https://ntrs.nasa.gov/search.jsp?R=19890001471 2019-04-11T21:53:57+00:00Z NASA Technical Memorandum 4062 Experimental Results for the Eppler 387 Airfoil at Low Reynolds Numbers in the Langley Low-Turbulence Pressure Tunnel Robert J. McGhee, Betty S. Walker, and Betty F. Millard Langley Research Center Hampton, Virginia National Aeronautics and Space Administration Scientific and Technical Information Division 1988 Contents Abstract ................................... Introduction ................................. 1 Symbols ................................... 1 Model, Apparatus, and Procedure ........................ 2 Instrumentation ............................... Tests and Methods .............................. Presentation of Results ............................ Discussion of Results ............................. Concluding Remarks ............................. Tables .................................... Figures ................................... 12 Appendix A Uncertainty Analysis ....................... 86 Appendix B Section Characteristics ...................... 90 Appendix C---Spanwise Drag Coefficients ................... 106 Appendix D--Chordwise Pressure Coefficients ................. 113 Appendix E--Spanwise Pressure Coefficients .................. 204 References ................................. 228 ¢)RECEDING PAGE BLANK NOT FILMED ,°° III Abstract This report presents only the pressure model re- sults obtained from this research program. Tests Experimental results have been obtained for an on a pressure model of the Eppler 387 airfoil have Eppler 387 airfoil in the Langley Low-Turbulence been conducted over a Mach number range from 0.03 Pressure Tunnel (LTPT). The tests were conducted to 0.13 and a chord Reynolds number range from over a Mach number range from 0.03 to 0.13 and a 60000 to 460000. Lift and pitching-moment data chord Reynolds number range from 60 000 to 460 000. were obtained from airfoil surface pressure measure- Lift and pitching-moment data were obtained from ments, and drag data were obtained from wake sur- airfoil surface pressure measurements, and drag data veys. Oil flow visualization was used to determine were obtained from wake surveys. Oil flow visualiza- laminar-separation and turbulent-reattachment loca- tion was used to determine laminar-separation and tions. Comparisons of these results with data on the turbulent-reattachment locations. Comparisons of Eppler 387 airfoil from two other facilities, as well these results with data on the Eppler 387 airfoil from as with predictions from the Eppler airfoil code, are two other facilities, as well as with predictions from included. A discussion of the most pertinent results the Eppler airfoil code, are included. from this test is reported in reference 5. The data are presented herein in both tabulated and plotted Introduction formats. Symbols Recent interest in low Reynolds number aerody- namics has increased for both military and civil ap- The symbols in parentheses are those used in plications with emphasis on providing better vehi- computer-generated tables in the appendixes. cle performance (ref. 1). Reynolds numbers below 500 000 are usually identified as being in this classifi- cation. Applications are varied and include remotely b (B) airfoil span, in. piloted vehicles, ultralight human-powered vehicles, Cp pressure coefficient, wind turbines, and propellers. qoo c (C) Although the design and evaluation techniques for airfoil chord, 6 in. airfoils at Reynolds numbers above 500 000 are well Cc section chord-force coefficient, developed, serious problems related to boundary- f cp layer separation and transition have been encoun- tered at lower Reynolds numbers. Presently avail- ¢d (CD) section profile-drag coefficient, able design and analysis methods generally do not fwake Ctdd(h/c) adequately model flow phenomena such as laminar point-drag coefficient (see separation bubbles. Experimental results obtained appendix A) on an Eppler 387 airfoil at low Reynolds numbers in the Model Wind Tunnel at Stuttgart (ref. 2) and Cl (CL) section lift coefficient, the Low-Turbulence Tunnel at Delft (ref. 3) have cncOsO_- CcsinOl shown large differences in airfoil performance. This Cm (CM) section pitching-moment is not surprising because of the sensitivity of the coefficient about quarter- airfoil boundary layer to free-stream disturbances, chord point, - f Cp(x/c - model contour accuracy, and model surface rough- 0.25) d(x/c) + f Cp z/c d(z/c) ness. Also, the model forces and pressure differences section normal-force coeffi- are small and difficult to measure accurately. c12 NASA Langley Research Center has initiated cient, -f Cp d(x/c) a research program to develop test techniques to h vertical distance in wake determine performance characteristics of airfoils profile, in. at low Reynolds numbers (R _< 500000) (ref. 4). M free-stream Mach number This experimental program uses the Langley Low- Turbulence Pressure Tunnel and consists of perfor- P static pressure, psi mance evaluation of both force and pressure models pt (PT) total pressure, psi of an Eppler 387 airfoil. Oil flow visualization and q dynamic pressure, psi surface-mounted thin-film gages were used to deter- mine laminar-separation and turbulent-reattachment R Reynolds number based on locations. Also, test-section turbulence and acoustic free-stream conditions and measurements were obtained. airfoil chord of 6 in. R/ft unit Reynolds number fabrication tolerance was maintained except on the model upper surface between chordwise locations of ratio of fluctuating velocity to x/c = 0.60 and z/c = 0.80. A surface finish of 64 #in. mean velocity in streamwise (rms) was specified. direction Grooves were machined in the surface of the steel x (x) airfoil abscissa, in. model and pressure tubing was routed through the grooves for orifice locations. The grooves were filled y (Y) spanwise distance along model with epoxy resin. Orifices were drilled through the from centerline, in. metal surface into the tubing with their axes per- Z airfoil ordinate, in. pendicular to the local surface. Each orifice had a diameter of 0.020 in. except at x/c = 0.95 where a (ALPHA) angle of attack, deg diameter of 0.010 in. was used. The locations of both Subscripts: upper and lower surface orifices are indicated in ta- ble II. The orifices were staggered to alleviate mu- des design tual interference, as illustrated by the photograph of diff difference figure 4. max maximum Wind Tunnel meas measured The test was conducted in the LTPT. This fa- cc (INF) free-stream conditions cility is described in detail in reference 6, and dy- Abbreviations: namic flow quality measurements are reported in ref- erence 7. The LTPT is a pressurized, closed-circuit, LS laminar separation from flow continuous-flow wind tunnel with an operating pres- visualization sure from approximately 0.10 to 10 atm. The test LTPT Low-Turbulence Pressure section was designed for two-dimensional testing of Tunnel airfoil sections and is 7.5 ft high, 7.5 ft long, and 3 ft wide. The contraction ratio is 17.6:1, and 9 anti- NT natural transition from flow turbulence screens are installed in the settling visualization chamber. This facility was selected to develop test tech- rms root mean square niques for low Reynolds number aerodynamics be- sep. separation cause of its good flow quality, precision pressure in- TR turbulent reattachment from strumentation, and variable pressure capability. The flow visualization tunnel operating envelope for a 6-in-chord airfoil model is shown in figure 5; test conditions for the Eppler 387 model are also indicated. In order to en- Model, Apparatus, and Procedure chance the resolution of model forces and pressure differences, it is desirable to operate at the higher Model end of the dynamic pressure envelope. The airfoil model was machined from stainless To supplement the turbulence measurements for steel. To provide structural integrity and room for the LTPT (see ref. 7) in the low Reynolds number pressure tubing, the trailing edge of the model was range, additional test-section turbulence was mea- thickened from 0 to 0.01 in. The additional thick- sured with a hot-wire anemometer by Gregory S. ness was blended into the Eppler 387 coordinates at Jones of the Langley Research Center. These prelim- x/c = 0.95. (See table I.) The basic camber distri- inary results, shown in figure 6, indicate that free- bution of the Eppler 387 airfoil was retained. The stream turbulence is increased for a constant unit model had a chord length of 6 in. and a span of 36 in. Reynolds number as the tunnel total pressure is de- A drawing of the Eppler 387 section shape is shown creased. For example, at a unit Reynolds number in figure 1. A photograph of the model mounted in of 200000 per foot, the test-section turbulence level the LTPT is shown in figure 2. The model design (frequency bandpass from 1 to 50000 Hz) increases contour accuracy was within +0.001 in. The differ- from about 0.06 percent to 0.18 percent as the to- ences between the design and measured coordinates tal pressure is reduced from 15 psi to 3 psi. It is are shown in figure 3 as a function of both chord- well known (ref. 1) that boundary-layer receptivity wise and spanwise locations. In general, the specified is strongly affected by the frequency content of the 2 disturbanceenvironmenatswellasbythemagnitude Instrumentation ofbothvelocityandpressurefluctuations. Measurements of pressure on the model sur- faces, wake-rake pressures, and basic tunnel pressures Wake Survey Rake were made with variable-capacitance precision trans- The wake survey rake (fig. 7) was mounted on ducers. These transducers have an accuracy of the tunnel survey apparatus and located 1.50 chords -t-0.25 percent of reading. An automatic pressure- scanning system was used to record the model behind the trailing edge of the airfoil. The rake con- tained seven total-pressure tubes, each 0.063 in. in pressures. The following full-scale ranges of pres- diameter, which were flattened to 0.020 in. (inter- sure transducers were used: Pt, 1000 mm Hg; q, nal height) over a length of 0.25 in. from the tip of 10 mm Hg; wake rake, 10 mm Hg; model upper the tube. The rake is equipped with both standard surface, 50 and 10 mm Hg; model lower surface, 10 mm Hg. and disc-type static-pressure probes. The standard probes were used to measure the static pressure in Model angle of attack was measured by a cali- the wake for the present test. The static probes brated digital shaft encoder driven by a pinion gear were 0.125 in. in diameter with eight flush orifices and rack attached to the pitch mechanism. Data (0.018 in. diameter) drilled 45° apart and located were obtained by a high-speed data acquisition sys- 8 tube diameters from the tip of the probe. The rake tem and recorded on magnetic tape. Real-time data also contained two claw-type flow-angularity probes, displays on cathode-ray tubes were available for tun- which consisted of two open-ended probes inclined nel parameters, model pressures, and wake profiles. 90° with respect to each other. These probes were used to obtain the mean flow direction of the wake. Tests and Methods The pressure model was tested at Reynolds num- Survey Apparatus bers based on airfoil chord from approximately 60 000 to 460000 and Mach numbers from 0.03 to 0.13. The The wake rake was positioned at various spanwise model was generally tested in a smooth condition ex- stations behind the model by means of the remote- cept for a strip of turbulator tape used at a Reynolds controlled survey apparatus (fig. 8). The apparatus number of 100 000. This tape was 0.008 in. thick and basically consists of an articulating arm mounted on an arc strut. Movement of the arm enables the wake 0.08 in. wide. The leading edge of the tape formed a zig zag pattern and was located at 0.22c on the surveys to be made for various angles of attack. model upper surface. The arm is composed of three movable compo- Laminar-separation and turbulent-reattachment nents: a main boom, an offset boom, and a forward locations were determined using the oil flow tech- pivoting head. Each component has a position con- trol device. The main boom is mounted on the strut nique reported in reference 8. These results are shown in table III and a typical result for a Reynolds with a pivot point allowing rotation in the vertical number of 300 000 is illustrated in the photograph of plane. Its motion is controlled by the linear actuator. The offset boom can be rotated about the main boom figure 9. by the roll actuator. This allows survey positions to The static-pressure measurements at the model be made at distances up to 12 in. from the tunnel surface were reduced to standard pressure coeffi- centerline. The forward pivoting head is mounted at cients and numerically integrated to obtain section the end of the offset boom and may be rotated in normal-force and chord-force coefficients and sec- the vertical plane by the (internally mounted) pitch tion pitching-moment coefficients about the quarter- adjustment mechanism. Figure 8 shows the survey chord point. Section profile-drag coefficients were apparatus with the wake rake mounted on the for- computed from the wake-rake total and static pres- ward pivoting head assembly. In addition, the en- sures by the method of reference 9. tire apparatus can be positioned vertically in the Standard low-speed wind-tunnel boundary cor- wind tunnel by using the movable strut that moves rections (ref. 9) have been applied to the section data. within the confines of fixed leading- and trailing-edge Corrections were applied to the free-stream dynamic fairings. Positioning and rate of movement of the pressure because of solid and wake blockage and ap- rake are controlled by a microprocessor controller. plied to lift, pitching moment, and angle of attack be- In general, wake surveys using this apparatus pro- cause of the effects of floor and ceiling constraints on vided good drag results with a survey rate of about streamline curvature. No blockage corrections have 0.10 in/sec or less. been applied to the pressure coefficient data. The magnitudeof these corrections for the Eppler 387 Effect of tunnel environment on airfoil are chordwise pressure distributions for R = 60 000 ............. 14 a corrected = a + 0.0083(c/+ 4Cm) Spanwise pressure data for c_= 5°; ct corrected = c/(0.9988 - 0.0333Cd) R = 60000 and 100000 ......... 15 Cm corrected = Cm(0.9997 - 0.0333Cd) + 0.0002ct Effect of Reynolds number on Cd corrected = Cd(0.9995 -- 0.0333Cd) section data .............. 16 It is important when measuring performance Effect of angle of attack on chordwise characteristics of airfoils to provide some indication pressure distributions; of the data accuracy. There are several areas in two- R = 60 000 to 460 000 .......... 17 dimensional airfoil testing at low Reynolds numbers Effect of Reynolds number on that contribute to the overall uncertainty of the re- chordwise pressure distributions; suits: tunnel flow quality, experimental apparatus, R = 60 000 to 460 000 .......... 18 and instrumentation accuracy. The major errors introduced by the apparatus are Variation of drag coefficient with confinement effects of the wind-tunnel walls, sidewall Reynolds numbers ........... 19 boundary-layer interaction, and large-scale vortices Variation of maximum lift coefficient in the wake if wake-rake surveys are used to deter- with Reynolds number ......... 20 mine drag. For the present tests, the confinement ef- fect of the wind-tunnel walls was minimized by test- Separation and reattachment ing a model with a chord-to-tunnel-height ratio of locations from oil flow data; about 0.07. The sidewall boundary-layer interaction R -- 100 000 to 300 000 ......... 21 effect was reduced by using a pressure model with Comparison of pressure data with orifices near the center of the model and a model oil flow results illustrating span-to-chord ratio of 6. To survey the spanwise laminar-separation and turbulent- flow structure in the wake, the wake rake was tra- reattachment locations; versed in the spanwise direction. However, the wake- R = 100 000 to 300 000 ......... 22 rake technique of determining drag is still subject to errors related to the changing flow direction in the Hysteresis effects on section data; unsteady wake. Figure 10 illustrates typical wake R = 60 000 to 300 000 ......... 23 profiles where two different total-pressure probes Hysteresis effects on chordwise pressure traversed through the complete wake. Note the un- distributions for R -- 60 000 ....... 24 steady wakes for R < 100000. (See ref. 1.) The degree of uncertainty associated with the instrumen- Hysteresis effects on chordwise pressure tation accuracy was minimized by using precision distributions for R = 100 000 ...... 25 pressure transducers. Effect of turbulator tape on section An estimate of the uncertainties in the section data; R = 100000 ........... 26 data for a = 4°, using the technique of reference 10, is shown in appendix A. Effect of turbulator tape on chordwise pressure distributions; R = 100 000 .... 27 Presentation of Results Data from LTPT and other facilities; The results of this investigation have been re- R = 60 000 to 200 000 .......... 28 duced to coefficient form and tabulated in appen- dixes B through E. Selected results are presented in Experimental data and predictions from the following figures: Eppler airfoil code; R -- 60 000 to Figure 460 000 ................ 29 Effect of tunnel environment on section data; R = 60000 and 100000 ...... 11 Discussion of Results Spanwise drag data; R = 100 000 to 300 000 ............... 12 Experimental Results Effect of tunnel environment on chordwise Effect of tunnel environment. Figures 11 through pressure distributions for 15 illustrate the effect of tunnel environment. It is R = 100 000 .............. 13 well known (ref. 1) that boundary-layer phenomena, suchaslaminar-separatiobnubblesc,anbeaffected tivity of the airfoil boundary-layer characteristics at by the tunnelenvironment.Theeffectsof several R = 60 000. free-streamconditionson the airfoil sectiondata Reynolds number effects. Figures 16 through 25 at R = 100000 are shown in figure ll(b). The illustrate Reynolds number effects. The effects of in- measured turbulence levels (fig. 6) vary from about creasing Reynolds number from 60 000 to 460 000 on 0.06 percent at Pt = 15 psi and M = 0.03 to the airfoil section data are shown in figure 16. The about 0.16 percent at Pt = 5 psi and M = 0.08. data presented are for the free-stream environment Increasing the tunnel turbulence level at constant where the lowest disturbance levels were measured Reynolds number showed no effect on the lift and (fig. 6). Increasing the Reynolds number results in pitching-moment data. However, some effect on large improvements in airfoil performance because of the drag data did occur as illustrated by the drag the decrease in size of the laminar-separation bub- polar of figure l l(b). Increasing the turbulence ble. The pressure data of figure 18 illustrate this level of the tunnel would be expected to have a favorable Reynolds number effect. For example, for beneficial effect on the bubble characteristics, similar -- 4° (fig. 18(d)), a decrease in the extent of the to that observed for surface roughness (ref. 1), and upper surface laminar-separation bubble from more hence, cause a reduction in drag. However, this than 0.50c to about 0.10c is indicated for an increase result is not clearly indicated. Significant spanwise in Reynolds number from 60 000 to 460 000. A cor- variations in cd are shown (fig. 12(a)) at R = 100000 responding decrease in cd of 0.0310 is indicated. As for these free-stream conditions. The lowest values discussed earlier, two flow regimes (laminar separa- of cd were measured at span station 3 in., which tion with and without turbulent reattachment) oc- is where the model surface pressure orifices were curred at R ----60 000 for several angles of attack. For located. Large improvements in spanwise variations Reynolds numbers greater than 60 000, when laminar of cd are shown (fig. 12(b)) at Reynolds numbers of separation occurred, turbulent reattachment always 200000 and 300000. The pressure data of figure 13 resulted. The pressure data (fig. 18) also indicate illustrate the effect of the tunnel environment on the the changes in airfoil loading because of increases in bubble characteristics for several angles of attack. Reynolds number (R = 60000 to 200000) and the The main effect of different free-stream conditions resulting decrease in the magnitude of the pitching- is the location of flow reattachment on the upper moment coefficients. Figures 19 and 20 summarize surface of the airfoil. These results illustrate the the effects of Reynolds number on drag coefficient sensitivity of the bubble phenomena to the free- and maximum lift coefficient. stream environment. A more detailed effect of Reynolds number and Figure ll(a) illustrates the effects of two free- angle of attack on the upper surface bubble charac- stream conditions on the section data at R = 60 000. teristics from the oil flow results is shown in figure 21 The tunnel turbulence levels were about 0.16 per- for Reynolds numbers from 100 000 to 300 000. The cent for pt = 5 psi and M = 0.05, and 0.20 per- pressure data and oil flow results are shown in com- cent for Pt = 3 psi and M = 0.09. For the data parison in figure 22. A decrease in bubble length is taken at Pt = 5 psi and M = 0.05, two differ- shown for either an increase in angle of attack at a ent flow phenomena (laminar separation with and constant Reynolds number or an increase in Reynolds without turbulent reattachment) were observed at number at a constant angle of attack. Increas- the same angle of attack. This unsteady flow oc- ing the Reynolds number resulted in only a small curred for angles of attack between about 3° and effect on the location of laminar separation com- 7°. The pressure data of figure 14 illustrate the two pared with turbulent reattachment. For example, for flow regimes for several angles of attack, and span- a = 4°,increasing the Reynolds number from 100 000 wise pressure data are shown in figure 15 for a = 5°. to 300000 produced only about 0.05c movement in It should be noted that the pressure data were ob- the laminar-separation point compared with about tained using an automatic pressure scanning system; 0.15c movement in the turbulent-reattachment loca- thus each pressure was measured at a different time. tion. At Reynolds numbers of 200 000 and 300 000 The data at Pt = 3 psi and M = 0.09 for the angle- and angles of attack between 7° and 8°, the flow of-attack range where the two flow regimes were ob- remained attached and natural transition occurred. served always resulted in laminar separation without This condition generally resulted in the best lift-to- flow reattachment. Consistent flow reattachment oc- drag ratio for the airfoil. curred at a = 7.5° (fig. ll(a)) for both tunnel con- The importance of hysteresis phenomena for air- ditions. Large increases in drag are shown in the foils at low Reynolds numbers is pointed out in ref- angle-of-attack range where flow reattachment did erence 1. The presence and extent of these phe- not occur. These results illustrate the extreme sensi- nomena are generally determined by the location of separationand/ortransitionin theboundarylayer. 100000 and 200000 (figs. 28(b) and 28(c)), generally HysteresidsatawereobtainedatReynoldsnumbers good agreement between the LTPT and Delft data from60000to300000byincreasintgheangleofat- is shown; the major discrepancy is in the lift data in tackfrom-3° tostallandthendecreasintgheangle the high-angle-of-attack range where the LTPT data of attackfromstallto about0°. Figure23illus- show higher values of lift coefficients. This difference tratesthehysteresiesffectonthesectiondataand may be attributed to the flow interference effects be- figures24and25showtheeffectsonthechordwise tween the tunnel sidewall and model end plates, since pressuredata. Generallyn,ohysteresisloopswere a balance was used for the Delft tests. However, large observedh;owevera,spreviouslydiscussedtw, oflow differences are shown between the Stuttgart data and regimewserepresenftorasmallangle-of-attacrkange data from the LTPT or Delft. The Stuttgart lift data for R = 60 000. are generally lower, particularly at the higher angles Effect of turbulator. Figures 26 and 27 illustrate of attack, and large differences in drag data are indi- the effect of the turbulator. Performance character- cated. The Stuttgart drag data, compared with the istics of airfoils at low Reynolds numbers are domi- other tunnels, indicate lower values of cd at lift coef- nated by laminar-separation bubbles. One approach ficients where the bubble has a large influence on Cd, to provide improvements is the introduction of suit- and generally higher values of cd in the low lift co- able disturbances in the boundary layer such that efficient range. (See fig. 28(b), R = 100000.) These transition occurs ahead of where laminar separation differences in drag data may be attributed to tunnel would normally occur. Thus, a boundary-layer dis- flow quality, or perhaps model contour accuracy and turbance or turbulator was employed. A spanwise surface roughness effects. The data from the three facilities at R = 60 000 is strip of tape was placed at 0.22c on the model up- per surface, and the results for R = 100000 are il- shown in comparison in figure 28(a). As previously lustrated in figures 26 and 27. The turbulator was discussed, the LTPT data displayed two flow regimes effective in reducing drag up to a lift coefficient of at several angles of attack and showed extreme sensi- about 1.0, as shown by the drag polar of figure 26. tivity to the tunnel environment at R -- 60 000. The LTPT and Delft data both indicate that laminar stall The pressure data for _ = 4° (fig. 27(h)) show typ- ical effects on the laminar-separation bubble due to near cl ,_ 0.6 occurred with large increases in Cd, and the turbulator. The turbulator tape did not elim- flow reattachment occurred near cI = 1.0. However, inate the bubble; however, turbulent reattachment the Stuttgart data do not display these phenomena. occurred further forward on the airfoil upper surface, Comparison of Results With Eppler Airfoil as indicated by the forward movement of the aft pres- Code sure recovery. A reduction in cd of about 17 percent resulted. For a = 7° (fig. 27(k)), transition occurred The Eppler airfoil code (ref. 11) has been one ahead of the turbulator tape (because of the adverse of the most useful codes for the design and analy- pressure gradient near the leading edge) and as ex- sis of low-speed airfoils. The most important and pected, no reduction in cd resulted. difficult part of the boundary-layer calculations for low Reynolds numbers is to account for the laminar- separation bubble. This code contains a bubble ana- Comparison With Results From Other Facilities logue that is evaluated from conventional computa- tional methods based on the integral momentum and The results of the present experiment are com- energy equations. pared with data obtained on an Eppler 387 air- Lift and pitching-moment coefficients are deter- foil model in the Model Wind Tunnel at Stuttgart mined from the potential flow. Viscous corrections and the Low-Turbulence Tunnel at Delft, where the are applied, including a correction for boundary- free-stream turbulence levels are 0.08 percent and layer separation. Drag coefficients are obtained by 0.03 percent, respectively. Data shown for the LTPT applying a modified Squire-Young formula to the are for the environment where the lowest turbulence boundary-layer characteristics at the trailing edge. levels were measured (0.06 percent for R = 100000 The prediction of separation is determined by the and R = 200 000, and 0.16 percent for R = 60000). shape factor based on energy and momentum thick- The lift data for the LTPT tests were obtained from nesses. The prediction of transition is based on an surface pressure measurements while the data from empirical criterion that contains the Reynolds num- the other facilities were obtained from force-balance ber (based on local conditions and momentum thick- measurements. Drag data for all three facilities ness) and the shape factor. The code predicts the ex- were obtained from pressure measurements by us- istence of significant laminar-separation bubbles and ing a wake survey rake. For Reynolds numbers of provides a warning to indicate that the predicted 6