Document
Background Oriented Schlieren (BOS) of a
Supersonic Aircraft in Flight
James T. Heineck, NASA Ames Daniel W. Banks, NASA Armstrong Edward T. Schairer, NASA Ames Edward A. Haering, Jr., NASA Armstrong Paul S. Bean, NASA Armstrong AIAA Flight Test Conference June 13 - 17, 2016 Washington, DC
Introduction
• Schlieren imaging for aerodynamics research was limited to ground test facilities • Weinstein introduced the first reliable method for flight test in 1994 • Retroreflective Background Oriented Schlieren was demonstrated to work for full - scale aircraft in flight in 2012 Weinstein’s sun - edge streak camera image of a T - 38 (DLR Goettingen) • NASA’s Commercial Supersonic Technology Program developed the next generation airborne schlieren methods for supersonic flight testing. First AirBOS flight in April, 2011 was successful, but restricted.
RBOS of a BO - 105 in slow forward flight Raffel, DLR Goettingen
The BOS Method
Speckle Background Wind tunnel reference, data image and result of an abort motor tower at M=1.3
The AirBOS Method
AOX - Experimental Aero - Physics Branch
AirBOS Implementation
• Fly in the Supersonic Corridor near Edwards AFB • Characterize the Mojave Desert flora in the Supersonic Corridor: Creosote bushes with scattered Joshua trees • Bushes average 10 feet (3.1 m) diameter; too few trees to be of concern • Dark green against light gray soil; red filter enhances contrast
Observer plane
NASA Beechcraft B - 200 Super King Air
• Fly at 30,000 ft MSL ( Highest practical altitude) • Low stall speed – 90 knots (75 knots at full - flap) • Already equipped with high - quality nadir port window • GPS navigation
Target plane
Air Force T - 38, operated by the Test Pilot School at Edwards AFB Supersonic flight achieved by full acceleration during a shallow dive, leveling for the flyby
Imaging system design
• Calculate the proper lens focal length to optimize speckle size • Phantom V641, with 2650 x 1600 pixel and 10 – micron pitch • Speckle distribution should be 2 - 5 pixels • Spreadsheet calculates pixel resolution and field of view on ground and at target location Lens Camera Half Angles Altitudes FOV at Target Aircraft FOV at Ground Observer Target Lens fl ccd nx ccd ny pixel size X Y a/c a/c Ratio Δ xfov Δ yfov Resolution Δ xfov Δ yfov Resolution (mm) (pixels) (pixels) (μ m) ( Deg ) (Deg) (Ft) (Ft) (Ft) (Ft) (pixels/ft) (Ft) (Ft) (pixel / ft) 105 640 512 25 4.36 3.49 27000 13500 0.5 2057.14 1645.71 0.31 4114.29 3291.43 0.16 180 2560 1600 10 4.07 2.54 30000 26000 0.75 568.89 355.56 4.50 4266.67 2666.67 0.60 180 2560 1600 10 4.07 2.54 30000 28000 0.9 284.44 177.78 9.00 4266.67 2666.67 0.60 Sample of table for two cameras and target aircraft separation distances
AirBOS Implementation
• Survey the Black Mountain SS Flight Corridor at 30,000 ft altitude • Photographically survey large area, find consistent flora • Test for cross correlation performance • Design flight pattern to hit the “sweet spot” where the acceleration can be achieved, but turn around is within the corridor Black Mountain SSC Area Survey area Edwards AFB 9 Google Earth view of area of Supersonic Corridor and Edwards AFB
Determining the Sweet - spot
Evaluate two successive frames from the reconnaissance flight using cross correlation
Determining the Sweet - spot
Assure SNR of 5 or higher in the cross - correlation product using the anticipated window size
Flight Plan
Cameras and Layout
Phantom V641 monochrome, 2560 x 1600 pixels, 10 micron pitch, 180 mm lens - 8 GB of internal memory, ~ 2 seconds of record time @ 1000 fps - #25 Red filter, enhance contrast of bushes against the bright soil Two cameras: redundancy and potential for stereo and multi - stream referencing Legacy camera for 2011 work: Goodrich SUI SU640 - SDWHVis - 1.7RT InGaS - 640 x 512 pixel sensor, 25 micrometer pixel pitch, and fitted with a 105 mm lens - Used mainly as real - time spotting camera Two cameras, mounted vertically Schematic of cabin layout
Data Processing
1. Reference - to - data registration: First - order projective transform - Aligns the displaced backgrounds caused moving observer - Corrects perspective distortion caused by pitch and roll during acquisition ′ ′ 𝑎 𝑥 + 𝑎 𝑦 + 𝑎 1 2 3 𝑥 = ′ 𝑐 𝑥 + 𝑐 𝑦′ + 1 1 2 ′ ′ 𝑏 𝑥 + 𝑏 𝑦 + 𝑏 1 2 3 𝑦 = ′ 𝑐 𝑥 + 𝑐 1 2𝑦′ 1 + - Four points at corner of images are chosen, large - window CC performed - Cross correlation between the two images yields ∆ 𝑥 and ∆ 𝑦 at each ′ ′ location, 𝑥 and 𝑦 are solved to then calculate the eight coefficients ′ 𝑥 = 𝑥 + ∆ 𝑥 ′ 𝑦 = 𝑦 + ∆ 𝑦 2. Image c ross correlation at defined grid nodes yields Dx and Dy due to density gradient shift
AirBOS Results
15 Sample movie of raw imagery, two frames skipped for brevity
AirBOS Results
T - 38 at 45 deg. r oll, M=1.05, s ingle frame, cross correlated to reference image
AirBOS Results
Results from 200 sequences aligned and averaged
Optical Flow: A new refinement
Movie of Optical flow - processed sequence, two frames skipped for brevity
AirBOS Results
Cross correlation average vs. Optical flow average
AirBOS Results
20 T - 38 at Mach=1.05, wings - level optical flow with 100 sequences aligned and averaged
Summary
Background Oriented Schlieren has been successfully adapted to full - scale supersonic flight The planning and system design permit predictable results Image processing has made the data of high quality – better than wind tunnel data Technique permits testing of maneuvers, monitoring tip vortex trajectories, and subsonic wakes
Acknowledgements
Nils Larson, NASA Armstrong Chief Pilot and the flight crews Maj. Jonathan Orso and Maj. Jeremy Vanderhal , Instructor Pilots from Air Force Test Pilot School Brett Pauer, for flight test management Peter Coen for years of support for this research Neal Smith and Mike Hill for development of the Optical Flow algorithms