Modeling Rework Cycle: Comparing Alternative Formulations

Rework cycle is at the heart of modeling projects, one of the major application areas of system dynamics. In this paper we introduce a new formulation for rework cycle in which multiple defects may exist in a task. We compare the performance of this model with three other formulations, two adopted from system dynamics literature and one agent-based formulation. This comparative study illustrates the impact of underlying assumptions about the nature of defects and homogeneity of tasks in different formulations on model behaviors and provides information needed for effective formulation selection decisions. The new formulation we introduce allows for capturing significant schedule over-runs due to a few tasks, with multiple defects, that may go through rework cycle multiple times. Its perfect mixing assumption, however, limits its precision in terms of simulating the final project quality. Sensitivity analysis informs the robustness of results to multiple projects parameters. We discuss the implications for selecting robust formulations in modeling project dynamics.