Flight Test Data Analysis

Transcription

1 Flight Test Data Analysis Edward Whalen University of Illinois 3-2

2 Flight Test Objective: Approach: To develop and evaluate the identification and characterization methods used in the smart icing system using flight data from clear air and in natural icing conditions. In cooperation with NASA, acquire detailed flight dynamics data on the Twin Otter with and without ice. Use data to develop and test ID and characterization methods including the effects of dynamic input, sensor noise, repeatability, uncertainty, IPS detection, etc. 3-3

3 The Twin Otter DeHavilland DH-6 High-wing, twin-engine commuter class aircraft Max Gross Weight:,000 pounds ruise Speed: 30 KIAS Range: approximately 300 nm 3-4

4 Twin Otter Instrumentation loud Instrumentation: FSSP-75: 3-47 micron particles OAP-2d-Grey: micron particles King LW (2) Rosemount Ice Detector Icing Documentation: Wing stereo photography system Video tape of right wing leading edge Hand-held 35mm camera Air Data: Rosemount OAT probe General Eastern Dew Point Hygrometer Rosemount five-hole probe: pressure altitude, airspeed, angle of attack and sideslip. 3-5

5 Twin Otter Instrumentation Aero-Performance Sensors: Linear Accelerometers (3) Angular Rate Gyros (3) Vertical Gyro Engine/Propeller Data Data Acquisition System: Science & Engineering Associates Lite 8 50 analog channels Digitized at up to 00Hz and 6 bits 3-6

6 Test Matrix ase Icing Flight Doublet Mag. Test Information. lear Air 0.25g Baseline.2 lear Air 0.0 to 0.50 Vary doublet magnitude.3 lear Air None Standard maneuvers.4 lear Air None lear air turbulence 2. Icing 0.25g Doublets during ice accretion 2.2 Icing 0.25g Doublets with selective deicing 2.3 Icing 0.25g Doublets with intercycle icing 2.4 Icing None Intercycle icing 2.5 Icing None Standard maneuvers 3-7

7 Flight Tests Phase I (February March 200) Eleven iced and 22 clear air flights were flown Standard doublets of varying magnitude used for lateral and longitudinal identification Phase II (February March 2002) Focused on collecting data in icing (2 flights) versus clear air conditions (3 flights) Experimented with different types of doublets and excitations (phase change, chirp, 2--) 3-8

8 Data Reduction Data cut into separate files containing doublet sets Filter used to remove nose boom oscillations Trim state extracted immediately before doublets using time average Doublet data passed to SIDPA stepwise regression scheme to calculate S& derivative values 3-9

9 3-20 Lift: Pitching Moment: Rolling Moment: Yawing Moment: Step-wise Regression Model ( ) ( ) ( ) e e q q e L Lq L L L α α α + + = ( ) ( ) ( ) e e q q e M Mq M M M α α α + + = ( ) ( ) ( ) p p a a lp l a l l l + + = β β β ( ) ( ) ( ) r r r r Nr N r N N N + + = β β β

10 Uncertainty Analysis SIDPA calculates ramer-rao bounds for the regression parameters accounting for the frequency content of the residuals 2σ error bars were used in the trim calculation to ensure 95% certainty Results exhibited good repeatability 3-2

11 lean Aircraft Results Doublet sets consisted of a pair of elevator doublets, an aileron doublet and a rudder doublet Multiple doublets at the same trim condition were averaged to obtain clean values 3-22

12 Lα Lα (/deg) Doublet Lα La Alpha Alpha Alpha (deg) Standard deviation for repeated doublets was, at worst, 2% of the mean. Reductions in Lα due to change in α are visible Estimate agreed well with previous tests (0.003 /deg) 3-23

13 Le L e (/deg) D oublet Standard deviation was less than 5% of the mean over all the trim conditions Error bars reflected the accuracy of the estimate Estimate agreed well with previous tests (0.08 /deg) 3-24

14 Mα Mα (/deg) Alpha (deg) Standard deviation was less than % of the mean hanges in Mα with α were visible Agreed well with previous tests ( and /deg) Mα Ma Alpha Alpha Doublet 3-25

15 Me M e (/deg) Me Mde Alpha Alpha Alpha (deg) Standard deviation was less than 2% of the mean Variations in the estimate with α are visible, but within error bars Estimates agreed well with previous tests (-0.03 and /deg) Doublet 3-26

16 Summary of Results* Derivative TM 4099 (deg - ) AIAA (deg - ) Illinois (deg - ) Lα Lq Le Mα Me Mq nβ nr nr * Estimates are for α=0 3-27

17 Flight 00302f2.30E-0 Vertical Tail and Struts Deiced.20E-0 Baseline lean Value Horizontal Tail Deiced Wings Deiced Intercycle Ice Residual Ice.0E-0 La (/deg).00e E-02 lean All Iced Residual Ice Residual Ice 8.00E-02 Out of loud In loud All Iced All Iced 7.00E E-02 7:6 7:3 7:45 8:00 8:4 8:28 8:43 8:57 Time 3-28

18 Turbulence Effects Results of parameter identification in turbulence compare poorly with established parameters Icing is usually accompanied by turbulence There is a direct relationship between estimated measurement scaling factors/biases and poor parameter estimates The turbulence effect introduces a bias error to the parameter estimates Further investigation is required to achieve good identification in turbulence 3-29

19 Lα Estimate in Turbulence 0.0 Baseline lean 0.05 La (/deg) α trim = Time (s) 3-30

20 Estimated Scaling Factors/Biases 0.2 orrelation of La with Alpha Scaling Factor 0. Lα (/deg) Alpha Scaling Factor 3-3

21 Summary A joint University of Illinois - NASA Glenn flight test program was conducted to assist in the development of the Smart Icing System The objective of this research was to develop and evaluate the identification and characterization methods used in the smart icing system using flight data The flight data were acquired in clear air and in icing conditions Stability and control derivatives were estimated through the use of SIDPA The icing and deicing of the aircraft was observed through changes in aircraft trim and stability and control derivatives 3-32

22 onclusions Stability and ontrol derivatives have been identified for the clean aircraft in smooth air with excellent results Turbulence significantly effects the parameter identification process Estimation results in icing conditions with atmospheric turbulence are, at this time, unreliable using SIDPA methods 3-33

23 Work in Progress More extensive flight data analysis planned Phase II flight test data analysis Other stability and control derivatives Investigating changes in aircraft trim as an indicator of ice accretion 3-34