Managing Projects with Stochastic Durations

This research addresses analyzing and responding to uncertainty in projects with stochastic task durations. First we examine the effect of contractor flexibility (or agility) on project completion times. We find that this impact can be significant depending on the size and the structure of the project network.Next we study a stochastic time-cost trade-off problem with penalties for exceeding a project deadline. In considering this problem, we take a contingency approach to decision making where crashing decisions are made dynamically throughout project execution. For serial projects we develop a dynamic programming algorithm as well as a variety of heuristic methods. Extending and modifying these methods for general projects allows us to deal with more complex network structures. Specifically, we propose hybrid dynamic programming/linear programming algorithms and simulation-based algorithms. We perform computational studies to assess the performance of each method and to compare the contingency approach with a static approach (where all crashing decisions are made before the project start time). Finally, we study the case with penalties, incentives, and overhead costs. We find that when the project cost function is not convex, the dynamic programming solution may become non-monotonic, which requires further modification of the methods. We show that the performance of our algorithms does not deteriorate with inclusion of additional parameters. In fact, the gaps between the case with perfect information and the methods presented herein seem to be smaller than in the penalty only case