| Abstract: |
| Accurate estimation of activity durations is fundamental in project scheduling, industrial planning, and service systems where uncertainty and variability affect decision-making. Traditional Program Evaluation and Review Technique (PERT) models rely on three subjective time estimates - optimistic, most likely, and pessimistic - to approximate expected activity duration. However, empirical activity durations are frequently skewed, limiting the accuracy of the classical PERT formulation. Existing literature studied approaches utilizing normal or lognormal approximations that worked for data with either symmetric or the right skewed underlying activity distributions, but their applications are limited for moderate or heavily left skewed data. Abdel-Raheem et al. (2018) introduced refinements with triangular distributions due to its comprehensibility to the project planner and proposed a framework allowing the determination of the contract time using deterministic and probabilistic scheduling techniques. This approach has its own limitations. To overcome these limitations and enable broader applications, this study proposes an enhanced method for activity duration estimation using maximum likelihood estimation applied to lognormal distributions. Using empirical production-rate data, we evaluate estimation accuracy and demonstrate improvements over existing methods. |
|