49th AIAA Aerospace Sciences Meeting AIAA 2011-875 4-7 January 2011, Orlando, FL Wind Tunnel Testing Airfoils at Low Reynolds Numbers Michael S. Selig,∗ Robert W. Deters,† and Gregory A. Williamson‡ University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA This paper describes the wind tunnel testing methodology that has been applied to testing over 200 airfoils at low Reynolds numbers (40,000 to 500,000). The ex- periments were performed in the 2.8 4.0 ft (0.853 1.219 m) low-turbulence wind × × tunnel in the Subsonic Aerodynamics Research Laboratory at the University of Illi- nois at Urbana-Champaign (UIUC). The test apparatus, methodology, and data reduction techniques are described in detail, and the measurements are validated against benchmark data. New results on the AG455ct airfoil with a large 30%-chord flap, deflected over a wide range, are presented. The results show a dramatic in- crease in drag with higher flap deflections, and the flap efficiency reduces with large deflections up to 40 deg. Also, tests on a flat-plate airfoil with leading edge serration geometries were conducted to explore the effects on stall characteristics. The results support the conclusions of other researchers that leading edge serrations (protuber- ances like those found on the fins/flippers of some aquatic animals) lead to higher lift and softer stall. The results suggest that these characteristics are accompanied by lower drag in the stall and post-stall range. I. Introduction Airfoil performance at low Reynolds numbers impacts the performance of a wide range of systems. The expanding role of unmanned aerial vehicles (UAVs) into unmanned aircraft systems (UAS) in military use1,2 hasledtogrowinginterestinsubsoniclowReynoldsnumberaerodynamics.3–5 LowReynoldsnumber aerodynamics of airfoils also apply to a host of other applications such as wind turbines,6–8 motorsports, high altitude aircraft and propellers, natural flyers,9 and subscale testing of many full scale systems. AccuratemeasurementsoflowReynoldsnumberairfoilperformanceiskeytounderstandingandimprov- ingtheefficiencyoflowReynoldsnumbersystems. Mostaerodynamicperformancemeasurementtechniques for airfoils rely on using balance systems or pressure systems, or a combination of both.10–12 The approach described in this paper uses a force balance approach to obtain lift and moment data and the wake rake methodtoobtaindrag. SectionsIIandIIIofthispaperdescribethisexperimentalapproachandvalidation. SectionIVpresentsfirst,dataonanairfoilwithlargeflapdeflections,andsecond,dataonaflat-plateairfoil as compared with one having a range of leading-edge serration geometries. The paper ends with conclusions that can be drawn from this research. II. Wind Tunnel Facility and Measurement Techniques ThissectionpresentsdetaileddescriptionsoftheUIUClow-turbulencesubsonicwindtunnelfacility, test section flow quality, lift, drag and moment measurement techniques, data acquisition equipment, and data reduction procedures that have been documented in Refs. 6,13–16. ∗AssociateProfessor,DepartmentofAerospaceEngineering,104S.WrightSt.,SeniorMemberAIAA. http://www.ae.illinois.edu/m-selig †GraduateStudent,DepartmentofAerospaceEngineering,104S.WrightSt.,StudentMemberAIAA. ‡GraduateStudent,DepartmentofAerospaceEngineering,104S.WrightSt.,StudentMemberAIAA. 1of32 AmericanInstituteofAeronauticsandAstronautics Copyright © 2011 by the authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission. Figure 1. UIUC low-speed subsonic wind tunnel. Figure 2. Photograph of wind-tunnel test section. A. Experimental Facility and Flow Quality Measurements The low Reynolds number airfoil performance measurements described here were conducted in the UIUC low-turbulence subsonic wind tunnel shown in Fig. 1. The wind tunnel is an open-return type with a 7.5:1 contraction ratio. The rectangular test section is nominally 2.8 4.0 ft (0.853 1.219 m) in cross section × × and 8-ft (2.438-m) long. Over the length of the test section, the width increases by approximately 0.5 in. (1.27 cm) to account for boundary-layer growth along the tunnel sidewalls. Test-section speeds are variable upto160mph(71.53m/s)viaa125-hp(93.25-kW)ACmotorconnectedtoafive-bladedfan. ForaReynolds number of 500,000 based on an airfoil chord of 1 ft (0.305 m), the resulting nominal test-section speed is 80 ft/sec (24.38 m/s). Photographs of the test section and fan are presented in Figs. 2 and 3 Since low Reynolds number airfoil performance is highly dependent on the behavior of the laminar boundarylayer,lowturbulencelevelswithinthewind-tunneltestsectionarenecessarytoensurethatlaminar flowdoesnotprematurelytransitiontoturbulentflow. Inordertoensuregoodflowqualityinthetestsection, the wind-tunnel settling chamber contains a 4-in. (10.16-cm) thick honeycomb in addition to four anti- turbulencescreens,whichcanbepartiallyremovedforcleaning. Theturbulenceintensitywasmeasuredand previouslyreportedtobelessthan0.1%,17 whichissufficientforlowReynoldsnumberairfoilmeasurements. Subsequent detailed measurements are show later in this section. TheexperimentalsetupisdepictedinFig.4. Forthecurrenttests, the12-in.(0.305-m)chord, 335/8-in. (0.855-m)longairfoilmodelsweremountedhorizontallybetweentwo3/8-in.(0.953-cm)thick,6-ft(1.829-m) f R longPlexiglas splitterplates toisolatetheendsofthemodelfromthetunnelside-wall boundarylayers and the support hardware. For clarity, the Plexiglas splitter plates and the traverse enclosure box are not shown in Fig. 4. Gaps between the model and Plexiglas were nominally 0.05 in. (1.27 mm). The left-hand side of 2of32 AmericanInstituteofAeronauticsandAstronautics Figure 3. Photograph of wind-tunnel exit fan. f R Figure 4. Experimental setup (Plexiglas splitter plates and traverse enclosure box not shown for clarity). the model was free to pivot (far side of Fig. 4). At this location, the angle of attack was measured using a precision potentiometer. The right-hand side of the airfoil model was connected to the lift carriage through two steel wing rods that passed through the wing-rod fixture and were anchored to the model through two set screws. At this side, the airfoil model was free to move vertically on a precision ground shaft, but not free to rotate. The lift carriage was linked to a custom load beam, as described later. Linear and spherical ball bearings within the lift carriage helped to minimize any frictional effects. The two-axis traverse can be seen in Fig. 4, positioned above the wind-tunnel test section. Not shown is the pressure-sealed box that encloses the traverse system. The traverse was manufactured by LinTech and consists of horizontal and vertical screw-type linear positioning rails that operate in combination with two computer-controlled stepper motors. The rails are equipped with linear optical encoders that supply a feedback signal to the motor controller allowing the traverse to operate with virtually no error. Attached to the traverse is a 3-ft (0.914-m) long boom that extends down into the wind tunnel test section to support the eight side-by-side wake pitot probes [spaced 1.5 in. (3.81-cm) apart in the spanwise direction]. 3of32 AmericanInstituteofAeronauticsandAstronautics Average Turbulence Intensity (Empty Test Section) 0.25 %) DC Coupled ensity ( 0.2 HHHPPPFFF === 031.H01HzHzz nt e I nc0.15 e ul b Tur 0.1 n a e M 0.05 100,000 200,000 300,000 400,000 500,000 Reynolds Number Average Turbulence Intensity (Test Apparatus Installed) 0.25 %) DC Coupled ensity ( 0.2 HHHPPPFFF === 031.H01HzHzz nt e I nc0.15 e ul b Tur 0.1 n a e M 0.05 100,000 200,000 300,000 400,000 500,000 Reynolds Number Figure 5. Turbulence intensity at tunnel centerline, empty test section and with rig in place. 1. Turbulence Intensity The turbulence intensity was previously documented;17 however, those measurements were with an empty tunnel test section. It is of interest to examine what effect, if any, the splitter plates and other test section components have on the turbulence intensity. The turbulence intensity was measured using hot-wire anemometry. In particular, the hot-wire system was a TSI Incorporated IFA 100 anemometer in conjunction with a TSI Model 1210-T1.5 hot-wire probe. The probe makes use of a 1.5 micron platinum-coated tungsten wire. The probe was mounted in the tunnel end-floworientationwiththewireperpendiculartothetunnelfloorinordertomeasuretheaxialturbulence intensity. A PC equipped with a data acquisition card was used to log the signal from the anemometer. A Hewlett-Packard HP 35665A Dynamic Signal Analyzer, which performed an FFT (Fast Fourier Transform) analysis, was employed to allow the turbulence spectrum to be monitored over a broad range of frequencies. The hot-wire probe was calibrated in the UIUC low-speed subsonic wind tunnel. The tunnel speed was setusingstaticpressureprobesinsidethetunnel,andthecorresponding(average)outputoftheanemometer was recorded. From these data, a curve fit was generated that was used to measure the fluctuating velocity with the hot-wire probe. Corrections were made to the signal to account for changes in temperature and density between the time the probe was calibrated and the time the measurements were made. A more detailed description of the methods used is found in Ref. 18. Theturbulenceintensitywascalculatedfromdatausingatotalof50,000sampleswithasamplefrequency of 10,000 Hz and is shown in Fig. 5 for the case in which the tunnel was empty and that in which the full measurement apparatus was installed. As compared with the baseline empty tunnel, turbulence levels observed with the test apparatus installed are relatively unchanged at Re= 100,000 but increase at higher Reynolds numbers. These effects all but disappear when the high-pass filter is set above 3 Hz. The main effect of the test rig appears to be added velocity fluctuations in the very low frequency range. Figure 6 shows the power spectra between 0 and 100 Hz for the Re = 350,000 case both for the empty tunnel and for that with the test apparatus installed. Measurements were taken over a wide range of frequencies (up to 6,400 Hz), but in all cases the interesting features ranged between 0 and 100 Hz. Apart from the peaks in power at 56 and 79 Hz, the turbulent power spectrum is similar in magnitude for both configurations. It is only in the range from 0 to 25 Hz that there is a noticeable offset between the empty-tunnel test section and the installed-apparatus section. In general, these turbulence levels are considered to be sufficiently low for taking low Reynolds number airfoil measurements. 4of32 AmericanInstituteofAeronauticsandAstronautics Turbulence Power Spectrum (Re/l = 350,000/ft) −20 Tunnel Empty −30 Test Rig Installed B)−40 d er (−50 w o P−60 −70 −80 0 10 20 30 40 50 60 70 80 90 100 Frequency (Hz) Figure 6. Power spectrum comparison between empty tunnel and installed test apparatus cases for Re = 350,000. 2. Freestream Velocity The variation of velocity in the test section of the UIUC low-speed subsonic wind tunnel was obtained by comparing the dynamic pressure (directly related to velocity) at a pitot-static probe mounted near the entrance of the splitter plates with that measured bya downstream probe. Theupstream probewas located atthecenterlineofthetunnelinthespanwisedirection(X =0),0.97ft(0.296m)belowthecenterlineofthe tunnel in the vertical direction [Y = 11.66 in. (0.296 m)], and 1.323 ft (0.403 m) upstream of the quarter- − chord location of the airfoil model when mounted in the test section. The downstream probe was traversed in the X-Y plane perpendicular to the freestream and coincident with the quarter chord. Measurements were made both with the test section empty and with the test apparatus installed. Themeasurementplaneextendedfrom5.5in.(13.97cm)abovethetunnelcenterlineto14.5in.(36.83cm) below in the vertical direction Y, and from 10.5 in. (26.67 cm) to the left of the tunnel centerline to 10.5 in. (26.67 cm) to the right in the horizontal direction X. A grid spacing of 1 in. (2.54 cm) was used for the measurements, resulting in a total of 462 measurement points for each case tested. Three differential-pressure transducers were used for the measurements. One transducer measured the upstream dynamic pressure Q by measuring the pressure difference across the total pressure and static u pressureportsofapitot-staticprobe. Asecondpressuretransducerwasconfiguredtomeasurethedifference between the upstream and downstream total pressure ∆P . A third transducer was configured to measure 0 the difference between the upstream and downstream static pressure ∆p. The change in dynamic pressure ∆Q is just ∆P ∆p. Thus, the local dynamic pressure at each point is therefore 0 − Q=Q +∆Q=Q +∆P ∆p (2.1) u u 0 − For each Reynolds number tested, the tunnel speed was set using the upstream probe as the reference. Differences in temperature and ambient pressure were accounted for. The percent difference at each point was calculated according to Q Q u ∆Q(%)= − 100% (2.2) Q × u Figure 7 shows contours of ∆Q for various Reynolds numbers plotted against its X and Y location for the case in which the wind tunnel was empty. For comparison, Fig. 8 shows ∆Q plotted against its X and Y location with the test rig installed. From Figs. 7 and 8, several observations can be made. First, for the empty test section case, there is a slight decrease in the test section flow speed at the location of the model relative to the upstream probe. When the test rig is installed, there is instead an increase in the flow speed. It is likely that the velocity measured at the location of the model is higher than the upstream velocity because of the growth of the boundary layer along the splitter plates, ceiling and floor as well as the blockage that occurs between the splitter plates and the tunnel sidewalls. This percentage increase in the flow speed grows larger as the Reynolds number is reduced, which is consistent with the thicker wall boundary layers at lower Reynolds numbers. As discussed later, this rise in velocity is accounted for in the airfoil-performance data-reduction procedure. Second, over the region where the model is located, the net change in flow speed is observed to 5of32 AmericanInstituteofAeronauticsandAstronautics ∆Q (%), Re/l = 100,000/ft ∆Q (%), Re/l = 200,000/ft Y (in) −0550−000.2 − 0.−40.4−−00.2.40 −000.0020 Y (in) −0550.40.200.200.4 −0.2−00−0.02.−20.20 0.0200.040.4 0 .2 −10 0.40.20 0.4 −000..022.4 0.02.6 −10 0.40.6 0.60 0.2 0.40.8 −10 −5 0 5 10 −10 −5 0 5 10 X (in) X (in) ∆Q (%), Re/l = 350,000/ft ∆Q (%), Re/l = 500,000/ft 5 0.40.4 0.2 0.40 0.4 5 0.20.2 0 00.20.2 Y (in)−−10500.20.4000.20−0000.020.40 0− 0.2.2 0.20.4000..46 Y (in)−−10500.20.4−0000..202 −0.2−0.2−0.−60.−4000.−2−00.−2.04.0.220 0.6 0.6 0.8 1 00..860.60 0.6 0.4 0.6 −10 −5 0 5 10 −10 −5 0 5 10 X (in) X (in) Figure 7. Dynamic pressure variation across the test section when empty. ∆Q (%), Re/l = 100,000/ft ∆Q (%), Re/l = 200,000/ft 52 22 1.8 1.8 2 5 1.18.8 1.61.8 1.8 Y (in) −05221.18.81.1.66 1.6 1.4111.6.6.411..68 22.2 Y (in) −05 1.81.1.61.168.6 1.4 1.4 11..8822 −102 2 2.4 −10 1.8 1.6 1220.1.82 2.26.8 2 2 0122.2 22.42.6 −10 −5 0 5 10 −10 −5 0 5 10 X (in) X (in) ∆Q (%), Re/l = 350,000/ft ∆Q (%), Re/l = 500,000/ft 5 5 8 0.8 0.8 Y (in) −05 1.4 11.21.16..44 1.2 1.1.241.611.21.21.14.611.6.16.82 Y (in) −05 0.40.060..0.42040..600.4.04.6 00.6.04.04.80.8 −10 1.4 1.86 −10 0.6 00..68 11.2 1.8 1. 102 2 2 2.2 0.81 11 10 11.41.4 −10 −5 0 5 10 −10 −5 0 5 10 X (in) X (in) Figure 8. Dynamic pressure variation across the test section with the experimental rig installed. be relatively small. For instance, Fig. 8 shows that at Re/l = 200,000/ft (656,168/m), the increase in the flow speed varies from approximately 1.4% to 1.8%, which is a relative difference of 0.2% in the working ± 6of32 AmericanInstituteofAeronauticsandAstronautics Figure 9. Illustration of the seven-hole probe used for flow angle measurements. range of the test section. As stated in Ref. 19, it is desirable for the variation in dynamic pressure in the working range of the test section to be less than 0.5% from the mean, i.e., 0.5%. The results show that ± theflowiswellwithinthe“ruleofthumb.” Athirdobservationistheexistenceofaslightasymmetryinthe flow, noticeable mainly in the +X: Y quadrant (bottom right corner in Figs. 7 and 8). The asymmetry is − presentwiththetunnelemptyandwiththetestriginplace; hence, itisunrelatedtothetestrig. Moreover, the lines of constant Q are parallel to the tunnel floor at X =0 (centerline), so the effect is negligible with respect to the performance-measurement quantities in the center region of the test section. 3. Freestream Flow Angularity Just as it is important to have uniform flow velocity in the wind-tunnel test section, it is equally important tohavetheflowparalleltotheaxialdirection.19 Forthemostpart,pitot-staticprobesareinsensitivetoflow anglesintherangeof 12deg,soalargeflowangleisrequiredtointroduceanerrorinthedynamicpressure ± measurements. Similarly, large flow angles are required to introduce errors into total head measurements. Apart from pressure measurements, a small change in pitch angle contributes to a change in the effective angleofattackoftheairfoilmodel,andtherebysuchanerrorcanskewtheliftanddragmeasurementswhen they are plotted versus the angle of attack. The flow angularity in the test section of the UIUC low-speed subsonic wind tunnel was measured using anAeroprobeCorporationModelS7TC317seven-holeprobeasshowninFig.9. Theprobehasatotal-head port located at the center, and six chamfered ports were equally spaced circumferentially around the center. Each port of the seven-hole probe was connected to the high-pressure side of an MKS Model 220CD 1-mm Hg pressure transducer. The reference side of each pressure transducer was left open to ambient pressure. Theprobewasmountedinthewindtunnelonaspecialtwo-beamstingattachedtothecomputer-controlled LinTech traverse. The flow measurements were all taken with the test rig installed in the wind-tunnel test section, without the model. A more detailed description of the use of the seven-hole probe is found in Ref. 20. The seven-hole probe was traversed in a plane perpendicular to the freestream flow over the range from X = 6.5 in. (16.51 cm) to Y = 10 in. (25.4 cm). The traverse was not extended to the edges of the test ± ± section because of equipment limitations. Traversing this central core was acceptable because one would expect to find the largest flow angle variation in the center of the test section rather than along the walls, where at a minimum the flow is parallel to the wall (yaw or pitch is thereby zero). A grid spacing of 1 in. (2.54 cm) was used, resulting in a grid of 252 sample locations for each case tested. The seven-hole probe tip was located approximately 1.5 chord lengths behind the quarter chord of the airfoil model. To set the tunnel speed, one pitot-static probe was located at X =0, Y = 11.66 in. (29.62 cm). For redundancy, an − additional probe was located at X = 5 in. (12.7 cm), Y = 11.66 in. (29.62 cm). Both pitot-static probes − were mounted at the same streamwise location, 1.323 ft (3.36 cm) upstream of the location of the quarter chord of the airfoil model. Calibration curves supplied by the probe manufacturer were used to determine the flow angle at each location. Three such curves were provided, each of which covers a particular angle of attack range, namely, 0 to 5 deg, 5 to 10 deg, or 10 to 15 deg. Because the flow angles measured never exceeded 1 deg, only the first curve was needed. Figures 10–12 show the measured flow angle at each point plotted against its X and Y coordinate. 7of32 AmericanInstituteofAeronauticsandAstronautics Pitch Angle (deg), Re/l = 100,000/ft Pitch Angle (deg), Re/l = 200,000/ft 10 10 50.300.02..011.10 0.10−000.1..10.200.5.34 0.3 05.2 0−−00..12−0.2−00.0.10.2300..340.30.40 Y (in)−−10050000..0.121.0.1200..3020.3.00.25.0.064.03.20000.00..0.0246..3.5450.04.00610.00..071000..8010...52.3.9.234 Y (in)−−100500000.−1.001..110.10000.20−−.00400.0.3.01...21320.002000..0..5660..710.43.510.10 −5 0 5 −5 0 5 X (in) X (in) Pitch Angle (deg), Re/l = 350,000/ft Pitch Angle (deg), Re/l = 500,000/ft 10 10 500.0.03.012.10−0−0.−030..2−10.200..32000..3.104 50.0.300.1.210−00.−.210−.100.20.10.20.03.01.40.03 Y (in)−−100500.1−00.100.100.01.10.010.−2000.2.0.13.100.0200..0.501..34600 Y (in)−−100500.010.10.100.10−000.0..3120..21 0.500.0.00460.30.2.1 −5 0 5 −5 0 5 X (in) X (in) Figure 10. Pitch angle variation across the test section with the experimental rig installed. Yaw Angle (deg), Re/l = 100,000/ft Yaw Angle (deg), Re/l = 200,000/ft Y (in)−−11000005500..020.4.012.0001..0.14300.10.2000000..2010.20−.−30−0−000...04.2−3−1.0010.1.00−.3.43.030−..200005..21.020 Y (in)−−11−000055−.−2000−.−2−.−00.30.30−−2..002−1.−.2−03−00.30..40.410−−−.−0100−0−.5−−.0−..7040600..06..3..3−07.890−2−0−.−1.10020−..−.4205−0.00−6..203.2 −5 0 5 −5 0 5 X (in) X (in) Yaw Angle (deg), Re/l = 350,000/ft Yaw Angle (deg), Re/l = 500,000/ft 1−050.−2−00.−3.03.20 0.10.2 −0−0−0.02..13 1−05−0−.020.2.3 0−0.2 0.−20.001.2−−0.01.2 Y (in)−−10005−−.1−0000..0.−231−00.02.2 −−−00.−04−.−.0−5−03−0.0.00.6.38..972−−−−0−1.30−000..26..54 Y (in)−−1005−−−0000...−1120.2−−00.2.3−−00−.−.3040−.−0−.5.06.70−−8−.0109.3−−.−020.02..42 −5 0 5 −5 0 5 X (in) X (in) Figure 11. Yaw angle variation across the test section with the experimental rig installed. The contour plots of flow angle in the test section show that pitch and yaw angles are smallest at Re= 500,000, becoming more pronounced at lower Reynolds numbers. Pitch angle, the more important angle for airfoil testing, is generally between 0 and 0.2 deg ( 0.1 deg) across the working region of the test section ± wheretheairfoilmodelislocated. AccordingtoRef. 19,aflowanglevariationof 0.2degisacceptable,but ± 0.1 deg or better is the preferred. The current measurements meet this latter desired level of flow quality. ± 8of32 AmericanInstituteofAeronauticsandAstronautics Combined Angle (deg), Re/l = 100,000/ft Combined Angle (deg), Re/l = 200,000/ft 10500..320.20.010.3.2000..5.440.60.05.3 1050.40.30.30000..2.1.430.50.20.0.024.30.4 Y (in)−−100005..5400.100.20..012..23000...5340.600.00..03064...54500..62000..0.91087.050.4..03.23 Y (in)−−1000050.2..2030.40.010..4.300.230.5.400.0.56.6000.0..0.84.97180.00310..056.7.0.230.2 −5 0 5 −5 0 5 X (in) X (in) Combined Angle (deg), Re/l = 350,000/ft Combined Angle (deg), Re/l = 500,000/ft Y (in) 10.02005.40.200.3.2 00.30..0302.3.010.00..442.20.200.300.03...432 Y (in) 10050.0400.0.2..1230.20.02.300.01..3200.04.0.013..3200..2030.02.0..3402.2 −−1005.200.1.20.30.400.005...866 00..197001..0059..23 −−1050.20.10.20.030..4500.09.8.14000..0.607.8.52 −5 0 5 −5 0 5 X (in) X (in) Figure 12. Combined pitch and yaw angle across the test section with the experimental rig installed. It is worth noting that downstream of the pitot-static probes placed near the floor, relatively large flow angles were recorded locally. Flow angle perturbations due to probes near the wall, such as the dynamic pressure probes used to determine the flow speed during the airfoil tests, are of no concern because the corresponding flow is well below the working region of the test section where the model is located. B. Airfoil Models In order to determine the accuracy of the wind-tunnel models, most all models that have been tested are digitized using a coordinate measuring machine (CMM) to determine the actual airfoil shape at the midsection of the model. Approximately 80 coordinates points are typically taken around the airfoil. The spacing is more or less proportional to the local curvature. Near the leading and trailing edges the spacing usedisrelativelysmallandthenlargerovertheairfoilmidsection. Measurementssuchasthesecanbefound in Refs.13–16. C. Performance Data Measurement Techniques As an overview to this section, the data acquisition process, which was largely choreographed by computer control, isbrieflydescribed. Twotypesofrunswereperformed: “liftruns”and“dragruns.” Fortheformer, only lift and moment vs angle of attack data were taken for a fixed Reynolds number; whereas for the latter, drag data were included. Lift runs were extended to high angle of attack and sometimes into stall; drag runs were set to take data nominally over the low drag range, which for this work was defined as C d approximately less than 0.05. For lift runs, data were taken for both increasing and decreasing angles of attack to document any aerodynamic hysteresis present. For drag runs, data were taken only for increasing angles of attack. For each angle of attack, the tunnel speed was checked and, if necessary, adjustments were made to maintain a fixed Reynolds number. In general, acquiring data during a lift run was a relatively quick process compared with acquisition during a drag run. Since no wake measurements were taken during a lift run, it was possible, for instance, to cover a full angle of attack range from 10 to 20 deg and back in − 1 deg increments in approximately 20 min. For a drag run, however, the time could range from 1 to 6 hr, depending principally on the width of the wake and desired angle of attack range. Details pertaining to the lift, drag, and moment measurements are described in more detail in the following sections. 9of32 AmericanInstituteofAeronauticsandAstronautics Figure 13. Lift beam balance assembly as viewed from the working side of the test section. 1. Lift Force Measurement Figure 13 depicts a schematic of the lift beam balance that was integrated into the test section and oriented according to Fig. 4. The lift apparatus consisted primarily of a fulcrum supported beam restrained in rotation by a strain-gauge load cell. The lift load acting through a pushrod was applied to the beam and transferred to the load cell via a lever arrangement, as shown. For correct operation of the lift balance, the load cell was required to be in tension. To ensure this outcome, spring tension on the lift side of the beam and counterweight on the opposite end were used, and the amount of each depended on the model weight as well as the range of loads expected. Moreover, the mix of spring tension and counterweight changed in going from a calibration to a performance data run. DependingontheReynoldsnumberofthetestandexpectedrangeinliftcoefficient,oneofthreepossible load cells (Interface Inc. Models SM-10, SM-25, and SM-50) was inserted into one of nine possible load cell attachment holes, allowing for a variation in the operational range of the lift beam balance. This approach offered the ability to measure the small forces present at the lower Reynolds numbers range (Re 60,000) ≈ while retaining the capability to handle the larger lift forces occurring at the higher Reynolds numbers (Re 500,000). The resulting lift measurements were repeatable to within 0.01% of the rated output of the ≈ particular load cell used. Calibrations were performed frequently to minimize the effects of drift. 2. Drag Force Measurement WhiletheliftforceonairfoilsatlowReynoldsnumberscanbeobtainedwithacceptableaccuracythrougha lift balance, drag force is often considerably much lower than the lift. As a result, profile drag is often best obtained by the momentum method instead of a force balance. For the current tests, the profile drag was determined through the method developed by Jones21 (taken from Schlichting22). Afterapplicationofthetwo-dimensionalmomentumandcontinuityequationstoacontrolvolumeshown in Fig. 14, the drag force per unit span can be calculated from ∞ d=ρ u1(V∞ u1)dy (2.3) Z−∞ − Assumingthatthelocationofthemeasurementsisfarenoughbehindtheairfoilsothatthestaticpressurehas returned to upstream tunnel static pressure (i.e., Ps,1 = Ps,∞ = Ps) and that the downstream flow outside the airfoil wake proceeds without losses (i.e., the total pressure remains constant along every streamline), the total pressure relationships from Bernoulli’s equation are 1 P + ρu 2 =P (2.4) s 1 0,1 2 1 Ps+ ρV∞2 =P0,∞ (2.5) 2 Applying the above relationships to Eq. (2.3) and simplifying yields ∞ d=2 P0,1 Ps q∞ Ps (P0,1 Ps) dy (2.6) Z−∞(cid:8)p − p − − − (cid:9) 10of32 AmericanInstituteofAeronauticsandAstronautics