Performance evaluation of two-worker operations in a worker-to-cell order assorting system

This paper studies an order assorting system (OAS) composed of multiple-parallel order assorting aisles (OAAs) in a retailer’s order fulfillment center. We evaluate two worker-assignment policies for a worker-to-part OAA: a dedicated-assignment policy and a shared-assignment policy. Our analytical models evaluate the mean value and variance of worker process time including assorting time, walk time, empty walk time, and blocking time when single-line products arrive randomly. From the worker process time, we identify the system flow time of product lines per each worker-assignment policy using G/G/1 and G/G/2 queuing models. We find that the shared-assignment policy performs better compared to the dedicated-assignment policy in spite of workers’ productivity losses incurred by blocking delays over high sort probability, slow walk speed, and high magnitude and variations of inter-arrival times. At the worker-centric break-even point of worker process time, where worker process time suggests selecting either policy, system flow time always suggests selecting the shared-assignment policy.