Operations systems with discretionary task completion

An extended abstract of a 2005 winner of the MSOM Society Student Paper Competition. Traditional manufacturing work and routine service work (e.g., bank tellers, checkout clerks) tend to have nondiscretionary task completion (NDTC) characteristics, while professional work and complex service work (e.g., engineers, physicians, financial analysts) tend to have discretionary task completion (DTC). In this paper, we show that discretionary task completion introduces quality as an additional factor for buffering variability. To gain insights into the managerial implications of this newly introduced variability buffer, we examine one-server and two-server systems with discretionary task completions. We first characterize the optimal control policy, which specifies how long the worker should spend on each task in a single job class system. We also introduce two simple threshold heuristics and show that they can be effective under certain conditions. Next, we compare our discretionary completion models with analogous nondiscretionary completion models in four different design and control aspects: (i) capacity, (ii) queue pooling, (iii) variability, and (iv) information. This leads us to two interesting and counterintuitive phenomena that are distinctive to systems with discretionary task completion. In sharp contrast to well-known NDTC system behavior, in DTC systems: (a) increasing capacity can intensify congestion and (b) adding process time variability may improve system performance.