlated risk-factors. The Critical Path Method (CPM) and Monte Carlo Simulation (MCS)-based-CPM have also been applied on the case of Gönen Project in order to compare the results. The findings of these applications showed that CSRAM operated well and produced realistic results concerning the extent of uncertainty inherent in the schedule of Gönen Project. Furthermore, comparison of the results revealed the significance of considering the correlation between activities and between risk-factors. We also examined the risk-factor sensitivity analysis capability of CSRAM, discussed the insufficiency of the bar chart schedule, and clarified the advantages and disadvantages of CPM, MCS-based-CPM and CSRAM. As previously mentioned, this study tries to reveal the importance of schedule risk analysis in irrigation and drainage projects that are being implemented in Turkey, as a country aiming at irrigating all the economically irrigable land of 8.5 million ha in the near future. Thank you for your attention.