SMU-19-0045

KYY: STORAGE ASSIGNMENT IN A UNIT-LOAD WAREHOUSE

It was June 2019, and Lillian Lim, a consultant working for Mackor Consultancy, a third-
party consultancy in Singapore, was at the newest warehouse of her client, KYY Group,
a major third-party logistics provider in Singapore.

KYY Group was founded in 1975, and was a leading integrated end-to-end supply chain
management and logistics service provider in Singapore. It provided three main services
to its customers: (a) setting up physical warehousing and distribution hubs, (b) integrating
backend information systems between itself and the clients for inventory management,
and (c) planning the supply-chain process flow of inventory and information. KYY’s
business activities spanned Asia, covering countries such as Australia, China, Hong
Kong, India, Indonesia, Korea, Malaysia, Thailand, the Philippines, and Vietnam.

In 2013, the company had started construction of a US$200 million logistics hub. With
two million square feet of warehouse space and a million square feet of office space, it
would house KYY Group’s headquarter, as well as research and warehousing facilities.
The facility was due to open in late 2019, and would comprise the latest supply-chain
and logistics automation technology.

Lim’s company had been hired to do a study on the improvements that could be made to
increase the profitability of KYY’s warehouse operations. To do this, she needed to
understand the pattern of supply and demand expected at the warehouse, and the current
process of storage and retrieval of inventory.

She had to make a presentation on her preliminary findings in two weeks’ time to KYY
Group’s senior management.

Supply Chain and Logistics Industry in Singapore

In the early part of the 21st century, the supply chain and logistics industry in Asia was
becoming more important due to two developing trends. The first was the pattern of
international trade moving from the US and Europe to Asia. The second was the
advantage of Asia in providing low-cost logistics services and ‘frugal engineering’.1,2

1
Frugal engineering refers to a product design approach that emphasizes using the bare minimum of resources to create basic, no-
frills products, http://www.edb.gov.sg/content/edb/en/industries/industries/logistics-and-supply-chain-management.html, accessed
June 2014.
2
Economic Development Board of Singapore, Industries - Logistics and Supply Chain Management Singapore,
http://www.edb.gov.sg/content/edb/en/industries/industries/logistics-and-supply-chain-management.html, accessed June 2014.

This case was written by Associate Professor Yun Fong Lim and Associate Professor Marcus Ang at the Singapore
Management University. The case was prepared solely to provide material for class discussion. The authors do not
intend to illustrate either effective or ineffective handling of a managerial situation. The authors may have disguised
certain names and other identifying information to protect confidentiality.

Copyright © 2020, Singapore Management University Version: 2020-01-20

SMU-19-0045 KYY: Storage Assignment in a Unit-load Warehouse

Singapore, located in the heart of Southeast Asia, was geographically poised to offer
excellent service in the supply chain and logistics industry. In 2012, it boasted one of the
largest cargo airports in the region, with a throughput of two million tonnes of cargo. It
was also the largest container transhipment port, having seen 31.26 million twenty-foot
equivalent units (TEUs)3 of containers pass through its customs channels, which were
well known globally for their efficiency.4 The country was ranked first, in the World
Bank’s 2012 Logistics Performance Index, based on six factors – customs, international
shipments, infrastructure, logistics quality and competence, tracking and tracing, and
timeliness.5

In 2012, 20 out of the top 25 global logistics companies such as DHL, Schenker, and
UPS had regional or global headquarters based in Singapore. Singapore was also a key
supply chain management hub for large manufacturing companies in different industries
such as Panasonic, Dell, Diageo, and Louis Vuitton Moet Hennessy.6

KYY’s New Facility

KYY’s new facility contained a unit-load warehouse that provided storage services for
its existing clients. All other activities such as demand forecasting, customer order
processing, and production scheduling were done by the client. All products were stored
and retrieved in unit-load (pallet) quantities. Each pallet carried items of the same kind
and was generally handled one at a time. The warehouse owned a fleet of 15 forklifts to
handle all the products. The warehouse paid its employees by the hour and charged the
client by volume for storing and handling products. For the warehouse to be able to
increase its profits, it needed to find a way to increase efficiency and reduce costs. The
key was to minimise the travel distance of the forklifts. This would result in manpower,
time, and fuel savings.

Daily Operations of the Warehouse

The warehouse operated six days a week from Monday to Saturday. Each day had two
shifts. It received and put away products during the day shift between 8:00 AM and 5:30
PM. Sixty percent of the arriving pallets were from the client's manufacturing plant
located nearby, and the rest were imported from foreign countries. The warehouse was
informed one week in advance about the arrivals of pallets with 98% accuracy. The day
shift had 30 employees, who not only received and put away pallets, but also batched
customer orders that were transmitted from the client. All orders arriving during the day
shift were retrieved during the following night shift, which had 10 employees working
between 8:00 PM and 6:00 AM. Lim defined each period as a day so that the warehouse's
business processes were in line with her assumption: In each period, all arriving pallets
were stored before any pallets were retrieved for demand occurring in the period.
3
A TEU refers to a unit of cargo capacity, especially for container ships. These ships carry cargo in standard metal boxes, called
containers, which can be transferred easily to trains or trucks. TEU is an abbreviation for "twenty-foot equivalent unit." One TEU
represents the cargo capacity of a standard container 20 feet long, 8 feet wide, and (usually) a little over 8 feet high, or half the
capacity of a similar container 40 feet long. One TEU equals about 12 register tons or 34 cubic metres.
http://www.unc.edu/~rowlett/units/dictT.html, accessed June 2014.
4
Economic Development Board of Singapore, Industries - Logistics and Supply Chain Management Singapore,
http://www.edb.gov.sg/content/edb/en/industries/industries/logistics-and-supply-chain-management.html, accessed June 2014.
5
Supply Chain Digest, Global Supply Chain News: Bi-Annual World Bank Logistics Performance Index Places Germany on Top,
US at Number 9 Position, May 20, 2014, http://www.scdigest.com/ontarget/14-05-20-1.php?cid=8085, accessed June 2014.
6
Economic Development Board of Singapore, Industries - Logistics and Supply Chain Management Singapore,
http://www.edb.gov.sg/content/edb/en/industries/industries/logistics-and-supply-chain-management.html, accessed June 2014.

2/5
SMU-19-0045 KYY: Storage Assignment in a Unit-load Warehouse

Layout of the Warehouse

The warehouse’s storage area was 65 metres wide with ten aisles. Each aisle was 73.125
metres long. There were 18 single-deep racks. Every aisle had a rack on each side except
the end aisles. Each rack contained 48 sections and each section had four to five levels.
Only one pallet could be stored in each level of a section (refer to Exhibit 1 for further
details on the warehouse’s layout). Thus, each level of a section represented a storage
location. Lim grouped the storage locations of the warehouse into N classes. Since all
levels of the same section had identical store cost and identical retrieve cost, they
belonged to the same class (however, each class may contain multiple sections). When
the warehouse was out of space, any arriving products would be stored in an emergency
storage area with unlimited space but with high store and retrieve costs.

Product Information of the Warehouse

Lim had the data for the actual numbers of arriving and departing pallets of each product
in each period for 52 weeks. Exhibit 2 contains a list of arriving products and a list of
departing products. Each row in the list contains data corresponding to an arrival or a
departure of a product, and the columns correspond to: (1) date of arrival or departure,
(2) product (stock-keeping unit or SKU) number, and (3) number of arriving or departing
pallets.

Since the warehouse operated six days a week, a natural choice for the length of planning
horizon was six periods (that is, weekly planning). Figure 1 in Exhibit 3 shows the
autocorrelation of the total daily demand for all products over time. The peaks at lags 6,
12, and 18 suggest a weekly seasonality pattern.

Solution Methods

Given the above information of KYY, Lim used a mathematical modelling approach to
improve the warehouse operations. To derive an efficient storage assignment policy, Lim
proposed a linear programming model (LPM). Lim wanted to compare the LPM policy
with some well-known heuristics: The static class-based turnover policy (TOS) and the
dynamic class-based turnover policy (TOD). She wanted to compute the actual cost of
each policy using the data available. The implementation of each policy required the
means of daily demands. To achieve this, Lim used the first 16 weeks of data to estimate
these parameters in each period of week 17.7 She then implemented the policy for week
17 to compute its actual cost using the actual demands of the week. Following a moving-
average forecasting method, the means of daily demands for week 18 could be estimated
based on the data from week 2 to week 17, and so on.

The LPM policy determined vijt representing the number of pallets of product i to be
stored to class j in period t, and wijt representing the number of pallets of product i to be
retrieved from class j in period t. Lim rounded any non-integral vijt and wijt to the nearest
integer. If the rounded solution did not match the actual number of arriving (departing)
pallets, then she proceeded to store (retrieve) any additional pallets to (from) the most
economic class available. For example, if the demand for a product was 20 but the
7
Due to the weekly seasonality pattern, we can estimate the mean demand for a product in a day based on its demand in the
corresponding day of each of the previous weeks. For example, we use the actual demand for product i on Mondays of the first 16
weeks to estimate its mean demand on Monday of week 17.

3/5
SMU-19-0045 KYY: Storage Assignment in a Unit-load Warehouse

rounded solution was to retrieve 19 pallets from class 1 and 0 from other classes, then she
retrieved 1 more pallet from the most economic class that contained the product.

Lim used ait and dit to denote the number of arriving pallets and the mean demand,
respectively, for product i in period t. For the TOS policy, Lim defined the static turnover
rate of each product i as

,
where xit was the mean number of pallets of product i at the start of period t, and T=6
representing a planning horizon of one week. Arriving products were ranked according
to their static turnover rates. The product with the highest static turnover rate was assigned
to the most economic class. When a product was requested it was retrieved from the most
economic class containing the product. The TOD policy was similar to the TOS policy
except in each period t the former ranked arriving products according to their dynamic
turnover rates (ait+dit)/xit. Finally, the cost of each policy was computed based on the
store cost sj and the retrieve cost rj of each class j.

Finding the Best Solution

Lim was supposed to submit a preliminary report to KYY Group on her findings. In order
to prepare the presentation, Lim partitioned the warehouse into N=4 classes, each with
roughly the same number of storage locations. Shared storage was adopted within each
class: When a pallet was assigned to a class, it was stored at an arbitrary location in the
class. For the preliminary report, she considered M=10 products that had the highest flow
rate (arrivals and departures) over the 52 weeks in the data file.

The presentation and report should address the following questions:

1. For each policy, Lim should determine the class for each arriving pallet and each
departing pallet so that the total travel distance in week 17 was minimised. She
used the actual demands in Exhibit 2 to evaluate each policy.
2. Following the moving-average forecasting method, repeat Question (1) for each
of the remaining weeks (week 18 to week 52). Which policy gives the smallest
average weekly travel distance?
3. For Question (2), could Lim reduce the average weekly travel distance for each
policy if she increased the number of classes N? When Lim changed the value of
N she needed to decide on how the storage locations were grouped into different
classes.
4. For Question (2), what was the performance of each policy if the number of
products M increased?

Lim was excited to analyse the data and present her findings to show exactly how much
cost savings KYY Group could gain.

4/5
SMU-19-0045 KYY: Storage Assignment in a Unit-load Warehouse

EXHIBIT 1: WAREHOUSE'S LAYOUT8

EXHIBIT 2: DATA ON THE ACTUAL NUMBERS OF ARRIVING AND DEPARTING PALLETS

OF EACH PRODUCT IN EACH PERIOD OVER 52 WEEKS9

EXHIBIT 3: SEASONALITY OF THE TOTAL DEMAND FOR ALL THE PRODUCTS

Figure 1

8
Handouts will be provided for PDF versions
9
Handouts will be provided for PDF versions

5/5