You are on page 1of 14

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/259633490

Application of a modified VIKOR method for decision-making problems in


lean tool selection

Article  in  The International Journal of Advanced Manufacturing Technology · March 2013


DOI: 10.1007/s00170-013-5520-x

CITATIONS READS

37 312

3 authors:

Alireza Anvari Norzima Zulkifli


Islamic Azad University, Shiraz Branch Universiti Putra Malaysia
23 PUBLICATIONS   411 CITATIONS    55 PUBLICATIONS   936 CITATIONS   

SEE PROFILE SEE PROFILE

Omid Arghish
Islamic Azad University, Gachsaran , iran
5 PUBLICATIONS   40 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Alireza Anvari on 03 November 2018.

The user has requested enhancement of the downloaded file.


Int J Adv Manuf Technol (2014) 71:829–841
DOI 10.1007/s00170-013-5520-x

ORIGINAL ARTICLE

Application of a modified VIKOR method for decision-making


problems in lean tool selection
Alireza Anvari & Norzima Zulkifli & Omid Arghish

Received: 27 September 2013 / Accepted: 19 November 2013 / Published online: 12 December 2013
# Springer-Verlag London 2013

Abstract Group decision making with multiple criteria is the 1 Introduction


most popular method for ranking a set of alternatives. In this
regard, all alternatives are compared based on a common The field of multiple criteria decision making (MCDM) con-
criteria set. Meanwhile, decision makers sometimes encounter cerns about the problems of how decision-makers (DMs)
special situations, for example, having to select from among a should ideally make choices when facing multiple conflicting
set of alternatives without a set of criteria or with a set of criteria [1]. The decision-making process involves identifying
criteria that are grouped/related to various alternatives. Thus, the problems, constructing a set of preferences, evaluating the
it may be impossible to select from among a set of alternatives alternatives, and determining the best alternative [2]. Decision
using typical methods. Hence, in this study, a new, modified making is extremely intuitive when considering single-criterion
VIKOR method is proposed to address the lean tool selection problems because we only need to choose the alternative with
problems in manufacturing systems. In this study, a model the highest preference rating [1]. Although the VIKOR method
was developed to help practitioners improve their ability to is a popular method in multi-criteria analysis, it has some
solve problems when the possible solutions have their own problems when addressing MCDM situations [4, 5].
individual criteria. In fact the modified VIKOR method can be MCDM techniques are the most well accepted and well
applied to rank alternatives in threefold: alternatives with known methods used to prioritize a set of alternatives [6].
common criteria, without common criteria, with integrated Typically, all alternatives are compared based on a set of
common, and uncommon criteria. This paper offers numerical common criteria. However, there is not simply a set of criteria;
examples of the model, using a case study to illustrate an there are a few criteria that are grouped into various alterna-
application of the proposed model and properly assess the tives. Thus, decision makers may not be able to select from
validity of this new method. The results demonstrate the among a set of alternatives using typical methods.
usefulness and effectiveness of the modified new method. In this paper, we propose a modified VIKOR
The model covers the lack existing in the current literature to (VlseKriterijumskaOptimizacoja I Komopromisno Resenje /
assess effectiveness of applying lean tools. multi-criteria optimization and compromise solution) model,
which is based on incompatible multi-criteria analysis, to over-
Keywords Lean manufacturing . Leanness . Lean tool come VIKOR problems [2]. The most important features of the
selection . MCDM . VIKOR . Compromise solution proposed model are that its levels of best and worst scores are
determined (if all alternatives have their own criteria) and calcu-
lated (if the normalized decision matrix is in different matrixes).
A. Anvari (*) : O. Arghish
Hence, in this paper, a modified VIKOR method is pro-
Department of Industrial Engineering, Gachsaran Branch, I.A.U,
Gachsaran, Iran posed to address lean tools selection problems in manufactur-
e-mail: anvar.ali67@gmail.com ing systems when the alternatives have their own individual
O. Arghish criteria. Numerical examples (for verification), including
e-mail: omid_arghashi@yahoo.com two resolved problems and a case study (for validation),
are presented to illustrate applications of the proposed
N. Zulkifli
model and properly validate this new method. The results
Department of Mechanical and Manufacturing Engineering,
University Putra Malaysia, Selangor, Malaysia demonstrate the usefulness and effectiveness of the modified
e-mail: norzima@eng.upm.edu.my new method.
830 Int J Adv Manuf Technol (2014) 71:829–841

2 Leanness in manufacturing systems Start

Lean tools selection is currently one of the main challenges


Literature
faced by managers in manufacturing. This is because lean review
tools selection is the most important process determining the
success or failure of lean manufacturing (LM) systems; with-
out implementing proper techniques, leanness cannot be eval- Problem
uated. Leanness is the performance measurement of LM prac- statement
tices, the essence of which is the efficient use of resources
through the minimization of waste [1, 7]. Success in
implementing relevant techniques and achieving desired goals
within LM can be related to a certain level of leanness. A
PDCA method used to determine leanness is illustrated in No
Is agreement on
Fig. 1.
the best selection?
In this study, in addition to discussing the attributes of
leanness (cost, lead time, defects, value), lean tools and tech- Yes
niques were identified to be prioritized by a modified VIKOR
method.

Lean tools selection In this study, more than 100 tools


and techniques were obtained from various sources (e.g., Fig. 2 Delphi method to select the properly lean tools
[8–15]).
Thirty-five of the most popular tools and techniques were 3 VIKOR method
identified by referring to the literature. Next, 13 tools and
techniques were selected based on majority voting (21 ex- The compromise ranking method VIKOR was established as
perts), i.e., the select tools and techniques were those garner- an MCDM method to resolve decision-making problems that
ing over 50 % of the votes [16]. This selection was performed feature non-commensurable and incompatible criteria. In pro-
by ten experts using the Delphi method (Fig. 2). Ultimately, viding such a compromise solution, VIKOR applies the con-
five main tools and techniques were chosen: Continuous cepts of “acceptable advantage” and “acceptable stability” to
flow, Poka-Yoke, Standard works, Synchronize, and determine the maximum “group utility of the majority” and
TPM. The five tools and techniques were designated the minimum “individual regret of the opponent” [6].
as important alternatives. A summary of descriptions Based on the above-mentioned concepts, the compromise-
[11, 14] of the mentioned practices are presented in ranking algorithm VIKOR consists of the following steps
Table 1. [3, 17].

Fig. 1 Leanness process based


on Deming Cycle
Plan
Determining Criteria to Leanness

Cost
Lead Time Value
Defects

Action Do
Running results Determining & implementing
& improving for Lean Tools & Techniques
next study

Check
Analyzing &
confirming results
Int J Adv Manuf Technol (2014) 71:829–841 831

Step 1: The alternatives are denoted as A 1, A 2, … A i , … A m . number of criteria. The rating (performance score)
W j is the weight of the jth criterion, expressing the of the jth criterion is denoted by fij for alternative A i .
relative importance of the criteria, where j =1, 2,…, Step 2: Determine the best fi* and the worst f j− values of all
n; m is the number of alternatives, and n is the criterion functions, i =1,2,…,n ;

   
fi ¼ max f ij ; j ¼ 1; …; J ; f −j ¼ min f ij ; j ¼ 1; …; J ; if the i‐th function is benefit;
    ð1Þ

fi ¼ min f ij ; j ¼ 1; …; J ; f j ¼ max f ij ; j ¼ 1; … J ; if the i ‐ th is cost :

If we assume the jth function represents a benefit, then; and 0≤v ≤1, where v is introduced as a weight for the
f j* =max fij (or setting an aspired level) and f j− =mini fij (or strategy of maximum group utility, whereas 1−v is the
setting a tolerable level). Alternatively, if we assume the jth weight of the individual regret. In other words, when v >
function represents a cost/risk, then f *j =mini fij (or setting 0.5, this represents a decision-making process that could
an aspired level), and f j− =maxi fij (or setting a tolerable use the strategy of maximum group utility (i.e., if v is big,
level). group utility is emphasized), or by consensus when v ≈0.5,
or with veto when v <0.5.
Step 3: Compute the values S j and R j (utility measure and
regret measure); j =1, 2,. . ., m , using the relations .
ðS i −S − Þ ðS  −S − Þ Closeness to ideal solution
Xn  .  .
  −
Sj ¼ j¼1
w j f j −f ij f j −f j ð2Þ ðS i −R− Þ ðR −R− Þ Closeness to anti‐ideal solution
h  . i
R j ¼ Max w f j − f ij f j − f −j ð3Þ
Step 5: Rank the alternatives, sorting by the values S i , R i ,
and Q i in decreasing order based on two conditions.
Therefore, mini Si expresses the minimization of the aver-
age sum of the individual regrets/gaps and min Ri ex- Condition 1 : Q A2Þ− Q ðA1 Þ ≥ ð1=ðn−1ÞÞ
ð5Þ
presses the minimization of the maximum individual Condition 2 : Q AmÞ− QðA1 Þ < ð1=ðn−1ÞÞ
regret/gaps for prioritizing the improvement. In other
words, min Si emphasizes the maximum group utility, Propose as a compromise the alternative (A 1),
whereas min Ri emphasizes selecting minimum among which is ranked first by the measure min {Q i |i =1,
the maximum individual regrets. 2, …, m} if the following two conditions are satisfied:
Actually, the obtained solution is compromised by a
maximum group utility (represented by min j S) of the Table 1 A summary definition of alternatives in the study
majority, and a minimum of the individual regret represent-
ed by min j R of the opponent. The j S (the maximum Alternatives Description
group utility) and j R (the minimum individual regret of the
Continuous Producing and moving one item/items at a time through a
opponent) values, j =1, 2, K , J, by the relations. In other flow series of processing steps as continuously as possible,
words, Sj indicates the relative distance of the option “i”, with each step making just what is requested by the
positive ideal solution (best combination) and Ri indicates next step.
the most uncomfortable/regret of the option “i ”, is the Poka-Yoke Refers to a mistake-proofing device or procedure used to
distance from the positive ideal solution. prevent a defect during the production process.
Standard Establishing precise procedures for each operator’s work
Step 4: Calculate the values Qj, j=1 …m, using the relation works in a production process, based on three elements: Takt
time, the precise work sequence, the standard
Qi ¼ v ðS i −S − Þ = ðS  −S − Þ þ ð1−vÞ ½ðS i −R− Þ=ðR −R− ފ inventory.
S − ¼ MinS i S  ¼ MaxS i R− ¼ MinRi R ¼ MaxRi Synchronize The bringing together of materials information and
anything else needed in a coordinated manner such that
ð4Þ no part is waiting long for another.
TPM A set of techniques, to ensure that every machine in a
where S∗=mini Si (or setting the best S∗=0), S −=maxi Si
production process always is able to perform its
(or setting the worst S −=1), R∗=mini Ri (or setting the required tasks.
best R ∗=0), R −=maxi Ri (or setting the worst R −=1),
832 Int J Adv Manuf Technol (2014) 71:829–841

Fig. 3 Flowchart of modified


VIKOR method for lean tools
selection

C1. A cc ep ta ble ad va nt ag e: Q (A 2 ) − Q −Q (A 1)<(1/(n −1)) for maximum N; the posi-


(A 1)≥(1/(n −1)), where A 1 is the alternative oc- tions of these alternatives are close to one another.
cupying the second position in the ranking list by
Q ; m is the number of alternatives.
C2. Acceptable stability in decision making:
Alternative A 1 must also be the best ranked by 4 A proposed modified VIKOR method
{Si or/and R i |i =1, 2, …m }.
If one of the conditions is not satisfied, then a In this subsection, the details of the proposed method are
set of compromise solutions is proposed: presented step by step. This is followed by an examination
of numerical examples, comparison with resolved problems,
& Alternatives A 1 and A 2, if only condition C 2 is and a case study. The VIKOR method generally features a set
not satisfied. of criteria for a number of alternatives. Thus, the alternatives
& Alternatives A 1, A 2, …, A m , if condition C 1 is not are ranked according to all criteria; in other words, the same
satisfied. A m is determined by the relation Q (A m) criteria are used for each alternative. Sometimes, only a few

Table 2 Nine-point scale for determining degree of interaction

Verbal judgment Equally Equally to Moderately Moderately Strongly Strongly to Very strongly Very strong to Extremely
of preference preferred moderately preferred to strongly preferred very strongly preferred extremely preferred

Numerical rating 1 2 3 4 5 6 7 8 9
Int J Adv Manuf Technol (2014) 71:829–841 833

Table 3 A matrix of paired-comparison among multi-criteria related to Table 5 A matrix of scored alternatives based on related criteria
multi-alternatives
C1 C2 C3 C4 C5 C6 C7 C8 C9
A1 C1 C2 C3 A2 C4 C5 C6 A3 C7 C8 C9
A1 5 4 6 – – – – – –
C1 1 4 2 C4 1 6 3 C7 1 2 5 A2 – – – 3 7 4 – – –
C2 0.25 1 2 C5 0.17 1 2 C8 0.5 1 0.5 A3 – – – – – – 6 2 8
C3 0.5 0.5 1 C6 0.33 0.5 1 C9 0.2 2 1

4.2 Secondary steps in modified VIKOR method


criteria are used for each alternative: The set of criteria for a
number of alternatives is not fixed, so perhaps some alterna- Step 1: Organize decision matrix and determine the best f j*
tives have a fixed set of common criteria and some do not. and the worst f j− values. Because each alternative
This means that each alternative has its own criteria. To is ranked according to its own criteria, an ideal point
resolve these problems, a modified VIKOR method is pro- and non-ideal point, as in the VIKOR method ( f j* =
posed. The details of this modified VIKOR method are illus- max ( f ij , j =1,…,J ); and f j− =min ( f ij , j =1,…,J )
trated as follows. cannot be set. Therefore, the benefits or costs
must be reset according to each criterion of each
alternative, and we call the best f ij* the greatest
4.1 Primary construction modified VIKOR method
level and the worst f ij− the acceptable (at least)
level. Therefore, the initial rating matrix of the new
The details regarding the construction of the new method,
method is as follows:
including determining criteria, collecting score and weights,
Here, A i represents the ith alternative, i =1, 2 …m;
and using the new method for lean tools selection, are ex-
C j represents the jth criterion, j =1, 2, …, n; and
plained here. The new method is used to implement successful
f ij is the individual performance of an alternative
lean tools selection using leanness criteria. The method’s
(A i with respect to the jth criterion)
structure is presented in Fig. 3; the figure illustrates how the
Step 2: Normalize. The normalized weight rating matrix
modified VIKOR method is twofold: It addresses situations in
(determine by using the relations weight rating)
which alternatives have common and uncommon criteria.
can be expressed as follows:
Thus, to improve lean tools selection under various conditions,
Where w ij is the weight of i (1, 2 …m) alternative
the VIKOR method was modified.
and j (1, 2 …n) criteria, and the performance scores
as normalized ratings (r ij ) are indicated as
Score and weights This study used the above-mentioned
criteria to design two questionnaires. The first questionnaire  .  o
rij ¼ f ij −f ij f ij − f −ij ; i ¼ 1; 2…m; j ¼ 1; 2…n :
investigated grades of criterion importance based on paired
comparison analysis, and the tools rating of the second ques- ð6Þ
tionnaire was based on four criteria according to the view-
points of a panel of lean experts. In the questionnaires, the f ij* is the best score for each criterion of each
terms shown in Table 2 were adopted (with respect to a nine- alternative;
level scale) for the various grades of importance of every f ij− is the worst score for each criterion of each
criterion and alternative. alternative.
The decision matrix/different matrixes are structured with
respect to alternatives and attributes. The decision matrix Step 3: Determine ideal solutions. The best f ij* and the worst
calculates a score that indicates the weight and importance f ij− values (each alternative has its own criteria)
of the criteria/alternatives. The evaluating decision matrix
allows the DMs to be structured and then solved.
Table 6 A matrix of determined highest and lowest levels

C1 C2 C3 C4 C5 C6 C7 C8 C9
Table 4 Weights of criteria related to alternatives
A1 5 4 6 – – – – – –
A1 A2 A3
A2 – – – 3 7 4 – – –
C1 0.566234 C4 0.655556 C7 0.585822 A3 – – – – – – 6 2 8
C2 0.241558 C5 0.193333 C8 0.190347 f i* 10 10 10 10 10 10 10 10 10
C3 0.047619 C6 0.043571 C9 0.07619 f i− 1 1 1 1 1 1 1 1 1
834 Int J Adv Manuf Technol (2014) 71:829–841

Table 7 The priority of alternatives without common criteria step by step

C1 C2 C3 C4 C5 C6 C7 C8 C9

A1 5 4 6 – – – – – –
A2 – – – 3 7 4 – – –
A3 – – – – – – 6 2 8
Step 1 F*ij 10 10 10 10 10 10 10 10 10 F*ij =maxf ij
Fij− 1 1 1 1 1 1 1 1 1 Fij− =minf ij
  
Step 2 r ij 0.56 0.67 0.44 0.78 0.33 0.67 0.44 0.89 0.22 f −f
rij ¼ f ij − f −ij
ij ij

Step 3 Wj* rij 0.32 0.16 0.02 0.51 0.06 0.03 0.26 0.17 0.02
!
f ij −f ij
Step 4 Utility Si 0.5 0.6 0.44 Si ¼ ∑nj¼1 w j
measure
f ij −f −j
h i
f − f
Step 5 Regret Ri 0.32 0.51 0.26 Ri ¼ Max w j f ij− f ij−
ij j
measure
S *i =0.6 S −i =0.44 R *i =0.51 S −i =0.26
Step 6 VIKOR Qi 0.31 1 0
index
Q i =[v(S i −S −)/(S*−S −)]+[(1−v)(R i −R −)/(R*−R −)]
Step 7 Preference ranking A 3> A 1 >A 2

n     o 2 3
A ¼ max f ij jε J or min f ij jε J ; i ¼ 1; 2… f 1 ; f 2 ; … f n f11 ⋯ f1n1    
  F ¼ 4⋮ ⋮ ⋮ 5 rij ¼ fij − fij = fij − f j−
max f ij ; j ¼ 1; …; J ; fm1 ⋯ fmnn
n    o ð8Þ
A− ¼ min f jε J or max fij jε J ; i ¼ 1; 2; …f − ; f − ; … f − !1=2
ij 1 2 n . Xm

ð7Þ f ij ¼ xij 2
xij
i¼1


fij ¼ max fij ; ðbest levelÞ and fij − ¼ min fij ðworst levelÞ: Xn Xn    
 Si ¼ j¼1
w j rij S i ¼ j¼1
w j fij − fij = fij −f j− ð9Þ
fij ¼ max fij ; for i ¼ 1; 2; …n; and j ¼ 1; 2; … J

fij ¼ min fij ðif the ith function represents a benefitÞ


 h    i
Ri ¼ max w j rij Ri ¼ Max w j fij − fij = fij − f j−

f j− ¼ min fij for i ¼ 1; 2; …n; and j ¼ 1; 2; … J fi ¼ 1; 2; …; m& j ¼ 1; 2; …; ng ð10Þ

f j− ¼ max fij ; ðif the ith function represents a costÞ
Where Si represents the utility measure, Ri represents the
regret measure, and wj is the weight of the jth criterion.
Step 4: Compute the values S i (utility measure) and R i Step 5: Compute the index values Q i , i =1, 2, …, m , using
(regret measure), i =1, 2, …, m, using the relations the relation; the VIKOR index can be as follows:
Qi ¼ ½vðS i −S − Þ=ðS  −S − ފ þ ½ð1−vÞðRi −R− Þ=ðR −R− ފ;
Table 8 The final result ranking of alternatives without common criteria
Where Qi represents the ith alternative VIKOR
AJ Si Ri Qi Final ranked value.

Score Rank Score Rank Score Rank S − ¼ minðSj; j ¼ 1; …; J Þ; S  ¼ maxðSj; j ¼ 1; …; J Þ;


R− ¼ minðRj; j ¼ 1; …; J Þ; R ¼ maxðRj; j ¼ 1; …; J Þ
A1 0.5 2 0.32 2 0.31 2 2
A2 0.6 3 0.51 3 1 3 3
ð11Þ
A3 0.44 1 0.26 1 0 1 1 and “V ” is the weight for the strategy of maximum
group utility and usually set to 0.5, whereas “1–V ” is
Int J Adv Manuf Technol (2014) 71:829–841 835

Table 9 Ranking the alternatives based on modified method—first example

Eq. 6 Eq. 8 Eq. 11 Eq. 12 Eq. 13 Eq. 14

Ci Wj f ij− f ij* f ij r ij W *j r ij S i =£W *j r ij R i =maxW *j r ij Qi Ranked

A1 C1 0.566 1 10 5 0.56 0.31696 0.5 0.32 0.31 2


C2 0.242 1 10 4 0.67 0.16214
C3 0.048 1 10 6 0.44 0.02112
A2 C4 0.656 1 10 3 0.78 0.51168 0.6 0.51 1 3
C5 0.193 1 10 7 0.33 0.06369
C6 0.044 1 10 4 0.67 0.02948
A3 C7 0.586 1 10 6 0.44 0.25784 0.44 0.26 0 1
C8 0.190 1 10 2 0.89 0.1691
C9 0.076 1 10 8 0.22 0.01672

the weight of the individual regret. Here, we define Q(A(M))−Q(A(1))<1/(m −1) for maximum M (the
the value “V =0.5.” positions of these alternatives are close).
Step 6: Rank the alternatives, sorting by the value of
{S i , R i , Q i }, in decreasing/increasing order.
Propose as a compromise the alternative (A(1)) which 4.3 Application of modified VIKOR method
is ranked first by the measure min{Ri|i =1, 2, …,m}
if the following two conditions are satisfied: The modified method resolves problems in two ways: for
alternatives “without common criteria” and for alternatives
Q ð A2Þ− Q ðA1 Þ ≥ ð1=ðm−1ÞÞ with “common criteria.”
ð12Þ
QðAmÞ− QðA1 Þ < ð1=ðm−1ÞÞ
C1. Acceptable advantage: Q(A(2))−Q(A(1))≥1/ 4.3.1 First situation, alternatives without common criteria,
(m −1), where A(2) is the alternative with second numerical examples
position in the ranking list by Q; m is the number
of alternatives. Example 1: Suppose there are three alternatives (A 1, A 2, A 3) and
C2. Acceptable stability in decision making: nine criteria (C 1, C 2, …, C 9); all of the criteria are categorized in
Alternative A(1) must also be the best ranked by three groups so that each of the three criteria are dependent on
{Si or/and Ri|i =1, 2, …,m}. special alternatives. In this example, (C 1, C 2, C 3) are related to
If one of the conditions is not satisfied, then a set of A 1; similarly, (C 4, C 5, C 6) are related to A 2; and (C 7, C 8, C 9)
compromise solutions is proposed, which consists of: are related to A 3. The related scores are shown in Table 3.
Alternatives A(1) and A(2) if only condition C2 is not
satisfied. Step 1: Organize decision matrix/different matrixes (Table 3)
Alternatives A(1), A(2), …, A(M) if condition C1 and determine f i* (desired/highest level) and f i− (low-
is not satisfied. A(M) is determined by the relation est level).

Table 10 Ranking the alterna-


tives based on revised method by Ci Wj f ij− f ij* f ij r ij W *j r ij S i =£W *j r ij Q i =max r ij Ri Ranked
Yang et al. (2009)–first example
A1 C1 0.566 1 10 5 0.56 0.31696 0.5 0.67 0.19 2
C2 0.242 1 10 4 0.67 0.16214
C3 0.048 1 10 6 0.44 0.02112
A2 C4 0.656 1 10 3 0.78 0.51168 0.6 0.78 1 3
C5 0.193 1 10 7 0.33 0.06369
C6 0.044 1 10 4 0.67 0.02948
A3 C7 0.586 1 10 6 0.44 0.25784 0.44 0.89 0.5 1
C8 0.190 1 10 2 0.89 0.1691
C9 0.076 1 10 8 0.22 0.01672
836 Int J Adv Manuf Technol (2014) 71:829–841

Table 11 Comparing ranking orders of alternatives revised VIKOR and to the proposed model (if all alternatives have their
proposed modified model (example 1)
own criteria), the levels of the best and worst scores
Project Criteria Yang et al. method Modified method are determined.
The major characteristic of the proposed model is
S Q R S R Q its levels of best and worst scores (Table 6) are
A1 C1 0.5 0.67 0.19 0.5 0.32 0.31
determined (if all alternatives have own criteria):
C2 Step 2: Normalize (Eq. 8), for example, according to Table 6
C3 a 11 =5/f 11=5; also, determine the maximum level de-
A2 C4 0.6 0.78 1 0.6 0.51 1 sired ( f i* =10) and minimum acceptable level ( f i− =1).
C5 Based on Eq. 8, r ij =( f ij* −f ij )/( f ij* −f j− ).
C6 Therefore, r 11 =(10−5)/(10−1)=0.56
A3 C7 0.44 0.89 0.5 0.44 0.26 0 Similarly, r ij should be calculated for all the ranking
C8 numbers from Table 6, as shown in Table 7.
C9 Step 3: Calculate wj ×r ij
Preference ranking A 3>A 1 >A 2 A 3>A 1 >A 2 For example, w 1 =0.566234, and a 11 =0.56 then,
w 1 ×r 11 =0.32
Step 4: Calculate S i and R i
According to the modified method (Fig. 3), the  . 
Xn
weight calculation and the overall operation   −
Si ¼ j¼1
w j f ij −f ij f ij −f j
are performed separately. For example, to calculate
S 1 ¼ ½ð0:566  0:56Þ þ ð0:242  0:67Þ þ ð0:048  0:44ފ ¼ 0:5
the weights of criteria, each of the matrixes in Table 3
ð13Þ
should be computed separately. h  . i

Similarly, to calculate weights/scores and priori- Ri ¼ max w j rij Ri ¼ Max w j f ij −f ij f ij − f −j ð14Þ
tize alternatives, each of the groups should be calcu-
lated separately (see Table 4).
R1 ¼ max½ð0:566  0:56Þ þ ð0:242  0:67Þ þ ð0:048  0:44ފ
As shown in Table 5, the rating of alternatives is
based on each alternative’s own criteria. R1 ¼ max½ð0:31696Þ þ ð0:16214Þ þ ð0:02112ފ ¼ 0:32
A matrix of score of the proposed model is that the
levels of the best and worst scores (Table 6) are
determined (if all alternatives have own criteria): Thus, S 1 =0.5, S 2 =0.6, S 3 =0.44; and R 1 =0.32, R 2 =0.51,
alternatives based on related criteria form a ma- R 3 =0.26
trix and determine f i* (desired/highest level) and Step 5: Calculate Q 1 (Eq. 13)
f i− (lowest level); in other words, f ij* is the best h i h i
score for each criterion of each alternative, and Qi ¼ vðS i −S − Þ=ðS  −S − Þ þ ð1−vÞðR1 −R− Þ=ðR −R− Þ
f ij− is the worst score for each criterion of each
alternative (Eq. 6). As mentioned before, according Here, we define the value “V =0.5”, and

S − ¼ MinS i ¼ 0:44 S  ¼ MaxS i ¼ 0:6; R− ¼ MinRi ¼ 0:26; R ¼ Max Ri ¼ 0:51


Q1 ¼½0:5ð0:5−0:44=ð0:6−0:44ފ þ ½ð1−0:5Þð0:32−0:26Þ=ð0:51−0:26ފ
Q1 ¼ 0:31

With this procedure, we obtain Q 2 =1, and Q 3 =0.


Table 12 A matrix of paired-comparison among multi-criteria—exam- The details of the operations and abstract are shown in
ple 2
Table 7.
C1 C2 C3 C4 C5
Table 13 A matrix of
scored alternatives based C1 C2 C3 C4 C5
C1 1 0.333 0.25 0.333 0.5
on related criteria
C2 3 1 0.333 0.5 5 A1 – 5 4 – 3
C3 4 3 1 2 5 A2 4 5 4 5 –
C4 3 2 0.5 1 4 A3 3 – – 4 2
C5 2 0.2 0.2 0.25 1 A4 5 6 – – 6
Int J Adv Manuf Technol (2014) 71:829–841 837

Table 14 A matrix of scored alternatives based on related criteria—example 2 (modified method)

Eq. 6 Eq. 8 Eq. 11 Eq. 12 Eq. 13 Eq. 14

Ci Wj f ij− f ij* f ij r ij W *j r ij S i =£W *j r ij R i =maxW *j r ij Qi Ranked

A1 C2 0.303287 5 6 5 1 0.303287 0.321187 0.303287 0.4053 3


C3 0.607049 4 4 4 0 0
C5 0.089664 2 6 3 0.2 0.0179
A2 C1 0.087597 3 5 4 0.2 0.01175 0.191986 0.180236 0.2414 2
C2 0.180236 5 6 5 1 0.180236
C3 0.459802 4 4 4 0 0
C4 0.272366 4 5 5 0 0
A3 C1 0.155979 3 5 3 1 0.155979 1 0.619662 1 4
C4 0.619662 4 5 4 1 0.619662
C5 0.22436 2 6 2 1 0.22436
A4 C1 0.153604 3 5 5 0 0 0 0 0 1
C2 0.640515 5 6 6 0 0
C3 0.205881 2 6 6 0 0
Preference ranking A 4>A 2 >A 1 >A 3

Step 6: sort (S I , R I , Q I ) in decreasing order and rank alter- According to the results shown in Table 11, the prioritiza-
natives. tion of alternatives (A 3 >A 1 >A 2) using the revised VIKOR
Finally, we should rank the alternatives, sorting (Yang et al. method) and that using the modified method
by the value of {S i , R i , Q i }, in decreasing order if are the same.
the following two conditions are satisfied (Eq. 14):

Q A2Þ− QðA1 Þ ≥ ð1=ðn−1ÞÞ 4.3.2 Second situation, alternatives with integrated common
Q AmÞ− QðA1 Þ < ð1=ðn−1ÞÞ and uncommon criteria

From the results of this example, it seems A 3 may be the Example 2: Suppose there are four machines that should be
best alternative because it has the lowest Q , S , and R prioritized based on five criteria as shown in the matrixes
values with respect to the other alternatives; A 1 is the next below (Table 12). Note that the alternatives feature common
best, followed by A 2. The final results for ranking the and uncommon criteria (Table 13).
alternatives without common criteria are shown in Table 8. The associated operations and results obtained after calcu-
The problem has been resolved based on the modified lating the related weights of the criteria are shown in Table 14
method (Table 9) and the revised method (Table 10) pro- (modified method).
posed by Yang et al. [2].
Table 16 Comparing ranking orders of alternatives revised VIKOR and
Table 15 Comparing ranking orders of alternatives revised VIKOR and proposed modified model (2)
proposed modified model (1)
Project Criteria Yang et al. method Modified method
Project Criteria Yang et al. method Modified method
S Q R S R Q
S Q R S R Q
A1 C 11 0.33 0.4 0.37 0.33 0.13 0.67
A1 C 11 0.33 0.56 0.45 0.33 0.13 0.18 C 12
C 12 C 13
C 13 C 14
A2 C 21 0.30 0.50 0.40 0.30 0.15 0.11 A2 C 21 0.24 0.3 0.27 0.24 0.12 0
C 22 C 22
A3 C 31 0.48 0.56 0.52 0.49 0.22 1 A3 C 31 0.32 0.5 0.41 0.32 0.15 0.94
C 32 C 32
C 33 C 33
Preference ranking A 2>A 1 >A 3 A 2>A 1 >A 3 Preference ranking A 2 >A 1 >A 3 A 2 >A 1 >A 3
838 Int J Adv Manuf Technol (2014) 71:829–841

Table 17 A matrix of points scored by five experts 5 Verification of the modified method
CTs C1 C2 C3 C4
In this section, two examples are presented to illustrate how
T1 (9,7,4,9,7) (7,6,3,9,6) – (5,4,1,8,4) the proposed method can improve the prioritization of lean
T2 (3,9,2,9,5) (5,5,3,9,6) (2,7,3,8,4) (3,7,2,8,4) tools selection in manufacturing systems. We attempt to dem-
T3 (5,6,2,8,7) (5,7,1,7,8) (3,4,2,8,6) (5,6,1,9,4) onstrate the validity of the suggested model and show how the
T4 – (3,2,4,8,6) (7,6,9,9,8) – new method overcomes the critical weak points of the VIKOR
T5 (3,4,1,7,5) (3,7,1,8,6) – – method. The validity of the method is assessed by exploring
criterion validity and methodology validity.

According to results shown in Table 14, the order of 5.1 Criterion validity
the alternatives (based on the modified method) is (A 4 >
A 2 >A 1 >A 3). Criterion validity refers to the ability of a measure to correlate
with other measures under the same concepts [18]. Validity
refers to the ability of a scale to correlate with criterion
4.4 Comparing modified VIKOR method and revised VIKOR measures of the same constructs when measured at the same
method time. This study presents a new approach for ranking lean
tools by considering the effects of defined criteria on the
In this section, two problems resolved by Yang et al. [2] are leanness of manufacturing systems.
treated by the modified method and the results are then Lead time, cost, defects, and value were identified as the
compared. most important components of leanness. These items are
According to Tables 15 and 16, the results of the highlighted by many authors, for example, for lead time and
proposed model are in agreement with those of the method of cost [19–21], for defects [22, 23], and for value [19, 24]. Thus,
Yang et al. [2]. these metrics can be useful criteria is assessing the application
of lean tools and techniques [25].

4.5 Application of modified VIKOR method—case study 5.2 Methodology validity

A case study was used to determine the preference order There are some obstacles to implementing all of the lean tools
among five “tools and techniques” (alternatives): of a company: (1) it takes a long time for all tools to be
Continuous flow, Poka-Yoke, Standard works, Synchronize, implemented; (2) some tools and techniques are not adopted
and TPM. The alternatives were compared according to four in manufacturing systems; and (3) some techniques are similar
criteria (lead time, cost, defects, and value) by a panel of lean to one another.
experts (five members). As shown in Table 17, the experts In this study, we addressed a large number of techniques,
believed that some of the lean tools and techniques perhaps do from which certain techniques are not easily selected. When
not have an effect on decreasing lead time, cost, defects, or voters select and rank more than one candidate in order of
increasing value (Table 18). preference, ranked voting data arise [26]. According to the
Table 19 shows the order of alternatives determined by the literature, it was found among all combination methods that
modified method: (T 2 >T 4 >T 3 >T 1 >T 5). In this method (see majority vote is the simplest to implement [27]. However, the
Table 19), the algorithm and results indicate the measurement majority vote method is just as effective as other more com-
and calculation of alternatives based on related criteria. plex systems in refining the identified rate for the data set used
[27, 28]. Consequently, 13 tools and techniques were selected
as alternatives based on majority voting.
Moreover, the ultimate validation of a ranking methodolo-
gy indicates how well the results of the methodology fit the
Table 18 A matrix of expectations of the target constituency, especially for the top-
points scored after nu- CTs C1 C2 C3 C4
ranked alternatives. To solve the problem of “lean tools selec-
meral average
T1 6.91 5.84 – 3.64 tion for alternatives that have their own criteria, ” a modified
T2 4.75 5.27 5.22 4.22 VIKOR method was utilized. VIKOR is a helpful tool in
T3 3.36 4.55 4.1 4.04 multi-criteria decision making, particularly in cases in which
T4 – 4.1 7.71 – the decision maker is not able to express his/her preference at
T5 3.35 3.99 – – the beginning of system design. It has been widely applied to
address MCDM problems in various fields.
Int J Adv Manuf Technol (2014) 71:829–841 839

Table 19 Ranking of alternatives based on modified method—case study

Eq. 6 Eq. 8 Eq. 11 Eq. 12 Eq. 13 Eq. 14

Ci Wj f ij− f ij* f ij r ij W *j r ij S i =£W *j r ij R i =maxW *j r ij Qi Ranked

T1 C1 0.266471 6.91 3.35 6.91 0 0 0.62 0.62 1 4


C2 0.115977 5.84 3.99 5.84 0 0
C4 0.617553 3.64 3.64 4.22 1 0.62
T2 C1 0.175902 4.75 3.35 6.91 0.62 0.11 0.29 0.15 0.01 1
C2 0.087615 5.27 3.99 5.84 0.31 0.03
C3 0.223401 5.22 4.1 7.71 0.69 0.15
C4 0.513082 4.22 3.64 4.22 0 0
T3 C1 0.175902 3.36 3.35 6.91 1 0.18 0.62 0.22 0.93 3
C2 0.087615 4.55 3.99 5.84 0.7 0.06
C3 0.223401 4.1 4.1 7.71 1 0.22
C4 0.513082 4.04 3.64 4.22 0.31 0.16
T4 C2 0.29394 4.1 3.99 5.84 0.94 0.28 0.28 0.28 0.14 2
C3 0.70606 7.71 4.1 7.71 0 0
T5 C1 0.745546 3.35 3.35 6.91 1 0.75 1 0.75 1.7 5
C2 0.254454 3.99 3.99 5.84 1 0.25

Two examples of resolved problems (for verification) [2] techniques will fit into manufacturing processes and how
and a case study (for validation) are presented to illustrate how several concepts can be followed as guidelines to help ensure
the proposed method can improve the prioritization of lean the success of implementation [29].
tools selection processes. The proposed model tries to illus- Many recent papers have proposed analytical models to
trate the importance of goals in lean tools selection; attempts provide solutions to questions that arise in conflict manage-
to demonstrate the validity of the suggested model; and final- ment situations. Among the numerous approaches available
ly, overcomes the critical weak points of VIKOR. This meth- for conflict management, one of the most prevalent is MCDM.
od not only resolves new problems but classic problems as Among MCDM methods, the VIKOR method is based on an
well. aggregating function representing the proximity to the ideal
In summary, this paper has presented the results of a conditions; VIKOR is derived from the compromise program-
survey of manufacturing performance measures for au- ming method. Therefore, in this study, a modified VIKOR
tomotive industries. Reliability and validity analysis method was developed to address lean tools selection prob-
were conducted to assess the survey results. Consequently, lems in manufacturing systems in which the alternatives have
according to the foregoing discussion, the developed their own individual criteria.
model’s ability to perform lean tools selection can be Classical methods prioritize and choose alternatives using
validated. all attributes. However, pragmatically, decision makers should
control the objectives of one or several alternatives, and they
only need to rank the unimproved objectives or aspects of one
6 Discussion or several alternatives separately, each according to its own
criteria.
In today’s competitive world, LM has become an important To resolve the two mentioned problems (best level and
“role model” for two groups: academics and practitioners. alternatives under various conditions), one only needs to
Many organizations around the world have attempted to im- aggregate the related criteria (not all criteria) to obtain the
plement LM but the lack of a clear understanding of the main objectives. Thus, this study proposed a modified VIKOR
tools and techniques of leanness contribute to the failure of method. Based on the proposed new method, the results can
lean practices. It therefore seems necessary to provide a way to be more logical and reliable. This method helps decision
evaluate the impact of lean tools using an approach to deter- makers analyze and determine the usefulness and effective-
mine the critical techniques of leanness. ness of all of the alternatives with which they are confronted.
The objectives of LM are defined by the prioritization and This is a general model developed for all types of companies.
comparison of possible solutions and overall strategy. The Each phase of the model, including the determination of
implementation of LM tools and technique development as- criteria and the assessment of tools and techniques, has to be
sists in determining how the implementation of tools and modified to suit every distinct case.
840 Int J Adv Manuf Technol (2014) 71:829–841

7 Conclusion also would like to express appreciation to the anonymous reviewers and
editors for their very helpful comments for improving the paper.

In recent years, understanding how to determine suitable tools


and techniques in lean manufacturing has become a key
objective. However, empirical results indicate that the pro-
posed approaches provide a comprehensive measure of the References
effects of incorporating tools and techniques, which in general
reflect the efficiency and effectiveness of manufacturing sys- 1. Vinodh S, Vimal KEK (2012) “Thirty criteria based leanness assess-
tems, respectively. A methodology that allows for lean tools ment using fuzzy logic approach”. Int J Adv Manuf Technol 1185-
selection to assist practitioners in defining the objectives and 1195. doi: 10.1007/s00170-011-3658
2. Yang Y-PO, Shieh H-M, Leu J-D, Tzeng G-H (2009) A VIKOR-
implementing of LM was developed. based multiple criteria decision method for improving information
The VIKOR method was developed to solve MCDM prob- security risk. Int J Inf Technol Decis Mak 8(2):267–287
lems with conflicting and non-commensurable criteria. 3. Huang J-J, Tzeng GH, Liu HH (2009) “A revised VIKOR model for
Assuming that compromise is acceptable for conflict resolu- multiple criteria decision making—the perspective of regret theory”,
Y. Shi et al. (eds.): MCDM 2009, CCIS 35, 761–768.
tion, the decision maker wants a solution that is the closest to 4. Chang CL (2010) A modified VIKOR method for multiple criteria
the ideal, and the alternatives are evaluated according to all analysis. Environ Monit Assess 168:339–344. doi:10.1007/s10661-
established criteria. The new VIKOR method is able to derive 009-1117-0
and rank all alternatives within common/the same criteria; 5. Shemshadi A, Shirazi H, Toreihi M (2011) A fuzzy VIKOR method
for supplier selection based on entropy measure for objective
moreover, each alternative, through its own criteria, can help weighting. Expert Syst Appl 38:12160–12167
practitioners strengthen their selection of tools and techniques. 6. Opricovic S, Tzeng GH (2004) Compromise solution by MCDM
Therefore, this study proposes a new VIKOR method that is methods: a comparative analysis of VIKOR and TOPSIS. Eur J Oper
suitable and effective for ranking the tools and techniques of Res 156:445–455
7. Bayou ME, Korvin A (2008) Measuring the leanness of manufactur-
LM. ing systems—a case study of Ford Motor Company and General
The analysis performed using this new method was able to Motors. J Eng Technol Manag 25:287–304
help decision makers understand the ranking orders of imple- 8. Ohno T (1998) Toyota production system: beyond large-scale pro-
mented LM tools. Consequently, the method proposed in this duction. Productivity Press, Portland
9. Womack JP, Jones DT (2003) LEAN THINKING: banish waste and
study is suitable for ranking lean tools to achieve leanness. create wealth in your corporation. Simon and Schuster, London
The main contribution of the model is the assessment of the 10. Liker JK (2004) The Toyota way: 14 management principles from the
effectiveness of lean tool implementation; thus, it addresses an world’s greatest manufacturer. McGraw-Hill, New York
important deficiency in models reported in the literature. 11. Smith R, Hawkins B (2004) “Lean maintenance,” Elsevier
Butterworth–Heinemann, Oxford, UK 99, 42-43.
Furthermore, the model can combine a variety of available 12. Dahlgaard JJ, Dahlgaard-Park S (2006) Lean production, six sigma
tools and techniques in different disciplines to resolve deci- quality, TQM and company culture. TQM Mag 18(3):263–281
sion problems. 13. Grewal C (2008) An initiative to implement lean manufacturing
The model starts with criteria analysis and tools analysis using value stream mapping in a small company. Int J Manuf
Technol Manag 15(3):404–417
and then evaluates the current state of a given system, deter- 14. Marchwinski C, Shoo J (2010) “Lean lexicon: a graphical glossary
mining objectives, applying the Delphi method, voting, and for lean thinkers”, Lean Enterprise Institute, ISBN: 0966784367.
the modified VIKOR method to make the decision. These 15. Steinlicht CL (2011) “Lean production and the organizational life
aspects have been demonstrated to be of high importance cycle: a survey of lean tool effectiveness in young and mature
organizations”, doctoral dissertation. CAPELLA UNIVERSITY, US
when decision makers want to find the most effective lean 16. Van Erp M, Schomaker L (2002) “An overview and comparison of
solution. Strength of this model compared with existing the- voting methods for pattern recognition”, in Proc. 8th Int. Workshop
oretical models is that it has been tested in a real case. Frontiers Handwriting Recogn., 195–200.
17. Liu P, Wang M (2011) An extended VIKOR method for multiple
attribute group decision making based on generalized interval-valued
trapezoidal fuzzy numbers. Sci Res Essays 6(4):766–776
8 Recommendations for future research 18. Zikmund WG (2013) Business research methods, 7th edn. South-
Western, Ohio
19. Wan H, Chen F (2007) “Leanness score of value stream
In the future, the methodology may be further enhanced in the
maps”, Proceedings of the 2007 Industrial Engineering
following areas. Research Conference G. Bayraksan, W. Lin, Y. Son, and R.
Wysk, eds.
– Experiential study of real-world systems 20. Bowen P, Cattell K, Jay I (2011) Value management in the South
African manufacturing industry: exploratory findings. Manag Decis
– Developing leanness targets for various industries
49(1):6–28
21. Li C (2011) A customised lean model for a Chinese aerospace OEM
Acknowledgments This paper is supported by the Dept. of (original equipment manufacturer). Cranfield University, UK, MS
Manufacturing and Industrial Engineering, UPM, Malaysia. The authors thesis
Int J Adv Manuf Technol (2014) 71:829–841 841

22. Taj S (2005) “Applying lean assessment tools in Chinese hi-tech 26. Lam L, Suen CY (1997) Application of majority voting to
industries”, Collage of Business Administration, University of pattern recognition: an analysis of its behavior and per-
Detroit Mercy, and Detroit. Michigan, USA formance. IEEE Trans Syst Man Cybern Syst Hum 27(5):
23. Herron C, Braiden PM (2006) A methodology for developing sus- 553
tainable quantifiable productivity improvement in manufacturing 27. Obata T, Ishii H (2003) A method for discriminating efficient
companies. Int J Prod Econ 104(1):143–153 candidates with ranked. Voting data. Eur J Oper Res 151:233–
24. Wilson L (2010) “How to implement lean manufacturing”, McGraw 237
Hill, New York Chicago San Francisco Lisbon London Madrid 28. Lee D-S, Srihari SN (1993) “Hand printed digit recognition: a com-
Mexico City Milan New Delhi San Juan Seoul Singapore-ISBN: parison of algorithms”, in Proc. 3rd Int. Workshop Frontiers Hand
978-0-07-162508-1. writing Recognition, Buffalo, NY, 153–162.
25. Rivera L, Chen FF (2007) Measuring the impact of lean tools on the 29. Mahapatra SS, Mohanty SR (2007) Lean manufacturing in
cost–time investment of a product using cost–time profiles║. Robot continuous process industry: an empirical study. J Sci Ind
Comput Integr Manuf 23(6):684–689 Res 66:19–27

View publication stats

You might also like