1、通過設(shè)計(jì)空間探索驅(qū)動(dòng)產(chǎn)品創(chuàng)新Driving Product Innovation Through Design Space ExplorationDriving Innovation Through Design Space Exploration“Any company that stays stationary with their technology will be exceeded by their competitors”Elon Musk Tesla-SpaceXThe Challenges of InnovationENVIRONMENTBRAND IMAGESPEED TO

2、MARKETCOST REDUCTIONPackagingCustomer NeedsConsumer InsightProcess EfficiencyEnergy EfficiencyPackagingAdapt to Market DemandsManufacturingSustainabilityConsistent QualityCompetitionLight weightingMaterials ManagementCustomer LoyaltyDevelopment TimeDecrease Hardware TestsEDAG Engineering GmbHFull Ve

3、hicle Weight ReductionChallenge:Minimize Body-In-White weightMaintain manufacturabilityObtain the same or better performanceSystem-level requirementsFull vehicle requirementsConstrain increases in costPrevious complex optimizations(with 484 variables) were not possibleView Study ObjectiveResults:Aft

4、er 250 evaluations, found 49 feasible designsIdentified the best design with an 18% reduction in weight and less than 5% increase in costTARGET WEIGHT(20% saving)TARGET COST (10% up)BODY NVHFRONT CRASH(all load cases)OCCUPANT RESTRAINTS(all load cases)SIDE CRASH(all load cases)REAR CRASH(all load ca

5、ses)ROOF CRUSH(all load cases)DURABILITYBODY STIFFNESS(global and local) MDO PERFORMANCE ASSESSMENTBASELIN E“HEEDS allowed us to meet DFSS requirements in record time.” Javier Rodriguez, EDAGDiscover Better Designs, FasterWhat taper, twist, height, cant?AerospaceWhat hole location and size?Automotiv

6、eWhat orientation for platform?Oil & GasElectronicsLife SciencesMarineWhere to place fan, heat sinks?What shape and draw rate?What size and shape of hull?How easy is it for you to answer these questions today?Would it help if this process was simpler and faster?Precision Castparts CorporationCasting

7、 Process DesignChallenge:Minimize wetted surface area during pourLoading conditions: sprue cup fill processDesign variables: 4 separate sprue cup radiiResults:75 evaluations with 4 in parallel on 24 cores2.5 days of computation (coarse mesh)Improved design saves 35% scrap metalCFD SimulationEulerian

8、 MultiphaseUnsteady VOFSegregated SolverLaminar FlowDesign geometries exploredMaximum Wetted Surface AreaInitial DesignBest DesignEvaluationsSprue CupPour CupInitial DesignMax Wetted Surface = 878.7 cm2Best DesignMax Wetted Surface = 572.4 cm2VALUEEXPLORE DIGITALLY, CONFIRM PHYSICALLYPROACTIVEREACTI

9、VEValidateTroubleshootPredictAutomateExploreRedefining Simulation Strategy for InnovationSIMULATION PURPOSEMost companies operate here to reduce time and costInnovative companies operate here to design better productsVirtual Prototype(CAD / CAE / Costing Simulations)ValueResourcesResponsesDesign Var

10、iablesVirtual TestsDiscover Better Designs, Faster with HEEDSBuildTestImproveAssessTime50%40%30%20%10%0%Process AutomationDistributed ExecutionEfficient SearchInsight& DiscoveryDesign ModificationsTraditional Process For Design ExplorationIntroduces ErrorsRequires IterationRequires Expertise“Too Dif

11、ficult”Problems Encountered“Too Costly”“Too Error-Prone”Reactions:Traditional ProcessDefine ObjectivesSimplify ModelSelect AlgorithmEither:Screen variables and/orFit Response Surface Model (DOE/RSM)Gradient searchGenetic algorithmParticle swarmAnt colonySimulated annealEtc.Tune AlgorithmConduct Sear

12、chInterpret ResultsPopulation size# of generationsCrossover rateMutation rateSelection typeEtc.PredictedActual1.510.50 Baseline model Variables Responses Objectives ConstraintsDesign ExplorationUsed sparinglyOnly by expertsOnly for toughest “optimization” problemsDelivers limited innovationResult:Ne

13、w Paradigm for Design ExplorationPredictedActual1.510.50Traditional ProcessDefine ObjectivesSimplify ModelSelect AlgorithmEither:Limit Variables and/orFit Response Surface Model (DOE/RSM)Gradient searchGenetic algorithmParticle swarmAnt colonySimulated annealEtc.Tune AlgorithmConduct SearchInterpret

14、 ResultsPopulation size# of generationsCrossover rateMutation rateSelection typeEtc. Baseline model Variables Responses Objectives Constraints“HEEDS surpasses anything on the marketin its ability to help us drive innovation.”- Anders Ahlstrm , Scania TruckDefine ObjectivesAutomated SearchInterpret R

15、esultsModern Process (HEEDS) Baseline model Variables Responses Objectives ConstraintsOptionally: Study robustness & sensitivity of optimal design(s) Select number ofevaluations Run optimizationTime SavingsDiscover Better Designs, FasterBaseline DesignBaseline DesignCoolant InletGasket HolesCustomer

16、 Success: Cooling Jacket Gasket Hole DesignChallenge:Minimize pressure drop across water jacket24 gasket hole geometric design variablesConstraintsPeak head temperature 250 CCylinder to cylinder variation in peak liner and dome temperatures 10 CPeak liner temperature 200 CPeak coolant temperature 15

17、0 CPeak velocity of coolant in head and block water jackets 9% Weight reduction $500,000 in annual launch cost reductionNXAbaqus/CAEAbaqusAEROSPACECustomer Success: Undersea Pipe SealabilityChallenge:Improve sealability of an undersea threaded pipe designConstraintsSeal contact length 0.02 in.Contac

18、t pressure 120 ksiDesign variables0.005 in. Seal interference 0.045 in.1.5 in. Seal radius 10% better than anyother algorithm and 30% better than the nearest hybrid algorithmGraph showing function for two variables (n=2)Average values for 25 optimizations from random baselinesx2x1Design space points

19、 evaluated by HEEDS (SHERPA)Competitive Hybrid Function : f x = -x sini xi i=1n- 500 xi 500 Minimum : f = - 418.9829nResults for n 20What Our Customers Say:“HEEDS has been an indispensable tool for Trek to continue redefining best-in-class products. Its algorithm has enabled us to explore themultimodal design space at unparalleled efficiency. Whether its optimizing the stiffness, reducing aerodynamic drag, and/or fine tuning the weight, HEEDS can help Trek engineering to convert fresh ideas into the next generation products that maximize performance.” M