towards virtual aircraft design and testing based on...
Post on 10-Mar-2018
219 Views
Preview:
TRANSCRIPT
www.DLR.de • DLR-AS 1
Towards Virtual Aircraft Design and Testingbased on High-Fidelity Methods- Recent Developments at DLR -
O. Brodersen, C.-C. Rossow, N. KrollDLR Institute of Aerodynamics and Flow Technology
Digital-X
Prof. A. Jameson 80th
Symposium Mathematics, Computing & Design– Where Analysis and Creativity Combine20-21 Nov. 2014, Stanford, USA
Society: European Aviation Vision “Flightpath 2050” & Aeronautics Strategy Germany
Environment: 75% CO2, 90% NOx, 65% noiseIndustrial Competiveness: innovation process frombasic research to full-scale demonstrators
Towards the Virtual AircraftMotivation & Strategic Objectives
www.DLR.de • DLR-AS 2
Industry:Robust & efficient design process with all disciplinesMore knowledge into processes
Need for DLR Virtual Aircraft Software PlatformSupport industrial and research activities in GermanyIdentify future options for HPC based aircraft designSimulate all Flight Physics aircraft propertiesrelevant for design & certificationTrade studies for technology evaluationEnhance and maintain aircraft design capabilities in DLR
Towards the Virtual AircraftChallenges & Capability Needs
Multi-DisciplinesLink of preliminary overall aircraft and “detailed” MDA / MDO
High-fidelity for relevant disciplines
Large number of design parameters
Identification of realistic & relevantload scenarios for structurallay-outs (metal, CFRP)
Representation of relevant systemproperties (mass, volume,performance, energy)
www.DLR.de • DLR-AS 3
DLR LamAir
AerodynamicsSeparated flowsTransition laminar/turbulentUnsteady effects
Towards the Virtual AircraftChallenges & Capability Needs
AIAA CFDDrag/High Lift Prediction
WorkshopsAirbus
www.DLR.de • DLR-AS 4
Software & HardwareMassively parallel computer hardwareWorkflow-management environment with “automatic” processesConsistent and comprehensive data formats
Towards the Virtual AircraftChallenges & Capability Needs
www.DLR.de • DLR-AS 5
Reliable Multi-Disciplinary Simulation and Optimization require:Comprehensive verification & validation efforts
Towards the Virtual AircraftVerification & Validation Needs
Wind Tunnel (ETW) Flight Test A/C DLR A320-ATRACFD
HINVA (High Lift Inflight Validation)German Aeronautics Research Program (LuFo) project1st flight tests performed in August 2012 in Toulouse2nd flight tests planned of end of 2014 in Braunschweig
www.DLR.de • DLR-AS 6
Previous WorkHistory
1982-1992
Digital-X
CevcatsCode
Next Generation Code
www.DLR.de • DLR-AS 7
Reynolds-averaged Navier Stokes CodeUnstructured grids, overlapping, adaptation Finite Volume method 2nd orderAdvanced turbulence models, e.g. RSMHybrid RANS/LESInterfaces for multidisciplinary coupling,e.g. FlowSimulatorContinuous verification & validation efforts
Applied in European aircraft industry,e.g. Airbus, Airbus D&S, Airbus Helicopter,RRD)Research platform for European universitiesand research organizations
Previous WorkTAU CFD Solver
www.DLR.de • DLR-AS 8
Courtesy: Airbus D&S
Previous WorkCFD-CSM Coupling
Flight test of high lift configurationMa=0.204, Re=~25M, = 10°
• Trimmed aircraft• Nastran-in-the-loop
Symbols = Flight test data
Wing bending
Simulation
Simulation@Front Spar
Structural Model
CFD grid
Simulation@Rear Spar
Flap bending
www.DLR.de • DLR-AS 9
Previous WorkCFD-FM Coupling
Air DropTime-consuming certification process of newairdrop systems or loads (costs/risks)Generic, blunt cargo bodies and parachutesAircraft wake characterized by strong vorticesRelative motion of cargo and parachute→ coupling with multi-body simulationExperimental validation in wind tunnel
www.DLR.de • DLR-AS 10
Preliminary Aircraft DesignDevelopment of a system forpreliminary aircraft design, CPACS
Previous WorkAerodynamic/Multidisciplinary Design & Optimization
RANS-based Design & OptimizationGradient-free:
High-fidelity methods (aerodynamics & structure)Low number of parameter and load casesHigh manual setup time &computational costs
www.DLR.de • DLR-AS 11
RANS-based Design & OptimizationGradient-based:
1st adjoint code bought from Prof. Jameson ca. 2000Flower adjoint developments (discr.)Development of adjoint in TAU (discr.) ca. 2004Laminar airfoils:Cp on upper side & CD min. @ CL=const
Previous WorkAerodynamic/Multidisciplinary Design & Optimization
www.DLR.de • DLR-AS 12
Preliminary Aircraft DesignDevelopment of a system forpreliminary aircraft design, CPACS
RANS-based Design & OptimizationGradient-based:
1st adjoint code bought from Prof. Jameson ca. 2000Flower adjoint developments (discr.)Development of adjoint in TAU (discr.) ca. 2004Laminar airfoils:Cp on upper side & CD min. @ CL=constBWB optimization, 110k param., Euler, 2011
Previous WorkAerodynamic/Multidisciplinary Design & Optimization
www.DLR.de • DLR-AS 13
Preliminary Aircraft DesignDevelopment of a system forpreliminary aircraft design, CPACS
Previous WorkAerodynamic/Multidisciplinary Design & Optimization
RANS-based Design & OptimizationGradient-based:
1st adjoint code bought from Prof. Jameson ca. 2000Flower adjoint developments (discr.)Development of adjoint in TAU (discr.) ca. 2004Laminar airfoils:Cp on upper side & CD min. @ CL=constBWB optimization, 110k param., Euler, 2011Multipoint shape optimization, 75 param., 2011Aero-elastic adjoint (RANS), deformation includedStructural thickness constant
www.DLR.de • DLR-AS 14
Preliminary Aircraft DesignDevelopment of a system forpreliminary aircraft design, CPACS
Long TermDevelopment of an integrated software platform for multidisciplinaryanalysis and optimization based on high-fidelity methods
Disciplines: aerodynamics, structures, flight mechanics & control
Development of a next generation CFD solver with innovative solutionalgorithms and high parallel performance on future HPC architectures
DLR Project Digital-XTowards Virtual Aircraft Design and Flight Testing
Short Term (2015)Realistic maneuver simulationsIntegrated aero/structural designDemonstration of new capabilities usingconfigurations relevant for industry & researchFirst release of next generation solver
DLR LamAir
Airbus XRF-1
www.DLR.de • DLR-AS 15
Simulations of the Flight EnvelopeCoupling of relevant disciplinesMoving control surfacesEfficient prediction of static and dynamic loads
CFD SolverImproved modelling of physicsRobust and efficient algorithmsSoftware design of next generation solver (HPC)
Multi-Disciplinary Optimization Parametric representation of structureIdentification of critical loads for structural sizingPlanform and airfoil shape optimizationIntegrated process of preliminary and “detailed” design
Simulation EnvironmentIntegration of tools/methods into a single platformEfficient usage of HPC resources
DLR Project Digital-XObjectives 2015
www.DLR.de • DLR-AS 16
DLR Project Digital-XContent and Partner
HAP 2CFD
Solver
Digital-X
HAP 3Reduced Order
Methods
HAP 4Multidisciplinary
Optimization
HAP 5Maneuver at Flight
Envelope Limits
HAP 6Numerical
Uncertainties
HAP 7HelicopterSimulation
HAP 8Automatic Processes
HAP 1Management
Partner / LinksAirbus Industrial boundary conditions Airbus research configuration
XRF-1 Test of methods/processes Research Center Jülich High Performance Computing
Project AeroStrukt (universities)
DLR Institutes Aerodynamics and Flow TechnologyAeroelasticityPropulsion TechnologyStructures and DesignComposite Structures and Adaptive SystemsFlight SystemsAir Transportation SystemsSystem Dynamics and ControlSimulation and Software Technology
Duration: 2/2012 – 12/2015Efforts: 120 PY
www.DLR.de • DLR-AS 17
Improved Modelling of PhysicsReynolds stress models (RSM)
As standard RANS method for all configurationsScale resolving simulations (SRS)
Targeted application for specific components of aircraftTransition prediction and modeling
Necessary for accurate results of turbulence modelsTurbulence modeling improvements
Targeted experimental investigations
DLR Project Digital-XCFD Solver TAU
Improved Robustness and EfficiencyAgglomerated multigrid for complex applicationsImplicit algorithms (hierarchy of pre-cond.)
Integration of advanced turbulence modelsAdaptive local grid refinement
Best practice guideAdjoint solver for efficient sensitivity analysis
www.DLR.de • DLR-AS 18
Improved Modelling of PhysicsReynolds stress models (RSM)
As standard RANS method for all configurationsScale resolving simulations (SRS)
Targeted application for specific components of aircraftTransition prediction and modeling
Necessary for accurate results of turbulence modelsTurbulence modeling improvements
Targeted experimental investigations
DLR Project Digital-XCFD Solver TAU
Improved Robustness and EfficiencyAgglomerated multigrid for complex applicationsImplicit algorithms (hierarchy of pre-cond.)
Integration of advanced turbulence modelsAdaptive local grid refinement
Best practice guideAdjoint solver for efficient sensitivity analysis
www.DLR.de • DLR-AS 19
Improved Modelling of PhysicsReynolds stress models (RSM)
As standard RANS method for all configurationsScale resolving simulations (SRS)
Targeted application for specific components of aircraftTransition prediction and modeling
Necessary for accurate results of turbulence modelsTurbulence modeling improvements
Targeted experimental investigations
DLR Project Digital-XCFD Solver TAU
Improved Robustness and EfficiencyAgglomerated multigrid for complex applicationsImplicit algorithms (hierarchy of pre-cond.)
Integration of advanced turbulence modelsAdaptive local grid refinement
Best practice guideAdjoint solver for efficient sensitivity analysis
www.DLR.de • DLR-AS 20
Future CFD
Current TAU not sufficiently suitedfor strong scaling on future HPC
Full exploitation of future HPC
Consolidation of current DLR CFD solvers
Basis for innovative algorithms & concepts,e.g. higher order finite element
Integration into multidisciplinary simulation
State-of-the-art software engineering methods,C++11, templates, multi-level parallelization
DLR Project Digital-XNext Generation Solver
Shared Memory &Communication Overlap
www.DLR.de • DLR-AS 21
DLR Project Digital-XReduced Order Models
Motivation & ActivitiesHuge aero & aeroelasticity data required for aircraft loads analysis
RANS computations for all cases of the flight envelope is still not feasible
Development of parametric models for static & dynamic aero-loads prediction
Development of correction methods for aeroelastic applications
www.DLR.de • DLR-AS 22
DLR Project Digital-XMDA – Gust Loads
Ma = 0.836 m = 150 to, Re = 77 x 106 , H = 8.2 km gust = 213.36 m, vgust = 10.52 m/s
ObjectiveAnalysis of gust load based on different methods and coupled disciplines
TasksCoupling of disciplines (CFD, CSM, FM)Integration of flight controlModelling / meshing of control surfacesIntegration into parallel simulation backbone FlowSimulator
www.DLR.de • DLR-AS 23
Need for High-Fidelity SimulationsPotential for weight reduction?
ObjectiveMDO of full aircraft based on multi-fidelity
DLR Project Digital-XMDO
MethodLow fidelity methods for preliminary designFast methods for identification of critical load casesRANS methods for aerodynamics and structures for selected load casesParallel software platform for high-fidelityInteractive workflow managementDemonstration case: Airbus XRF-1
www.DLR.de • DLR-AS 24
DLR Project Digital-XMDO – Implementation
www.DLR.de • DLR-AS 25
1st Test of Implemented ProcessOptimize wing with min. fuel burn (cruise) Wing-body geometry, XRF-1 as baselineRANS for aerodynamics, FEM for structureFully-stressed design for structural thickness, 2.5g, -1g loads5 parameters: AR, sweep, twistTarget lift, wing area=const. (parameterization)
No explicit constraintsCruise-climb: 3 altitude segmentsComparison based on ODE integration & Breguet range eq.
DLR Project Digital-XMDO – Implementation
www.DLR.de • DLR-AS 26
Project continues until end of 2015
Progress achieved on disciplinary level as well as in linking all aeronautical institutes of DLR
“Virtual Product” is one of DLR “Leading/Key Concepts”
Follow-on project/s planning process started
DLR Project Digital-XFuture
www.DLR.de • DLR-AS 27
www.DLR.de • DLR-AS Chart 28
Professor Jamesonhas significantly influenced our research!
Thank you very much!
DLR Institute of Aerodynamics and Flow Technology
top related