J Syst Sci Syst Eng (Sep 2008) 17(3): 353-366

DOI: 10.1007/s11518-008-5080-z

ISSN: 1004-3756 (Paper) 1861-9576 (Online)

CN11-2983/N

AN ALGORITHM FOR THE DETERMINATION OF THE ECONOMIC

ORDER QUANTITY IN A TWO-LEVEL SUPPLY CHAIN WITH
TRANSPORTATION COSTS: COMPARISON OF DECENTRALIZED WITH
CENTRALIZED DECISION
Armand BABOLI1

Mohammadali Pirayesh NEGHAB2

Rasoul HAJI2

Laboratoire LIESP, INSA de Lyon, 19 avenue Jean Capelle, 69621Villeurbanne Cedex, France
[Link]@[Link] ( )

Department of Industrial Engineering, Sharif University of Technology, Azadi Av. Tehran, Iran
a_pirayesh@[Link], haji@[Link]

Abstract
This paper considers a two-level supply chain consisting of one warehouse and one retailer. In this
model we determine the optimal ordering policy according to inventory and transportation costs. We
assume that the demand rate by the retailer is known. Shortages are allowed neither at the retailer nor
at the warehouse. We study this model in two cases; decentralized and centralized. In the decentralized
case the retailer and the warehouse independently minimize their own costs; while in the centralized
case the warehouse and the retailer are considered as a whole firm. We propose an algorithm to find
economic order quantities for both the retailer and the warehouse which minimize the total system cost
in the centralized case. The total system cost contains the holding and ordering costs at the retailer and
the warehouse as well as the transportation cost from the warehouse to the retailer. The application of
this model into the pharmaceutical downstream supply chain of a public hospital allows obtaining
significant savings. By numerical examples, the costs are computed in MATLAB to compare the costs
in the centralized case with decentralized one and to propose a saving-sharing mechanism through
quantity discount.
Keywords: Inventory control, transportation, medical supply chain, saving-sharing mechanism

1. Introduction

in a supply chain.

The goal of many research efforts related to

supply

chain

supply

chain

costs.

a criterion to decide replenishment policy. In the

Inventory holding and transportation costs are

replenishment process, beside the inventory cost,

regarded as the most important operational costs

the transportation cost is a major cost factor

reduce

to

in

management only consider the inventory cost as

to

is

researches

propose

mechanisms

management

Many

operational

The original version was presented on ICSSSM06.

Systems Engineering Society of China & Springer-Verlag 2008

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

354

which affects the shipment size. In this paper,

affected by the shipment size and vice versa. So,

our purpose is to investigate the inventory cost

it is important to determine the economic order

as well as the transportation cost in a two-level

quantity to minimize the overall logistic costs.

supply chain and to compare centralized with

Ghanshan (1999) introduces a three-level supply

decartelized decision making.

chain consisting of a number of identical

The paper is organized as follows. Section 2

retailers, one central warehouse, and a number

is devoted to literature review. Section 3

of identical suppliers. In this model, the

describes the model. Sections 4 and 5 present

objective function consists of ordering, holding

cost

and

and the transportation costs. Specifically, the

decentralized cases, respectively. Section 6

transportation cost is considered as a function of

introduces our algorithm to find optimal solution

the quantity to order, but ignores the capacity of

in the centralized case. Numerical results are

the vehicle.

evaluation

for

the

centralized

described in section 7; and finally section 8

Swenseth and Godfrey (2002) demonstrates

presents the conclusions and suggestions for

that straightforward freight rate functions can be

further research in this area.

incorporated

into

inventory

replenishment

decisions without compromising the accuracy of

2. Literature Review

the

decision

and

without

adding

undue

Literature review is presented in two parts;

complexity to the decision process. Ertogral et al.

first, the inventory models in a supply chain

(2007) consider a vendor-buyer supply chain

with

second,

model and incorporate the transportation cost.

centralized and decartelized models in supply

The transportation cost depends on shipment

chains.

size.

transportation

costs,

and

All-unit-discount

transportation

cost

structures with and without over declaration

2.1 Inventory Models in a Supply Chain

with Transportation Cost

have been considered in their work. Huang et al.

(2005) consider a two-level supply chain in

Many researches have been undertaken in

which a warehouse delivers its products to many

supply chains which consider the transportation

retailers. Each retailer faces a constant and

cost as part of the ordering cost and, thus,

deterministic demand. The objective is to

assume it to be independent of the size of the

determine an optimal stationary ZIO (Zero

shipment; e.g. Hill (1997), Goyal and Nebebe

Inventory

(2000), and Hoque and Goyal (2000). These

warehouse and retailers in which the average

papers deal with determining the economic

transportation and inventory cost is minimized.

policy of production and shipment for a

The transportation cost consists of a fixed cost

single-vendor

integrated

and a variable cost which is linearly proportional

production-inventory system. The objective is to

to the amount of quantity ordered. Zhao et al.

minimize the total joint costs incurred by both

(2004) propose an algorithm to find the optimal

the vendor and the buyer.

ordering

single-buyer

In practical cases, the transportation cost is

Ordering)

quantity

policy

and

for

both

frequency

for

the

supplierretailer logistic system in which the

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

355

fixed and the variable transportation as well as

and at the retailer as well as the holding costs at

multiple use of the vehicle cost are considered.

the warehouse and the retailer.

heuristic

In supply chains in which the sites belong to

decomposition method to solve the problem of

independently owned companies, decentralized

an inventory-transportation system, minimizing

decision is made. Axster (2005) considers a

the long-run total average costs (major- and

two-echelon distribution inventory system with

minor-ordering, holding, backlogging, stopover

a central warehouse and a number of retailers.

and travel). The decomposition algorithm works

The system is controlled by continuous review

by using separate calculations for inventory and

installation stock (R, Q) policies with given

routing decisions, and then coordinating them

batch quantities. He develops an approximate

appropriately.

technique for determining the backorder cost of

Qu

et

al.

(1999)

propose

the warehouse, and let the warehouse choose its

2.2 Centralized and Decartelized Models

in Supply Chains

reorder point so that the sum of expected

holding and backorder costs are minimized.

Some researchers consider a supply chain as

Given the resulting warehouse policy, the

a single firm; i.e. all sites belong to the same

retailers similarly optimize their costs with

organization. Thus, they investigate centralized

respect to the reorder points.

models. Axster (1993, 1996), Forsberg (1995,

Andersson and Marklund (2000) study

1996), Matta and Sinha (1995), Das and Tyagi

decentralized inventory control in a two-level

(1997) Seifbarghi and Akbari (2006), and

distribution system. They consider a two-level

Mrklund (2002) study centralized models in the

distribution system with a central warehouse and

inventory system of two-level supply chains.

N non-identical retailers. All installations use

They investigate different models with respect to

continuous review installation stock (R, Q)

the type of demand distribution function, the

policies. They present an approximate cost

type of shortage (lost-sale or backorder) and the

evaluation technique, where the retailers replace

inventory replenishment policy. These models

their stochastic lead times by correct averages.

include a central warehouse and a number of

They introduce a modified cost-structure at the

retailers. The retailers face stochastic demand.

warehouse

They develop an exact or an approximate

inventory control problem into N+1 single-level

method to evaluate the total system cost. The

sub-problems, one problem per installation.

to

decompose

the

multi-level

total system cost consists of the holding cost at

Abdul-Jalbar et al. (2003) compare the effect

the warehouse and at the retailers as well as the

of centralization versus decentralization on the

shortage cost at the retailers.

total system cost. They consider a two-level

Silver et al. (1998) investigate a supply chain

supply chain consisting of one warehouse and a

consisting of one warehouse and one retailer in

number of retailers. By numerical examples,

which the external demand rate is known with

they show that as the number of retailers

certainty. In their model, the total cost consists

increases so does the number of instances in

of fixed replenishment costs at the warehouse

which the decentralized policy is better. In

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

356

addition, given a number of retailers, sensitivity

strategy.

analysis indicates that under specific conditions

Viswanathan and piplanis model to allow for a

Mishra

(2004)

generalizes

the

of the unit replenishment and holding costs at

selective discount policy that excludes some

the warehouse, the centralized policy can make

buyers to minimize the suppliers total cost.

Li and Liu (2006) consider a supplier-buyer

better solutions.
In some two-level supply chain inventory

system with probabilistic customer demand.

systems (containing warehouse and retailer), a

They show that quantity discount policy is a way

coordination mechanism is a way to gain lower

to achieve coordination. They propose a method

total system cost. Coordination usually generates

to divide the profit due to joint decision between

savings for the warehouse and losses for the

the buyer and supplier and the optimal quantity

retailer. In these situations, researchers propose

discount policy is obtained by using this profit

saving-sharing mechanisms to interest the

sharing method. We present now our model,

retailer in accepting the coordination. The

which is based on this previous work.

quantity discount policy is a saving-sharing

3. Model

mechanism to achieve coordination.

Chen and Chen (2005) present a joint

In this paper we consider a two-level supply

replenishment arrangement with a two-echelon

chain consisting of one warehouse and one

supply

retailer, as illustrated by Figure 1.

chain,

having

one

supplier

or

manufacturer and one buyer or retailer, facing a

Warehouse

deterministic demand and selling a number of

External

products in the marketplace. They considered a

Supplier

situation involving a major setup cost and minor

Flow

processing cost. The retailer also has major

Information

setup costs due to economies of scale in

minor

setup

cost

for

each

Customer
of

Figure 1 A two-level supply chain

transportation and distribution expenses and an

item-specific

Retailer

We obtain the economic order quantity (EOQ)

additional item involved in the order. They

for our retailer and our warehouse in two

proposed both centralized and decentralized

different cases; centralized and decartelized. In

decision models to determine the best solution to

the decentralized case the retailer and the

minimize costs and by a search algorithm and

warehouse independently minimize their own

numerically illustrated present the benefits

costs. In the centralized case the warehouse and

generated from such an arrangement.

the retailer are considered as a whole firm and

Viswanathan and Piplani (2001) study a one

the objective is to minimize the total system cost.

vendor, multi-buyer supply chain. They analyze

Conversely to the most similar models which

the benefit of coordinating supply chain

determine EOQ just based on inventory costs, in

inventories

common

this model, the transportation cost from the

replenishment epochs. The vendor offers the

warehouse to the retailer is also taken into

price discount to entice the buyers to accept this

account.

through

the

use

of

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

357

It is necessary to specify that the proposed

the warehouse to the retailer is made by a single

model can be applied when the inventory control

vehicle without splitting. Each type has its own

policy is:

fixed and variable costs and capacity size. Table

Periodic review: order point, order quantity

1 shows the context of our transportation

(R, s, Q); every /R/ unit of time, we check

scheme.

the inventory position, if it is at or below

With the notations in Table 1, it is assumed

the reorder point /s/, we order the quantity

that

/Q/.

F2=F1+q1(v1-v2),

Continuous review: order point, order

equations

quantity, (s, Q); the fix quantity Q is

over-declaration. Hence, the transportation cost

ordered whenever the inventory position

varies according to the order quantity as shown

drops to the recorder point s or lower.

in Figure 2.

Table 1 Computational results for small problems

Vehicle
Type

Capacity

Fixed
Cost

Variable Cost

q1

F1

v1

q2

F2

v2

q3

F3

v3

F1<F2<F3,
are

v1>v2>v3,

q1<q2<q3,

F3=F2+q2(v2-v3).
supposed

to

and
These

avoid

any

Assumptions:
1.

The retailer faces deterministic demand

with a constant demand rate.

2.

The demand is confined to a single

3.

Shortage is allowed neither at the

item.
retailer nor at the warehouse.
4.

The

warehouse

incurs

the

transportation cost from the warehouse

Transportation

to the retailer.

Cost (TC)

5.

F3

The transportation time for an order to

arrive

at

the

retailer

from

the

warehouse is constant.

F2

6.

Total TC

The lead time for an order to arrive at

the warehouse is constant.

TC per unit

7.

There

is

no

lot-splitting

at

the

8.

Order quantity of the retailer does not

warehouse.
q1

q2

q Quantity

exceed the capacity of large vehicle.

Order

Figure 2 Variations of transportation cost

Notations:
1.

D: Demand rate at the retailer.

In this model, we suppose three types of

2.

Ar: Ordering cost for the retailer.

vehicles, namely, small (S), medium (M) and

3.

Aw: Ordering cost for the warehouse.

large (L), and the delivery of each order from

4.

hr: Rate of holding cost at the retailer.

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

358

5.

hw: Rate of holding cost at the

economic

warehouse.

respectively as follows:

6.

Qr: Order quantity of the retailer.

7.

Qw: Order quantity of the warehouse.

8.

Cr (Qr ) =

Tr: The time interval between any two

consecutive orders of the retailer.

9.

Tw: The time interval between any two

consecutive orders of the warehouse.

10. LTr: Lead Time (Transportation Time)

from the warehouse to the retailer.

We now present the first method to make

model,

that

is,

retailer

are

DAr hr Qr
+
Qr
2

(1)

2 DAr
hr

(2)

We assume that lot-splitting of retailer order

is not authorized at the warehouse. Furthermore,
shortage is not allowed at the warehouse. Thus,

retailer. For optimal solution the arrival of an

13. CTi: Total system cost.

this

of

integer multiple (n) of the order quantity of the

12. Cwi: Total cost of the warehouse.

in

Q*r =

quantity

the order quantity of the warehouse must be an

11. Cr: Total cost of the retailer.

decisions

order

in

decentralized way.

order to the warehouse corresponds to the

delivery of an order to the retailer. Thus, the
maximum inventory level at the warehouse is
QwQr. Changes in the inventory level at the
warehouse are shown in Figure 4.

4. Decentralized Case
In the decentralized case, the retailer and the

Inventory
Level

warehouse intend to optimize their own costs

independently. Demand rate at the retailer and

Qw-Qr

transportation time to the retailer are supposed

to be constant. Shortage is not allowed at the

Tr

Qr

retailer, therefore, the retailer uses a simple EOQ

model and inventory level behaves as depicted
in Figure 3.

Tw

Inventory
Level

Time

Figure 4 Inventory level at warehouse

LTr

In the decentralized case, the total cost of

Qr

warehouse is the sum of the holding and

ordering costs plus the transportation cost from
Time

Figure 3 Inventory level of the retailer

The total cost of retailer is the sum of
holding and ordering costs. The total cost and

the warehouse to the retailer. Thus, the total cost

of the warehouse is:
Cwi (Qw ) =

DAw hw (Qw Qr )
+
Qw
2

DFi
+ Dvi ,
Qr

i = 1, 2,3

(3)

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

q(i 1) < Qr qi

(3.1)

Qw = nQr

(3.2)

n is int, q0 = 0

(3.3)

Index i denotes the vehicle type; 1, 2 and 3

respectively for S, M and L. For the shipment of
retailers orders, one of these types must be

359

transportation cost from the warehouse to the

retailer. Thus, the total system cost can be
written as:
DAw hw (Qw Qr ) DAr hr Qr
+
+
+
Qw
2
Qr
2

CTi (Qw , Qr ) =

chosen based on the size of the order quantity.

Hence, the ordering-size is restricted by the
vehicle capacity.
We substitute nQr for Qw in (3). Thus, we can
rewrite (3) as:

DFi
+ Dvi ,
Qr

i = 1, 2,3

CTi (n, Qr ) =

(4)
(4.1)

n is int, q0 = 0

(4.2)

n equal to zero, we obtain the approximate

optimal value of n as follow:
2 DAw
hwQr 2

(5)

Or, from (2) we can rewrite:

hr Aw
hw Ar

q(i 1) < Qr qi

(7.1)

Qw = nQr

(7.2)

n is int, q0 = 0

(7.3)

(6)

We now present the other approach of

decision making, that is, the centralized case.

DAw hw (n 1)Qr DAr hr Qr

+
+
+
nQr
2
Qr
2

Setting the derivatives of Cwi with respect to

n* =

(7)

(7) as:

q(i 1) < Qr qi

n* =

i = 1, 2,3

Substituting nQr for Qw in (7) we can rewrite

DAw hw ( n 1)Qr
+
nQr
2

Cwi ( n) =

DFi
+ Dvi ,
Qr

DFi
+ Dvi ,
Qr

i = 1, 2,3

(8)

q(i 1) < Qr qi

(8.1)

n is int, q0 = 0

(8.2)

6. Algorithm to Find the Optimal

Solution in the Centralized Case
Figure 5 graphically shows the total system
cost for a given value of n. We analyze the
properties of this function with respect to the
order quantity of the retailer. We present the
following 3 propositions to show that for a given
value of n the total system cost function is a
piece-wise convex function.
Proposition 1 For a given value of n the total

system cost function, CTi, is convex, i=1, 2, 3.

5. Centralized Case
In the centralized case, the objective is to
find the economic order quantities for both the
retailer and the warehouse in order to minimize
the total system cost. The total system cost is the
sum of holding and ordering costs at both the
retailer and the warehouse plus the

Proof. For i=1, the first interval of Qr, the total

system cost function is:

CT 1 ( n, Qr ) =

DAw hw (n 1)Qr DAr

+
+
nQr
2
Qr

hr Qr DF1
+
+ Dv1
Qr
2

(9)

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

360

Total

of CT1 at q1 is greater than the slope of CT2 at q1,

also the slope of CT2 at q2 is greater than the
slope of CT3 at q2.
Proof. The derivative of CT1 minus the

derivative of CT2 at q1 is:

Qr

Figure 5 Total system cost for a given value of n

Taking the second derivation we have:

2 CT 1
2Qr

2 DAw
nQr 3

It is clear that

2 DAr
Qr 3
2CT 1
2 Qr

DF1
Qr 3

(10)

> 0 , thus CT1 is a

convex function. Similarly we can prove that CTi

is convex for i=2, 3.
Proposition 2 For a given value of n,
CT1(q1)=CT2(q1) and CT2(q2)=CT3(q2).
Proof. CT1 at point q1 is as follow:
DAw hw (n 1)q1 DAr
+
+
nq1
2
q1

CT 1 ( q1 ) =

hr q1 DF1
+
+ Dv1
q1
2

(11)

Based on the chosen transportation scheme

we know that F2=F1+q1(v1-v2). By substituting
F2 for F1 we obtain:
CT 1 (q1 ) =

hr q1 DF2
+
+ Dv2
q1
2

(13)

The transportation scheme assumes F2>F1,

which means the slope of CT1 at q1 is greater
than the slope of CT2 at q1. Similarly, it can be
shown that the slope of CT2 at q2 is greater than
the slope of CT3 at q2. As a consequence,
Proposition 3 indicates that the minimum of the
total system cost function cannot occur at the
break points q1 and q2.
As a result, we can conclude from
Propositions 1, 2 and 3 that for a given value of
n the method of obtaining the optimal value of
Qr is the same as the one described by Hadley
and Whitin (1963) for their incremental quantity
discount model.
In order to obtain the optimal values of n,
and Qr, we develop a search algorithm based on
the model presented by Ertogral et al. (2007). As
mentioned above, we apply the incremental
quantity discount method for a given value of n.
To create our search algorithm we need a lower
bound and an upper bound for n. Clearly 1 is a

DAw hw (n 1) q1 DAr
+
+
nq1
2
q1

CT 1 ( q1 ) CT 2 ( q1 ) D

= 2 ( F2 F1 )
Qr
Qr
q1

(12)

The left side of the above equation is CT2(q1).

Similarly we can prove CT2(q2)=CT3(q2).
We note that, proposition 2 indicates that the
total system cost function is a continuous

lower bound for n. The following proposition

generates the upper bound for n.
Proposition 4 The upper bound of n is:

A (h hw )
nup = w r

hw ( Ar + F1 )

function.

Proof. In the first interval of Qr, the total system

Proposition 3 For a given value of n, the slope

cost function is:

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

CT 1 (n, Qr ) =

DAw hw ( n 1)Qr DAr

+
+
nQr
2
Qr

hr Qr DF1
+
+ Dv1
Qr
2

4
(14)

If we set the derivatives of CT1 with respect

to Qr and n equal to zero we obtain:
Q

2D(

n* =

Aw
n

+ Ar + F1 )

hw (n 1) + hr

1
Qr

2 DAw
hw

CT1 (n) =

(16)

+ Ar + F1 )(hw (nI 1) + hr ) + Dv1

Aw ( hr hw )
hw ( Ar + F1 )

From (16), it is clear that n and Qr have an

inverse relation. The value of Qr obtained from
(15) is a lower bound on Qr, because there is no
gain to decrease Qr less than Q*r. Hence, the n*
in (18) would be an upper bound on n.
In summary, the procedure to obtain the
values of n and Qr is as follow:
Algorithm:
1
2

hr
A (h hw ) A
w
=
Set nup = w r
.
u
n
F
hw ( Ar + F1 ) h
w
rA
For n=1, 2,,nup find the corresponding
optimal value of Qr by incremental quantity
discount method to minimize total system

costs.
For n=1, 2,,nup and the corresponding
optimal value of Qr calculate the minimum
total system cost.

to compare the centralized method presented in

this section with the decentralized method
presented in the previous section.

7. Numerical Results

(17)
The value of n which optimizes CT1(n) is
obtained as:
n* =

The solution which has the minimum total

system cost among the solutions in step 3 is
the overall optimal solution.
The next section employs simulation in order

(15)

Substituting Q*r in (14) we have:

A
2D( nw

361

In this section we apply our model to the real

case of the pharmaceutical downstream supply
chain of a public hospital, in which the hospital
pharmacy (HP) is considered as a retailer
because it delivers the pharmaceutical products
to several care units. HP orders these products
from the central pharmacy (CP), considered here
as the warehouse that delivers to several
hospitals (2006). We target a part of the
pharmaceutical products, which are ordered
regularly with fixed quantities and at fixed
periods. By numerical examples, the costs are
computed in MATLAB to compare the costs in
the centralized case with the decentralized one
and to propose a saving-sharing mechanism
through quantity discount.
Tables 2 and 3 give the transportation data
and four types of pharmacy product, respectively.
The capacity and variable costs of vehicles
depend on volume. In order to apply our
algorithm we need to transform these parameters
(capacity and variable cost of vehicles) into
units of product.
Table 4 presents the capacity and variable
costs of the vehicles in terms of a unit of
product.
The ordering cost of HP (retailer) and CP
(warehouse) are 1 and 3 respectively, i.e., Ar=1
and Aw=3. At both CP and HP, the rate of the

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

362

holding cost is presented in terms of volume.

According to the results in Table 6, for all

The values of this rate at both pharmacies are

100 and 70 Euros per volume (m3) per year

products centralized decision increases the

retailer cost, but decreases both the warehouse

respectively. We transform these values into

units of product. Table 5 contains the rates of

and the total system costs.

Table 5 Rate of holding cost

holding cost that are used in our model.

Table 2 Transportation data
Vehicle
Type

Capacity
(m3)

Fixed
Cost ()

S
M
L

8
14
20

169
179
196

Variable
Cost
(/m3)
5
3.75
2.5

Table 3 Data of pharmacy products

Product
No.
1
2
3
4

Demand
(unit/year)
19518
15106
1685
1864

Price
(/unit)
0.07
0.11
1.74
6.26

Volume
(m3)
0.011
0.017
0.073
0.045

Table 4 Transportation data according to product unit

Product
No.
1

Vehicle
Type

Capacity
(unit)

S
M
L
S
M
L
S
M
L
S
M
L

727
1273
1818
471
824
1176
110
192
274
178
311
444

Variable
Cost
(/unit)
0.06
0.04
0.03
0.09
0.06
0.04
0.37
0.27
0.18
0.23
0.17
0.11

Table 6 shows the numerical results. In these

examples, we obtain EOQ for HP and CP by our
centralized and decentralized methods. We
compare the cost of HP, the cost of CP and the
total system cost obtained by these two methods.

Product
No.
1
2
3
4

hr
(/unit/year)
1.1
1.7
7.3
4.5

hw
(/unit/year)
0.77
1.19
5.11
3.15

Table 7 shows the amount of loss or saving

and the percentage of them which are gained in
the centralized case. If HP and CP are
considered as a single firm, then it is cost
efficient to apply the centralized model; for
these four products the total system cost is lower
in the centralized case. If the pharmacies are
independent firms, then HP prefers to make
decentralized decisions but CP prefers
centralized decisions. Since the saving of CP is
more than the loss of HP, it is possible for CP to
propose a saving-sharing mechanism in order to
motivate HP to make centralized decisions.
Discount quantity is a common and simple
saving-sharing mechanism that CP can apply
here. The discount price offered by CP should be
large enough to cover the increases in cost of HP.
Table 8 presents the minimum discount and its
percentage by which HP is motivated to accept
centralized decisions.
Table 8 Minimum discount price
Product
No.
1
2
3
4

Discount
(/unit)
0.05 (71.4%)
0.06 (54.5%)
0.42 (24.1%)
0.31 (5.0%)

Price after
discount
0.02
0.05
1.32
5.95

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

363

Table 6 The results for four pharmacy products

Product
No.
1
2
3
4
Total cost (for 4
products)

Model

Retailer Cost
(/year)

Warehouse Cost
(/year)

Total System Cost

(/year)

DeCent.
Cent.
DeCent.
Cent.
DeCent.
Cent.
DeCent.
Cent.
DeCent.
Cent.

207.22
1010.6
226.63
1012.4
156.85
854.68
129.52
705.79
720.22
3583.47

18909
2689.8
20759
3160.4
14050
1678.6
11516
1331
65234
8859.8

19116
3732.6
20986
4172.9
14206
2533.3
11645
2036.8
65953
12475.6

Table 7 Loss or saving in centralized case

Product
No.
1
2
3
4
Total (for 4 products)

Retailer
(/year)
0.803.38
(387.69%)
785.77
(346.72%)
697.83
(444.9%)
576.27
(444.93%)
2863.25
(397.55%)

For some products the percentage of

discount is high, for example 71.4% for Product
1. At first glance, it might seem unreasonable
that CP propose this amount of discount.
Table 9 Saving in central pharmacy after discount
Product No.
1
2
3
4

Saving after Discount (/unit)

15243.3 (80.6%)
16692.2 (80.4%)
11663.7 (83.0%)
9607.2 (83.4%)

Warehouse
(/year)
+16219.2
(+85.78%)
+17598.6
(+84.78%)
+12371.4
(+88.05%)
+10185
(+88.44%)
56374.2
(86.41%)

System
(/year)
+15383.4
(+80.47%)
+16813.1
(+80.12%)
+11672.4
(+82.17%)
+9608.2
(+82.51%)
53477.4
(81.08%)

8. Discussion
Through this example, we compare the costs
generated by the centralized and the
decentralized strategies. The interest of the
centralized policy has been illustrated and we
show that the warehouse has a better benefit
from the retailer. To incite retailers to take part
in this type of decisions, it is necessary to use a
mechanism to share savings (win-win strategy).
Several solutions such as consignment stocks,

However, Table 9 shows that there is still a

saving for CP after discount. In the other words,
there is a saving for CP despite implementing

the division of security stocks or discount

policies could be considered. We illustrate the
last solution and we show that in spite of very

discounts.

important discounts for low-price products,

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

364

expensive products do not require an important

discount. Finally, as illustrated by Table 10, the
total discount cost is not very important (more
than 10%) when compared to savings because,
in our results, the transportation and inventory
costs are much more important than the price of
the products.
Table 10 Discount cost and saving in centralized case
Product

Saving per
product

1
2
3
4

Discount
cost per
product
975.9
906.36
707.7
577.84

Total

3167.8

37491.4

1713.9
19852.64
2452.7
13472.16

9. Conclusions and Suggestions for

Further Research
In this paper, we have looked for the optimal
ordering policy in a two-level supply chain in
two cases; centralized and decentralized.
Conversely to the most similar models which
determine the economic order quantity just
based on inventory costs, our model
incorporates transportation costs into inventory
replenishment decisions. An algorithm was
proposed to find economic order quantities for
both the retailer and the warehouse which
minimize the total system cost in the centralized
case. The numerical results indicated that the
centralized case leads to savings for the
warehouse and the whole system as compared to
the decentralized case. However, centralized
decisions induce losses for the retailer. These
losses are the reason for which we proposed a
mechanism for the warehouse to share its saving
through discount quantity in order to cover

retailer's losses.
For future research we intend to complete
this model by including the multi-item lot-sizing
problem. Next, we think that our model can be
more practical if the number of vehicles are a
decision variable.

References
[1] Abdul-Jalbar, B., Gutirrez, J., Puerto, J. &
Sicilia, J. (2003). Policies for inventory/
distribution systems: the effect of
centralization
vs.
decentralization.
International
Journal
of
Production
Economics, 81-82: 281-293
[2] Andersson, J. & Marklund, J. (2000).
Decentralized inventory control in a
two-level distribution system. European
Journal of Operational Research, 127:
483-506
[3] Axster, S. (1993). Exact and approximate
evaluation of batch-ordering policies for
two-level inventory systems. Operations
Research, 41 (4): 777-785
[4] Axster, S. (1996). Optimization of
order-up-to-S policies in two-echelon
inventory systems with periodic review.
Naval Research Logistics, 40: 245-253
[5] Axster, S. (2005). A simple decision rule
for decentralized two-echelon inventory
control. International Journal of Production
Economics, 93-94: 53-59
[6] Chen, J.M. & Chen, T.H. (2005). The
multi-item replenishment problem in a
two-echelon supply chain: the effect of
centralization
versus
decentralization.
Computers & Operations Research, 32:
3191-3207
[7] Das, C. & Tyagi, R. (1997). Role of
inventory and transportation costs in

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

determining the optimal degree of

centralization. Transportation Research Part
E: Logistics and Transportation Review, 33
(3): 171-179
[8] Ertogral, K., Darwish, M. & Ben-Daya, M.
(2007). Production and shipment lot sizing in
a vendor-buyer supply chain with
transportation cost. European Journal of
Operational Research, 176: 1592-1606
[9] Forsberg, R. (1995). Optimization of
order-up-to-S
policies
for
two-level
inventory systems with compound Poisson
demand. European Journal of Operational
Research, 81: 143-153
[10] Forsberg, R. (1996). Exact evaluation of (R,
Q) policies for two-level inventory systems
with Poisson demand. European Journal of
Operational Research, 96: 130-138
[11] Ganeshan, R. (1999). Managing supply
chain inventories: a multiple retailer, one
warehouse, multiple supplier model.
International
Journal
of
Production
Economics, 59: 341-354
[12] Goyal, S.K. & Nebebe, F. (2000).
Determination of economic productionshipment policy for a single-vendor-single
-buyer system. European Journal of
Operational Research, 121: 175-178
[13] Hadley, G. & Whitin, T.M. (1963).
Analysis of Inventory Systems. Prentice-Hall
Inc
[14] Hassan, T., Baboli, A., Guinet, A.,
Leboucher, G. & Brandon, M.T. (2006). Reorganizing the pharmaceutical supply chain
downstream:
implementation
a
new
pharmacy. In: 12th IFAC Symposium on
Information
Control
Problems
in
Manufacturing, 17-19, May 2006

365

[15] Hill, R.M. (1997). The single-vendor

single-buyer integrated production-inventory
model with a generalized policy. European
Journal of
493-499

Operational

Research,

97:

[16] Hoque, M.A. & Goyal, S.K. (2000). An

optimal policy for a single-vendor singlebuyer
integrated
production-inventory
system with capacity constraint of the
transport equipment. International Journal of
Production Economics, 65: 305-315
[17] Huang, H.C., Chew, E.P. & Goh, K.H.
(2005). A two-echelon inventory system
with transportation capacity constraint.
European Journal of Operational Research,
167: 129-143
[18] Li, J. & Liu, L. (2006). Supply chain
coordination with quantity discount policy.
International
Journal
of
Production
Economics, 101: 89-98
[19] Marklund, J. (2002). Centralized inventory
control in a two-level distribution system
with Poisson demand. Naval Research
Logistics, 49 (2): 798-822
[20] Matta, K.F. & Sinha, D. (1995). Policy and
cost
approximations
of
two-echelon
distribution systems with a procurement cost
at the higher echelon. IIE Transactions. 27
(5): 638-645
[21] Mishra, A.K. (2004). Selective discount for
supplier-buyer coordination using common
replenishment epochs. European Journal of
Operational Research, 153: 751-756
[22] Qu, W.W., Bookbinder, J.H. & Iyogun, P.
(1999).
An
integrated
inventory
transportation system with modified periodic
policy for multiple products. European
Journal of Operational Research, 115:
254-269

Baboli et al.: An Algorithm for the Determination of the Economic Order Quantity
J Syst Sci Syst Eng

366

[23] Seifbarghy, M. & Jokar, M.R.A. (2006).

Cost evaluation of a two-echelon inventory
system with lost sales and approximately
Poisson demand. International Journal of
Production Economics, 102: 244-54
[24] Silver, E.A., Pyke, D.F. & Peterson, R.
(1998).
Inventory
Management
and
Production Planning and Scheduling. John
Wiley & Sons, Third Edition
[25] Swenseth, S.R. & Godfrey, M.R. (2002).
Incorporating transportation
inventory
replenishment

costs into
decisions.

International
Journal
of
Production
Economics, 77: 113-130
[26] Viswanathan, S. & Piplani, R. (2001).
Coordinating supply chain inventories
through common replenishment epochs.
European Journal of Operational Research,
129: 277-286
[27] Zhao, Q.H., Wang, S.Y., Lai, K.K. & Xia,
G.P. (2004). Model and algorithm of an
inventory problem with the consideration of
transportation cost. Computers & Industrial
Engineering, 46: 389-397
Armand Baboli is currently an Associate
Professor in the Industrial Engineering
Department of the National Institute of Applied
Sciences (INSA) of Lyon, France, and member
of LIESP laboratory. He received his [Link].
degree in Industrial Engineering from Iran
University of Science & Technology in 1988. In
1994 he received his [Link]. degree in Industrial
Engineering and then in 1999 his Ph.D. degree
in Industrial Engineering both from Grenoble
Polytechnic National Institute (INPG). His

current research interests include supply chain

design and organization, inventory management
and facility layout planning.
Mohammadali Pirayesh Neghab received his
[Link]. & [Link]. degrees in Industrial Engineering
from Sharif University of Technology, Tehran,
Iran, in 2000 and 2002, respectively. Currently
he is a Ph.D. candidate in the Industrial
Engineering Department of Sharif University.
His Ph.D. thesis is in the area of inventory
control and supply chain management.
Rasoul Haji is currently a professor of
Industrial Engineering at Sharif University of
Technology in Tehran, Iran. He received a [Link].
degree from University of Tehran in Chemical
Engineering in 1964. He also earned an [Link].
degree from University of California-Berkeley
in Industrial Engineering in 1967. In 1970 he
received his Ph.D. degree from Berkeley in
Industrial Engineering. He is the Editor-in-Chief
of the Journal of Industrial and Systems
Engineering and he is a member of Irans
Academy of Sciences. He is recognized as a
co-founder of a fundamental and important
relation in queuing theory known as
Distributional Littles Law His research
interest includes inventory control, stochastic
processes, and queuing theory. He has published
papers in different technical journals such as
Journal of Applied Probability, SIAM Journal of
Applied Math, European Journal of Operations
research, Computers and Industrial Engineering,
Journal of Production Planning and Control,
and Applied Mathematics and Computation.