Abstract:
Resources for development projects are often scarce in the real world. Generally, many projects are to be completed that rely on a common pool of resources. Besides resource constraints, there exist precedence constraints among tasks within each project. Beyond the feed-forward dependencies between tasks, it is common in development projects the existence of feedback dependencies that constitute a new level of scheduling complexity for these projects. In this thesis, two genetic algorithm (GA) approaches (Variable Sample GA and Variable Length GA) are proposed for scheduling project activities in order to minimize the overall duration or makespan of development projects in a resource constrained, multi project environment without violating inter-project resource constraints or intra-project precedence constraints. Additionally, the proposed GAs allow for the existence of stochastic feedback between activities or rework of activities. These proposed GAs, with several variants of GA parameters, are tested on sample scheduling problems with and without stochastic feedback. The algorithms provide quick convergence to a global optimal solution and detect the most likely schedules, makespan range, as well as the minimum makespan and its schedule. Two objectives functions were used in this study: project lateness and portfolio lateness. Using several measures for project and portfolio scheduling problems (with feedback) characteristics, we conducted a comparative analysis between 31published priority rules and the proposed GAs. Test problems were generated to the specifications of project, activity, and resource-related characteristics including network complexity, resource distribution and contention and rework probability. The GAs performed better than the PRs as the level of iteration increases as well as the three other factors increased, including project complexity, resource utilization and resource loading. I close the thesis by providing managers with a decision matrix showing when (i.e. under what project-port
Description:
Thesis. M.E.M. American University of Beirut. Engineering Management Program, 2014. ET:6117
Advisor : Dr. Ali Yassine, Professor, Engineering Management ; Members of Committee: Dr. Bacel Maddah, Associate Professor, Engineering Management ; Dr. Walid Nasr, Assistant Professor, Engineering Management.
Includes bibliographical references (leaves 81-85)