Abstract for: Using a Task-Level SD Approach to Help Create More Realistic Schedule and Budget Estimates for a Large Development Project

Development projects are notorious for being late and over-budget. Many times these problems arise, not from a lack of project management capabilities of the project managers, but instead from the use of simple scheduling and planning tools that can provide misleading or even incorrect schedule and budget estimates. There have been several examples of SD applied to projects at the project-level (i.e., treating the entire project at once with work “units” moving through the model). However, no significant work has been done using SD models at the task-level with intricate dependencies among the tasks (e.g., finish-to-start, start-to-start). The approach used on this project contains task-level SD models that allow for true project planning (e.g., Microsoft Project, Oracle Primavera). Results from the task-level SD simulation provided more realistic and believable estimates than the original commercial tool used for the defense project, which was Microsoft Project. The task-level SD model was able to point out where MS Project fell short with unrealistic calculations, insufficient information, or impossible scenarios. Furthermore, the task-level SD model was able to show how the schedule for the program could be accelerated to get the project back on track. The use of task-level SD models significantly improved the ability of the project to estimate budget and schedule considerations. Because of the real-world feedback loops related to losses of productivity due to overtime fatigue and overmanning on tasks, the team’s confidence in the estimates were much higher than traditional project planning and estimating tools (e.g., MS Project). Agentic AI (agent-based simulation)