SKU arrangement on a unidirectional picking line

• DOI: http://doi.org/10.1111/itor.12550
• Author: Stephan E. Visagie,Jason Matthews
• Published date: 01 January 2019
• Date: 01 January 2019

Intl. Trans. in Op. Res. 26 (2019) 100–130

DOI: 10.1111/itor.12550

INTERNATIONAL

TRANSACTIONS

IN OPERATIONAL

RESEARCH

SKU arrangement on a unidirectional picking line

Jason Matthews and Stephan E. Visagie

Department of Logistics, University of Stellenbosch, PrivateBag X1, Matieland 7602, South Africa

E-mail: matthews.jsn@gmail.com [Matthews]; svisagie@sun.ac.za [Visagie]

Received 1 February2018; received in revised form 16 February 2018; accepted 24 March 2018

Abstract

An order picking system consisting of independent unidirectional picking lines has been investigated in this

study. A picking line forms a type of unidirectional carousel as multiple pickers walk in a clockwise direction

around a conveyor belt running down the centre of the picking line.There are three decision tiers in this order

picking system, namely the assignment of stock keeping units (SKUs) to different waves, the arrangement of

SKUs into locations on a picking line for a wave of picking and the sequencing of orders. Each tier has an

objective to minimise the total travel distance of pickers. The focus of this paper is on the arrangement of

SKUs on a picking line for a wave of picking. An exact mathematical formulation is introduced which is not

solvable to optimality for large real-life instances. Two heuristics known to be optimal for certain carousel

systems as well as two further heuristics using SKU correlationsare therefore tested. The heuristic approaches

are compared against lowerbounds generated by relaxing the exact formulation and a set of randomsolutions.

Known algorithms for carousel systems are not optimal in this carousel system and all heuristic approaches

perform approximately equally well. It is shown thatthe gains from addressing the order sequencing decision

tier outweighs that of the SKU arrangement decision tier. Moreover, the complexity of addressing the entire

system of decision can be reduced by arranging SKUs with a simple heuristic having a negligible impact on

solution quality.

Keywords:combinatorial optimisation; order picking methods; SKU assignment; unidirectional carousels

1. Introduction

Distribution centres (DCs) play a key role in many supply chains. DCs typically match supply with

demand by consolidating product, resulting in buffers of stock and a reduction in transportation

costs (Bartholdi and Hackman, 2011). The order pick operation plays a significant role within most

DCs and typically accounts for 60% of all DC costs (van den Berg and Zijm, 1999). de Koster et al.

(2007) describe order picking as the process of retrieving products from storage or buffer areas in a

response to customer requests.

From the review by van Gils et al. (2018b) and the follow-up study by van Gils et al. (2018a), it is

clear that multiple problemsplay a role and should be considered in designing efficient order picking

C

2018 The Authors.

International Transactionsin Operational Research C

2018 International Federation ofOperational Research Societies

Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA02148,

USA.

J. Matthews and S. E. Visagie/ Intl. Trans. in Op. Res. 26 (2019) 100–130 101

Fig. 1. A schematic representation of a picking line with mlocations.

systems. One of these problems is the effect of storage location assignment on the picker routing.

In the majority of the cases, storage location assignments have a statistical significant impact on

the routing of pickers and thus the efficiency of the total order picking system. In van Gils et al.

(2018b), it was found that in 21 of 24 papers that specifically investigated the interaction between

storage location and routing, there was a measurable relation between these two problems.

The problem investigated in this paper also considers the effect of storage location assignments

on the overall efficiency of an order picking system. However, we consider a very specific order

picking setup in which pickers walk in unidirectional cycles. No similar order picking setups could

be identified between the 24 studies reported in van Gils et al. (2018b), and we therefore provide a

detailed discussion of the background and setup for this order picking system.

The order picking operations for the DCs owned by Pep Stores Ltd. (PEP), a major retailer in

South Africa, is considered in this paper. PEP has three DCs in Southern Africa, in addition to 14

distribution hubs. Together, the DCs occupy more than 230,000 m2and distribute over 600 million

items per year across Southern Africa (https://www.pepstores.com/).

PEP preponderantly sells apparel, but has also been growing in the home d´

ecor and cellular

device market. They serve a target market consisting mainly of the low-income populationin South

Africa. PEP requires a large footprint of approximately 1500 retail outlets (stores) to reach its

market. PEP adopted a central inventory planning and management approach to keep costs low

with such a large number of stores. All inventory levels for the stores are managed centrally by the

planning department at the central office. Stock is thus planned at a stock keeping unit (SKU) level

and pushed to stores by a central planner rather than pulled by a store manager placing orders.

The order pick operation in PEP’s DCs is greatly influenced by this central planning approach.

All store requirements for the subset of SKUs scheduled to be picked during a specific operations

window (typically weekly) will be releasedby a central planner to a DC for all stores. This allows the

DC to process all store requirements fora single SKU in a single operation (or wave). PEP therefore

batches SKUs for collective picking in waves, rather than batching orders as is often the case in the

literature (Pan and Liu, 1995; Gademann and van de Velde, 2005; Hsu et al., 2005; Henn, 2012).

PEP uses 12 picking lines operating independently and in parallel to process waves of picking.

Figure 1 schematically illustrates the layout of a single picking line with m(typically 56) locations.

Figure 2 illustrates the layout of the set of picking lines in the DC. Each picking line processes SKUs

for a single wave at a time. Packed cartons from individual picking lines areconveyed onto the main

central conveyor which conveys them to the dispatch area where they are closed and shipped. Each

carton therefore holds stock from a single wave and no consolidation is done.

C

2018 The Authors.

International Transactionsin Operational Research C

2018 International Federation of OperationalResearch Societies