aerodynamics and aeroelasticity methodologies …the evolution of modern vertical lift design by fox...
TRANSCRIPT
AERODYNAMICS AND AEROELASTICITY METHODOLOGIES FOR FUTURE CONCEPTS IN VERTICAL LIFTMarilyn J. Smith, PhDProfessorDirector, Vertical Lift Research CenterGeorgia Institute of TechnologyAtlanta, GA US
Oct 15, 2019Aerodynamics Tools and Methods in Aircraft Design
The Evolution of Modern Vertical Lift Design
https://www.lockheedmartin.com/en-us/products/sikorsky-black-hawk-helicopter.html
Main Rotor
Tail Rotor
Fuselage
Empennage
Aeromechanics of a Traditional Helicopter
V∞
W
FXC
Fuselage Flow• Drag• Component Loads
Dynamic Stall
• Loads• Performance
Rotor Fuselage Interactions• Vibration
Rotor-Wake Interactions
• Vibration• Noise• Loads• Performance
Transonic Flow
M < 1• Loads• Noise• Performance
M > 1
Tail-rotorInteractions With:
• Empennage• Main Rotor• Main-Rotor Wake
Performance, Handling-Qualities
The Evolution of Modern Vertical Lift Design
By FOX 52 - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=35908368
Main Rotor
Fuselage
Empennage
RecirculationEffects
Conversion
Multiple RotorsProprotors
Bell FlightJaunt Air Mobility
Urban Air Mobility
FLRAA and FARA Unmanned Vehicles/DronesLow Reynolds Numbers Coaxial/Corotating Rotors
Distributed Electric Propulsion
Shrouded/Ducted Rotors
Vertical Lift in the 21st Century
Aeromechanics Prediction Requirements• Vehicle performance• Blade loads• Airframe & drive train loads• Vibration (rotor and fuselage)• Aeroelastic stability • Flight dynamics • Handling qualities• Noise
UH-60F Hover
UH-60A forward flight
TRAM proprotor
Figures courtesy of US Army
Aerodynamic Tools for 21st Century VL
Conceptual and Preliminary Design- NDARC- GTABB/COMPASS
Detailed Design and Engineering Analysis- Comprehensive Codes: CAMRAD, RCAS, HOST- Panel-Based Methods- Dual Solver Hybrid Methods- Adjoint Design: FUNtoFEM
Physics and High Fidelity Analysis- Helios, FUN3D, OVERFLOW- FLOWer, ELSA, TAU- Academic Codes: Liverpool/Glasgow, UMD
21st Century Aerodynamic Prediction Goals
Johnson et al., “Requirements for Next Generation Comprehensive Analysis of Rotorcraft,” AHS Specialist's Conference on Aeromechanics, San Francisco, CA, January 23-25, 2008.
Multiple Core/Processor CapabilitiesAbility to model or include the effects of
New Technology
Active Flow Control
Active Surfaces/DampersMultiple Rotors/Propellers
New Designs
What Capabilities are Needed?
Multiple Core/Processor CapabilitiesAbility to model or include the effects of
New TechnologyNew Designs
What Capabilities are Needed?
Incorporate the effects ofEnvironment Urban and Nap of the EarthNear-ground OperationsIn-Flight Operations
Slung Loads
Robotic Landing Gear
Large Gusts in Urban Canyons
Gust Responses: Urban Canyon Typical
What Capabilities are Needed?
Reduce the scope of wind tunnel testing during design and flight testing for certification
• Digital Threads• Uncertainty Quantification• Removal of User-generated Errors
Quantitative Data Analysis Techniques
• Coherence: Linearity between response of two data sets, illuminates missed harmonics. Coh < 0.6 →missed harmonic
• Cost Function: Scalar measure of difference between response of two data sets
J < 100 → Very little differenceJ < 50 → Virtually identical
where !",!$, %&' !( are weighting factors for coherence, magnitude, and phase, respectively
) = 20& -
./
.0!" !$ 123 − 2 5 +!( ∠123 − ∠2
5
J= 10 40 1000
CIFER• Developed for system identification (USRA)• Computes difference between frequency response of two datasets
Cho et al. "System Identification and Controller Optimization of a Coaxial Quadrotor UAV in Hover," AIAA, 2019Smith et al., “Towards Certification of CFD as Numerical Experiments for Rotorcraft Applications,” Aeronautical Journal,122(1247), 104-130.
Quantitative Data Analysis Techniques
J= 10 40 1000
Cho et al. "System Identification and Controller Optimization of a Coaxial Quadrotor UAV in Hover," AIAA, 2019Smith et al., “Towards Certification of CFD as Numerical Experiments for Rotorcraft Applications,” Aeronautical Journal,122(1247), 104-130.
Aerodynamics for Design
and Modeling & Simulation
Aerodynamic Tools for
Designand
Modeling & Simulation
NASA Design and Analysis of Rotorcraft (NDARC)Capabilities for:
Off-design mission analysisFlight performanceVehicle sizing
Flexibility for non-conventionalvehicles through • Synthesis of component modules• Propulsion system options (electric,
turbojet, turbofan, reaction drives, fuel cells, etc.)• Surrogate models for trade-space analysis
• Silva et al, “Multidisciplinary Conceptual Design for Reduced-Emission Rotorcraft,” AHS Specialists Conf., San Francisco, California, USA, January 16-18, 2018.
• Johnson, “NDARC: A Tool for Synthesis and Assessment of Future Vertical Lift Vehicles,” Vertiflite, Nov-Dec, 2014, pg. 26-27. Prior slide figure from this work.
Optimization with NDARC
A New Approach for Design and M&S : GTABB and COMPASS
• Development of physics-based reduced order models for aerodynamic and bluff bodies
• Extensible to new configurations: Not an interpolation• New features:
Shadowing orientation effects GUIRotor and engine effects Multiple body types
• Validated with CFD, wind tunnel, and flight test• Adopted/in adoption by US Army, US Navy, Drone Racing
League, academia• Slung loads• Control law and autonomous algorithm design• Virtual reality-based training (M&S)
• Prosser, D. and Smith, M. J., “Physics-Based Aerodynamic Simulation Models Suitable for Dynamic Behavior of Complex Bluff Body Configurations,” American Helicopter Society 71st Annual Forum, May, 2015. See also JAHS.
• Koukpaizan et al. “Rapid Vehicle Aerodynamic Modeling for Use in Early Design,” in Proceedings of the AHS Aeromechanics Design for Transformative Vertical Flight Conference, San Francisco, CA, January 16–19, 2018
GT Aerodynamics of Bluff Bodies (GTABB)
17
External Control Module
Unsteady Aerodynamics in Real Time • Unsteady phase lag• Large vortex shedding
fluctuations• Three-dimensional effects
(finite bodies)
No vortex shedding
Withvortex
shedding
US Army Blackhawk Flight Test Validation
Captures Onset of Instability
COMPlex Aerodynamic Shape Simulation (COMPASS)
• Use canonical shapes with corrections to estimate the characteristics of complex shapes.
• Add corrections for shadowing (feature blockage) and shear layers
COMPASS quadrotor representation (without rotors)
Nondimensional Aerodynamic CharacteristicsFigures from Prosser and Smith (2016)
Predict the Cp distributionabout the body
Integrate to get force andmoment coefficients
Nondimensional Aerodynamic Characteristics
Prosser, D. and Smith, M. J., “Aerodynamics of Finite Bluff Bodies,” Journal of Fluid Mechanics, Vol. 799, No. 6, pp. 1–16, 2016,
Quasi-Steady Predictions
Drag Coefficient Lift Coefficient
Typical Control Law Design:• Flat plate wetted area used for forces• Moments are neglected
trim (Deg) Range (Km) Flight Time (Min)0
5
10
15
20
25
30Flat PlateUnsteady
FP
FPFP
Vehicle Performance
Flat plate aerodynamics overpredicts performance by ~10%and underpredicts rotor trim by ~15%
Split-S Trajectory predicted by GTABB
HeightAbove
Ground (m)
Flight Test Validation
https://www.youtube.com/watch?v=UNoGxq8pQGE, Courtesy Drone Racing League
Real Flight Vehicle
Simulated Flight Vehicle
Hybrid CFD Analyses
CFD + Free Wake Model
• Data is exchanged between the CFD and free wake code at fixed intervals
• CFD near-body solver calculates the sectional loads along the blades as the solution advances
• The vortex filaments which model the wake are updated based on the sectional loads
• The outer boundary conditions of the CFD domain are updated from the wake-induced flow field to reflect the influence of the wake
Outer Boundary CFD
Wake Vortex Filaments
Hybrid CFD Methods
• Promise of CFD-level accuracy without the costs, up to 90% savings in some cases• Many approaches through the decades, but with
limitations:• Single blade• No fuselage (except DLR)• Very mixed results with significant errors in pitching
moment, structural bending, and hub forces• Some have known numerical formulation errors• Many are “academic” codes without formal version
controlPast (US) engineering experience not positive to
adoption in work flow
CDI-GT Hybrid CFD Approach• CFD/CFD-CSD provide highest accuracy at highest cost• Hybrid CFD methods mitigate costs by 50%-90% while maintaining accuracy
• Increase CFD-based applications to earlier design and additional analysis
Cost Accuracy
• Carefree: Couple CFD solvers with commercial wake solvers
• Flexible: Takes advantage of near-body capabilities
• Physics: Multiple level of physics
• Cost-effective: Can use same meshes
• Expanded Capabilities: Multiple components, full vehicles
Non-contiguous Methodology
Contiguous vs non-continguous
• Remove the inertial background grids• Allow boundary motion and arbitrary boundary shapes• Velocity at all unblanked outer boundaries determined by
the free wake code
Blade Vortex Interaction
Conglomeration of vortex filaments modeling the tip vortex
Illustration of tip vortex passing between domains
• Mtip = 0.65, trimmed to CT/σ=0.09• 3.9M nodes per blade (15.4M total)• O-grid topology
S-76 Hover
Helios Predictions with OVERFLOW as near-body solver (Full CFD, 448M nodes) Mark Potsdam, Rohit Jain (Army ADD)
Noncontiguous Meshes
Hover is complex computation with significant wake interactions
S-76 Integrated Loads
Torque coefficient vs. Thrust coefficient Figure of Merit
Figure of Merit
Exp. uncertainty ~0.6 countsHelios ~ 1 count OF-Charm ~ 2 countsOF-alone >2 counts
Comparable or better to all US and International participants
Lateral Wake
VerticalWake
Non-contiguous wake predictions (lines)comparable to those in Contiguous CFD domain(symbols)
Computational cost
The full grid non-contiguous simulations cost about 7.6% of the number of core-hours as the stand-alone
engineering OVERFLOW simulations
Simulation Number of grid nodes (millions)
Core-hours Cost ratio
OVERFLOW-Helios (Jain 2015) 448 122,880 57.5
OVERFLOW (Narducci 2015) 63.4 28,080 13.14
OVERFLOW-CHARM with background grids
28 14,400 6.74
OVERFLOW-CHARM non-contiguous grids (5 revs)
15.4 2,140 1
Jacobson, K. and Smith, M. J., “Performance and Physics of a S-76 Rotor in Hover With Non-Contiguous Hybrid Methodologies,” AIAA SciTech, AIAA-2016-0302, San Diego, CA, January, 2016
Evaluation in Forward Flight• Baseline mesh was provided by Boeing-PHL • Represents a typical near-body mid-size mesh• 3.7M point mesh on each of four blades• 32.9 M background Cartesian mesh (orig)• Menter kw-SST turbulence model with full viscosity in
the near-body region and Euler terms in the off-body region
• Sensitivity to mesh sizes, coupling updates, turbulence model
• Aeroelastic coupling• Examination of different rotor blade structural
properties on same rotor (experimental set-up)• Evaluation of fuselage and wind tunnel floor effects
• Wilbur et al., “Complex Vehicle Design and Analysis with Hybrid Methodologies,” AHS Aeromechanics Conference, San Francisco, CA, January 16–19, 2018. Coming soon: Journal of Aircraft
• Min, B.-Y. et al., “Toward Improved UH-60A Blade Structural Loads Correlation,” AHS 74th Annual Forum, Phoenix, AZ, May 14-17, 2018
NFAC Tunnel Test
Dual-Solver Vortex Interactions
Correlation with Industry CFD-aloneNormalized, means removed
r/R=0.4
r/R=.92
• Not a correlation with experiment – correlation with “best industry mesh/practices”• Reduction in background mesh by 40% nodes – no difference in loads• Convergence occurs at 75% of full CFD convergence requirements• None of the poor correlations with structural, hub or pitching moments as
observed with other hybrid approaches
Full CFD AnalysesFull CFD Analysis
Rotorcraft simulation and adjoint-based derivative evaluation framework
FUNtoFEM: Adjoint-enabled aeroelastic analysisFUN3D
Surface integration Flow solver
Mesh warping
FUNtoFEM
Load transferLoad adjoint
Displ. transferDispl. adjoint
FEM/TACS
Forward pathAdjoint path
I General interface for aeroelastic load and displacement transferbetween FUN3D and a finite-element code
I E�cient hand-coded partial derivatives terms and transposeJacobian-vector products that enable adjoint-based derivativeevaluation
I Di↵erent transfer schemes enabled through an abstract interfacelayer: MELD and RBF (for surface-to-surface transfer), beam-specifictransfer
I Di↵erent bodies can be assigned di↵erent transfer schemes
3 / 29
Time-accurate Aeroelastic Rotorcraft Simulation
I On-going work to verify against HART-II case
23 / 29
• FUNtoFEM: A general interface for aeroelastic simulation and adjoint-based gradient evaluation
• Enables use of either loosely or tightly coupled simulation• Separation of time integration schemes for CSD/CFD• Different bodies can employ their own load/displacement
transfer methods• Efficient hand-coded derivative terms enable efficient
computation of coupled Jacobian-vector products
Comparison of natural frequencies with DYMORE
I Fan plot for HART-II type model
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2�/�ref
123456789
10111213141516
�/�
ref
TACS flap TACS lag TACS torsionTACS flap TACS lag TACS torsion
25 / 29
(Graeme Kennedy at Georgia Tech)
Helios Overview
Steady forward flight
Sikorsky X2™
Unsteady maneuver
Wake resolution
Interactional aerodynamics
Elastic bending
• Rotary-wing product of CREATE™-AV− Jointly developed by CREATE and Army ADD
• Multi-mesh/Multi-solver- Strand, unstructured, structured curvilinear
- Cartesian high-order with AMR
- Automated domain interpolation
• Interfaces to rotorcraft comprehensive analysis codes for CSD and trim
• Scalable execution on HPC systems
• 100+ active user licenses- DoD organizations (Army, Navy)
- Various US rotorcraft companies
- Various academic institutions
• Helios integrates simulation codes from CREATE, Army, and outside organizations
• Software integrated through an extensible Python infrastructure
UH-60 main/tail simulationCourtesy U.S.Army AED
H-47 tandem simulationCourtesy Boeing
JMR “Defiant”Sikorsky/Boeing
JMR “Valor” Bell Helicopter
CH47 – Block II Boeing/AED/ADD
AH-64 Apache Courtesy Boeing
Cap
abili
ty
V2
AMR
V6V5
SIF, PUNDIT, SAMARC, NSU3D, RCAS, SAMRAI
V3
Rotor-Fuselage
V4
Multi-rotor
PARAVIEWOVERFLOW
V1
Dualsolver
CAMRADIICSI MELODI
SAMCartFUN3D
V7
mStrandKCFD
Developed under CREATE
Third-party packages
2010 2011 2012 2013 2014 2015 2016
Propellers/Propulsers
Maneuver/Strand slvr
Annual releases
COVIZ
DES Wake
V8
Automated Strands
20182017
V9
3D Structures
US Army Helios Development
AH-64 E Apache, Hover:
Narducci & Tadghighi, BoeingAIAA-2016-0564
• High-fidelity, time-accurate rotor and fuselage combinations
• Boeing Mesa
Apache INSTALLED ROTOR PERFORMANCE
V2?B1? V4
V3B2 B3
B4
B1T3iT1i
T4i W4T1o W1
Blade 4
Blade 3
Blade 2
UH-60A Rotor Modeling & Validation§ Blade tip vortex (B)§ Twist/planform vortex (V)
§ Trim tab vortex (T)§ Wake sheet (W)
Blade 1CFD
Helios wake velocities, vortex locations and strength predictions are within experimental errors
PIV
NASA/Army UH-60A NFAC test
Sikorsky X-2™ TD
Helios models the interactional aerodynamics between lift-offset coaxial compound rotors, fuselage, and propulsor configuration
Main rotor unsteady lift
Reed, Egolf, “Coaxial Rotor Wake and Prop Induction Impact on a Horizontal Tail using Helios”, AIAA-2015-0554, 53rd AIAA Aerospace Sciences Meeting, AIAA SciTech Forum, Jan 2015
Nor
mal
ized
Am
plitu
de Unsteady fuselage lift
Unsteady propeller thrust
82 kts
1 rev
1 rev
6 revs
Validation with Bell XV-15 Tiltrotor
Reingestion
Reingestion
Wake impingement
Fountain flow
!"/$Download/Thrust (%)
Wing Fuselage Nacelle Total
CFD 0.1195 10.31 5.41 1.07 16.79
Flight test (Arrington et al.)
0.1271 – 0.1296 - - - 13.42 – 14.50
Config 1FUN3D-
OVERFLOW
Config 2mStrand
OVERFLOW
FUN3D
mStrand
mStrand
Felker et al, ‘87
Download Validation with JVX
FUN3D-OVERFLOW mStrand
0.01360.0138
0.0133
0.007410.007540.0075
0.0000
0.0025
0.0050
0.0075
0.0100
0.0125
0.0150
0.0175
0.0200
Rotor Thrust CT
Low thrustq = 6o
High thrustq = 12o
Test
FUN3D/OFLOW
mStrand
0.001350.001320.00134
0.0008460.000852
0.000813
0.00000
0.00025
0.00050
0.00075
0.00100
0.00125
0.00150
0.00175
0.00200
Wing/Flap Download CZ
10.2%9.6%10.1%
11.4%11.3%
10.8%
0% 2% 4% 6% 8% 10%12%
14%16%
18%20%
Download - CT/CZ
JVX DOWNLOAD ANALYSIS
• 40 revs, 9 sec (real-time)
• Flight path angle 3 deg at start to 35 deg at end
Manuever: UH-60A UTAAS Pull-Up
UH-60A UTTAS ManeuverSection Pitching Moment Cm M2
86.5 % R
UH-60A UTTAS Maneuver : Pushrod loads
Helios demonstrates big improvement over RCAS for pushrod load prediction
Pushrod load
Joint CFD-Experimental Validation!• Full-scale UH-60A rotor in Air Force 40- by 80-foot Wind Tunnel under
NASA/Army Airloads Wind Tunnel Test Program (2010)• Extensive database of performance, hub loads, air loads, structural loads, wake
PIV, blade deformation, RBOS on highly instrumented rotor for validating analytical tools
• PIV phase• Wake measurements at 90 deg azimuth over 50% of outer blade
• Vortex characteristics extracted: circulation, size, position
NASA/Army UH-60A NFAC test
PIV camera portslaser launch port
PIV ROI
mirror port
smoke generators
airflow
Future Research Directions• Continuous improvement in speed and efficiency on multi-core processors for
all levels of aerodynamic prediction tools• Integration of reduced-order modeling that is physics-based for earlier design
of maneuvering and operational effects• Advanced design capabilities for rapid, highly-accurate design• Using high-fidelity methods, exploration of physics to understand complex
interactional phenomena and shortcomings of current algorithms• Improved prediction of acoustics (interior and exterior) from unsteady aerodynamics• Jointly designed experimental datasets with sufficient information for CFD validation!
• International collaboration to solve complex problems
Multi-rotor trim
3D Structures
Acknowledgements• This effort was partially sponsored by the U.S. Government under
• U.S. Army/Navy/NASA Vertical Lift Research Center of Excellence under the directionof Mahendra Bhagwat of AFDD, Agreement No. W911W6-11-2-0010.
• Other Transaction number W15QKN-10-9-0003 between Vertical Lift Consortium, Inc. and the Government
• Navy STTR contract N68335-09-C-0335 with guidance from technical monitors Jennifer Abras and Mark Silva
• US DOE STTR DESC0004403
• The US Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright notation thereon.
• The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the U.S. Government.
• The authors would like to thank• Andy Wissink and Roger Strawn of CCDC Aviation Command for the Helios figures
and slides• Glen Whitehouse and Dan Wachspress of Continuum Dynamics, Inc.• Rohit Jain and Mark Potsdam for their insights and figures• Ted Meadowcroft of Boeing-Philadelphia for the UH-60A computational mesh and input
decks
• All slides that require approval have been previously approved for prior presentations.
Questions?