Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Numerical Aeroelastic Simulation of Aircraft Holger Hennings, DLR – Inst. of Aeroelasticity MUSAF II COLLOQUIUM - AROUND AIRCRAFT & WITHIN ENGINES 18-20 SEPTEMBER 2013, TOULOUSE, FRANCE
www.DLR.de • Chart 1 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 2
Flutter stability
Limit cycle oscillations
Forced Response
Gust response Wake vortex Tail buffeting
Vibration due to flow instabilities
Seperated flow
Bluff bodies Shock buffet
Aeroelasticity (relevant to fixed-wing aircraft)
Chart 3
Flutter stability
Forced Response
Vibration due to flow instabilities
Aeroelasticity (relevant to fixed-wing aircraft)
Linear Frequency Domain Solver Laminar Flow, Shock Buffet Flutter Speed Flutter Analysis
1
2
4
3
Chart 4
Linear Frequency Domain (LFD) Solver
Thormann / Widhalm IFASD 2013
www.DLR.de • Chart 4 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 5
Motivation LFD
Flutter, Gust Response
- analysed in frequency domain
- many parameter combinations have to be simulated
Accuracy of RANS required
RANS expensive
LFD could save time
www.DLR.de • Chart 5 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 6
Validation LFD
Assuming small amplitudes of motion
RANS equations can be linearized
Transformed into frequency domain
Validation of LFD:
LFD solution and RANS time domain solution
should be the same for small amplitudes
www.DLR.de • Chart 6 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 7
Validation case
- Fermat-Model
- 4 · 106 grid points
- Re = 5 · 106
- Spalart-Allmaras turbulence model
- Rigidly trimmed to CL = 0.5
- 2 Mach numbers considered:
- 0.85 and 0.89
steady flow characteristics
www.DLR.de • Chart 7 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 8
Unsteady results: Transfer function of CL
- Good agreement between LFD and time-domain results for the attached case
- Differences for the detached case - LFD and time-domain solver should agree!
Mach = 0.85 Mach = 0.89
www.DLR.de • Chart 8 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 9
Computational Effort
Time Saving Factors:
Memory requirement:
no. of CPUs: 48
Case TD (h) LFD (h) ζCPU
Ma = 0.85, 2.85 periods 33.4 0.38 88
Ma = 0.89, 3.09 periods 40.4 0.55 73
solver memory (GB) M / Msteady
steady RANS 7.20 1.00
Mme domain RANS 7.58 1.05
LFD 161.00 22.36
www.DLR.de • Chart 9 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 10
Intermediate Results
- Good agreement between LFD and time domain results
for the attached case
- Larger deviations for the separated case
- LFD and time domain solution should be same for
small amplitudes
www.DLR.de • Chart 10 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 11
Flutter stability
Forced Response
Vibration due to flow instabilities
Aeroelasticity (relevant to fixed-wing aircraft)
Linear Frequency Domain Solver Laminar Flow, Shock Buffet Flutter Speed Flutter Analysis
1
2
4
3
Chart 12
Influence of transitional flows
on flutter speed
Michael Fehrs IFASD 2013
www.DLR.de • Chart 12 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 13
Background and Relevance
- Increasing the laminar flow regime can reduce friction drag and improve the overall flight performance.
- But: Influence of the laminar-turbulent boundary layer transition on the aeroelastic behavior is not known.
www.DLR.de • Chart 13 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 14
Investigation Set-up Aerodynamic Model: CAST 10-2 Airfoil - CAST 10-2: supercritical (laminar) airfoil
- CFD investigation with the DLR TAU code:
- Ma = 0.500 up to Ma = 0.800 - Rec = 2·106 with c = 0.3 m
- unsteady analysis for αmean = 0.0°
- Wind tunnel data shows large transitional regions for Re = 2·106 .
www.DLR.de • Chart 14 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 15
Flutter Diagram
How does the flutter boundary change in a transitional flow?
www.DLR.de • Chart 15 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 16
Transition Modeling in CFD
RANS models and transition prediction methods turbulence closure transition prediction flow type
SST k-ω model γ-Reθ model transitional
flow
SST k-ω model fully turbulent
flow
www.DLR.de • Chart 16 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 17
Investigation Set-up Aerodynamic Model: CFD grid ~ 60 000 nodes ~ 290 nodes on upper surface ~ 210 nodes on lower surface far field 100 chords away
www.DLR.de • Chart 17 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 18
Results Steady CFD Results: Ma = 0.720
non-linear lift region
lift gain through laminar flow
Lift curve
www.DLR.de • Chart 18 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 19
Results Steady CFD Results: Ma = 0.720
laminar bucket
drag penalty for fully turbulent flow
Drag polar
Chart 20
Results Steady CFD Results: Ma = 0.720
strong influence of transition behavior on lift curve
www.DLR.de • Chart 20 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 21
Results Unsteady CFD results: Overview - Focus on pitch results for:
- Mach number: 0.720 - 6 reduced frequencies k - fully turbulent (blue) and transitional flow (red)
- Results in terms of magnitude |clα| and phase angle Φlα
www.DLR.de • Chart 21 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 22
Results Unsteady CFD results: clα Ma = 0.720: - Different magnitude and phase
response for transitional flow
- Aerodynamic resonance only found for transitional flow
transitional
fully turbulent
www.DLR.de • Chart 22 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 23
Results Unsteady CFD results: clα Resonance found in the non-linear lift region
www.DLR.de • Chart 23 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 24
Investigation Set-up Flutter Analysis and Structural Model
- 2 degree-of-freedom system: - pitch motion around
quarter chord (x0 = 0) - heave motion
- classical bending-torsion flutter
www.DLR.de • Chart 24 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 25
Results Flutter Analysis
www.DLR.de • Chart 25 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 26
Results Flutter Analysis
Turbulent / Transitional flow: - higher flutter stability for
given structural model outside the transonic dip with BL transition
- but deeper transonic dip at lower Mach number with BL transition
www.DLR.de • Chart 26 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 27
Intermediate Results
- CFD based flutter analysis for a supercritical airfoil (CAST 10-2): - Fully turbulent and transitional simulations - Re = 2e6, transonic flight regime
- The transonic dip gets deeper and moves to a lower Mach number for a transitional flow.
- The transitional flows show an aerodynamic resonance connected to an instability of the transitional region.
- For the transitional flows the magnitude and phase response to the prescribed motion differ significantly from the turbulent ones.
- A similar resonance is found in wind tunnel data.
www.DLR.de • Chart 27 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 28
Future Work
- Comparison to wind tunnel data from ongoing experiments with the CAST 10-2 airfoil at the DLR.
- Validation of the transition model’s prediction capabilities for
unsteady transonic flows.
www.DLR.de • Chart 28 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 29
Flutter stability
Forced Response
Vibration due to flow instabilities
Aeroelasticity (relevant to fixed-wing aircraft)
Linear Frequency Domain Solver Laminar Flow, Shock Buffet Flutter Speed Flutter Analysis
1
2
4
3
Chart 30
Shock Buffet
Jens Nitzsche STAB 2012
www.DLR.de • Chart 30 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 31
www.DLR.de • Chart 31 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 32
www.DLR.de • Chart 32 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 33
Shock Buffet - NACA0010, Ma=0.69, Re=10×106, 2-d (U)RANS, SAO - Pitch around c/4
𝜕𝑐↓𝑙 /𝜕𝛼 |↓𝛼=6° <0
www.DLR.de • Chart 33 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 34
Shock Buffet Aerodynamic Resonance
www.DLR.de • Chart 34 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 35
Shock Buffet Experiment in Transonic Windtunnel Göttingen (TWG):
www.DLR.de • Chart 35 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 36
Ma 0.80 Re 1.95×106
Chart 37
Ma 0.80 Re 1.95×106
Chart 38
Ma 0.80 Re 1.95×106
Chart 39
Ma 0.80 Re 1.95×106
Chart 40
Deduction
Shock buffet onset problem seems to be treatable as a linearized stability problem
www.DLR.de • Chart 40 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 41
Flutter stability
Forced Response
Vibration due to flow instabilities
Aeroelasticity (relevant to fixed-wing aircraft)
Linear Frequency Domain Solver Laminar Flow, Shock Buffet Flutter Speed Flutter Analysis
1
2
4
3
Chart 42
Flutter Analysis
www.DLR.de • Chart 42 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 43
Challenges " Applicability of validated CFD methods to a/c
design and aeroelastic certification process
" Analyze aeroelastic stability in full operational and flight domain
" High number of flight conditions, load cases, modes, reduced frequencies, …
" Generalized airloads calculation as fast as by DLM
" Identify critical couplings & conditions
Fast and Reliable Flutter Analysis
by frequency domain methods
www.DLR.de • Chart 43 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 44
Fast and Reliable Flutter Analysis
" Simulations in time-domain to identify influence of nonlinearities at selected flight conditions " from structure, or controls laws " from aerodynamics
by frequency domain methods
www.DLR.de • Chart 44 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 45
Fast and Reliable Flutter Analysis for Complete Aircraft
Strategies 1. Reduction of CPU time for CFD runs
" Massive parallelization of process running nonlinear CFD " Time-linearized RANS " Reduced order modelling (ROM)
" DLM correction methods 2. Efficient approximation methods models to investigate structure and
system nonlinearities in time-domain 3. Uncertainty analysis methods 4. Combination of strategies 1 to 3
www.DLR.de • Chart 45 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013
Chart 46
Conclusion
Linear Frequency Domain Solver validated for attached case
Laminar Flow, Flutter Speed
transonic dip deeper and at a lower Mach Shock Buffet
treatable as a linearized stability problem Flutter Analysis
reduction of CPU time
www.DLR.de • Chart 46 > Numerical Aeroelastic Simulation > Holger Hennings > MUSAF II 2013 > 18.09.2013