Analytical models and optimal strategies for automated storage/retrieval system operations

The objective of this research is to provide exact reliable expressions for use in designing and operating automated storage/retrieval systems. We focus on the efficient operation of dual command cycles. The main performance criteria are travel time and system throughput.;We begin by developing a general analytical baseline for automated storage/retrieval system performance analysis. The baseline is a closed-form expression for the mean and variance of single and dual command cycle time. The model can be effectively used for evaluating any storage policy, i.e., random, dedicated or class-based storage. We give examples to illustrate how the formulas can be used for evaluating each storage policy.;With an evaluation baseline, our attention turns to specific improvement strategies. Contour line configurations for storage location assignment are developed first. We develop a general scheme to generate contour line configurations for dual command operations. To investigate the effects of alternative contour line configurations on system performance, a series of experiments are performed. The storage policies considered are random storage, priority-based open location (POL) storage, turnover-based storage, and 2-class storage. The performance of each contour line configuration is measured in terms of the expected dual command travel time.;Next, we develop optimal dwell point policies for automated storage/retrieval systems. Based on the fact that dwell point policies minimize the completion time of the first transaction after the storage/retrieval machine becomes idle, we show that there is a unique optimal dwell point policy, regardless of other system parameters. Then a variety of return paths to the dwell point are introduced and studied.;Finally, an end-of-aisle order picking system with inbound and outbound buffer positions is studied. This is usually referred to as a miniload system with a horse-shoe "front-end". The system is modeled as a two-stage cyclic queueing system consisting of one general and one exponential server with limited capacity. The cyclic queueing system is then analyzed by using the customer-hole duality concept. Closed-form expressions for the stationary probability and system throughput are developed. We also obtain the proportion of picker-idle and storage/retrieval machine-idle time by noting that the throughput of the picker is equal to that of the storage/retrieval machine. Then, a design problem to determine the optimal number of inbound and outbound buffer positions and a control problem to determine the optimal number of storage containers in the system are studied. The effect of buffer size on system throughput is also investigated.