Aerospace Testing 2011, Hamburg, Germany, April Jan Debille Solutions Manager Aerospace & Defense

Transcription

1 Industrial solutions for in-flight & offline experimental flutter analysis A. Lepage, P. Naudin, J. Roubertier, A. Cordeau ONERA M.A. Oliver-Escandell, S. Leroy, AIRBUS Jan Debille, LMS Aerospace Testing 2011, Hamburg, Germany, April Jan Debille Solutions Manager Aerospace & Defense

2 Presentation outline Flutter testing: What, When and How? Required technology Industrial implementation Validation Conclusions 2 copyright LMS International

3 What is Flutter? Flutter is an aero-elastic phenomenon Unstable self-excited vibration Structure extracts energy from the air stream Flutter starts to occur at a certain speed Negative damping start to occur at flight points where two modes are coupled in an unstable way Typical coupling: wing bending/torsion, wing torsion/control surface, wing/engine 3 copyright LMS International

4 Stages of Aircraft Development & Flutter: When? Feasibility Concept Definition Development GVT/Flutter/FEC In Service Market Study Concept Selected Authority To Offer Program Launch Agreement With Primary Partners Component Design Major Assemblies Major Body Sections First Flight Certification Entry Into Service Full Aircraft Subsystem CAE Component CAE Component Physical Test Full Virtual Prototype Full Physical Test Performance Performance Exploration Exploration Certification Certification Models & Loads Subsystem Component Subsystem Physical Test Concept Validation & Target Cascading Upfront Engineering Detailed Engineering Refinement Engineering 4 copyright LMS International

5 Flutter in the design process flow Feasibility Concept Definition Development Gvt/Flutter/FEC In Service Market Study Concept Selected Authority To Offer Program Launch Agreement With Primary Partners Component Design Major Assemblies Major Body Sections First Flight Certification Entry Into Service Virtual Prototype FE Model Analytical Modal Model Pre-Test & De-Risking Update / refine Models Flutter Simulation & Prediction Flight Envelope Opening Physical Prototype Ground Vibration Test Identify & Validate Modes Correlate Model GO-NO/GO First Flight Define Flight Envelope Flight Envelope Clearance Flight Envelope Expansion LMS Flutter Analysis 5 copyright LMS International

6 Aero-elastic simulation and in-flight flutter testing Traditional FEM, GVT-updated FEM, or direct GVT M&& x( t) + Cx& ( t) + Kx( t) F ( x) = a 0 Aerodynamic panel method Due to presence of aero-dynamic term, modes of structural system are changing with airspeed and altitude Flutter analysis = assessing evolution of modes (zero-crossing of damping value) FE Model Test Model (GVT) Aerodyn. Panel Model Physical prototype 6 copyright LMS International

7 Flutter procedure: extract from NASA technical memo Fly at several stabilized speeds Increasing dynamic pressure Increasing MACH number Ref: NASA Technical Memorandum 4720, A Historical Overview of Flight Flutter Testing, October copyright LMS International

8 Flight flutter testing & in-flight modal analysis Background Testing Analysis Damping Flutter (m/s2) Real s Airspeed Telemetry link g 2 Amplitude Hz 8 copyright LMS International

9 Flutter testing procedure Find frequency and damping of critical modes For increasing Speeds increases the dynamic load At different Altitudes the lower the altitude, the higher the dynamic load At different MACH values Altitude (feet) 40,000 30,000 MACH 0.85 MACH 0.90 MACH ,000 10, True Air Speed (knots) 9 copyright LMS International

10 Presentation outline Flutter testing: What, When and How? Required technology Industrial implementation Validation Conclusions 10 copyright LMS International

11 Flutter testing requirements Modal Analysis Modal Analysis on operational (output-only) data Accuracy Get accurate damping estimate in an operational situation Speed Waiting Time is Money aircraft is airborne during the analysis Waiting Time is Dangerous during the analysis time, the aircraft may be exposed to near-flutter conditions! 11 copyright LMS International

12 EUREKA project FLITE2 Structural testing and modal analysis for aeronautics and space applications Airbus France Dassault Aviation Lambert Aircraft Engineering PZL Mielec LMS LMS Net Funding in FLITE2: 378 keur ILOT INRIA ONERA SOPEMEA University of Brussels (VUB) University of Krakau (AGH) University of Leuven (KUL) University of Manchester (UMAN) 12 copyright LMS International

13 EUREKA project FLITE2 Structural testing and modal analysis for aeronautics and space applications / Faster testing for GVT Smart combination of broadband / sweep / stepped Assessment of non-linear behavior using multi-sines Modal parameter estimation: iterative methods using noise information and yielding uncertainty bounds on estimates (PolyMAX results as starting values) Flight flutter testing: OMAX identification framework, i.e. combination of known and unknown excitation (EMA + OMA) Use of GVT for flutter safety prediction 1.00 Aerodynamic panel model ((m/s2)/n) db F F B FRF0 Variance0 COH0 Amplitude 0.00 Hz 13 copyright LMS International

14 Stable, robust and reliable modal analysis on operational data Identification of modal parameters from response data (accelerations) measured in operating conditions Eigenfrequencies Damping ratios Mode shapes Operational modal analysis = identifying H Based on Y U Without knowing U (BUT white noise assumption) White noise H Y White noise + harmonic 14 copyright LMS International

15 Output only: artificial vs. natural excitation Operational Modal Analysis: Output-only analysis no FRFs but Crosspowers between responses and reference responses Reference responses: wing tips, tail tips, nose; in general: well excited points Operational PolyMAX Requires natural, operational excitation! 0.10 ( g/n) Log NASA OMA with artificial excitation Only operational responses are considered but: all modes are well-excited due to force input! and: additional operational excitation used Phase Hz PolyMAX 15 copyright LMS International

16 16 copyright LMS International Modal analysis LMS PolyMAX - theory & implementation Step 1: Denominator matrix polynomial (in z-domain) Poles and participation factors Step 2: Stabilisation diagram Step 3: LSFD to estimate mode shapes and upper/lower residues from selected poles [ ] [ ][ ] [ ] [ ] [ ] " " ) ( ) ( ) ( z z z A B H p p p p β + + β + β = ω ω = ω K z z z p p p p α + + α + α K [ ] [ ] [ ] [ ] 0 ) ) (, ( 1 0 = α α α ω ω p H M L [ ] { } { } UR LR l v l v H n H i i T i i + > < + > < = ω * ) ( j j i i i ω ω λ ω λ = 2 1 *

17 Modal analysis LMS PolyMAX vs. LSCE Same 3-step procedure Time MDOF Step 1 differs LSCE uses impulse responses PolyMAX uses FRFs Big difference in stabilization diagram PolyMAX LMS PolyMAX excels in both high and low damping cases! 17 copyright LMS International

18 Modal analysis LMS PolyMAX vs. Other Frequency-Domain Methods Frequency domain methods p p 1 p p 1 [ H ( ω) ] = ([ β ]( jω) + [ β ]( jω) + K+ [ β ]). ([ α ]( jω) + [ α ]( jω + K+ [ α ]) 1 Powers of the frequency axis Numerical conditioning problems Consequences: Not in PolyMAX! z p Limited frequency range (ω) Limited model order (p) jω t e = ω = 2π( f f1) t = 1 2( f f ) LMS PolyMAX excels in broadband, high model order analyses! p 1 0 p p 1 ) end 1 f end Im f 2 f 1 0 Re 18 copyright LMS International

19 Modal analysis LMS PolyMAX Modal analysis, an area where no substantial advances were to be expected? Extremely clear stabilization Easy pole selection Faster analysis User-independent results More modes found General purpose method Single broadband analysis High & low damping Noisy data LMS PolyMAX, A Revolution in Modal Parameter Estimation! 19 copyright LMS International

20 LMS Test.Lab automatic modal parameter selection Speed up modal analysis One push instead of manual selections Rule-based method Not affected by ability of human mind to treat information High accuracy on pole selection Reduce uncertainty Improve productivity NASA Guidance tool for all All physical poles selected at a glance in stabilization diagram Extensible to automatic modal analysis Analyze multi-patches measurement Low modal density cases (ex. Bodyin-white car) Flight qualification of aircraft Structural damage detection / Structural health monitoring 20 copyright LMS International

21 Presentation outline Flutter testing: What, When and How? Required technology Industrial implementation Validation Conclusions 21 copyright LMS International

22 Flutter testing procedure: Data acquisition Telemetry ground station Flight parameters (Altitude, CAS, MACH, etc.) TCP/IP In-flight data recorder & telemetry transmitter Dynamic data D/A ONLINE data preparation OFFLINE data preparation Flight data LMS Scadas Mobile data acquisition system 3 rd party software: Data tape/card reader On-board flight data recorder and telemetry system Ground station: receive data, split fast/slow channels Data acquisition with Scadas F/E or 3 rd part software 22 copyright LMS International

23 Flutter testing procedure: cyclic Measure with Spectral Testing from telemetry Average flight parameters fix the flight point Automate OMA (minimal interaction) Evaluate evolution of f and ζ for each mode in display Flight envelope definition: Determined by real-time flight parameters. After 1 cycle, the pilot is instructed to move on to the next flight point Complete offline processing possible: allows in-depth analysis afterwards, accounts for telemetry-based data errors The Flutter application consists of 2 dedicated GUIs: A Flutter Progress window that is always on top: The dedicated Flutter worksheet which 23 copyright LMS International

24 Flutter analysis: viewing results Measurements are displayed in the flight envelope Selected poles are added to the Pole Table, based on a match in frequency using pole tolerance parameter. Poles are displayed in U/L display with amplitude in the upper display and damping in the lower display. The Displays group allows to select a different x-axis, or to fix against a slow channel parameter (e.g. MACH value) Support for military damping standard g=2*ζ 24 copyright LMS International

25 Flutter testing: Overview Analysis Operational Modal Analysis with artificial excitation & PolyMAX parameter estimation method Automated AMPS: automatic modal parameter selection in stabilization diagram Repeatability AMPS: deterministic method! Always yields the same results with the same parameters Expert interaction Experts can interact with the automated procedure Speed Flutter condition can be determined in as fast as 10 seconds 25 copyright LMS International

26 Presentation outline Flutter testing: What, When and How? Required technology Industrial implementation Validation Conclusions 26 copyright LMS International

27 ONERA flutter data set generator Altitude Speed Transonic flutter simulator Time domain vibration response to wing aileron impulse Aim: Simulate data for testing Classical wind-bending and torsion coupling Procedure: Input altitude and speed Run simulation to get time histories Output: Time domain data of 4 simulated sensor responses Model (state-space): Structural: GVT result: first 7 symmetrical wing vibration modes Aerodynamic: Generalized aero-elastic forces - Doublet Lattice Method Input for time-response calculation: impulsion on the command of a wing aileron 27 copyright LMS International

28 ONERA Flutter simulator dataset Constant MACH Measured MACH 0.8 Available data set: 8 flight points constant MACH # decreasing altitude Altitude (m) increasing airspeed => increasing dynamic pressure Speed (km/h) 28 copyright LMS International

29 ONERA Flutter Simulator sample time data (1) 4 channels per flight point 16 seconds per flight point 32 seconds for nearflutter 4000 m set 29 copyright LMS International

30 ONERA Flutter Simulator sample time data (2) Evolution vs. altitude Decrease in damping clearly visible in response 30 copyright LMS International

31 ONERA Flutter Simulator evolution of modes vs. flight conditions Flutter analysis 40 Frequency (Hz) CAS (knots) Mode 1: 9.79Hz Mode 2: 10.05Hz Mode 3: 16.76Hz Mode 4: 19.03Hz Mode 5: 27.75Hz Mode 6: 34Hz Mode 7: 34.02Hz 7 modes in model Modes 1 & 2 couple Structural frequencies: 8.98 Hz ; 5.88 % Hz ; 1.81 % Flutter analysis CAS (knots) Damping (%) Mode 1: 9.79Hz Mode 2: 10.05Hz Mode 3: 16.76Hz Mode 4: 19.03Hz Mode 5: 27.75Hz Mode 6: 34Hz Mode 7: 34.02Hz copyright LMS International

32 LMS PolyMAX results 8000 m altitude First bending mode: 8.9 Hz ; 5.9 % First torsion mode: 10.5 Hz ; 1.8 % 32 copyright LMS International

33 LMS PolyMAX results 4000 m alt. near-flutter condition Strong coupling between first bending and first torsion mode First bending mode: 9.8 Hz ; 12 % VERY HIGH damping First bending mode: 10.1 Hz ; 0.16 % VERY LOW damping 33 copyright LMS International

34 Quality check of PolyMAX modal parameter extraction Synthesis of Cross Powers 0.09 V 2 (1/s)s Log 2.22e-3 V 2 (1/s)s Log Green = Synthesized Red = measured CrossPower Aile:Point11:+Z/Aile:Point15:+Z Synthesized Crosspow er Aile:Point11:+Z/Aile:Point15:+Z CrossPow er Aile:Point11:+Z/Aile:Point15:+Z Synthesized Crosspow er Aile:Point11:+Z/Aile:Point15:+Z 15.1e e Synthesized Crosspow er Aile:Point11:+Z/Aile:Point15:+Z 0.00 Hz Synthesized Crosspow er Aile:Point11:+Z/Aile:Point15:+Z 0.00 Hz m safe flight point 4000 m dangerous flight point 34 copyright LMS International

35 LMS PolyMAX results overview Evolution of frequency & damping as a function of altitude Decreasing altitude increase of dynamic pressure 35 copyright LMS International

36 Final check: comparison analytical modes vs PolyMAX result Frequencies Frequency vs Altitude Frequency (Hz) Analytical Mode 1 Analytical Mode 2 Calc. Mode 1 Calc. Mode Altitude (m) 36 copyright LMS International

37 Final check: comparison analytical modes vs PolyMAX result Damping Damping vs Altitude Damping (%) Analytical Mode 1 Analytical Mode 2 Calc. Mode 1 Calc. Mode Altitude (m) 37 copyright LMS International

38 Flutter testing at Airbus Fly-by-wire: excitation via control surfaces Sweep: detailed engineering / Pulse: crew/aircraft safety ONERA: MEFAS (Methodes et Exploitation des essais de Flottement de l Avion Souple) Pierre Vacher, Alain Bucharles: A Multi-Sensor Parametric Identification Procedure in the Frequency Domain for the Real-Time Surveillance of Flutter, SYSID copyright LMS International

39 Accelerometers General Group Only primary control surface accelerometers have been removed Otherwise control surfaces modes identified rather than structural ones 150 accelerometers Distributed over the main aircraft structure 39 copyright LMS International

40 Geometry representation (adding slave DOFs) 40 copyright LMS International

41 Accelerometer sub-set Reduit Symetrique Group Used for in-flight real-time analysis (MEFAS) Fast and accurate identification Accelerometers with best SNR Wing tips Elevator tips Engines Some on fuselage 41 copyright LMS International

42 Symmetric sweep Left wing tip and right wing tip Time WINL:951:+Z Time WINR:951:+Z Time WINL:951:+Z Time WINR:951:+Z Amplitude Amplitude s s 42 copyright LMS International

43 FRFs fuselage (g/n) db Real / (g/n) db Real / F B FRF 2: :+Z/2: :+Z Coherence 2: :+Z/2: :+Z F B FRF 2: :+Z/2: :+Z Coherence 2: :+Z/2: :+Z Hz Hz (g/n) db Real / (g/n) db Real / F B FRF 2: :+Z/2: :+Z Coherence 2: :+Z/2: :+Z F B FRF 2: :+Z/2: :+Z Coherence 2: :+Z/2: :+Z Hz 0.00 Hz 0.00 nose rear Central bottom front 43 copyright LMS International

44 FRFs engines (g/n) db Real / (g/n) db Real / F B FRF 2: :+Z/2: :+Z Coherence 2: :+Z/2: :+Z F B FRF 1: :+Y/2: :+Z Coherence 1: :+Y/2: :+Z Hz Hz (g/n) db Real / (g/n) db Real / F B FRF 2: :+Z/2: :+Z Coherence 2: :+Z/2: :+Z F B FRF 1: :+Y/2: :+Z Coherence 1: :+Y/2: :+Z Hz 0.00 Hz 0.00 Outer Z Outer Y Inner Z Inner Y 44 copyright LMS International

45 FRFs wings (g/n) db Real / (g/n) db Real / F B FRF 0: :+X/2: :+Z Coherence 0: :+X/2: :+Z F B FRF 2: :+Z/2: :+Z Coherence 2: :+Z/2: :+Z Hz Hz (g/n) db Real / (g/n) db Real / F B FRF 0: :+X/2: :+Z Coherence 0: :+X/2: :+Z F B FRF 2: :+Z/2: :+Z Coherence 2: :+Z/2: :+Z Hz 0.00 Hz 0.00 Left X Left Z Right X Right Z 45 copyright LMS International

46 FRFs elevator (g/n) db Real / (g/n) db Real / F B FRF 2: :+Z/2: :+Z Coherence 2: :+Z/2: :+Z F B FRF 2: :+Z/2: :+Z Coherence 2: :+Z/2: :+Z Hz 0.00 Hz 0.00 Left Right 46 copyright LMS International

47 Coherence & geometry 128 coherence functions in excitation frequency band Averaged over excitation frequency band Averaged over DOFs / node Averaged coherence color scale from Geometry mapping / Amplitude 0.00 Hz 47 copyright LMS International

48 FRF PolyMAX Accelerometer sub-set (g/n) db F F Sum FRF SUM Synthesized FRF SUM Hz 48 copyright LMS International

49 FRF PolyMAX All sensors (g/n) db Phase Hz Sum FRF SUM Synthesized FRF SUM 49 copyright LMS International

50 Influence of pre-processing Frequency variations Varying block size N, N/2, N/4, N/8 Other FRF estimation parameters constant (Hanning window, overlap) PolyMAX results Frequency variations small (±2%) Dramatic damping ratio variations (+200%) Trade-off Smaller block size: Hanning window bias larger Larger block size: noise variance larger (few averages) Damping variations 50 copyright LMS International

51 Influence of pre-processing Workaround for trade-off 1 st step: large block size FRFs suffering from noise 2 nd step: IRFs truncated by rectangular window Influence on modal parameter estimates Biased participation factors (closed-form expression describing bias exists) Alternatives Exponential window Frequency-averaging 51 copyright LMS International

52 Output-only modal parameter estimation procedure Spectrum estimation: leakage-free and Hanning window-free Weighted correlogram High-speed estimation of correlations with positive time lags Exponential window Reduces the effect of leakage Reduces the influence of the higher time lags having a larger variance Compatible with the modal model ( Hanning window with biased damping) DFT of windowed correlation sequence Practical: selection of references Left wing tip, right wing tip, tail plane 52 copyright LMS International

53 PolyMAX: EMA vs. OMA (g/n) db Sum FRF SUM Synthesized FRF SUM Phase Hz db Sum Crosspow er SUM Synthesized Crosspow er SUM Phase Hz 53 copyright LMS International

54 Pulse excitation and wing response Raw Filtered / decimated F B Time Force:ref:+Z Time WINR:952:+Z F B Time Force:ref :+Z Time WINR:952:+Z N Real Real g s s F B Spectrum Force:ref:+Z Spectrum WINR:952:+Z N db db g N db db g N Real Real g F B Spectrum Force:ref:+Z Spectrum WINR:952:+Z Freq. domain Time domain Hz Hz 54 copyright LMS International

55 Pulse excitation OMA results AutoPow er FIN:091:+Y Synthesized Crosspow er FIN:091:+Y AutoPow er WINL:952:+Z Synthesized Crosspow er WINL:952:+Z g 2 (1/s)s db Phase copyright LMS International

56 In-flight OMA mode shape (1/4) 56 copyright LMS International

57 In-flight OMA mode shape (2/4) 57 copyright LMS International

58 In-flight OMA mode shape (3/4) 58 copyright LMS International

59 In-flight OMA mode shape (4/4) 59 copyright LMS International

60 Conclusions Extensive study Standard EMA vs. Operational Modal Analysis Parameter estimation: MEFAS, PolyMAX, TimeMDOF, logarithmic decrement Windowing, SNR, sensor groups, frequency resolution, OMA Only response signals used in analysis, but artificial excitation was used Use of output cross-correlations Good performance wrt. noise Eliminate windowing problems: exponential window leads to unbiased damping estimates 60 copyright LMS International

61 Presentation outline Flutter testing: What, When and How? Required technology Industrial implementation Validation Conclusions 61 copyright LMS International

62 Conclusions OMA: some important flutter-critical modes not excited EMA: some modes mainly excited by the turbulences may not be identified Conclusion: beneficial to use artificial excitation, but data analysed with stochastic methods that also take into account the unknown excitation Airbus flight test team evaluated LMS Test.Lab using large-aircraft data We actually achieved better results using operational techniques than with classical EMA. We found more modes. The synthesis was better with higher correlation and fewer errors. And the in-flight mode shapes looked much nicer! We found that the exponential window, which allowed for cross-correlation calculations was a good de-noising tool for our in-flight data. Altitude (feet) 40,000 30,000 20,000 10,000 MACH 0.95 MACH 0.90 MACH True Air Speed (knots) 62 copyright LMS International

63 Thank you Aerospace Testing 2011, Hamburg, Germany, April Jan Debille Solutions Manager Aerospace & Defense