Quality functiondeployment(QFD)isausefulcustomer-drivenproductdevelopmenttoolfortranslatingtheneeds
of thecustomerintoefficientcommunicationthroughthevariousstagesofproductplanning,design,engineering,and
manufacturingtoachievehighercustomersatisfaction.QFDwasconceivedinJapaninthelate1960sbyAkaoand
then disseminatedworld-wide [1]. Ithasbeensuccessfullyintroducedinmanyindustriestoimproveproductdesign,
decision makingprocess,andcustomersatisfaction [12,16,22]. AtypicalQFDprocessconsistsoffourphases.Inthe
newproductdesignstage,aQFDteamisorganizedtoimplementthefirsttwophases,i.e.,productplanningandpart
deployment,ofQFDprocesses,whichmirrortheproductdesignprocesstoachievegreatcustomersatisfaction.The
twophasesarecloselyrelatedatthedesignstage,sincetheoutcomefromthelatterphaseshouldmakethedecisions
from theformerphaseapplicable.However,mostoftheexistingliteratureonlyfocusesonthefirstphaseofQFD,so
this studywillconsiderthetwophasesinQFDactivitiesforanewproductdesign.The houseofquality(HOQ)isanimportanttoolforQFDactivities,containinginformationon“what”,“how”,
relationship between“what”and“how”,and,therelationshipbetweenthe“how”factorsthemselves.Forexample,for
the developmentofphase1,theQFDteamdeterminesasetof“what”,i.e.,customerrequirements(CRs),and“how”,
i.e., designrequirements(DRs),thataffectCRs,andthenconstructstherelationstrengthbetweenCRsandDRsandthe
relationships betweentheDRsthemselves.TheobjectiveofthefirstphaseistodeterminethefulfillmentlevelofDRs
to maximizethecustomersatisfaction.Inphase2,theDRsareconsideredas“what”,andpartscharacteristics(PCs)
are “how”factorsdesignedtomaximallyachievetherequirementlevelofDRs.IntraditionalQFD,themeasurements
of importanceof“what”andtheassociatedrelationshipsareassessedonascalesystemsuchas1–3–9,or1–5–9,
indicating “weak”,“moderate”,and“strong”,respectively [11,16]. Nevertheless,gatheringsufficientknowledgeand
information aboutthemeasurementsisusuallydifficult [9,10]. Thishasledtotheintroductionoffuzzyapproaches
into QFDstudies [5,6,9,10,18]. However,theseexistingstudiesstillsufferfromanumberofdrawbacksinapproaches
and methodologies,asmentionedbyChenandWeng [9].
In practice,effectivedesignprocessescanleadtotheachievementofmaximumcustomersatisfaction.Forinstance,
in phase1ofQFD,DRsareabletoachievebettereffectivenessofthedesignplanningwhentheirperformanceishighly
correlated withCRstomeetexternalcustomersatisfaction.TheeffectofPCsonDRsinphase2iscomparabletothat
of DRsonCRs.However,thefulfillmentlevelsofDRsandPCsarenotnecessarilyproportionallyconsistentwithCRs’
and DRs’satisfaction,respectively.Kanoetal. [17] proposed thatqualityperformanceofaproductorservicehasthree
differentrelationships(i.e.,attractive,one-dimensional,andmust-be)withcustomersatisfaction.Someresearchers
incorporated Kano’sideaintoQFDforassigningweightstodifferent“what”toexplainhowtheyimpactcustomer
satisfactionindifferentways [15,25,30–32]. Instead,recentlyChenandKo [8] adopted Kano’sconcepttocategorize
DRs intheirfuzzymodels.Referringto [8], thisstudyappliesKano’sconcepttoclassifyvarious“how”intothree
categoriesbasedontheirimportanceranking,sincetheperformanceofvarious“how”achievesdifferentlevelsof
external/internalcustomersatisfaction.
Furthermore, inordertodecreaserisksinherentinofnewproductdesign,theriskanalysisofDRsisnecessaryatthe
design stageofnewproductdevelopment.Theoutcomesofphase1,i.e.,thefulfillmentlevelsofDRs,areappliedto
phase 2astheconstraintfactorsindeterminingtheachievementlevelsofPCs.Withthisconsideration,failuremodeand
effectanalysis(FMEA)isappliedforriskanalysis.TheinfluenceofDRs’risklevelsonPCsinphase2isconsideredin
the decisionmodels.FMEAisasystematictechniqueforidentifying,prioritizingandactingonpotentialfailuremodes
before thefailuresoccur.SeveralresearchershavediscussedFMEAanditsapplicationstoQFD [2,13,29,30]. However,
these studiesareonlylimitedtodescriptiveanalysestoobtainthequalityandresourcebenefits.Themethodstocarry
out theaggregationoftheQFDandFMEAarenotmentioned,andtheuncertaintyatthenewproductdesignstageis
not considered,either.Inthisstudy,weconstructthefuzzynonlinearmathematicalprogrammingmodels,referringto
Kano’sconceptinphase1;furthermore,weintroduceFMEAintothefuzzyQFDapproach,andproposeanideato
link thefunctionrelationshipbetweenphases1and2oftheQFDindeterminingtheachievementlevelsof“how”in
phase 2tomaximizetheinternalandexternalcustomersatisfactionundertherelevantconstraints.Anexampleofa
semiconductor packingcaseisgiventoillustratetheproposedmodels.
In thefollowingsection,afuzzyapproachtoQFDisintroduced.InSection3,afuzzynonlinearmodelinphase1
of theQFDprocessisproposedbasedonKano’sconcepttocategorize“how”intothreecharacteristics.Inthefourth
section, thefuzzyFMEAapproachisintroducedandafuzzynonlinearprogrammingmodelisdevelopedtodetermine
the achievementdegreeofPCs,constrainedbytheneedofDRsinphase1andtheriskratingofPCsaccordingto
the developmentofFMEAonDRs.Anexampleofasemiconductorpackingcaseispresentedtodemonstrateour
approaches inSection5.Finally,theconcludingremarksareprovidedinSection6.
Determining thefulfillmentlevelsofthe“how”isanimportanttaskinthefirsttwophasesoftheQFDprocesses,
when QFDisappliedtothenewproductdesign.However,mostrelatedstudieshaveonlyfocusedonthepriorityor
achievementlevelsofDRsinphase1.Unliketheexistingresearchworks,bothphases1and2areconsideredinthis
paper.ConsideringKano’sconcept,thispaperproposesafuzzynonlinearmodeltodeterminetheperformancelevels
of the“how”basedontheexistingfuzzylinearmodel.The“how”canbecategorizedintothreeclassesbasedonKano’s
concept. Throughthemodificationofthedecisionvariablesdenotingtheperformancelevel,thefuzzynonlinearmodels
can beconstructedforphases1and2inafuzzyenvironment.Comparedwithfuzzylinearmodelswithoutconsidering
Kano’sconcept,fuzzynonlinearmodelscanachievebettersatisfactiondegree.
Toreducethedesignrisk,riskanalysisofDRs,namelyafuzzyFMEA,istakenintoaccountinthephase2modelas
a constraintfactor.Inaddition,thispaperconsidersthecloselinkbetweenthetwophasestobuildupafuzzynonlinear
programming modelforphase2indeterminingthefulfillmentlevelsofPCs.ThesatisfactionofDRj determined in
phase 2isgreaterthanorequaltothecontributionofDRj in phase1inordertorealizecustomersatisfaction.The
proposed modelisillustratedwithanumericalexampletodemonstrateitsapplicabilityinpractice.Theresultingranges
of satisfactiondegreesandthepossiblerangesoffulfillmentlevelsofthe“how”inbothphasescanprovidetheQFD
team withusefulinformationfornewproductdesign.BesidesKano’sconceptandFMEA,infutureresearchother
considerations, suchasmeans-endchain(MEC),theintegrationofmorephasesofQFDactivity,andsoon,couldbe
modeled intheQFDprocessesforthenewproductdesignanddevelopment.Inaddition,moreelaborateapproaches
could bedevelopedtosolvetheproblem,inwhichallfuzzyevaluationsareusedasinputsinthemodelsinorderto
obtain completeinformation.
