Fuzzy EOQ model with stockout

Wen-Kai K. Hsu and Jun-Wen Chen

Department of Shipping Transportation & Management
National Kaohsiung Marine University, Taiwan, R.O.C.
khsu@mail.nkmu.edu.tw

ABSTRACT several parameters such as cost coefficients, demands

etc. Accordingly, most of the inventory problems
For the fixed order size inventory models, the
under fuzzy environment can be addressed by fuzzing
economic order quantity (EOQ) model is most
these parameters. For instance, Park[11] discuss the
well-known. However, sometimes, the coefficients of
EOQ model with fuzzy cost coefficients. Ishii and
model may be fuzzy. This paper firstly discusses the
Konno[5], Petrovic et. al. [12], and Kao and Hsu [6]
EOQ model with fuzzy demands. Two cases of
investigate the Newsboy inventory model with fuzzy
stockout, backorder and lost sales, are then
cost coefficients and demands respectively. Roy and
considered. The Yager’s ranking method for fuzzy
Maiti [13] develop a fuzzy EOQ model with a
numbers is utilized to find the optimal inventory
constraint of fuzzy storage capacity. Chang [1]
policies. A set of numerical data is employed to
construct a fuzzy EOQ model with fuzzy defective
analyze the characteristics of proposal models.
rate and fuzzy demand. Yao and Chiang[17] compare
Finally, for completeness of model, this paper also
the EOQ model with fuzzy demand and fuzzy
fuzzies holding cost and shortage cost coefficients.
holding cost in different solution methods. Kao and
Hsu [7] find the lot size-reorder point model with
Key Words: inventory, fuzzy, EOQ, backorder, lost
fuzzy demand. Besides, there are another kind of
sales
studies which fuzzes the decision variables of
inventory models. For example, Lee and Yao [9] and
1.Introduction
Yao and Lee [18,19] develop the EOQ model with
Within the context of traditional inventory
fuzzy ordering quantities; Chang [2] discusses the
models, the pattern of demands are either
production quantities of EPQ model as a fuzzy
deterministic or uncertain. In practice, the latter
number; Chang and Yao [3] investigate the EOQ
corresponds more to the real-world environment. To
model with fuzzy order point; Lee and Yao [8] show
solve these inventory problems with uncertain
the EPQ model with fuzzy demands and fuzzy
demands, the classical inventory models usually
production quantities.
describe the demands as certain probability
This purpose of this paper firstly is to discusses
distributions and then solve them. However,
the EOQ model with fuzzy demands. Two cases of
sometimes, demands may be fuzzy, and more suitably
stockout, backorder and lost sales, are then
described by linguistic term rather than probability
considered. In general, the stockout is described by
distributions. If the traditional inventory theories can
ether backorder or lost sales. With a fuzzy sense, the
be extended to fuzzy senses, the traditional inventory
latter problem was discussed few in the previous
models would have wider applications.
studies. Because demands are fuzzy, the quantities of
Usually, inventory systems are characterized by

1
stockout, which depend on the demands and the calculated as:
period time of stockout, will also be fuzzy. Most of
the previous paper directly supposes the model s = t ⋅λ (2)
parameters as a triangular fuzzy number and then
Inventory
finds their optimal solutions. However, this may not level

be suitable, because they should be justified

previously. Instead of the quantities of backorder, this
paper suggests the time period of stockout which is a
crisp number as the economic order point. In other
word, this paper tries to find the optimal ordering Q

quantities, Q, and time period of stockout, t, for the t

EOQ model with backorder and fuzzy demands. Time

t' = (Q − s) / λ s
Besides, the other stockout case, lost sales, which has
t1 t2 t3
been considered few, will also be discussed in this
paper. Finally, for completeness of model, the holding Figure 1. The EOQ model with backorder
cost and shortage cost coefficients will also be
Let p be the backordering cost per unit and
fuzzied.
The structure of this paper is organized as assumed to be directly proportional to the length of
follows. First, the traditional EOQ model with
backorder will be introduced. Next, with fuzzy the time period of stockout. The annual cost then can

demands, the models of backorder case and lost sales be derived as [14]:
case will be discussed respectively. Finally, the
characteristics of proposed models will be illustrated C B (Q , s ) = a λ / Q + h (Q − s ) 2 / 2 Q
(3)
and some conclusions will be made. + p ⋅ s 2 / 2Q

2.The EOQ model with backorder

Since s = t ⋅ λ , if λ is fuzzy then s, the quantities
Let λ denote the demand rate for the EOQ of backorder, will also be fuzzy. In the study of
Chang and Yao [3], λ is a constant and s is supposed
mode, then the annual cost function can be described
to be a triangular fuzzy number. This is the same as
as: assuming that t is a fuzzy number. However, for
managers, t should be more easily controlled than λ
C (Q) = aλ / Q + hQ / 2 (1)
which usually depends on the external environment.
where a is the ordering cost for each inventory cycle Hence, it should be more practical to treat λ as a
fuzzy number and t as a crisp number. Accordingly,
and h is the holding cost per unit per year. If
by replacing s with tλ , model (3) can be rewritten
backorder is permitted, then the model can be as:
C B (Q , t ) = a λ / Q + h (Q − tλ ) 2 / 2 Q
depicted as Figure 1. The time period of stockout is t (4)
+ p ⋅ t 2 λ 2 / 2Q
and the quantities of backorder denoted by s can be in which t is treated as a decision variable, the order

2
point, instead of s. ~
the annual cost function, T (α λ ) is a fuzzy number,
By taking the derivatives of C B ( Q , t ) with
respect to Q and t, it is not difficult to show that the ~
we can compare T (α λ ) of different Q by using
optimal solutions for models (1) and (4) are
respectively as: some ranking methods to find the optimal solution
Q * with a minimal total cost. In the literature, a lot
Q * = 2aλ / h (5)
of ranking methods have been proposed and
and discussed. Of which, not every method is applicable

~
Q * = (h + p) / p ⋅ 2aλ / h . (6.1) to rank T (α λ ) of all possible Q. The reason is

t * = (Q * / λ ) ⋅ h /(h + p) (6.2) most methods need to know the membership

Note, when p → ∞ , t → 0 and Equation
*
functions of all fuzzy number to be compared, and
(6.1) will degenerate to Equation (5). this is not possible in this case. The method proposed
by Yager [16], which is also named signed distance[1,
3.The EOQ model with backorder and fuzzy 17], does not need to know the explicit form of the
demands membership functions, and can thus be applied here.
~
Let λ be a normal fuzzy number with The Yager’s ranking index ranks the fuzzy

~ numbers by an area measurement defined as:

parameters λ = (l , m , n , u ) , then the membership ~ ~ ~
I (T ) = I L (T ) / 2 + I R (T ) / 2 (10)
~ ~
function of λ can be defined by a left shape function where I L (T ) represents the area bounded by the
L ( x ) and a right shape function R ( x ) as: ~
left shape function of T (α λ ) , the x axis, the y axis

⎧ L ( x ), l≤x≤m
⎪ and the horizontal line u T~ = 1 ; and
µ λ~ ( x ) = ⎨1, m≤x≤n (7)
⎪ R ( x ), n≤x≤u ~
⎩ I R (T ) represents the area bounded by the right
The above equation can also be described by the ~
~ shape function of T (α λ ) , the x axis, the y axis and
terms of α -level cut of λ as:

λ (α ) = [min . µ λ~−1 (α ), max . µ λ~−1 (α )] the horizontal line u T~ = 1 . The Yager’s ranking
(8)
= [ L−1 (α ), R −1 (α )], 0 ≤ α ≤ 1
~
index of T (α λ ) thus can be calculated as:
Firstly, we discussed the EOQ model with fuzzy
demand, namely, the stockout is neglected. 1 1
~
I (T ) = ∫ C[Q | λ = Lλ−1 (α λ )]dαλ
According to the extension principle, the model can 2 0 (11)
be described by terms of α as: 1 1
+ ∫ C[Q | λ = Rλ−1 (αλ )]dαλ
~ 2 0
T (α ) = {C [ Q | λ = L− 1 (α )],
(9) Let
C [ Q | λ = R −1 (α )]}, 0 ≤α <1
1 1 −1
where C (Q) is just defined as the model (1). Since
K 1 (αλ ) =
2 ∫0
[ Lλ (αλ ) + Rλ−1 (αλ )]dαλ (12)

3
 ~ h K (α )
, taking the partial derivative of T (α λ ) with t * = {( )⋅ 1 } ⋅ Q* (18.2)
h + p K 2 (α )
respect to Q and setting to zero, the necessary

~ ~
condition of optimal solution of T (α λ ) can be The sufficient conditions for the I (TB ) to attain

found as: ~
the minimum are I Q " (TB ) > 0 and
2a
Q* = ⋅ K 1 (α λ ) (13)
h ~ ~ ~
I Q " ( T B ) ⋅ I t " ( T B ) − [ I Q , t " ( T B )] 2 > 0
~
T * (α ) = {C [ Q * | λ = L− 1 (α )],
(14) , which can be derived respectively as follows
C [ Q * | λ = R −1 (α )]}, 0 ≤α <1
~
I Q "(TB ) = (4a / Q3 ) ⋅ K1 (α ) + [(h + p)t 2 / Q3 ] ⋅ K2 (α)
Now, if backorder for stockout is permitted,
~ ~ ~
from Equation (4), the model can be described by I Q " (TB ) ⋅ I t " (TB ) − [ I Q ,t " (TB )]2 = (2a / Q 3 ) ⋅
terms of α as K1 (αλ ) ⋅ [(h + p) / Q] ⋅ K 2 (α )
~ Because of K 1 (α ) > 0 and K 2 (α ) > 0 , the
T B (α ) = {C B [ Q , t | λ = L− 1 (α )],
(15) sufficient conditions are clearly hold from the above
C B [ Q , t | λ = R −1 (α )]}, 0 ≤α <1
two equations. Consequently, the optimal solutions,

~ (Q * , t * ) , then can be found from Equations (18.1)

The Yager’s index of T B (α ) then can be derived as:
and (18.2), and, from Equation (15), the optimal

1 1 anneal cost can be calculated as:

~
I (TB ) = ∫ C B [Q , t | λ = L−1 (α )]dα
2 0 ~
(16) TB* (α ) = {C B [Q * , t * | λ = L−1 (α )],
1 1 (19)
+ ∫ C B [Q , t | λ = R −1 (α )]dα C B [Q * , t * | λ = R −1 (α )]}, 0 ≤ α < 1
2 0
Let Note, as p → ∞ , for Equations (18.1) and

(18.2), it is clear that t → 0 and the former will

*
1 1
K 2 (α ) = ∫ {[ L−1 (α )]2 + [ R −1 (α )]2 }dα . (17)
2 0 degenerate to the optimal ordering quantities of EOQ

~ model with fuzzy demands i.e. Equation (13). These

, the necessary conditions for I (T B ) = to attain the
results are the same as the traditional EOQ model
~ ~ with backorder and crisp demands.
minimum are I Q′ (T B ) = 0 and I t′ (T B ) = 0 , which

can be calculated as follows.

4. The lost sales case
Q 2 = (1 / h) ⋅ [2a ⋅ K1 (α ) + (h + p)t 2 ⋅ K 2 (α ) ]
To consider the lost sales case for stockout, the
h K (α )
t = [( )⋅ 1 ]⋅Q EOQ model can be depicted as Figure 2. Let the
h + p K 2 (α )
decision variables be S and t, where the S is defined
To solve the above equations, we have:
as the sum of ordering quantities and the stockout
2a h K (α) 0.5 2
quantities, i.e. S = Q + s . Here, the decision
Q* ={[ ⋅ K1(α)] [1- ⋅ 1
]} (18.1)
h (h + p) K2 (α) variables S is a little bit like the S of the famous <s,
S> inventory model. The annual cost function,

4
C L ( S , t ) , thus can be derived as: ~
TL (α ) = [C L ( S , t ) | λ = L−1 (α ),
(23)
C L ( S , t ) = a λ / S + h ( S − tλ ) 2 / 2 S C L ( S , t | λ = R −1 (α ))], 0 ≤α <1
(20)
+ p ⋅ t 2 λ2 / 2 S
where C L ( S , t ) is defined as the model (20). As
Inventory
level the case of C L ( S , t ) , by replacing Q with S in
Equations (18.1) and (18.2), the global optimal

~
solutions of T L (α ) can be found as:

Q 2a h K 2 (α) 0.5
S * = {[ ⋅ K1 (α)] [1- ⋅ 1 ]} (24.1)
S h (h + p) K2 (α)
t

Time h K (α )
t' = (S − s) / λ s t * = {( )⋅ 1 }⋅ S * (24.2)
h + p K 2 (α )
t1 t2 t3

Figure 2. The EOQ model with lost sales Furthermore, from Equation (23), the optimal annual
cost then can be found as:
Notes, if we replace Q with S in Equations (4), the
~
Equation (20) will be same as Equation (4). Hence, T L* (α ) = {C L [ S * , t * | λ = L− 1 (α )],
(25)
substitute S for Q in Equations (6.1) and (6.2), the C L [ S * , t * | λ = R −1 (α )]}, 0 ≤ α < 1
optimal solutions of C L ( S , t ) then can be derived
as:
The above Equation will be equivalent to Equation

~* ~*
S * = (h + p) / p ⋅ 2aλ / h . (21.1) (15), i.e. T L (α ) = T B (α ) .

~
t * = ( S * / λ ) ⋅ h /(h + p) (21.2) Since the quantities of stockout, tλ , is fuzzy,
from Equation (22), the optimal ordering quantities,
Since S = Q + s and s = t λ , the optimal
~
Q * , will also be fuzzy and can be found as:
~ ~
ordering quantities then can be found as: Q * = S * − tλ (26)
Q = S − tλ
* *
(22) The above equation indicates that the ordering
From the above three Equations, it is easy to show quantities is not the same for each order. It depends
~
hat as p → ∞ , t → 0 and on the current tλ . This is different from the
*

backorder case in which the ordering quantities is

Q * = S * = 2aλ / h which is the optimal
fixed for each order. However, no matter how much
ordering quantities of EOQ model. Furthermore, the the stockout quantities of each inventory cycle, by
optimal annual cost of Equation (19) will be the same using S * which is a crisp number minus the current
as Equation (4), i.e. C L ( S * , t * ) = C B ( Q * , t * ) . stockout quantities, the ordering quantities still can be
Now, If λ is treated as a fuzzy number, the decided definitely.
anneal cost function then can be describe by terms of By the way, from Equations (24.1), (24.2) and
α as:
5
(26), it is clear that if p → ∞ then t → 0 and
*
also be a trapezoidal fuzzy number and with
parameters as:
Q * = S * = 2a ⋅ K1 (α ) / h , which is same as

Equation (13). In other words, the model degenerates ~ l Q* m Q*

C (Q * ) = [a * + h , a * + h ,
to the EOQ model with fuzzy demands. Q 2 Q 2
(30)
n Q* u Q*
a * +h , a * +h ]
Q 2 Q 2
5. Discussion

To show the characteristics of proposed models, Equation (30) implies that the shape of membership
~
a trapezoidal fuzzy demand is employed. Let λ be ~ ~
function of C (Q * ) is the same as the λ , but
the trapezoidal fuzzy demands with parameters:
~
λ = [l, m, n, u] (27) with a different scale. Accordingly, the spread of

It is easy to find that K 1 (α ) = (l + m + n + u ) / 4 . ~ ~

C (Q * ) will vary according to the spread of λ .
~
For the EOQ model with λ , from Equation (13),
we have Fuzzier demands only yield fuzzier annual costs.
Furthermore, a set of numerical data with a
2a 1
Q* = ⋅ [ (l + m + u + u )] symmetrical triangular demand,
h 4
~
~
If λ is a symmetrical fuzzy number then λ = [400, 800, 800, 1200] ,and several coefficients
(u − n ) = ( m − l ) , i.e. (u + l ) = ( m + n ) . Let of sortage cost, p, are employed to analyze the effect
of p. Table 1, firstly, shows the solution results of
~
λ 0 = ( m + n ) / 2 , the mean of λ , then we have backorder case, which indicates that increasing in p is

2a 1 2a associated with decreasing in (Q * , t * )

Q* = ⋅ [ ( m + u )] = λ0 . (28)
h 2 h
Equation (28) indicates Q * is the conventional simultaneously. In practice, these results imply that as

EOQ with crisp demands, ( m + n ) / 2 . This result p increases, the t will be shortened to save the

implies that no matter what the spreads of fuzzy shortage cost. Since t is a part of the time of ordering

demands, as long as the fuzzy demands are cycle, i.e. (Q / λ ) , a decreasing t will lead to a

~
symmetric with the same mean, the Q * will be the decreasing Q. Next, the optimal annual cost C B*

same and equal to the conventional EOQ with the which can be illustrated as Figure 3 implies that as p

mean of fuzzy demands. In addition, from Equation ~

gets decreasing, the membership function of C B*
(14), the fuzzy number of annual cost can be
calculated as: shifts to the left, i.e. decreases, and skews to the right

~ with a decreasing spread. On the contrary, as p gets

T * (α ) = {C [Q * | λ = l + α ( m − l )],
(29) ~
C [Q * | λ = u − α (u − n )]}, 0 ≤ α < 1 increasing, the membership function of C B* shifts

to right and gets close to be symmetrical with an

The above equation shows that the annual cost will increasing spread. When p → ∞ , the membership

6
 ~ and illustrated as Figure 4. From the view point of
function of C B* is just a symmetrical triangular,
~
Yager’s index, Q * increases according with p. In
~
which is the same as the C (Q * ) of EOQ model
~ ~
~ addition, for the optimal annual cost C L* , since C L*
with λ .
~
is same as the C B* of backorder case, the above
Table 1 The results of numerical examples for
backorder case with different p values ~
characteristics of C B* thus can also be held for
Case p ~
(Q * , t * ) C B*
~
C L* .
1 0.1 (545.108, 0.314) [24.917, 28.384, 37.658]

2 0.3 (456.070, 0.132) [27.526, 34.678, 44.258] µQ~

3 0.5 (434.846, 0.084) [28.368, 36.537, 46.251]

4 1.0 (417.919, 0.044) [29.118, 38.150, 47.991] 1.0

p = 0.1 0.3 0.5 1.0 ∞
5 ∞ (400.000, 0.000) [30.000, 40.000, 50.000]

µC~ B

1.0
p = 0.1 0.3 0.5 1.0 ∞
Q
150 200 250 300 350 400

~
Figure 4 The membership function of Q * with
different p values for the case of lost sale

6. Conclusion
The purpose of this paper is to study the EOQ
CB
25 30 35 40 45 50 model with stockout and fuzzy demands. Because
Figure 3 The membership function of optimal demands are fuzzy, the quantities of stockout, s,
anneal cost with different p values for usually treated as a decision variable (the order point)
the case of backorder. will be also fuzzy. This paper suggests to replace the
s with the time period of stockout, t, as one of the
For the lost sales case, since the optimal decisions variables. The other decision variable is Q
for backorder case and S, the sum of Q and s, for lost
solutions ( S * , t * ) are same as the (Q * , t * ) of
sales case.
backorder case, increasing in p will also be associated
The results of this study indicate that, for the
with decreasing in (S , t )
* *
simultaneously. EOQ model, no matter what the spreads of fuzzy
demands, as long as the fuzzy demands are
Furthermore, from Equation (32), the fuzzy number
symmetric with the same mean, the optimal ordering
~
of optimal ordering quantities, Q * , can be calculated quantities will be the same and equal to the

7
conventional EOQ with the mean of fuzzy demands. the fuzzy number of holding cost and shortage cost of
Fuzzier demands only yield fuzzier annual costs. If models, and be defined as:
stockout is permitted, as the shortage cost per unit
gets infinite, both of the backorder case and lost sales p(α p) = [ min . µ ~−p 1(α p), max . µ ~−p 1(α p)]
(34.1)
case will degenerate to the EOQ model with fuzzy = [L p −1(α p), R p −1(α p)], 0 ≤ α p ≤ 1
demands. This result is the same as the EOQ model
and
with stockout and crisp demands.
In practice, it is possible the shortage cost is h (α h ) = [min . µ h~−1 (α h ), max . µ h~−1 (α p )]
(34.1)
evaluated by the maximum quantities of shortage in = [ L h −1 (α h ), R h −1 (α h )], 0 ≤ α h ≤ 1
each inventory cycle. In such case, the annual cost for
Let
backorder case can be derived as:
1 1
2 ∫0
C B' (Q , t ) = a λ / Q + h (Q − tλ ) 2 / 2Q K2 (αλ) = {[Lλ−1 (αλ)]2 + [Rλ−1 (αλ)]2 }dαλ (35)
(31)
+ p ⋅ tλ 2 / Q
1 1 1
K3 (αλ,αp) = {∫ Lp −1 (αp)dαp ⋅ ∫ [Lλ−1 (αλ)]2 dαλ
4 0 0
(36)
1 1
With fuzzy demands, the annual cost function then + ∫ Rp −1 (αp)dαp[∫ [Rλ−1 (αλ)]2 dαλ
0 0
can be described by the terms of α as:
1 1 −1
2 ∫0
~
TB' (α ) = [C B' (Q , t ) | λ = L−1 (α ), K 4 (αh) = [ Lh (αh) + Rh −1 (αh)]dαh (35)
(32)
C B' (Q , t | λ = R −1 (α ))], 0 ≤α <1
1 1 1
~ K5 (αh,αλ) = [∫ Lh−1(αh)dαh ⋅ ∫ Lλ−1(αλ)dαλ
Unfortunately, the Yager’s index of I (T B' ) can not 4 0 0
(37)
1 1
be proven to be a convex function. However, we can + ∫ Rh (αh)dαh ⋅ ∫ Rλ (αλ)dαλ]
−1 −1
0 0
~
still show that I Q " (TB' ) > 0 . Hence, let 1 1 1
K6 (αh, αλ) = {∫ Lh −1 (αh)dαh ⋅ ∫ [Lλ−1 (αλ)]2 dαλ
4 0 0
(38)
~ 1 1
I Q ' (T B' ) = 0 , the conditional optimal solution + ∫ Rh (αh)dαh ⋅ ∫ [Rλ (αλ)] dαλ}
−1 −1 2
0 0

, then the Yager’s ranking index can be derived as:

(Q * | t ) then can be derived as:
Q ={ 2a ⋅ K1 (α) / h + (2 pt / h + t 2 ) ⋅ K2 (α)}0.5 ~ a Q
(33) I (T ) = ⋅ K1 (αλ ) + ⋅ K4 (αh) − t ⋅ K5 (αh, αλ )
Q 2 (39)
t 2
t2
+ ⋅ K6 (αh, αλ ) + ⋅ K3 (αλ , αp)
By enumerating the t with a practical interval to 2Q 2Q
~
compare I (T B' ) of different (Q * | t ) , the optimal
, and the optimal solution can be found as:
~
solution (Q * , t * ) with a minimal I (T B' ) can be
1
found. By the same way, by replacing Q with S, the ⎧2aK (αλ) K 25 (αh, αλ) ⎫2 (40)
Q =⎨ 1
*
1− ⎬
optimal inventory policy of lost sales case can also be ⎩ K4 (αh) K4 (αh) ⋅ [K6 (αh, αλ) + K3 (αλ,αp)]⎭

found.
⎡ K 5 (αh , αλ ) ⎤ *
Furthermore, besides demands, the other t* = ⎢ ⎥ ⋅Q (41)
⎣ K 6 (αh , αλ ) + K 3 (αλ , αp ) ⎦
~
coefficients may also be fuzzy. Let h and ~
p be

8
 Reference 11. Park, K. S, “Fuzzy-set theoretic interpretation of
economic order quantity,” IEEE Trans. System,
1. Chang, H. C. “An application of fuzzy theory to Man, Cybernetics, SMC-17, 1082-1084(1987).
the EOQ model imperfect quality items,” 12. Petrovic D., R. Petrovic and M. Vujosevic,
Computers and Operations Research, 31, “Fuzzy model for the newsboy problem,”
2079-2092(2004). International Journal of Production Economics,
2. Chang, S. C. “Fuzzy production inventory for 45, 435-441(1996).
fuzzy product quantity with triangular fuzzy
number,” Fuzzy Sets and Systems, 107, 13. Roy, T. K., and M. Maiti. “A fuzzy EOQ model
37-57(1999). with demand-dependent unit cost under limited
3. Chang, S. C. and J. S. Yao, “Economic reorder storage capacity,” European Journal of
point for fuzzy backorder quantity,” European Operational Research, 99, 425-432(1997).
Journal of Operational Research, 109, 14. Tersine, R. J., “Principles of inventory and
183-202(1998). materials management,” 1988, North-Holland,
4. Chen, S. J. and C. L. Hwang, “Fuzzy multiple NY.
attribute decisions making methods and 15. Vujosevic M., “D. Petrovic and R. Petrovic, EOQ
applications,” 1992, Spring-Verlag, NY. formula when inventory cost is fuzzy,”
5. Ishii, H and T. Konno, “A stochastic inventory International Journal of Production Economics,
problem with shortage cost,” European Journal of 45, 499-504(1996).
Operational Research, 106, 90-94(1998). 16. Yager, R. R., “A procedure for ordering fuzzy
6. Kao. C. and W.K. Hsu, “A Single-period subsets of the unit interval, Information
inventory model with fuzzy demand,” Computer Sciences,” 24, 143-161(1981).
and Mathematic with Application, 17. Yao, J. S. and J. Chiang, “Inventory without
43,841-848(2002). backorder with fuzzy total cost and fuzzy storing
7. Kao. C. and W.K. Hsu, “Lot size-reorder point costn defuzzified by centroid amd sign distance,”
inventory model with fuzzy demand,” Computer European Journal of Operational Research, 148,
and Mathematic with Application, 401-409(2003).
43,1291-1302(2002). 18. Yao, J. S. and H. M. Lee, “Fuzzy inventory with
8. Lee, H. M. and J. S. Yao, “Economic production backorder for fuzzy order quantity,” Information
quantity for fuzzy demand and fuzzy production Sciences, 93, 283-319(1996).
quantity,” uropean Journal of Operational 19. Yao, J. S. and H. M. Lee, “Fuzzy inventory with
Research, 109, 203-211(1998). or without backorder for fuzzy order quantity
9. Lee, H. M. and J. S. Yao, “Economic order with trapezoid fuzzy number,” Fuzzy Sets and
quantity in fuzz sense for inventory without Systems, 105, 311-337(1999).
backorder model,” Fuzzy Sets and Systems, 105,
13-31(1999).
10. Liou, T. S. and M. J. Wang, “Ranking fuzzy
numbers with integral value,” Fuzzy Sets and
Systems, 50, 247-255(1992).

9
10