Application Example for Evaluating Networks Considering Correlation

Abstract

Current approaches to network scheduling do not consider the correlation between activity durations. When activity durations are correlated, the variability of path and project durations may be increased. High variability in a project's duration increases the uncertainty of completing the project by a target date. The model NETCOR (NETworks under CORrelated uncertainty) has been developed to evaluate schedule networks when activity durations are correlated. The NETCOR model builds upon a factor-based procedure to indirectly elicit correlation. An activity duration model disaggregates the effect of uncertainty by factors from a duration distribution (grandparent) for each activity. Correlation is captured by a child-distribution approach that further breaks down the factor-subdistribution (parent) based on the factor conditions. This paper demonstrates the practical application of NETCOR to a current construction project. Using the same inputs, the program evaluation and review technique and several simulation analyses that do not consider correlation also are evaluated. Comparison of the results shows the significance of considering correlation in scheduling analysis.

Risk Modeling of Dependence among Project Task Durations

“…The expected delay penalty (EDP) can be computed as follows: where f ( T ) = PDF of completing the project with a delay of T days and R = daily delay penalty. Because the integral in may be difficult to compute, one can use the following formula to approximate the result (Wang and Demsetz, 2000): where p ( T ) = probability of completing the project with a delay of T days.…”

“…where f (T) = PDF of completing the project with a delay of T days and R = daily delay penalty. Because the integral in Equation (6) may be difficult to compute, one can use the following formula to approximate the result (Wang and Demsetz, 2000):…”

“…More recently, Wang and Demsetz (2000) presented NETCOR to distribute the uncertainty in task durations to factor subdistributions (e.g., equipment, crew, and weather) that are further broken down into several conditions (e.g., better‐than‐expected, expected, and worse‐than‐expected). NETCOR captures the correlation between task durations by sampling from the same condition.…”

“…Thus, the task durations are not independent and with certain correlations (Woolery and Crandall, 1983). Ignoring such correlation can significantly distort the realization of the actual project duration (Wang and Demsetz, 2000). The need of recognizing the correlation between task durations in PERT analysis leads to the development of the following models.…”

Evaluating Schedule Uncertainty in Unit-Based Repetitive Building Projects

“…Schedule risk analysis methods such as PERT, MCS, and PNET are capable of analyzing uncertainty but they are insufficient in identifying the sensitivity of activities individually or the network as a whole to risk-factors. Furthermore, they ignore the correlation effect between activities [28,29]. They approach the uncertainty problem through accepting the activity durations between some estimated boundary values and trying to measure the variance of project completion time.…”