Throughput maximization in serial production lines with worksharing

In the traditional serial line model, tasks are assigned to unique workstations. We consider the case of partially cross-trained workers where some tasks may be performed at either of a pair of adjacent workstations. We investigate dynamic decision rules for guiding the operators' decision regarding when to complete the shared tasks and when to pass a unit along to the next workstation. The objective is to maximize throughput in constant work-in-process systems with low levels of work in process. A simple rule based on a buffer threshold level is shown to be nearly optimal for the two-stage line. Extensions to longer lines are provided and we show that myopic strategies that only consider adjacent workstations continue to perform well.