S. S. KaraInt. J. Environ. Sci. Tech., 8 (2), 291-304, Spring 2011ISSN: 1735-1472© IRSEN, CEERS, IAU
*Corresponding Author Email: ssoner@ yildiz.rdu.ir Tel./Fax: + 90 212 383 2902
Received 10 February 2010; revised 29 April 2010; accepted 11 February 2011; available online 1 March 2011
Evaluation of outsourcing companies of waste electrical and electronic equipment recycling
S. S. Kara
Department of Industrial Engineering, Mechanical Faculty, Yildiz Technical University, Istanbul, Turkey
ABSTRACT: An increasing number of companies have focused on reducing the amount of waste properly or gainingvalue from used products. Facilitating the reverse flow of used products from consumers to manufacturers is a difficultand expensive process depending on the product and transportation type and distance. Another alternative is tooutsource these activities. Outsourcing management helps companies for better using of time, energy, labor, technology,capital, resources etc. Moreover, working with wrong partners effects manufacturers’ financial and operational situations.In order to get the best services, manufacturers usually invite several outsourcing companies for providing their tendersand then select the best offer. In this stage, using mathematical decision making techniques may help decision makers toget realistic results. In this paper the proposed methodology integrates two multi-criteria decision methods for rankingalternatives. This methodology is applied to a mid-sized firm operating in the field of electrical and electronic equipment.The results indicate that the most important criterion is cost for determining the best alternative. Besides, as it can beseen from the results, the best alternative for the manufacturer is the second alternative. These results propose aguideline for manufacturers for selecting the best alternative. From the results it can easily be seen that this approachshows its potential advantage in selecting suitable alternative due to its sound logic and easily programmable computationprocedure.
Keywords: Electronic recycling; Fuzzy analytical hierarchy process; Outsourcing management
INTRODUCTIONThere is an increased interest in the products which
are at the end of their life phase (EOL). The amount ofproducts that reaches their end of life is growing dueto changes in consumer attitude. Waste electrical andelectronic equipment recycling market is growing 8.1% every year in Turkey. The current threatening levelof environmental problems, along with related customerpressure and governmental regulations, motivatescorporations to undertake environmentally consciousinitiatives (Tuzkaya and Gülsün, 2008). Electricalequipments contain hazardous materials: lead, mercury,cadmium, chromium, phosphorus, barium, beryllium,etc (Bicheldey and Latushkina, 2010; Karapidakis etal., 2010). One way of preventing hazardous effects ofelectrical equipments is to treat them in a proper wayand to recycle valuable parts and materials (Lin et al.,2010). This also prevents depletion of resources. Suchan approach results in cost and waste reduction.Theincreasing global competition for primary raw materials
and the increasing price volatility will enforcecompanies to pay more attention to recyclingactivities. The manufacturers can benefit from wasteelectrical and electronic equipment (WEEE) recyclingactivities by using cheaper secondary materialsinstead of using expensive primary raw materials intheir production (Liu et al., 2010). However, someobstacles make recycling challenging for today’smanufactured products. First, it is difficult to gainall the information necessary to plan for the recyclingevaluation. Another problem in recycling EOLproducts is lack of technologies to handle the verycomplex products that are being discarded today,because the knowledge of how to do so is owned bythe recycler (Kuo, 2010). Outsourcing has becomean important business approach and a competitiveadvantage may be gained as products or servicesare produced more effectively and efficiently byoutside providers (Yang et al., 2007). The purposeof outsourcing is helping firms to reduce costs andconcentrate on their core competency (Gottfredson
S. S. Kara
292
et al., 2005). Outsourcing also plays a strategic roleby helping firms to acquire new capabilities, to bringabout fundamental changes to managerial strategyand organizational structure and to facilitate thetransformation of business models (Linder, 2004; Caoand Wang, 2007; ).
Manufacturers can make contract with WEEEoutsourcing companies in order to take advantagesof outsourcing. The contract can cover differentprocess such as collection, inspection, testing anddisassembling of used products and gaining rawmaterials. The quality of gained raw materials is relatedto manufacturer’s WEEE outsourcing company. Priorto making contracts with outsourcing companies, adetailed survey has to be done during the evaluationprocess.
Many criteria must be considered in the selectionprocedure. Thus, this problem can be viewed as amultiple criteria decision making (MCDM) problem inthe presence of many quantitative and qualitativecriteria (Tuzkaya et al., 2009). Many researchers haveattempted to use PROMETHEE (Preference RankingOrganization METHod for Enrichment Evaluations),in waste management. Mergias et al. (2007) usedPROMETHEE in selecting the best compromisescheme for the management of End-of-Life Vehicles(ELVs). Kapepula et al. (2007) utilized PROMETHEEII in ranking nine areas of the city for household solidwaste management in the urban community of Dakarwith respect to multiple criteria of nuisance. Vego etal. (2008) provided new insights to waste managementplanning at strategic level. Two multi-criteria decision-making (MCDM) methods, PROMETHEE and GAIA(Geometrical Analysis for Interactive Assistance) wereapplied to assist with the systematic analysis andevaluation of the alternatives.
Although, there were a number of publications inthese subjects, few of them have been interested inWEEE recycling. Queiruga et al. (2008) described amethod for ranking Spanish municipalities accordingto their appropriateness for the installation of WEEErecycling plants. In order to rank the alternatives, thediscrete multi-criteria decision method PROMETHEE,combined with a survey of experts, was applied.Rousis et al. (2008) examined alternative systems forthe WEEE management in Cyprus. These systems areevaluated by developing and applying the MCDMmethod PROMETHEE. Moreover, some researchershave used PROMETHEE while concentrating in
outsourcing management. Dulmin and Mininno (2007)proposed a PROMETHEE based approach which isapplied to a mid-sized Italian firm operating in thefield of public road and rail transportation for supplierselection. Araz et al. (2007) developed an outsourcerevaluation and management system for a textilecompany using fuzzy goal programming. The existingoutsourcers of the company are evaluated byPROMETHEE. Araz and Ozkarahan (2007) describeda supplier evaluation and management methodology,based on PROMETHEE methodology, for strategicsourcing. In their work, the suppliers are assessedconsidering to the supplier’s co-design capabilitiesand categorized based on overall performances.Additionally, the potential reasons for differences inperformance of supplier groups are identified andperformances of the suppliers are improved byapplying supplier development programs. Wang andYang (2007) proposed the use of analytic hierarchyprocess and preference ranking organization methodfor enrichment evaluations PROMETHEE as aids inmaking IS (Information System) outsourcingdecisions. The AHP is used to analyze the structureof the outsourcing problem and determine weights ofthe criteria and PROMETHEE method is used for finalranking, together with changing weights for asensitivity analysis. Tuzkaya (2009) appl iedPROMETHEE to choose the environmentallyconvenient transportation mode with respect to thedetermined evaluation criteria in Marmara Region ofTurkey. In addition, other multi criteria techniquesare used for waste management. Karamouz et al. (2006)introduced a framework in which to develop a masterplan for industrial solid waste management. SalmanMahini and Gholamalifard (2006) described a type ofmulti-criteria evaluation (MCE) method calledweighted linear combination (WLC) in a GISenvironment to evaluate the suitability of the studyregion for landfill. Hsu and Hu (2008) examined theconsistency approaches by factor analysis thatdetermines the adoption and implementation of greensupply chain management in Taiwanese electronicindustry.
There are many weight calculation procedures, butthe AHP has some advantages. One of the mostimportant advantages of the AHP attributes to its pair-wise comparison. Also PROMETHEE is a widelyaccepted multi-criteria decision-making technique dueto it s sound logic and easily programmable
S. S. Kara
293
Int. J. Environ. Sci. Tech., 8 (2), 291-304, Spring 2011
computation procedure. Besides, there is no evidence inthe literature that any of the work were prepared with theaim of selecting the suitable outsourcing companies forWEEE recycling using Fuzzy AHP and PROMETHEE. Thisstudy proposes a combined Fuzzy AHP andPROMETHEE methodology for evaluating and selectingthe most suitable company which offers WEEE recyclingoutsourcing service for Electrical and ElectronicEquipment Manufacturers.
The manufacturer in this study aims to evaluate thevarious companies’ offers that operate in WEEE recyclingin Istanbul. Compared to the previous evaluationresearches, the proposed method makes followingcontributions. Firstly, there is not any evaluation researchabout companies which offer WEEE recyclingoutsourcing service for Electrical and ElectronicEquipment Manufacturers. Secondly, a new integratedmethodology, Fuzzy AHP-PROMETHEE, is developed forevaluation process. For this reason, this integratedmethodology is applied to this problem. The integratedmethodology is based on Fuzzy AHP and PROMETHEE.Fuzzy sets are used in describing uncertainties in thepairwise comparison of criteria in the proposed method.
The rest of the paper is organized as follows: Thecontents of the fuzzy AHP process and PROMETHEEmethodology are described in Section 2. Application ofthe integrated model to the outsourcing companies’evaluation problem as a real world case study is presentedin Section 3. The results of the application and sensitivityanalysis are discussed in this section. In section 4,conclusions, main findings and contributions are drawnand future developments are suggested.
MATERIALS AND METHODSFuzzy AHP Procedure
In the proposed methodology, AHP with its fuzzyextension, namely fuzzy AHP, is applied to obtain moredecisive judgments by prioritizing the evaluationcriteria and weighting them in the presence ofvagueness. There are various fuzzy AHP applicationsin the literature that propose systematic approachesfor evaluation of alternatives and justification ofproblem by using fuzzy set theory and hierarchicalstructure analysis. Decision makers usually find itmore convenient to express interval judgments thanfixed value judgments due to the fuzzy nature of thecomparison process (Bozdag et al., 2003). This studyconcentrates on a fuzzy AHP approach introducedby Chang (1992), in which triangular fuzzy numbersare preferred for pairwise comparison scale. Extentanalysis method is selected for the synthetic extentvalues of the pairwise comparisons. The outlines ofthe extent analysis method for fuzzy AHP are given inthe following. A linguistic variable is a variable whosevalues are expressed in linguistic terms. The conceptof a linguistic variable is very useful in dealing withsituations, which are too complex or not well definedto be reasonably described in conventionalquantitative expressions (Kaufmann and Gupta,1991; Zadeh, 1965; Zimmermann, 1991).
In this study, the linguistic variables that areutilized in the model can be expressed in positiveTFNs (Triangular fuzzy numbers) for each criterionas in Fig. 1. The linguistic variables matching TFNsand the corresponding membership functions are
Fig. 1: Linguistic variables for the importance weight of each criterion
1 3 5 7 90
1
0.5
µ
Equally Moderately Strongly Very Strongly Extremely
~M
S. S. Kara
294
Evaluating recycling companies
provided in Table 1. Proposed methodology employs aLikert scale of fuzzy numbers star tingfrom 1~ to 9~ symbolized with tilde (~) for the fuzzy AHPapproach. Table 1 depicts AHP and fuzzy AHPcomparison scale considering the linguistic variablesthat describes the importance of attributes andalternatives to improve the scaling scheme for thejudgment matrices.
Let X = { 1x , 2x ,…, nx } be an object set, whereasU = { 1u , 2u , … , mu } is a goal set. According to fuzzyextent analysis, the method can be performed withrespect to each object for each corresponding goal,resulting in m extent analysis values for each object,given as 1
giM , 2giM , … , m
giM , i = 1, 2, … , n, where allthe j
giM (j = 1, 2, … , m) are triangular fuzzy numbersrepresenting the performance of the object ix withregard to each goal ju .The value of fuzzy syntheticextent with respect to the ith object is defined as (Chang, 1992).where the degree of possibility of 1M ≥ 2M is defined as:
Linguistic scale for importance Fuzzy numbers for fuzzy AHP Membership function Domain Triangular fuzzy scale
(l, m, u)
Just equal (1.0, 1.0, 1.0)
Equal importance 1~
)13()3()( −−= xxMµ
31 ≤≤ x (1.0, 1.0, 3.0)
Weak importance of one over another 3
~
)13()1()( −−= xxMµ
31 ≤≤ x (1.0, 3.0, 5.0)
)35()5()( −−= xxMµ
53 ≤≤ x
Essential or strong importance 5~ )35()3()( −−= xxMµ
53 ≤≤ x (3.0, 5.0, 7.0)
)57()7()( −−= xxMµ
75 ≤≤ x
Very strong importance 7~ )57()5()( −−= xxMµ
75 ≤≤ x (5.0, 7.0, 9.0)
)79()9()( −−= xxMµ
97 ≤≤ x
Extremely preferred 9~
)79()7()( −−= xxMµ
97 ≤≤ x (7.0, 9.0, 9.0)
Intermediate values between the two adjacent judgments
If factor i has one of the above numbers assigned to it when compared to factor j, then j has the reciprocal value when compared with I
Reciprocals of above
)/1,/1,/1( 1111
1 lmuM ≈−
Table 1: Linguistic variables describing weights of attributes and values of ratings
When a pair ( x , y ) exists such that x ≥ y and)(
1xMµ = )(
2yMµ , the equality equation V ( 1M ≥ 2M )
= 1 holds. Since 1M and 2M are convex fuzzy numbersand can be expressed as in Eqs. (3) and (4):
where, d is the ordinate of the highest intersectionpoint D between
1Mµ and 2Mµ (Fig. 2). When 1M =
( 111 ,, uml ) and 2M = ( 222 ,, uml ), the ordinate ofD is given by the following equation:
To compare 1M and 2M both values of V ( 1M ≥2M ) and V ( 2M ≥ 1M ) are required. The degree
iS =
1
1 1 1
−
= = =∑ ∑∑ ⎥
⎦
⎤⎢⎣
⎡⊗
m
j
n
i
m
j
jgi
jgi MM
V ( 1M ≥ 2M ) =
yx≥sup [ ])(),(min(
21yx MM µµ
V ( 1M ≥ 2M ) = 1 iff 1m ≥ 2m ,
V ( 1M ≥ 2M ) = hgt ( 1M ∩ 2M ) = )(1
dMµ
(1)
(2)
(3)
(4)
S. S. Kara
295
Int. J. Environ. Sci. Tech., 8 (2), 291-304, Spring 2011
possibility of a convex fuzzy number to be greater thank convex fuzzy numbers iM (i= 1, 2, … , k) can bedefined by Eq. (6).
Assume that:
For k = 1, 2, … , n; k ≠ i. Then, the weight vector isgiven by as in Eq. (8):
where iA (i = 1, 2, … , n) has n elements. Thenormalized weight vectors are defined as:
where W is a nonfuzzy number..
PROMETHEE MethodologyIn the following section, some basic important
definitions of PROMETHEE from Araz et al. (2007),Mergias et al. (2007); Vego et al. (2008) are reviewedand summarized. PROMETHEE, which is developedby Brans et al. (1986), is a non-parametric outrankingmethod for a finite set of alternatives. It is based onpositive and negative preference flows for eachalternative in the valued outranking relation to rankthe alternatives according to the selected preferences(weights). Positive flow expresses how much thespecific alternative is dominating other alternatives andnegative flow expresses how much that alternative is
V ( 2M ≥ 1M ) = hgt ( 1M ∩ 2M ) =
)()( 1122
21
lmum
ul
−−−
−
V (M ≥ M1, M2, … , Mk) = V [( M ≥ M1) and (M ≥ M2) and … and (M ≥ Mk)] = min V ( M ≥ Mi), i = 1, 2, 3, … , k.
)( iAd ′ = min V ( Si ≥ Sk)
W ′ = TnAdAdAd ))(),...,(),(( 21 ′′′
dominated by the others. Like all outranking methods,PROMETHEE proceeds to a pair of wise comparisonof alternatives in each single criterion in order todetermine partial binary relations denoting thestrength of preference of an alternative a overalternative b. Preference function ( )baPj , is calculatedafter evaluation matrix is formed. It is applied to decidehow much the outcome a is preferred to b. It translatesthe difference between the evaluations obtained bytwo alternatives (a and b) in terms of a particularcriterion, into a preference degree ranging from 0 to 1.
bad , is the difference of the value of criterion betweena and b. In order to facilitate the evaluation of aspecific preference function, six basic types have beenproposed. Table 2 summarizes various preferencefunctions.
If a is better than b according to jth criterion,0),( >baPj otherwise 0),( =baPj . Using the
weights jw assigned to each criter ion(where∑ jw =1), one can determine the aggregatedpreference indicator as follows:
Overall ranking is done by aggregating themeasures of pair wise comparisons. For eachalternative Aa∈ , the following two outrankingdominance flows can be obtained with respect to allother alternatives Ax∈ :
The leaving flow is the sum of the values of the arcsleaving node a and therefore provide a measure of theoutranking character of a. The higher the ( )a+φ , thebetter the alternative a.
W = ( TnAdAdAd ))(),...,(),(( 21
( ) ( )baPwba jj ,, ∑=Π (10)
( ) ( )∑∈
+ Π−
=Ax
xan
a ,1
1φ (11)
Fig. 2: Intersection point “d” between two fuzzy numbers M1 and M2
M2 M1
V (M2 = M1)
l2 l1m 2 d u2 m 1 u1
1
(7)
(6)
(5)
(8)
(9)
S. S. Kara
296
Function Shape Mathematical justification
Type 1 Usual criterion 1
da,b
Pj
( )⎩⎨⎧
>≤
=0 if 10 if 0
,,
,
ba
baj d
dbaP
Type 2 Quasi-criterion (U-Shape)
1
da,b
P j
q
( )⎩⎨⎧
>≤
=qdqd
baPba
baj
,
,
if 1 if 0
,
Type 3 Criterion with linear preference (V-Shape) 1
da,b
P j
p
( )⎪⎩
⎪⎨⎧
>
≤<=
pd
pdp
dbaP
ba
baba
j
,
,,
if 1
0 if ,
Type 4 Level criterion
1
da,b
Pj
q p
0.5
( )⎪⎩
⎪⎨
⎧
>≤<
≤<=
pdpd
qdbaP
ba
ba
ba
j
,
,
,
if 1q if 0.5
0 if 0,
Type 5 Criterion with linear preference and indifference area 1
da,b
Pj
pq
( )⎪⎪⎩
⎪⎪⎨
⎧
>
≤<−
−≤
=
pd
pdqpqd
qd
baP
ba
baba
ba
j
,
,,
,
if 1
q if
if 0
,
Type 6 Gaussian criterion
bad ,is the difference value of
alternative a and b in each criterion 1
da,b
Pj
( )⎩⎨⎧
+= x
x
j eebaP
1,
Table 2: Different Preference Functions in PROMETHEE (Vego et al,. 2008)
The entering flow measures the outranked character.The smaller ( )a−φ , the better the alternative a.
The two main PROMETHEE tools can be used toanalyse the evaluation problem:
• the PROMETHEE I partial ranking,• the PROMETHEE II complete ranking.
The PROMETHEE I partial ranking provides a ranking
of alternatives. In some cases, this ranking may beincomplete. This means that some alternatives cannotbe compared and, therefore, cannot be included in acomplete ranking. This occurs when the first alternativeobtains high scores on particular criteria for which thesecond alternative obtains low scores and the oppositeoccurs for other criteria. The use of PROMETHEE I,then, suggests that the decision-maker should engagein additional evaluation efforts. PROMETHEE II
S. S. KaraInt. J. Environ. Sci. Tech., 8 (2), 291-304, Spring 2011
297
provides a complete ranking of the alternatives fromthe best to the worst one. Here, the net flow is used torank the alternatives. Leaving flow and entering floware used in PROMETHEE I.
There are two basic rules for outranking inPROMETHEE I.
One other parameter, net flow, is used for resultingPROMETHEE II.( )aφ quantifies the position of alternative a accord-
ing to criterion j with respect to all the other alterna-tives in the set A. The larger the single criterion netflow the better alternative a on criterion j.
RESULTS AND DISCUSSIONAppl ication of the integrated Fuzzy AHP-PROMETHEE methodology to a case study
The technologic industry in Turkey has gonethrough a number of significant changes in the lastyears. Technological advances and customer demandare the main reasons of this development. In order toget sustainable competitiveness in the sector,manufacturers need to take strategic decisions. Oneof the most important decisions is recycling activitiesto gain value from reused products. Since thisdecision requires a long term investment, to outsourcethese activities make manufacturers more compatible.
As a real world case study, WEEE recyclingprovider evaluation problem is proposed to verify ourmethodology. The company in the case study isoperating in the Electrical and Electronic Equipment
( ) ( )∑∈
− Π−
=Ax
axn
a ,1
1φ (12)
(13)
(14)
a outranks b (aPb) if: ( ) ( )ba ++ ≥ φφ and ( ) ( )ba −− ≤ φφ or
( ) ( )ba ++ > φφ and ( ) ( )ba −− = φφ or
( ) ( )ba ++ = φφ and ( ) ( )ba −− < φφ a and b (aRb) are incomparable if:
( ) ( )ba ++ > φφ and ( ) ( )ba −− > φφ or
( ) ( )ba ++ < φφ and ( ) ( )ba −− < φφ a and b (aIb) are indifference if:
( ) ( )ba ++ = φφ and ( ) ( )ba −− = φφ
(16)
(17)
( ) ( )∑∈
−−
=Ax
jj axPxaPn
a ),(),(1
1φ (18)
(15)
industry in Turkey. Its application area is in homeappliances sector and international electronicmanufacturers. Its products are televisions, DVDplayers, digital satellite receivers, air-conditioningproducts, air conditioners, white goods and washingmachines. This case study is a proposal for an Electricaland Electronic Equipment manufacturer which has anobjective to gain 300,000 ton of recycled WEEE. Themanufacturer decides to outsource its WEEE recyclingactivities and wanted to contract with the optimalrecycler to get the best service. The manufacturer wantsto evaluate various companies’ offers that operate inWEEE recycling. For this reason, this integratedmethodology is applied to this problem. The integratedmethodology is based on Fuzzy AHP andPROMETHEE. Fuzzy sets are used in describinguncertainties in the pairwise comparison of criteria.
A detailed survey is conducted through thedistribution of a comprehensive questionnaire to themanagers and the related authorities in themanufacturing company. The questionnaire related withthe data regarding the qualitative and quantitativecriteria is formed for the evaluation model. Furthermorea lot of face-to-face interviews are held to develop solidinformation on the selected criteria and alternatives.After determining all selected criteria and alternativecompanies, the paired comparisons in the questionnaireare made by using the triangular fuzzy numbers totackle the ambiguities involved in the process of thelinguistic assessment of the data. Fig. 3 summarizesthe integrated methodology. The stepwise procedureof the methodology contains three phases:
Step 1. Pre-research phaseStep 2. Fuzzy AHP phaseStep 3. PROMETHEE phase
Pre-research phaseIn the pre-research phase the WEEE recyclers in themarket are investigated. Then the related criteria aredetermined with related authorities. In this study sixcriteria are determined by the company and five WEEErecycling companies (A1, A2,…, A5) are taken intoconsideration for evaluation. It is essential to identifya set of evaluation criteria that evaluate all of theproposed alternatives. The following criteria areselected in this particular case:
1. Cost: The fee of one year contract with eachcompany.
S. S. KaraEvaluating recycling companies
298
2.Capacity utilization: The percentage of the capacitythat can be assigned for the manufacturer’s products.3. Existing collection networks: The number of existingcollection networks of each company in Turkey.4. Capability of disassembly infrastructure: To measureif each company’s disassembly infrastructure isadequate or not. Five points scale is used to expresseach company’s capability.5. Availability of new waste processing programs:Waste Processing Programs related with environmentallegislation help these kinds of companies in recyclingprocess. To measure the existence of these programs,a value of one was given to a company with a wasteprocessing program and zero to a company with nosuch a program.6. Land requirement: To measure if each companyneeds area for the installation of the mechanicalequipment as well as the auxiliary infrastructures. Avalue of 1 was given to a company which needs areaand 0 to a company with no requirement.
Fuzzy AHP phaseIn the second phase, decision-makers do pairwise
comparison in linguistic form in the questionnaire for
obtaining criteria weights. The linguistic forms areconverted into triangular fuzzy numbers for Fuzzy AHPevaluations. Fuzzy comparisons are defuzified withChang’s extent analysis and the criteria weights areobtained in Fuzzy AHP phase. Criteria weights arecalculated by using Fuzzy AHP. Table 1 is used forpairwise comparison. Table 3 depicts the pairwisecomparison matrix set by TFNs (Triangular FuzzyNumbers) that matches linguistic statements of data.The fuzzy values of paired comparison are converted tocrisp values via the Chang’s extent analysis (1992)mentioned as before. The obtained priority weight vectorof criteria is figured out in the last column of Table 3.
PROMETHEE PhaseIn the last phase the data related with alternatives’
decision matrix is obtained from related authorities.PROMETHEE is used to rank potential companieswhich offer WEEE recycling outsourcing service forElectrical and Electronic Equipment Manufacturers.The last phase of the study starts establishingevaluations of the alternative companies with respectto the individual criteria. This is a decision matrix forranking alternatives and indicates the performance
The manufacturer decides tooutsource its WEEE recycling
activities
The WEEE recycling companiesin the market are searched
The criteria are determined
WEEE recycling companiesare determined
Decision-markers` opinion areexpressed in linguistic form
The linguistic form areconverted into fuzzy numbers
The fuzzy number are defuzzifiedwith chang`s exert analysis
The criteria weight aredetermined
The decision matrix is formed forWEEE recycling companies
Performance function types aredetermined for all criteria
The leaving and entering flowsare calculated
The ranking is done withPROMETHEE I and II
Pre- Research phase Fuzzy AHP phase PROMETHEE phase
Fig. 3: Proposed approach
S. S. KaraInt. J. Environ. Sci. Tech., 8 (2), 291-304, Spring 2011
299
Weights Land
requirement (C6)
Availability of new waste processing
programs (C5)
Capability of disassembly
infrastructure (C4)
Existing collection networks
(C3)
Capacity utilization
(C2) Cost (C1)
0.523 (5,7,9) (3,5,7) (5,7,9) (3,5,7) (5,7,9) (1,1,1) Cost (C1) 0.304 (3,5,7) (1,3,5) (1,3,5) (1,3,5) (1,1,1,) (0.11,0.14,0.2) Capacity
utilization (C2) 0.409 (0.14,0.2,0.33) (1,1,3) (1,3,5) (1,1,1) (0.2,0.33,1) (0.14,0.2,0.33) Existing
collection networks (C3)
0.064 (1,1,3) (1,3,5) (1,1,1) (0.2,0.33,1) (0.2,0.33,1) (0.11,0.14,0.2) Capability of disassembly infrastructure (C4)
0.011 (1,3,5) (1,1,1) (0.2,0.33,1) (0.33,1,1) (0.2,0.33,1) (0.14,0.2,0.33) Availability of new waste processing programs (C5)
0.049 (1,1,1) (0.2,0.33,1) (0.33,1,1) (3,5,7) 0.14,0.2,0.33) (0.11,0.14,0.2) Land requirement (C6)
ratings of the alternatives according to the criteria.Before constituting decision matrix, preferencefunction types are determined. Vego et al. (2008) choselinear preference for cost (C1) or criteria that areexpressed with real numbers (C3). Araz et al. (2007)used linear preference function for “Capacity utilization(C2)”. Also they chose level preference function for 5-point scale expressions (C4). The criteria: “Availabilityof new waste processing programs (C5)” and “Landrequirement (C6)”, were measured with ‘‘yes’’ (1) or ‘‘no’’(0). For that reason a usual preference function, whichexpresses only indifference or strict preference, wasapplied (Queiruga et al. (2008). (C1) and (C6) should beminimized and (C2), (C3), (C4) and (C5) should bemaximized. Threshold values for preference functionsdepend on decision-makers opinions and expertise. Inthis study these threshold values are determined withcompany managers. Decision matrix and criteriacharacteristics are shown in Table 4. Decision LAB.2000 commercial software is used for PROMETHEEcalculations.
Indifference threshold is the maximum value thatrepresents two alternatives’ indifference. Preferencethreshold is the maximum value that represents onealternative’s preference to another. If the difference oftwo alternatives ( bad , ) is greater than preferencethreshold so alternative a is strictly preferred to b. To
explane the calculations, Table 5 is presented for thecomparison of A1with A2.
C1’s preference is minimization so the smallest valueis the best. A1’s value is greater than A2’s value in C1,so 0)2,1(1 =P . C2’s, C3’s, C4’s and C5’s preferencesare maximization and A1’s value is smaller than A2’svalue in C2, so 0)2,1(2 =P . In C3, 78152,1 =−=dand =−−= )210/()27()2,1(3P 0.625 (In Table 2Type 5). A1’s value is smaller than A2’s value in C4,so 0)2,1(4 =P . AA1’s value is equal to A2’s value in C5,so 0)2,1(5 =P . At last, AA1’s value is greater than A2’svalue in C6, so 0)2,1(6 =P because of minimizationpreference. ( ) 03.0049.0*625.02,1 ==Π (Eq.(10)).This comparison is done with all other alternatives andby using Eq. (11), Eq. (12) and Eq. (18). The leavingflows, the entering flows and the net flow of eachalternative are calculated.
Results and sensitivity analysisThe results of the leaving flows ( )a+φ , entering
flows ( )a−φ and net flow ( )aφ for PROMETHEE Iand II are presented in Table 6. Also the partial rankingin PROMETHEE I is presented in Fig. 4, while thecomplete ranking of alternatives in PROMETHEE IIfrom best to worst in terms of comparison with respectto Eqs. (13 - 17) is presented in Fig. 5.
In the PROMETHEE representation, if a outranks b,
V (Sc1 > = Sc2, Sc3, Sc4, Sc5, Sc6) = 1.00; V (Sc2 > = Sc1, Sc3, Sc4, Sc5, Sc6)= 0.59;V (Sc3 > = Sc1, Sc2, Sc4, Sc5, Sc6) = 0.09; V (Sc4 > = Sc1, Sc2, Sc3, Sc5, Sc6)= 0.12;V (Sc4 > = Sc1, Sc2, Sc3, Sc4, Sc6) = 0.02; V (Sc5 > = Sc1, Sc2, Sc3, Sc4, Sc5)= 0.09;
Table 3: Pairwise comparisons of criteria
S. S. Kara
300
C1 C2 C3 C4 C5 C6
Min/Max
Preference
Min Max Max Max Max Min
Function Linear Linear Linear Level Usual Usual
Indifference
Threshold (q)
5,000 0.03 2 0.5 - -
Preference
Threshold((p)
70,000 0.2 10 1.5 - -
Unit TL/year % No. Networks 5-point scale Yes/No Yes/No A1 250,000 0.6 15 1 1 1 A2 225,000 0.75 8 3 1 0 A3 240,000 0.45 5 4 0 0 A4 195,000 0.55 20 2 1 1 A5 320,000 0.7 14 3 0 1
Table 4: Decision matrix for PROMETHEE
C1 C2 C3 C4 C5 C6 Min Max Max Max Max Min
The smallest is the best
The largest is the best
The largest is the best
The largest is the best
The largest is the best
The smallest is the best
A1 250,000 0.6 15 1 1 1 A2 225,000 0.75 8 3 1 0 d1,2 - - 15-8=7 - - - P(1,2) 0 0 0.625 0 0 0
the representation is as ba → . As it can be seen inFigs. 4 and 5, the best alternative for the manufactureris the second alternative. A4 is the second bestalternative. However, A1 and A3 are incomparable (A1R A3) because ( ) ( )13 AA ++ > φφ and ( ) ( )13 AA −− > φφ . AsAsit can be seen in Fig. 5, in net flow ( ) ( )31 AA φφ > soA1 is better than A3. Lastly A5 is the worst alternativewith respect to both PROMETHEE I and PROMETHEEII.
Another analysis is represented in Fig. 6. It presentsthe criteria’s effects on the alternatives. The x-axis iscriteria. The scores are between +1 (being the best)
and “1 (being the worst). The strong and the weaksides of each alternative can be seen in this figure. Thesecond, the third and the fourth alternatives arepositively impacted from C1 whereas the fifth alternativeis negatively impacted. The second and the fifthalternatives are positively impacted from C2 whereasthe first, the third and the fourth alternatives arenegatively impacted. The first, the fourth and the fifthalternatives are positively impacted from C3 whereasthe second and the third alternatives are negativelyimpacted. The second, the third and the fifthalternatives are positively impacted from C4 whereasit has negative impact on the first and the forthalternatives. The first, the second and the forthalternatives are positively impacted from C5 whereasthe third and the fifth alternatives are negativelyimpacted.
The second and the third alternatives are positivelyimpacted from C6 whereas the first, the forth and thefifth alternatives are negatively impacted. Besides A2
Table 5: The explanation of the comparison of A1 to A2
Table 6: Leaving, entering and net flows
(a) Φ - (a)Φ + (a)Φ -0.102 0.320 0.218 A1
0.389 0.084 0.473 A2 -0.157 0.382 0.226 A3
0.247 0.195 0.442 A4
-0.377 0.578 0.202 A5
S. S. Kara
301
Fig. 5: The complete ranking in PROMETHEE II
3
21
4
5A2 A4
A1
A3
A5
1 3
2 4
5
A2
A4
A1
A3
A5
Creiteria exchange New criteria wieghts Results of PROMETHEE II Base 0.523, 0.304, 0.049, 0.064, 0.011, 0.049 2 4 1 3 5 1 2 0.304, 0.523, 0.049, 0.064, 0.011, 0.049 2 4 5 1 3 1 3 0.049, 0.304, 0.523, 0.064, 0.011, 0.049 2 5 4 1 3 1 4 0.064, 0.304, 0.049, 0.523, 0.011, 0.049 4 5 1 2 3 1 5 0.011, 0.304, 0.049, 0.064, 0.523, 0.049 2 1 4 5 3 1 6 0.049, 0.304, 0.049, 0.064, 0.011, 0.523 2 5 4 1 3 2 3 0.523, 0.049, 0.304, 0.064, 0.011, 0.049 4 2 1 3 5 2 4 0.523, 0.064, 0.049, 0.304, 0.011, 0.049 2 4 3 1 5 2 5 0.523, 0.011, 0.049, 0.064, 0.304, 0.049 4 2 1 3 5 2 6 0.523, 0.049, 0.049, 0.064, 0.011, 0.304 4 2 3 1 5 3 4 0.523, 0.304, 0.064, 0.049, 0.011, 0.049 2 4 1 3 5 3 5 0.523, 0.304, 0.011, 0.064, 0.049, 0.049 2 4 1 3 5 3 6 0.523, 0.304, 0.049, 0.064, 0.011, 0.049 2 4 1 3 5 4 5 0.523, 0.304, 0.049, 0.011, 0.064, 0.049 2 4 1 3 5 4 6 0.523, 0.304, 0.049, 0.049, 0.011, 0.064 2 4 1 3 5 5 6 0.523, 0.304, 0.049, 0.064, 0.049, 0.011 2 4 1 3 5
Table 7: Sensitivity analysis results
Fig.4: The partial ranking in PROMETHEE I
Int. J. Environ. Sci. Tech., 8 (2), 291-304, Spring 2011
S. S. KaraEvaluating recycling companies
302
Fig. 6: The criteria’s effects on the alternatives y-axis the scores x-axis the criteria
A1 A2
A3 A4
A5
1
0
-1
1
1 1
1
0
0
0 0
-1-1
-1
-1
is the best alternative according to all criteria, it is quitelow in C4 and C6. The second alternative A4 is quite lowin C3. The third alternative A1 is quite low in C3 and C5.The forth alternative A3 is quite low in C2, C4 and C6.The last alternative A5 is quite low in C1, C5 and C6.
Besides, sensitivity analysis is done for gettingaccurate results. The idea of sensitivity analysis is toexchange each criterion’s weight with another criterionweight. Thus, 15 different calculations are formed. Table7 summarizes sensitivity analysis results. As it can beseen from Table 7, 40 % of the calculations give theexactly same ranking of our result. 73% of thecalculations choose the second alternative as the bestalternative. However, 27 % of the calculations choosethe forth alternative as the best alternative. However,decision maker can decide on the importance of criteriaand can choose any weight ranking.
CONCLUSIONThe objective of this study is to analyze the potential
alternatives and to choose the best alternative by usinga multi-criteria approach. In this paper a Fuzzy-AHPand PROMETHEE based method for ranking potentialcompanies which offer WEEE recycling outsourcingservice for Electrical and Electronic EquipmentManufacturers is proposed. This evaluation is basedon the comparisons of alternatives according to theirperformances with respect to relevant criteria. Theranking of the approach has the purpose of discoveringthe worst and best alternatives. Fuzzy method is usedfor dealing uncertainty and improving lack of precisionin evaluating criteria. Using fuzzy numbers enablesdecision makers to get better results in the overallimportance of criteria. In other words, using linguisticpreferences can be very useful for uncertain situations.
S. S. KaraInt. J. Environ. Sci. Tech., 8 (2), 291-304, Spring 2011
303
Triangular numbers are applied into traditional AHPmethod in this study. PROMETHEE calculations andanalyses are done by Decision Lab 2000 software. Theresults indicate that the second alternative hassignificant advantages over all alternatives. As a resultof the study, it is found that the proposed method ispractical in ranking alternatives with respect to multipleconflicting criteria. Compared with the previousevaluation researches, following contributions aremade with the proposed method. Firstly, there is notany evaluation research about companies which offerWEEE recycling outsourcing service for Electrical andElectronic Equipment Manufacturers. Secondly; a newintegrated methodology, Fuzzy AHP-PROMETHEE, isdeveloped for evaluation process. Such a frameworkhas never being found before in the literature. Fromthe case study, it can be seen that this approach showsits potential advantage in selecting suitable alternativedue to its sound logic and easily programmablecomputation procedure. For further research,developing a group decision making system can bevery useful. In this way different authorities’ opinionscan be taken into account. Also, different hierarchicaland detailed objectives can be incorporated into thestudy. Lastly, mathematical models or metaheuristicscan be combined with the existing method.
ACKNOWLEDGEMENTSThe author is partially supported by the TUBITAK-
Turkish Scientific and Technologic ResearchAssociation and wishes to thank its financial support.
REFERENCESAraz, C.; Ozfirat, P. M.; Ozkarahan, I., (2007). An integrated
multicriteria decision-making methodology for outsourcingmanagement. Comput. Oper. Res., 34 , 3738-3756 (19pages) .
Araz, C.; Ozkarahan, I. , (2007). Supplier evaluation andmanagement system for strategic sourcing based on a newmulti-criteria sorting procedure. Int. J. Prod. Econ., 106,585-606 (22 pages).
Bicheldey, T. K.; Latushkina, E., (2010). Biogass emissionprognosis at the landfills. Int. J. Environ. Sci. Tech., 7 (4),623-628 (6 pages).
Brans, J. P.; Vincke, B. H.; Mareschal, B., (1986). How toselect and how to rank projects: The PROMETHEE method.Eur. J. Oper. Res., 24, 228–238 (11 pages).
Bozdag, C. E.; Kahraman, C.; Ruan, D., (2003). Fuzzy groupdecision making for selection among computer integratedmanufacturing systems. Comput. Ind., 51 (1), 13-29 (17pages) .
Cao, Q.; Wang, Q., (2007). Optimizing vendor selection in atwo-stage outsourcing process. Comput. Oper. Res., 34,3757-3768 (12 pages).
Chang, D. Y., (1992). Extent analysis and synthetic decision.Optim. Tech. Appl., 1, 352-355 (26 pages).
Dulmin, R.; Mininno, V., (2003). Supplier selection using amulti-criteria decision aid method. J. Purchasing Sup.Manage., 9 (4), 177-187 (11 pages).
Gottfredson, M.; Puryear, R.; Phillips, S., (2005). Strategicsourcing: from periphery to the core. Harvard Bus. Rev., 83(2), 132–139 (8 pages).
Hsu, C. W.; Hu, A. H., (2008). Green supply chain managementin the electronic industry. Int. J. Environ. Sci. Tech., 5 (2),205-216 (12 pages).
Kapepula, K.; Colson, G.; Sabri, K.; Thonart, P., (2007). Amultiple criteria analysis for household solid wastemanagement in the urban community of Dakar. WasteManage., 27 (11), 1690-1705 (16 pages).
Karamouz, M.; Zahraie, B.; Kerachian, R.; Mahjouri, N.;Moridi, A., (2006). Development of a master plan forindustrial solid waste management. Int. J. Environ. Sci.Tech., 3 (3), 229-242 (14 pages).
Karapidakis, E. S.; Tsave, A. A.; Soupios, P. M.; Katsigiannis,Y. A., (2010). Energy efficiency and environmental impactof biogas utilization in landfills. Int. J. Environ. Sci. Tech.,7 (3), 599-608 (10 pages).
Kaufmann, A.; Gupta, M. M., (1991). Introduction to fuzzyarithmetic: Theory and applications, New York: VanNostrand Reinhold.
Kuo, T. C., (2010). The construction of a collaborative-designplatform to support waste electrical and electronicequipment recycling., Robotic and Computer-IntegratedManufacturing, 26, 100-108 (9 pages).
Lin, K. H.; Shih, L. H.; Lee, S. C., (2010). Optimization ofproduct line design for environmentally conscioustechnologies in notebook industry. Int. J. Environ. Sci. Tech.,7 (3), 473-484 (12 pages).
Linder, J. C., (2004). Transformational outsourcing. MITSloan. Manage. Rev., 45 (2), 52-58 (7 pages).
Liu, R. R.; Tian, Q.; Yang, B.; Chen, J. H., (2010). Hybridanaerobic baffled reactor for treatment of desizingwastewater. Int. J. Environ. Sci. Tech., 7 (1), 111-118 (8pages) .
Mergias, I.; Moustakas, K.; Papadopoulos, A.; Loizidou, M.,(2007). Multi-criteria decision aid approach for the selectionof the best compromise management scheme for ELVs: Thecase of Cyprus. J. Hazard. Mater., 147, 706-717 (10 pages).
Queiruga, D.; Walther, G.; Gonzalez-Benito, J.; Spengler, T.,(2008). Evaluation of sites for the location of WEEErecycling plants in Spain. Waste Manage., 28 (1), 181-190(10 pages).
Rousis, K.; Moustakas, K.; Malamis, S.; Papadopoulos, A.;Loizidou M., (2008). Multi-criteria analysis for thedetermination of the best WEEE management scenario inCyprus. Waste Manage., 28 (10), 1941–1954 (14 pages).
Salman Mahini, A.; Gholamalifard, M., (2006). Siting MSWlandfills with a weighted linear combination methodologyin a GIS environment. Int. J. Environ. Sci. Tech., 3 (4),435-445 (11 pages).
Tuzkaya, G.; Gülsün, B., (2008). Evaluating centralized returncenters in a reverse logistics network: An integrated fuzzymulti-criteria decision approach. Int. J. Environ. Sci. Tech.,5 (3), 339-352 (14 pages).
Tuzkaya, U.R., (2009). Evaluating the environmental effectsof transportation modes using an integrated methodology
S. S. Kara
How to cite this article: (Harvard style)Kara, S. S., (2011). Evaluation of outsourcing companies of waste electrical and electronic equipment recycling. Int. J. Environ. Sci.Tech., 8 (2), 291-304.
AUTHOR (S) BIOSKETCHESKara, S. S., Ph.D. Research assistant, Department of Industrial Engineering, Yildiz Technical University, Yildiz, Turkey.Email: [email protected]
304
and an application. Int. J. Environ. Sci. Tech., 6 (2), 277-290 (14 pages).
Tuzkaya, G.; Ozgen, A.; Ozgen, D.; Tuzkaya, U. R., (2009).Environmental performance evaluation of suppliers: A hybridfuzzy multi-criteria decision approach. Int. J. Environ. Sci.Tech., 6 (3), 477-490 (14 pages).
Vego, G.; Kucar-Dragicevic, S.; Koprivanac, N., (2008).Application of multi-criteria decision-making on strategicmunicipal solid waste management in Dalmatia, Croatia.Waste Manage., 28, 2192–2201 (10 pages).
Wang, J. J.; Yang, D. L., (2007). Using a hybrid multi-criteriadecision aid method for information systems outsourcing.Comput. Oper. Res., 34, 3691-3700 (10 pages).
Yang, D. H.; Kim, S.; Nam, C.; Min, J. W., (2007). Developinga decision model for business process outsourcing. Comput.Oper. Res., 34 (12), 3769-3778 (10 pages).
Zadeh, L. A., (1965). Fuzzy sets. Info. Control, 8, 338-353(16 pages).
Zimmermann, H. J. , (1991). Fuzzy set theory and itsapplications, 2nd ed., London: Kluwer Academic Publisher.