Just‐in‐Time Retail Distribution: A Systems Perspective on Cross‐Docking

• Author: J.(Hans) C. Wortmann,Hans W. Danhof,Paul Buijs
• DOI: http://doi.org/10.1111/jbl.12135
• Published date: 01 September 2016
• Date: 01 September 2016

Just-in-Time Retail Distribution: A Systems Perspective on Cross-

Docking

Paul Buijs

1

, Hans W. Danhof

2

, and J.(Hans) C. Wortmann

1

1

University of Groningen

2

KPMG Management Consulting

Cross-docking is a just-in-time strategy for distribution logistics. It is aimed at reducing inventory levels and distribution lead times by cre-

ating a seamless flow of products from suppliers to customers. Prior supply chain literature has argued that creating such a seamless pro-

duct flows requires a holistic view on cross-docking management, aimed at synchronizing cross-docking operations at the distribution center

with its inbound and outbound network logistics. This paper provides an in-depth case study illustrating how cross-docking operations can be

managed more holistically in a retail distribution context. A discrete event simulation model has been developed to understand and improve the

cross-docking operations of a large grocery retailer in the Netherlands. The model is used to quantitatively evaluate two proposed changes that

exploit opportunities in the design and control of the retailer’s distribution network. An extensive real-world data set is used as input to the

model. Overall, the case and simulation results show that a holistic cross-docking management approach can indeed improve system-wide per-

formance, which further stresses the importance of making cross-dock operational decisions making and network decisions together.

Keywords: cross-docking; supply chain design; logistics planning and control; simulation

INTRODUCTION

Cross-docking is a just-in-time strategy for distribution logistics.

It is aimed at realizing a seamless flow of products from suppli-

ers to customers by moving products through the distribution

network without storing them as inventory in distribution centers.

Therefore, cross-docking should result in improved distribution

lead times and reduced inventory levels. Dating back to the

1990s, cross-docking has been applied in a range of industrial

settings, predominantly in parcel delivery (Forger 1995), less-

than-truckload trucking (Gue 1999), retail distribution (Stalk

et al. 1992), and the automotive industry (Witt 1998). In a recent

survey, over two-thirds of the respondents—embodying a cross-

section of the logistics industry—stated that cross-docking is part

of their distribution strategy portfolio (Saddle Creek Logistics

Services 2011). Another 15% expected to start cross-docking in

the near future. Following its sustained popularity in industry,

cross-docking has become a prominent topic in academic litera-

ture (Van Belle et al. 2012; Buijs et al. 2014).

This paper considers cross-docking in the context of retail

distribution. In that context, cross-docking is often deployed in

combination with the more traditional warehousing strategy.

Warehousing is part of a make-to-stock replenishment strategy,

in which distribution centers are used as a push-pull boundary.

In contrast with cross-docking, products are stored as inventory

to accommodate for demand variability and enable the consoli-

dation of products for transportation. Inventory levels are con-

trolled according to demand forecasts. The concept of holding

products in inventory allows for decoupling: the warehousing

operations inside the distribution center can be managed largely

in isolation from the inbound and outbound network logistics

processes. Accordingly, distribution center operations and net-

work logistics are often the responsibility of different managers

—each with a distinct set of performance indicators. At the dis-

tribution center level, the performance indicators are typically

focused on material handling efficiency. At the network level,

the focus is on delivery service levels and cost-effectiveness of

transportation.

Cross-docking fundamentally differs from warehousing. In

cross-docking, distribution centers are purposefully not used as a

push-pull boundary. Products are either moved directly from

inbound to outbound trailers or temporarily placed on the floor

(Apte and Viswanathan 2000). Moreover, cross-docked products

are not stored as inventory. As a result, the cross-docking opera-

tions inside a distribution center are tightly coupled with the cor-

responding inbound and outbound logistics processes.

Accordingly, supply chain management literature has stressed the

need for a holistic cross-docking management approach (Napoli-

tano 2000), aimed at synchronizing the cross-docking operations

inside a distribution center with inbound and outbound network

logistics (Apte and Viswanathan 2000; Vogt 2010).

Despite these fundamental differences, cross-docking manage-

ment in retail distribution is often organized according to tradi-

tional warehousing principles, that is, with separated

management responsibilities and distinct performance indicators

for cross-docking operations at the distribution center versus the

network level. Given the just-in-time nature of cross-docking,

this comes as a surprise. From lean and just-in-time production

theories, one would expect cross-docking performance indicators

to be geared toward the flow of goods, distribution lead times,

and the amount of work in progress (e.g., Hopp and Spearman

2011). A reason for the contrary could be that retail distribution

centers are often used for both cross-docking and warehousing.

Arriving truckloads can either be partially or entirely put away in

storage or cross-docked. At the outbound side of the distribution

center, the warehousing and cross-docking product flows are

consolidated just before shipment to the retail stores. In this

paper, we focus on the internal distribution center and network

Corresponding author:

Paul Buijs, Faculty of Economics and Business, Department of Opera-

tions, University of Groningen, P.O. Box 800, 9700 AV Groningen,

the Netherlands; E-mail: p.buijs@rug.nl

Journal of Business Logistics, 2016, 37(3): 213–230 doi: 10.1111/jbl.12135

© Council of Supply Chain Management Professionals

logistics processes that are directly related to the cross-docking

product flow, that is, those products that do not enter inventory

records.

The need for a holistic view on cross-docking management is

grounded in systems thinking (e.g., Ashmos and Huber 1987).

Thus far, academic cross-docking studies have not used real-

world data sets to empirically verify this need in detail. Studies

with a supply chain or operations management focus have relied

on logical argumentation and anecdotic evidence (e.g., Apte and

Viswanathan 2000; Napolitano 2000; Vogt 2010). In Operations

Research literature, cross-docking papers have focused on effi-

ciently solving well-defined isolated subproblems—addressing

either the design and planning of cross-dock facilities or the

design and planning of distribution networks with cross-docks

(Van Belle et al. 2012; Buijs et al. 2014). As a result, it remains

unclear how managers in retail distribution can organize and

manage their cross-docking operations more holistically. Owing

to the limitations in academic literature and practice, in this

paper, we address the following research objectives:

1 Investigate how cross-docking operations can be managed

holistically for a representative real-world case.

2 Empirically test if a more holistic management approach can

indeed improve cross-docking operations.

3 Propose performance measures that can reflect system-wide

changes in cross-docking performance.

Specifically, we study the case of a grocery retailer in the

Netherlands. Within that case, we introduce two changes to the

current cross-docking operations that illustrate how cross-docking

operations can be managed more holistically. First, we propose a

policy that dynamically assigns trailers to dock doors at a distri-

bution center, while carefully considering the inbound and out-

bound transportation planning characteristics. Second, we study

the relocation of preparatory cross-docking activities from a dis-

tribution center to a logistics facility upstream in the distribution

network. Both proposed changes exploit opportunities in the

design and control of the retailer’s distribution network to realize

system-wide cross-docking performance improvements. A dis-

crete event simulation model is developed to evaluate the effects

of the proposed changes.

The paper is organized as follows. First, the methodology

section justifies the use of a case study with discrete event simu-

lation modeling to attain our research objectives. The paper con-

tinues with a description of the case and conceptual model.

Descriptions of the simulation model, experimental factors, and

performance measures are provided in the simulation design sec-

tion, which is followed by a section presenting the simulation

results. More detailed information about the simulation model

and the real-world data used are presented in a separate Online

Supplement. In the conclusions and discussion section, we dis-

cuss the practical and theoretical implications that can be derived

from our study.

METHODOLOGY

In order to attain the research objectives described in the Intro-

duction, a large grocery retailer in the Netherlands was

approached for a case study on the cross-docking operations in

their distribution network. The case company—henceforth

referred to as “the retailer”—is considered to be leading with

regard to the design and control of its distribution network, in

which the broad implementation of cross-docking plays an

important role. During the research project, the retailer facilitated

many interviews and observation sessions and allowed unre-

stricted access to operational data and archival documents.

Accordingly, our case selection can be justified by the unique

research opportunity it provided (Yin 1994; Eisenhardt and

Graebner 2007) to identify, investigate, and describe examples

illustrating the need for a holistic cross-docking management

approach. A detailed description of the case is provided in the

subsequent section, which also elaborates how the case is repre-

sented in a conceptual model.

The main purpose of our case study is to understand and

improve the retailer’s cross-docking operations while maintaining

a systems perspective. For this purpose, we adopted a discrete

event simulation research approach. Robinson (2004) defines

simulation as the “experimentation with a simplified imitation

(on a computer) of an operations system as it progresses through

time, for the purpose of better understanding and/or improving

that system.”Simulation is a research method that is particularly

well-suited to represent the variability, interconnectedness, and

complexity often encountered in such systems (Law and Kelton

2000; Robinson 2004; Evers and Wan 2012). Accordingly, sev-

eral prior cross-docking studies have used simulation methods

(e.g., McWilliams et al. 2005; Wang and Regan 2008; Yang

et al. 2010).

In order to address our first research objective, discrete event

simulation modeling has been used to propose and test new

holistic cross-docking management solutions. To that end, the

model simulates the retailer’s cross-docking operations inside a

distribution center as well as the inbound and outbound logistics

processes. This simulation model also allows to address the sec-

ond research objective, that is, investigate if holistic cross-dock-

ing management can indeed lead to better performance. A large

real-world data set has been used as input to the simulation

model. For example, these data included a full year of actual

product flows through one of the retailer’s distribution centers

(including truck arrival and departure times). Assessing the need

for more holistic cross-docking management, requires perfor-

mance indicators that can reflect system-wide changes in cross-

docking performance. The definition of these performance indica-

tors constitutes our third research objective. While developing

the case, we introduce new performance indicators related to the

flow of products through the distribution network.

The validity of our simulation design has been determined by

assessing the conceptual model validity and experimental validity

and by performing white-box and black-box tests on the simula-

tion model (Robinson 2004). To that end, we visited several

logistics facilities throughout the retailer’s distribution network

and conducted interviews with employees and managers that play

a key role in the cross-docking operations—with due attention

being given to triangulation with the collected quantitative data.

In total, we logged 110 hr of observations and performed 11

interviews (which lasted between 1 and 2 hr each). Conceptual

model validity has been assessed through interviews discussing

the scope, level of detail, and correctness of the conceptual

214 P. Buijs et al.