Managing a single warehouse, multiple retailer distribution center

Journal of Business Logistics, 2000 by Yang, Kum Khiong

A challenge facing many producers and distributors of consumer goods is the task of distributing their products to many small retailers in different cities. To create a local presence, most of these companies maintain at least one warehouse in each city. A single-warehouse, multiple-retailer (SWMR) system is therefore common.

The popularity of such systems is clear from the many studies of SWMR systems.1 None, however, examines the effects of the environment, although Krajewski et al., suggest that shaping the environment may be more important than choosing the right systems or policies.2

This article examines the relative effect of various policies and environmental factors on the performance of an SWMR system. The objective is to identify the key areas that affect performance. This is an important issue because most companies have limited resources. Consequently, it is important to understand the effect of different policies and environmental factors so that limited resources can be allocated to the right areas.

A second objective is to identify the environments in which choosing the right policies is important. Although many studies have focused on finding the right inventory and vehicle-scheduling rules, none has examined how the environment affects policy choices. This study will help identify and define the environments in which selecting the right policies is critical.

This paper begins with a description of a simulation model of an SWMR system. Next, the environmental factors and policies are described. A full factorial simulation experiment is then proposed, and the simulation results are analyzed. Finally, the conclusion provides some suggestions and implications for managing SWMR systems.

SIMULATION MODEL OF AN SWMR SYSTEM

An SWMR system was modeled using computer simulation. The characteristics of the simulation model were carefully chosen to reflect as realistically as possible actual SWMR systems found in cities such as London, New York, and Singapore. The objective is to build a model that produces results that can be applied directly to a real-world setting with few qualifications. To achieve the level of realism, a literature search and company visits were conducted to gather information on the environments and policies found in practice.3

The city is modeled as a square of 50 by 50 kilometers with the retailers generated randomly over the square area. The travel distances between the retailers are assumed and computed as rectilinear distances. Following the travel speed of delivery vehicles observed in a previous study, the average speed of the delivery vehicles is fixed at 35 kilometers per hour.4

Previous studies have found that unloading time at a retail store increases with the size of the delivery.5 Trial simulation, however, indicated little performance difference between the use of constant and varying unloading times. Therefore, the unloading time at each store is assumed fixed at 0.25 hours per order.

To model deliveries from one warehouse to many small retailers, it is assumed that each delivery vehicle has an infinite load-carrying capacity. This assumption was affirmed by private correspondence with a number of distributors. In a city where retail space is expensive, retailers often order in small quantities to conserve the use of space. In addition, many suppliers report a growing trend of paying retailers for the use of their space. Both distributors and retailers therefore prefer small deliveries to conserve the use of space. Consequently, an urban delivery vehicle is often constrained by the total travel time and distance between stores rather than by its load-carrying capacity. An earlier study by Martin made a similar assumption for the deliveries of products from a central bakery to a pool of small customers.6

The literature review and company visits suggest seven important environmental factors and policies that may affect the performance of an SWMR system. The environmental factors are the number of stores, order processing time, and demand variability. The policies are the warehouse location, vehicle-scheduling rule, inventory rule, and order size.

Number of Stores (NS)

Intuitively, the number of stores affects the inventory and vehicle-scheduling rules. Many stores mean a large amount of inventory in the system, which may increase the value of a good inventory rule. Similarly, many stores increase the complexity of the delivery problem, which may increase the value of a good vehicle-scheduling rule.

The numbers of stores were examined at 40 and 200 stores. These two levels were selected to cover the numbers of stores operated by the distributors known to the author. For each number of stores, their locations were generated randomly within a city of 50 by 50 kilometers.

Order Processing Time (OT)

Order processing time is the interval between receiving an order from a retailer and dispatching the shipment from the warehouse. This interval is often cited as 50% to 70% of the total order cycle time between order placement and actual receipt by the customer.7 Order processing time obviously can be shortened by automation or by processing the order entry, credit check, and order filling activities in parallel.