ENVIRONMENT ORIENTED INVENTORY MODEL AND BENEFITS OF

INCINERTION AS WASTE DISPOSAL METHOD

A.Shanthi
Idhaya College of Arts and Science for
Women,Pondicherry
Abstract
It is well known that industrial sectors are the backbone of any nations economy. The
motive of the establishment of any industry is to satisfy the demands of the needy
instantaneously, which explicitly states that it has social responsibilities. To make the
fulfillment of the demands in right proportion and at the right time, inventory models
(EOQ/EPQ [Economic order/production quantity]) were formulated. On the other hand in
a production / inventory situation items which are, received or produced are not of perfect
quality. Thus the presence of defects which is inevitable in a produced/ordered lot is
sorted out by the process of screening. The imperfect items cannot be discarded as waste,
since the industries possess environment responsibilities, remanufacturing tactics and
waste management techniques have been implemented. Several literatures have discussed
about inventory models with production, remanufacturing and waste disposal, but
disposal methods and techniques were not discussed in particular. In this paper an
inventory model is formulated in which the defective items in a lot are made to undergo
seperate screening which in turn are categorised based on the Life Cycle Assessment
(LCA) methodology as reusable, recyclable and final residues are subjected to
incineration one of the waste disposal methods which is both economically and
environmentally beneficial.
1. Introduction
Inventory models have been a great support for the successful run of the industries which
undergo a number of modifications to act as the remedial measure to the problems that
arise in the production /inventory situations. Demands of the customers must be satisfied
instantaneously for which the production/ordering rate is fostered to avoid shortages. It is
vivid that quantity of items is focussed on, but in addition to it the quality of the items
must also be given significance as customers consistency depends on it. The quality of
the produced / received lot is determined by the proportion of defective items present.
The process of screening is carried out to isolate the perfect items from the imperfect
ones. To transform the imperfect items as perfect, remanufacture is undertaken.
Remanufacture accounts to recycle, repair and rework. The items that are
imperfect have defects, but the percentage of defects present in each item is distinct. A
few numbers of defect can be mended manually within a short span of time so that they

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

explained by Bernadette, but these effects could be minimized by proper installation of

mitigating techniques to treat the emissions. The cost of setting an incineration plant and
its associated cost parameters such as operating, maintenance, residue disposal,
transportation waste haulage and fuel surcharge are included, and the revenue earned out
of it is briefed in the literature. After the analysis of the various disposal methods
incineration rank first, because even though the incineration operational tasks are
expensive; it yields revenue in terms of energy and materials. This advantage of
incineration has encouraged many industrial sectors to construct small incinerators within
their firms for waste management to promote environment sustainability.
4. Model Formulation
4.1 Problem Description
Consider the case where a lot of size Y is delivered instantaneously with a
purchasing price of c per unit and an ordering cost of K. It is assumed that each lot
received contains a known percentage of defectives p, after 100% percent of screening
imperfect items is separated from the good items. To make profit out of the defective
items, they are made to undergo second screening i.e the LCA methodology is used to
rank the defective items as reusable, reworkable and the residues to be incinerated based
on their proportion which is expressed as the percent of imperfect items. It is classified
based on the degree and nature of defects present in each imperfect item.
4.2
Notations
The following notations will be used throughout the paper when developing the
mathematical model.
Y

order size

unit variable cost

fixed cost of placing an order

percentage of defective items in Y

percentage of defective items those are suitable for reuse.

percentage of defective items those are suitable for recycle.

percentage of defective items that are subjected to incineration.(1-q-r)

unit selling price of items of good quality.

unit selling price of defective items that are categorized as reusable.

(v < s)

unit cost of recycling

R1

unit selling price of recycled items.(R1 > R)

unit cost of disposal by the method of incineration

I1

revenue earned per unit.

screening rate

unit screening cost

d1

unit screening cost of the defective items.

Emission caused by incineration mitigating cost

cycle length.

3.3

Assumptions

1.

The demand is constant.

2.

No shortages are allowed.

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.

Lead time is assumed to be Zero.

6.

Planning horizon is infinite.

4.4

Mathematical Model.

As in Salameh [30] model to avoid shortages it is assumed that

Dt N(Y,p) , where N(Y,p) is the number of good items in each order, also in
addition to that p, the percentage of defective items in y is restricted to
1i.e p 1.

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

; and can be written as

vq-

+
+

D*

(1)
D*
+
D*

(s

vq+

TPU(Y)

(2)

The objective is to determine the optimal quantity. The necessary condition is

(3)

Note that when p = 0, Eq.(3) reduces to the traditional EOQ model.

The reason why E vanishes is, if there is no defective items present in
the produced lot then there is no need of the waste disposal method incineration so the
costs associated with emission mitigation is excluded.
5. Numerical results.
To illustrate the usefulness of the model developed in Section 4, let us consider the
inventory situation where a stock is replenished instantly with Y units of which not all are
of the desired quality.
The parameters needed for analyzing the above inventory situation are given below:
D = 50,000 unit / year; c = $ 25 / unit; K = $100 / cycle; s = $ 50/unit; v=$40/unit; h = $ 5
/unit / year; R=$5/unit;E = $ 500 / cycle R1 = $ 15 / unit ; I = $ 2 / unit ; I1 = $ 5 / unit ; x
= 1 / unit / time ; d = $0.5 /unit; d1 = $0.5 / unit; p =0.05; q = 0.5; r = 0.3, i = 0.2.
Assume that the inventory operation operates on an 8 hours/day, for 365 days a year, then
the annual screening rate, x=1*60*8*365=175 200 units/year.
Then the optimum value of Y that maximizes Eq.(1) is given from Eq.(2)

= 31583 units.

Substituting

units in Eq.(2) the maximum profit per

unit year is given as

TPU(Y) = 1290321.9/year.
6.

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

considerations,Comput. Ind. Eng. 58 (3) (2010) 352362.

[9] A.M.A. El Saadany, M.Y. Jaber, The EOQ repair and waste disposal model
with switching, Comput. Ind. Eng. 55 (1) (2008) 219233.
[10] A.M.A. El Saadany, M.Y. Jaber, Production/remanufacture EOQ model
with returns of subassemblies managed differently, Int. J. Prod. Econ. 133
(1)(2011) 119126.
[11] A.M.A. El Saadany, M.Y. Jaber, M. Bonney, How many times to
remanufacture? Int. J. Prod. Econ., doi:10.1016/j.ijpe.2011.11.017.
[12]Goran.et.al.,Life cycle assessment of energy from solid waste- part I :
general methodology and results.,Journal of Cleaner Production 13 (2005)
213- 229.
[13]Finnveden, G., Ekvall, T., 1998. Life cycle assessment as a decision support
toolthe case of recycling vs. incineration of paper.Resource,Conservation
and Recycling 24, 235256.
[14]Finnveden, G., Johansson, J., Lind, P., Moberg, A ., 2000. Life Cycle
Assessments of Energy from Solid Waste. Stockholms Universitet, Sweden.
[15] M. Fleischmann, J.M. Bloemhof-Rwaard, R. Dekker, E. vander Laan,
J.A.E.E. van Nunen, L.N. van Wassenhove, Quantitative models for reverse
logistics,Eur. J. Oper. Res. 103 (1) (1997) 117.
[16]Frijriksson, G. B., Johnsen, T., Bjarnaso ttir, H. J., Slentnes, H., 2002.
Guidelines for the use of LCA in the waste management sector. Nordtest
Project no. 153701.
[17]Harris F.W., 1913.How many parts to make at once, factory?,Mag Manage.
10, 135136 (p. 152).
[18]Harrison, K.W., Dumas, R.D., Barlaz, M.A., 2000. Life-cycle inventory
model of municipal solid waste combustion. Journal of the Air and Waste
Management Association 50, 9931003.
[19] M.Y. Jaber, M.A. Rosen, The economic order quantity repair and waste
disposal model with entropy cost, Eur. J. Oper. Res. 188 (1) (2008) 109
120.

[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

[32] L.A. San-Jose, J. Sicilia, J. Garcia-Laguna, An economic lot-size model

with partial backlogging hinging on waiting time and shortage period,
Appl. Math. [6] R.H. Teunter, Lot-sizing for inventory systems with
product recovery, Comput. Ind. Eng. 46 (3) (2004) 431441.
[33] D.A. Schrady, A deterministic inventory model for repairable items, Naval
Res. Log. Quart. 14 (3) (1967) 391398.
[34]TEI, 2003. Final Report for the Project on Life Cycle Assessment for Asian
CountriesPhase III. Thailand Environmental Institute, Thailand.

[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.

ENVIRONMENTALLY RESPONSIBLE REPAIR AND WASTE

DISPOSAL INVENTORY MODELS
K.Elavarasan
SCMS School of engineering and technology, Kerala -683 582

Abstract : Traditional inventory models involve different decisions that attempt to

optimize material lot sizes by minimizing total annual switching costs. However, the
increasing concern on environmental issues stresses the need to treat inventory
management decisions as a whole, by integrating economic and environmental
objectives. Recent studies have underlined the need to incorporate additional criteria in
traditional inventory models in order to design "responsible inventory systems". This
paper explores the problem of determining the optimal batch sizes for production and
recovery in an EOQ (economic order/production quantity) repair and waste disposal
model context. This paper assumes that a first shop is manufacturing new products as
well as repairing products used by a second shop. The used products can either be stored
in the second shop and then be brought back to the first shop in an approach used to
reduce inventory costs, or be disposed outside the system. The works available in the
literature assumed a general time interval and ignored the very first time interval where
no repair runs are performed. This assumption resulted in an over estimation of the
average inventory level and subsequently the holding cost. These works also have not
accounted for switching costs when alternating between production and recovery runs,
which are common when switching among products or jobs in a manufacturing facility.

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).

2. NOTATION AND ASSUMPTIONS :

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

number of newly manufactured batcher in are interval of length T.

number of repaired batches in an interval of length T

demand rate (unit per unit of time)

holding cost per unit per unit of time for shop 1.

holding cost per unit per unit of time for shop 2.

waste disposal rate, where 0 < < l

repair rate of used items, where + = I and 0 < < 1.

batch size for interval T which includes n newly manufactured and m

repaired batches.

repair setup cost per batch

manufacturing set up cost per batch

fixed cost per trip(monetary unit)

variable cost per unit transported per distance travelled

distance travelled

proportion of demand returned (0 < < 1)

social cost from vehicle emission

average velocity (km/h)

cost to dispose waste to the environment(mu/unit)

proportion of waste produced per lot Q.

Assumptions :

1.

Infinite manufacturing and recovery rates.

2.

Repaired items are as good as new.

3.

Demand is known, constant and independent.

4.

Lead time is zero.

5.

Single product case.

6.

No shortage.

7.

Infinite planning horizon.

8.

Unlimited storage capacity is available.

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.

Setup cost for shop 1 and shop 2 is

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

Holding cost for the first shop is

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

Holding cost for the second shop is

u Q 2
Q 2d

uQ 2
2d
=

(m 1)
2

1 - m (m 1)

Transportation cost per cycle is

Ct(Q) = 2a + bDQ + bDQ
Emission cost from transportation per cycle is
Ce (Q) = 2

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)

= setup cost + holding cost + transportation cost + emission cost

+ waste disposal cost.

mr ns +

h Q2 2 Q2 2
+
2 dn
dm

uQ 2

+
1 (m 1)
2d
m

+ [2a + bDQ + bDQ +

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)

Now diff (1), w.r.t Q,

d2K
dQ 2
(ie)

for every Q > 0 and it has a unique minimum and derived by setting its first
dK
= 0.
dQ

derivative equal to zero (ie)

dK
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)

Subs the value of Q in (3), we get

4Dd
2 0d
V

4Dd
2d mr ns + 4ad +
2 0d
V

2 2

h
+ u 1
m

n m
2d mr ns + 4ad +

dK
dQ2

The optimal value of Q0 is

2d mr ns + 4ad +
2 2

+ u

n m

Q0

4Dd
2 0d
V

1
m

3/2

>0

Subs the value of Q0 in (1)

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)

Subs the value of l1, l2, l3 in (3) we get

2

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

The optimal total cost is

K(Q) =

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.

Hall, R. W. (1988). Graphical techniques for planning changes in production capacity.

International Journal of Production Research, 26(4), 675-689.
Han, Y.H., Zhou, C Bras, B., McGinnis, L., Carmichael, C., & Newcomb, P. (2003).
Paint line color change reduction in automobile assembly through simulation. In
Proceedings of the 2003 winter simulation conference, December 7-10, Vol. 2, pp. 12041209.
Hur, S., Kim, J., & Kang, C. (2003). An analysis of the M/G/I system with N and T
policy. Applied Mathematical Modelling, 27(8), 665-675.
Inderfurth, K., Lindner, G., & Rachaniotis, N. P. (2005). Lot sizing in a production
system with rework and product deterioration. International Journal of Production
Research, 43(7), 1355-1374.
Inman, R. R. (1999). Are you implementing a pull system by putting the cart before the
horse? Production and Inventory Management Journal, 40(2), 67-71.
Jaber, M. Y., & Rosen, M. A. (2008). The economic order quantity repair and waste
disposal model with entropy cost. European Journal of Operational Research, 188(1),
109-120.
Kamrad, B., & Ernst, R. (2001). An economic model for evaluating mining and
manufacturing ventures with output yield uncertainty. Operations Research, 49(5), 690699.
Kella, O. (1991). Optimal manufacturing of a two product mix. Operations Research,
39(3), 496-501.
Khouja, M. (2005). The use of minor setups within production cycles to improve product
quality and yield. International Transactions in Operations Research, 12(4), 403-416.
Khouja, M., & Mehrez, A. (2005). A production model for a flexible production system
and products with short selling season. Journal of Applied Mathematics and Decision
Sciences, 2005(4), 213-223.

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.

Watkins, H. R. W. (1957). The cost of rejecting optimum production runs. Operations

Research, 8(4), 200-205.
Weber, A. (2006). Some change is good; too much is bad. Assembly, 49(2), 58-64.
Wolsey, L. (1997). MIP modelling of changeovers in production planning and scheduling
problems. European Journal of Operational Research, 99(1), 154-165.
Wu, R., Song, C., & Li, P. (2004). Receding horizon control of production systems with
aging of equipment. In Fifth world congress on intelligent control and automation
(WCICA 2004), June 15-19, Vol. 3, pp. 2763-2767.
Yan, H., & Zhang, Q. (1997). A numerical method in optimal production and setup
scheduling of stochastic manufacturing systems. IEEE Transactions on Automatic
Control, 42(10), 1452-1455.s

An analysis of Help-Seeking Behaviour among

Women with Urinary Incontinence using Fuzzy
Cognitive Map Model
Dr.Joseph,
Christ university,Bangalore
Abstract. Female urinary incontinence is a major health problem and has a significant
detrimental impact quality of life. Feelings of low self-esteem, embarrassment and helplessness
are common. Between 20-50% of women suffer from incontinence at some time in their lives and
the prevalence of symptoms and demand for services is rising substantially. The present study
was designed with the objectives to study the factors of not seeking a help of a doctor for
urinary incontinence. Hence, this research investigates the most contributing / impactful factor of
not seeking help of a doctor due to urinary incontinence using Fuzzy Cognitive Maps Model.
Fuzzy Cognitive Map Model is a fuzzy-graph modeling approach based on experts opinion. This
is the non-statistical approach to study the problems with imprecise information.
Keywords: Fuzzy cognitive maps (FCMs), Urinary Incontinence , Help-seeking Behaviour.

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.

3.2.3. Strength of Relationships Matrix of Success (SRMS)

The SRMS is an [n x n] matrix. The rows and the columns of the matrix are the
factors and each element Sij in the matrix indicates the relationship between factor i and factor
j. Sij can accept values in the interval [_-1, 1]. Each factor is represented as a numerical vector
Si containing n components, one for each concept to be represented in the map. There are three
possible relations between two concepts i and j (Sij):
Sij > 0 indicates direct (positive) causality between concepts i and j. That is,
increase in the value of concept i leads to an increase in the value of concept j.
Sij < 0 indicates inverse (negative) causality between concepts i and j. That is,
increase in the value of concept i leads to a decrease of the value of concept j.
Sij = 0 indicates no relationship between concepts i and j.
Therefore, three parameters have to be considered when assigning values to S ij:
The sign (or polarity) of Sij, which indicates whether the relationship between concepts
i and j is direct or inverse. The strength of Sij that indicates how strongly concept i
influences concept j. The direction of causality that indicates whether concept i causes
concept j or vice versa.
Determining the polarity of the relationships.
The numerical vectors of the IMS are transformed into fuzzy sets in the FZMS. Given
V1 and V2, the vectors associated with factors 1 and 2 respectively, and X 1(Vj) and X2(Vj), the
grades of membership of the jth component in vectors V1 and V2; vectors V1 and V2 will be
monotonically increasing related (direct relationship between concepts 1 and 2, and S ij > 0) if
X1(Vj) is similar to X2(Vj) for all or most of the corresponding elements of the two vectors; and
vectors V1 and V2 will be monotonically decreasing related (inverse relationship between
concepts 1 and 2, and Sij < 0) if X1(Vj) is similar to (1 - X2(Vj)) for all or most of the
corresponding elements of the two vectors.
Determining the strength of the relationships.
The closeness of relation between two vectors V1 and V2 given by the computation of
the similarity between those two vectors, determines the strength of the relationship between
the concepts 1 and 2 associated with these two vectors, which will be represented by the module
of the component S12 to be introduced in the SRMS. The closeness of relation between two
vectors is based on the concept of distance between vectors (Kosko, 1985). The mathematical
procedure to calculate the similarity between two vectors that is shown hereafter in this paper
follows the approach proposed by Schneider et al. (Schneider, Shnaider,Kandel,& Chew, 1998).
Different computation is required for those vectors that are directly related and those
which are inversely related.
If vectors V1 and V2 are directly related, then the closest relation between them is
when for each j (j = 1, . . . , m), X1(Vj) = X2(Vj). Let dj be the distance between the corresponding
jth elements of V1 and V2 such that
and

(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

5.APPLICATION OF FCM METHODOLOGY

5.1. The Initial Matrix of Success (IMS)
The identified proposed factors mentioned above are given to the panel of experts
consisting of 9 experts. So IMS matrix is of 9x9 type matrix, as the number of factors = number
of people interviewed = 9.
Each element Oij of the matrix represents the importance that individual j gives to
factor i, within the scale from 0 ( the factor does not contribute at all to the success of the
project) to 100 ( the contribution of that factor in very much essential for the task). The IMS
Matrix is given by
Individuals

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

5.2. Fuzzified Matrix of Success (FZMS)

The fuzzification of numerical vectors in IMS can be done using the equations (1) to (5).
For example,

.
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

5.3 Strength of Relationship Matrix (SRMS)

Using the equations (6) (9), the Sij values are calculated from Xij values for each i,j.
For example, for S12 ,
dj values are 0.33
0.33
0.34
0.34
0
0.33
0.33
0
0.17

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

0.74 0.72 0.67

0.69 0.67 0.57
0.74 0.68 0.70
0.65 0.74
0.65
0.83
0.74 0.83
0.59 0.76 0.74
0.67 0.61 0.59
0.78 0.61 0.78

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

5.5 Targeted FCM Graph Representation

0.81

0.84

2
0.76

0.74

0.78

5
4

9
0.83

0.78

6
7

0.76

Figure 2. FCM Graph

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]

Luis Rodriguez-Repiso, Rossitza Setchi , Jose L. Salmeron, Modelling IT projects

success with Fuzzy Cognitive Maps, Expert Systems with Applications 32 (2007) 543
559.
Ahmed S El-Azab1, Omar M Shaaban, Measuring the barriers against seeking
consultation for urinary incontinence among Middle Eastern women , BMC Womens
Health 2010, 10:3 .
Nezihe Kizilkaya Beji, Ayfer Ozbas, Ergul Aslan, Dilek Bilgic, Habibe Ayyildiz Erkan,
Overview of the Social Impact Of Urinary Incontinence with A Focus on Turkish
Women, 2010 Society of Urologic Nurses and Associates Urologic Nursing, pp. 327335.
Teunissen D, van Weel C, Lagro-Janssen T. Urinary incontinence in older people living in
the community: examining help-seeking behaviour. Br J Gen Pract 2005;55:776-782.
[PubMed: 16212853].
M. ODonnella,, G. Loseb, D. Sykesc, S. Vossc, S. Hunskaara, ,Help-Seeking Behaviour
and Associated Factors among Women with Urinary Incontinence in France, Germany,
Spain and the United Kingdom , European Urology 47 (2005) 385392 .
Shaw C. A review of the psychosocial predictors of help-seeking behaviour and impact
on quality of life in people with urinary incontinence. J Clin Nurs 2001;10:15. [PubMed:
11820232].
Hannestad YS, Rortveit G, Hunskaar S. Help-seeking and associated factors in female
urinary
incontinence. Scand J Prim Health Care 2002;20:1027.
Kinchen KS, Burgio K, Diokno AC, et al. Factors associated with womens decisions to
seek treatment for urinary incontinence. J Womens Health (Larchmt) 2003;12:687.
[PubMed: 14583109].
Hagglund D, Walker-Engstrom ML, Larsson G, Leppert J. Reasons why women with
long-term urinary incontinence do not seek professional help: a cross-sectional
population-based cohort study. Int Urogynecol J 2003;14:296304.
Holst K, Wilson PD. The prevalence of female urinary incontinence and reasons for not
seeking treatment. NZ Med J 1988;101:7568.
Hunskaar S, Lose G, Sykes D, Voss S. The prevalence of urinary incontinence in women
in four European countries. BJU Int 2004;93:32430.
R. Axelrod, Structure of Decision, the cognitive maps of Political Elites, Princeton, NJ:
Princeton
University Press, 1976.
Bart, Fuzzy Cognitive Maps Int. J. Man-Machine Studies 24 (1986) 6575.
Gerald J. Calais, Fuzzy Cognitive Maps Theory: Implications for Interdisciplinary
Reading, National
Publications FOCUS On Colleges 2 (2008) 115.
Kosko, B., January, 1986, Fuzzy Cognitive Maps, International journal of manmachine
studies, pp.62-75.
W.B. Vasantha Kandasamy and Florentin Smarandache, Fuzzy Cognitive Maps and
Neutrosophic
Cognitive Maps, Copyright (2003), 510E, Townley Ave, USA.
Kardaras, D., & Karakostas, B. (1999). The use of fuzzy cognitive maps to simulate the
information systems strategic planning process. Information and Software Technology,
41, 197210.
Kinchen KS, Burgio K, Diokno AC, et al. Factors associated with womens decisions to
seek treatment for urinary incontinence. J Womens Health (Larchmt) 2003;12:687.
[PubMed: 14583109].

[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.

OVERAGE MANAGEMENT WITH CUSTOMER

ACQUSITION
A.Mathuravalli
Pondicherry Engineering College,Pillaichavady, Puducherry - 605 014.

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

Inventory an idle resource, which is maintained by every business

system whose ultimate aim is profit maximization. To run a healthy business
inventory planning and execution must have high synchronization, but
inspite of well and organized planning the business sectors face the
challenge of handling the problems of shortage and overage. The problem
of shortage can be managed by the maintenance of safety stocks and other
suggestive measures are also proposed by several researchers. To mention a
few Harris (1913), Wilson (1934), Murdoch (1965), Lewis (1970),Gardner
and Dannenbring (1979), Silver.et.al(1979), Brown(1982), Shah (1992) have
researched the effective management of inventory at times of shortage. The
other situation where the management attention is required is at times of
overage. Overage is any inventory above the maximum planned stock level
Geoff ( 2003). Shah (1992) was the first person who used the term overage
in his classification when describing excess inventory. Bocking (1996) and
Hadley (1999) in their work demonstrated that overage have a positive
effect. Geoff in his paper has argued that overage is important because there
is evidence that, even in well-managed business a significant proportion of
the inventory is in overage at any time; evaluated twenty inventory profiles
from companies in different business sectors which showed that inventory
values in overage is between 10% and 98%; described causes and
measurement of overage The remaining of the paper is organized as follows.
Section 2 discusses the reasons for incompatible transport services which
lead to overage. Section 3 elucidates the novelty approach and customer
acquisition. Section 4 describes the model formulation. Section 5 presents a
numerical example. Section 6 presents the sensitivity analysis and the last
section concludes the proposed work.

2 Incompatible Transport Services

Geoff in his paper has briefly discussed the occurrence of overage due to
independent events, non-compliance, damping and time fence which are due
to lack of proper planning and the ignorance of corrective actions. Therefore
these causes explicitly state that overage is the enterprises inability to react
to the reduction in demand in short term. Another cause for the occurrence
of overage is the incompatibility of the transport services. In recent years the
expansion of commerce is assisted by transportation, as it facilitates to
distribute the products to a number of customers. Generally the modes of
transportation are three in number among which roadways is preferred..
Consider a situation where the manufacturer produces a product and delivers
to the retailers, who has ordered for it. The manufacturer transports the
ordered products to the retailer to satisfy his demand. Suppose if there arise
certain problems in transporting, accumulation of the products takes place
which leads to the cause of overage. This accumulation of products is called
as stagnation of products. There are many hurdles faced by an enterprise at
times of transporting the required products to the customers which are listed
as follows.
i
ii
iii
iv

Climatic barrier
Occurrences of calamity
Improper transport services
Disputes and Strikes

Climate change is almost invariably considered an issue of global interest.

The factors those leads to climate change are global warming, ozone
depletion, emission of pollutants etc. A sector that receives fairly little
(explicit) attention, however, is the transport sector. Still, it is widely known

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

takes time to restore back into normal condition transportation stoppage

takes place. This is also responsible for stagnation which leads to overage.
(iv). Disputes and Strikes
In many countries strikes hit the transport sector from time to time. Strikes
are the outcome of disputes and disagreements. The reasons for strikes are
numerous in number like increase of prices of commodities, political
problems, economic crisis, social problems and so on. The magnitude of the
effect of these strikes on transportation varies strongly according to the
circumstances of action. Carrying out transportation at times of strike is a
risky task. Therefore happenings of strikes perturb transportation.
3. The Novelty Approach and Customer Acquisition.
To maintain steady business status, the manufacturers should employ various
tactics to reduce overage to maximum stock level. But practically many
manufacturers feel that it is not possible to exercise control over overage as
it burdens them. A new approach to make benefit out of overage is proposed
in this paper. The products that raise the maximum stock to overage must be
segregated and then should be channelized by the manufacturers to make
profit out of it. One of the ways to yield gain from the products added to
overage is by using as a tool to drag new customers. Increasing the number
of customers is very significant as it enable to expand the business. To run
any sort of business, customers are very important. Attracting the customers
is in fact an art as it requires many skills. The manufacturers therefore
segregate the products that add to overage from the maximum stock and
maintains as a special inventory to sell it to the new customers at a feasible
cost. This renders two benefits to the manufacturers, one is to avoid the

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

1. The maximum stock level reaches overage due to any oneSafety

of the causes
mentioned above. 2. The acquisition cost is the cost of Stock
acquiring new
0.0

customer.3. The products are not of deteriorating type. 4. Planning horizon is

infinite.
4.2 Model Formation

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

we obtain the following

expression for total revenue per unit of time.

TRU(Q) = s(1-p)D + vpD
PC(Q) is the sum of procurement cost per cycle, Acquisition cost per cycle,
AC(Q) Renovating cost RC(Q) and holding cost.
Based on the shape of the areas in Fig.1.b. the holding cost per cycle for the
items in maximum stock is h times the area of the lower trapezoid with
parallel sides equal to

.It is given by

The holding cost for the items in special inventory

triangle of Fig 1 is

times the area of the

Multiplying the sum of the

above holding costs by the number of cycles per year,

, we obtain the

following expression of the total holding costs per unit of time.

Therefore, the total inventory cost per unit time is

TCU(Q)

The total profit per unit time TPU(Q) is obtained by TRU(Q) - TCU(Q)
TPU(Q) = s(1-p)D + vpD -

} (1)

Since p is a random variable with a known probability density by definition,

f(p), then the expected value of (1), ETP(Q), is given as
ETPU(Q)=s[1-E(p)]D

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)

0, is negative for all values of Q which implies that there

exists a unique value

. (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) =

Therefore, we get E(p) as follows from (5)

E(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

= 2000 units in Eq.(2) the maximum profit per year is

given as
ETPU(Q)=

50

=2366987 / year

0.9

50,000

30

0.1

50,000-

Suppose if we wish to compare the case of not taking

the overage into consideration, that is if p =0 (p is the

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

computers.Institute of Operations Management Annual Conference

1996 Proceedings, June 1996.
2. Brown, R., 1982. Advanced Service Parts Inventory Control.Materials
Management Systems, Inc., Vermont, USA.
3. Gardner, E., Dannenbring, D., 1979. Using optimal policy surfaces to
analyse aggregate inventory tradeoffs. Management Science 25 (8).

4. Geoff Relph., Peter Barrar.,2003., Overage inventoryhow does it

occur and why is it important? Int. J. Production Economics 8182
(2003) 163171
5. Hadley, R., 1999. Inventory management control. The Journal of the
Institute of Operations Management 25 (2).
6. Harris, F., 1913. How many parts to make at once, FactoryThe
Magazine of Management 10 (2), 135136, 152.
7. Ibrahim, A. and Hall, F. (1994) Effect of Adverse Weather Conditions
on Speed-pow-occupancy Relationship, Transportation Research
Record 1457. Transportation Research Board, Washington, DC.
8. Jones, B., Janssen, L. and Mannering, F. (1991) Analysis of the
frequency and duration of freeway accidents in Seattle. Accident
Analysis and Prevention 23(4), 239-255.
9. Khattak, A., Schafer, J. and Wang, M. (1995b) A simple procedure for
predicting freeway incident duration. IVHS Journal 2(2), 113-138.
10.Lewis, C., 1970. Coverage Analysis, Scientific Inventory Control.
Butterworths, London.
11.Murdoch, J., 1965. Coverage analysisnew technique for optimising
the stock ordering policy. Proceeding from One day Conference,
Cranfield, UK.
12.Pablo

Suarez,

William

Anderson,

Vijay

Mahal

T.R.

Lakshmanan .,2005., Impacts of flooding and climate change on urban

transportation: A systemwide performance assessment of the Boston
Metro Area Transportation Research Part D 10 , 231244
13.Shah, S., 1992. Setting parameters in MRP for effective management
of bought out inventory in a JIT assembly environment, Ph.D. Thesis,
Aston University, Birmingham,January 1992, unpublished.

14.Silver, E., Peterson, R., 1979. Decision Systems for Inventory

Management and Production Planning. Wiley, New York,1979, pp.
237 & 541.
15.Silver, E., Peterson, R., Pyke, D., 1998. Inventory Management and
Production Planning and Scheduling. Wiley, New York.
16.Wight, O., 1993. ABCD Checklist for Operational Excellence,Wiley,
New York.
17.Wilson, R.H., 1934. A scientific route for stock control,Harvard
Business Review, Number 13, 1934.

Risk Factors of Meningitis in Adults-An analysis using Fuzzy

Cognitive Map with TOPSIS
D.Manoharan
Arignar Anna Government Arts College
Karaikal

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

Meningitis is inflammation of the meninges. The meninges are the collective

name for the three membranes that envelope the brain and spinal cord (central nervous
system), called the dura mater, the arachnoid mater, and the pia mater. The meninges'
main function, alongside the cerebrospinal fluid is to protect the central nervous system.
Meningitis is generally caused by infection of viruses, bacteria, fungi, parasites, and
certain organisms. Anatomical defects or weak immune systems may be linked to
recurrent bacterial meningitis. In the majority of cases the cause is a virus. However,
some non-infectious causes of meningitis also exist [4, 9].
Bacterial meningitis is generally a serious infection. It is caused by three types of
pneumonia bacteria. Meningitis caused by Neustria meningitides is known as
meningococcal meningitis, while meningitis caused by Streptococcus pneumonia is
known as pneumococcal meningitis. People become infected when they are in close
contact with the discharges from the nose or throat of a person who is infected. About
80% of all adult meningitis are caused by N. meningitides and S. pneumonia. People over
50 years of age have an increased risk of meningitis caused by L. monocytogenes [10].
Meningitis is not always easy to recognize. In many cases meningitis may be
progressing with no symptoms at all. In its early stages, symptoms might be similar to
those of flu. However, people with meningitis and septicemia can become seriously ill
within hours, so it is important to know the signs and symptoms. Early symptoms of
meningitis broadly include: Vomiting, Nausea, Muscle pain, High temperature (fever),
Headache, Cold hands and feet, a stiff neck, severe pains and aches in your back and
joints, sleepiness or confusion, a very bad headache (alone, not a reason to seek medical
help) ,a dislike of bright lights, very cold hands and feet, shivering, rapid breathing , a
rash that does not fade under pressure. This rash might start as a few small spots in any
part of the body - it may spread rapidly and look like fresh bruises. This happens because
blood has leaked into tissue under the skin. The rash or spots may initially fade, and then
come back. 10-12% of meningitis cases in the industrialized countries are fatal.
20% of meningitis survivors suffer long-term consequences, such as brain damage,
kidney disease, hearing loss, or limb amputation. There are 2,300 cases of meningitis and
meningococcal septicemia in the UK each year. 70% of meningitis patients are aged
fewer than 5 or over 60 [9, 13].

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

[0, 1], where i is the cause node and j the effect

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

is the value of the concept i = 1 at instant t = 0.

The new values of the nodes are computed in an iterative vector-matrix

multiplication process with an activation function, which is used to map monotonically
the node value into a normalized range [0, 1]. The sigmoid function is the most used one
(Bueno & Salmeron, 2009) when the concept (node) value maps in the range [0, 1]. The
vector state Xt+1 at the instant t + 1 would be
-------------------- (3)
where Xt is the vector state at the t instant,

is the value of the i concept the t

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).

General TOPSIS process is briefly explained in the next section.

2.2.2. TOPSIS process
Let us define the set of alternatives as

criteria as

and the set of

Furthermore, let us assume a decision matrix,

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

A set of alternatives then can be preference ranked according to the

descending order of

then larger

means better alternative.

3. Selected Risk Factors of Meningitis

Previous studies on predicting Meningitis were focused either on rules to classify
a patient to a group of risk of getting pneumonia or data mining techniques that extract
rules from data to predict Meningitis risk [9]. Artificial neural networks and machine
learning techniques were investigated to predict the outcomes of patients with meningitis.
However, the previous works that have been done to predict meningitis state using FCMs
[9].

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

- Dislike of bright colours

C7

- Very sleepy

C8

- Confused/delirious

C9

- Seizures

C10

- Possibility of Meningitis

4. Implementation of the methodology to the study

Based on the experts opinion, the directed diagram (figure1.) is drawn with ten
nodes and twelve edges. The corresponding connection matrix A is given as follows:

C1

C9

C2

C8

C3

C10

C7

C4
C6

C5

Fig.1 Fuzzy Cognitive Map

The adjacency matrix A is given by
0
0

0
0
A
0
0

0
0

0 0.4 0 0 0 0 0 0.6 0.7

0 0 0 0 0 0 0
0
0
0 0 0 0 0 0 0.3 0
0.6

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

FCM dynamic results

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

w 0.5667 0.7 0.45 0.5 0.82 0.7 0.7 0.75 0.81 0

Table4

0.2660
0.0379

V 0.3012

0.3548
0.2064

0.3286 0.0845 0.0035 0.0057 0.0048 0.3286 0.0438 0.2287 0

0.0468 0.0301 0.2854 0.4681 0.0468 0.0468 0.0035 0.0091 0
0.0030 0.0962 0.0022 0.0035 0.0030 0.3721 0.0504 0.2591 0

0.0042 0.1143 0.0055 0.0089 0.0042 0.0021 0.0596 0.3081 0

0.1025 0.0659 0.1821 0.2986 0.2549 0.2549 0.0339 0.1774 0

According to the TOPSIS methodology, the positive-ideal scenario (PIS) is

calculated by the higher scores of each node and the negative-ideal scenario (NIS) is
calculated by the lower scores of each node. After applying TOPSIS algorithm, the results
are shown in table 5.

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]

Axelrod, R. (1976), Structure of Decision: The Cognitive Maps of Political Elites,

Princeton, NJ: Princeton University Press.

[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]

De Cauwer H et al. Differential diagnosis between viral and bacterial meningitis

in children, Eur J Emergency Med 2007,14(6):343.

[5]

Hwang, C. L., & Yoon, K. (1981). Multiple Attribute Decision Making-Methods

and Applications. Berlin, Heidelberg: Springer.

[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]

Kosko, B. (1986). Fuzzy cognitive maps. International Journal on Man Machine

Studies, 24.

[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.

[ 10 ] Saez-Llorens X, McCracken G:Bacterial meningitis in children. Lancet 2003,

361(9375):2139-2148.

[11]

Salmeron, J.L., Rosario Vidal, & Angel Mena, (2012). Ranking fuzzy cognitive
map based scenarios with TOPSIS. Expert Systems with Applications, 39,24432450.

[12]

Wang, Y. M., & Elhag, T. M. S. (2006). On the normalization of interval and

fuzzy weights. Fuzzy Sets and Systems, 157(8), 24562471.

[13]

World Health Organisation, Report on global surveillance of epidemic-prone

infectious disease. Tech-rep. WHO/CDS/CSR/ISR/20001127, 2000.

[14]

World Health Organisation: Meningococcal meningitis. Tech-rep. 2011,

[http//www.who.int/mediacentre/factsheets/fs141/en/].

[15]

Zadeh, L.A. (1965). Fuzzy Sets. Information and Control, 8,338-353.

OPTIMIZATION OF A SINGLE-VENDOR SINGLE-BUYER FUZZY

INVENTORY MODEL WITH DISCRETE DELIVERY ORDER,
RANDOM MACHINE UNAVAILABILITY TIME AND LOST SALES
A.George Mary
Marivanios College, Tiruvananthapuram-695 015 ,,Kerela

Abstract : Supply chain management is concerned with the coordination of

material and information flows in multi-stage production systems. Inventory
management has long been treated as an isolated function solely focused on
individual entities, taking into account concerned with single vendor-single
buyer and single-vendor multiple buyer models. Integrated single-vendor
single-buyer inventory model with multiple deliveries has proved to result in
less inventory cost. However, many researchers assumed that the production
run is perfect and there is no production delay. In reality, production delay is

prevalent due to random machine unavailability and shortages. This study

considers lost sales, and two kinds of machine unavailability distributions
uniformly and exponentially distributed. A classical optimization technique
is used to device an optimal solution and a numerical example is provided to
illustrate the theory. In this model we consider the fuzzy total cost under
crisp order quantity or fuzzy order quantity in order to extend the traditional
inventory model to the fuzzy environment. We use Function Principle as
arithmetic operators of fuzzy total cost and use the Graded Mean Integration
Representation Method to defuzzify the fuzzy total cost. Then we use the
Kuhn-Tucker Method to find the optimal order quantity of the fuzzy order
inventory model. The results show that delivery frequency has significant
effect on the optimal total cost and a higher lost sales cost will result in a
higher delivery frequency.

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

distribution models represent two different types of distribution. Uniformly

distributed means constant number of machine unavailability over a period
of time while exponentially distributed means machine unavailability may
increase with time. Both cases can occur in real life. Similar distribution
types were used by Abboud et al. (2000) and Giri and Dohi (2005).
In section 2, the methodology is introduced. In section 3, discuss with
fuzzy EOQ inventory models with different situation. A numerical example
is shown in section 4 and section 5 concludes.

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

is a generalized fuzzy number as shown in Fig.1. It is

described as any fuzzy subset of the real line R, whose membership function
A~

1.
2.

satisfies the following conditions.

A~ x

A~ x

is a continuous mapping from R to [0, 1],

= 0, - < x a1,

3.
4.
5.
6.

A~ x

A~ x

A~ x

A~ x

= L(x) is strictly increasing on [a1, a2],

= WA, a2 x a3,
= R(x) is strictly decreasing on [a3, a4],
= 0, a4 x < ,

where 0 < WA 1 and a1, a2, a3 and a4 are real numbers.

This type of generalized fuzzy numbers are denoted as
a4 ; A)LR when A = 1, it can be formed as

A%

A%

= (a1, a2, a3,

= (a1, a2, a3, a4 ; A)LR. Second,

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%

= (a1, a2, a3, a4 ; A)LR is given by

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

. . .

where 0 < h wA and 0 < wA 1.

Throughout this paper, we only use popular trapezoidal fuzzy number
as the type of all fuzzy parameters in our proposed fuzzy production
inventory models. Let

B%

B%

be a trapezoidal fuzzy number and be denoted as

= (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

Fig.1. The graded mean h-level value of generalized fuzzy number

a2

A%

R-1(h) a4

= (a1, a2, a3, a4 : wA)LR.

2.2. The Fuzzy Arithmetical Operations under Function Principle

Function Principle is introduced by Chen (1985) to treat the fuzzy
arithmetical operations by trapezoidal fuzzy numbers. We will use this
principle as the operation of addition, multiplication, subtraction and
division of trapezoidal fuzzy numbers, because (1) the Function Principle is
easier to calculate than the Extension Principle, 92) the Function Principle
will not change the shape of trapezoidal fuzzy number after the
multiplication of two trapezoidal fuzzy numbers, but the multiplication of
two trapezoidal fuzzy numbers will become drum-like shape fuzzy number
by using the Extension Principle (3) if we have to multiply more than four
trapezoidal fuzzy numbers then the Extension Principle cannot solve the
operation, but the function principle can easily find the result by pointwise
computation. Here we describe some fuzzy arithmetical operations under the
Function Principle as follows.
Suppose

%
A

= (a1, a2, a3, a4) &

%
B

= (b1, b2, b3, b4) are two trapezoidal

fuzzy numbers. Then

(1)

%
A%
B
The addition of and is
~
~
A B

= (a1 + b1, a2 + b2, a3 + b3, a4 + b4)

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)

where T = {a1b1, a2b2, a3b3, a4b4}

T1 = {a2b2, a2b3, a3b2, a3b3}
C1 = min T1, C2 = min T1, C3 = max T1, C4 = max T1
If a1, a2, a3, a4, b1, b2, b3 and b4 are all zero positive real numbers then

% 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

and b4 are any real numbers.

(4)

1
% B%-1
B
=

1
1
1
1

,
,
,
b 4 b 3 b 2 b1

where b1, b2, b3, b4 are all positive real

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)

2.3. The Kuhn-Tucker Conditions

Taha (1997) discussed how to solve the optimum solution of nonlinear
programming problem subject to inequality constraints by using the Kuhn
Tucker conditions. The development of the KuhnTucker conditions is based
on the Lagrangean method.
Suppose that the problem is given by
Minimize y = f(x)
Subject to gi(x) 0, i = 1, 2, . . . , m.
The nonnegativity constraints x 0, if any, are included in the m
constraints.
The inequality constraints may be converted into equations by using
nonnegative surplus variables. Let

Si2

be the surplus quantity added to the ith

constraint gi(x) 0. Let = (1, 2, . . . , m), g(x) = (g1(x), g2(x), . . . gm(x)) .

S , S , . . . , S .
2
1

. . and S2 =

2
2

2
m

The KuhnTucker conditions need x and to be a stationary point of

the minimization problem, which can be summarized as follows :

 0,
f(x) - g(x) = 0,

i g i (x) = 0, i = 1, 2, . . . , m.
g i (x) 0, i = 1, 2, . . . , m.

3. THE EOQ INVENTORY MODEL WITH DISCRETE DELIVER

ORDER, RANDOM MACHINE UNAVAILABILITY TIME AND
LOST SALES
In this section, we develop a sequential optimization method using
Kuhn-Tucker method. We use this method to find the Optimal Economic
Order Quantity (EOQ) of the fuzzy inventory model.

3.1. Model Formation

Assume that the unavailability time t is a random variable uniformly
distributed over the interval [0, 6]. The probability density function f(t) is
given as

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 =

Throughout this, we use the following variables in order to simplify

the treatment of the fuzzy inventory models

% 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

arithmetical operators under function principle.

Suppose

A%v

(Av1, Av2, Av3, Av4)

are the fuzzy

D%
h%
v

P%

%
C
t
h%

A%

(D1, D2, D3, D4)

(hv1, hv2, hv3, hv4)

(P1, P2, P3, P4)

(Ct1, Ct2, Ct3, Ct4)

(h1, h2, h3, h4)

(A1, A2, A3, A4)

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

Third we can get the optimal order quantity

q*NL ,

when P[

minimization. In order to find the minimization of P[

derivative of P[

%
P TUC
NL
q

] with q* is

NL

= 0.

Hence we find the optimal order quantity

] 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

3.2. Fuzzy Inventory EOQ Model with Fuzzy Order Quantity

In this section, we introduce the fuzzy inventory EOQ models by

q%
changing the crisp order quantity

be a trapezoidal fuzzy number

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

q4 q3 0 and q1 > 0. The Kuhn-Tucker conditions are i 0,

f(P[

]) - (qi) = 0

NL

igi(qi) = 0
gi(qi) = 0

These conditions simplify to the following 1, 2, 3, 4 = 0 &

f(

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

- 1(q2 q1) - 2(q3 q2) - 3(q4 q3) - 4q1 = 0

which implies

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

Because q1 > 0 and 4 q1 = 0 then 4 = 0. If 1 = 2 = 3 = 0 then q4 < q3 < q2 <

q1, it does not satisfy the constraints 0 < q1 q2 q3 q4. Therefore q2 = q1,
q3 = q2 & q4 = q3. (ie) q1 = q2 = q3 = q4 = q*.

Hence from we find the optimal order quantity

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%

= (18800, 19000, 19400, 19600)

Fuzzy annual demand rate D is more or less than 4800

D%

= (4200, 4500, 5100, 5400)

Fuzzy setup cost Av for vendor per cycle is more or less than 600

A%v

= (400, 500, 700, 800)

Fuzzy ordering cost of buyer per order A is more or less than 25

A%

= (15, 20, 30, 35)

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

= (30, 40, 60, 70)

If the number of delivery k is 6, then fuzzy order quantity

q*NL

= (179.49, 185.75, 199.14, 206.04)

The optimal total cost when the production down time is bigger than the
upper bound of the machine unavailability time is

%*
TUC
NL

= (4641.012821, 6136.485317, 9305.045038, 10942.08398)

CONCLUSION

In this paper, the machine unavailability time is assumed to be

uniformly distributed. The numerical example illustrates how the multiple
deliveries result in a lower cost than the single delivery model. The
stochastic machine time model results in a higher cost and more delivery
frequencies when compared to a perfect machine model. The proposed
model helps enterprises to optimize their profit by coordinating the number
of deliveries for various machine unavailability time and lost sales cost. For
each fuzzy model, a method of defuzzification namely the Graded Mean
Integration Representation is employed to find the estimate of total cost
function in the fuzzy sense and then the corresponding optimal order lotsize
is derived from Kuhn-Tucker Model.

REFERENCES
1.

Abboud, N.E., Jaber, M.Y., Noueihed, N.A., 2000. Economic lot

sizing with consideration of random machine unavailability time.

2.

Computers & OperationsResearch 27 (4), 335-351.

Abboud. N.E., 2001. A discrete-time Markov production-inventory
model

3.

with

machine

breakdowns.

Computers

&

Industrial

Engineering 39, 95-107.

Aghezzaf, E.H., Jamali, M.A., Ait-Kadi, D., 2007. An integrated
production and preventive maintenance planning model. European

4.

Journal of Operational Research 181, 679-685.

Banerjee, A., 1986. Adjoint economic lot size model for purchaser and

5.

vendor. Decision Sciences 17, 292-311.

Ben-Daya, M Hariga, M 2004. Integrated single vendor single buyer
model with stochastic demand and variable lead time. International
Journal of Production Economics 92,75-80.

6.

Ben-Daya, M., Darwish, M., & Ertogral, K. (2008). The joint

economic lotsizing problem. Review and extensions. European

7.

Journal of Operational Research, 185, 26-742.

Campos, L., Verdegay, J.L., 1989. Linear Programming problems and

8.

ranking of fuzzy numbers. Fuzzy sets and systems. 1-11.

Chen, S.H., 1985. Operations on fuzzy numbers with function

9.

principle. Tarnkang Journal of Management Sciences 6(1) 13-25.

Chen, S.H., Hsieh, C.H., 1998. Graded Mean Integration
Representation of Generalized fuzzy numbers. In the sixth conference
on Fuzzy Theory and its applications (CDROM), filename : 031.wdl,

10.

Chinese Fuzzy Association, Taiwan 1-6.

Chen, Y.C., 2006. Optimal inspection and economical production
quantity strategy for an imperfect production process. International

11.

Journal of Systems Science 37 (5), 295-302.

Chiu, Y.S.P, Wang, S.S, Ting, C.K., Chuang, H.J Lien, Y.L, 2008.
Optimal run time for EMQ. model with backordering, failure-inrework and breakdown happening in stock-piling time. WSEAS

12.

Transaction on Information Science & Applications 4, 475-486.

Christoph Glock. A multiple-vendor single-buyer integrated inventory
model with a variable number of vendors. Computer sand Industrial

13.

Engineering 60(201 1) 173-182.

Chung, C.J., Widyadana, G.A., Wee, H.M., 2011. Economic
production quantity model for deteriorating inventory with random
machine unavailability and shortage. International Journal of

14.

Production Research 49, 883-902.

El-Ferik. S., 2008. Economic production lot-sizing for an unreliable
machine under imperfect age-based maintenance policy. European
Journal of Operational Research 186.150-163.

15.

Ertogral. K., Darwish, M., Ben-Daya. M., 2007. Production and

shipment lot sizing in a vendor-buyer supply chain with transportation

16.

cost. European Journal of Operational Research 176, 1592-1606.

Giri, B.C.. Dohi, T., 2005. Computational aspects of an extended
EMQ. model with variable production rate. Computers & Operations

17.

Research 32, 3143-3161.

Glock, C.H. (2010a). Supply chain coordination via integrated

18.

inventory models. A review, working paper. University of Wuerzburg.

Goyal., S.K., Nebebe, F., 2000. Determination of economic
production-shipment policy for a single-vendor-single-buyer system.

19.

European Journal of Operational Research 121, 175-178.

Hill, R.M (1997). The single-vendor single-buyer integrated
production-inventory model with a generalized policy. European

20.

Journal of Operational Research. 97, 493-499.

Huang, C.K., 2004. An optimal policy for a single-vendor singlebuyer

integrated

production-inventory

problem

with

process

unreliability consideration. International Journal of Production

21.

Economics 91, 91-98.

Hui Ming Wee, Gede Augs Widyadana, 2011. Single vendor Single
Buyer Inventory Model with discrete delivery order, random machine
unavailability time and lost sales, International Journal of Production

22.

Economics doi : 10.1016/j.ijpe.2011.11.019.

Jaber, M.Y., Osman, I.H 2006. Coordinating a two-level supply chain
with delay in payments and profit sharing. Computers & Industrial

23.

Engineering 50, 385-400.

Kaufmann, A., Gupta, M.M., 1991. Introduction to fuzzy arithmetic

24.

theory and applications. Van Nostrand Reinhold.

Lee, H.M., Yao, J.S., 1998. Economic production quantity for fuzzy
demand quantity and fuzzy production quantity. European Journal of
Operational Research, 109, 203-211.

25.

Lu, L. (1995). A one-vendor multi-buyer integrated inventory model.

26.

European Journal of Operational Research, 81,312-323.

Meller, R.D., Kim, D.S 1996, The impact of preventive maintenance
on system cost and buffer size. European Journal of Operational

27.

Research 95, 577-591.

Omar. M., 2009. An integrated equal-lots policy for shipping a

28.

vendor's final
production batch to a single buyer under linearly decreasing demand.

29.

International Journal of Production Economics 118, 185-188.

Taha, H.A., 1997. Operations Research, Prentice Hall, New Jersey,

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.

Optimum Shipment Quantity for Three-Echelon Joint Model

with Fuzzy Parameter
H.Lourdu sagayarani
Santhi Nivas College,Bangalore

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.

Inventory control of goods is a very important part of logistic

systems, common to all economic sectors such as agriculture, industry,
trade and business.
Very large costs are incurred as a result of
replacement actions, shortages and the use of managerial and clerical
time in making and routinely implementing inventory management
decisions. Thus, properly designed decision rules, based on
mathematical modeling, can lead to substantial benefits. The major
problem in inventory control can be summed up by the two
fundamental questions (i) when should a replenishment order be
placed? And (ii) how much should be ordered? In the real world, the
parameters and variables in inventory model may be almost uncertain
datum. In 1987, Park [12] used fuzzy set concept to treat the inventory
problem with fuzzy inventory cost under arithmetic operations of
Extension Principle. In 1996, Chen et al. [6] introduced backorder fuzzy
inventory model under Function Principle. In [5], Chen et al. discussed
fuzzy inventory model for crisp order quantity or for fuzzy order
quantity with generalized trapezoidal fuzzy number. Chen and Wang

[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.

Economic lot size models have been studied extensively. Since

Harris [8] presented the eminent EOQ formula in 1913. Five years later,
the economic production quantity (EPQ) inventory model was
anticipated by Taft [14]. However, in recent years, both academicians
and researchers have shown an increasing level of interest in finding
alternative ways to solve inventory models. In this paper, we introduce
the fuzzy all joint optimal model in which fuzzy parameters and fuzzy
joint shipment quantity are all trapezoidal fuzzy numbers.

1.1

symbols Used

The retailers demand and cost parameters are:

~
D
~
A
~
h
~
Z

: fuzzy annual demand quantity of the retailer

: fuzzy retailers ordering cost per contract

: fuzzy stock holding cost per unit per year for the retailer

: fuzzy fixed cost of receiving a shipment from the distribution

center
~

: fuzzy cost of losing flexibility per unit per year

The distribution centers cost parameters are:

~
A
~
h

: fuzzy distribution centers ordering cost per contract

: fuzzy stock holding cost per unit per year for the distribution

center
~
Z
~
Z

1D

2D

: fuzzy fixed cost of receiving a shipment from the manufacturer

: fuzzy cost of a shipment from the distribution center to the

retailer

: fuzzy cost of losing flexibility per unit per year

The manufacturers production and cost parameters are:

~
P

: fuzzy annual production rate of the manufacturer

~
A
~
Z

: fuzzy fixed production setup cost per lot

: fuzzy cost of a shipment from the manufacturer the distribution

centre
~
h
n

: fuzzy stock holding cost per unit per year for the manufacturer

: number of shipments from the distribution cener

m2 : number of shipments per lot from the manufacturer

qR : shipment quantity using the retailers optimum model
j

: Three echelon index (j=1,2,3)

~
q
~
q

: fuzzy shipment quantity using the all-joint optimum model

: fuzzy optimal value of qj

~
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:

Demand for the item is constant over time.

Shortages are not allowed.
Time horizon is infinite.

 Ordering cost/setup cost, unit inventory cost, and the cost of a

shipment are known.
The annual production rate is larger than the demand rate.
Lead time is constant.
Costs of losing flexibility per unit per year are positive.
The distribution centers order quantity is an integer multiple of
the retailers order quantity, and the distribution center follows
the (qR, (n - 1)qR) policy.
Replenishment of the retailers inventory is instantaneous, and
the retailer follows the (s, qR) policy. Whenever the inventory
position falls to s, the retailer places an order for qR units.
2. Methodology
2.1. Fuzzy Numbers
Any fuzzy subset of the real line R, whose membership function
A satisfied the following conditions, is a generalized fuzzy number

%
A

A%(x)

is a continuous mapping from R to the closed interval [0, 1].

(1)
A%(x)

(2)

= 0, - < x a1,

A%(x)

= L(x) is strictly increasing on [a1, a2]

(3)
A%(x)

(4)

= wA, a2 x a3,

A%(x)

= R(x) is strictly decreasing on [a3, a4],

(5)
A%(x)

(6)

= 0, a4 x < ,where 0 < wA 1, and a1, a2, a3 and a4 are real

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.

2.2. Trapezoidal Fuzzy Number

The fuzzy number

a4 and defined

%
A

= (a1, a2, a3, a4), where a1 < a2 < a3 <

on R is called the trapezoidal fuzzy number, if the membership

%
A
function is given by

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

= (a1 + b1, a2 + b2, a3 + b3, a4 +

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

{a1b1, a2b2, a3b3, a4b4}

3.%
B

%
B

%
%
% A
A
B
= (-b4, -b3, -b2, -b1), then the subtraction of
and is

= (a1 - b4, a2 - b3,

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

, where b1, b2, b3 and b4 are all positive real

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

(ka 4 , ka 3 , ka 2 , ka1 ), if k < 0

2.4. Graded Mean Integration Representation Method

%
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

L-1 (h)+R -1 (h)

h
dh
2

A%
If

by graded mean integration representation

wA

h dh
0

with 0 < h wA and 0 < wA 1.

= (a1, a2, a3, a4) is a trapezoidal number, then the graded

mean integration representation of

1

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

2.5. Extension of the Lagrangean Method

In [19], Taha discussed how to solve the optimum solution of

non-linear programming problem with equality constraints by using
Lagrangean method and showed how the Lagrangean method may be
extended to solve inequality constraints. The general idea of extending
the Lagrangean procedure is that if the unconstrained optimum
problem does not satisfy all constraints, the constrained optimum must
occur at a boundary point of the solution space.
Suppose that the problem is given by
Minimize y = f(x)

Subject to gi(x) 0, i = 1, 2,. . . m.

The non-negativity constraints x 0, if any, are included in the m
constraints. Then the procedure of Extension of the Lagrangean
method involves the following steps.

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.

3. Fuzzy Mathematical Model

In this method retailers demand, manufactures production rate
and the cost parameters are represented by trapezoidal fuzzy
numbers. Let the fuzzy numbers D, AR, hR, ZR, R, AD, hD, Z1D,
Z2D,D,P,AM, ZM, hM and qj are represented as
%
D

= (d1, d2, d3, d4),

(zr1, zr2, zr3, zr4),

%
A
R

= (ar1, ar2, ar3, ar4),

h%R

= (hr1, hr2, hr3, hr4),

%
Z
R

%
R

= (r1, r2, r3, r4),

%
A
D

= (ad1, ad2, ad3, ad4),

h%D

= (hd1, hd2, hd3, hd4),

Z%1D

= (z1d1, z1d2, z1d3,z1d4),

Z%2D

= (z2d1, z2d2, z2d3, z2d4),

%
D

= (d1, d2, d3, d4),

P%

= (p1, p2, p3, p4),

%
A
M

= (am1, am2, am3, am4),

%
Z
M

= (zm1, zm2, zm3, zm4),

h%M

q%j

= (hm1, hm2, hm3, hm4),

= (qj1, qj2, qj3, qj4).

By using function principle method, fuzzy model can be expressed

as

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

By Graded Mean Integration, solve the unconstrained problem

Minimize
% n,m
P TRC
j
2
=
a d1

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

qj2 qj3 qj4.

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

(1) subject to qj 2 - qj1 = 0 by the Lagrangean

method. We have Lagrangean function is L(qj 1, qj2, qj3, qj4, ) =

% n,m
P TRC
j
2
- (qj2 - qj1). Let all the partial derivatives equal to zero
and solve qj1, qj2, qj3 and qj4, then we get

 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

2 m2 z m2 d 2 d 2 +z1d 2 +nz 2d2 d 2

nzr2 d 2
m2
m2

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 :

Convert the inequality constraint qj2 - qj1 0 and qj3 - qj2

0into equality constraints qj2 - qj1 = 0 and qj3 - qj2 = 0. We optimize
% n,m
P TRC
j
2
subject to qj2 - qj1 = 0 and qj3 - qj2 = 0 by the
Lagrangean method. Then the Lagrangean function is L(qj1, qj2, qj3,
% n,m
P TRC
j
2
qj4, ) =

- (qj2 - qj1) - (qj3 - qj2). Let all the

partial derivatives equal to zero and solve qj1, qj2, qj3 and qj4, then
we get

a m4
a
a r4

z m4 d 4 d4 +z1d 4 +nz 2d4 d 4

nzr4 d 4

m2
m2

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 :

Convert the inequality constraints qj2 - qj1 0, qj3 - qj2 0 and

qj4 - qj3 0 into equality constraints qj2 - qj1 = 0, qj3 - qj2 = 0 and qj4 % n,m
P TRC
j
2
qj3 = 0. We optimize

subject to qj2 - qj1 = 0, qj3 - qj2 =

0 and qj4 - qj3 = 0 by the Lagrangean method. The Lagrangean

% n,m
P TRC
j
2
function is given by L(qj1, qj2, qj3, qj4, ) =
(qj2 - qj1) - (qj3 - qj2) - (qj4 - qj3).

In order to find the

minimization of L(qj1, qj2, qj3, qj4, ) we take the partial

derivatives of L(qj1, qj2, qj3, qj4, ) with respect to qj1, qj2, qj3,
qj4, andand let all the partial derivatives equal to zero to solve
qj1, qj2, qj3 and qj4 then we get

 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

z m2 d 2 d2 +z1d 2 +nz 2d2 d 2

nzr2 d 2
m2
m2
m2

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

= (qj1, qj2, qj3, qj4) satisfies all

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

receiving a shipment from the distribution center is about $15, the

~
fuzzy cost of losing flexibility ( R)per unit per year is about $18, the
~
fuzzy distribution centres ordering cost ( A D) per contract is about
~
$400, the fuzzy stock holding cost( h D) per unit per year for the
~
distribution centre is about $84.6, the fuzzy fixed cost( Z 1D) of
receiving a shipment from the manufacturer is about $30, the fuzzy
~
cost of a shipment( Z 2D) from the distribution centre to the retailer is
~
about $50, the fuzzy cost of losing flexibility ( D)per unit per year is
~
about $12.69, the fuzzy annual production rate ( P ) of the
~
manufacturer is 2000 units, the fuzzy fixed production setup cost( A
~
Z M )from the
M) per lot is about $1600, the fuzzy cost of a shipment (
manufacturer the distribution centre is about $300, the fuzzy stock
~
holding cost ( h M) per unit per year for the manufacturer is about
$48.6, the number of shipments (n) from the distribution centre is 3,
the number of shipments (m 2) per lot from the manufacturer is 4.How
many car units should FORD manufacturing company produce in each
batch?

Here we use a general rule to transfer the linguistic data, about X,

into trapezoidal fuzzy numbers as about X = (0.95X, X, X, 1.05X).By
using the above rule, the fuzzy parameters in this example can be
transferred as follows:

~
D

~
= (1425, 1500, 1500, 1575); A

= (380,400,400,420);

~
h
~

~
h
~
Z

~
Z

= (17.1, 18, 18, 18.9);

= (14.25, 15, 15, 15.75);

~
A

= (47.5, 50, 50, 52.5);

= (1900, 2000, 2000, 2100);

M

= (285,300,300,315);

~
h

= (380,400,400,420);
~
Z

= (80.37, 84.6, 84.6, 88.83);

2D

~
P
~
Z

= (228,240,240,252);

1D

= (28.5, 30, 30, 31.5);

= (12.0555, 12.69, 12.69, 13.3245)

~
A

= (1520, 1600, 1600, 1680);

= (46.17, 48.6, 48.6, 51.03); n =3

and m2 =4
Replace the above fuzzy parameter values into formula (2), we find the
optimal fuzzy
Shipment quantity

~
q

= (101.97, 101.97, 101.97, 101.97)

(102,102,102,102)
Conclusion

In the fuzzy environment, it may be possible and reasonable to discuss

the fuzzy three-echelon joint model with trapezoidal fuzzy number for
fuzzy three echelon joint shipment quantity. In addition, we find that
*
*
*
*
*
q%
j = qj, qj, qj, qj

the fuzzy joint optimal shipment quantity

is the

special type of trapezoidal fuzzy number. Hence the optimal solution of

our suggested model can be effective to meet the conventional threeechelon joint model. Thus this fuzzy three-echelon joint model is
practical and very useful in the real world. The three-echelon-joint
model will result in tremendous savings in the total system cost for the
value chain. When the manufacturer, the distribution center, and the
retailer belong to the profit center of the same company, they can
choose the three-echelon-joint model. This model can be expanded it
to multiple manufacturers, one distribution, and multiple retailers. The

proposed integrated model is suitable for either an independent just in

time (JIT) system or profit centers from the main office. The result of
the example applying this model is quite feasible and satisfactory. This
model could be applied in more practical and sophisticated cases. For
a special case that all variables are set as real numbers, the result will
be the same as the traditional non-fuzzy model.

References
[1] Apte, U. M., & Viswanathan, S. (2000). Effective cross docking for
improving
distribution efficiencies. International Journal of Logistics, 3, 91
302.

[2] Arora, V., Chan, F. T. S., & Tiwari, M. K. (2009). An integrated

approach for logistic
and vendor managed inventory in supply chain. Expert Systems with
Applications,8,8. doi:10.1016/j.eswa.2009.05.016.

[3] S.H.Chen,C.H.Hsieh, Graded Mean Integration Representation of

Generalized fuzzy
number, Journal of Chinese Fuzzy Systems. 5(2) (1999) 1 - 7.

[4] S.H. Chen, Operations on fuzzy numbers with function principle,

Tamkang Journal
of Management Sciences 6(1) (1985), 13-26.

[5] S.H. Chen, C.C. Wang, A. Ramer, Backorder fuzzy inventory model
under

function principle, Information Sciences, an International Journal 95

(1996) 7179.

[6] S.H. Chen, C.H. Hsieh, Optimization of fuzzy simple inventory

models,In: IEEE International Fuzzy System Conference Proceedings,
Seoul, Korea,vol. 1,1999, pp. 240244.

[7] Chih Hsun Hsieh, (2002). Optimization of fuzzy production

inventory models,
Information, Sciences, 146. P.N.29-40.

[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

[10] Kreng, B.V,&Chen, F. T. (2007). Three echelon buyersupplier

delivery policy A
supply chain collaboration approach. Production Planning and
Control, 18(4),338 349.

[11] Mirko Vujosevic, Dofrila Petrovic and Radivoj Petrovic, EOQ

Formula
whenInventoryCost is fuzzy, Int. J. Production Economics 45(1996)
499-504

[12] K.S. Park, Fuzzy set theoretic interpretation of economic order

quantity, IEEE
Transactions on Systems, Man, and Cybernetics SMC-17 (1987)
10821084.

[13] Shan-Huo Chen and Chien-Chung Wang,Backorder Fuzzy Inventory

Model under
Function Principle, Information Science 95,(1996) 71-79

[14] E.W.Taft, The most economical production lot, The Iron Age101
(1918)1410 1412.[15]

[15]H.A. Taha in : Operations Research, Prentice Hall, Englewood

Cliffs, NJ, USA,1997, pp.753777.

[16] Toni, A. D., & Nassimbeni, G. (2000). Just-in-time purchasing: an

empirical study of
operational practices, supplier development and performance.
Omega, 28,

[17] Van Houtum, G. J., Inderfurth, K., & Zijm, W. H. M. (1996).

Materials coordination
instochastic multi-echelon systems. European Journal of
Operational Research,95,1-23.

[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

[19] Wagner, S. M. (2006). A firms responses to deficient suppliers and

competitive
advantage. Journal of Business Research, 59, 686695.631651.

[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.

MACHINE REPAIR PROBLEM UNDER UNCERTAIN ENVIRONMENT

Sreelekha menon.B**

**Department of Mathematics, SCMS School of engineering and technology,

Kerala -683 582

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.

2. FUZZY MACHINE REPAIR MODEL

Consider a conventional machine interference model that consists of K
machines
and R repair crews. At any instant in time, a particular machine is in either
good or bad condition. When a machine breaks down, one of R available
repair crews is called upon to do the repair. Normally a repair person is in
charge of more than one machine (i.e., R < K), but a repair person can repair
at most only one machine at one time. When a machine breaks at the period
of time that all repair persons are busy, it has to wait and is interfered with by
the machine being repaired. Once a machine is repaired, it returns to good
condition and is again susceptible to breakdown.
The length of time that a machine remains in good condition follows an
exponential distribution with breakdown rate x (or the rate of breakdown per
machine is x breakdowns per unit time), and the time it takes to complete
repairs on a broken machine is assumed exponential with service rate y (or
mean repair time is 1/y). In fact, the machine interference problem is a
typical example of the queuing models with finite call ing population, which is
denoted as M/M/R/GD/K/K using KendallLee_s notation Suppose the
breakdown rate x and service rate y are approximately known and can be
represented by convex fuzzy sets

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.

Mixed integer programming approach

~~ ~
f ( , , )

One approach to construct the membership function

~
~
~
~

deriving the cuts of , Denote the cut of ,

are

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 ( , , )

Clearly defined membership function of

is also parameterised by .
Consequently we can use its cut to construct the corresponding membership function.
~~ ~
f ( , , )
is also parameterised by . We can use it as cut to construct the corresponding
membership function.
~ ( x ), ~ ( y)

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)

, which is equivalent to find the lower boundary

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 )

, which is equivalent to find the lower bound

bound

of the cuts of

L ( z)

and upper

x x L

~ (x)

x x U

Since the requirement of

represented by
or
L
U
x 1x (1 1 )x
can be formulated as the constraint of
where 1 = 0 or 1.

this

Similarly,
2 y L (1 2 )y U
y =
x
by

Moreover, from the definition of in (1) and x y can be replaced

x L , x U

y y L , y U

upper bound
L

= min
L
1

L
L1

of cuts of
L

, L 2

can be constructed via finding the lower bound

. 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

. Consequently, considering both of these two cases above,

L ( z)

the membership function

, 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-decreasing with respect to and

is non-

and

LS (z) = L

are invertible with respect to , then the left shape function

and a right shape function

L ( z)
Membership function
is constructed as

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

Other membership functions of performance measures can be derived in the similar

manner.
4. Numerical example

To demonstrate the validity of the proposed approach, a numerical example

inspired by Gross and Harris is solved.

The W.E .Finish machine shop company has four turret lathes.These machines
break down periodically and the company has two repair man to service the
lathes when they break down.When a lathe is fixed , the time until the next
breakdown is exponentially distributed with a fuzzy rate that can be
represented by a trapezoidal fuzzy number

= [ 1, 2 , 3 , 4] per day. The service time for each service is exponentially

distributed with a fuzzy rate that can be represented by a trapezoidal fuzzy
number

= [ 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.

The interest performance measures FM/FM/2/GD/4/4 are given as

j

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 )

Let the break down and service rate be

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]

Thus following mixed integer non linear parametric 0 1 programming is

constructed deriving the membership function of

~
L

4xy3 + 12x 2 y 2 + 18x 3y + 12x 4

y 4 4xy 3 + 6x 3 y + 6x 2 y 2 + 3x 4

LL
= Min [

1 x 4

s.t

6 y 10

4xy 3 + 12x 2 y 2 + 18x 3 y + 12x 4

y 4 4xy3 + 6x 3 y + 6x 2 y 2 + 3x 4

LU
= Max [

1 x 4

s.t

6 y 10

L
The shortest

occurs when x = + 1 and y = 10-

LL
=

4( 1)(10 ) 3 12( 1) 2 (10 ) 2 18( 1) 3 (10 ) 12( 1) 4

(10 ) 4 4( 1)(10 ) 3 6( 1) 3 (10 ) 6( 1) 2 (10 ) 2 3( 1) 4

In the maximum response ,on the contrary , x = 4 - and

y=6+

LU
=

4(4 )(6 ) 3 12(4 ) 2 (6 ) 2 18(4 ) 3 (6 ) 12(4 ) 4

(6 ) 4 4(4 )(6 ) 3 6(4 ) 3 (6 ) 6( 4 ) 2 (6 ) 2 3(4 ) 4

It is a challenging task to find their inverse functions analytically. Owing to

the complicated form of the objective function,it is impossible to represent
L

L
the optimal solution

L
and

interms of .Consequentely,a closed form

~ (z )

L
membership function for

cannot be obtained. It However we

~ (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]

K.E. Stecke, J.E. Aronson, Review of operator/machine interference

models, International Journal of Production Research 23 (1985)
129151.

[2]

D.I. Cho, M. Parlar, A survey of maintenance models for multi-unit

systems, European Journal of Operational Research 51 (1991) 1
23.

[3]

D. Gross, C.M. Harris, Fundamentals of Queueing Theory, third ed.,

John Wiley, New York, 1998.

[4]

E.A.

Elsayed,

An

optimum

repair

policy

for

the

machine

interference problem, Journal of the Operational Research Society

32 (1981) 793801.
[5]

D.

Gross,

J.F.

Ince,

The

machine

repair

problem

with

heterogeneous populations, Operations Research 29 (1981) 532

549.
[6]

K.H. Wang, Profit analysis of the machine-repair problem with a

single service station subject to breakdowns, Journal of the
Operational Research Society 41 (1990) 11531160.

[7]

K.H. Wang, Profit analysis of the M/M/R machine repair problem

with spares and service breakdowns, Journal of the Operational
Research Society 45 (1994) 539548.

[8]

G.I. Falin, A multiserver retrial queue with a finite number of

sources of primary calls, Mathematical and Computer Modelling
30 (1999) 3349.

[9]

J.C. Ke, K.H. Wang, Cost analysis of theM/M/R machine repair

problem with balking, reneging, and server breakdowns, Journal of
the Operational Research Society 50 (1999) 275282.

[10]

J.B. Atkinson, Some related paradoxes of queuing theory: new

cases and a unifying explanation, Journal of the Operational
Research Society 51 (2000) 921935.

[11]

Y.F. Hung, I.H. Chen, Dynamic operator assignment based on

shifting machine loading, International Journal of Production
Research 38 (2000) 34033420.

[12]

W.K. Ching, Machine repairing models for production systems,

International Journal of Production Economics 70 (2001) 257266.

[13]

M.J. Armstrong, Age repair policies for the machine repair

problem, European Journal of Operational Research 138 (2002)
127141.

[14]

C.R. Schultz, Spare parts inventory and cycle time reduction,

International Journal of Production Research 42 (2004) 759776.

[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]

J.J. Buckley, Elementary queueing theory based on possibility

theory, Fuzzy Sets and Systems 37 (1990) 4352.

[18]

D.S. Negi, E.S. Lee, Analysis and simulation of fuzzy queues, Fuzzy
Sets and Systems 46 (1992) 321330.

[19] J.B. Jo, Y. Tsujimura, M. Gen, G. Yamazaki, Performance evaluation

of network models based on fuzzy queuing system, Japanese
Journal of Fuzzy Theory and Systems 8 (1996) 393 408.
[20] C. Kao, C.C. Li, S.P. Chen, Parametric programming to the analysis
of fuzzy queues, Fuzzy Sets and Systems 107 (1999) 93100.

[21]

J.J. Buckley, T. Feuring, Y. Hayashi, Fuzzy queueing theory

revisited, International Journal of Uncertainty, Fuzziness and
Knowledge-Based Systems 9 (2001) 527537.

[22]

R.R. Yager, A characterization of the extension principle, Fuzzy

Sets and Systems 18 (1986) 205217.

LINEAR PROGRAMMING APPROACH FOR TWO ECHELON SUPPLY CHAIN

WITH MULTI-PRODUCT AND MULTI-TIME PERIOD

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.

2. Problem Description, Assumptions and Notations

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.

All the objective functions and constraints are linear equations.

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.

Toll fees considered per shipment.

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

inventory level of product l by warehouse i in period h.

TWhlij

units distributed of product l from warehouse i to distributor j in period h.

IDhlj

Inventory level of product l by distribution centers j in period h.

TDhljk

units distributed of product l from distribution centers j to retailer k in period h.

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,

for light weight 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,

for light weight

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

Capacity of ith warehouse in period h.

Capacity of jth distribution centers in period h.
Capacity of kth retailer in period h.

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

Figure 1 : Two-echelon Supply Chain Model

First Stage Problem

The objective function for the first stage is given below :
N w N pr

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

ICW hli , TCW hlji , PC hli , TFW hij

where

denote the cost coefficients.

In the first stage we have the following constraints

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.

ie. Inventory at the warehouse should be less than or equal to the

warehouse capacity in period h.

. . . (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)

ICD hlj , TCD hljk , TFD hj

where

denote the cost coefficients.

In the second stage we have the following constraints

1)

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 distribution centers ND

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

Warehouse Capacity in Units

Packing Charge

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

Transportation Cost in Rupees from Warehouse to

D1

D2

1000

1400

Table 3 : Toll fees from Warehouse to Distribution Centers per shipment in Rs.
Distribution Centers

Toll Fees from Warehouse to Distribution Centers

D1

D2

1770

2090

Table 4 : Distribution Centers Demand

Period

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

Table 5 : Input Data for Distribution Centers

Distribution Centers
D1

D2

Inventory carrying cost per unit in Rs.

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

Table 7 : Toll fees from Distribution Centers to Retailers per shipment

in Rs.

Distribution Centers
D1

Retailers
Nr1

Nr2

Nr3

Nr4

170

250

430

560

D2

510

595

170

340

Table 8 : Retailer Demand

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

Table 8a : Input Data for Retailers

Retailers

Retailers Capacity in Units

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

Table 10 : Optimal number of units transferred from warehouse to distribution centers

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

Table 11 : Optimal number of units transferred from Distribution Centers to Retailers

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

Table 12 : Optimal inventory in units at distribution centers

Period

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

Table 13 : Optimal number of shipments from warehouse to distribution centers

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

Table 14 : Optimal number of Shipments from Distribution Centers to Retailers

Distribution

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

Table 15 : Optimal Total Cost in Rs.

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

4.3. Computation Analysis

The Optimal Total Cost in Table 15 shows that there is 2.25% significant decrease in total cost of
the two echelon, multi-product, multi-time period supply chain due to the usage of eco-friendly light weight
plastic pail rather than the traditional one. The following graph proves that the total cost incurred while
using light weight pail is comparatively less than the traditional one. So it is better to adopt eco- frielndly
light weight pail in future.

totalcost comparision between traditional and light weight pail

10
9.5
totalcost

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

Engineering 56, pp. 708-723.

Wang H.S., 2007, A two-phase ant colony algorithm for multi echelon defective supply chain network design, European
Journal of Operation Research, doi: 10.1016/j.ejor.2007.08.037.
Williams J.F., 1981, Heuristic techniques for simultaneous scheduling of production and distribution in multi-echelon
structures: theory and empirical comparisons, Management Science 27 (3), pp. 336-352.
Xu N., Nozick L., 2009, Modeling supplier selection and the use of option contracts for global supply chain design,
Computers & Operations Research, 36, pp. 2786-2800.
Yan H., Yu Z., Cheng T.C.E, 2003, A strategic model for supply chain design with logical constraints: formulation and
solution, Computers & Operations Research 30 (14), pp. 2135-2155.
Ylmaz P., 2004, Strategic level three-stage production distribution planning with capacity expansion, Unpublished
Master Thesis, Sabanc University Graduate School of Engineering and Natural Sciences, pp. 1-20, in Turkish.
You, F. and E. Grossmann, 2008, Design of responsive supply chain under demand uncertainty. Computers & Chemical
Engineering 32 (12), pp.3090-3111.

EPQ MODEL WITH IMPERFECT QUALITY ITEMS WITH ALLOWABLE

SHORTAGES USING FUZZY NUMBERS

Dr.Senthil Thilak and D.Swathi,

NIT ,Bangalore

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.

Lead Time is Zero.

6.3. Notations
K

Production rate

Demand rate

Percentage of defective

Screening rate

Screening cost per item

selling price of perfect quality item

selling price of imperfect quality item s > v.

production quantity

y2

shortage quantity

shortage cost / unit

inventory holding cost / unit / unit time

set up cost

production cost / unit

cycle length

fuzzy inventory holding cost

fuzzy production rate

fuzzy demand rate

fuzzy defective rate

fuzzy shortage cost

fuzzy set up cost

h%

K%
r%

p%
b%

K%


P TP(y)

TP(y)
- defuzzified value of

6.4. Crisp Mathematical Model

Based on the assumptions, the situation is shown on inventory-time diagram in Fig.6.4.1.

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

(K + pr)y x(K + pr)

. . . (6.3)

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

6.5. Fuzzy Mathematical Model

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)

Differentiating (6.9) partially with respect to y2 and equating to zero we get

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 =

=cy (say) (6.10)

P TP(y)
y 22

y 02 = cy 0
is negative.

y 02
To determine y0, we substitute the value of

in (6.9) and differentiate with respect to y and equate to zero.

(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)

Differentiating (6.11) with respect to y and equating to zero we get

y2 =

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 )

(K1 -r3 (1-p1 )

K1

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(K 3 -r1 (1-p3 )

(K1 -r3 (1-p1 )

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

K11 = K12 = K13 = K1 ; K1 = K 2 = K 3 = K ; h1 = h 2 = h 3 = h ;

If we take

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)

and equation (6.12) becomes

y =

which are exactly the results of the crisp model.

6.6. Numerical Example

Consider a situation with the following parameters.
Production rate, K

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.

Selling price of good quality item, s=Rs.50/unit.

Selling price of imperfect quality item, v
Fraction of imperfect quality item, p =

=Rs.20/unit.
0.02

%
K
The optimum order quantity is found to be 4829 units. In fuzzy model if

P%
16400),

= (0.015, 0.02, 0.025),

K%

%
V
= (29950, 30000, 30050),

= (16200, 16300,

h%
= (990, 1000, 1010),

= (3.24, 3.25, 3.26) then optimum order quantity y 0 = 4828. There is

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.