SysML2.0RFP-Analysis

SysML2.0WG 1

SysML2.0RFPWG

Lastupdated:Feb8,2017Curator:ManasBajaj(manas.bajaj@intercax.com)

Team:ManasBajaj(Intercax),AhsanQamar(Ford),GeorgeWalley(Ford),BjornCole(JPL),BillBailey(Ford)

1 REQUIREMENTS

1.1 ServiceRequirementsAnalysis isthesystematicinvestigationofarealorplannedsystemtocompare,evaluate,andselect candidate systemarchitectures, and/ordetermine causes& resolutionsof failuresandexceptions.[SEBoK,NASASEHandbook2007].ThescopeofanalysisforSysML2.0includes:(a)quantitative analyses, such as evaluating KPPs and MoEs, (b) qualitative analyses, such asassessing the impact of changes in the system, and (c) system V&V, such as checking if allrequirementsareallocatedtofunctional/structuralelementsofasystem.SeeGlossary(section2)formoredetails.SysML 2.0 based System Modeling Environment (SME) shall provide the following analysisservices:1. Setupanalysis–Thisincludesidentifying:(a)objectives,type,andfidelityoftheanalysisto

beperformed,(b)keymetrics(MoEs,KPPs,PoIs)thatarebeingcomputedintheanalysis,(c) the representation of the system (model, prototype, or deployed) which is beinganalyzed, (d) the cases/scenarios for which the analysis will be performed, and finallysettinguptheanalysismodel/methodandtheenvironment(tools–software/hardware)toperformtheanalysis.

2. Executeanalysis–Thisincludescarryingouttheanalysis,suchasexecutinganalysismodelsor running experiments/tests on the system (real or prototype), for the variouscases/scenariosidentifiedduringsetup.

3. Review analysis results – This includes reviewing the information generated during the

analysisusingmultipleviews(tables,charts,cause-effectdiagrams,etc.).

SysML2.0AnalysisWG

SysML2.0RFP-Analysis

SysML2.0WG 2

4. Store analysis information – This includes versioning and persistence of analysisinformation in a system/analysis repository, such as system representation analyzed,analysismodels,analysisscenarios,solvers/tools,andanalysisresultsandviews.

5. Queryanalyses–This includessearchingandretrievinganalysesfromthesystem/analysis

repositoryforreview.

6. Trackdecisionsbasedonanalyses–Thisincludeslinkinganalysesasrationalefordecisionstaken during system lifecycle processes. Note that thiswould only apply to decisions forwhichoneormoreanalysiswasperformedtoaidinthedecision-makingprocesses.

1.2 ConceptRequirementsSysML2.0 languagemustsupportthefollowingcoreconceptsrelatedtotheanalysisdomain,which shall be directly used by the services defined above. See theAnalysis ConceptModel(section3.3)forabackgroundmeta-modelandexplanationoftheseconcepts.1) Analysis2) Analysisobjectives3) Analysisassumptions4) Analysiscase/scenario5) Analysismodel6) Analysiscomposition7) Analysisrun/execution8) Analysisresult9) Rationale10) Decision11) Metadatasuchasaccuracyoftheanalysis,whoperformedtheanalysis,etc.Severalof theseconceptsaresimilarasconceptsproposedbyotherworkinggroups, suchasmodel management and verification. The overall SECM meta-model will be normalized toincludetheseconcepts.**Toresolve:Theconcept“AnalysisContext”hasoftenbeenusedduringourdiscussionsandSysML1.xmeetings.Itisnotpreciseenoughsincetheword“context”isoverloaded.Weneedtoresolvewhatthisconceptmeansand if it isnotalreadycoveredbytheconcepts listedabove.Forexample,theconcept“Analysis”listedaboveissimilarto“AnalysisContext”anditreferstotheSystemrepresentation(model,prototype,realsystem)beinganalyzed.

SysML2.0RFP-Analysis

SysML2.0WG 3

In addition to the high-level concepts, SysML 2.0 should also support representation ofconceptsinthefollowingareas,asneededforsupportingrigorousanalysisofcomplexsystems.12) UniversalIDsystem(UUID)13) Versioningandconfigurationcontrolofanalysisinformation14) Datastructures

a) Arrays,Lists(ordered/unordered),Sets,…b) Matricesc) Map(key-valuepairs)d) TensorsandVectorse) MutableandImmutableobjects(constants)f) DateandTimeg) Temporal model for representing time-based representation of properties, behavior,

andstructure.h) Geographicmapi) ProbabilityDistributions

15) Unitsandquantities(extensiblelibrary)a) LibraryofunitssupportingbothSIandFPSsystemsb) Frameworkfordefiningnewunitsandsupportingautomatedunitconversions

16) OperatorsandFunctions(extensiblelibrary)a) Logarithmic,Trigonometric,Calculus,andothers.b) We should have a tool-independent representation of math expressions that can be

mappedtovariousexecutablelanguages(e.g.Modelica,Mathematica,andMATLAB)17) Geometry

a) Coordinatesystemsb) Primitivegeometricshapesc) ReferencingcomplexgeometricshapesdefinedinISOSTEPAP242ornativeCADmodels

18) Model transformations – concepts to specifymodel transformations to generate analysismodel in various formalisms from system representation. Formalisms include equation-basedmodels,state-basedmodels,flow-basedmodels.

2 GLOSSARYAnalysis isthesystematicinvestigationofarealorplannedsystemtocompare,evaluate,andselect candidate systemarchitectures, and/ordetermine causes& resolutionsof failuresandexceptions. [SEBoK,NASASEHandbook2007].Ananalysis activitymay includeevaluationbymeansofmodeling/simulation,inspection,demonstration,test,oracombinationofthese.KeyParameterofInterest(KPI))KeyPerformanceParameter(KPP)

SysML2.0RFP-Analysis

SysML2.0WG 4

Measure-of-effectiveness(MoE)Quantitative analysis - Computing KPPs and MoEs to evaluate system alternatives, orquantitativeimpactanalysestoassessifandbyhowmuchasystemarchitecture(structureorfunction)mayneedtochangeifarequirementchanges.

Qualitativeanalysis,suchasdoneusinggraphtraversalstoassesshowchanges inpartofthesystem“may”affectotherspartsofthesystem.SystemV&V, suchaschecking ifall requirementsareallocated to thesystemdefinition,or ifthereisatestcaseassociatedwitheveryrequirement,ormatchingpatternsandanti-patternsthatmayberelevantforaspecifictypeofsystemordiscipline.

3 BACKGROUND

3.1 AutomotiveExample–AnalysisScenariosOurgoalistodeveloptheworkflowsforthe3challengeproblemsrelatedtoHSUV.WewillthenderivetheservicesrequiredintheSystemModelingEnvironment(SME)tosupportthatworkflow.Fromthere,we will derive the analysis conceptmodels and capabilities required in the SME. To summarize, ourprogressionwillbeasfollows:AnalysisWorkflows(HSUVChallengeProblem)→SMEServices→Conceptmodelsandcapabilities

3.1.1 Scenario # 1: Govt. regulation to improve fuelefficiency

#Aerospace-Forspacecraft,thiscouldbeaneedtogathermoredatapertimeperiod,whichpushesonefficienciesofdownlink,poweruse,ortaskingofinstruments.(1) Locatingandgatheringthebaselinemodelpackage—TotalSystemModel

a. Checkout the latest version (baseline) of the Total SystemModel package. This includes thearchitecture model (SysML), connected design models (CAD/PLM/ALM), connected analysismodels (simulation/FEA/CFD), and connected requirement models of the vehicle (system ofinterest)andtheenvironment.TheTotalSystemModelisaconsistentsetofconnectedmodelsthatrepresentthevehicle.

b. Getthelatestversionsofanalysismodelsusedforcomputingfuelefficiency,theresultsofthelastanalyses,andthedecisionstakenthereof.

c. Wherearethelateston-fieldtestresultsforfuelefficiency?d. Whatisthediscrepancybetweentheanalysismodelresultsandon-fieldresults?e. Havewe identifiedfactorsthatcontributetothisdiscrepancy?Aretheyformalizedsothatwe

canaccountforthoseinthenextsetofanalyses?

SysML2.0RFP-Analysis

SysML2.0WG 5

(2) Reviewingthearchitecture,createnewdesignalternatives,andanalyzea. Understanding factors that influence fuel efficiency at the system level, such as mass, size,

engine,powertrainefficiency,drivingenvironment,driverbehavior,andotherfactors.b. Identify the parametric changes versus the topological changes that can bemade to improve

fuelefficiency.i. Parametricchanges(canaffecttopologybeyondacertainrange)ii. Topologicalchangesiii. Combination

c. Parametricchangesandrelatedanalyses

i. Examples - Engine maps, Torque/speed correlation to fuel economy, Gear ratios, andControlscalibration.Thesecouldbeavailableas:(1)multivariatefunctionsthatneedtobeevaluated simultaneously, and/or (2) tables from handbook, such asfuelrate=f(torqueload,requestedspeed).

i. Needtoperformatradestudyoftorque,speedandfueleconomyii. Multiplecontrolalgorithmshavetobeinsyncforthefuelefficiencytobeoptimum.Lossof

synchronizationbetweenashiftcontrolalgorithmandtheenginecontrolalgorithmwouldleadtoafuelefficiencyloss.Ithinkthisistrueforalotofbigsystem-leveloptimizations…lotsofcomponentsandsubsystemsneedtobecoordinated.

#Issue-Analysismodelsfromthesuppliersareusuallyblackboxmodels,notparameterizedoryoudonothavecontrolovertheknobsthatcontrolthem.

b. Architectural/Topologychanges-Identifyifthecurrentarchitectureiscapableofmeetingfuel

efficiency goals, if not, perform an architectural exploration study to find an architecturalsolutionthatwouldmeetthefuelefficiencygoal.i. Exampleofarchitecturalchanges--TurbochargedengineVsnaturallyaspiratedVsHybrid;

ControlsArchitecture(insteadofcontrollingspeedatthewheel,youcontroltorqueatthedriveshaft)

ii. What architectural changes will require rebuilding models (beyond the scope of existinganalysismodels)versusreusing?

#Issue–Howdoyouknowthatthechanges(parametricortopological)willnotbreakexistinganalysismodels?Howdoyoucapturetheanalysisintent,whenmodelisbuilt,sothatwecanverify if the assumptions in the model still hold true for the new scenarios? Capture theanalysis intentdescribing theassumptionsandgoalsof theanalysis. Forexample,aparticularanalysisusedafixedstepsolverwith2msecintervalandifthisisnotcommunicatedtodomainexperts,theymaybuildanalysismodelsthatrunwithadifferentconfiguration,leadingtoissueswhileintegrating.Similarly,ifadomainexpertmakesanenginemodelanddoesnotspecifythatit isnottobeusedforcoldcyclesimulation,could leadtoerrors. InSimulink,assumptionsbythemodelerareeithernotmadeexplicit,orare specified sometimes twelve levelsdown.Weneedaproperty-based/math-baseddescriptionoftheassumptionandnottext-based,e.g.allanalysismodelsinthechainofanalysestocomputefuelefficiencyarevalidfrom-10Cto45Coperatingtemperaturerange.

c. Analysis planning – Planning the analyses to compute fuel efficiency for parametric and

topologicalchanges.i. Whatistheorderinwhichwewillexploretheparametricandtopologicalchanges?

SysML2.0RFP-Analysis

SysML2.0WG 6

ii. What is the level at which changes will be affected – single component, sub-assembly,wholesystem,hardware-only,software-only,etc.

iii. Whatistheorderinwhichwewillruntheanalysesforeachchange?Forexample,runninganalysisatdifferentscales–isolatedcomponentversussub-assemblyversusvehicle,

iv. Whatadditionalanalyseswouldweruntocheckthatwedon’tfailotherrequirementswhilewetrytoanalyzetheimpactofchangestofueleconomy.

v. Whenareyoudone?It would be helpful we had an “Architectural Distance” model to give a sense of degree ofchanges, such as how extensive are they in terms of number of element affected, disciplinesaffected,andtherangeofanalyses thatwehavetobere-run.This issimilar towhat iscalledplatformorprogramortechnologyscalebasedondegreeofchangetounderbody,upper-body,etc.elementsofthevehicle.Maybeifcomputedthiscouldhelpconfirmoralertwhenchangeshavebeenmadebeyondtheintendedscaleofchangeallowed,etc.

d. Executingtheanalysismodelsbasedonthetheanalysisplan

i. Parametricanalyses-Performasensitivitystudytofigureoutwhichparameterseffectthefuel efficiency, based on a given architecture description, e.g. given a hybrid-electricalvehiclesarchitecture,findtheparametersthataffectfuelefficiency.

ii. Tradestudies–Performparametricandarchitecturaltradestudiespertheanalysisplantocomputefueleconomyforvariousscenarios.

e. Visualizationandreviewofanalysisresults

i. Stack analysis results against each other to view the “analyzed” impact of parametric andtopologicalchangestothefueleconomy.

ii. Comparethegainsinfueleconomytotheestimatedcostandcompletiontimeofthevehicleprogram.

f. Whendoyoustoptheanalysis?

i. Regulatoryrequirementsvsvehicleprogramattributes,dictatewhentheanalysisisstopped.A regulatory requirement must be met, vs a market demand, fo which we can be in thenearby/ballpark figure. In short, the analysis is stopped when it is no longer valuable tofurthercontinuebasedonthecost/benefittradeoff.

(2) Performfieldtests

a. Orderandinstallnewpartsinthetestvehicleb. Performfieldtestsundersameoperatingconditionsasanalysisc. Compareon-fieldresultsversusas-analyzedresults,andrecheckthefactord. Ifon-fieldresultsagreewithas-analyzedresults,planabroadertest,andeventualrolloute. Setup formal test cases, e.g. a specific parameter is within a specific percentage, validated

throughaparticulartestcase.f. Comparingtestdatawiththeparametersintheanalysis,e.g.transientsignals.g. Queryatestrepositoryforatestcasedata,pulldowndata,butsometimescertainsignalsthat

arerequiredtocomparetheanalysisarenot inthedata.Needtoestimatethosesignals fromtheonesthataremeasurable.

SysML2.0RFP-Analysis

SysML2.0WG 7

3.1.2 Scenario # 2 -- I&T provides data stating thatvehicle fails to meet its fuel efficiencyrequirement

(1) Locatingthebaselineresultsandmodels

a. Get I&T results, and identify as-built test vehicle, fuel efficiencymeasurement approach, andtestenvironment(drivingconditions,driverbehavior,....)

b. Gather as-designed information (BOM / architecture), analysis models and as-analyed fuelefficiencyresults

(2) Reviewtheresultsandplanningoutnextstepsa. Dothefuelefficiencymeasurementapproacheson-fieldandas-analyzedmatch?b. What is the differentmargin betweenon-field and as-analyzed results? Is thismargin normal

basedonpastvehicledesigns?Ifyes,didthedesignteamnotaccountforthis?

(3) Updatingthedesigna. Followstep2inScenario1

(4) Performnewon-fieldtests

a. Followstep3inScenario1

SysML2.0RFP-Analysis

SysML2.0WG 8

3.2 AnalysisServicesBreakdownSysML2.0basedSMEshallprovidethefollowingservices.Analysisimplies:

1) Quantitativeanalysis,suchascomputingKPPsandMoEstoevaluatesystemalternatives,or

quantitativeimpactanalysestoassessifandbyhowmuchasystemarchitecture(structureorfunction)mayneedtochangeifarequirementchanges.

2) Qualitativeanalysis, suchasdoneusinggraphtraversals toassesshowchanges inpartofthesystem“may”affectotherspartsofthesystem.

3) ModelV&V,suchascheckingifallrequirementsareallocatedtothesystemdefinition,orif

there is a test case associated with every requirement, or matching patterns and anti-patternsthatmayberelevantforaspecifictypeofsystemordiscipline.

An analysis activity may include evaluation by means of modeling/simulation, inspection,demonstration,test,oracombinationofthese.Analysisservicebundleincludesservicesrelatedtoanalysissetup,analysisexecution,analysisdataandmodelmanagement,andanalysisdecisionmanagement.Theseareelaboratedbelow.Thehigh-levelanalysisservicedefinition (Level1):Setup,validate,andexecutemodels (e.g.,systemmodels,analysismodels,validationrules)includesthefollowingspecificservices(Level2),asanexample:

• Setupanalysis• Executeanalysis• Saveanalysisresults• Takedecisions• Queryandcompareanalysisresults

EachofthespecificLevel2servicesareelaboratedbelow,asanexample.1. SetupAnalysis

a. Modelthetypesofanalysesthatneedtobeperformedonthesystemrepresentation

b. Modeltheanalysisobjectivesmathematically#Properties_Expressions – Objectives defined as expressions/constraints usingproperties

c. Definethekeyparameters(KPPs/MoEs)beingcomputedorpatterns/anti-patternstobematched.#Properties_Expressions–KPPsandMOEsarespecialtypesofproperties

SysML2.0RFP-Analysis

SysML2.0WG 9

#Model_Management – Patterns and anti-patterns are also models that are versionmanagedandconfigurationcontrolledinSMEandrelatedrepositories

d. Definemapping/transformationfromsystemmodeltoanalysismodel#Model_Construction–Definingamodeltransformation#Model_Management–Storingthemapping/transformationforlateruse#Model_Interoperability–Iftheanalysismodelwillbedefinedinaspecializedanalysistool(ormathengine),e.g.X,thendefiningthetransformationwillneedmeta-modelsofSysMLandX.

e. Execute the model transformation - Create or generate analysis model based onmapping/transformation(tool-neutralortool-dependent)#Model_Construction–transformationsandexecutethemtogenerate/updatemodels.

This includes both transformations within or between SysML models, and between SysMLandnon-SysMLmodels.

#Model_Interoperability–IftheanalysismodelwillbedefinedoutsideSysML,thenwehavetoverifythattheresultinganalysismodelisvalid,e.g.didwegeneratesyntacticallycorrectsetofmathequationsinMathematicaorModelicaorMATLAB?

Notes–Patterns,Transformationsare typesofmodels, similar to systemmodel,analysismodels,inthegeneralsense.So,theyareconstructed,managed,visualized.

2. Executetheanalysisa. Define the analysis execution/process – What are the inputs, what is the

context/scenario?#Model_Construction-Creatingmodelstorepresentoperationalscenariosforanalyses#Model_Management – Retrieve existing scenarios from a model library, e.g. rainyconditions,icyconditions,mountainterrainforautomotive.#Model_Interoperability – Scenarios and boundary conditions defined in the systemmodel (SysML)mayneedtobecommunicatedtoanalysistools/engines,e.g.set inputvariablesandboundaryconditionsforasystemofequationsinMathematica.

b. Executetheanalysismodelinasolvertool.Analysisexecutionproduceslotofdata,e.g.fuel economy values for different automotive designs evaluated under differentoperationalscenariosinatradestudy.#Model_Visualization–Visualizeanalysisresultsdata–tables,charts,graphs,etc.#Model_Interoperability–BriningviewsofanalysisresultstotheSME

3. Commitmodelsandresultstotheanalysisrepository

a. Committheanalysismodel,runconditions,analysisresults,andvisualizationstogetherasasettotheanalysisrepository

b. Relatetheanalysismodelruntothesystemrepresentation

SysML2.0RFP-Analysis

SysML2.0WG 10

#Model_Management – Storing analysis and related models for future queries, linked/connectedtothesystemrepresentation.

4. DecisionsandChangeRequest

a. Modeldecisions/changerequestsbasedontheanalysisresultb. Connectthedecision/changerequesttotheanalysismodel/resultset

#Model_Construction – Modeling system engineering decisions and relating them toanalysisresultsastherationaleforthosedecisions.

5. Queryanalysisresults

a. Formulateaquerytoretrievetheanalysesi. Basedonagivensystemii. Basedonaspecificdecisioniii. Basedonaspecificanalysismodeliv. …andmore

b. Fetchtheanalysismodelsetfromtherepositoryc. Visualizeandbrowsethroughtheanalysismodelset#Model_Construction–Formulateandexecutequeries#Model_Visualization – Visualization of queried results, comparison ofmultiple analysesperformedonasetofsystemrepresentations

3.3 AnalysisConceptModelSysML2.0shallprovideanalysisconceptsdefinedintheAnalysisConceptModel.The Analysis Concept Model provides a foundational meta-model for system analyses. Itincludesconcepts relevant for formulatingandsolvingsystemanalysismodels,andthe inter-relationships between these concepts.We present the core aspects of the Analysis ConceptModelhere.In this section, we present the Analysis Concept Model as a foundational meta-model forcomputer-based representationof systemanalyses. Figure1, Figure2, andFigure3 illustratetheSysMLmodelfortheAnalysisConceptModel,whichisalsoincludedwiththisreport.TheconceptmodelwasdevelopedbyagrantprojectbetweenNIST(ConradBock)andIntercaxin2014-2016timeframe.SystemAnalysisisthefundamentalconceptintheAnalysisConceptModel,asshowninFigure1 below. It represents the information required to formulate and solve system analysisproblems. The various types of system analyses, as discussed in section are modeled assubtypesoftheSystemAnalysisconcept.TheclassificationoftheSystemAnalysisconceptintothesubtypes isbasedon (1)aspect/measureof thesystembeinganalyzed,e.g.performance

SysML2.0RFP-Analysis

SysML2.0WG 11

and other MoEs for the Effectiveness Analysis or cost measures for Cost Analysis; and (2)computational process for the analysis, e.g. trade-off analysis requires computation andcomparison of multiple MoEs versus cost, and optimization searches for a system designalternativethatminimizes/maximizesobjectivefunctionsdefinedusingMoEsandcost.

Figure1:Typesofsystemanalyses(SysMLBDDwithtaxonomyofanalysistypes)

TheSystemAnalysis concept ismodeled indetail, as shown in Figure2below.The followingconceptsaredefined.AnalysisObjective conceptrepresents theobjectiveof theanalysis.Theobjective ismodeledusingasetofmathexpressionsthatcanbeformallyevaluatedandadescription.Theobjectiveof an analysis is met if the expressions can be successfully evaluated by the informationgenerated during the analysis. For example, if the analysis is being performed to verify aperformanceormassrequirement, thenthemathrepresentationof that requirementwillbeusedfortheexpressionsthatthatneedtobeevaluated,suchasacceleration>10m/s2ormass<1000kg.

SysML2.0RFP-Analysis

SysML2.0WG 12

Figure2:SystemAnalysisconcept(SysMLBDD)

SystemRepresentationconceptrepresentsthesubjectoftheanalysisbeingperformed.Sincethe scope of system analysis spans across the lifecycle, the subject of the analysis could beeitherofthefollowing,asshowninFigure3.

(1) design representation of the system, such as a digital mockup of a spacecraft beingdeveloped

(2) prototypeofthesystem,suchasascaledorrealprototypeofthespacecraft(3) deployedsystem,suchastheactualspacecraftdeployedinorbit

SysML2.0RFP-Analysis

SysML2.0WG 13

Figure3:Systemrepresentation(SysMLBDD)

Property concept represents attributes of a system, in any of its representations. Everyproperty has a name, set of values, and a unit if the property represents a quantity such asmass.Apropertymaybediscreteorcontinuous—takevaluesfromadiscreteorcontinuoussetofvalues.Apropertymayhavedeterministicorstochasticvalues(expressedwithprobabilities).Asystemrepresentationmayhaveoneormoreproperties,asshown inFigure3.Foragivenanalysis,someorallofthesepropertiesmaybeofinterest,asshowninFigure2.Analysis Model concept represents the computation model used to calculate the systemproperties (relevant to theanalysis) tomeet theanalysisobjectives.Analysismodel couldbecomputer-basedexecutablemodel (e.g.Mathematica/MATLABcodeorFEA/CFDmodel),oramodelrepresentingphysicalmeasurementonaprototypeoractualsystem.Foreveryanalysismodel,thefollowingcharacteristicsaremodeled:

(1) Languageinwhichthemodelisformulated(2) Softwareusedtoformulatethemodel(3) Typeofmodel(4) Resultdatafromexecutingthemodel(5) Relationship to the system representation, e.g. design model. This relationship

embodiesthemodeltransformationsrequiredtogenerateorupdatetheanalysismodelfromthesystemrepresentation

AnalysisRelationconceptrepresentstherelationshipsbetweenanalyses.Agivenanalysismayberelatedtomultipleanalyses,suchasinthefollowingscenarios:

(1) Ananalystmayperformthesametypeofanalysis(sameobjective)withvaryingdegreesof fidelity,suchasone-, two-,or three-dimensionalanalyses.Foreachanalysis,anewanalysis and corresponding analysismodel would be created, and all analyses of thesametypecanalsobegroupedtogetherusingtheanalysisrelationconcept.

SysML2.0RFP-Analysis

SysML2.0WG 14

(2) Basedontheresultsofananalysis(sayA1),newanalyses(sayA2,A3,andA4)maybe

formulatedforthesystem.Inthisscenario,theanalysisrelationshipisusedtorelateA2,A3,andA4asbeingderivedfromA1.

(3) Asingleanalysismaybedecomposed intomultipleanalyses.Theanalysis relationshipconceptcanalsobeusedtorepresentthisdecomposition.

AnalysisResultconceptrepresentstheresultoftheanalysis intermsoftheevaluationsofalltheexpressionsintheAnalysisObjective.Ananalysisissuccessfulifitsobjectivehasbeenmet.Since the objective is represented as a set of expressions, an analysis is successful if all theexpressionscanbeevaluated.Theanalysisshouldgenerateenoughinformationtobeabletoevaluate theexpressions in theanalysisobjective. Forexample, if theexpressionsassociatedwiththeanalysisobjectivewere(1)mass<1000kg,and(2)powergenerated>1200hp,thentheanalysisresultwouldcontaintheevaluationsoftheseexpressions:(1)true,and(2)false.Decisionconceptrepresentsthedecisionsthataretakenrelatedtotheanalysis.Thisincludesdownstreamdecisions—basedontheresultoftheanalysis,andtheupstreamdecisionsthatledtoperforming theanalysis.Modeling thedecision isnot in the scopeof thisproject.WewillleveragetheOMGDecisionModelingNotation[DMN1.0,2015]forthispurpose.Itisimportantto capture the decisions taken during a system engineering process such that they can betraceddownstream,especiallyfordesignreviewsandproblemresolutions.Itisalsoimportanttocapturetherationaleandthefollow-upactionsforeachdecision.Thisneedisaddressedbythe relationships between the System Analysis concept and theDecision concept—resultedfromforupstreamdecisions,andresultinginfordownstreamdecisions.

4 REFERENCES• NASASEHandbook(2007)-http://goo.gl/iVBVES• SEBoK-http://goo.gl/RCtAKt