Efficient sensitivity analysis of PERT network performance measures to significant changes in activity time parameters

Using a Program Evaluation and Review Technique network model of a project schedule, a method is presented to estimate the effects of changes to the probability distribution for any activity time on several project schedule measures. Computational results show this method to be significantly faster and more accurate than a previously published approach, which estimated the effects of changes to the means of normally distributed activity times on expected project completion time and activity criticality. In addition, the new method allows more flexibility to model changes to activity times, including independent changes to parameters and even changes in the distributional form. Finally, the new method estimates the effects of these changes on several additional performance measures, including the probability of meeting a specified due date and a project (penalty) cost function. All desired estimates are obtained from a single set of simulation runs.