AIAA 99-0925 Oscillatory Excitation of Unsteady Compressible Flows over Airfoils at Flight Reynolds Numbers Avi Seifert and LaTunia G. Pack NASA Langley Research Center Hampton, VA 37th AIAA Aerospace Sciences Meeting and Exhibit January 11-14, 1999/ Reno, NV For permission to copy or republish, contact the American Institute of Astronautics and Aeronautics 1801 Alexander Bell Drive, Suite 500, Reston, VA 20191 AIAA 99-0925 Oscillatory Excitation of Unsteady Compressible Flows over Airfoils at Flight Reynolds Numbers Avi Seifert* and LaTunia G. Pack= Flow Modeling and Control Branch, NASA Langley Research Center Hampton, Va. 23681-0001 Abstract C total drag coefficient d An experimental investigation, aimed at delaying C form drag coefficient dp flow separation due to the occurrence of a shock-wave- C zero-lift drag coefficient do boundary-layer interaction, is reported. The experiment C lift coefficient l was performed using a NACA 0012 airfoil and a NACA ( ) C pressure coefficient, ” P- P q 0015 airfoil at high Reynolds number incompressible and p s compressible flow conditions. The effects of Mach and C ' fluctuating pressure coefficient, ” p' q p Reynolds numbers were identified, using the capabilities f frequency [Hz] of the cryogenic-pressurized facility to maintain one GN gaseous nitrogen 2 parameter fixed and change the other. Significant F+ reduced frequency, ” fX /U Reynolds number effects were identified in the baseline te ¥ h slot height or width compressible flow conditions even at Reynolds number M Mach number, ” U¥ a of 10 and 20 million. The main objectives of the P pressure experiment were to study the effects of periodic excitation on airfoil drag-divergence and to alleviate the q free stream dynamic pressure, ” 1/2r U¥2 severe unsteadiness associated with shock-induced Rc chord Reynolds number, ” U¥ c/n separation (known as “buffeting”). Zero-mass-flux T temperature oscillatory blowing was introduced through a U, u averaged and fluctuating velocity downstream directed slot located at 10% chord on the X/c normalized streamwise location upper surface of the NACA 0015 airfoil. The effective Xte distance from actuator to TE frequencies generated 2-4 vortices over the separated Y/c distance normal to airfoil surface region, regardless of the Mach number. Even though the Z spanwise location excitation was introduced upstream of the shock-wave, a airfoil angle of attack, deg due to experimental limitations, it had pronounced effects n kinematic viscosity downstream of it. Wake deficit (associated with drag) and r density unsteadiness (associated with buffeting) were Abbreviations significantly reduced. The spectral content of the wake LE airfoil leading edge pressure fluctuations indicates of steadier flow TE airfoil trailing edge throughout the frequency range when excitation was < > phase locked values applied. This is especially important at low frequencies Subscripts which are more likely to interact with the airframe. j conditions at blowing slot max conditions at maximum lift Nomenclature s tunnel static conditions a speed of sound, ” gRT ¥ free stream conditions cm oscillatory blowing momentum coefficient, Superscripts ” h/c(1+T¥ /Tj)( u' /U¥ )2 ‘ crit rcoriotti cmale Mana scqhu naurem obfe frluctuating value c airfoil chord res acoustic resonance frequency * NRC researcher, on leave from Tel-Aviv University, member AIAA. = Research engineer, Flow Modeling and Control Branch. (cid:211) A. Seifert (NRC) and NASA, 1999. Printed by AIAA with permission. 1 American Institute of Aeronautics and Astronautics AIAA 99-0925 1. I n troduction the airfoil geometry is almost unchanged, due to the Low Reynolds and Mach number studies [Ref. 1-3 narrow blowing slots. and references therein] have shown that periodic vortical Flow separation at compressible speeds typically excitation introduced into a separating boundary layer, occurs downstream of a shock-wave/boundary-layer slightly upstream of the average separation location, can interaction. The pressure jump across the shock either effectively delay boundary layer separation. The causes immediate separation or thickens the boundary improved ability of the boundary layer to overcome an layer and reduces its momentum such that it separates adverse pressure gradient is attributed to enhanced further downstream. Once the flow separates downstream mixing between the low momentum fluid near the wall of the shock, the unsteady separation and subsequent and the external high momentum flow. The successful reattachment (if it occurs) induce unsteadiness both in the application of the method increases the lift while shock position and strength. This phenomenon is known maintaining low drag. At low Mach numbers, where as buffeting. The low frequency oscillations can cause high-lift for take-off, landing or loiter is required, the structural damage, if coupled with the resonance delay of boundary layer separation allows increased frequencies of the structure. Porous strips and wall loading of a multi-element high-lift airfoil system. bumps7 are effective in reducing the strength of the It was recently demonstrated4 that periodic excitation shock-wave. Vortex generators (mechanical8 or canted of the boundary layer upstream of separation can delay jets9) as well as suction through slots, are effective at the occurrence of the adverse effects associated with controlling shock-induced separation and alleviating boundary layer separation and significantly enhance the buffet. performance of airfoils at flight Reynolds numbers and Triangular, ramp-like, vortex generators were used in incompressible speeds. Low Reynolds number a numerical simulation to demonstrate passive control of experiments, where control was applied from the LE shock-wave/boundary-layer interaction and buffet region of the airfoils, were repeated at a chord Reynolds alleviation10. This was achieved by energizing the number of 37.6x106. Using a flapped NACA 0015 airfoil, boundary layer upstream of the shock-wave by pairs of where control was applied at the flap shoulder, it was counter rotating vortices. The vortices were generated 15 shown that the method is essentially independent of boundary layer thicknesses (50 devices heights) upstream Reynolds number4, as long as the appropriate of the shock position. Although placed upstream, the dimensionless control parameters are applied. main effect was a smaller separation bubble downstream A recently published numerical simulation5 shows of the shock. As a result of the thinner boundary layer the that oscillatory excitation of a separated boundary layer, shock strength increased, as did the rate of pressure at low Reynolds and Mach numbers, can significantly recovery downstream of it. The calculated pressures are increase post-stall lift at excitation frequencies that are in qualitative agreement with experiment8. Additional 0.3 to 4 times the natural vortex shedding frequency aspects such as skin friction, device drag and flow + (F =0.4 in this case). By using frequencies that are at unsteadiness were not investigated. least twice the shedding frequency, which correspond to The effects of suction through slots and a porous our definition of F+» 0.8, the lift enhancement is surface, on the performance of a supercritical airfoil at accompanied by a significant reduction in drag and drag off-design conditions were tested experimentally11. excursions [Fig. 6 and 7 in Ref. 5]. Indications of these Suction through a single slot with a mass flow coefficient effects were experimentally identified in Ref. 4. Using an of 60x10-5 (slot width unknown therefore momentum appropriate combination of frequency (F+» 1) and coefficient unknown) was very effective in maintaining magnitude (<cm >= 10 to 50x10-5) the flow should be the upstream shock position, re-establishing the pressure steadier, even if it is intermittently separated. Similar recovery and positive C , and increasing post-stall lift p,TE trends, at least for the lift increment and the excitation but was less effective in reducing drag. The double slot + F , were identified numerically6. Oscillatory addition of and perforated plate (8% porosity over 7.5% chord) were momentum is two orders of magnitude more efficient less effective than the single slot at increasing post-stall than tangential steady blowing for separation control2,4. lift, but more effective in reducing drag (when using the Neither forced transition, thickened turbulent boundary same mass flow coefficient, i.e. significantly smaller layer nor elevated Reynolds numbers4 (up to 37.6*106 momentum coefficient). The double slot and perforated tested) had an adverse effect on the efficiency of the plate were effective even without the suction. This was method. The nature of the device used to generate the possible because the cavity underneath allowed mass unsteady disturbances is not important, as long as the end transfer from the downstream side of the shock (high result (i.e. the vortical excitation) is similar. Two pressure) to the upstream side of it. In the steady sense, dimensional, wall tangential oscillatory blowing2,4 is the suction downstream of the shock reduced the currently used since the magnitude, the frequency and the boundary layer tendency to separate while the upstream average mass flux are conveniently controllable, while bleed from the surface, thickened the boundary layer and 2 American Institute of Aeronautics and Astronautics AIAA 99-0925 reduced the shock strength. This in turn reduced the methodology of data processing are presented in the tendency for separation downstream. The double slot was following sections. A conventional NACA 0012 airfoil especially effective in buffet alleviation, completely was tested in order to validate the new rake and to eliminating the shock-wave/separated flow coupling. The acquire unsteady wake data over a smaller chord, smooth physical mechanism producing this effect was not airfoil. The straight NACA 0015 airfoil was tested with identified. control applied from the LE region. To be efficient the above mentioned methods need to be actively controlled. Bump position, height and shape 2.2 T h e Wind Tunnel need to be tailored to each specific flow condition. The experiment was conducted in the 0.3-meter Porous strips need to cover about 10% of the chord, cause Transonic Cryogenic Wind Tunnel at the NASA Langley transition on laminar airfoils and increase skin friction Research Center. It is a closed loop, fan driven tunnel drag. Porosity and slot locations also need to be mission with a test cross section of 0.33m x 0.33m. Gaseous tailored. While these devices could be designed to widen nitrogen (GN ) is the test medium. The tunnel operates at 2 the flight envelope to a certain extent, their application stagnation pressures ranging from 1.2 bar up to 6 bar and for guidance and control, which requires fast response12, total temperatures from 78K up to 327K 13, 14. The floor is doubtful. Furthermore, efficiency considerations rule and ceiling of the tunnel were diverged by 0.3(cid:176) to 0.4(cid:176) in out the use of tangential blowing and stealth the vicinity of the airfoil to reduce blockage resulting considerations rule out the use of mechanical vortex from boundary layer growth on the test section walls. A generators. Therefore, fast responding, active methods for wake survey rod extends from the left tunnel sidewall to management of high-lift as well as compressible unsteady vertically traverse the airfoil wake (see Fig. 1). Details flows are studied. about the wake data acquisition system are provided in The objectives of the present investigation are the section 2.7. following: Wall pressures were acquired together with the airfoil 1. To reduce flow unsteadiness in incompressible pressures at all test conditions. This information could be speeds and to determine the relationship between the used later for numerical simulation of the flow taking into controlled excitation and the naturally occurring consideration wall interference effects. unsteady flow in the presence of separated flows. 2. To suppress shock-induced separation and provide 2.3 T h e Airfoils buffet alleviation by using periodic excitation. The experiments were conducted on a standard NACA 0012 airfoil (c=165 mm) and a NACA 0015 2. E x perimental Set-up airfoil (c=254 mm). The NACA 0015 airfoil (Fig. 2) was equipped with a blowing slot at 10% chord, suitable for 2.1 O v erview the control of separation near the leading edge. The slot The experiment was conducted in the 0.3-meter was about 0.2% chord wide (0.5 mm), and allowed an Transonic Cryogenic Tunnel at NASA Langley Research almost tangential streamwise introduction of the Center, using Gaseous Nitrogen (GN2) as the test fluid. excitation. Each airfoil was equipped with some 50 static The execution of an active flow control experiment in a pressure taps. The airfoil and test section wall pressures pressurized cryogenic wind tunnel has advantages and were recorded by the tunnel pressure scanning system. disadvantages. For example, a cryogenic pressurized facility allows independent control ofR and M at a 2.4 O s cillatory Blowing System c fixed free stream velocity. With this type of control, the A rotating, siren type, valve was used to generate the + effective frequencies are clearly indicated because F pressure oscillations inside the airfoil cavity. The can be held fixed whenR is varied and M is held oscillatory blowing valve was upgraded and is presently c constant orR is fixed while M is varied. Another capable of generating frequencies up to 800 Hz and c advantage of testing in a cryogenic pressurized facility is fluctuating pressure levels of 5 psi. For safety reasons the the ability to generate a zero mass flux disturbance when valve was rated to 300 psi. GN was supplied to the valve 2 using an oscillatory blowing valve. One of the by converting a portion of the liquid nitrogen available disadvantages of testing in a cryogenic pressurized for operating the tunnel using an ambient temperature facility is that an in-situ determination of <cm > is very vaporizer. The use of ambient temperature GN2 difficult. However, using atmospheric bench-top tests and simplified the valve design. A pressure regulator a simplified flow model it is possible to estimate the controlled the GN entering the valve and the variable 2 <cm > used4. speed drive of the valve motor controlled the frequency A new diagnostic tool was used in this experiment. It of the pressure oscillations. The oscillatory blowing valve consists of a wake rake that was instrumented with was attached to the right tunnel plenum door at the center dynamic pressure transducers. Its calibration and the of rotation of the turntable (Fig. 1). A 49mm I.D. pipe 3 American Institute of Aeronautics and Astronautics AIAA 99-0925 connected the valve to the leading edge cavity. The dynamic pressure, it was decided to use a smaller exhaust side of the airfoil was connected to the tunnel diameter tube (1.60 mm O.D. and 1.03 mm), about 30 boundary layer removal system. The valves in the mm long, from the transducer’s face to the measuring boundary layer removal system controlled the flow rate location. The length and internal volume of this tube can out the exhaust side of the airfoil cavity (Fig 1). Any change the frequency response of the tube-transducer relevant combination of steady and oscillatory blowing system due to acoustic resonance. A calibration was could be generated with this type of control. However, performed by placing a sound source in front of the wake all the results presented in this paper are for zero-mass- rake, inside the wind tunnel test section at atmospheric flux excitation. conditions, with the tunnel flow off. The sound was either white noise or discrete tones. A microphone was 2.5 B e nch -Top Experiments positioned next to the tubes. The signals from the two The phase-locked pressure fluctuations ( <p'> ) at the dynamic pressure transducers were recorded along with entrance to the airfoil cavity were measured both in-situ the microphone output. These results are used to correct and with bench-top tests. The correlation between <u'2> the signals measured under various tunnel conditions, and <p'>/r (derived from the bench-top tests) is used to assuming that the only variant is the flow temperature calculate the <cm > in the cryogenic tests4. (which affects the speed of sound). The velocity fluctuations exiting the slots of the Fig. 4 presents the results of the white noise and airfoil were measured with the airfoil outside the tunnel discrete tone testing. The white noise results show that using a hot-wire mounted on a 3D traverse system. The the microphone and the dynamic pressure transducer GN supplied to the oscillatory blowing valve during the measure similar noise levels (– 2dB) between 150Hz and 2 wind tunnel test was replaced with compressed air. All of 1KHz. The discrete tone testing was performed at a the pipe accessories were identical in both the wind resolution of about 100 Hz and at significantly higher tunnel and the bench-top experiments in order to SPL levels (about 104 dB) than the white noise test. maintain similarity between the two experiments. While Figure 4 also presents the SPL difference between the any desired combination of oscillatory and steady flow dynamic pressure transducer and the microphone. rates were obtainable in the wind-tunnel experiment, Transducer #1 (position #2 in the wake-rake see Fig. 3) steady suction could not be applied in the bench-top and the microphone measure similar SPL until 0.5 kHz. experiments because of the atmospheric test conditions. For frequencies above 0.6 kHz the pressure transducer However, even in the bench-top experiments, reverse output increases monotonically with respect to the flow in the slot was encountered due to the instantaneous, microphone output. The difference peaks at about 1.5 sub-atmospheric pressures created by the inertia of the kHz where the pressure transducer output is about 50 dB continuous flow along the airfoil cavity. The signals from higher than the microphone output. Examining again the the dynamic pressure transducer and the hot-wire were results of the white noise test and normalizing the acquired using a 16-bit high speed A/D converter, pressure transducer output by the microphone output (the coupled with an anti-aliasing filter. In the cases where solid line in Fig. 5) reveals a flat response between 150 reverse flow was expected at the slot exit, the hot-wire Hz and about 1 kHz. The transducer-tube system signal was de-rectified to account for the reverse flow4. resonates at about 1.4-1.5 kHz, at ambient temperature. The SPL difference between the pressure transducer and 2.6 T h e Dynamic Wake-Rake microphone outputs is plotted in Fig. 4. It shows that at The wake of the airfoils was traversed using the f/f =0.95 the difference is 20 dB, and a smooth line fits res tunnel standard wake survey rake and a modified wake the data very well. The required pressure fluctuations rake, instrumented with two dynamic pressure attenuation factor was calculated using a second order transducers and seven total pressure tubes. The modified polynomial fit to the inverse of the SPL difference in the wake rake enables the recording of unsteady wake data, form phase locked to the controlled excitation. Its LE was Factor=- 1.076(f f )2 +0.1343(f f )+0.905 (1) res res machined to a semi-elliptic shape. The dynamic wake The acoustic resonance frequency is temperature rake was located 3.1 and 2 chords downstream of the dependent and was found to scale, as expected, inversely NACA 0012 and 0015 airfoils mid chord, respectively. with the square root of the temperature ratio. This point is Fig. 3 presents a picture of the dynamic wake rake. illustrated by the additional data shown in Fig. 5. The The dynamic pressure transducers were installed inside M=0.28 and M=0.55 data were acquired outside the tubes 2 and 4 of the rake. The tubes were displaced 12.7 airfoil wake. The resonance frequencies correlate very mm in the spanwise direction, while tube #1 was located well with the expected changes in the resonance 38.1 mm from the tunnel centerline. Due to the relatively frequency based on the tunnel flow temperature. large outer diameter of the transducer (2.36mm), which The experiments were conducted at Mach numbers of could adversely affect the measurement of the wake 0.2 to 0.65 and chord Reynolds numbers ranging from 4 American Institute of Aeronautics and Astronautics AIAA 99-0925 1.5x106 to 23.5x106 using the standard NACA 0012 section also provides guidelines for interpreting the airfoil and the NACA 0015 airfoil. controlled NACA 0015 data. Figure 6a presents the lift-angle-of-attack data of the 2.7 E x perimental uncertainty NACA 0012 for a range of Reynolds and Mach numbers. The incompressible experiments are conducted at At M=0.3, a strong R effect is evident only up to c conditions close to the limits of the wind tunnel operating R =3x106. At R ‡ 3x106 and M=0.3, the maximum lift c c envelope. For example, the very low temperatures (about as well as post-stall lift curves are similar. A small 100K), and Mach numbers (0.28) are close to the limits increase of the C -a slope with increasing R can be l c of the tunnel capability. Most of the baseline data were noticed. Strong compressibility effects exist at M=0.65, acquired with separated regions or shock-waves present compared to M=0.3, at R = 6x106. The C -a slope c l on parts of the airfoil. Tunnel wall interference will be increases from 0.101 at M=0.3 to 0.133 at M=0.65 while considered at a later stage, as the tunnel wall pressures the airfoil stalls at a lower by 7(cid:176) . Compressible, post-stall were recorded along with the airfoil and wake data. The lift is flat, in contradiction to the 30% lift drop observed movable walls of the test section were slightly diverged at the incompressible stall. Strong R effects on the c to reduce the effects of boundary layer growth. The airfoil stall can be viewed in the compressible data at following tables contain most of the relevant information R =3x106 and 6x106. c regarding experimental uncertainties. These values were Airfoil drag (C , filled symbols, solid lines) and calculated using – 3 standard deviations of the various d trailing edge pressures (C , empty symbols, broken experimental conditions and calculated parameters p,TE lines) atR =6x106 and at two Mach numbers, M=0.3 and (including repeated runs). All test instruments were c calibrated prior to use. M=0.65, are plotted in Fig. 6b. The data indicate that C is insensitive to the Mach number. The d,0 Item Uncertainty Full scale and incompressible drag increases very slowly, owing to the [% F.S.] condition boundary layer thickening as a result of the increased Slot 5 0.5 mm adverse pressure gradient, until C is reached. Then, l,max width the drag increases almost ten-fold while the lift drops by Static 0.3 300K 30% (Fig. 6a). The increase in drag is accompanied by a Temperature decrease in C (note the inverted scale on the right p,TE Static 1 70 psi, M=0.3 hand side of Fig. 6b), which indicates the tendency to Pressure & 100K R develop TE separation. Incompressible stall is c 2.0 M<0.3 Rc 1.0 M>0.3 accompanied by a change in the sign of Cp,TE, from positive to negative. The striking similarity between the M 0.5 M>0.3 F+ 1 2 drag and TE pressure plots demonstrates how well <cm > 25 local values correlated these two quantities are in incompressible turbulent flow that is characterized by TE separation. The The uncertainty regarding the calculated airfoil compressible drag divergence (Fig. 6b) occurs as a aerodynamic parameters is listed in the following table shock-wave develops over the airfoil, while the boundary (in absolute values and related to flow condition): layer downstream of the shock remains attached. The fully stalled controlled doubling of the compressible drag between a =5(cid:176) and 6(cid:176) attached is not accompanied by a similar increase in C , p,TE Cl 0.01 0.03 0.015 indicating that TE separation did not develop yet. The Cdp 0.001 0.003 0.0015 strong shock-wave that resides over the airfoil is the main Cd 0.001 0.003 0.002 cause for the drag rise. Additionally, a local and possibly unsteady separation bubble exists downstream of the 3. D i scussion of results shock. The severe drag divergence occurs only between 8(cid:176) and 10(cid:176) , as C becomes negative. The measured p,TE 3.1 N ACA 00 1 2 Airfoil and unsteady lift and drag are close to those presented in Refs. 15 and wake pressures 16. Some deviations in the stall and post stall data were The NACA 0012 airfoil which does not contain any identified, but are within the scatter of the data in Ref. 15. surface slots, and the dynamic wake rake (located at A further demonstration of compressibility effects on X/c=3.1 downstream of the airfoil mid-chord) were tested the stall process of the NACA 0012 airfoil are provided in order to validate the wake rake, to evaluate Reynolds by comparing airfoil pressure distributions, steady and and Mach number effects and to correlate the unsteady fluctuating wake pressures, and spectra at two angles of wake pressures and the airfoil pressure distributions. This 5 American Institute of Aeronautics and Astronautics AIAA 99-0925 + attack, corresponding to pre-stall and post-stall conditions peak. The compressible spectra is noisier at low F . One with approximately the same lift coefficients. possible source for the F+=0.7 peak, measured in the Fig. 7a presents airfoil surface pressure coefficients at wake of the M=0.65, a =6(cid:176) and Y/c=0.04, is the Cl=0.83 (a =8(cid:176) at M=0.3 and a =6(cid:176) at M=0.65). It unsteadiness of the shock-wave. A strong similarity demonstrates the distinctly different way that the exists between the spectra of the post-stall wakes, compressible lift is generated versus its incompressible regardless of the Mach number. An incompressible counterpart. While the incompressible pressure spectral peak is found around F+=0.25, while the distribution contains a sharp suction peak at the LE that is compressible spectra were monotonic. At greater F+, followed by a steep pressure recovery, the compressible both post-stall spectra show a continuous decay. C indicates that the flow turns supersonic very close to p the LE and ceases to accelerate at C » -2.1. A strong 4. C o ntrol of separation at compressible speeds p shock-wave is located between 0.2<X/c<0.3. 4.1 O v erview Downstream of X/c=0.4 the rate of pressure recovery is almost independent of the Mach number. The main challenges of active flow control at The wake data corresponding to C =0.83 is presented compressible speeds are: drag reduction, delay of drag l divergence, and suppression of unsteady aerodynamic in Fig. 7b. The narrow mean wake at M=0.3 is loading along with its possible interaction with the accompanied by a double hump distribution of the structure (“buffeting”). The generation of high-lift is less pressure fluctuations. This shape could be attributed to important than in low Mach numbers (take-off, landing the upper and lower surface boundary layers and also to a and loiter flight). Very little knowledge is available on sinuous wake mode. active control of flow separation at compressible speeds. At M=0.65 the mean wake pressure indicates a One encouraging result is reported in Ref. 17. It shows significantly larger drag in the compressible flow, while that by rotating a flat plate downstream of an airfoil the peak level of the wake pressure fluctuations is similar experiencing shock-induced separation, the extent of to its incompressible counterpart. The signature of the separated flow could be affected. The reduced forcing shock-wave appears in the upper section of the mean frequency was close to the airfoil buffeting frequency, compressible wake (0<Y/c<0.15). This secondary hump in the average wake pressure deficit, is accompanied by a F+» 0.1. These frequencies are too low for effective gradual decrease of the level of unsteadiness, indicating separation control. that the position and strength of the shock-wave are There are a number of possible obstacles* to the stationary. successful application of mixing enhancement using Fig. 8 presents a comparison of the pressure periodic excitation to transonic flows. First, it is not clear distributions of the NACA 0012 airfoil at post-stall how the excitation evolves in the supersonic flow. angles of attack, a =16(cid:176) at M=0.3 and a =10(cid:176) at M=0.65, Second, it is not known how vorticity fluctuations are and C » 0.87. The incompressible pressure distribution transported through shock-waves. Third, it is not clear l how downstream flow modification will affect the contains a sharp and narrow suction peak at the LE. upstream supersonic flow. Due to the many unknown Thereafter, it is almost flat from X/c=0.05 downstream, as the flow separates. The compressible C shows a aspects of the problem, it was decided to start exploring p active separation control in wholly incompressible flow supersonic flow region from the LE to X/c» 0.25 (the (M=0.28) and to gradually increase the Mach number, shock does not turn the flow subsonic) and a pressure while maintaining constant Reynolds number, wherever recovery exists all the way to the TE even though the possible. These flow conditions provided shock-free flow flow is separated. The effects of severe unsteadiness of (M=0.28), a weak shock upstream of the excitation slot the airfoil pressures on its time mean appearance are the (M=0.4) and a strong shock-wave downstream of the slot subject of an on-going investigation. (M=0.55). Control was applied from the slot located at The spectra of the wake pressures at selected points in X/c=10% and the excitation was of zero-mass-flux. the upper region of the wake are presented in Fig. 9. These data correspond to local maximum of p’/q (with 4.2 S eparation control over the straight respect to Y/c). NACA 0015 airfoil The frequency ordinate was normalized to present a Fig. 10 presents the lift data of the NACA0015 airfoil Strouhal number based on the airfoil chord and the free at Mach numbers of 0.28 and 0.4 and atR =12.7x106. + c stream velocity (F ). The pressure fluctuations were The baseline lift is presented by empty symbols and corrected according to the method described in section broken lines. Significant Mach number effects can be 2.6 of the paper and frequencies above 0.95f are not res seen even at the marginally compressible flow of M=0.4. presented. The data show that the spectra of the incompressible pre-stall a do not contain any distinct * For brevity we list here only three important obstacles. 6 American Institute of Aeronautics and Astronautics AIAA 99-0925 The baseline C -a slope increases by 7%, C drops Fig. 13a presents the steady and fluctuating wake l l,max from 1.31 to 1.17 and a drops from 12(cid:176) to 9(cid:176) . When pressure distributions measured at M=0.28, a =14(cid:176) and max R =12.7x106, with and without excitation. The baseline periodic excitation was applied to the airfoil at M=0.28, c with F+=2 and <cm >=30x10-5, stall was delayed from wake could not be scanned in its entirety, due to mechanical limitation on the travel of the wake rake. 12(cid:176) to 14(cid:176) and C increases from 1.31 to 1.47. When l,max Therefore it had to be extrapolated in order to estimate + excitation was applied at M=0.4, with F =2.1 and the total drag. When a linear extrapolation was used to <cm >=25x10-5 (Fig. 10), a max increases from 9(cid:176) to evaluate the baseline drag, the total drag coefficient was 11(cid:176) , C increases from 1.17 to 1.29 and post-stall lift about 0.18. When excitation was activated the drag was l,max became flat. The lift increment due to the excitation is reduced to 0.085. These values compare favorably with also plotted on Fig. 10. It shows that significant lift the measured reduction in form-drag, from 0.18 to 0.09, increment, due to the excitation persists as M is increased as shown in Fig. 11 for <cm >=30x10-5. The fluctuating from 0.28 to 0.4. One small difference could be observed part of the baseline wake is skewed and presents a much at a =8(cid:176) and M=0.4, where the lift decreases slightly higher fluctuation level on the upper side, corresponding below the baseline as the excitation was activated. to the separated flow on the upper surface of the airfoil. The baseline form-drag for M=0.4 (not shown) is A significant reduction of the wake unsteadiness was + higher than that of the M=0.28 data, as stall is obtained due to the excitation at F =2. An integration of approached. The small decrease in the controlled lift prior the fluctuating wake momentum, in the form ¥ to stall at M=0.4 (fig. 10), is accompanied by a small 1 (cid:242) p'dY (2) increase in form-drag. These detrimental effects were not qc-¥ observed at M<0.3. indicates a reduction of 34% in the controlled flow The effect of the excitation magnitude on the airfoil (without extrapolating the fluctuating part of the baseline lift and form-drag at M=0.28 is presented in Fig. 11. wake). Reproducing the effects found in the lowR experiments3 c Fig. 13b presents the baseline and controlled wake as well as the controlled flap data at high R 4, the lift c pressure spectra, measured at M=0.28, a =14(cid:176) , X/c=2.1 increases and the drag decreases proportionally to the and Y/c=0.2. This position was selected because it excitation <cm >. Saturation has not been reached at the resides where the baseline pressure fluctuations reach a available momentum coefficients. As indicated by the maximum and dP/dY is high. It also corresponds to the straight lines, the trends are almost logarithmic. The airfoil upper surface, where the blowing slot is located. increased Cl-<cm > slope for <cm > > 10x10-5 is The baseline spectra show some peaks between F+=0.3 probably a result of increased interference with the tunnel to 0.7. When excitation is applied the level of the wake walls. The decrease rate of form-drag is initially slower pressure fluctuations decreases over the entire frequency than the increase rate of the lift. range. The most significant reduction is in the frequency + Fig. 12a presents the baseline and controlled airfoil range F <0.9. The level of the pressure fluctuations at surface pressure distributions at a =14(cid:176) and M=0.28. The the excitation frequency, F+=2, is not significantly baseline Cp indicates that the flow separates at X/c=0.15. higher than the background noise in the controlled wake. When excitation is applied from the slot located at This indicates that the oscillatory momentum that was X/c=0.1, the suction peak became significantly stronger introduced through the excitation slot was transferred and an effective pressure recovery was restored, mainly to the mean flow and also to smaller scales that downstream of the slot. C becomes more positive, dissipate rapidly in the attached regions of the turbulent p,TE boundary layer. indicating the increased tendency of the flow to reattach. The global effect of the excitation frequency on the The flow at M=0.28 is entirely subsonic, since the critical unsteadiness of the wake, as expressed in C excursions, C (= -8.06) is not reached. Fig. 12b presents baseline d p can be evaluated from examining the amplitude and controlled pressure distributions at M=0.4 and distribution of the fundamental excitation frequency in a =11(cid:176) . The baseline flow separates at X/c» 0.2. When the entire wake. Fig. 13a also presents the phase-locked excitation is applied, a pressure distribution similar to the pressure fluctuations measured in the wake, one measured at M=0.28 is obtained. The only significant corresponding to the excitation frequency. It has the difference is that the upper surface flow, close to the LE, shape of the wake sinuous mode, which can be becomes supersonic as the excitation is activated interpreted as alternate sign vortex shedding every half (C =-3.66). Since the flow turns subsonic again p,crit cycle from the upper and lower surfaces of the airfoil. If upstream of the excitation slot, one should not expect one assumes the theoretical shape of the phase significant compressibility effects on the controlled flow. distribution corresponding to the sinuous mode18 and calculates the phase-locked drag excursions, it results in 7 American Institute of Aeronautics and Astronautics AIAA 99-0925 only 0.6% of the controlled drag, that is lower than the This could be a result of two possible mechanisms19. The baseline drag by 53%. The random wake fluctuations are excitation and the accompanied enhanced mixing thins assumed to average out and have no significant effect on the boundary layer upstream of the shock. In turn the C excursions. This points needs to be verified boundary layer accelerates and a stronger shock is d experimentally and numerically. required to turn the flow subsonic again. The stronger Data were also acquired at M=0.45 and M=0.5 at shock also causes immediate flow separation downstream R =12.7x106 and 19x106. Significant R effects were of it. The other mechanism is enhanced mixing c c identified in these mildly compressible flow conditions downstream of the shock. This allows the turbulent that make it difficult to identify changes in the boundary layer to close the separation bubble faster, effectiveness of the excitation. Another difficulty stems resulting in a stronger pressure recovery and an enhanced from the fact that the shock position varied but was capability to handle the stronger shock without always downstream of the fixed blowing slot (located at catastrophic separation. It is assumed that both X/c=0.1). These results will be presented elsewhere. Here mechanisms are presently active. Indeed, the shock we chose to present the results for M=0.55 at R =19x106 strength is stronger and its foot is less smeared. The c that contain strong compressibility effects and shock pressure distribution downstream of the shock indicates position closer to X/c=0.1, but still downstream of it. that a large unsteady separation bubble exists between Fig. 14 presents compressible baseline and controlled 0.2<X/c<0.6. A more favorable pressure recovery can be lift data. It shows that the baseline C for M=0.55 seen at X/c>0.6 and Cp,TE becomes more positive. l,max dropped to 1 at a =6(cid:176) from 1.31 at a =12(cid:176) for Clearly, the controlled flow at compressible speeds is max max M=0.28 (Fig. 11a) and 1.17 at 9(cid:176) for M=0.4 (also in Fig. significantly more complex than its incompressible 11a). Post-stall lift decreases relative to C , in counterpart. l,max The averaged and fluctuating wake pressures for contrast to the NACA 0012 data measured at M=0.65 M=0.55 and a =9(cid:176) are presented in Fig. 16. Here also, the andR =6x106, where post-stall lift remained fairly c mechanical limitations on the movement of the wake rake constant. Unlike the wholly incompressible flow, where prevented measurement of the entire baseline wake. The surface tangential oscillatory excitation did not have any width of the wake decreased significantly, while the detrimental effect, a small decrease in lift as a result of maximum wake deficit increased somewhat, with active activating the excitation can be observed at pre-stall excitation. This could be a result of a more stationary angles of attack, at M=0.55 (note the lift change due to wake. The wake fluctuating pressure also restored a more the excitation which is also plotted in Fig. 14). An symmetrical, double peak structure, due to the excitation. examination of the airfoil pressure distributions (not A major reduction in the wake unsteadiness can be shown) reveals that the reason for this performance observed on the upper side of the wake (i.e. Y/c>0). degradation is a secondary shock-wave that was created Fig. 17a presents the drag, Fig. 17b presents the at the excitation slot, upstream of the naturally occurring trailing edge, C , and Fig. 17c presents the shock-wave. At post stall a ’s the excitation maintains C p,TE l integrated wake unsteadiness as defined in Eq. 2. for the at about 1 while the baseline lift drops gradually to 0.85. compressible baseline and controlled data. The severe As mentioned before, high lift is seldom required at drag divergence starts at a >4(cid:176) , as supersonic flow turns transonic speeds, since the sizing of the wing is made to subsonic through a strong shock-wave. The drag meet take-off requirements. The important aspects of increases five fold (Fig. 17a) and the integrated wake active separation control at compressible speeds are delay unsteadiness increases six fold (Fig. 17c) between a =4(cid:176) the onset and alleviate the effects of buffet. and 8(cid:176) . C becomes negative only between a =6(cid:176) and Fig. 15 presents the baseline and controlled airfoil p,TE surface pressure distributions at post-stall angle of attack 7(cid:176) , indicating the development of TE separation. The (a =9(cid:176) ). The baseline C shows a supersonic flow region majority of the drag divergence, certainly up to 6(cid:176) , was p due to wave-drag and was not caused by TE flow near the LE which turns subsonic via a strong and separation. possibly unsteady shock-wave. The unsteadiness of the For a£ 8(cid:176) the excitation results in a minor increase in shock strength and position can be evaluated from drag (Fig. 17a) as well as in the wake unsteadiness (Fig. observing the smeared foot of the shock. The average 17c), and a small but detrimental effect on C (Fig. position of the shock foot is at X/c=0.15. The separated p,TE flow is manifested as a mild pressure recovery 17b) . These effects accompany the pre-stall reduction in downstream of the baseline shock position and negative the controlled lift, observed in Fig. 14. It remains to be C . As excitation was applied from the X/c=0.1 slot seen if these effects will disappear once excitation is p,TE introduced closer to or downstream of the shock-wave. (i.e. upstream of the averaged shock position and inside For a >8(cid:176) the excitation reduces the drag by about 11- the supersonic flow), a number of changes can be 17%. A similar decrease can be observed in the observed. The flow upstream of the shock accelerates. 8 American Institute of Aeronautics and Astronautics AIAA 99-0925 integrated wake unsteadiness and in the tendency of 5. C o nclusions C to increase due to the excitation. Again, these Active separation control was applied from the p,TE benefits were obtained when excitation was applied leading edge region of a straight NACA 0015 airfoil, upstream of the shock position, inside the supersonic initially in incompressible flow. Then the Mach number flow. The efficiency of the excitation, in terms of <cm >, was gradually increased to cover compressible flows at flight Reynolds numbers. seems to be lower than in the incompressible flow Strong Reynolds number effects were identified in conditions but it is not known what effects the passage the airfoils baseline performance at moderately through the shock-wave has on the vorticity of the compressible flow conditions and post stall angles of excitation. attack. It makes the identification of clear trends in the The spectra of the baseline and the controlled wake controlled data difficult. TheR effects weaken as the pressure fluctuations at a =9(cid:176) are presented in Fig. 18. It c Mach number increases and a stronger shock develops. demonstrates that the baseline wake at M=0.55 does not It was demonstrated, in accordance with low contain a distinctive shedding frequency. The controlled Reynolds number experiments, that incompressible wake shows a reduction in the level of wake pressure C can be increased by 15%, post-stall lift can be fluctuations, especially in the low frequency range l,max + increased by as much as 50% and post stall drag can be (F <1). When the level of the pressure fluctuations reduced by more than 50%. The controlled wake is also between 1 and 800 Hz (that includes the forcing steadier. The main application of the present control frequency) is calculated, a reduction of about 50% is method to these flow conditions is a reduction in the obtained (see also Fig. 16). The controlled data shows + number of elements and their complexity in high-lift two distinctive peaks. One at F =1 and the other at the + systems while maintaining the performance and steady excitation frequency (F =1.65). flow. These benefits should allow simpler and lighter In order to check the effect of the wake motion on the high-lift systems. time dependent drag excursion, we have calculated and The significant increase in lift and lift to drag ratio, plotted the wake pressure fluctuations, phase-locked to obtainable in incompressible speeds, should not be the excitation frequency, and the corresponding phase expected at compressible speeds. The global effect of the distribution of the fundamental excitation frequency + method is to accelerate the upstream boundary layer due (F =1.65, see Fig. 19). The amplitude and phase to the delay of boundary layer separation. In distributions are similar to the theoretically predicted compressible speeds this could lead to a stronger shock- distribution for a sinuous wake mode18. The phase is less wave that in turn causes a more severe separation, that is coherent on the upper edge of the wake, corresponding to harder to control. This process saturates the effect. Also the controlled flow, but the shape is close to the at increased Mach number the required frequencies theoretical prediction and to experiments. The phase- increase at least linearly with the flow speed while the locked Cd excursions are calculated using the mean available <cm > is inversely proportional to the dynamic wake profile with the amplitude and phase distributions pressure. In transonic flow, the method could be used to superimposed. We found that the C excursions are less d alleviate buffet rather than generating higher lift. than 0.3%, compared to 11% drag reduction which is When excitation was applied well upstream of the obtained due to the excitation. Here also, the reduction in shock-wave it had a detrimental effect on lift, drag and C excursions needs to be validated by unsteady airfoil d wake steadiness. This is due to the creation of a surface pressures and balance data. secondary shock-wave at the blowing slot. This effect is Fig 20 presents the improvement in the airfoil integral not present in low Mach number experiments. There, the parameters as the excitation magnitude, <cm >, is introduction of wall-tangential excitation, far upstream of increased, at M=0.55 and a =9(cid:176) . The data shows that the the boundary layer separation, resulted in a smaller relative lift to drag ratio increases by 35%, in an almost performance increment (when compared to excitation linear manner with the increase in <cm >. This trend is that was introduced immediately upstream of the highly desirable for the construction of a linear controller. separation), but did not result in absolute performance The wake gradually becomes steadier as the lift to drag degradation. ratio increases as a result of increased excitation It seems that the ability of similar <cm > levels to magnitude. The averaged pressure gradient downstream reduce drag and wake unsteadiness decreases as M of the active excitation slot increased by 50% and Cp,TE increases. But due to the great sensitivity of the blowing increased by 30% due to the excitation with F+=1.65 slot location with respect to the mean shock position, it and <cm >=0.015% (not shown). This indicates that the remains to be seen how the efficiency will vary once this parameter is better controlled. above benefits were obtained as a result of an increased The excitation became effective when it was tendency for flow reattachment. introduced only slightly upstream of the shock-wave, 9 American Institute of Aeronautics and Astronautics