Abstract Project control, as an iterative process crucial for managing project uncertainties, demands continual adaptation throughout the project’s lifespan. Activity duration unpredictability, inherents in project networks, poses challenges to meeting deadlines. In response, project managers collect real-time data and employ corrective actions to ensure timely and budget-compliant project delivery. Despite extensive research on project control methodologies, explicit methods for determining the optimal size of corrective actions in stochastic environments have received limited attention. This study addresses this gap by integrating probabilistic optimization models into project control strategies, emphasizing the minimization of corrective action costs. The model ensures that a specified percentile of the project duration distribution aligns with a predefined deadline, offering a probability of meeting the promised project duration. Two control strategies are proposed: preventive, involving a one-time optimization before project initiation, and protective, with continuous re-optimization at specific tracking periods during project execution. Computational experiments on artificial datasets reveal that protective strategies outperform preventive strategies, showing faster convergence and reduced variability in project duration distribution. Results indicate that, on average, a minimum of 6 re-optimizations during project execution is required to minimize both the cumulative cost of corrective actions and the average deviations between project duration realizations and the promised deadline.
Vaseghi et al. (Tue,) studied this question.