• Aucun résultat trouvé

A decision support system to manage the quality of End-of-Life products in disassembly systems

N/A
N/A
Protected

Academic year: 2021

Partager "A decision support system to manage the quality of End-of-Life products in disassembly systems"

Copied!
5
0
0

Texte intégral

(1)

Any correspondence concerning this service should be sent to the repository

administrator:

staff-oatao@listes-diff.inp-toulouse.fr

Open Archive TOULOUSE Archive Ouverte (OATAO)

OATAO is an open access repository that collects the work of Toulouse researchers and

makes it freely available over the web where possible.

This is an author-deposited version published in:

http://oatao.univ-toulouse.fr/

Eprints ID: 16631

To cite this version:

Colledani, Marcello and Battaïa, Olga A decision support system to

manage the quality of End-of-Life products in disassembly systems. (2016) CIRP Annals -

Manufacturing Technology, vol. 65 (n° 1). pp. 41-44. ISSN 0007-8506

Official URL:

http://dx.doi.org/10.1016/j.cirp.2016.04.121

(2)

A

decision

support

system

to

manage

the

quality

of

End-of-Life

products

in

disassembly

systems

Marcello

Colledani

(2)

a,

*

,

Olga

Battaı¨a

b

aPolitecnicodiMilano,DepartmentofMechanicalEngineering,VialaMasa,1,20156Milan,Italy b

InstituteSuperieurdel’Ae´ronautiqueetdel’Espace,ISAE-Supaero,Toulouse,France

1. Introduction,motivationandobjectives

Circular Economy has been recently proposed as a new paradigm for sustainable development, showing potentials to generatenewbusinessopportunitiesinworldwideeconomiesand to significantly increase resource efficiency in manufacturing [1]. However, a sustainable transition to Circular Economy businesseswillneedtobesupportedbysubstantialtechnological improvements. Remanufacturing is one of the key enabling technologiesforanefficientimplementationofCircularEconomy, inseveralhigh-techindustries,suchastheautomotive, aeronau-tics, and electronics. Remanufacturing includes the set of technologies, tools, and knowledge-based methods to recover and re-use functions and materials from post-consumer high-value products. Advantages of remanufacturing include, on average, 80–90%savingsin raw materialsand energy,avoiding thedisposal costs that manufacturers by law have to support. Moreover,remanufacturingcanallowageneralpricereductionof 35–40%withanaveragemarginof20%intheaftermarket,which nowadaysrepresentsabout50%oftheautomotiveentirebusiness [2].Duetothesefeatures,remanufacturingisachievingincreasing importanceintheworldwidepolitical andresearchagendas,as reportedinthe2015G7SummitDeclaration[3].

Aremanufacturingprocess-chaintypicallyincludesdisassembly, cleaning,re-conditioningandre-assemblystages.Theperformance ofremanufacturingsystemsisstronglyboundedbythenegative impactofvariabilityanduncertaintyduetothehighvariabilityof returnproductconditions,mainlycausedbydifferentage ofthe return products, variable use modes by the customers, and environmentalconditions.Theeffectishighvariabilityin disassem-blyprocessing timesand risk ofencountering technical process

feasibility issuesin disassembly [4]. To overcomethis problem, remanufacturersusuallycarryoutpre-processvisualandfunctional inspectionsandassesstheconditionsoftheinputreturnproductsto identify remanufacturable and damaged units. This decision is extremelycriticalasitentailsarisktoeitherprocessapartthat showstobedamagedonlyafterexpensivedisassemblytaskshave beenalreadyperformed,ortodiscardpotentiallyre-usablereturn products.However,qualityinformationaboutreturnproductsis rarelyusedinindustrytoadaptthedisassemblystrategy.

In the scientific literature, the benefit of sorting the return productshasbeenshownandgradingthemforqualitytoimprove the disassembly has been recommended [5]. For example, the experimentalstudyreportedin[6]showsthatincreasingthe pre-process inspection effort may be beneficial in decreasing the disassemblytime.Inspiteoftherelevanceofthiscorrelationinthe remanufacturingindustry,thedevelopmentofspecific methodol-ogies to plan the disassembly process in presence of quality uncertainty of return products has received relatively low attentionintheliterature.Theproblemsofdefiningthebestlevel of disassembly [7] and balancing disassembly lines have been traditionally approached under a deterministic point of view [8]. Recently, attempts to include variability in the task time duration,duetotheeffectofmanualoperations,havebeenmade [9]. A methodto solve the line-balancing problem considering complete productdisassemblyandmultiplequality classeswas proposed[10].Theproblemoffindinganoptimizedre-assembly strategyunderuncertaintyinthequalityofreturnproductshas beenconsideredin[11].Asamatteroffact,amethodtojointly define the best disassembly level and line balancing when disassembly task times depend on the input product quality classification outcomes has never been proposed. Important questions like ‘‘what is the economic impact of adapting the disassemblystrategy basedonthequalityofthespecificreturn

ARTICLE INFO

Keywords: Disassembly Partquality

Decisionsupportsystem

ABSTRACT

Thequalityofpost-consumerproductsisoneofthemajorsourcesofuncertaintyindisassemblysystems. This paperdevelopsamethodology todesigndisassemblylines undervariabilityof theEnd-of-Life productquality,withtheobjectivetomaximizetheprofit.Thisdecisionsupportsystemhelpstotake decisionsaboutthedepthofdisassemblyandtheorganizationofthedisassemblysystem,dependingon thepre-processmeasurementofthekeyqualitycharacteristicsoftheproducttobedisassembled.The industrial benefits are demonstrated in a real industrial case focused on the remanufacturing of mechatronicpartsintheautomotiveindustry.

(3)

productunderprocessing?’’remainunsolved,thusundermining theefficiencyofremanufacturingsystems.

Thispaperproposesforthefirsttimeadecisionsupportsystem based on a methodology to optimize the specific disassembly strategyandprocess,consideringvariablequalityconditionsofthe inputreturnproducts,intheremanufacturingplanningphase.The decisionsupportsystemcanbeusedatshopfloorleveltosupport theoperatorsinadjustingthedisassemblysequenceonthebasisof the measured key quality characteristics of the input return products.Theproposedmethodologyisappliedtoarealcasestudy in the automotive part remanufacturing sector, to show its industrialbenefits.

2. Disassemblyproblemformulation

Inthispaper,weconsideraremanufacturingprocesswherethe disassemblytasktimedurationanddistributiondependson the qualityoftheincomingreturnproducts.Itisassumedthatasetof criticalproductcharacteristics,definedforthereturnproduct,can beobservedatthebeginningofthedisassemblyprocessby pre-process inspection. For example, characteristics such as the corrosiononthepart,corrosiononthescrews,ortheweightof thepartcanbeconsidered.Accordingtotheoutcomeofinspection, thereturnproductscanbeclassifiedintoqualityclasses.Aquality classrepresentsapropercombinationofcriticalproduct charac-teristicvalues that is usefulto aggregatereturn products with similardisassemblychallenges.Thepartsineachqualityclassare characterized by the same task time distributions. However, differentqualityclassesmaypossessdifferenttimedistributions foridenticaldisassemblytasks.

Thejointdisassemblylevelandlinebalancingproblems,under definitionofthetasksfordifferentqualityclassesoftheproduct,can beformulatedasfollows.Theobjectiveistodesignadisassembly lineconsistingofasequenceofworkstationsJ.Thesetofpossible disassemblytasksIpisgivenforeachproductp2P,butitispossible

toassignonlyasubsetI

pofthesetIpifcompletedisassemblyisnot

economicallyprofitable.ThesetIrepresentsthewholesetofall possibledisassemblytasksforallproductsI¼Sp2PIp.Notethat

eachqualityclassismodelledasadifferentproductp.Theobjective istomaximizetheprofitofthedisassemblyline,whichisdefinedas thedifferencebetweenthenetrevenueoftherecoveredpartsofthe End-of-life(EOL)productsandthelineoperationcost.Thelatter comprisestwotypesofcostsdefinedasfollows:

Fc,afixedcostforoperatingatimeunitofastation,

Ch,anadditionalfixedcostforoperatingatimeunitofastation

handlinghazardousparts.

Theprecedencerelationsaremodelledbyaprecedencegraph whichiscreatedforeachproducttobedisassembled.Foreachtask iforproductp,Ppiisdefinedasthesetofdirectpredecessorsoftask

i2Ip,withp2P.Allpredecessorsoftaskishouldbeassignedbefore

taskiinthesequenceoftasks. Disassembly task times t

˜pi;p2P;i2Ip, are assumed to be

random variables with known probability distributions. A disassemblytask is not preemptive andcan beperformed by any, but only one, workstation. A task i2HI is called hazardous ifitgenerates a hazardous part;H is theset of all hazardoustasks.

Itshouldbealsonotedthatthesetuptimebetweendifferent disassemblytasksshouldbetakenintoaccount.Thissetuptimeis sequence-dependent.Forthisreason,thetasksshouldnotonlybe assignedtostationsbutalsosequencedateachstation.

3. Descriptionofthedisassemblyplanningprocedure 3.1. Adoptednotation

Tomodelthedisassemblylinedesignproblem,thefollowing notationsareintroduced.

Parameters:

P:setofproductstobedisassembled;

Qp:thequantityofproductsoftypeptobedisassembledduring

theplanningperiod; M:setofparts’indices;

Mpi:setofrecoveredpartsiftaski2Ipisperformedonproductp,

MpiM;i2Ip;

rm:revenuegeneratedbyapartm;m2M;

diq:setuptimenecessarytoswitchfromtaskitoq;

T:lengthofplanninghorizon; TT:takttime;

hi=1,iftaskiishazardous,hi=0,otherwise,i2I

Ppi:setofdirectpredecessorsoftaski2Ipp2P;

q0:maximumnumber oftasksauthorizedtobeassigned toa

station.Eachstationcannotcontainmorethanq0tasks;

j0:upperboundonthenumberofstations;

l0=q0*j0:maximallengthofthesequenceoftasks(longerthan

thenumberoftaskssincesomepositionsforsomestationscan beempty);

S(j):setofpossiblepositionsfortasksatstationj.Thissetisgiven byanintervalofindexesandthemaximumpossibleintervalis S(j)={q0(j1)+1,q0(j1)+2,...,q0*j},j=1,2,...,j0;

J(i):setofstationswheretaskicanbeprocessed,J(i){1,2,..., j0},calculatedonthebasisoftheprecedenceconstraintsandthe

takttime;

Lp(i): setof possible positions for taski for productp in the

sequenceofalltasks,Lp(i){1,2,...,q0j0};

Np(j):setoftasksforproductpwhichcanbeprocessedatstationj;

Ap(l):setoftasksforproductpwhichcanbeassignedtopositionl.

Binarydecisionvariables yj¼

1 ifstation j isopen; 0 otherwise:



xpil¼

1 iftask i forproduct p isassignedatposition l; 0 otherwise:



zj¼ 10 ifotherwise:ahazardoustaskisassignedtostation j;



Auxiliaryrealdecisionvariables



t

pl setup time required between tasks assigned to the same

stationatpositionlandl+1forproductp;

Prjminimalprobability(amongallproducts p)torespecttakt

timeatstationj;Prj2[0,1],i.e.theprobabilitythatthevalueof

thesumoftasktimesatstationjisinferiororequaltoagiven valueoftakttimeTT.

3.2. Objectivefunctionandconstraints

Theobjectivefunctionaimsatmaximizingthedisassemblyprofit takingintoaccounttherevenuefromretrievedpartsandthecostof all opened stations, including supplementary cost for treating hazardoustasks.Theobjectivefunctionisformulatedasfollows:

max X p2P X i2Ip X l2LpðiÞ X m2Mpi QprmxpilT FC X j2J yjþCh X j2J zj 0 @ 1 A 8 < : 9 = ; (1)

Theintroducedconstraintsaredescribedinthefollowing.Each disassemblytaskisassignedatmostonce(atonepositionl)andfor each productp, a disassembly task can benot assigned if the disassemblyispartial.

X

l2LpðiÞ

(4)

Eachpositionisoccupiedbyatmostonetask: X

l2ApðlÞ

xpil1;

8

p2P; l¼1;2;...l0 (3)

Eq.(4)assuresthatataskisassignedtostationjtoaposition l+1onlyifanothertaskisalreadyassignedtotheprecedingplace (l)ofthesequencetothesamestation(thereisnoemptyplacein thesequenceofassignedoperationsinsideastation).

X i2ApðlÞ xpil X i2Apðlþ1Þ xpiðlþ1Þ;

8

l2SðjÞnmaxfSðjÞg; j ¼1;2;...;j0

8

p2P (4)

Eq.(5)assuresthatstationsareopenedinanincreasingorder, withoutemptystations:

X i2Aðl0Þ xpil0 X i2Apðl00Þ xpil0; l0¼q0ðj1Þþ1; l00¼q0jþ1; j ¼1;2;...;j01;

8

p2P (5)

Eq.(6)assuresthatastationisopenedifatleastoneoperationis assignedtoit:

yj

X

i2Apðl0Þ

xpil0; l0¼q0ðj1Þþ1;

8

p2P; j¼1;2;...;j0 (6)

Moreover,theprecedenceconstraintsfortasksisformulated:

1þ X l2LpðqÞ l xpql X l2LpðiÞ l xpil;

8

p2P;

8

i2Ip;

8

q2Pip (7)

Astationisconsidered‘‘hazardous’’ifatleastonehazardous taskisassignedtoit.Therefore,thisconstraintfollows:

zj

X

l2LðjÞ

hi xpil;

8

p2P; j¼1;2;...;j0;

8

i2Ip\NðjÞ (8)

Eq.(9)calculatestheadditionaltimebetweentaskiandtaskq whentaskqisprocesseddirectlyaftertaskiforproductpatthe samestation:

t

pl X q2Apðlþ1Þnfig diq ðxpilþxpqðlþ1Þ1Þ;

8

p2P;

8

i2ApðlÞ;j2J;

8

l2SðjÞnmaxfSðjÞg (9) The probability to respect the assigned takt time at each workstationiscalculatedfortheworstcaseamongallproducts: PrjPr X l2SðjÞnmaxfSðjÞg

t

plþ X i2NpðjÞ X l2SðjÞ tpið

j

˜Þ xpilTT 0 @ 1 A;

8

p2P; j2J (10) Finally,thecycletimeconstrainthastobejointlysatisfiedwith atleastaprobability(1

a

),where

a

isfixedbythedecisionmaker:

PrðPrjTT

8

j¼1;2;...;j0Þ1a (11)

Thisproblemissolvedusingstochasticprogramming techni-ques(formoredetailsonthemethodsee[12]).Itshouldbenoted that in order to speed up the optimization process the task assignment to workstations for different products can be parallelizedwithjointverificationofthetakttimeconstraints. 4. Applicationtoarealremanufacturingindustrialcase

Theproposedapproachhasbeenappliedtothe remanufactur-ing of automotive mechatronic products at Knorr-Bremse, a leadingglobalproviderofbrakingsystemsforrailandcommercial vehicles. Remanufacturing is an established and profitable business for the company [13]. One of the most successfully remanufacturedproductsistheElectronicBreakingSystem (EBS-2). Recently, the next generation EBS-5 breaking system has startedtobecollectedbytheaftermarketlogisticsnetwork.The companyis thereforeinterestedin thedefinitionofa profitable disassemblystrategyforthisproduct.Inparticular,theanalysisis

focused on one critical mechatronic component of the EBS-5, whichistheTrailerControlModule(TCM).AviewofaTCMandits sub-componentsisreportedinFig.1.

Todesignanefficientdisassemblyprocessforthisproduct,the case studywasconductedfollowingtwomain phases.Thefirst phase aimedat analysingthe real return products in order to determine the number of different quality classes and the discriminatingfactorstobeinspectedinordertoclassifyareturn productinoneoftheclasses.Inthesecondphase,theinformation aboutthequalityclasseswasprovidedininputtothemodel(1)– (11)inordertodeterminethenumberofstationsandthetasksto be performed at each station for each quality class, with the objectivetomaximizetheoverallsystemprofit.Theperformance oftheoptimalconfigurationsolutionwascomparedwiththeone ofasystemconfigurationobtainedwithoutdifferentiatingquality classes.Thetwomainphasesaredescribedinthefollowing.

Definitionofqualityclasses.Asampleof60post-consumerTCMs hasbeen processedin theMechatronicsDemanufacturing Pilot Plant at ITIA-CNR, Milan [14]. The parts in the sample were analyzedandclassifiedintermsofquality,accordingtoasetof qualitative and quantitative classification criteria, designed in cooperation with the company and reported in Fig. 2. These classification criteria could be either observed on the external surfaceofthereturnproductormeasuredin-linebytheoperator. Inordertoanalyzetheimpactofeachclassificationcriterionon thedisassemblyprocess,completedisassemblyhasbeenperformed foreachpartinthesample.Theinformationonthefeasibilityof disassemblytasks,thequalityoftheobtainedcomponentsinterms ofcorrosionlevels,andthedisassemblytimeshasbeencollected. TheANOVAanalysisofthecollecteddatashowedthatonlythree out of the eight considered quality features proved to be significantlyaffectingthedisassemblyprocess,amongwhichthe returnproductage,itslevelofdirtinessandthelevelofcorrosionon screws. By analysing alsothe interactions among these quality features,sixqualityclassesthat,accordingtotheanalyzedsample, significantlyaffectthedisassemblyprocesshavebeendetermined. Disassemblylinedesign.Inthesecondphaseoftheactivity,the methodproposedinthispaperhasbeenusedtojointlyoptimize the disassembly level and to balance the line, considering the characterizedqualityclasses.Thelistoftasksandtheestimated tasktimes,foreachqualityclass,Q1,Q2,...,Q6,arereportedin

Table1.Inthistable, thetimeunit(TU)durationis omittedfor

confidentialityreasons.Theprobabilityofapartbeingassociated to a specific quality class was estimated by the experimental resultsdescribedaboveandwasrespectivelyequal to0.15, 0.1, 0.25,0.15,0.2,0.15.Anytaskcouldbeassignedtoanyworkstation undertherespectoftheprecedenceconstraints.Setuptimesexist betweentasksandvarybetween0.5and1TU.

Fig.1.EBS-5,TCM(left)anditssub-components(right).

(5)

The optimal designs of the disassembly line, considering a targettakttimeof60TUarereportedinTable2.Inparticular,two configurations are compared. Configuration 1 is obtained by solving the disassembly line design problem by considering a single aggregated quality class with task times obtained by averagingthevalue of the6 quality classes. Configuration 2 is obtainedbyadoptingtheapproachproposedinthispaper,which considersthedifferentqualityclassesintheproblem.

As it can be noticed, if the quality classes are neglected (configuration1)alltasksareassignedtothe2workstations.Ifthe qualityclassesareconsidered(configuration2),theoptimalsolution is the same for four classes while it differs for qualityclasses 3and 4. In particular,taskT6becomesuneconomicalanditisnotallocated.The relatedperformance,intermsofprobabilityofnotexceedingthe targettakttime,isreportedinTable 3.Asitcanbeobserved,by applying a design method that neglects the quality classes, workstation2willfailtorespectthetakttimein50%ofcaseswhere returnproductsofqualityclass3aretreated,andin22.8%ofcases wherereturnproductsofqualityclass4aretreated.Onthecontrary, thedevelopedmodelprovidesadifferentpartialdisassemblyplanfor differentpartqualityclassesgivingtheprioritytothedisassembly tasks with a higher recovering value, thus providing a higher probabilityofmeetingthetargettakttime,foreachqualityclass.

Decision support system. The obtained results confirmed theimportanceofclassifyingreturnproductsinqualityclasses.A decisionsupportsystemwasdevelopedtoenabletheoperatorto(i) characterizeandsortthepartsinqualityclassesbyusingautomatic datagatheringsystems,and(ii)tovisualizetotheoperatorsinthe

workstationthespecificstrategydependingontheclassifiedquality class of the return product. To this purpose, specific inspection technologiesandmethodsincludinganopticalcharacterrecognition approachandahyperspectralimagingsystemhavebeendeveloped forautomaticbarcodereadingandformakinginferenceonthelevel ofthecorrosionontheexternalsurfaceofthescrews.Allinall,the developedsolutionprovidedthecompanywithanewmethodology andnewenablingtechnologiesforadaptingthedisassemblyprocess dependingonthequalityclassofthereturnproduct.

5. Conclusionsanddiscussions

In this paper, a new method to support robust disassembly planningunderuncertaintyoftheinputproductqualityhasbeen proposed.Thedevelopedmethodhasbeensuccessfullyappliedona realcasestudyintheauto-partremanufacturingindustry.Numerical results show that the proposed methodology can improve the efficiencyandrobustnessofthedisassemblyprocess,thus support-ingremanufacturingandthe transitiontonew circulareconomy orientedbusiness.Futureresearchwillconcerntheintegrationof disassemblystrategydefinition,linebalancingandbufferdesignin ordertoreducetheinventoryinthesystem,whilekeepingthesame advantagesoftheproposedapproach.Moreover,thepossibilityof integratingdistributed,inlineinspectionstationstorefineproduct informationalongtheprocess-chainwillbeconsidered,underazero defectremanufacturingviewpoint.Furthermore,themethodwillbe extendedtointegratemanual,andsemi-automatictaskswithinthe planning problem, following the recentinnovations in cognitive roboticsforfuturedisassemblysystems[15].

Acknowledgements

Thisresearch hasbeenpartially supported by theEuropean Union7thFrameworkProgrammeProjectNo:NMP2013-609087, Shock-robustDesignofPlantsandtheirSupply ChainNetworks (RobustPlaNet).TheauthorswouldliketothankDr.DanielKohler, Mr.FrankMerwerthandMr.ThomasMeyerfromKnorrBremsefor theirsupportinthisresearch.

References

[1]McKinseyCenterforBusinessandEnvironment.EllenMacarthurFoundation (2015)GrowthWithin:ACircularEconomyVisionforaCompetitiveEurope. [2]WeilandFJ(2012)EuropeanAutomotiveRemanufacturing–TechnicalTrends&

MarketDevelopments.

[3]G7-AllianceonResourceEfficiency,Leaders’DeclarationG7Summit7–8June 2015.

[4]DuflouJR,SeligerG,KaraS,UmedaY,OmettoA,WillemsB(2008)Efficiency andFeasibilityofProductDisassembly:ACase-basedStudy.CIRPAnnals– ManufacturingTechnology57(2):583–600.

[5]TeunterRH,FlapperSDP(2010)OptimalCoreAcquisitionand Remanufactur-ing PoliciesUnder Uncertain Core QualityFractions. EuropeanJournalof OperationalResearch.

[6]IjomahWL,RidleyS(2015)ANovelPre-processingInspectionMethodologyto EnhanceProductivityinAutomotiveProductRemanufacture:An Industry-basedResearchof2196Engines.JournalofRemanufacturing5:8.

[7]KaraS,PornprasitpolP,KaebernickH(2006)SelectiveDisassembly Sequenc-ing:AMethodologyfortheDisassemblyofEnd-of-LifeProducts.CIRPAnnals– ManufacturingTechnology55(1):37–41.

[8]SantochiM,DiniG,FailliF(2002)ComputerAidedDisassemblyPlanning:Stateof theArtandPerspectives.CIRPAnnals–ManufacturingTechnology51(2):507–529. [9]BentahaM,BattaıaO,DolguiA,HuJ(2014)DealingWithUncertaintyin DisassemblyLineDesign.CIRPAnnals–ManufacturingTechnology63(1):21–24. [10]RiggsRJ,Battaı¨a O,HuJ(2015)DisassemblyLineBalancingUnderHighVariety ofEndofLifeStatesUsingaJointPrecedenceGraphApproach.Journalof ManufacturingSystems37(3):638–648.

[11]JinX,Ni J,KorenY(2011) OptimalControlofReassemblyWithVariable Quality Returns in a Product Remanufacturing System. CIRP Annals – ManufacturingTechnology60:25–28.

[12]BentahaML,Battaı¨aO,DolguiA,HuSJ(2015)SecondOrderConic Approxi-mationforDisassembly Line DesignWith Joint ProbabilisticConstraints. EuropeanJournalofOperationalResearch247(3):957–967.

[13]Ko¨hler DCF, MerwerthF, MechatronicRemanufacturing atKnorr-Bremse CommercialVehiclesSystems(CVS).

[14]ColledaniM,CopaniG,TolioT(2014) De-manufacturingSystems,Keynote paperatthe47thCIRPConferenceonManufacturingSystems,28–30April, Windsor,Ontario,Canada.

[15]VongbunyongS,KaraS,PagnuccoM(2013)ApplicationofCognitiveRoboticsin DisassemblyofProducts.CIRPAnnals–ManufacturingTechnology62(1):31–34. Table3

Lineperformancecomparison.

Configuration Station Qualityclasses

Q1 Q2 Q3 Q4 Q5 Q6 1(averagevalues) WS1 99 99.8 78 85 100 99.99 WS2 99.9 99.9 50 77.2 99.99 99.99 2(qualityclasses) WS1 99 99.8 99.9 99.9 100 99.99 WS2 99.9 99.9 98.7 99.5 99.99 99.99 Table2

Optimallinedesigns.

Conf. Qualityclass Station Tasksequence Meantotal stationtime Var 1 / WS1 T1,T10toT16 50 – WS2 T2toT9 50 – 2 Q1,Q2,Q5,Q6 WS1 T1,T10toT16 51,49,42,45 15 WS2 T2toT9 47,46,45,45 16.2 Q3 WS1 T1,T12toT16 49 12.5 WS2 T2toT5,T7 52 12.7 Q4 WS1 T1,T12toT16 49 12.5 WS2 T2toT5,T7,T8 50.5 13.2 Table1

Taskdescriptionforthestudycase.

ID Disassembledpart Meantasktime Var P1i

Q1 Q2 Q3 Q4 Q5 Q6 1 PCBcase 19 18 21 22 13 14 4 – 2 PCB 22 21 35 32 20 20 11 1 3 Pressuresensor 2 2 2 2 2 2 1 2 4 Fixingplate 5 5 5 5 5 5 0.2 2 5 3solenoids 5 5 5 5 5 5 0.2 4 6 Connectorcase 3 2 3 3 3 3 1 – 7 Connector 2 1 2 2 2 2 0.3 2 8 Airfilter 1 1 1 1 1 1 0.5 – 9 Pneumaticconnectors 4 4 4 4 4 4 2 – 10 Cap 2 2 4 3 1 2 0.5 – 11 Silencer 2 2 2 2 2 2 1 10 12 Mainbody 2 2 2 2 2 2 1 – 13 Cover 14 13 16 15 12 13 3 12 14 Lowerpistonpart 2 2 2 2 2 2 1 13

15 Sealing 2 2 2 2 2 2 1 14

Figure

Fig. 1. EBS-5, TCM (left) and its sub-components (right).

Références

Documents relatifs

As result, this test proved that this model allows the dynamic information representation of the current situation thanks to factual agents organisation.. Moreover we could study

Because the CCRC method was used to merge expert opinions and statistics and CRC and DRC mech- anisms were used to instrument inference in directed evidential networks, determining

It described a decision making problem in textile manufacturing (color fading ozonation optimization) to the Markov Decision Process in terms of the tuple of {S, A,

Pempa can be used to investigate the impacts of different payment schemes and quality incentives on growers’ revenue, value sharing between growers and millers and value sharing

The originality of the approach lies in (1) the combination of functionality-analysis models and continuous resilience assessment following several dimensions of systems,

Larva agents life cycle is composed of three steps: a growing step, when the agent updates its properties (development rate, stage, speed, length, etc.), a

Si l’on peut distinguer des étoiles individuelles dans un amas ou une galaxie éloignés, on peut chercher des céphéides, mesurer leur période et leur luminosité apparente,

The differential cross-section, assuming unpolarised J/ψ production, is measured as a function of the transverse momentum of the lower-p T J/ψ meson, di-J/ψ p T and mass, the