1

DecisiontreesforimplementingRapidManufacturingforMassCustomisation

DominikDeradjat*andTimMinshall

InstituteforManufacturing,UniversityofCambridgeDepartmentofEngineering,17,CharlesBabbageRoad,CambridgeCB30FS,UK

*Correspondingauthor:dominik.deradjat@gmail.com

Abstract

This paper aims to (1) compare implementation considerations and challenges formetal andpolymerrapidmanufacturing (i.e. theuseof additivemanufacturing technologies for final part production) formass customisation and (2) derive decision trees for firms seeking to implement such an approach.Implementationdatafrom10casestudiesfromthedentalandhearingaidindustrieshasbeencapturedandusedas thebasis for thecomparisonanddesignof thedecision trees.Ourobjective is toprovideevidenceontheuseofadditivemanufacturingtechnologiesasenablersformasscustomisationandtoprovidepractitionersinindustrywithguidelinesfordecision-makingonhowtoinstallandramp-upmasscustomisation production with these technologies. Common considerations and challenges for bothmetal and polymer applications have been identified. Based on these insights, eight implementationdecisiontreeshavebeencreatedandrepresentthemaincontributionofthispaper.

KeywordsAdditiveManufacturing, RapidManufacturing,Mass Customisation, Ramp-upManagement, AdvancedManufacturingImplementation

Acknowledgements

TheresearchpresentedinthispaperwassupportedbyfundingfromtheR&DManagementAssociation,theUK Engineering and Physical Sciences ResearchCouncil and theUK Economic and Social ResearchCouncil(EP/K039598/1).

brought to you by COREView metadata, citation and similar papers at core.ac.uk

provided by Apollo

2

1 IntroductionMass customisation (MC) – i.e. the production of individualised objects at mass productionlevels–canberealisedthroughrecentadvancesintheperformanceofadditivemanufacturing(AM)technologies(DeradjatandMinshall,2017;Fogliattoetal.,2012).Theseadvancesenablethe direct production of objects using AM, and this application of AM is known as ‘rapidmanufacturing’ (RM). The reason why RM is said to enable MC stems from the inherentadvantages offered by AM technologies such as superiority in customisation and productionflexibility (Weller et al., 2015). However, to date there have been very few successfulapplicationsofRMforMC(Fogliattoetal.,2012;Melloretal.,2014;Sandström,2015)andasaresult,knowledgeoftheimplementationprocessofsuccessfulRMsystemsthatrealiseMChasnot been widely diffused. In particular, the production ramp-up stage comprising the timebetweenR&Dandfullyworkingmassproductionisanareathathasnotbeenwellresearched.

TherearethreetypesofchallengesthatcompaniesencounterwhenseekingtoapplyRMasaproductionprocess(Ruffoetal.,2007):

1. Manufacturingprocessesandmaterials2. Design3. Management,organisationandimplementation

Thispapertargetsthethirdcategorywithanemphasisonimplementation.

ThereareseveralindustriesapplyingRM(WohlersandCaffrey,2016)butveryfewareusingitfor mass customisation. ‘True’ mass customisation (MC) has to fulfil the requirement ofproviding highly individualised products at high output numbers (Piller, 2008). Two mostsuccessful areashavebeendental andhearing aid applications (Deradjat andMinshall, 2015;Mellor,2014).BuildingontheworkofDeradjatandMinshall(2017)thatfocusedspecificallyonmetal RM implementation for MC in the dental industry, this paper seeks to develop moregeneralisedanddetailedinstructionsonhowtoimplementRMforMC.Inaddition,inordertocoverthemostrelevantAMprocessforMC,comparableinsightsforpolymerRMcasesforMCneed to be established. One of the most promising applications for this type of AM is thehearing aid industry as indicated by Sandström (2015). The hearing aid industry provides anappropriatenumberofcompaniesthathavesuccessfullyimplementedRMforMCtowarrantanin-depthanalysissimilartotheonecarriedoutbyDeradjatandMinshall(2017).

Our researchtargetsgaps in literatureonMCenablersbyprovidinganunderstandingofhowenterprises implementmetal and polymer RM forMC. There is a distinct lack of research inliterature regarding in-depth implementation instructionsof how to implement thediscussedprinciples. Inaddition,thepaperprovidesdetailed instructionsfordecisionmakersonhowtoimplementRMforMCintheformofdecisiontrees.Withinthiscontext,thefollowingresearchquestionandsub-questionhasbeenformulated:

3

• Researchquestion:WhichaspectsdocompaniesseekingtoimplementRMforMChavetoconsiderintheirdecision-makingprocess?

• Researchsub-question:Whichcommonconsiderationsandchallenges1existduringtheimplementationprocessofRMforMCformetalandpolymerRM?

In the following section, a review of literature onMC and RM is presented. The subsequentsectionpresentstheresearchmethodologyandcontextual informationregardingthepolymerRMforMCcases.Next,theresearchresultsfromthehearingaidindustrycasesarepresentedandcontrastedagainsttheresultsobtainedfromDeradjatandMinshall(2017).Theresultswillbe presented according to the framework categories of strategic, technical, operational,organisational and external considerations. Based on these insights a decision tree will bederivedforeachofthestatedcategories.Thepapercloseswiththeconclusions,limitationsoftheresearchandsuggestionsforfurtherresearch.

1Thepaperwillmerge‘considerationsandchallenges’since‘considerations’canrepresentchallengesandviceversa.

4

2 Literaturereview

2.1 AdditiveManufacturingtechnologiesandRapidManufacturing

RapidManufacturing describes the use of AdditiveManufacturing technologies for the production offinalparts.AccordingtoHopkinsonetal.(2006,p.1),itisdefinedas:

“the use of a computer aided design (CAD)-based automated Additive Manufacturing process toconstructpartsthatareuseddirectlyasfinishedproductsorcomponents”.

AdditiveManufacturing(AM)isdefinedas:

“the process of joining materials to make objects from 3D model data, usually layer upon layer, asopposedtosubtractivemanufacturingmethodologies”(ASTM,2012,p.2).

WongandHernandez(2012)classifyAMaccordingtothemethodofmaterialsupply into liquidbased,solidbasedandpowderbasedsystems(SeeFigure1).ProcessesutilisedinourcasestudiesareSelectiveLaser Sintering (SLS), Selective LaserMelting (SLM) andDirectMetal Laser Sintering (DMLS) formetalpartsandStereolithography(SLA)andDigitalLightProcessing(DLP)forpolymers.

Figure1:ClassificationofadditivemanufacturingtechnologiesbasedonWongandHernandez(2012)

RM evolved from Rapid Prototyping, which describes the creation of prototypes through AM. RapidPrototyping was first deployed in the 1990s (Atzeni et al., 2010). Over time AM technologies haveevolved to realise the production of moulds and tooling inserts (Pham and Dimov, 2001). Increasedinvestment in R&D forAM technologies over thepast 10 years have resulted in improvements in theperformanceofAMsuchthatithasbecomecapableforfinalpartproduction(RM)incertainapplications(Melloretal.,2014).UtilisingAMfordirectproductionhasseveraladvantagescomparedtotraditionalmanufacturingtechnologies(Holmströmetal.,2010):

1. Absenceoftoolingrequirementsreducesproductiontimeandexpenses

AMprocesses

SolidbasedLiquidbased Powderbased

Melting Polymerisation

FusedDepositionMelting

Stereo-litho-graphy

Polyjet

LOM Melting Binding

Selectivelaser

sintering

Selectivelaser

melting

3Dprinting Pro-metal

Directlasermetal

sintering

Electronbeammelting

Laminatedengineerednet

shaping

DLP

5

2. Smallproductionbatchesbecomefeasibleandeconomical3. Quickdesignchangesarepossible4. Productioncanbeoptimisedinregardtofunctionalpurposes5. Customproductsbecomeeconomicallyviable6. Wasteisreduced7. Supplychainscanbesimplified8. Designofproductscanbecustomised

In particular, the benefit of design customisability has been said to realise the production strategy ofMass Customisation (MC) (Fogliatto et al., 2012). However, the literature on RM forMC (such as theworkbyReevesetal. (2011)andGibsonetal. (2010))eithermerelydescribepotentialapplicationsorenumerate existing applications without providing in depths technical and industrial insights. OnlyDeradjatandMinshall(2017)provideaframeworkofRMforMCimplementationderivedfromin-depthindustrial insightsand researchbyMelloretal. (2014).However,DeradjatandMinshall (2017)donotprovideclearguidanceonhowcompaniesshouldproceedwhenattemptingto implementRMforMC.Theirresearchisadditionally limitedtometalRM.Thus,thereisaneedtoexpandtheworktoincludepolymerAMtechnologiesinordertodevelopabroaderunderstandingofhowRMcanbeimplemented.WhileinsightsontheindustrialadoptionofpolymerRMhasbeencarriedoutbySandström(2015),hisresearch takes amore industry centric viewanddoesnot provide sufficient data to capture technicalimplementationaspects.InlightofthefoundationalworkontheimplementationofRMforMCprovidedbyDeradjatandMinshall (2017), there isaneed tosupplement thedatabase forpolymerRMand tofurtherdevelopthese insights toprovidemoreadvanced implementation instructions thatcanbenefitbothliteratureandpractitionersseekingtoimplementRMforMC.

2.2 MassCustomisation

“Masscustomisation” isatermintroducedbyDavis (1987)andPine(1993)todescribetheproductionstrategy of realising themanufacture of customised objects atmass production level. The concept isbasedonHayesandWheelwright's(1979)productprocessmatrix(Duray,2011).MCrepresentsahybridversionofone-offandmassproduction.WithintheproductprocessmatrixillustratedinFigure2,MCislocated between individual production in the top left-hand corner andmass production in the lowerright-handcorner(Tucketal.,2008).

6

Figure2:Productvariety-volumematrix(Tucketal.,2008)

‘Customisation’ denotes the input of the customer during the design process (Lampel andMintzberg,1996).AccordingtoLampelandMintzberg(1996),therearevaryingdegreesofcustomisationspanningfrom the highest degree of customisation, namely ‘pure customisation’, and ‘standardisedcustomisation’. In literature there is no prescribed level of customisation andproduction volume thatdefinesMC (BatemanandCheng, 2006). Some scholars believe thatMConly exists if the customer isable to completely customise the product in every aspect (Silveira et al., 2001). Others such asWestbrook andWilliamson (1993) accept limited customer involvement and influenceon the productdesigntoqualifyasMC.Equally,thereisnoagreementontherequiredlevelofproductionoutputthatqualifiesforMC.Instead,Durayetal.(2000)statethatinanoptimalcaseofMCproductionefficiencyofMCshouldbeclosetothoseofmassproduction.

Fogliattoetal.(2012)conductedareviewofMCliteratureandhavecategoriseditinto(i)theeconomicsof the principle, (ii) success factors, (iii) MC enablers and (iv) customer-manufacturer interaction.Consideringtheabove-mentionedadvantagesofAM,thisresearchfocusesonthebodyofliteratureonMCenablers.AparticularlackofresearchonimplementationmodelsofmanufacturingtechnologiesforMCenablershavebeenidentifiedbyFogliattoetal.(2012)andDeradjatandMinshall(2017).Giventhenoted advances of RM within recent years, there is a need to study how AM technologies can helprealiseMC.Welleretal. (2015) support thisendeavourbysuggesting thatAMallowsanenterprise toincrease profits by capturing value through flexible production of customised objects. Deradjat &Minshall (2017)proposea framework for implementationofRM forMC taking intoaccount strategic,technological, operational, organisational and external considerations (Figure 3).While their researchoutputspresentaconceptualframeworkandalistofrelevantchallenges,adetailedcontributiononhowtoproceedwiththeimplementationprocessofRMforMCisabsentinliterature.

7

Figure3:FrameworkofRapidManufacturingimplementationforMassCustomisation(DeradjatandMinshall,2017)

Inthispaper,MCisdefinedasaproductionstrategywhichrelatestothemanufactureofindividualisedand unique objects at output levels characteristic ofmass production. Each product is different fromanotherinshapeandsize.

Our review has revealed that with the benefits and advances of AM technology, RM can become anenablerforMC.WhilethereisresearchonRMandMC,thereisalackofresearchwhereandhowthesetwo fields intersect. This paper seeks to build on the foundations provided by Deradjat andMinshall(2017)andextendthemtocreatedetailedinstructionsofhowRMforMCcanberealised.Thisresearchwill supplement existing research for metal RM for MC with four cases for polymer RM and derivedecisiontreesforRMimplementationofMC.TheconceptofimplementationhasbeendefinedbyVoss(1988)alongathreestagelife-cyclemodel(Figure4).

Figure4:ImplementationprocessaccordingtoVoss(1988)

The first ‘pre-installation’ phase comprises all variables pertaining to the success or failure of theimplementationprocess.Inthesecondphase,theinstallationandcommissioningphase,aworkingorderof the applied technology on a consistent level is guaranteed. The last step, the post-commissioningphase,consistsofimprovementoftechnicalandbusinessoperations.AccordingtoVoss(1988),thisfinalphaseshouldneverendsinceaneffectivecompanyshouldseektocontinuouslyimprove.Inthecontextofthispaper,thedefinitionofimplementationstatedbyVoss(1988)willbeadopted.

PRE-INSTALLATION INSTALLATION AND COMMISSIONING

POST-COMMISSIONING

GO/NO GO GO/NO GO

8

3 ResearchmethodologyThe research is divided into two parts. Firstly, an in-depth analysis of the hearing aid industry wasconducted in an identical manner as carried out by Deradjat & Minshall (2017). The results aresummarisedinatableatthebeginningofeachsub-chapterinsection4(e.g.Figure6Figure6)togetherwith the relevant insights from Deradjat & Minshall (2017). This serves the purpose of enhancingunderstanding of RM for MC to include polymer RM applications allowing for more generalisedconclusions.Awithin-caseanalysisandasubsequentcross-caseanalysisidentifiedchallengesthatwerereoccurringamongthemajorityofthepolymerRMcases.

Secondly, the research results formetal andpolymerRMare comparedandprovide thebasis for theformulation of decision trees for the implementation of RM forMC. These decision trees are derivedprimarilyfromthedataprovidedinourcases,i.e.frominterviews,productiondataandliterature.Whilepolymer case insights are derived from primary case data (e.g. interviews, production data of thecompanies),metalcasereferencesinthispaperaretakenfromDeradjat&Minshall(2017).Thedecisiontreesrepresentthechallengesandinsightsthatcompaniesinthestatedcasestudieswerefacing.

Toaccountforthenoveltyandexploratorynatureofourresearchobjective,acasestudybasedresearchdesign ismostappropriate.Amulti-casedesignhasbeenchosen toenable intercasecomparisonandrobustnessandgeneralisabilityof the findings (HerriottandFirestone,1983).Since thepaperseeks toanalyseseveralaspects inordertocaptureacomprehensivepictureof implementationofRMforMC,multiple units of analysis are required with an embedded multi-case study design according to Yin(2009).Fourcompaniesfromthehearingaidindustryareanalysed.TheframeworkprovidedbyDeradjat&Minshall (2017)will servethepurposeofprovidingastructure for thedatagatheringphaseandforderiving decision trees for the implementation of RM for MC. The framework categories comprisecorporate strategy, technical (overall RMprocess, process front-end, AMmachine, process back-end),operational, organisational and external considerations. In order to ensure comparability of resultsbetween this research and the insights from the hearing aid industry and the dental industry fromDeradjat&Minshall(2017),thesamedataacquisitionmethodologywasused.

Awide rangeof data sources havebeenutilised. In each case, productiondata andpublicly availablefinancialdatafromannualreportsandgovernmentdatabaseshavebeenprocessed,supplementedwithinterviews with company representatives who are familiar with the RM implementation process.Additionally,toincreasevalidityofthecases,datafromAMmachineandsoftwareprovidershavebeenincluded.Table1givesanoverviewofthecasecompanies.

9

Table1:Informationonhearingaidcasecompanies

Company AMProducts Number of unitsproduced peryear

AMProcess Size ofcompany

Other Informant

Company1 Hearing aidshells

>1,000,000 SLA,DLP Germanybasedmultinationalcompany,> € 835 mannualsales

First company toimplementAMinthehearing aid industry,collaboration withCompany 3 duringpre-installationphase

Head of customhearingaidproduction

Company2 Hearing aidshells

>1,000,000 DLP US basedmultinationalcompany,> €770 mannualsales

Late follower of AMadoption

Headofproduction

Company3 Hearing aidshells

>1,000,000 DLP Switzerlandbasedmulitinationalcompany,

€2.23bannualsales

Collaboration withCompany 1 duringpre-installationphase

Director of customproducts

Company4 Hearing aidshells

>500,000 SLA US basedmultinationalcompany

Latefollower Head of productionNorthAmerica

10

4 Results and discussion The challenges involved with the implementation of RM for MC gathered fromMinshall & Deradjat(2017) and four cases from the hearing aid industry are presented in this section. Based on theseinsights,decision-makingtreesofhowtoimplementRMforthefollowingfivecategoriesasidentifiedbyMellor (2014) and Deradjat and Minshall (2017) are synthesised. Figure 5 illustrates the differentcategories of strategic, technological, operational, organisational and external and their contextualrelationtoeachother.

Figure5:Overviewdecision-makingcategories

Eachsub-sectionisstructuredtofirstanswertheresearchsub-questionandthencontributetothemainresearch question. Each subsequent section starts with an identification of considerations that havebeendeemedrelevantforeachframeworkcategory.EachofthesearethencomparedwitheachotherforbothmetalandpolymerRMforMC.Onlyconsiderationsthathaveoccurredinthemajorityofcaseswerecaptured in the followingsections.This rationalofdataanalysis corresponds to the logicofhowdatawascapturedbyDeradjat&Minshall(2017)formetalRM/dentalcases.Suchauniformwayofdataanalysisensuredcomparabilityof results.Decision-making treesare thenderived foreach sub-sectionfrom the above-mentioned insights, thus directly addressing the main research question of whichaspectscompanieshavetoconsiderintheirdecision-makingprocesswhenimplementingRMforMC.

Withinouranalysesandsubsequentderivationofdecisiontrees,eachofthefiveframeworkfactorswillbedividedaccordingtotheimplementationphasessuggestedbyVoss(1988).Toreducethecomplexityofpresentation,theresultsforthepre-installationandinstallationphasehavebeenmergedsincemanyconsiderationsoverlapped.

4.1 Strategic considerations Forcorporatestrategicconsiderations,bothmetalandpolymercasesshowedthatRMgenerallyofferedacompetitiveadvantagebecauseofhighprocessconsistencyandreducedscraprate(Table2).Easeofproductionscalabilityandcheaperproductioncostsperunitproducedwereobservedforthehearingaidcases but not for the dental refinement cases. For the dental industry, RM was primarily adoptedbecauseoftheindustry’strendtowardsdigitisedoperationsindentistry.

2.Technological

3.Operational

4.Organisational5.External

1.Strategic

11

Table2:Summaryofstrategicconsiderationsandchallenges

Figure6illustratesadecisiontreeforstrategicconsiderationsofRMforMCimplementation.Actionsandconsiderationsarehighlightedinlightgrey.ItiscrucialforcompaniesseekingtoimplementRMforMCto first determine their key objectives and then to verify whether AM is in line with their businessstrategy and whether advantages of RM compared to the traditional process are compelling. Shouldthesequestionsbenegated,AMwouldnotpresentaviableoption.Wefoundthatallcasecompanieshadtogothroughthesementionedsteps.Inallcases,theyfoundthatAMwasinlinewiththeirbusinessstrategyofmovingtowardsdigitaldentistryortoaimforproductionramp-upandbusinessexpansionasobserved in the hearing aid cases. Additionally, RM for MC offered the advantage of high processconsistencycomparedtotraditionalmanufacturingprocesses(e.g.casting).Theinvestigatedcasesshowthatthesecancomprisehigherproductionscalability,betterprocessconsistency,betterdesignfreedom,better integration possibilities with digital production, cost advantages and superior productionflexibility.

As anext step, companieshave todecidewhether theywould like tobe first to adoptRM forMCorwhether they like to follow at a later point in time. For the later option, they have to execute acomprehensive technological assessment of available components and adopt the best possible RMsystemsonthemarket.IfcompaniesarethefirsttoapplyRMforMCintheirrespectivefield,theywillhave to determinewhether they can cope financiallywith the high up-front costs and risk of failure.Should the costs and risks be acceptable, they will have to proceed to test whether technical,operational, organisational and external factors are favourable. Alternatively, collaborations can be ofuseifR&Dcostscannotbecoveredinternally.Herethecompanyneedstocarefullyassesswhethertherisksareacceptable.Company1,forexample,hadtospendsignificantresourcesasthefirstcompanyinthehearingaidindustrytoadoptAMfortheirparticularapplication.However,smallercompaniesinthedentalindustryreliedoncollaborationstoreducerisksandcosts.IncaseofRMforMCimplementation,the first-movercompanycandecidetokeeptheknowledgeor IP internalandtry to leverageAMasacompetitiveadvantageasinthecaseofSiemens(LiebermanandMontgomery,1988).Asecondoptionistopatentandlicenseelementsoftheprocess,material,etc.Thisoptionallowsincomethroughroyaltyfees, awareness of the market and opportunities for collaboration with software companies, AM-machine andmaterial providers. A third option is to file patents andnot to license the IP in order toensure a competitive advantage. This practice, however, has not been observed in any of the casecompanies.IftheupfrontR&DcostsarenotaffordableforanenterpriseseekingtoapplyRMforMCfirstin their field, it needs to verify the feasibility of collaborations. Without collaborations, RM for MCimplementationisnotrealisableatthisstage.Shouldcollaborativeworkbepossible,anevaluationoftheriskshastobeexecuted.Datasensitivityandhighfailurepotentialcanterminatefurtherdeliberation.Ifthe risks are acceptable, the other implementation decision trees need to be checked as mentionedabove.

Category Considerationsandchallenges HearingAid

Dental

Corporatestrategy

Allimplementationphases • RMofferedacompetitiveadvantagebecauseofhighprocessconsistency X X

• RMwasadoptedinthecontextofanindustrytrendtowardsdigitalgeometrycaptureandproduction X

12

Figure6:Strategicconsiderations.DecisiontreeforRMforMCimplementation

DoesAMofferanadvantageoverotherproduction

methods?

DoesAMfitintotheoverallbusinessstrategyandkeyobjectives?(Synergies

withotherdepartments,etc.?)

No No

Firstmoverorlatefollowerinthatparticularindustry?

Yes[*] Yes

ArelargeupfrontR&Dcostsaffordable?

Movetotechnicaldecisiontrees(4.2)

Arecollaborationspossibletoreducerisksandcosts?

Assessrisksofcollaboration

Aretheotherdecisiontrees

positive(4.3-4-5)?

+Royaltyfees+Awarenessofmarket

+Opportunitiesforcollaborations

- AllowingothercompaniestoleverageAMforMC(/mitigationofcompetitive

operationaladvantages)

Keepknowledgecompanyinternalor

utiliseIP?

+Potentialrealisationofproductionanddesign

superiorityovercompetitorsforacertaintime

- Othercompaniesmayfilepatentsandmitigatethe

advantage

DiscardRMforMC

DiscardRMforMC

DiscardRMforMC

DiscardRMforMC

[*]- Higherscalability- Higherprocess

consistency- Moredesign

freedom- Betterintegration

withdigitalproduction/Existingproductioninfrastructure

- Costadvantages- Higherorder

processingability- Productionflexibility

Executecomprehensivetechnologicalassessmentofavailabletechnological

componentsandadoptmostappropriate/besttechnologicalsystemson

themarket

Firstmover Latefollower

Yes No

YesNo

No Yes

Internal Patent&license Patent&don‘tlicsense

+Preventingcompetitorsfromadoptingthesameproductionprocess

- CompetitorsmaybeabletocircumventtheIP

- Informingcompetitionincountrieswithnon-stringentlawsystemstocopy

Determinekeyobjectives

Risksacceptable?

DiscardRMforMC

NoYes

Prepareimplementation

Corporatestrategy

13

4.2 Technological factors

4.2.1 Overall process

For the technological overall process considerations, setting up a production planning system andproduct identification system was pivotal during the pre-installation and installation phase for casesfrombothindustries(Table3).

Table3:Summaryofoverallprocessconsiderationsandchallenges

Thetopicofprocessconsistency inthecontextofRMforMCwascrucial forallcasecompanies.Mostcase companies spent significant resources on ensuring that the process is consistent. This includesadaptations of front-end, AM machines, post-processing and operational factors. The example ofCompany 4 which already had a consistent RM (SLA) for MC production system in place and whichrefusedtoswitchtoacheaperandmoreaccurateAMprocess(DLP)illustratesthehighlevelofresourceinput required to establish a consistent RM for MC process. Considering the importance of processconsistencyforthesuccessofRMforMCimplementation,companiesseekingtoimplementRMforMCinother industries,willhavetospendasignificantamountof resourcestoensureprocessconsistencyandproduct traceability:Morespecifically, thisentails thesuccessfulsynchronisationof front-end,AMmachineandback-endfactors.Thescopeofresourcestobespentonthisendeavourdependsheavilyonthetechnologymaturityofeachofthesecomponentsasobservedinthecasestudies.Figure7depictsthe decisions and actions that have to be taken into account regarding the overall process whenimplementingRM forMC. In thepre-installationand installationphase, it is crucial todeterminehowmuchadaptationeach technological component requires. Inmost cases, especiallywhen technologieswerenotmature enough for an application, attainingprocess consistency required collaborationwithexternalparties.CostsofoperatingRMforMCvariesdependingontheAMprocessusedandthelevelofdesignthatacompanyexecutes.Forpolymerprocesses,allcasesshowacostdistributionof50%,25%and 25% respectively for front-end, AMmachine related and back-end factors in all implementationphases.FormetalAM,thisdistributionishigherforback-endfactorswith35-40%andsignificantlylowerfor front-end ones. Companies seeking to implement polymer AM have to be aware that producthandlingcanpresentachallengeduringallimplementationstagessincethepartleavestheAMmachinein a ‘green’ state, a state in which parts require further treatment in order to gain their intendedmaterialproperties(e.g.tolowerporosity).

Category Considerationsandchallenges HearingAid

Dental

Technical(Overall process)

Pre-installation, installationphase • SettingupaproductionplanningsystemandproductidentificationsystemwasconsideredachallengeandcrucialforthesuccessfulimplementationofRMforMC X X

• Ensuringprocessconsistencywasamajorchallenge X X

Allimplementation phases • ProducthandlingafterAMprocesswasachallengeaspartsarestillinagreenstate X

• Costdistribution:Front-end,machinerelated,back-end:50%,25%,25% X

14

Figure7:Technologicaloverallprocessconsiderations.DecisiontreeforRMforMCimplementation

4.2.2 Front-end

For front-end considerations, both companies in the hearing aid sector and the dental sector had tocollaborate with software providers or machine producers to develop software during the pre-installationphaseandinstallationphase(Table4).3Shapesoftwarereducedfilepreparationstepsfromthepre-installationand installationphase to thepost-commissioningphase. Scanningofan imprintorintra-oralscanprovidedapossibilityforgeometricalerrorintheRMproducedpart(aswellasincreasedthetolerancestack-up)inallimplementationphases.

DoesRMforMChavetobeadapted?

No

TypeofAMprocessanddegreeofdesign?

Yes

Costs:Back-end:35-40%

Polymer,highdegreeofdesign Metal,lowdegreeofdesign

DetermineifRMforMCneedstobeadaptedfortherespectiveapplicationandwhetherallcomponentsarecommerciallyavailable

Technological– Overallprocess

Choosecomponentsthatareinlinewithproduction

strategy

Significantresourcesandtimewillbeneededto

attainprocessconsistencyandproducttraceability

Costs:Frontend:50%,AM-machin:25%Back-end:25%

Producthandlingingreenstatecanpresentachallenge(nameAM

processes)

Allimplem

entatio

nph

ases

Pre-installatio

nandinstallatio

nph

ase

15

Table4:Summaryoftechnologicalfront-endconsiderationsandchallenges

Itisimportanttoalsonotethatthemorefront-endactivitiesacompanydecidedtoincorporateintheirbusinessmodel,themoredifficultRMimplementationforMCbecame.Thisobservationcanbeclearlymade when contrasting hearing aid companies which executed the entire file design and the dentalcompanies,whichdidnotexecuteanyextensivefiledesignoperations.Asthereiscurrentlynosoftwarethatallowstofullyautomatefiledesign,scalabilityofRMforMCishighlydependentonhumanlabourforcompanieschoosingtoofferfiledesignservices.Fromtheseobservations,itbecomesapparentthatduringthepre-installationand installationphasecompaniesseekingto implementRMforMCneedtofirstascertainwhichfront-endprocessesaretobeexecutedin-houseandwhetherthenecessaryfront-endcomponentssuchasscannersandsoftwareareavailable.Figure8 illustratesadecisiontreebasedontheseidentifiedchallengesandinsightsderivedfromthe10casestudies.Shouldthecomponentsnotbe available, the company seeking implementation can either develop these in-house if it makeseconomicsenseorenteracollaborationwithexternalpartiestodevelopthese. If thecomponentsareavailable,however,thenextstepwillbetoassesstowhatdegreeautomationoffront-endprocessescanberealised.Incasethatautomationcanberealised,ananalysisoftheprocessstepsandthenumberofsteps that canbeautomatedneeds tobeascertained.Witha lowdegreeof automationwithexistingfront-end technologies, collaborations and development with software companies can potentiallyincrease automation. Should the analysis yield a high level of attainable automation, only theappropriatefront-endtechnologyremainstobechosen. If,however,front-endprocessesaregenerallynotautomatable,outsourcing to lowerwagecountries couldenhanceeconomicviabilityof anRM forMC concept. Weighing the cost of these outsourced front-end services is crucial. During allimplementationphases,itisimportanttoaccountforandcontrolthetolerancestack-upthatmayoccurifscanning is involved. Inmanymedicalapplicationsscanningofabodypartorscanningofan imprintandsubsequentdatamanipulationmayleadtoanaccumulationofgeometricalerrors.

Category Considerationsandchallenges HearingAid

Dental

Technical (Front-end)

Pre-installation, installationphase • Co-developmentofsoftwareandcollaborationwithsoftwareprovidersandmachinemanufacturerswasnecessary X X

• Softwarerequiredhighlabourinput X X

Post-commissioningphase • Softwarerequiredlesslabourinputthaninthepre-installationandinstallationphaseduetoautomatedfiledesignfeaturesofferedby3Shapebutfullautomationwasnotpossible

X X

Allimplementation phases • Scanaccuracycouldvaryleadingtoinaccuricesintheproducedpart(esp.withtolerancestack-up) X X

16

Figure8:Technologicalfront-endconsiderations.DecisiontreeforRMforMCimplementation

Areallcomponentsavailable?

Arefront-endprocessesautomatable?

Yes Determineifinternalresourcesaresufficienttodevelopscannersand

software

Yes

No

1.Determinewhichfront-endprocessesaretobeexecutedbyyourcompany2.Determineifallfront-endcomponents(scanners,software)areavailable

Technological– Front-end

Determineiftheseareautomatable

Determinetowhatdegreetheyareautomatable:1. Breakdownthefile

design/generationintodifferentsteps;

2. Determinehowmanystepscanbe

automatedbycurrentsoftwareandscanners

Allimplem

entatio

nph

ases

Pre-installatio

nandinstallatio

nph

ase

Areinternalresourcessufficient?

Highdegreeofautomation?

No Yes

Collaboratewithsoftwareproviderstodevelopsoftware

Chooseappropriatesoftware

No Yes

Iscollaborationpossible?

No Yes

Developcollaborativeopportunities

DiscardRMforMC

Executeacost-benefitanalysis

No

Isfilepreparationoutsourcabletolowerwagecountries?

No

YesCalculatefront-endcostsforscaledupproduction

DiscardRMforMC

Yes

Accountandcontrolfortolerancestack-upifscanningisinvolved

Yes

17

4.2.3 Additive manufacturing equipment

AMmachinerelatedfactorscomprisethefactthatallAMmachines,bothmetalandpolymer,havetobeadaptedtospecificapplicationsinregardtoprocessparameter.Thiscanalsocorrelatewithadaptationsanddevelopmentofrawmaterial,whichinourfocusindustriesneedtobebiocompatible.MetalRMorevenpolymerpowderbasedprocessescausemorechallengesthanthediscussednon-powderpolymerprocesses. Issuessuchasmaintenance,materialaccountabilityandsievingandrecyclingpresentmajorchallenges to implementers of RM for MC (Table 5). For polymer RM for MC, machine ‘openness’regarding modifiability of machine parameters and free choice of material supply and maintenanceservicepresentedchallenges.

Table5:SummaryoftechnologicalAMmachinerelatedconsiderationsandchallenges

Thus, in the pre-installation and installation phase it is crucial to first narrow down the potential AMprocesstypebasedonmaterialthat istobeused,therequiredproductqualityandaccuracy,speedofproduction and flexibility, cost and maintenance aspects (Figure 9). If appropriate machines areavailable,fullworkingsolutionswouldbepreferableasitsignificantlyshortensimplementationtime(aswasobservedinCompanyBbyDeradjatandMinshall(2017)).Ifsuchsolutionsarenotavailable,thenumber of adaptations have to be determined, specifically regarding raw material and the chosenprocess.A costbenefit analysis shouldbeexecuted.Another subsequent consideration thathas tobetaken into account is the above identified degree ofmachine ‘openness’. An open system allows forpotentialcostsavingsbycheaperrawmaterialormaintenanceservicesfromthirdparties.However,itisimportant to note that cheapermaterialmay result in inconsistentmaterial quality. Derivingmaterialfrommachinesuppliersdirectlyoftenensuresconsistentlevelsofquality,asobservedinbothmetalandpolymerRMcases.Insomecases,itcanbepossibletonegotiateanopensystemplatformwiththeAMmachineprovider if theordervolumehasa certain size.Duringall implementation stages, it is crucialthata company implementingmetalRMforMChas toaccount for the fact that somemachineshavenon-ideal material accountability. Powder often has to be manually removed after each run, whichrequireslabourandequipment.

Category Considerationsandchallenges HearingAid

Dental

Technical (Machinerelated)

Pre-installation, installationphase • AMmachinestendtobeadjustedtospecificapplication X X

• Rawmaterialhadtobedevelopedandprocesshadtobeadjusted(materialhadtobebiocomptible) X X

Post-commissioningphase • -

Allimplementation phases • Maintenancewaslabourintensive X

• Materialaccountabilitywasnotideal X

• Sievingandrecyclingpresentedachallenge X

• Machineplatformopenenessregardingmachineparametermodifiabilityandrawmaterialandmaintenanceservicechoicecanberestrictive X

18

Figure9:TechnologicalAM-machineconsiderations.DecisiontreeforRMforMCimplementation

IsAMmachineavailable?

ArethereAMmachineswithafullworkingsolutionforthe

application?

Yes

Yes

No

NarrowdownAMprocessofchoicebasedon:- Rawmaterial

- Requiredqualityandaccuracy- Speedofproductionandflexibilityneeded

- Costs- Maintenance

Technological– Additivemanufacturingmachine

Determinehowmuchadaptationisrequiredandruncost-benefitanalysis

Allimplem

entatio

nph

ases

Pre-installatio

nandinstallatio

nph

ase

DoestheAMmachinehaveanopenplatformregardingparametermodifiability,materialsupplyand

maintenance?

No Yes

Considerwhether

ordervolumeisbigenoughtonegotiateanopensystem

Consideralternative

materialsupplyand

maintenanceoptions*

DiscardRMforMC

No

Consideracquiringsolution

Metal:Accountforpotentialproblemswithmaterial

accountabilityandensuingsievingprocesswhichrequires

labour

*derivingmaterialfromthirdpartymaybecheaperbutatrade-offwithrawmaterialqualitycanberelevant

19

4.2.4 Back-end

Settingupanappropriatepost-processingsystemduringthepre-installationandinstallationphasewasoneof themost challenging factors thatwerepresent inboth industries regardingback-endvariables(Table6).Companieshavebeen trying toautomateasmanypost-processingstepsaspossiblebut fullautomationwasneverattainedduetothehighlevelofuniquenessofeachproductwhichnecessitatedmanual interaction.AcomparisonofresultsbetweenmetalRMcasesfromDeradjat&Minshall (2017)andthepolymerRMcasesinthispaperhaveshownthatmetalAMprocessesdisplaymorechallengesinregardtopost-processingthanpolymerbasedprocesses,e.g.removalofsupportstructures,annealingprocesses.Post-processingchallengesforpolymerprocessespresentedlessofachallengethanformetalRM. The hearing aid cases comprised different AM processes and thus displayed differentconsiderations.FortheinvestigatedpolymerAMprocesses,SLAandDLP,Companies1-4hadtoaccountforproducthandlingdirectlyaftertheAMprocessbecausethiscouldpresentariskfactorduetothefactthattheproductwasinagreenstateandwassubjecttodeformation.WhileremovalofproductsaftertheAMprocessismorecomplicatedforSLAprocesses,itrequiredlesseffortsforDLPprocesseswhereawater jetwasapplied tosimply removesupportmaterial.Thesechallenges forpolymerRM,however,weremerelyminor andwereassociatedwith less resourceallocation thanback-end factors formetalRMsuchasmechanicalpost-processing,supportandpowderremoval.

Table6:Summaryoftechnologicalback-endconsiderationsandchallenges

Inthepre-installationandinstallationphase,companiesseekingtoimplementmetalRMforMChavetothus establish a product and support removal procedurewhich requires labour (Figure 10). Secondly,theyneedtosetupaheattreatmentstepforannealingforsomeapplications.Itisimportanttoacquiretheappropriateovenandtoascertaintheappropriateannealingparameterswhichvarydependingonthe batch and product sizes. Lastly, mechanical post-processing requires trained personnel andtools/machinery.Asdiscussedabove,duringallimplementationstages,itisnecessarytoreducetheriskof employee’s damaging parts during the removal and mechanical post-processing step by ensuringappropriate training. ForpolymerAMprocesses, thepost-processing stepsvarydependingon theAMprocess and the products produced. Hence, therewill be no further elaboration on the specific post-processing steps at this stage. In all cases, however, human interaction is always required to at leastmoveobjectsfromoneprocessingstationtoanother.

Category Considerationsandchallenges HearingAid

Dental

Technical (Backend)

Pre-installation, installationphase • Settingupapost-processingsystemwasoneofthemostchallengingactionsintheimplementationprocess X X

Post-commissioningphase • -

Allimplementation phases • Post-processingcouldnotbefullyautomatedduetouniquenessofproducts(whichrequiremanualinteraction) X X

• Removalofsupportstructurescouldberisky• Annealingcouldpresentachallenge X

20

Figure10:Technologicalback-endconsiderations.DecisiontreeforRMforMCimplementation

Polymerormetal?

Metal Polymer

Technological– Back-endAllimplem

entatio

nph

ases

Pre-installatio

nandinstallatio

nph

ase

*DLP,forinstance,

Establishproductandsupportremoval

Establishheattreatment(ifnecessary)

Establishmechanical

post-processing

Trainorrecruitlabour1. Ascertain

annealingparameters

(variesdependingon

lotsize)2. Buyfurnace

1. Trainorrecruitlabour

2. Invest intools/

machinery

Ensureemployeesdon‘tdamagepartduringremovalorpost-processing

Post-processingvariesheavily

dependingonprocess,settingupa

semi-automatedprocessrequiressignificantresources

Humanresourcesarerequiredatminimumforthetransfer

betweenpost-processingstations

21

4.3 Operational Operationally setting-up a production planning system was a challenge for all companies in bothindustriesduring thepre-installationand installationphase (Table7). Short-delivery timesofproductsduring all implementation phases and fluctuating daily demand put strain on production planningthroughout all implementation phases. For larger companies, synchronisation of global productionpresentedanadditionalchallenge.Companiesinthehearingaidindustry,forinstance,utilisedcheaperlabour in China and Ukraine to execute the file design and preparation, which accounted for 50% ofproductioncosts.ProductiononAMmachineswassubsequentlyexecutedintheUSorEU.Additionally,orders could be processed almost immediately, irrespective of the time of the day, because facilitieswerelocatedindifferenttimezones.

Table7:Summaryofoperationalconsiderationsandchallenges

Based on the identified challenges and the contextual insights gained from the cases, the followingimplementation tree canbederived foroperational considerations (Figure11). Firstly, during thepre-installation and installation phase, it is crucial to determine the type of demand that the enterpriseseekingtoimplementRMforMCisfacing:Shouldtherebestableandpredictabledemand,maximisingmachine utilisation and delaying production in favour of maximising a production batch is advised.Additionally, if the demand is large enough, utilising an AMmachinewith a large build platform canincrease production efficiency. During all implementation phases, AM production should be executedoutside of employeeworking hours in order to utilise labourmore efficiently. If the demand, on theotherhand, isfluctuatingcoupledwithshortdeliverytimes,AMmachineswithsmallerbuildplatformstendtobemoresuitable.Theirbuildplatformfillsupmorequicklyandincomingorderscanbestartedmuchquickerandmorefrequently.Thus,duringthepre-installationandinstallationphase,itshouldbedeterminedwhichmachinesaretobeimplemented.Duringallimplementationphases,itcanbeprudenttoallocateacertainslotforurgentordersineachproductionrunortodedicatea“fast-trackproductionline”inwhichsuchorderscanbeaccommodated.Iftheenterpriseisproducingatdifferentsites,designandproduction resources inall sites shouldbe synchronised.Takingadvantageof cheaper labouranddifferenttimezonesinwhichorderscanbeprocessedcanbebeneficial.

Category Considerationsandchallenges HearingAid

Dental

Operational

Pre-installation, installationphase • SettingupaproductionplanningsystemwasconsideredachallengeandcrucialforthesuccessfulimplementationofRMforMC X X

Post-commissioningphase • -

Allimplementation phases • Shortdeliverytimesputpressureonefficientproductionplanning X X

• Synchronisingproductionplanningofproductionindifferentplantsaroundtheworldhasbeenachallenge(coupledwithseasonalanddailyfluctuationsindemand)

X (X)

22

Figure11:Operationalconsiderations.DecisiontreeforRMforMCimplementation

Thesecondoperationalconsiderationthathastobetakenintoaccountistheoptionofwhetherproductdesigncanbe leveraged. In thecaseofahighly customiseddesign,onehas toaccount forpotentially

Stablepredictabledemandorfluctuatingdemandwithshort

deliverytimes?

Stablepredictabledemand

OperationalAllimplem

entatio

nph

ases

Pre-installatio

nandinstallatio

nph

ase

Maximisemachineutilisationanddelayproductioninfavourofmaximisingbatchwherepossible

UtiliseAMmachineswithsmallerbuildplatform

Runovernightproductionifpossible

Usefasttracklinesforurgentorders

Utiliseglobalcapabilities(iftheyexist)toprocess

orders;takeadvantageoftimezonesandcheaperlabour

Automateoperationalactivitiesinthedesignphaseasmuchaspossible

1.Determinetypeofdemand 2.Determineifproductdesigncanbeleveraged

Shoulddemandbesufficientlylarge,utiliseanAM

machinewithalargerbuildplatform

Fluctuatingdemandwithshortdeliverytimes

Accountforback-endchallengesinprocessingnewdesigns(thinwall

thickness,complicated

structures,etc.)

Accountforpotentiallyhighlabourinputfordesign

operations

Functionaldesignorhighlycustomised?

Highlycustomised

Functionaldesign

Establishfunctionaldesignthatminimisesoperationallabourinputonfront-andback-

end

23

high labour input for design operations. Thus, operational activities in the design phase should beautomatedwherepossible.Inthecaseoffunctionalandcomplexdesign,itisimportanttoensureduringthepre-installationandinstallationphasethatthedesignminimiseslabourinputonfront-andback-end.Duringallimplementationphasesthenewdesigncanpotentiallycausechallengessuchasthefactthatthinwallthicknesscanincreasetheriskofdamagingthepartinthepost-processing.

4.4 Organisational Onanorganisationallevel,existingemployeeshadtoberetrainedornewonesacquiredduringthepre-installationandinstallationphase(Table8).Forsmallerenterprises inthedental industries,availabilityof qualified personnel could present a limitation for production ramp-up in the post-commissioningphase. This was less relevant for the hearing aid companies as all of the case companies were largemulti-national enterprises. For hearing aid companies, the major challenge was the acquisition ofemployeesinvolvedwiththefiledesignandpreparation.Productionstrategiesconsistedofautomationcoupledwitheithermaximisationofmachineutilisationorprioritisationforproductionflexibility/speedofproductionforallcasecompaniesinboththehearingaidanddentalindustry.

Table8:Summaryoforganisationalconsiderationsandchallenges

Thus,therearethreedifferentfactorsthathavetobeaccommodatedwhenconsideringorganisationalaspects involved with the implementation of RM for MC (Figure 12). During the pre-installation andinstallationphase,implementersneedtoevaluatetheirexistinghumanresourcesandascertainwhetherthesearesufficientorre-trainable.Shouldtheybere-trainablethecostandtimerequiredforretrainingshould be assessed. Additionally, it is important to note that in later stages of the implementationprocess, i.e. thepost-commissioningphase, it canbea challenge toexpand theproductportfolio intoareas in which the retrained personnel has no knowledge. This was observed for certain dentalcompanieswhichretraineddentaltechniciansandhaddifficultiesexpandingtheirproductlinesintonon-dentalapplications. Ifexisting staff isnot re-trainable,appropriatepersonnelneeds tobe recruitedorproductionstepswherein-housecapabilitiesarelackingneedtobeoutsourced.

Category Considerationsandchallenges HearingAid

Dental

Organisational

Pre-installation, installationphase • Humanresources:RetrainingandacquiringemployeeswithmoretechnicalandCADknowledgewasachallenge X X

Post-commissioningphase • Availabilityofhumanresourcescanpresentalimitingfactorforproductionscale-up X

Allimplementation phases • SizeofcompanyinfluencedtheimplementationofRMforMC X X

• Productionstrategiesconsistedofautomationcoupledwitheithermaximisationofmachineutilisationorprioritisationforproductionflexbility/speedofproduction X X

24

Figure12:Organisationalconsiderations.DecisiontreeforRMforMCimplementation

Isexistingstaffsufficientandretrainable?

Yes

OrganisationalAllimplem

entatio

nph

ases

Pre-installatio

nand

installatio

nph

ase

Considertimerequiredfortraining

Hirenewpersonnelor

outsourcecertainproductionstepswherecapabilities

arelacking

Considerpotentiallimitationtoexpandproductionportfoliointootherindustries

Procurehumanresourcesintimeforproductionscale-up(orlesslikely:changebusinessmodeltoexcludeoperationsthatrequirequalified

personnel)

Highermanagementhastodriveimplementat-ionofRMfor

MC

1.Evaluatehumanresources 2.Assesssizeofcompany

No

Utilisecapabilities

fromsubsidiaries

Largecompany?

Yes No

Initiatecollaborationwithothercompanies

ConsiderAMmachine

leasingoptions

Takeadvantageofnationalresearchschemes

3.Determineproductionstrategy:Striveforautomation

andreducelabourwherepossible

Prioritisespeedandflexibilityorcost-efficiencyandhighmachineutilisation?

AMmachinewithsmaller

buildplatform

AMmachinewithbigger

buildplatform

Speedandflexibility Costefficiency

Post-com

missioning

phase

25

Generally,humanresourcescanbeachallengeforproductionramp-upinthepost-commissioningphaseandneedtobeaccountedfor.Limitingthescopeofcomplexandtimeconsumingoperationssuchasfilegenerationinthebusinessmodelcanbeanalternativesolutiontoaddressthispotentialchallenge.

Similarly, it is importanttobeawareofthesizeandcapabilitiesofacompanywhenimplementingRMforMC.Largecompaniescanutilise internal resourcesandknowledgetoexecutethe implementation.Duetothebureaucraticnatureoflargerenterprises,itisthusessentialtohavehighlevelmanagementdrivetheimplementationofRMforMC.Asseeninthecasecompanies,especiallyforcompaniesbeingpioneers in applying RM in their industry, a strong drive from higher management is essential. ForCompany 1, senior management had to proactively support and initiate AM implementation despiteinitial failures. For smaller companies, collaborationwith AMmachine producers, software providers,material supplier and/or competitors has to be initiated in the pre-installation and installation phase.FinancingoptionssuchasleasingandcollaborativeR&Dsupportedbynationalresearchinitiativescanbehelpfulintheimplementationprocess.

Lastly, it is essential to determine the production strategy that needs to be adopted. All case studiesshowthatstrivingforautomationandreductionoflabourwasdeemedmostappropriateforRMforMC.Aprioritisationhas tobedetermined,however, forwhetheracompanyprefersproductionspeedandflexibilityorcostefficiency.Intheformercase,theenterpriseshouldchooseanAMsystemwithsmallerbuildplatformsothatorderscanbeexecutedinshorterintervalsasasmallernumberofproductunitsarerequiredtowarrantamachinerun.Inthelatercase,executingabiggernumberofordersinalargermachineisfinanciallymorebeneficialthanexecutingseveralsmallerrunsinsmallermachines.However,individualordersmaytakemoretimetobeprocessedasthewaitingtime isprolongeduntil thereareenoughorderstowarrantaproductionrun.Thedecisionastowhichapproachtoadoptdependsonthetypeofdemandasillustratedaboveintheoperationalsection(Figure11).

4.5 External External considerations comprised the need to collaborate with software providers, AM machineproducers and raw material suppliers during the pre-installation and installation phase in order toimplement RM forMC (Table 9). Companies in the dental industry, however, relied on collaborationthroughout all phases firstly because themajority of companieswere smaller than in the hearing aidindustryandsecondlybecausethereweremoreproducttypesandmaterials toresearch inthedentalarea.Duringtheearlystagesof implementingAMproducts inboth industries,customeracceptanceofAM products needed to be gained. In both industries, the first companies to implement RM haveintroduced products of insufficient quality. Thus, they and subsequent implementers of RM had toconvincecustomersofadditivelyproducedparts.

26

Table9:Summaryofexternalconsiderationsandchallenges

Basedontheidentifiedchallengesandinsightfromthecasestudiesthefollowingdecisionshavetobemade in regard to external considerationswhen implementing RM forMC (Figure 13). As a first stepduringthepre-installationandinstallationphase,itiscrucialtoassesswhetherregulatoryrequirementscan pose a challenge for RM production. High quality standards and regulatory approval of certainproducts can prevent implementation. Should the regulation be appropriate for implementation,customeracceptanceofAMproductsshouldbeascertainedbeforeproductionramp-up.Inmanycases,asseeninboththehearingaidandthedentalcases,resourcesandtimehavetobespendtofamiliarisecustomers with the quality of AM products if AM has not been applied to the industry before.Conversely, failureof providing convincingproduct quality can result in additional resources and timerequiredtoimplementRMforMCasseeninthecasesofCompanyC(DeradjatandMinshall,2017)andCompany 1. The next step requires establishing collaboration opportunities with different companiessuchasscanning,software,AMmachine,materialandmaintenancesuppliersorcompetitors.Duetothehigh level of adaptations that is needed to create a successful RM for MC system collaboration isessential.Unlessthecompanyalreadyownscrucialpartsrequiredforimplementationandunlesstheriskof exposing sensitive data is perceived as too high, collaboration is unavoidable. In all the 10 casestudies, the benefits of collaboration far outweighed the disadvantages. Collaboration often occurs inthepre-installationand installationphasebutcanalsoberelevant inthepost-commissioningphasetoexplorenewapplicationsandoptimisationofproduction.

Category Considerationsandchallenges HearingAid

Dental

External

Pre-installation, installationphase • CollaborationwithAMmachineproducers,softwarecompaniesandmaterialsupplierstodevelopandadjusttechnicalaspectswascrucial;theimplementationrequiredthedevelopmentorextensiveadaptationsofcommerciallyavailablecomponentstotheRMsystem

X X

• Customeracceptanceofadditivelyproducedpartswasachallengeinthebeginning X X

Post-commissioningphase • -

Allimplementation phases • Collaborationthroughoutallphaseswasrelevantformostdentalcompanies X

27

Figure13:Externalconsiderations.DecisiontreeforRMforMCimplementation

DocustomersacceptAMproducts?

Yes

ExternalAllimplem

entatio

nph

ases

Pre-installatio

nandinstallatio

nph

ase

EvaluatecustomeracceptanceofAM

No

Evaluateoptionsforcollaborationswithscanning,software,AMmachine,materialand

maintenancesuppliersorcompetitors

Howsensitiveistheinformation?Howhigharethebenefits?

Setupcollaboration

AssesswhetherinternalresourcesaresufficienttoimplementRMforMC

Notsensitiveinformation,highbenefits

Highsensitivityand/orlowbenefitsofcollaboration

Assessregulatoryconditionsforproductqualityandpotentialfuturechangesinregulation

Regulationappropriate?

Yes No

DonotimplementRMforMC

SpendresourcesandtimeonconvincingcustomersofAMproductquality

Areinternalresourcessufficient?

Yes No

DonotimplementRMforMCAssessstrategicandeconomicbenefitsofimplementingRMforMC

28

5 Conclusion and future work Thepaperaimstoimprovetheunderstandingofwhichconsiderationscompanieshavetoaccountforintheir decision-making process when seeking to implement RM for MC for the most common AMtechnologies.Case results fromthedentalandhearingaid industryhavebeencontrastedandused todevelopeightimplementationdecisiontrees.

The insights from the case studies show that the framework and list of considerations byDeradjat&Minshall (2017) are valid for hearing aid applications. Additionally, the influence of implementationstages according to Voss (1988) has been shown for polymer applications. This is illustrated by theidentification of varying considerations at different points during the implementation process. Theresearchfound22commonconsiderationsthatwereforbothmetalandpolymerapplications.Fourtosix aspects for both dental and hearing aids were technology and application specific. These insightsdirectlyaddresstheresearchsub-questionofidentifyingcommonconsiderationsformetalandpolymerRM.Basedontheseinsightsandthecontextofthecasestudies,eightdetailedimplementationdecisiontrees for RM forMCwere derived. These decision trees represent an expansion of work initiated byDeradjat & Minshall (2017), Sandström (2015) and Mellor et al. (2014) and directly contribute toaddressing the main research question. The findings address a gap in literature on MC enablers inmanufacturing as identified by Fogliatto et al. (2012) by providing successful case examples and byincreasing the understanding of how to approach the implementation process. The cases from thehearing aid industry illustrate that similar to the dental industry RM enables ‘pure customisation’ inmanufacturing,thehighestdegreeofcustomisationwhereeveryproductisuniqueandspecifiedtothecustomers’requirements(LampelandMintzberg,1996).

ThelistofconsiderationsforpolymerRManddecisiontreesprovideusefulinformationformanagementin industry on ascertaining whether RM for MC is appropriate for their business and on how tospecificallyimplementRMforMC.Additionally,itincreasesunderstandingofthescopeofapplicabilityofRM andMC for governmental initiativeswhich have suggested this approach to strengthen domesticproduction(DFG,2017;InnovateUK,2016;TheWhiteHouse,2016).

There are limitations to our research that should be noted. The number of case studies limits thegeneralisability of the research. In addition, general shortcomings inherent with case study basedresearchapproachesasdescribedbyYin(2009)areprevalent.TheresultsanalysedtheimplementationofRMforMConlyfordentalandhearingaidapplications.Inordertoaddresstheseshortcomings,futureworkshouldincludeotherindustryapplicationsandalargersamplesizeonceindustryhasadoptedRMforMCmorewidely.Additionally,amoreevenlyspreaddistributionofSMEsandlargesizedenterprisesinthedatasetcanaccountforpotentialimplementationdifferencespertainingtofirmsize.

29

6 References ASTM,2012.StandardTerminologyforAdditiveManufacturingTechnologies.

Atzeni, E., Iuliano, L., Minetola, P., Salmi, A., 2010. Redesign and cost estimation of rapidmanufacturedplasticparts.RapidPrototyp.J.16,308–317.

Bateman,R.J.,Cheng,K.,2006.Extendingtheproductportfoliowith“devolvedmanufacturing”:methodologyandcasestudies.Int.J.Prod.Res.44,3325–3343.

Davis,S.,1987.FuturePerfect,Reading.Addison-Wesley.

Deradjat, D., Minshall, T., 2017. Implementation of Rapid Manufacturing for MassCustomisation.J.Manuf.Technol.Manag.28.

Deradjat, D., Minshall, T., 2015. Implementation of additive manufacturing for masscustomisation, in: Pretorius, L. (Ed.), IAMOT 2015-24th International Association forManagement of Technology Conference: Technology, Innovation and Management forSustainableGrowth.CapeTown,pp.2079–2094.

DFG,2017.IntegrativeProductionTechnologyforHigh-WageCountries[WWWDocument].URLhttp://www.dfg.de/en/research_funding/programmes/list/projectdetails/index.jsp?id=25065172(accessed2.16.17).

Duray, R., 2011. Process Typology of Mass Customizers, in: Fogliatto, F.S., da Silveira, G.J.C.(Eds.), Mass Customization. Engineering and Managing Global Operations. Springer,London,pp.29–44.

Duray, R., Ward, P.P.T., Milligan, G.W.G., Berry, W.W.L., 2000. Approaches to masscustomization:configurationsandempiricalvalidation.J.Oper.Manag.18,605–625.

Fogliatto, F.S., da Silveira, G.J.C., Borenstein, D., 2012. The mass customization decade: Anupdatedreviewoftheliterature.Int.J.Prod.Econ.138,14–25.

Gibson, I., Rosen, D., Stucker, B., 2010. Additive manufacturing technologies. Springer, NewYork.

Hayes,R.H.,Wheelwright,S.C.,1979.LinkManufacturingProcessandProductLifeCycles.Harv.Bus.Rev.57,133–140.

Herriott,R.,Firestone,A.,1983.Multisitequalitativepolicyresearch:Optimisingdescriptionandgeneralisability.Educ.Res.12,14–19.

Holmström,J.,Partanen,J.,Tuomi,J.,Walter,M.,2010.Rapidmanufacturinginthesparepartssupply chain:Alternativeapproaches to capacitydeployment. J.Manuf. Technol.Manag.21,687–697.

Hopkinson,N.,Hague,R.,Dickens,P.,2006.IntroductiontoRapidManufacturing,in:Hopkinson,N., Hague, R., Dickens, P. (Eds.), RapidManufacturing - An Industrial Revolution for theDigitalAge.JohnWiley&Sons,Ltd,Chichester,UK,pp.1–4.

30

InnovateUK,2016.MappingUKResearchandInnovationinAdditiveManufacturing.AreviewoftheUK’spubliclyfundedR&Dactivitiesinadditivemanufacturingbetween2012and2015.London.

Lampel,J.,Mintzberg,H.,1996.CustomizingCustomization.SloanManage.Rev.38,21–30.

Lieberman,M.B.,Montgomery,D.B.,1988.First-MoverAdvantages.Strateg.Manag.J.9,41–58.

Mellor, S., 2014. An Implementation Framework for Additive Manufacturing. University ofExeter.

Mellor,S.,Hao,L.,Zhang,D.,2014.Additivemanufacturing :Aframeworkforimplementation.Int.J.Prod.Econ.149,194–201.

Pham,D.T.,Dimov,S.S.,2001.RapidManufacturing:TheTechnologiesandapplicationsofRapidPrototypingandRapidTooling.Springer,London.

Piller, F.T., 2008. Observations on the present and future ofmass customization. Int. J. Flex.Manuf.Syst.19,630–636.

Pine, B., 1993. Mass customization: The new Frontiers in Business Competition, HarvardBusinessSchoolPress.HarvardBusinessSchoolPress,Boston.

Reeves, P., Tuck, C., Hague, R., 2011. Additive Manufacturing for Mass Customization, in:Fogliatto, F.S., da Silveira, G.J.C. (Eds.), Mass Customization. Engineering and ManagingGlobalOperations.Springer,London,pp.275–289.

Ruffo, M., Tuck, C., Hague, R., 2007. Make or buy analysis for rapid manufacturing. RapidPrototyp.J.13,23–29.

Sandström, C.G., 2015. The non-disruptive emergence of an ecosystem for 3D Printing —Insights from the hearing aid industry’s transition 1989–2008. Technol. Forecast. Soc.Change102,160–168.

Silveira,G.Da,Borenstein,D.,Fogliatto,H.S.,2001.Masscustomization :Literaturereviewandresearchdirections.Int.J.Prod.Econ.72,1–13.

TheWhiteHouse,2016.RevitalizingAmericanManufacturing.Washington,D.C.

Tuck, C., Hague, R., Ruffo,M., Ransley,M., Adams, P., 2008. Rapidmanufacturing facilitatedcustomization.Int.J.Comput.Integr.Manuf.21,245–258.

Voss,C.,1988.Implementation:Akeyissueinmanufacturingtechnology:Theneedforafieldofstudy.Res.Policy17,55–63.

Weller, C., Kleer, R., Piller, F.T., 2015. Economic implicationsof 3Dprinting:Market structuremodelsinlightofadditivemanufacturingrevisited.Int.J.Prod.Econ.164,43–56.

Westbrook,R.,Williamson,P.,1993.Masscustomization: Japan’snewfrontier.Eur.Manag. J.11,38–45.

Wohlers,T.,Caffrey,T.,2016.WohlerReport2016.FortCollins.

31

Wong,K.V.,Hernandez,A.,2012.AReviewofAdditiveManufacturing.ISRNMech.Eng.2012,1–10.

Yin,R.,2009.CaseStudyResearch.DesignandMethods,4thed.SAGEPublications,London.