ARTICLE IN PRESS

Building and Environment 40 (2005) 12441254 www.elsevier.com/locate/buildenv

Judgmental risk analysis process development in construction projects

ztas nder O kmen Ahmet O - , O
Civil Engineering Department, University of Gaziantep, Gaziantep, Turkey Received 16 June 2003; accepted 22 October 2004

Abstract Various risks and uncertainties exist in construction projects. These may not only prevent the projects to be completed within budget and time limit, but also threaten the quality, safety and operational needs. In this context, risk analysis processes are the systematic methods to analyze the potential project risks and develop risk response strategies in order to cope with risks and achieve the desired objectives. This study proposes a new schedule risk analysis method named as judgmental risk analysis process (JRAP) and offers a different project duration equation through JRAP. The process (JRAP) can be dened as a pessimistic risk analysis methodology or a hypothesis based on Monte Carlo simulation that is effective in uncertain conditions due to its capability of converting uncertainty to risk judgmentally in construction projects. A case study has also been developed to show how the proposed process is applied on a construction project and to prove its validity. r 2004 Published by Elsevier Ltd.
Keywords: Risk; Risk analysis; Monte Carlo simulation; Schedule analysis; Construction management

1. Introduction Each construction project has unique features that differentiate it from even resembling projects. Construction techniques, design, contract types, liabilities, weather, soil conditions, politic-economic environment and many other aspects may be different for every new commitment. This fuzzy atmosphere has been represented with the terms uncertainty or risk by construction managers and researchers, and they tried to control this systematically through risk management and analysis methods since the early 1990s [1]. Some researchers like Flanagan et al. [2] and Pilcher [3] put differentiation between these two terms. They have mentioned that uncertainty represents the situations in which there is no historical data; and risk, in contrast, can be used for situations where success or failure is determined in probabilistic quantities by beneting from
Corresponding author.

ztas E-mail address: aoztas@gantep.edu.tr (A. O -). 0360-1323/$ - see front matter r 2004 Published by Elsevier Ltd. doi:10.1016/j.buildenv.2004.10.013

the previous data available. Since such a separation is regarded as meaningless in the construction literature, risk turns out to be the most consistent term to be used for construction projects because some probability values can be attached intuitively and judgmentally to even the most uncertain events [2]. The uncertainty represented quantitatively at some level is not the uncertainty any more; rather it is the risk henceforth and needs to be managed. In this context, this research aims to prove this hypothesis by developing judgmental risk analysis process (JRAP) and applying this process in the assessment of schedule risks of a construction project through a case study. One important way of controlling risks in construction projects is to develop reliable project estimates and schedules [4]. Probabilistic scheduling through conduction of simulations on prepared models provides more powerful results fundamental to making decisions with regard to the control of project risks. Barraza et al. [5] has brought a stochastic approach to traditional S-curve project performance control method by using simulation

ARTICLE IN PRESS
ztas . O kmen / Building and Environment 40 (2005) 12441254 A. O -, O 1245

techniques and probability distributions. Chehayeb et al. [6] has developed a different simulation-based scheduling method with continuous activity relationships to enable effective use of systems simulation in scheduling of construction projects. Similarly, Senior et al. [7] has discussed a new scheduling system based on statistical simulation. The reliability of probability distribution functions used as input data in stochastic scheduling systems has attracted the interest of researchers like Maio et al. [8] and Fente et al. [9]. Different methods have been used so far for measuring the variations in activity durations and modelling schedule risks of projects, e,g. Program Evaluation and Review Technique (PERT) [10], Probabilistic Network Evaluation Technique (PNET) [11], Narrow Reliability Bounds (NRA) [12] and Monte Carlo simulation (MCS) [2,10,13]. Activity durations in construction projects are likely to vary because of the inherent risks. For this reason, Dawood [13] and Ranashinghe [14] have also developed equations for quantifying and modelling the variation in activity durations and nding the overall project duration.

tion and classication techniques are present such as brainstorming, check lists, organizational charts and mapping [1,16]. Risk analysis part is dominantly quantitative and, therefore, it includes mathematical and statistical operations [17]. Risk response is the last risk management step at which the results of the preceding steps are discussed and suitable risk mitigation actions are taken before facing risks.

3. JRAP methodology The JRAP proposed in this research consists of a number of managerial steps to be carried out and an equation that offers the variation in each activitys duration in the schedule network. The characteristics of JRAP make this methodology effective in uncertain conditions of which there is no or little previous data and increases its applicability for converting high level of uncertainty to risk judgmentally. Since, judgmental decisions based on experience and intuition would insert additional risks to the project network model, a pessimistic way has been followed in analyzing the overall project duration. In other words, the pessimistic characteristic of JRAP decreases the effect of planning engineers making inaccurate data estimation during risk modelling. This pessimism is created through the computing characteristics of the proposed equation that is introduced within JRAP. Another important point that should be mentioned about JRAP is that it is not a whole risk management system, rather it is considered as an analysis stage that is performed during the conguration of a risk management system. JRAP can be dened as a pessimistic risk analysis methodology or a hypothesis that is effective in schedule risk modelling of even the most uncertain situations in construction projects. Fig. 1 illustrates the logic and steps of JRAP and claries the processs relationship with the other risk management stages. The tasks labelled as D, E, F and G represent the steps of JRAP. They make up the risk management system together with the other tasks, which are B, C, H and I. The task labelled as A, in contrast, represents the classical approach, i.e. the arrangement of project content by deterministic schedule planning. JRAP methodology comprises the analysis part of overall risk management system as seen in Fig. 1. In other words, JRAP can only be applied just after identifying and classifying the risks that would inuence the activity durations of the shedule. Deterministic way of schedule analysis prior to setting up a risk management system (refer to Fig. 1) leads to a more appropriate risk-modelling environment for any project, even though deterministic approach avoids the use of statistical data and assumes certain xed values. The need for the deterministic model arises from its

2. Risk analysis and risk management Each construction project has its own technical characteristics that vary according to the construction type, execution time and its environment. This leads to a different risk atmosphere for each construction project. In this context, risk management can be dened as a systematic controlling procedure of risks that are predicted to be faced in an investment or project [15]. However, it is neither an insurance system nor a magical risk elimination method. It only aims to identify the potential risks as early as possible and manage them for preventing the harmful effects of the risks to the project aims. The system utilizes the historical data, statistical knowledge, computer modelling-running power, human intuition, judgment, experience, gut feel, common sense and willingness up to a full extent [2]. In the course of time, the management of risk has become a key element for the completion of projects within time schedule and planned budget. It is now a common opinion that controllable and uncontrollable risks can only be responded by utilizing risk management process over the entire project, i.e. prior to the tender and subsequently, by controlling and updating the system periodically during the application of the predetermined plan. Like every systematic procedure, risk management is a stepwise phenomenon. These steps can be summarized basically as risk identication, risk classication, risk analysis, and risk response [2]. Risk identication and classication comprise the qualitative investigation of the risks. In the literature, a number of risk identica-

ARTICLE IN PRESS
1246 ztas . O kmen / Building and Environment 40 (2005) 12441254 A. O -, O

Identify risks -B-

Model deterministic time schedule plan of the project -A-

Classify risks -CRISK MANAGEMENT FRAMEWORK

Decide the critical risks effective on activity durations -DUpdate the model

Assign probability distributions and max-min durations -E-

Control the progress -I-

Establish activity-risk factor matrix -F-

Model stochastic time schedule plan of the project and Run Monte Carlo Simulation -GJRAP METHODOLOGY

Respond to risks -H-

Fig. 1. The steps of judgmental risk analysis process (JRAP). Note: JRAP steps are labelled as D, E, F and G.

power of highlighting the potential project risks. In other words, it helps the planner especially during the risk identication step. 3.1. Step 1deciding of main risks The rst task undertaken during the conduction of JRAP is to nd out the critical risks that may be effective on activity durations. This is achieved through examining the critical activities of the deterministic schedule network and then selecting the risks which inuence these critical activities from the risk list

predetermined in the risk identication step (step B in Fig. 1) of risk management framework. For the sake of simplication, the critical risks established in this way are assumed to be effective on also non-critical activities. 3.2. Step 2assignment of probability distributions and maximum minimum durations The second step of JRAP is to attach probability distributions to the identied risks in the preceding step. This can be achieved through utilizing experience and engineering judgment in case there is not enough past

ARTICLE IN PRESS
ztas . O kmen / Building and Environment 40 (2005) 12441254 A. O -, O 1247

record to conduct detailed statistical analysis. This is already the advantage and practicality of JRAP during conditions when there is no previous data. The random values between 0 and 1 (refer to Eq. (1)) will be selected by the utilized software according to the character of the assigned probability distributions and correlation coefcients during MCS. While distributions are being assigned, maximum and minimum durations of activities should also be determined. Eventually, all of these values will be used in calculating activity durations. Common distributions like beta, triangle and lognormal can be used for representing the critical risks. However, negative skewness is not offered in JRAP, because it is less probable to face with a situation that the real execution time of a construction activity would be below the appraised most likely duration with a greater chance than being above the most likely duration. This is one of the characteristics of JRAP, which makes it a pessimistic approach. The corner values such as maximum, likely and minimum durations of an activity can be determined according to the type of key resources consumed for that activity. The activities in construction projects can be categorized as production-related, procurement-related and administrative activities. The production-related activities can either be labor-driven or equipmentdriven. According to this classication system, a variety of methods like heuristics, statistical studies, cycle time analysis, queuing theory and simulation are available to calculate activity durations [10]. However, engineering judgment, experience and intuition affect the consistency of the results obtained by these techniques. It is also possible to determine the corner duration values of the activities judgmentally as it is applied in the case study. 3.3. Step 3establishment of activity-risk factor matrix In this step of the process, activity-risk factor matrix is established. Activity-risk factor is the percentage effect of each risk over each activity (refer to Table 2). In activity-risk factor matrix, the varying effect of each risk over each activity is quantied in percentages. However, the total inuence of all risk factors should be 100% on any given activity [13]. This assessment is performed mainly in a judgmental way. In other words, if previous data is unavailable and a fully uncertain situation is encountered, a fully judgmental solution will be produced. 3.4. Step 4modelling and simulation The fourth JRAP step is modelling and simulating the schedule network of the project. Deterministic schedule model developed in project planning software, package

programs performing spreadsheet modelling and MCS are the main tools of this step. At this step, MCS utilizes the activity-risk factor matrix developed at the third step in order to calculate the variation in activity durations. In order to enable MCS to calculate the activity durations in a stochastic manner, the following pessimistic equation (Eq. (1)) is developed after some reformation on Dawoods [13] equation (Eq. (2)). Duration of activity MAXMin:Time Risk Factor Affect; Max:Time Risk Factor Effect; Duration of activity MAXMin:Time Max:Time Min:Time RF1 Random1 RF2 Random2 RFn Randomn ; Max:Time Max:Time Min:Time RF1 Random1 RF2 Random2 RFn Randomn ; 1

where Min.Time is the minimum duration that an activity can be completed, Max.Time, the maximum duration that an activity can be completed, RFn, the percent effect of nth risk factor on an activity (taken from activity-risk factor matrix), Randomn is a random number, between 0 and 1, generated during MCS to represent the nth risk factors probability distribution. Duration of activity Min:Time Max:Time Min:Time RF1 Random1 RF2 Random2 RFn Randomn : 2 Eq. (1) (activity duration equation), which is an important fragment of the JRAP model, consists of two main parts of which one of them is already Eq. (2) itself. For this reason, the nal result of Eq. (1) is the maximum of the results of these two parts. While Eq. (2) calculates the duration of the activity by adding the risk factor effect to the minimum activity time; Eq. (1), in contrast, calculates also the subtracted value of the same risk factor effect from the maximum duration and eventually gives the maximum of the two results as the activity duration. This approach provides to stand at the worse side and creates a pessimistic view. This is one of the cornerstones of the JRAP. Now, as an example, assume that an activity of a construction project has a maximum duration of 60 days and a minimum duration of 30 days. Next, assume that

ARTICLE IN PRESS
1248 . O kmen / Building and Environment 40 (2005) 12441254 ztas A. O -, O Table 1 Computations showing four manually conducted Monte Carlo iterations Iteration number 1 Data Random1 1.00 Random2 0.00 Random3 0.60 Random4 0.25 Random1 1.00 Random2 1.00 Random3 0.00 Random4 0.70 Random1 0.50 Random2 0.35 Random3 0.40 Random4 0.90 Random1 0.00 Random2 0.10 Random3 0.30 Random4 0.80 RF1 20% RF2 20% RF3 40% RF4 20% RF1 20% RF2 20% RF3 40% RF4 20% RF1 20% RF2 20% RF3 40% RF4 20% RF1 20% RF2 20% RF3 40% RF4 20% Computation

Risk factor Effect 14.7 Duration of activity 45.3 days

Risk factor effect 16.2 Duration of activity 46.2 days Max. duration 60 days Min. duration 30 days Risk factor effect 15.3 Duration of activity 45.3 days

Risk factor EFFECT 9.0 Duration of activity 51.0 days

the random numbers produced from the distributions assigned to four risks are (1, 0, 0.6, 0.25), (1, 1, 0, 0.7), (0.5, 0.35, 0.4, 0.9) and (0, 0.1, 0.3, 0.8) to be used during four manual iterations of MCS. Besides, from activityrisk factor matrix, the weather has 20%, design changes has 20%, soil conditions has 40% and labor productivity has 20% effect on the duration of the assumed activity. All of this information is enough to calculate the duration of the activity using Eq. (1). The computations are summarized in Table 1. Nevertheless, manual application of MCS is inconsistent while computer programs are capable of doing the same job faster, more precisely and with much more running preferences. Furthermore, such programs provide an easy and effective environment for data entrance and worksheet modelling. The assessment of correlation between risk factors is as important as the identication of risks and is more important than deciding on probability distributions of risk elements. For this reason, any risk model should consist of the correlation coefcients, if correlation exists. Otherwise, the simulation results would not be realistic.

4.2. Project denition The project is the decoration of a bank ofce, which is located at a rural area of the South-Eastern part of Turkey. The contract used is an admeasurement type with xed unit-price including specic clauses offered by the promoter. The promoter is the States biggest bank. The project start date was 30th November 1999 and the completion date was 25th January 2000 in the contract. In other words, the contractual project duration was 57 days. However, the project has been completed in 122 days, i.e. 65 days after the project deadline negotiated within the contract documents. 4.3. Scope and major project risks Since the project is the decoration of a bank ofce, the usage of some special decorative materials is unavoidable. These special materials might be related to electrical or mechanical services, or related to internal or external nishes according to the content of the architectural design. This is a risk for the contractor from both cost and duration point of views because the location of the construction site is far from the main manufacturers, which exist in the western industrialized regions. The second important risk creating characteristic of the project is that the possible material changes requested by the contractor after the approval of the contract is not permitted. This is due to the image sensitivity of the architects of the bank, i.e. a standard image style has been determined for the bank and all ofces have to be constructed according to this standard all over the country. This situation creates an inexible environment for the contractor.

4. Case study 4.1. Purpose The purpose of this case study is to give an example to the application of JRAP, to show how it works and to prove its validity. For achieving these goals and for the sake of comprehensiveness, a small and real completed construction project at which one of the authors involved throughout the whole process has been chosen.

ARTICLE IN PRESS
ztas . O kmen / Building and Environment 40 (2005) 12441254 A. O -, O 1249

Next, the contract is an admeasurement type with xed unit price, which can be considered as a risky contract strategy from the contractors point of view. The main reason of this is that the payment is at the end of each month after the measurement of the completed parts but the promoter does not compensate cost escalation, although there is high ination in the country and interest rates are very high. Furthermore, transporting cost is not paid to the contractor. The states bill of quantities and the banks own unit prices for special activities are used in measuring payment certicate values. In case of late completion of the project, 0.2% of tender price is cut off as a liquidated damage for each day and this issue has been mentioned strictly in contract documents. All of these oblige the contractor to complete the project as early as possible before the project deadline, which is committed within the contract, because of the time value of money and the reputation of the company. Fortunately, weather conditions are not very effective on the construction activities because most of them will be carried out in a closed environment; at the ground and base oors of a multi-storey building. Besides, soil condition is not a risk factor for the same reason. Although the construction site is at a region that has a high human-resource potential, working quality is not high. The abundance of this resource means that there are no human-resource allocation and levelling problems during scheduling. In other words, the activities are mainly material driven, not labor driven. 4.4. Schedule characteristics The project start date was 30th November 1999 and the completion date was 25th January 2000 in the contract. The completion date has been determined by negotiation between the promoter and the contractor without preparing any deterministic schedule! In this context, rst, deterministic schedule of the project is developed using MS Project planning software. Likely durations of the activities (refer to Table 5) are determined whether judgmentally or by examining previous similar projects. Experience has also made it easier to construct the network logic and activity interrelationships. Finally, the deterministic plan has shown that the likely project completion date was 12th February 2000, which was 17 days after the contractual completion date. This shows that even the deadline determined without risk analysis is a later date than the contractual completion date. It should be noted that JRAP will also provide the opportunity of comparing this deterministic deadline with the other stochastic deadline values and show whether this deterministic value is optimum or not.

4.5. JRAP application The rst task undertaken in application of JRAP on this project is determining the critical risks of the project because the rst step of JRAP is the identication of risks that are effective on critical activities explored during the critical path analysis (refer to Fig. 1). The critical risks of the project are determined as: defective design, design changes, subcontractors default, uctuation in labor productivity, delay in resolving disputes, difculties/delays in availability of materials, equipment and labor, inadequate quality of work and need for correction, promoter delays (unable to get approvals, lack of payment, etc.), and changes in quantity/scope of work [13,18]. Three different computer package programs namely MS Project, MS Excel and Crystal Ball have been used during this JRAP application. Before producing the spreadsheet risk model, activity-risk factor matrix has been formed. Table 2 presents activity-risk factor matrix of the bank decoration project. Afterwards, the remaining steps of JRAP, which are assigning probability distributions and maximumminimum durations, and modelling and simulating have been carried out subsequently. The statistical distributions assigned to risks and the correlation coefcients between them are given in Tables 3 and 4, respectively. Experience and engineering judgment have been utilized for setting up this data. All the data gathered and processed up to here have been utilized in setting the spreadsheet risk model so that it has become possible to perform MCS on the produced model by the risk analysis software, Crystal Ball. Table 5 shows the main spreadsheet model developed by the spreadsheet program (MS Excel) for the current case study. Tables 24 contain and provide the supportive data to Table 5. For instance, the cells just near the means row in Table 3 are the assumption cells as it is called in the program and here they represent the probability distributions assigned to the critical risks. The cells at the last column and the cell at the bottom of the late nish time column of Table 5 are called forecast cells and represent the resultant values that vary with respect to the relevant assumption cells. After running preferences and correlation coefcients had been introduced into the risk analysis model, simulation was conducted. The report produced by Crystal Ball at the end of MCS mainly consists of information about the assumption cells and forecast cells. Indeed, desired and necessary answers mostly exist among the parts related to forecast cells in the simulation report. The spreadsheet model in Table 5 has two kinds of forecast cells: total project duration and critical activity query cells. The cell at the bottom of the late nish time column gives the total project duration and it is related

1250

Table 2 Activity-risk factor matrix

Activity number Risk no. Activity name 1 Defective design (%) 2 Design changes (%) 3 Subcontractors default (%) 4 Fluctuation in labour productivity (%) 5 Delay in resolving disputes (%) 6 Promoter delays (unable to get approvals, lack of payment, etc.) (%) 7 Difculties/ delays in availability of materials, equipment and labour (%) 8 Inadequate quality of work and need for correction (%) 9 Changes in quantity/scope of work (%) Total (%)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37

Tender Sanction Site delivery Order materials Procure materials Remove existing old materials Detail measurement; by aluminum subcontractor Order and procure aluminum Manufacture aluminum joineries Approval of weight records of aluminum joineries Detail measurement; by glazier Order and procure joinery glasses Install electrical tting Install suspended ceiling Install HVAC, sanitary and mechanical tting Level slab cover sub-layer Cover slab Paint interior walls and columns Cover WC walls and slab Install aluminum joineries Install joinery glasses Install security alarm Reinforcement and framework; cash room walls Approval and carriage of cash room door Approval of reinforcement and framework Pour concrete Remove frameworks Cure concrete Install cash room door Paint walls and cover slab of cash room Delivery of money machine Install money machine Finish exterior works Cover mechanical and electrical units Temporary commissioning Approval of extra project duration Approval of temporary commissioning

70 70 20 20 70 40 40 80 10 20 80 80 30

30 30

20 20 20

100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100

ztas . O kmen / Building and Environment 40 (2005) 12441254 A. O -, O

10

10

20 80 40 10 20 10

10

20

ARTICLE IN PRESS

40 20 10 20

10

10 10 10 60 30 40 40 30 20 20 30

80 30 20 30 40 40 20 30 20 20 30 20

20 20 20

20 20 10 10 10

10 10 10 20 30 30 30 20% 20 20

10 20 10 30 30 40 30 80

10 10 10

30

60 60 60 20 40

40

30

40 40 40 10 60

100 10 10 30 40 30 30 40 40 40 60 60 60 40 30 60 10 10

Table 3 Probability distributions assigned to the risks Risks: 1 Defective design 2 Design changes 3 Subcontractors default 4 Fluctuation in labour productivity 5 Delay in resolving disputes 6 Promoter delays (unable to get approvals, lack of payment, etc.) 7 Difculties/ delays in availability of materials, equipment and labour Lognormal 0.44 8 Inadequate quality of work and need for correction 9 Changes in quantity/scope of work ztas . O kmen / Building and Environment 40 (2005) 12441254 A. O -, O

Distribution Type Mean

Triangle 0.46

Lognormal 0.39

Triangle 0.37

Triangle 0.40

Triangle 0.37

Lognormal 0.44

Triangle 0.42

Triangle 0.41

ARTICLE IN PRESS

Table 4 Correlation coefcients assigned to the risks Risks 1 Defective design 2 Design changes 3 Subcontractors default 4 Fluctuation in labour productivity 5 Delay in resolving disputes 6 Promoter delays (unable to get approvals, lack of payment, etc.) 7 Difculties/delays in availability of materials, equipment and labour 8 Inadequate quality of work and need for correction 9 Changes in quantity/scope of work

1 2 3 4 5 6 7 8 9

1.0 0.6 0.0 0.0 0.3 0.1 0.0 0.0 0.2

1.0 0.0 0.0 0.3 0.2 0.3 0.0 0.1

1.0 0.0 0.0 0.0 0.0 0.3 0.0

1.0 0.0 0.0 0.0 0.0 0.0

1.0 0.1 0.0 0.1 0.0

1.0 0.0 0.0 0.0

1.0 0.0 0.1

1.0 0.0

1.0

1251

1252

Table 5 Spreadsheet risk model of the project

Time schedule risk analysis Activity number Activity name Predecessor and relationship Lag (days) Minimum duration (days) Likely duration (days) Maximum duration (days) Simulated duration (days) Early start time Early nish time Late start time Late nish time Total oat (days) 0 critical activity 1 non-critical activity

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37

Tender Sanction Site delivery Order materialsa Procure materials Remove existing old materials Detail measurement; by aluminum subcontractor Order and procure aluminum Manufacture aluminum joineries Approval of weight records of aluminum joineries Detail measurement; by glazier Order and procure joinery glasses Install electrical tting Install suspended ceiling Install HVAC, sanitary and mechanical tting Level slab cover sub-layer Cover slab Paint interior walls and columns Cover WC walls and slab Install aluminum joineries Install joinery glasses Install security alarm Reinforcement and framework; cash room walls Approval and carriage of cash room door Approval of reinforcement and framework Pour concrete Remove frameworks Cure concrete Install cash room door Paint walls and cover slab of cash room Delivery of money machine Install money machine Finish exterior works Cover mechanical and electrical units Temporary commissioning Approval of extra project duration Approval of temporary commissioning

1 2 2 4 3 6

FS FS FS FS FS FS

0 2 1 1 10 2 1 15 15 1 1 10 2 3 4 1 1 3 1 3 2 1 2 2 1 1 1 4 1 1 20 3 3 3 1 1 1

0 2 1 1 15 3 1 20 20 1 1 15 3 4 5 2 2 4 2 4 3 2 2 2 1 1 1 4 2 2 30 3 4 4 1 1 1

0 4 2 1 40 4 2 45 30 1 2 30 5 10 8 3 4 10 3 8 6 3 4 4 1 1 1 4 3 3 60 7 7 6 2 2 2

0 3.3 1.6 1 27.8 3.1 1.7 32.7 24.2 1 1.7 21.8 3.7 7.3 6.3 2 2.6 6.6 2.1 5.9 4.7 2.2 3.1 3.3 1.0 1.0 1.0 4.0 2.4 2.2 43.5 5.3 5.2 4.8 1.7 1.5 1.5

0 0 5.3 3.3 4.3 6.9 10 11.7 44.3 68.6 69.6 71.3 10 13.7 10 21 32.1 37.7 32.1 69.6 93.1 97.7 10.0 4.3 13.1 14.1 18.1 26.1 30.1 32.5 4.3 47.8 53.1 32.1 99.9 101.6 103.1

0 3.3 6.9 4.3 32.1 10 11.7 44.3 68.6 69.6 71.3 93.1 13.7 21 16.3 23 34.7 44.3 34.1 75.5 97.7 99.9 13.1 7.5 14.1 15.1 19.1 30.1 32.5 34.7 47.8 53.1 58.3 36.9 101.6 103.1 104.7

0 0 5.3 44.9 47.2 6.9 10 11.7 44.3 68.6 69.6 71.3 61.9 65.6 66.7 73 75 80.6 85.2 87.2 93.1 97.7 75.2 92.1 78.3 79.3 83.3 91.3 95.3 97.7 45.9 89.4 94.7 95.1 99.9 101.6 103.1

0 3.3 6.9 45.9 75 10 11.7 44.3 68.6 69.6 71.3 93.1 65.6 73 73 75 77.6 87.2 87.2 93.1 97.7 99.9 78.3 95.3 79.3 80.3 84.3 95.3 97.7 99.9 89.4 94.7 99.9 99.9 101.6 103.1 104.7

0 0 0 41.7 42.9 0 0 0 0 0 0 0 52 52 56.7 52 42.9 42.9 53.1 17.6 0 0 65.3 87.8 65.3 65.3 65.3 65.3 65.3 65.3 41.7 41.7 41.7 63.1 0.0 0.0 0.0

0 0 1 1 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0

ztas . O kmen / Building and Environment 40 (2005) 12441254 A. O -, O

7 FS 8 FS 9 FS 10 FS 11 FS 6 FS 13 FS 6 FS 14,15 FS 5,16 FS 17 FS 5,15 FS 10,18,19 FS 12,20 FS 21 FS 6 FS 4 FS 23 FS 25 FS 26 FS 27 FS 24,28 FS 29 FS 4 FS 31 FS 32,5 FS 15,5 FS 22,33,34,30 FS 35 FS 36 FS

ARTICLE IN PRESS

3 7

ARTICLE IN PRESS
ztas . O kmen / Building and Environment 40 (2005) 12441254 A. O -, O Table 6 Probable durations in percentiles and critical activities with 100% certainty Forecast: total project duration Critical activities (with 100% certainty) Activity no. 2 3 6 7 8 9 10 11 12 21 22 35 36 37 1253

Percentile (%) 0 10 20 30 40 50 60 70 80 90 100

Days 97.5 101.1 102.6 104.0 105.5 106.9 108.4 110.1 112.1 114.6 124.8

to the other varying cells with the formula within the frame of the activity network logic. Moreover, Eq. (1) is utilized in the simulated duration column. On the other hand, the remaining forecast cells (the cells at the last column) show how often each activity becomes critical so that it can be decided on which activities managerial effort should be focused and necessary precautions should be taken in advance. 4.6. Discussion of the results Table 6 shows the summary of the results of the simulation performed with 10,000 trials. It consists of the project duration frequency values and the critical activities with 100% certainty. For instance, the project can be completed in 110.1 days with 70% probability according to these results. If at least this extra information had existed prior to the sanction, project overrun would not have occurred. In other words, the risk percentiles of the project duration would have helped the parties agree on a more suitable and realistic contractual deadline. Unfortunately, not carrying out deterministic and stochastic network analysis of the project prior to the sanction resulted in late completion of the current project. That is, the project was completed in 122 days, i.e. 65 days after the project deadline written in the contract documents.

considered as a different approach to the analysis stage of risk management systems. Judgmental risk analysis process is comprised of a number of steps as shown in Fig. 1. The methodology is mainly based on MCS and an equation that is proposed in the preceding sections. This equation (Eq. (1)) enables the planner to observe the variation in activity durations during MCS. These simulated values regarding the activity durations are then utilized in the spreadsheet risk model through arithmetic formulas that are produced according to the logical relationships between the activities of the time schedule network. The reason for the proposed risk analysis method to be qualied as judgmental is that the lack of historical data related to the activities of the time schedule network does not interfere with performing risk analysis on the network, i.e. engineering judgment, experience and intuition of the planner have all been utilized through JRAP. Obviously, such an analysis strategy would create additional risks due to the subjective data used. Nevertheless, the pessimistic character of the proposed activity duration equation (Eq. (1)) and skewness in probability distributions assigned to critical risks compensate this undesired situation. The other objective of the present research was to illustrate the application of JRAP using a construction project data. For this, JRAP was applied on a bank decoration project using the computer package programs, which are MS Project, MS Excel and Crystal Ball. The results of the case study have shown that JRAP has been capable of answering the questions such as:

  

In how many days and with how much probability it is possible to complete the project? Which activities are the most critical activities? Which risk elements are more effective on project duration?

5. Conclusion The aim of this paper was to propose a time schedule risk analysis methodology that is applicable on construction projects. This methodology, which is named as Judgmental Risk Analysis Process (JRAP), can be

The answers to such questions are obviously useful to make sound decisions prior to the contracting. As in the case study, the project completion duration specied in the contract, which is 57 days, came out to be only a big mistake because the results of JRAP has shown that the total project duration is 114.6 days with 90% and 124.8 days with 100% probability. The commissioning documents of the project have already supported this result because the construction has been completed in exactly 122 days, i.e. 65 days after the negotiated contract deadline. As a general concluding remark, JRAP stands as a practical schedule risk analysis methodology in the overall risk management framework and it also contributes to the dynamism of the risk management system through its stochastic characteristics. However, it is open to development through its application over more

ARTICLE IN PRESS
1254 ztas . O kmen / Building and Environment 40 (2005) 12441254 A. O -, O of Construction Engineering and Management 2000;126(4): 28592. Fente J, Schexnayder C, Knutson K. Dening probablity distribution function for construction simulation. ASCE Journal of Construction Engineering and Management 2000;126(3): 23441. Grifs FH, Farr JV. Construction planning for engineers. Singapore: McGraw-Hill; 2000. Illumoka AA. A tolerance analysis approach to network scheduling for engineering project management. International Journal of Production Research 1987;25(4):53147. Chapman CB. A risk engineering approach to project risk management. International Journal of Project Management 1990;8(1). Dawood N. Estimating project and activity duration: a risk management approach using network analysis. Construction Management and Economics 1998;16:418. Ranasinghe M. Qualication and management of uncertainty in activity duration networks. Construction Managemant & Economics 1994;12(1):1529. kmen O . Risk analysis and management of construction projects O tendered under design-build (turnkey) contract system. M.Sc. Thesis. Gaziantep: University of Gaziantep, 2002. Kartam NA. Risk and its management in the Kuwaiti construction industry: a contractors perspective. International Journal of Project Management 2001;19(4):32535. ztas kmen O . Risk analysis in xed-price design-build O - A, O construction projects. Building and Environment 2004;39:22937. Gursel AP. Risk perception and trends of Turkish construction companies. M.S. Thesis. Ankara: Middle East Technical University, 1998.

complex construction projects of longer durations from different categories. Furthermore, the methodology can be modied and converted into a more data (past record)-dependent characteristic by adaptation of modern techniques like genetic algorithms, articial neural networks and expert systems. References
[1] Edwards L. Practical risk management in the construction industry. London: Thomas Telford; 1995. [2] Flanagan R, Norman G. Risk management and construction. Cambridge: Backwell Scientic; 1993. [3] Pilcher R. Project cost control in construction. London: Collins; 1985. [4] Isidore LJ, Back WE. Multiple analysis for probabilistic cost and schedule integration. ASCE Journal of Construction Engineering and Management 2002;128(3):2119. [5] Barraza GA, Back WE, Mata F. Probabilistic monitoring of project performance using SS-curves. ASCE Journal of Construction Engineering and Management 2000;126(2):1428. [6] Chehayeb NN, AbouRizk SM. Simulation-based scheduling with continuous activity relationships. ASCE Journal of Construction Engineering and Management 1998;124(2):10715. [7] Senior BA, Halpin DW. Simplied simulation system for construction projects. ASCE Journal of Construction Engineering and Management 1998;124(1):7281. [8] Maio C, Schexnayder C, Knutson K, Weber S. Probablity distribution functions for construction simulation. ASCE Journal

[9]

[10] [11]

[12]

[13]

[14]

[15]

[16]

[17] [18]