computer-aided materials selection during structural design

179
title: Computer-aided Materials Selection During Structural Design author: publisher: National Academies Press isbn10 | asin: 0309051932 print isbn13: 9780309051934 ebook isbn13: 9780585030753 language: English subject Structural design--Data processing, Materials--Data processing. publication date: 1995 lcc: TA658.2.C64 1995eb ddc: 624.1/7 subject: Structural design--Data processing, Materials--Data processing.

Upload: others

Post on 11-Sep-2021

6 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Computer-Aided Materials Selection During Structural Design

title: Computer-aidedMaterialsSelectionDuringStructuralDesign

author:publisher: NationalAcademiesPress

isbn10|asin: 0309051932printisbn13: 9780309051934ebookisbn13: 9780585030753

language: English

subject Structuraldesign--Dataprocessing,Materials--Dataprocessing.

publicationdate: 1995lcc: TA658.2.C641995ebddc: 624.1/7

subject: Structuraldesign--Dataprocessing,Materials--Dataprocessing.

Page 2: Computer-Aided Materials Selection During Structural Design

NATIONALRESEARCHCOUNCILCOMMISSIONONENGINEERINGANDTECHNICALSYSTEMS

NATIONALMATERIALSADVISORYBOARD

ThepurposeoftheNationalMaterialsAdvisoryBoardistheadvancementofmaterialsscienceandengineeringinthenationalinterest.

CHAIRMAN

Dr.JamesC.WilliamsGeneralManagerEngineeringMaterialsTechnologyLaboratories,MailDropH85GeneralElectricCompanyINeumannWayCincinnati,OH45215-6301

MEMBERS

Dr.JanD.AchenbachDirector,CenterforQualityEngineering&FailurePreventionNorthwesternUniversity2137N.SheridanRoadEvanston,IL60208-3020

Dr.BillR.AppletonAssociateDirectorOakRidgeNationalLaboratory4500N,MS-6240P.O.Box2008OakRidge,TN37831-6255

Mr.RobertR.BeebeConsultantP.O.Box32048

Page 3: Computer-Aided Materials Selection During Structural Design

Tucson,AZ95751-2048Tucson,AZ85715)

Dr.1.MelvinBernsteinVicePresidentforArts,ScienceandTechnologyBallouHallTuftsUniversityMedford,MA02155

Dr.J.KeithBrimacombeDirector,TheCentreforMetallurgicalProcessEngrg.TheUniversityofBritishColumbia#305-6350StoresRoadVancouver,BCV6T124CANADA

Dr.JohnV.BuschPresidentIBISAssociates,Inc.55WilliamStreet,Suite220Wellesley,MA02181

Dr.HarryE.CookDepartmentofMechanicalandIndustrialEngineeringUniversityofIllinois,Urbana1206WestGreenStreetUrbana,IL61801

Dr.RobertEaganDirector,EngineeredMaterialsProgramsandMetrologyCenterSandiaNationalLaboratoriesP.O.Box5800,Org.1700Albuquerque,NM87185-0336

Dr.CarolynHanssonProfessorandHead,DepartmentofMaterialsandMetallurgical

Page 4: Computer-Aided Materials Selection During Structural Design

EngineeringQueen'sUniversity-NicolHallKingston,Ontario,Canada,K7L3N6

Dr.KristinaM.JohnsonDirector,OptoelectronicsComputingUniversityofColoradoCampusBox425Boulder,CO80309

Dr.LionelC.KimerlingThomasLordProfessorofMaterialsScienceandEngineeringMassachusettsInstituteofTechnology,Room13-5094Cambridge,MA02139

Dr.JamesE.McGrathDirector,NSFScienceandTechnologyCenterVirginiaPolytechnicInstituteandStateUniversityBlacksburg,VA24061-0344

Dr.RichardS.MullerDirector,BerkeleySensorandActuatorCenterUniversityofCalifornia,BerkeleyBerkeley,CA94720

Dr.ElsaReichmanisDepartmentHead,PolymersandOrganicMaterialsResearchAT&TBellLaboratories600MountainAve.,Rm1A261MurrayHill,NJ07974

Dr.EdgarA.StarkeOglesbyProfessorofMaterialsScience&EngineeringThorntonHall

Page 5: Computer-Aided Materials Selection During Structural Design

UniversityofVirginiaCharlottesville,VA22903

Dr.JohnStringerDirectorofAppliedResearchElectricPowerResearchInstitute3412HillviewAvenueP.O.Box10412PaloAlto,CA94303

Dr.KathleenC.TaylorHead,PhysicalChemistryDepartmentGeneralMotorsCorporation30500MoundRoadWarren,MI48090

Dr.JamesWagnerProfessorandChairmanEngineeringDepartment,BiomedicalEngineeringTheJohnsHopkinsUniversity3400NorthCharlesStreetBaltimore,MD21218

Dr.JosephWirthSeniorVicePresidentandChiefTechnicalOfficerRaychemCorporation300ConstitutionDriveM.S.122/8505MenloPark,CA94025-1164

NMABDIRECTOR

Dr.RobertE.Schafrik2101ConstitutionAvenue,N.W.Washington,DC20418

Page 6: Computer-Aided Materials Selection During Structural Design
Page 7: Computer-Aided Materials Selection During Structural Design

Pagei

Computer-AidedMaterialsSelectionDuringStructuralDesign

CommitteeonApplicationofExpertSystemsto

MaterialsSelectionDuringStructuralDesign

NationalMaterialsAdvisoryBoardCommissiononEngineeringand

TechnicalSystemsNationalResearchCouncil

NMAB-467NationalAcademyPressWashington,D.C.1995

Page 8: Computer-Aided Materials Selection During Structural Design

Pageii

NOTICE:TheprojectthatisthesubjectofthisreportwasapprovedbytheGoverningBoardoftheNationalResearchCouncil,whosemembersaredrawnfromthecouncilsoftheNationalAcademyofSciences,theNationalAcademyofEngineering,andtheInstituteofMedicine.Themembersofthecommitteeresponsibleforthereportwerechosenfortheirspecialcompetenciesandwithregardforappropriatebalance.

ThisreporthasbeenreviewedbyagroupotherthantheauthorsaccordingtoproceduresapprovedbyaReportReviewCommitteeconsistingofmembersoftheNationalAcademyofSciences,theNationalAcademyofEngineering,andtheInstituteofMedicine.

ThisstudybytheNationalMaterialsAdvisoryBoardwasconductedunderARPAOrderNo.8475issuedbyDARPA/CMOunderContractNo.MDA972-92-C-0028withtheU.S.DepartmentofDefenseandtheNationalAeronauticsandSpaceAdministration.

Theviewsandconclusionscontainedinthisdocumentarethoseoftheauthorsandshouldnotbeinterpretedasrepresentingtheofficialpolicies,eitherexpressedorimplied,oftheDefenseAdvancedResearchProjectsAgencyortheU.S.Government.

LibraryofCongressCatalogCardNumber94-69233InternationalStandardBookNumber0-309-05193-2

Availableinlimitedsupplyfrom:NationalMaterialsAdvisoryBoard2101ConstitutionAvenue,NWWashington,[email protected]

Additionalcopiesareavailableforsalefrom:NationalAcademyPress2101ConstitutionAvenue,NW

Page 9: Computer-Aided Materials Selection During Structural Design

Box285Washington,D.C.20055800-624-6242202-334-3313(intheWashingtonMetropolitanArea)

Copyright1995bytheNationalAcademyofSciences.Allrightsreserved.

PrintedintheUnitedStatesofAmerica.

Page 10: Computer-Aided Materials Selection During Structural Design

Pageiii

AbstractTheselectionofthepropermaterialsforastructuralcomponentisacriticalengineeringactivity.Itisgovernedbymany,oftenconflictingfactorsthattypicallyincludeservicerequirements,designlife,materialsavailability,databaseaccessibility,manufacturingconstraints,repairandreplacementstrategies,clientpreferences,andcost.Theincorporationofcomputer-aidedmaterialsselectionsystemsintocomputer-aideddesignandcomputer-aidedmanufacturingoperationscouldassistdesignersbysuggestingpotentialmanufacturingprocessesforparticularproductstofacilitateconcurrentengineering,recommendingvariousmaterialsforaspecificpartbasedonagivensetofcharacteristics,orproposingpossiblemodificationsofadesignifsuitablematerialsforaparticularpartdonotexist.Thisreportreviewsthestructuraldesignprocess,determinestheelementsandcapabilitiesrequiredforacomputer-aidedmaterialsselectionsystemtoassistdesignengineers,andrecommendstheresearchanddevelopmentareasofmaterialsdatabase,knowledgebase,andmodelingrequiredtodevelopacomputer-aidedmaterialsselectionsystem.

Page 11: Computer-Aided Materials Selection During Structural Design

Pageiv

TheNationalAcademyofSciencesisaprivate,nonprofit,self-perpetuatingsocietyofdistinguishedscholarsengagedinscientificandengineeringresearch,dedicatedtothefurtheranceofscienceandtechnologyandtotheiruseforthegeneralwelfare.UpontheauthorityofthechartergrantedtoitbytheCongressin1863,theAcademyhasamandatethatrequiresittoadvisethefederalgovernmentonscientificandtechnicalmatters.Dr.BruceAlbertsispresidentoftheNationalAcademyofSciences.

TheNationalAcademyofEngineeringwasestablishedin1964,underthecharteroftheNationalAcademyofSciences,asaparallelorganizationofoutstandingengineers.Itisautonomousinitsadministrationandintheselectionofitsmembers,sharingwiththeNationalAcademyofSciencestheresponsibilityforadvisingthefederalgovernment.TheNationalAcademyofEngineeringalsosponsorsengineeringprogramsaimedatmeetingnationalneeds,encourageseducationandresearch,andrecognizesthesuperiorachievementsofengineers.Dr.RobertM.WhiteispresidentoftheNationalAcademyofEngineering.

TheInstituteofMedicinewasestablishedin1970bytheNationalAcademyofSciencestosecuretheservicesofeminentmembersofappropriateprofessionsintheexaminationofpolicymatterspertainingtothehealthofthepublic.TheInstituteactsundertheresponsibilitygiventotheNationalAcademyofSciencesbyitscongressionalchartertobeanadvisortothefederalgovernmentand,uponitsowninitiative,toidentifyissuesofmedicalcare,research,andeducation.Dr.KennethI.ShineispresidentoftheInstituteofMedicine.

TheNationalResearchCouncilwasorganizedbytheNationalAcademyofSciencesin1916toassociatethebroadcommunityofscienceandtechnologywiththeAcademy'spurposesoffurthering

Page 12: Computer-Aided Materials Selection During Structural Design

knowledgeandadvisingthefederalgovernment.FunctioninginaccordancewithgeneralpoliciesdeterminedbytheAcademy,theCouncilhasbecometheprincipaloperatingagencyofboththeNationalAcademyofSciencesandtheNationalAcademyofEngineeringinprovidingservicestothegovernment,thepublic,andthescientificandengineeringcommunities.TheCouncilisadministeredjointlybybothAcademiesandtheInstituteofMedicine.Dr.BruceAlbertsandDr.RobertM.Whitearechairmanandvicechairman,respectively,oftheNationalResearchCouncil.

Page 13: Computer-Aided Materials Selection During Structural Design

Pagev

CommitteeonApplicationofExpertSystemstoMaterialsSelectionDuringStructuralDesign

FRANKW.CROSSMANChair,Director,MaterialSciences,LockheedPaloAltoResearchLaboratory,PaloAlto,California

JAND.ACHENBACH,Director,CenterforQualityEngineeringandFailurePrevention,NorthwesternUniversity,Evanston,Illinois

HAROLDL.GEGEL,Director,ProcessingScienceDivision,UniversalEnergySystems,Dayton,Ohio

RICHARDN.HADCOCK,VicePresident,RNHAssociates,Inc.,Huntington,NewYork

THOMASS.KACZMAREK,NorthAmericanOperationManufacturingCenter,GeneralMotorsCorporation,Warren,Michigan

J.GILBERTKAUFMAN,VicePresident,Technology,TheAluminumAssociation,Washington,D.C.

MICHAELORTIZ,EngineeringDepartment,BrownUniversity,Providence,RhodeIsland

FRIEDRICHB.PRINZ,RodneyH.AdamsProfessorofEngineering,DepartmentsofMechanicalEngineeringandMaterialsScience,StanfordUniversity,Stanford,California

JANSCHREURS,WestinghouseScienceandTechnologyCenter,WestinghouseElectricCompany,Pittsburgh,Pennsylvania

VOLKERWEISS,ProfessorofEngineeringandPhysics,Chairman,DepartmentofMechanicalAerospaceandManufacturingEngineering,SyracuseUniversity,Syracuse,NewYork

LIAISONREPRESENTATIVES

Page 14: Computer-Aided Materials Selection During Structural Design

RALPHP.I.ADLER,Chief,MetalsResearchBranch,ArmyResearchLaboratoryMaterialsDirectorate,Watertown,Massachusetts

WILLIAMBARKER,DefenseSciencesOffice,DefenseAdvancedResearchProjectsAgency,Arlington,Virginia

ANDREWCROWSON,Director,MetallurgyandMaterialsScienceDivision,ResearchTrianglePark,NorthCarolina

WALTERM.GRIFFITH,DeputyDirector,MetalsandCeramicsDivision,MaterialsDirectorate,WrightPattersonAirForceBase,Ohio

CRAIGMADDEN,ResearchEngineeringGroup,DavidTaylorResearchCenter,Bethesda,Maryland

RONALDG.MUNRO,Physicist,CeramicsDivision,NationalInstituteofStandardsandTechnology,Gaithersburg,Maryland

NMABSTAFF

ROBERTM.EHRENREICH,SeniorStaffOfficer

PATWILLIAMS,SeniorSecretary

Page 15: Computer-Aided Materials Selection During Structural Design

Pagevi

AcknowledgementsThecommitteewouldliketoexpressitsappreciationtothefollowingindividualsfortheirpresentationstothecommittee:J.HendrixofHerculesIncorporated,O.RichmondofALCOA,andD.MarinaroofPDAEngineering.ThecommitteewouldalsoliketothankLarryIlcewiczforhostingasitevisittotheBoeingfacilityinSeattle,Washington,andthefollowingindividualsfortheirpresentations:H.Shomber,A.Falco,T.Lackey,T.Richardson,B.Das,B.Backman,A.Miller,J.Boose,P.Rimbos,andG.Swanson.ThecommitteeacknowledgeswiththanksthecontributionsofRobertM.Ehrenreich,SeniorStaffOfficer,andPatWilliams,SeniorSecretary,totheproject.

Page 16: Computer-Aided Materials Selection During Structural Design

Pagevii

PrefaceTheDepartmentofDefenseandtheNationalAeronauticsandSpaceAdministrationrequestedthattheNationalMaterialsAdvisoryBoardconveneacommitteetostudytheapplicationofexpertsystemstomaterialsselectionduringstructuraldesign.Theobjectivesofthestudyweretodetermine(1)thecomponentsneededforaneffectivecomputer-assisted,concurrent-engineeringdesignsystem,(2)thebarrierspreventingthedevelopmentofsuchasystem,and(3)theresearchanddevelopmentrequiredtoconstructsuchasystem.

ThecommitteemetsixtimesbetweenJune1991andSeptember1992.Thefirstmeetingfocusedondevelopingaperspectiveonthestudyscopeandanapproachforassessingtheprimary,underlyingtechnologiespertinenttothestudyviacasestudiesandanindustrysitevisit.1Thesecondmeetingprovidedanopportunityforeachcommitteeandliaisonmembertodescribehisorherresearchandtechnicalexperiencepertinenttothecommitteecharter.Committeemembersalsopresentedcasestudiesofcomputer-aidedmaterialsselectionsystemswithwhichtheyhadexperience.Thishelpeddeterminethestateoftheartofsuchsystemsandprovideexamplesofdesigndecisionsinvolvinggeometricrelations;designrulesassociatedwithperformance,processing,manufacturing,andsupportability;andadvancedcomputerconceptsandtechnologiesthataidthedesignoptimizationprocess.Thecommitteethenheldthreestudysessionsfocusedonproductdesign,materialssupplyanddevelopment,andstate-of-the-artcomputer-aidedsystemstechnologyformaterialsselection.ThefirststudysessionconsistedofasitevisittotheBoeingCommercialAirplaneCompanyinSeattle,Washington,tolearnaboutmaterialsselectionwithintheairplanedesignprocess.Thesecondsessionconsistedofpresentationsbyrepresentativesof

Page 17: Computer-Aided Materials Selection During Structural Design

ALCOAandHerculesonmaterialsmodeling,certification,andthesupplier-designerinteraction.Thethirdsessionfocusedoncomputerdemonstrationsofsomestate-of-the-artsystemsthataidthematerialsselectionprocess,todeterminecurrentcapabilitiesandidentifybarrierstothedevelopmentofanoptimalsystem.

Thisreportisdividedintofivechapters.Chapter1definesthestudyscopeandcommitteecharge.Chapter2placesmaterialsselectionwithinthecontextofthedesignprocess,usingtheBoeingCommercialAirplaneGroupasacasestudy.Chapter3presentsthecommittee'svisionofafull-function,computer-aidedmaterialsselectionsystem.Chapter4reviewstheunderlyinginformationtechnologiespertinenttothematerialsselectionprocess,determinedbytheexaminationofthecasestudieslistedinAppendixB.Chapter5discussestheissuespreventingthedevelopmentofcomputer-aidedmaterialsselectionsystemsandoutlinestherecommendationsfortheresearchanddevelopmentrequiredtoattaintheenvisionedsystem.Theappendicesinclude(1)aglossaryofacronyms,(2)

1Thenumberofcomputer-aidedsystemsonthemarketisrapidlygrowing,withnewproductsbeingintroduceddaily(seeSchorrandRappaport,1989;RappaportandSmith,1991;SmithandScott,1991).Sinceanycompilationofsystemswouldberapidlyout-of-date,thecommitteedeterminedthecurrentcapabilitiesofcomputer-aidedsystemsbyexaminingthe30casestudieslistedinAppendixB.

Page 18: Computer-Aided Materials Selection During Structural Design

Pageviii

acompletelistofthecasestudiesreviewedbythecommittee,(3)areviewofsomeoftheknowledge-representationtoolsandtechnologiesdiscussedinthereport,and(4)twoexamplesofthecasestudiesreviewedbythecommitteethattypifythematerialsselectionanddatabasesystemscurrentlyavailable.

CommentsorsuggestionsthatreadersofthisreportwishtomakecanbesentviaInternetelectronicmailtonmab@nas.eduorbyFAXtotheNationalMaterialsAdvisoryBoardat202/334-3718.

FrankW.Crossman,Chair

Page 19: Computer-Aided Materials Selection During Structural Design

Pageix

Contents

EXECUTIVESUMMARY 1

VisionofaComputer-AidedMaterialsSelectionSystem 1

StrategiesforOvercomingBarriers 2

GeneralConclusionsandRecommendations 4

1INTRODUCTION 7

Benefits 7

Definitions 8

StudyObjectivesandScope 8

2MATERIALSSELECTIONINSTRUCTURALDESIGN 11

ConcurrentEngineeringandDesignOrganization 11

MaterialsSelectionDuringComputer-AidedDesign 13

SummaryofMaterialsInformationRequirementsinDesign

15

3ENHANCINGTHEMATERIALSSELECTIONPROCESSINDESIGN:AVISION

19

IntegratedEngineeringSupportinIntegratedEnterprises 19

SupportingStrategicMaterialDecisions 20

SupportingRoutineMaterialDecisions 21

SupportingInnovativeMaterialsSelectioninDesign 21

Summary 22

Page 20: Computer-Aided Materials Selection During Structural Design

4INFORMATIONTECHNOLOGIESPERTINENTTOTHEMATERIALSSELECTIONPROCESS

25

DatabasesandKnowledgeBases 25

ModelingandAnalysisSystems 29

5CONCLUSIONSANDRECOMMENDATIONS 37

StrategiesforOvercomingBarriers 37

GeneralConclusionsandRecommendations 39

References 41

AppendixA:GlossaryofAcronyms 45

AppendixB:CaseStudiesReviewedbytheCommittee 47

Page 21: Computer-Aided Materials Selection During Structural Design

Pagex

AppendixC:ReviewofSelectedKnowledge-RepresentationTechniquesandTools

49

AppendixD:Knowledge-BasedIntegratedDesignSystem 57

AppendixE:AnIntelligentKnowledgeSystemforSelectionofMaterialsforCriticalAerospaceApplications

63

AppendixF:BiographicalSketchesofCommitteeMembers 69

Page 22: Computer-Aided Materials Selection During Structural Design

Pagexi

ListofFigures

2-1Thesequentialengineeringapproachtostructuraldesign 11

2-2Atypicalcomparisonofsequentialandconcurrentengineering

12

2-3AnexampleofatypicalstructuresDBThierarchy:Boeing777horizontalstabilizerDBTs

12

2-4Theexpertiseinastructuraldesignconcurrentengineeringteam

13

2-5TheinteractionsofatypicalDBT 13

2-6AmodelofthewingoftheGrummanX-29andassociatedfiniteelementanalysis

14

2-7Structuresandmaterialsdesigninteractions 15

3-1TheconceptualarchitectureofaComputer-AidedMaterialsSelectionSystem

24

4-1Automobilesidemarker 30

4-2Concurrentengineeringenvironmentincludinginspectability

33

4-3PODcurvesfortwoscanningplans 35

C-Thedifferencesbetweenwireframe,surfacemodel,and 52

Page 23: Computer-Aided Materials Selection During Structural Design

1 solidmodelrepresentationaldomains

C-2AnexampleoftheLOOSsystemtodefinethestructureor''topology''ofalayout

55

D-1InformationflowinIPD 58

D-2Methodsdevelopers'frameofreference 59

D-3Controlflowbetweenrolesinthebladedesignassistant 60

D-4Bladedesignassistant 61

E-1Modelofanintelligentknowledgesystemapplicabletothematerialselectionproblem

64

E-2SpecificsystemarchitecturefortheprototypeIKSMAT 64

Page 24: Computer-Aided Materials Selection During Structural Design

Pagexii

ListofTables

2-1ExamplesofMaterialsInformationRequiredDuringProductDesign

16

2-2TypicalProductDesignRequirementsforAircraftStructureDevelopment

16

2-3SummaryofDesigner"Wants" 17

3-1SummaryoftheMaterials-SpecificInformationTechnologiesandSomeofthePrimaryComputerTechnologiesRequiredforaComputer-AidedMaterialsSelectionSystem

23

4-1RepresentativeApplicationsBasedonKnowledgeofMaterials

27

4-2StepsintheDevelopmentofaProcessModel 31

Page 25: Computer-Aided Materials Selection During Structural Design

Page1

ExecutiveSummarySelectingthepropermaterialsforastructuralcomponentiscriticaltoengineeringdesign.Materialsselectionisgovernedbymanyfactors,someofwhichareinopposition.Theprincipalselectionfactorsincludetheservicerequirementsanddesignlifeoftheproduct;theavailabilityofcandidatematerialsandtheappropriatedataonapplication-specificpropertiesforthem;thecompany'smakeorbuydecisionforthesystemcomponents;thecustomerpreferences;and,mostimportantly,thetotallife-cyclecost.

Theuseofcomputer-aidedsystemscouldreducecostanddesignreworkandrequalificationbyprovidingengineeringdesignteamswiththemostcurrentmaterials-propertydata,knowledgeoffactorssuchasmaterialsoptionsandlife-cyclecosts,andavailablematerialsforadesignbasedonexperiencederivedfrompreviousproductdevelopments.AComputer-AidedMaterialsSelectionSystem(CAMSS)withlearningcapabilitieswouldalsoensuretheproperarchivingofmaterialsselectiondecisionsforfuturereferenceandacceleratetheapplicationofnewmaterialsandprocessingtechnologiesbyprovidingdesignerswithanexpandedrangeofpossiblematerialsandmanufacturingmethodsforagivensetofproductcharacteristicsandcost-performancecriteria.

Thisstudyconcentratedonthematerials-specificknowledgeelementsofacomputer-aidedsystem.TheCommitteeontheApplicationofExpertSystemstoMaterialsSelectionduringStructuralDesigndeterminedthatthedevelopmentofgenericcomputer-aidedsystemsisalreadyreceivingagreatdealofattentionwithinthecomputerscienceandengineeringcommunity.Thebasicinformationrequirementsforacomputer-aidedsystemformaterialsselectionarereceivinglittleattentionwithinthematerialscommunity,however.Thus,the

Page 26: Computer-Aided Materials Selection During Structural Design

committeeassessedthatthisstudywouldhavethegreatestimpactifit(1)detailedthecapabilitiesrequiredforcomputer-aidedsystemstobeofvaluetothematerialsselectionprocessduringconcurrentengineering;(2)identifiedtheissuesinhibitingthedevelopmentofsuchasystem;and(3)recommendedmaterials-specificapplicationsanddevelopmentsindatabase,knowledgebase,andmaterialsmodelingthatwouldaidtheproductionofaknowledgeelementappropriateforcomputer-aidedsystemsformaterialsselectionduringdesign.

VISIONOFACOMPUTER-AIDEDMATERIALSSELECTIONSYSTEM

ThecommitteedevelopedaconceptualarchitectureforaCAMSSthatdepictsthesupportingmaterials-specificinformationtechnologiesrequired.TheobjectiveofaCAMSSshouldbetoprovidedesignoptionsforconsiderationbythedesignengineeringteam.Designandmaterialsadvisortoolsshouldbeavailablethroughouttheconcurrentengineeringprocess.Significantmaterialpropertiesaswellasemergingconsiderations,suchasprocessingandproductrecyclingcosts,willbeincreasinglysupportedbytheinformationinfrastructure.Materialsknowledgeshouldbemadeaccessibletotheengineerasreferencedatathroughdesignadvisorsthatinteractwithproductandprocessmodelsthatanalyze,critique,improve,oroptimizethedesign.

Majortoolsintheintegratedenvironmentwillprovidethefollowingmaterials-specificcapabilities:managingelectronicrepositoriesofdataanddocuments,searchingpastdevelopmenthistoriestofindsimilaroranalogousproducts,managingrequirements,analyzingperformancecharacteristics,modelingmanufacturingandmaintenancecharacteristics,estimatingcosts,suggestingimprovementstotheproposedproductorprocessdescription,andstoringtherationaleformaterialsselectiondecisionsforfuturereference.Thealternativeselectedduringconceptevaluationwould

Page 27: Computer-Aided Materials Selection During Structural Design

thenbeavailableforfurtherrefinementbythedesignerinacoarse-to-finedevelopmentprocess.Toaccomplishthis,theCAMSSshouldmakeuseofavailablecomputer-aidedsystemstechnologies.Computer-aidedsystemsconsistingofbothheuristicandquantifiabledesignrulescanbedevelopedforsubsetsofthedesignknowledgebase.

ComputingtechnologynolongerpresentsabarriertothedevelopmentofaCAMSS.Thewiderangeofbothhardwareandsoftwarecapabilitiesisrapidlyreducingthecostofrepresentingandimplementingcomputer-aided

Page 28: Computer-Aided Materials Selection During Structural Design

Page2

systemlogicandprocesssimulationwithinaffordablelimits.Advancesinreducedinstructionsetcomputer(RISC)chiptechnologyallowinexpensiveworkstationstoperformbothdesignlayoutandanembeddedstructuralanalysisormaterialsprocessingsimulation.Visualizationtechniquescoupledwithsimulationofsystembehavioratmanylevelscanbeapowerfulmeansofconveyinginformationtodesignteammembers.Thecontinuedevolutionofcost-effective,high-performancecomputinginconjunctionwithanationalinformationsuperhighwayinfrastructurewillfurtherassistthenation'smanufacturingsectorinbecomingmorecompetitiveintheinternationalmarket-place.

STRATEGIESFOROVERCOMINGBARRIERS

ThecommitteeidentifiedtwomainareasthatarecurrentlypreventingthedevelopmentofaCAMSS:(1)databaseandknowledge-basedesign,implementation,instantiation,andmanagementand(2)structuraldesignmodelingtechnologies.

DatabaseandKnowledge-BaseBarriers

Thedesign,implementation,instantiation,andmaintenanceofmaterialspropertiesdatabasesandknowledgebasesareintegraltothedevelopmentofaneffectiveCAMSS.Forexample,adesignengineercannotuseasystemiftheunderlyingdatabasescontainobsolete,extraneous,unverified,orincompleteinformation.Thecommitteehasfoundthatthedatabaseandknowledge-baseareaiscurrentlyinhibitedbyfivebarriers.

1.StandardizationofdatabasesandknowledgebasesConstructingdatabasesandknowledgebasesthatcontaintherelevantinformationrequiredforthedesignprocessanddevelopingsystemsthatlocateandpresentthisdataaretwodifficultproblemsbecauseoftheamountofextraneousinformationavailableandthelackofstandardsinthecontentofdatabasesandknowledgebases.Toovercomethese

Page 29: Computer-Aided Materials Selection During Structural Design

contentofdatabasesandknowledgebases.Toovercomethesebarriers,thecommitteerecommendsthat(1)standardsandguidelinesbedevelopedforelectronicdataquality,capture,storage,analysis,andexchange(followingtheComputer-AidedAcquisitionandLogisticsSupportandtheStandardfortheExchangeofProductapproaches)andknowledge-basecontentandconstruction;(2)CAMSSbedesignedtoacceptavarietyofdatabasetaxonomiesthroughtheuseofactive,"intelligent"datadictionariesthataidtheidentificationandconversionofthecontentsofdifferentdatabasesforuseinthesystem;(3)linksbetweenmaterialsdatabasesandknowledgebasesbeimprovedandcomputernetworksformaterials-specificinformationcommunicationbecreated(e.g.,anelectronicJournalofMaterialsSelectioninStructuralDesign,anationalmaterialsbulletinboardonInternet,oralinkednetworkofworldwidematerialsdatasystems);and(4)electronictechnicalassistancebeprovidedtodesignteamsinelectronicformats.

2.StatusofknowledgecaptureMethodsforknowledgecapturearerequiredtoenhancethelessons-learnedsegmentofCAMSS.Theseincludeestablishingknowledge-representationtaxonomies,technicalcontextstandards,andtechniquestoupdateandaccessthisinformationrapidly.Toovercomethisbarrier,thecommitteerecommendsthat(1)materialsandcomputerscientistscollaborateinthedevelopmentofsuitableknowledge-capturesystemsforuseinCAMSS;(2)industrydesignteamsbeencouragedtoestablishelectronictechnicaldatabasesbyelectroniccaptureofalldesigndiscussions,decisions,andlessonslearnedinfreetext,spread-sheet,computer-aideddesign(CAD)standards,andothermultimediaformats;and(3)industrydesignteamsbeencouragedtoassignspecificfunctionswithintheteamtospecialize,categorize,index,andfiltertheaccumulateddesignknowledgebaseandlocateandaccessotherdesignknowledgebases.

3.DiffuseresponsibilityforgeneratingdatabasesTheissueofhowtocoordinatematerialsdevelopers,componentusers,andmaterialssocietiestogenerateandintegratematerialspropertydatabases

Page 30: Computer-Aided Materials Selection During Structural Design

societiestogenerateandintegratematerialspropertydatabasesrequiresresolution.Materialssupplierspredominantlyleavematerialsqualificationprogramstotheuserbecauseofconcernsthattheywillbeheldliableforsystemmalfunctionscausedbyfailuresandthatuserswillonlyemploy

Page 31: Computer-Aided Materials Selection During Structural Design

Page3

materialsthattheythemselveshavequalified.Materialssocietiesgenerallydonothavetheresourcesnecessaryforlargeprojects.Componentmanufacturerstypicallyonlyqualifymaterialsforagivenapplicationandtreatthedataasproprietary.Toovercomethisbarrier,thecommitteerecommendsthat(1)nationalteameffortsofusers,suppliers,materialssocieties,andstandardsorganizationsdevelopintegratedmaterialqualificationprogramsthatrelatetodesignrequirementsandeventualuseand(2)theresultantappropriate,independentlyverifieddatabemadeavailableinanationalinformationinfrastructuretoprovidearealistic,initialappraisaloftheadvantagesofamaterial.

4.DisclosureofmaterialsdataIngeneral,companiesprotectasproprietaryalldatabasesandknowledgebasesthatcontainmaterialspropertiesandproduction-relateddata,suchas(1)state-of-the-artinformation,projections,orforecasts;(2)manufacturinglaborstandards,rates,andpricedata;and(3)weight,performance,andcosttradeoffdataandcriteria.Toovercomethisbarrier,thecommitteerecommendsthatCAMSSbedesignedtoassurethatproprietaryportionsofdatabasesandknowledgebasesarefullyprotected.

5.InvestmenttomaintaindatabasesItisimportantthattheinformationwithinadatabasebeconstantlymonitored,verified,andupdatedtoensurethatthebestpossibleinformationisavailable.Toovercomethisbarrier,organizationsmust(1)assigntheresponsibilityformaintenanceofdatabasestoacentralizedfunction,suchasadataadministratorwithdomainexpertsidentifiedtoactascuratorsoftheknowledgebase,and(2)providelong-termsupportfordatabasemaintenanceoncetheprogramisestablished.

StructuralDesignModelingTechnologyBarriers

ModelinginstructuraldesignwillbeanimportantcomponentofanyCAMSSbothtoprovidevaliddetailsonwhichtobasetradeoffdecisions

Page 32: Computer-Aided Materials Selection During Structural Design

CAMSSbothtoprovidevaliddetailsonwhichtobasetradeoffdecisionsandtoreducerelianceonlyonmaterials-propertiesdatabases.Modelingtechniquesarerequiredforgeometricreasoning,materialresponsesonmultiplescalelevels,materialsprocessing,manufacturingprocessingperformance,productperformance,andlife-cycleissuessuchasinspectability.Modelingtechniqueswillalsoberequiredthatsimulatenewmaterialsbysuccessiveextrapolationfromthepropertiesofexistingmaterialsorbycalculationfromfirstprinciples.Thecommitteeidentifiedtwobarrierstothedevelopmentofmodeling.

1.OptimizationmodelingAsopposedtosimplyshowingtradeoffsbetweendesignparametersinputbyusers,modelingtechniqueswillberequiredthatcansuggestmodificationstooptimizedesignsandmanufacturingprocesses.Processoptimizationisanimportantingredientofintegratedproduct-processdesignandwillbeusedmoreandmoreinthefutureastheindustryfullyadoptsconcurrentengineeringtoreducemanufacturingcostsandconvergeonmanufacturingsolutionsinashortertime.Tobeuseful,modelingmustalsobedonerapidlyandaccurately,usingnormaldesignparametersandinformationfrommultipleknowledgebases.Ifmodelingtechniquesaretooslow,untrustworthy,orunabletoaccesstheproperinformation,theywilllanguish.Toovercomethesebarriers,thecommitteerecommendsthat(1)materialsscientistsandcomputerengineersfromindustryanduniversitycollaboratetodevelopadvancedmodelingtechniquestoreducerelianceonstraightmaterialsdata,introduceexpertknowledge,provideacrediblebasisfortradeoffdecisions,andincreasetrustinCAMSS;and(2)materialsscientistsparticipateinbasicandappliedresearchthatestablisheslinksbetweenmaterialsmodelsatseveralscales(e.g.,atomic,molecular-crystal,cluster-grainsize,polycrystal-aggregate,substructure,structure,andsystem).

2.CulturalandeducationalbarrierstoimplementingmodelingandanalysistechnologyThedesignprocessistraditionallyaheuristictrial-and-errorapproach.Increasedrelianceonmodelingtechniquesrequiresestablishingconfidencethattheimproveddesignsolutions

Page 33: Computer-Aided Materials Selection During Structural Design

requiresestablishingconfidencethattheimproveddesignsolutionscanbedevelopedinashortertimeperiod.Currentengineeringprogramsdonotstresstheimportanceoftrainingineithermaterialssynthesisandprocessingorcomputerscience.Formodelingand

Page 34: Computer-Aided Materials Selection During Structural Design

Page4

analysissystemsinaCAMSStobeusefulandeffective,futureengineersmustreceivetrainingincomputersystems,modelingandanalysissystemstheory,andtheirapplicationtothedesignprocess.Toovercometheculturalandeducationalbarriers,thecommitteerecommendsthatinstitutionsofhigherlearningdevelopinterdisciplinaryprogramsledjointlybyexpertsinmaterialsscienceandengineering,design,andcomputersciencethat(1)exposestudentteamstobasicapproachestocomputer-assistedconcurrentengineeringdesignsystemsinordertoproduceknowledgeableworkerswithabroadunderstandingofthescienceofprocessing,(2)trainjourneymenormastertechnologiststousethisnewtechnologytopushacceptanceofprocessmodelingtechniquestotheshopfloor,and(3)encourageyoungerfacultymemberstocollaboratewithcolleaguesinotherdepartments(e.g.,materialsscience,thetraditionalengineeringfields,andcomputerscience)oninterdisciplinarydesignprojectsandcomputer-assistedconcurrentengineeringdesignsystems.

GENERALCONCLUSIONSANDRECOMMENDATIONS

TheareasinhibitingthedevelopmentandimplementationofCAMSSdiscussedabovecanonlybeovercomebyamultiprongedinitiativewithfullparticipationandsupportbytheintegratedproductdevelopmentteams(IPDTs)andmaterialsandcomputerscientistsandengineersinthegovernmentresearchanddevelopment(R&D)agencies,universities,andindustrialorganizations.Theimplementationofthisvisionwillrequire(1)thedevelopmentofsignificantdemonstrationsofCAMSSanddisseminatingtheresults;(2)thecontinuedexpansionofelectronicstorageofmaterialsinformation;(3)therapidadoptionandapplicationofdevelopingmethodsofcomputerscienceandtechnologytoenhancetherepresentationofmaterialsdesignknowledge;(4)thecontinueddevelopmentofmultilevel(atomistictomacroscopic)materialsprocessingandconstitutivebehaviormodelsthatreliablypredict

Page 35: Computer-Aided Materials Selection During Structural Design

performanceandmanufacturabilityatthescaleofapplication;and(5)theimplementationofmethodstoaddressinspectability,reliability,andmaintainability.

Adherencetouniformcomputingandmaterialsdescriptionstandardsinsuchprogramsisessentialtothenetworkedlinkingofindividualtoolsintomuchlargerdesignknowledgeandsupportsystemsinthefuture.Thecommitteerecommendsahigherlevelofcommunicationamongmaterials-specificinformationsystemsresearchersanddevelopersthroughamoreformalelectronicinterchangeofresearchinformation,network-linkeduseofcomputer-aidedsystemtools,andaccesstoelectronicmaterialsknowledgebases.

RecommendationsspecifictodevelopersandusersofCAMSSare

Governmentpolicymakersshouldpromote(1)thedevelopmentofpre-competitiveR&Dprogramsthatencourageindustry,university,andgovernmentlaboratoriestoleverageexpertiseandknowledgetoreducethetimetodevelop,standardize,andimplementproductdesignsupportsystemsandmaterials-specificinformationtechnologiesand(2)theuseoftheinformationsuperhighwayasameansforexpeditingthesharingoftechnicalinformationandmemoryamongfederalagencies,industries,andmaterialssocieties.GovernmentR&Dorganizations(DepartmentofDefense,AdvancedResearchProjectsAgency,NationalAeronauticsandSpaceAdministration,NationalInstituteofStandardsandTechnology)shouldpromotedatabaseandknowledge-baseconstructionandstandardization,design-knowledgetooldemonstrations,andpilotprojectsaspartoftheirfuturesystemsprograms.Theseprogramsshouldintegrateexistingcomputer-aidedsystemtools.Twopotentialwaysinwhichthismightbeaccomplishedaretoprovide(1)fundingfordemonstrationprogramswithcreativeproblemsolvinganddesignconceptstoteamsofuniversityfacultyandstudentscomposedofcomputerscientists,engineeringdesignspecialists,materialsscientists,andcognitivepsychologistsand(2)financialincentivestoindustryforsharingmaterialspropertydata

Page 36: Computer-Aided Materials Selection During Structural Design

(2)financialincentivestoindustryforsharingmaterialspropertydatawhereinputtopublicandlimitedaccessmaterialsknowledgebasescanbecontrolled.Industriesanduniversitiesshouldbeencouragedtocollaboratein:

Page 37: Computer-Aided Materials Selection During Structural Design

Page5

1.developingandusingwell-definedstandardsforelectronicinformationsharingtoenableselectiveprotectionoforganizationalprivatedata,companyproprietarydata,andindustryrestricteddatafromthepublicdomaindata;

2.improvingcontactbetweenresearcher,designer,andsupplierondesignteams;

3.increasingrateofgeneration,validation,andexchangeofmaterialsdata;

4.developingpowerfulprogramsforservicelifepredictionofstructuralcomponentsfrommaterialsdata,constitutivemodels,andin-servicenondestructivetesting;

5.developingmodelsofpracticalsignificancetoproductdevelopment;6.providingmaterialsdevelopmentdatainmachinereadableelectronicformat;

7.preparingstandardsforknowledgerepresentationofmaterialsinformation(e.g.,propertiestables,graphs,andpictorialdescriptionsofmicrostructures);

8.publicizingsuccessstorieswhereexperiencedengineersselectmaterialsshowingthatproperrepresentationstogetherwithreasoningexampleswillpromoteeffectivematerialcomputer-aidedsystemsdevelopment;and

9.developinganinformationbaseonavailable(networkaccessible)materialsdatabasesandcomputer-aidedsystems.

Page 38: Computer-Aided Materials Selection During Structural Design

Page7

1IntroductionTheselectionofthepropermaterialsforastructuralcomponentiscriticaltoengineeringdesign.Existingdesignproceduresmaycurrentlybesufficient,especiallywhereexperienceexists,butfierceindustrialcompetitionisspurringthesearchforimprovedmethodsandtools.Themaindriversarequality,life-cyclecost,andtime-to-market.Improveddesignefficiencyandaccuracymayhaveanenormousimpactontheeconomicviabilityofthefinalproduct.

Materialsselectionisgovernedbymanyfactors,someofwhichareofteninopposition.Theprincipalselectionfactorsincludetheservicerequirementsanddesignlifeoftheproduct;theavailabilityofcandidatematerialsandtheappropriatedataonapplication-specificpropertiesforthem;thecompany'smakeorbuydecisionforthesystemcomponents;thecustomerpreferences;andmostimportantly,thetotallife-cyclecost.

BENEFITS

Manydesignsinitiallyfailbecauseofalackofrelevantexperienceorbecausethedesignteamdidnotincludeappropriateexpertswho"couldhavetoldusso."Intheend,thereworkassociatedwithdesignrequalificationsignificantlyincreasescostandtime-to-market.Thus,theuseofcomputer-aidedsystemsthatassistdesignteamscouldpotentiallyreduceproductcostandtime-to-market.Computer-aidedsystemsformaterialsselectioncouldassistconcurrentengineeringactivitiesbyhelpingtoresolvemanyofthematerialsdilemmaspresentedduringtheinitialdesignphaseandbyhelpingtoguidetheselectionprocessbasedonthedataandexperiencecompiledfrompreviousproductdevelopment.Advancedcomputertechnologieswouldalsomakeitfeasibletoarchivedesignexperienceascasesinacorporateknowledgebaseforsubsequentre-use,tailoring,andevolution.

Page 39: Computer-Aided Materials Selection During Structural Design

baseforsubsequentre-use,tailoring,andevolution.

Thedevelopmentofacomputer-aidedsystemtosupportmaterialsselectioncouldalsoacceleratethegeneralacceptanceofnewmaterialsandprocessingtechnologies.Thequalityandefficiencyofthematerialsselectionprocesswouldbeenhancedbyincreasingaccesstoknowledgeoffactorssuchasmaterialsoptionsandlife-cyclecosts.Forinstance,designerscouldbeprovidedwitharangeofpossiblematerialsandmanufacturingmethodsforaproposedpart,basedonagivensetofcharacteristicsandcost-performancecriteria.Thus,themembersofthedesignteamwouldnotbetotallyreliantontheirownpersonalexperienceandlimiteddesign-handbookinformationduringthematerialsselectionprocessbutwouldhaveaccesstoinformationonpromisingnewmaterialsandprocessingtechnologiesthatcouldbeexploited.

Computingtechnologyisnolongerabarriertothedevelopmentofcomputer-aidedsystemsformaterialsselection.Advancesinreducedinstructionsetcomputer(RISC)chiptechnologyalreadyallowhigh-performance,inexpensiveworkstationstoperformdesignlayout,structuralanalysis,andmaterialsprocessingsimulation.Itwasgenerallybelievedintheearly1980sthattheuseofadvancedmodelingtechniques,suchasthree-dimensionalmodeling,wasnotpracticalbecauseofthelargeamountofcomputertimerequiredforanalyticalsimulations.Sincethen,computerspeedshavedramaticallyincreased.Accuratemodelingandsimulationofaunitprocessiscurrentlybecomingthenorm.Arangeofnewcomputerproductsarenowavailablethatenablethedevelopmentofcomputer-aidedsystemsformaterialsselection:

high-performancemicrocomputers;high-performanceworkstations(minicomputers);workstationclusters;RISCparallelsystems(e.g.,16computerprocessingunits);mainframe-workstationnetworks;

Page 40: Computer-Aided Materials Selection During Structural Design

vectorsupercomputing;andmassivelyparallelcomputing.

Thewiderangeofhardwarecapabilitieswillsoonbringthecostofimplementingcomputer-aidedsystemlogicandprocesssimulationwithinaffordablelimits.The

Page 41: Computer-Aided Materials Selection During Structural Design

Page8

continuedevolutionofcost-effective,high-performancecomputinginconjunctionwithanationalinformationsuperhighwayinfrastructurewillfurtherassistthenation'smanufacturingsectorinbecomingmorecompetitiveintheinternationalmarketplace.

DEFINITIONS

Computer-aidedsystemsarebroadlyinterpretedinthisreportasadvancedcomputingtechnologiesthataccessvariousmodulestoprovidespecificinformationwhenrequestedbyuserinput.Acomputer-aidedsystemhasthreeprimaryelements:(1)aninterfacewiththeuser,(2)areasoningelementthattriggerssystemaction,and(3)aknowledgeelementintheformofdatabases,knowledgebases,andmodelingmodulesthatprovidetheinformationandanalysestobeapplied.Computer-aidedsystemsformaterialsselectionindesignwillcontainonlyasubsetofthetotalproductknowledgeappliedduringdesign.

Knowledgerepresentationistheforminwhichfactsandrelationshipsareencodedandstoredintheknowledgecomponentofacomputer-aidedsystem.Knowledgerepresentationservesfivedistinctroles.First,knowledgerepresentationisasurrogateforknowledge.Second,itservesasetofontologicalcommitmentsortermswithwhichacomputer-aidedsystemcanreason.Third,itisapartialtheoryofintelligencethatexpressesthefundamentalconceptofintelligence,theinferencesthatarepossible,andtheinferencesthataremade.Fourth,itisamediumforpragmaticcomputations.Fifth,itisamediumforhumanexpression(Davisetal.,1993).

Concurrentengineeringis"asystematicapproachtotheintegrated,concurrentdesignofproductsandtheirrelatedprocesses,includingmanufactureandsupport,[that]isintendedtocausethedevelopers,fromtheoutset,toconsiderallelementsoftheproductlifecyclefrom

Page 42: Computer-Aided Materials Selection During Structural Design

conceptionthroughdisposal,includingquality,cost,schedule,anduserrequirements"(Winneretal.,1988).Theprocessofconductingdesigntradeoffscanbedonesequentiallyorinparallel.Inthisreport,thecommitteefocusesoncomputer-aidedsystemsthatcansupportadesignprocessinwhichdesigndecisionsaremadeinparallel,orconcurrently,byseveralmembersofadesignteam.Whenthedesignteamcontainsmembersthathaveaccesstoallknowledgepertinenttothecreationoftheproduct,itsuse,anditsultimateretirement,thatteamiscalledan"integratedproduct-developmentteam"(IPDT).

Topracticeconcurrentengineeringeffectively,allknowledgerelatedtothemanufactureofacomponentanditsmaintenanceinadeliveredsystemmustbeavailabletotheIPDT.Life-cycledataandexperienceknowledgeisthusanimportantprerequisiteforthefullapplicationofcomputer-aidedsystemsinwhichdesignchoicesareevaluated.Becauseoftheirimportance,theestablishmentoflife-cycledatabasesisnowrequiredbythemajorDepartmentofDefenseinitiativeonComputer-AidedAcquisitionandLogisticsSupport.1

STUDYOBJECTIVESANDSCOPE

Thisstudyconcentratesonthematerials-specificknowledgeelementsofacomputer-aidedsystem.TheCommitteeontheApplicationofExpertSystemstoMaterialsSelectionduringStructuraldesigndeterminedthatthedevelopmentofgenericcomputer-aidedsystemsisalreadyreceivingagreatdealofattentionwithinthecomputersciencecommunity.Thebasicinformationrequirementsforacomputer-aidedsystemformaterialsselectionarereceivinglittleattentionwithinthematerialscommunity,however.Thus,thecommitteeassessedthatthisstudywouldhavethegreatestimpactifit(1)detailedthecapabilitiesrequiredforcomputer-aidedsystemstobeofvaluetothematerialsselectionprocessduringconcurrentengineering,(2)identifiedtheissuesinhibitingthedevelopmentofsuchasystem,and(3)recommendedmaterials-specificapplicationsanddevelopmentsin

Page 43: Computer-Aided Materials Selection During Structural Design

database,knowledgebase,andmaterialsmodelingthatwouldaidtheproductionofaknowledgeelementappropriateforcomputer-aidedsystemsformaterialsselectionduringdesign.

Duringthestudy,thecommitteeexaminedengineering-relateddesigndecisionsinvolvinggeometryorspatialrelationshipstodeterminehowdesignrulesincorporatedmaterialsdataandhowassessmentsofperformance,

1Computer-AidedAcquisitionandLogisticsSupportisathree-phaseprogramthatrequires:(1)theadherencebycontractorstodataexchangestandards;(2)thelinkingofcontractorandgovernmentagencysystemsdatabaseswithstrongemphasisondemonstrationofconcurrentengineeringduringsystemdesign;and(3)thedevelopmentandautomationofdesignknowledgebases(DOD,1986).

Page 44: Computer-Aided Materials Selection During Structural Design

Page9

processing,manufacture,andsupport(e.g.,reliabilityandmaintainability)weremade.Thestate-of-the-artcomputer-aidedsystemsthatassistthematerialsoptimizationordesigntradeoffprocesseswerereviewed.Thestudyalsoconsideredtheuseofmaterialsmodelingsinceitcanpotentiallyreducethequantityofdatarequiredbythesystem.Thecommitteeusedcasestudiestoprovideinstancesofdesigndecisionsinvolvinggeometryorspatialrelations;designrulesassociatedwithperformance,processing,manufacture,andsupportability;andadvancedcomputertechnologiesandconceptsthataidintheoptimizationordesigntradeoffprocess.Thecommitteerecommendedtheresearchanddevelopment(R&D)thatisrequiredinmaterials-specificdatabases,knowledgebases,andmodelingtofacilitateconcurrentengineeringdesign.

Thecommitteecomposedthisreportforthestudy'ssponsors:theDepartmentofDefenseandNationalAeronauticsandSpaceAdministration.However,thecommitteealsorecognizedthatthereport'saudienceincluded:

structuralengineers,materialstechnologists,andcomputertechnologistsinthesponsoringgovernmentR&Dagencieswhofundresearch,reducetechnologicalbarrierstoagencyprojectsandmissions,andenablethetransitionoftechnologyintoproductsandprocesses;structural,materials,andcomputerscientistsandengineersengagedinuniversityandindustrialR&Dwhostrivetoinnovateinordertoovercometechnologicalbarriers,demonstratetechnologicaladvancements,andenablethetransitionoftechnologyintoproductsandprocesses;andproductdesignteamswhoaspiretomaximizethequalityofthedesignprocessandtheresultantvalueoftheproducttotheultimatecustomer:theproduct'suser.

Page 45: Computer-Aided Materials Selection During Structural Design
Page 46: Computer-Aided Materials Selection During Structural Design

Page11

2MaterialsSelectioninStructuralDesignThischapterdiscussesthestructuraldesigncontextformaterialsselection,thematerialsselectionprocess,theevolutionofcomputersystemsthatsupportthedesignprocess,andtheneedsformaterialsinformation.Thecommitteechosetodescribethestructuraldesignprocessusingaerospacevehicledesignasacasestudy.Thischapterisbased,inpart,oninformationgatheredbythecommitteeduringatwo-daysitevisittotheBoeingCommercialAirplaneGroup1andcomparativedataontheGrummandesignprocess.However,thedesignprocessandinformationneedsdetailedherearegenericandapplicabletostructuraldesignsinmanyindustries(e.g.,buildings,bridges,oilrigs,automobiles,ships,andspacecraft).

CONCURRENTENGINEERINGANDDESIGNORGANIZATION

Thedesignanddevelopmentofastructurelikeanaircraftisenormouslycomplex.Theoriginalsequentialapproachtoaircraftdesignwastobreakthestructureandsystemsintomanageablesections.Preliminarydesignsofeachsectionwerethenevaluatedsequentiallyandmodifiedbyamultitudeofdifferentengineering,manufacturing,quality-assurance,andoperations-supportexperts(Figure2-1).Thissequentialapproachledtoextensivechangesanderrorsduringandfollowingthedesignprocess,problemswithcommunicationsbetweenthedifferentdisciplines,increasesindevelopmentcosts,andextensionsindesignandmanufacturingschedules.Consequently,theamountofneededreworkandredesignaccountedforasignificantproportionofproductioncosts.

Theconcurrentengineeringapproach,supportedbycentralizeddigitaldatabasesforgeometry,materials,fabrication,andassemblyprocesses

Page 47: Computer-Aided Materials Selection During Structural Design

andpaperlessdrawings,wasproposedtoimprovethedesignprocessandreducereworkandredesign(Winneretal.,1988).Figure2-2comparesthesequentialandconcurrentengineeringapproaches.

Boeinghasimplementedconcurrentengineeringthroughanapproachthatusesdesignbuildteams(DBTs).TheDBTapproachestablishesanIPDTfordesigningnewproductsandsystemsandexecutingaconcurrentengineering/manufacturingprocess.Theteamgoalsaretoproduceerror-freedesignsthatareoptimizedintermsofperformance,weight,andproductionandoperatingcosts.

Figure2-1Thesequentialengineeringapproachtostructuraldesign.ReprintedcourtesyofRNHAssociates,Incorporated.

1TheBoeingsitevisitcomprisedtestimonyfrommaterialsandstructuresexperts,designmanagers,andautomatedsystemsspecialistswhowereimprovingthedesignprocessviaorganizationalandtechnologicalinnovations.(Note:TheselectionofBoeingasacasestudyshouldnotbeinterpretedasastatementthattheirdesignprocessissuperiortothoseofothercompanies.)

Page 48: Computer-Aided Materials Selection During Structural Design

Page12

Figure2-2Atypicalcomparisonofsequentialandconcurrentengineering.VariationsofthisillustrationarepresentedinWinneretal.(1988),

Whitneyetal.(1988),andNRC(1991).

ThefirststepintheDBTapproachistodividethesystemsintomajorcategories(e.g.,structure,avionics,flightcontrols,mechanicalsystems,environmentalsystems,hydraulics,flightdeck,andpayload)aswellasgenericintegrationareas(e.g.,airworthiness,reliability,andmaintainability).Thesecategoriesarethenfurthersubdivided.Forinstance,thebasisstructuredivisionsarebody,wing,empennage,andpropulsionsystem.These,inturn,aresubdividedevenfurtherintomanageablecomponentsandsubcomponents,eachofwhichistheresponsibilityofaseparateDBT.Forexample,themainbodycomponentsarecockpit,forwardsection,centersection,rearsection,andtailfuselage,aswellasdoors,doorcutouts,floors,andfloorbeams.AtypicalhierarchicalrelationshipbetweentheIPDTandtheDBTsisshowninFigure2-3.

BoeinginitiallyimplementedtheDBTsysteminanattempttoremaincompetitiveintheglobalmarketplace(NRC,1993).TheBoeing777programpeakedatatotalofapproximately250DBTs,including97DBTsrelatedtostructures.

ThestructuralDBTs(Figure2-4)arecomposedofdesign,structures,materials,manufacturing(e.g.,toolingandmachining),qualitycontrol,andcostanalysisexperts;someteamsalsoincludeliaisonrepresentativesfromkeysubcontractors.Additionalsupportmayalso

Page 49: Computer-Aided Materials Selection During Structural Design

beprovidedasrequiredbyrepresentativesfromothercompanydivisionsorbyspecialistsonapart-timebasis.Theteammembersfromthevariousdisciplinesresponsibleforcreatingaspecificcomponentorsubsystemfromconceptionthroughfinaldesignarecollocated,andeachteammemberisexpectedtoparticipatefullyintheDBTdecision-makingprocess(BoeingCommercialAirplaneGroup,1991).Afterthetotaldesign(includingtooldesign)iscompleted,manufacturingisempoweredtoreviewandapproveengineeringdatasheetsverifyingproducibilitypriortodrawingrelease.AsimplisticrepresentationoftheinteractionswithinatypicalDBTisshowninFigure2-5.Animportantaspectoftheprocessindicatedinthefigureisthatoptimizationdecisionsaremadefromtheperspectiveoftheentiresystem,notfromthatofaparticularsubsystem.

Althoughconcurrentengineeringhasconsiderablyreducedrework,structuraldesignandmaterialselectionremainiterative,cyclicalprocesses.Structuralanalysesareperformedoncandidatepreliminarydesign,andmodificationsaremadetosatisfystructuralrequirements.Weightandcostestimatesareusedfortradeoffstudiesto

Figure2-3AnexampleofatypicalstructuresDBThierarchy:Boeing777HorizontalStabilizerDBTs.Source:

BoeingCommercialAirplaneGroup.

Page 50: Computer-Aided Materials Selection During Structural Design
Page 51: Computer-Aided Materials Selection During Structural Design

Page13

Figure2-4Theexpertiseinastructuraldesignconcurrentengineeringteam.ReprintedcourtesyofRNHAssociates,

Incorporated.

identifyandselectthebestmaterialsanddesignconfiguration.TheDBTapproachalsoaddressesmostoftheinter-disciplinaryproblemsassociatedwithcandidatedesignconceptsandmaterialselectionearlyinthepreliminarydesignphase.CompromisesolutionscanthenbeidentifiedandselectedbytheDBTmembersbeforethecompletedesignisfinalized.

EachDBTrecordsteamnotes,memoranda,andsummariesofprojectreviewsinDBTlibraries.ThesecanbeaccessedbyotherDBTstoobtaininformationanddigitaldesigndata.Thisallowsrapiddisseminationofchangesthataffecttheinterfacebetweencomponents,facilitatestradeoffsusingglobalcriteria,andensuresstorageoflessons-learneddataforfuturedesigns.Theserecordsareprimarilyfoundinhard-copyform.Althoughsomearefiledelectronically,theyarenotavailableforon-linereference.

MATERIALSSELECTIONDURINGCOMPUTER-AIDEDDESIGN

Page 52: Computer-Aided Materials Selection During Structural Design

ADBTteamrequiresanenormousamountofdetailedinformationtodevelopstructuresthatwillsatisfyperformance,reliability,safety,weight,anddurabilityrequirementsateconomicalproduction,operation,andmaintenancecosts.Inthe1960s,structuraldesignandanalysisconsistedofslide-ruleandadding-machinecalculationsusingformulaeandtablesfromhandbooksincombinationwithnumerousassumptionsbasedonpriorexperience.Theresultingdesignswerethenevaluatedbymaterials,manufacturing,andcost-estimatingpersonnelwhofedbacktheirrecommendationsfordesignchanges.DesignandengineeringoperationsarecurrentlyperformedrapidlyandaccuratelybyDBTmembersusinginteractivecomputer-aidedengineeringorcomputer-aideddesignandcomputer-aidedmanufacturing(CAD/CAM)programs.Manydifferentintegratedcomputer-aidedengineeringandCAD/CAMsystemsarecurrentlyavailable.Eventhemostadvancedofthesefocusonlyonfiniteelementanalysis(FEA)orboundaryelementanalysiscomputerprogramsandcurrentlyhavelittlematerialsselectionexpertsystemcapabilities.

Computer-aidedengineeringandCAD/CAMsystemsgenerallyuseamixtureofbars,panels,andsolids,whichareutilizedfrompreliminarydesignthroughdrawingrelease.Thestructureispredominantlymodeledusingcombinationsofbarsandpanelsforthestructuralanalysisandoptimizationprogramsbecauseofthesignificantlylongercomputertimesneededtomodelthestructureandcompletetheanalysisoroptimizationusingsolids.Solidelementsareonlyusedwhenthestructurecannotbe

Page 53: Computer-Aided Materials Selection During Structural Design

Figure2-5TheinteractionsofatypicalDBTduringinitialconceptdevelopment.Source:

BoeingCommercialAirplaneGroup.

Page 54: Computer-Aided Materials Selection During Structural Design

Page14

realisticallymodeledusingthesimplerelementsorwhenmoreaccuratedeterminationofthethree-dimensionalstateofstressorstraininthecomponentisneeded.Forexample,GrummanusedthemodelofawingfortheX-29andassociatedFEA(Figure2-6)incombinationwithafuselagemodeltodeterminetheloadsinthestructureandthedynamicandaeroelasticbehaviorofthewingrequiredtoprecludedivergenceandflutter.Aerodynamicallyinducedstructuraldivergencewasavoidedbydesigningthecarbon-epoxycoverstoprovidebending-twistingcouplingtothewing,takingfulladvantageoftheanisotropicpropertiesofthecompositematerial.Thismodelwasiterativelyappraisedbystructuralanalysis,weightoptimization,anddivergenceanalysiscomputerprogramstodeterminethegeometryandorientationofthecarbon-epoxytapeforeachofthe148pliesintheupperwingskinandthe158pliesinthelowerwingskin.Thesamemodelandcomputerprogramswerethenusedforselectionofthematerialsandthesizingofthecapareasandwebthicknessesfortheotherwingcomponents.AsshowninFigure2-6,thewingcoversarecarbon-epoxy.Theothermaterialsusedinthewingcomponentaresteel,6A1-4Vtitanium,2024aluminum,anwovenglass-epoxy(Hadcock,1985).

Three-dimensionalmodelsofforgingsormachinedpartsareusedformoredetailedanalysisandsizingofcomponents,suchascomplexwing-to-fuselageattachmentfittingsandcontrolsurfacehinges.ThesemodelspredicttheboundaryloadsandconstraintsfromtheoverallFEA.TheinformationfromtheseprogramscanbeelectronicallytransferredtoCAD/CAMsystemstogeneratethedrawingsofthedetailpartsandassembliesformanufacturingengineering.

Inalltheseprograms,materialpropertiesandexternalgeometryaregenerallyinputdata.Structuraloptimizationisdoneiteratively.Structuralgeometry,whichdependsonmaterialproperties,panelthicknesses,andstiffenersizes,

Page 55: Computer-Aided Materials Selection During Structural Design

Figure2-6AmodelofthewingoftheGrummanX-29andassociatedFEA.Source:NorthropGrumman.

Page 56: Computer-Aided Materials Selection During Structural Design

Page15

canbeautomaticallyadjustedduringtheiterations.Theeffectsofchangesinmaterialsselectioncanbeevaluatedbyexecutingtheprogramswithdifferentmaterials-propertiesdatasets.Thisdesigntradeoffanalysisprocesscanbeverytimeconsuming,particularlywhentherearelargenumbersofcandidatematerialsforeachpartandarangeofstructuralanalysistests,suchasthermalstrains;dynamicbehavior;fatigue;fracture;durability;and,inthecaseofcombataircraft,survivability.However,optimizationprogramsareemergingthatwillallowtheselectionofbestchoicesgiventheconstraintparametersspecifiedbythedesignengineer.

Theaerospaceindustryhastraditionallyadoptedarigorous,yetconservative,materialsselectionprocesstominimizetheriskassociatedwiththeintroductionofnew,andthereforeless-provenmaterials.RiskasafactorinmaterialsselectionwillbediscussedinChapter3.

Someintegratedcomputerprogramsareavailablefordesign,structuralanalysis,andproductionofcomplex-shapedcastingsandinjection-moldedplasticparts.Theseprogramsincludethermalandflowanalysisoftheliquidmaterial,designofpatternsandmoldsthatmayincludecoolingpassagestoeliminatedistortionandcrackingduringcooling,anddeterminationofresidualstrains(seeAppendixB).

SUMMARYOFMATERIALSINFORMATIONREQUIREMENTSINDESIGN

Table2-1providesapartiallistingofmaterials-relatedinformationthatisneededinthematerialsselectionprocess.Materialsselectionisstronglyinfluencedbyoverallproductdesign,manufacturing,andcostrequirements.SomeoftheproductdesignrequirementsforaircraftstructuraldesignarepresentedinTable2-2.

Page 57: Computer-Aided Materials Selection During Structural Design

ThemajorstructuresandmaterialsdesigninteractionsareshowndiagrammaticallyinFigure2-7.Materialselectionisdirectlyorindirectlydefinedbythecombinationofthesedesigninteractions.Theseinteractionsincludemostoftheinformationneedsofateamtodesignandselectthematerialsforaprimarystructurecomponent.

AsummaryofdesignerwantspertinenttotheapplicationofexpertsystemsinthematerialsselectionprocessduringstructuraldesignislistedinTable2-3.Thistableprovidesthebasisforestablishingtherangeofinformationtechnologiespertinenttothematerialsselectionprocessthatwillbeassessedinthenexttwochapters.

Figure2.7Structuresandmaterialsinteractions.ReprintedcourtesyofRNHAssociates,Incorporated.

Page 58: Computer-Aided Materials Selection During Structural Design

Page16

Table2-1ExamplesofMaterialsInformationRequiredDuringProductDesignMaterialidentificationMaterialclass(metal,plastic,ceramiccomposite)MaterialsubclassMaterialindustrydesignationMaterialproductformMaterialconditiondesignation(temper,heattreatment,etc.)MaterialspecificationMaterialalternativenamesMaterialcomponentdesignations(composite/assembly)MaterialproductionhistoryManufacturabilitystrengthsandlimitationsMaterialcomposition(s)Materialcondition(fabrication)MaterialassemblytechnologyConstitutiveequationsrelatingtopropertiesMaterialproperties&testproceduresDensitySpecificheatCoefficientofthermalexpansionThermalconductivityTensilestrengthYieldstrengthElongationReductionofareaModuliofelasticityStressstraincurveorequationHardnessFatiguestrength(definetestmethods,load,andenvironment)Temperature(cryogenic-elevated)Tensilestrength,yieldstrengthCreeprates,rupturelifeatelevatedtemperaturesRelaxationatelevatedtemperaturesToughnessDamagetolerance(ifapplicable)Fracturetoughness(definetest)

Page 59: Computer-Aided Materials Selection During Structural Design

Fatiguecrackgrowthrates(defineenvironment,andload)TemperatureeffectsEnvironmentalstabilityCompatibilitydataGeneralcorrosionresistanceStresscorrosioncrackingresistanceEnvironmentalstabilityToxicity(atallstagesofproductionandoperation)Recyclability/disposalMaterialdesignpropertiesTensionCompressionShearBearingControlledstrainfatiguelifeProcessabilityinformationFinishingcharacteristicsWeldability/joiningtechnologiesSuitabilityforforging,extrusion,androllingFormability(finishedproduct)CastabilityRepairabilityFlammabilityJoiningtechnologyapplicableFusionAdhesivebondingFastenersWeldingparametersFinishingtechnologyapplicableImpregnationPaintingStabilityofcolorApplicationhistory/experienceSuccessfulusesUnsuccessfulusesApplicationstobeavoidedFailureanalysisreportsMaximumlifeservice

Page 60: Computer-Aided Materials Selection During Structural Design

MaximumlifeserviceAvailabilityMultisource?Vendors?SizesFormsCost/costfactorsRawmaterialFinishedproductorrequireaddedprocessingSpecialfinishing/protectionSpecialtooling/toolingcostsQualitycontrol/assuranceissuesInspectabilityRepairRepeatability

Table2-2TypicalProductDesignRequirementsforAircraftStructureDevelopmentPerformanceDesignloadsandconditionsAssociatedairloadsandaccelerationsFuelusageCabinandcargoholdloadingsTemperaturesandassociatedenvironmentaldataFatiguespectraFail-safeandsafe-lifedesignAeroelasticityrequirementsAirworthinessstandardsanddesignrequirements(FederalAviationAdministration:FederalAviationRegulations,AdvisoryCirculars,etc.)CostDesignProductionPreparationsMaterialhandlingSafetyEnvironmentalandwastedisposalInterfacesGeometricaltolerances

Page 61: Computer-Aided Materials Selection During Structural Design

GeometricaltolerancesStructuralassemblySurfacesmoothnessandtolerancesAvionicsPropulsionEnvironmentalcontrolPassengeraccommodationsTestingLoad-temperature-environmentspectrafatigueQualityRepairandreinspectionAutomatedandnonautomatedquality-controlequipmentVendor/supplierqualificationfornewmaterialspartfabrication

Page 62: Computer-Aided Materials Selection During Structural Design

Page17

Table2-3SummaryofDesignerWantsDesignToolsMaterial/processing/manufacturingtradeoffsinconceptdesignCompositematerialsstructuresdesigntoolsQualitymaterials-selectionaidsDesignKnowledgeInformationonthecompetitionLessons-learnedknowledgebaseMaterials-usecasebaseindexedbymultipleattributesCostKnowledgeCostmodelsLife-cyclecostsManufacturingcostsMaterialpricesDesignCycleTime/TimetoMarketMoretradeoffsconsideredingiventimeIterationforrealisticmaterialstargetsReducecycletimetomarketRapiddeploymentofnewmaterialRiskReductionTrusteddesignandmaterialsdataReducedriskinselectingnewmaterials/processesProductionCapabilitiesFacilityavailabilityEquipmentavailabilityWorkforceexperiencecapabilityandavailabilityViablesupplieroptionsExpertAgentsGatherpertinentdesigninformationfrommultiplesourcesSpecificexpertsystemsforeachcomponentdesignteam

Page 63: Computer-Aided Materials Selection During Structural Design

Page19

3EnhancingtheMaterialsSelectionProcessinDesign:AVisionToidentifytheinformationtechnologiesrequiredforacomputer-aidedsystemtosupportmaterialsselection,thecommitteearticulatedafuturevisionofafull-functionComputer-AidedMaterialsSelectionSystem(CAMSS)basedontheinformationsummarizedinChapter2.Inthefuture,materialsselectionisenvisionedinabusinesscontextthathasseveralmajordifferencescomparedtocurrentenvironments.Theengineeringdesignprocessisevolvingfromastageofemphasisonconcurrentengineering(i.e.,thesimultaneousdesignofproductsandmanufacturingprocesseswithinacompany)tooneonconcurrententerpriseprocessing(i.e.,thesimultaneousdesignofproductsandprocessesthattakesintoaccountinternalandexternalpartnering,preferredsupplierrelationships,andcorporatealliances).Thereisanever-increasingpressureforaccuracy,flexibility,speed,andcompetitiveleadership.Theinformationsystemssupportingtheconcurrententerpriseprocesswillbemoreubiquitous,powerful,andintegratedintothebusinessprocess.

Asaresultofthesechanges,theimpactonmaterialsselectionissubstantial.Materialsselectionwillbebasedonamuchbroaderrangeofconcernsandnotonisolated,sub-optimizedsteps.Theconcurrententerpriseprocessdemandsthatmaterialselectionisnotonlybroadbased,butdonefast,right,atthecorrecttime,andonce.Softwaretosupportmaterialsselectionwillbepartofanintegratedcomputingenvironmentthatspanstheconcurrententerpriseprocessandmakesuseofembeddedassistanceformanyaspectsoftheproductlifecycle.

Amidsttheevolutionofthebusinesscontext,materialsselection

Page 64: Computer-Aided Materials Selection During Structural Design

continuestooccurintwoforms:strategicmaterialsselectionandroutinematerialsselection.Therearenosharpdistinctionsbetweenthese,butstrategicdecisionsareprimarilyinresponsetocorporateobjectives,high-visibilitycustomerrequirements,orstrategictechnologyplanning.Theintroductionofnewmaterialsorprocessesforaparticularproductapplicationisnearlyalwaysastrategicdecision.Suchdecisionsarestrategicbecauseofthetimerequiredandcapitalcostsassociatedwithvalidationandinvestmentinnewprocesscapabilities.

Thepressuresforincreasedagilityinresponsetoglobalcompetitionisamixedblessingfortheintroductionofnewmaterials.Thecompetitivepressureplacesdemandsonleadershipbutstripsawaythetimetoreact.Theadvancedcomputingsupportforstrategicandroutinematerialsselectiondiffersbutsharesacommoninfrastructure.Thefollowingfoursectionsexplainthevisionforthiscommonbusinessandinformationprocessingenvironment,discusstheuniquecapabilitiesrequiredforstrategicandroutinematerialsselection,andexaminethebasisforinnovativematerialsselectionindesign.

INTEGRATEDENGINEERINGSUPPORTININTEGRATEDENTERPRISES

Enterprisesconsiderationarecausingorganizationstochangethewaytheyviewthemselves.Newrelationshipswithinternalandexternalunitsareemerging.Theshifttoanemphasisonconcurrentengineeringisevidenceoftheshiftthatconcentratesoninternalandexternalpartnering.Externalpartneringhasledtofavoringpreferredsupplierrelationshipsoverlow-bidcompetition.Technology''food-chains''arebeingaddressedwithcorporatestrategiesforstrategictechnologyplanning.Allianceswithtechnologysuppliersareincreasing.

Asmorecooperationemergesbetweenunits,moreunifiedcommunicationandcomputingenvironmentsarerampinguptomeet

Page 65: Computer-Aided Materials Selection During Structural Design

theneed.Increasedemphasisonstandardsgivesevidencetothisshift.Forexample,theInitialGraphicsExchangeStandard,whichisusedtoexchangegeometry,isexpectedtobesupplantedbytheevolving,internationalStandardfortheExchangeofProductDefinitionData,thegoalofwhichistheexchangeofcomplete,unambiguouscomputer-interpretabledefinitionsofthephysicalandfunctionalcharacteristicsofaproductthroughoutitslifecycle.Theseshiftsprovidean

Page 66: Computer-Aided Materials Selection During Structural Design

Page20

infrastructuretosupportfutureconcurrententerpriseprocesses.Inthemeantime,computinghardwareandsoftwarecapabilitiescontinuetoevolvewithanemphasisonintegratedcomputingenvironmentsandopensystems.Partnershipsbetweensoftwaresupplierspermitengineeringorganizationstoconsiderasuiteofapplicationsthatcollectivelycoversubstantialacreageintheart-to-partlandscape.Standardsinuserinterfacetechnology(e.g.,X-windows)arebreakingdownconceptualbarriersbetweencomputerapplications.Inafewplacescurrently,andmoresointhefuture,theengineerissupportedbyacomputingenvironmentfortherapidtransmissionofshareddatathatlinkstootherengineeringandmanufacturingorganizationsbothwithinthecompanyandwithsuppliersandvendors.

Inthefuture,functionalcapabilitiesofsoftwarewillbeintegratedsothatconceptualdesignalternativescanbedevelopedandevaluatedforanynumberofcriteriaduringearlyproductplanningphases.Thealternativeselectedinthisprocessmusthaveahighprobabilityofbeingmanufacturableatthetargetcostsnegotiatedbytheproductteam.Theremustbeanequallyhighprobabilitythattheproductmeetstherequirementsofthecustomerandisalignedwiththetechnologyplansoftheenterprise.

Thedevelopmentandevaluationofthesealternativesbydesignteamscouldbeassistedbyintegratedcomputer-aidedsystems.Theseknowledge-basesystemsarewovenintothecomputingframework.Becauseoftheintegrationoftheenvironment,theydonotappeartousersasseparatesystemsbutratheraddtothefunctionalitythatthesystemprovides.Thus,fromtheuserperspective,knowledge-basetoolsareundifferentiatedfromanalysisanddesignautomationtools.

Typically,computer-aidedsystemsaretoprovidedesignadvice,leavingfinaldecisionstotheengineer.Theadvicegivenbythesystemcanbeassimpleasselectingadefaultmaterialspecification.Itcan

Page 67: Computer-Aided Materials Selection During Structural Design

telltheengineerwheretofindsuppliersofrelevantmaterialoritcanretrievethatmaterial.Itcananalyzeadesignandprovideaquantitativeorqualitativejudgment.Itcansuggestanimprovementorgenerateanalternativethatincludestheimprovement.Itcansearchavarietyofalternativesandsuggestthebestorthebestfewalternativesbasedonquantitativeorqualitativejudgmentsanduser-suppliedcriteria.Thealternativeselectedduringconceptevaluationisthenavailableforrefinementinacoarse-to-finedevelopmentprocess.Thedetailsofthecomputermodeloftheproductthenevolve,aidedbytheuseofanumberofdesignadvisortoolsthatprovidereferenceinformation,analyzeadesign,critiqueit,improveit,oroptimizeitforagivensetofdesignmetrics.Analysisanddesignautomationtoolsalsohelpintherefinementoftheproductdescription.

Availablethroughouttheprocessareadviceandknowledgeaboutmaterials.Significantmaterialpropertiesaswellasemergingconsiderations,suchaslife-cyclecosts,aretobeincreasinglysupportedbytheinformationinfrastructure.Materialsknowledgeistobeaccessibletotheengineerasreferencematerialthroughdesignadvisorsthatinteractwiththeuseraswellasproductandprocessmodelstoanalyze,critique,improve,oroptimizeit.Majortoolsintheintegratedenvironmentshouldprovidethefollowingcapabilities:managingelectronicrepositoriesofdataanddocuments;searchingthroughpastdevelopmenthistoriestofindsimilaroranalogousproducts;managinginteractionswithotherpartsoftheenterprise;managingrequirements;predictingperformancecharacteristics;predictingmanufacturingandmaintenancecharacteristics;estimatingcosts;suggestingimprovementstotheproposedproductorprocessdescription;andreleasingmaterial,product,andprocessdescriptionstoothercomponentsoftheenterprise.

SUPPORTINGSTRATEGICMATERIALDECISIONS

Asindicatedearlier,strategicmaterialsselectionalmostalwaysoccurs

Page 68: Computer-Aided Materials Selection During Structural Design

fornewmaterialintroduction.Italsooccurswhenthereareseveralmaterialalternativesthatrepresentsignificanttradeoffsincriticalcustomerrequirements(e.g.,appearance,durability,cost,andrisk).

Inthefuture,materialsexpertsshouldusesimulationsofmaterialperformanceatbothmicro-andmacro-structurallevelstoreducethecostofmaterialvalidation.Materialmodelsmustincludemanufacturingprocessperformanceaswellasproductperformance.Majorcostsavingsarefoundinthereductionofdecision-timeandreworkrequired.Materialsupplierandusersmustregularlyjointogethertodevelopandspecifymaterialsandprocesses.

Strategicdecisionsaretobemadeusingformaldecisionmethodologiesandcomputertoolsforsupport.QualityFunctionalDeploymentandDecisionandRiskAnalysis

Page 69: Computer-Aided Materials Selection During Structural Design

Page21

aretwoexamplesofmethodssupportedbytools(seeChapter4).Materialsarethefocusofmanystrategicdecisionsbutonlyoneofmanyfactorsinfarmorestrategicdecisions.Tosupportthedecisionprocess,performanceandcostmodelsformaterialsandprocessingaretobeusedforstrategicdecisionsthatincludelong-rangebusinessplanning.

Strategicdecisionmakingcannotbehandedovertocomputers;rathercomputersandinformationsystemsmustbereliedontoprovideaccesstodocumentedinformationandmodelsofthebusiness,product,processingcapabilities,andprocessinginfluencesonmaterials.Theycanalsomanagethecomplexityofrelateddecisionvariablesandkeeptrackofalternativesthatareunderconsideration.Broaderaccesstosuchinformationcontributestoabetterunderstandingandquantificationoftheriskfromtheintroductionofnewmaterials.

Itisimportanttorecognizethatmaterialsdesignfor"structurecritical"applicationstendstoberatherconservative.Designerscannotaffordtotakeunnecessaryriskswithnewmaterialsbuttheycangainexpertisewithprocessingandperformanceofnewmaterialsinnoncriticalordevelopmentapplications(e.g.,compositefishingrods,nitanoleyeglassframes,orceramicscissors).Suchexperienceisvitalingatheringdataandconfidenceforcriticalpurposeapplications.However,itisimperativethatfuturesystemsbeabletocollect,organize,anddistributesuchlessons-learnedexperience.

SUPPORTINGROUTINEMATERIALDECISIONS

Routinematerialdecisionshappeneverydayforeverycomponentdevelopedbytheenterprise.Thereisincreasingemphasisontheprocessformakingsuchdecisionstoassureconsistency,accuracy,andreliability.Computersystemscanprovideassistancetothe

Page 70: Computer-Aided Materials Selection During Structural Design

engineeringcommunitytofollowestablishedprocessesbutshouldnotlocktheuserintoarigorousframeworkthatstripstheuserofopportunitiestoexercisecreativity.

Thereispressuretoincludemorefactorsinalldecisions.Materialsselectionisinfluencedbyfactorssuchasmanufacturing;assembly;service;andenvironmentalimpactofmaterialproduction,use,anddisposalorrecycling.Computer-aidedadvisorscanhelpmanagethecomplexityofthemanyconcerns.Supportingtheproductteaminthematerialsselectionprocessareelectronicdocuments;costestimationtools;trade-studytools;material,product,andprocessdatabases;andknowledge-basesystemsthatprovideanalysis,critiques,andproductimprovementsuggestions.Materialsselectionfallswithinthescopeofsuchtools.

SUPPORTINGINNOVATIVEMATERIALSSELECTIONINDESIGN

Aprospectivecomputer-aidedsystemshouldalsobecapableofassistinginnovativedesign.Itshouldnotjustprovidealimitedseriesofconventionalmaterialorprocessingchoices.ThissectionaddressesthecharacteristicsofCADsupportsystemsforsolvingdifficult,nonroutinedesignproblems.TheconceptspresentedherearedrawnextensivelyfromtherecentpublicationbyStevenKimentitledTheEssenceofCreativity:AGuidetoTacklingDifficultProblems(Kim,1990).

Adesignproblemcanbecharacterizedbyitsdomain,difficulty,andsize.Domainreferstotheapplicationareaorareas,sizereferstotheamountofworkneededtoanalyzeandimplementthedesignsolution,anddifficultyreferstothelevelofconceptualchallengetoidentifyanacceptablesolution.Adifficultproblemisoneinwhichresolutionisnotreadilydiscernible.Designproblemscanbeill-structured.Theyaregenerallynotboundedbyalgorithmicmodelsandmaylackcompletesetsofheuristicstobeappliedtothedesignspace.

Page 71: Computer-Aided Materials Selection During Structural Design

Therefore,aninnovativedesignisthecreativeresolutionofadifficultproblem.

AcreativesolutionexhibitscertainfeaturesthatarecloseinconceptualspaceandothersthataremoredistantinthatspaceaconceptKimterms"theMultidistancePrinciple."Thoseaspectsofthesolutionthatareclosesttotheknowledgeorexperienceofthedesignteammaybeclearlyevident.Thoseaspectsthataremoredistantinthesolutionspacearetheonesthatoftenrequireinsightfulthinking.TheMultidistancePrinciplehasimplicationsforthedevelopmentofsoftwaretoolssuchascomputer-aidedsystemsthatenhancefindingcreativedesignsolutions.Thesetoolsmustbeabletoestablishlinkstooneormoreattributesofthedistantelementsofthedesignsolutionaswellasprovidingaccesstothemoreroutine,detaileddesignfeatures.

Page 72: Computer-Aided Materials Selection During Structural Design

Page22

Thereareseveralfactorsthatcontributetothegenerationofacreativedesignsolution.Theobjectiveofthedesignmustbedefinedanddistilledintoitselementsinordertobeginthedesignprocess.Theelementsofthedesignobjectivecanbeviewedasalogicalhierarchyofdesignalternativesanddecisionsregardingthealternatives(Weberetal.,1991).

Thecreativesolutionprocesshastwoelements:(1)astructure,characterizedbydiversityandrelationships,and(2)vehiclesforenhancingtheidea-generatingprocessinvolvingimageryandexternalization.Diversityreferstothefusionofdisparateideas(i.e.,theMultidistancePrinciple).Itcanbeaidedbymemoryenhancers,suchasaccesstoknowledgebasesandhistoricalarchives.Relationshipsdefinethepatternamongdesignspaceobjectsthatcanbediscoveredfromreferencetorelatedproblemsandrapidenumerationofalternativesolutions.Imageryisthegenerationofideasthroughsensoryimages(auditoryandtactileaswellasvisual).Themostpowerfulimageryforhumansisvisual.Itispossiblethroughimagerytorepresentnumerousobjectsandtheirrelationshipssimultaneously.Externalizationistheexpressionorcommunicationoftheideastoothersthroughtext,models,anddiagramsoftheprocess.Externalizationhelpstoclarifytheideaandisoftenthemostimportantstepinachievingacreativedesignsolution.Theuseofvirtualrealityisanexampleofthecombinationofimageryandexternalization.

Thecreativesolutionprocessstructurerepresentstheimportantingredientsforinnovativedesignandproblemsolving.Thestrategyforenhancinginnovationisthedevelopmentoftoolsthatpromoteorenhanceelementsofthisstructure,especiallymemory,imagery,andexternalization.Withthisstrategy,onecanidentifyseveraldomainsofknowledgethatcancontributetotheenhancementofinnovativedesign:

artificialintelligence(learning,inferencing,andknowledgerepresentation);

Page 73: Computer-Aided Materials Selection During Structural Design

representation);computerandnetworktechnologies(parallelprocessing,storagemedia,networkcommunications,andworkstations);humaninterfacetechnologies(graphics,vision,touch,animation,simulation,andvoiceorspeech);andcognitivepsychology(perception,memory,reasoning,andinsight).

Humanmemory,eitherinanindividualorwithinagroup,isbothastoreofarchivalknowledgeandworkareaforthedevelopmentandexaminationofdesignalternatives.Thecontentsofhumanarchivalmemory,enhancedbycomputerrecallofdetailsorrelatedconcepts,facilitatesthegenerationofnovelelementsofapossibledesignsolution.Theworkingmemoryoftheindividualorgroupisabasisbywhichtocraftthefullsolution.Imageryandexternalizationthataidinrepresentingvarioussolutionsarekeytobringingawiderangeofinformationtobearonthedesignproblemathand.Whilecomputer-aidedsystemshavebeenusedtoenhancelogical,rule-basedthinkingandneuralnetworkscanlearnperception,theelementofcognitivepsychologycalled"insight"isthekeytothediscoveryofcreativesolutionstodifficultdesignproblems.Computertechnologiesandtoolsmaynotbeabletoreplacehumaninsightbutcouldenhanceit.Thisareaneedsresearchemphasisasacriticalcomponentofdesigntechnology.

SUMMARY

Thematerials-specificinformationtechnologiesthatdesignersrequireinaCAMSSandsomeofthecomputertechnologiesthatareneededtobuildthissystemaresummarizedinTable3-1.Figure3-1specifiesthehigh-levelconceptualarchitectureandsomeofthecontentsofafull-functionCAMSSbasedonthevisionpresentedinthischapter.Thestateoftheartoftheinformationtechnologiespertinenttothematerialsselectionprocessisdiscussedinthenextchapter.

Page 74: Computer-Aided Materials Selection During Structural Design
Page 75: Computer-Aided Materials Selection During Structural Design

Table3-1SummaryoftheMaterials-SpecificInformationTechnologiesandSomeofthePrimaryComputerTechnologiesRequiredforaCAMSS

MATERIALSSELECTIONCAPABILITIESREQUIRED PRIMARYCOMPUTERTECHNOLOGIES

REQUIRED

RoutineMaterialsSelectionStandardselectionprocessforeverycomponentdevelopedthatconsistently,accurately,andreliablyfollowsestablishedprocedureswithouteliminatingopportunitiestoexercisecreativity.Systemrequirestoolsthatmanagethecomplexityofmanufacture,assembly,inspection,service,andenvironmentalimpactconsiderationsofmaterialproduction,use,anddisposal/recycling,andsuggestsproductimprovements.

MaterialsdatabasesandknowledgebasesElectronicdocumentationofpreviousdesignsHeuristicsandselectionreasoningtracebackCostestimationmodeling

StrategicMaterialsSelectionDecisionstointroducenewmaterialsbasedonunderstanding/quantificationofrisk,impactoncustomerrequirements,andcorrespondencewithenterpriseobjectivesandstrategictechnologyplanning.Systemshouldnotmakestrategicdecisionsbutshould(1)providedesignadvice;(2)developandevaluateconceptualdesignalternativesthatmeetcustomerrequirements,manufacturingtargetcosts,andenterprisestrategicplans;(3)provideaccesstomaterial-performancemodelsanddecisionmethodologies;and(4)collect,organize,anddistributelessons-learnedexperiencetoassistfuturedecisions.

Modelingsystemsfortradeoffanalysis:ConstitutiveAnalysisProcessmodelingPerformancesimulationCostestimationRiskassessmentLessons-learnedcollectionandsearchingsystems:Object-orienteddatabasesNeuralnetworksElectronicdocumentationCase-basedreasoning

IntegratedEnterpriseProcessesMaterialsselectionbasedonbroadrangeofindustrialcompetitivenessconsiderations,includinginternalandexternalpartneringandpreferredsupplierrelationshipsandalliances.Requiresubiquitous,powerful,andwell-integratedsystemswithstandardizedcomputingenvironmentsanddatastructurestopromoteunifiedcommunicationandpermitrapidassembly/transmissionofshareddatabothwithincompaniesandwithsuppliers/vendors.

StandardizeddatastructuresordatadictionariesInter-andintra-companynetworking/communicationsystems

Page 76: Computer-Aided Materials Selection During Structural Design

InnovativeMaterialsSelectionMustassistinnovativedesignandhelpsolvedifficult,nonroutineproblemsasopposedtojustprovidinglimitedseriesofconventionalmaterialorprocessingchoices.Toolsthatenhance/stimulatecreativeprocessesandinsightviahumaninterfacetechnologiesthatpromotelearning,inferencing,andcognition.

Artificialintelligence(learning,inferencing,andknowledgerepresentation)Databaseandknowledge-baseacquisitionHumaninterfacetechnologies(graphics,vision,touch,animation,simulation,andvoiceorspeech)Linkingofmultipleknowledgebases

Page 77: Computer-Aided Materials Selection During Structural Design

Page24

Figure3-1TheconceptualarchitectureofaCAMSS.

Page 78: Computer-Aided Materials Selection During Structural Design

Page25

4InformationTechnologiesPertinenttotheMaterialsSelectionProcessThischapterdiscussesthekeymaterials-specificinformationtechnologiesrequiredtoproducetheCAMSSdiagramedinFigure3-1.Thischapterisdividedintotwosections:"DatabasesandKnowledgeBases,"whichpertainstothefirsttwoboxesatthetopofFigure3-1,and"ModelingandAnalysisSystems,"whichpertainstothethirdboxatthetopofFigure3-1.Thischapterfocusesontheuseofcomputer-aidedsystemsastoolsoraidestodesignteams.Full-functionautomatedsystemscurrentlyrequiresignificantbreak-throughsinareasoffrontierresearchandareparticularlyweakintasksdemandingcreativeinsight.

DATABASESANDKNOWLEDGEBASES

AsshowninChapter3,aCAMSSrequiresaccesstoandapplicationofmaterialsdatabasesandknowledgebasesateverystageofuse.Intheidealcase,electronicallystoredknowledgeaboutmaterialsanddesigndetailscouldbeprovidedautomaticallytothedesignteamfromdatabasesandknowledgebasesatappropriatelevelsofsophistication.ThissectionprovidesabriefoverviewofthelevelsofknowledgerepresentationintheautomationoftechnicalmemoryanddiscussestheprincipalmethodsforrepresentingmaterialsknowledgewithinaCAMSStofacilitatethedesignprocess.AppendixAcontainsabriefoverviewofsomeoftheknowledgerepresentationtechniquesdiscussedinthischapter.

LevelsofRepresentation

Thebasiclevelofelectronicknowledgerepresentationisanelectronic

Page 79: Computer-Aided Materials Selection During Structural Design

library(i.e.,databasesandknowledgebases).Inthisscenario,ahumandesignerwouldperformessentiallythesametasksaspreviously,butinsteadofsearchingforinformationthroughwrittenmaterial,thesearchwouldbeconductedthroughscreens.Thescreenscouldpresentpriordesigns,lessonslearned,designguidelines,orstandards.Whileelectronicsearcheshaveadvantagesovermanualsearches,thecostofimplementingallthenecessaryreferencematerialelectronicallywouldbehighandprobablycouldnotbejustifiedbasedsolelyonproductivitygains.However,sincemostdocumentationiscurrentlybeingcreatedelectronically,thisisasignificantissueonlyforolderreferencematerial.

Initssimplestconception,theelectronicdatabaseorknowledgebasewouldhavenomoreembeddedreasoningpowerthanbooks(i.e.,theusersuppliesallthereasoning).Threeadvantagesofthisbasictypeofknowledgerepresentationarethat(1)itiseasy,inprinciple,toimplement;(2)itisrepresentedinnaturallanguage,withallitsflexibility;and(3)itautomaticallymakesthemostrecentversionsofmaterialavailable.

Ahigherlevelofsophisticationwould,continuingtheanalogy,consistofareferencebookthatopensautomaticallytothedesiredpageandthen,basedonauserrequest,highlightsthatpartofthepagewhichtheuserneeds.Thisrequiresamechanismorprocessfortheusertodescribethereasonandcriteriaforsearchingthereferencebook.

Inalimitedsense,suchsearcheshavebeenavailableforalongtimeindatabasesandknowledgebasesusingkeywordsandindexingschemes.Whilekeywordsandindexschemesareusefulinrestrictingthematerialthatwilllaterhavetobescannedbythehumanexpert,theseschemesarenotveryintelligentbecauseagreatdealofinformationisoftenpresentedthatistotallyirrelevanttotheproblemathand.Keywording,moreover,worksonalphanumericinformationbutisnoteasilyadaptedtootherinformationtypessuchasshapes,

Page 80: Computer-Aided Materials Selection During Structural Design

colors,andgraphs.Thetechnologytosupportmultimedia,interactivereferencebooksisnowemerging(ACM,1993;IEEE,

Page 81: Computer-Aided Materials Selection During Structural Design

Page26

1993).Thiscapabilityisanadvanceinthattheauthorhaspre-programmedexpectedsearchrequests.

Ahigherlevelofsophisticationofelectronicrepresentationoftechnicalmemorywouldbetheequivalentofaneducatedassistantortechnicianthatcansearchthelibraryandretrievethepertinentinformationinthebackgroundwithoutdirectuserinvolvement.Representingtheknowledgethatanelectronicassistantcontainsandrepresentingtheknowledgethattheuserrequiresaretwodifferentproblems,however.Forthematerialsselectionprobleminstructuraldesign,theintelligentelectronicassistantwouldhavetounderstand,atsomelevelofcompetence,theinformationprovidedbythehumanexperts.Thatinformationwouldconsistnotonlyofconceptsusedinmaterialsscienceandengineeringbutconceptsrelatedtotheentirelifecycleofdesign,manufacturing,inspection,anddisposalorrecycling.Intheultimatecase,theelectronicassistantwillhavetoknowthelanguagesofmanydifferentpertinentdatabasesandthenbecapableofrepresentingthatknowledgeinaconsistentform.Thisleadstotheneedfortheintelligentintegrationofinformationfromthesemultiplesources.

Thehighestlevelofsophisticationenvisionedwouldprovideafull-function,computer-aidedelectronicassistantortechnicianwhocouldnotonlyfindthecorrectreferencematerialbutalsoapplytheresultstothequerytothedesignproblemathand.Justaswecanimaginehumanassistantsofdifferentlevelsofskill,sowecanalsoimagineelectronicassistantsatdifferentlevelsofutility.Asmentionedearlier,fullautomationofdatabasesorknowledgebasestoperformthecompletedesigntaskisnotcurrentlyfeasible.However,certainselectroutinesorcomputationallyintensivetasksnowperformedbypeoplecanbeperformedbyelectronicassistantstoprovideparticularadviceortocritiqueselectedaspectsofthedesign.Forinstance,therehasbeenconsiderableresearchintoAgentTechnology,wherebyauser

Page 82: Computer-Aided Materials Selection During Structural Design

canspecifyanagenttoroamtheInternettoobtainappropriateinformation(ACM,1994).Thistechnologyrequirestheuseofdatadictionariesormediators,however,torecognizeandtranslatetherelevantinformationindifferentdatabasesandknowledgebasesandtointegratepossiblyconflictingdatafrommultiplesources.Thislevelofcapabilityhasbeenshowntogreatlybenefittheoverallperformanceofthedesignteamsinseverallimitedinstances(Klahretal.,1987;Familietal.,1992).Forexample,thethreeareasofexpertiseproductdesign,materialsselection,andmanufacturingcannotbeentirelyseparatedforhighlyengineeredproducts.Materialspropertiesdependtoanextentontheprocessingroute,andprocessingconsiderationscanbeinfluencedbydesignconstraints.Similarly,thedesignmustreflecttherealityofavailablematerialproperties,andthepropertiesarenotcompletelyindependentofthedesignapplication(e.g.,highloadingratescanreduceamaterial'sfracturetoughness).

IssuesConcerningKnowledge-BaseDevelopment

Relianceonstandarddatabasesthatcontainphysicalandmechanicalpropertiesisinadequatetosupportmaterialsselectionprocessesfully.Knowledgebasesarerequiredthatcapturetheadvantagesandlimitationsofmaterials,theirprocessability,andtheirapplicationhistoriesallofwhicharecriticaltothedesignprocess.Thereareproblemsrelatedtothedefinition,development,andconstructionofknowledgebases,however.

DefinitionofKnowledgeBases

Whilemuchhasbeenwrittenaboutknowledgebases,thereislittleagreementonthescopeofwhatexactlyconstitutesone.Therearewidevariationsintheliteraturedescribingdesignsandimplementationsofcomputer-aidedsystems.

Manycompanieshavecustom-builtbasicsystemsfortheirownapplications.Table4-1summarizesrepresentativecomputer-aided

Page 83: Computer-Aided Materials Selection During Structural Design

systemapplicationareasthatrelatetothematerialsinformationusedinthedesignprocessinamannerconsistentwiththevisiondiscussedinChapter3.Theseapplicationsareconsideredstateoftheartinthesensethatexamplescanbefoundeitherinuseorunderdevelopmentatmajorgovernmentandindustrialsites.

Table4-1showsthatthereisawidebreadthofknowledge-baseapplications.Thelistisalsoincomplete,sinceitonlyrepresentswhatiscurrentlypossibleinthedesignandengineeringphaseofproductdevelopment.Ifthescopewerebroadenedtootherphasesoftheproductlifecycle,moreapplicationscouldbelistedthatwouldrequireknowledge(e.g.,diagnosisofthemanufacturingprocess).Furtherworkisrequiredtodeterminewhatconstitutesaknowledgebaseandhowitdiffersfromsimpledatabases.

Page 84: Computer-Aided Materials Selection During Structural Design

Table4-1RepresentativeApplicationsBasedonKnowledgeofMaterialsApplication Description KnowledgeRequired

SelectingMaterials

Assisttheproductdesignerinselectingthebestmaterialfortheproduct.

Materials,performance,andprocessingknowledge.

CostEstimation Determinethemanufacturedcostofaproductbasedonaparametricdescriptionofit.Couldincludefulllife-cyclecosts.

Costofrawmaterialandcostsassociatedwithdesign,validation,manufacture,anddisposalmethodsforvariousmaterials.

ManufacturingProcessPlanning

Developaprocessorassemblyplanforaproduct.Thedepthofdetailcanvaryfromprocessroutingtoinstructionsforcontrollers.

Manufacturingmethodsandprocessingdataassociatedwithvariousmaterials.

PredictiveAnalysis

Predicttheperformanceofaproduct.Manydimensionsofperformance(e.g.,stress,wear,andkinematics)canbeanalyzed.

Materialspropertiesandcharacterizationofusageofmaterials.

ManufacturabilityAdvisor

Critiqueaproductdesign,processplan,oroperationsplanregardingtheefficiency,productquality,andcostofproduction.

Manufacturingknowledgeaboutmaterialsandthemanufacturingprocessesassociatedwiththematerials.

RequirementsAllocationandBalancing

Provideassistancetotheproductdesignertoallocateandbalancerequirementsinformalrequirementsprocesses

Knowledgeoftherelationshipofmaterialstothecharacteristicsdefinedintherequirementsflowdownprocess.Examplesincludedurabilityofmaterials,surfacequalityforfinishing,andmass.

Page 85: Computer-Aided Materials Selection During Structural Design

suchasQualityFunctionDeployment.

SearchforPriorDesigns/Products

Assistingproductdesignerinfindingmostsimilarproductsfromalibraryofproductstotakeadvantageofpriorexperience.Priorproductsusingsimilarmaterialsmaybeinterestingevenifproductwasquitedissimilar.

Anymaterialpropertiesmaybeofinterest.Knowledgeofsimilaritiesbetweenmaterialsandrelationshipsbetweenmaterialsandproductandprocessperformanceandcost.Repositoryofpriordesignsandproductinformation.

SearchingStandardComponents

Assistinselectingastandardcomponentfromalibraryofstandardcomponents.

Anymaterialpropertiesofthestandardcomponentsthatarerelevanttothecostandperformanceoftheproduct.

ToleranceAnalysisandAllocation

Predictthevariabilityofthemanufacturedpartwithrespecttogeometricdimensionsandotherkeyproductcharacteristics.

Knowledgeabouttherelationshipsbetweenmaterialsandthemanufacturingprocessesandmanufacturingequipment.

Prototyping Assistancecanincludeselectingthebestmaterialfortheprototypeandtheprototypetoolingbasedonthematerialsusedintheproduct.Theprototypecanbeconstructedwithvaryingdegreesofproduction-intenttooling.

Knowledgeofanymaterialpropertiesandsimilaritiesofmaterialstobeusedforprototypeandprototypetoolingtothosetobeusedinproduction.

RobustDesign Aformalthree-stepprocessforproducinghigh-quality,low-costproducts.

Knowledgeofmaterialpropertiesareusedinthesystemdesignphaseindecidingrequiredfeatures,functions,andproductparameters.Knowledgeofmaterialsisusedintheparameterdesignprocesstoconsiderreliabilityandmanufacturabilityofthe

Page 86: Computer-Aided Materials Selection During Structural Design

reliabilityandmanufacturabilityofthecomponents.Finally,inthetolerancephase,materialpropertiesareusedtoadjustproductparameterstoachievebroadrangesformanufacturability.

Trade-StudyMethods

Assisttheproductdesignerinmakingchoiceswithrespecttoproductfeatures,function,andmanufacturing.

Knowledgeofmaterialsandhowtheyrelatetokeyproductcharacteristicscanbeusedintheevaluationofalternatives.

Page 87: Computer-Aided Materials Selection During Structural Design

Page28

DevelopmentofKnowledgeBases

Theabsenceofspecificguidelinesforthebuildingofdatabasesandknowledgebasesaboutmaterialsandthehighersoftwaredevelopmentcostsnecessarytoobtaingeneralityandrobustperformancearebarriertothequickandeffectiveproliferationoftheuseofdatabasesandknowledgebasesinmaterialselection(orotherfieldsaswell).Knowledge-baseapplicationsarecurrentlydevelopedinseveraldifferentways.

Somebasic,commercial,knowledge-baseapplicationsexistthattheusercommunitycanacquire,install,andcustomizebyloadingsite-specificmodels,informationmodels,andinformation.Oneexampleofsuchanapplicationisinthedomainofcostestimation.Thisapplicationallowstheusertobuildparameterizedproductdescriptionsthatincorporaterelevantattributesthatinfluencecostandmanufacturability,materialmodelsthatincludeparametersthatimpactcost,andprocessdescriptionsthatlinkmodelsofprocessestotheproductandmaterialattributes.Theusercanthenemployasimplespreadsheet-likelanguagetomatchprocessesandmaterialtoproductsandtocomputecostestimates.

Moreadvancedknowledge-basesystemscanalsobedevelopedbyusingapplicationshellsorknowledge-engineeringtoolsthatimplementknowledge-basetechniquesdevelopedbytheartificialintelligencecommunitytosolveavarietyofproblems(e.g.,diagnosis,simulation,andscheduling).Anapplicationshellisoftenalibraryofmodulesthatcanbeusedtoassembleanapplicationinsomebroadareaandrapidlyprovideanenvironmentforcapturingandrepresentingexpertiseintheformofrulesandtheknowledgestructures.Theadvantageofusingshellsisthattheypermittheusertoconcentrateonrepresentingtheknowledgeratherthanattendingtolow-levelprogrammingtasks.Applicationshellsdorequiretailoring,

Page 88: Computer-Aided Materials Selection During Structural Design

modification,andextensionbeforeuse.Typically,aknowledge-engineeringeffortmustfirstbeundertakentoperformknowledgeacquisitionandroutineprogrammingtasks.Forapplicationshells,thecostofcustomizationisoftenoffsetbythereducedcostforgeneralityinthesoftware.Aswithcompleteapplications,manycompanieshaveinternalproductsthattheydistributetoanumberofsites.

ConstructionofKnowledgeBases

Twofinalbarrierstotheconstructionofknowledgebasesarethehigherhardwarecostsandtheinherentnatureofexpertknowledge.Theconstructionofknowledgebasesrequiretechnologiesbeyondthestandardmanualenteringofpertinentpiecesofinformation.Forinstance,digitalscanningcombinedwithcharacter-recognitiontechnologycanbeusedtoenterapplicationhistory,suchasfailureanalysisresults.Digitizingtablesandfiguresusingdatacapturesoftwarecanalloweachsemanticelementandsemanticrelationexpressedinthetablesandfigurestobestoredinasearchablefile.Graphscanbemadesearchableforspecificdata,interpolateddata,andextrapolateddata.Audio-visualandmultimediadigitalstorageandpresentationarebecomingcommon-placeonengineeringworkstations.Theappropriateassignmentofrecordingsofdesignengineeringdiscussionsinknowledgebaseswillposeasignificantresearchchallenge.Annotationstodiagramsrelatingkeydesigndecisionsandconstraintswouldalsoassistothersinunderstandingthereasonsforaparticularchoice.

IssuesConcerningDatabaseDevelopment

ThematerialdatabasesneededduringthedesignprocessarenotopenlyavailabletoU.S.industry.Somegovernmentandprivatelysponsoredorganizations,suchastheDepartmentofDefense'sInformationAnalysisCentersandtheNationalMaterialsPropertyDatabaseNetwork(AppendixC),havemadeastart,butindustrialdatalimitationshaveresultedintheirfallingshortoftheneedsfor

Page 89: Computer-Aided Materials Selection During Structural Design

knowledge-baseapproachestodesign.

Thesituationiscurrentlylittlebetterthanitwasin1983whenaNationalMaterialsAdvisoryBoardreportstated:"ThereisnonationalpolicyintheUnitedStatesdirectedtowardarationalsystemofmaterialspropertiesdatamanagementandthesituationinthisareaisbestdescribedaschaotic"(NRC,1983).Thereisstillnoleadershipforcollecting,generating,validating,andupdatingthedataneededforstructuraldesign.Mostorganizationswillagree,however,thatacertifieddatabaseisavaluableasset.AsrecommendedinamorerecentNationalMaterialsAdvisoryBoardstudy(NRC,1993):"afederalagency,suchastheNationalInstituteofStandardsandTechnology,couldestablishaforumto

Page 90: Computer-Aided Materials Selection During Structural Design

Page29

developthestandardsthroughtimely,activeparticipationbyindustryandotherinterestedparties.''

MODELINGANDANALYSISSYSTEMS

AsshowninChapter3,aCAMSSmusthavemodelingandanalysissystemstoanalyzetheinformationavailableindatabasesandknowledgebases.Thissectionexaminesthemodelingandanalysissystemspertinenttothematerialsselectionprocess:GeometricReasoning,ProcessModeling,andModelingfortheAbstractionofDownstreamConstraints.

GeometricReasoning

MostCADsystemscurrentlyinindustryareusedprimarilyfordraftingpurposeswithsomeanalyticalsupport.Thelevelofgraphicrenderinghasreachedanimpressivelevel,butalldesigndecisionsareessentiallymadebythedesignengineerwithlittleornodecisionsupport,orreasoning,otherthanvisualfeedback.Reasoningfallsintothetwomajorcategoriesofsynthesis(i.e.,thesystematiccreationofalternativesastheproductdesignandthedesignprocessbecomeincreasinglymorespecific)andabstraction(i.e.,theeliminationofpossibledesignalternativesbydownstreamconcernsandconstraints,suchasperformanceandmanufacturability).Thetermabstractionisusedsince,inmostinstances,detaileddownstreamconcerns(e.g.,manufacturability,maintainability,orrecyclability)needtobeeithersimplifiedorabstractedtobecomemorereadilyunderstandablebyanengineeratanearlystageofthedesignprocess.Asdiscussedintheprevioussection,thedesignershouldreceivefeedbackabouttheconsequencesofadecisionatavarietyofdifferentsupportlevelsandfromavarietyofdifferentviewpoints(e.g.,materialsconsiderations,productperformance,manufacturing,cost,service,andreliability).

Geometryplaysanimportantroleindesign,yetmanyinitialdesign

Page 91: Computer-Aided Materials Selection During Structural Design

decisionsaremadeindependentlyofgeometricconsiderations.Thefinaldecisionsinproductdesignalmostalwaysinvolveformandgeometricconstraints.Anexampleofsuchaninformationflowinanindustrialsettingistoimagineacardesignteamthatissynthesizinganewcarbody.Thecurrentmanufacturingapproachwouldbetospot-weldseveralhundredsofpiecesofsheetmetaltogethertofitthepartgeometry.Apossiblealternativeisanassemblyoflightweightaluminumbeamsonwhichplasticpanelsaremounted.Thisdesigniscalledaspaceframe.However,tomakeanintelligentdecisionaboutwhetherthisalternativedesignhasmerit,adesigngroupmustunderstandtheessenceofthealuminumcastingandextrusionprocesses,themicrostructuresthatresult,thepropertiesofthebeams,andthemethodsforjoiningthepieces.Thismeanssimplifyingthedetailsoftheprocesstotheextentthatsomeonewithlittlematerialsexpertisecantakethisinformationasthebasisforperformingdesigntradeoffs.Thefasterandmoreefficientlythemanufacturingprocessandmaterialsperformanceconstraintsarepresentedtothedesigners,thefasterandmoreefficientlytheycansynthesizeanewproduct.

Whenreasoningaboutmaterialsandassigningmaterialpropertiestocertaingeometricobjects,severalthingsareimportant.Inprinciple,allgeometricmodels,whetherbasedonelementsofone,two,orthreedimensions,canbelinkedtoattribute-valuepairs,suchasmaterialsproperties,likeYoung'sModulus,withaspecificvalueassigned.Someoftheseattributesmayberelatedtomaterialpropertiesortothespecificsofacertainmicrostructure.Intheevaluationofthegeometry,theattribute-valuepairmechanismmustbeabletoinheritpropertystructures.Furthermore,whentheattributeoritsassociatedvalueassumeaspecificcharacteristic,themechanismshouldbeabletotriggereventsautomatically.MostcommercialCAD/CAMsystemsdonotcurrentlyhavethiscapability,althoughitisanactiveresearcharea.1Significantresearchneedstobeperformedtocreatesoundrepresentationalschemeswiththedescribedbehavior,however.

Page 92: Computer-Aided Materials Selection During Structural Design

Formalmodelsarealsonotavailableatanylevelofrepresentationthatallowsthederivationofstructure-propertyrelationshipsfromfirstphysicsprinciplestoanextentthatadesignengineercanbenefit.Oneexampleisthefieldofdislocationtheory(defectstructurelevel).Theflowstress(microandmacromaterialpropertylevel)ofsubgrainformingmaterialsisunderstoodtobeproportionaltothesquarerootofthedislocationdensity,butthemodelscurrentlyavailablegiveonlyanorderofmagnituderelationfortheproportionalityfactorinthisrelation.

1Onesuchareaisthatofactivedatabasesystems,whichcontainrulestomonitorthestateofthedatabaseandtriggeroperationstoupdatethesystemoralerttheusertocertainsituations(ACM,1994).

Page 93: Computer-Aided Materials Selection During Structural Design

Page30

Adesignengineercouldbenefitfromsuchamaterialsscienceinsight,however.ItwouldhelpthedesignertobetterunderstandthechoiceofaconstitutiverelationforFEAthatisfrequentlyusedtodeterminethedimensionsofloadbearingcomponents.Thedesignengineerneeds(computer-aided)decision-supporttoolsthatprovideinsightintomaterialsscienceissuestoconsidermaterialalternativesandprocessingtradeoffseffectively.Reasoningaboutmaterialsneedstobecloselycoordinatedwithdecisionsregardingshapeandgeometry.Consideringtheearlierexampleofthecarspaceframe,anoveldesignwascreatedbycombiningmaterialpropertiesandprocessingknowledgewithspatiallayout.

Decision-supportmethodologiesandtoolsthataidinsynthesizingandfindingconstraintsearlyoninthemanufacturingprocessarealsokeytoimprovingthequalityandspeedofproductcreation.Considerthefollowingtoolusedfordesigningasidemarker,arelativelysimplecomponentforacar(Figure4-1).Onesideofthiscarsidemarkerhasstrengtheningribstoincreasetheresilience.Thesystemshownhererecognizescertaingeometricfeaturesthatareimportantfromamanufacturingmaterialsviewpoint.Withinthatsystem,thedesignerreceivesfeedbackaboutwhetherthedimensionschosenarecompatiblewithgoodmanufacturingpractice.Decision-supporttoolssuchasthiscansignificantlyreduceerrors,cost,anddevelopmenttime.Theirrealizationwilllargelydependonresearchperformedtoconstructmethodologiesforrepresentationinwhichgeometricinformationcanbeproperlylinkedwithnongeometricinformation,likematerialsandprocessingknowledgeanddatabases.

Page 94: Computer-Aided Materials Selection During Structural Design

Figure4-1Automobilesidemarker.

Page 95: Computer-Aided Materials Selection During Structural Design

Page31

ProcessModeling

Aprocessmodelisamathematicalrepresentationorsimulationofaprocessthatallowsproblemstobesolvedinthecomputerratherthanbyempiricalorexperimentalmethods,especiallytrial-and-errortechniques.Simulationisanimportantconceptinthemodernmanufacturingorganization.Itusesmathematicalmodelsofrealsystemstotestorpredicttheactualperformanceofthesystemsundervariousconditions.Throughsimulation,engineersandmanufacturingpersonnelcantestadesign,analyzeaprocedure,orassessaprocessperformancebeforeimplementingtherealthing.Processmodelingdiscussedinthecontextofthisreportdealswithunitprocessessuchascasting,forging,rolling,hotisostaticprocessing,heattreating,machining,chemicalvapordeposition,andcompositematerialfabrication.Constitutivemodelingmaybeoneofseveralelementsintheoverallprocessmodel.Constitutivemodelsfocusonpredictingthemechanicalresponseofamaterialasafunctionofpriorprocessinghistoryandinternalstructuralparametersinresponsetoexternallyappliedforces.ProcesssimulationcouldjustaswellapplytomodelingthebehaviorofaFlexibleManufacturingCenterorthesimulationoftheflowofinformationinaprocessplan.Asimulationcapabilityinthemanufacturingsettingcansubstantiallydecreasetheenergy,materialwaste,andtimerequiredtoproduceaproductorimplementaprocess.

Anumberofsequentialstepsareinvolvedinanyprocessmodelingactivity.Thesestepscanbeformalizedandimplementedinthecomputerasanactivitymodelthatisbasedonhowtheparticularmanufacturingenterprisedoesbusiness,ortheycanbeaccomplishedinalessformalmodeofproblemsolving.ThevariousstepsinvolvedinprocessmodelingarelistedinTable4-2.

Processmodelingisgenerallyusedtounderstandunitprocessesthatrequirecouplingofdisparatephysicalphenomena.Forinstance,virtuallyallunitprocessesaregovernedbyheatflow,fluidflow,plasticflow,

Page 96: Computer-Aided Materials Selection During Structural Design

allunitprocessesaregovernedbyheatflow,fluidflow,plasticflow,stress,andphasetransformations.Theseallcanbemodeledbyavarietyofnumericaltechniques.Severalapproximatenumericaltechniquesforsimulatingmaterialshapingandformingprocessesunderarbitraryconditionshavebeenused,including(1)theslabmethod(approximatestressanalysis),(2)theslip-linemethod(methodofcharacteristics),and(3)theupper-boundmethod(methodutilizinganenergyprinciple).Althoughthesetechniquescontribute

Table4-2StepsintheDevelopmentofaProcessModel1.Definetheproblemandstatetheproblem-solvingobjective.2.Developthemathematicalmodelinaccordancewiththeproblem.3.Collectmodelinputdataandspecifications.4.Implementprocessmodelinthecomputer.5.Establishthatthedesiredaccuracyorcorrespondenceexistsbetweenthesimulationandtherealsystem.

6.Establishboundaryconditionsforusingthemodel.7.Runsimulationstoobtainoutputfile.8.Post-processtheoutputvaluestodrawinferencesandmakerecommendationstosolvethedefinedproblem.

9.Implementanddocumentthedecisionsresultingfromthesimulationanddocumentingthemodelanditsuse.

significantlytowardunderstandingthemechanicsofdeformationinmetalworking,theylackgeneralityandoftendonotprovideaccurateestimatesoftherequiredforcesandenergy.

Forfifteenyears,thefiniteelementmethod(FEM)hasbeenappliedtomodelawiderangeofmetalworkingoperations.FEMdividesthevolumeoftheplasticallydeformingmaterialintoatwo-dimensionalorthree-dimensionalnetworkofdiscreteelements(finiteelements).Thedeformationatselectedpoints(nodes)isdeterminedbytheapplicationofsolidmechanicsprinciples.

SpecialtyFEManalysiscodesforprocessmodelingcurrentlyhavebeendevelopedforanalyzingalmosteveryclassofunit-manufacturingprocess.Theprocessesthathavebeensimulatedincludemachining,heat

Page 97: Computer-Aided Materials Selection During Structural Design

process.Theprocessesthathavebeensimulatedincludemachining,heattreating,sheetmetalforming,shaperolling,ringrolling,extrusion,forging,powderconsolidationandforging,superplasticforming/diffusionbonding,cogging,andradialforging.Atleastoneofthecommerciallyavailablethree-dimensionalcodesiscapableofconcurrentlymodelingtheequipment,thedies,andtheworkpiece'sresponsetotheboundaryconditions,includingtheeffectsofdifferentheatsourcessuchasinductionandresistanceheatingonthematerialflowbehaviorandthediereaction(Kiridenaetal.,1989).

AlmosteverydetailofaunitprocesscanbemodeledbyFEManalysis,includingpredictingtheevolutionofmicrostructureandpropertiesinthefinishedshape.Thelatterismadepossiblethroughtheuseofatechnique

Page 98: Computer-Aided Materials Selection During Structural Design

Page32

knownasdynamicmaterialmodeling(DMM),whichusesconstitutiverelationshipsthatdefinetheevolutionofmetallurgicalstructureonascalethatrangesfromthemicroscaleofdislocations,precipitates,dislocationnetworks,andgrainstothemacroscaleoflaps,shearbands,andgrainflowlines(Richmond,1992).

DMMdefinestheintrinsicworkabilityoftheworkpiecematerialatthemacroscalelevelofstructureintermsofmechanicalandstructuralstability.Thismaterialmodelenablestheprocessdesignengineertodefinethecontrolspaceforastableprocess,includingthenumberofpreformshapes,thedievelocityandtemperatureranges,andthediegeometries.Withinthedomainsofthestablecontrolspace,wheretheactivationenergyisfairlyconstant,microscopicmodelscanbeusedtopredictstructuralevolution.Theendresultisaproducthavingacontrolledsetofstructuresandpropertiesinthefinishedshape.

Asanexample,thegroupofprocessesthatcanbedescribedbroadlyascastingprocessesarenowbeingsimulatedbyseveralnumericalmethods.Processesofthistypeareinvestmentcasting,permanentmoldcasting,diecasting,squeezecasting,andplasticinjectionmolding.Theapproachtodesigningtheseprocessesbyprocesssimulationisfundamentallythesameasforanyotherprocess.

Achoiceofnumericalmethodsareavailableforthisclassofproblem.Thefinitedifferencemethod,boundaryelementmethod,andFEMhavebeenusedformodelingthebehaviorofcastingprocessesthatarecontrolledbycoupledthermal,fluid,andstressphenomena.Boththefinitedifferencemethodandtheboundaryelementmethodcanbeusedwhenthematerialpropertiesarelinearandtheproductgeometryisrelativelysimple.However,whentheproblemcouplesthermal,fluid,andstressphenomena,theFEMmodelingtechniqueissuperior.

Oneexampleofincorporatingmaterialbehaviorduringprocessingconditionsistheuseofexperimentalcastabilitymapsthatare

Page 99: Computer-Aided Materials Selection During Structural Design

expressedintermsofthefundamentalvariablespredictablebytheprocessmodel.Thesemapsprovideaframeworkformodelingthecastingprocessatthemacroorcontinuumlevelofanalysis.LikeDMM,whichpredictsstableplasticflow,thecastabilitymapsdefinethedomainswherecertainmicrostructuresordefectswillformwhencertainratiosofR/Goccur,whereRistheinterfacialvelocityandGisthetemperaturegradientatthesolid/liquidfront.Thesemodelingparameterswerederivedfromfirstprincipleunderstandingofthenucleationgrowthkinetics.

Micromodelscanbeintegratedwiththemacromodelsforpredictingtheevolutionoroccurrenceofmicrostructuralfeatures.Moreeffortisrequiredtoexpandthenumberofmicromodelstocoverallpossibilities,however.Mechanicalpropertypredictionsarenotyetpossiblebecauseveryfewcorrelationshavebeenmadewithmicrostructuresandthermomechanicalhistories.Thesemodelssufferfromnotincorporatingtheknowledgeofthebasicphysicalmechanismsinvolved.

Modelscanbeusedtobetterinformsuppliersontheirprocessrequirements.Forexample,castproductsareusuallyconsideredtohaveinferiorpropertiestowroughtproductsbecauseofthelargevariationinmechanicalpropertyvaluesthatcanbefoundinthesamecastproductproducedbydifferentvendors.Processmodelingcanreducethescatterfromvendortovendorbyspecifyingtoeachvendorthedesiredthermomechanicalhistoryforagivencomponent.

Thermophysicalpropertydataplayakeyroleinmodelingmanyunit-processessuchasinvestmentcasting,welding,crystalgrowth,glassmaking,microwaveprocessing,andcompositesproduction.Acriticalneedexistsformeasuringandarchivinghigh-temperaturethermophysicalpropertydataformaterialsintheliquid,solid,andbiphasicstates.Theimportantpropertiesincludeemissivity,heatcapacity,heatoffusion,meltingtemperature,density,surfacetension,

Page 100: Computer-Aided Materials Selection During Structural Design

thermaldiffusivity,andmaterialsviscosityasafunctionoftemperatureandshearstrainrate.

Comprehensivethermophysicalproperties,constitutivemodelsfornonlinearbehavior,intrinsicprocessingmaps,anddatabasesofmicrostructure-propertyrelationshipsareneededforindustrialprocessmodelingasinfluencedbypriorthermomechanicalhistory.Thisdatabaseshouldrepresentastandardizationofmaterialsprocessdesigndatathatarecertifiableinorderforthemtobeusefultomanufacturers.Thedatabaseinformationcurrentlyusedinprocessmodelingisgenerallytypicaldata,andthestatisticalassuranceassociatedwithsuchvaluesisnotknown.Accurateerrorestimatesinprocessmodelpredictionsrequireinformationregardingthestatisticaldistributionoftheinputdata.

Page 101: Computer-Aided Materials Selection During Structural Design

Page33

ModelingofAbstractionofDownstreamConstraints

Asstatedpreviouslyinthischapter,thedesigntrade-offstobetakenintoaccountinconcurrentengineeringalsomustmeetperformancecapabilityconstraintssuchasinspectability,maintainability,andreliability.Inspectabilityduringmanufacturingandproductservicearenotyetgenerallyconsideredinthedesignprocess.However,recentdifficultieswithinspectabilityofagingaircraft(AchenbachandThompson,1991)haveclearlyindicatedtheneedforincorporatinginspectabilityatanearlydesignstage.

Figure4-2showsthediagramofaconcurrentengineeringenvironmentthatlinksdesigntoinspectabilityandtheotherdownstreamcapabilityconstraints,aswellastoareassuchasqualityassurance,life-cyclecosts,andmaterialsandprocesses.Atthepresenttime,theonlylinksthathavebeendevelopedaretheCAD/CAMlinksbetweendesignandmanufacturingmethodsandprocesses.

Figure4-2Concurrentengineeringenvironmentincludinginspectability.ReprintedcourtesyofD.ThompsonandL.Schmerr,

CenterforNondestructiveEvaluation,IowaStateUniversity.

ComputermodelsforotherdownstreamcapabilityconstraintscanprovidekeyingredientsforimplementingthecompleteconcurrentengineeringenvironmentofFigure4-2.Inanearlystageofthedesignprocess,modelscanbeusedtodeterminetheroleofsuchproceduresasnondestructiveevaluation(NDE)forin-processcontrolof

Page 102: Computer-Aided Materials Selection During Structural Design

importantparametersinthemanufacturingprocessandforin-serviceuseandin-the-fieldinspections.Theyalsoplayanessentialroleinadamage-tolerantdesignphilosophyandinquestionsofin-servicereliabilityandlife-cyclecosts.SignificantprogresshasbeenmadeinestablishingNDEmodelsandinbuildingtheotherconcurrentengineeringlinks,suchasthroughthejointNationalInstituteofStandardsandTechnology,IowaStateUniversity,andNorthwesternUniversityPrograminIntegratedDesign,NDE,andtheManufacturingSciences.

MeasurementModeling

Theavailabilityofameasurementmodelhasmanybenefits.Numericalresultsbasedonareliablemodelareveryhelpfulinthedesignandoptimizationofefficienttestingconfigurations.Agoodmodelisindispensableintheinterpretationofexperimentaldataandtherecognitionofcharacteristicsignalfeatures.Therelativeeaseofparametricstudiesbasedonameasurementmodelfacilitatesanassessmentoftheprobabilityofdetectionofanomalies.Ameasurementmodelisavirtualrequirementforthedevelopmentofaninversetechniquebasedonquantitativedata.Iftestedforaccuracybycomparisonwithexperimentaldata,itprovidesapracticalwayofgeneratingatrainingsetforaneuralnetworkoraknowledgebaseforacomputer-aidedsystem.Finally,andmostimportantlyinthepresentcontext,thesemodelscanbeincorporatedintoaconcurrentengineeringdesignprocess.

Oneofthemostsignificantadvancesinnondestructiveevaluationoverthelastdecadehasbeentheevolutionofquantitativenondestructiveevaluation(QNDE)fromaconglomerationofempiricaltechniquestoawell-definedfieldinterdisciplinaryscienceandengineering.Inthecourseofthisdevelopment,ithasbecomewellrecognizedthatafundamentalapproachtoQNDEmustbebasedonquantitativemodelsofthemeasurementprocessesofthevarious

Page 103: Computer-Aided Materials Selection During Structural Design

inspectiontechniques.Amodel'sprincipalpurposeistopredict,fromfirstprinciples,themeasurementsystem'sresponsetospecificanomaliesinagivenmaterialorstructure(e.g.,cracks,voids,distributeddamage,

Page 104: Computer-Aided Materials Selection During Structural Design

Page34

corrosion,ordeviationsinmaterialpropertiesfromspecifications).Thus,ameasurementmodelmustincludetheconfigurationofprobeandcomponentbeinginspectedandadescriptionofthegeneration,propagation,andreceptionoftheinterrogatingenergy.Forexample,inthecaseofultrasoundasinterrogatingenergy,thisdescriptionrequirescomputationsofthetransducerradiationpattern,refractionofthebeamatthepart'ssurface,thebeamprofile,andthepropagationcharacteristicsinthehostmaterialincludingeffectsofmaterialanisotropy,attenuation,anddiffractionlosses.Detailedmodelingofthefield-flawinteractionsthatgeneratethemeasurementsystem'sresponsefunctionarealsoincluded,aswellasinformationonmaterialpropertiesandotherconditionsthatincreasevariabilityandadduncertaintytothemeasurementresults.Awell-constructedmeasurementmodelshouldbeabletopredictspecificinstrumentalresponsestoanyanomaliesincomplexmaterialsandstructuresaswellastoanystandardflawsplacedinvariouscalibrationblocks.Thestatusofmodelsforultrasonics,eddycurrentmethods,andradiographictechniqueshasrecentlybeendiscussedbyGrayetal.(1989).

QuantitativeNondestructiveEvaluation

ThissectiondiscussesQNDEasameasurementmodelanditsapplicationtoadamagetolerantdesignphilosophyanddetectionprobability.ThecouplingofmeasurementmodelstoCADisalsoreviewed.

Theload-bearingcapacityofastructuralsystemcanconventionallybedeterminedbyapplyingincreasinglylargerloadsuntilthestructurefails.Suchprooftestingispartofthedesignprocess.Onceastructureisinservice,aprooftestisobviouslynotapracticalwaytoassessapart'scondition.Afeasibleapproachtoobtainingstrengthinformationunderin-serviceconditionsisbyusingaQNDEtechnique,wherebyamaterialorastructureisevaluatedthroughinteractionwithsomeformofinterrogatingenergy.Manyformsofradiatedenergyhavebeenusedin

Page 105: Computer-Aided Materials Selection During Structural Design

interrogatingenergy.ManyformsofradiatedenergyhavebeenusedinQNDE(e.g.,laserlight,ultrasound,eddycurrents,andxrays).Othertechniquesarebasedonthepenetrationofneutronandthermalwaves.TheQNDEapproachincludesthedevelopmentofnondestructivemeasurementprocedurestodeterminematerialpropertiesandtodetectflawsandotherfailure-relatedconditions.QNDEalsoencompassesthedesignofinstrumentation,dataprocessing,theuseofmeasurementmodels,andtheinterpretationofdatatodeterminewhetherapartshouldberejectedorastructuralsystemshouldberepaired.QNDEproceduresshouldbeconsideredinthedesignstageaspartofqualityassurance,maintainability,andreliabilityanalysis.

Fracturemechanicsandfailuremechanicshavemadegreatstridesintheunderstandingandpredictionoftheintegrityofstructuralcomponents.Foracomponentmadeofamaterialofknownpropertiessubjectedtoagivensetofloads,itispossibletocalculatethecriticalsizeofacrackataspecifiedlocation.Acomponentisjudgedtobesafeifthecrackissmallerthanacriticalsizeandisnotexpectedtogrowtocriticalsizeduringtheservicelifeorpriortothenextinspection.Reliablemethodsmustbeavailabletodetectandcharacterizecracks,includingthoseofsubcriticalsize.QNDEprovidesthetechnologytodetectcracks(ormoregenerallyflaws)largerthanthedetectabilitylimitandtodeterminelocation,size,shape,andorientation.

Damage-TolerantDesign

Inadamage-tolerantapproach,subcriticalflawsjustbelowthedetectionlimitareassumedtoexistateveryfracture-criticallocation.Aspartoftheanalyticalevaluationthefollowingquestionsmustbeanswered:

Whatisthecriticalflawsizethatwillcausecomponentfailurewhensubjectedtoknownserviceloadsandtemperatureconditions?Whatarethedrivingforcescausingcrackgrowth?Howfastwillasubcriticalcrackgrowunderserviceloadandtemperature,andhence,howlongcanacomponentcontainingasubcriticalflawbesafelyoperatedin-service?

Page 106: Computer-Aided Materials Selection During Structural Design

Whatinspectionmustbeperformedtodetectacrackbeforecatastrophicfailureofthecomponentoccurs?

Subcriticalcrackinitiationandpropagationoccursinhigh-stressareasandinlocationswherecomponentscontainmaterial-andmanufacturing-relatedinhomogeneitiessuchasvoids,inclusions,machiningmarks,orsharpscratches.Currentprogramsrequireaninspectionathalfthetimerequiredforapotentialcracktogrowtocritical

Page 107: Computer-Aided Materials Selection During Structural Design

Page35

size.Theinspectionisassumedtodetectanyflawlargerinsizethanadefinedlimit(Cowie,1989).

AsystematicapproachtotheoverallinspectionrequirementsofstructuresisrequiredwithinaCAMSSsystemtoadvisethedesigneraboutpotentialproblemswithinspectionandpossibledesignalternatives.Thisapproachshouldtakeintoaccountthestatisticsoftheoccurrenceofflaws,thecrackgrowthmechanisms,andthevariousnondestructivedetectiontechniques.Theprobabilityofdetectionofcertainclassesofdefectsalsodependsontheload,damagedeteriorationpropertiesofthematerial,inspectionintervals,humanfactors,andreplacementandrepairmethodologies.

ProbabilityofDetection

Theimplementationofameasurementmodelshouldbecoupledtotheconceptofprobabilityofdetection(POD).Thisisastatisticalrepresentationoftheprobabilitythatagivenmeasurementsystemwillbeabletodetectaspecificflaw(orcondition)inagivenmaterialorstructure.Itincorporatesknowledgeofthesignaldetectedbythemeasurementsystemtogetherwithstatisticalinformationconcerningflawdistributions,instrumentalnoise,andthresholdlevels.APODcurveshowstheprobabilityofaflaw'sdetectionasafunctionofflawsizeforaspecificinspectiontechnique.Foranidealtechnique,thePODofflawssmallerthanasizepredeterminedbyperformancerequirementsandmaterialpropertiesiszero,whilethePODforanyflawgreaterthanthissizeisunity.Inthiscase,thereareneitherfalserejectionsofgoodpartsnorfalseacceptancesofdefectiveones.However,NDEtechniquesinpracticeareneverassharpandasdiscriminatoryasindicatedbytheidealcurve.Thus,thereareregionsofuncertaintywithfalserejectionsandfalseacceptances.

Figure4-3,whichwastakenfromGrayandThompson(1986),showstheresultsofsimulatingtheultrasonicPODofcircularcracksat

Page 108: Computer-Aided Materials Selection During Structural Design

differentdepthsbelowacylindricalcomponentsurfaceandfortwodifferentscanplans.TheplotontheleftillustratestheuseofthePODmodeltoquantifythedetectioncapabilityofanNDEsystem.Forthespecificparametersinthatsimulation,cracksthatareotherwiseidenticalhavesignificantlydifferentdetectabilitylevelsdependingontheirdepthbelowthesurfaceofthepart.ThisexampleillustratestheuseofthePODmodelbothforquantifyingthecapabilityofaflawdetectionsystemandforsuggestingimprovementsineitherthesystemoritsoperationthatcanimproveitscapability.

Figure4-3PODcurvesfortwoscanningplans.Source:GrayandThompson,1986.

Page 109: Computer-Aided Materials Selection During Structural Design

Page36

IssuesConcerningImplementingModelingandAnalysisSystems

IfmodelingandanalysissystemsinaCAMSSaretobeusefulandeffective,futureengineersmustreceivesufficienttraininginboththetheorybehindthesesystemsandtheirapplicationtothedesignprocess.ApreviousNationalMaterialsAdvisoryBoardreportentitledEnablingTechnologiesforUnifiedLife-CycleEngineeringofStructuralComponentsstatedthateducationinmaterialssynthesisandprocessingisabarrierthatmustbemutuallyaddressedbyindustryandU.S.institutionsofhigherlearning(NRC,1991).ThetimetointroduceCAMSSandtoperfecttheskillsinusingthetechniquewillbedependentontheavailabilityofindividualswithexpertiseincomputerandmaterialsscience.Someengineers,suchasmanufacturingmechanicalengineers,arehighlyskilledatusingcomputersanddoingFEAbutdonothavesufficientknowledgeaboutthebehaviorofmaterialsunderprocessingconditions.Incontrast,materialsscientistsandengineershaveabetterunderstandingofmaterialbehaviorbutlacksufficienttrainingintheuseofcomputersandcomputer-aidedsystemsinmanufacturing.Also,nondegreedtechnicalpersonnelperformmanycrucialtasksthroughoutengineeringandmanufacturingthatwillbedependentonthesenewtechnologies.Institutionsofhigherlearningwillhavetodevelopinterdisciplinaryprogramsledjointlybyexpertsinmaterialsscienceandengineering,design,andcomputerscienceifaCAMSSistobeproperlyimplemented.

Page 110: Computer-Aided Materials Selection During Structural Design

Page37

5ConclusionsandRecommendationsThisreporthasdiscussedthestructuralengineeringdesignprocess(Chapter2),thevisionforaCAMSS(Chapter3),andthekeymaterials-specificinformationtechnologiesthatcouldimpactthedevelopmentofaCAMSS(Chapter4).

ThecommitteehasidentifiedtwomainareasthatarecurrentlypreventingthedevelopmentofaCAMSS:(1)databaseandknowledge-basedesign,implementation,instantiation,andmanagementand(2)structuraldesignmodelingtechnologies.ThebarrierswithintheseareasandtherecommendedR&Drequiredtoovercomethesebarriersarediscussedinthefirstsectionofthischapter.Generalrecommendationsforgovernment,industry,anduniversitycollaborationtoforwardthedevelopmentofCAMSSarediscussedinthesecondsectionofthischapter.

STRATEGIESFOROVERCOMINGBARRIERS

ThecommitteeidentifiedtwomainareasthatarecurrentlypreventingthedevelopmentofaCAMSS:(1)databaseandknowledge-basedesign,implementation,instantiation,andmanagementand(2)structuraldesignmodelingtechnologies.

DatabaseandKnowledge-BaseBarriers

Thedesign,implementation,instantiation,andmaintenanceofmaterialspropertiesdatabasesandknowledgebasesareintegraltothedevelopmentofaneffectiveCAMSS.Forexample,adesignengineercannotuseasystemiftheunderlyingdatabasescontainobsolete,extraneous,unverified,orincompleteinformation.Thecommitteehasfoundthatthedatabaseandknowledge-baseareaiscurrentlyinhibitedbyfivebarriers.

Page 111: Computer-Aided Materials Selection During Structural Design

byfivebarriers.

1.StandardizationofdatabasesandknowledgebasesConstructingdatabasesandknowledgebasesthatcontaintherelevantinformationrequiredforthedesignprocessanddevelopingsystemsthatlocateandpresentthisdataaretwodifficultproblemsbecauseoftheamountofextraneousinformationavailableandthelackofstandardsinthecontentofdatabasesandknowledgebases.Toovercomethesebarriers,thecommitteerecommendsthat(1)standardsandguidelinesbedevelopedforelectronicdataquality,capture,storage,analysis,andexchange(followingtheComputer-AidedAcquisitionandLogisticsSupportandtheStandardfortheExchangeofProductapproaches)andknowledge-basecontentandconstruction;(2)CAMSSbedesignedtoacceptavarietyofdatabasetaxonomiesthroughtheuseofactive,''intelligent"datadictionariesthataidtheidentificationandconversionofthecontentsofdifferentdatabasesforuseinthesystem;(3)linksbetweenmaterialsdatabasesandknowledgebasesbeimprovedandcomputernetworksformaterials-specificinformationcommunicationbecreated(e.g.,anelectronicJournalofMaterialsSelectioninStructuralDesign,anationalmaterialsbulletinboardonInternet,oralinkednetworkofworldwidematerialsdatasystems);and(4)electronictechnicalassistancebeprovidedtodesignteamsinelectronicformats.

2.StatusofknowledgecaptureMethodsforknowledgecapturearerequiredtoenhancethelessons-learnedsegmentofCAMSS.Theseincludeestablishingknowledge-representationtaxonomies,technicalcontextstandards,andtechniquestoupdateandaccessthisinformationrapidly.Toovercomethisbarrier,thecommitteerecommendsthat(1)materialsandcomputerscientistscollaborateinthedevelopmentofsuitableknowledge-capturesystemsforusein

Page 112: Computer-Aided Materials Selection During Structural Design
Page 113: Computer-Aided Materials Selection During Structural Design

Page38

CAMSS;(2)industrydesignteamsbeencouragedtoestablishelectronictechnicaldatabasesbyelectroniccaptureofalldesigndiscussions,decisions,andlessonslearnedinfreetext,spreadsheet,CADstandards,andothermultimediaformats;and(3)industrydesignteamsbeencouragedtoassignspecificfunctionswithintheteamtospecialize,categorize,index,andfiltertheaccumulateddesignknowledgebaseandlocateandaccessotherdesignknowledgebases

3.DiffuseresponsibilityforgeneratingdatabasesTheissueofhowtocoordinatematerialsdevelopers,componentusers,andmaterialssocietiestogenerateandintegratematerialspropertydatabasesrequiresresolution.Materialssupplierspredominantlyleavematerialsqualificationprogramstotheuserbecauseofconcernsthattheywillbeheldliableforsystemmalfunctionscausedbyfailuresandthatuserswillonlyemploymaterialsthattheythemselveshavequalified.Materialssocietiesgenerallydonothavetheresourcesnecessaryforlargeprojects.Componentmanufacturerstypicallyonlyqualifymaterialsforagivenapplicationandtreatthedataasproprietary.Toovercomethisbarrier,thecommitteerecommendsthat(1)nationalteameffortsofusers,suppliers,materialssocieties,andstandardsorganizationsdevelopintegratedmaterialqualificationprogramsthatrelatetodesignrequirementsandeventualuseand(2)theresultantappropriate,independentlyverifieddatabemadeavailableinanationalinformationinfrastructuretoprovidearealistic,initialappraisaloftheadvantagesofamaterial.

4.DisclosureofmaterialsdataIngeneral,companiesprotectasproprietaryalldatabasesandknowledgebasesthatcontainmaterialspropertiesandproduction-relateddata,suchas(1)state-of-the-artinformation,projections,orforecasts;(2)manufacturinglaborstandards,rates,andpricedata;and(3)weight,performance,andcosttradeoffdataandcriteria.Toovercomethisbarrier,thecommitteerecommendsthatCAMSSbedesignedtoassurethatproprietaryportionsofdatabasesandknowledgebasesarefullyprotected.

Page 114: Computer-Aided Materials Selection During Structural Design

portionsofdatabasesandknowledgebasesarefullyprotected.5.InvestmenttomaintaindatabasesItisimportantthattheinformationwithinadatabasebeconstantlymonitored,verified,andupdatedtoensurethatthebestpossibleinformationisavailable.Toovercomethisbarrier,organizationsmust(1)assigntheresponsibilityformaintenanceofdatabasestoacentralizedfunction,suchasadataadministratorwithdomainexpertsidentifiedtoactascuratorsoftheknowledgebase,and(2)providelong-termsupportfordatabasemaintenanceoncetheprogramisestablished.

StructuralDesignModelingTechnologyBarriers

ModelinginstructuraldesignwillbeanimportantcomponentofanyCAMSSbothtoprovidevaliddetailsonwhichtobasetradeoffdecisionsandtoreducerelianceonlyonmaterials-propertiesdatabases.Modelingtechniquesarerequiredforgeometricreasoning,materialresponsesonmultiplescalelevels,materialsprocessing,manufacturingprocessingperformance,productperformance,andlife-cycleissuessuchasinspectability.Modelingtechniqueswillalsoberequiredthatsimulatenewmaterialsbysuccessiveextrapolationfromthepropertiesofexistingmaterialsorbycalculationfromfirstprinciples.Thecommitteeidentifiedtwobarrierstothedevelopmentofmodeling.

1.OptimizationmodelingAsopposedtosimplyshowingtradeoffsbetweendesignparametersinputbyusers,modelingtechniqueswillberequiredthatcansuggestmodificationstooptimizedesignsandmanufacturingprocesses.Processoptimizationisanimportantingredientofintegratedproduct-processdesignandwillbeusedmoreandmoreinthefutureastheindustryfullyadoptsconcurrentengineeringtoreducemanufacturingcostsandconvergeonmanufacturingsolutionsinashortertime.Tobeuseful,modelingmustalsobedonerapidlyandaccurately,usingnormaldesignparametersandinformationfrommultipleknowledgebases.Ifmodelingtechniquesaretooslow,untrustworthy,orunabletoaccesstheproperinformation,theywilllanguish.Toovercomethesebarriers,the

Page 115: Computer-Aided Materials Selection During Structural Design

information,theywilllanguish.Toovercomethesebarriers,thecommitteerecommendsthat(1)materialsscientistsand

Page 116: Computer-Aided Materials Selection During Structural Design

Page39

computerengineersfromindustryanduniversitycollaboratetodevelopadvancedmodelingtechniquestoreducerelianceonstraightmaterialsdata,introduceexpertknowledge,provideacrediblebasisfortradeoffdecisions,andincreasetrustinCAMSS;and(2)materialsscientistsparticipateinbasicandappliedresearchthatestablisheslinksbetweenmaterialsmodelsatseveralscales(e.g.,atomic,molecular-crystal,cluster-grainsize,polycrystal-aggregate,sub-structure,structure,andsystem).

2.CulturalandeducationalbarrierstoimplementingmodelingandanalysistechnologyThedesignprocessistraditionallyaheuristictrial-and-errorapproach.Increasedrelianceonmodelingtechniquesrequiresestablishingconfidencethattheimproveddesignsolutionscanbedevelopedinashortertimeperiod.Currentengineeringprogramsdonotstresstheimportanceoftrainingineithermaterialssynthesisandprocessingorcomputerscience.FormodelingandanalysissystemsinaCAMSStobeusefulandeffective,futureengineersmustreceivetrainingincomputersystems,modelingandanalysissystemstheory,andtheirapplicationtothedesignprocess.Toovercometheculturalandeducationalbarriers,thecommitteerecommendsthatinstitutionsofhigherlearningdevelopinterdisciplinaryprogramsledjointlybyexpertsinmaterialsscienceandengineering,design,andcomputersciencethat(1)exposestudentteamstobasicapproachestocomputer-assistedconcurrentengineeringdesignsystemsinordertoproduceknowledgeableworkerswithabroadunderstandingofthescienceofprocessing,(2)trainjourneymenormastertechnologiststousethisnewtechnologytopushacceptanceofprocessmodelingtechniquestotheshopfloor,and(3)encourageyoungerfacultymemberstocollaboratewithcolleaguesinotherdepartments(e.g.,materialsscience,thetraditionalengineeringfields,andcomputerscience)oninterdisciplinarydesignprojectsandcomputer-assistedconcurrentengineeringdesignsystems.

Page 117: Computer-Aided Materials Selection During Structural Design

systems.

GENERALCONCLUSIONSANDRECOMMENDATIONS

TheareasinhibitingthedevelopmentandimplementationofCAMSSdiscussedabovecanonlybeovercomebyamultiprongedinitiativewithfullparticipationandsupportbytheIPDTsandmaterialsandcomputerscientistsandengineersinthegovernmentR&Dagencies,universities,andindustrialorganizations.

Theimplementationofthisvisionwillrequire(1)thedevelopmentofsignificantdemonstrationsofCAMSSanddisseminatingtheresults;(2)thecontinuedexpansionofelectronicstorageofmaterialsinformation;(3)therapidadoptionandapplicationofdevelopingmethodsofcomputerscienceandtechnologytoenhancetherepresentationofmaterialsdesignknowledge;(4)thecontinueddevelopmentofmultilevel(atomistictomacroscopic)materialsprocessingandconstitutivebehaviormodelsthatreliablypredictperformanceandmanufacturabilityatthescaleofapplication;and(5)theimplementationofmethodstoaddressinspectability,reliability,andmaintainability.

Adherencetouniformcomputingandmaterialsdescriptionstandardsinsuchprogramsisessentialtothenetworkedlinkingofindividualtoolsintomuchlargerdesignknowledgeandsupportsystemsinthefuture.Thecommitteerecommendsahigherlevelofcommunicationamongmaterials-specificinformationsystemsresearchersanddevelopersthroughamoreformalelectronicinterchangeofresearchinformation,network-linkeduseofcomputer-aidedsystemtools,andaccesstoelectronicmaterialsknowledgebases.

RecommendationsspecifictodevelopersandusersofCAMSSare

Governmentpolicymakersshouldpromote(1)thedevelopmentofpre-competitiveR&Dprogramsthatencourageindustry,university,andgovernmentlaboratoriestoleverageexpertiseandknowledgetoreducethetimetodevelop,standardize,andimplementproductdesignsupport

Page 118: Computer-Aided Materials Selection During Structural Design

thetimetodevelop,standardize,andimplementproductdesignsupportsystemsandmaterials-specificinformationtechnologiesand(2)theuseoftheinformationsuper-highwayasameansforexpeditingthesharingoftechnicalinformationandmemoryamongfederalagencies,industries,andmaterialssocieties.GovernmentR&Dorganizations(DepartmentofDefense,AdvancedResearchProjectsAgency,

Page 119: Computer-Aided Materials Selection During Structural Design

Page40

NationalAeronauticsandSpaceAdministration,NationalInstituteofStandardsandTechnology)shouldpromotedatabaseandknowledge-baseconstructionandstandardization,design-knowledgetooldemonstrations,andpilotprojectsaspartoftheirfuturesystemsprograms.Theseprogramsshouldintegrateexistingcomputer-aidedsystemtools.Twopotentialwaysinwhichthismightbeaccomplishedaretoprovide(1)fundingfordemonstrationprogramswithcreativeproblemsolvinganddesignconceptstoteamsofuniversityfacultyandstudentscomposedofcomputerscientists,engineeringdesignspecialists,materialsscientists,andcognitivepsychologistsand(2)financialincentivestoindustryforsharingmaterialspropertydatawhereinputtopublicandlimitedaccessmaterialsknowledgebasescanbecontrolled.Industriesanduniversitiesshouldbeencouragedtocollaboratein:1.developingandusingwell-definedstandardsforelectronicinformationsharingtoenableselectiveprotectionoforganizationalprivatedata,companyproprietarydata,andindustryrestricteddatafromthepublicdomaindata;

2.improvingcontactbetweenresearcher,designer,andsupplierondesignteams;

3.increasingrateofgeneration,validation,andexchangeofmaterialsdata;

4.developingpowerfulprogramsforservicelifepredictionofstructuralcomponentsfrommaterialsdata,constitutivemodels,andin-servicenondestructivetesting;

5.developingmodelsofpracticalsignificancetoproductdevelopment;6.providingmaterialsdevelopmentdatainmachinereadableelectronicformat;

7.preparingstandardsforknowledgerepresentationofmaterialsinformation(e.g.,propertiestables,graphs,andpictorialdescriptionsofmicrostructures);

8.publicizingsuccessstorieswhereexperiencedengineersselect

Page 120: Computer-Aided Materials Selection During Structural Design

8.publicizingsuccessstorieswhereexperiencedengineersselectmaterialsshowingthatproperrepresentationstogetherwithreasoningexampleswillpromoteeffectivematerialcomputer-aidedsystemsdevelopment;and

9.developinganinformationbaseonavailable(networkaccessible)materialsdatabasesandcomputer-aidedsystems.

Page 121: Computer-Aided Materials Selection During Structural Design

Page41

ReferencesAchenbach,J.D.,andD.O.Thompson.1991.Towardsquantitativenon-destructiveevaluationofagingaircraft.Pp.114inStructuralIntegrityofAgingAirplanes,S.N.Atluri,S.G.Sampath,andP.Tong,eds.NewYork:Springer-Verlag.

ACM(AssociationforComputingMachinery).1993.Multimediaintheworkplace.CommunicationsoftheAssociationforComputingMachinery36(1).

ACM(AssociationforComputingMachinery).1994.SpecialIssueonIntelligentAgents.CommunicationsoftheAssociationforComputingMachinery37(7).

Allen,B.P.1994.Case-basedreasoning:Businessapplications.CommunicationsoftheAssociationforComputingMachinery,37(3):4142.

Ambler,E.1985.EngineeringPropertyData:ANationalPriority.ASTMStandardizationNews,August:4650.

Annis,C.G.,M.C.VanWanderham,J.A.Harris,andD.L.Sims.1981.Gasturbineenginediskretirement-for-cause:AnapplicationoffracturemechanicsandNDE.JournalofEngineeringforPower103(1):198200.

Ashby,M.F.,andD.R.H.Jones.1980.EngineeringMaterials:AnIntroductiontoTheirPropertiesandApplications.NewYork:PergamonPress.

BoeingCommercialAirplaneGroup.1991.PresentationstotheCommitteeonApplicationofExpertSystemstoMaterialsSelectionDuringStructuralDesign,Seattle,December1011.

Page 122: Computer-Aided Materials Selection During Structural Design

Brathwaite,K.S.1988.Analysis,Design,andImplementationofDataDictionaries.NewYork:McGraw-Hill.

Burte,H.M.,andC.L.Harmsworth.1989.DatabaseR&Dforunifiedlifecycleengineering.Pp.197199inComputerizationandNetworkingofMaterialsDataBases,J.S.GlazmanandJ.R.Rumble,eds.ASTMSpecialTechnicalPublication1017.Philadelphia,Pennsylvania:ASTM.

Clocksin,W.F.,andV.S.Melish.1981.ProgramminginPROLOG.NewYork:Springer-Verlag.

Cowie,W.D.1989.FractureControlPhilosophy.Pp.666673inMetalsHandbook:NondestructiveEvaluationandQualityControl,Vol.17.MetalsPark,Ohio:ASM,International.

Coyne,R.F.1991.ABLOOS:AnEvolvingHierarchicalDesignFramework.Ph.D.dissertation.DepartmentofArchitecture.CarnegieMellonUniversity,Pittsburgh,Pennsylvania.

Coyne,R.F.,andU.Fleming.1990.Planningindesignsynthesis:Abstraction-basedLOOS.Pp.91111inProceedingsoftheFifthInternationalConferenceonApplicationsofArtificialIntelligenceinEngineering,Vol.1,J.S.Gero,ed.Southampton,England:ComputerMechanicsPublications.

Davis,R.,H.Shrobe,andP.Szolovits.1993.Whatisknowledgerepresentation?AIMagazine14(1):1733.

DOD(DepartmentofDefense).1986.MilitaryStandardizationHandbook:MetallicMaterialsandElementsforAerospaceVehicleStructures,MIL-HDBK-5E.Washington,D.C.:DepartmentofDefense.

Eberhard,E.,andD.D.Vvedensky.1987.Localizedgrain-boundaryelectronicstatesandintergrannularfracture.PhysicsReviewLetters58(1).

Page 123: Computer-Aided Materials Selection During Structural Design

Famili,A.,D.S.Nau,andS.H.Kim,eds.1992.ArtificialIntelligenceApplicationsinManufacturing.Cambridge,Massachusetts:MassachusettsInstituteofTechnologyPress.

Gray,T.A.,andR.B.Thompson.1986.UseofmodelstopredictultrasonicNDEreliability.Pp.911918inReviewofProgressinQuantitativeNondestructiveEvaluation,Vol.5,D.O.ThompsonandD.E.Chimenti,eds.NewYork:PlenumPress.

Gray,J.N.,T.A.Gray,N.Nakagawa,andR.B.Thompson.1989.ModelsforPredictingNDEReliability.Pp.702715inMetalsHandbook:NondestructiveEvaluationandQualityControl,Vol.17.MetalsPark,Ohio:ASMInternational.

Gruber,T.,C.Baudin,J.Boose,andJ.Weber.1991.Designrationalecaptureasknowledgeacquisition:Tradeoffsinthedesignofinteractivetools.Pp.312inMachineLearning:ProceedingsoftheEighth

Page 124: Computer-Aided Materials Selection During Structural Design

Page42

InternationalWorkshop,L.A.BirnbaumandG.C.Collins,eds.SanMateo,California:MorganKaufmann.

Gursoz,E.L.,andF.B.Prinz.1990.Apointsetapproachingeometricmodelling.Pp.7387inProceedingsoftheInternationalSymposiumonAdvancedGeometricModellingforEngineeringApplications,R.L.Krause,andH.Jansen,eds.NewYork:ElsevierSciencePublications.

Hadcock,R.N.1985.X-29CompositeWing.PresentedattheAmericanInstituteofAeronauticsandAstronauticsSymposiumontheEvaluationofAircraft/AerospaceStructuresandMaterials,AirForceMuseum,Dayton,Ohio,April2425.

Hadcock,R.N.1986.StructuresandMaterialsTechnology.PresentedattheGrummanAircraftSystemsDivisionForum,Bethpage,NewYork,August26.

Harmon,P.,andD.King.1985.ExpertSystems.NewYork:JohnWiley&Sons.

Harrison,R.J.,andI.Hulthage.1987.Specificationofknowledgebasesystems.InProceedingsofthe34thSagamoreArmyMaterialsResearchConference.Watertown,Massachusetts:MaterialsTechnologyLaboratory.

Hayes-Roth,F.,andN.Jacobstein.1994.Thestateofknowledge-basedsystems.CommunicationsoftheAssociationforComputingMachinery37(3):2739.

IEEE(InstituteofElectricalandElectronicEngineers).1993.IEEESpectrumSpecialReportonInteractiveMultimedia30(3).

Kaufman.J.G.1986a.Standardizationformaterialspropertydatabasesandnetworking.ASTMStandardizationNews.February:2833.

Page 125: Computer-Aided Materials Selection During Structural Design

Kaufman,J.G.1986b.Technicalchallengesandthestatus:thenationalmaterialspropertydatanetwork.Pp.159166inTheMaterialPropertyData:ApplicationsandAccess,Vol.1.NewYork:AmericanSocietyofMechanicalEngineers.

Kaufman,J.G.1988.AnIntelligentKnowledgeSystemforSelectionofMaterialsforCriticalAerospaceApplications(IKSMAT).FinalTechnicalReport,U.S.AirForceContractF33615-87-C-5305.Columbus,Ohio:DepartmentofDefense.

Kim,S.H.1990.TheEssenceofCreativity:AGuidetoTacklingDifficultProblems.NewYork:OxfordUniversityPress.

Kiridena,V.,A.B.Chaudhary,andJ.Gunasekera.1989.Analysisoflargedeformation3-Dmetalforminganalysiswithintermediateremeshing.InNumericalMethodsinIndustrialFormingProcess,E.G.Thompson,ed.Boston,Massachusetts:A.A.Balkema.

Klahr,D.,P.Langley,andR.Neches,eds.1987.ProductionSystemModelsofLearningandDevelopment.Cambridge,Massachusetts:MassachusettsInstituteofTechnologyPress.

McCarthy,J.1968.Programswithcommonsense.InSemanticInformationProcessing,M.Minsky,ed.Cambridge,Massachusetts:MassachusettsInstituteofTechnologyPress.

Minsky,M.,ed.1968.SemanticInformationProcessing.Cambridge,Massachusetts:MassachusettsInstituteofTechnologyPress.

Mobley,C.E.1992.QualityCastings:CT17CastabilityMaps.Kettering,Ohio:EdisonMaterialsTechnologyCenter.

NRC(NationalResearchCouncil).1983.MaterialsPropertiesDataManagementApproachestoaCriticalNationalNeed.NationalMaterialsAdvisoryBoard,NRC.Washington,D.C.:NationalAcademyPress.

Page 126: Computer-Aided Materials Selection During Structural Design

NRC(NationalResearchCouncil).1989.MaterialsScienceandEngineeringforthe1990s:MaintainingCompetitivenessintheAgeofMaterials.NationalMaterialsAdvisoryBoard,NRC.Washington,D.C.:NationalAcademyPress.

NRC(NationalResearchCouncil).1991.EnablingTechnologiesforUnifiedLife-CycleEngineeringofStructuralComponents.NationalMaterialsAdvisoryBoard,NRC.Washington,D.C.:NationalAcademyPress.

NRC(NationalResearchCouncil).1993.CommercializationofNewMaterialsforaGlobalEconomy.NationalMaterialsAdvisoryBoard,NRC.Washington,D.C.:NationalAcademyPress.

Parsaye,K.,M.Chignell,S.Khoshafian,andH.Wong.1989.IntelligentDatabases:Object-Oriented,Deductive,HypermediaTechnologies.NewYork:JohnWiley&Sons.

Piatetsky-Shapiro,G.,andW.J.Frawley,eds.1993.KnowledgeDiscoveryinDatabases.Cambridge,Massachusetts:MassachusettsInstituteofTechnologyPress.

Rappaport,A.,andR.Smith.1991.InnovativeApplicationsofArtificialIntelligence,Vol.2.Cambridge,

Page 127: Computer-Aided Materials Selection During Structural Design

Page43

Massachusetts:MassachusettsInstituteofTechnologyPress.

Reynard,R.1987.Computerizedmaterialsdata:TheSDIview.InProceedingsofWorkshoponComputerizedAerospaceMaterialsData,J.H.WestbrookandL.R.McCreight,eds.NewYork:AmericanInstituteofAerospaceandAeronautics.

Richmond,O.1992.ConstitutiveModel-BasedMaterialProductDesign.PresentedtotheNationalMaterialsAdvisoryBoardCommitteeontheApplicationofExpertSystemstoMaterialsSelectionDuringStructuralDesign,April27,Washington,D.C.

Robinson,S.1987.TheSpangRobinsonReport,Vol.3,Number3.PaloAlto,California:SpangRobinson.

Robinson,J.A.,andE.E.Sibert.1981.TheLOGSLISPUser'sManual.TechnicalReport.Syracuse,NewYork:SchoolofComputerandInformationScience,SyracuseUniversity.

Rumelhart,D.E.,B.Widrow,andM.A.Lehr.1994.Thebasicideasinneuralnetworks.CommunicationsoftheAssociationforComputingMachinery37(3):8792.

Schurr,H.,andA.Rappaport.1989.InnovativeApplicationsofArtificialIntelligence,Vol.1.Cambridge,Massachusetts:MassachusettsInstituteofTechnologyPress.

Smith,R.,andA.C.Scott.1991.InnovativeApplicationsofArtificialIntelligence,Vol.3.Cambridge,Massachusetts:MassachusettsInstituteofTechnologyPress.

Thompson,D.O.1993.ToolsforaNDEEngineeringBasis:Flight-VehicleMaterials,StructuresandDynamicsTechnologyAssessmentandFutureDirections.NewYork:AmericanSocietyofMechanicalEngineers.

Page 128: Computer-Aided Materials Selection During Structural Design

Weber,J.C.,B.K.Livezey,J.G.McGuire,andR.N.Pelavin.1991.Integratingspecializedrepresentationsforspreadsheet-likedesign.Pp504511inProceedingsofFourthInternationalConferenceonIndustrialandEngineeringApplicationsofArtificialIntelligence.Tulahoma,Tennessee:UniversityofTennesseeSpaceInstitute.

Weiss,V.,andD.W.Aha.1984.Materialsselectionwithlogicprogramming.Pp.506518inProceedingsoftheSocietyfortheAdvancementofMaterialandProcessEngineeringSymposiumonTechnologyVectors,Vol.29.Covina,California:SocietyfortheAdvancementofMaterialandProcessEngineering.

Wertz,C.J.1989.TheDataDictionaryConceptsandUses,SecondEdition.Wellesley,Massachusetts:QEDInformationSciences.

Whitney,D.E.,J.L.Nevins,T.L.DeFazio,R.E.Gustavson,R.W.Metzinger,J.M.Rourke,andD.S.Seltzer.1988.Thestrategicapproachtoproductdesign.Pp.200223inDesignandAnalysisofIntegratedManufacturingSystems,W.D.Compton,ed.Washington,D.C.:NationalAcademyPress.

Widrow,B.,D.E.Rumelhart,andM.A.Lehr.1994.Neuralnetworks:Applicationsinindustry,business,andscience.CommunicationsoftheAssociationforComputingMachinery37(3):93105.

Winner,R.I.,J.P.Pennell,H.E.Bertrand,andM.M.Slusarczuk.1988.RoleofConcurrentEngineeringinWeaponsSystemsAcquisition.IDAReportR-338(unclassified).Alexandria,Virginia:InstituteforDefenseAnalysis.

Page 129: Computer-Aided Materials Selection During Structural Design

Page45

AppendixA:GlossaryofAcronyms

AI Artificialintelligence

CAD Computer-aideddesignCAD/CAMComputer-aideddesignandmanufacturingCAM Computer-aidedmanufacturingCAMSS Computer-aidedmaterialsselectionsystemDBT DesignbuildteamDMM DynamicmaterialsmodelingFEA FiniteelementanalysisFEM FiniteelementmethodIKSMAT Intelligentknowledgesystemforselectionofmaterialsfor

criticalaerospaceapplicationsIPDT IntegratedproductdevelopmentteamKIDS Knowledge-basedintegrateddesignsystemNDE NondestructiveevaluationOEM OriginalequipmentmanufacturerPOD ProbabilityofdetectionQNDE QuantitativenondestructiveevaluationR&D ResearchanddevelopmentSME Small-to-mediumenterprise

Page 130: Computer-Aided Materials Selection During Structural Design

Page47

AppendixB:CaseStudiesReviewedbytheCommitteeAMaterialsSelectionExpertSystemforCorrosiveAqueousEnvironments,V.Weiss,SyracuseUniversity,NewYork

DesignforHigh-SpeedCivilTransportApplications,P.Rimbos,HSCTStructuresTechnologyDevelopment,TheBoeingCompany,Seattle,Washington

TheRoleofMaterialsEngineersinHardwareDesign,T.Richardson,777AirplaneDevelopment,TheBoeingCompany,Seattle,Washington

AllowablesPerspectiveofMaterialsSelectionintheDesignProcess,B.F.Backman,TheBoeingCompany,Seattle,Washington

DesignBuildTeam(DBT)ApproachtoProductDevelopment,H.Shomber,Design777Division,TheBoeingCompany,Seattle,Washington

DBTExperiencesinCurrentHardwarePrograms,A.FalcoandT.Lackey,DesignEngineers777Empennage,TheBoeingCompany,Seattle,Washington

Computer-AidedDesignToolsCurrentlyinPractice,T.S.Kaczmarek,ArtificialIntelligence,GeneralMotorsCorporation,Warren,Michigan

QualityAssurancePerspectiveofDBTNewAirplaneProgram,B.Das,777QualityAssurance/QualityEngineering,TheBoeingCompany,Seattle,Washington

NASAFundedACTProgram:COINSandCOSTADEDesignTools,L.

Page 131: Computer-Aided Materials Selection During Structural Design

Ilcewicz,AdvancedTechnologyCompositeAircraftStructures,TheBoeingCompany,Seattle,Washington

CompositeMaterialsSelection:ASuppliersViewPoint,J.Hendrix,HerculesIncorporatedWashington,DC

ConstitutiveModel-BasedMaterialProductDesign,O.Richmond,ALCOA,AlcoaCenter,Pennsylvania

IntelligentProcessingandMaterialsModeling,W.Barker,ARPA,DepartmentofDefense,Arlington,Virginia

DesignKnowledgeCaptureforCorporateMemory,J.Boose,BoeingAdvancedTechnologyCenter,ComputerScience,TheBoeingCompany,Seattle,Washington

DesignCostModelStudiesforAdvancedCompositeFuselage,G.Swanson,AdvancedTechnologyCompositeAircraftStructures,TheBoeingCompany,Seattle,Washington

MaterialSelectionforCommercialAirplanes,A.Miller,BeoingMaterialsTechnology,TheBoeingCompany,Seattle,Washington

IntegrationofMaterialService/Life-CycleConsiderationsintheDesignProcessA.Miller,BoeingMaterialsTechnology,TheBoeingCompany,Seattle,Washington

ArtificialIntelligenceinDesignofMaterialsandStructures,F.Crossman,LockheedPaloAltoResearchLaboratory,PaloAlto,California

LifePredictionData/MethodologyforaMSES,J.Schreurs,WestinghouseElectricCompany,WestinghouseScienceandTechnologyCenter,Pittsburgh,Pennsylvania

KnowledgeEngineeringandRepresentationMethodsforMaterialSelection,T.S.Kaczmarek,ArtificialIntelligence,GeneralMotorsCorporation,Warren,Michigan

Page 132: Computer-Aided Materials Selection During Structural Design

SuperconductorSearch:AnExpertSystemfortheDevelopmentofHigh-TemperatureSuperconductors,J.Schreurs,WestinghouseElectricCompany,WestinghouseScienceandTechnologyCenter,Pittsburgh,Pennsylvania

DesignSpecificationsforKnowledge-BasedSystemsforMaterialsDesign,I.Hulthuse,RoboticsInstitute,Carnegie-MellonUniversity

AnIntelligentKnowledgeSystemforCriticalAerospaceSystems,W.M.Griffith,WrightPattersonAirForceBase,Ohio

AdvancedMaterialsDatabaseSystem,W.M.Griffith,WrightPattersonAirForceBase,Ohio

ExpertSystemforFailureAnalysisofAircraftMetallicMaterialsandGroundSupportEquipment,W.M.Griffith,WrightPattersonAirForceBase,Ohio

ADatabaseManagementSystemforMMCs,W.M.Griffith,WrightPattersonAirForceBase,Ohio

Page 133: Computer-Aided Materials Selection During Structural Design

Page48

AdvancedCeramicsInformationSystems,R.G.Munro,CeramicsDivision,NationalInstituteofStandardsandTechnology,Gaithersburg,Maryland

M/VisionMaterialsSelectionSystem,D.Marinaro,SoftwareEngineeringGroup,PDAEngineering,CostaMesa,California

DesignInformationSystem,F.Crossman,LockheedPaloAltoResearchLaboratory,PaloAltoCalifornia

Knowledge-BasedIntegratedDesignSystem(KIDS),H.L.Gegel,Director,ProcessingScienceDivision,UniversalEnergySystems,Dayton,Ohio

AnIntelligentKnowledgeSystemforSelectionofMaterialsforCriticalAerospaceApplications(IKSMAT),J.G.Kaufman,VicePresidentofTechnology,AluminumAssociation,Washington,D.C.

Page 134: Computer-Aided Materials Selection During Structural Design

Page49

AppendixC:ReviewofSelectedKnowledge-RepresentationTechniquesandToolsExpertsystemimplementationsemploymanydifferentknowledge-representationtechniquesandtools.Eachtechniqueprovidesanabstractionthatisusefulindescribingsomeaspectofexpertbehaviororanimprovedimplementationofanabstractionconcept.Justaswords,numbers,graphs,andsketchesaredifferentbutusefulabstractions,thetechniquesdescribedinthisappendixarevariouswaysofdescribingrelationshipsandreasoning.Toolsareimplementationsofknowledge-representationtechniques.Thisappendixreviewsseveralofthecurrentlyusedrepresentationtechnologiesandtoolsdiscussedinthereportandisnotmeanttobeexhaustive.Referencesareprovidedsothatinterestedreaderscanfurtherexplorethetechniquesdiscussedhere,aswellasmanyothers.

CASE-BASEDREASONING

Case-basedreasoningisamethodfordecisionmakingbasedontheretrievalandadaptationofpriorrecordedcases.Toolfunctionalitycanrangefromretrieval,whichonlyfindsrelevantcasesinresponsetoauser'sinput,toanalogicalreasoning,whichfindsandadaptsapriorsolutiontothecurrentsituation.Assuch,case-basedsystemsprovideatleastaprimitiveformoflearning.Commercialtoolsforassociativeretrievalandcase-basedretrievalareavailableandsignificantapplicationsarebeginningtoemerge.

CONSTRAINT-BASEDREASONING

Inconstraint-basedreasoning,knowledgeisencodedasconstraintsthatexpressqualitativeorquantitativerelationshipsbetweendesign

Page 135: Computer-Aided Materials Selection During Structural Design

parameters.Variousalgorithmsexisttoprovidevaryingsupportofconstraintsrangingfromviolationdetection,toenforcement,topropagation,tosatisfaction.

Whenreasoningaboutconstraints,theexpertsystemmustdecidewhichconstraintsarerelevanttotheproblemandtheninterpretthem.Constraintsandconstraintreasoningcansupportdesignanalysisbyidentifyingproblemareas.Duringdesignsynthesis,constraintscanbeexploitedtoproposeasolutionthatisacceptablewithintheproblemdomain,providedthattheproblemisnotoverlyconstrained.Asanaidtosearch,constraintscanbeusedtoconfinethesearchspace.Systemsthatsymbolicallysolvemathematicsproblemshavebeeninvestigatedsincetheearlydaysofartificialintelligence(McCarthy,1968;Minsky,1968).Constraintreasoningisamorerecenttechnologythathasevolvedinseveraldifferentstyles.Inparticular,reasoningaboutgeometricinvarianceiscriticaltospatialreasoning.Linkinggeometricreasoningtosymbolicreasoningwillbecriticalforexpertsystemtechnologyinmaterialselection.

Constraint-basedalgorithmsvaryincomplexityassupportrangesfromdetectionofviolationsthroughsatisfaction.Violationdetectioncanbeandhasbeendonewithavarietyofrule-basedlanguagesaswellasproceduralcode.Constraintpropagationisavailableinseveralcommercialproducts.Constraintsatisfactionorsolutionisanactiveresearcharea,althoughsomealgorithmswithlimitedcapabilitiesareavailableandinuse.

ACTIVEDATADICTIONARIES

Datadictionariesareorganizedreferencestodatacontainedinotherprograms,systems,databases,orcollectionsoffiles.Whereasdatabasesstoreandprocessordinarydataaboutobjects,datadictionariescontaindataaboutdata,ormetadata.Activedatadictionariesareusedtocoordinateandsupportdataretrievalandanalysisbetweendifferentsystemsordatabases.Althoughthe

Page 136: Computer-Aided Materials Selection During Structural Design

implementationofactivedatadictionariesispredominantlyaresearcharea,somelimitedcapabilitiesarecurrentlyavailableandinuseaspartofdatabasesystems.

Page 137: Computer-Aided Materials Selection During Structural Design

Page50

DATABASES

Simplydefined,databasesstoreinformationaccordingtoaspecifiedschema.Relationaldatabasesarecommonlyusedtoday,andtheystoredatalikethatrepresentedbytablesinreferenceworks.Databasemanagementsystemsareanimportantcomponentofmostexpertsystems.Theysupportdynamicfactualrecall,updating,anduseraccesscontrol,whichmaybethoughtofasaformofintelligentbehavior.Furthermore,manyoftherecentadvancesinstate-of-the-artdatabasemanagementsystemshaveincorporatedadvancedconceptsfromexpertdatabasesystems.

Relationalelectronicdatabasesareorganizedasrigidtables,whereeachrecordofthedatabaseisassignedanequalnumberoffields,eachcontainingaspecifiedtypeofentry.Suchrigidformattingisnolongernecessarywiththeflexibilityaffordedbyimplementationssuchasassociationlistsorstructuresandevenarraysofstructures.Free-formatdatabasescansavestoragespacewhenmostrecordsinthedatabasecontainentriesonlyforafewofthemanypossiblefields.Free-formatdatabasesarealsopreferredwhenthedatabaseisunstableandupdatedfrequently,notonlybyaddingrecordsbutalsobyaddingfieldsorbymodifyingtherequirementsonafield.However,rigidformattingsavesspaceandquerytimewhenmostrecordsofthedatabasecontainvaluesforthesamenumberoffieldsandallowsnonproceduralqueriestobemadethatautomaticallylinktogethermultipletables.

Moderndatabasepackagesaresoversatileandeasytousethatthematerialsscientisthardlyneedstoworryaboutdatabaseformatsaslongasthedataarewelldefinedquantitiesorarbitrarytext.However,analysismustbedoneinthedesignofadatabaseorknowledgebase.Factorstobeconsideredincludeexpectedqueryscenariosandunusualdatasets(e.g.,defaultvalues,multipleentries,dataranges,

Page 138: Computer-Aided Materials Selection During Structural Design

incidentalinformationsuchaswarningsandcomments,derivedvalues,constraintranges,quality,unitconversions,andeducatedguessesformissingentries).

Mostexpertsystemapplicationsinvolveextractinginformationfromexistingdatabases.Someexistingdatabasepackagesmaybeabletohandlecertaintypesofreasoningaboutdatawithinthedatabasestructureitself,butinmostcasestheknowledgeengineermusteitherselectadifferentformofknowledgerepresentationtosupportreasoningrequirementsordesignanewdatabasetohandlespecialproblems.Newtechnologyforknowledgediscoveryindatabasesshowspromiseformakingtheinformationmoreusefulintheconstructionofmoreadvancedreasoningsystems(Piatetsky-ShapiroandFrawley,1993;see''ActiveDataDictionaries'').Object-orienteddatabasesareaddingthesecurityandaccessfacilitiesofdatabaseswiththeflexibilityofartificialintelligencedatastructuresandmayovercomeperformancelimitationstobewidelyusedinthefuture.

Astrengthofcomputersistheabilitytoretrieveitemsstoredinadatabaseforuseinotherformsofknowledgerepresentationorfordisplaytoinformtheuser.However,thequantityofdataavailableformostadvancedmaterialsisinadequateforthetypesanddepthofanalysesneededforknowledge-basesystems,includingstatisticallybaseddesignvalues.Databasetechnologyhasoutstrippedtheefforttobuildanddistributereliabledata.

FUZZYLOGIC

Fuzzylogicisamethodfordealingwiththeinherentambiguitiesinconceptsandanattempttobuildaformallogicforplausibility.Fuzzylogicdoesnotdealwithprobabilitiesbut,rather,withthetypeofreasoningpeopleusewhenfacedwithinconclusiveorcontradictoryevidence.Fuzzylogicinvolvesfourbasicelements:(1)schemestoconvertstimulisignalstostrengthofbelief,(2)simplerulesexpressedinlogicalterms,(3)algorithmsforcomputingstrengthofbelieffor

Page 139: Computer-Aided Materials Selection During Structural Design

theconclusionofrules,and(4)outputfunctionstoconvertthebeliefinconclusionstoacontrolsignal.Unliketherulesfoundintypicalexpertsystems,whicharecomplexanddesignedtosupportdeeplogicalchains,fuzzylogicrulesareverysimpleanddonotinvolvethecombinationofconclusionstoinferotherconclusions.

Recentcomputingadvanceshaveledtosuccessfulapplicationsoffuzzylogicinareasrangingfrommanufacturingcontrolstoconsumerelectronicproducts.Themainadvantageoffuzzylogicisthatitprovidesasimpleformulationofsimplereasoningprocesses.Thesimplicityofthereasoningrestrictstheapplicationoffuzzylogictonarrowlyscopedproblems,however.

Page 140: Computer-Aided Materials Selection During Structural Design

Page51

GEOMETRICANDMICROSTRUCTURALINFORMATIONREPRESENTATION

Therearethreelevelsofgeometricmodelingcommoninindustry.One-dimensional(wireframemodeling),two-dimensional(surfacemodeling),orthree-dimensional(solidmodeling)analyticalelementsareusedinconstructingspatialrepresentations.FigureC-1displaystheessentialdifferenceintherepresentationaldomainofeachofthethreelevelsofmodeling.Itshouldbeclearthatsolidmodelinghasthemostexplicitrepresentationoftheobjectsinrealworldandincludesthelowerlevelsofrepresentation.Thisimplicitembodiment,however,doesnotnecessarilymeanthatanysolidmodelingapproachcanuniformlymanipulateentitiesofthelowerdimensionsaswellassolids.

Severaltaxonomiesforsolidmodelinghavebeenproposed,butonewayofcategorizingmostoftheexistingsolidmodelapproachesistoperceivetheminthreeclasses:

cell-basedrepresentations;constructivesolidgeometryrepresentations;andsurfaceboundaryrepresentations.1

Twodistinctapproachesinthecategoryofcellbasedrepresentationsarethecellenumerationtechniqueandtheoctreeapproach.Inbothofthesecases,asolidisdefinedasaunionofaselectionofspace-basedcubicalvolumes.Inconstructivesolidgeometryschemes,objectsareobtainedbycombiningasetofsolidprimitiveswithbooleanoperators.Insurfaceboundaryrepresentations,theobjectsarerepresentedbytheirenclosingshell.

HYPERDOCUMENTS

Hyperdocumentsaremultimediafilesinwhichthepiecesofthe

Page 141: Computer-Aided Materials Selection During Structural Design

documentsarelinkedtooneanothertocaptureimportantrelationshipsbetweenconceptspresentedinthedocuments.Severalcommercialsystemshaveledtonumeroussuccessfulapplications.Thecapturingoftherelationshipsbetweenobjectsactsasaprimitiveformofsemanticnetwork(see"ObjectsandTaxonomies").

MACHINELEARNING

Machine-learningtechniquesallowasystemtoacquireknowledgeautomatically.Somesimpletechniqueshavebeensuccessfullyappliedandarecommerciallyavailable,suchasintheareasofcase-basedreasoningandneuralnetworks,butmostarestillintheresearchphase.

MATHEMATICALRELATIONS

Mathematicalrelationsareequations,inequalities,approximations,anditerationsthatdesignersusetodeterminethepropertiesofmaterialsundercertainconditions.Mathematicalrelationsappearinexactlythesameformatinelectronicknowledgebasesastheydoinbooks.Thegreatadvantagethatcomputershave,however,isthattheycanactuallycomputevaluesusingequations,whereasbookscanonlydescribehowtocomputethevalues.Onedisadvantageofcomputersisthatthecovertuseofequationstorepresentknowledgecanbedangerous.Manyequations,particularlyinthematerialsanddesignfields,arebasedonapproximationsthatareonlyvalidforoneapplicationorwithinaspecificrangeofvariables.Goodknowledgerepresentationdemandstheexistenceofamechanism,commonlyreferredtoas"explanation,"topermittheusertoinspecttheequationsthatareinvokedandtheassumptionsthatareinherentinthechoiceoftheequationsusedinthecalculationtoavoidpotentialproblems.

Whendealingwithmathematicalrelations,theknowledgeengineermustdecidewhethertheserelationsaretobeusedassymbolicexpressionsorsimplecomputations.Iftheyaretobeusedascomputationsonly,thenconventionalprogramminglanguagescanbeusedtoperformthe

Page 142: Computer-Aided Materials Selection During Structural Design

1Surfaceboundaryrepresentationofasolidreferstotheclosingsurfaceofanotherwiseopensolid.Surfacemodeling,ingeneral,maycontainsurfacesthatdonotnecessarilyenclosesolids.Anexampleistwosurfacesthatmayinteractbutnonotformanenclosedshell.

Page 143: Computer-Aided Materials Selection During Structural Design

Page52

FigureC-1Thedifferencesbetweenwireframe,surfacemodel,andsolidmodelrepresentationaldomains

(Source:GursozandPrinz,1990;ReprintedcourtesyofElsevierSciencePublications).

arithmetic.Frequently,engineeringproblemsrequireusingmathematicalrelationsassymbolicexpressionsofconstraints,however.Allconstraintsarenottypicallythoughtofasmathematicalrelationshipsandanexpertsystemreasoningaboutconstraintsmustalsoadmitconstraintsexpressedinamorelogic-basedformalism(see"Constraint-BasedReasoning").

NEURALNETWORKS

Neuralnetworksattempttomimicbrain-likesystemsviasimplifiedmathematicalmodels.Researchershavefoundthatsimplemathematicalstimulus-responseequationscanbeusedtosimulatethebehaviorofneuronsinthebrain.Likethebrain,themostbasicprocessingunitofneuralnetworksistheneuron,whichischaracterizedby"anactivitylevel(representingthestateofpolarizationofaneuron),anoutputvalue(representingthefiringrateoftheneuron),asetofinputconnections(representingsynapsesonthecellanditsdendrite),abiasvalue(representinganinternalresting

Page 144: Computer-Aided Materials Selection During Structural Design

leveloftheneuron),andasetofoutputconnections(representinganeuron'saxonalprojections)"(Rumelhartetal.,1994).Neuralnetworksanalyzedatabymappinginputdataintooutputpatternsbasedonmapsproducedbypreviousruns.

Amajoradvantageofneuralnetworksisthatthesimplemathematicalrepresentationlendsitselftolearningalgorithms.Usingfeedback,thesealgorithmsadjustthesetcoefficientsusedtoreinforceandcombinestimulitominimizeanerrorscore.Neuralnetworklearningalgorithmsrequireverylargetrainingsetsandtypicallyworkbestwhenthenetworkconnectivityhasbeenproperlyorganizedinadvancebyanexpert.Successfulapplicationshavebeendevelopedmainlyinpatternrecognition.

Therearetwomaindrawbackswithneuralnetworks,however.Inadditiontorequiringlargetrainingsets,neuralnetworksdonothavestrongmechanismsforexplainingtheresultsofacomputation.Thelatterproblemisparticularlytroublesomeinareasofengineeringwheretractabilityofdesigndecisionsisarequirement,suchasinthedesignofproductsthataffectpublicsafety.

Whilemanyoftheconceptsofneuralnetworkshavebeeninvestigatedforquitesometime,thistechnologyisinitsearlystagesofapplication.Applicationshaveonlynowbecomefeasiblebecauseoflow-costcomputingdevelopments.

Page 145: Computer-Aided Materials Selection During Structural Design

Page53

OBJECTSANDTAXONOMIES

Objectsandtaxonomiesareknowledge-basetoolsthatallowprogrammerstorepresentknowledgeofphysicalorconceptualentitieswithmanyattributesinanabstractmannerthatmimicsthewaypeopleorganizeknowledgeaboutconceptsandclassesofobjects.Objectsandtaxonomiesareprobablythemostgeneralandflexibleformofknowledge-captureschemeavailable.Theycanhandledatabases,mathematicalrelations,rules,andanythingthatcanbeclassified,includingdesignfeaturessuchasshapesandcolors.Object-orientedprogrammingrequiresadifferentsoftwaredesignapproachthanconventionalprogrammingandisstillevolving.

Therearemanyconceptsthathavebeenexploredinartificialintelligenceandprogramminglanguageresearchthataresimilartoobjectsandtaxonomies,someofwhicharecommerciallyavailableinmanyformsaswellasembeddedinknowledge-baseengineeringtools.Abstractdatatypes,frames,schema,relationaltables,andsemanticnetworksarethemostcommonlyreferredtovariantsofthetechnology.Alloftheseprovideameanstodescribefactsandmeaningfulrelationshipsbetweenfacts.Theydifferfromdatatypesfoundinconventionalprogramminglanguagesanddatabasesintheexpressivepowerregardingrelationships.Thepricepaidforthisislessefficientprogramminganddifficultyinprovidingsharedaccesstodata.Objectsandabstractdatatypesprovideanadditionalbenefittotheprogrammerorknowledgeengineerbyassociatingtheprocessingorfunctionalelementsoftheimplementationwiththekindsofdatathatthefunctionscanmanipulate.Thus,theyprovidemorestructureandunderstandability.

Onedisadvantageofmaximizinggeneralityandflexibilityinasystemisthatagreatdealofexpertiseisusuallyrequiredtooperatethesystem,sothattheusereffectivelybecomesacomputerspecialistas

Page 146: Computer-Aided Materials Selection During Structural Design

wellasamaterialsscientistordesignspecialist.

REASONINGWITHUNCERTAINTY

Reasoningwithuncertaintyisaknowledge-representationtechniqueforcombiningcontradictory,incomplete,orinconclusiveknowledge.Thisisnotthesameasfuzzyknowledgeorfuzzylogic,however.Expertsystemscanaccommodateuncertaintybyseveralapproaches,includingmaintainingmultipleproblemformulations,qualitativemethods,andquantitativemethodsinvolvinguncertaintymeasures.Manyapplicationshaveusedsomeformofuncertaintylogic.

RULE-BASEDREASONING

Rulesarerepresentationsofknowledgeaboutwhichpatternsofinformationexpertsusetomakedecisionsandwhatarethedecisionsthatfollow.Rule-basedreasoningprovidesautomaticcombinationofrulestochaintoaconclusion.Onepopularwaytorepresentknowledgeisthe"if-then"rule.Arulecanformallyberepresentedasthelogicalrelation:

p q

prepresentsasetofconditionsorpremises,andqrepresentsasetofconsequencesorconclusions.Manydifferentalgorithmshavebeendevelopedtoimplementandsupportthebasicnotionofrule-basedreasoning.Thedifferencesbetweenvariousapproachesareinthedomainofknowledgeengineering.Forexample,forwardchainingrulesfacilitateprogrammingsynthesis,whilebackwardchainingrulesaremoresuitedforanalysisorsearch.

Rulesarewellsuitedforthetypeofreasoningthatcantypicallyberepresentedbyatreeoraflowdiagram.Rulestypicallyrepresentreasoningaboutfactsanddataratherthanthefactsordatathemselves(i.e.,metadata).Expertsystemsbasedonrulesincludeanimplementationofanalgorithmthatgovernswhattherulescando,

Page 147: Computer-Aided Materials Selection During Structural Design

whentheyareactivatedortriggered,andwhatorderofprioritytheyarecheckedandexecuted.Thesoftwarecomponentcontrollingtherulesiscommonlyreferredtoasaninferenceengine,sinceitcontrolstheinferencesofthesystem.Knowledgeaboutmaterialscanusuallybestatedinthe"if-then"form.Rule-basedknowledgerepresentationscanalsohandlelimitedformsofuncertainreasoning,suchasbyaddingorsubtractingconfidencewhileappraisingahypothesisorbyprovidingmechanismstohandlealternativelinesofreasoning.

Manycommercialtoolsareavailablethatprovideforwardorbackwardchainingorbothtypesofrules.

Page 148: Computer-Aided Materials Selection During Structural Design

Page54

Rule-basedtoolsareoftencharacterizedasexpertsystemshells.Manysuccessfulapplicationshavebeendevelopedincombinationwithothertools(e.g.,objects).Themainadvantagesofsoftwarepackagesthatrepresentknowledgeinruleformarethattheyallowtheusertoinspecttherulesinnear-naturallanguageandprovideanexplanationofwhyadecisionwasmade.Althoughitiseasyforahumanexperttounderstandaruleaboutmaterialpropertiesandtojudgewhethertheruleisacceptable(adefinitepluswhenoneneedstoknowwhatknowledgehasbeenbroughttobear),knowledgeengineerstendtocluttertheirruleswithcomputingtricksthatultimatelymakereading,managing,modifying,orupdatingtherulesbytheuserextremelydifficult.Thispracticehasledtounfaircriticismoftheunderlyingtechnology.Thus,rulesshouldbeusedasappropriateinconjunctionwithotherknowledge-representationforms.

SPATIALSYNTHESISANDLAYOUT

Conceptuallayoutisusuallyoneofthefirststepsincreatingastructure.Thesestructuresmaybeintheelectrical,mechanical,architectural,ormicrostructuraldomain.Thenatureoftheproblemofformulatingalayoutisdiscussed,andsomeoftheemergingcomputertechnologiesaredescribedthatcaneitherautomaticallyorwithuserguidancesynthesizestructuresintwoorthreedimensions.Theearlierdescribedgeometricrepresentationscanbeusedtoimplementsuchalgorithms.

Layoutdesigndealswithmanyofthecomplexissuesthattypicallyariseinthedesignofartifactsthathavetosatisfyspecifiedconstraintsandarecomposedofpartsthathaveshapeandoccupyspace.Alarge(potentiallyinfinite)numberoflocationandorientationcombinationsareavailableforplacinganysingleobject.Ineachcombination,designobjectsinteractinintricatewaysthroughtheirshapes,sizes,andthespatialortopologicalrelationsthatexistbetweenthem.These

Page 149: Computer-Aided Materials Selection During Structural Design

characteristicsalsointeractincomplexpatternswithmultipleperformancecriteriaorfunctionalattributesdemandedoftheartifactbeingdesigned.Layoutdesigndecisionsmustsimultaneouslysatisfyglobalrequirements(e.g.,usageofspace)andlocalrequirements(e.g.,adjacenciesbetweenpairsofobjectswithcertainmicrostructures,asrequiredinthedesignofslidingcomponents);acceptablespatialarrangementoftenexhibitsacomplexpatternoftradeoffs.

Forthesereasons,thereisnoknowndirectmethodthatisguaranteedtoproducefeasiblesolutionswithoutiterationsoftrialanderrorformostapplicationdomains.Someamountofexplorationofthestructure,formulationofthelayouttask,andsearchingforcandidatesolutionsisrequired.However,duetocognitivelimitations,humandesignersdonothavethecapabilityformakingsystematicexplorationsofalternativearrangements.Thisshortcominginhumanperformancehasmotivatednumerousattemptstoapplycomputationalmethodstolayout.Whatisdesiredisastructuredmethodforproducingmultiplealternatives,eachofwhichembodiestradeoffsthatcanbeunderstood,justified,andindicativeofarangeofpossiblevariationswithinwhichoptimizationcantakeplace.

Attemptstoarriveatsuchamethodconfrontthechallengesmentionedabove.Consequentlythereisalonghistoryofattemptstodevelopaneffective,"closed-solution"computational-basedmethodreflectingavarietyofrepresentations,systemarchitectures,andplanningstrategiesforlayoutdesign.Findinganeffectiverepresentationtosupporttheefficientgenerationandevaluationofdesignalternativeshasbeenadifficultundertakingandhasdominatedtheevolutionofthefield.Therepresentationmustsupportthecreationofaspaceofpossibledesignsbycapturingmeaningfuldifferencesbetweendesignalternativesatamanageablelevelofdetail(orabstraction).Layoutsforagivendesignproblemaretypicallyverylarge;therefore,therepresentationmustallowfortheemploymentof

Page 150: Computer-Aided Materials Selection During Structural Design

effectiveplanningandsearchstrategiestoenablereasonableexaminationofthebestalternatives(e.g.,throughtheevaluationofpartialsolutionsandtheincrementalspecificationofdesigns).

Thelayoutoperatingsystem(LOOS;CoyneandFlemming,1990),forexample,enablesthesystematicgenerationoflayoutalternativesandtheirevaluationovermultipleperformancecriteria.Thesystemutilizesagraph-basedrepresentationthatseparatestopologicalissues(spatialrelationsbetweenobjects)frommetricalissues(dimensionsanddimensionalpositionsofobjects)inlayout.Therepresentationusesbasicspatialrelations(i.e.,above,below,totherightof,andtotheleftof)todefinethestructureortopologyofalayoutasasetofrelationsbetweenpairsofrectangles.Itrepresentsthisstructureformallythroughanarc-coloreddirectedgraph,theverticesofwhichrepresenttherectanglesinalayoutandthearcsofwhichrepresentthespatialrelationsbetweentherectangles.FigureC-2showsanexamplein

Page 151: Computer-Aided Materials Selection During Structural Design

Page55

whichsolidarrowsindicateabove/belowrelations,dashedarrowsindicateleft/rightrelations,andErepresentstheminimumenclosingrectanglethatisabove,totherightof,totheleftof,andbelowallotherrectanglesinthelayout.Usingthisrepresentation,asetofrulesoroperationsaredefinedthatcangenerateallpossiblearrangementsofrectanglesinaplanebyinsertionofonerectangleatatime.ThelayoutsproducedbytheLOOSarelooselypackedarrangementsofrectangles(e.g.,therectanglesarenonoverlappingandneednotfillthesurroundingrectangle).Therefore,theapproachisgeneralenoughtoencompassabroadclassoflayoutsandisusefuloverawiderangeofdomains.Theserectangulararrangementsaregivenmeaningaslayoutsinaparticulardomainbyattributingthelayoutobjectsorcomponentsfromthedomaintorespectiverectangles.Inaddition,testsorperformancerequirementsforthelayoutareattachedtotheseobjectsenablingthelayoutsproducedtobecomparativelyevaluated.Thosethatfailrequirementsmaybediscarded,whilethosethatshowpromisecanbefurtherdeveloped.

Layingoutabstractobjectsdoesnotmakeadesign;itonlygivesaspatiallyfeasibleconfigurationoftheobjectsconsidered.Thenextstepinvolvesincorporatingalldetailedfeaturesofthedesign,bothgeometricaswellasnongeometricones.Tofacilitatethisstep,itisconvenienttointroduceaformallanguagewithagrammartoexpresstheintentionsofwheretogeneratewhatentityinwhatshapeandsize,andtodeterminewhatothernongeometricentityshouldbeassignedtoit.

Solidscanbedescribedthroughthesurfaceboundaryrepresentationaspreviouslyintroduced.Boundarysolidgrammarprovidesameansofgeneratingcomplexmodelsofrigidsolidobjects.Solidsarerepresentedbytheirboundaryelements(i.e.,vertices,edges,andfaceswithcoordinategeometryassociatedwiththevertices).Labelsmaybeassociatedwithanyoftheseelements.Rulesmatchconditionsofa

Page 152: Computer-Aided Materials Selection During Structural Design

solidorcollectionofsolidsandmaymodifythemorcreateadditionalsolids.Aboundarysolidgrammarusesaninitialsolidandasetofrulestoproducealanguageofsolidmodels.

Mountaingrammarisdefinedbyusingalaminaastheinitialsolidandeightrulestomodifythemountain'ssurface.Aruleofthegrammarsubdividesanexistingfaceandrandomlymovesthepositionoftheverticesoftheface.Thisproducesrandomvariationofthesurfaceofthemountain,whiletherulesrecursivelysubdivideitsfaces.

Althoughthisapplicationmayinitiallynotsoundveryuseful,itmaybequiteattractiveifthecreationofnovelmaterialstructuresandcompositionsareimaginedwithinstructuresthataredesignedtobehaveinapredefinedway.Variationsonthistechniquemaybeusedtoproduceawidevarietyoftexturesonthesurfaceoftheinteriorof

FigureC-2AnexampleoftheLOOSsystemtodefinethestructureortopologyofalayout

(Coyne,1991;reprintedcourtesyofRobertF.Coyne).

Page 153: Computer-Aided Materials Selection During Structural Design

Page56

anysolids.Toolssuchastheonesjustdescribedbecomeevenmoreimportantifyouimaginethatmanyofthestructuresthatwillbebuiltinthefuturemaybebuiltonaverysmallscale(e.g.,micromachinesandnanotechnology).

STRUCTURESELECTION

Structureselectionisatechniqueforselectingcomponentsfromafinitelistofcandidatesandensuringcompatibility.Manyproblemshavebeenformulatedwithapproachinmindandthereareavarietyoftechniquesthatareusedtoprovidestructuretotheprocessandtheinformationneededtoperformthetask.Manyoftheearlysuccessofexpertsystemsemployedstructuredselection.

TRUTHMAINTENANCE

Truthmaintenanceisaknowledge-representationtechniquethatrecordsthejustificationforinformationsothatthefactisremovedifthesupportforafactisnegatedorremoved.Truthmaintenancetechniqueshavebeenincludedinseveralcommercialknowledge-engineeringtools.Thetechniqueisparticularlyusefulinexploringmultipleoptionsbuthasnothadtheimpactthatwasexpected.

Page 154: Computer-Aided Materials Selection During Structural Design

Page57

AppendixD:Knowledge-BasedIntegratedDesignSystem1

INTRODUCTION

Engineeringdesignandmanufacturingprocessdevelopmentarecrucialcomponentsoftheproductrealizationprocess.Theyarethemeansbywhichnewproductsareconceived,developed,andbroughttomarket.Theabilitytodevelopnewproductsofhighqualityandlowcostthatmeetcustomerneedsisessentialtoincreasingprofitabilityandnationalcompetitiveness.Improvingthepracticeofengineeringdesignandmanufacturingisessentialtoachievingindustrialexcellence.

Competitivenessdemandshigh-qualityproductstosatisfycustomer'sperformanceneeds(e.g.,easeoffinaluse,expendability,aestheticappeal,andfreedomfromdefects).Higherqualityproductsrequirehigherqualitycomponentsandmanufacturingprocesses.Approximately70percentormoreofthelife-cyclecostsofaproductisdeterminedduringdesign.Fixingdefectsanderrorsduringdesigntoachieveaqualityproductisinexpensive.Itismuchmoreexpensivetofixdefectsifthecustomerfindsthemintheproductafterdelivery.

Today,productrealizationisaseriesofsequentialactivities.Duringtheproductdesignphase,thereisaminimumfocusonproducibility.Theabilitytomanufacturetheproductisnotconsidereduntilaftertheproducthasbeendesigned.Communicationbetweenproductdesignandmanufacturingislacking,andcollaborationamongsubassemblysuppliersandpartmanufacturers(vendors)israre.Asaresult,time-to-marketislongerthannecessary,andfinalproductqualitymaybepoorrelativetowhatitcouldbe.

Page 155: Computer-Aided Materials Selection During Structural Design

Thissectionreportsacasehistorythatsubstantiatestheuseofadvancedcomputer-basedtechnologiesandsupportforworkflowprocesstoincreasethecompetitivenessoforiginalequipmentmanufacturers(OEM)andsmall-to-mediumsizesupplierenterprises(SME).TheresultsreportedarebasedonanAirForceManufacturingScienceprogram.

Thedesignsystemisfocusedonreducingthedesign-to-buildtimefornewproducts.Thissystemwouldbeaknowledge-based,integrateddesignsystemforhelpingOEMsandSMEstobuildandmaintaintechnologicalleadershipintheworldmarketplace.Theoverallmissionofsuchasystemistoreducethetimefordeliveringthefirst-qualityproductionpartstothemarketplacebydramaticallyreducingthetimeforproductdesignandmanufacturingprocessdevelopment.SuchasystemwouldenablevirtualmanufacturingorganizationstobeeasilyassembledtoprovidethesupportrequiredbytheOEMforbringingasuccessfulproducttomarket.

Companiesofthefutureareenvisionedtobemorelikesolarsystems,whereOEMswillbesurroundedbyaplethoraofhighlyefficientSMEsinaflexiblenetwork.Thisflexiblenetworkwouldincludebanks,communitycolleges,technologyproviders,andSMEsuppliers.Thisnewwayofbeingcompetitivewouldstillhavetheadvantagesofaverticallyintegratedorganizationaswellastheflexibilityandloweroverheadofsuchanetwork.Theseflexiblenetworkswillbebothlocalandregionalandwillutilizenationalservernodesformaterials,productdesign,andprocessdevelopmentwithaccessbyOEMsandtheirsuppliersprovidedoverthenationalelectronicsuperhighway.

THECASESTUDY

ThecasehistoryisthedevelopmentofaBladeDesignAssistantfortheAllisonDivisionofGeneralMotors,withthecollaborationofIBMandUES,Incorporated,todemonstratethebenefitsofanintegrateddesignsystemthat(1)streamlinestheworkflow;(2)integratesthe

Page 156: Computer-Aided Materials Selection During Structural Design

applicationtoolsusedinengineeringdesignandmanufacturing;and(3)integratestheSMEsupplierindustrieswiththeircustomers,theOEMs.TheAllisonGasTurbinecasehistorywasaccomplishedbyapplyingthemethodologiesdevelopedunderanAirForcemanufacturingscienceprogramforprocessdesigntobladedesign(i.e.,productdesignatanOEM).

1CasestudysuppliedbyHaroldL.Gegel,committeemember,asatypicalexampleofthematerialsselectionsystemscurrentlyavailable.

Page 157: Computer-Aided Materials Selection During Structural Design

Page58

Theknowledge-basedintegrateddesignsystemwasdesignedtohaveaclient-serverarchitecture,wheretheserverwasintendedtobeamassivelyparallelcomputer.Usingthisarchitecture,aglobalmethodologywasdevelopedfordesigningunitfabricationprocessesstartingattheproductspecificationstageoftheproductrealizationprocess.Thedesignactivitywasstructuredasfourstages:(1)designclarification,(2)conceptualdesign,(3)embodimentdesign,and(4)detaileddesign.

Duringthedesignclarificationstage,thefunctionalrequirementsthatwillsatisfythecustomersrequirementsareestablished.Thefunctionalrequirementsarethenelectronicallypassedtothedesignteamresponsibleforconceptualdesign.Inthisdesignstage,allofthealternativesforsatisfyingtheareconcurrentlyanalyzedtogeneratearesponsesurface,whichthenisanalyzedtoobtainasetofnearoptimaldesignparameters.Thisisthenreanalyzedtomakecertainthatthisnearoptimalsetofparameterssatisfiesthefunctionalrequirements,whichwereoriginallyagreeduponbyboththeOEMandallelementsofpartmanufacturing(includingthelowertiertoolingvendors).

Intheembodimentdesignstage,theprocessdesignmaybefurtheroptimizedbyperformingparametricstudiesonthevariousdesignparameters.Therefinedprocessdesignisthenelectronicallypassedtothedetaildesignstage,wheremostoftheengineeringeffortisspent.

Thedesignactivitythatwasbrieflydescribedaboveintroducesanewdesignconceptcalledsoftoptimizationandsoftautomation.Theconceptualdesignstagecanbeautomatedtoadegreebyutilizingartificialintelligencetechniquessuchasneuralnetworkanalysisthatautomaticallyprovidealistofpossibleprocess-designalternatives,dependingonthefunctionalrequirementsdefinedduringthedesignclarificationstage.

Page 158: Computer-Aided Materials Selection During Structural Design

Thisapproachallowsthedesignertodealwithreal-worldproblems,wherethebestisonlyatheoreticalidealthatisoftenunattainableornotcost-effective.ThroughtheuseofsoftoptimizationtechniquesamodestgoalofbeingjustgoodenoughcanbeachievedevenforproblemsinmanufacturingcurrentlyconsideredtobebeyondreachbyCalculus-basedmethods.Thisapproachtoconceptualdesignreducesthetimeforarrivingatasetofdesignparametersthatwillsuffice,sincedesignproblemsareoftenincomplete(i.e.,itishighlyunlikelythatefficientalgorithmsforthesolutionoftheseproblemsofarbitrarysizewillbefound).

Thedesignapproachusedinthisresearchwasacomponentofatotalintegratedproduct/processdevelopmentstrategythatrequiresthesimultaneousandintegrateddevelopmentandqualificationofalltheelementsofatotalsystem,ascontrastedtoasequentialdevelopmentprocess.Integratedproduct/processdevelopmentrequiresatwo-wayflowofinformationbetweenthecustomer(theOEM)andthelowertierSMEsuppliers.ThisisillustratedinFigureD-1.Integratedproduct/processdevelopmentincreasesthefocusonproductsandprocesses,improveshorizontalcommunications,establishesclearlinesofresponsibility,delegatesauthority,establishesclearinterfaceswithindustry,andchangestheacquisitionprocessexpectations.

Theaimoftheknowledge-basedintegrateddesignsystemwastodevelopanadvancedprocessdesignsystem.Aglobaldesignmethodologywasdevelopedfordesigningawiderangeofunitprocesses(e.g.,casting,forging,extrusion,andsheetmetalforming),startingattheproductspecificationstageofthedesignprocess(i.e.,thestageoftheproductrealizationprocesswheretheproductdesignershavedefinedasetoffunctionalrequirementsforthepart).

Theprogramwasateameffort.TheteamconsistedoftwoOEMs,asoftwaredeveloperandsystemintegrator,severalvendors,anduniversities.Astructureddesignprocessthatsystematicallymoves

Page 159: Computer-Aided Materials Selection During Structural Design

fromqualitativetoquantitativeprocessdefinitionswasdeveloped.FigureD-2

FigureD-1Informationflowinintegratedproduct/processdevelopment.

Page 160: Computer-Aided Materials Selection During Structural Design

Page59

illustratesthemethodsdeveloper'sframeofreferencefordevelopingandnegotiatingdesigncriteria.Theproductdefinition,whichpassesfromtheproductdesignactivities,communicatesacrossstandardinterfacesandbecomestheinitialcondition(i.e.,thefunctionalrequirements)fordevelopingtheformalprocessdesigndefinition.

Theprocessdefinitionmethodconsistsofproceduresandrules(axioms)foreachdesignactivityandsubactivitytoensurethat"whattheproductdesignerwantsisthesameaswhattheproductwillpossessafterprocessing."Themethodologythatwasimplementedwasaformalaxiomaticdesignprocedureforthecreationofsynthesizedsolutionsintheformofproducts,processes,orsystemsthatsatisfyperceivedneedsthroughthemappingbetweenthefunctionalrequirementsinthefunctionaldomainandthedesignparametersinthephysicaldomain,throughtheproperselectionofdesignparametersthatsatisfythefunctionalrequirements.Thismappingprocessisnonunique,andmorethanonedesignmayresult.Therefore,theconceptofsoftoptimizationorbeinggoodenoughbasedonheuristicorrule-of-thumbmethodsofdesignareemphasizedforachievingcost-effectivedesigns.

AttheAllisonGasTurbineDivisionofGeneralMotors,aknowledge-basedintegrateddesignassistantwascreatedbasedonflowchartsoftheworkflowprocessalreadydevelopedbythecustomer.Inaddition,thecustomerhadidentifiedalloftheFORTRANapplicationprogramsandhowtheywereusedinthedesignprocess.Beforethebladedesignassistantwasdeveloped,the

Page 161: Computer-Aided Materials Selection During Structural Design

FigureD-2Methodsdeveloper'sframeofreferencefordevelopingandnegotiatingcriteria.

compressorbladeswerebeingdesignedbyagroupofengineersusingtheseveralworkstationandmainframeFORTRANapplicationprogramsillustratedinFigureD-3.

TheimplementeddesignmethodologycombinestheexistingFORTRANapplicationprogramsandtheturbinebladedesignprocessintoabladedesignassistant.Thismethodassiststheengineersinthedesignofabladeanditsassociatedattachments.Thefourmajorengineeringrolesforbladedesignareaerodynamics,stressanalysis,dynamics,andmechanicaldesign.Individualroleshavebeencreatedforeachoftheseactivities.Withineachrole,designactivitiesaresupportedthatincludedesignsteps,dataentry,coordinationbetweenotherengineersanduseofsoftwareapplicationprograms.NoneoftheseFORTRANprogramswerealtered.ThebladedesignassistantbuildstheappropriateinputparameterfileandJCLStreamtoinvokeeachoftheseapplications.

Theengineerwasnotconcernedwiththeformatofthisfile,onlythecontent(i.e.,thevalueoftheparameters).Wherepossible,designactivitieswereperformedinparallel;however,dependentactivitieswerepreventedfrombeingexecuteduntilallrequisiteinformationandapprovalswereavailable.Thisprocedurewasperformedtoensurethatengineerswerenotexpendingeffortoninappropriateactivities.

Page 162: Computer-Aided Materials Selection During Structural Design

Thisprocedureenforcementwasaccomplishedusingthecoordinationfeaturestotheassistant.Theoverallgoalofthisassistantwastoshortenthecycletimeforbladedesignandtodecreasetheeffortexpendedbythedesignengineers.

FlowchartsdevelopedbyAllisonGasTurbinewereconvertedtoaworkflowprocessmodelandimplementedinthecomputerviaaknowledge-integrationshell(i.e.,TheKIShellTM).Theshelldevelopmentenvironmentwasusedtocreatetheworkflowprocessmodel;toimplementanalyticalcodetoanalyzeapplicationoutputandapplydesignconstraints;toprepareinput;monitorstatus;toretrieveoutputofapplicationonheterogeneouscomputers;andtosuspendorinitiateworkflowprocessfordifferentspecialistsbasedonthecurrentstateofdesign.

AnoverviewofthemethodforbladedesignwiththesubprocessesforthedifferentrolesisgiveninFigureD-4.Theactivitiesassociatedwitheachroleweregroupedor"framed"intoprocessframesindifferentways.Forexample,theactivitiesassociatedwiththedynamicsengineerrolehadtobeperformedinsequence,whereasachoiceintheorderofactivityexecutionwas

Page 163: Computer-Aided Materials Selection During Structural Design

Page60

FigureD-3Controlflowbetweenrolesinthebladedesignassistant.

allowedforthedesignreviewrole.Datathatwascommontoallroleswasmaintainedininformationframes(e.g.,materialtypewasdatathatmustbeaccessedandupdatedbyallroles).Informationframesdonothaveanyimpliedcontrolsequencing.

Alsoassociatedwitheachactivityarerulesthatgovernthecircumstances(i.e.,faileddesignparameter)underwhichacopyorinstanceofarolewascreatedforanotheruser.Theremaybemanyinstancesofthesamerolecreatedduringonedesign.Inbladedesign,thiswasusedtotryoutdifferentdesignparameters,whichweremaintainedinthedatabase.Moregenerally,asdecisionsweremadeduringactivities,theyweremaintainedinthedatabase.

Finally,themechanicalengineerroleinFigureD-4alsoillustrateshowanactivitycaninturnrequireactivitiesinanotherframetobeexecuted.Thesubactivitylinkwasusedtolinkanactivitytoanotherframe.Applicationinterfaceswereimplementedtosubmitbatchjobsviathecommunicationsnetworkexistingbetweentheworkstationandthehost.

Page 164: Computer-Aided Materials Selection During Structural Design

TheKI-Shellfeaturesusedinthebladedesignassistantwere:(1)multipleroles;(2)rolesthatcreateinstancesorotherroles;(3)rolesthatwaitforotherrolestofinish;(4)abilitytoexecutedifferentrolesfrommultipleworkstations;(5)abilitytostorehistoryofiterations(i.e.,processinstancesandotherbladedesigns);(6)abilitytostoresets,matrices,etc.;(7)persistentstorageofdesignstateanddata;and(8)displayof2-dimensionalgraphs.

Thetechnologiesinvolvedinthisprojectinclude:

variousartificialintelligencemethods,geneticalgorithms,heuristicandotherrandomizedstrategiesforsoftoptimizationandautomationoftheengineeringdesignactivity;processmodelingsoftwarethatcouplesheat,fluids,andstresswithmaterialsscienceforpredictingmicrostructureandpropertyevolutionduringpartmanufacturing;high-performancecomputingtocalculateoptimizedproductdesignandmanufacturingprocessalternatives;

Page 165: Computer-Aided Materials Selection During Structural Design

Page61

FigureD-4Bladedesignassistant.

Page 166: Computer-Aided Materials Selection During Structural Design

Page62

high-performancestoragesystemsandcommunicationstomovelargedatafiles(e.g.,modelingresults)amongstoragedevices,massivelyparallelcomputers,andhigh-performanceworkstations;hypermediatechnologyenvironmentsthatallowsausertodiscover,retrieve,anddisplaydocumentsanddatabyclickingonhyperlinks-terms,icons,orimagesindocumentsthatpointtootherrelateddocuments;andmaterialpropertydatabasestosupportprocessmodelingthatusethefiniteelementmethod.

Page 167: Computer-Aided Materials Selection During Structural Design

Page63

AppendixE:AnIntelligentKnowledgeSystemforSelectionofMaterialsforCriticalAerospaceApplications1

INTRODUCTION

Akeyrequirementforthesuccessfulimplementationoftheunifiedlife-cycleengineeringconceptforaerospacestructuresdesignistheavailabilityofawell-developedintelligentknowledgesystemfortheselectionofmaterialsforspecificcomponents(BurteandHarmsworth,1989).Therearetworeasonsforthis:(1)theincreasingcomplexityoftherequirementsformaterialperformanceforanycomponentsand(2)thewiderangeofcandidatematerials,particularlythenewerandmore-sophisticated,advancematerialsclasses.Therearetwoadditionalcomplicatingaspectstothispartoftheproblem.First,thenewer,high-performancepolymers,ceramics,andcompositesaredifficulttoidentifyandcomparebecauseofthelackofstandardnomenclaturesandtestprocedures.Second,dataarebecomingavailablesorapidlyonsomanymaterialsthatthetaskofkeepingadatabasecurrentisenormous.

Acomputerizeddiagnosticprogramtoensurethatalloftheimportantpropertiesandcharacteristicsofalllogicalcandidatematerialsareconsideredandthattheyareanalyzedwithappropriateprioritieswouldbeofgreatvalueforreliablematerialselection.Astudydemonstratedthetechnicalandeconomicfeasibilityofdevelopingacomputerizedintelligentknowledgesystemformaterialsspecialistsanddesigners(IKSMAT)inthescreeningandselectionofawiderangeofmaterialsforcriticalaerospaceapplications(Kaufman,1988).FurtherithasbeendemonstratedthattheIKSMAThasthepotentialtoprovidegreatflexibilityinquery,search,andanalysisoptions,tobeveryeasyforengineersandscientiststouse,andtobeeasilyandeconomically

Page 168: Computer-Aided Materials Selection During Structural Design

expandedtoincludemanyadditionalapplications.

TheprogramdescribedbelowcoveredtheproductionofaprototypeIKSMATthatprovidedmaterial-searchcapabilitiesforawidevarietyofaircraftcomponents.

VISIONOFTHESYSTEM

ThegoaloftheprogramwastodevelopandbuildaprototypeversionofIKSMAT.Itwastoprovidevitalguidanceintheselectionofalloystomeetsophisticateddesignrequirementsforsparapplicationsandalsoformthebasisofasystemthatcouldbeexpandedtoencompassabroadrangeofmaterials(e.g.,polymers,ceramics,andcomposites),components(e.g.,enginesandempennage),andapplications(e.g.,helicoptersandmissiles).Thevariouselementsundertakenwere

knowledge-basedevelopment,dataqualification,andinterfacerefinementforthefollowingaerospacecomponents:wingspar,bulkhead,upperwingskin,lowerwingskin,fuselage,landinggear,andpivot/swivelfitting;programmingthesystemlogic,queryrules,andresponseoptions;systemdesignandassembly;andestablishmentofamasterdatabasebasedprimarilyonMIL-HDBK-5F(DOD,1986).

TECHNOLOGIESINCLUDEDINTHESYSTEM

ConceptualModelofanIntelligent-Knowledge-System

OnegeneralizedmodelofanIKSMATapplicabletothematerialselectionproblemdefinedaboveisillustratedinFigureE-1(Kaufman,1988).Inthismodel,theknowledgebaseisthecatalogofdesignandperformancecriteriaforspecificstructuresandtherelativeimportanceoftheindividualcriteriaintheperformanceofthosestructures,whichinterfaceswiththematerialpropertiesdatabasecoveringtherangeof

Page 169: Computer-Aided Materials Selection During Structural Design

materialsandpropertiesofinterest.Thisknowledgebaseisinterfacedwithprogramspermittinguserstocompile(knowledgeacquisition)andutilize(inferenceengine)knowledgeanddatatosolveproblems.Thesystemmaybeusedindependentlytoaid

1CasestudysuppliedbyJ.gilbertKaufman,committeemember,asatypicalexampleofthematerialsdatabasesystemscurrentlyavailable.

Page 170: Computer-Aided Materials Selection During Structural Design

Page64

FigureE-1ModelofanIKSMATapplicabletothematerialselectionproblem.

inthetrackingandselectionofmaterialsforspecificapplicationsorinterfaceddirectlywiththeearlystagesofthedesignprocesstoillustratetheimpactoftheutilizationofadvancedmaterialsoncomponent/vehicleperformance.

Itisimportanttonotethatwhiletheknowledgebaseitselfisthefoundationofthesystem,itisessentialtohaveawell-developeddatabaseofreliable,well-documentedinformationonwhichtheinferenceenginecanoperate.Nomatterhowsophisticatedthelogicincorporatedintothesystem,ithaslittlevalueunlessuserscanhaveahighlevelofconfidenceinthecompletenessandqualityoftheunderlyingknowledgebase(Ambler,1985;Kaufman,1986a,Reynard,1987).

IKSMATArchitectureandOperatingCapabilities

ThespecificsystemarchitecturerequiredfortheprototypeisillustratedinFigureE-2.Itiscomposedoftwogroupsofcomponents,oneintheuser'sfacilityandtheotherina''masterdatabasefacility.''Theuserinterface,controller,andsupportingdatabaseandknowledgebasesaremaintainedwiththeirowndatabasemanagementsystemat

Page 171: Computer-Aided Materials Selection During Structural Design

theuser'ssite.The"master"databasefacilitycontainstheevaluateddataprescribedbymaterialsspecialistsfromtheAirForceandaerospaceindustrythroughactivitiessuchastheMIL-HDBK-5CoordinationCommitteeandmaintainedinMIL-HDBK-5itself(DOD,1986).Italsocontainsother"external"sourcesofdata,suchasthosemadeaccessiblebytheNationalMaterialsPropertyDataNetworkandtheScientificandTechnicalInformationNetwork(Kaufman,1986b).

Theuserinterfacehandlesinteractionswiththeuserandprovidesthedisplayscreencontrol.Thecontroller,theinferenceengineintheconceptualmodeldiscussedabove,carriesouttheexpertsystemfunctions,mostofwhichwillbebroadlyapplicabletoothercomponentsandotherapplications.Therulesformaterialselectionanddesignofspecificcomponentsconstitutetheknowledgebaseinthismodel;extensionofthesystemtohandleadditionalcomponentsandapplicationsinvolvesaddingtothelogicinthisknowledgebase.Basedoninformationintheknowledgebase,thecontrollerpassesinformationtotheuserinterfacetodeterminewhatactionsarerequiredandtheninterpretstheuserinputtogeneratequeriesinthesearchquerygenerator.

Thelocaldatabasemanagementsystemdealswiththeinformationin"temporary"datafilessuppliedbytheuserinamannercompletelycompatiblewiththepermanentdatabases,andtheresponsestotheuserarecompletelytransparentinthisrespect(i.e.,theuserwillnotseetheoperationastwoseparatesystemsinteracting),yettheintegrityofthepermanentdatabasesismaintained.Twouserfilesarekept:thepersonaldatabasecontainsuser-specificinformationandwhiletheupdatedatabasecontainsdataagreedonbytheentireusergroupas"pre-standard."

Asnotedabove,thepermanentdatabasescontainingevaluateddata,suchasthosebasedonMIL-HDBK-5,andtheprincipalsearch

Page 172: Computer-Aided Materials Selection During Structural Design

softwarewouldresideataremote

FigureE-2SpecificsystemarchitecturefortheprototypeIKSMAT.

Page 173: Computer-Aided Materials Selection During Structural Design

Page65

location(theNationalMaterialsPropertyDataNetworkonScientificandTechnicalNetwork,Columbus,inthiscase),andalladditionsanddeletionstothesedatabaseswouldbecarefullycontrolledbytheappropriateagencies(liketheMIL-HDBK-5CoordinationCommittee).

Thesystemoperatesbyapplyingrulesbasedonapplicationorcomponentinthecomparisonandrankingofindividualmaterialsinprescribedsequences,graduallyeliminatingcandidatematerialsbasedontheirinabilitytomeetstatedcriteriaandthepresentationinpriorityorderofsurvivingcandidates.Thesystemmustbeflexibleanddynamicinthesensethatnewmaterialoptionsmaybeincorporatedatanytimeandtherulesmaybealteredasnecessarytoreflectchangingvehicularorstructuralperformancerequirements.

Tobemorespecific,userapproachestoIKSMATmustbeofseveralgeneraltypesofvaryingsophistication,includingbutnotnecessarilylimitedtothefollowing:

userworkswithinexistingdatabaseandpredefinedmaterialselection/designcriteriaandlogictoidentifyoptimumcandidatematerialsforspecificapplication;useraddsnewmaterialstodatabaseandthenperformsanalysesbasedonpresetmaterialselection/designcriteriaandlogictoselectoptimumcandidatematerials;useraddsnewpropertiesformaterialsintheknowledgebaseandthenperformsanalysesbasedonexistingornewcriteriainvolvingthosenewpropertiestoidentifycandidatematerials;userredefinesprioritiesassociatedwithexistingpropertiesandcriteriaformaterialselectionandperformsnewanalysestodeterminetheeffectofthechangesonthepreferredcandidatematerials;userinputsnewcriteria(specificpropertiesordesign-relatedparameters)anddefinestheirprioritiesandthencarriesoutanalysestodeterminecandidatematerials;oruserconductsgeneralunstructuredsearchwithself-generatedqueries

Page 174: Computer-Aided Materials Selection During Structural Design

userconductsgeneralunstructuredsearchwithself-generatedqueriesbasedonmaterial,property,andparametercriteria.

Functionally,thereareseveraladditionalfeaturesbeyondthosewithintheinternalIKSMATlogicbaseandknowledgebasethatareconsideredimportanttoavaluablematerialsinformationretrievalandanalysissystem.Theseare

easeofunderstandingfortheoccasionaluserwhoisnotaninformationprofessionaltrainedinthelanguageandcommandstructureoftraditionalon-linesearchsystemslikeDialogue;flexibilitywithregardtotheuseofmaterialsnomenclatureandpropertyterminology,permittingtheusertouseanytechnicallycorrectnamesorterms(aliases)andstillbeabletolocatethedesiredinformation;inaddition,theusershouldhavetheabilitytoeasilyquerythesystemaboutthemeaningsofthetermsorabbreviationsencounteredintheprocessofsearching;andeasyaccesstomanyothersourcesofmaterialspropertydata,beyondthoseuponwhichtheprogrammedmaterialsselectionprocessisbased,sothatnewlygeneratedcorroborativeorcontrastingdatamaybelocated,retrieved,andanalyzedquicklyandefficiently.

ThematerialstobeincludedintheprototypeIKSMATwillincludeallsteels,aluminumalloys,magnesiumalloys,andhigh-temperaturealloysincludedinMIL-HDBK-5.Amongthesetofcandidatecriteriaforsearchingforaerospacematerialsarethefollowing:

criticalcracksizeindexsquareofratioofplanestrainfracturetoughnesstoyieldstrength,anindexofthecriticalcracksizeattheyieldstrength;stress-corrosioncrackingsusceptibilityratioofmaximumtensilestressforresistancetotensileyieldstrength(asanalternative,theratioofstressintensitythresholdforstress-corrosioncrackingtoplanestrainfracturetoughnessmightbeused);stiffnessefficiencyratioofmodulusofelasticitytodensity;tensileoryieldstrengthefficiencyratioofultimatetensileortensileyield

Page 175: Computer-Aided Materials Selection During Structural Design

tensileoryieldstrengthefficiencyratioofultimatetensileortensileyieldstrengthtodensity;fatiguecrackinitiationresistancefatiguestrengthatonemillioncyclesoflife,withstressratio,R=0.0,withsmooth(Kt=1.0)andnotched(Kt=3.0)specimens;

Page 176: Computer-Aided Materials Selection During Structural Design

Page66

rateoffatiguecrackpropagationfatiguecrackgrowthrateatanappliedstressintensity(R=0.0)of10ksi*in**0.5(preferablybasedonspectrumloading,butnoconsistentstandardexists);andfabricability/costindexincludingfactorssuchasinitialcostperpound;specialfabricationrequirementslikefinishingorjoining;ormultiplesources(materialswithhighproductioncostsortimelineswouldhavelowindexnumbers).

STATUSOFDEVELOPMENTANDBARRIERSTOIMPLEMENTATION

KnowledgeBase/DatabaseContentDevelopment

ThecontentoftheknowledgebasewasestablishedbasedontheinformationobtainedfromsynthesisoftheguidelinesprovidedbyseveralaerospacedesignersandfinalizedindiscussionswithGeneralDynamics,FortWorth.Whileitwasdifficulttoestablishaconsensusdesignapproach,oncetheapproachwaspresentitwasnotdifficulttobuildtheassociatedrulestoparalleltheanalyticalprocessandmaptherelatedseriesofdecisioncriteria.

Compilationofthecontentdataneededforthemasterdatabaseprovedtobeamuchmoredifficulttaskthananticipatedbecauseofthepaucityofreliable,statisticallymeaningfulpropertydataavailableforanybutthesimplestofMIL-HDBK-5designdata.Forexample,evenwithinMIL-HDBK-5,notablyintheareasoffatigue,fracturetoughness,andstresscorrosion(keyelementstocriticalaircraftdesign),thereareveryfewconsistentandstatisticallybaseddata.Thisneedcouldalsonotbesatisfiedfromothersources;mostarefarlessreliablethanMIL-HDBK-5insofarasqualityandconsistencyofdataareconcerned.

IksmatDesign

TheoverallIKSMATdesignwassatisfactorilycompleted.Theknowledgebaseandcontrollerdesignswerepractical,andrule

Page 177: Computer-Aided Materials Selection During Structural Design

knowledgebaseandcontrollerdesignswerepractical,andruleimplementationwascompleted,includingastrategyforprogrammingrankinglogic.

Programming

ProgrammingoftheIKSMATwascompletedtothepointwherefull-scaleinteractionswiththemasterIKSMATdatabaseassembledcouldbetested.Menuinterfaceandpresentationformatswerealsoprogrammed,alltobeoperationalwithintheNationalMaterialsPropertyDataNetworkandtheScientificandTechnicalInformationNetwork,International,MESSENGERmainframesoftware.

ElectronicDataAcquisitionandLoading

This,likelocatingtheoriginaldata,provedtobeoneofthemostdifficultandexpensivetasks.OnlyapartiallycompleteversionofMIL-HDBK-5couldbecreatedbecauseofthecomplexityandvariabilityofdataanddataformats,evenwithinMIL-HDBK-5.Inaddition,developingmachine-readableupdatestoMIL-HDBK-5inaprotocolneededtomatchthemasterdatabasehostintheMESSENGERlanguageonScientificandTechnicalInformationNetwork,International,wasnecessaryroughlyeverysixmonthsandprovedsoexpensiveastobeprohibitive,becausethehandbookwasatthetimeproducedasahard-copyproduct,andthemachine-readableupdatesweregeneratedafterthefact.Developmentofamachine-readablemasterversionofMIL-HDBK-5willsolvethisproblem.

METHODSTOOVERCOMINGBARRIERS

ThebarriersidentifiedabovepreventedtheproductionofawhollysatisfactoryprototypeIKSMATandprecludedanyplanstocommercializeIKSMATatthattime.Theprocessdemonstratedthatwhileitmaybepossibletoconceive,design,andcreatetechnicallycapableandlogicalartificialintelligencesystemsforconcurrentengineering,thesystemmaybeofverylimitedvaluebecauseof(1)thelimitationsoftheavailabledata,bothinquantityandquality,and(2)the

Page 178: Computer-Aided Materials Selection During Structural Design

limitationsoftheavailabledata,bothinquantityandquality,and(2)thehighcostofplacingnumericdatainusefulmachine-readableformatsfortheextensivemanipulationneededinsuchsystems.

Theactionsneededtoeliminatethesebarriersinclude:

Page 179: Computer-Aided Materials Selection During Structural Design

Page67

placementofmuchmoreemphasisbygovernmentandindustryinbuildingreliable,statisticallymeaningfulmaterialpropertydatabasesthatmayserveasthefoundationofintelligentmaterialsselectionanddesignsoftware;maintenanceofmasterversionsofmaterialsdatabasesinmachine-readableform,readilyupdatableandreadilyduplicatedanddistributedforbroaduse;andutilizationofflexiblesoftwaresystemscapableofrapidmanipulationandvariedpresentation(e.g.,graphicalanalysisandpresentationofcomplexnumericdatacomplementedwithengineering-oriented,intuitivemenu-driveninterfaces).