Professional Documents
Culture Documents
can be reused immediately, some defects need technical assistance along with a long span
of time to get it recycled and certain defects cannot be reworked and those items that are
unfit for reuse and recycle are subjected to waste disposal category. The defective items
are identified after screening the entire production/received lot. But again the defective
items are made to undergo screening to classify them as reusable, recyclable and residue.
Thus generation of waste takes place in production/inventory situation at the end of
second screening. Many inventory models were formulated to bring co ordination
between the inventory/ production situations along with remanufacture and waste
disposal but so far the disposal methods were not taken into account.
There are many waste disposal methods such as incineration, landfill,
decomposing and few others which are employed in MSW (Municipal solid waste). After
various analyses incineration seemed to be the suitable method as it is both economically
and environmentally good. Industry is one of the major causes of generation of municipal
waste which is the outcome of developmental activities. To minimize the risk of waste
management and to abide by the legislations, many industries have constructed smaller
incinerators
within
their
own
site
for
their
personal
use
(www.eea.europa.eu/publications). One of the attitudes of these waste management
techniques is the conversion of waste to energy. Waste is sometimes considered as a
renewable fuel[12]. Thus it is clear that benefits in terms of revenue could be earned out
of the installation of such constructions to which incineration is not an exception.[3]
Bernadette in his paper has mentioned the incineration cost parameters and revenue
gained out of it. He has also finally concluded that use of incineration facility to manage
a portion of waste was environmentally better than landfilling which ranks next
immediate to incineration in the waste hierarchy suggested and used in waste policy
making which are again examined by LCA methodology[12].
The remainder of the paper is organized as follows, Section 2 presents the
literature review of the previous works cited. Section 3 describes about incineration
briefly. Section 4 elucidates the model formation. Section 5 concludes the proposed work.
2. Literature Review
The Economic Order Quantity (EOQ) model has been extensively used in
inventory management for a long time. The economic order quantity (EOQ) model of
Harris in 1913 is the foundation of modern-day inventory models [17]. The economic
production quantity (EPQ) model is the oldest and most useful in production and
inventory management which is regarded as the generalized form of the Economic Order
Quantity [25].A common unrealistic assumption in using the EOQ is that all units
produced are of good quality[31]Zhang and Gerchak considered the inclusion of
inspection policy in a EOQ model where a random proportion of units are defective.
[43]Rosenblatt and Lee in their paper have formulated a model where the defective items
are reworked[31].As the concept of repair of items gained much significance it stimulated
many researchers to work on it. Schrady [33]developed an EOQ model with
instantaneous production and repair rates and no disposal. As environmental issues
started surfacing and becoming a public concern, Richter [27,28] picked up the problem
of Schrady [33] and turned it into a production, remanufacturing and waste disposal
problem. The industrial sectors must bear the cost of disposal in addition to the cost of
production and remanufacture.
The items that cannot be remanufactured are disposed. A number of disposal methods are
available, but the proper method has to be selected to minimize the cost of disposal and
the environmental effects for which LCA is preferred due to its systematic approach and
the frame work of waste hierarchy which is widely accepted by the decision makers in
industries[2].The international standard ISO 14040-43 defines LCA as a compilation and
evaluation of the inputs, outputs and the potential environmental impacts of a product
system throughout its life cycle [2].Life-cycle assessments were initially developed for
the purpose of analysing products, although recently, it has also been applied to the
treatment of waste. LCA studies the environmental aspects and potential impacts
throughout a products life. The structure of LCA is discussed briefly in [12]
Salameh in his paper has formulated a simple inventory model in which the
defective items which are expressed as the percentage (p) of lot size y with a known
probability density function are sold as a single batch at a discounted price after 100%
percent of screening and also the total profit per unit time is determined. In this paper this
model is modified by the inclusion of screening cost of the defective items, cost of
remanufacture and the cost of incineration. Also in addition, it is assumed that defectives
of a known proportion were present in produced / received lot. In this paper an integrated
inventory model which includes production, remanufacture and waste disposal method
incineration is formulated.
3. Incineration
Waste is considered as fuel [14] and Waste to energy has become the major goal to be
achieved in the implementation of any waste disposal activity. Amidst a number of waste
disposal options incineration is judged to be desirable after the analysis of social cost.
Incineration is the most important way of utilizing the energy content of the waste which
is the conclusion made by Bernadette [3] in his paper and also he has explained about the
incineration method briefly as follows. Incineration is a controlled process which
involves oxidative conversion of combustible solid material to harmless gases suitable for
atmospheric release. It converts a waste to a less bulky, less toxic, or less noxious
material. It also reduces sludge to an ash residue which can be more easily handled for
ultimate disposal. Incineration has its own environmental effects which have also been
order size
R1
I1
screening rate
d1
cycle length.
3.3
Assumptions
1.
2.
3.
The items to be reused, reworked and incinerated are expressed as the percentage
of imperfect items.
4.
The cost of mending the items that are categorized as reusable is neglible.
5.
6.
4.4
Mathematical Model.
Let TR(Y) and TC(Y) be the total revenue and the total cost per
cycle,respectively. TR(Y) is the sum total sales volume of good quality, reusable items,
reworked items and the revenue from the incineration of non- repairable units.
TR(Y) = sY(1-p) + vqpY + R1rpY + I1ipY
TC(Y) is the sum of procurement cost per cycle, screening cost per cycle, holding
cost per cycle,Screening cost of defective items per cycle, Remanufacturing cost per
cycle,cost of incineration per cycle.
TC(Y)
=
K+cY+dY+d1
pY+
E
+h*
The total profit per cycle is the total revenue per cycle less the total cost per cycle,
TP(Y) = TR(Y) TC(Y), and it is given as
TP(Y) = sY(1-p) + vqpY + R 1rpY + I1ipY { K+cY+dY+d1 pY+E+h*
The total profit per unit of time is given by dividing the total profit per cycle by
the cycle length, TPU(Y) =
TPU(Y)
D*
(s
vq-
+
+
D*
(1)
D*
+
D*
(s
vq+
TPU(Y)
(2)
(3)
= 31583 units.
Substituting
Conclusion
This paper presents an environmental oriented inventory model which discusses
about the benefits that are gained out of using incineration, the waste disposal method.
This paper also presents a numerical example to validate the above model also it discuss
about the waste disposal method in particular. This model also assists the decision makers
in industries to expel the waste more economically and also enables the industries in
fulfilling its environmental responsibilities.
References.
[1] AECOM. (2009). Management of Municipal Solid Waste in Metro
Vancouver A Comparative Analysis of Options for Management of Waste
after Recycling.
http://www.metrovancouver.org/services/solidwaste/planning/Thenextsteps/
SDD_3_AECOM_FULL_REPORT.pdf.
[2]Arena, U., Mastellone, M.L., Perugini, F., 2003. The environmental
performance of alternative solid waste management options: a life cycle
assessment study. Chemical Engineering Journal 96, 207222.
[3] Bernadette and Yuri.,The environment comparison of landfilling vs.
Incineration of MSW accounting for waste diversion., Waste Management
(2011)
[4] S-S. Chung, C-S. Poon, Accounting for the shortage of solid waste disposal
facilities in Southern China, Environ. Conserv. 28 (2) (2001) 99103.
[5] I. Dobos, K. Richter, A production/recycling model with stationary demand
and return rates, Central Eur. J. Oper. Res. 11 (1) (2003) 3546.
[6] I. Dobos, K. Richter, An extended production/recycling model with
stationary demand and return rates, Int. J. Prod. Econ. 90 (3) (2004) 311
323.
[7] I. Dobos, K. Richter, A production/recycling model with quality
consideration, Int. J. Prod. Econ. 104 (2) (2006) 571579.
[8] A.M.A. El Saadany, M.Y. Jaber, A production, repair and waste disposal
inventory model when returns are subject to quality and price
[20] M.Y. Jaber, A.M.A. El Saadany, The production, remanufacture and waste
disposal with lost sale, Int. J. Prod. Econ. 120 (1) (2009) 115124.
[21] I. Konstantaras, S. Papachristos, Lot-sizing for a single-product recovery
system with backordering, Int. J. Prod. Res. 44 (10) (2006) 20312045.
[22] I. Konstantaras, K. Skouri, M.Y. Jaber, Lot sizing for a recoverable product
with inspection and sorting, Comput. Ind. Eng. 58 (3) (2010) 452462.
.
[23] M.C. Mabini, L.M. Pintelon, L.F. Gelders, EOQ type formulations for
controlling repairable inventories, Int. J. Prod. Econ. 28 (1) (1992) 2133.
[24]Miranda, M.L., Hale, B., 1997. Waste not, want not: the private and social costs of
waste-to-energy production. Energy Policy 25,587 600.
[25]Morteza Rasti Barzoki, Morteza Jahanbazi, Mehdi Bijari.,Effects of imperfect
products on lot sizing with work in process inventory., Applied Mathematics and
Computation 217 (2011) 83288336.
[26] N. Nahmias, H. Rivera, A deterministic model for a repairable item
inventory system with a finite repair rate, Int. J. Prod. Res. 17 (3) (1979)
215221.
[27] K. Richter, The extended EOQ repair and waste disposal model, Int. J.
Prod. Econ. 45 (13) (1996) 443447.
[28] K. Richter, Pure and mixed strategies for the EOQ repair and waste
disposal problem, OR Spectrum 19 (2) (1997) 123129.
[29] D.S. Rogers, R.S. Tibben-Lembke, Going Backwards: Reverse Logistics
Trends and Practices, Reverse Logistics Executive Council Press,
Pittsburgh, PA,1999.
[30] A. Roy, K. Maity, S. Kar, M. Maiti, A productioninventory model with
remanufacturing for defective and usable items in fuzzy-environment,
Comput.Ind. Eng. 56 (1) (2009) 8796.
[31]M.K Salameh and M.Y.Jaber.,Economic production quantity model for
items with imperfect quality.,International Journal of Production Economics
64 (2000) 59-64
[35] R.H. Teunter, Economic order quantities for stochastic discounted cost
inventory systems with remanufacturing, Int. J. Log. 5 (2) (2002) 161
175.Model. 31 (1) (2007) 21492159.
[36]UNEP, 2003. Environmental management tools: life cycle
assessment,available from: /http://www.uneptie.org/pc/pc/tools/lca.htmS.
[37] US EPA. (2005). First-Order Kinetic Gas Generation Model Parameters for
WetLandfills. Contract EP-C-04-023, US Environmental Protection
Agency.Washington, DC.
[38]US EPA. (2006). Solid Waste Management and Greenhouse Gases. A Lifecycle Assessment of Emissions and Sinks, 3rd ed. US Environmental
Protection Agency. Washington, DC.
[39]US EPA. (1996). AP 42, Fifth Edition, Volume I. Chapter 2: Solid Waste
Disposal:Refusecombustion.<http://www.epa.gov/ttn/chief/ap42/ch02/final/c0
2s01.pdf>.
[40]US EPA. (2008). Background Information Document for Updating AP42
Section 2.4 for Estimating Emissions from Municipal Solid Waste Landfills.
<http://www.epa.gov/ttn/chief/ap42/ch02/draft/db02s04.pdf>.
[41]Wanichpongpan, W., Gheewala, S.H., 2007. Life cycle assessment as a
decision support tool for landfill gas-to energy projects. Journal of Cleaner
Production 15,18191826.
[42] D. Waters, Inventory Control and Management, Wiley, New York, 2003.
[43] Zhang, Y. Gerchak, Joint lot sizing and inspection policy in an EOQ model with
random yield, IIE Transactions 22(1) (1990) 41.
This paper addresses these two limitations. Mathematical models are developed with
numerical examples presented and results discussed.
Keywords : EOQ model, Production/recovery, Reuse, Waste disposal. Switching cost.
I. INTRODUCTION :
In competitive markets products are pushed faster and faster to customers along
supply chains resulting in faster generation of waste and depletion of natural resources.
These environmental issues that many governments around the world are confronted with
led their legislators to devise laws that require manufacturers to initiate product recovery
programs that collect used items of products from their customers once these products
reach their economic or useful lives. This gave rise to reverse logistics (RL) as a business
term (e.g., [l,2]). Carter and Ellarm (1998) have collected a number of definitions of
reverse logistics. I will cite one of the definitions. The more general definition is
Reverse logistic is such activity which helps to continue an environmental effective
policy of firms with reuse of necessary materials, remanufacturing, and with reduction of
amount of necessary materials." This efficiency touches the personal in production
supply and consumption process. Carter and Ellarm (1998) approach reverse logistics
from point of view of environmental protection. Environmental consciousness occurs at
three level of activity of firms : governmental regulation, social pressure and voluntary
self restriction. Reverse Logistic Executive Council (RLEC) has given a more general
definition of reverse logistics which summarizes the above definitions : Reverse logistics
is a movement of materials from a typical final consumption in an opposite direction in
order to regain value or to dispose of wastes. This reverse activity includes take back of
damaged products, renewal and enlargement of inventories through product take back
remanufacturing of packaging materials, reuse of containers, and renovation of products,
and handling of obsolete appliances. Reduction of use of raw materials, decrease of costs
of waste disposal, and value added through reuse.
Reverse logistics is an extension of logistic, which deals with handling and reuse
of reusable used products withdrawn from production and consumption process. Such a
reuse is eg recycling or repair of spare parts. An environmental conscious materials
management and / or logistics can be achieved with reuse. It has an advantage from
economic point of view, as reduction of environmental load through return of used items
in the manufacturing process, but the exploitation of natural resources can be decreased
with this reuse that saves the resources from extreme consumption for the future
generation.
Reverse logistics (RL) extends the traditional forward flow of raw materials,
components, finished products to account for activities such as reusing, recycling,
refurbishing or recovering of these products, components and raw materials. Rogers and
Tibben-Lembke (2001) defined RL as the process of planning, implementing, and
controlling the efficient and cost effective flow of raw materials, in-process inventory,
finished goods, and related information from the point of consumption to the point of
origin for the purpose of recapturing value or proper disposal. Fleischmann et al. (1997)
provided a survey addressing the logistics of industrial reuse of products and materials
from an Operational Research perspective. They subdivided RL into three main areas,
namely distribution planning, inventory control, and production planning. Inventory
management of returned items (including repaired/recovered items) will be the focus of
this paper. The importance of the repairable/recoverable inventory problem was
recognized back in the 1960s. Schrady (1967) and Sherbrooke (1968) reported that
repaired/recovered items accounted for more than 50% of the dollars invested in
inventory. Schrady (1967) was the first to address the inventory problem for repairable
(recovered) items. He developed an EOQ model for repaired items which assumes that
the manufacturing and recovery rates are instantaneous with no disposal cost. Nahmias
and Rivera (1979) extended
Schrady's model to allow for a finite repair rate with the assumption of limited
storage in the repair and production shops.
Along the same line of research, Richter (1996a, 1996b) assumed the production
and repair system to consist of two shops. The first shop manufactures new items of a
product and repairs used items of the same product collected by the second shop. The
used items are collected at a repair rate , 0 < 1. Richter (1996a, 1996b) also assumed
that not all the collected items are repairable, and therefore some are disposed as waste
outside the system at a rate = 1 - , 0 < 1, which may display the ecological
behavior of the system. This assumption is different from that adopted in Schrady (1967)
who assumed a continuous flow of used items to the manufacturer. Richter extended his
work cited above in several directions. He has done so either individually or in
collaboration with Dobos. Richter (1997) investigated the cost analysis developed in
Richter (1996a, 1996b) for extreme waste disposal rates. He showed that the pure (bangbang) policy of either no waste disposal (total repair) or no repair (total waste disposal)
dominates a mixed waste disposal and repair strategy. Richter and Dobos (1999) further
examined the bang-bang policy where they showed that the properties of the minimal
cost function and the optimal solution known for the continuous EOQ repair and waste
disposal problem (Richter, 1996a, 1996b, 1997) could be extended to the more realistic
integer problem. The characteristics of the cost function developed in Richter and Dobos
(1999) were further studied in Dobos and Richter (2000). They showed that the minimum
cost function is partly piecewise convex and partly piecewise concave function of the
waste disposal rate. In a follow-up paper, Dobos and Richter (2003) extended their earlier
work (Dobos & Richter, 2000) assuming finite repair and production rates with a single
repair cycle and a single production cycle per time interval. In a subsequent paper, Dobos
and Richter (2004) generalized the work they presented in Dobos and Richter (2003) by
assuming that a time interval to consist of multiple repair and production cycles. Steering
their earlier work into a new direction, Dobos and Richter (2006) investigated the
production-recycling model in Dobos and Richter (2004) for quality considerations.
In this work, they assumed that the quality of used collected (returned) items is not
always suitable for further recycling, i.e., not all used items can be reused. Dobos and
Richter (2006) showed that when considering quality, a mixed strategy is more
economical than the pure strategies of their earlier work. Apart from the above surveyed
works, other researchers have developed models along the same lines as Schrady and
Richter and Dobos, but with different assumptions. Some of the recent works, but not
limited to, are those of Teunter (2001, 2004), Inderfurth, Lindner, and Rachaniotis (2005),
Konstantaras and Papachristos (2006), and Jaber and Rosen (2008). The work of Richter
(1996a, 1996b) has limitations. Two of these limitations are addressed in this paper.
First, this paper modifies the method that Richter (1996a, 1996b) adopted for
calculating the holding costs. Richter assumed that collected items are transferred from
shop I to shop 2 in m batches to be repaired at the start of each time interval. Richter
considered a general time interval ignoring the effect of the first time interval. The first
time interval has no repair batches, as there is nothing manufactured before that is to
be collected and repaired in this interval. This assumption results in a residual inventory
and thus overestimates the holding cost. Second, this paper accounts for switching costs
(e.g., production loss, deterioration in quality, additional labor) when alternating between
production and recovery runs. When shifting from producing (performing) one product
(job) to another in the same facility, the facility may incur additional costs referred to as
switching costs (e.g., Glazebrook, 1980; Paul, Law, & Leong, 1980; Kella, 1991; Yan &
Zhang, 1997; Teunter & Flapper, 2003; Hajji, Gharibi, & Kenne, 2004; Song, Wang, &
Li, 2004; Khouja & Mehrez, 2005).
In this section, the present study develops EOQ repair and waste disposal
inventory model. The following notations and assumptions are used throughout this
paper.
Notation :
n
distance travelled
Assumptions :
1.
2.
3.
4.
5.
6.
No shortage.
7.
8.
3. MODEL FORMULATION
Ritcher (1996a, 1996b) introduced repair and waste disposal model in which
demand is satisfied by manufacturing new and repairing used items of a certain
product. There are n batches of newly manufactured items and m batches of repaired
items in some collection time interval T. Ritcher assumed instantaneous manufacturing
and repair rates and a considered repaired items to be as good as new. As a Ritcher
(1996a, 1996b), this paper assumes demand is supplied by n newly manufactured batches
and m repaired batches in some collection time interval of length T. Ritchers work, as
well as the other studies in the literature did not account for T, when there are no items to
be repaired, perhaps because all these studies assumed an infinite planning horizon.
Modelling the total cost expression for shops 1 and 2 and considering the cost to dispose
waste to the environment, the optimal total cost consists of the following elements.
Repair setup cost per batch = r. Therefore setup cost per m repair batches is mr.
Manufacturing setup cost per batch = s, setup cost for n manufacturing batches is ns.
Total setup cost = mr + ns
hQ
T
h Q T
2 n . n . n + 2 m . m . m
hQ
T
h Q T
2 n . n . n + 2 m . m . m
h Q 2 2 Q 2 2
+
2 dn
dm
Q
since T =
QQ 1 Q Q
(m 1)m
2 m dm
d 2
u Q 2
Q 2d
uQ 2
2d
=
(m 1)
2
1 - m (m 1)
D
V
The number 2 refers to a roundtrip waste produced by the inventory system per cycle is
Cw(Q) = 0 + Q( + )
where 0 is the fixed cost per waste disposal activity.
The total cost K(Q)
mr ns +
h Q2 2 Q2 2
+
2 dn
dm
uQ 2
+
1 (m 1)
2d
m
D
V
+ 0 + Q( + )]
The total cost per unit of time is given by dividing the above expression by T.
K(Q) =
1
h Q2 2 Q2 2
mr
ns
+
+
T
2 dn
dm
uQ 2
+
1 (m 1)
2d
m
2a bDQ bDQ 2
where T =
K(Q) =
D
0 Q( )
V
Q
d
d
d
h Q2 2 Q2 2
+
mr ns + .
Q
Q 2d dn
dm
uQ d2
+
1 (m 1)
2d Q
m
2ad
d
d
Dd
d
+ bDQ + bDQ 2
0 ( )
Q
Q
Q
VQ
Q
K(Q) =
2
u
d
Q 2
+ +
mr ns + h
Q
2 n
m
2
2ad
1
m
Q
+ bdD + bD 2
Dd
d
0 ( ) d
VQ
Q
. . . (1)
d2K
dQ 2
(ie)
for every Q > 0 and it has a unique minimum and derived by setting its first
dK
= 0.
dQ
d
h 2 2 u
mr
ns
+
+ +
Q2
2 n m
2
2ad
1 2 0 0
m
Q
2Dd
d
- 2 0 0 02
VQ
Q
1
Q2
0h
2Dd
d 2+
d mr ns + 2ad + VQ
1
+
2
2 2
u 1
n m
=0
m
. . . (2)
1
Q2
2Dd
d mr ns + 2ad + dV =
0h
2Dd
2 d mr ns + 2ad + d =Q 0 h
V
1
+
2
2 2
u 1
m
n m
2 2
u 1
m
n m
Q2 =
2d mr ns + 4ad +
2 2
+ u
n
m
Q=
2d mr ns + 4ad +
2 2
+ u
n m
4Dd
2 0d
V
1
m
4Dd
2 0d
V
1
m
. . . (3)
Now diff (2) w.r.t. Q,
d2K
dQ2
2d
4ad
4Dd
d
mr ns + 3 +
2 0 3
3
3
Q
Q
VQ
Q
=
1
Q3
2d mr ns +4ad +
4Dd
2 0d
V
. . . (4)
4Dd
2d mr ns + 4ad +
2 0d
V
2 2
h
+ u 1
m
n m
2d mr ns + 4ad +
dK
dQ2
+ u
n m
Q0
4Dd
2 0d
V
1
m
3/2
>0
K(Q) =
0
2
2
d
Q
u
2ad
mr
ns
+
h
+
+
1 0
0
2 n
m
2
m
Q
Q
+ bdD + bD 2
D d
d
0 0 ( ) d
0
VQ
Q
=
1
Q0
0
2 u 2
Dd
Q
d
mr+ns
+2ad+2
d
+
+
h
0
1 + +bdD+bdD
V
2 n m 2
K(Q) = l1 + l2 + l3
1
Q0
where l1 =
l2 =
. . . (5)
Dd
d
mr
+
ns
+
2ad
+
2
0 d
0
2
2
Q
+ + u 1
h
2 n
m
l3 = bdD + bdD
Dd
0 d
V
4Dd
2d mr ns + 4ad +
2 0d
V
2 2
h
+ u 1
m
n m
d mr + ns + 2ad + 2
l1
2
2
+ u 1
h
m
m
1/2
Dd
d mr + nr + 2ad + 2
0d
Dd
2 d mr + nr + 2ad + 2
0 d
V
1/2
1/2
2
2
h
+
u 1
m
m
2 n
Dd
d mr + nr + 2ad + 2
0d
1/2
1 2
Dd
h
+
0d
u 1 d mr + nr + 2ad + 2
2 n
m
m
V
l1
(6)
...
Now
Dd
2d mr + ns + 4ad + 4 V 20 d
l2
2 2
h
+ u 1
m
n m
1
=
+ u 1
h
m
2 n m
1/2
1/2
1 2
h
+ u 1
2 n m
m
1/2
Dd
d mr + ns + 2ad + 2
0d
V
1/2
1 2
Dd
h
+
0d
u 1 d mr + ns + 2ad + 2
2 n m
m
V
l2
. . . (7)
1 2
Dd
+ u 1 d mr + ns + 2ad + 2
0d
h
2 n m
m
V
K(Q) =
1 2
Dd
h
+
0d
u 1 d mr+ ns + 2ad +2
2 n m
m
V
+ bdD + bdD
1 2
Dd
h
+
+ 0d +bdD +bdD
u 1 d mr+ns + 2ad+2
2 n m
m
V
2 2
Dd
2 h
+ u 1 d mr+ns + 2ad+2
+ 0d
m
V
n m
1/2
+ bdD + bdD
2 2
Dd
2 h
+ u 1 d mr+ns + 2ad+2
+ 0d
m
V
n m
1/2
+ bdD + bdD
. . . (8)
CONCLUSION
This paper studies the environmentally responsible repair and waste disposal
inventory models. This model is very helpful to reduce the cost of remanufacturing
option provides inventory cost benefits for a wide range of system by giving the
manufacturer the opportunity to balance the workload between two alternative ways to
produce a product.
REFERENCES :
Brahimi, N., Dauzere-Peres, S., Najid, N. M., & Nordii, A. (2006). Single item lot sizing
problems. European Journal of Operational Research, 168(1), 1-16.
de Matta, R., & Miller, T. (2004). Production and inter-facility transportation
scheduling for a process industry. European Journal of Operational Research, 158(1), 7288.
Dobos, I., & Richter, K. (2000). The integer EOQ repair and waste disposal model Further analysis. Central European Journal of Operations Research, 8(2), 173-194.
Dobos, I., & Richter, K. (2003). A production/recycling model with stationary demand
and return rates. Central European Journal of Operations Research, 11(1), 35~6.
Dobos, I., & Richter, K. (2004). An extended production/recycling model with stationary
demand and return rates. International Journal of Production Economics, 90(3), 311-323.
Dobos, I., & Richter, K. (2006). A production/recycling model with quality consideration.
International Journal of Production Economics, 104(2), 571-579.
Fleischmann, M., Bloemhof-Ruwaard, J. M., Dekker, R., Van Der Laan, E., Van Nunen,
J. A. E. E., & Van Wassenhove, L. N. (1997). Quantitative models for reverse logistics: A
review. European Journal of Operational Research, 103(1), 1-17.
Gascon, A., & Leachman, R. C. (1988). A dynamic programming solution to the
dynamic, multi-item, single-machine scheduling problem. Operations Research, 36(1),
50-56.
Galvin, T. M. (1987). Economic lot scheduling problem with sequence dependent setup
costs. Production and Inventory Management Journal, 28(1), 96-105.
Glazebrook, K. D. (1980). On stochastic scheduling with precedence relations and
switching costs. Journal of Applied Probability, 17(4), 1016-1024.
Hajji, A., Gharibi, A., & Kenne, J. P. (2004). Production and set-up control of a failureprone manufacturing system. International Journal of Production Research, 42(6), 11071130.
Kim, J. H., & Han, C. (2001). Short-term multiperiod optimal planning of utility systems
using heuristics and dynamic programming. Industrial & Engineering Chemistry
Research, 40(8), 1928-1938.
Kim, E., & Van Oyen, M. P. (1998). Dynamic scheduling to minimize lost sales subject to
set-up costs. Queuing Systems, 29(2-4), 193-229.
Konstantaras, I., & Papachristos, S. (2006). Lot-sizing for a single-product recovery
system with backordering. International Journal of Production Research, 44(10), 20312045.
Koulamas, C. (1993). Operation sequencing and machining economics. International
Journal of Production Research, 31(4), 957-975.
Kumar, R. L. (1995). An options view of investments in expansion-flexible
manufacturing systems. International Journal of Production Economics, 38(2-3), 281291.
Lahmar, M., Ergan, H., & Benjaafar, S. (2003). Resequencing and feature assignment on
an automated assembly line. IEEE Transactions on Robotics and Automation, 19(1), 89102.
Larsen, C. (2005). The economic production lot size model extended to include more
than one production rate. International Transactions in Operational Research, 12(3), 339353.
Liaee, M. M., & Emmons, H. (1997). Scheduling families of jobs with setup times.
International Journal of Production Economics, 51(3),
Teunter, R. H., & Flapper, S. D. P. (2003). Lot-sizing for a single-stage single-product
production system with rework of perishable production defectives. OR Spectrum, 25(1),
85-96.
Tsubone, H., & Horikawa, M. (1999). A comparison between machine flexibility and
routing flexibility. International Journal of Flexible Manufacturing Systems, 11(1), 83101.
1. INTRODUCTION
Urinary Incontinence (UI) is a common but neglected problem of women. Globally,
urinary incontinence affects the quality of life of at least one third of women. Also, UI affects
general well-being, self-esteem, and social functioning. Many women are too embarrassed to
talk about it and some believe it to be untreatable even in western countries. This problem is
more pronounced in India, where women usually do not seek treatment for their reproductive
health problems and do not vocalize their symptoms. There is a culture of silence and low
consultation rate among Indian women regarding such problems. Women in India have been
reported to have a high tolerance threshold for seeking treatment. The condition can be
effectively treated with significant improvement of clinical and quality of life parameters[5].
Previous research has indicated that many women with UI do not consult a doctor about
their condition [5,7,8]. Reasons for not consulting a doctor include: seeing UI as a minor
problem which can be self-managed[9], not regarding symptoms as abnormal or serious[10-12],
hoping the condition improves by itself[13], regarding incontinence as a normal part of
aging[14], being embarrassed to discuss it with a doctor[13], and lack of knowledge of possible
treatments[10,11,13].
Previous research has shown that lack of money/time, fear of surgery or hospital and
pain are usually the reasons given by women for non-consultation[2,3,6].
UI is a condition that poses considerable human and social complications, bringing
physical discomfort, economic burden, shame, and loss of self-confidence, all diminishing
quality of life. Appropriate management can reduce the suffering that accompanies UI. Many
studies have reported that conservative treatments (therapies that do not involve pharmacological
or surgical intervention) can be helpful in managing the condition (Wilson et al., 2002).
However, most women accept UI as an ailment connected with childbearing and age, and
believe symptoms should not be considered serious (Hagglund, Walkwers, Larsson, & Leppert,
2003). Although urinary symptoms are extremely common, with as many as one-third of the
population over 40 years of age experiencing a clinically significant disorder (Perry et al., 2000),
and the impact on quality of life can be substantial, relatively few women seek medical help
(Shaw, Brittain, Tansey, & Williams, 2008; Shaw, Das, Williams, Assassa, & McGrother, 2006).
Only one of four eventually seeks help from a physician (Addis, 2008).
In a population-based prevalence study conducted in Canada, Germany, Italy, Sweden,
and England, the vast majority (80%) who had UI for at least one year, as well as 49% who had
UI for three years, had been suffering from symptoms. Only 60% of females with UI have
consulted with a doctor, and only 27% have had continuous treatment (Milsom et al., 2001). In
many studies, it has been determined that females with UI seek help for its symptoms in very
low percentages: 14% (Hagglund, Walker-Engstrm, Larsson, & Leppert, 2001), 27.8% (Yu,
Wong, Chen, & Chic, 2003), and 38% (Kinchen et al., 2003).
This paper attempts to identify some important factors for not seeking a doctor for UI.
After analyzing the factors using FCM tool, the predominant factors are found. The resultant
factors are the more impactful factors for not consulting a doctor due to UI.
This paper is organized as follows: Section 2 presents FCMs and their foundations on
fuzzy logic. Section 3 introduces the FCM methodology developed. Section 4 presents the
important factors for the task. Section 5 gives the calculation of FCM model for the given data.
Section 6 discusses the results in a detailed way. Section 7 gives the limitation of the study in
brief. Finally, Section 8 outlines the conclusions.
2. FUZZY COGNITIVE MAPS (FCMs)
A Fuzzy Cognitive Map (FCM) is a graphical representation consisting of nodes
indicating the most relevant factors of a decisional environment and links between these nodes
representing the relationships between those factors [2]. FCM is a modeling methodology for
complex decision systems, which has originated from the combination of fuzzy logic and neural
networks. A FCM describes the behavior of a system in terms of concepts; each concept
representing an entity, a state variable, or a characteristic of the system [3].
Kosko(1986) introduces the concept of Fuzzy Cognitive Maps (FCMs) which are
weighted cognitive maps with fuzzy weights [6].
Fuzzy logic considers that everything is a matter of degree. The variables in a system are
not true or false in a 100%; they can be true or false to some extent under certain conditions. A
membership of a variable to certain universe. The mathematical models provided by fuzzy logic
allow the study in a systematic manner of the linguistic descriptions and explanations that human
observes provide. Fuzzy modeling transforms linguistic description into mathematical models [20
].
One characteristic of fuzzy logic that makes itself very interesting for its application to
FCM is that fuzzy logic allows a membership of more than one set of concepts and consequently,
sets of statements overlap and merge with one another.
3. FCM METHODOLOGY
3.1. Introduction to the methodology
The methodology developed uses four matrices to represent the results that the
methodology provides in each one of its stages. These are Initial Matrix of Success (IMS),
Fuzzified Matrix of Success (FZMS), Strength of Relationships Matrix of Success (SRMS) and
Final Matrix of Success (FMS) [1].
The methodology uses data gathered from interviews with people whose knowledge
and background enable them to identify and evaluate under solid criteria those factors that
according to their understanding. The data are then transformed into numerical vectors associated
with each one of the factors identified. Each component in a vector represents the value that the
corresponding factor has for each one of the individuals interviewed. This generates an Initial
Matrix of Success (IMS). Thresholds are applied to the IMS in order to adapt the information
contained to the real world and maintain the logical integrity of data.
Next, the numerical vectors associated with each factors are fuzzified and treated as
a fuzzy set. The fuzzification of the IMS produces the Fuzzified Matrix of Success (FZMS)
containing the grades of membership of each component of a vector to a fuzzy set. The factors
are treated as variables, sometimes also referred as concepts. The degree of similarity between
any two variables is evaluated to estimate the strength of the relationship between them and the
polarity of that relationship. The strength of the relationship is given by a fuzzy weight preceded
by a positive or negative sign indicating whether that relationship is direct or inverse respectively.
The result of these computations is a new matrix called Strength of Relationships
Matrix of Success (SRMS). At this point, an expert determines whether there is a relationship of
causality between each pair of variables analysed or not. The result is a Final Matrix of Success
(FMS) which contains only those numerical fuzzy components which represent relationships of
causality between the factors. The graphical representation of the Final Matrix of Success in the
form of a FCM produces the targeted FCM for mapping factors for Help-seeking behavior of
women with urinary incontinence. Once the graphical representation of the factors is produced,
an in depth analysis of the map should be undertaken in order to extract all information contained
into it.
3.2. Constructing the FCM
The procedure for creating the targeted FCM is shown in Fig.1.
Fuzzy Cognitive Map (FCM)
Construction Steps
Data acquisition
from domain
experts
Data fuzzification:
Thresholds &
grades of
memberships
Degree of
similarity between
concepts and
polarity evaluation
Causality
evaluation
Graphical
Representation
of the FMS
IMS
FZMS
SRMS
FMS
FCM
Figure.1
3.2.1. Initial Matrix of Success (IMS)
The IMS is an [n x m] matrix, where n is the number of factors identified, also
referred as concepts or variables, and m is the number of people interviewed to obtain the data.
Each element Oij of the matrix represents the importance that an individual j gives to a certain
factor(concept) i within a scale [0,100]. The elements O i1, Oi2, . . . ,Oim are the components of
the vector Vi associated with the factors belonging to row i of the matrix.
3.2.2. Fuzzified Matrix of Success (FZMS)
The numerical vectors Vi are transformed into fuzzy sets, where each element of the
fuzzy set represents the degree of membership of component O ij of vector Vi to the own vector
Vi. The numerical vectors are converted into fuzzy sets with values within the interval [0, 1] in
the following way.
FZMS = [Xij]nxm, where
i)
Xi = 1, for the maximum value in Vi
i.e., Max(Oiq)
(1)
ii)
Xi = 0, for the minimum value in Vi
i.e., Min(Oip)
iii)
(2)
For the other values in Vi ,
(3)
where Xi(Oij) is the degree of membership of the element O ij to the vector Vi.
Directly projecting values into the interval [0, 1] may result in assigning grades of
membership that do not reflect the real world and are unsupportable by common sense reasoning.
In those cases, the introduction of an upper and/or lower threshold value by the expert analysing
the data is required.
Therefore, if Vi is the numerical vector of m elements associated to concept i, and O ij,
with
j = 1, 2, . . . ,m, are the components of Vi , the upper and lower threshold values (u and
l respectively) are as follows:
.. (4)
. (5)
The remaining elements of the vector are projected into the interval [0, 1] proportionally.
All threshold values introduced during the process to generate the FZMS are further explained
and justified in the discussion of the case study included in this paper.
3.2.3.1.
3.2.3.2.
(6)
AD is the average distance between vectors V1 and V2,
. (7)
The closeness or similarity S between the two vectors is given by the equation
S = 1 - AD
.. (8)
S = 1 indicates perfect similarity and S = 0 indicates the maximum degree of dissimilarity.
If vectorsV1 and V2 are inversely related, then the method to calculate the similarity
between them is similar to the previous one with the exception that in this case the equation to
calculate the distance between the corresponding elements of the inversely related vectors V 1 and
V2 is
(9)
The remaining equations to calculate the average distance between the two vectors (AD)
(Eq. (7)) and their similarity (S) (Eq. (8)) are the same.
In this case S = 1 indicates perfect inverse similarity and S = 0 indicates perfect inverse
dissimilarity between the two vectors. However, when studying relations between numerical
vectors represented as fuzzy sets neither perfect similarity, nor perfect dissimilarity is expected.
The most likely representation of the relation between vectors is in terms of similarity to a certain
degree.
For each pair of vectors Vi and Vj, the method proposed calculates the similarity between
the two vectors twice, once based on direct relation and then based on inverse relation. Higher
degree of similarity will determine the polarity of the relationship between the factors i and the
factors j positive (direct) or negative (inverse) and the strength of that relationship, in
definitive the value Sij to be introduced in the SRMS.
3.2.4. Final Matrix of Success (FMS)
Once the SRMS matrix is completed, some of the data contained in it could be
misleading data. Not all factors represented in the matrix are related, and not always there is a
relationship of causality between them. An expert opinion is required to analyse the data and
convert the SRMS matrix into the FMS matrix, which contains only those numerical fuzzy
components representing relationships of causality between the factors. When analysing the data
in the SRMS matrix, two vectors can be related by coincidence. The vectors might display close
mathematical relation, while logically the two concepts/indicators might be totally unrelated.
These irrelevant relations could be easily identified by a domain expert.
3.2.5. Graphical representation of Fuzzy Cognitive Map (FCM)
The graphical representation of the Final Matrix of Success in the form of a FCM
produces the targeted FCM for mapping factors. In the final representation, each arrow joining
factors i and j has a signed weight xij. This value, representing the strength of the direct or
inverse relationship of causality between both factors, is that value contained in the Final Matrix
of Success (FMS) in the cell given by row i and column j.
4. IDENTIFIED FACTORS FOR THE TASK
Using the survey and the experts (Doctors) opinion, We have taken the following nine
concepts(factors) as {1 ,2,3,4,5,6,7,8,9}. The following factors are taken as the main nodes for
our studies:
The factors/components/concepts are:
1. Seeing UI as a minor problem which can be self-managed.
2. Not regarding symptoms as abnormal or serious.
3. Hoping the condition improve by itself.
4. Regarding incontinence as a normal part of aging.
5. Being embarrassed to discuss it with a doctor.
6. Lack of knowledge of possible treatments.
7. Fear of hospital visits and investigations.
8. Low consultation rate and the lag in consultation.
9. Financial problems
IMS =
90 60 40 60 50 20
60 100 60 40 50 40
90 100 40 30 70 30
70 100 30 70 70 80
100 40 100 100 100 40
50 80 100 100 70 50
50 30 80 80 100 100
40 40 50 60 100 50
40 80 50 30 70 100
Factors
100 70 70
60 70 90
70 70 80
100 60 70
100 100 90
80 80 100
30 100 100
30 40 60
100 100 100
9x9
.
Similarly for all the vectors, Xi values calculated.
Therefore, FZMS Matrix is obtained as
Individuals
FZMS =
1
0.67
0.83
1
0.5
0.5
0.33
0.33
0.67
1
1
1
0.33
1
0.17
0.33
1
0.33
0.67
0.33
0.17
1
1
1
0.5
0.5
0.67
0.33
0.17
0.83
1
1
1
0.67
0.17
0.83 0.83
0.83
1
0.83
1
0.67 0.83
1
1
1
1
1
1
0.33 0.67
1
1
0.5
0
1
0.5 0.33 0.67
0.83 0.17 0.83
0.83 1
1
1
0.33
1
0.83 0.5
1
1
1
0.17
1
0.5 0.17
0.83 1
1
Factors
Likewise all the values Sij have been calculated. The resultant SRMS matrix is as follows:
1
2
3
4
5
6
7
8
9
0.76
0.81
0.74
0.72
0.67
0.48
0.59
0.63
0.76 0.81
0.84
0.84
0.69 0.74
0.67 0.68
0.57 0.70
0.57 0.48
0.61 0.56
0.76 0.78
0.63
0.76
0.78
0.78
0.61
0.78
0.63
0.70
0.59
0.63 0.59
0.48
0.57
0.48
0.59
0.76
0.74
0.59
0.61
0.56
0.67
0.61
0.59
0.70
SRMS =
5.4 The Final Matrix of Success (FMS)
Once the SRMS Matrix is found, it has been sorted out according to domain expert with
the fact that only the factors are included which have causal relationship mathematically and
logically.
According to the Experts opinion (Doctor), The FMS Matrix is formed in the following
way:
1
1
2
0.76
3
0.81
FMS =
2
3
4
5
6
7
8
9
0.76
0.81
0.84
0.84
0.74
0.78
0.78
0.74
0.83 0.76
0.83
0.76
0.78
0.78
0.84
2
0.76
0.74
0.78
5
4
9
0.83
0.78
6
7
0.76
Fig.2. shows the strong influence that certain factors have over others, according to the
data supplied by the expert in the relevant field.
It should be mentioned here that there is no negative fuzzy value in the directed arc. It
shows that no factor influences negatively other factor. The reason is that the factors mentioned
in this papers are well analysed recommendations for the help-seeking behaviour of women with
urinary incontinence.
6.DISCUSSION
The results achieved are logically coherent and according to human sense, no
contradictions are observed. The analysis of the FCM not only identifies the factors but also
shows how far the causal relation exists between them clearly and gives overall concrete picture
before our eyes. The discussion is mainly based on the FCM graph.
The factor that Not regarding symptoms as abnormal or serious directly influences the
third factor that Hoping the condition improve by itself with the degree +0.84 . It means that
people not considering symptoms of urinary incontinence as a serious one may think that the
condition can be improved by itself.
The fifth factor Being embarrassed to discuss UI with a doctor greatly influences Lack
of knowledge of possible treatments with the degree +0.83. That is, a person with UI
embarrassed to discuss it with a doctor may not know the possible treatments of the urinary
incontinence and the way in which it must be treated effectively.
The aim of this paper is to identify high impactful factors for the help-seeking behavior
of the UI. This aim is fulfilled successfully here by just looking at the FCM graph. In the graph ,
the factors 2,3,and 5 have outgoing arcs with high arc weight namely above +0.80.
Also, the second, fourth and ninth factor influences the third factor directly with high
degree. So, a person not considering symptoms of UI as a serious one, a person considering UI as
a normal part of aging, and a person with a financial problems may hope the condition of urinary
incontinence improve by itself and they dont want to consult a doctor for their condition. Hence
the most impactful factor for the help-seeking behaviour for not to consult a doctor is that Hope
the condition improve by itself .
7. LIMITATIONS OF THIS STUDY
In the case of this study, when transforming data into fuzzy sets, a variation of 20% has
been considered. It has been understood that those degrees of importance between 80 and 100
mean that the interviewee considers the contribution of that factor very important for this project,
and that those degrees of importance between 0 and 20 mean that the interviewee considers the
contribution of that factor irrelevant for this study. In this point of the methodology application,
different experts analysing the data could have different criteria, all of them valid if accompanied
with a logical explanation and justification.
One of the main drawback of this methodology is that the SRMS matrix contains plenty
of misleading data. Not all indicators of success represented in the matrix are related, and much
less there is a relationship of causality between them. They may present a close mathematical
relation while being logically totally unrelated. To clean up the SRMS matrix and keep only those
causal relations in the FMS matrix, the opinion of an expert is required. The perceptions of the
expert could be not 100% accurate and could lead to erroneous results. In addition, different
experts may have different perceptions working with the same data, which will lead to different
Fuzzy Cognitive Maps and different conclusions. In most cases, the opinion of more than one
expert should be considered.
The solution described in this paper is only one of the several solutions available. A
further analysis of the data may lead to a different, more complex and perhaps more accurate
solution.
8. CONCLUSION
A fuzzy model, like FCM, represents a system in a form that corresponds closely to the
way humans perceive it. Therefore, the model is easily understandable, even by a nonprofessional audience and each parameter has a perceivable meaning.
Several researchers have attempted research on Help-seeking Behaviour among
women with urinary incontinence. However, they identified to assess the proportion of women
who consult their doctor about urinary incontinence, and explore factors associated with helpseeking in all countries.
In this study, the most impactful factor for not seeking a help of a doctor with urinary
incontinence and relation between the factors with corresponding degree were found. This
proposed study believes that it has been taken one more step forward in this perspective research.
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
C.D. Stylios, V.C. Georgopoulos, P.P. Groumpos, The Use of Fuzzy Cognitive Maps in
Modeling Systems, Proceeding of 5th IEEE Mediterranean Conference on Control and
Systems, Paphos, Cyprus, 2123 July 1997.
Xirogiannis, G., Stefanou, J., & Glykas, M. (2004). A fuzzy cognitive map approach to
support urban design. Expert Systems with Applications, 26, 257268.
Ashworth PD, Hagan MT: The meaning of incontinence: a qualitative study of nongeriatric urinary incontinence sufferers. Journal of Advanced Nursing 1993, 18(9):14151423.
Hannestad Y, Rortveit G, Sandvik H, Hunskaar S: A community based epidemiological
survey of female urinary incontinence. The Norwegian EPINCONT study. Journal of
Clinical Epidemiology 2000, 53:1150-1157.
Abstract
Inventory planning is essential for the sequential execution of the business
activities at all levels. It is also equally important for the smooth
maintenance of the customer relation. Lack of co-ordination between
inventory planning and implementation leads to two problems namely
overage and shortage. Surplus and deficit of inventory above and below the
maximum stock level results in overage and shortage respectively. The latter
is discussed by quite a lot of researchers where as the concept of overage has
not much focused. This paper addresses on several aspects of overage and
proposes a new approach which simultaneously restores overage to
maximum stock level and assists to gain new customers who are worthier
than money. A mathematical model is also developed and numerical
examples are provided to illustrate the solution procedure.
1 Introduction
Climatic barrier
Occurrences of calamity
Improper transport services
Disputes and Strikes
that transport systems on the whole perform worse under adverse and
extreme weather conditions. This is especially true in densely populated
regions, where one single event may lead to a chain of reactions that
influence large parts of the transport system. In current days the effects of
climate change are numerous in number which causes trouble in
transporting.
(ii). Occurrences of calamity
Nowadays the occurrences of calamities are quite common. Earthquakes,
floods, storm are some of the calamities that affect the normal life of the
people. Consider the situation where a calamity (flood) has affected a place,
this calamity will certainly disturb the transportation of goods between the
supplier and the customer. Though the supplier tries his level best to
transport, the goods will not reach its destination at right time due to some
reasons as follows.
1. The transport network would have ruined. 2. The transporter might have
been forced to take a route longer than the usual one. 3. Either the origin
location or the destination location is flooded. 4. The persons engaged in
charge of transportation might have faced direct effect of flood. These are
some of the reasons discussed by Pablo.et.al.
(iii). Improper transport services
Presently the impact of calamities causes road damage, breakage of bridges
which highly interrupt the flow of transportation. Consider the happening of
landslides in hilly regions, felling of huge trees on the roads during storm
that widely causes degradation of the quality of the transport services. As it
chance of the products becoming obsolete and other is to extend the business
circle.
4. Mathematical Model - 4.1 Problem Description
Consider the situation where an enterprise maintains maximum stock level
to fulfill the demand of its customers but due to certain reasons as discussed
earlier the inventory level exceeds the maximum stock which leads to
overage. Let us consider the items that add up to overage in the n th cycle is
expressed as the percentage (p) of the inventory level in the (n-1) th cycle,
where p follows a known probability distribution. The overage must be
restored back to maximum stock level to regulate the steady flow of
inventory for which segregation of the overage items takes place. Overage
items are nothing but the percent of accumulated items in the (n-1) th cycle
which increases the maximum stock level to overage in the n th cycle. This is
represented in the fig 1.a. Geoff (2003) has given a wide explanation for the
figure. He has differentiated the classic inventory saw tooth plan to
maximum stock and inventory plan to including overage. He has mentioned
the word planned inventory level in both the cases which is the level of
inventory that is desired by the enterprise to be maintained, but due to many
external and internal barriers the enterprise cannot attain the planned
inventory level. (In the classic inventory it is equal to maximum stock where
as in the case of inventory with overage it is above the overage)The
occurrence of the causes of overage are uncertain ,so the enterprise does not
alter the replenishment rate with respect to the presence of overage items in
stock, as it will affect the entire system, because the quality of the overage
items is not reliable at all times. As the accumulation of overage items in one
cycle affects the next immediate cycle and also the consecutive cycles, the
items that increase the maximum stock to overage in each cycle are
segregated from the maximum stock and they are maintained as special
inventory (fig 1.b).This special inventory is utilized in acquiring new
customers who are worthier more than money. To determine the expected
profit of the enterprise which channelizes the overage to customer
acquisition a mathematical model is developed with the following notations
and assumptions.
Fig 1 a Maximum stock level to Overage.
Planned inventory
1.0
Overag
e
0.7
Max.Sto
ck
0.5
Min.Sto
ck
4.1 Assumptions.
0.2
Let us define TR(Q) and TC(Q) as the total revenue and the total cost per
cycle, respectively. TR(Q) is the sum of total sales volume of items in
maximum stock and overage items.
TR(Q) = s(1-p)Q + vpQ ( 1)
The total revenue per unit of time is obtained by multiplying the total
revenue by the number of cycles per year
.It is given by
triangle of Fig 1 is
, we obtain the
The total profit per unit time TPU(Q) is obtained by TRU(Q) - TCU(Q)
TPU(Q) = s(1-p)D + vpD -
} (1)
vE[p]D
......... (2)
The concavity of the expected total profit per unit time function, as an
optimality condition, is demonstrated by finding the first derivatives of Eq
(2)
+
. (3)
The second derivative of Eq (3)
. (4)
5.Numerical Example
To illustrate the usefulness of the model developed in section 4, let us
consider the inventory situation where a stock is replenished instantly with
Q units. The parameters needed for analyzing the above inventory situation
are given below:
Demand rate, D = 50,000 units/ year ,Ordering cost, K = 100/ cycle ,Holding
cost for maximum stock, h = 4/ unit/year ,Holding cost for special inventory,
h0 = 5/ unit / year , Selling price of items in maximum stock, s = 50/ unit ,
Selling price of items in overage, v = 30/ unit ,Cost of acquiring new
customers, A = 30/ customer
Number of new customers acquired, say n= 2 ,Renovating cost per item in
overage , r = 5 / item We assume that p, is uniformly distributed with its p.d.f
as
f(p) =
Also
E(1-p)
E(p2
Then the optimum value of Q that maximizes Eq.(4) is given from Eq.(3)
= 2000 units.
Substituting
given as
ETPU(Q)=
50
=2366987 / year
0.9
50,000
30
0.1
50,000-
percentage
of
items in Q that adds to overage in the next cycle) in the above problem we
get Q* = 310 and we get total profit per unit to be negative.
6.Sensitivity Analysis
The total costs per unit in the case where the overage is not taken into
consideration are only the ordering costs per unit and the holding costs per
unit, as the renovation costs , acquisition costs and holding costs of the
special inventory are excluded as p becomes zero. The optimal order
quantity is greater in the case where the overage is included (first case) than
the optimal order quantity obtained in the case where the overage is not
taken into consideration (second case). Also the total profit per unit in the
first case is positive where as in the second case it is negative). Therefore we
conclude that this inventory model with overage is applicable in the case
when the enterprises face the problem of overage. This model is uniquely
designed to deal the issues of overage. If the overage is not taken into
consideration then one can employ the classical method to determine the
optimal order quantity.
References
1. Bocking,
C.,
1996.
Inventory
management
at
Husky
Suarez,
William
Anderson,
Vijay
Mahal
T.R.
Abstract: Meningitis is inflammation of the lining around the brain and spinal cord. It is
usually caused by an infection. The infection occurs most often in children, teens, and
young adults. Also at risk are older adults and people who have long-term health
problems, such as a weakened immune system. It is a serious disease which can be lifethreatening may result in permanent complications if not diagnosed and treated early.
Early diagnosis and treatment is crucial for a positive outcome, yet identifying meningitis
is a complex process involving an array of symptoms. Based on these symptoms,
decision-makers are able to explore different courses of action. In recent years, the
number of potential scenario methods and applications of fuzzy cognitive map to assist in
the modeling are increasing. The proposed methodology aims to use the scenarios
assessment and rank the scenarios using fuzzy cognitive maps and multicriteria
techniques (TOPSIS).
Keywords: Fuzzy cognitive maps, TOPSIS, Scenarios, risk factors, meningitis
1. Introduction
A number of studies have shown that the diagnosis and treatment management of
meningitis is a complex and challenging problem for government and health care
agencies requiring novel approaches to its management and intervention [13, 14]. In this
paper, we are proposing a modeling approach to understanding meningitis which focuses
on capturing the various symptoms associated with the disease. The main scope of this
work is the construction of a knowledge based tool for modeling meningitis diagnosis for
adult. This paper proposes a TOPSIS based methodology for ranking FCM based
scenarios.
This paper is structured as follows: Section 2 presents the methodology. Section 3
gives the selected risk factors of meningitis. Section 4 explains the calculation of the
methodology for the given data. Section 5 discusses the experimental result. Finally,
Section 6 outlines the conclusion.
2. Proposed Methodology
Scenarios describe events and situations that would occurred in the future realworld. The whole methodology proposal is composed of three blocks [11].
1. Building FCM models using experts opinion.
2. Scenarios simulation. It is composed of two stages. The first one is the
scenarios definition and the second one is the FCM inference.
3. Ranking the scenarios with TOPSIS. The closer scenario to the positive-ideal
scenario is the best solution.
2.1. Fuzzy Cognitive maps
2.1.1. FCM Fundamentals
Cognitive maps (Axelrod, 1976) are a signed digraph designed to capture the
casual assertions of an expert with respect to a certain domain and then use them to
analyze the effects of alternatives. A fuzzy cognitive map (FCM) is a graphical
representation consisting of nodes indicating the most relevant factors of a decisional
environment and links between these nodes representing the relationships between those
factors (Kosko, 1986). FCM has two significant characteristics. The first one, casual
relationships between nodes have different intensities. These are represented by fuzzy
numbers. The second one, the system is dynamic, it evolves with time. It involves
feedback, where the effect of change in a concept node may affect other concept nodes,
which in turn can affect the node initiating the change.
After an inference process, the FCM reaches either one of two states following a
number of iterations. It settles down to a fixed pattern of node values, the so-called
hidden pattern or fixed-point attractor. Alternatively, it keeps cycling between several
fixed states, known as a limit cycle. Using a continuous transformation function, a third
possibility known as a chaotic attractor exists. This occurs when, instead of stabilizing,
the FCM continues to produce different results (known as state-vector values) for each
cycle. The relationships between nodes are represented by directed edges. An edge
linking two nodes models the causal influence of the causal variable on the effect variable
(e.g. the influence of the price to sales). Since FCMs are hybrid methods mixing fuzzy
logic (Bellman & Zadeh, 1970; Zadeh, 1965) and neural networks (Kosko, 1992), each
cause is measured by its intensity wij
one.
2.1.2. FCM dynamics
An adjacency matrix A represents the FCM nodes connectivity. FCMs measure
the intensity of the causal relation between two factors and if no causal relation exists it is
denoted by 0 in the adjacency matrix.
---------------------- (1)
FCMs are dynamical systems involving feedback, where the effect of change in a node
may affect other nodes, which in turn can affect the node initiating the change. The
analysis begins with the design of the initial vector state (X 0), which represents the value
of each variable or concept (node). The initial vector state with n nodes is denoted as
--------------------- (2)
where
instant, f(x) is the sigmoid function and A the adjacency matrix. The state is changing
along the process.
The sigmoid function is defined as
---------------------- (4)
where is the constant for function slope (degree of normalization). The value of = 5
provides a good degree of normalization (Bueno & Salmeron, 2009) in [0, 1].
The FCM inference process finish when the stability is reached. The final vector
state shows the effect of the change in the value of each node in the FCM. After the
inference process, the FCM reaches either one of two states following a number of
iterations. It settles down to a fixed pattern of node values, the so-called hidden pattern
or fixed-point attractor.
2.2 TOPSIS method
2.2.1. Concept
The technique for order performance by similarity to ideal solution (TOPSIS) is a
multicriteria method to detect the best alternative from a finite set of ones (Hwang &
Yoon, 1981). The chosen alternative should has the shortest distance from the positive
ideal solution and the farthest distance from the negative ideal solution. The positive ideal
solution is composed of all best values attainable from the criteria, whereas the negative
ideal solution consists of all worst values attainable from the criteria (Wang & Elhag,
2006).
criteria as
D and be defined as
-------------------- (5)
where D is composed of n alternatives (scenarios in this proposal) and m attributes
(nodes values in this proposal); xij denotes the value of the ith alternative with respect to
the jth criterion or attribute.
The procedure of TOPSIS technique can be expressed in the following stages.
Stage 1. Determine the normalized decision matrix (R = [rij]). The raw decision matrix is
normalized for criteria comparability. The normalized value of x ij, rij, can be
obtained by
,
j=1,2,,m,
i=1,2,,n
-------------------- (6)
Stage 2. Compute the weighted normalized decision matrix (V = [v ij]). The weighted
normalized value of rij will be denoted by vij and can be computed by
------------------- (7)
Note that wj is the weight of the jth criterion and
Stage3. Define the positive-ideal and negative-ideal alternatives. The values of the
criteria in the positive-ideal alternative correspond to best level. On the other
hand, the values of the criteria of the negative-ideal correspond to the worst
level.
Denote the positive-ideal alternative, A+, and the negative-ideal alternative, A-, as
=
-------------- (8)
And
=
--------------- (9)
where I+ and I- are the criteria sets of the benefit and cost type, respectively.
Stage 4. Compute the distance measures with the well-known Euclidean distance for mdimensional vectors. The separation of each alternative to the positive-ideal
alternative,
, is denoted as
i=1, 2 n
--------------- (10)
In addition, the distance to the negative-ideal alternative,
, is
denoted as
,
i=1, 2 n
--------------- (11)
Stage 5. Compute the relative closeness to the ideal alternative and rank the preference
order. The relative closeness of the ith alternative, ,
, is
defined as
,
i=1, 2 n
--------------- (12)
Since
and
, then
then larger
Now we illustrate the dynamical system by a very simple model from the
symptoms of meningitis for adults. At the first stage we have taken the following ten
arbitrary attributes (concepts) (C1, C2... C10). It is not a hard and fast rule we need to
consider only these ten attributes but one can increase or decrease the number of
attributes according to needs. The following attributes are taken as the main nodes for
study. An expert system spells out the ten major concepts relating to the meningitis. Of
the 10 concept nodes, 9 represent a list of the symptoms and risk factors considered by
the experts (physicians) and the central node Meningitis is the basic decision concept
which gathers the cause-effect interactions from all other input nodes. The selected nodes
for FCM are as follows:
C1
- Fever
C2
- Vomiting
C3
- Headache
C4
- Rash
C5
- Stiff neck
C6
C7
- Very sleepy
C8
- Confused/delirious
C9
- Seizures
C10
- Possibility of Meningitis
C1
C9
C2
C8
C3
C10
C7
C4
C6
C5
0
0
A
0
0
0
0
0 0 0 0 0 0 0
0
0.5
0 0 0 0 0 0 0
0 0.82
0 0 0 0 0 0 0
0
0.7
0 0 0 0 0 0 0
0
0.7
0 0 0 0 0 0 0
0 0.75
0 0 0 0 0 0 0
0 0.81
0 0 0 0 0 0 0
0
0
Furthermore, five initial stimuli have been defined as follows (Table 1). Each of initial
stimuli vector is used for generating FCM-based scenarios. Next stage is the FCM
dynamics. The results are shown in Table 2.
In addition, the final scenarios are represented graphically at Figure.2. Note
that the figure suggests the fifth scenario as the best one, but there is not more
information about the preference between the different scenarios.
After FCM dynamics, the next stage is to rank scenarios with TOPSIS. The
normalized decision matrix R is given in Table 3. The weighted normalized decision
matrix V is given in Table 4.
Table 1
Table 2
Initial Stimuli
N Initial
Scenarios(Si)
o stimul
d i (li)
e
s
C l l l l l
i
C 10111
1
C 10001
2
C 10001
3
C 01001
4
C 01001
5
C 00001
6
C 10101
7
C 01001
8
C 00001
9
C 00101
1
0
Nodes
Ci
S1
S2
S3
S4
S5
C1
C2
C3
C4
C5
C6
C7
C8
C9
C10
1.0
1.0
0.4
0.0148
0.0148
0.0148
1.0
0.1243
0.6013
1.0
0.117
0.117
0.117
1.0
1.0
0.117
0.117
0.0083
0.0197
1.0
1.0
0.008
0.4021
0.008
0.008
0.008
1.0
0.1265
0.6019
1.0
1.0
0.0096
0.4058
0.0174
0.0174
0.0096
0.0048
0.1270
0.6075
1.0
1.0
1.0
0.402
1.0
1.0
1.0
1.0
0.1240
0.6013
1.0
Table 3
0.4694
0.0668
R 0.5315
0.6261
0.3642
0.4694
0.0668
0.0043
0.0060
0.3642
0.1878
0.0668
0.2137
0.2541
0.1464
0.0069
0.5708
0.0043
0.0109
0.3642
0.0069
0.5708
0.0043
0.0109
0.3642
0.0069
0.0668
0.0043
0.0060
0.3642
0.4694
0.0668
0.5315
0.0030
0.3642
0.0584
0.0047
0.0672
0.0795
0.0452
0.2823
0.0112
0.3199
0.3804
0.2190
0.4694
0.5708
0.5315
0.6261
0.3642
0.2660
0.0379
V 0.3012
0.3548
0.2064
Table 5
TOPSIS results
i
Ci
Rank
S1
0.5429
0.5634
0.5093
3
S2
0.4772
0.5494
0.5352
2
S3
0.6822
0.5247
0.4348
4
S4
0.7708
0.4473
0.3672
5
S5
0.3826
0.5618
0.5949
1
5. Discussion
Finally the simulated scenarios are ranked as S5
S2
S1
S3
S4.
From this analysis, the 5th scenario, that is, when all the symptoms are present, possibility
of getting the meningitis is very high. The second rankings is for the second scenario, that
is, when the symptoms corresponding to rashes, stiff neck and confused states are present
then the possibility of getting meningitis disease is high. Likewise, we can conclude that
third ranking is for the first scenario that corresponds to the symptoms of fever, vomiting,
headache and very sleepy. If these symptoms are present, the risk of getting the disease is
moderate.
6. Conclusion
Scenarios describe the symptoms and situations that would occurred in the real
world. This approach is a supplication of a more complex reality, in which different
entities interact with each other. This study presents the results from research which
sought to model experts knowledge based on FCM decision support system with
TOPSIS among adults. More specifically, this work proposes the application of a decision
support tool based on the soft methodology of FCM with TOPSIS to diagnose meningitis.
References:
[1]
[2]
Bellman, R.E., & Zadeh, L.A. (1970), Decision making in a fuzzy environment,
Management Science,17,141-164.
[3]
Bueno, S., & Salmeron, J. L. (2008). Fuzzy modeling enterprise resource planning
tool selection. Computer Standards & Interfaces, 30(3), 137147.
[4]
[5]
[6]
Jetter, A. & Schweinfort, W., 2010. Building scenarios with fuzzy cognitive maps:
An exploratory study of solar energy, Futures, forthcoming,
<http://dx.doi.org/10.1016/j.futures.2010.05.002>.
[7]
[8]
Lee, K. C., Kim, J. S., Chung, H. N., & Kwon, S. J. (2002). Fuzzy cognitive map
approach to web-mining inference amplification. Expert Systems with
Applications, 22(3), 197211.
[9]
Mago et al. (2012). Supporting meningitis diagnosis amongst infants and children
through the use of fuzzy cognitive mapping. BMC Medical Informatics and
Decision Making, 12:98.
[11]
Salmeron, J.L., Rosario Vidal, & Angel Mena, (2012). Ranking fuzzy cognitive
map based scenarios with TOPSIS. Expert Systems with Applications, 39,24432450.
[12]
[13]
[14]
[15]
1. INTRODUCTION
Due to unreliable production system, vendors may not deliver
products to the buyers when needed, resulting in lost sales. However
excessive supplies to fulfill customers requirement results in higher
inventory cost. The inventory cost is one of the dominant costs for many
industries. Industries should plan their strategy to provide products and
services to the customers at a minimum cost. The order quantity and the time
to order are critical decisions for both the manufacturing and the service
industries. Some industries implement Just In Time (JIT) System to reduce
their inventory cost. In order to support an efficient JIT system, it is
important to ensure the reliability of the vendors production system.
But in reality, there are possibilities that the production process is
delayed due to machine unavailability and shortages of materials and
facilities. Abboud et al. (2000) developed EPQ models by considering
random machine unavailability with backorders and lost sales. the models
were extended by Jaber and Abboud (2001) who assumed learning and
forgetting in production. later Chung et al (2011) by considering
deteriorating items In this study, we assume a JIT system where the buyer
who pays the transportation cost, decides the order quantity size of items and
requests items delivery in multiple shipments. The vendor products the items
using an Economic Production Quantity (EPQ) model. Ideally, the machine
starts a production run when the inventory level is equal to zero. In some
periods, there is a possibility that the machine may not be available. If this
situation occurs, the vendor cannot deliver the predetermined quantity
ordered by the buyer, resulting in the buyers lost sales. We consider two
distribution models for the random machine unavailability case. The
2. METHODOLOGY
2.1. Graded Mean Integration Representation Method
Chen and Hsieh introduced Graded mean Integration Representation
Method based on the integral value of graded mean h-level of generalized
fuzzy number for defuzzifying generalized fuzzy number. Here, we fist
define generalized fuzzy number as follows :
Suppose
~
A
described as any fuzzy subset of the real line R, whose membership function
A~
1.
2.
A~ x
3.
4.
5.
6.
A~ x
A~ x
A~ x
A~ x
A%
A%
by Graded Mean Integration Representation Method, L-1 and R-1 are the
inverse functions of L and R respectively and the graded mean h-level value
of generalized fuzzy number
h -1
L (h) + R -1 (h) .
2
(see
fig.2).
A%
Then
the
graded
Mean
Integration
A%
Representation of P( ) with grade wA, where
A
2L
0
-1
(h) + R -1 (h) dh
A
A%
P( ) =
(1)
h dh
0
. . .
B%
B%
= (b1, b2, b3, b4). Then we can get the Graded Mean Integration
B%
Representation of by the formula (1) as
1
h
2 b +b h b - b
1
b 4 b3 dh
h dh
B%
P( ) =
b1 2b 2 2b3 b 4
6
. . .
(2)
1
~
A
F(x)
R(x)
WA
a1 L-1(h) a2
h -1
L (h) + R -1 (h) .
2
a2
A%
R-1(h) a4
%
A
%
B
%
A%
B
The addition of and is
~
~
A B
where a1, a2, a3, a4, b1, b2, b3 and b4 are any real numbers.
(2)
The multiplication of
~
A
%
B
and is
% B
%
A
= (C1, C2, C3, C4)
% B
%
A
= (a1b1, a2b2, a3b3, a4b4)
(3)
%
B
%
%
A
B
= (-b4, -b3, -b2, -b1) then the subtraction of and is
%
A
%
B
= {a1 - b4, a2 - b3, a3 - b2, a4 - b1} where a1, a2, a3, a4, b1, b2, b3
(4)
1
% B%-1
B
=
1
1
1
1
,
,
,
b 4 b 3 b 2 b1
numbers. If a1, a2, a3, a4, b1, b2, b3 and b4 are all nonzero positive real
%
%
A
B
numbers then the division and is
%
A
(5)
~
B
a1 a 2 a 3 a 4
,
,
,
b
b
b
b
4
3
2
1
Let R, then
(i)
0,
%
A
= (a1, a2, a3, a4)
(ii)
0,
%
A
= (a4, a3, a2, a1)
Si2
S , S , . . . , S .
2
1
. . and S2 =
2
2
2
m
0,
f(x) - g(x) = 0,
i g i (x) = 0, i = 1, 2, . . . , m.
g i (x) 0, i = 1, 2, . . . , m.
f(t) =
1
, 0 t b
b
0, Otherwise
The vendor and buyer total cost when the production down time is
bigger than the upper bound of the machine unavailability time is
TUCNL = TVUCNL + TBUCNL
TUCNL
TUCNL =
2
A v D h vq K K(1 - (D/p) + 1 D AD C t KD hq 2KD
qK
2DqK
qK
qK
2DqK
A v D h vq K(1 - (D/p) + 1
AD C t D hq
qK
2
qK
q
2
. . .
(1)
Taking the derivative of (1) with respect to q and set the value equal to
zero we have
2 A v C t K+A D
K h v K(1 - (D/p) + 1 h
q*NL =
% D,
% h%, P,
%h%
A%v , C%t , A,
v
are fuzzy
parameters.
This fuzzy total cost of the system is
D
h v1 q K(1 - 1 + 1
A v1 D1
p 4
2
qK
A1D1 C t1 D1 h1q
,
qK
q
2
TUCNL
D
h v2 q K(1 - 2 + 1
A v2 D 2
p3
qK
2
A 2 D 2 C t2 D 2 h 2q
,
qK
q
2
D
h v3 q K(1 - 3 + 1
A v3 D3
p 2
qK
2
D 4
h v4 q K(1 -
+ 1
A v4 D 4
p1
qK
2
A 3 D 3 C t 3 D 3 h 3q
,
qK
q
2
C t D 4 h 4 q
AD
4 4 4
qK
q
2
which implies
NL
P))
= ((Av
D)
K)) + ((hv
q)/2)
(K
(1 (D
((Ct
D)
q) +
1))
((A
(h
(q
D)
(q
K)
q)/2)
where
and
A%v
D%
h%
v
P%
%
C
t
h%
A%
are non negative trapezoidal fuzzy numbers. Then we solve the optimal
order quantity as the following steps. Second, we defuzzify the fuzzy total
inventory cost, using the Graded Mean Integration Representation Method.
The result is
D
h v1 q K(1 - 1 + 1
A v1 D1
P 4
qK
2
A1D1 C t1 D1 h1q
qK
q
2
P[
]=
NL
D 2
h
q
K(1
+
v
1
P
A v2 D 2 2
3
qK
2
A 2 D 2 C t 2 D 2 h 2q
qK
q
2
D
h v3 q K(1 - 3 + 1
A v D3
P2
3
qK
2
D 4
h v4 q K(1 -
+ 1
A v D 4
P1
4
qK
2
A 3D3 C t3 D3 h 3q
qK
q
2
C t D 4 h 4 q
AD
4 4 4
qK
q
2
q*NL ,
when P[
derivative of P[
%
P TUC
NL
q
] with q* is
NL
= 0.
] is
NL
] the
NL
A v D1 2A v D 2 2A v D3 A v D 4 A1D1 2A 2 D 2 2A 3D3 A 4 D 4
1
2
3
4
K h v1 K
C t1 KD1 2C t 2 KD 2 2C t 3 KD3 C t 4 KD 4 2
3
D
D
D
1 1 1 2h v2 K 1 2 1 2h v 3 K 1
P4
P3
P
2
D
h v4 K 1 4 1
P1
q%*
=
h1 2h 2 2h 3 h 4
q%
changing the crisp order quantity
q%
= (q1, q2, q3, q4) with 0 q1 q2 q3 q4. Then we get the fuzzy total cost
function as
D1
h v1 q1 K(1 - + 1
A v1 D1
P4
2
q 4K
NL
D 2
2h
q
K(1
+
1
v2 2
P
2A v2 D 2
3
qK
2
3
A1D1 C t1 D1 h1q1
,
q 4K
q4
2
2A 2 D 2 2C t 2 D 2 h 2
q2
,
q 3K
q3
2
D3
2h
q
K(1
+
v3 3
1
2A
D
P2
v3 3
qK
2
2
D4
h v4 q 4 K(1 -
+ 1
A v4 D4
P1
qK
2
1
2A 3D3 2C t3 D3 h 3q 3
,
q 2K
q2
2
A 4 D 4 C t 4 D4 h 4q 4
q1K
q1
2
Secondly we defuzzify the fuzzy total cost function using the Graded Mean
Integration Representation Method with 0 < q1 q2 q3 q4.
It will not change the meaning of formula (4.3) if we replace
inequality conditions with 0 q1 q2 q3 q4 into the following inequality
constraints q2 q1 0, q3 q2 0, q4 q3 0, q1 > 0.
Thirdly the Kuhn-Tucker condition is used to find the solutions of
q1, q2, q3, q4 to minimize P[
], subject to q2 q1 0, q3 q2 0,
NL
]) - (qi) = 0
NL
igi(qi) = 0
gi(qi) = 0
D1
h v1 q1 K(1 - + 1
1 A v1 D1
P4
6 q 4K
2
(TUCNL(qi)) - g(qi) = 0
A1D1 C t1 D1 h1q1
q 4K
q4
2
D
h v2 q 2 K(1 - 2 + 1
A v2 D 2
P3
2
q 3K
2
D3
h v3 q 3 K(1 -
+ 1
A v3 D3
P2
2
q 2K
2
D 4
h v4 q 4 K(1 -
+ 1
A v4 D 4
P1
q1K
2
A 2D2
q 3K
Ct2 D2
q3
C t 3 D3
A 3D 3
q 2K
q2
h 2
q2
2
h q
3 3
2
A 4 D 4 C t 4 D 4 h 4q 4
q1K
q1
2
D
h v1 K(1 - 1 + 1
A
D
P
1 h1
1
A
D
v
4
4
4 4 C t 4 D4 4=0
2
6 2
2
q1
K
K
D
h v2 K(1 - 2 + 1
P
A
D
2 h2
1
A
D
v
3
3 3 C t3 D3 1=0
2
6 2
2
q2 K
K
D
h v3 K(1 - 3 + 1
A
D
P
2 h3
1
A
D
v
2
2
2 2 C t 2 D2 2=0
2
6 2
2
q3
K
K
D
h v4 K(1 - 4 + 1
1 h4
1 A v1 D1 A1D1
P1
C t1 D1 =0
6 2
2
q4 K
K
1(q2 q1) = 0
2(q3 q2) = 0
3(q4 q3) = 0
4q1 = 0
q*NL
as
A v D1 2A v D 2 2A v D 3 A v D 4 A1D1 2A 2 D 2 2A 3D 3 A 4 D 4
1
2
3
4
K h v1 K
q*NL
C t1 KD1 2C t 2 KD 2 2C t 3 KD3 C t 4 KD 4 2
3
D
D
D
1 1 1 2h v2 K 1 2 1 2h v 3 K 1
P4
P3
P
2
D
h v4 K 1 4 1 h1 2h 2 2h 3 h 4
P1
4. NUMERICAL EXAMPLES
In this section, a numerical example is shown to illustrate the model.
Let the
Production rate, P = 19,200 units/year,
Demand Rate, D = 4,800 units/year,
Vendor setup cost, Av = $600/cycle,
Ordering cost of buyer, A = $25/order,
Vendor holding cost, hv =$6/unit/year,
Buyer holding cost, h = $7/unit/year,
Buyers transportation cost, Ct = $50/delivery,
Number of delivery, K = 6
The optimal solution is q = 192.354
TUCNL = 7694.15
Suppose Fuzzy production rate P is more or less than 19200
P%
D%
Fuzzy setup cost Av for vendor per cycle is more or less than 600
A%v
A%
Fuzzy holding cost hv for vendor per unit per year is more or less than 6
h%
v
= (4, 5, 7, 8)
Fuzzy holding cost h for buyer per unit per year is more or less than 7
h%
= (3, 5, 9, 11)
Fuzzy transportation cost Ct for buyer per delivery is more or less than 50
%
C
t
q*NL
The optimal total cost when the production down time is bigger than the
upper bound of the machine unavailability time is
%*
TUC
NL
CONCLUSION
REFERENCES
1.
2.
3.
with
machine
breakdowns.
Computers
&
Industrial
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
integrated
production-inventory
problem
with
process
22.
23.
24.
25.
26.
27.
28.
vendor's final
production batch to a single buyer under linearly decreasing demand.
29.
30.
USA.
Wang, S., Sarker, B.R 2006. Optimal models for multi-stage supply
chain system controlled by kanban under just-in-time philosophy.
European Journal of Operational Research 172,179-200.
Abstract
This paper considers a three-echelon joint model without
backorders in a fuzzy situation. Fuzziness is introduced by allowing the
traditional cost components to a certain extent. Trapezoidal fuzzy
numbers are used to represent the costs components. Extension of
Lagrangean method is used to defuzzify the trapezoidal fuzzy numbers.
The purpose of this paper is to find the optimal shipment quantity of
the total cost in the fuzzy sense. Numerical example is provided to
illustrate the model.
Keywords:
Fuzzy numbers, Lagrangean method, three-echelon inventory system,
joint model.
1. Introduction
Kreng and Chen [10] develop a two-integrated model (TIM) and
focus on the linkage of the two integrated models in the supply chain.
They explored that if all parties in a supply chain follow the joint
optimal shipment size, the total relevant cost within the supply chain
can be significantly reduced. This paper introduces specifically on the
decisions of economic order quantity made under a three echelon
supply chain among manufacturer, distribution centre and retailer. The
three echelon joint optimal strategy is proposed, where all parties are
ready to cooperate and agree upon a joint optimal shipment size.
When the stages of the supply chain are handled by multiple firms with
equal power, or they are the profit centers of the main office, the firms
may form alliances to better compete in the market. The manufacturer,
the distribution centre, and the retailer are about to cooperate, they
should agree upon a joint optimal shipment size.
[13] used trapezoidal fuzzy number to fuzzify the order cost, inventory
cost and backorder cost in the total cost of the inventory model
without backorder. Then, they found the estimate of the total cost in
the fuzzy sense by functional principle. Mirko Vujosevic et al. [11] used
trapezoidal fuzzy number to fuzzify the order cost in the total cost of
the inventory model without backorder. Then, they got fuzzy total cost.
They obtained the estimate of the total cost through centroid to
defuzzify. In [9], Yao and Lee fuzzified the order quantity q as the
trapezoidal fuzzy number for the inventory without backorder and
obtained the fuzzy total cost.
In the crisp inventory models, all the parameters in the total cost
are known and have definite values without ambiguity, as well as the
real variable of the total cost is positive. But in the reality, it is not so
sure. Hence it is essential to consider the fuzzy inventory models. In
order to simplify the calculation of trapezoidal fuzzy number, we use
Chens Function Principle [4] instead of Extension Principle to calculate
the fuzzy joint total cost of our planned model. Function Principle is
projected as the fuzzy arithmetical operations of fuzzy numbers in
1985.
Also the principle is making that it does not change the type of
membership function under fuzzy arithmetical operations of fuzzy
number. In the fuzzy sense, it is reasonable to discuss the grade of
each point of support set of fuzzy number for representing fuzzy
number. Therefore, Chen and Hsiehs Graded Mean Integration
Representation method [3] adopted grade as the important degree of
each point of support set of generalized fuzzy number. With this
reason, we use it to defuzzify the trapezoidal fuzzy joint total cost. In
fuzzy three-echelon joint model for crisp joint shipment quantity, the
first derivative of fuzzy joint total cost is used to solve the optimal joint
shipment quantity. Furthermore, the algorithm of Extension of the
Lagrangean method [15] is used to solve inequality constraints in fuzzy
joint model for joint shipment quantity.
1.1
symbols Used
: fuzzy stock holding cost per unit per year for the retailer
center
~
: fuzzy stock holding cost per unit per year for the distribution
center
~
Z
~
Z
1D
2D
retailer
~
A
~
Z
centre
~
h
n
: fuzzy stock holding cost per unit per year for the manufacturer
~
T RCj (n, m2) :joint total cost of the three echelon model
~
~
P{ T RCj (n,m2)}: Defuzzified value of fuzzy total cost P[ T RCj (n,m2)]
for three echelon model
1.2 Assumptions:
%
A
A%(x)
(1)
A%(x)
(2)
= 0, - < x a1,
A%(x)
(3)
A%(x)
(4)
= wA, a2 x a3,
A%(x)
(5)
A%(x)
(6)
numbers. Also
%
A
this type of generalized fuzzy number be denoted as = (a1, a2, a3,
a4; wA)LR.When
%
A
wA = 1,it can be simplified as = (a1, a2, a3, a4)LR.
%
A
A%(x)
x < a 1 or x > a 4
a1 x a 2
1
(x - a 4 )
(a 3 - a 4 )
a2 x a3
a3 x a 4
0
(x - a )
1
(a 2 - a1 )
A%(x)
a1
a2
a3
a4
=
2.3. The Function Principle
The function principle was familiarized by Chen [4] to treat fuzzy arithmetical
operations. This principle is used for the operation of addition, multiplication, subtraction
%
%
A
B
and division of fuzzy numbers.Suppose
= (a1, a2, a3, a4) and
= (b1, b2, b3, b4) are two
trapezoidal fuzzy numbers. Then
1.The addition of
b4),
%
A
%
%
A
B
and is
%
B
where a1, a2, a3, a4, b1, b2, b3 and b4 are any real numbers.
%
A
2.The multiplication of
{a1b1, a1b4, a4b1,
% B
% A
%
B
and
is
= (c1, c2, c3, c4),whereT =
a4b4}, T1 = {a2b2, a2b3, a3b2, a3b3}, c1 = min T,c2 = min T1, c3 = max
T1, c4 = max T.
If a1, a2, a3, a4, b1, b2, b3 and b4 are all non zero positive real
numbers, then
%
A% B
%
B
%
%
% A
A
B
= (-b4, -b3, -b2, -b1), then the subtraction of
and is
a3 - b2, a4 b1), where a1, a2, a3, a4, b1, b2, b3 and b4 are any real
numbers.
4.
1
%
B
%
B
1 1 1 1
,
,
,
b 4 b3 b 2 b1
= -1 =
numbers. If
a1, a2, a3, a4, b1, b2, b3 and b4 are all non-zero positive real numbers,
then the division of
a1 a 2 a 3 a 4
, , ,
%
% B
% A
% b 4 b3 b 2 b1
A
B
and
is =
5. Let k R then k
A%
(ka1 , ka 2 , ka 3 , ka 4 ), if k 0
%
A
If = (a1, a2, a3, a4; wA)LR is a generalized fuzzy number then the
%
P A
defuzzified value
method is given by
wA
%
P A
A%
If
h dh
0
by above formula is
a1 + a 4 + (a 2 - a1 a 4 + a 3 )h
dh
2
%
P A
%
A
h dh
0
%
P A
a1 + 2a 2 + 2a 3 + a 4
6
Step (1) :
Solve the unconstrained problem. Minimize y = f(x). If the
resulting optimum satisfies all the constraints, stop because all
constraints are redundant. Otherwise, set k = 1 and go to step 2.
Step (2) :
Activate any k constraints (ie. convert them into equality) and
optimize f(x) subject to the k active constraints by the Lagrangean
method. If the resulting solution is feasible with respect to the
remaining constraints, stop ; it is a local optimum. Otherwise, activate
another set of k-constraints and repeat the step. If all sets of active
constraints taken k at a time are considered without encountering a
feasible solution, go to step 3.
Step (3) :
If k = m, stop: no feasible solution exists. Otherwise, set k = k +
1 and go to step 2.
%
A
R
h%R
%
Z
R
%
R
%
A
D
h%D
Z%1D
%
D
P%
%
A
M
h%M
q%j
a d1
a r1
1 a m1
z m1 d1
+z1d1 +nz 2d1 d1
nzr1 d1
m2
m2
q j4 m 2
% n,m
TRC
j
2
n-1
h d1 + d1 m 2 h + nm
h m1 m 2
d1
1
2d1
r
r
2
n
1
q j1
+ 1
1
,
2
p
m
p
m
2
2n
4
2
4
2
a r2
a d2
1 a m2
z m2 d 2
+z1d2 +nz 2d2 d 2
nzr2 d 2
q j3 m 2
m2
m2
n-1
h d 2 +d2 m 2 h + nm
h m2 m 2
d2
1
2d 2
r2
r2
2
n
q j2
+
1
,
2
p 3 m 2 p 3m 2
2
2n
1
q j2
a m3
a d3
a r3
z m3 d 3
+z1d3 +nz 2d
d
nz
d
3
r
3
3
3
m2
m2
m2
n-1
h d3 +d3 m 2 h + nm
h m3 m 2
d3
2d 3
1
r3
r3
2
n
q j3
+
1
,
2
p 2 m 2 p 2 m 2
2
2n
a d4
a r4
1 a m4
z m4 d 4
+z1d 4 +nz 2d4 d 4
nzr4 d 4
q j1 m 2
m2
m2
n-1
h d 4 +d 4 m2 h + nm
h m4 m 2
d4
1
2d 4
r
r
2
n
4
q j4
+ 4
1
2
p
m
p
m
2
2n
1
2
1
2
1 1 a m1
z m1 d1
+z1d1 +nz 2d1 d1
6 q j4 m 2
m
2
nzr1 d1
m2
a r1
n-1
h d1 +d1 m 2 h + nm
h m1 m 2
d1
1
2d1
r
r
2
n
1
q j1
+ 1
1
2
p
m
p
m
2
2n
4
2
4
2
a d2
a r2
1 a m2
z m2 d 2
+z1d2 +nz 2d
d 2
nzr2 d 2
2
q j m2
m2
m2
h m2 m 2
q j2
2
1
qj
2
n-1
h d 2 +d 2 m2 h + nm
d2
1
2d 2
r2
r2
2
n
+
1
p 3 m 2 p3 m 2
2
2n
a m3
a d3
a r3
z m3 d 3
+z1d3 +nz 2d3 d 3
nzr3 d 3
m2
m2
m2
n-1
h d3 +d3 m 2 h + nm
d3
2d 3
1
r3
r3
2
n
+
1
p 2 m 2 p 2 m 2
2
2n
h m3 m 2
q j3
2
a d4
a r4
1 a m4
z m4 d 4
+z1d4 +nz 2d4 d 4
nzr4 d 4
q j1 m 2
m2
m2
h m4 m 2
q j4
2
n-1
h d 4 +d4 m2 h + nm
d4
1
2d 4
r4
r4
2
n
+
1
p1 m 2 p1m 2
2
2n
.. (1)
With 0 qj1
Differentiate (1) partially with respect to qj 1, qj2, qj3, qj4 and all the
partial derivatives equal to zero, and solve them. Those are
q j1 =
a d4
a r4
a m4
z m4 d 4
+z1d 4 +nz 2d4 d 4
nzr4 d 4
m2
m2
m2
n-1
h d1 +d1 m 2 h + nm
h m1 m 2
d1
1
2d1
r
r1
2
n
1
+ 1
2
p 4 m 2 p 4 m 2
2
2n
q j2 =
a d3
a r3
a m3
2
z m3 d 3
+z1d3 +nz 2d3 d 3
nzr3 d 3
m2
m2
m 2
n-1
h d 2 +d 2 m 2 h + nm
h m2 m 2
d2
1
2d 2
r2
r2
2
n
1
2
p3 m 2 p3 m 2
2
2n
q j3 =
q j4 =
a m 2
a d2
a r2
2
z m2 d 2
+z1d 2 +nz 2d
d 2
nzr2 d 2
2
m2
m2
m 2
n-1
h d3 +d3 m 2 h + nm
h m3 m 2
d3
2d 3
1
r
r3
2
n
+ 3
1
2
p 2 m 2 p2 m 2
2
2n
a d1
a r1
a m1
z m1 d1
+z1d1 +nz 2d1 d1
nzr1 d1
m2
m2
m2
n-1
h d4 +d 4 m2 h + nm
h m4 m 2
d4
1
2d 4
r
r4
2
n
1
+ 4
2
p1 m 2 p1m 2
2
2n
The above show that with qj1 > qj2 > qj3 > qj4, it does not satisfy
the constraint with 0 qj1 qj2 qj3 qj4, set k = 1 and go to step
2.
Convert the inequality constraint into equality constraint qj 2 - qj1
= 0 and optimize
a m3
a d3
a r3
2
z m3 d 3
+z1d3 +nz 2d3 d 3
nzr3 d 3
m2
m2
m 2
q j1 q j2
a m4
a d4
a r4
z m4 d 4
+z1d 4 +nz 2d4 d 4
nzr4 d 4
m2
m2
m 2
n-1
h d1 +d1 m 2 h + nm
h m1 m 2
d1
1
2d1
r1
r1
2
n
1
2
p 4 m 2 p 4 m 2
2
2n
n-1
h d2 +d 2 m 2 h + nm
h m2 m 2
d2
1
2d 2
r
r2
2
n
+ 2
+ 2
1
2
p
m
p
m
2
2n
3
2
3
2
q j3 =
a
a
a r2
m 2
,
n-1
h d3 + d3 m 2 h + nm
h m3 m 2
d3
1
2d 3
r
r3
2
n
1
+ 3
2
p 2 m 2 p 2 m 2
2
2n
and
q j4 =
a d1
a r1
a m1
z m1 d1
+z1d1 +nz 2d1 d1
nzr1 d1
m2
m2
m2
n-1
h d4 +d 4 m2 h + nm
h m4 m 2
d4
1
2d 4
2
n
1
+ r4 r4
2
p1 m 2 p1m 2
2
2n
The above show that qj3 > qj4, it does not satisfy the constraint 0 <
qj1 qj2 qj3 qj4. Therefore it is not a local optimum. Set k = 2 and
go to step 3.
Step 3 :
partial derivatives equal to zero and solve qj1, qj2, qj3 and qj4, then
we get
a m4
a
a r4
m2
a m3
a d3
a r3
2
z m3 d 3
+z1d3 +nz 2d3 d 3
nzr3 d 3
m2
m2
m 2
q j1 q j2 q j3
a m2
a d2
ar
+2
z m2 d 2
+z1d 2 +nz 2d2 d 2 2 nz r2 d 2
m2
m 2
m2
n-1
h d1 +d1 m2 h + nm
h m1 m 2
d1
1
2d1
r
r
2
n
1
+ 1
1
2
p
m
p
m
2
2n
4
2
4
2
n-1
h d 2 +d 2 m2 h + nm
h m2 m 2
d2
1
2d 2
r
r
2
n
2
+ 2
+ 2
1
2
p
m
p
m
2
2n
3
2
3
2
n-1
h d3 +d3 m2 h + nm
h m3 m 2
d3
2d 3
1
r
r
2
3
n
2
+ 3
1
2
p
m
p
m
2
2n
2
2
2
2
q j4
a d1
a r1
a m1
z m1 d1
+z1d1 +nz 2d1 d1
nzr1 d1
m2
m2
m2
n-1
h d 4 + d4 m 2 h + nm
h m4 m 2
d4
1
2d 4
r4
2
n
+ r4
1
2
p1 m 2 p1m 2
2
2n
and
The above result show that qj1 > qj4, it does not satisfy the
constraint 0 qj1 qj2 qj3 qj4, therefore it is not a local optimum.
Set k = 3 and go to step 4.
Step 4 :
a m1
a d1
a r1
z m1 d1
+z1d1 +nz 2d1 d1
nzr1 d1
m2
m2
m2
+2
2
q j1 q j2 q j3 q j4
a
a r2
a m3
a d3
a r3
z m3 d 3
+z1d3 +nz 2d3 d 3
nz r3 d 3
m2
m2
m2
a d4
a r4
a
m 4 z m4 d 4
+z1d 4 +nz 2d4 d 4
nzr4 d 4
m2
m2
m2
n-1
h d1 +d1 m2 h + nm
h m1 m 2
d1
1
2d1
r1
r1
2
n
1
2
p 4 m 2 p 4 m 2
2
2n
n-1
h d 2 +d2 m2 h + nm
h m2 m 2
d2
1
2d 2
r
r
2
n
2
+ 2
+ 2
2
p
m
p
m
2
2n
3
2
3
2
n-1
h d3 +d3 m2 h + nm
h m3 m 2
d3
2d 3
1
r
r
2
3
n
+2
+ 3
1
2
p
m
p
m
2
2n
2
2
2
2
n-1
h d 4 +d 4 m2 h + nm
h m4 m 2
d4
1
2d 4
r
r
2
n
4
+ 4
1
2
p1 m 2 p1m 2
2
2n
a m2
. . .
(2)
q%j
Because the above solution
q%j
inequality constraints, the procedure terminates with
as a local
optimum solution to the problem. Since the above local optimum
solution is the only one feasible solution of graded mean integration
formula, so it is an optimum solution of the inventory model with fuzzy
joint shipment quantity according to extension of the Lagrangean
method.
Let qj1 = qj2 = qj3 = qj4 = qj, then the optimal fuzzy joint shipment
*
*
*
*
*
q%
j = q j , q j , q j , q j , , where
quantity is
a m1
a d1
a r1
z m1 d1
+z1d1 +nz 2d1 d1
nzr1 d1
m2
m2
m 2
*
q%
j
a d2
a r2
a m2
+2
z m2 d 2
+z1d 2 +nz 2d2 d 2
nzr2 d 2
m2
m2
m 2
a d3
a r3
a m3
2
z m3 d 3
+z1d3 +nz 2d3 d 3
nz r3 d 3
m2
m2
m 2
a d4
a r4
a m4
z m4 d 4
+z1d 4 +nz 2d4 d 4
nzr4 d 4
m2
m2
m2
n-1
h d1 +d1 m2 h + nm
h m1 m 2
d1
1
2d1
2
n
1
+ r1 r1
2
p
m
p
m
2
2n
4
2
4
2
n-1
h d 2 +d2 m 2 h + nm
h m2 m 2
d2
1
2d 2
2
n
+ 2
+ r2 r2
1
2
p
m
p
m
2
2n
3
2
3
2
n-1
h d3 +d3 m2 h + nm
h m3 m 2
d3
2d 3
1
r
r
2
3
n
+2
+ 3
1
2
p
m
p
m
2
2n
2
2
2
2
n-1
h d 4 +d4 m 2 h + nm
h m4 m 2
d4
1
2d 4
r
r
2
n
4
+ 4
1
2
p1 m 2 p1m 2
2
2n
4. Numerical Example
FORD manufacturing company produces commercial car units in batch.
~
The firm estimated that the fuzzy annual demand quantity ( D ) of
~
the retailer is 1500 units, the fuzzy retailers ordering cost ( A R) per
~
contract is about $400, the fuzzy stock holding cost( h R) per unit per
~
year for the retailer is about $240, the fuzzy fixed cost ( Z
) of
~
D
~
= (1425, 1500, 1500, 1575); A
= (380,400,400,420);
~
h
~
~
h
~
Z
~
Z
~
A
= (285,300,300,315);
~
h
= (380,400,400,420);
~
Z
2D
~
P
~
Z
= (228,240,240,252);
1D
and m2 =4
Replace the above fuzzy parameter values into formula (2), we find the
optimal fuzzy
Shipment quantity
~
q
(102,102,102,102)
Conclusion
is the
References
[1] Apte, U. M., & Viswanathan, S. (2000). Effective cross docking for
improving
distribution efficiencies. International Journal of Logistics, 3, 91
302.
[5] S.H. Chen, C.C. Wang, A. Ramer, Backorder fuzzy inventory model
under
[8] F.W. Harris, How many parts to make at once, factory?, Mag
Manage. 10 (1913) 135136 (p. 152).
[9] Jing-Shing Yao and Huey-Ming Lee, Fuzzy Inventory with or without
Backorder for
Fuzzy Order Quantity with Trapezoid fuzzy number,Fuzzy sets and
Systems 105,(1999) 311-337
[14] E.W.Taft, The most economical production lot, The Iron Age101
(1918)1410 1412.[15]
[18] Victor B. Kreng a, Fang-Tzu Chen b (2011) The benefits of a threelevel coordinated
distribution policy in the value chain, expert systems with
applications , 38,835-842
[20] Xu, K., Dong, Y., & Evers, P. T. (2001). Towards better coordination
of the supply
chain. Transportation Research Part E, 37, 3554.
[21] Zimmer, K. (2002). Supply chain coordination with uncertain justin-time delivery.
International Journal of Production Economics, 77, 115.842.
Sreelekha menon.B**
Abstract
Machine repair is an important problem frequently encountered in production
and manufacturing operations such as semiconductor manufacturing and
maintenance operations. Due to uncontrollable factors, parameters in the
machine repair problem may be fuzzy. Uncertainty associated with the input
parameter are solved using Fuzzy set theory .This paper proposes a integer
non linear programming 0 - 1 approach to construct the membership
function of the performance measure of the machine repair problem with the
machine breakdown rate and the service rate being fuzzy numbers. Using
extension principle and
cuts , a pair of mathematical programs is
formulated to calculate the lower and upper bounds of the fuzzy performance
measures. By enumerating different values of , the membership function of
the system performance measure is constructed.
Key words : Fuzzy set theory, machine interference system, integer non
linear programming , extension principle , cuts.
1. Introduction
Machine repair models have wide applications in many practical situations,
such as production line systems, clientserver computing,maintenance
operations, and manufacturing systems Consider a manufacturing system
consisting of K machines subject to breakdowns from time to time. When a
machine breaks down, it is repaired by one of a crew of R repair persons, thus
this repair person cannot repair other broken machines for a period of time.
Thus, during this busy period of time, it is possible that there are broken
machines have to wait and are interfered with by the machine being
repaired. Many studies have been published on the machine interference
problem and its cost/profit analysis recent studies include , and so on.
Efficient methods have been developed for analysing the machine
interference problem when its parameters, such as the breakdown rate and
service rate, are known exactly. One commonly used type of solution
methods is the queuing theory approach in that the machine interference
problem is modeled as a finite calling population queuing system. The
machine breakdowns are treated as the customers and the repair persons
are servers in the system. On the basis of traditional queuing theory we can
derive the system performance measures (e.g., the expected number of
broken machines, the expected time a machine spends waiting for repair, and
the fraction of the time a particular repair
person is idle) of the machine interference problem and its variants when
their parameters are known exactly. However, there are cases that these
parameters may not be presented precisely due to uncontrollable factors.
Specifically, in many practical applications, the statistical data may be
obtained subjectively; e.g., the breakdown pattern and repair pattern are
more suitably described by linguistic terms such as fast, moderate, or slow,
rather than by probability distributions based on statistical theory. Imprecise
information of this kind will undermine the quality of decisions via
conventional machine interference models. To deal with imprecise
information in making decision, Bellman and Zadeh and Zadeh introduced
the concept of fuzziness.Today, fuzzy set theory is a well-known means for
modelling imprecision or uncertainty arising from mental phenomena.
Specifically, fuzzy queues have discussed by several researchers. Since the
machine interference problem with fuzzy parameters is more realistic than
the conventional machine interference problem, it deserves further
investigation. However, little research has been published on fuzzy machine
interference problems. Buckley investigated multiple-channel queuing S.-P.
Chen systems with finite or infinite waiting capacity and calling population, in
that the arrivals and departures are followed by some possibility distributions.
Recently, Buckley et al applied the previous results to the machine servicing
problem for determining the optimal size of crews. Clearly, when the machine
breakdown rate or service rate are fuzzy, the system performance measures
of the machine interference problem will be fuzzy as well. To conserve the
fuzziness of input information completely, the fuzzy performance measure
should be described by a membership function rather than a crisp value.
and
respectively.
x, ~ (x)
xX
y, ~ (y)
yY
where X and Yare the crisp universal sets of the breakdown rate and the
service
rate, respectively. Without loss of generality, in this model the breakdown
rate and the service rate are assumed to be fuzzy numbers, as crisp values
can
be represented degenerated membership functions that only one value in
their
domain. This model will hereafter be denoted by FM/FM/R/GD/K/K, where F
denotes fuzzy time and FM denotes fuzzified exponential time.
Clearly, when the breakdown rate and the service rate are fuzzy, the
system
performance measures in steady state are fuzzy as well; thus, they should be
described by membership functions to conserve the fuzziness of input
information
completely. If the obtained results are crisp values, then it may lose some
useful information.
~~ ~
f ( , , )
x L , x U
y L , y U
max
min
x X x ~ (x) , x X x ~ (x)
max
min
y Y y ~ (y) , y Y y ~ (y)
is on the basis of
(1)
(2)
These interval indicate where the batch arrival rate, service rate and vacation rate
lie at possibility level . Note that , , are crisp sets rather than fuzzy sets. By the
concept of a cuts, the imbedded fuzzy Markov Chain in FM/FM/R/GD/K/K the can be
decomposed into the family of ordinary Markov Chains with different transition
probability matrices, which are also parametrised by . Consequently, the
FM/FM/R/GD/K/K queue can be reduced to a family of crisp queue M/M/R/GD/K/K
0 1
0 1
queues with different level sets
and
. These sets cause
nested structures for expressing the relationship between ordinary sets and fuzzy sets and
they represent sets of shifting boundaries.
By the convexity of fuzzy number the boundaries of these interval are function of and
can be obtained as
x L min ~1
( )
x U
y L
y U
max ~1
( )
min 1 ( )
=
max 1 ( )
respectively.
~~ ~
f ( , , )
L (z ) Sup min
~~ ~
~ ~
Z f ( ,
) L ,
.( 3 )
~ (x) ~ (y)
L ( z)
is the minimum of
,
we need either
~ (x) ~ (y)
~ (x) ~ (y)
; or or
; ;
L (z )
such that to satisfy
=.
L ( z)
To find the membership function
, it will suffice to find the left shape function and
right shape function of
L (z)
upper boundary
and the
L ( z)
of the cuts of
.
L (z)
To find the membership function
it suffices to find the left shape function and right
shape function of
L (z )
bound
of the cuts of
L ( z)
and upper
x x L
~ (x)
x x U
this
Similarly,
2 y L (1 2 )y U
y =
x
by
y y L , y U
upper bound
L
= min
L
1
L
L1
of cuts of
L
, L 2
. In that we see
= max
U1
, L 2
respectively where
= min
=t1
y L
x L
+(1t1)
x U
y U
t1= 0 or
(4)
~~ ~
f ( , , )
= min
such that y =
x L
t2
L U1
~
L
~~ ~
f ( , , )
such that x
L L2
L ( z)
, 2 = 0 or 1
t2
~~ ~
f ( , , )
= max
y L
+(1t2)
y U
x U
or
(5)
and
such that x
=t3
y L
x L
+(1t3)
y
t3= 0 or
x U
y U
~~ ~
f ( , , )
U2
(6)
= max
such that y =
t4
x L
t4
y L
+(1t4)
y U
x U
or
..(7)
From the knowledge of calculus, a unique minimum and a unique maximum of the
L
objective models ( 4 ) ( 5 ) ( 6 ) ( 7 ) are assured which shows that the lower bound
U
~
L
L
and upper bound
of the cuts of can be found by solving these six models. In fact
these four models are MINLP with 0-1 variables. There are several effective and efficient
methods for solving these problems .Moreover, they involve the systematic study of how
x L x U y L y U
the optimal solutions change as
,
,
,
, vary over the interval [0,1]; they
fall into the category of parametric programming .
L
L
L1 L 2
Therefore, for two values 1 and 2 such that 0 < 2 < 1 < 1 we have
and
U
in other words
increasing with respect to .
L
If both
L1 L 2
is non-
and
LS (z) = L
U
RS (z) = L
can be obtained.
LS=0
(z)
(L)L
(L)L=1 z (L)U=1
(z)
RS=1
(L)U
(z)
L
z (L)L
=1
z (L)U
=0
= [ 6, 7, 9,10] per day. The shop manager wants to know the average
~ (z )
L
number of lathes operational at any time
. Clearly, this system can be
described by the FM/FM/2/GD/4/ 4 model, and the performance measures can
be constructed by using the proposed approach.
L(x, y) =
4 x
j
j y j!
2
4
4 x
j i j-R
2!2
j = 0 j y
j=3
j
4
4 x
j
j!
2
4 x
j = 3 j y
2 ! 2 j-R
j = 0 j y
j=3
j
x
j - 2 4 !
y
4
x
x
4
!
y
y 4 !
j=0
j = 3
j ! (4 - j) !
(4 - j)! 2 ! 2 j-2
2
Lq
L
x ( 4 L)
W =
Wq
Lq
x (4 L q )
x L x U
= [ 1, 2 , 3 , 4]
] = [6 + 10 -
L(x, y)
] = [ + 1 , 4 - ]
y L y U
x
x
x
x
4 + 12 + 18 + 12
y
y
y
y
2
3
x
x x
x
1 4 + 6 + 6 + 3
y
y
y y
= [ 6, 7, 9,10]
~
L
LL
= Min [
1 x 4
s.t
6 y 10
LU
= Max [
1 x 4
s.t
6 y 10
L
The shortest
LL
=
y=6+
LU
=
L
the optimal solution
L
and
~ (z )
L
membership function for
~ (z )
enumerate
different values of able presents the cuts at 11
distinct values 0,0.1,0.2,........1.0.The membership functions for the
expected queue length ,the expected system waiting time and the expected
queue waiting time can be derived similarly.
6(yx 3 + x 4 )
y 4 4xy 3 + 6x 3 y + 6x 2 y 2 + 3x 4
Lq =
Similarly , the lower bounds and upper bounds of
LLq
=
6( 1) 3 (10 ) 6( 1) 4
(10 ) 4 4( 1)(10 ) 3 6( 1) 3 (10 ) 6( 1) 2 (10 ) 2 3( 1) 4
6( 6)( 4 ) 3 6(4 ) 4
(6 ) 4 4(4 )(6 ) 3 6(4 ) 3 (6 ) 6( 4 ) 2 (6 ) 2 3(4 ) 4
U
q
WL
L
L
( 1)( 4 L )
=
WU
U
U
(4 )( 4 L )
=
Lq
WqL
L
L
( 1)( 4 Lq )
=
Lq
WqU
U
U
(4 )( 4 Lq )
=
Table lists the a-cuts of their values at 11 distinct a values: 0, 0.1, 0.2, . . . ,
1.0.
LL
LU
LLq
LUq
WL
WU
WqL
WqU
0.36772
1.800
0.004
5
0.340
4
0.010
1
0.204
5
0.0011
2
0.023
0.40538
1.700
0.005
5
0.318
0
0.102
5
0.0012
6
0.022
1
0.1
0.199
5
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0.44322
1.703
0.007
7
0.48101
1.653
0.009
8
0.275
9
0.105
1
0.190
3
0.0019
0
0.019
0
0.51986
1.603
0.012
0
0.256
0
0.106
6
0.185
7
0.0021
4
0.018
9
0.55855
1.550
0.015
0
0.236
9
0.108
1
0.180
7
0.0025
0
0.017
9
0.59754
1.504
0.018
3
0.218
6
0.109
7
0.177
2
0.0028
8
0.017
0
0.63671
1.454
0.021
9
0.201
1
0.111
3
0.173
0
0.0032
5
0.016
0
0.67632
1.405
0.026
0
0.184
5
0.113
0
0.169
1
0.0036
3
0.015
1
0.71622
1.356
0.030
6
0.168
6
0.114
7
0.165
4
0.0040
5
0.014
1
1.307
0.035
6
0.153
7
0.116
5
0.161
8
0.0044
8
0.013
3
0.75600
0.296
6
0.103
8
0.195
1
0.0016
1
0.019
6
Conclusions
Machine interference models play an important role in many real world
situations,
such as maintenance operations and operations/manufacturing systems. In
real world applications, parameters in the machine interference problem may
be fuzzy such that its system performance measures are fuzzy as well. This
paper proposes a mixed integer non linear mathematical programming
approach to find the membership function of the fuzzy performance measure
of the machine interference problem with breakdown rate and service rate
being fuzzy numbers. The idea is based on Zadehs extension principle. The
lower and upper bounds of the -cuts of the fuzzy performance measure are
calculated via a pair of mathematical programs parametrized by . From
different values of , the obtained lower and upper bounds of these a-cuts
are adopted to construct the corresponding membership function.
References
[1]
[2]
[3]
[4]
E.A.
Elsayed,
An
optimum
repair
policy
for
the
machine
D.
Gross,
J.F.
Ince,
The
machine
repair
problem
with
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15] L.A. Zadeh, Fuzzy sets as a basis for a theory of possibility, Fuzzy
Sets and Systems 1 (1978) 328.
[16] R.J. Li, E.S. Lee, Analysis of fuzzy queues, Computers and
Mathematics with Applications 17 (1989) 11431147.
[17]
[18]
D.S. Negi, E.S. Lee, Analysis and simulation of fuzzy queues, Fuzzy
Sets and Systems 46 (1992) 321330.
[21]
[22]
D.Usharani ,
College of Engineering , Kidangoor,Kerala
Abstract :
In todays rapidly evolving business dynamics, the study of the supply chain model is gaining
phenomenal importance around the globe. This work presents a linear programming approach to make
strategic resource planning decisions in multi-product and multi-time period two-echelon supply chains.
The objective of the proposed model is to provide an optimal inventory level for the warehouses and
distribution centers and also, minimizing the total cost related to transportation, inventory, packing and
toll fees of the entire supply chain for a finite planning horizon. An industrial case demonstrates the
feasibility of applying the proposed model to real world problem in a two-echelon supply chains. In
addition to that, if the efficiency of the transportation will be increased by using the light weight ecofriendly plastic pail rather than the traditional one, then the reduction of the total cost in the supply chain is
also analyzed.
Keywords : Supply chain management, linear programming, multi-product, multi-time period, two-echelon
supply chains
1. Introduction
Supply Chain Management (SCM) issues have long attracted interest from both practitioners and
academics because of its ability to reap more benefits by efficiently managing it. (Anamaria & Rakesh,
1999 ; Beamon, 1998 ; Bilgen & Ozkarahan, 2004, Chen, Lin & Huang, 2006, Erengiic, Simpson &
Vakharia, 1999 ; Gen & Syanf, 2005, Petrovic, 2001, Petrovic, Roy & Petrovic, 1998). Supply Chain
Management is a set of synchronized decisions and activities utilized to efficiently integrate suppliers,
manufacturers, warehouses, transporters, retailers and customers so that the right product or service is
distributed at the right quantities, to the right locations and at the right time, in order to minimize systemwide costs while satisfying customer service level requirements. The objective of SCM is to achieve
sustainable competitive advantage. To stay competitive, organizations should improve customer service,
reduction of costs across the supply chain and efficient use of resources available in the supply chain.
In this field, numerous researches are conducted. Williams, 1981, developed seven heuristic
algorithms to minimize distribution and production costs in supply chain. Cohen and Lee, 1989, presented
a deterministic, mixed integer, non-linear programming with economic order quantity technique to develop
global supply chain plan. Pyke and Cohen, 1993, developed a mathematical programming model by using
stochastic sub-models to design an integrated supply chain involving manufacturers, warehouses and
retailers. zdamar and Yazga, 1997, developed a distribution/production system involving a manufacturer
centers and its warehouses. They try to minimize total costs such as inventory; transportation costs etc
under production capacity and inventory equilibrium constraints. Petrovic et al., 1999, modeled supply
chain behaviors under fuzzy constraints. Their model showed that, uncertain customer demands and
deliveries play a big role about behaviors. Syarif et al., 2002, developed a new algorithm based genetic
algorithm to design a supply chain distribution network under capacity constraints for each echelon. Yan et
al., 2003, tried to contrive a network which involves suppliers, manufacturers, distribution centers and
customers via a mixed integer programming under logic and material requirements constraints. Yilmaz,
2004, handled a strategic planning problem for three echelon supply chain involving suppliers,
manufacturers and distribution centers to minimize transportation, distribution, production costs. Chen and
Lee, 2004, developed a multi-product, multi-stage, and multi-period scheduling model to deal with multiple
incommensurable goals for a multi-echelon supply chain network with uncertain market demands and
product prices. The uncertain market demands are modeled as a number of discrete scenarios with known
probabilities, and the fuzzy sets are used for describing the sellers and buyers incompatible preference on
product prices. The supply chain scheduling model is constructed as a mixed-integer nonlinear
programming problem to satisfy several conflict objectives, such as fair profit distribution among all
participants, safe inventory levels, maximum customer service levels, and robustness of decision to
uncertain product demands, therein the compromised preference levels on product prices from the sellers
and buyers point of view are simultaneously taken into account. Nagurney and Toyasaki, 2005, tried to
balance e-cycling in multi tiered supply chain process. Gen and Syarif, 2005, developed a hybrid genetic
algorithm for a multi period multi product supply chain network design. Paksoy, 2005, developed a mixed
integer linear programming to design a multi echelon supply chain network under material requirement
constraints. Lin et al., 2007, compared flexible supply chains and traditional supply chains with a hybrid
genetic algorithm and mentioned advantages of flexible ones. Wang, 2007, explained the imbalance
between echelons with pecans supply chain by changing chains perfect balanced. He used ant colony
technique to minimize costs in pecans imbalanced supply chains. Azaron et al., 2008, developed a multiobjective stochastic programming approach for supply chain design under uncertainty. Demands, supplies,
processing, transportation, shortage and capacity expansion costs are all considered as the uncertain
parameters. Their multi-objective model includes (i) the minimization of the sum of current investment
costs and the expected future processing, transportation, shortage and capacity expansion costs, (ii) the
minimization of the variance of the total cost and (iii) the minimization of the financial risk or the
probability of not meeting a certain budget. You and Grossmann, 2008, addressed the optimization of
supply chain design and planning under responsive criterion and economic criterion with the presence of
demand uncertainty. By using a probabilistic model for stock-out, the expected lead time is proposed as the
quantitative measure of supply chain responsiveness. Schtz et al., 2008, presented a supply chain design
problem modeled as a sequence of splitting and combining processes. They formulated the problem as a
two-stage stochastic program. The first-stage decisions are strategic location decisions, whereas the second
stage consists of operational decisions. The objective is to minimize the sum of investment costs and
expected costs of operating the supply chain. Tuzkaya and nt, 2009, developed a model to minimize
holding inventory and penalty cost for suppliers, warehouse and manufacturers based a holononic
approach. Sourirajan et al., 2009, considered a two-stage supply chain with a production facility that
replenishes a single product at retailers. The objective is to locate distribution centers in the network such
that the sum of facility location, pipeline inventory, and safety stock costs is minimized. They used genetic
algorithms to solve the model and compare their performance to that of a Lagrangian heuristic developed in
earlier work. Ahumada and Villalobos, 2009, reviewed the main contributions in the field of production and
distribution planning for agri-foods based on agricultural crops. Through their analysis of the current state
of the research, they diagnosed some of the future requirements for modeling the supply chain of agrifoods. Gunasekaran and Ngai, 2009, have developed a unified framework for modeling and analyzing
BTO-SCM and suggested some future research directions. Xu and Nozick, 2009, formulated a two-stage
stochastic program and a solution procedure to optimize supplier selection to hedge against disruptions.
Their model allows for the effective quantitative exploration of the trade-off between cost and risks to
support improved decision-making in global supply chain design. Shin et al., 2009, provided buying firms
with a useful sourcing policy decision tool to help them determine an optimum set of suppliers when a
number of sourcing alternatives exist. They proposed a probabilistic cost model in which suppliers quality
performance is measured by inconformity of the end product measurements and delivery performance is
estimated based on the suppliers expected delivery earliness and tardiness. Fahimnia, B.; Lee Luong;
Marian, R, 2009, developed a mixed integer formulation for a two-echelon supply network considering
the real-world variables and constraints. A multi-objective genetic algorithm (MOGA) is then designed for
the optimization of the developed mathematical model. P. Subramanian, N. Ramkumar, T.T. Narendran,
2010, formulated a generalised multi-echelon, single time-period, multi-product, closed loop supply chain
as an integer linear programme (ILP). K. Balaji Reddy et al.,2011,developed single-echelon supply chain
two stage distribution inventory optimization model.
Supply Chain Management is a field that is usually been studied from a market and product
perspective rather than from transport point of view. Transport process are essential parts of the supply
chain. They perform the flow of materials that connects an enterprise with its suppliers and with its
customers. The integrated view of transport, production and inventory holding process is the characteristic
of the modern supply chain management concept (B. Fleischmann, 2005). Only a good coordination
between each component would bring the benefits to a maximum.
Transportation occupies one-third of the amount in the logistics and transportation systems
influence the performance of logistics system hugely. Without well developed transportation systems,
logistics could not bring its advantages into full play.
Some authors have recently worked in the development of the supply chain and transport
relationship. (Stank and Goldsby, 2000, Potter and Lalwani, 2005 , Disney and Jowill, 2003 , Childhouse P.,
Towill D. R., 2003 , Bask A.H., 2001). However, Supply Chain Management and transport are areas that
should be discussed more in depth, in order to do so it is necessary to develop a framework that allow
holistic analysis from a system perspective.
Packaging is a co-ordinated system of preparing goods for safe, efficient, cost effective transport,
distribution, storage, retailing, consumption and recovery reuse or disposal combined with maximizing
consumer value, sales and hence profit proper packaging is required by all freight carriers to ensure safety
shipment. In supply chain packaging costs represents a significant part.
A toll is one of the fairest revenue sources. For the use of elevated road, user fee shall be collected
from all vehicles. David Jackson of Business Day, in an article published on 19 th March 2011, said that, the
new tolling system put strain on business. Logistics costs will increase by about 1% following the increased
toll road tariffs, companies may have to review their entire distribution network strategy as they seek viable
and cost efficient alternative to the toll road system.
According to our survey on the current literature, none of the previous models has considered the
major cost elements packing and toll fees. while taking this characteristic into consideration makes the
developed model adaptable to a wider manufacturing and distribution scenarios
This paper presents a linear programming approach to solve multi-product and multi-time period
two-echelon supply chain models.The objective of the proposed model is to provide an optimal inventory
level for the warehouses and distribution centers and also minimizing the total cost of the entire supply
chain for a finite planning horizon.
Also this study utilizes one of the leading paint company as a case study to demonstrate the
practicality of the proposed approach. In addition to that, if the efficiency of the transportation will be
increased by using the light weight eco-friendly plastic pail rather than the traditional one, then the
reduction of the total cost in the supply chain is also analyzed.
This paper is organized as follows : Section 2, describes the problem, details the assumptions and
notations. Section 3 formulates the multi-product and multi-time period two-echelon supply chain model.
Section 4 represents an industrial case for implementing the feasibility of applying the proposed approach
to real situations. Finally, section 5 presents the conclusion.
The proposed model is a two-echelon supply chain model with N w warehouses, ND distribution
centers and Nr retailers with limited capacities. In this model multi-product is being distributed from the
warehouse to distribution centers in the system. The demand for the product will be forecasted before
beginning of every period and will be used as the reference for warehouses to transfer stocks from them to
distribution centers in a particular period h. Further the multi-product is being distributed from the
distribution centers to retailers according to their demand for the product before beginning of every period
and will be used as the reference for distribution centers to transfer stocks from them to retailers in a
particular period h. This paper focuses on presenting a linear programming approach that optimizes a twoechelon supply chain models with multi-product and multi-time period .The aim of this approach is to
minimize the total cost associated with inventory holding cost, transportation cost, packing charge and toll
fees.
The following list of assumptions is the basics for this studys mathematical programming model.
1.
2.
Transportation costs on a given route are directly proportional to the units shipped.
3.
Packing charges vary considerably depending on the item and how much material it takes to get it
packed.
4.
5.
Capacity of the warehouse and distribution centers in each period cannot exceed their maximum
available levels.
Assumptions 1 to 4 indicate that linearity, proportionality properties must be technically satisfied
as a standard LP form. Assumption 5 represents the limit on the maximum available warehouse and
distribution centers capacities in a normal business operation.
This study uses the following notation.
Index Sets
i
Warehouses
i = 1, 2, . . . , Nw.
Distribution centers
Retailers
K = 1, 2, . . . , Nr.
Product Type
l = 1, 2, . . . , Npr.
Time period
Stages
j = 1, 2, . . . , ND.
h = 1, 2, . . . , H.
S = 1, 2.
Objective Functions
TCh1
Total cost in the first stage of period h
TCh2
Total cost in the second stage of period h
TC h
TCh1
Total cost in period h, is equal to sum of
TC h2
and
Decision Variables
IWhli
TWhlij
IDhlj
TDhljk
Parameters
ICW hli
unit inventory carrying cost of product l by warehouse i in period h.
TCW hlij
unit transportation cost of product l from warehouse i to distribution centers j in period h.
PChli
packing charge of product l by warehouse i in period h.
TFW hij
Toll fees from warehouse i to distribution centers j in period h.
NSW hij
Number of shipments form warehouse i to distribution centers j in period h is equal to
1 Npr
X TWhlij
10 l 1
1 Npr
X TWhlij
12 l 1
for traditional pail,
ICD hlj
unit inventory carrying cost of product l by distribution centers j in period h.
TCD hljk
unit transportation cost of product l from distribution centers j to retailer k in period h.
TFD hij
Toll fees from distribution centers j to retailer k in period h.
NSD hjk
Number of shipments form distribution centers j to retailer k
1 Npr
X TDhljk
10 l 1
in period h is equal to
1 Npr
X TDhljk
12 l 1
for traditional pail,
pail
FDlhj
Forecasted demand of product l of distribution centers j in period h.
ADlhj
Actual demand of products of distribution centers j in period h.
Dlhk
Demand of product l of retailer k in period h.
Chi
Chj
Chk
3. Problem Formulation
3.1. Multi-product and multi-time period two-echelon supply chain model
The proposed model deals with optimizing the inventory levels of the two-echelon multi-product
and multi-time period supply chain models. The diagrammatic representation of the proposed model for the
period h is shown in Fig.1.
Warehouse
Distribution centers
...
Nw
...
ND
Retailers
...
Nr
IW
hli
TCh1
x ICW hli +
i=1l=1
Nw N D N pr
TW
hlji
x TCW hlji +
i = 1 j = 1l = 1
N w N D N pr
TW
hlji
i = 1 j = 1l= 1
x PChli +
1 Nw ND
NSWhij x TFWhij
10 i = 1 j = 1
The sum of the units of product l transferred from a warehouse to all distribution centers should be
less than or equal to the warehouse inventory for a particular period h.
ND
TW
hlji
IWhli , i = 1, 2, . . . N w , l = 1, 2, . . . , N pr
j=1
. . . (2.2.2)
Npr
IW
hli
C hi ,
l=1
2.
. . . (2.2.3)
3.
The sum of the units of product l transferred from all warehouses to a particular distribution
centers should be greater than or equal to the forecasted demand of that particular distribution
centers in period h.
Nw
TW
FDlhj , l = 1, 2, . . . N pr , j = 1, 2, . . . , N D
hlji
l=1
. . . (2.2.4)
4.
The total distribution centers forecasted demand for a period h should be less than or equal to all
warehouse inventory of product l, in that particular period
Nw
IW
hli
l=1
ND
FD
lhj
, l = 1, 2, . . . N pr
j=1
. . . (2.2.5)
5.
The total number of units transferred from all warehouse to a particular distribution centers should
be less than or equal to that distribution centers capacity
N w N pr
TW
hlji
Chj , j = 1, 2, . . . N D
i=1l=1
. . . (2.2.6)
Second Stage Problem
The objective function for the second stage is given below :
N D N pr
ID
TCh2
hlj
x ICD hlj +
j = 1l = 1
N D N r N pr
TD
hljk
x TCD hljk +
j = 1k = 1l = 1
N D Nr
NSD
jk
x TFDjk
j = 1 k 1
=
. . . (2.2.7)
The inventory of product l at a particular distribution centers at the end of stage one is equal to the
total number units received from all warehouse by that distribution centers minus the actual
demand of that particular distribution centers in period h.
ID hlj =
Nw
TW
hlji
- AD lhj , l = 1, 2, . . . N pr
i=1
. . . (2.2.8)
(If this value is negative, IDhlj = 0)
2)
The sum of the units of product l transferred from all distribution centers to a particular retailer
should be greater than or equal to the demand of that particular retailer in period h.
ND
TD
FD hlk , l = 1, 2, . . . N pr , K = 1, 2, . . . N r
hljk
j=1
. . . (2.2.9)
3.
The sum of the units of product l transferred from a particular distribution centers to all retailers
should be less than or equal the actual demand of that particular distribution centers in period h.
Nr
hljk
ADhlj, j = 1, 2, . . . N D , l = 1, 2, . . . , N pr
K=1
. . . (2.2.10)
4.
The total number of units transferred from all distribution centers to a particular retailer should be
less than or equal to that retailer capacity
N D N pr
TD
hljk
C hk , K = 1, 2, . . . N r
j= 1l= 1
. . . (2.2.11)
4. IMPLEMENTATION AND COMPUTATIONAL ANALYSIS
4.1. Case Description
The company chosen for the application of the proposed methodology in this work is a leading
paint company located in the Southern part of India. This study focuses only on the emulsion paints which
is the fast moving paint product. Emulsion paints are coming in 3 different packs such as small, retailer,
bulk. The forecasted demand of these three products are different in three different time periods. The
company has planned to build a mathematical model to minimize the total cost of the supply chain. The
total cost can be minimized by optimizing the inventory levels at warehouse and distribution centers. Also
optimizing the transportation cost packing charge and toll fees in the supply chain. This can be done by
transporting the product with proper packaging through the optimum route and by maintaining an optimum
inventory level at warehouse and distribution centers. This study assumes the holding cost, transportation
cost, toll fees for warehouse and distribution centers and packing charge for warehouse are same for all the
three products in three different time periods. Usually a truck can carry maximum load of 10 tons per
shipment. But if 16% eco friendly light weight plastic pail is used rather than the traditional one, then a
truck can carry maximum load of 12 tons per shipment. So the number of shipments can be reduced. In
this study, the reduced cost in the supply chain due to this is also analysed. The following data is collected
for validating the above proposed model.
Number of warehouse Nw
Number of Retailers Nr
Number of Products
Number of Timeperiods
The input data for warehouse distribution centers and retailers are given in Tables 1 8.
Table 1 : Input Data for Warehouse
Inventory carrying cost per unit in Rs.
Rs.30/ton
650 ton
Small Pack
Rs.3/pack
Retail Pack
Rs.5/pack
Table 2 : Transportation Cost from Warehouse to Distribution Centers per unit in Rs.
Distribution Centers
D1
D2
1000
1400
Table 3 : Toll fees from Warehouse to Distribution Centers per shipment in Rs.
Distribution Centers
D1
D2
1770
2090
Product
Distribution Centers
D1
D2
h1
h2
h3
Forecasted
Actual
Forecasted
Actual
125
120
150
142
100
90
120
108
25
15
30
20
75
60
90
78
180
172
200
189
25
18
30
15
35
28
45
34
140
130
165
158
90
84
100
93
D2
20
25
Capacity in Units
300
400
Table 6 : Transportation cost from Distribution Centers to Retailers per unit in Rs.
Retailers
Distribution Centers
R1
R2
R3
R4
D1
150
234
468
540
D2
444
537
120
189
Distribution Centers
D1
Retailers
Nr1
Nr2
Nr3
Nr4
170
250
430
560
D2
510
595
170
340
Period
h1
h2
h3
Retailer
Product
R1
R2
R3
R4
68
52
78
64
52
38
60
48
10
12
30
30
47
31
100
72
113
76
10
10
11
17
23
11
59
71
102
56
39
45
60
33
R1
R2
R3
R4
150
140
190
130
4.2. Results
After solving the model for two stages, using Lindo Software, the optimal solutions for the case
study are obtained. These results are tabulated in tables 9 15.
Table 9 : Optimal Warehouse Stock
Period
Product
Number of Units
h1
IW111 = 275
IW121 = 220
h2
h3
IW131 = 55
IW211 = 165
IW221 = 380
IW231 = 55
IW311 = 80
IW321 = 305
IW331 = 190
Period
h1
h2
h3
Number of Units
Product
D1
D2
TW1111 = 125
TW1112 = 150
TW1211 = 100
TW1212 = 120
TW1311 = 25
TW1312 = 30
TW2111 = 75
TW2112 = 90
TW2211 = 180
TW2212 = 200
TW2311 = 25
TW2312 = 30
TW3111 = 35
TW3112 = 45
TW3211 = 140
TW3212 = 165
TW3311 = 90
TW3312 = 100
Distribution
Retailer
Centers
Perio
d
Produc
t
D1
h1
TD1111 = 68
TD1112 = 52
TD1113 = 0
TD1114 = 0
TD1211 = 52
TD1212 = 38
TD1213 = 0
TD1214 = 0
R1
R2
R3
R4
h2
h3
h1
D2
h2
h3
TD1311 = 10
TD1312 = 5
TD1313 = 0
TD1314 = 0
TD2111 = 30
TD2112 = 30
TD2113 = 0
TD2114 = 0
TD2211 = 100
TD2212 = 72
TD2213 = 0
TD2214 = 0
TD2311 = 10
TD2312 = 8
TD2313 = 0
TD2314 = 0
TD3111 = 11
TD3112 = 17
TD3113 = 0
TD3114 = 0
TD3211 = 59
TD3212 = 71
TD3213 = 0
TD3214 = 0
TD3311 = 39
TD3312 = 45
TD3313 = 0
TD3314 = 0
TD1121 = 0
TD1122 = 0
TD1123 = 78
TD1124 = 64
TD1221 = 0
TD1222 = 0
TD1223 = 60
TD1224 = 48
TD1321 = 0
TD1322 = 0
TD1323 = 12
TD1324 = 8
TD2121 = 0
TD2122 = 0
TD2123 = 47
TD2124 = 31
TD2221 = 0
TD2222 = 0
TD2223 = 113
T2224 = 76
TD2321 = 0
TD2322 = 0
TD2323 = 10
T2324 = 5
TD3121 = 0
TD3122 = 0
TD3123 = 23
TD3124 = 11
TD3221 = 0
TD3222 = 0
TD3223 = 102
TD3224 = 56
TD3321 = 0
TD3322 = 0
TD3323 = 60
TD3324 = 33
h1
h2
h3
Product
Distribution Centers
D1
D2
ID111 = 5
ID112 = 8
ID121 = 10
ID122 = 12
ID131 = 10
ID132 = 10
ID211 = 15
ID212 = 12
ID221 = 8
ID222 = 11
ID231 = 7
ID232 = 15
ID311 = 7
ID312 = 11
ID321 = 10
ID322 = 7
I331 = 6
I332 = 7
Type
Traditional
Light Weight
Distribution Centers
Period
D1
D2
h1
25
30
h2
28
32
h3
26.5
31
h1
20.75
24.90
h2
23.24
26.56
h3
21.99
25.73
Retailer
Type
Period
R1
R2
R3
R4
h1
13
9.5
h2
12
11
h3
10.9
13.3
h1
10.79
7.885
h2
9.96
9.13
h3
9.047
11.039
h1
15
12
h2
17
11.2
h3
18.5
10
h1
12.45
9.96
h2
14.11
9.296
h3
15.355
8.299
Traditional
Centers
Traditional
Lightweight
D1
Lightweight
D2
Type
Period
Traditional
Light Weight
h1
8,88,737.40
8,68,649.85
h2
9,61,343.40
9,39,595.74
h3
9,29,746.20
9,08,765.19
Traditional
Light Weight
8.5
8
1
period
5. CONCLUSION
This work presents a linear programming approach for solving multi-product, multi-time period
two-echelon supply chain models. The proposed model provided an optimal inventory level for warehouses
and distribution centers and minimizing the total costs related to inventory, transportation, toll fees and
packing charge of the entire supply chain. An industrial case study is utilized to demonstrate the feasibility
of applying the proposed approach to practical problems in a supply chain. The proposed approach yields
more efficient solution and several significant managerial implications for practical applications. The
reduction of the total cost in the supply chain due to the usage of eco-friendly lightweight plastic pail,
rather than the traditional plastic pail is also analyzed in the present work. In particular, the proposed
computational methodology can be easily extended to any other situation and can handle real life decision
making problems in supply chains. In future, uncertainty of costs and demands can be considered in the
model and new solution methodologies including uncertainty can be developed via fuzzy models.
REFERENCES
S. Ana Maria, N. Rakesh, 1999,A review of integrated analysis of production and distribution systems, IIE Transactions
31 10611074.
Ahumada O., Villalobos J.R., 2009, Application of planning models in the agri-food supply chain: a review, European
Journal of Operational Research 196 (1), pp. 1-20.
Azaron, A., Brown, K.N., Tarim, S.A. and M. Modarres (2008). A multi objective stochastic programming approach for
supply chain design considering risk. International Journal of Production Economics 116 (1), pp. 129-138.
K.Balaji Reddy, S.Narayan and P.Pandian,2011,Single-Echelon supply chain two stage distribution inventory
optimization model for the confectionery industry,Applied Mathematical sciences,vol.5,no.50,2491-2504
Bask, A. H. ,2001, Relationships among TPL providers and members of supply chains: a strategic perspective. Journal
of Business & IndustrialMarketing, 2001, 16(6), 470-486.
B.M. Beamon, 2004, Supply chain design and analysis models: models and methods, International Journal of
Production Economics 55 (1998) 281294
Chen, B. Bilgen, I. Ozkarahan, Strategic tactical and operational production-distribution models: a review, International
Journal of Technology Management 28 151171.
Childhouse P, Towill DR ,2003, Simplified material flow holds the key to supply chain integration, OMEGA, 31(1):17
27.
C.T. Chen, C.T. Lin, S.F. Huang, 2006 A fuzzy approach for supplier evaluation and selection in supply chain
management, International Journal of Production Economics 102 289301.
Cohen M.A., Lee H.L., 1989, Resource deployment analysis of global manufacturing and distribution networks,
Journal of Manufacturing and Operations Management 2, pp. 81-104.
Disney S.M. and Towill D.R. ,2003, Vendor-managed inventory andbullwhip reduction in a two-level supply chain,
International Journal ofOperations and Production Management, Vol. 23 No. 6, pp. 625-651
S.S. Erengc, N.C. Simpson, A.J. Vakharia, 1999Integrated production/distribution planning in supply chains: an
invited review, European Journal of Operational Research 115, 219236.
Fahimnia, B.; Lee Luong; Marian, R., 2009, Optimization of a Two-Echelon Supply Network Using Multiobjective Genetic Algorithms Computer Science and Information Engineering, WRI World Congress
B.Fleischmann,2005,Distribution and transport planning ,supply chain management and advanced planning ,229-244
Gen M., Syarif A., 2005, Hybrid genetic algorithm for multi-time period production distribution planning, Computers
and Industrial Engineering 48, pp. 799-809.
Gunasekaran A., Ngai E., 2009, Modeling and analysis of build-to-order supply chains,European Journal of Operational
Research 195 (2), pp. 319-334.
L., Lee, W., 2004, Multi objective optimization of multi echelon supply chain networkswith uncertain product demands
and prices, Computers and Chemical Engineering 28, pp. 1131-1144.
Lin L., Gen M., Wang X., 2007, A hybrid genetic algorithm for logistics network design with flexible multistage
model, International Journal of Information Systems for Logistics and Management 3 (1), pp. 1-12.
Nagurney A., Toyasaki F., 2005, Reverse supply chain management and electronic waste
recycling: a multitired network equilibrium framework for e-cycling, Transportation Research Part E, pp. 1-28.
Ozdamar L., Yazga T., 1997, Capacity driven due date settings in make-to-order production systems, International
Journal of Production Economics 49 (1), pp. 29-44.
Paksoy T., 2005, Distribution network design and optimization in supply chain management: under material
requirements constraints a strategic production-distribution model, Journal of Selcuk University Social Sciences
Institute 14, pp. 435-454, in Turkish.
D. Petrovic, 2001 Simulation of supply chain behavior and performance in an uncertain environment, International
Journal of Production Economics 71,429438.
D. Petrovic, R. Roy, R. Petrovic, 1998, Modelling and simulation of a supply chain in an uncertain environment,
European Journal of Operational Research 109,299309.
Petrovic D., Roy R., Petrovic R., 1999, Supply chain modeling using fuzzy sets, International Journal of Production
Economics 59, pp. 443-453.
Potter, A. and Lalwani, C. ,2005, Supply chain dynamics and transport management: A review, Proceedings of the 10th
Logistics ResearchNetwork Conference, Plymouth, 7th-9th September.
Pyke D.F., Cohen M.A., 1993, Performance characteristics of stochastic integrated production distribution systems,
European Journal of Operational Research 68 (1), pp. 23-48.
Schtz P., Tomasgard A., Ahmed S., 2008, Supply chain design under uncertainty using sample average approximation
and dual decomposition, European Journal of Operational Research, doi:10.1016/j.ejor.2008.11.040.
Shin H., Benton W.C., Jun M., 2009, Quantifying suppliers product quality and delivery performance: A sourcing
policy decision model, Computers & Operations Research 36, pp. 2462-2471.
N.C. Simpson, A.J. Vakharia, 1999,Integrated production/distribution planning in supply chains: an invited review,
European Journal of Operational Research115, 219236.
Stank T and Goldsby T J ,2000, A framework for transportation decision making in an integrated supply chain. Supply
Chain Management: AnInternational Journal, Vol 5: No 2
Sourirajan K., Ozsen L., Uzsoy R., 2009, A genetic algorithm for a single product network design model with lead time
and safety stock considerations, European Journal of Operational Research 197 (2), pp. 599-608.
P. Subramanian, N. Ramkumar, T.T. Narendran, 2010,Mathematical model for multi-echelon, multi-product, single
time-period closed loop supply chain, International Journal of Business Performance and Supply Chain Modelling
,Vol. 2, No.3/4 pp. 216 - 236
Syarif A., Yun Y., Gen M., 2002, Study on multi-stage logistics chain network: a spanning tree-based genetic algorithm
approach, Computers and Industrial Engineering 43 (1), pp. 299-314.
Tuzkaya U., nt S., 2009, A holonic approach based integration methodology for transportation and warehousing
functions of the supply network, Computers and Industrial
This chapter investigates production inventory model with imperfect quality items, allowing shortages. Cent
percent screening process is performed during the production stage itself so that imperfect quality items are
picket up and are sold as a single batch due to economies of sale. The objective is to determine the optimal
production quantity and shortage quantity to maximize the total profit. Holding cost, Setup cost, Production
rate and Supply rate, defective rate are taken as triangular fuzzy numbers to develop a fuzzy model to maximize
the total profit and graded mean integration representation method is used to defuzzify the result. Optimal lot
size, optimal shortage quantities are derived. Numerical examples are given to illustrate the results of the
proposed model.
6.1. Introduction
In production inventory scenario, the manufactured products are assumed to be, rather expected to be cent percent perfect.
But in real life situations, imperfect quality items are inevitable. The producers wish to
supply their customers only perfect quality items. So a screening process is adopted to pick out imperfect items. Salameh and Jaber
[45] incorporated some (or) structure of imperfect quality items and screening process by taking imperfect quality items as a random
variable. Many research works are carried out to avoid shortages using a human phenomenon known as learning.
Chang H.C (2004) [11] has applied fuzzy set theory to EOQ with imperfect quality items. In the present study, assume
finite rate of production of K units and finite rate of supply of r units per unit time. The unit production, cost is Rs. C and setup cost is
K per run. Each lot produced contains p percentage of defectives. Defective items are picked up by 100% screening process at a rate x
and are sold at the rate of v per unit prior to next production schedule back ordering (shortages) is permissible and backorder cost is b
per unit. y2 is the optimum unit of backordering. It is also assumed that only good quality items are alone supplied to the customer.
The behavior of inventory level is illustrated in Fig.6.1, where T is the total cycle length.
6.2. Assumptions
1.
Inventory is building up at a constant rate of K r (1 - p) units per unit of time during T11.
2.
No replenishment during time T12 and inventory is decreasing at a rate of r units per unit time.
3.
Shortage is building up at a constant rate of r units per unit of time during T21.
4.
Shortages are being filled immediately at the rate of K (1 - p)r units per unit of time during T22.
5.
6.3. Notations
K
Production rate
Demand rate
Percentage of defective
Screening rate
production quantity
y2
shortage quantity
set up cost
cycle length
h%
K%
r%
p%
b%
K%
P TP(y)
TP(y)
- defuzzified value of
Inventory
py
T2
y
T21
T11
T22
T12
Time
T1
T
Fig. 6.1
The total cost per unit time is given by
1
K
2y K - r(1 - p)
K - (1 - p)r
y - y 2
K
K Kr
hpyKr
+ by 22
(K+pr)y X(K+pr)
TC =
. . (6.1)
sy(1-p)+vyp-cy-dy
T
Total selling price TS =
. . . (6.2)
Total Profit, TP = TS TC
K - (1 - p)r
(s(1-p) + vp - c - d)Kr
K
y - y 2
h
K + pr
2y(K - r(1 - p))
K
+ by 22
TP =
K Kr
hpKry
Differentiating (6.3) partially with respect to y2 so as to maximize TP and equating to zero we have
h (K - (1 - p)r)
y
(b + h)
K
y2 =
. . . (6.4)
substituting the value of y2 in (6.3) and differentiating (6.3) with respect to y and equating to zero we have
K Kr
hpKr
(K - (1 - p)r)
bh
K + pr
2K(b + h)
x(K + pr)
y=
d2
(TP) < 0 y = y 0 =
2
dy
K1Kr
hpKr
(K - (1 - p)r)
bh
K + pr
x(K
+
pr)
2K(b
+
h)
. . . (6.5)
h (K - (1 - p)r) 0
y 02 =
y
b+K
K
. . . (6.6)
Validity : If we put p = 0 in (6.5) we get
2(b + h) KKr
bh(K - r)
y0 =
2K hr(K - r)
(h + b) bK
y 02
which is the value of y0 in production inventory (with all perfect items) with shortages and
h%= (h1 , h 2 , h 3 )
%= (K , K , K ) K
% = (K , K , K )
K
1
2
3
1
2
3
b%= (b1 , b 2 , b 3 )
p%= (p1 , p 2 , p 3 )
Let
%%
%
% + vp%- c - d)Kr
(s(1 - p)
K
%+ pr
%- r(1
%
%
% - p)
%
K
2y(K
TP(y)
=
r%= (r1 , r2 , r3 )
%- (1 - p)r
K
%%
h%
y
y
%
K
2
%
+ by 2
%%%%
K r%
K%%
hpKry
%% x(K%+ pr)
%%
(K%+ pr)y
.. (6.7)
K3
(s(1 - p3 ) + vp1 - c - d)K1r1 -
K 3 + p3 r3
2y(K1 -r3 (1-p1 )
K 3 -r1 (1-p3 )
2
h3
y-y
+
b
y
2
3 2
K1
K1 K1 r1
h p K ry
1 1 11
(K 3 + p3r3 )y x(K 3 + p3r3 )
,
(s(1 - p ) + vp - c - d)K r
2
2
2 2
K
+
p
r
2
2 2
K2
-
2y(K 2 - r2 (1 - p 2 )
K 2 - r2 (1 - p 2 )
h2
y
y
K2
+ b 2 y 22
K2 K 2 r2
h pK ry
2 2 22
(K 2 + p 2 r2 )y x(K 2 + p 2 r2 )
,
(s(1 - p ) + vp - c - d)K r
K1
1
3
3 3
-
K1 + p1r1
2y(K 3 - r1 (1 - p3 )
K1 - r3 (1 - p 1 )
h1
y
y
K3
(6.8)
P TP(y)
+ b1y 22
K3 K 3 r3
h pK ry
3 3 33
(K1 + p1r1 )y x(K1 + p1r1 )
..
K3
(s(1 - p3 ) + vp1 - c - d)K1r1
-
K 3 + p 3r3
2y(K1 - r3 (1-p1 )
K - r (1 - p 3 )
h3 3 1
y
y
K1
1
6
+ b3y
2
2
K1 K1 r1
h pK ry
1 1 11
(K 3 + p3r3 )y x(K 3 + p3r3 )
(s(1 - p ) + vp - c - d)K r
K2
2
2
2 2
4
-
K 2 + p 2 r2
2y(K 2 - r2 (1 - p 2 )
K 2 - r2 (1 - p 2 )
2
h2
y - y 2
+ b 2 y 2
K2
K2 K 2 r2
h pK ry
2 2 22
(K 2 + p 2 r2 )y x(K 2 + p 2 r2 )
(s(1 - p ) + vp - c - d)K r
K1
1
3
3 3
K1 + p1r1
2y(K 3 - r1 (1 - p3 )
K1 - r3 (1 - p 1 )
h1
y
y
K3
+ b1y 22
K3 K 3 r3
h pK ry
3 3 33
(K1 + p1r1 )y x(K1 + p1r1 )
. . (6.9)
K h (K - r (1 - p1 ))
h 3 (K 3 - r1 (1 - p 3 ))K 3
4h 2 1 1 1 3
(K1 r3 (1 p1 )K1
K 3 (K 3 r1 (1 p3 )
y
(b
h
)K
4(h
b
)K
(b
h
)K
3
3
3
2
2
2
1
1
1
(K1 r3 (1 p1 ) (K 2 r2 (1 p 2 )) (K 3 r1 (1 p 3 )
y2 =
P TP(y)
y 22
y 02 = cy 0
is negative.
y 02
To determine y0, we substitute the value of
(6.9) becomes
P TP(y)
K3
(s(1-p3 )+vp1 -c-d)K1r1
-
K 3 + p3r3
2y(K1 -r3 (1-p1 )
1
6
K -r (1-p3 )
2 2
h3 3 1
y-cy
+
b
c
y
K1
K1 K1 r1
h p K ry
1 1 11
(K 3 + p3r3 )y x(K 3 + p3r3 )
(s(1 - p ) + vp - c - d)K r
K2
2
2
2 2
4
-
K 2 + p 2 r2
2y(K 2 - r2 (1 - p 2 )
K 2 - r2 (1 - p 2 )
2 2
h2
y-cy
+ b 2c y
K2
K2 K 2 r2
h pK ry
2 2 22
(K 2 + p 2 r2 )y x(K 2 + p 2 r2 )
(s(1 - p ) + vp - c - d)K r
K1
1
3
3 3
K1 + p1r1
2y(K 3 - r1 (1 - p3 )
K - r (1 - p1 )
h1 1 3
y - cy
K1
+ b1c 2 y 2
K3 K 3 r3
hpKry
3 3 33
(K1 + p1r1 )y x(K1 + p1r1 )
. . .(6.11)
K 3K 3r3
K1 K1r1
4K2 K 2 r2
K 3 p3r3
K 2 +p 2 r2
K1 +p1r1
(K 3 - r1 (1 p3 )
K
K2
2
3
c
+b
c
4
3
3
K1
2(K 2 r2 (1 p 2 )
2(K1 -r3 (1 p1 )
2
2
(K 2 -r2 (1-p 2 )
2
2
h2
c b 2 c
h1
c b1c
K
2(K
-r
(1-p
)
K
3 1
3
3
1 h1p1K1r1 4h 2 p2 K 2 r2 h 3p3 K 3 r3
x K 3 p3 r3 K 2 p 2 r2 K1 p1r1
2 P TP(y)
y
Also
is negative.
\ y0 =
K3 K3 r3
K1 K1r1
4K2 K 2 r2
K 3 p3r3
K 2 +p 2 r2
K1 +p1r1
(K 3 - r1 (1 p3 )
K
K2
2
3
4
h
c
+b
c
3
3
K1
2(K1 -r3 (1 p1 )
2(K 2 r2 (1 p 2 )
(K 2 -r2 (1-p 2 )
K1
2
h2
b
c
2
K2
2
h1
b
c
1
K3
1 h p K r 4h p K r h p K r
+ 1 1 1 1 2 2 2 2 3 3 3 3
x K 3 p3r3 K 2 p 2 r2 K1 p1r1
. . . (6.12)
y02 = cy0
Validity
b1 = b 2 = b 3 = b ; r1 = r2 = r3 = r ; p1 = p 2 = p 3 = p ;
h(K - r(1 - p)
y
(b + h)K
equation
(6.10) becomes y2 =
K + pr
K Kr
hpKr
(K - (1 - p)r)
+
bh
x(K + pr)
2K(b + h)
y =
30000 units/year.
Supply rate,
16000
Setup cost,
K'
Rs.1000/setup
Holding cost, h
Rs.3.25/unit/year.
Shortage cost, b
Rs.30/unit/year.
Screening rate, x
Rs.175200/unit/year.
Screening cost, d
Re.0.5/unit.
Production cost, c
Rs.25/unit.
units/year.
=Rs.20/unit.
0.02
%
K
The optimum order quantity is found to be 4829 units. In fuzzy model if
P%
16400),
K%
%
V
= (29950, 30000, 30050),
= (16200, 16300,
h%
= (990, 1000, 1010),
a difference of only one unit between crisp and fuzzy optimal order quantity. Sensitivity analysis on the above parameters is presented
in Table 6.1.
Table 6.1
y0(fuzzy)
y0(crisp)
difference
y 02
990,1000,1010
3.24,3.25,3.26
29950,30000,30050
16200,16300,16400
.015,.02,.025
29,30,31
4828
4829.1
221
990,1000,1010
3.24,3.25,3.26
29950,30000,30050
16200,16300,16400
.024,.025,.026
29,30,31
4803
4803
221
990,1000,1010
3.24,3.25,3.26
29950,30000,30050
16200,16300,16400
.029,.03,.031
29,30,31
4778
4779
221
990,1000,1010
3.24,3.25,3.26
29950,30000,30050
16200,16300,16400
.034,.035,.036
29,30,31
4753
4754
221
990,1000,1010
3.24,3.25,3.26
29950,30000,30050
16200,16300,16400
.034,.035,.036
14,15,16
4976
4977
422
990,1000,1010
3.15,3.2,3.25
29900,30000,30100
16200,16300,16400
.015,.02,.025
24,25,26
4907
4909
260
990,1000,1010
3.1,3.2,3.3
29900,30000,30100
16100,16200,16300
.015,.02,.025
24,25,26
4874
4878
260
990,1000,1010
4.9,5,5.1
29950,30000,30050
24900,25000,25100
.039,.04,.041
14,15,16
7745
7752
387
990,1000,1010
4.9,5,5.1
29950,30000,30050
24900,25000,25100
.039,.04,.041
19,20,21
7515
7522
387
990,1000,1010
3.9,4,4.1
29950,30000,30050
16400,16500,16600
.039,.04,.041
19,20,21
4487
4488
353
From Table 6.1, we infer the following (i) when percentage of defective items increases, there is decrease in optimum order quantity,
keeping all parameters fixed (ii) when backorder cost decreases, there is an increase in the optimum and back order quantity keeping other values
fixed (iii) when holding cost increases and backorder cost decreases, the optimum order quantity and backorder quantity increases (iv) when
holding cost, rate of supply, percentage of defective items increase and backorder cost decreases keeping the other values fixed, the optimal order
quantity and backorder quantity get almost doubled (v) when percentage of defective items and holding cost increase and backorder cost
decreases, keeping the other parameters unchanged, there is considerable increase in optimum order quantity and backorder quantity.
But in all the cases we see that the optimal values of order quantity and backorder are very much closer to the crisp values.