Experimental investigation of shared storage assignment policies in automated storage/retrieval systems

IIE Transactions, August, 1999 by Sadan Kulturel, Nur E. Ozdemirel, Canan Sepil, Zafer Bozkurt

In this study, two shared storage assignment policies are compared in an Automated Storage/Retrieval System (AS/RS) by using computer simulation. The AS/RS is assumed to operate under a continuous review, order quantity, reorder point inventory policy. The average travel time of the AS/R machine for storing and retrieving products is used as the main performance measure. Sensitivity of the system to product variety, inventory replenishment lead time, and demand rate is investigated, as well as the effects of the inventory policy and the product classification technique used. Our main conclusion is that the turnover-based policy, in general, outperforms the duration of stay-based policy. However, the difference between the performance of the two policies becomes insignificant under certain conditions.

1. Introduction

Automated Storage/Retrieval Systems (AS/RS) have given industry the benefits of dramatically faster retrieval, condensed and safe storage, and increased productivity, as well as the capability to integrate storage with inventory control and material requirements planning systems. Some characteristics of AS/RS such as high initial investment, inflexible layout and limited capacity, force managers to highly utilize and efficiently operate these systems. The chosen method to operate an automated warehouse has a considerable effect on its efficiency and design.

A number of interesting problems arise in designing and operating an AS/RS. Early research efforts considered the optimal scheduling of automatic warehousing systems, with eventual feedback to their design. Basic elements of scheduling are pallet assignment, interleaving and storage assignment. Pallet assignment is the assignment of multiple items or product types to the same pallet. This problem is often ignored by assuming each pallet contains only one item. Such systems are called unit load systems. Interleaving is concerned with the rules for sequencing storage and retrieval requests. Interleaving allows both a store and a retrieve operation on a single trip from/to the Input/Output (I/O) point. The last element is the storage assignment, i.e., the assignment of unit loads to storage locations. A storage assignment policy is a set of rules that determines where the unit loads of different products will be located in a warehouse.

Our aim in this study is to compare two storage assignment policies in a unit load AS/RS where interleaving is allowed. The policies we consider are turnover-based and Duration Of Stay (DOS)-based storage assignment policies with three classes. The turnover-based policy, which has been intensively studied and widely used in the last decade, assumes that the items (product types) are grouped into three classes based on their mean turnover (or demand) rates. The AS/RS is also partitioned into three respective regions. The class of items having the highest turnover rate is then assigned to the region closest to the I/O point. The DOS-based policy is a relatively new approach proposed by Goetschalckx and Ratliff [1]. Assuming that an item is ordered in a quantity of Q units and individual units are retrieved at a rate of [lambda], the expected DOS of the ith unit in the lot is i/[lambda] for i = 1,2, ... , Q. Hence DOS is a unit characteristic rather than an item characteristic. The DOS-based policy classifies t he units of all items according to their expected DOS. Then the class of units having the shortest DOS is assigned to the closest AS/RS region.

In our AS/RS, the continuous review, order quantity, reorder point (Q,r) inventory model is used. The most important contribution of this study to the AS/RS literature is incorporation of the (Q,r) model in the ordering policy of the AS/RS, and comparison of the storage assignment policies under a stochastic inventory model.

Using computer simulation, we have experimented with turnover- and DOS-based storage assignment policies by changing the levels of four factors: ordering cost; backorder cost; ABC classification of parts; and AS/RS capacity shrinkage rate. Regression analysis is used to quantify the relationship between these factors and each of the two response variables, namely average travel time of the AS/R machine and balance of the system. Our main objective is to minimize the average travel time of the AS/R machine for storage and retrieval of unit loads.

The paper is organized as follows. Section 2 gives an overview of the AS/RS literature. In Section 3, the AS/RS environment under consideration is described including our assumptions. Experimental results for instant replenishment with arbitrary order quantities are given in Section 4. Design and analysis of the experiment with the (Q,r) inventory policy are discussed in Section 5. Section 6 is dedicated to the effects of the inventory policy and the product classification technique on system performance. Finally, the main conclusions of this study are summarized in Section 7.

2. Earlier work on storage assignment policies