J Ind Eng Int (2015) 11:485–494

DOI 10.1007/s40092-015-0115-9

ORIGINAL RESEARCH

Application of queuing theory in production-inventory

optimization
Reza Rashid1 • Seyed Farzad Hoseini1 • M. R. Gholamian1 • Mohammad Feizabadi2

Received: 4 December 2014 / Accepted: 29 May 2015 / Published online: 4 July 2015
 The Author(s) 2015. This article is published with open access at Springerlink.com

Abstract This paper presents a mathematical model for Day by day, the number of researchers, attracted to
an inventory control system in which customers’ demands production-inventory systems in supply chain, is increas-
and suppliers’ service time are considered as stochastic ing. In inventory management point of view, a manufac-
parameters. The proposed problem is solved through turer with a limited production capacity needs to hold
queuing theory for a single item. In this case, transitional finished goods inventory as safety stocks. For this inven-
probabilities are calculated in steady state. Afterward, the tory system, if demand is less than service capacity, the
model is extended to the case of multi-item inventory manufacturer needs to use different policies for different
systems. Then, to deal with the complexity of this problem, inventory levels.
a new heuristic algorithm is developed. Finally, the pre- For instance, it should use two points of inventory level:
sented bi-level inventory-queuing model is implemented as one for starting and the other for finishing the production
a case study in Electroestil Company. line. For better description of the model, Fig. 1 illustrates
the inventory level for the system when production rate l
Keywords Production inventory  Queuing theory  and demand rate D are considered deterministic. Actually,
Multi-item inventory  Heuristic algorithm in a real-world system, different stochastic parameters may
affect the model and increase the complexity of the system.
In summary, it is clear that in spite of many contribu-
Introduction tions to the stock control, there is little consideration
regarding production-inventory models in stochastic envi-
Nowadays, supply chains play an important role to meet ronments. For this reason, we represent a mathematical
diverse needs of customers. A supply chain can be defined model in which the main contributions of this paper can be
as a network of organizations that collaborate to control summarized as follows:
and manage materials and information flow from suppliers
• A production inventory system is developed in an
to customers (Aitken 1998). One of the challenging issues
uncertain environment.
in supply chain management is to find optimal policies for
• Queuing theory is used to provide a stochastic model.
inventory systems. The main objective of inventory man-
• The proposed model is extended for multi-item inven-
agement is to balance conflicting goals such as stock costs
tory systems.
and shortage costs (Arda and Hennet 2006).
• A new heuristic algorithm is proposed to solve the
model.
& Seyed Farzad Hoseini • All the steady-state equations are solved in one state.
[email protected]
This model can be probably applied to inventory systems
1
Department of Industrial Engineering, Iran University of where demand and production time are uncertain. Company
Science and Technology, Narmak, 16844 Tehran, Iran has setup costs and it also holds finish goods inventory.
2
Department of Industrial Engineering, Sharif University, The remainder of this paper is organized as follows. In
Azady, Tehran, Iran Sect. 2, a brief literature review is presented. In Sect. 3, an

123
486 J Ind Eng Int (2015) 11:485–494

D
μ- D
r …

Fig. 1 Inventory level of the production model

inventory model is presented and then a branch and bound inventory control problem in which periods between two
algorithm is proposed. Afterward, the model is validated in replenishments were assumed to be independent random
Sect. 4. Then, conclusions are made in Sect. 5. variables.
Generally, in a stock control system, most of the
parameters are not deterministic. Thus, some researchers
Literature review used queuing theory to construct their models. In this area,
there are some researchers who provided a model for
In the literature, there are some researchers who have dealt stochastic demand. Some researchers presented a model for
with production-inventory systems. For example, Yu and stochastic lead time. Parlar (1997) presented an inventory
Dong (2014) considered a two-stage production lot sizing model which was combined with queuing theory to con-
problem which used an inventory system with random sider demand and lead time as stochastic parameters. Ha
demand arrivals. To solve the problem, they proposed a (1997) considered Poisson distribution for demands and
numerical approach for the problem. Baek and Moon (2014) exponential for production times in a single item make-to-
considered a lost sales production-inventory system in an stock production system. He proposed an M/M/1/S queuing
uncertain environment. They also used queuing theory to system for modeling the system. Arda and Hennet (2006)
present a stochastic model for the system. Also, some pro- addressed inventory control of a multi-supplier strategy in
duction-inventory papers in the literature assumed that the a two-level supply chain. They considered random arrivals
lead time is negligible or it can be ignored in practice when for customers and random delivery time for suppliers and
it is short in contrast to other time factors (Karimi-Nasab represented the system as a queuing network.
and Seyedhoseini 2013). Most of such research papers did Sapna Isotupa (2006) considered lost sales of (s, Q) in-
not consider lost sales. Many of them, however, considered ventory system with two customer groups and illustrated
back orders as a rational managerial policy. the Markov processes. Boute et al. (2007), in a two-echelon
Chang and Lu (2011) considered a serial production supply chain, show that by including the impact of the
system controlled by the base-stock policy. They presented order decision on lead times, the order pattern can be
a phase-type approximation for a controlled base-stock smoothed to a considerable extent without increasing stock
serial production system. They also proposed a cost model levels. Karimi-Nasab and Konstantaras (2013) considered
to determine the optimal base-stock level. special sales offer for a single item from the supplier under
Jewkes and Alfa (2009) considered a production system stochastic replenishment intervals.
in which a supplier produces semi-finished items on a Jain and Raghavan (2009) studied batch ordering in
make-to-stock basis for a manufacturer that customizes the multi-echelon supply chains and used queuing theory to
items on a make-to-order basis. The manufacturer attempts capture the behavior of the manufacturing supply chain
to determine the optimal point of differentiation and its network. Babai et al. (2011) investigated stochastic demand
optimal semi-finished goods buffer size. They used matrix and lead time and analyzed a single item inventory system
geometric methods to evaluate performance through vari- through queuing theory. Bahri and Tarokh (2012) assumed
ous measures for this system. that the delivery lead time is stochastic and follows an
Some researchers worked on multi-item inventory sys- exponential distribution. Also, the shortage during the lead
tems. For instance, Shavandi et al. (2012) proposed a new time is permitted and completely back ordered for the buyer.
constrained multi-item pricing and inventory model. They Seyedhoseini et al. (2014) considered Poisson demand
covered three categories of perishable products in their for customer in a cross docking problem. They employed
model. Taleizadeh et al. (2012) dealt with a multi-product queuing theory to provide a stochastic model. Salameh et al.

123
J Ind Eng Int (2015) 11:485–494 487

(2014) combined the separate works on substitution and Model description

joint replenishment and introduced a solution procedure for
solving the joint replenishment model with substitution for In this paper, a historical production-inventory model is
two products within the framework of the classical eco- developed. Moreover, a new stochastic multi-item pro-
nomic order quantity model. Karimi-Nasab and Sabri- duction-inventory model in which shortages have been
Laghaie (2014) extended classical economic production considered lost sales is proposed. In this strategy, the
quantity model to the case of stochastically generated poor- production cycle starts when the manufacturer’s inventory
quality items, while an imperfect screening scheme was falls to or below the reorder point r and stops when its on-
devised to recognize such items from healthy ones. hand inventory reaches its optimal level R.
Krommyda et al. (2015) studied an inventory control Inspired by Sapna Isotupa (2006) who developed a
problem in which demand was satisfied using two mutually stochastic inventory model for (R, Q) systems, we also
substitutable products. Their aim was to determine the order developed a new stochastic model for the historical
quantity for each product that maximizes the joint profit production-inventory model. Thus, the customers’ arri-
function. val is assumed to be Poisson distribution with rate of D,
Seyedhoseini et al. (2015) applied queuing theory to and production time is negative exponentially distribu-
propose a mathematical model for inventory systems with tion with rate of l. Also, it has been assumed that if
substitute flexibility. (Rashid et al. 2015) also considered a one type of products is not under production, its pro-
location-inventory model. To prepare a stochastic inven- duction capacity would equally divide to the other
tory model, they used bi-level Markov process. Consider- products. To achieve a comprehensive model, queuing
ing the effectiveness of queuing theory in inventory theory is employed to describe the inventory system.
problems, we have also used queuing theory to develop a The transition diagram for this model is illustrated in
stochastic stock control model. Fig. 2.
As demonstrated in Table 1, there are some researchers For this diagram, states of the system have been
who considered (R, Q) policy in stochastic programming. decomposed into two sets: A and B. In set A, the system is
However, some of them did not consider multi-item inven- in production mode and in set B system in demand mode;
tory, and some of them did not consider production inven- in demand mode, no production would occur. Also, states
tory. Also, few researchers have analyzed their model with of the system are represented by (I, J), where
queuing theory. Hence, we propose a mathematical model I 2 {0, 1, …, R} demonstrates the level of inventory and
for the production-inventory problem in the stochastic J 2 {P, D} demonstrates the mode of the system. Hence-
environment. forth, let pi,j denote the steady-state probability of state (I,
J). For this queue, three lemmas are proposed to set steady-
state probabilities.
Model development
Lemma 1 Steady-state probabilities in terms of state
In this section, first, the proposed model is developed. pR1;P for the system can be calculated as follows:
Then, it is extended for the multi-item production-inven-  
lþD
tory system. Afterward, a new algorithm is proposed to pR2;P ¼  pR1;P ; ð1Þ
l
solve the model. As for notifications, parameters are ! 
Xl3   j   l2
described as follows: D Dþl D
pRl;P ¼ þ  pR1;P
j¼0
l l l
r Reorder level (I,J) State for I inventory and 8l 2 ð3; . . .; R  rÞ; ð2Þ
being at mood of J !
 k X D j
Rr3
R Optimal level pI,J Steady-state probability D
of state (I, J) prk;P ¼ 
l j¼0
l
D Demand S Amount of shortage    Rr2 ! ð3Þ
l Service rate M Number of product types
Dþl D
þ   pR1;P
SE Setup cost Ci,I Holding costs l l
Ci,S Shortage costs ri Reorder level for product i 8k 2 ð1; . . .; rÞ;
pr?1,D,i Steady-state probability Ri Optimal level for product i l
of being in state (r?1, pi;D ¼  pR1;P 8i 2 ðr þ 1; . . .; RÞ: ð4Þ
D) for product i
D

123
488 J Ind Eng Int (2015) 11:485–494

Table 1 Literature of the stochastic inventory and production inventory

References Shortage Lead time Demand Replenishment Multi-item Production
policy inventory inventory
Lost Back Did not
sales order consider

Sapna Isotupa (2006) H Stochastic Stochastic (R, Q) No No

Arda and Hennet (2006) H Stochastic Stochastic (S-1, S) No Yes
Boute et al. (2007) H Stochastic Stochastic (T, S) No No
Hill et al. (2007) H Deterministic Stochastic (S-1, S) No No
Hennet and Arda (2008) H Stochastic Stochastic (S-1, S) No No
Chang and Lu (2011) H H Stochastic Stochastic (S-1, S) No Yes
Babai et al. (2011) H Stochastic Stochastic (S-1, S) No No
Taleizadeh et al. (2012) H Stochastic Dynamic (T, s, S) Yes No
Tlili et al. (2012) H Deterministic Stochastic (T, s, S) No No
Bahri and Tarokh (2012) H Stochastic Deterministic (R, Q) No Yes
Shavandi et al. (2012) H Deterministic Deterministic (T, s, S) Yes Yes
Alimardani et al. (2013) H Stochastic Stochastic (S-1, S) Yes No
Guerrero et al. (2013) H Deterministic Stochastic (T, s, S) Yes No
Karimi-Nasab and Konstantaras (2013) H Stochastic Deterministic (R, Q) No No
Karimi-Nasab and Seyedhoseini (2013) H Deterministic Deterministic (T, S) Yes Yes
Karimi-Nasab and Sabri-Laghaie (2014) H Deterministic Deterministic (R, Q) No Yes
Salameh et al. (2014) H Deterministic Deterministic (R, Q) Yes No
Yu and Dong (2014) H Deterministic Stochastic (R, Q) No Yes
Rashid et al. (2015) H Stochastic Stochastic (R, Q) No No
Seyedhoseini et al. (2015) H Deterministic Stochastic (R, Q) No No
Krommyda et al. (2015) H Deterministic Deterministic (R, Q) Yes No
Our model H Stochastic Deterministic (R, Q) Yes Yes

μ μ μ μ μ μ μ

0,P 1,P ...

... r-1,P r,P r+1,P ...
... R-1,P

D D D
D D D D

Set A

r+1,D r+2,D ...

... R-1,D R,D
D D D
D

Set B

Fig. 2 Transition diagram for a product in the proposed model

Regarding Fig. 1 and Eq. 5, Eq. 1 is true. Using Eq. 1,

Proof For a Markov process, Eq. 5 is true:
for L between 3 and R-r, Eq. 6 could emerge, which
X
N X
N
proves Eq. 2,
kjx  pj ¼ kxj  px ; ð5Þ
j¼1 j¼1 l  pRL;P ¼ D  pRLþ1;P þ l  pR1;P : ð6Þ
where N is the number of all steady states, and kjx is the When I is smaller than r, then Eq. 7 is true:
probability rate from state j to state x.

123
J Ind Eng Int (2015) 11:485–494 489

Fig. 3 Transition diagram for

state (0, p)
μ μ μ
μ

... g+1 g ... 1 0

D D D D D
D

 j For this queue, if psg represents a steady-state probability

D
prj;P ¼ pr;P  : ð7Þ of g shortages, then it would be clear that:
l
ðD þ lÞpsgþ1 ¼ ðDÞpsg 8g  0: ð11Þ
Using Eqs. 2 and 7 could result in Eq. 3 . As for set B, it is
clear that the values of steady-state probabilities are equal Considering Eq. 11, psg could be calculated as follows:
and also Eq. 8 is true:  g
l D
pR;D ¼  pR1;P : ð8Þ psg ¼ ps0 : ð12Þ
D Dþl

So, Eq. 4 is true.Considering Lemma 1, pR-1,P is equal to By considering Eq. 12, Eq. 13 could be obtained as
1
F , where F can be calculated through Eq. 9: X1  g
D
0 0  Rr1 11 ps0 ¼ p0 : ð13Þ
D D g¼0
Dþl
l l þ DB 
Bl l CC
F ¼ ðR  r Þ  þ1þ @1 þ @ AA
D D 1  Dl Also, the average shortage is equal to:
0  Rr1 1    2  k !
D
 D D D D
þ
1 B l l C S¼ ps0 þ2 þ þ k þ. . .
@ð R  r  2 Þ  A Dþl Dþl Dþl
1  Dl 1  Dl
 
0  rþ1 10  Rr2 1 ¼ ps0 z 1 þ 2ðzÞ1 þ    þ kðzÞk1 þ. . . ; ð14Þ
D
 D
1  D  Rr2
Bl l CB l lþD D C
þ@ A@ þ  A:
1  Dl 1  Dl D l where z is equal to D
Dþl. In addition, it is known that:
 
ð9Þ 1 þ 2ðzÞ1 þ    þ kðzÞk1 þ. . .
Lemma 2 The expected value for shortage can be com-
d 
puted through the following equation: ¼ z þ ðzÞ2 þ    þ ðzÞk þ. . .
  dz
D  
S ¼ p0  : ð10Þ d z 1
l ¼ ¼ : ð15Þ
dz 1  z ð1  zÞ2
Proof The shortage occurs when the producer has no
Considering Eqs. 14 and 15 could result in Eq. 10.
inventory and also a demand is announced to the retailer. In
this section, g is used for representing a state of shortage. Lemma 3 For this inventory system, the expected value
For computing the expected value of shortage, we of the inventory level can be calculated by A.pR-1,P, where
decompose state (0, P) as in Fig. 3. A can be calculated by Eq. 16.

123
490 J Ind Eng Int (2015) 11:485–494

0  Rr1 1
    D
 Dl
ð R  2Þ  l l Rþrþ1 lþD Bl C
ð R  1Þ þ þ ðR  r Þ   þR @ A
lþD D 2 D 1  Dl

2
 Rr    Rrþ1 3
 l  l þ D6 1  ðR  r þ 1Þ  l  
D
1  Dl þ Dl  Dl
6 D 7
7
  4  2  1þ2
D D D l 5
1l
2 0  Rr1 13
! D D !  
1  1 ðR  r þ 3ÞðR  r  2Þ
6 Bl l C7
þR  4R  r  2  @ A5  
1  Dl 1  Dl 1  Dl 2

2 ð16Þ
 Rr    Rrþ1 3
! 1  ð R  r þ 1 Þ  D
1  D
þ D
 D  
1 l 6 l l l l D 7
þ  6
4  2  1þ2 7
1  Dl D D l 5
1l
0  Rr2 1
1 D  Rr2
B l lþD D C
þ@ þ  A
1  Dl D l

2 0  rþ1 1 0  r    rþ1 13
D D D D D D
6  Bl l C D B 1  ðr þ 1Þ  l 1 l þ 
l l C7
4r  @ A @  2 A5
1  Dl l
1  Dl

Proof For this inventory system, Eq. 17 is true: Also, for this queue the next three equations are true:
X
R1 XR Rr 
X   l2 Rr 
X   l2  
 lþD D lþD D lþD
I¼ pj;P  j þ pk;D  k ¼   ðR  lÞ ¼ R 
j¼1 k¼rþ1 l¼3
D l D l D
l¼3
0  Rr1 1
l3   j
Rr X
X D
 Dl  l  l þ D
D Bl C
 ðR  lÞ þ ðR  lÞ @ A  
l¼3 j¼0
l 1  Dl D D

r  i   Rr2 ! 2  Rr    Rrþ1

X D X
Rr3
D
j 
Dþl D
þ  þ  1  ðR  r þ 1Þ  Dl 1  Dl þ Dl  Dl
6
i1
l j¼0
l l l 6
4  2
1  Dl
 ðr  iÞ:  
D
ð17Þ  1þ2 ; ð18Þ
l

123
J Ind Eng Int (2015) 11:485–494 491

0  l2 1
l3  j D
X B1  l
Rr X X
Rr
D C
ðR  lÞ ¼ @ A  ðR  lÞ;
l¼3 j¼0
l l¼3
1  Dl

2 0  Rr1 13
! D D !  
1  1 ð R  r þ 3Þ ð R  r  2Þ
6 Bl l C7
¼R  4R  r  2  @ A5  
1  Dl 1  Dl 1  Dl 2

2  Rr    Rrþ1 3
! 1  ð R  r þ 1 Þ  D
1  D
þ D
 D  
1 l 6 l l l l D 7
þ  6
4  2  1þ2 7; ð19Þ
1  Dl D l 5
1  Dl

r  i
X X
Rr3 j    Rr2 !
D D Dþl D
 þ   ðr  i Þ
i1
l j¼0
l l l
0  Rr2 12 0  rþ1 1 0  r    rþ1 13
D  Rr2 D D D D D D
B1  l lþD D C6 Bl  l C DB 1  ðr þ 1 Þ  l 1  l þ l  l C7
¼@ þ  A4r  @ A  @  2 A5 :
1  Dl D l 1  Dl l
1  Dl
ð20Þ

"   X l3   j 
Rr X   l2
Using Eqs. 17, 18, 19, and 20, Eq. 16 can emerge. ll D Dþli D
1þ þ þ 
ll þD l¼3 j¼0
ll ll ll
Multi-product inventory model         #
Xr X D j
D i Rr3 Dþll D l2
þ  þ  pR1;P :
In this section, we considered multi-item inventory to i¼1
ll j¼0
ll ll ll
determine near optimal Ri, ri and li for each product. This ð23Þ
objective is denoted through minimizing Eq. 21:
XM  Figure 4 illustrates the inventory level, rate of production
min z ¼ Ii  Ci;I þ Si  Ci;S þ prþ1;D;i  SEi : ð21Þ and rate of demand for the two products.
i¼1 In this study, li is considered to be bigger than Di. For
For the inventory system, M types of products are fixed values of different variables, it is clear that increase in
considered. To present our heuristic algorithm, it is Ri causes an increase in Ii . Furthermore, it causes prþ1;D;i
assumed that on demand mood the product has a capacity and Si to be decreased. So, for fixed values of different
(production rate) that can be equally divided into the other variables, the long-run expected cost is pseudo-convex in R.
products. So, Eq. 20 could be used to approximate the On the other hand, for fixed values of different variables,
service rate for the type k product: increase in ri would increase the probabilities of states
P ! which have more than ri inventory level, and it causes
l l6¼k ð1  pl Þ prþ1;D;i and Ii to be increased. Moreover, it causes a
lk ¼ 1þ P ; ð22Þ
M l6¼k ðpl Þ decrease in Si . Consequently, for a fixed R, the long-run
expected cost is pseudo-convex in ri.
where n represents the number of products and pl is
PR1
equal to i¼0 pi;P;l , which represents the probability of
being in production mood for product l. It can be cal- Heuristic algorithm
P
culated from Eq. 23. In Eq. 22, l6¼k ð1  pl Þ approxi-
mates the number of products in the demand mood and Regarding the complexity of our model, simple models
P
l6¼k ðpl Þ approximates the number of products in the would take big computational times. For this reason, we
production mood. proposed a two-phased algorithm.

123
492 J Ind Eng Int (2015) 11:485–494

Fig. 4 Illustration of the RR11

inventory level for two products RR22 --DD11
μμ//22
DD22

22
μμ--DD
11
μμ--DD
DD1
1

D22
2--D
μμ//2
rr11
rr22


Phase one zi ¼ Ii  Ci;I þ Si  Ci;S þ prþ1;D;i  SEi : ð24Þ

In continuous space, z is convex due to ri and Ri. We Step 3 Choose the nodes which have the minimum zi for
suppose ri and Ri to be continuous and optimal and these each product.
were calculated by the steepest ascent method. For better
description, it is assumed that there is only one product and
if rli and Rli denote the solution of the steepest ascent Computational results
method, then Fig. 5 demonstrates the optimal solution in
discrete space for just one product. In this research, the proposed algorithm is coded in
Considering Fig. 5, for the discrete problem, the nearest C?? software. For evaluation of the algorithm, we pro-
points in any direction to the optimal solution could be duced five examples for M equal to 10, where all param-
found in a space in which ri is |rli| or |rli| ? 1, and Ri is |Rli| eters are selected randomly while creating the data set. The
or |Rli| ? 1. Consequently, a search among these nodes can parameters are set as follows: Ci;I 2 ½1; 10, Ci,S 2 [1, 10],
provide a perfect local solution. Assuming a number of Di 2 [1, 10], l = 10 M and SEi 2 [30, 60]. In this case,
product equal to M, there would be 22M nodes in the space instead of letting R go to infinity, the maximum value of
that needs to be searched. So, for big M, phase two is R was set to be 10,000. Therefore, the maximum value of
proposed. s was 9999. We also limited the computational time to 5 h
and mean solving time for our algorithm was 223.097 s,
Phase two while none of the examples could be solved through
searching all feasible solutions.
Considering the optimal solution resulting from the steep- In this model, an example is generated to analyze the
est deepest algorithm, we proceed with the following steps: parameters of the model in which CI,CS, D, SE and l are
Step 1 Calculate li for each of the products for the equal to 5, 20, 3, 200 and 4, respectively. Also, the
continuous solution. behavior of inventory costs has been studied for different
Step 2 For each product, search the nodes where | rli| or values of R and r in Fig. 6.
|rli| ? 1, and Ri is |Rli| or |Rli| ? 1, and then calculate It is proposed to use this model for companies using
Eq. 24, independently, workshop system and reorder point for replenishing their
finished goods inventory. To provide a better description of
the model, an example is presented in Table 2.
A near optimal solution for this example is 164 where
(r1 ; r2 ; r3 ; R1 ; R2 ; R3 ) is equal to (1, 1, 1, 4, 3, 3), respec-
tively. If each product is to be planned separately, the
|rl|+1 optimal cost would be equal to 211, 28 % more than our
model’s costs. We considered Product 3 and analyzed costs
|rl| related to D in Fig. 7. To analyze the performance of our
heuristic, R and Q are limited to be less than 100, and the
optimal solution for different Ds are prepared.
As in Fig. 7, the same pattern has been demonstrated for
|Rl| |Rl|+1
shortage and holding costs. On the other hand, the pattern
of setup cost is totally different. Furthermore, there is not a
Fig. 5 Solution of the discrete problem great difference among heuristic and optimal costs.

123
J Ind Eng Int (2015) 11:485–494 493

25
20
15
R
10
r
5
0
4 13 32 64 128 258 512 D

Fig. 8 Relation between R, r and D for ESR320

day, and demands of shop refrigerators for different pro-

duct types are equal and they are equal to 7 per day. They
Fig. 6 Behavior of the model regarding r and R ðr\RÞ have defined CI to be equal to 100,000 tomans for each
product, CS to 2 million tomans—Iran’s currency—for
Table 2 An example for describing the proposed model each product, and SE to 300,000 tomans for each product.
For this company, Table 3 illustrates the performance of
l = 15 CI Cs D SE
the proposed model. Significant changes in inventory costs
Product 1 10 100 4.5 50 have been determined if the proposed model is used.
Product 2 20 250 4 70 As shown in this table, there is a big difference between
Product 3 30 300 4 100 the optimal solution and the current situation of the com-
pany. In the current situation, shortage costs and setup
costs are small, but holding costs are big. For this company,
400
setup costs(near
the rate of production is much more than the rate of
opmal) demand. This could be a reason for the low quantity of r.
300
holding costs(near In this model, the demand rate has significant effects on
cost

200 opmal) r and R. Therefore, we only considered ESR320 which is a

shortage costs(near shop refrigerator. For this product, R and r were analyzed
100 opmal)
according to the demand rate in Fig. 7. The same pattern
setup cost (opmal)
0 was displayed for them. Also, it has been discovered that as
2 4 6 8 10 12 14 16
D the demand rate becomes bigger, when it is bigger than the
service rate, the optimal costs decrease. For example, the
Fig. 7 Sensitivity of different costs due to the demand rate costs for D = 64 is equal to 30 which is significantly less
than for D = 7. However, for small demand rates, its
This model could be implemented properly for work- increase makes optimal costs bigger. For example, in
shop systems. The Electroestil Company was established in D = 13, the optimal cost is equal to 302.
1980 and now it is one of the most popular brands of
refrigerator production in Iran. Electroestil produces dif-
ferent kinds of refrigerators for houses and shops. It has 20 Conclusion and future directions
types of refrigerators for houses and 15 types for shops.
In this company, there are four workshops and 1500 In this paper, a new production-inventory model is devel-
employees. This company has a production capacity of 800 oped (Fig. 8). To prepare a cohesive model, demand and
products per day for house refrigerators and 120 products production time are considered as stochastic parameters
per day for shop refrigerators. Experts in Electroestil and queuing theory is used to calculate long-run inventory
approximate that demands of house refrigerators for dif- costs. The proposed model also has been extended for
ferent product types are equal and they are equal to 30 per multi-item inventory systems. The inventory model has

Table 3 Comparison between our model and the current planning for each product (91000 tomans)
House refrigerator Shop refrigerator
r R Inventory holding costs Shortage costs Setup costs r R Inventory costs Shortage costs Setup costs

Multi-item model 1 4 175 82 41 1 3 144 49 73

Current planning 8 20 1229.1 0.001 8.638 5 12 610.63 44.010 9.907

123
494 J Ind Eng Int (2015) 11:485–494

been analyzed concerning r and R. These parameter system with batching and ordering constraints. Eur J Oper Res
behaviors are investigated. We also proposed a heuristic 231:98–108
Ha AY (1997) Stock-rationing policy for a make-to-stock production
algorithm to solve the problem. The results demonstrate the system with two priority classes and backordering. Naval Res
efficiency of the proposed algorithm. To prepare a better Logist 44:457–472
description of the model, a real-world example is studied Hennet J-C, Arda Y (2008) Supply chain coordination: a game-theory
and analyzed. Respecting the collected data, it has been approach. Eng Appl Artif Intell 21:399–405
Hill RM, Seifbarghy M, Smith DK (2007) A two-echelon inventory
observed that there is a significant relation between the model with lost sales. Eur J Oper Res 69:753–766
demand rate and production rate. If demand is smaller than Jain S, Raghavan NS (2009) A queuing approach for inventory
service rate, increase of demand causes an increase in planning with batch ordering in multi-echelon supply chains.
optimal costs. On the other hand, when it is bigger than the Central Eur J Oper Res 17:95–110
Jewkes EM, Alfa AS (2009) A queueing model of delayed product
service rate, a small increase may reduce the optimal costs. differentiation. Eur J Oper Res 199:734–743
Lost sales shortage may be considered as one of the Karimi-Nasab M, Konstantaras I (2013) An inventory control model
weaknesses in our model. So, in future, the model can be with stochastic review interval and special sale offer. Eur J Oper
extended through considering back-order shortage. This Res 227:81–87
Karimi-Nasab M, Sabri-Laghaie K (2014) Developing approximate
may increase the complexity of the problem while the algorithms for EPQ problem with process compressibility
model would become more realistic. Another extension of and random error in production/inspection. Int J Prod Res
this research is possible by covering corruption rates for 52:2388–2421
perishable inventories. Karimi-Nasab M, Seyedhoseini S (2013) Multi-level lot sizing and
job shop scheduling with compressible process times: a cutting
plane approach. Eur J Oper Res 231:598–616
Open Access This article is distributed under the terms of the Krommyda I, Skouri K, Konstantaras I (2015) Optimal ordering
Creative Commons Attribution 4.0 International License (http:// quantities for substitutable products with stock-dependent
creativecommons.org/licenses/by/4.0/), which permits unrestricted demand. Appl Math Model 39:147–164
use, distribution, and reproduction in any medium, provided you give Parlar M (1997) Continuous-review inventory problem with random
appropriate credit to the original author(s) and the source, provide a supply interruptions. Eur J Oper Res 99:366–385
link to the Creative Commons license, and indicate if changes were Rashid R, Bozorgi-Amiri A, Seyedhoseini S (2015) Developing a new
made. stochastic competitive model regarding inventory and price.
J Ind Eng Int, pp 1–10
Salameh MK, Yassine AA, Maddah B, Ghaddar L (2014) Joint
References replenishment model with substitution. Appl Math Model 38:
3662–3671
Aitken JM (1998) Supply chain integration within the context of a Sapna Isotupa K (2006) An (s, Q) Markovian inventory system with
supplier association: case studies of four supplier associations. lost sales and two demand classes. Math Comput Model 43:
Cranfield University, UK 687–694
Alimardani M, Jolai F, Rafiei H (2013) Bi-product inventory planning Seyedhoseini S, Rashid R, Teimoury E (2014) Developing a cross-
in a three-echelon supply chain with backordering, Poisson docking network design model under uncertain environment.
demand, and limited warehouse space. J Ind Eng Int 9:22 J Ind Eng Int, pp 1–12
Arda Y, Hennet J-C (2006) Inventory control in a multi-supplier Seyedhoseini S, Rashid R, Kamalpour I, Zangeneh E (2015)
system. Int J Prod Econ 104:249–259 Application of queuing theory in inventory systems with
Babai MZ, Jemai Z, Dallery Y (2011) Analysis of order-up-to-level substitution flexibility. J Ind Eng Int, pp 1–8
inventory systems with compound Poisson demand. Eur J Oper Shavandi H, Mahlooji H, Nosratian NE (2012) A constrained multi-
Res 210:552–558 product pricing and inventory control problem. Appl Soft
Baek JW, Moon SK (2014) The M/M/1 queue with a production- Comput 12:2454–2461
inventory system and lost sales. Appl Math Comput 233:534–544 Taleizadeh AA, Niaki STA, Seyedjavadi SMH (2012) Multi-product
Bahri M, Tarokh MJ (2012) A seller-buyer supply chain model with multi-chance-constraint stochastic inventory control problem
exponential distribution lead time. J Ind Eng Int 8:1–7 with dynamic demand and partial back-ordering: a harmony
Boute RN, Disney SM, Lambrecht MR, Van Houdt B (2007) An search algorithm. J Manuf Syst 31:204–213
integrated production and inventory model to dampen upstream Tlili M, Moalla M, Campagne J-P (2012) The trans-shipment problem
demand variability in the supply chain. Eur J Oper Res 178:121–142 in a two-echelon, multi-location inventory system with lost sales.
Chang K-H, Lu Y-S (2011) Inventory management in a base-stock Int J Prod Res 50:3547–3559
controlled serial production system with finite storage space. Yu AJ, Dong Y (2014) A numerical solution for a two-stage
Math Comput Model 54:2750–2759 production and inventory system with random demand arrivals.
Guerrero W, Yeung T, Guéret C (2013) Joint-optimization of inven- Comput Oper Res 44:13–21
tory policies on a multi-product multi-echelon pharmaceutical

123