proceedings of the 2nd annual smacc reserch seminar 2017 · digital channels of data in industrial...
Post on 29-Sep-2020
0 Views
Preview:
TRANSCRIPT
Tampere University of Technology
Proceedings of the 2nd Annual SMACC Research Seminar 2017
CitationAaltonen, J., Koskinen, K., Virkkunen, R., & Kuivanen, R. (2017). Proceedings of the 2nd Annual SMACCResearch Seminar 2017. Tampere: Tampere University of Technology.
Year2017
Link to publicationTUTCRIS Portal (http://www.tut.fi/tutcris)
Take down policyIf you believe that this document breaches copyright, please contact cris.tau@tuni.fi, and we will remove accessto the work immediately and investigate your claim.
Download date:03.02.2021
Proceedingsof
the2ndAnnualSMACCResearchSeminar2017
JussiAaltonen,RiikkaVirkkunen,KariT.Koskinen&RistoKuivanen(eds.)
Tampereenteknillinenyliopisto-TampereUniversityofTechnology
JussiAaltonen,RiikkaVirkkunen,KariT.Koskinen&RistoKuivanen(eds.)
Proceedings of the 2nd Annual SMACC Research Seminar 2017
TampereUniversityofTechnology
Tampere2017
ISBN978-952-15-4040-0
TABLEOFCONTENTS
FOREWORD.................................................................................................................................5
ILIFECYCLEMANAGEMENT......................................................................................................11. DIGITALASSETMANAGEMENTSERVICEOFFERINGASANINTEGRATEDPARTOFCUSTOMERS’BUSINESS...............................................................................................................2ToniAhonen,JyriHanski,PasiValkokari2. MACHINEPERFORMANCEANDAVAILABILITYPREDICTIONTHROUGHTHELIFETIME.........5PäiviKivikytö-Reponen,RistoTiusanen,JariHalme,KalleRaunio,OlliSaarela3. ENHANCEDMODELLINGCONCEPTANDCOMPUTINGPLATFORMFORSYSTEMLIFECYCLEANALYSES....................................................................................................................................8Jussi-PekkaPenttinen,ArtoNiemi,KariT.Koskinen,EricCoatanéa4. LIFETIMEASSESSMENTFORMECHANICALCOMPONENTS................................................12SatuTuurna,PerttiAuerkari,TimoJ.Hakala,AnssiLaukkanen,SanniYli-Olli5. DYNAMICSERVICEDECISIONSUPPORTPLATFORM..........................................................16HenriVainio,AnandaChakraborti,JussiAaltonen,KariT.Koskinen6. IMPACTOFMAINTENANCEONCIRCULARECONOMY.......................................................20PasiValkokari,JyriHanski,ToniAhonen
IIDIGITALANDADDITIVEMANUFACTURING..........................................................................247. DIGIMATURITYINMANUFACTURINGINDUSTRY...............................................................25Simo-PekkaLeino,Juha-PekkaAnttila8. IDENTIFICATIONOFFUNDAMENTALREQUIREMENTSFORIDEALMETAMODELINGFRAMEWORKINADDITIVEMANUFACTURING..........................................................................29HosseinMokhtarian,EricCoatanéa,HenriParis,JormaVihinen,MohammadHosseini9. INDUSTRIALIZATIONOFHYBRIDANDADDITIVEMANUFACTURINGOFMETALS-CHALLENGESINIMPLEMENTATION..........................................................................................32VeliKujanpää,JormaVihinen,TuomasRiipinen,PasiPuukko10. IDENTIFYING3D-PRINTABLESPAREPARTSFORADIGITALIZEDSUPPLYCHAIN.............37JoniReijonen
IIISMARTMACHINESANDROBOTICS.....................................................................................4111. AWARENESSINHUMAN-ROBOTINTERACTIONFORCOLLABORATIVETASKS...............42AlexandreAngleraud,RoelPieters12. ANOPEN-SOURCE,CONFIGURABLEIOTPLATFORMFORREMOTEMONITORINGOFSMARTMACHINES.....................................................................................................................45AnandaChakraborti,OlliUsenius,HenriVainio,JussiAaltonen,KariT.Koskinen13. SAFETYCONSIDERATIONSINTHEDESIGNANDVALIDATIONOFAUTONOMOUSMACHINESYSTEMS...................................................................................................................49EetuHeikkilä,RistoTiusanen,HeliHelaakoski14. SIMULATINGTHEDRAGCOEFFICIENTOFASPHERICALAUTONOMOUSUNDERWATERVEHICLE.....................................................................................................................................53ArttuHeininen,JussiAaltonen,KariT.Koskinen15. AUTONOMOUSANDCOLLABORATIVEOFFSHOREROBOTICS.......................................57JoseVillaEscusol,JussiAaltonen,KariT.Koskinen16. FUNCTIONALHIERARCHYINAUTONOMOUSROBOT....................................................60SoheilZavari,TuomasSalomaa,JoseVillaEscusol,JussiAaltonen,KariTKoskinen
LISTOFAUTHORS......................................................................................................................63...................................................................................................................................................65
FOREWORD
TheAnnual SMACCResearchSeminar is a forum for researchers fromVTTTechnicalResearchCentre of Finland Ltd, Tampere University of Technology (TUT) and industry to present theirresearch in theareaof smartmachinesandmanufacturing.The2nd seminar isheld in7thofNovember2017inTampere,Finland.
Theobjectiveoftheseminaristopublishresultsoftheresearchtowideraudiencesandtoofferresearchersa forumtodiscuss their researchandto findcommonresearch interestsandnewresearchideas.
SmartMachines andManufacturingCompetenceCentre– SMACC is joint strategic allianceofVTT LtdandTUT in theareaof intelligentmachinesandmanufacturing. SMACCoffersuniqueservices for SME`s in the field of machinery and manufacturing – key features are rapidsolutions, cutting-edge research expertise and extensive partnership networks. SMACC ispromotingdigitalizationinmechanicalengineeringandmakingscientificresearchwithdomesticandinternationalpartnersinseveraldifferenttopics(www.smacc.fi).
Tampere26thofOctober,2017
Editors
ILIFECYCLEMANAGEMENT
DIGITALASSETMANAGEMENTSERVICEOFFERINGASANINTEGRATEDPARTOFCUSTOMERS’BUSINESS
-2-
1. DIGITAL ASSETMANAGEMENT SERVICEOFFERINGAS AN INTEGRATED PARTOFCUSTOMERS’BUSINESS
ToniAhonen,JyriHanski&PasiValkokariVTTTechnicalResearchCentreofFinlandToni.ahonen@vtt.fi
ABSTRACT
LargeFinnish technologycompanieshadalreadystarteddevelopingknowledge-basedserviceslongbefore anywidespreaddiscussionof the topicsof digitalizationor the industrial Internethadappeared.However,onekeyproblemisthattheservicedevelopmenthas includedratherproduct-centricpracticesandrealintegrationtocustomers’processeshasnotoftentakenplace.Whilecustomer’sdecision-makingprocesseshaveremainedunknownandtherealpotentialoftheserviceofferingsunderexploited,manycompanieshavealsodeveloped ICTsolutions fromtheperspectiveofbilateralpartnerships,neglecting thecustomers’multi-vendorproblematicsand needs for integrated solutions. Current paper discusses the challenges of digital assetmanagementservicedevelopmentfromtheperspectiveofhowtheseservicesareconnectedtothecustomers’assetmanagementprocessesandsystems.Furthermore,wediscuss theneedsfor finding the optimal development practices, thus finding the balance between long-termvisionsandrapidexperimentsanddevelopingthenewdigitalcapabilitiesin-house.
INTRODUCTION
The impact of the Industrial internet and digitalization is expected to be systemic in manyindustries.Therefore,understandingthebusinessenvironmentandthenetworkednatureofthebusinessiscrucial.Althoughtheliteraturedoesincludeanumberofexamplesofdigitalizationinasset management (e.g. Baines & Lightfoot, 2013; Lee et al. 2015), there are relatively fewexamples of thorough transformations involving the efficient integration and exploitation ofdigitalchannelsofdatainindustrialenvironments.
While many of the impacts of digitalization are difficult to predict beforehand due to theirsystemicnature,manycompanieshaveadoptedproductivity, leadtimes, features,qualityandcostasthedriversbehindthecreationanddevelopmentofdigitalconcepts(Sommarberg2016).In order to effectively develop services, one needs to understand how value related to thesetopics can to be created and how customers’ assets are to be managed. There is a risk ofindividual tools being isolated from the customers’ business processes and remainingunderexploited, despite their potential of creating customer value. Thus, creation of noveldigitalsolutionsrequiresthoroughunderstandingatstrategic,tacticalandoperationallevels.
CONSIDERATIONSONCREATINGDATA-BASEDDIGITALSERVICEOFFERINGS
Developmentofdigitalassetmanagementservicescallsfornewcapabilitiesintheorganization.Understanding the customer needs and the requirements related to the networked businessenvironment is in the core of the development. The transition from bilateral partnershipstowardsbusinessecosystemsischallengingthecompaniestothinkdifferently.Furthermore,thecompanynetworksproducelotsofdata,whicharerarelyefficientlyintegratedinthedecision-making processes in the ecosystem. Thus, in addition to the understanding of the networked
DIGITALASSETMANAGEMENTSERVICEOFFERINGASANINTEGRATEDPARTOFCUSTOMERS’BUSINESS
-3-
businessenvironment,we find the capabilities to integrate thevarious sourcesofdataas thecorecompetitiveadvantageofthefutureassetmanagementecosystems.
In addition to theorganizational capabilities, understandingof thephenomena related to thecustomer’s production environment is crucial. According to our findings, service developmentneedstoaddressthefollowinggenerictrendsandtodefinetheapproachtorespondtothem:
• increaseinautomationlevelandthestep-wiseapproachtowardsautonomoussystems• value-basedbusinessmodelsandearninglogicsinabusinessnetwork• from analytics-based and consultative services towards AI-based automated decision-
makingprocesses
FROMDISTINCTSERVICESTOWARDSINTEGRATEDASSETMANAGEMENTSERVICEOFFERING
Companies often collaborate in large business ecosystems in order to maximise the valuecreation.Customersareofferedavarietyofservices,platformsanddigitaltools.Makingallthesolutions work together and integrating all the new solutions with the older ones and thestrategic,tacticalandoperationalpracticesofthecompanyisachallengingtask.Developmentof data-oriented digital tools is often seen as a straight-forward software development effortwhereagilepracticesandquickexperimentshavebeenfoundimportant.However,OEMshavelimitedknowledgeofthecomplexityofthecustomers’process.Therefore,developmentistoooftencarriedoutwithanarrowperspective.Fromcustomers’pointofview,theselectionanddevelopmentof theserviceplatformshouldbedoneverycarefully inordernot to reducethefutureopportunities.Table1presentsthescenariosthatdefinetheOEM’srolewithrespecttoserviceofferingandITsolutions.
Table1.ScenariosforOEMwithrespecttoserviceandITsolutiondevelopment(basedonKortelainenetal.2017).
Servicestrategy ITsolutionstrategy
OEM companies develop their information orknowledge based offering mainly to supportthecustomer´sdailyoperations.
OEM focuses on the provision ofmeasurement solutions, on-board analyticstoolsanddatatransfertechnologytoenabletheprovisionofassetleveldata.
OEMs and service providers develop theexcellence in refining data in a way thatdeliversmorevaluetotheassetowner.
OEM companies build the analyticscapabilities and integrate the resultedinformation-based service in the platformusedbytheendcustomer.
OEM companies take the responsibility ofmanaging and optimising the performance ofcustomers’assetsinthebusinessecosystem.
OEMbuildsthecapabilitiesfortheprovisionof platform formanaging the data and theservicesinthenetwork.
Thereareanumberofoperationalassetmanagementprocesses,suchassparepartsorderanddelivery processes, maintenance processes and customer support, that can be supported bydigital technologies.Both largerandsmallerOEMshavestarteddevelopmentofnewtools for
DIGITALASSETMANAGEMENTSERVICEOFFERINGASANINTEGRATEDPARTOFCUSTOMERS’BUSINESS
-4-
supportingtheseprocesses.Thismaynothavetobedependentonthechoicesmaderelatedtothe above-described service and IT strategies. The fact that customers are different in theirneeds,requiresthatOEMsneedtobeadaptiveinhowtheyintegratetocustomers’processes.Thus,whileOEMdevelopsthedigitaltoolset, itneedstobecarefullyplannedhowthesetoolsareintegratedinthecustomers’processesandITinfrastructure.Figure1showsanexampleofthevarioussourcesofdataincustomers’productionenvironmentandtheneedsforintegrationat1)operational,2)tacticaland3)strategiclevels.
Figure1.ExampleoftheneedforintegratingdifferentsourcesofdataandtheservicesofOEMatstrategic,tacticalandoperationallevels.
CONCLUSIONS
In order to develop successful digital asset management services, OEM company needs tounderstandthecomplexityrelatedtothebusinessecosystem,varioussourcesofdata,decision-makinglevelsandserviceandITstrategies.Therefore,OEMcompanyneedstounderstandhowdigital asset management services are integrated into customers’ decision-making, assetmanagementprocessesandITinfrastructure.
REFERENCES
Baines, T. and Lightfoot, H.W. (2013), “Servitization of themanufacturing firm”, InternationalJournalofOperations&ProductionManagement,Vol.34Iss.1,pp.2-35.Kortelainen, H., Happonen, A. and Hanski, J. (2017). ”From asset provider to knowledgecompany-transformationinthedigitalera”,WCEAM2017Conference,Brisbane,Australia,2.-4.August2017.Lee,J.,Bagheri,B.andKao,H-A.(2015),“ACyber-PhysicalSystemsarchitectureforIndustry4.0-basedmanufacturingsystems”,ManufacturingLetters,Vol.3,pp.18–23.Sommarberg,M. (2016), “Digitalization as a ParadigmChanger inMachine-Building Industry”,(TampereUniversityofTechnology.Publication;Vol.1436).TampereUniversityofTechnology
Data management in end customer’s IT
systems
Control system and automation
data
CMMS event data
Data fromonline/offline
conditionmonitoring
Assetavailability & performance
data
Offlinequality
assurancedata
Online quality
control data
Locationdata
Processevent data /
Diaryremarks
Productionplanning
data
Online process data
RAMS analysesand OEM
information
Cost and business
intelligenceinformation
1) Integration of OEM’s service at operationallevel; to enable efficient processes for e.g. spareparts management, service calls, online support.
2) Integration of OEM’sservice at tactical level, to enable efficient processes for e.g. maintenance planningand performance optimisationwith the new knowledgederived from the data and related analytics.
3) Integration of OEM’sservice at strategic level, to enable efficient processes for e.g. investment decision-making support, and sustainable lifecyclemanagement of the assets.
MACHINEPERFORMANCEANDAVAILABILITYPREDICTIONTHROUGHTHELIFETIME
-5-
2. MACHINE PERFORMANCE AND AVAILABILITY PREDICTION THROUGH THELIFETIME
PäiviKivikytö-Reponen,RistoTiusanen,JariHalme,KalleRaunio,OlliSaarelaVTTTechnicalResearchCentreofFinlandLtd,MachineHealthTeampaivi.kivikyto-reponen@vtt.fi
ABSTRACT
Machine performance, availability and long lifetime play important role in operations andmaintenanceplanningforindustrialplayerse.g.withheavyworkingmachines.Recently,circulareconomy discussions have even more highlighted the role of the maintenance and lifecycledesign from the efficiency and sustainability perspective. [1] The new technologies areconsideredenablersforcirculareconomychangee.g.additivemanufacturingenablesadvancedsolutions forboth functionaloptimizationand localmanufacturing, furthermore the increasedintelligenceoftheproductsandthedataavailabilityduringthewholelifetimeenablereliabilityandpredictabilityofmachineperformance.
The manufacturing companies have a crucial role for implementing these changes into theproductsandmachines theyproduce.Furthermore, inmachinery industry,change fromsinglemachines to automated machinery fleets brings new challenges for safety, reliability andmaintenance; autonomy in mobile machine applications will increase and operator-assistingtechnologies will be widely utilized. New energy supply systems in machinery, like hybridtechnologies,fullelectricandfuelcellsystemsarecomingtomobileworkmachinesaswell.Weconcentrate in this paper to the importance of early state concept design, design formaintenance, reliability, availability and safety and how significant advantages in OverallEquipment Effectiveness (OEE) can be gained through hybrid modelling and an increasedunderstandingofthesystem’sbehaviourthroughoutitswholelifecycle.
INTRODUCTION
Increased awareness of sustainability and new profitable circular (economy) businessmodelsare currently under discussion. Circular (economy) design strategies heavily include productlongerlife,lifeextensionstrategiesandproductmaintenancestrategies.Stilllongerproductlifestrategiessuchas, reparability, refurbishment, remanufacturingarecurrentlyunderdeveloped.[2]
Data-drivenmodelling, including “Deep Learning” and “Big Data Analytics” receives currentlysignificant business focus and publicity. It has proven successful in several application areas.Furtherchallengeistointegratedetailedknowledgeanddatathroughthelifecycleinall levels(e.g. application domain knowledge with measurement data etc.) for product lifetimeoptimization,maintenance and operation optimization and in order to identify new risks andprobablefailures,andtoprovesystemsafetyandavailability.Newsystemicconceptsandtoolsincreaseunderstandingof thesystemicchallenges,e.g. inSystemIntegratedMetalProduction[3]showexampleforsolvingrecyclingrelatedchallenges.
MACHINEPERFORMANCEANDAVAILABILITYPREDICTIONTHROUGHTHELIFETIME
-6-
NEWCONCEPTDESIGN-NEWOPPORTUNITIESANDNEWRISKS
Novel technology access to market and customer satisfaction, with confidence of safety andreliability in the first of the kind implementationswill create new business opportunities andimproves profitability in themachinery business. Especially safety needs to be ensured in allcircumstanceswhendevelopingnew technology. Systematic concept design and systemic riskassessment approach supports the development of new technology, including traceabledocumentationthroughoutthedevelopmentanddecisionmakingprocess.Itisessentialtofindnew opportunities for longer machine lifetime, to identify uncertainties and risks in newtechnologies involved and find solutions to control risks already in concept design level formanaginglifecyclecostsandoptimizingprofit.
DESIGNANDMANUFACTURINGFORMAINTENANCE
Investment in the predictive maintenance increases the considerably safety, reliability andavailabilityofthemachineorfleet.Thegoalofmaintenanceistoavoidthefailurestypicallybyusing sensor data to monitor a system. The knowledge and data from maintenance andoperation conditions are vital information for system understanding.Machine design processgenerally combine the data related to lifecycle (design-manufacturing-use/operations-remanufacturing-reuse/operations-end-of-life), unfortunately this data can be partly lacking,especiallyinnewtechnologylaunch.Stilldeepapplicationdomainknowledge,understandingofthemachineperformanceduringmultipleoperationconditionsandavailableon-linemonitoringhistorydatewouldgiveadditionalinputtowardsthedesignandengineering.
Additionally design and manufacturing lay ground for maintenance and machine operationoptimization. Novel product data and operational data management including predictivemaintenancemayrequirere-designstartingfromearlystepsofproductdesignprocessinordertore-builtandmanufacturebuilt-inmaintenanceandpredictivecapabilitiesseamlesslyintothemachine.Thesecouldbee.g.embedded intelligence,on-linemeasurementsystems inside thestructuresetc.integratedintostructuresinamanufacturingphase.
KNOWLEDGEDRIVENOPERATIONANDMAINTENANCEPREDICTION
Measurement data together with application domain knowledge provide a key basis foroptimizingbothmachineoperationandmaintenance (O&M).Significantadvantages inOverallEquipment Effectiveness (OEE) can be gained through an increased understanding of thesystem’sbehaviourthroughoutitswholelifecycleandanticipatingthewear,corrosion,fatigueand other failures of machines, structures, and process parts. In practise for example thefracturemechanicsequationsandinterpretationofacousticemissionsignalsfromthestructuresandthematerialdislocationmovementsarecombinedforincreasingphenomenaunderstandingformaintenancepurposes.
Data-driven models are in general fast and cheap to develop, and can consequently bedeveloped for most applications with a fixed amount of resources. For machinery O&M,however, the data-driven approach utilizes only part of available information. In hybridmodelling, data analytics and knowledge ofmachine design, its operating characteristics, anddegradation mechanisms are used to supplement each other. This combination of theapplication data through lifecycle provides a significantlymore reliable basis for analytics by,e.g.,reducingsensitivitytodataqualityandrepresentativityissues.Eventhoughlargeamountsof data are continuously accumulated from machine fleets, the problem space spanned by
MACHINEPERFORMANCEANDAVAILABILITYPREDICTIONTHROUGHTHELIFETIME
-7-
possible operating and load histories of mobile machinery, different machine designs andinstrumentationsetups,andpotential(possiblymultiplesimultaneous)failuremodescannotbecovered by any practical data set. The relative strengths of the modelling approaches aredepictedinFigure1.
Figure1.Hybridmodellingcombinesdata-drivenandfirstprinciplesmodelling,balancingtherelativeadvantagesofeach.
CONCLUSIONS
Systemicthinkinganddata-driventoolswithapplicationdatathroughthelifecyclecanbeseenas enabler for solving and understanding systemic challenges in design, operations andmaintenanceinmacro,mesoandevenmicrolevel.Data-drivenmodellinghasprovensuccessfulinseveralapplicationdomains.Furtheredwithsystemapproach,combiningthedomaindatatothemeasureddata,thenextstepinperformanceunderstandingandpredictionscanbereachedby utilizing hybrid modelling. In hybrid modelling, data analytics and knowledge of machinedesign,itsoperatingcharacteristics,anddegradationmechanismsareusedtosupplementeachother.Integrationofdetailedknowledgeanddatathroughthelifecyclewillincreasethesafety,reliabilityandmoreaccuratefailurepredictability.
REFERENCES
[1]CircularEconomySystemDiagram14.10.2017https://www.ellenmacarthurfoundation.org/circular-economy/interactive-diagram[2]ConnyBakkeretal.JournalofCleanerProduction,Volume69,15April2014,Pages10-16,http://www.sciencedirect.com/science/article/pii/S0959652614000419[3]MarkusReuter,Plenarylecture,RecyclingMetalsfromIndustrialWaste.AShortCourseandWorkshopWithEmphasisonPlantPracticeJune23-25,2015.
Advantages
Data-driven modelling
First principles modelling
Hybridmodelling
Ease and speed of development
Reliability of computed results
Ability to provide new insight
Insensitivity to data quality
Ease of modelling multi-variate non-linearities
Insensitivity to validityof assumptions
A priori knowledge
Measurement data
ENHANCEDMODELLINGCONCEPTANDCOMPUTINGPLATFORMFORSYSTEMLIFECYCLEANALYSES
-8-
3. ENHANCED MODELLING CONCEPT AND COMPUTING PLATFORM FOR SYSTEMLIFECYCLEANALYSES
Jussi-PekkaPenttinen,ArtoNiemi,KariKoskinen,EricCoatanéaTampereUniversityofTechnology,MechanicalEngineeringandIndustrialSystemsjussi-pekka.penttinen@tut.fi
ABSTRACT
Thispaper introducesanopenmodellingconcept for systemavailabilityand reliability studiesandacompatibledistributedcomputingplatform.Uniquelyourapproachallowscombiningandconnecting risk assessment and operation models defined with different techniques. Thisensuresahighdegreeoffreedomtoaccuratelydescribethemodelledsystem.Theconceptalsosupports to include custom process performance indicators for example to measuremanufacturing productivity.WeuseMonte-Carlomethods to create versatile analysis results.Thevastandcomplexmodelsrequireheavycalculationsespeciallyforsensitivityanalyses.Asasolutionwe create aparallel computingenvironment topermit veryefficient simulations.Weseehighpotentialinusingourapproachinperformance,energyefficiencyandsystemlifecycleanalysesofindustrialprocesses.
INTRODUCTION
Riskassessmentandprobabilisticsafetyanalyseshavealonghistoryinsystemsengineering[1].Same techniques can be used for assessing lifecycle performance of production andmanufacturing processes. For these analyses industry favours Overall Equipment Efficiency(OEE)[2]astheKeyPerformanceIndicator(KPI).TheOEEdependsonreliabilityandavailabilitybutmightnotbedirectresultofthem.ThiswasthecaseinCERNcollideroperations[3],whichmotivated us to define an OpenModelling concept for Availability and Reliability of Systems(OpenMARS). The concept is created in collaboration with CERN, Tampere University ofTechnology(TUT)andriskanalysisexpertcompanyRamentorOy.ThespecialneedsrecognizedinmodellingCERNparticleacceleratorsarecombinedwiththedecadelongexperience[4,5]inmodelling of concrete use cases that go beyond the scope of traditional risk assessmenttechniques.
The OpenMARS concept allows defining models with the most common risk assessmentmodelling techniques [6], such as Fault Tree Analysis (FTA), Reliability Block Diagram (RBD),Markovanalysis,FailureModeandEffectsAnalysis(FMEA)andPetriNet(PN).Furthermore,ourapproach is scalable and open to support additional modelling techniques and theircombinations. This allows to collect all the information to a comprehensive Reliability,Availability,MaintainabilityandSafety(RAMS)model.Figure1illustrateshowtheRAMSmodel,technicalperformanceanddesignrequirementscanaffecteachotherandformtheinformationfor the systemdesign anddevelopment. This information allows to comparedifferentKPIs ofalternativescenariosanddesigns.
ENHANCEDMODELLINGCONCEPTANDCOMPUTINGPLATFORMFORSYSTEMLIFECYCLEANALYSES
-9-
Figure1.RAMSmodelisanintegralpartofthesysteminformation
Various data formats exist for the definition of RAMSmodels [7, 8]. TheOpenMARS uses anobject-orientedapproachthatallowstousesingledataformatfordefinitionofall techniques.AsillustratedinFigure2,thismakesitpossibletocreatesinglecalculationengineforStochasticDiscreteEventsimulation(DES)toproducedetailed,concreteandexplicitresults frommodelsmadewithanytechnique.
Figure2.Singledataformatandcalculationengineforvariousmodellingtechniques
OPENMODELLINGCONCEPTFORAVAILABILITYANDRELIABILITYOFSYSTEMS
TheOpenMARSmodels consistofelements thathaveclasses.Theclassdefineswhichkindofattributes the element has. Each risk assessment modelling technique has a catalogue ofavailable classes. Thegoal is that themost cases shouldbe coveredwith them.To guaranteethat the framework is always applicable, the expert users are allowed to extend the modelspecific classes to tailor them for their specific needs. For example newKPI can be added byintroducinganewclassfortheframework.
TheOpenMARSdefinesmodels in tabular format.Tablesarenaturalway to store themodelsintoadatabaseand canbeused forexample toefficientlydefinevastmodelswith repetitivestructures [9]. Themodel tables areplatform independentandhuman readable. TheFigure3illustrateshowtheuserselects thestudiedapproachandanalysesthemodel.TheOpenMARSdatabase collects the expert knowledge and links to other databases containing for examplehistorydata.Optimallytheresultsofthecalculationjobsarestoredagaintothedatabase.
Informationforsystemdesignanddevelopment
RAMSModelAvailability
Reliability(Failures)
Maintainability(Preventive,Corrective)
Safety(Hazards,Harms)
Costs(Break,Downtime,Repair)
Risks/Damages(Business,Property,Environment,Human)
Supportability
Technicalperformance(Operationprofile)
Designrequirements(Legislation,Customers)
ENHANCEDMODELLINGCONCEPTANDCOMPUTINGPLATFORMFORSYSTEMLIFECYCLEANALYSES
-10-
Figure3.Thecomputinganddatastorageenvironment
ThemodelforCERNcollideroperations[3]requiredtocombinesemi-Markovoperationsmodelwith a fault tree. The active operational phase can change the failure probabilities and thefailureeventscanactivateaMarkovtransition.InourmodeltheKPIisthecollisionproduction,whichdependsontimespendoncertainoperationphase.However,thisisnotalinearrelation.AcustommathematicalfunctionisrequiredtocalculatethisspecificKPI.OuraimforOpenMARSistoformanobject-orientedapproachthatcansupportcommonstandardtechniques.Wealsogivetheexpertusersapossibilitytotailorthemethods,whichmakestheapproachexpandable.
CALCULATIONENGINEFORSTOCHASTICDISCRETEEVENTSIMULATION
The object-oriented framework is designed to decouple the model specification and thesimulation engine from user-interfaces and custom user-supplied code. The Monte Carlosimulation based assessment of a system is parallelizable in a distributed computing clusterleading to simulation speedup by a factor 10 to 100. Very efficient calculations can bemaderemotelyfromanylocationwiththelayoutillustratedinFigure4.
Figure4:Distributedcomputingenvironment
CONCLUSIONS
OpenMARS approach allows combiningmodelsmadewith common standard risk assessmenttechniques and permits tailored features for special needs such as defining specific KPIs. Theapproach is suitable for distributed computing environment, which can vastly speed up theMonteCarlosimulations.Theadaptabilityofourapproachfordifferentmodellingcasesmakesitvery efficient for complex system life cycle analyses and we see high potential in furtherapplicationsofourapproach.
ENHANCEDMODELLINGCONCEPTANDCOMPUTINGPLATFORMFORSYSTEMLIFECYCLEANALYSES
-11-
REFERENCES
[1] W. Vesely, F. Goldberg, N. Roberts, and D. Haasl, Fault Tree Handbook (U.S. NuclearRegulatoryCommission,Washington,1981).[2] T. Kanti Agustiady and E.-A. Cudney, Total Productive Maintenance: Strategies andImplementationGuide(CRCPress,BocaRaton,2015),p.111.[3]A.Niemietal.Phys.Rev.Accel.Beams19(2016)121003.[4]S.Virtanen,P.-E.HagmarkandJ.-P.Penttinen,Proc.oftheRAMS’06.AnnualReliabilityandMaintainabilitySymposium,(IEEE,2006),pp.506–511.[5] J.-P. Penttinen and T. Lehtinen, Proc. of the 10th World Congress on Engineering AssetManagement(WCEAM2015),(Springer,Cham,2016),pp.471–478.[6] European Committee for Electrotechnical Standardization, Risk management. Riskassessmenttechniques,EN31010:2010,(2010).[7]E.StevenandR.Antoineeds.Open-PSAModelExchangeFormat (TheOpen-PSA Initiative,2017).[8]InternationalOrganizationforStandardization,High-levelPetrinets-Part2:Transferformat,(2011).[9] O. Rey Orozko et al. Proc. 7th Int. Particle Accelerator Conf. IPAC’16, (JaCoW, 2016), pp.4159–4162.
LIFETIMEASSESSMENTFORMECHANICALCOMPONENTS
-12-
4. LIFETIMEASSESSMENTFORMECHANICALCOMPONENTS
SatuTuurna,PerttiAuerkari,TimoJ.Hakala,AnssiLaukkanen,SanniYli-OlliVTTTechnicalResearchCentreofFinlandLtd.Satu.tuurna@vtt.fi
ABSTRACT
Lifetimeassessmentofcomponentsunderextremeconditions,suchashightemperaturesandhighpressures,requirecharacterization,testingandmodelling.Byextendingthelifetimesavingscanbeachievedandunpredictedshutdownsprevented.Hereweshowtwoindustriallyrelevantcasesforlifetimeassessment;hightemperaturecomponentinpowerplant,reheater,andgear,andhowtheirlifetimecanbeevaluatedandextended.Herewehavecalculatedtheremaininglifetime for the components with defined design and condition parameters. At the time ofinspection,thereheatertubehadbeen inserviceforabout180000happroachingthedesignlifetime and based on the evaluations carried out, the estimated minimum residual life ofcomparabletubeswaspredictedtobe50000h.Thelifetimeofcoatedgearcanbeextendedbyalmost300%whensurfacetexturingused.
INTRODUCTION
The aim is to present which kind of tools can be used for the life assessment of differentindustrial cases.The firstaims to showhowtheappearanceofcreepdamage in ferritic steelscanprovideusefulcluesfor lifeassessmentundervaryinginitialandoperationalconditions. Ingeneral,theexpectedlifeisnotastrongfunctionofthematerialgradeforthesameextentofdamageordegradation,mainlybecausetheoperationalconditionsshould largelycompensateforthedifferencesinmaterialperformance.Theexamplecaseforboilerinternalswasselectedto demonstrate the long term impact of microstructural degradation and oxidation, and theeffectsontheestimatedservicetemperature.
TheFEmodellingwasusedtoshowwhythecoatingswithlowersurfaceroughnesshavebetterwear resistance in the gear applications, considering the location and propensity of weardamageinitiationatdifferentscalesofroughnesstopographies.Theinitiationofsurfacedamagein gear contacts is quantified for surfacedesignpurposesby introducingawear resistanceFEmodel.TheFEmodelcanbeappliedtoestimatethetimethegearcanoperatebeforeinitiationofdamagebecomeslikelyfordifferentsurfaceloadingsandtopographies.
LIFETIMEASSESSMENTOFCOMPONENTINPOWERPLANT
Observations of in-service thermal degradation and oxidation arewidely used to support lifeassessmentofsuperheatersandreheaters,particularlytoestablishtheeffectivetubematerialtemperatures. In parallel, observations of creep cavitation damage in hot headers and steamlines are commonly providing indications of the current and expected integrity of weldedcomponents.Thoughrelativelycommon,themethodsarenotfullystandardized,butguidelinesexistforevaluatingcreepcavitationdamage.
Oneoftheprincipalquantitiesofinterestinlifeassessmentofsuperheatersandreheatersistheeffectivemetaltemperatureduringservice.Ascreepandotherdamagemechanismsarehighlysensitivetotemperature,uncertaintiesinassessingthistemperaturecanresultinconsiderableerrors in the expected life. Sourcesof uncertainty (error) for different temperature indicatorsareshowninTable1.
LIFETIMEASSESSMENTFORMECHANICALCOMPONENTS
-13-
Table1.Errorsourcesinestimatingtubemetaltemperatures.
Indicator Sources of error Notes
Microstructure Limited t-T range Insensitive at low temp.
Hardness Limited range Mainly for 10CrMo9-10
Internal oxide Spalling, porosity From layer imbalance
Thermocouples Drift, temp. limits Often K-type
Diffusion couples Non-standard -
Superheaters and reheaters often have metal temperatures exceeding that of the steam bysome15-50°C. For a fixedoutlet steam temperature such a rangeofmetal temperatures canmeanawiderangeintheexpectedtubelife,as10°Cincreasemayapproximatelyhalvethetubelife. The tubes are subjected to the fireside environment with through-wall temperaturegradients and non-constant material temperatures in addition to wall thinning by externaloxidation, hot corrosion and erosion in the flue gas. Also, thermal insulation by the growinginternal oxide is resulting in progressively increasing material temperature of the tube, ifapproximately constant heat flow is extracted from the superheater/reheater section. Tocountertheeffectoftheinternaloxide,thislayermayberemovedbychemicalcleaning.
ThemicrostructureandinternaloxideforareheaterofacoalfiredplantisshowninFig1.Thisreheatertube(10CrMo9-10orT22steel)hadbeeninserviceforabout180000h.Lifepredictionforthetubeswasbasedonasimplifiedeffectivetemperaturehistory,essentiallyreducingthethreepastoxidecleaningepisodestoasingleeffectivecycleandthentoanaturaloxidegrowthprediction from the time of assessment onwards. An in-house software tool was used tocombinetheeffectsofstress, temperatureandcreepstrengthevolutiontoobtainapredictedlife(ref.Auerkariet.al.2010). Inthiscasethethicknessof internaloxidefailedtoindicatethemetal temperatures due to oxide spalling, and estimates were based on microstructuralevolution and hardness. A Larson-Miller type of expression was applied to provide a time-temperature master curve with constant features in the microstructures. The estimatedminimumresiduallifeofcomparabletubeswaspredictedtobe50000h.
Figure1.Degradedmicrostructureofareheatertube,suggestinganeffectiveservicetemperatureof575-600°Candobservedinternaloxidelayer
LIFETIMEASSESSMENTFORMECHANICALCOMPONENTS
-14-
LIFETIMEASSESSMENTOFCOATEDGEAR
Afinite-element(FE)modelofgearcontactwascreated(Fig.2).Themeshsizeofthemodelwasone millimeter. The model was operated with certain parameters and the generated forceswereappliedintoanothermodelwheretworoughsurfaceswereslidingoneachother.
Figure2.FE-modelofgearcontact
Basedon the simulations (Fig. 3) there are both tensile (red) and compressive (blue) stressesformedontheteethsurface.Thevaluesofthesestresseswereappliedintoanothermodel(ref.Holmberget.al.2017)wheretworoughsurfaceswereslidingoneachother.
Figure3.Simulationofstressesformedingearcontact
LIFETIMEASSESSMENTFORMECHANICALCOMPONENTS
-15-
Thestressesformedonthesimulationwereappliedintothemodelwherethestressesbetweentworoughsurfacesweresimulated.Thesurface roughsurfaceshad threedifferent roughnesslevels;smooth,averageandrough.Thetexturewasformedbygrindingmarks.Theaverageandsmoothsurfaceswereformedbypolishingtheroughsamples.
Fromthesimulationresultsitcanbecalculatedthelifetime(cycles)ofthegearifthefailureofthe gear origins from the coating cracking. As shown in Fig. 4 the lifetime of the gear issignificantlyextendedwhen the slidingof the surfaces is in45degreesdirection towardeachother. Inthatcasethere isacomplexstressfieldformed intothecoatingsandnohightensilestressesareformed.
Figure4.Lifetime(cycles)ofthegearin0,45and90degreesdirectiontowardthegrindingmarks
CONCLUSIONS
The estimatedminimum residual life of comparable tubeswas predicted to be 50000 h. Thereheater section was included in a follow-up inspection program, and no incident hascontradictedthepredictionaftermorethan20000hoffurtherservice.
Finite-element(FE)-modelforgearcontactwascreated.Bysimulationsthelifetimeforcoatedgearscanbecalculated.Thelifetimecanbeextendedby300%whenslidingoccursindirectionof45degreescomparedtogrindinggrooves.
REFERENCES
Auerkari,P.,Salonen,J.,Holmström,S.,Laukkanen,A.,Laaksonen,L.;Mäkinen,S.;Väänänen,V.;Nikkarila, R. High temperature damage as an indicator of operating conditions and residualcreep life.12th InternationalConferenceonCreepandFractureofEngineeringMaterials,May27-31,2012,Kyoto,JapanLaukkanen,A.,Holmberg,K.,Ronkainen,H.,Stachowiak,G.,Podsiadlo,P.,Wolski,M.,Gee,M.,Gachot, C., Li, L. Topographical orientation effects on surface stresses influencing onwear inslidingDLCcontacts,Part2:Modellingandsimulations(2017)Wear,388-389,pp.18-28.
DYNAMICSERVICEDECISIONSUPPORTPLATFORM
-16-
5. DYNAMICSERVICEDECISIONSUPPORTPLATFORM
HenriVainio,AnandaChakraborti,JussiAaltonen,KariT.KoskinenTampereUniversityofTechnology,MechanicalEngineeringandIndustrialSystemshenri.vainio@tut.fi
ABSTRACT
Connectingmultiplesourcesofdatafromdifferentlevelsofamanufacturingprocessopensuppossibilitiesforanalysisandoptimizationoftheprocess.Thesedatasourcesincludeworkpiecelocations, sensor readings frommanufacturingequipment,processdata suchasERPandMESdata, and eventually expanding to subcontractor data as well as other third party data, e.g.weatherdata.Combiningthesesourcesinameaningfulwayenablesthediscoveryofsystem-of-systems level phenomena and patterns and thereby creates knowledge to support processoptimization and service delivery. The technology required for a decision support system likethisincludethedatagathering,transfer,storage,preprocessingandanalysismethodsaswellaspresenting the relevant results to the end user in a clear and intuitive manner. This workconcentrates on processing, analyzing and visualizing data gathered from storages anddatabases of a number of stakeholders. A Decision Support Platform is created using opensourcesoftwareandan industrial case ispresented.Themaincontributionof thiswork is theutilization of interlinked open source analytics to perform data fusion, analysis and decisionsupportcreationforamanufacturingprocess.
INTRODUCTION
The Industrial InternetofThingsparadigmenablesnewandmoreefficientwaysofsupportingproductsandofferingindustrialservicesintheProduct-ServiceSystem(Goedkoopetal.,1999)concept.Gatheringdataandinformationfrommultiplesourcesrelatedtoanindustrialprocessmakesitpossibletocreatenewknowledgeusingmultiplemutuallysupportinganalysismethods(Assahlietal.2017).Thesemethodscanbelinkedtorunonaserverandautomaticallyanalyzeand predict the states of the process, offering near-real time decision support for industrialservices.SuchaDecisionSupportPlatformwastestedbyconnectingamanufacturingprocessofamachinepart factory to cloud serversand combiningproduction information in the formoflead time plans. These data were gathered in a database and then used as inputs for theDecision Support Platform, with an Agent BasedModel as its basic structure, enhancedwithinternal analytical functionalities. The platformwas able to provide near-real time situationalinformationaswellasbreakdownsofpasteventchains.TheresultswerevisualizedusingbothHTML-baseddashboardsanda3D-modelofthefactory,displayedonawebsite.Allofthetoolsusedinthestorageofdata,analysisandvisualizationwerecreatedwithopensourcesoftware.
DECISIONSUPPORTPLATFORMSTRUCTURE
Datafromanindustrialprocessisgatheredinvariouswaysandbyseveralorganizations.Sensordata, process data andmaintenance data are saved to a number of locations by the processownerandserviceprovidersandoftensomeformofpreprocessingisperformed.Thedatacanbeaccessedusinganumberofprotocols,suchasHTTPqueriesornetworksocketconnectionsincaseofacloudserverdatabase,orevenperiodicalmanuallyactedtransmissionsincaseofon-sitedata. Ineachofthesecasestherearewaysof limitingtheaccesstothedata,sothatonlythe data relevant to the application requesting it is available. Accessing the data using thesemethods,aserviceproviderisablesavethedatatoadatabaseforpreprocessing.Timestamps
DYNAMICSERVICEDECISIONSUPPORTPLATFORM
-17-
andmeasurementfrequenciesmustbematched;outliersandclearmeasurementerrorsaswellasmissingdatamustbecleaned(Tsaiet.al,2014).Afterpreprocessing,datafusionandanalysiscan be performed. The approach taken in thiswork uses an Agent BasedModel as the basicstructure of theDecision Support Platform. The various data are used as inputs for theMultiAgentSystem(Weiss,2013),whichrepresentsthecomponentsintheindustrialprocess,suchaswork orders, NC-machines ormachine subsystems. These agents, aware of their current andpastsstates,arealsoabletoaccessprocessinformation,e.g.leadtimeplans,andcomparetheiractual situation to their intended situation. They are able to trace delays in the past andassociate themwitherror reports fromotheragents, creatingaholistic viewof the causesofproductiondelaysanderrors.Onceenoughdataaccumulates,machinelearningmethodscanbedirectly added to the model and bottleneck detections, delay predictions and learningoptimizersbecomepossible.
The output of themodel is visualized on awebsite using HTML-based dashboards and a 3D-model of the factory. The 3D-model offers an intuitive view of the situation at a glance,featuring color codes formachine and subsystem statuses as well as location information ofworkpiecesandtheirpastandplannedpaths.Thedashboardsoffermoredetailedresultsfromtheagents,suchasrootcauseanalysesfordelays,usageratesofmachinesandsubsystemsandleadtimebreakdowns.
TheDecisionSupportPlatformcanbecreatedusingopensourcesoftware.ThetoolsetisbuiltusingPython.DatapreprocessingandfusionisperformedusingPANDASandSciPy,AgentBasedModels are createdwithMesa,whilemachine learning tools are provided by Scikit-learn andKeras. Analytical models built with OpenModelica can be interfaced. The results of theseanalyticscanbeshownonawebsiterunonaNodeJs-serverwithaHTML-baseduserinterfaceaswellas3D-visualizationscreatedusingBlenderandBlend4Web.Theseopensourcemethodsare extremely versatile and can be tailored tomeet the requirements of vastly different usecases.
RESULTS
TheDecision Support Platformwas tested in a factory thatmanufacturesmachine parts. Theworkordersthatmovethroughtheprocesswerefittedwithradiotagsthatallowtheirlocationstobemonitored.SomeoftheNCmachinesintheprocesswerealsofittedwithradiosensortagsmonitoring vibration and heat. The radio tag data was collected and preprocessed by thecompany providing the tags and saved on their cloud server. The PLC data of a storagemanagementsystemintegraltotheprocesswasalsoaccessedthroughthecloudserverofthecompanythathadmanufacturedthesystemandactedasan industrialserviceprovider inthiscase. Lead time plans of the manufacturing process were sent by the company owning thefactory.
TheAgentBasedModelincludedagentsrepresentingtheworkorders,theNC-machinesandthesubsystems of the storage system. The work order agents were able to calculate delays andcreatelistsofdelayswithrelatederrorreports.ThestructureoftheDecisionSupportPlatformispresentedinFigure1.
DYNAMICSERVICEDECISIONSUPPORTPLATFORM
-18-
NodeJsServer
OpenModelicaAnalyticalmodels
NumPy
SciPy
Scikit-learn
KerasPandas
Python-basedAnalytics
Blend4Web3D-
visualization
HTMLUserInterface
Datacollectingandfusion
MongoDBDatabase
Figure1:ThestructureoftheDecisionSupportPlatform
Thevisualizationincludeda3D-modelofthefactoryinquestion,createdfromaSolidWorksfileusingBlender andBlend4Web.Changing status colors for themachines and subsystemswereaddedaswellasannotationsdisplayingthecurrentstatusofthemachineinmoredetail.Workorder locationsanddelay informationwasdisplayedaswell as thepastandplannedpathsofthe work orders in the process. The model could be scaled upwards to present the entireproductionnetwork includingmultiplesitesownedbytheprocessowningcompanyaswellassubcontractor statuses, though no subcontractorswere involved in the case at this time. TheHTML-based dashboards could be used to offer more detailed historical analyses of thisinformation.Thedifferentviewsofthevisualizationarepresentedinfigure2.
Figure2:Machinelevelstatusvisualizations,workpiecelocationinformationandproductionnetworkvisualization
CONCLUSIONS
Gathering data and information from an industrial process enables efficient processoptimization. Such optimization models can be created using open source software, which
DYNAMICSERVICEDECISIONSUPPORTPLATFORM
-19-
allowsthecustomizationofthesolutionstomeettheneedsoftheenduser.Sinceitislikelythata real world scenario includes both software provided by companies, aswell as open sourcetools, it would be advisable for the software companies to provide standard interfaces andtransparentmethodsinordertofacilitatewhatevercustomizationtheircustomersneed.Futureworkincludesdataanalyticswithmachinelearningmethodsandextendingthemodeltoincludesubcontractorsforfullnetworkvisibility.Addingdatafromthirdpartysourcessuchasweatheragencieswouldenablethedetectionandanalysisoftheeffectsofevenlargerscalephenomena.
REFERENCES
Assahli, S., Berrada, M., Chenouni, D. 2017. Data preprocessing from Internet of Things:Comparative Study, Wireless Technologies, Embedded and Intelligent Systems (WITS), 2017InternationalConferenceon.Goedkoop,M.,vanHalen,C.,teRiele,H.,Rommens,P.1999.ProductServiceSystems,EcologicalandEconomicBasics.Tsai,C.,Lai,C.,Chiang,M.,Yang,L.T.2014.DataMiningforInternetofThings:ASurvey.IEEECommunicationsSurveys&Tutorials,vol.16,No.1.Weiss,G. IntelligentRoboticsandAutonomousAgents:MultiagentSystems.2013.MITPress,2ndedition.917p.
IMPACTOFMAINTENANCEONCIRCULARECONOMY
-20-
6. IMPACTOFMAINTENANCEONCIRCULARECONOMY
PasiValkokari,JyriHanski,ToniAhonenVTTTechnicalResearchCentreofFinlandpasi.valkokari@vtt.fi
ABSTRACT
Novel business models that leverage the largely untapped potential of circular economy aregradually emerging in several industries. The widespread use of circular economy hasimplicationsforassetmanagementinmanufacturingindustry,astheybringnewchallengesandopportunitiesformanagingandmaintainingmachineryandequipment.TheimportanceofassetmanagementandmaintenanceforthecirculareconomyisstudiedintheTekesfunded“DatatoWisdom”project.
Circulareconomycanbedefinedasasystemwhichcreatesvaluebyminimizingwaste,energyand the use of natural resources. It also utilises solutions that aim at slowing, closing andnarrowing loops of material and energy. These solutions include, for instance, long-lastingdesign,designforeasydisassemblyandmaintainability,maintenanceactionssuchasrepairandrefurbishing,reuse,remanufacturingandrecycling.
In this paper, we analyse circular economy archetypes from the life cycle managementperspective.Thisisdonebycomparingcirculareconomyarchetypesagainstthefundamentalsofasset management. Opportunities and threats produced by the circular economy for assetmanagementandmaintenanceareaddressedwithrealcasespresentedinSitra’slistofthemostinterestingcompaniesinthecirculareconomyinFinland.
INTRODUCTION
Asset management (AM) is a “coordinated activity of an organization to realize value fromassets” (ISO 55000, 2014). Considering physical assets, AM aims at optimising the life cyclebusiness impact of cost, performance, risk exposures and ensuring the availability, efficiency,quality, longevity and regulatory/safety/environmental compliance (IAM, 2008). Assetmanagement realises asset value through operation, maintenance and investment activities(Komonen,2008).
Circulareconomy(CE)isasystemwhichcreatesvaluebyminimizingwaste,energyandtheuseof natural resources (Geissdoerfer et al., 2017). CE solutions aim at slowing, closing andnarrowing loops of material and energy. These solutions include, for instance, long-lastingdesign,designforeasydisassemblyandmaintainability,maintenanceactionssuchasrepairandrefurbishing,reuse,remanufacturingandrecycling.Figure3visualisestheconnectionbetweenAMandCE.
IMPACTOFMAINTENANCEONCIRCULARECONOMY
-21-
Figure3.Assetmanagementandcirculareconomy.
FromtheperspectiveofCE,figureaboveillustratesmainlytechnicalcyclessuchaslifecycleofmachineryandequipment.AMactivities takeplace indifferent lifecyclephasesand feedbackloops.EventhoughthegoalsofCE,maintenanceandAMarelargelycongruent,theresearchonthe impactofCEonmaintenanceandassetmanagementhasbeen scarce.DescriptionsofCEsolutionsgenerallydonotconsidertheopportunitiesandthreatsforAM.
Therefore,ouraim is toanalysecirculareconomyarchetypes fromthe life cyclemanagementperspective. We compare circular economy archetypes against the fundamentals of assetmanagementdescribedinISO55000(2014).Opportunitiesandthreatsproducedbythecirculareconomy for assetmanagementandmaintenanceareaddressedwith real casespresented inSitra’slist(Sitra,2017)ofthemostinterestingcompaniesinthecirculareconomyinFinland.
OPPORTUNITIESANDTHREATSOFCIRCULARECONOMYFORMAINTENANCE
TheimpactofmaintenanceandassetmanagementonCEcanbeviewedthroughthefourassetmanagementfundamentals:1.valuethatisprovidedtotheorganisation,2.alignmentofplansandactivities, 3. leadership and culture that ensures that employees in theorganisationhaveclearroleandresponsibilitiesandarecompetentandempowered,and4.assurancethatassetsfulfil their purpose (ISO 55000, 2014). Sitra (2017) classifies CE solutions into five categories:product-life extension, product as a service, sharing platform, renewability, and resourceefficiency and recycling. Other often referred classifications are presented in Bocken et al.(2016)andLacyandRutqvist(2015).
Table1.ImpactofmaintenanceonCEsolutions.
IMPACTOFMAINTENANCEONCIRCULARECONOMY
-22-
CEmodel/example Threats Opportunities
Renewability
Closingresourceloops
Example
St 1: Petrol substitutefromorganicwaste
Assettype
Fleet of productionunits
Value
Treatment of hazardous materials and occupationalsafety
Potential quality challenges for the end product(heterogeneousrawmaterial)
Uncertainty about the durability of machinery andequipment due to changing characteristics of rawmaterialandlackofusagehistorydata
Alignment,leadership&assurance
More actors and stakeholders in the value chainincrease thecomplexityof theproductionsystemandthe organisation and cause new challenges fordecision-making, leadership and the assurance ofperformance
Demand for interoperability of technical andinformation systems of different actors in theecosystemtoshowthesustainabilityofproducts
Value
Several small production units increase deliveryreliabilityincomparisontoonelargeunit
Fleet of production units may have a commonsparepartstock
Alignment,leadership&assurance
Increased control and flexibility for the executionof planned shutdowns due to small productionunits
Accumulatedfleetdatacanbeusedtosupporttheassuranceofthemachineryreliability
Product-lifeextension
Slowingresourceloops
Example
Valtra:Remanufacturingtractorgearboxes
Assettype
Tractorgearbox
Value
Recognizing when it is not feasible to extend thelifetimeofgearboxes
Alignment,leadershipandassurance
In order to use the concept, there is a need forextensive information about the utilization, conditionandlocationofmachinery
Remanufactured machinery needs to be compliantwiththerequirements
Newkindoflifecyclephases(duetoremanufacturing)require the implementation of new assuranceprocesses
Value
Using same equipment for remanufacturingreducescosts(e.g.reductionofnewmaterialsandenergyconsumptioninproduction)
Alignment,leadershipandassurance
Predictivemaintenanceenablesearlydetectionofwear/defectsbeforefailureoccurs
Remanufacturing demands an advancedperformancemonitoringandsupportprocessestobe successful. This enables indirect benefits suchas decreased failures and disturbances, and newserviceopportunities
CONCLUSIONS
AssetmanagementandmaintenanceplayamajorrolewhenCEsolutionsareimplementedandtargetedstrategicobjectivesarepursued.Challengesrelatedtothemaintenanceofmachineryincreasedue to theheterogeneousmaterials.Additionally, CE solutions increase theneed fordata of usage history. On one hand, the implementation of CE solutions often requires therenewalofthedatacollectionandanalysis.Ontheotherhand,increasedutilisationoffleetdataenablesmoreefficientdecision-makinginAMactivities.
REFERENCES
Geissdoerfer,M.,Savaget,P.,BockenN.M.P.andHultink,E.J. (2017)TheCircularEconomy-Anewsustainabilityparadigm?Journalofcleanerproduction,143,pp.757-768.Bocken, N.M.P., de Pauw, I., Bakker, C. and van der Grinten, B. (2016) Product design andbusiness model strategies for a circular economy. Journal of Industrial and ProductionEngineering,33:5,pp.308-320.
IMPACTOFMAINTENANCEONCIRCULARECONOMY
-23-
IAM(2008)InstituteofAssetManagement.London.ISO55000(2014)Assetmanagementstandard.Komonen,K.(2008)AStrategicAssetManagementModel:Aframeworkofaplantlevelmodelfor strategic choices and actions. Proceedings of Euromaintenance 2008 Conference on assetmanagement&productionreliability,Bryssel,8-10April2008,pp.1-10.Lacy, P. and Rutqvist, J. (2015)Waste toWealth: The Circular Economy Advantage. PalgraveMacmillan.Sitra (2017) Themost interesting companies in the circular economy in Finland. Available at:https://www.sitra.fi/en/projects/interesting-companies-circular-economy-finland/.
IIDIGITALANDADDITIVEMANUFACTURING
DIGIMATURITYINMANUFACTURINGINDUSTRY
-25-
7. DIGIMATURITYINMANUFACTURINGINDUSTRY
Simo-PekkaLeino,Juha-PekkaAnttilaVTTTechnicalResearchCentreofFinlandLtd.Simo-Pekka.Leino@vtt.fi
ABSTRACT
The genericVTTModel ofDigimaturity helps inunderstanding and structuring the conceptofdigitalisation.Thispaperdiscusseshowitshouldbeextendedandmademoreprofoundtowardsthespecificneedsofmanufacturingindustryandproductprocesses.
INTRODUCTION
WeareinmiddleofthefourthindustrialrevolutionwhichisbasedonInternettechnologies,realtimeinformationflowandcombinationofavailabletechnologyinanewway(Kagermannetal.,2013).This isnotbasedontotallyneworunknownconcepts,but thescaleandcomplexityofdigitalization will be in totally new level in the future. Therefore frameworks and holisticperspectives on business models, strategies, processes, and enterprise architectures forstructuring the complex concept of digitalization are required. Germany based Industrie 4.0(Kagermannetal.,2013)isonenameforthefourthindustrialrevolution.However,Industrie4.0onlydescribesthevision,butdoesnotprovideacleardefinition.EvenmoreunclearishowitfitstoFinnishcompanies.Ourhypothesis is that Industrie4.0will rulealso theFinnish industry innear future, which requires radical improvement of data and information flow within valuenetworks.VTTModel of Digimaturity (http://digimaturity.vtt.fi) helps in understanding and structuringthe concept of digitalisation (Leino et al., 2017). Additionally, it gives an estimate oforganisations’ current capabilities andmaturity aswell as generaldirections towardadesiredmaturitylevel.Sinceitisagenericmodel,itshouldbeextendedtowardsmorespecificneedsofmanufacturingindustry.Thispaperdiscusseshowit isdevelopedandmademoredeep,takingmanufacturing, product processes and the Industry4.0 vision into account. The extension ispartly based on the ongoing Tekes-funded ”Digital Extended Enterprise – Dexter” –project(http://smacc.fi/projektit/dexter,fundedbyTekes,VTT,TUTandparticipatingcompanies),andits preliminary results and conclusions, particularly the “Dexter-model”. Dexter builds on theChrysler Corporation based notion of Extended Enterprise (EE) and studies how digitalizationenablesbettercollaboration,flexibility,informationflow,transparencyandtotalproductivity.
RESEARCHAPPROACH
Theresearchapproachisconstructive.ItbuildsontopoftheVTTDigimaturityModeladdingtherecognized sub-dimensions of theDexter-model that are important inmanufacturing industryand networks ofmanufacturing companies. TheDextermodel is relatively rich having almost2000 different mentions/topics highlighted in the case companies. The extended sub-dimensions were analysed in Dexter-project case studies with a network of one large OEMcompanyand four SME suppliersof them. The sub-dimensionswere reflectedwith literature.Additionally, the proposed extendedmodel has been tested and discussed in interviews andworkshopswithseveralotherSMEmanufacturingcompanies.Thesewereselectedfromthelistof companies that have applied the VTT Digimaturity Model web tool, and indicated theirinterestinfurtherdiscussions.
DIGIMATURITYINMANUFACTURINGINDUSTRY
-26-
RESULTSANDANALYSIS
Analysis of the Dexter-model raises the following themes in order to build a future digitalmanufacturingextendedenterprise:Strategicalcollaborationanddevelopmentofdigitalizationtogether,digitalandmodelbased integrationandfeedbackofengineeringdesignandproductdevelopment, transparency on every level, systematization and discipline of high qualityprocesses,developmentof reliableandpredictableengineeringchangeprocedures.Ascanbeseen,thesethemestouchallsixdimensionsoftheoriginalVTTDigimaturityModel,butascanbe guessed, the product related models and processes are emphasized. Additionally,digitalizationmaturityreferstotwodifferent,butinterrelatedthings(Leinoetal.,2017):Firstly,requiredcapabilitiesandorganisationreadinessfordigitalisation.Thismeansanorganisation’scapability and willingness (mindset) to change its function, processes, organisation andcapability to effectively adopt new technology. Secondly,maturity refers to an organisation’sperformance based on digital technology. Thus, the sub-dimensions and elements of theextended models are categorized within these two perspectives (Figure 1).Currently the extended model includes the recognized (Dexter-model and other industrialexperience, literature) elements categorized as sub-dimensions of the sixmain dimensions ofthe VTT Digimaturity Model. Practically speaking, the sub-dimensions work as more detailedstructureandquestions thatcanbeused for instance inworkshopsaiming tohelpcompaniescreating stepson their futuredigitalizationpath.The sub-dimensions cannotbe introduced indetailsinthislimitedspace.
Figure1.Requireddigitalizationcapabilitiesareprerequisitesforthepotentialincreasedbusinessperformance.Technologyisjustoneofsixdigitalizationdimensions.
Asanexample,therequiredcapabilitiesoftheOrganization&Processesdimensionstudiesforinstance the importance, definition and development of processes, types and modes ofproduction,typesofproductdevelopment,productandprocessqualityandKPIs,anddefinitionand typology of information streams. Potential performance of O&P includes e.g. levels ofinformationdigitationand integration,data flowand transformation towards information andknowledge,levelofrealtimeandtransparentflows,totaloptimizationofproductprocessesandnewrolesinorganizations.
DIGIMATURITYINMANUFACTURINGINDUSTRY
-27-
DISCUSSION
Maturitymodels have been a topic in academic research for decades.Maturitymodels havebeendevelopedandutilized for instance in softwaredevelopmentandPLM implementations.Our extendedmodel ofmanufacturing digitalizationmaturity contributes to the discussion ofdigitalizationasabusinessconceptandmaturitymodelasamethodandconsultancytool.Fromthe practical viewpoint it is proposed to boost digitalization in manufacturing industry byenabling understanding and structuring the extensive, complex and systemic concept ofdigitalization.The firstexperiencesand feedback fromcompanies ispositiveandencouraging.Thereseemstoberealdemandforsuchaframework.Samplesizeofcompaniesisstillsmallandresults are partly interpretations, so very solid conclusions cannot be drawn. However, themodel will be developed continually in dialogue with companies.Discussions with companies show that starting point for digitalization is mostly good:Commitment fromownersandtopmanagement,optimismandpositiveattitudetowardsnewtechnology,openculture,willingnesstoshareinformation,readinessfororganizationalchanges.Alsodigital tools suchasGooglemarketing, socialmedia (LinkedIn, Facebook) are surprisinglywell utilized also in SMEs. Many companies are interested in novel concepts and pieces oftechnologyaspartoftheirmarketing,sales,productdevelopmentanduserexperience:VR/AR,AI, model-based processes, Digital Twin, IIoT.ForFinnishmanufacturingcompanies,customization,flexibilityandagilityaretypicallyessentialcompetitive edges,which create challenges for digitalization.On one hand digitalization is anenabler,butontheotherhandburdenandbiginvestmentparticularlyforSMEs.Currentlydataand information do not typically flow optimally causingwaste in processes. Some companieshave realized thatdigitalizationcould result innewwaysof creatingvalueandnovelbusinessmodelswith opportunities to provide downstream services, such asmanagement, processingand infusion of data to knowledge. Digitalization may change the strategic position in valuenetworks,andcreatesopportunitiesforstart-upstoseamdigitalvaluenetworksinanewway.Digitalization is essentially a networked and systemic concept. It requires development ofcapabilitiesandinterfacesbetweenorganizationalfunctions,departments,suppliers,customers,partners,etc.Itshouldbeunderstoodthatnewdatabasedbusinessmodelsusuallyalsorequiresystemicchanges for instancetoproductprocesses,productdatamanagement,andcustomerinterface.
CONCLUSIONS
Largecompaniesareoftenseenasdigitalizationforerunners.However,aspartoftheirsuppliernetworks, SMEs should be strategically prepared for the digital transformation. Actually,digitalization capabilities increase credibility as part of global networks. On the other hand,starting point for digitalization is mostly good also in SMEs. Digitalization is recognized instrategicallevelandthereiscommitmentfromownersandtopmanagement,butitisnotclearhowtoproceed.Thereforeaframeworkfordigitalizationisneededincludingconceptdefinitionsandterminology-togetherwithexternalperspectivesandexpertise.TheprototypeofextendedDigimaturitymodel formanufacturing industry has gained positive feedback from companies,anditwillbefurtherdevelopedindialoguewithcompanies.
REFERENCES
Kagermann, H., Wahlster, W., & Helbig, J. (2013). Recommendations for implementing thestrategicinitiativeIndustrie4.0.FinalreportoftheIndustrie4.0WorkingGroup.
DIGIMATURITYINMANUFACTURINGINDUSTRY
-28-
Leino, S-P., Kuusisto, O., Paasi, J., Tihinen, Mt. 2017. VTT Model of Digimaturity. In: Towards a new era in manufacturing. Final report of VTT's For Industry spearhead programme. VTT Technology: 288. http://www.vtt.fi/inf/pdf/technology/2017/T288.pdf
IDENTIFICATIONOFFUNDAMENTALREQUIREMENTSFORIDEALMETAMODELINGFRAMEWORKINADDITIVEMANUFACTURING
-29-
8. IDENTIFICATIONOFFUNDAMENTALREQUIREMENTSFORIDEALMETAMODELINGFRAMEWORKINADDITIVEMANUFACTURING
HosseinMokhtarian,EricCoatanéa,HenriParis,JormaVihinen,MohammadHosseiniHossein.mokhtarian@tut.fi
ABSTRACT
Although Additive Manufacturing (AM) has existed for several decades and its industrialapplicationsare increasing, itstillsuffersfromareliablepartqualificationmethod. It ismostlydueto lackofthoroughlyunderstandingthebehaviorofAMTechnologiesandthereasonsforvariabilityamongtheadditivelymanufacturedparts.Advancement incomputationalhardwareand software are continually improving the modeling and simulation capabilities. However,recent advances inmodeling and simulation techniqueswerenot sufficient yet topredict theparts’ properties and to certify the manufactured parts. This refers to highly complexinterconnections between the different physical phenomena in each AM technology, hencebetweendevelopedmodels.Inotherwords,thereliabilityandaccuracyofthemodelswiththeirspecific scope, purposes, and constraints are increasing, but the links describing theirinterconnection are missing. Therefore, to link different types of models and hence betterunderstandingAMprocesses,arobustmetamodelingframework is required.Towardthisend,the following paper aims at addressing and extracting the fundamental requirements for anidealmetamodelingframeworkforadditivemanufacturingfromtheliteraturereviewanalysis.
INTRODUCTION
Incomparisontoconventionalprocesses,AMprocessesaremorecomplex,lesspredictableandmoredifficulttounderstandduetotheirmulti-physicsnature[1,2].Inspiteoftheadvancementin computational hardware and software and thereforemodeling and simulation reliability inAM, the adoption of AM models are still impractical for the industry practitioners [1, 3].Presentingasuitablemetamodeling technique isbeneficial to link themodelsandreplace theexpensive simulation models. According to Yang et al., although the importance ofmetamodeling in AM iswidelymentioned andwell understood in the literature, only limitednumberofresearchhasbeenarticulatedaroundit[1].Tomentionafewofresearchesfocusingondevelopingmetamodels inAM,ontology-basedmetamodeling ispresentedbyWitherell etal., and Roh et al. in separate research papers [3,4]. Yang et al. focused their research onPolynomial Regression and Kriging method on constructing metamodels [1]. Toward thisdirection, the prior publications of the authors focus ondeveloping amathematicalmodelingframeworktoestablishthesimulablecasualmodelbetweenthevariables[5-7].Theintegratedmodel is then achieved by linking the causal models. The existence of different approachesmotivatetheauthorstohighlightsomeofthemostimportantrequirements(butnotlimitedto)thatasuccessfulAMmodelingframeworkshouldfulfill,infollowingshortarticle.
FUNDAMENTALREQUIREMENTS
ForovercomingthechallengesandcomplexitiesinthemodelingofAMprocessesandthereforebeing able to certify the additivelymanufactured parts, a robustmetamodeling framework isrequired [1,3]. Obviously, the idealmetamodeling framework should be tailored for AM. Thefollowing list represents the minimum set of requirements (not limited to) for an ideal AMmetamodelingframework:
IDENTIFICATIONOFFUNDAMENTALREQUIREMENTSFORIDEALMETAMODELINGFRAMEWORKINADDITIVEMANUFACTURING
-30-
Requirement1:ApplicabletoallAMtechnologywithasystematic,repeatableapproach
The framework should ideally represent a systematic, repeatable approach tomodel any AMtechnologyandtheirassociatedfunctionsandphenomena.Itenablescomparativestudyamongdifferent AM technologies, AMmachines, and different parameter settings and enhances thedecision-makingprocess.
Requirement2:Multi-Domain,Multi-ScaleModeling
Each AM process involves withmulti-physics phenomena taking place whilemanufacturing apart. Therefore, the framework should be capable of developing models interacting withdifferent energy domains, and linking models with different level of details. Toward thisdirection, Williams et al. presented a functional classification framework for the conceptualdesignofAMtechnologies[8].DimensionalAnalysisConceptualModeling(DACM)Frameworkisanothereffortformulti-domain,multiscalemodeling[5,6].
Requirement3:DataandDifferentModelIntegration
Themodelingprocedureshouldprovidethepossibilitytolinkdifferenttypeofmodelsanddata.An effective predicting model might be the combination of experimental models and theexistingvalidatedtheoreticalmodels.Towardthisdirection,Shawetal.proposedanapproachto decompose the AM processes for representing process-related data [9]. Ontology-basedmetamodeling is another approach suggested by researchers for data reconciliation [4].Moreover, another level of model integration is refereeing to integrating the part’s modeldescribingitsgeometricalcharacteristicsintoAMprocessmodel.
Requirement4:EnhancementandEnablingMachineLearningorDataMiningTechniques
TheInter-laboratoryroundrobinstudyamongdemonstratedthatAMprocessesarefacingwithhugevariabilityinmechanicalpropertiesofthepartsmanufacturedwiththeparametersettings[10].ThevariabilityinmechanicalpropertiesnotonlyexistedbetweendifferentlaboratoryusingsameparametersettingsandsameAMprocessbutalsobetweendifferentpartsmanufacturedbyeachlaboratory.Itstronglyhighlightstheimportanceofusingmachinelearningtechniquestoconstantly adapt and converge the developedmodel to the desired AM process that we aremodeling.Furthermore,themachinelearningtechniquescanenabletheautomaticextractionofunknownrelationshipsbetweenthevariables.
Requirement 5: Qualitative and Quantitative Simulation, Multi-criteria Simulation andContradictionDetection
Enhancing the quality of the additively manufactured parts with less trial and error and lessexperimental costs requires that the framework support the simulation of the processmodelqualitativelyandquantitatively.Anotherimportantaspectistheabilityofdetectionofsystem’sweaknesses and contradiction while simulating the model, which systematically leads tounderstanding where the innovation or optimization is required. Toward this ability, DACMFrameworkperfectlyprovidesthemathematicalbasisandtheintegrationofvalidatedtheoriesandmethod.[5-7]
IDENTIFICATIONOFFUNDAMENTALREQUIREMENTSFORIDEALMETAMODELINGFRAMEWORKINADDITIVEMANUFACTURING
-31-
CONCLUSIONANDPERSPECTIVE
In this paper, some of the most important requirements that a successful AM modelingframework should fulfill were highlighted. Concerning the mentioned fundamentalrequirements,theauthorsdevelopedtheDACMFrameworkandtailoreditforAM.AuthorsarecurrentlyattemptingtointegrateanovelKnowledge-basedArtificialNeuralNetwork(K-ANN)tothe framework. This knowledge-based ANN is not only enabling themachine learning in theframeworkbutalsoboosttheconvergenceofmodelincomparisontotheconventionalANN.
REFERENCES
[1] Yang, Zhuo, et al. "Investigating predictive metamodeling for additive manufacturing."Proceedings of the ASME 2016 International Design Engineering Technical Conferences &ComputersandInformationinEngineeringConference.2016.[2]Kim,DuckBong,etal."Streamliningtheadditivemanufacturingdigitalspectrum:Asystemsapproach."Additivemanufacturing5(2015):20-30.[3] Witherell, Paul, et al. "Toward metamodels for composable and reusable additivemanufacturingprocessmodels."JournalofManufacturingScienceandEngineering136.6(2014)[4] Roh, Byeong-Min, et al. "Ontology-Based Laser and ThermalMetamodels forMetal-BasedAdditiveManufacturing." ASME 2016 International Design Engineering Technical Conferencesand Computers and Information in Engineering Conference. American Society of MechanicalEngineers,2016.[5] Mokhtarian, Hossein, et al. "A network based modelling approach using the dimensionalanalysis conceptual modeling (DACM) framework for additive manufacturing technologies."ASME 2016 International Design Engineering Technical Conferences and Computers andInformationinEngineeringConference.2016.[6]Coatanéa,Eric,etal."Aconceptualmodelingandsimulationframeworkforsystemdesign."ComputinginScience&Engineering18.4(2016):42-52.[7] Mokhtarian, Hossein, Eric Coatanéa, and Henri Paris. "Function modeling combined withphysics-based reasoning for assessing design options and supporting innovative ideation." AIEDAM31.4(2017):476-500.[8]Williams,ChristopherB., FarrokhMistree, andDavidW.Rosen. "A functional classificationframework for the conceptual design of additive manufacturing technologies." Journal ofMechanicalDesign133.12(2011).[9] Feng, ShawC., et al. "FundamentalRequirements forDataRepresentations in Laser-basedPowder Bed Fusion." ASME 2015 International Manufacturing Science and EngineeringConference.AmericanSocietyofMechanicalEngineers,2015.[10] Brown, Christopher U., et al. "Interlaboratory Study for Nickel Alloy 625Made by LaserPowder Bed Fusion to Quantify Mechanical Property Variability." Journal of MaterialsEngineeringandPerformance25.8(2016):3390-3397.
INDUSTRIALIZATIONOFHYBRIDANDADDITIVEMANUFACTURINGOFMETALS-CHALLENGESINIMPLEMENTATION
-32-
9. INDUSTRIALIZATION OF HYBRID AND ADDITIVEMANUFACTURING OFMETALS -CHALLENGESINIMPLEMENTATION
VeliKujanpää,JormaVihinen*,TuomasRiipinen,PasiPuukkoVTTTechnicalResearchCentreofFinlandLtd.*TampereUniversityofTechnologyveli.kujanpaa@vtt.fi
ABSTRACT
Both additive and hybrid manufacturing open many new possibilities and perspectives tomanufacturing but until now the systems have been stand-alone technologies without realindustrialimplementationonworkshopfloor.Theresearchisconcentratedmainlyindesignandprocess optimization, and separate part printing. Therefore, automatic and flexible industrialproductionhasnotbeenpossiblewithcurrenttechnologies.Mostoftheprocessstepsincludemanualworkanddevelopmentworkisneededespeciallyinpre-andpost-processingoftheAMmachines.Sometrialshavebeenmadee.g.onpowderhandlingandpartremovalfromsystem,but real solutionsarenot seen inmarketsyet.Alsopost-processes, i.e. stress reliefandotherpost-heat treatments, as well as machining steps, grinding, polishing etc. need solutions. Inadditionqualityinspectionshouldbeincludedinthetotalsystem.
Inthepresentationthesystemsinthemarketareanalyzedgivingthepossibilitiesfordifferentand the current barriers hindering the controlled automatic production are presented to giveperspectivetothefuturework.
INTRODUCTION
ThereareafewkeyfactorswhytheautomaticandcontrolsystemsaregoingtobecomemoreimportantintheintegrationandautomationofAMprocesses
• IntegrationofAMtechnologyintofactoryfloor• Demandforsmallbatchproduction• Highvariabilitybetweenproducts• Demandforensuringconsistentqualityforfinishedparts• HowtolinkAMmachinestotheexistingmachinery/workflow?• CapabilityofAMtobeintegrated?
Theimportantquestionsarealso:
• Whataretheprocessesthatshouldbeautomated?• Whatisthelevelofautomationthatcanberealisticallyachievedinanearfuture?
PROCESSPHASES
In metal additive manufacturing there are different process phases that can be separatelyanalyzed. These arepresented in Figure1; design,material handling,AMpartmanufacturing,separatepostprocessingandqualitycontrol.
INDUSTRIALIZATIONOFHYBRIDANDADDITIVEMANUFACTURINGOFMETALS-CHALLENGESINIMPLEMENTATION
-33-
Fig.1.Processphasesofadditivemanufacturing(PowderbedfusionandDirectedenergy
deposition).
Designphase
In the design phase a 3D model of the component is created and modified. In some cases modifications to the original 3D model are necessary to improve the printability of the part. The orientation of the part is chosen and the support structures are created based on the geometry and design criteria of the part. The level of automation at this phase is low and user input is required especially for new designs. The design phase can be streamlined for parts with little complexity using best practices in AM design. Critical process parameters include layer thickness, energy input, scanning strategy, platform heating, etc. They affect the quality of printed parts causing issues as high porosity, high surface roughness, thermal distortion and failed prints. Currently the existing solution to these issues is to use parameters from machine manufacturer or parameters that have been obtained experimentally. The challenge in industrial production is the matter of ensuring parameters that function properly between different powder batches, machines and part geometries Powderhandling
The necessary steps for powder processing and handling depend on the type and properties of the powder. The quality of the powder material is typically ensured by the supplier who processes the powder to meet the requirements for chemical composition, particle size distribution, morphology, humidity and flow properties. These properties
INDUSTRIALIZATIONOFHYBRIDANDADDITIVEMANUFACTURINGOFMETALS-CHALLENGESINIMPLEMENTATION
-34-
affect the behavior of the powder and are controlled to ensure good quality. If the powder is ordered directly from manufacturer, the customer could be responsible for the processing and inspection of the powder. Information should be collected from all of the process steps to ensure comparability between different powder batches. Challenges in automation lie in handling the procurement and quality control. Some commercial machines have powder-sieving stations with semi-automatic functionality. Powder removal from the build chamber is automated in some SLM machines that have a modular design, but in practice this step is done manually. AMPartmanufacturing
The build process itself is usually automated, but many of the preparations before and after printing are manual. Build chamber must be cleaned before the next print job. All of the information required for the printing can be transferred between the design phase and the machines. The major challenge in the build process phase is quality monitoring for detecting defects and machine-related problems during the build process. The process parameters, build chamber conditions and powder properties all affect the build process and the quality of printed parts. For example improper process parameters can result in thermal distortions during the build causing re-coater damage and a failed print. The proper function of the machines are ensured by performing routine maintenances. There are also a number of commercial quality monitoring systems (powder bed, melt pool, laser, etc.) under development for the AM machines. The existing monitoring solutions are passive systems, but there is a growing demand for systems that adaptively monitor the build process and make changes to the process parameters when needed. Postprocessing
Post-processing is probably the most challenging part of automation for AM. The technique for removing parts from the platform depend on the print layout geometry, platform size, surface quality etc. The parts are usually removed after stress relief heat treatment. Other heat treatments are also very important steps, e.g. different annealing and hardening processes. In most cases the product must also be machined after the printing and annealing steps. The dimensional accuracy of the printed parts compared to the CAD model depend on material, process parameters and post treatments. The challenge is to predict the microstructure and mechanical properties. In addition, the more complex geometries pose challenges for machining. Currently parts and support structures are removed manually. Conventional heat treatment procedures and machining steps are used. In the future the post processing procedures must be adapted to match the high variability of small batch production. Qualityassurance
All manufactured parts should meet the required quality standards. A key challenge in adopting AM for industrial use is ensuring consistent quality for finished products. AM machines that are used for large scale printing typically use only one material, which makes quality control measures more straightforward to implement compared to small batch production, where the material is changed frequently. The quality aspects are important in all of the process phases.
INDUSTRIALIZATIONOFHYBRIDANDADDITIVEMANUFACTURINGOFMETALS-CHALLENGESINIMPLEMENTATION
-35-
Hybridmanufacturing
Hybridmanufacturingarecalledthesystemswhereadditiveandsubtractivemanufacturingarecombinedinthesamemachine.Theyarebasedoneitherpowderbedfusionordirectedenergydepositionsystemsmakingthemverydifferentinpractice.
Themostusualsystemsinthemarketare
• Directedenergydepositioncombinedwith5-axismilling(Fig.2)• Directedenergydepositioncombinedwithturnmill(Fig.2)• Powderbedfusioncombinedwith3-axismilling• Coldmetalspraycombinedwith5-axismilling
The advantages of theDED-hybrid process over other solutions are that it can be utilized forrepairingpartsorforaddingadditionalfeaturestocomponentsusingrelativelyhighbuildrates,high geometrical flexibility and the ability to use multiple materials in one build. It can alsoproduce high dimensional accuracy of milled surfaces. A disadvantage is that cutting fluidscannotbeused,limitingthemillingcapabilitiesofthemachines.
InPBFandmillinghybrid the lasermeltingprocess is repreated for several layers followedbyhigh speedmilling of selected contours. The advantages are a high dimensional accuracy andlowsurfaceroughnessandthattheprocessmakesmillingofinternalchannelspossible.Possiblythe entire surface area of the part can be milled, which can be difficult or impossible forcomplexpartsifdoneafterprocessing.Thedisadvantagesarelongbuildtimesperpartandthatcuttingchipresiduethatisleftinthebuildchamber(howeversmallinsize).Manufacturedpartsmightrequireheattreatmentsdependingonthematerial.
In the hybrid systems the additive process as well as the machining steps are quite wellautomatedbutinasimilarmannertothestandalonePBFandDEDsystems,thedesign,powderhandlingandpost-processing,especiallyheattreatmentisnotfullyautomated.Therefore,theseaspectsrequiredevelopment..OneimportantquestionishowtoplantheheattreatmentsforHybrid manufacturing processes.Namely, the heat treatment is not included in the hybridsystem itself. The systems have also quite many limitations, which are not informed by thesystemmanufacturers. Therefore, it looksmore or less so thatmany of the systems are bestsuitedforcertainproducttypes.Thismustbeoneofthemajordevelopmentsinthefuture.
Fig.2.Directedenergydeposition/Lasercladdingcombinedwith5-axismilling/Turnmill(Mazak).
[1] [2]
INDUSTRIALIZATIONOFHYBRIDANDADDITIVEMANUFACTURINGOFMETALS-CHALLENGESINIMPLEMENTATION
-36-
Controlsystems
Controlsystemisakeyissueonthefactoryautomation.MostoftheAMandhybridsystemsinthe market are stand-alone machines and are not ready for automated factories. Manydevelopments are goingonand in thenear future the systemsbetter suitable for automatedcontrolsystemswilllikelybeavailable.
CONCLUSIONS
Additive and hybridmanufacturing systems are currentlymore or less stand-alonemachines,whichdonothavethereadinessforautomaticproduction.Thebuildingprocessitselfisusuallyautomated, but pre- and post-processes are not. Design phase, material handling, AM partmanufacturing, separatepostprocessing andquality control needa lotof development tobereadytobecombinedinafullyautomatedfactoryconcept.
REFERENCES
[1] Flynn et al (2016). International Journal of Machine Tools and Manufacture, 101, 79-101. [2] Yamazaki, T. (2016). Procedia CIRP, 42, 81-86.
IDENTIFYING3D-PRINTABLESPAREPARTSFORADIGITALIZEDSUPPLYCHAIN
-37-
10. IDENTIFYING3D-PRINTABLESPAREPARTSFORADIGITALIZEDSUPPLYCHAIN
JoniReijonenVTTTechnicalResearchCentreofFinlandjoni.reijonen@vtt.fi
ABSTRACT
InOEM’sproducingcapitalgoods,aftersalesservicesisamajorpartofthebusiness.Managingsparepartscost-efficiently,whileprovidinghighservice level for thecustomers isachallenge.Digitalsparepartsproject(DIVA)studiesadigitalsupplychain,wheresparepartsare3Dprintedondemandclosetotheenduser.Thiscouldsimultaneouslyreducecostsand increaseservicelevel. The first challenge is how to identify parts suitable for 3D printing from the tens orhundredsofthousandsofsparepartsthatcompaniesmanage,asnotallaretechnologicallyoreconomicallyviable to3Dprint. In this study,a similarmethodasdevelopedbyKnofiusetal.(2016) isusedto identifysuchspareparts intwoOEMsproducingcapitalgoods.Basedonthecasestudies,itisestimatedthatupto20%ofthesparepartscouldbe3Dprintableintheory.This is reduced to around 6 % having good technological potential to be 3D printed.Furthermore,around2%wouldalsobeeconomicallyfeasibleto3Dprinttoday.Toimprovetheaccuracyandautomatetheidentificationprocess,morestructuredsparepartdataisneeded.
INTRODUCTION
In OEMs producing capital goodswith lifecycles spanning several decades, aftersales servicescouldaccountover50%oftheirrevenue.Thetraditionalsupplychainiscompromisingbetweenalargeinventoryformaximumservicelevelversussmallinventoryforminimumcosts.(Deloitte,2013).Adigitalsupplychain,wherepartsaremanufacturedondemandvia3Dprintingclosetothe end user, has the potential to disrupt the aftersales service business by simultaneouslyreducing costswhile increasing service level.Digital spareparts couldhelp reduce lead-times,costsrelatedtowarehousingandlogistics,manufacturingcostsandsecurethesupplyoflegacyparts. (Holmströmet al., 2010). Thegoal of this study is to identify thoseparts that couldbedigitalspareparts.Assessingtheeconomicbenefitthatcouldbegainedvia3Dprintingislimitedbyreadilyavailabledata.Asforidentifyingsparepartsthatevencanbe3Dprinted,thelimitingfactor is also theavailabilityand structureof relevant sparepartdata.Themost fundamentalclassificationattributesarematerialandsize,asonlyrelativelysmallpartsconsistingofasinglematerial,metallic orplastic, canbe3Dprintedwith today’s technology.However, theexplicitpartinformationsuchasmaterial,geometryandtolerancesarestoredindifferentfilesrelatedtothesparepart.Suchscatteredandnon-unifieddataisnoteasilyextractedfromadatabase.Therefore, fora first stage identificationof sparepartswithhighest3Dprintingpotential, themethodologyhastobeadaptedforthereadilyavailable,companyspecificsparepartattributesthatarestoredinastructureddatabase.
Therehasnotbeenmanystudiesabouthowtoidentifyallthepotential3Dprintablepartsfroma large spare part assortment and how much such parts exists today. Knofius et al. (2016)proposedandtestedsuchamethodologyforonecompanyworkinginthefieldofaviationwitha selected sparepart assortmentof 40330. In the studied company, 15.3%of the scopewasidentified as technologically possible to 3D print and further, 2.8 % were consideredeconomicallyviablebusinesscases.Theaimofthisstudyistoapplysimilarmethodtoidentify3Dprintablesparepartsoftwoadditionalcompaniesproducingdifferentcapitalgoods.
IDENTIFYING3D-PRINTABLESPAREPARTSFORADIGITALIZEDSUPPLYCHAIN
-38-
METHODS
Spare parts of two companies, referred here as Company A and Company B, were classifiedbasedonthe3Dprintingpotential.Thetechnologicalclassificationattributeswerematerialandsize.Thematerialinformationwasindirectinformationderivedfromcompanyspecificmaterialgroupingandcommoditycodes.Asforsize,weightanddimensionswereusedwhenapplicable.Itwasrecognizedthatafterthisclassification,thepotentialgroupstillcontainedpartsthatareimpossibleortechnologicallynotfeasibleto3Dprint.Toexcludethese,imageswereextractedfromthecompanies’databasesforthosesparepartsthathadbeenphotographed(samplesize30%forCompanyAand15%forCompanyB)andavisualclassificationwasdone.Theseresultswere extrapolatedwith the assumption that partswithout imageswould classify in the samemannerasthosewithanimage.Theeconomicclassificationof3Dprintingpartsthatpassedthetechnological classification was done based on a ranking matrix, shown in table 1. First, thesparepartsweredivided intometals andplastics and further into stock andnon-stock items.Spare partswithout specified resupply lead-timewere classified into a separate group, calledspecial.Basedonthematerialandtowhichof thethreecategoriesthesparepartbelongsto,therankinglimitsandrelatedattributesweredefined.
The economic classification attributes were price, resupply lead-time, demand rate andminimumorderquantity (MOQ).Highprice,resupply lead-timeandMOQor lowdemandrateindicates increased 3D printing potential. Resupply lead-time is only considered relevant fornon-stock items. Itwasassumedthatstock itemscouldnotyetbemadeintodigitalnon-stockitemswith the current status of 3D printing technology. Tomake stock items into non-stockitems (making the resupply lead-time a critical factor), would drastically reduce warehousingcosts,butwouldrequirethat3Dprintingondemandcouldmeettheshort(fromhourstofewdays)responsetimesspecifiedinservicelevelagreements.Therankinglimitsareshowforranks1-3,whichwereconsideredeconomicallyfeasible.Partsnotfulfillinganyofthesecriteriawereconsideredunfeasible.Thecompleteprocesscanbesummarizedtofoursteps:definethescopeandclassificationattributes,excludesparepartsthatcannotbe3Dprinted,excludesparepartsthat arenot technologically feasible to3Dprint and rank the remaining sparepartsbasedontheireconomicfeasibility.
Table1.Economicfeasibilityrankingmatrix
Non-stock Stock Special Material MOQ*Price
[€]Resupply lead-time[day]
Price[€] MOQ /demand+1
MOQ *Price[€]
Rank
Metal >1000 >1000 >500 1500-1000 >14 >50 >100 60-100 >10 >100 100-500 30-60 500-1000 100-500 2>10 >30 100-500 1-10 10-100 10-50 500-1000 <14 100-500 nohelp 10-100 3>10 60-100 10-100 1-10 100-500 10-100 30-60
IDENTIFYING3D-PRINTABLESPAREPARTSFORADIGITALIZEDSUPPLYCHAIN
-39-
Plastic >500 >500 >100 1100-500 >14 >50 >10 60-100 100-500 >1 >1 >100 100-500 <14 100-500 10-100 210-100 30-60 10-100 1-10 1-10 10-50 10-100 14-30 10-100 nohelp 1-10 3>1 60-100 1-10 1-10
RESULTSANDDISCUSSION
Figure1illustratestheresultsfromthetwocasestudies.Atstep1,thescopewasdefinedandrelateddataretrievedfromthecompanies’databases.Steps2and3areinterlinked,assomeofthe technologically not feasible spare parts were already identified and excluded at step 2,basedon sparepartattributes. In the samemanner, somespareparts that shouldhavebeenexcludedasnotprintablealreadyinstep2wereonlyidentifiedatstep3.Largestofsuchgroupswasmechanicalparts consistingofmultiplematerials.At step4,parts that received rank1-3,wereconsideredeconomicallyfeasible.Step4wasnotconductedwithCompanyB.
Figure1.ResultsfromthecasestudieswithcompanyAandB.
Thelargedifferencebetweenstep2and3comesfromexcludingspareparts,whichcouldbe3Dprinted in theory,but itwouldnotmake sense toproduce thesewith3Dprinting.Thisgroupconsistsmostlyofbulk items (forexamplebolts,nutsetc.)andsimplegeometrypartssuchassheetmetalparts.Inthisanalysis,thesewereexcludedinstep3astechnologicallynotfeasibleand therefore the economic feasibility was not further assessed for these in step 4.Counterintuitively, this group may contain some parts, which would actually make a viablebusinesscase,ifexaminedinmoredetail.Themostcommonfactorinthesparepartsidentifiedastechnologicallyfeasibleisthattheyarecomplexgeometryparts,typicallycast/moldedtoday.Thisindicatesfurthereconomicbenefitfor3Dprinting,asexpensivemoldsrelatedtothesparepartscouldbeeliminated.Ontheotherhand,thisstudydoesnotconsiderthepossibleneedofpost-processingorother finishingafter3Dprinting tomeet tolerancesorother specifiedpartproperties. This factor was not considered, as this information for the spare parts was notreadilyavailable.Itisbelievedthatifconsidered,thiswouldrendersomepartsunfeasible.Thesecase studies and the one conducted by Knofius et al. (2016) indicate that regardless of thecompany,similarresultscanbeexpected.
CONCLUSIONS
Basedonthecasestudiesincompaniesproducingcapitalgoods,itisestimatedthatuptoover20%of the spareparts couldbe3Dprintable in theory. This is reduced toaround6%when
IDENTIFYING3D-PRINTABLESPAREPARTSFORADIGITALIZEDSUPPLYCHAIN
-40-
partswithextremelysimpleorstandardizedgeometriesareexcluded.Furthermore,around2%wouldalsobeeconomicallyfeasibleto3Dprinttoday.To improvetheaccuracyandautomatetheidentificationprocess,morestructuredsparepartdataisrequired.
REFERENCES
Deloitte. 2013. Driving Aftermarket Value: Upgrade Spare Parts Supply Chain. Deloitte ChinaAutoIndustrySparePartsManagementBenchmarkSurveyWhitePaper.23p.Holmström,J.,Partanen,J.,Tuomi,J.&Walter,M.2010.Rapidmanufacturinginthesparepartssupply chain: Alternative approaches to capacity deployment. Journal of ManufacturingTechnologyManagement,21(6),pp.687-697.Knofius,N.,VanderHeijden,M.&Zijm,W.2016.Selectingpartsforadditivemanufacturinginservicelogistics.JournalofManufacturingTechnologyManagement,27(7),pp.915
IIISMARTMACHINESANDROBOTICS
AWARENESSINHUMAN-ROBOTINTERACTIONFORCOLLABORATIVETASKS
-42-
11. AWARENESSINHUMAN-ROBOTINTERACTIONFORCOLLABORATIVETASKS
AlexandreAngleraud,RoelPietersTampereUniversityofTechnology,LaboratoryofAutomationandHydraulicsalexandre.angleraud@tut.fi
ABSTRACT
Ahumananda robotworking inpairproducesabetterand faster result than if theyworkedindividually[1].Thiscollaborationisparticularlyadaptedtoexplicittaskswhichmakesitagreatasset formanufacturing lines.However, these interactionscomewithchallengessuchasstricttimeconstraintsinplanning,advancedperceptionabilities,andofcourse,anabsoluteneedforsafety.Morepractically, thismeans thathumanand robotneed tobeawareofeachother inorder to make the collaboration possible. To investigate these novelties in human-robotinteraction, we propose to build a demonstrator composed of one or two robotic armsreproducinganindustrialcontextwithacollaborativetasktoperform.Thisdemonstratorwillbeequippedwith different sensors allowing it to be conscious of its environment and a displaysystem that will enable the communication with a human agent. By developing a cognitivesystemon top of this systemwewant to showcase the different improvements that it couldbringuponthemanufacturingworld.
CONCEPT
Althoughrobotsarealreadypresentinourindustries,theiruseislimitedtoarestrictednumberoftasksanditisstillhardtochangeandreprogramthemforanothersetofactions.Moreover,in certain cases, humans just tend to performbetter than robots at a specific task. For thesereasons, research about cobots, i.e., robots meant for interaction and collaboration withhumans,isanongoingtopicallaroundtheworld.Suchsystemshavetheopportunitytomixthebestfromtherobotsidebutalsofromthehumansideandinthissense,cobotscouldbeawaytorethinkourindustries,aspartoftheindustry4.0movement[2].
However, two keys for this change to happen are the cost and the ease of use of the finalsystem.We can already take lessons from other ongoing projects such as the two robots ofRethinkRoboticsintheUSA,BaxterandSawyer[3].Thefirstone,Baxter,isatworoboticarmssolutionforthemanufacturinglinesfeaturingforcesensingand7DoFforeacharmaswellasasafebehaviornearhumans.AsforSawyeritoffersthesameflexibilitybutiscomposedofonlyone arm having a longer reach and taking less space than Baxter. They both use a softwarecalled Interawhichconsists ina train-by-demonstration interface for theeaseofdeployment.WecanalsothinkabouttheGermanprojectiMRKconductedincollaborationwithVolkswagen[4]which resulted in a dual armmanipulation platform that could be controlledwith humangestures. This shows that it is possible to produce robots for collaborationwith humans. Thenextchallenge isnowtomakethesoftwareevolve inawaythatmakesrobotsabletohandlemoretaskswhilekeepinginmindtheimportanceofhavinganintuitiveinteraction.
Onamoretechnical level,previousworkswithreasoningthroughontologiesandcollaborativerobotics include [5] in which a framework has been developed to tackle the human activityrecognitionchallengesthatwecouldneedintheprocessofteachingnewtaskstothecobot.In[6], human activity recognition is also addressedwith a focus onhow tohandle thedifferentstylesthatcouldbeencounteredamonghumanswhendoingthesamekindofaction.Thisworkbrings the idea of really understanding what is being demonstrated. More recently, in [7]
AWARENESSINHUMAN-ROBOTINTERACTIONFORCOLLABORATIVETASKS
-43-
learning isconsideredfromtheperspectiveofmemoryskillstotake intoaccountthepreviousexecutionsofatask.Ourideasaresimilarbutaimatdevelopingthepotentialinteractionsinthereasoning processwhereas the describedworkswere investigating automated solutions fromartificial intelligence.Of course, oneof our optionwould be to build an hybrid system takingadvantageofbothsides.
The object of our research is the construction of a demonstrator to investigate collaborativetasks between humans and robots through various interactions with a certain degree ofawareness from the robot. Thechallengesaddressedhereare the following. Firstwewant todevelop the existing solutions for the perception interface that is needed to gather enoughinformationabout theenvironment.Our secondobjective is to rethink the interfacebetweenthe robot and human agents allowing an efficient exchange of information adapted to theindustrialenvironmentinwhichtherobotwillbedeployed.Finally,wewanttohaveasoftwarethat can use ontologies to build a knowledge base about the different components of theenvironmentandbringawarenessinthecollaborationbetweenthehumanandtherobot.Thiswouldallowustotakeintoaccountsthetasktoperformandplananefficientwaytoperformitusingthedifferentagentsavailable.Thissoftwarewouldalsocontainawaytoenablehumanstoteach a task in an intuitive way using the latest technologies such as augmented reality forexample.
A transversal goal in thiswork concerns the safety issues that occur during the collaborationbetweenahumananda robot.Recently, standardshavebeenestablished andwe thushaveprecisecriterias toqualify the robustnessofour system[8].Asanexample, the ISO/TS15066specifiesthefollowing:“safetyrequirementsforcollaborativeindustrialrobotsystemsandtheworkenvironment,andsupplementstherequirementsandguidanceoncollaborativeindustrialrobotoperation”. Furthermore, toensure theoperationalityof ourproduct, testing is also animportantpartforusandourfuturepartners.In[9]isdescribedanapproachforthis,toallowmoretransparencyinthetestingprocess.
TovalidateourresearchwewillmainlyusetheFrankaEmikaarmandits2-fingersgrippereventhoughusingasecondarmtoallowmorepotentialapplicationsisalsoconsidered.Inafirsttimetheproductwillbedevelopedincollaborationwithalimitedamountofindustrialpartnersbutenlargingthisnumberwouldbeagoalforthenextpartofourresearchplan.
Whenworkingwith robotseveryday, it iseasy to forget“'Whatcan Idowith this robot?''asexplainedin[10]byanengineeroftheRobotiqcompany.However,hereishowheclassifiesthedifferent applications of collaborative robotics. First would be machine tending, secondpackagingandfinallymaterialhandling.Ifthesetermscanberepresentativeofalotofdifferenttasks, the precise use cases and how to adapt the system in the best way according to itspurposeissomethingthatwewilldiscusswithourindustrialpartners.
GOALS
Bybuildingthisdemonstratorourmaingoalsareasfollows.First,tomakeprogressinwhatwebelievecould increase theperformancesof themanufacturingenvironment. Second, toboostthecooperationbetweenacademiaandindustry,especiallytowardsSMEs.Asaresult,byhavingwaystoproducefasterandbetter,companieshavingpartoftheiractivitiesinothercountriestoreduce the productions costs could be able to come back to a local production. The possible
AWARENESSINHUMAN-ROBOTINTERACTIONFORCOLLABORATIVETASKS
-44-
applications are wide, as the industry presents an important amount of tasks that could beaffectedbyhuman-robotcollaboration.
REFERENCES
[1] Frank Tobe (30 Dec 2015).Why Co-Bots will be a huge innovation and growth driver forrobotics industry. Retrieved from https://spectrum.ieee.org/automaton/robotics/industrial-robots/collaborative-robots-innovation-growth-driver[2] Cencen, Argun & Verlinden, Jouke & Geraedts, Jo. (2016). Qualifying the performance ofhuman-robotcoproductionatarelablingstation.[3]Duhameletal.“RethinkRobotics-FindingaMarket”StanfordCasePublisher204-2013-1.20May2013.[4] José de Gea Fernández, Dennis Mronga, Martin Günther, Malte Wirkus, Martin Schröer,StefanStiene,ElsaKirchner,VinzenzBargsten,TimoBänziger,JohannesTeiwes,ThomasKrüger,Frank Kirchner (2017). iMRK: Demonstrator for Intelligent and Intuitive Human-RobotCollaborationinIndustrialManufacturing[5]KarinneRamirez-Amaro,MichaelBeetzandGordonCheng,Understanding the intentionofhumanactivitiesthroughsemanticperception:observation,understandingandexecutiononahumanoidrobot.AdvancedRobotics29(5),2015,345-362[6] Karinne Ramirez-Amaro, Emmanuel Dean-Leon, and Gordon Cheng, Robust SemanticRepresentations for Inferring Human Co-manipulation Activities even with DifferentDemonstrationStyles.15thIEEE-RASInternationalConferenceonHumanoidRobots,2015.[7] Ferenc Balint-Benczedi, Zoltan-Csaba Marton, Maximilian Durner, Michael Beetz, "StoringandRetrievingPerceptualEpisodicMemoriesforLong-termManipulationTasks",InProceedingsof the 2017 IEEE International Conference on Advanced Robotics (ICAR), Hong-Kong, China,2017.FinalistforBestPaperAward[8]RobertaNelsonSheaandSeungminBaek(11Nov2016).Demystifyingcobotsafety:Thefourtypesofcollaborativeoperation.Retrievedfrom:https://blog.universal-robots.com/demystifying-cobot-safety-the-four-types-of-collaborative-operation[9] Florian Weisshardt, Jannik Kett, Thiago de Freitas Oliveira Araujo, Alexander Bubeck,AlexanderVerl,EnhancingSoftwarePortabilitywithaTestingandEvaluationPlatform,VerbandDeutscher Elektrotechniker e.V. -VDE-, Berlin: ISR/Robotik 2014, Joint Conference of 45thInternational Symposium on Robotics and 8th German Conference on Robotics. Proceedings.CD-ROM:2-3June2014,Munich,Germany.[10] Mathieu Bélanger-Barrette (28 Oct 2015). 3 Best Collaborative Robot Applications.Retrievedfromhttps://blog.robotiq.com/3-best-collaborative-robots-applications
ANOPEN-SOURCE,CONFIGURABLEIOTPLATFORMFORREMOTEMONITORINGOFSMARTMACHINES
-45-
12. ANOPEN-SOURCE,CONFIGURABLEIOTPLATFORMFORREMOTEMONITORINGOFSMARTMACHINES
AnandaChakraborti,OlliUsenius,HenriVainio,JussiAaltonen,KariKoskinenMechatronicsResearchGroup,MechanicalEngineeringandIndustrialSystems,TUTananda.chakraborti@tut.fi
ABSTRACT
IndustrialInternetofThings(IIoT)andCyberPhysicalSystems(CPS)havebroughtforthseveralconnecteddeviceswithweb-baseddatacollectionandmonitoringplatforms,whichcanextractmachine data from sensors and provide valuable insights to the working of smartmachines.Nowadays,useof such IoTplatformsaregettingmorecommon inmobile industrialmachinesand manufacturing systems. However, no open-source, fully configurable IoT platform existswhich can be easily integrated with the control system of any mobile machinery includingmachines with legacy systems to remotelymonitormachine specific details in real-time. Themain contribution of this article is to provide systematic guidelines in integrating variousenabling technologies to develop such a platform,which collect, store andprocess data frommobilemachineries to gain insights on its condition and represent informationwith enduserdevices.
INTRODUCTION
Industrial Internet of Things philosophy has resulted in a rapidly growing ecosystem of opensource IoT platforms for monitoring physical processes in industrial systems and mobilemachineries. The open source nature of these platforms has attracted large communities ofdevelopers,enthusiastandbusinesses tocontributeandenhance featuresof theseplatforms.However, adoption of these platforms by industries are rare because of the complexity inintegrating variousheterogeneous components tohave seamless interoperationbetween realand virtual environments (Joseph Boman, 2014). These heterogeneous components aredifferenthardwareandsoftwareusedindevelopingtheplatform.Itischallengingtodevelopamessagingmechanismtocommunicatebetweentheseheterogeneouscomponentswithcross-platformtechnologiesinautonomousindustrialsystems(ZhaozongMeng,2016).Therestofthearticle is organized as follows. First the enabling technologies are introduced which areleveragedtobuildtheplatform.Thentheplatformanditscapabilitiesarediscussedandfinallytheconclusionisdrawn.
ENABLINGTECHNOLOGIES
IIoT depend on enabling technologies like smart sensors, embedded devices, communicationandconnectivityprotocols,publicandprivatecloudanddataanalytics.SmartsensorshelpinIoTsensingwhichmeans gathering data from the physical environment and sending it to a datawarehouseorcloudwherethisdataisprocessed.Inindustrialmobilemachineries,interchangeof digital information between sensors and electronic control units is done with serialcommunication protocol ISO 11898 (Controller AreaNetwork) (ISO, 2006). Embedded devicesare critical to IIoT platform operation. Sometime these embedded devices are Single BoardComputers (SBCs)withbuilt-inTCP/IPandsecurity functionalitiestogather,process,andsenddata to remote storage (Ala Al-Fuqaha, 2015). Connectivity protocols put the heterogeneouscomponents together. The connectivity protocol can be wired or wireless. Example of wiredconnectivityprotocolisIEEE802.3(Ethernet)(Healey,2017).Examplesofwirelessconnectivity
ANOPEN-SOURCE,CONFIGURABLEIOTPLATFORMFORREMOTEMONITORINGOFSMARTMACHINES
-46-
protocolsare IEEE802.11(WirelessLocalAreaNetwork)(Stephens,2017),BLE(BluetoothLowEnergy), Zigbee, LowPowerWideAreaNetwork like LoRAand SigFoxorCellular technologieslikeGSM,3G,LTEetc.Communicationprotocolsenablesmessagingbetweendifferenthardwareand software components. There are two communication patterns widely used betweencomponents in IoT platforms for real-time data exchange. They are Request-Responsemodel(e.g.HTTP)andPublish-Subscribemodel(e.g.MQTT)(LaurentPellegrino,2013).
Opensourcesoftwarecomponentsenableeasierinteroperabilitybetweentheplatformandthephysicaldevice.NodeJSisanopen-sourcecross-platformJavaScriptruntimeenvironment,whichisapopularplatformforserver-sidedevelopment.NoSQLdatabaseallowsstorageandqueryofunstructureddatafromdifferentdevices.IoTplatformsarecapableofdoingremotemonitoringbecause of data collection strategy from sensors and storing them in cloud repositories orremotedatabases.Useofpubliccloudservicesofferedbypopularcloudserviceproviderscanprovide on-demand pay-per-use data storage, analytics, devices and data managementfunctionalities but at the cost of security. Private clouds on the other hand provide highestdegreeofcontroloverperformance,reliabilityandsecurity.However,privatecloudservicesarebuiltbyorganizationsfortheirownusage(QiZhang,2010).Computationallyintensiveanalyticsaremoreexpensiveonprivatecloud.SecurityisanincreasingconcernforCPS.ModernCPSarepervasively monitored systems capable of networking with each other. These systems areresourceconstrainedandsafetycritical.Hence,thethreatisnotjustfromexternalcyber-attackstryingtocompromiseinformationsecuritybutalsowithinthesystemlikedataintegrity.
THEPLATFORM
Thiscross-platformtechnologybringstogetherheterogeneoushardwareandsoftware.Processdataiscollectedfromin-builtsensorsofthesmartmachinewithCANBusbasedcontroller.ThiscontrolleruseslowlevelfirmwaredevelopedwithCfollowingMISRACguidelines(C,2016)forsafety critical applications. It supports standard TCP/IP protocol to communicatewith remoteserver.GSMdevice is usedas a gateway.GSMhas longest rangeand reasonabledata rate incellular networks suitable for near real-time remote monitoring applications in mobilemachineries. The firmware post data from the device to a single-server running on a publicnetworkinstanceoverHTTP.Theserver-sidedevelopmentisdonewithNodeJS.AnopensourceNoSQL database runs as a service in the server instance. The platform remotely monitorsprocessdata fromthesmartmachineaswellasdisplayhistoricdata fromthesmartmachinesensors. The platform also tracks and displays the location of the machine. The platform iscapable of doing simple analytics such as regression analysis ofmachine data, online in real-time.Figure1showsasnapshotoftheplatformanditscapabilities.
ANOPEN-SOURCE,CONFIGURABLEIOTPLATFORMFORREMOTEMONITORINGOFSMARTMACHINES
-47-
Figure1.TheopensourceIoTplatform
Asstandardopensourcetechnologiesareusedtobuildthisplatform,itisfullycustomizableandconfigurable. It is also easy to add new features to this platform and remove unwantedcomponents.
CONCLUSION
Because of the simplicity in hardware-software integration and communication, this platformcanbeeasilyinstalled,configuredandputtouseinanymobilemachineryorindustrialsystemwithCANBusbasedcontrolsystem.Thisplatformcanbebeneficialforfleetownerswhowanttomonitortheirmachinesaswellasother industrialserviceproviderssuchasmaintenanceorlogisticservices.Smartmachineplatformslikethiscanalsoserveastest-benchforresearchersto study and integrate future technologies such as AI and human-machine interaction withindustrialmachineries.
REFERENCES
AlaAl-Fuqaha,M.G. (2015). InternetofThings:ASurveyonEnablingTechnologies,Protocols,andApplications.IEEECOMMUNICATIONSURVEYS&TUTORIALS.C,M.(2016).MISRACGuideline.UK:HORIBAMIRA.Healey,A.(2017).IEEE802.3ETHERNETWORKINGGROUP.IEEE.ISO.(2006).Roadvehicles—Controllerareanetwork(CAN)—Part3:Low-speed,fault-tolerant,medium-dependentinterface.ISO.JosephBoman,J.T. (2014).Flexible IoTMiddlewarefor IntegrationofThingsandApplications.10thIEEEInternationalConferenceonCollaborativeComputing.Maimi,Florida:IEEE.Karl Koscher, A. C. (2010). Experimental Security Analysis of a Modern Automobile. IEEESymposiumonSecurityandPrivacy.IEEE.Laurent Pellegrino, F. H. (2013). A Distributed Publish/Subscribe System for RDF Data.International Conference on Data Management in Cloud, Grid and P2P Systems (ss. 39-50).Berlin:Springer.Liikanen, E. (2003). Commission recommendation. Retrieved 9 17, 2016, from http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32003H0361&from=ENMark Yampolskiy, P. H. (2012). Systematic analysis of cyber-attacks on CPS-evaluatingapplicabilityofDFD-basedapproach.ResilientControlSystems(ISRCS).SaltLakeCity,USA:IEEE.
ANOPEN-SOURCE,CONFIGURABLEIOTPLATFORMFORREMOTEMONITORINGOFSMARTMACHINES
-48-
PratimRay, P. (2016). A survey of IoT cloud platforms. Future Computing and InformaticsJournal,35-46.Qi Zhang, L. C. (2010). Cloud computing: state-of-the-art and research challenges. Journal ofInternetServiceApplication,7-18.Quintana, V., Rivest, L., Pellerin, R., Venne, F., & Kheddouci, F. (2010). Will Model-basedDefinitionreplaceengineeringdrawingsthroughouttheproduct lifecycle?Aglobalperspectivefromaerospaceindustry.ComputersinIndustry,2010(61),497-508.Ruemler, S. P., Zimmermann, K. E., Hartman, N. W., Hedberg, T. J., & Feeny, A. B. (2016).Promoting Model-based Definition to Establish a Complete Product Definition. Journal ofManufacturingScienceandEngineering,2016.Stephens,A.(2017).IEEE802.11TMWIRELESSLOCALAREANETWORKS.IEEE.Xiao-Hong, L. (2014). Research and Development of Web of Things System Based on RestArchitecture. Intelligent Systems Design and Engineering Applications (ISDEA). Hunan, China:IEEE.ZhaozongMeng, Z.W. (2016).AData-OrientedM2MMessagingMechanism for Industrial IoTApplications.IEEEInternetofThingsJournal,236-246.
SAFETYCONSIDERATIONSINTHEDESIGNANDVALIDATIONOFAUTONOMOUSMACHINESYSTEMS
-49-
13. SAFETY CONSIDERATIONS IN THE DESIGN AND VALIDATION OF AUTONOMOUSMACHINESYSTEMS
EetuHeikkilä,RistoTiusanen,HeliHelaakoskiVTTTechnicalResearchCentreofFinlandLtd,MachineHealtheetu.heikkila@vtt.fi
ABSTRACT
The advances in robotics, artificial intelligence, and communication technology are makingindustrial applications increasingly autonomous. For manufacturing companies this meanstransformation towards digitized enterprise that requires the development of core elements,such as building the data ecosystem, choosing the right techniques and tools, integratingtechnology into workplace processes, and adopting an open, collaborative culture. The rapidintroductionofnewtechnologies,combinedwithincreasinghuman-machineinteraction,bringsalongnewandmodifiedsafety risks.Formost industrialmachinesystems, the lackofexistingstandardsandbestpracticesfurthercomplicatesthedesignandvalidationoftheseapplications.Inthispaper,weidentifythefocalchallengesinthedesignprocessofsafeautonomousmachinesystemsfromasystemsengineeringperspective.Asaresult,themethodologyandtoolstobeused in various phases of the design process are discussed. The needs for extensions to thesystemsengineeringapproacharealsoconsidered.
INTRODUCTION
Increasing autonomy is a global trend in industry, with industries like transportation andmanufacturing leading the way in the change. Autonomous systems are seen as means toincrease productivity, cost efficiency and safety – not only by reducing the work done byhumans, but also by enabling completely new business models. The concept of autonomy,however, has no single agreed definition and it is common that the terms “automation” and“autonomy”areusedinterchangeably.Autonomy,however,goesbeyondautomationbyaddingself-governing behavior and requiring intelligent decision-making abilities (NFA, 2012). Theconcept of autonomy also involves operation in complex domains, often in co-operationwithhumans.Autonomyisnotadefinitesystemcharacteristic.Instead,thetransformationtowardsmore autonomous systems is an incremental process and various autonomy levelcategorizations are available in different fields to demonstrate the gradual change. In thesecategorizations, the autonomy levels usually range from fully human operated to fullyautonomous,withseveralsemi-autonomouslevelsinbetween.Inthispaper,wefocusonlyonhigh levels of autonomy, where the system operates autonomously with possible humansupervision, but without the need for human interference in the majority of operationalscenarios. Additionally, we focus on the safety risks in the intended performance of thesesystems,whilecybersecuritythreatsareexcludedinthiscontext.
Increasing autonomy is not a limited to increasing the autonomy levels of individual devices.Instead,inmanyindustries,itcancausedisruptionintheentirebusinessbyenablingtheusageof novel approaches, such asby replacingheavymachinerywith fleets of lighterdevices. Theresulting changes in the ways of working bring along new and modified safety risks. Whilegeneral guidelines for safety considerations exist (e.g. IEEE, 2016), detailed standards andlegislationareonlyslowlybecomingavailableinlimitedapplications,suchasintheautomotiveindustry – but also in limited industrial applications, such as. earth-moving machinery andmining (ISO17757:2017). Instead of the traditional prescriptive nature of standards, however,
SAFETYCONSIDERATIONSINTHEDESIGNANDVALIDATIONOFAUTONOMOUSMACHINESYSTEMS
-50-
thenewstandardsfocusmoreonprovidinggeneralperformanceguidelines (Danks&London,2017). Thus, the technologydevelopersbearan increasing responsibility forensuring the safeoperationofautonomoussystems.Thisfurtherincreasestheimportanceofasystematicdesignandvalidationapproach.
CHALLENGESINDESIGNANDVALIDATION
The challenges in design and validation were studied by following a systems engineering V-model,whichisatypicalmethodtofacilitateatop-downdesignprocessofacomplexsystem.Several versions of the model exist for different purposes, but generally they represent thesystemdevelopmentprocessmovingfromconceptdevelopmenttowardsmoredetaileddesign,implementation, and testing phases. In Figure 1, a simple V-model is presented with focalchallengesbroughtalongbyincreasingautonomydescribedineachstep.
In the design of complex autonomous systems, the focus of activities is likely to concentratetowardstheearlieststagesofthedesignprocess.Organizationsareputtingincreasingefforttounderstandingtherisksinvolvedintheearliestpossiblephase.Thissetsincreasingdemandforhigh-quality system description and requirements management methods. For autonomoussystems, the system definition needs to be broader than in traditional system design. Thedefinition needs to include the relevant information regarding the operating environment,stakeholders involved in theoperation,and thedifferent interfaces throughwhich thesysteminteracts with its environment. The requirements for the autonomous system’s performancewithin all relevant situations need to be documented. For example, a Concept of Operationsapproach(seee.g.Osborneetal.2005)canbeusedtoproduceacomprehensivedescription.
Figure1.AsystemsengineeringVmodel(adoptedfromOsborneetal.,2005),withselectedfocalchallengesfordesignandvalidationofautonomoussystems.
Conceptdesign
- system description &requirementsmanagement
Architecturedesign
- handling of large datamasses
Detaileddesign
- development of safelearningalgorithms
Integration,test,andverification
-testingofthelearningqualities
-transparencyoflearning
Implementation
- teaching the machinelearningsystem
System Verification andValidation
-lackofprescriptivestandards
OperationandMaintenance
-newoperationallogicVerificationandValidation
-demonstrating&communicatingthesafety
SAFETYCONSIDERATIONSINTHEDESIGNANDVALIDATIONOFAUTONOMOUSMACHINESYSTEMS
-51-
Inthearchitecturedesignphase,thelargestchallengesaresetbythelargedataamountsthatneedtobecreatedandprocessedreliablytoenableautonomousoperation.This involvesalsotheplanningofhowthedataisusedtoteachtheautonomoussystembyusingmachinelearningmethods. The learning capabilities of the autonomous system also need to be consideredthroughout the rest of the development process. This includes validation of the system’slearninginvariousscenarios,whichislikelytoincreasetheneedforadvancedsimulatortesting.The transparencyof themachine learningprinciplesadds furtherchallenges,as thedevelopermaybeunawareofthedetailsoftheactualmachinelearningprocess.Toaddresstheseneeds,certainexpansionshavebeenproposedtotheV-modeltoincorporatetheadaptiveandlearningcharacteristics. For example, Falcini & Lami (2017) propose a “W-model” to account for theeffectsofmachinelearningcapabilitiesinautonomoussystems.IntheW-modelmethodology,aparallel process is introduced to facilitate data-driven training and validation of themachinelearningapplication.
In the verification & validation phase, the lack of prescriptive standards sets increasingrequirementsalsoforthesafetycommunication.Agoal-basedapproachhasbeenproposedasamethod to communicate the relationship between the system requirements and safetyvalidation results (Heikkilä et al., 2017). Finally, in the operation phase, the role of well-facilitated change management procedures is likely to increase when modifications areintroducedintothesystem.
CONCLUSIONS
The increasing autonomy inmachine systems imposes increasing demands on the design andvalidation processes. Focal challenges for design and validation were identified as follows:quality of systemdescription and requirementsmanagement, design& validationof safe andtransparent machine learning, and safety communication with respect to the systemrequirements.
Whileindividualtoolsareavailableformanagingtheseindividualaspects,andextensionshavebeen proposed to the V-model approach, a holistic approach for the design of autonomoussystemshasnotyetbeendeveloped.Developmentofsuchmethodologyisseenasanimportantarea for future research to ensure the safe and sustainable introduction of increasinglyautonomous applications. A comprehensive methodology could also be incorporated into abroaderfutureframeworkfortheassessmentofentirelifecycleofautonomoussystems.
REFERENCES
DanksD.,LondonA.2017.Regulatingautonomoussystems:Beyondstandards. IEEE IntelligentSystems.Vol.32:1.IEEE.Falcini F., Lami G. 2017. Deep Learning in Automotive: Challenges and Opportunities.CommunicationsinComputerandInformationScience,vol770.Springer.IEEE.2016.EthicallyAlignedDesign:AVisionforPrioritizingWellbeingwithArtificialIntelligenceandAutonomousSystems,Version1.IEEE.ISO17757:2017. Earth-moving machinery and mining – Autonomous and semi-autonomousmachinesystemsafety.ISO.HeikkiläE.,etal.2017.SafetyQualificationProcessforanAutonomousShipPrototype–aGoal-based Safety Case Approach. Marine Navigation: Proceedings of the 12th InternationalConferenceonMarineNavigationandSafetyofSeaTransportation(TransNav2017).
SAFETYCONSIDERATIONSINTHEDESIGNANDVALIDATIONOFAUTONOMOUSMACHINESYSTEMS
-52-
NFA Norwegian Society of Automatic Control. 2012.Autonomous Systems: Opportunities andChallengesfortheOil&GasIndustry.NFA.OsborneL.etal.,2005.Clarus:ConceptofOperations,WashingtonD.C.,USA:FederalHighwayAdministration.
SIMULATINGTHEDRAGCOEFFICIENTOFASPHERICALAUTONOMOUSUNDERWATERVEHICLE
-53-
14. SIMULATING THE DRAG COEFFICIENT OF A SPHERICAL AUTONOMOUSUNDERWATERVEHICLE
ArttuHeininen,JussiAaltonen,KariT.KoskinenTampereUniversityofTechnology,MechanicalEngineeringandIndustrialSystemsarttu.heininen@tut.fi
ABSTRACT
In this paper, an AUV pressure drag simulations with different velocities are presented. TheAUV’s main hull is spherical but its instrumental components extend out of the hull, andinterferewiththeflow.Apressuredragcoefficientofaspherewassimulatedtodecidetheusedturbulencemodel(SSTk-ω),bycomparingsimulationresultstothosefoundinliterature.Then,simulationsweredonewithdifferent velocities tounderstand thebehaviourof theAUVdragcoefficient. These results can be used to improve the control system of the AUV and are animportantfactorintheAUVenergyconsumption.
INTRODUCTION
One of themost important design factors of an autonomous underwater vehicle (AUV) is itsshape.Shapeaffectsdirectlytothedragcoefficient.Alongwithotherhydrodynamicderivatives,thedragcoefficientdictatestheAUVmotionandenergyconsumptionwhilemoving.Thelargerthedrag themoreenergy is required fora constantmotion.This is the reasonwhy thereareplethoraoftorpedo-shapedAUVsfoundintheliterature.
Conventionally, the drag coefficient is measured using a tow-tank. At present, the dragcoefficient canbe simulatedusing computational fluiddynamics (CFD). The simulationusuallyrequiresvalidation, soboth themeasurementsandsimulationareusedtogether.However, intheearlydesignstage,simulationprovidesinformationthatisnotavailableotherwiseandthisreducestheamountofrequiredprototypesandmeasurements.
Usually the flow around an AUV is turbulent. Turbulence is often modelled using so-calledturbulencemodels. In this paper, it is shownhow an early stage simulation process could bedonebychoosingaturbulencemodelandthencalculatingthedragcoefficientofanAUV.
CHOOSINGTHETURBULENCEMODEL
Thedragcoefficient,Cd,ofaspherecanbeestimatedusingequation[1,p.624]:
÷øö
çèæ+
÷øö
çèæ
+
÷÷ø
öççè
æ+
÷÷ø
öççè
æ
+
÷øö
çèæ+
÷øö
çèæ
+=-
-
6
6
00.8
94.7
52.1
10Re1
10Re25.0
000263Re1
000263Re411.0
0.5Re1
0.5Re6.2
Re24
dC (1)
whereRe is theReynoldsnumber [1,p.622]. Fora spherewithadiameterof0.6m thedragcoefficientis0.28,withflowvelocityof2km/h,whichistheAUVtargetvelocity.Whentheflowvelocityisfastenough,flowisseparatedfromthesurfaceofthesphere.Accordingto[2,p.92]SST k-ω turbulence model is suitable for separating flows. Using this turbulence model, asimulationwasperformedforthespherewiththesamevelocityasabove,showninFigure4.
SIMULATINGTHEDRAGCOEFFICIENTOFASPHERICALAUTONOMOUSUNDERWATERVEHICLE
-54-
Figure4.Aseparatedflowaroundasphere
Thedrag force is calculated fromthepressure fieldaround thesphere,and thepressure fieldfromthevelocityfield.Thedragcoefficientis:
P
DD
AVF
C2
21
¥
=r
(2)
andfromthesimulationit is0.29,whichisclosetotheonecalculatedearlier. Inthiscase,theSSTk-ωturbulencemodelperformssufficiently.
THEDRAGCOEFFICIENTOFTHESPHERICALAUV
TheAUVhascylindricalinstrumentsoutofthesphericalmainhull.Theseinstrumentsaffectthedragcoefficientbydisturbingtheflow,andareshowninFigure5,whichalsoshowstheCFDgridthat was generated. The grid has 978000 volumes and most of those are near the AUV tocapture the flow field in detail. Using this grid, simulations were run with different flowvelocities.Thedragcoefficientwascalculatedforeachcase.
Figure5.ThegridgeneratedfortheAUV.
Aftertheinitialsecondsofthesimulation,thedragcoefficienttendstowardstoasinglevalue.The protruding instruments cause some variation, and therefore, a time average was takenbetweencertaintimestepsofthesimulation.Theresultsareshownbelow.
SIMULATINGTHEDRAGCOEFFICIENTOFASPHERICALAUTONOMOUSUNDERWATERVEHICLE
-55-
Table2.Dragcoefficientsfordifferentvelocities.
Velocity: Dragcoefficient:
1.0km/h 0.35
1.5km/h 0.33
2.0km/h 0.30
3.0km/h 0.42
Velocity field and vortices around theAUV (2 km/h) are shown inVirhe. Viitteen lähdettä eilöytynyt..Changingvelocitychangesalsovorticesthattheprotrudinginstrumentscauses.
Figure6showstheEquation1plottedwiththesimulatedvaluesmarked.ItcanbeseenthattheAUVdragcoefficientbehavescloselytothatofasphere.Moresimulationsandmeasurementsto validate these simulationsare required to fullyunderstand thehydrodynamicsof theAUV.Gridindependencemustbestudied,astheboundarylayergridaroundtheAUVusedherewasdesigned for the targetvelocityandmightnotbe suitable forothervelocities.Also, thereareotherhydrodynamicderivativesthatneedtobestudied.
Figure6.Equation1andthesimulationresults.
CONCLUSION
The simulated drag coefficient of the AUV is in close agreement with that of a sphere. Theturbulencemodelchosenprovedtoworkwellforthisapplication.However,moreinformationfrom simulation and measurements are required to fully understand how the flow developsaroundtheAUV.Intheearlydesignstage,whenmeasurementsarenotavailable,simulationisausefultooltoaidthedesignprocess.TheinformationgainedherecanbeusedtodeveloptheAUV control system and to calculate the required batteries inside the AUV, as the energyconsumptionduringmovementisdirectlyproportionaltodrag.
SIMULATINGTHEDRAGCOEFFICIENTOFASPHERICALAUTONOMOUSUNDERWATERVEHICLE
-56-
REFERENCES
[1]Morrison,F.,A.,Anintroductiontofluidmechanics,CambridgeUniversityPress,NewYork,USA,2013.[2] Versteeg, H., K., Malalasekera, W., An introduction to computational fluid mechancics,PearsonEducationLimited,Harlow,England,2007.
AUTONOMOUSANDCOLLABORATIVEOFFSHOREROBOTICS
-57-
15. AUTONOMOUSANDCOLLABORATIVEOFFSHOREROBOTICS
JoseVillaEscusol,JussiAaltonen,KariT.KoskinenMechanicalEngineeringandIndustrialSystemsTampereUniversityofTechnologyjose.villa@tut.fi
ABSTRACT
Autonomous robotic systems are becoming increasingly important in many industries. Manydifferent industries includeworkandtasks thatcouldbedone fasterandmoresafelywithoutanyhuman interaction. These tasks vary fromunderwatermining andmaritime toborderlineguarding.Thisresearchopeningiscreatingthebasicmethodologyforanautonomousoffshorerobotic system and open demonstration in a working environment. The autonomous roboticsystem includes collaborative features between aerial subsystems (Unmanned AerialSystem/Drone) and waterborne subsystems (Autonomous Surface Vessel and AutonomousUnderwaterVehicle) to carry out inspections or other actionswithout a need for any kind ofhuman interaction. These research results of the new cooperation principles will providepossibilitytofindnewbusinessopportunitiesintheautonomousmaritimeecosystem.
INTRODUCTION
Inthispaper,theresearchopening“AutonomousandCollaborativeoffshorerobotics” isgoingtobeexplained.Themainobjectiveof the researchopening is todevelopnovel collaborativemethods between a complete autonomous robot system. Additionally, this robot systemoperatesinanoffshoreenvironmentandcoversallthreepossiblescenarios(air,watersurfaceandunderwater)byanUnmannedAerialSystem(UAS),anUnmannedSurfaceVessel(USV)andan Autonomous Underwater vehicle (AUV) respectively. The ultimate goal is a completedemonstrationoftheoffshoresysteminanopen-realenvironment.
The research openingwill focus on developingmethodology for shared intelligence, includingsystemandmission control, situational awareness and evolving capabilities tasks. In addition,offshorefieldtestingandbuildingademonstrationwillbehighlyfocusedinthisresearch.
IMPLEMENTATION
Theaccomplishmentof theresearchopening is thedemonstration inacompleteautonomousoperationoftheoffshoreroboticsystemconsistingofanUAS,anUSVandanAUV,allworkingincollaboration. The USV is used as an autonomous launch and recovery platform, and as adockingunit,wheretheUAS/DroneandAUVbatteriescanberechargedandthedatatheyhaveacquiredcanbeuploadedtocloudstorage.TheUSVdockingstationsystemisabletoprocessdata,monitor and control operations to enablemore reliable, faster andmore efficient datauploads. Furthermore, the data consistency can automatically be evaluated by the given ifmissionsneedtoberepeatedinanypossiblescenario.TheUASisusedasasensorplatformfortheUSV. This gives the systema completely newperspective and a lotmore range formanysensorsthatcannotbeinstalledintheUSV.Moreover,theUASgivestheUSVsomeshortrangeformappingcapabilities.Finally, theUAVisusedasanautonomousunderwaterexplorer.Thissubsystemisabletooperateininspection-classorresearch-classmissions.
AUTONOMOUSANDCOLLABORATIVEOFFSHOREROBOTICS
-58-
Figure1.SchematicoftheaCOLORsystem
Apart from the ultimate goal of the complete demonstration, the main point is the sharedintelligenceresearch.Thissharedintelligencecomprises:
• Systemandmissioncontrol:Formedbylow-levelcontrolofeachrobotsubsystem,
theworkingprinciplesofeachvehicle toworkautonomouslyand incollaboration,
andmissionplanning,navigationandsensordataobtaining.
• Situational awareness: Sensor data obtaining and obstacle avoidance of the three
subsystems.
• Evolving capabilities: Ability to develop when the robotic systemmakes common
activities.Thesystemworksfasterasitalreadylearnthowtofacecomparabletasks.
Itcomprisesartificialintelligence,situationalawarenessandmissioncontrol.
• Demonstration: Includes the implementationof theautonomousandcollaborative
features and a pilot scale test and demonstrations in close and open-real
environment. Concerning about the needed infrastructure for the Pilot and
Demonstrationphases,fieldtesting,pilotinganddemonstrationswillutilizeexisting
researchanddevelopment(R&D)infrastructure.
Figure2.ExistingR&Dinfrastructure(Alamarin-JetOy)
AUTONOMOUSANDCOLLABORATIVEOFFSHOREROBOTICS
-59-
Regardingtheresearchcases,responsetounknownenvironments,suchasriverenvironmentsornosatellitedata,situationawareness(monitoringthepositionandworkingconditionsofallsubsystems),commonmissions,suchassearchandrescueofthesubsystems,andredundancyofthesystem(replacementradarwithvisionfeatures)havetoberesearchedinthisoffshoreroboticsystem.Furthermore,accordingtoend-usercases,e.g.underwatermining,variousactivitiesincludingmappingorequipmentmaintenancecanbestudiedandtestedinthisresearch.
SOCIETALIMPORTANCE
These research results of the new cooperation principles will provide possibility to find newbusinessopportunities in theautonomousmaritimeecosystem.Theresultswillalsoenable tomake current operations safer and more efficient in demanding environments. Autonomousrobotic systems are becoming increasingly important in many industries. In this researchopening, the multi-component offshore robotic system can answer needs of many differentindustriessuchasocean/seatransportation (e.g.emergingautonomousshipping),underwatermining,offshorerenewableenergyorneedsofpublicauthorities(customsorborderguard).
CONCLUSIONS
Thisresearchopeningcreatesatechnologicalandmethodologicalbasisforunmannedsystemsrequired in several applications servicesneeded in the futuremaritime industries.As theUSVused in this research opening works as an autonomous carrier platform, docking unit andcommunicationrelaystation,thereisnoneedforanykindofhuman-beinginteraction,makingeveryactivitymoreefficientandsafer.
In general, offshore inspections andmany other offshore operations still remain very labourintensive and expensive activities, typically requiring at least support surface vessels beingmanned. According to that, this research contributes in making any marine and offshoreinspectionoperationsautonomousinanyscenario(air,watersurfaceandunderwater).
FUNCTIONALHIERARCHYINAUTONOMOUSROBOT
-60-
16. FUNCTIONALHIERARCHYINAUTONOMOUSROBOT
Soheil Zavari,Tuomas Salomaa, Jose Villa Escusol, Jussi Aaltonen, Kari T KoskinenMechanicalEngineeringandIndustrialSystemTampereUniversityofTechnologysoheil.zavari@tut.fi
ABSTRACT
The process of design and manufacturing of the UX-1 concern with considering a variety ofdesignrequirementsforthepurposeofcreatinganautonomousexplorerrobot.However,dueto restrictedandharshoperatingenvironment, thenon-functional requirement suchas spaceand weight highly determine overall design and operation of the system, and therefore theprocess of developing andbuilding the robot comeswith several trials and challenges.Hencethe implemented FEM analysis demonstrates the optimize solution by considering the non-functionalrequirementwhichinturndirectlyeffectonthedesignandpropertiesoftherobot.
INTRODUCTION
The process of design and development of autonomous robot interacts with numerousparameters thatneed tobe fulfilledandevolveduring theprocessofdesign.Althoughrobotsare designed and developed for the different purpose, some common grounds fundamentalsneed to be taken to account. Despite the fact that design requirements, general hierarchicaldecompositionhasbeenthetopicofsystemengineeringinmechanicaldesign,therehavebeenfew research (refer to [1] and [2]) in robotic due to interacting with complex systems. InfollowingthemainrequirementsoftheUX-1(referto[3])iscategorizedanditisdescribedthathowNon-FunctionalRequirement(NFR)parameterscaneffectonmultiplefunctionalitiesofthesystem. Later general hierarchy for an autonomous robot is shown, while there could be aninteractionbetweenparametersatdifferentlevelsorthesamelevel.
Since the design of an autonomous underwater robot interacts with multiple disciplines,evaluatingtherequirementofthesystem[referto4]becomecomplex.Inthiswork,therearerequirements which are not directly related to the technical design of the system. However,theyaffecttoahighdegreeonthesystemindirectly.Forinstance,adaptabilityandportabilityinthe context of robot operating environment (mines channels and tunnels) effect on themodeling and design of the robot. However, there is no unit or mathematical relation tomeasuring the effect exactly. ButAnalyticHierarchy Process (AHP) is onemethod in order toevaluateandverifythedesignmethod
REQUIREMENTCATEGORIZATIONOFTHESYSTEM
ThefollowingtabledemonstratestheNFRandFRparameterspriororduringthedevelopmentphaseoftheproject.
Table3:RequirementsSpecification
FunctionalRequirement Non-Functionalrequirement
Speed2km/h Endusers(stakeholders)
5hoursofoperation Pilotsites(internalstakeholder)
FUNCTIONALHIERARCHYINAUTONOMOUSROBOT
-61-
Operatingdepthof500m Degreeoffreedom
Manoeuvre in vertical shafts andtunnels
Weightandvolume(size)
Oneofthecrucialnon-functionalrequirementofthesystemistheverticalstabilityoftherobotinthewater,whichcanbederiveddirectlyfromthebuoyancyandweightoftherobot.Astherobotiscompletelybuoyant,thentheweightofrobotwouldbemaximum107kg.Thereforethedifferencebetweentwoparametersdeterminecriteriawhichfurthercanevaluatetheoperationoftherobot.Inourcase,itisdecidedthatrobotmustbecompletelybuoyantandthereforetherobotweightisexactly107kg.
GENERALHIERARCHYDECOMPOSITIONOFTHESYSTEM
Thedecompositions in thispaper follow thegeneralhierarchicaldecomposition. It shouldbenoted that the robot includes multiple complex functions however for the purpose ofdecompositiononeobjectivecanincludeseveralfunctions.Inbrief,themaingoaloftherobotcanbecondensedto ‘Autonomousvehicle forextractingmineral information’whichwouldbeontopofthehierarchy.Follwing,severalconceptscanbeextractedfromtheformerstatement,which leadtoseveralsubsystems.Solely,byconsidering themechanicalaspectof theproject,themain functionalityof the systemasanautonomous robot forms themainvarious levelofhierarchyasmaneuverability,structure,feedback,andpower.Eachlayercanrelatetoanotherlayeronewayorbothway.
Motion VisionsensorsStabilityShape MotorsVolume&
weight
Autonomy
Structure maneuverability Power Feedback
Figure7: General Hierarchy Decomposition
Thedegreeof freedomas functional requirementof the robot indicates thepropertiesof thesystem. In this case heave, surge, heading, pitch, roll, and sway are the common degree offreedomforanunderwatervehicles.Thedecompositionofmaneuverabilitydemonstrates thenecessarydegreeoffreedomwhichishorizontalandverticalmotion(surge,heave),whichcanbeoperatedindependently(decoupledsystem),andthereforeitdemonstratesspecificfunctionofthesystem.Ontheotherhand,degreesoffreedomsuchassway,roll,andpitchmotionisnotprioritized, and therefore they don’t directly effect on accomplishing the goal of the project.Hence, the roll and sway motion is added into the system only after finding an optimizedsolutionforthrustersconfigurationthatsupportsmaneuverabilityrequirement(Thefunctional
FUNCTIONALHIERARCHYINAUTONOMOUSROBOT
-62-
analysis can determine the relation between sway and surge (or heave) motion that can bedeliverasmathematicfunction.Indeed,furtherdecompositionfromfeedbacksystem,indicatesthattherobotmusthavepitchanglemotion,whichleadstothedesignofpendulumsystem.
REQUIREMENTANDDECOMPOSITION
Figure 2 demonstrates how the non-functional requirement of the system effect on thehierarchy. For instance, since the site entrance tunnel is limited, the maximum robot size isrequiredas60cmindiameterandconsequentlyitindirectlyeffectonvolumeandweightastwomajorfunctionalparametersoftherobot.Theseparametersdeterminetherobotstabilitywhichis sub level hierarchy decomposition of maneuverability. Simultaneously, the volume candetermine the shape of the robot which in turn is sub level hierarchy decomposition of thestructure.Ontheotherhand,thefigureindicatestheflowdirectioninwhichNFRcandeterminethesub-functionsandnottheotherwayaround.
NFR:Pilotsite Constraintinsize Volume Wight Stability
Structure
Manauverabiity
Figure8:NFRtoHierarchy
CONCLUSION
Since during the process of design, numerous parameters need to be taken to account, thispaperaimtoanalyzedtherequirementsoftheprojectinordertofindtherelationshipbetweenparameters.AsfutureworkAHPmethodcanbeappliedtoverifythefinalsolutionanddesignoftherobot.
REFERENCES
[1]Yamashita,Yuichi,andJunTani."Emergenceoffunctionalhierarchy inamultipletimescaleneuralnetworkmodel:ahumanoidrobotexperiment."PLoScomputationalbiology4.11(2008):e1000220[2 ] Kirschman, C. F., G.M. Fadel, and C. Jara-Almonte. "Classifying functions formechanicaldesign."TRANSACTIONS-AMERICAN SOCIETY OF MECHANICAL ENGINEERS JOURNAL OFMECHANICALDESIGN120(1998):475-482.[3] Zavari, Soheil, et al. "Early stagedesignof a sphericalunderwater robotic vehicle."SystemTheory,ControlandComputing(ICSTCC),201620thInternationalConferenceon.IEEE,2016.[4]Morita,Toshio,Hiroyasu Iwata,andShigekiSugano. "Humansymbiotic robotdesignbasedon division and unification of functional requirements."Robotics and Automation, 2000.Proceedings.ICRA'00.IEEEInternationalConferenceon.Vol.3.IEEE,2000.
LISTOFAUTHORS
1. AaltonenJussi2. AhonenToni3. AhonenToni4. AngleraudAlexandre5. AnttilaJuha-Pekka6. AuerkariPertti7. ChakrabortiAnanda8. CoatanéaEric9. HakalaTimoJ.10. HalmeJari11. HanskiJyri12. HeikkiläEetu13. HeininenArttu14. HelaakoskiHeli15. HosseiniMohammad16. Kivikytö-ReponenPäivi17. KoskinenKariT.18. KujanpääVeli19. LaukkanenAnssi20. LeinoSimo-Pekka21. MokhtarianHossein22. NiemiArto23. ParisHenri24. PenttinenJussi-Pekka25. PietersRoel26. PuukkoPasi27. RaunioKalle28. ReijonenJoni29. RiipinenTuomas30. SaarelaOlli31. SalomaaTuomas32. TiusanenRisto33. TuurnaSatu34. UseniusOlli35. VainioHenri36. VainioHenri37. ValkokariPasi38. VihinenJorma39. VillaEscusolJose40. Yli-OlliSanni41. Zavari Soheil
#SMACCFinland
www.smacc.fi
top related