Available online at www.sciencedirect.

com

Applied Mathematical Modelling 32 (2008) 21062116

www.elsevier.com/locate/apm

A deterministic, multi-item inventory model with

supplier selection and imperfect quality
Jafar Rezaei

a,*

, Mansoor Davoodi

Department of Industrial Management, Vali-e-Asr University of Rafsanjan, Rafsanjan, Iran

Department of Applied Mathematics and Computer Sciences, Amirkabir University of Technology, Tehran, Iran
Received 1 July 2006; received in revised form 1 March 2007; accepted 4 July 2007
Available online 19 July 2007

Abstract
This paper considers the scenario of supply chain with multiple products and multiple suppliers, all of which have limited capacity. We assume that received items from suppliers are not of perfect quality. Items of imperfect quality, not necessarily defective, could be used in another inventory situation. Imperfect items are sold as a single batch, prior to receiving
the next shipment, at a discounted price. The demand over a nite planning horizon is known, and an optimal procurement
strategy for this multi-period horizon is to be determined. Each of products can be sourced from a set of approved suppliers, a supplier-dependent transaction cost applies for each period in which an order is placed on a supplier. A productdependent holding cost per period applies for each product in the inventory that is carried across a period in the planning
horizon. Also a maximum storage space for the buyer in each period is considered. The decision maker, the buyer, needs to
decide what products to order, in what quantities, with which suppliers, and in which periods. Finally, a genetic algorithm
(GA) is used to solve the model.
 2007 Elsevier Inc. All rights reserved.
Keywords: Inventory control; Integer programming; Mathematical programming; Optimization; Supplier selection; Genetic algorithms

1. Introduction
In the past several decades, supplier selection problem has gained great attention in the business management literature and practice (cf. Burton [1], Degraeve and Roodhooft [2], Degraeve et al. [3], Dickson [4],
Jayaraman et al. [5], Patton III [6], Weber et al. [79]). Under the business environment of global sourcing,
core-competence outsourcing strategy, supply base reduction, strategic buyersupplier relationship, crossfunctional purchasing program, Internet and e-commerce and so forth, the supplier selection decision is
becoming ever important and complicated decision.
Recently, the issue of supplier selection integrated with inventory management. Bua and Jackson [10] presented a schedule purchase from mix of vendors over a dened planning horizon via goal programming model.
*

Corresponding author. Tel.: +98 9133919628; fax: +98 3913202009.

E-mail addresses: Rezaei@vru.ac.ir, Jafarrezaei@gmail.com (J. Rezaei), MDmonfared@aut.ac.ir (M. Davoodi).

0307-904X/$ - see front matter  2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.apm.2007.07.009

J. Rezaei, M. Davoodi / Applied Mathematical Modelling 32 (2008) 21062116

2107

Bender et al. [11], described, but not developed, a mixed integer optimization model, to minimize the sum of
purchasing, transportation and inventory costs over multiple time periods, constraints of vendor capacity and
policy. Degraeve and Roodhooft [2] put forward a mixed integer non-linear programming model to consider
vendor selection and inventory management decision together, under conditions of multiple sourcing, multiple
criteria and capacity constraints.
Basnet and Leung [12] presented a model for optimal procurement lot-sizing with supplier selection. Their
model is an extension of Wagner and Whitins seminal work [13], which is a lot sizing model for a single product, to the multi-product case. In addition, it is a multi-period model, considering the possibility of exploiting
economies of scale in the procurement process in exchange for accruing inventory costs from one period to the
next. Multi-period models also oer the opportunity to change suppliers for a product from one period to the
next. Many supplier selection models (cf. Ganeshan [14], Jayaraman et al. [5], Kasilingam and Lee [15], Rosenthal et al. [16]) are single period models. Capacities on resources, in general, were considered in the important
work of Manne [17], in which he introduced a mixed-integer linear programming model in the presence of
labor limitations. Benson [18] by introducing capacities for the suppliers, considered a supply chain with multiple suppliers, all of which have limited capacity and determined an optimal procurement strategy for this
multi-period horizon.
In a real production environment, it can often be observed that there are defective items being produced.
These defective items must be rejected, repaired, reworked, or, if they have reached the customer, refunded. In
all cases, substantial costs are incurred. Therefore, it is more appropriate to take the quality-related cost into
account in determining the optimal ordering policy. In the literature, Porteus [19] and Rosenblatt and Lee [20]
are among the rst to explicitly elaborate on the signicant relationship between quality imperfection and lot
size. Porteus [19] describes a system that begins each production run in control. As each unit is produced, there
is a probability p that the system goes out of control, at which time all subsequent units are defective. He used
this model to study the optimal setup investment in relation to reducing the probability p of the process going
out of control. His work has encouraged many researchers to deal with modeling the quality improvement
problems that we refer to some of them.
Rosenblatt and Lee [20] assumed that the time between the beginnings of the production run until the process goes out of control is exponential and that defective items can be reworked instantaneously at a cost. They
concluded that the presence of defective products motivates smaller lot sizes. In a subsequent paper [21], they
considered a joint lot sizing and inspection policy is studied under an economic order quantity model where a
random proportion of units are defective. Salameh and Jaber [22] hypothesized a production/inventory situation where items, received or produced, are not of perfect quality. Items of imperfect quality, not necessarily
defective, could be used in another production/inventory situation, that is, less restrictive process and acceptance control. They extended the traditional EPQ/EOQ model by accounting for imperfect quality items when
using the EPQ/EOQ formulae. Chan et al. [23] provided a framework to integrate lower pricing, rework and
reject situations into a single EPQ model. A 100% inspection is performed in order to identify the amount of
good quality items, imperfect quality items and defective items in each lot. Papachristos and Konstantaras [24]
looked at the issue of non-shortages in models with proportional imperfect quality, when the proportion of the
imperfects is a random variable and revised the papers of Salameh and Jaber [22] and Chan et al. [23]. Hayek
and Salameh [25] presented an inventory model of shortage and backlog that considers rework of defective
products. Zhang and Gerchak [26] considered a joint lot sizing and inspection policy studied under an
EOQ model where a random proportion of units are defective. Wee et al. [27] developed an optimal inventory
model for items with imperfect quality and shortage backordering. They assumed that all customers are willing to wait for new supply when there is a shortage. Ouyang et al. [28] investigated the lot size, reorder point
inventory model involving variable lead time with partial backorders, where the production process is imperfect. Francis Leung [29] proposed an EPQ model with a exible and imperfect production process. He formulated this inventory decision problem using geometric programming. Freimer et al. [30] investigated the eect
of imperfect yield on economic production quantity decisions. Ouyang and Chang [31] investigated the impact
of quality improvement on the modied lot size reorder point models involving variable lead time and partial
backorders. Chiu [32] considered the eects of the reworking of defective items on the economic production
quantity (EPQ) model with allowed backlogging. Urban [33] proposed a nite replenishment inventory model
in which the demand of an item is a deterministic function of price and advertising expenditures. The formu-

2108

J. Rezaei, M. Davoodi / Applied Mathematical Modelling 32 (2008) 21062116

lated models also incorporate learning eects and the possibility of defective items in the production process.
Ben Daya [34] proposed multi-stage lot sizing models for imperfect production processes. Lee [35] developed a
cost/benet model for supporting investment strategies about inventory and preventive maintenance in an
imperfect production system. The eect of such investments on the return is expressed as a function of measurable variables. Using this model, the decision maker can decide whether investments in inventory and preventive maintenance are necessary and how much to invest. Ben Daya [36] presented an integrated model for
the joint determination of economic production quantity and preventive maintenance (PM) level for an imperfect process having a general deterioration distribution with increasing hazard rate.
Goyal et al. [37] developed a simple approach for determining an optimal integrated vendor-buyer inventory policy for an item with imperfect quality. Ouyang et al. [38] investigated the integrated vendor-buyer
inventory problem, they assumed that an arrival order lot may contain some defective items, and the defective
rate is a random variable. They derived an integrated mixture inventory model with backorders and lost sales.
In this paper, we introduce imperfect items and storage capacity in the lot sizing with supplier selection
problem and formulate the problem as a mixed integer programming model. Then the model is solved with
a Genetic Algorithm.
2. Formulation
To develop the proposed model, we adopt the following assumptions and notation:
2.1. Assumptions and notations
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.

Transaction cost Oj for supplier j does not depend on the variety and quantity of products involved.
Holding cost Hi of product i per period is product-dependent.
Demand Dit of product i in period t is known over a planning horizon.
Items received, are not of perfect quality, not necessarily defective, kept in stock and sold prior to the
next period as a single batch.
Each lot of product i received from supplier j contains an average percentage Pij of defective items.
Purchasing price of product i from supplier j is bij. Good-quality items i have a selling price Sgi per unit
and defective items i are sold as a single batch at a discounted price Sdi. Also Sgi > Sdi.
A 100% screening process of the lot is conducted with a unit screening cost vi of product i.
Each supplier has a limited capacity.
All requirements must be fullled in the period in which they occur: shortage or backordering is not
allowed.
Product i needs a storage space wi and available total storage space is limited by W.

3. Mathematical modeling
Based on the above assumptions a mathematical model is developed. The behavior of the model is illustrated in Fig. 1.
Fig. 1 considers the scenario of supply chain with multiple products and multiple suppliers, all of which
have limited capacity. We assume that received items from suppliers are not of perfect quality. Items of imperfect quality, not necessarily defective, could be used in another inventory situation. Imperfect items are sold as
a single batch, prior to receiving the next shipment, at a discounted price. The demand over a nite planning
horizon is known, and an optimal procurement strategy for this multi-period horizon is to be determined.
Each of products can be sourced from a set of approved suppliers; a supplier-dependent transaction cost
applies for each period in which an order is placed on a supplier. A product-dependent holding cost per period
applies for each product in the inventory that is carried across a period in the planning horizon. Also a
maximum storage space for the buyer in each period is considered. In order to maximize the total prot,
the decision maker, the buyer, needs to decide what products to order, in what quantities, with which suppliers, and in which periods (xijt).

J. Rezaei, M. Davoodi / Applied Mathematical Modelling 32 (2008) 21062116

Supplier 1

Supplier 2

Product 1

sale
imperfect
items

good items
from
period t-1

sale good
items

Product 2

Product i

..

sale
imperfect
items

good items
from
period t-1

sale good
items

good items
to period
t+1

Supplier j

..

good items
from
period t-1

sale
imperfect
items

sale good
items

good items
to period
t+1

Supplier n

..

..

good items
to period
t+1

2109

Product m

good items
from
period t-1

sale
imperfect
items

sale good
items

good items
to period
t+1

Fig. 1. Behavior of the model in period t.

Total revenue (TR) = revenue of selling good quality products + revenue of selling imperfect quality
products
XXX
XXX
TR
X ijt 1  P ij S gi
X ijt P ij S di :
1
i

Total cost (TC) = purchase cost of the products + transaction cost for the suppliers + screening cost of the
products + holding cost for remaining inventory in each period
XXX
XX
XXX
TC
X ijt bij
Oj Y jt
X ijt vi
i

XX
i

Hi

t
X

k1

X ijt 1  P ij 

t
X

Dik :

k1

The total prot over a planning horizon is the total revenue less the total cost, p = TR  TC and it is as
follows:
"
#
XXX
XXX
X ijt 1  P ij S gi
X ijt P ij S di
p
i

"
XXX
j

XX
i

Hi

X ijt bij

XX
j

t X
X
k1

Oj Y jt

X ijt 1  P ij 

XXX
i

t
X

!#

Dik

X ijt vi

k1

There are four restrictions for this objective function:

(1) All requirements must be fullled in the period in which they occur: shortage or backordering is not
allowed: Eq. (5).
(2) There is not an order without charging an appropriate transaction cost: Eq. (6).

2110

J. Rezaei, M. Davoodi / Applied Mathematical Modelling 32 (2008) 21062116

(3) Suppliers have limited capacities: Eq. (7).

(4) Available total storage space is limited: Eq. (8).
The problem is to nd number of product i ordered from supplier j in period t so as to maximize the total
prot function subject to restrictions and boundary conditions:
"
# "
XXX
XXX
XXX
max p
X ijt 1  P ij S gi
X ijt P ij S di 
X ijt bij
i

XX
j

subject to

Oj Y jt

t X
X
j

k1
T
X

XXX
i

X ijk 1  P ij 

t
X

X ijt vi

Dik P 0

Hi

t
X

k1

X ijt 1  P ij 

t
X

!#
Dik

k1

for all i and t;

for all i; j and t;

k1

Dik Y jt  X ijt 1  P ij P 0

wi

XX

kt

t X
X

X ijk 1  P ij 

k1

0 6 X ijt 6 C ij

t
X

!
Dik

6W

for all t;

k1

for all i; j and t:

Using classical optimization methods, we maintain a single solution and improve it until an optimal solution is found. Often, because some higher level information (such as social and cultural consideration) could
not be included in the model, the decision maker cannot apply this only solution to a real world problem. Such
higher level information is non-technical, qualitative and experience-driven. However, if a set of many solutions are already worked out or available, one can evaluate the pros and cons of each of these solutions based
on all such non-technical, qualitative, yet still important, consideration and compare them to make a choice
[39]. Genetic Algorithms dier in that they maintain and manipulate a family or population of solutions in the
search for an optimal solution. This is one of the reasons to apply a GA to solve this problem. On the other
hand, although using classical optimization methods we can obtain the single optimal solution but it is true
when the problem is small. According to problem (p), it is observed that the scale is mainly determined by
the number of constraints. So the number of constraints S(I, J, T) can be used to describe the scale of problem
(p). The number of constraints S(I, J, T) can be calculated as follows:
SI; J ; T I  T I  J  T T 2  I  J  T :

Such a bottleneck diculty can be shown clearly in Table 1. Table 1 shows that S(I, J, T) will increase with
the number of products I and the number of suppliers J and the number of periods T. When I, J and T are all
large, solving such a large optimal problem using conventional methods is impossible. In order to obtain a
Table 1
Number of constraints to problem (p): S(I, J, T)
I=2

I=5

I = 10

I = 20

I = 50

I = 100

I = 200

T=6

J=2
J=5
J = 10
J = 20

90
196
378
738

216
486
936
1836

426
966
1866
3666

846
1926
3726
7326

2106
4806
9306
18,306

4206
9606
18,606
36,606

8406
19,206
37,206
73,206

T = 12

J=2
J=5
J = 10
J = 20

180
396
756
1476

432
972
1872
3672

852
1932
3732
7332

1692
3852
7452
14,652

4212
9612
18,612
36,612

8412
19,212
37,212
73,212

16,812
38,412
74,412
146,412

J. Rezaei, M. Davoodi / Applied Mathematical Modelling 32 (2008) 21062116

2111

population of solutions in respect of the rst reason and to overcome this bottleneck problem, a GA approach
is proposed to solve the problem as follows.
4. Genetic algorithm
GAs mimic the evolutionary process by implementing a survival of the ttest strategy. GAs are probabilistic search and optimization techniques guided by the principles of evolution and natural genetics. This
method provides global near-optimal solutions of an objective or tness function by striking a remarkable balance between exploration and exploitation in complex, large, and multi-modal landscapes.
A more complete discussion of GAs including extensions to the general algorithm and related topics can be
found in books by Davis [40], Goldberg [41], Holland [42], and Michalewicz [43].
In each evolution step of a standard genetic algorithm, a new population is created from the preceding one
using the selection, crossover and mutation operators that explain these in the following.
4.1. Chromosome representation
In this study, we take each chromosome as a solution for the problem. Assume there are T periods, I products and J suppliers, then each chromosome is developed based on two single dimensional arrays. The rst
single array (we mark it by X) consists of real values by lengths of (I J T) and the second array (we mark
it by Y) consists of binary values representing by lengths of (J T). For example, the representation of a chromosome is illustrated in Fig. 2.
4.2. Initialization of the population
For any GA it is necessary to initialize the population. The most common method is to randomly generate
solutions for the entire population. Since GAs iteratively improves existing solutions the beginning population
can be seeded by the decision maker with individuals from other algorithms or from an existing solution to the
problem. The remainder of the population is then seeded with randomly generated solutions. All experiments
discussed in this paper employ a completely random seeding of the beginning population. Population size
depends only on the nature of problems and it must balance between time complexity (consume for computing
tness function and genetic operators) and search space measure. In this paper, the population size is set at
100.
4.3. Evaluation or tness function
The tness value is a measure of the goodness of a solution with respect to the original objective and the
amount of infeasibility. The tness function is formed by adding a penalty to the original objective function
in MIP model [44,45].
Since nding a feasible solution (which will satisfy all constraints in MIP model) is a major problem and has
diculty in nding even one feasible solution, we use penalty approach to decrease tness of infeasible solutions and toward the feasible region.
For each solution (or each chromosome) S, the penalty function is calculated as follows:
PenaltyS

C
X

Rc Xc S;

10

c1

X111

X112

Xijt

XIJT-1

XIJT

Y11

Y12

Yjt

YJT-1

YJT

Fig. 2. Each chromosome consists of two real and binary single arrays.

2112

J. Rezaei, M. Davoodi / Applied Mathematical Modelling 32 (2008) 21062116

where C is the number of constraints (in this model C equals to (I T + I J T + T)), Rc is penalty parameter for cth constraint and Xc(S) is constraint violation from cth constraint. However, we check upper and
lower bounds of variables in random initial step and after every crossover and mutation operations. Before
the penalty function is calculated, all constraints are normalized. Most studies in evolutionary optimization
use chosen static values of Rc [46]. We choose static values of Rc too. Since the MIP model objective function
is maximization, we take negative value for Rc. After calculating the penalty function, we add it to original
objective function and reach tness function
FitnessS pS PenaltyS:

11

It is important to note that the corresponding Penalty(S) for a feasible solution S is zero.
4.4. Selection strategy
The selection of parents to produce successive generations plays an extremely important role in the GA.
The goal is to allow the ttest individuals to be selected more often to reproduce. However, all individuals
in the population have a chance of being selected to reproduce the next generation. Each individual is assigned
a probability of being selected with better individuals having larger probabilities. There are several schemes for
determining and assigning the selection probability, e.g. roulette wheel selection, tournament selection and its
extensions scaling techniques and ranking methods [41,43]. For other non-probabilistic selection procedures,
including tournament selection and elitist models refer to [41,43].
In this paper, the roulette wheel selection method is used. This selection operator inherently maximizes the
tness function.
4.5. Crossover and mutation operators
Reproduction is carried out by the application of genetic operators to the selected parents. Crossover and
Mutation are the two basic types of genetic operators [43]. Crossover operators combine information from
two parents such that the two children have a resemblance to each parent. Mutation operators alter or mutate
one parent by changing one or more variables in some way or by some random amount to form one ospring.
The application of these types and their derivatives depend on the chromosome representation. In the proposed GA, since each chromosome contains two types of variables real and binary (correspond to X and Y
arrays). We use linear crossover operator for real variables and two-point
crossover
operator for binary vari

 1 
ables. For mutation, we use from bit-wise mutation by probability lengths 1of string , in our study J T
for binary
variables and use uniform mutation for real variables by probability 10% as follows:
h
i
u
u
l
xijt r xijt  xijt ; r is random in 0; 1;
u

12

where ijt is selected random index in real array. xijt and xijt are upper and lower bounds for xijt, respectively
u
l
[39]. In this model, xijt equals Cij for all t and xijt equals zero for all i, j and t.
Moreover, we use an external population by xed size (10% of the population size which in this study
10
becomes 100
10 for elite-preserving operator [39].
4.6. Termination criterion
The GA moves from generation to generation selecting and reproducing parents until a termination criterion is met. The most frequently used stopping criterion is a specied maximum number of generations.
Another termination strategy involves population convergence criteria. In general GAs will force much of
the entire population to converge to a single solution. When the sum of the deviations among individuals
becomes smaller than some specied threshold the algorithm can be terminated. The algorithm can also be
terminated due to a lack of improvement in the best solution over a specied number of generations. Alternatively a target value for the evaluation measure can be established based on some arbitrarily acceptable

J. Rezaei, M. Davoodi / Applied Mathematical Modelling 32 (2008) 21062116

2113

threshold. Several strategies can be used in conjunction with each other. In this paper, a maximum number of
generations are employed.
5. Numerical example
In this section, we solved a numerical example of the proposed model using the above-described GA. We
consider a scenario with three products over a planning horizon of four periods whose requirements are as
follows: demands of three products over a planning horizon of four periods are given in Table 2.
There are three suppliers and their prices and transaction cost and average percentage of the defective items
are shown in Tables 3 and 4, respectively.
In Table 5, selling price of good product i (Sgi), selling price of defective product i(Sdi), storage space of
product i (wi), holding cost of product i per period (Hi) and screening cost of product i(vi) are shown.
Table 2
Demands of three products over a planning horizon of four periods
Products

1
2
3

Planning horizon (four periods)

1

170
85
280

155
90
255

160
80
290

140
105
300

Table 3
Price of three products supplied by three suppliers 1, 2, 3 and transaction cost of them
Products

Price (supplied from three supplier)

1

1
2
3
Transaction cost

25
30
54

27
32
50

24
33
49

3000

2700

3500

Table 4
Average percentage of the defective item for three suppliers
Products

1
2
3

Average percentage of defective items

1

0.03
0.02
0.04

0.02
0.03
0.04

0.03
0.05
0.01

Table 5
Sgi, Sdi, wi, Hi and vi for three products 1, 2 and 3
Products

Sgi
Sdi
wi
Hi
Vi

50
20
0.2
5
2

34
25
0.18
3.5
1.5

60
40
0.5
8
1.8

2114

J. Rezaei, M. Davoodi / Applied Mathematical Modelling 32 (2008) 21062116

Also we considered the total storage space (W) equal to 20.

To solve this Mixed Integer Programming (MIP) problem we used both Lindo and GA. Results are shown
in Table 6. Column 2 shows the single solution obtained from Lindo and columns 36 represent the varied
results obtained from GA. Using classical optimization tools, we can obtain just a single optimal solution.
On the other hand, GAs can obtain several optimal solutions and this is one of their most important
characters.
Table 6
The results of the numerical example
Variable

Lindo results

GA results

Y11
Y12
Y13
Y14
Y21
Y22
Y23
Y24
Y31
Y32
Y33
Y34
X111
X112
X113
X114
X121
X122
X123
X124
X131
X132
X133
X134
X211
X212
X213
X214
X221
X222
X223
X224
X231
X232
X233
X234
X311
X312
X313
X314
X321
X322
X323
X324
X331
X332
X333
X334

0
0
0
1
1
1
1
0
1
1
1
1
0
0
0
103.09
0
0
0
0
175.25
159.79
164.94
144.32
0
0
0
107.14
80
75
82.47
0
7.78
18.15
0
0
0
0
0
75.31
85.41
59.37
54.58
0
200
200
240
230

0
0
0
1
1
1
1
0
1
1
1
1
0
0
0
120
0
0
0
0
176.66
159.27
164.94
124.83
0
0
0
99
80
75
90
0
10
16
0
0
0
0
0
76
84.44
60
54.98
0
200
200
240
230

0
0
0
1
1
1
1
0
1
1
1
1
0
0
0
100
0
0
0
0
178.89
168.59
164.17
136.24
0
0
0
99.01
46.56
55
90
0
46.73
34.37
0
0
0
0
0
55.31
85.38
89.06
65.49
0
200
180
240
230

0
1
0
0
1
0
1
1
1
1
1
1
0
20
0
0
0
0
20
33.5
211.54
123.03
159.94
136.32
0
78.4
0
0
.22
0
90
70
100
3.37
0
32
0
89
0
0
98.41
0
132.58
71
188
176
164
230

0
1
0
0
1
0
1
1
1
0
1
1
0
130
0
0
0
0
34.09
98.79
205
0
139.91
127.07
0
130
0
0
36
0
40.45
0
54
0
14.03
100
0
280
0
0
170.41
0
144
71.9
120
0
141
230

16910.77

16854.16

16543.21

15651.77

15196.82

J. Rezaei, M. Davoodi / Applied Mathematical Modelling 32 (2008) 21062116

2115

6. Conclusion
We have considered the problem of supply chain with multiple suppliers and multiple products. Major
assumptions include that the suppliers have limited capacity, received items from suppliers are not of perfect
quality, the demand over a nite planning horizon is known and the buyer has a maximum storage space in
each period. We then formulate the problem as a mixed integer programming in which we have combined supplier selection with traditional inventory lot sizing and imperfect items. Finally, a numerical example of the
proposed model is solved by Lindo and GA. By solving the problem via classical methods we can just obtain
an optimal solution. Most of the time, the decision maker (DM) cannot apply this only solution because of
some higher level information that could not be included in the model. Therefore we applied a GA to solve
the problem to obtain more solutions. Although these solutions are not necessarily optimal and are almost
near-optimal, it can be possible for DM to select one of them that matches with the real world condition.
Results indicate what products to order, in what quantities, with which suppliers, and in which periods.
References
[1] T.T. Burton, JIT/repetitive sourcing strategies: typing the knot with your suppliers, Prod. Inventory Manage. J. (1988) 3841.
[2] Z. Degraeve, F. Roodhooft, Improving the eciency of the purchasing process using total cost of ownership information: the case of
heating electrodes at Cockerill Sambre S.A., Eur. J. Oper. Res. 112 (1989) 4253.
[3] Z. Degraeve, E. Labro, F. Roodhooft, An evaluation of vendor selection models from a total cost of ownership perspective, Eur. J.
Oper. Res. 125 (2001) 3458.
[4] G.W. Dickson, An analysis of vendor selection systems and decisions, J. Purchas. 2 (1) (1966) 517.
[5] V. Jayaraman, R. Srivastava, W.C. Benton, Supplier selection and order quantity allocation: a comprehensive model, J. Supply Chain
Manage. 35 (1989) 5058.
[6] W.E. Patton III, Use of human judgment models in industrial buyers vendor selection decisions, Ind. Market. Manage. 25 (1986)
135149.
[7] C.A. Weber, J.R. Current, A multi-objective approach to vendor selection, Eur. J. Oper. Res. 68 (1983) 173184.
[8] C.A. Weber, J.R. Current, W.C. Benton, Vendor selection criteria and methods, Eur. J. Oper. Res. 50 (1980) 218.
[9] C.A. Weber, J.R. Current, A. Desai, Non-cooperative negotiation strategies for vendor selection, Eur. J. Oper. Res. 108 (1988) 208
223.
[10] F.P. Bua, W.M. Jackson, A goal programming model for purchase planning, J. Purchas. Mater. Manage. (1983) 2734.
[11] P.S. Bender, R.W. Brown, M.H. Isaac, J.F. Shapiro, Improving purchasing productivity at IBM with a normative decision support
system, Interfaces 15 (1985) 106115.
[12] C. Basnet, J.M.Y. Leung, Inventory lot-sizing with supplier selection, Comput. Oper. Res. 32 (2005) 114.
[13] H.M. Wagner, T.M. Whitin, Dynamic version of the economic lot size model, Manage. Sci. 5 (1958) 8996.
[14] R. Ganeshan, Managing supply chain inventories: a multiple retailer, one warehouse, multiple supplier model, Int. J. Prod. Econ. 59
(1989) 341354.
[15] R.G. Kasilingam, C.P. Lee, Selection of vendors a mixed-integer programming approach, Comput. Ind. Eng. 31 (1986) 347350.
[16] E.C. Rosenthal, J.L. Zydiak, S.S. Chaudhry, Vendor selection with bundling, Decision Sci. 26 (1985) 3548.
[17] A.S. Manne, Programming of economic lot sizes, Manage. Sci. 4 (1958) 115135.
[18] H.Y. Benson, Optimal pricing and procurement strategies in a supply chain with multiple capacitated suppliers, Comput. Oper. Res.
32 (1) (2005) 114.
[19] E.L. Porteus, Optimal lot sizing, process quality improvement and setup cost reduction, Oper. Res. 34 (1986) 137144.
[20] M.J. Rosenblat, H.L. Lee, Economic production cycles with imperfect production processes, IIE Trans. 18 (1986) 4855.
[21] H.L. Lee, M.J. Rosenblatt, Simultaneous determination of production cycles and inspection schedules in a production system,
Manage. Sci. 33 (1987) 11251137.
[22] M.K. Salameh, M.Y. Jaber, Economic production quantity model for items with imperfect quality, Int. J. Prod. Econ. 64 (2000) 59
64.
[23] W.M. Chan, R.N. Ibrahim, P.B. Lochert, A new EPQ model: integrating lower pricing, rework and reject situations, Prod. Plan.
Control 14 (7) (2003) 588595.
[24] S. Papachristos, I. Konstantaras, Economic ordering quantity models for items with imperfect quality, Int. J. Prod. Econ. 100 (1)
(2006) 148154.
[25] P.A. Hayek, M.K. Salameh, Production lot sizing with the reworking of imperfect quality items produced, Prod. Plan. Control 12 (6)
(2001) 584590.
[26] X. Zhang, Y. Gerchak, Joint lot sizing and inspection policy in an EOQ model with random yield, IIE Trans. 22 (1) (1980) 4147.
[27] H.M. Wee, J. Yu, M.C. Chen, Optimal inventory model for items with imperfect quality and shortage backordering, Omega 35 (1)
(2007) 711.
[28] L.Y. Ouyang, C.-K. Chen, H.-C. Chang, Quality improvement, setup cost and lead-time reductions in lot size reorder point models
with an imperfect production process, Comput. Oper. Res. 29 (2002) 17011717.

2116

J. Rezaei, M. Davoodi / Applied Mathematical Modelling 32 (2008) 21062116

[29] K.-N. Francis Leung, A generalized geometric-programming solution to An economic production quantity model with exibility and
reliability considerations, Eur. J. Oper. Res. 176 (1) (2007) 240251.
[30] M. Freimer, D. Thomas, J. Tyworth, The value of setup cost reduction and process improvement for the economic production
quantity model with defects, Eur. J. Oper. Res. 173 (1) (2006) 241251.
[31] L.Y. Ouyang, H.-C. Chang, Impact of investing in quality improvement on (Q, r, L) model involving the imperfect production
process, Prod. Plan. Control 11 (6) (2000) 598607.
[32] Y.P. Chiu, Determining the optimal lot size for the nite production model with random defective rate, the rework process, and
backlogging, Eng. Optimiz. 35 (4) (2003) 427437.
[33] T.L. Urban, Deterministic inventory models incorporating marketing decisions, Comput. Ind. Eng. 22 (1) (1982) 8593.
[34] M. Ben-Daya, Multi-stage lot sizing models with imperfect processes and inspection errors, Prod. Plan. Control 10 (2) (1989) 118126.
[35] H.-H. Lee, A cost/benet model for investments in inventory and preventive maintenance in an imperfect production system,
Comput. Ind. Eng. 48 (2005) 5568.
[36] M. Ben-Daya, The economic production lot-sizing problem with imperfect production processes and imperfect maintenance, Int. J.
Prod. Econ. 76 (2002) 257264.
[37] S.K. Goyal, C.K. Huang, K.C. Chen, A simple integrated production policy of an imperfect item for vendor and buyer, Prod. Plan.
Control 14 (7) (2003) 596602.
[38] L.Y. Ouyang, K.S. Wu, C.H. Ho, The integrated inventory models with defective items and controllable lead time, Prod. Plan.
Control 14 (2003) 562578.
[39] K. Deb, Multi-Objective Optimization Using Evolutionary Algorithms, Wiley, New York, 2001.
[40] L. Davis, The Handbook of Genetic Algorithms, Van Nostrand Reinhold, New York, 1991.
[41] D.E. Goldberg, Genetic Algorithms in Search, Optimization, and Machine Learning, Addison-Wesley, Reading, MA, 1989.
[42] J.H. Holland, Adaptation in Natural and Articial Systems, The University of Michigan Press, Ann Arbor, IL, 1975.
[43] Z. Michalewicz, Genetic Algorithms + Data Structures = Evolution Programs. AI Series, Springer, New York, 1994.
[44] A. Homaifar, S.H.-V. Lai, X. Qi, Constrained optimization via genetic algorithms, Simulation 62 (4) (1994) 242254.
[45] Z. Michalewicz, M. Schoenauer, Evolutionary algorithms for constrained parameter optimization problems, Evol. Comput. J. 4 (1)
(1996) 132.
[46] K. Deb, Evolutionary algorithm for multi-criterion optimization in engineering design, in: K. Miettinen, P. Neittaanmaki, M.
Makela, J. Periaux (Eds.), Evolutionary Algorithms in Engineering and Computer Science, Wiley, Chichester, 1999, pp. 135161.