You are on page 1of 18

Information Sciences 168 (2004) 7794

www.elsevier.com/locate/ins

Measuring exibility of computer


integrated manufacturing systems using
fuzzy cash ow analysis
Cengiz Kahraman a, Ahmet Beskese b,*
, Da Ruan c

a
Department of Industrial Engineering, Istanbul Technical University, 34367 Macka,
Istanbul, Turkey
b
Department of Industrial Engineering, University of Bahcesehir, 34538,
Bahcesehir, Istanbul, Turkey
c
Fuel Research Unit, Belgian Nuclear Research Centre (SCK CEN), Boeretang 200,
2400 Mol, Belgium
Received 16 June 2003; received in revised form 29 August 2003; accepted 10 November 2003

Abstract
There exist a number of methods proposed in the literature to quantify manufac-
turing exibility in monetary terms and to use a nancial evaluation model with a
decision criterion based on present worth. However, most of these methods are unable
to handle problems with incomplete and uncertain data. To obtain a sensible result in
quantifying the manufacturing exibility in computer integrated manufacturing sys-
tems, this paper proposes some fuzzy models based on fuzzy present worth. The fuzzy
models based on present worth are basically engineering economics decision models in
which the uncertain cash ows and discount rates are specied as triangular fuzzy
numbers. To build such a model, fuzzy present worth formulas of the manufacturing
exibility elements are formed. Flexibility for continuous improvement, exibility for
trouble control, exibility for work force control, and exibility for work-in-process
control are quantied by using fuzzy present worth analysis. Formulas for both ina-
tionary and non-inationary conditions are derived. Using these formulas, more reliable
results can be obtained especially for such a concept like exibility that is described in
many intangible dimensions. These models allow experts linguistic predicates about
computer integrated manufacturing systems.
 2003 Elsevier Inc. All rights reserved.

*
Corresponding author. Tel.: +90-212-669-6523/1251; fax: +90-212-669-4398.
E-mail address: beskese@bahcesehir.edu.tr (A. Beskese).

0020-0255/$ - see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/j.ins.2003.11.004
78 C. Kahraman et al. / Information Sciences 168 (2004) 7794

Keywords: Fuzzy cash ow analysis; Computer integrated manufacturing systems;


Fuzzy numbers; Fuzzy intervals; Manufacturing exibility

1. Introduction

Computer integrated manufacturing (CIM) systems can be viewed as a total


system, which provides an automatic link between product design, manufac-
turing engineering, and the factory oor. Eective implementation of advanced
manufacturing technologies through a CIM system is the cornerstone of fac-
tory modernization. CIM systems provide many important benets such as
greater process exibility, reduced inventory, reduced oor space, faster re-
sponse to shifts in market demand, lower lead times, and a longer useful life of
equipment over successive generations of products. Manufacturing exibility
may be dened as the ability to cope with changing circumstances or instability
caused by the environment. Flexibility is emerging as one of the key compet-
itive strengths in todays manufacturing systems, since it provides a critical
measure of total manufacturing performance. Hence, measuring the exibility
of manufacturing systems is important to operations managers engaged in
decision making on strategic issues related to exibility. Flexibility is widely
recognized as a multidimensional attribute. Most of the studies reported in the
literature have focused on measuring separate dimensions independently and
also have tended to be non-nancial in nature. As a result, many of these
measures have only limited application in strategic decision making.
There exist various methods to measure the value of exibility by using
surrogate measures to represent the intangible parts. However, managers still
prefer the methods by which all intangible parts of exibility can be quantied
as far as it is possible in monetary terms. This preference leads the decision-
makers into an uncertain economic decision environment of the contemporary
business world, where an experts knowledge about the cash ow information
usually consists of a lot of vagueness instead of randomness. For example, to
describe a sales prot that may be implicitly forecasted from past incomplete
information, linguistic descriptions like around one million are often used.
The major contribution of the fuzzy approach used in this paper is its capa-
bility of representing vague knowledge in such an economic environment. In
real word applications, precise data concerning exibility factors of CIMs are
not available or very hard to be extracted. In addition, decision-makers prefer
natural language expressions rather than sharp numerical values in assessing
exibility parameters. So, manufacturing exibility is an inherently fuzzy no-
tion, which can be measured by the synthesis of its constituents. Fuzzy logic
oers a systematic base in dealing with situations, which are ambiguous or not
well dened. Indeed, the uncertainty in expressions such as low exibility or
C. Kahraman et al. / Information Sciences 168 (2004) 7794 79

high utilization, which are frequently encountered in the exibility literature,


is fuzziness.
Tsourveloudis and Phillis [29] are the only researchers using a fuzzy logic
framework to measure manufacturing exibility. They propose a fuzzy rule-
based method that handles imprecise data and knowledge about a production
system. However, there is no paper published in refereed international journals
aimed at measuring exibility in terms of fuzzy cash ows.
With the rapid expansion of CIM and exible manufacturing systems
(FMS), there are hundreds of studies on the denition, classication, and
quantication of manufacturing exibility and on decision-making tools for
investment analysis of exible automation. Nonetheless production managers
of many rms, under pressure to make automated manufacturing systems more
exible, are nding that there is still a considerable gap between the benets
promised by the nancial evaluation tools and the benets realized in practice.
As a value-added to the literature on the topic, this paper aims at providing
practitioners with a fuzzy point of view to the traditional cash ow analysis
method for dealing quantitatively with imprecision or uncertainty and at
obtaining a fuzzy exibility quantication from this point of view which will
close this gap considerably. Fuzzy intervals will be used to represent the cash
ows and the other non-monetary parameters of exibility. Beskese et al. [1,2]
provide a basis for this study.
The paper is organized as follows: Section 2 gives a literature review for the
quantication of manufacturing exibility. Section 3 includes a general
knowledge about fuzzy present worth analysis, two applications of this analysis
for measuring manufacturing exibility, and a numerical example. Section 4
concludes the current research results and future plan of the study reported
in the paper.

2. Quantication of manufacturing exibility

The economic justication of CIM has received increased attention in recent


years. Arbitrarily high hurdle rates, comparison with the status quo, and
insucient benet analysis are cited as major eorts in application. However,
traditional methods for the economic justication of new manufacturing
technologies fail to include benets such as better quality, greater exibility and
reduced work-in-process, since these benets are dicult to measure and
therefore hard to quantify. Application of traditional capital budgeting
methods, for example, does not fully account for the benets arising from
increased exibility. Some special tools must be used to facilitate a thorough
analysis of tangible and intangible benets.
In the literature, many researchers have tried to quantify the manufacturing
exibility and to integrate it with decision-making tools. Their methods of
80 C. Kahraman et al. / Information Sciences 168 (2004) 7794

quantifying the exibility can be classied into three groups: (1) there are many
intangible parts of exibility, so it should be considered as blackbox [11,18,21];
(2) the intangible parts of exibility that cannot be quantied in monetary
terms can be measured by a surrogate value [8,27,30,32]; and (3) all intangible
parts of exibility can be quantied as far as it is possible in monetary terms
[12,25,26,28].
Pyoun and Choi [25] distinguish the exibility that is inherent in the man-
ufacturing system from the exibility that the user can attain after imple-
mentation. They introduce the concepts of potential exibility and realizable
exibility to obtain a deeper insight into exibility and to give consideration to
all possible aspects of user experience and capability. Potential exibility is the
exibility inherent in a manufacturing system from the system manufacturers
point of view, before it is implemented and operated by the user. Realizable
exibility is the exibility that will be realized by operating the manufactur-
ing system using both potential exibility and the users engineering and
management capability. Pyoun and Choi [25] classify potential exibility into
four elements and realizable exibility into 11 elements. They propose a sys-
tematic procedure for quantifying realizable exibility in monetary terms and
use a nancial evaluation model with a decision criterion based on present
worth.
Although such eorts are highly appreciated, one should be aware that there
are some diculties to quantify manufacturing exibility by only using crisp
methodologies. Some of them can be mentioned as:  there is no eective
method for the synthesis of the functional parameters aecting each type of
exibility and the exibility types, which are observed on dierent hierarchical
levels and are crucial in the determination of manufacturing exibility,  there
is no correspondence between exibility and the physical characteristics of the
production system, and  certain operational characteristics have contradictory
eects on exibility. To overcome such diculties, this paper proposes a fuzzy
exibility quantication method using fuzzy present worth analysis based on
fuzzy interval arithmetic.

3. Fuzzy present worth analysis in exibility measurement

3.1. Fuzzy present worth analysis

Most economic decision problems involve the uncertainty feature of cash


ow modeling. If sucient objective data is available, probability theory is
commonly used in modeling cash ows and performing decision analysis.
Unfortunately, decision-makers rarely have enough information to perform
the decision analysis, since probabilities can never be known with certainty and
the economic decision is attributable to many uncertain derivations. In this
C. Kahraman et al. / Information Sciences 168 (2004) 7794 81

p (x)

1.0
f1 (.) f2 (.)
y

0.0 a b c X

a+(b-a) y c+(b-c) y
Fig. 1. A triangular fuzzy number, Pe a; b; c.

situation, most decision-makers rely on experts knowledge in modeling cash


ows.
In the literature, triangular and trapezoidal fuzzy numbers that are the
special forms of LR-type fuzzy numbers are usually used to capture the
vagueness of the parameters related to the topic. The arithmetic operations of
these types of fuzzy numbers can be found in [33]. In this paper, triangular
fuzzy numbers (TFNs) will be used to consider the fuzziness of the exibility
parameters. A TFN is designated as Pe a; b; c. It is graphically depicted in
Fig. 1 in which f1  is the left side, and f2  is the right side representation of
the TFN.
To deal quantitatively with imprecision or uncertainty, fuzzy set theory is
primarily concerned with vagueness in human thoughts and perceptions. As an
alternative to conventional cash ow models where cash ows are dened as
either crisp numbers or risky probability distributions, Chiu and Park [7]
propose an engineering economics decision model in which the uncertain cash
ows and discount rates are specied as triangular fuzzy numbers. They
examine deviation between exact present worth (PW) and its approximate form
(PWA) and perform the fuzzy project selection by applying dierent domi-
nance rules as shown in Eqs. (1) and (2) respectively. The result of the exact
present worth is also a fuzzy number with a non-linear membership function. It
is in complex non-linear representations that require tedious computational
eort [7]. For the reason of simplicity, a TFN can be used as an approximate
form of the complex (exact) present worth formula in Eq. (1).
" !
X N
maxfFt ; 0g
ly ly
minfFt ; 0g
PW Qt ry
Qt ly
;
t0 t0 0 1 Rt0 t0 0 1 Rt0
!#
X N
maxfFt ; 0g
ry ry
minfFt ; 0g
Qt ly
Qt ry
1
t0 t0 0 1 Rt0 t0 0 1 Rt0

ly ry
where Ft is the left side representation, Ft is the right side representation of
the fuzzy cash ow Fe at time t, and Rt0 is the left side representation, Rt0 is
ly ry
82 C. Kahraman et al. / Information Sciences 168 (2004) 7794

the right side representation of the fuzzy interest rate R e at time t0 . N is a crisp
number denoting the project life.
ly
When the degree of membership (y) in Eq. (1) is equal to 0, Ft ft0 ,
ry ly ry
Ft ft2 , Rt0 rt0 , and Rt0 rt2 . When the degree of membership (y) in Eq.
ly ry ly ry
(1) is equal to 1, Ft Ft ft1 , and Rt0 Rt0 rt1 . Substituting these to
the exact present worth formula, the approximate form of the present worth
formula can be derived as in Eq. (2). PWA is represented using its three
parameters and it is easier to implement because they are in linear represen-
tations.
X N  X N
maxfft0 ; 0g minfft0 ; 0g ft1
PWA Qt Qt ; Qt ;
t0 t0 0 1 r 0
t2 t0 0 1 r 0
t0 t0 t0 0 1 rt 0 1
X N  !
maxfft2 ; 0g minfft2 ; 0g
Qt Qt 2
t0 t0 0 1 rt0 0 t0 0 1 rt0 2

Chiu and Park [7] compute the maximum deviation as a measure of the tness
between PW and PWA. They use very small increments of y as the measure-
ment method instead of derivative method since the latter is dicult to cal-
culate. Using a simulation software, they calculate the deviations for dierent
ranges of cash ows and discount rates, and nd out that the deviations are not
signicant unless the condent width of discount rate is larger than an absolute
range of 4%. In the real world applications, when the discount rates are
usually estimated within the width of 4%, PWA can be used in project
analysis. The deviations of PW and PWA are depicted in Fig. 2.
In literature, some other approaches to calculate fuzzy PW can be found.
For instance, Buckley [4] forms the membership function for the fuzzy present
worth, P fW N , as in Eq. (3)
h i
W N pwN 1 ; fN 1 yjP f
lxjP f W N =pwN 2 ; pwN 2 =fN 2 yjP f
W N ; pwN 3 3

where N is the crisp useful life of the project, pwNi is the least, the most, and the
largest possible values of P f W N respectively for i 1; 2; 3, pwN 1 < pwN 2 <
pwN 3 , f1 yjP fW N is a continuous monotone increasing function of y
lxjP fW N for 0 6 y 6 1 with f1 0jP f W N pwN 1 and f1 1jP f
W N pwN 2 , and
f
f2 0jP W N is a continuous monotone decreasing function of y for 0 6 y 6 1
with f2 0jP fW N pwN 3 and f2 1jP f W N pwN 2 .
lxjP fW N is determined by
h i
fNi yjP f
W N fi yj Fe 1 fk yj~r N 4

for i 1; 2 where k i for negative Fe and k 3 i for positive fuzzy future


worth, Fe [4]. In Eq. (4), the fuzzy interest rate (~r) is assumed to be kept con-
C. Kahraman et al. / Information Sciences 168 (2004) 7794 83

PW
(PW)
PWA
1.0
dl dr
y
l

0.0 a c PW
b

Fig. 2. Deviation between PW and PWA.

stant whereas it is a parameter changing from year to year in Chiu and Parks
[7] formula in Eq. (1). In other words, ~r in Eq. (4) represents the average annual
fuzzy interest rate along N years.
Assume that the project life N e is fuzzy. If Fe is the nal amount in the
account, then its membership function is dened by [4]
lxj Fe suph; 5
Cx

where
e ; lvj~r; lwj N
h minlujP W e 6
w
Cx fu; v; w j u1 v xg 7
The denition of Fe is simply the extension principle applied to Eq. (8) [4].
N
PW1 r F 8
For some relevant publications on fuzzy capital budgeting techniques, the
reader is referred to [3,5,9,13,1517,19,22].

3.2. Application of fuzzy present worth analysis for measuring exibility

In the literature, some publications can be found on measuring exibility


using crisp present worth analysis. A signicant contribution to this research
area by Pyoun and Choi [25], and the fuzzication of this model are given in
the following sub-sections.

3.2.1. Pyoun and Chois quantication of exibility value


Pyoun and Choi [25] classify realizable exibility into three categories: (1)
the internal control group, (2) the investment policy group, and (3) the mar-
keting adaptation group. The exibility value is the sum of the values in these
three categories. All of the formulas of the elements of these three groups
are based on present worth. Therefore, we have decided to develop fuzzy
present worth formulas for the elements of the internal control group only, to
84 C. Kahraman et al. / Information Sciences 168 (2004) 7794

constitute an example. Similar transformations can be applied easily to the


elements of the other two groups. The exibility formulas of the internal
control group are given in Table 1. In the formulas in Table 1, t is the annual
income tax rate, r is the interest rate and N is the project life. Pyoun and Choi
[25] use the exibility value (FV) in their nancial evaluation model as shown
in Eq. (9).
( )
X
N
1 tRn In 1 tOn tDn
PWr n n n
n1
1 r 1 rn 1 1 r 1 r
tBN 1 tS
FV 9
1 rN 1 rN

where t is the income tax rate, N is the project life, R revenue, I investment
cost, O operating cost, D depreciation cost, B book value, and S salvage
value.
In this paper, only the fuzzy exibility formulas in the internal control group
are obtained since the other modications are similar. The internal control
group is categorized by the users integrated operational capability to cope with
their internal changes and is calculated as the sum of four elements, exibility

Table 1
Flexibility formulas of internal control group
Internal control group Flexibility formula
P
Flexibility for continuous CI 1 t Nn1 1r
sn cn
n

improvement (CI) where sn is the average number of parts revisions, and cn is the
average cost required for alteration of tooling and software in
the nth year P
Flexibility for trouble control TC 1 t Nn1 1r
Tn bn
n

(TC) where TC 0 if Ca 0 and, Tn ed minfCa Mn ; Qn g c2 , Tn is


the average increased production revenue per breakdown in the
nth year, after subtracting o the rescheduling cost, bn is the
average number of breakdowns in the nth year, e is the average
revenue per item, d is the average period required for repair per
breakdown, Ca is the average manufacturing capacity available
during breakdown (% of Mn ), Mn is the maximum manufac-
turing capacity at the beginning of the nth year, Qn is the fuzzy
average demand in the nth year, and c2 is the rescheduling cost
P /Jn Wn
Flexibility for workforce WC 1 t Nn1 1r n

control (WC) where / is the average benet per job, Jn is the average number
of dierent jobs, Wn is the average number of operators
P
Work-in-process control (IP) IP 1 t Nn1 1r
ucv Vn
n

where u is the average revenue per item, cv is the inventory


carrying-cost rate, and Vn is the average work in process in the
N th year
C. Kahraman et al. / Information Sciences 168 (2004) 7794 85

for continuous improvement, exibility for trouble control, exibility for


workforce control, and exibility for work-in-process control [25].
Pyoun and Choi [25] do not consider that some deviations in the estimates of
exibility parameters, such as the average number of parts revisions in the nth
year (sn ), may occur. Our aim in this paper is to adapt their formulation of
realizable exibility to the uncertainty of real-world conditions by applying
Fuzzy Sets Theory. So here, in this paper, the terms sn , cn , and r are assumed to
be fuzzy parameters because in a CIM environment, the decision-makers
usually have imprecise data about these three. The income tax rate, t, is dened
by the government and usually is a xed value (kept unchanged) for a very long
time. So, it is assumed to be a crisp value. The project life, N , is also assumed to
be a crisp number here in the formulation. However, the fuzziness for this
parameter will be incorporated in the numerical example for the practical
purposes.
We may dene the parameters that Pyoun and Choi used in their paper [25]
as triangular fuzzy numbers such as ~sn sn0 ; sn1 ; sn2 . We may also dene the
discount rate during time n such as ~rn rn0 ; rn1 ; rn2 . The indices 0, 1, and 2 in
the representations of ~sn and ~rn stand for the smallest possible value, the most
possible value, and the largest possible value, respectively. The exact form of
the fuzzy present worth of the exibility for continuous improvement (P f WoCI)
under non-inationary conditions can be calculated as follows:
" #
XN
sly ly XN
sry ry
f n cn n cn
P WoCI 1 t Qn ry
; 1 t Qn ly
n1 n0 1 1 Rn0 n1 n0 1 1 Rn0

10
Under inationary conditions, it is very important to take the ination rate
into consideration. Hence, the average cost required for alteration of tooling
and software (cn ), which is the only monetary parameter in Eq. (10) changes
from year to year. In this case, the present worth formula for geometric cash
ows, which is frequently used in Engineering Economics, can be used [23].
Using Buckleys notation and assuming that c1 will increase at a fuzzy constant
rate (~
g), and ~s is xed along N years, Eq. (10) will change to,
" #
N N
1 f yj~
r f yj~
g
fNi yjP f
k k
W N 1 tfi yj~sc1 11
fk yj~r fk yj~ g

Other monetary parameters in the formulas of the following elements of


internal control group are Tn for trouble control, / for workforce control, and
u for work-in-process control. Similar modications derived by inationary
conditions are applied to Pyoun and Chois formulas in Eqs. (12), (19) and (21)
to yield with formulas in Buckleys notations in Eqs. (18), (20) and (22),
respectively. Please note that, to achieve a more precise present worth under
86 C. Kahraman et al. / Information Sciences 168 (2004) 7794

inationary conditions forcing us to deal with geometric growth rates, we


substitute the arithmetic average values / with /n , and u with un for
n 1; 2; . . . ; N , in Eqs. (20) and (22), respectively.
The second element of the internal control group is exibility for trouble
control (TC). It is the users capability to adapt the manufacturing system to
handle a breakdown of the system. This exibility is applicable only when the
breakdown period is long enough to reschedule the manufacturing system and
the user also needs to keep producing during this period with reduced capacity.
The terms Tn , bn , r, e, d, Ca , Mn , Qn , and c2 are assumed to be fuzzy
parameters. Only t and N are assumed to be crisp numbers for the reasons
mentioned in the fuzzication of CI above. Considering these, exibility for
trouble control P f WoTC is measured by
" #
XN
T la la
b XN
T ra ra
b
PfWoTC 1 t Qn
n n
ra
; 1 t Qn
n n
la
n1 0
n 1 1 R n0 n1 0
n 1 1 Rn0
12
where Ten is obtained by using the following equation:
ea M
Ten ~e d~ min C e n ~c2
e n; Q 13
In Eq. (13), it is necessary to use a ranking method to compare the fuzzy
numbers, C ea M
e n and Qe n . There are many ranking methods in the literature
and these methods may give dierent ranking results. Some of them are
Changs [6], Chiu and Parks [7], Dubois and Prades [10], Jains [14], Kauf-
mann and Guptas [20], Liou and Wangs [24], and Yagers [31] methods. For
example, Liou and Wang [24] propose the total integral value method with an
index of optimism x 2 0; 1. Let A e be a fuzzy number with left membership
function f L and right membership function f R . Then the total integral value is
eA eA
dened as
e xER A
Ex A e 1 xEL A
e 14
where
Z b
e
ER A R
xfe x dx 15
A
a

and
Z d
e
EL A L
xfe x dx 16
A
c

e a; b; c,
where 1 < a 6 b 6 c 6 d < 1. For a triangular fuzzy number, A
the total integral value is obtained by
e 1xa b 1 xb c
Ex A 17
2
C. Kahraman et al. / Information Sciences 168 (2004) 7794 87

Under inationary conditions, assuming that T1 will increase at a fuzzy con-


stant rate (~g) and ~b is xed along N years, Eq. (12) becomes
" #
N N
1 f yj~
r f yj~
g
fNi yjP f ~ 1 k k
W N 1 tfi yjbT 18
fk yj~r fk yj~ g

The third element of the internal control group is exibility for workforce
control (WC). It is the users capability to manage the size and the technical
and managerial capability of the workforce required for operation of the
manufacturing system. This exibility is measured by the size and the technical
and managerial capability of the required workforce and by the average benet
that the user can attain per job.
The terms /, Jn , Wn , and r are assumed to be fuzzy parameters. Only t and N
are assumed to be crisp numbers for the reasons dened in the fuzzication of
CI above. Considering these, the fuzzy present worth of exibility for work-
force control (P f
WoWC) can be formulized as in Eq. (19).
" #
XN
/ la la
J W la XN
/ ra ra
J W ra
Pf
WoWC 1 t Qn
n n
ra
; 1 t Qn
n n
la
n1 0
n 1 1 R n0 n1 0
n 1 1 R n0

19
Under inationary conditions, assuming that /1 will increase at a fuzzy con-
stant rate (~g) and J~ and W e are xed along N years, Eq. (19) becomes
" #
N N
f 1 f yj~
r f yj~
g
fNi yjP W N 1 tfi yj e e /1 k k
J fi yj W 20
fk yj~r fk yj~ g

The fourth element of the internal control group is work-in-process control


(IP). It is the users capability to minimize the work in process required for
operation of the manufacturing system. This exibility is measured by the
average amount of work in process required for operation of the manufac-
turing system and the inventory carrying-cost rate.
The terms u, Cv , Vn , and r are assumed to be fuzzy parameters. Only t and N
are assumed to be crisp numbers for the reasons dened in the fuzzication
of CI above. Considering these, the fuzzy present worth of exibility for work-
in-process control (P fWoIP) can be formulized as in Eq. (21).
" #
X N
u la la la
c V XN
u ra ra ra
c V
PfWoIP 1 t Qn
v n
ra
; 1 t Qn
v n
la
n1 n0 1 1 Rn0 n1 n0 1 1 Rn0

21
Under inationary conditions, assuming that u will increase at a fuzzy constant
g) and Ve and ~c are xed along N years, Eq. (21) becomes
rate (~
88 C. Kahraman et al. / Information Sciences 168 (2004) 7794

" #
N N
1 fk yj~r fk yj~g
fNi yjP f
W N 1 tfi yj Ve fi yj W
e u1 22
fk yj~r fk yj~
g

3.2.2. Numerical example


EGEY, a Turkish Motors Company, produces transmission gearboxes for
cars. The fuzzy average numbers of part revisions ~sn , the fuzzy maximum
manufacturing capacities M e n , the fuzzy average demand Q e n , and the fuzzy
~
average number of breakdowns bn in the years 20032008 are expected to be
as in Table 2. However, the analysis period is not certain since the company is
not sure that it will produce these products for the next six years. The pro-
duction will continue for N e years where lni j N
e is given in Table 3. The fuzzy
interest rate is expected to be xed as (8%, 10%, 12%) along this period while
the fuzzy average cost required for alteration of tooling and software is also
expected to be xed as (20, 30, 50) ($1000). The fuzzy average manufacturing
capacity available during breakdown is (85%, 90%, 95%), the fuzzy average
revenue per item is $(6.0, 6.5, 7.0) ($10), the fuzzy average period required for
repair per breakdown is (0.015, 0.032, 0.038) years, the fuzzy rescheduling cost
is $(180, 210, 240) and they are all assumed to be xed through N e years. The
crisp annual income tax rate is 35%.
Using Eq. (10), the exact fuzzy present worth of the exibility for continuous
improvement is calculated as follows:
8 2 1y 3
2y 2y
< 1:12 0:02y 2
1:12 0:02y 1:12 0:02y3
Pf
WoCI 0:65  20 10y  4 5;
: 3
1:12 0:02y 4
3y 3y
1:12 0:02y
1:12 0:02y5 6
2 4 y
39
2 3
1:080:02y 2 =
1:080:02y 1:080:02y3
0:65  50 20y  4 5
4 y
1:080:02y4
4 5 y
1:080:02y ;
1:080:02y5 6

Since the discount rate is estimated within the width of 4%, P f


WA can be used
in the problem under consideration [7]. To obtain the approximate form of

Table 2
Fuzzy expected values for e e n , and ~bn
e n, Q
Sn, M
Year, n e
Sn e
Mn en
Q ~bn
2003 (1, 2, 2) (3600, 3800, 4000) (3100, 3400, 3900) (3, 4, 5)
2004 (2, 3, 3) (3900, 4000, 4200) (3500, 3650, 3900) (4, 5, 6)
2005 (2, 3, 4) (4000, 4200, 4300) (3600, 3700, 4000) (4, 5, 7)
2006 (3, 3, 4) (4300, 4500, 4600) (3700, 3900, 4400) (5, 6, 7)
2007 (3, 4, 4) (4500, 4700, 4800) (4000, 4200, 4600) (6, 7, 7)
2008 (3, 4, 5) (4600, 4800, 4900) (4100, 4300, 4700) (7, 8, 8)
C. Kahraman et al. / Information Sciences 168 (2004) 7794 89

Table 3
Possibility table for the project life
Ni 4 5 6
e
lNi j N 0.7 1.0 0.8

PfWoCI, for n 4, 5, and 6, y will be assigned a value of 0 and 1 in the left side
representation, and then a value of 0 in the right side representation in the
equation above. Thus, the results in Table 4 are obtained.
After obtaining both the exact and approximate equations for P f WoCI, the
values of these two functions were calculated by using very small increments of
0.001 on y axis. Spearman correlation coecient was used to decide whether
the approximate form can be used instead of the exact form, or not. The
coecient was calculated as 0.9984 for the left sides, whereas it was 0.9993 for
the right sides. Since these coecients are close enough to 1.00, it was proven
for our example that the approximate form can be used instead of the exact
form.
In Fig. 3, the approximate fuzzy present worth of the exibility for con-
tinuous improvement is shown by its TFNs.
The numeric results of the fuzzy exibility for trouble control are given in
the following:
For the year 2003, C ea Me 1 3060; 3420; 3800 and Q e 1 3100;
3400; 3900. Using Liou and Wangs ranking method [24], for a moderately
optimistic decision-maker, x 0:5, the minimum of E0:5 C ea Me 1 3425 and
e 1 3450 is 3425. So we select C
E0:5 Q ea Me 1 value and use this value in Eq.
(13). By calculating all Ten s in the same way, we obtain

Te1 2514; 6904; 9928; Te2 2744; 7278; 10433


Te3 3000; 7486; 10460; Te4 3090; 8602; 11524
Te5 3203; 8588; 11950; Te6 3279; 8776; 12202

By using these values, the exact fuzzy present worth of exibility for trouble
control is obtained as

Table 4
Pf
WoCI values for n 4; 5; 6
Life, n fn;i yjP f
WoCIn
y 0 and i 1 y 1, i 1 or i 2 y 0 and i 2
4 $75,625.67 $167,709.86 $342,527.79
5 $97,755.32 $216,141.72 $431,003.61
6 $117,513.93 $260,170.69 $533,406.17
90 C. Kahraman et al. / Information Sciences 168 (2004) 7794

PW~ oCI
1.0

0.8
0.7

~
PWoCI
260,170.69
117,513.93

167,709.86

216,141.72

342,527.79

431,003.61

533,406.17
75,625.67
97,755.32

Fig. 3. The approximate fuzzy present worth of the exibility for continuous improvement shown
by its TFNs.

8 2 3 9
4390y25143y
>
>
> 4534y27444y >
>
7 >
1:12 0:02y 2
>
> 6 1:12 0:02y >
>
>
> 6 4486y30004y
5512y30905y 7 >
>
> 6
> 0:65  6 7 >
7 >
3 4
> 1:12 0:02y 1:12 0:02y ; >
>
> 4 5385y32036y 5497y32797y 5 >
>
>
> 1:12 0:02y5 1:12 0:02y6 >
>
>
< >
=
f
P WoTC 2 3
9928 3024y5 y 10433 3155y6 y
>
> 1:080:02y2 >
>
>
> 6 1:080:02y 7>>
>
>
> 0:65  6
> 1:080:02y4 7 >
10460 2974y7 2y 11524 2922y7 y
> 6 1:080:023 7 >
>
>
> 6 7 >
>
>
> 4 11950 3362y7 11202 3426y8 5 >
>
> 5 >
>
> 1:080:02y 1:080:02y 6
>
>
: ;

To obtain the approximate form of P f WoTC, y will be assigned a value of 0 and


1 in the left side representation, and then a value of 0 in the right side repre-
sentation in the equation above. Thus, the least possible value, the most pos-
sible value, and the largest possible value of the fuzzy present worth of the
exibility for trouble control are calculated as $33,844.19, $118,552.69, and
217,048.22 for n 4; $44,749, $155,880, and $273,979 for n 5, and 56,377.70,
190,994.79, and 330,452.28 for n 6, respectively. In Fig. 4, the approximate
fuzzy present worth of the exibility for trouble control is shown by its TFNs.
The FV in Eq. (9) will be the sum of the results which are obtained for the
internal control group, the investment policy group, and the marketing
adaptation group. Here, in this numerical example, we include only two of the
C. Kahraman et al. / Information Sciences 168 (2004) 7794 91

PW~oTC
1.0

0.8
0.7

273,979.00
118,552.69

190,994.79

217,048.22

330,452.28
44,749.00
56,377.70

155,880.00

~
33,844.19

PWoTC

Fig. 4. The approximate fuzzy present worth of the exibility for trouble control shown by its
TFNs.

elements of the internal control group, which are P fWoCI and P f WoTC, just to
show the way how the total FV amount is calculated. Thus, the result of the
summation of these two elements is shown in Fig. 5.
Using Fig. 5, the possibility of any FV can be found easily. In case of having
incomplete information, this possibility distribution provides the decision-
maker with a detailed information source. For example, the most possible
value in Fig. 5 is $372,021.72. If the decision-maker wants to know the interval
of exibility values having a possibility larger than 0.90, he/she should make
the calculations below:

1. Derive the functions for the area above the desired possibility value (0.90 for
our example) using the bold lines in Fig. 5.
x1 142; 504:32 229; 517:4y
x2 704; 982:61 332:960:89y
2. Substitute the desired possibility value into these equations and nd the
extreme points of the interval.
x1 142; 504:32 229; 517:4  0:90 349; 069:98
x2 704; 982:61 332:960:89  0:90 405; 317:81
According to the results of the calculations above, the interval of exibility
values having a possibility larger than 0.90 is (349,069.98; 405,317.81).
92 C. Kahraman et al. / Information Sciences 168 (2004) 7794

PW~oFV
1.0

0.8
0.7

559,576.01

704,982.61
P W oFV
142,504.32
109,469.86

173,891.63

286,262.55

372,021.72

451,165.48

863,858.45
Fig. 5. The approximate fuzzy present worth of the exibility value shown by its TFNs.

4. Conclusion

In this paper, the fuzzy present worth formulas for optimal level of exibility
and the exibility elements in the internal control group, which is categorized
by the users integrated operational capability to cope with their internal
changes, are obtained. Using these formulas, as shown in the given numeric
example, more informative results of exibility measures can be achieved. The
exibility modeling using triangular fuzzy numbers allows experts linguistic
predicate about computer integrated manufacturing systems. When there are
more than one alternative of CIM systems to compare in terms of exibility,
using the fuzzy present worth formulas of several types of exibility and
applying some dominance rules on triangular fuzzy numbers, the fuzzy exi-
bility evaluation process is completed. This study involves discrete com-
pounding for calculating the fuzzy present worth of exibility. Further research
can be aimed at including continuous compounding for the same calculation.
Also, trapezoidal fuzzy numbers can be used instead of TFNs when the deci-
sion-maker uses between instead of around to express his/her estimations
about the exibility parameters.

References

[1] A. Beskese, C. Kahraman, Z. Irani, Quantication of exibility in advanced manufacturing


systems using fuzzy present worth analysis, in: Proceedings of the 2nd International
C. Kahraman et al. / Information Sciences 168 (2004) 7794 93

Conference on Systems Thinking in Management, University of Salford, UK, April 35, 2002,
pp. B1.1724.
[2] A. Beskese, C. Kahraman, D. Ruan, Flexibility quantication in computer integrated
manufacturing systems based on fuzzy cash ow analysis, in: Proceedings of the 5th
International FLINS Conference, Gent, Belgium, September 1618, 2002, pp. 171178.
[3] A.H. Boussabaine, T.M.S. Elhag, Applying fuzzy techniques to cash ow analysis, Construc-
tion Management and Economics 6 (1999) 745755.
[4] J.J. Buckley, The fuzzy mathematics of nance, Fuzzy Sets and Systems 21 (1987) 257273.
[5] J.J. Buckley, E. Eslami, T. Feuring, Fuzzy Mathematics in Economics and Engineering,
Physica-Verlag, 2002.
[6] W. Chang, Ranking of Fuzzy Utilities with Triangular Membership Functions, in: Proc. Int.
Conf. of Policy Anal. and Inf. Systems, 1981, pp. 263272.
[7] C. Chiu, C.S. Park, Fuzzy cash ow analysis using present worth criterion, The Engineering
Economist 2 (1994) 113138.
[8] J.G. Demmel, R.G. Askin, A multiple objective decision model for the evaluation of advanced
manufacturing system technologies, Journal of Manufacturing Systems 3 (1992) 179194.
[9] A.D. Dimitrovski, M.A. Matos, Fuzzy engineering economic analysis, IEEE Transactions on
Power Systems 1 (2000) 283289.
[10] D. Dubois, H. Prade, Ranking fuzzy numbers in the setting of possibility theory, Information
Sciences 30 (1983) 183224.
[11] J.H. Hodder, H.E. Riggs, Pitfalls in evaluating risky projects, Harvard Business Review 1
(1985) 128135.
[12] G.K. Hutchinson, D. Sinha, A quantication of the value of exibility, Journal of
Manufacturing Systems 1 (1989) 4757.
[13] K. Iwamura, B. Liu, Chance constrained integer programming models for capital budgeting in
fuzzy environments, Journal of the Operational Research Society 8 (1998) 854860.
[14] R. Jain, Decision making in the presence of fuzzy variables, IEEE Transactions on Systems
Management Cybernet 6 (1976) 698703.
[15] C. Kahraman, Capital budgeting techniques using discounted fuzzy cash ows, in: D. Ruan, J.
Kacprzyk, M. Fedrizzi (Eds.), Soft Computing for Risk Evaluation and Management:
Applications in Technology, Environment and Finance, Physica-Verlag, Heidelberg, 2001, pp.
375396.
[16] C. Kahraman, D. Ruan, E. Tolga, Capital budgeting techniques using discounted fuzzy versus
probabilistic cash ows, Information Sciences 142 (2002) 5776.
[17] C. Kahraman, E. Tolga, Z. Ulukan, Justication of manufacturing technologies using fuzzy
benet/cost ratio analysis, International Journal of Production Economics 1 (2000) 4552.
[18] R.S. Kaplan, Must CIM be justied by faith alone?, Harvard Business Review 2 (1986) 8795.
[19] E.E. Karsak, Measures of liquidity risk supplementing fuzzy discounted cash ow analysis,
Engineering Economist 4 (1998) 331344.
[20] A. Kaufmann, M.M. Gupta, Fuzzy Mathematical Models in Engineering and Management
Science, Elsevier Science Publishers, 1988.
[21] I. Krinsky, J. Miltenburg, Alternate method for the justication of AMT, International
Journal of Production Research 9 (1990) 9971015.
[22] D. Kuchta, Fuzzy capital budgeting, Fuzzy Sets and Systems 111 (2000) 367385.
[23] M. Kurtz, Calculations for Engineering Economic Analysis, McGraw-Hill, 1995.
[24] T.-S. Liou, M.-J. Wang, Ranking fuzzy numbers with integral value, Fuzzy Sets and Systems
50 (1992) 247255.
[25] Y.S. Pyoun, B.Y. Choi, Quantifying the exibility value in automated manufacturing systems,
Journal of Manufacturing Systems 2 (1994) 108118.
[26] Y.K. Son, C.S. Park, Economic measure of productivity, quality, and exibility in advanced
manufacturing systems, Journal of Manufacturing Systems 3 (1987) 193206.
94 C. Kahraman et al. / Information Sciences 168 (2004) 7794

[27] A. Stam, M. Kuula, Selecting a FMS using multiple criteria analysis, International Journal of
Production Research 4 (1991) 803820.
[28] N.C. Suresh, Towards an integrated evaluation of exible automation investments, Interna-
tional Journal of Production Research 9 (1990) 16571672.
[29] N.C. Tsourveloudis, Y.A. Phillis, Manufacturing exibility measurement: a fuzzy logic
framework, IEEE Transactions on Robotics and Automation 4 (1998) 513524.
[30] R.N. Wabalickis, Justication of FMS with the AHP, Journal of Manufacturing Systems 3
(1988) 175182.
[31] R.R. Yager, On choosing between fuzzy subsets, Kybernetes 9 (1980) 151154.
[32] M.S. Zahir, Incorporating the uncertainty of decision judgements in the AHP, European
Journal of Operational Research 53 (1991) 206216.
[33] H.-J. Zimmermann, Fuzzy Set Theory and its Applications, Kluwer Academic Publishers,
Boston, MA, 1994.