See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/4368911

A warehouse design decision model — Case

study

Conference Paper · July 2008

DOI: 10.1109/IEMCE.2008.4618004 · Source: IEEE Xplore

CITATIONS READS

5 65

3 authors, including:

Guilherme A. B. Pereira
University of Minho
38 PUBLICATIONS 56 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Guilherme A. B. Pereira
letting you access and read them immediately. Retrieved on: 01 September 2016
 1

A Warehouse Design Decision Model – Case

Study
Carla A. S. Geraldes, M. Sameiro F. B. S. Carvalho, Guilherme A. B. Pereira

advantage of economies of scale; to provide value-added

processing and to reduce response time. Thus, warehouses
Abstract—Today’s competitive and volatile market requires will continue to be an important node at the logistic network.
flexibility, quality and efficiency from the logistics operations. In In distribution logistics where shorts lead-times and
this context, warehouses are an important link of the logistic
flexibility are essentials warehouse design and operations
chain and warehouse management plays an important role over
customer's service. Throughout this work we analyze a become more important and complex. Furthermore, the ever-
mathematical model aiming to support warehouse management increasing variety of products and the constant changes in
decisions. A case study is used for that purpose and the model customer demand has placed a tremendous emphasis on the
jointly identifies product allocation to the functional areas in the ability to establish efficient logistic operations. Warehouse
warehouse, as well as the size of each area. This case study also design and operations will be determinant to have quality,
evaluates the performance of the model when real data is used.
efficiency and flexibility.
Model is solved using LINGO 9.0 mixed-integer commercial
solver, and potential savings achieved using the proposed Typically, a design runs from a functional description,
technique are discussed. through a technical specification, to equipment selection and
determination of the layout. A layout must be modular,
Index Terms— Warehouse Operations, Facilities Planning and adaptable, compact, accessible and flexible and must be
Design, Mixed-integer programming, Case study. capable to respond to changing conditions, to improve space
utilization and to reduce congestion and movement. For these
reasons, warehouse design is a highly complex task, where,
I. INTRODUCTION sometimes, trade-offs have to be made between conflicting

W AREHOUSE management is, nowadays, a great challenge

in the field of supply chain management. Inventory
level management, warehouse design, warehouse operations
objectives.
According to Hassan [13] an important aspect of designing
a warehouse is its layout. The design of the layout should be
and space optimization and costumers’ requirements are some concerned with the arrangement of the functional areas,
examples of important challenges in this context. Warehouses determining the number and location of input/output (I/O)
must be flexible structures to provide quality, efficiency and points, determining the number of aisles, their dimensions and
effectiveness of the logistics operations in a very demanding, orientation, estimating space requirements, designing the flow
competitive and uncertain market. pattern, and defining picking zones.
On the other hand, modern supply chain management The majority of scientific research studies address isolated
principles compel companies to reduce or eliminate inventory problems. However, most real problems are unfortunately not
levels. Additionally a warehouse requires labour, capital and well-defined and often cannot be reduced to multiple isolated
information technologies, which are expensive resources. So, sub-problems. Therefore, design often requires a mixture of
why do we still need warehousing? According to Bartholdi analytical skills and creativity. Anyhow, research aiming an
and Hackman [3] there are four main reasons why warehouses integration of various models and methods is badly needed in
are useful: to consolidate products in order to reduce order to develop a methodology for systematic warehouse
transportation costs and to provide customer service; to take design (Rouwenhorst et al. [23]).
In this paper we tackle problems found during the redesign
Manuscript received February 1, 2008. A Warehouse Design Decision process of a warehouse in a Portuguese company. In
Model – Case Study.
Carla A. S. Geraldes is with the Department of Industrial Management,
particular, we adapt a mixed-integer programming model to
Polytechnic Institute of Bragança, Campus de Santa Apolónia, 5301-857 support two warehouse management decisions. The model,
Bragança, Portugal (phone: +351 273 31 30 50; fax: +351 273 31 30 51; e- developed by Heragu [14], jointly determines the size of
mail: carlag@ ipb.pt).
functional areas and allocates products to them. The results of
M. Sameiro F. B. S. Carvalho is with the Department of Production and
Systems Engineering, University of Minho, Campus de Gualtar, 4710-057 applying this model to this case study are analyzed and
Braga. (e-mail: [email protected]). potential savings achieved are discussed.
Guilherme A. B. Pereira is with the Department of Production and Systems
Engineering, University of Minho, Campus de Gualtar, 4710-057 Braga. (e-
mail: [email protected]).
 2

II. WAREHOUSES most important functions in most warehouses. SKU are

Warehousing is concerned with all the material handling retrieved from their storage positions based on customers’
activities that take place within the warehouse. They include orders and moved to the accumulation and sorting area or
the receiving of goods, storage, order-picking, accumulation directly to the shipment area. The picked units are then
and sorting and shipping. Basically, we can distinguish two grouped by customer order, packaged and stacked on the right
types of warehouses: distributions warehouses and production unit load and transferred to the shipping area.
warehouses. According to Berg [5], a distribution warehouse The design of a warehouse is a highly complex problem. It
is a warehouse in which products from different suppliers are includes a large number of interrelated decisions involving
collected (and sometimes assembled) for delivery to a number warehouses processes, warehouse resources and warehouses
of customers. A production warehouse is used for the storage organizations (Heragu [14]). Rouwenhorst et al. [23] classify
of raw materials, semi-finished products and finished products management decisions concerning warehousing into strategic
in a production facility. decisions, tactical decisions and operational decisions.
There are many activities that occur at a warehouse. Strategic decisions are long term decisions and always mean
Typically, distribution warehouses receive products - Stock high investments. The two main issues are concerned with the
Keeping Units (SKU) - from suppliers, unload products from design of the process flow and with the selection of the types
the transport carrier; store products, receive orders from of warehousing systems. Tactical management decisions are
costumers, assemble orders, repackage SKU and ship them to medium term decisions based on the outcomes of the strategic
their final destination. Frequently, products arrive packaged decisions. The tactical decisions have a lower impact than the
on large scale units and are packaged and shipped on small strategic decisions, but still require some investments and
units. For example, SKU may arrive in full pallets but must be should therefore not be reconsidered too often. At the
shipped in cases. operational level, processes have to be carried out within the
Typically, in a warehouse there are several functional areas constraints set by the strategic and tactical decisions made at
and flows. Next, we briefly describe some of the most the higher levels. At this level, the concern includes the
common areas and product flows (Fig. 1): operational policies such as storage policies and picking
operations.
Receiving area
Gray et al. [11] developed an integrated approach to the
pallets
design and operation of a typical order-consolidation
mixed unit loads
pallets, cartons

warehouse. This approach included warehouse layout,

equipment and technology selection, item location, zoning,
cartons

Pallet reserve area

picker routing, pick generation list and order batching. Due to
Auxiliary
the complexity of the overall problem, they developed a multi-
Carton pick area
process area
stage hierarchical decision approach. The hierarchical
cross docking operation

approach used a sequence of coordinated mathematical

models to evaluate the major economic trade-offs and to
cartons

Item pick area

reduce the decision space to a few alternatives. They also used

pallets

items
simulation technique for validation and fine tuning of the
rush orders

Sorting area resulting design and operating policies.

After determining warehouse location and its size, layout
decisions must include areas definition and what size should
Consolidation area be allocated to each functional area. Although addressing this
cartons mixed unit
problem is a strategic decision problem, it is strongly
loads
associated upon some tactical problems such as how the items
Shipping area will be distributed among the functional areas. Thus a joint
solution is desirable. However, the approach usually adopted
Fig. 1. Functional structure of a warehouse (Salvendy [24]). is to solve the problems sequentially by generating multiples
alternatives for the functional area size problem and then
At the receiving area products are unloaded and inspected determine how the products can be allocated for each of the
to verify any quantity and quality inconsistency. Afterwards alternatives. Heragu [14] developed a higher-level model that
items are transferred to a storage zone or are placed directly to jointly determines the functional areas size and the product
the shipping area (this is called a cross-docking operation). allocation in a way that minimizes the total material handling
We can distinguish two types of storage areas: reserve storage cost.
area and forward or picking area. The reserve area is where A case study is presented in the next section. The use of this
products stay until they are required by costumers’ orders. The case example allowed us to evaluate the model’s performance
picking area is a relatively small area typically used to store when real data was used and simultaneously redesign the
fast movers products. Most of the flows between areas are the warehouse.
result of replenishment processes. Order picking is one of the
 3

III. CASE STUDY (10%) and to walk through the entire warehouse to process an
F&F, located at Guimarães, north of Portugal, manufactures order (60%). This was the result of several factors such: the
and distributes house appliances to about 1200 customers lack of any arrangement of items on picking route lists, the
around the country (hypermarkets (85%), retail outlets (10%), absence of an efficient layout of items and the lack of an
hotels (1%), and others (1%)) and also some customers in integrated information system.
Spain and Africa (3%). Over 25% of the products are acquired Furthermore, inefficient operations planning and the
locally (including cutlery supplied from a factory owned by incapacity to establish demand forecasts was the source of
F&F), 20-30% are shipped from China and the remaining wrong inventory management policies. It was also observed
products are supplied from European countries. F&F that codification was not used, some of the fast movers’ items
processes an average of 40 customer orders per day (20 items were located on difficult access positions and the different
per order, on average). The company warehouse is a 4000 m2 functional areas were not well defined.
facility. Currently the firm owns and utilizes approximately The described situation confirmed the need of a redesign of
1200 m2 of warehouse storage space in a 3 km distant facility. the warehouse. Priorities were the elimination of obsolete
Items in the main warehouse are stored in pallets over 4000 items, sizing the functional areas and allocating products to
m2 of storage aisles (with 4 storage levels in the vertical the areas.
direction), corresponding to 2000 storage spaces. The facility As already mentioned we used a mixed-integer
is divided into two main sections: picking area and assembly programming model, developed by Heragu [14], to redesign
area. the warehouse. This model addresses sizing functional areas
A team of 15 workers are involved in warehouse operations and allocates items within the areas. The model uses data
which include receiving, storage, picking consolidation, readily available to the warehouse manager.
sorting and shipping and some value-added activities
(labelling, repackaging, etc). Receiving, storage operations
and picking operations were usually carried out during the IV. Decision Model
morning period. During the afternoon, while picking The model, adapted from Heragu [14], includes the
operations are still in progress, orders were checked for following functional areas: receiving, shipping, reserve and
completeness (including searching for missing items) for 3 picking. Thus, the three following material flow patterns are
teams and then packed and prepared for shipping. F&F possible (Fig. 2):
outsource the transportation activity and therefore it is
performed by a logistics provider.
The company was facing increasing difficulties in Flow 1
remaining competitive due to the high operation costs and low
levels of productivity to maintain existing throughput rate reserve
receiving

shipping
(output) as result of changes in customers demand pattern Flow 2
over the last years: increasing need of extra labour hours,
constant need of outsourcing labelling and repackaging tasks. picking

Additionally a decrease in service level was also reported as a Flow 3

result of i) difficulty in meeting customers demands of shorter
delivery times; ii) high level of errors in order processing
(some as result of stock-out situations) Fig. 2. Flows in the warehouse.
Finally, a major problem was the increasing need of extra
storage space. This problem was mainly due to the increasing • Flow 1 refers to a pattern that represents a typical
variety of products available and to large inventory levels warehouse operation. Items are stored in a reserve
required to face long lead-times (supplier lead-times can be up area and order picking is performed as required. It
to 4 months long). is assumed that, typically, only those items that
An ABC Pareto´s curve showed that, in the last few years, remain for long periods of time or items used for
only 43% of the existing SKUs were processed (positive replenishment will be allocated to this area.
turnover) and from the total set of existing customers only 8 • Flow 2 refers to items stored initially in the reserve
of them were responsible for 70% of the sales. area and then moved to the picking area. This
The lack of space and the lack of operations efficiency were pattern is considered for fast picking operations,
mainly due to the large number of obsolete items and really order consolidation or even to perform value-
slow moving stock items which were eating up square foot added operations.
after square foot with pallets laying down the aisles • Flow 3 refers to items that go directly to the
preventing pickers from reaching stock positions. picking area.
The study also showed that the largest portion of the
warehouse operations time was required to find the products The mathematical mixed-integer programming model that
 4

determines the flow pattern for each product and consequently area if the item is assigned to material flow 2 was given by the
the size of each of the functional areas within the warehouse warehouse manager.
is: All items are stored in pallet unit loads. Thus, considering a
1 m3 average volume pallet, we have a storing cost of 20€ per
( ) ( )
n 3 n 3
min ∑∑ H ij λij X ij + ∑∑ Cij Qi 2 X ij pallet per year to the reserve area and a cost of 35€ per pallet
i =1 j =3 i =1 j =3
3
per year to the picking area. In general, the storing cost is
∑X
j =1
ij = 1 ∀i given by

n n ⎧ 20 j =1
∑ (Q / 2 X ) + ∑ ( p Q / 2 X ) ≤ βTS
i =1
i i1
i =1
i i i2
⎪
Ci j [€ / year ] = ⎨20 × pi + 35 × (1 − pi ) j=2
⎪ j =3
n n ⎩ 35
∑ ((1 − p ) (Q / 2 X )) + ∑ (Q / 2 X ) ≤ γTS
i =1
i i i2
i =1
i i3

The cost of handling a unit load of each item depends on

β +γ ≤1 the size of the item, its handling characteristics, as well as the
LLR ≤ β TS ≤ ULR material-handling systems used in the respective flow pattern.
LLP ≤ γTS ≤ ULP Thus, handling costs tend to be more expensive for items
β, γ ≥ 0 allocated to the reserve area then to items allocated to the
picking area. Assuming items arrive on pallets and picking is
X ij = 0 or X ij = 1 ∀i , j
done in case unit loads we defined an average time for
receiving a pallet of 1 min. We also considered an average
Where the following notation is used: picking time of 1.5 min/carton to items assigned to flow 1, 0.5
min to items allocated to flow 2 and flow 3 and an average
⎧1 if item i is assigned to flow j replenishment time of 1 min to items at flow 2. Assuming a
X ij = ⎨
⎩0 otherwise cost of a picker of 25€/hour the handling cost is given by
β : proportion of availabe space assigned to reserve area
⎧0.417 + 0.625 × (cartons / pallets ) j =1
γ : proportion of availabe space assigned to picking area ⎪
H i j [€ / pallet ] = ⎨0.833 + 0.208 × (cartons / pallets ) j=2
λi : annual demand of item i ⎪0.417 + 0.208 × (cartons / pallets ) j=3
⎩
H ij : handling cost of a unit load of item i in flow j
Cij : storing cost of a unit load of item i in flow j We solved the model using LINGO 9.0 – a commercial
branch-and-bound based mixed-integer programming. Results
Qi 2 : average number of unit loads in inventory
are presented in next section.
TS : total available storage space
pi : average proportion of time a unit load of item i B. Results
spends in reserve area if item is assigned to flow 2 The model was solved including the 1500 items with
LLP , ULP : lower and upper storage space limit for the positive turnover, an average inventory level of 2257 unit
picking area loads and 2000 stock locations.
LLR , ULR : lower and upper storage space limit for the Results produced by the model defined a reserve area with
880 stock location units (44% of total space) and a picking
reserve area
area with 300 stock locations (15% of total space). Over 43%
of items were assigned to Flow 1, 45% were allocated to Flow
The above objective function minimizes the handling cost
2 and only 12% of items to Flow 3 (Fig. 3).
of each product and the corresponding storage cost. The
model’s integrity is observed by ensuring that each item is
assigned to only one type of material flow pattern and by
43%
keeping the space allocated to reserve and picking areas
within the limits imposed. The output of this model will play reserve
44%
receiving

an important role in the process of redesigning the warehouse.

shipping

45%

picking
A. Application to the F&F case study 15%
In this case study, an estimate for annual demand and for 12%
the average number of unit loads in stock was calculated
based on data from the past two and a half years. The estimate
for average proportion of time a unit load remains in reserve Fig. 3. Model’s optimal solution.
 5

[6] Caron, F. et al. 1998. “Routing Policies and COI-Based Storage Policies
in Picker-to-Part Systems”. International Journal of Production
As a result of this study F&F decided to redesign their Research 36, 3, 713-732.
warehouse: storage areas were identified and products were [7] Carvalho, J. et al. 2001. “A Auditoria Logística - Medir para Gerir”. 1st
allocated to the areas. Within the picking area items with ed. Edições Sílabo, Lisboa.
[8] Frazelle, E. 2002. “World-Class Warehousing and Material Handling”.
flows types 2 and 3 were located at lower levels improving
McGraw-Hill.
picking operations of fast movers products. Replenishment [9] Geraldes, C. 2007 “Modelos de programação linear aplicados à gestão
stock and items with low rotation were allocated at the upper das operações de um armazém – caso prático”. M.S. thesis, Dept.
levels. Production and Systems Engineering, University of Minho, Braga,
Portugal.
Additionally, other measures were also suggested aiming to [10] Goetschalckx, M. and Ratlift, H. 1990. “Shared Storage Policies Based
improve warehouse operations performance such as: on the Duration Stay of Unit Loads. Management Science 36, 9, 1120-
codification of stock positions, development of an information 1132.
[11] Gray, et al. 1992. “Design and operation of an order-consolidation
system and disposal of obsolete stock. warehouse: models and application”. European Journal of Operation
Research 58, 14-36.
[12] Hackman, S. and Platzman, L. 1990. “Near-optimal solution for
generalized resource allocation problems with large capacities”.
V. CONCLUSIONS
Operations Research 38, 5, 902-910.
The design of a warehouse is a complex problem due to the [13] Hassan, M. 2002. “A framework for the design of warehouse layout”.
Facilities 20, 13/14, 432-440.
large number of interrelated decisions sometimes with
[14] Heragu, S. et al. 2005. “Mathematical model for warehouse design and
conflicting objectives. To have a single decision model where product allocation”. International Journal of Production Research 43, 2,
all the decisions concerning warehouse design are integrated 432-440.
is still a challenging objective. This paper deals with [15] Koster, R. de and van der Poort, E. 1998. “Routing orderpickers in a
warehouse: a comparison between optimal and heuristic solutions. IIE
warehouse operation problems faced by a company that Transactions 30, 469-480.
manufactures and distributes house appliances. A wide and [16] Liu, C. 1999. “Clustering techniques for stock location and order-
complex set of problems were identified and some solutions picking in a distribution center”. Computers & Operations Research 26,
989-1002.
have already been implemented. In particular, this research [17] Mulvey, J. and Crowder, H. 1979. “Cluster analysis: an application of
addresses the redesign of the distribution warehouse. The Lagrangian relaxation”. Management Science 25, 4, 329-340.
mixed-integer programming model used considered both [18] Oliveira, M., Carvalho, M. and Pereira, G. 2003. “Warehouse
Operations: a case study”. BEX’03, Business Excellence I, Performance
functional area size determination and item assignment Measures, Benchmarking and Best Practises in New Economy.
problems. The approach revealed to be very useful in defining Universidade do Minho, Portugal, pp 577-579.
the new layout given that a significant reduction (about 40%) [19] Petersen, C. 1997. “An Evaluation of order Routeing Policies”. Journal
of the necessary storage area was achieved. Therefore, there of Operations and Production Management, 17, 1, 1096-1111.
[20] Rao, A. and Rao, M. 1998. “Solution procedures for sizing of
was no need of extra storage space. warehouses”. European Journal of Operation Research 108, 16-25.
Another advantage is that input data requirement are [21] Ratliff, H. and Rosenthal, A. 1983. “Order-Picking in a rectangular
usually readily available in most warehouses and the fact of Warehouse: A Solvable Case of the Travelling Salesman Problem.
Operations Research 31, 3, 409-421.
the model considers realistic constraints. [22] Roodbergen, K. and Koster, R. de. 2001. “Routing methods for
However, many others decisions are not included. The warehouses with multiple cross aisles”. International Journal of
output of this model may be taken as a starting point for Production Research 39, 9, 1865-1883.
[23] Rouwenhorst, B., et al. 2000. “Warehouse design and control:
further detailed warehouse operations decisions. Despite of Framework and literature review”. European Journal of Operation
the implemented solutions future research is needed on the Research 122, 515-533.
storage policy and on the order picking process. Afterwards it [24] Salvendy, G. ed. 2001. “Handbook of Industrial Engineering:
Technology and Operations Management”. 3rd ed. John Wiley & Sons,
will be possible to compare other performance measures such
Inc. Chapter 81, 2083-2109.
as the time required to find the products and the time required [25] Tompkins, J. et al. 2003. “Facilities Planning”. John Wiley & Sons.
to process an order.

REFERENCES
[1] Ashayeri, J. and Gelders, L. 1985. “Warehouse design optimization”.
European Journal of Operation Research 21, 285-294.
[2] Ballou, R. 1999. “Business Logistics Management”. 4th ed. Prentice
Hall. Upper Saddle River, New Jersey.
[3] Bartholdi, J. and Hackman, S. 2006. “Warehouse & Distribution
Science”. Release 0.76. Available: http://www.warehouse-science.com.
[4] Berg, J. van der et al. 1998. “Forward-reserve allocation in a warehouse
with unit-load replenishments”. European Journal of Operation
Research 111, 98-113.
[5] Berg, J. van der et al. 1999. “Models for warehouse management:
Classification and examples”. International Journal of Production
Economics 59, 519-528.