Airfoil Separation Control with Plasma Actuators. Shawn Fleming

Transcription

1 Airfoil Separation Control with Plasma Actuators By Shawn Fleming Bachelor of Science in Mechanical Engineering Oklahoma State University Stillwater, OK, USA 2008 Submitted to the Faculty of the Graduate College of Oklahoma State University in partial fulfillment of the requirements for the Degree of MASTER OF SCIENCE May, 2010

2 Airfoil Separation Control with Plasma Actuators Thesis Approved: Dr. Jamey Jacob Thesis Advisor Dr. Andrew Arena Dr. David Lilley Dr. A. Gordon Emslie Dean of the Graduate College ii

3 TABLE OF CONTENTS Chapter Page 1 INTRODUCTION Importance of Problem Objective Methodology Thesis Outline Previous Work Low Speed Airfoils Boundary layer Characterization Active Flow Control Plasma Actuator Flow Control Experimental Set-Up Plasma Actuators PIV Measurements Bench-Top Testing Wind Tunnel Testing Results Actuator Development X-Foil Wind Tunnel Flow Control Tests iii

4 5 Discussion and Conclusions Discussion Conclusions Recommendations A High Re Testing 114 A.1 Wind Tunnel Setup A.2 Subsonic Wind Tunnel Results: La203a Performance B Input File and MATLAB Codes 125 B.1 WaLPT Algorithm Input File B.2 MATLAB Codes B.2.1 Mask Generation Code B.2.2 PIV Post-Processing BIBLIOGRAPHY 157 iv

5 Figure LIST OF FIGURES Page 1.1 Progression of Stall Across an Airfoil [24] Flight Reynolds-number Spectrum [21] Low-Reynolds-number airfoil performance [21] Laminar Separation Bubble Geometry [21] Overall Categorization of Flow Control Approaches [7] Passive and Active Flow Control Devices Schematic of the Typical Plasma Actuator Teflon Plasma Actuator with 1/2 inch Copper Electrodes Acrylic Plasma Actuator with 1/2 inch Copper Electrodes Alumina Plate Plasma Actuator with 1/2 inch Copper Electrodes LabView Block Diagram for 1 Channel Output LabView Block Diagram for 2 Channel Output Schematic of PIV Setup Schematic of bench-top Setup Liebeck La203a CAD Model Liebeck La203a Airfoil Wind Tunnel of Setup Plasma Actuator Parameters Plasma Actuator Benchmarking: Varying Frequency (Teflon with 1/4 in. Copper Electrodes) v

6 4.3 Plasma Actuator Benchmarking: Varying Modulation Frequency (Teflon with 1/4 in. Copper Electrodes) Velocity Vectors for Teflon with 1/4 in. Copper Electrodes Vorticity for Teflon with 1/4 in. Copper Electrodes Velocity Profile for Teflon with 1/4 in. Copper Electrodes Plasma Actuator Benchmarking: Varying peak-to-peak Voltage (Acrylic with 1/2 in. Copper Electrodes) Plasma Actuator Benchmarking: Varying Frequency (Acrylic with 1/2 in. Copper Electrodes) Velocity Vectors for Acrylic with 1/2 in. Copper Electrodes Vorticity for Acrylic with 1/2 in. Copper Electrodes Velocity Profile for Acrylic with 1/2 in. Copper Electrodes Plasma Actuator Benchmarking: Varying Frequency (Alumina with 1/2 in. Copper Electrodes) Velocity Vectors for Alumina with 1/2 in. Copper Electrodes Vorticity for Alumina with 1/2 in. Copper Electrodes Velocity Profile for Alumina with 1/2 in. Copper Electrodes Plasma Actuator Benchmarking: Varying Frequency (Teflon with 1/2 in. Copper Electrodes) Maximum Velocity Comparison for each Dielectric XFoil Separation Point Tracking Reynold s Number of 100,000 at an Angle of Attack of 10 Degrees XFoil Separation Point Tracking Reynold s Number of 650,000 at an Angle of Attack of 10 Degrees XFoil Separation Point Tracking Wind Tunnel Plasma Actuator Test Matrix vi

7 4.22 Actuators Off (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Actuators Off Reverse Flow Probability within the Flow Field Leading Edge Actuator, Constant Activation (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Leading Edge Actuator, Constant Activation Reverse Flow Probability within the Flow Field Leading edge Actuator, Pulsed Activation with an F + = 1 (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Leading edge Actuator, Pulsed Activation with an F + = 1 Reverse Flow Probability within the Flow Field Aft Actuator, Constant Activation (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Aft Actuator, Constant Activation Reverse Flow Probability within the Flow Field Aft Actuator, Pulsed Activation with an F + = 1 (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Aft Actuator, Pulsed Activation with an F + = 1 Reverse Flow Probability within the Flow Field Steady Activation on the LE and Aft Actuators (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Steady Activation on the LE and Aft Actuators Reverse Flow Probability within the Flow Field Pulsed Activation with an F + = 1 on the LE and Aft Actuators (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity vii

8 4.35 Pulsed Activation with an F + = 1 on the LE and Aft Actuators Reverse Flow Probability within the Flow Field Out-of-Phase, Pulsed Activation with an F + = 1 on the LE and Aft Actuators (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Out-of-Phase, Pulsed Activation with an F + = 1 on the LE and Aft Actuators Reverse Flow Probability within the Flow Field Velocity Profiles for 50,000 Reynolds Number Cases: Solid Black, No Control; Solid red, LE Steady; Dashed Red, LE Pulsed F + = 1; Solid Blue, Aft Steady; Dashed Blue, Aft Pulsed F + = 1; Solid Green, Both Steady; Dashed Green, Both Pulsed F + = 1 in-phase Velocity Profile Comparison of In-Phase and Out-of-Phase Actuator Activation at 50,000 Reynolds Number: Solid Black, No Control; Solid Red, Both Pulsed F + = 1 in-phasae; Solid Blue, Both Pulsed F + = 1 out-of-phase Actuators Off (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Actuators Off Reverse Flow Probability within the Flow Field Leading edge Actuator, Constant Activation (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Leading edge Actuator, Constant Activation Reverse Flow Probability within the Flow Field Leading edge Actuator, Pulsed Activation with an F + = 1 (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Leading edge Actuator, Pulsed Activation with an F + = 1 Reverse Flow Probability within the Flow Field viii

9 4.46 Leading edge Actuator, Pulsed Activation with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Leading edge Actuator, Pulsed Activation with F + = Reverse Flow Probability within the Flow Field Leading edge Actuator, Pulsed Activation with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Leading edge Actuator, Pulsed Activation with F + = Reverse Flow Probability within the Flow Field Steady Activation on the LE and Aft Actuators (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Steady Activation on the LE and Aft Actuators Reverse Flow Probability within the Flow Field Pulsed Activation on the LE and Aft Actuators with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Pulsed Activation on the LE and Aft Actuators with F + = Reverse Flow Probability within the Flow Field Pulsed Activation on the LE and Aft Actuators with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Pulsed Activation on the LE and Aft Actuators with F + = Reverse Flow Probability within the Flow Field Velocity Profiles for 75,000 Reynolds Number Cases: Solid Black, No Control; Solid Red, LE Steady; Dashed Red, LE Pulsed F + = 1; Dash- Dot Red, LE Pulsed F + = 0.198; Dotted Red, LE Pulsed F + = 0.297; Solid Blue, Both Steady; Dashed Blue, Both Pulsed F + = 0.198; Dash- Dot Blue, Both Pulsed F + = ix

10 4.57 Actuators Off (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Actuators Off Reverse Flow Probability within the Flow Field Leading edge Actuator, Constant Activation (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Leading edge Actuator, Constant Activation Reverse Flow Probability within the Flow Field Leading edge Actuator, Pulsed Activation with an F + = 1 (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Leading edge Actuator, Pulsed Activation with an F + = 1 Reverse Flow Probability within the Flow Field Leading edge Actuator, Pulsed Activation with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Leading edge Actuator, Pulsed Activation with F + = Reverse Flow Probability within the Flow Field Leading edge Actuator, Pulsed Activation with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Leading edge Actuator, Pulsed Activation with F + = Reverse Flow Probability within the Flow Field Steady Activation on the LE and Aft Actuators (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Steady Activation on the LE and Aft Actuators Reverse Flow Probability within the Flow Field Pulsed Activation on the LE and Aft Actuators with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity x

11 4.70 Pulsed Activation on the LE and Aft Actuators with F + = Reverse Flow Probability within the Flow Field Pulsed Activation on the LE and Aft Actuators with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Pulsed Activation on the LE and Aft Actuators with F + = Reverse Flow Probability within the Flow Field Velocity Profiles for 100,000 Reynolds Number Cases: Solid Black, No Control; Solid Red, LE Steady; Dashed Red, LE Pulsed F + = 1; Dash- Dot Red, LE Pulsed F + = 0.148; Dotted Red, LE Pulsed F + = 0.222; Solid Blue, Both Steady; Dashed Blue, Both Pulsed F + = 0.148; Dash- Dot Blue, Both Pulsed F + = δ and θ Vs. x/c for 50,000 Reynolds Number, Actuators Off δ and θ Vs. x/c for 50,000 Reynolds Number, Aft Actuator, Constant Activation δ and θ Vs. x/c for 50,000 Reynolds Number, Leading Edge Actuator, Constant Activation δ and θ Vs. x/c for 75,000 Reynolds Number, Constant Activation on the LE and Aft Actuators δ and θ Vs. x/c for 100,000 Reynolds Number, Pulsed Activation on the LE and Aft Actuators with F + = F + Vs. Reynold Number for Pulsed Activation Tests A.1 La203a Large Wind Tunnel Wing CAD Model A.2 La203a Large Wind Tunnel Wing A.3 OSU Large Low-Speed Wind Tunnel xi

12 A.4 OSU Wind Tunnel Test Section with Lift/Drag Balances and Pitot- Static Tube for Wake Surveys A.5 OSU Wind Tunnel Test Section with Lift/Drag Balances and Pitot- Static Tube for Horizontal Sweeps A.6 Manometer Bank A.7 La203a Airfoil Performance Curves (taken from Liebeck [10]) A.8 La203a Experimental Coefficient of Pressure Data for Different Angles of Attack Ranging from -6 degrees to 16 Degrees A.9 Experimental, Computational, and Reference for the Lift Curve of a La203a Airfoil A.10 Comparison of the Multiple Lift Curves at Several Different Reynold s Number xii

13 CHAPTER 1 INTRODUCTION 1.1 Importance of Problem The main purpose of wings on aircraft is to produce lift. Wings produce lift by changing their angle of attack, but there is a point when that angle of attack becomes too large for the wing to continue to produce lift. At that time the air flow over the top of the wing starts to separate and become detached from the suction side of the wing. This angle of attack that causes separation over the entire top surface of the wing is called the stall angle or α stall. Generic stall pattern to general wing platforms is a much easier thing to discuss. Certain stall patterns can make it easier for a pilot to recover from a stall. Straight wings tend to stall from the centerline outward, while highly tapered wings tend to stall from the tips inward. When the stall starts at the inboard section of the straight wings, it generates less downwash on the tail, allowing for an easier nose down behavior. Also when the stall starts at the inboard side of the ailerons the pilot has better ability to maintain roll control. When the stall starts at the outboard part of the wing during a turn, one wing is moving faster than the other. The wing stalls in the direction of the slower wing and will fall out of the turn in that direction. The plane then enters a spin and the pilot has to recover from a stall and a spin. So the real question is What makes an airfoil stall? At α stall the flow detaches from the upper surface of the airfoil. The pressure in this separated region can no longer maintain suction, which leads to a loss in lift. The lower surface of an airfoil will not stall because the free stream flow is along the lower length of the airfoil 1

14 because it has a favorable pressure gradient. At the point of separation, there is a loss in lift and an increase in drag. There is also a change in momentum due to the loss in suction along the top surface. The progression of stall, as an airfoil is taken to higher angles of attack, can be seen in Fig. 1.1[24]. A favorable pressure gradient is where there is a decreased pressure in the direction of the flow, allowing the flow to accelerate. An adverse pressure gradient is an increase in pressure against the flow direction, making the flow decelerate. Figure 1.1: Progression of Stall Across an Airfoil [24] In general stall characteristics for an airfoil can be characterized in part by the leading edge (LE) radius and the thickness of the airfoil. An airfoil is considered fat if it has a rounded LE and a thick cross-section (t/c > 14%). Fat airfoils stall gently from the trailing edge (TE) and stall progresses forward. The moment changes slightly forward on a fat airfoil due to a loss in lift at the TE promoting a nose down behavior. A moderate airfoil has a thinner thickness (6% < t/c < 14%) and unlike a fat airfoil, a moderate airfoil stalls abruptly from the LE. Moderately thick airfoils create a separation bubble at the LE. The bubble reattaches and then collapses. This collapse is what marks the stall of a moderate airfoil. Stall happens suddenly and over the entire upper surface of the airfoil. There is a large shift in moment for a moderate airfoil leading to a large nose down behavior. A thin airfoil has a thickness less than that of the other two types of airfoils (t/c < 6%) and sharp LE. The thin airfoil generates a separation bubble near the LE then reattaches and grows toward the TE. The moment shifts slightly for a thin airfoil 2

15 and causes a moderate nose down behavior. Now if we consider a wing the same question still applies, What makes a wing stall? The cause is still much the same. The wing stalls when a significant portion of airfoil sections stalls, where each wing section is just an airfoil. C LMax for the wing is not achieved until almost all of the wing is stalled. Once the wing has reached its C LMax most of the sections are well past their c LMax for sections, C LMax. At this point the wing has started to lose lift. Stall is inevitable when trying to max out lift. Part of the problem is that once a wing stalls, there is a large decrease in lift, a large increase in drag. This decrease in lift, and increase in drag lowers the L/D ratio of the wing and the aircraft as a whole. The problem is how can we increase the operational range of airfoils so that we can take wings to higher angles of attack before the flow separates from the upper surface? In this thesis, we performed a series of experiments where we looked at using plasma actuators to add energy into a fluid flow for the purpose of reattaching flow in a separated region on a highly cambered wing. Several different dielectrics were tested on a set of bench-top tests to show how different dielectric materials can produce higher or lower velocity plasma jets. The materials used were Teflon, acrylic, and alumina plate. Two different widths of copper tape were examined when testing the Teflon dielectric to see if there was a difference in plasma jet velocities between the two different size electrodes. The plasma actuators were configured in such a way that two electrodes were placed on either side of a dielectric material with a small gap between one edge of one electrode and another edge of the second electrode. One electrode was exposed to air. The other was placed on the opposite side of the dielectric so that it was not exposed to air. 3

16 1.2 Objective In certain situations it is desirable to increase your coefficient of lift or alleviate drag by controlling separation on an airfoil. In this work, plasma actuators were used to control the separation of an airfoil at high angles of attack. The purpose of this thesis was to examine the use of various configurations of single dielectric barrier discharge (SDBD) or plasma actuators for control of stall on low speed airfoils. A plasma actuator consists of two electrodes separated by a dielectric medium, one exposed on the top surface of the dielectric material and one embedded on the underside of the dielectric material, and a high voltage low current signal input to the electrodes. The dielectric materials used in this investigation were: Teflon, acrylic and alumina. The electrodes were a 1/2 inch and 1/4 inch copper tape. There were two objectives examined in this investigation. The first objective was to examine how different geometries effect each of the three different dielectrics in bench-top testing. These results were implemented in further wind tunnel testing. The second objective was to demonstrate separation control on an airfoil in a wind tunnel at a high angle of attack using a plasma actuator near the LE of the airfoil and another actuator located near the mid-chord to reattach a separated flow. Various configurations were tried during this portion of the investigation: One with just the LE actuator on; one with the downstream actuator on; and then with both actuators on, both synchronously and asynchronously. These three configuration were looked at for an input signal giving a steady plasma activation and a pulsed plasma activation. For the pulsed activation a variety of frequencies were examined. 1.3 Methodology Several parameters were used to determine how each of the dielectrics would behave, parameters such as operating frequency, modulation frequency, and max peak-to- 4

17 peak voltage. An optimum frequency was found by stepping the frequency up from 3,000 Hz to 15,000 Hz and keeping the duty cycle at 100% to maintain a plasma jet in steady operation. The modulation frequency and peak-to-peak voltage were held constant. The measurements to determine velocity were performed using particle image velocimetry (PIV). Once a maximum velocity was obtained by varying the operating frequency, the modulation frequency was varied and the duty cycle was then held constant at 50%. The modulation frequency was varied from 5 Hz to 1,000 Hz while holding the operating frequency at 15,000 Hz, and the peak-to-peak voltage constant. The maximum velocity was again found using PIV to determine the optimum modulating frequency. The last parameter investigated was the maximum peak-to-peak voltage for a given material. Once the best parameters were found using one material, the same procedure was used to determine the optimum operating frequency and the maximum peak-to-peak voltages for the other two materials. Using the results from the bench-top test, we ran a series of experiments in a wind tunnel to see how the plasma actuators performed in a freestream flow. First the stall angle of attack was determined. The test wing was positioned at an angle of attack of 22 degrees, and a Reynolds number of 50,000, 75,000, 100,000 and 150,000. Two plasma actuators were attached to the upper surface of the airfoil, one near the leading edge and another at the 40% chord. These actuators were controlled much in the same fashion as the bench-top tests. The wind tunnel experiments were run both in a steady state, or with the duty cycle at 100%, and unsteady or pulsed state, with the duty cycle at 50%. The plasma frequency was held constant at 10,000 Hz and for the steady activation the duty cycle was held at 100%. During the steady state activation, one run was performed with the rear actuator on but the front actuator off. Then a second run was performed with the front actuator on and the rear actuator off. A third run was performed in steady state with both actuators on at the same time. Several runs were performed in an pulsed activation with both actuators on at 5

18 a phase angle of 0 degrees and 180 degrees. During the pulsed activation the duty cycle was set to 50% and a variety of modulation frequencies. 1.4 Thesis Outline This thesis is arranged as follows. In Chapter 2 an investigation was also done on the previous work that has been done with plasma actuators for aerodynamic uses. Chapter 3 goes into depth of how the apparatus and diagnostic experiments were setup and performed throughout this investigation using plasma actuators to control separation on an airfoil. Chapter 4 shows the results for all of the bench-top, computational, and wind tunnel tests performed. Chapter 5 discusses the conclusions reached and a short discussion on what other future work should be investigated. 6

19 CHAPTER 2 Previous Work In Chapter 2 an investigation was also performed on previous work that has been done with plasma actuators for aerodynamic uses. Research in Low speed airfoils was done to provide knowledge on what to expect from the wind tunnel tests performed at lower Reynolds numbers. Boundary layer characteristics were examined here to provide an incite on how flow behaves along a given surface. Research was then preformed to investigate the different methods of flow control and how these control methods are used. Further literature review was done to examine the different approaches to the method of flow control that is described within this thesis. 2.1 Low Speed Airfoils There are a wide range of Reynolds numbers that todays airfoils operate in, this range can be seen in Fig. 2.1[21]. A particularly interesting area of flight is the low Reynolds number flight regime. Lissaman [21] discusses this particular flight region extensively. The choice of the right airfoils for a particular mission has always seemed to be a bit mystical, but there has always been prerequisites for choosing an airfoil, that has satisfactory performance across the flight envelope. The shape of airfoils is very important, and the teardrop or paisley motif-like shapes have a universal aesthetic appeal. There is an ideal shape for an airfoil. It depends on the size and speed of the wing. This dependence is called the scale effect. The scale effect was first observed in the 30s. It was seen that the excellent qualities of an insect or bird wing doesn t scale up when you try to use the same 7

20 Figure 2.1: Flight Reynolds-number Spectrum [21] shape for an airplane wing and vise versa. This all goes back to the fact that an airfoil is designed dependent on the size and speed at which the wing is traveling, so a different size leads to a different shape. Lissaman states, This scale effect is characterized by the chord Reynolds number, R, defined by R = V c/ν, where V is the flight speed, c is the chord, and ν is the kinematic viscosity of the fluid in which the airfoil is operating. This Reynolds number is of importance because it quantifies two important effects to airfoil behavior, internal (fluid momentum) and viscous (fluid stickiness) effects. The viscous effect tends to have more effect on airfoil behavior because it determines how much drag and the maximum lift an airfoil is capable of producing. Lift and drag are usually described in a non-dimensional coefficient form C L and C D respectively. Lift and drag coefficients are defined as L/qc and D/qc, respectively, where L is lift per unit span and D is the drag per unit span, q is the dynamic pressure of the fluid (with the dynamic pressure q = 1/2ρV 2, ρ is the density of the fluid), and c is the chord. C L and C D are both dependent on the Reynolds number and the angle of attack at which the airfoil is operating. Lately the development of small flying vehicles has brought us into the low Reynolds number flight regime. Being at Reynolds number of half a million and below leads 8

21 to the need for a new low Reynolds number airfoil that was previously not needed. The operating range for such an airfoil is usually between sea level and 30 km. Usually the function of an airfoil is to produce lift that is perpendicular to the flight direction. Drag is the bi-product connected to the force needed to propel the lifting surface. A parameter is used to describe the effectiveness of an airfoil. This parameter is the lift-to-drag ratio C L /C D, where C L /C D max is a measure of the airfoil s max performance. Designing airfoil to have a C L /C D max that occurs at a high C L, this minimizes the size of the lifting surface. While operating at the lower Reynolds numbers, viscous effects are large and result in a high drag and limit the max lift that can be produced. But at higher Reynolds numbers, C L /C D improves, and the viscous effects have less of an impact on airfoil performance. There is a Reynolds number, where the performance of the airfoil changes. The critical Reynolds number is about 70,000. Smooth airfoils have the greatest benefit from the critical Reynolds number while rough turbulated airfoils do not gain the same performance increase. At the low Reynolds numbers rough airfoils actually benefit from the surface discontinuities. These surface discontinuities actually promote attachment. Smooth airfoils do not have the same benefit that the rough ones do in this low Reynolds numbers flows. As the Reynolds number increases, the smoothness of the airfoil begins to become more important. This is apparent by examining Fig. 2.2[21]. While discussing the fundamental fluid mechanics, Lissaman states, All airfoils have region of lower-than-static pressure. For most airfoils, this lower-than-static region is on the suction surface. For a symmetrical non-lifting airfoil that does not have this types of suction surface the thickness alone induces a lower pressure and accelerates the flow over the airfoil. Once the flow has moved past this low pressure region it must slow down to about freestream velocity at the TE. This slow down region is called an adverse pressure gradient or pressure recovery region. In low Reynolds number flows the boundary layer maybe be laminar and attached, but might not be 9

22 Figure 2.2: Low-Reynolds-number airfoil performance [21] able to handle an adverse pressure gradient, therefore laminar flow has a tendency to have poor separation resistance. Often times when a laminar flow separates there is a rapid transition to a turbulent flow. In a turbulent flow, the increased entrainment leads to the reattachment of the separated flow as a turbulent boundary layer. This detachment of the laminar boundary layer and reattachment as a turbulent boundary layer is called a laminar separation bubble. The structure of a laminar separation bubble can be see in Fig. 2.3[21]. When a laminar separation bubble forms, the flow separates from the surface at a near constant separation angle. At this point the transition occurs and the turbulence starts to develop. If the entrainment due to the turbulent flow is strong enough the flow will reattach to the surface and a the turbulent boundary layer rearranges itself to a normal turbulent profile. At Reynolds numbers greater than Re c, reattachment can occur creating a laminar separation bubble. During this occurrence, it is the airfoil characteristics that determine the type of laminar bubble. The bubble that is formed can be either long or short. At Reynolds numbers around 100,000, long bubbles can exist and tend 10

23 Figure 2.3: Laminar Separation Bubble Geometry [21] to extend over about 20-30% of the airfoil s lifting surface. With this much of the lifting surface being covered with a laminar bubble, it greatly changes the pressure distribution over the surface, basically changing the airfoil s shape and performance. At Reynolds numbers greater than 100,000, short laminar bubbles tend to form instead of long laminar bubbles. A short bubble s length is usually only a few percent and therefore does not tend to change the pressure distribution. The short laminar bubble usually represents the transition-forcing mechanism. As the angle of attack is increased, the airfoil requires a much greater pressure recovery. Because of the need for greater pressure recovery, a short bubble can burst, and at this point the short bubble becomes a long one. A sudden stall results from the severe loss in airfoil performance. At Reynolds numbers of about 200,000, the laminar bubble can be avoided. Avoiding the laminar bubble is possible because the transition point happens far enough upstream from the adverse pressure gradient that the bubble is avoided. The transition actually occurs in a turbulent boundary layer that can withstand the adverse 11

24 pressure gradient. When Reynolds numbers approach the 500,000 range airfoil performance improves even more. The laminar separation bubble is not solely linked to the chord-wise Reynolds number, but is also influenced by the local boundary layer Reynolds number. This local boundary layer Reynolds number is associated with the region where the pressure recovery starts. However, if an adverse pressure gradient is severe enough and close enough to the leading edge, a bubble like characteristic can still be seen. These occurrences can even happen at Reynolds numbers of a few million. This is usually the case with a thin, small nose radius airfoil. 2.2 Boundary layer Characterization Boundary layers are characterized by several parameters, including boundary layer thickness, δ, displacement thickness, δ, momentum thickness, θ, and shape factor H. According to Munson [25], δ represents the outward displacement of the streamlines caused by the viscous effects on the plate. The momentum thickness is the height of the freestream flow needed to compensate for lack of momentum flux inside the boundary layer because of the shear force near the surface. The shape factor is used to determine what type of flow you are in. These parameters are defined as such: δ = h 0 (1 u Ue ) dy (2.1) θ = h 0 u U e (1 u Ue ) dy (2.2) H = δ θ (2.3) where u is the stream-wise velocity, U e is the freestream velocity, y is the height above the surface, and h is a large distance away from the surface. The thickness of a laminar boundary layer is predicted by theory as: 12

25 δ x = 5.0 (2.4) Rex where x is the distance from the leading edge and Re x = U e x/ν. So it is seen that δ is proportional to x. 2.3 Active Flow Control Flow control can be considered anything that is done to alter the flow to a more desirable behavior. The flow is often altered by the addition of mass, momentum, energy, vorticity, or even actively changing the shape of the surface the flow is moving over. Flow control can be broken up into two different types: active or passive, based on weather the mechanism is actively adding energy to the flow or not. The difference between these two classifications depends on several factors: whether energy is added, whether the flow control is a steady control or unsteady, or if the system can be modified after it is built. Active flow control can be either steady or unsteady but some form of external energy source, whether it is electrical or mechanical, is needed to operate the device. On the contrary, passive flow control requires no external energy added. An overall view of flow control can be seen in Fig. 2.4[7]. Fig. 2.5 is a further classification of flow control broken up into the types of devices that would be found under each of the two flow control categories. According to Gad-el-Hak [3], passive devices do not need the addition of external energy to work but they often come with an associated penalty to the amount of drag they produce. The drag increase is caused because passive devices often trip the flow intentionally allowing for a transition between laminar flow (LF) to turbulent flow. The aim of passive devices is to make this transition in flow upstream of the natural laminar flow (NLF) separation point. Devices like boundary-layer fences are used to help prevent the separation of flow at the tips of swept wings. Vortex generators are used by placing them on a body in order to raise the energy and momentum of the 13

26 Figure 2.4: Overall Categorization of Flow Control Approaches [7] Figure 2.5: Passive and Active Flow Control Devices 14

27 flow near the surface. Other means of achieving a passive flow control device is to change the geometric shape of the body or fabricate features into the body to achieve the desired flow characteristics. Active devices use some form of energy or fluid to add energy or momentum to the flow. Active flow control devices expend energy. For them to be truly useful, the energy gained from the postponed separation needs to exceed the amount of energy expended. Unlike passive flow control devices, active flow control does not suffer from the same drag penalty. There are three types of active flow control devices: Micro Electric Mechanical System (MEMS), Mass Flux (MF), and Zero Net Mass Flux (ZNMF). Gad-el-Hak, MEMS are devices that require a moving device or wire to induce a change in the boundary layer to cause a transition from laminar to turbulent flow. Many investigations have been performed on MEMS devices using flaps. There has been proof that you can manipulate a separated flow so that reattachment can improve a flight region. MF devices are designed to remove or add mass to the flow field. Suction and blowing are examples of MF devices. Both chord and span wise laminar flow control (LFC) can greatly improve with the addition of suction and blowing devices [1]. These improvements have been seen in both wind tunnel testing, as well as flight tests. The last type of active flow control devices are ZNMF. ZNMF devices, like synthetic jets, remove some fluid from the flow and then that same fluid is injected back into the flow at a higher energy level and momentum. The synthetic jets of Glezer et al. [2] are unique flow control devices because the jets are formed entirely from the working fluid. Another ZNMF flow control device that is being investigated include plasma actuators. Unlike synthetic jets, plasma actuators energize a region of flow and with this added energy the momentum of the flow is increased. 15

28 2.4 Plasma Actuator Flow Control Single Dielectric Barrier Discharge (SDBD) plasma actuators generate a body force due to the plasma that is generated. The plasma is generated along the electrode interface, when a low-current, high-voltage, and high-frequency AC signals are sent to an exposed electrode, thereby ionizing the surrounding air. The ionized air then creates a body force on its surroundings, inducing a near wall plasma jet. Investigations have shown that the usefulness of the plasma actuator depends on the type of dielectric, the electrode width, the gap between the exposed and embedded electrode, and the input signal. A schematic of a plasma actuator can be see in Fig Figure 2.6: Schematic of the Typical Plasma Actuator Bolitho [5] investigated what the effects of input power and actuator geometry would have on the type of plasma jet or the vortex generation. bench-top and wind tunnel testing was done to see what the different effects would have in a static test case as well as with low Reynolds number flows. The purpose of the bench-top testing was to see what types of plasma jets could be produced and at what angle. The benchtop tests were run by changing the input power, operating frequencies, duty cycle, spacing between the exposed electrodes, and the modulation frequency. The wind tunnel tests focused on changing the duty cycle, modulation frequency, and sideslip angle. 16

29 During bench-top testing Bolitho showed that the strength of the plasma jet is proportional to the input power. By varying the input power to the exposed electrodes asymmetrically, an angled plasma jet can be varied over a 180 degree range. Bolitho also showed that the strength of the plasma actuator can be controlled by varying the operating frequency, a similar level of plasma control to that of varying the input power. An adverse reaction to lowering the operating frequency to one side of the plasma actuator was that the momentum decreased linearly. When investigating the effects of varying duty cycle, a degree of control over the plasma jet angle could be achieved. Bolitho experimented with asymmetric duty cycle by leaving one of the exposed electrodes at a duty cycle of 50% and varying the other electrode between 20% and 50%. The varying duty cycle allowed for the angle to vary between a span of -90% to 90%. Varying the modulation frequency was investigated by varying the frequency to each of the electrodes simultaneously to see what effects this would have on the types of plasma jet produced. At the lower frequencies, two independent jets were seen near the wall in either direction. As the frequency was increased, vortices were formed by the exposed electrodes. As the frequency neared the critical frequency the vortices that were produced impinged on each other and nearly canceled each other out. Once past the critical frequency, the plasma jet created resembles that of a plasma jet under steady operation. During construction of the plasma actuators the spacing of the two exposed electrodes gave each of the actuators a unique critical frequency. Bolitho showed that modulation frequency does not affect the amount of momentum induced, the amount of momentum induced increases as the spacing between the exposed electrodes is increased. Bolitho varied the duty cycle, the effects of sideslip angle, and the wind tunnel speed. During the varying duty cycle investigation, the focus was to see how much vorticity can be created. One side of the actuator was kept at a duty cycle of 50% while the other side was allowed to vary between 10% and 50%. The vorticity did 17

30 not significantly change with duty cycle, but there were changes in the location and size of the vortices. When investigating the effects of sideslip, the maximum vorticity increased for all cases when the duty cycle was between 10% and 50% and the sideslip angle was nonzero. The effectiveness of the vectoring actuators was investigated as speed was increased. During both of the two cases that were being investigated, that the effectiveness of the jet actuators diminished. Öztürk [6] investigated the use of plasma actuators as micro-thrusters. These thrusters were investigated by doing both bench-top and wind tunnel testing. benchtop testing was done by varying several parameters: inner diameter, forcing frequency, and duty cycle. Wind tunnel testing was done by varying the wind tunnel speed. During the bench-top testing, different velocities were achieved by varying the inner diameter of the thruster. There were three different size inner diameters used: cm, cm, and 1.27 cm. That the 1.27 cm had the largest velocity and produced the maximum amount of thrust. The maximum thrust decreased as the diameter of the thruster decreases. The thruster with an inner diameter of cm had a larger velocity than the one with the thruster with an inner diameter of cm. Further investigation was done varying the duty cycle for each of the three plasma thrusters at different forcing frequency. The velocities were affected by the change in duty cycle and forcing frequency. As the duty cycle increased, the velocity also increased, but the opposite was true with forcing frequency. The peak thrust, maximum velocity and average velocity for each of the three plasma thrusters was found to be at all different values. The velocity profiles of each of the smaller diameter plasma thrusters were seen to be practically parabolic with a maximum velocity occurring near the centerline. As the inner diameter of the plasma thruster was increased, the centerline velocity decreased and the velocity near the wall was seen to increase. Öztürk then conducted wind tunnel tests on the plasma thrusters by varying the 18

31 wind tunnel speeds. Three different wind tunnel speeds were used to test the plasma thrusters: 0.62 m/s, 1.28 m/s and 2.32 m/s. The cm diameter thruster showed the best results as the tunnel speed was increased from 0.62 m/s to 2.32 m/s. The effects started to decrease as the speed got higher and the velocity profiles became harder and harder to distinguish. The larger thrusters were seen to have a smaller effect especially as the tunnel speed increased to 2.32 m/s. Öztürk also looked at jet vectoring plasma actuators. The jet vectoring plasma actuators were investigated by doing wind tunnel testing. These tests were done by varying duty cycle and wind tunnel speed. Many of these tests used and expanded on the work done by Bolitho. No quiescent measurements and the only part of the input signal that was being varied was the duty cycle throughout all the tests. Duty cycle was the only parameter investigated at because it has the greatest sensitivity and has the greatest effect on the plasma jet vectoring angle. During these tests a maximum vorticity was seen when one exposed electrode had a duty cycle of 50% and the other electrode had a duty cycle of 0%. These jets formed along the wall and were considered to be linear cases. For vectoring cases a duty cycle of 50% was placed on one electrode and a varying duty cycle between 20% and 40% was placed on the other. Lower vorticity was observed from these tests because of the asymmetric plasma strength on each of the electrodes. Jet vectoring plasma actuators were placed on a NACA 0012 in the chord-wise direction and placed at an angle of attack of 10 degrees. Plasma jet vectoring actuators proved to be effective in the controlling of separation at low tunnel speeds around 0.62 m/s. Three cases were run using three different duty cycle configurations: 0%/50%, 30%/50% and 50%/50% for the left and right electrodes respectively. For all three cases, appreciable separation control was achieved. Öztürk also investigated the effect of pulsing the actuator at a duty cycle of 50% on both electrodes. For the three cases stated above, the strongest vortex was generated when the electrodes are 19

32 pulsed at a 50%/50%, followed by 30%/50% then 0%/50%. The effectiveness of a jet vectoring actuator decreases with an increase in tunnel speed. At higher tunnel speeds a simple linear actuator is a better choice for flow control. He et al. [8] did experiments using weakly ionized plasma actuators to help with flow control separation on a wing to try and replace the leading edge slats and trailing edge flaps. A SDBD plasma actuator was set up to make an array of actuators at the leading edge and trailing edge. Setting up the plasma actuators this way would effectively eliminate hinge gaps that are created by the moving control surfaces. Hinge gaps directly affect the drag component in viscous drag calculation on a given wing. Hinge gaps are also a large source of radar signal reflection. Removing the hinge gap would reduce the radar signature of a given wing, making a hinge-less wing more desirable in many military applications. The benefit of plasma actuators would come from being able to tile a generic wing to focus on regions of separation, to be able to control aerodynamic forces produced within these regions, and by the wing itself. Being able to control these forces would allow us to be able to change the wing s aerodynamic performance to suit various flight conditions. The SDBD plasma actuators used were made up of two electrodes separated by a dielectric material, one electrode exposed to air and the other being fully embedded in the dielectric material. A high voltage AC input was supplied to the electrodes, and when the input amplitude was large enough, the air ionized. The ionization begins at the edge of the exposed electrode and extends over the region that is covered by the embedded electrode. The ionized air creates a body force which is given by f b = ( ) ɛ0 λ 2 D ϕe (2.5) where ϕ is the electric potential, E is the electric field, λ D is the Debye field, and ɛ 0 is the permittivity of air. The body force can be manipulated by changing the arrangement of the electrode and dielectric. This way a wide verity of arrangements 20

33 can be made to handle different situations. These SDBD plasma actuators were investigated in both a quasi-steady or unsteady activation. During steady activation the frequency used was well above the fluid response frequency. With the activation frequency above the fluid frequency there was a constant body force that was sensed by the fluid. During unsteady activation a driving frequency was switched on and off to excite the region of instabilities in the fluid flow. By activating the plasma actuators in an unsteady fashion, the power consumed by the actuator was reduced. This investigation looked at using a 10% duty cycle that showed a 90% reduction in power consumption over the steady operation. The unsteady case also showed a better overall flow control. These experiments showed that a SDBD plasma actuator, located on the leading edge of a NACA 0015 wing, was able to suppress stall well past the stall angle of attack. There was also an increase in the lift-to-drag ratio at higher post-stall angles of attack. The plasma actuator used for this experiment was arranged and oriented so the plasma jet was toward the suction side of the airfoil when the airfoil was at positive angles of attack. The actuator was operated at both quasi-steady and unsteady states. During both cases there was an increase in the operational angle of attack. When the plasma actuator was operated in the quasi-steady state, the actuator drew more power than the unsteady case but was less effective than the unsteady actuator. For the unsteady case, an optimum frequency, was found to be: or could also be written as: St = fc U = 1 (2.6) F + = fc x U = 1 (2.7) where c is the chord length. The Strouhal number (St) equal to one is seen to be an optimum for various operations involving plasma actuators in unsteady operations. 21

34 F + is the non-dimensional frequency, where x c is the distance from the actuator to the trailing edge. During unsteady operation plasma actuators produce periodic vortices that flow into the fluid in the direction of the flow. The unsteady frequency f has a correlation with the wavelength of the vortices. This wavelength, λ = cr f the convection speed. The Strouhal number then becomes: where c r is St = fc ( ) cr L = = 1 (2.8) U λu With a Strouhal number of one this seems to be optimum for separation control because this would maintain a pair of vortices in the separated region. The unsteady plasma actuators worked the best when oriented to have a wall jet in the direction of the fluid flow. Plasma actuators placed near the trailing edge for roll control were also investigated. When placing plasma actuators near the trailing edge, the lift coefficient shifted in a way that would increase the wings chamber. Simulations were run, and plasma actuators placed near the trailing edge generated C L larger than that generated with actuators located near the leading edge. When several actuators were placed and operated in conjunction, their effect on C L was additive. This investigation also looked into replacing moving aileron surfaces with plasma actuators for roll control. One actuator was placed on the upper surface of the wing while another actuator of the same length L a, was placed on the bottom. The wing had dimensions of wing span b and the chord length c, on a standard NACA 0015 rectangular wing. This configuration was designed so that the actuator on the upper surface would produce an increase in lift while the actuator on the lower surface would produce a decrease in lift. This change in lift would affect the entire wing area S w. This arrangement resulted in a positive roll moment L R. The lift force is uniform in span over the area of the plasma actuator S a. The magnitude of the roll moment is L R = 2 ( C L ) qs a r a where r a is the moment arm from the center of the span to the lift center. The roll 22

35 moment coefficient becomes C LR = L R qs w b = 2 ( C L) S a r a S w b (2.9) With a leading edge plasma actuator, the flow could be reattached on a NACA 0015 airfoil up to 18 degrees angle of attack, which is 4 deg past the normal stall when operated in a quasi-steady state. During unsteady operation stall was postponed by 8 deg. The leading edge plasma actuator resulted in an increase in both C Lmax and α stall, and also improved L/D by as much as 340%. When placing a plasma actuator at the trailing edge and operating it in a quasi-steady state, the actuator influenced the flow like a plain trailing edge flap. Mabe et al. [9] investigated the use of SDBD plasma actuators to improve airfoil performance. A NACA 0021 was used during the testing with a plasma actuator located near the leading edge, where it could affect the transition point, the leading edge separation bubble, and at the flap shoulder, where it could affect the flow over a detached flap and possibly reattach the flow on the flap. The plasma actuators used during the experiments were operated at ± 5 kv at a frequency of 5 khz a 10% cycle, and an F + = 1. A control experiment was done on an airfoil with no plasma actuators at two Reynolds numbers (100,000 and 200,000). The slope of the life curve (dc L /dα) was 2π. This slope is only viable over a small range of angles between 6 to 8 degrees for a Reynolds number of 200,000 and between 2 to 6 degrees for a Reynolds number of 100,000. A plasma actuator was placed at the leading edge but not activated to see what effect the presence of the actuator itself would have on the flow. The presence of the actuator caused a decrease in the maximum lift (C Lmax ) even though the maximum stall angle (α stall ) was increased by 2 to 4 degrees. The non-active actuator actually created a discontinuity on the surface of the airfoil that in effect shortened the leading edge separation bubble and reduced the lift generated by the 23

36 airfoil. When the plasma actuator was activated in an unsteady state with F + = 1, the surface disturbance was increased and further reduced the initial dc L /dα and in doing so increased α stall. Increasing α stall generated some lift that helped recover some of the lift lost by the presence of the plasma actuator. With the addition of the passive actuator at x/c = 0.05, there was a decrease in drag by 20% on the clean airfoil at a Reynolds number of 100,000. Was also noticed that there was no noticeable effect at a Reynolds number of 200,000. When the plasma actuator was activated, it eliminated the separation bubble over most of the suction side of the airfoil, but the vortices that were created by having an F + = 1 increased the skin friction. It was observed that the active plasma actuator actually accelerated the transition earlier from laminar flow to turbulent flow and decreased the amount of lift generated at given angle of attack. A plasma actuator was also placed near the flap shoulder (x/c = 0.65) to see if flow could be reattached if the flow is detached over the flap. Leaving the actuator passive did not have any effect on the leading edge separation bubble. The passive actuator actually caused earlier separation over the flap and therefore increased the drag. Even when the plasma actuator was activated and the flap was never deflected there was no observable benefit from the active actuator. Even in the presence of the active plasma actuator, the momentum generated was so insignificant that the suction upstream was not noticeable. The wing actually stalled earlier than it would have if there was no actuator present. Yurchenko et al. [12] did an investigation on boundary layer control through the use of localized plasma generation. During the generation of plasma, the thermal patterns that were generated in the boundary layer proved to be an innovative method for controlling the boundary layer. The boundary layer control was realized using span-wise-regular microwave-initiated discharges. The behavior and characteristics of an airfoil was shown to improve during the plasma actuation. During a numerical 24

37 simulation of the flow control system using span-wise arrays of plasma actuators, a boundary layer that has transitioned to a turbulent boundary layer is less susceptible to the thermal patterns generated during plasma activation. Further experiments were done using a wind tunnel to test the span-wise array of plasma actuators and take aerodynamic measurements such as: lift, drag, pitch moment, and pressure coefficients. Wind tunnel tests showed that the plasma actuators adjusted the pressure around the airfoil in a favorable manner, was able to delay separation when the angle of attack was increased, as well as increasing C L without increasing drag when the angle of attack was increased past the stall angle. Yurchenko et al. [13] also did an investigation on localized plasma generation using wind tunnel tests. These tests were done to investigate the ability to control a boundary layer using plasma actuation generated by using microwave radiation. These tests were preformed in the Aerodynamic Facility for Interdisciplinary Research (AFIR). The test results showed that there was a delay in separation by 15% at prestall angles of attack, and during activation there was a decrease in drag by about 5%. These improvements can be considered as a method of flow stabilization. This allows for the delay of separation at post-stall angles of attack as well as an increased probability of flow reattachment once separation occurs. Little et al. [11] investigated the use of plasma actuators to help control separation on the flap of a high-lift airfoil. The test was designed to evaluate the efficacy of a single dielectric barrier discharge (DBD) plasma actuator. The actuator was placed on the shoulder of the detached flap in a position that would allow for flap reattachment in flow velocities between the Reynolds numbers of 240,000 (15 m/s) and 750,000 (45 m/s). During these experiments the moment coefficients that were calculated were of an order of magnitude lower than those seen in previous tests. The control authority for these tests is still maintained due to the amplification of the natural shedding frequency of vorticites from the flap shoulder. This behavior has a tendency 25

38 to transfer momentum between the freestream and separated regions of flow around the airfoil. This change in momentum changes the circulation around the airfoil and can enhance the lift of the airfoil. The activation of the DBD demonstrated how it could control the flow over a detached TE flap of a high-lift airfoil. Another set of experiments in active flow control using plasma actuators was done by Vey et al. [15]. This set of experiments was focused on the low Reynolds number speeds of less then 100,000. The reason for this region of flight speeds was to concentrate on the use of plasma actuators for the development of micro air vehicle (MAV) applications. During the testing, force measurements and frequency response of lift was investigated for different wing-actuator combinations at different angles of attack. Many of these measurements and flow field visualization was done using a PIV system. Active flow control using plasma actuators is very productive while operating in low Reynolds numbers flow. Testing showed C L increased up to C L = 0.45 through the use of periodic actuation. These results were found for actuator placement at the leading edge or at wing tips. The forcing frequency, for use during the periodic activation, was dependent on the wing s angle of attack. When this frequency was optimized for a certain angle of attack, there was a greater increase in lift. Vey also demonstrated that leading edge flow control was more effective when the aspect ratio was increased. Control authority was seen to decrease as the Reynolds number is increased. Burman et al. [16] examined separation control over Low Pressure Turbines (LPT) using plasma actuators. DBD plasma actuators were evaluated in quiescent air as well as in air flow with a Reynolds number of 50,000. The plasma actuators were tests in both a steady actuation as well as pulsed. During both tests cases, velocity and total pressures were measured and then studied to examine the effects of excitation frequency and amplitude on the flow, and to see how pulsed operation effected the separation. The same measurements were made for actuators orientated oppo- 26

39 site and aligned, as well as, downstream and span-wise plasma discharges. When the actuators were tested in quiescent flow, the momentum generated scaled with the increase to excitation frequency and amplitude. For the cases where the plasma actuator operation was pulsed, separation control increased monotonically with increased modulation frequency and duty cycle. When examination of the orientation of the plasma actuators was done, even though the reversed orientation successfully demonstrated separation control, the aligned orientation was a much better means of flow control. Guo et al. [18] investigated the effect of a new plasma actuator configuration on thrust. This new design was introduced to try and take advantage of discharge asymmetry. This new design focused on controlling the surface charge by the addition of a third electrode. The new configuration was seen to produce about 70% more thrust than that of the conventional plasma actuator design. Thomas et al. [20] investigated the use of SDBD plasma actuators for use in active aerodynamic flow control. Experiments were run to try and optimize the body force that was produced by these types of actuators. This study was focused on being able to improve control authority of plasma actuators while at higher Reynolds numbers. Actuator parameters such as dielectric material, dielectric thickness, applied voltage, applied frequency, voltage waveform, exposed electrode geometry, covered electrode widths, and multiple actuator arrays were tested. The limiting factor to the amount of body force you get is in the formation of plasma streamers. Investigations were preformed to figure out a way to gain a higher control authority by delaying the formation of streamers. By using a dielectric with a higher dielectric strength, lower dielectric constant, and using a thicker material, that plasma streamer development was delayed. Another method that was discovered to reduce streamer formation was to lower the AC frequency that was applied to a given actuator. Thomas et al. discovered that a plasma actuator with a serrated TE, rather than a conventional 27

40 straight TE, actually produced a higher body force. When examining the construction of not only single actuators but multiple actuator arrays, Thomas found that if your embedded electrode is not wide enough, it actually constrains the body force, therefore limiting the actuator and not allowing for optimum body for production. When multiple actuators were arranged in an array there was an increase to the total body force but the body force does not sum linearly. At with multiple actuators the wall jet thickened with the addition of the other actuators due to the increase of momentum flux close to the wall. Overall lift enhancement and drag reduction could be achieved by using plasma actuators in the fuselage and wing Reynolds number of 6.8X10 6 and 1.2X10 6 respectively. 28

41 CHAPTER 3 Experimental Set-Up Chapter 3 goes into depth of how the apparatus and diagnostic experiments were setup and performed throughout this investigation using plasma actuators to control separation on an airfoil. The construction method and guidelines for constructing a plasma actuator is outlined in this section. The program used to generate plasma along the actuator is also discussed here. A discussion on how PIV measurements were taken is also provided to give a background on how the results were achieved. The experimental setup for both the bench-top and wind tunnel tests is provided to help with experimental reproduction for future experiments. 3.1 Plasma Actuators Plasma actuators are constructed using a dielectric material and copper electrodes. The dielectric materials that were investigated are Teflon, acrylic, and alumina. The Teflon used was a 1/32 inch thick, generally about 1 inch wide and ran the span of the wings being tested Fig The acrylic was 1/8 inch thick, 2 inches wide and about 12 inches long Fig The alumina plates were inches thick and about a 4 inch by 4 inch squares Fig Two different widths of copper electrodes were looked at and compared when the Teflon was tested, 1/4 inch and 1/2 inch wide copper tape. One set of Teflon actuators, the acrylic actuators and the alumina actuators all used the 1/2 inch copper tape. The 1/4 inch copper tape was compared with the 1/2 copper tape on the Teflon. Two strips of copper tape were used on each plasma actuator. The exposed electrode was connected to the high voltage lead and 29

42 the embedded electrode was connected to ground. The high voltage lead was supplied with an alternating current (AC) with an average peak-to-peak voltage of 50 kv for Teflon. The AC signal was generated by a computer using a LabView program to produce the square wave input signal and plasma frequencies between 3 khz and 15 khz (a block diagram of the LabView program can be seen in Fig. 3.4). For situations when two transformers were used, a different LabView program was used (the block diagram of the Labview program can be seen in Fig. 3.5). The signal was then sent to a QSC RMX 1450 amplifier and out to a CMI 5012 transformer. The output from the transformer was monitored for voltage with a North Star PVM :1 high voltage probe. The voltage probe was then connected to an oscilloscope to monitor and maintain the voltage necessary. Figure 3.1: Teflon Plasma Actuator with 1/2 inch Copper Electrodes Figure 3.2: Acrylic Plasma Actuator with 1/2 inch Copper Electrodes 30

43 Figure 3.3: Alumina Plate Plasma Actuator with 1/2 inch Copper Electrodes 31

44 Figure 3.4: LabView Block Diagram for 1 Channel Output Figure 3.5: LabView Block Diagram for 2 Channel Output 32

45 3.2 PIV Measurements Particle image velocimetry (PIV) was used to measure flow field that was produced while an active plasma actuator was being used. The flow field was a 2-D cross section of the plasma actuator. The field was saturated with particles of about 1 micron in size from a Turbofog fog generator. A thin laser sheet produced by a dual-head Nd: YAG laser from Big Sky Lasers, was projected over the 2-D cross section of the plasma actuator and the plasma jet being produced. Three lenses were used to alter the original laser projection into a thin sheet spanning the cross section. The first lens was a converging lens that focuses the beam coming from the laser heads into a fine horizontal sheet. A second diverging lens, at the focal length, was used to turn the sheet 90 degrees and gives a thinner more concentrated laser beam. The final lens used was a cylindrical lens to spread the beam into the fine thin sheet that is needed to be projected across the 2-D cross section. The lasers were pulsed in sync with a Kodak Megaplus ES 1.0 CCD, high speed, and high resolution camera in combination with a Quantum Composer timing box. Every time a PIV run was completed, the Epix frame grabbing software captured 63 pairs of images, each measuring 1008 x 1018 pixels. Fig. 3.6 shows a schematic of this setup. The PIV program we used to analyze each of the runs was a Wall adaptive Lagrangian Parcel Tracking algorithm (WaLPT). This algorithm was developed by Sholl and Savas [4], it takes all the particles that were saturated in the region by the fogger and treats them as fluid parcels. WaLPT then analyzed the movement and deformation of the fluid field and then compared it to the previous snap shot to determine the individual velocities, vorticity, and acceleration of each particle. The WaLPT algorithm was used to get very accurate measurements for the velocities near surfaces. 33

46 Figure 3.6: Schematic of PIV Setup 3.3 Bench-Top Testing The bench-top testing phase of these experiments were done in quiescent flow. The Plasma actuator was placed in a 20 in x 10 in x 12 in clear glass aquarium. Fog was then injected into the aquarium from the Turbofog fog generator to allow for PIV measurements to be taken by the process described above. The laser was positioned so that the laser sheet passed over the midsection of the plasma actuator at a perpendicular angle to the plasma jet. The three lenses were positioned to produce a thin laser sheet to allow for better definition when PIV measurements were being taken. The high speed Kodak Megaplus camera was placed perpendicular to the laser sheet so that it was pointed down the long direction of the plasma actuator. The placement of the camera in this fashion allowed the capturing of the wall jet that was produced by the plasma actuator as well as the flow structure downstream of the actuator. A block schematic of the bench-top setup is pictured in Fig

47 Figure 3.7: Schematic of bench-top Setup 35

48 3.4 Wind Tunnel Testing For the wind tunnel tests a Liebeck La203a airfoil was used to create a test wing. A profile of this airfoil is illustrated in Fig Using SolidWorks 3D CAD program, a wing with a span of 6 inches and a chord of 6 inches was modeled. The CAD model is illustrated in Fig The test wing was also designed to have two one-inch wide, 3/64 inch deep cuts located near the leading edge and at the 40% chord. These cuts were designed so that the plasma actuators could be embedded into the surface of the wing, to avoid artificially tripping the air flow over the actuator, causing separation. The CAD drawing was then sent to a rapid prototyping machine to create two wing sections made of an SLA plastic material, so that the final test wing had a span of 12 inches and a chord of 6 inches. The wing sections were then glued together with epoxy and a piece of 1/4 x 20 all thread was fed through the 1/4 chord to allow a pivot and mounting location. Plasma actuators were then secured into the grooves on the wing surface by tape and the wing was mounted into the wind tunnel. The wind tunnel used for these plasma actuator tests is a GDJ FLOTEK 1440 wind tunnel, with a test section that is 12 in. x 12 in. x 36 in.. This wind tunnel is an Eiffel or Open-Loop style wind tunnel. The air flow is pulled through a large inlet and through the test section by the motor and fan in the exhaust section. This tunnel has a top speed of about 16m/s at 1200 rpm. The laser was positioned on the outside of the wind tunnel much like that used in the bench-top tests. The camera was positioned above the tunnel in such a way that a large portion of the upper surface of the wing could be seen. The Turbofog fog generator was placed at the inlet of the wind tunnel so that when the tunnel was turned on the fog could be ingested into the inlet, pass through the test section over the wing, and be exhausted out the back end of the tunnel. While the tunnel was running PIV measurements were done as described above. The 36

49 test runs done in the wind tunnel will take many of the best parameters from the bench-top tests and examine them in a free stream environment. A schematic of the wind tunnel test setup is illustrated in Fig Figure 3.8: Liebeck La203a CAD Model 37

50 Figure 3.9: Liebeck La203a Airfoil Figure 3.10: Wind Tunnel of Setup 38

51 CHAPTER 4 Results Chapter 4 shows the results for all of the bench-top, computational, and wind tunnel tests performed. Between the bench-top testing and the wind tunnel testing there were approximately 200 runs recorded. During the bench-top investigation roughly 100 tests were done to illustrate which dielectric material produces the strongest plasma jet and which dielectric material would be best to use in further wind tunnel testing. An investigation was performed using XFoil to help determine the optimum location to place plasma actuators to provide separation control on a La203a airfoil. With the results from the previous two investigations, wind tunnel testing was performed at different Reynolds numbers to test for active separation control using plasma actuators. Reattachment was achieved for three of the four Reynolds numbers tested. 4.1 Actuator Development As discussed earlier, the basic actuator consists of two electrodes separated by a dielectric medium. During bench-top testing three different dielectric materials were investigated while varying several parameters during their operation. The three different materials were Teflon, acrylic, and alumina. The parameters that were varied included operating frequency, modulation frequency, and peak-to-peak voltage. The duty cycle was set to 100% while the operating frequency was varied to have a steady output of plasma. Once the optimum operating frequency was found the duty cycle was reduced to 50%; this gave the plasma actuator a pulsing unsteady behavior. The 39

52 physical parameters that were varied can be seen in Fig Figure 4.1: Plasma Actuator Parameters The first set of experiments was performed using Teflon, which has a dielectric constant of about 2.1, as the dielectric material, with 1/4 inch wide copper electrodes. To establish an optimum operating frequency 14 runs were performed. The first set of runs varied the frequency between 3,000 Hz and 15,000 Hz by steps of 1,000 Hz. The peak-to-peak voltage was held constant at 50 kv and the duty cycle was set at 100%. It was observed that there seemed to be two optimum frequencies for these cases; one at about 8,000 Hz and another at 15,000 Hz. 15,000 Hz was a much higher frequency than had been previously investigated, so 15 more runs were performed to investigate the behavior of the plasma actuator at and near 15,000 Hz. As can be seen in Fig. 4.2 the optimum operating frequency was 15,000 Hz and had a maximum velocity of 110 cm/s, and this occurred about 28 mm downstream of the actuator according to Fig The PIV measurements for the optimum run are seen in the next several figures. Fig. 4.4 illustrates the velocity vectors generated by the plasma actuator and it can be seen that a strong wall jet was produced. Figs. 4.4 and 4.5 show that a standing vortex was generated above the plasma actuator, and this illustrates that the actuator was pulling the surrounding air inward and projected the air outward in the form of a wall jet. Fig. 4.6 shows the velocity profiles in relation to the position of the plasma actuator. Fig. 4.6 shows how the velocity varied both a little upstream and downstream of the actuator. The standing vortex that was generated by the actuator can account for the variation in velocities. The swirling motion that was 40

53 generated by the actuator could be the reason for the slight rearward flow that is seen in Fig The jet that was produced stretched forward of the actuator about 30 mm. Figs. 4.2 and 4.3 were plotted with a maximum velocity of 180 cm/s to allow for comparison with other cases that were run later. The next parameter that was examined was how varying the modulation frequency changes the effect of the plasma actuator. To see what modulation frequency had the biggest effect on the plasma jet, the frequency was varied between 5 Hz and 1,000 Hz by semi-logarithmic steps. The voltage was once again held constant at 50 kv and the duty cycle was reduced to 50% throughout the 23 runs, to generate a pulsed plasma jet. The first 13 runs swept the whole range between 5 Hz and 1,000 Hz and it was found that a modulation frequency of 50 Hz produced the greatest plasma jet velocity as seen in Fig During this phase of experiments an interesting phenomenon was observed in the PIV measurements, there was a standing vortex formed when the modulation frequency reached 200 Hz. Another 10 runs were performed to try and see how this vortex changed when the modulation frequency was changed around 200 Hz. It was observed that the vortex would grow or shrink in size depending on how close the frequency was to 200 Hz. The generation of the vortex was somewhat inconsistent, and the velocities varied a little due to the inconsistency. Once the frequency was beyond the 200 Hz range, the velocity produced by the plasma actuator decreased well below that of the lower frequency range. The last thing performed during this part of the investigation was to test and see what the highest peak-to-peak voltage the Teflon could handle before it burned out. Several runs were performed by increasing the peak-to-peak voltage by steps of 10 kv from 50 kv to determine the burn out limit of 70 kv. Once this power setting was reached there was a saturation of plasma streamers that led to the dielectric material breaking down and burning out. Once the actuator burned out in any spot all plasma generation stopped. From this set of experiments we were able to say that the Teflon produced the strongest plasma jet 41

54 when the following parameters were met: the plasma frequency was set to 15,000 Hz, the peak-to-peak voltage was set to about 50 kv, and when operating the actuator in a pulsed fashion the modulation frequency was set to 50 Hz. Figure 4.2: Plasma Actuator Benchmarking: Varying Frequency (Teflon with 1/4 in. Copper Electrodes) Figure 4.3: Plasma Actuator Benchmarking: Varying Modulation Frequency (Teflon with 1/4 in. Copper Electrodes) 42

55 Figure 4.4: Velocity Vectors for Teflon with 1/4 in. Copper Electrodes Figure 4.5: Vorticity for Teflon with 1/4 in. Copper Electrodes 43

56 Figure 4.6: Velocity Profile for Teflon with 1/4 in. Copper Electrodes 44

57 The second set of experiments were performed using acrylic, with a dielectric constant ranging for about , as the dielectric material, with 1/2 inch wide copper electrodes. Acrylic was tested to see how it compared to the Teflon. The main tests that were run on the acrylic were with a varying operating frequency between 3,000 Hz and 15,000 Hz. 13 runs were performed at a peak-to-peak of 50 kv. No favorable results were seen while the actuator was operated at this voltage level. Several runs were then performed to see the highest peak-to-peak voltage the acrylic could withstand before it burned out. Acrylic s highest peak-to-peak voltage was observed to be 90 kv when the acrylic started to flex and burn out. Fig. 4.7 shows how the plasma jet velocity increased as the peak-to-peak voltage increased. Having found a good operating voltage, the 13 runs that were performed at the lower voltage, were re-performed, with the peak-to-peak voltage set at 80 kv, to see if a more favorable set of results could be found while varying the plasma frequency. Fig. 4.8 shows the results of these last 13 runs, and an optimum frequency of 7,000 Hz was found with a corresponding maximum velocity of 146 cm/s. Throughout the runs the acrylic mostly had a constant velocity over the frequency range. The maximum velocity of the acrylic was observed to be higher than that of the Teflon. Fig. 4.9 shows the velocity vectors for the acrylic plasma actuator. As with the Teflon a strong wall jet is observed, but when examining Fig there was no standing vortex like was observed in Fig. 4.5 for the Teflon dielectric material. Fig illustrates the velocity profile of the plasma jet produced by the acrylic plasma actuator downstream of the actuator. There was no upstream flow produced by the acrylic actuator, because of the possible lack of a standing vortex. The maximum jet velocity for the acrylic is found somewhere between 13 mm and 19 mm downstream of the actuator. Figs. 4.7 and 4.8 were plotted with a maximum velocity of 180 cm/s to allow for comparison to other cases. 45

58 Figure 4.7: Plasma Actuator Benchmarking: Varying peak-to-peak Voltage (Acrylic with 1/2 in. Copper Electrodes) Figure 4.8: Plasma Actuator Benchmarking: Varying Frequency (Acrylic with 1/2 in. Copper Electrodes) 46

59 Figure 4.9: Velocity Vectors for Acrylic with 1/2 in. Copper Electrodes Figure 4.10: Vorticity for Acrylic with 1/2 in. Copper Electrodes 47

60 Figure 4.11: Velocity Profile for Acrylic with 1/2 in. Copper Electrodes 48

61 The next set of experiments were performed with alumina, with a dielectric constant of 4.5, as the dielectric material, with 1/2 inch wide copper electrodes. Alumina is a common material used when dealing with plasma actuators because it can produce a strong plasma jet and has a high dielectric constant. For our purpose it was used to compare with the Teflon dielectric material tested above. Like the other dielectric materials, the alumina was tested with a varying operating frequency. The alumina was first tested with a peak-to-peak voltage of 50 kv but burned out before any tests were finished running. The peak-to-peak voltage was then reduced to 40 kv and the material fared much better. The alumina was then tested with a varying operating frequency between 3,000 Hz to 15,000 Hz which was performed before on the other dielectric materials. 13 runs were performed over this operating frequency range and the results can be seen in Fig A maximum velocity of 162 cm/s was found for an operating frequency of 8,000 Hz. Unlike the characteristics of the acrylic, the alumina had a very non-constant velocity range. It was observed that once past the peak operating frequency the velocity dropped off sharply. Fig shows the velocity vectors for the alumina at its optimum frequency and it is seen that it too produces a strong wall jet forward of the actuator. Fig shows a strong vorticity in the direction of the wall jet, but there was no standing vortex present like there was in the Teflon. The velocity profile is seen in Fig where the jet velocity produced upstream and downstream is illustrated. The maximum velocity is located beyond 15 mm of the actuator, so peak velocity may not have been determined. The upstream component of velocity that was produced is believed to come from the plasma jet produced from the electrode on the bottom surface of the actuator. Fig was plotted with a maximum velocity of 180 cm/s to compare to Teflon and acrylic. 49

62 Figure 4.12: Plasma Actuator Benchmarking: Varying Frequency (Alumina with 1/2 in. Copper Electrodes) Figure 4.13: Velocity Vectors for Alumina with 1/2 in. Copper Electrodes 50

63 Figure 4.14: Vorticity for Alumina with 1/2 in. Copper Electrodes Figure 4.15: Velocity Profile for Alumina with 1/2 in. Copper Electrodes 51

64 The last set of tests was performed with Teflon as the dielectric material again, but with 1/2 inch wide copper electrodes instead of the 1/4 inch wide copper electrodes. The 1/2 inch wide tape was examined to see if it had a large impact on the plasma jet produced by the actuator. This actuator was tested by varying the operating frequency between 3,000 Hz and 15,000 Hz. The results are seen in Fig and it was clear that the Teflon with the 1/2 inch wide copper electrodes behaved similar to that of the Teflon with the 1/4 inch wide copper electrodes. The 1/4 inch copper electrodes seem to have a better mid-range velocity. The two different sized electrodes had roughly the same velocity at the upper end of the plasma frequency range. Figure 4.16: Plasma Actuator Benchmarking: Varying Frequency (Teflon with 1/2 in. Copper Electrodes) 52

65 When comparing the three types of dielectrics and their maximum velocity, it was seen that alumina was the dielectric that produced the highest plasma jet. This comparison is illustrated in Fig Even though the alumina produced the highest plasma jet velocity, it was a very stiff and brittle material and cannot be flexed without cracking. The acrylic produced the second highest velocity, and even though it was not brittle like the alumina, it was not flexible. Both the Teflon with the 1/4 inch wide copper electrodes and the Teflon with the 1/2 wide copper electrodes preformed about the same. Both sets of Teflon produced a plasma jet with a velocity a little over 100 cm/s. The Teflon may not produce the highest plasma jet velocity, but it was flexible enough to be conformed to an airfoil for further testing. Figure 4.17: Maximum Velocity Comparison for each Dielectric 4.2 X-Foil To determine the best location to place the plasma actuators on the surface of the La203a wing an extensive separation point investigation was performed. This investigation was performed using XFoil. Several different Reynolds numbers were examined to see how different Reynolds numbers would affect the point of separation. The La203a airfoil was run through a wide range of angles of attack to see 53

66 how the separation progressed along the airfoil. Each run varied the angle of attack from -6 to 16 degrees, for each of the following Reynold s numbers: 50,000, 100,000, 175,000, 250,000,375,000, 500,000 and 650,000. For each Reynolds number there were 12 graphs made examining the coefficient of skin friction in comparison to the percent chord location. Figs and 4.19 illustrate how the separation point changed as Reynolds number changed. By tracking the point where the coefficient of skin friction goes to zero, we could track the progression of separation over an airfoil as the angle of attack was increased. Examining Fig it can be observed that a leading edge separation bubble was formed, but the flow then reattached further downstream. By examining Fig there was a similar trend to the previous run at the lower Reynolds number but the flow does not actually detach until close to the trailing edge. The results of the separation tracking is pictured in Fig From these investigations it was seen that the La203a airfoil stalls at the trailing edge and progressed forward as the angle of attack increased to the point where almost the entire airfoil was separated. It was decided to place two actuators on the surface of the airfoil. The first actuator was placed close to the leading edge to help prevent the development of any separation bubbles, and the second was located about the 40% chord, where throughout this investigation it was seen that this was the next most likely location where separation would occur. Fig. 3.8 illustrates the plasma actuator placement. 54

67 Figure 4.18: XFoil Separation Point Tracking Reynold s Number of 100,000 at an Angle of Attack of 10 Degrees Figure 4.19: XFoil Separation Point Tracking Reynold s Number of 650,000 at an Angle of Attack of 10 Degrees 55

68 Figure 4.20: XFoil Separation Point Tracking 56

69 4.3 Wind Tunnel Flow Control Tests Wind tunnel tests were performed using a La203a wing model. Two grooves cut into the suction surface to enable the embedding of plasma actuators flush to the rest of the surface. The wind tunnel tests were designed to test the plasma actuators under different Reynolds numbers and different plasma activation states. The plasma actuators were tested for flow control at 50,000, 75,000, 100,000, and 150,000 Reynolds numbers. The different plasma activation cases consisted of testing the LE actuator solo, aft actuator solo (placed at the 40% chord as stated above), and both actuators being used together. For the different activation scenarios, two different things were performed: one was to test steady activation of the plasma actuator, then the actuators were pulsed at different frequencies to test unsteady activation. For all the wind tunnel tests the wing was placed at an angle of attack of approximately 20 to 22 degrees to achieve full separation from the airfoil. This α stall was varied to provide a deep stall condition. Fig illustrates the different plasma actuator configurations tested during the wind tunnel investigation. For a Reynolds number of 50,000, eight runs were performed. For all eight runs the angle of attack was held constant at 20 degrees to achieve the deep stall condition we were looking for. The first run for any case performed in this investigation was a baseline test to compare between actuator off and actuator on conditions. For the first run Fig shows a set of four graphs showing various measurements taken. Fig (a) shows the velocity flow field with lines coming from the airfoil surface to indicate where measurements were taken, and also shows clear flow separation and deep stall. Fig (b) show the vorticity in the flow field. Fig (c) measures U rms and (d) is a measurement of the Turbulent Kinetic Energy (TKE) within the flow. By examining these for graphs you can also visually see the shear layer formed 57

70 Figure 4.21: Wind Tunnel Plasma Actuator Test Matrix 58

71 when the flow separated. Another measurement taken during testing was the reverse flow probability. Reverse Flow Probability (RFP) is measured when a velocity vector is divided by the mean velocity vector within the flow field. Fig shows the RFP within the flow, and as can be seen, there was a high RFP along the airfoil, illustrating a separated region. Since the angle of attack was placed so that a deep stall condition was achieved, the flow separated close to the leading edge. The next test that was preformed was to activate the LE actuator with a steady activation. During this test the duty cycle was placed at 100% to achieve a constant plasma activation. Fig shows that the flow was reattached by looking at Fig (a). It can also be observed that there was a standing vortex that was generated by the plasma actuator, Fig (b). The presence of the standing vortex demonstrated that energy was injected into the flow to promote reattachment. To further show that the flow was reattached Fig shows that the high probability of reverse flow along the airfoil surface was gone. The third test performed was to pulse the LE actuator at modulation frequency of F + = 1. To achieve the pulsed activation the duty cycle was reduced to 50%, actively lowering the power sent to the actuator. Both Fig and Fig showed when pulsing the plasma, attachment can still be achieved. Examination of Fig (c) and (d) clearly shows where the plasma actuator was injecting energy into the flow. The next two tests were performed using only the aft plasma actuator placed at the 40% chord. These two tests were performed in the same fashion as the tests performed on LE actuator. When this actuator was tested it was seen that it had no effect on separation and results in Fig through Fig confirm this. Fig shows that the actuator was generating energy but because the flow was so far separated, it had no effect, and this is confirmed further by Fig The last three tests were all preformed with both the LE and aft actuators on. The first two tests were performed the same way as the tests were performed on the LE or aft actuators. The first test was performed with a steady activation and the second 59

72 test being the pulsed case where F + = 1. The results for the steady activation can be seen in Fig and Fig Fig (a) illustrates that the flow is reattached to the surface. Fig (c) and (d) demonstrates where the two actuators were injecting energy into the flow. Fig demonstrates that when the actuators were activated in this configuration the RFP was decreased. It was observed that at the 40% chord location, there was a high RFP in this region, which would indicate that the aft plasma actuator was active. The pulsed activation results are seen in Fig 4.34 and Fig Fig 4.34 (a) shows that the flow was reattached to the surface of the airfoil just as it was in the previous cases. Fig further confirms that reattach had been achieved when the actuators were activated in this manner. The final test performed with both actuators on was to see if activating the plasma actuators outof-phase from each other had more of an effect than the regular pulsed case. The initial results are pictured in Fig and Fig The initial findings presented in Fig demonstrated that the flow had been reattached and Fig confirmed it. When comparing Fig and Fig it was observed that there was a lower RFP in the out-of-phase case than there was in the in-phase case. 60

73 Figure 4.22: Actuators Off (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Figure 4.23: Actuators Off Reverse Flow Probability within the Flow Field 61

74 Figure 4.24: Leading Edge Actuator, Constant Activation (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity Figure 4.25: Leading Edge Actuator, Constant Activation Reverse Flow Probability within the Flow Field 62

75 Figure 4.26: Leading edge Actuator, Pulsed Activation with an F + = 1 (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE 63

76 Figure 4.27: Leading edge Actuator, Pulsed Activation with an F + = 1 Reverse Flow Probability within the Flow Field 64

77 Figure 4.28: Aft Actuator, Constant Activation (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE 65

78 Figure 4.29: Aft Actuator, Constant Activation Reverse Flow Probability within the Flow Field Figure 4.30: Aft Actuator, Pulsed Activation with an F + = 1 (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 66

79 Figure 4.31: Aft Actuator, Pulsed Activation with an F + = 1 Reverse Flow Probability within the Flow Field 67

80 Figure 4.32: Steady Activation on the LE and Aft Actuators (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE 68

81 Figure 4.33: Steady Activation on the LE and Aft Actuators Reverse Flow Probability within the Flow Field Figure 4.34: Pulsed Activation with an F + = 1 on the LE and Aft Actuators (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 69

82 Figure 4.35: Pulsed Activation with an F + = 1 on the LE and Aft Actuators Reverse Flow Probability within the Flow Field Figure 4.36: Out-of-Phase, Pulsed Activation with an F + = 1 on the LE and Aft Actuators (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 70

83 Figure 4.37: Out-of-Phase, Pulsed Activation with an F + = 1 on the LE and Aft Actuators Reverse Flow Probability within the Flow Field 71

84 A set of velocity profiles was generated for each run performed at seven locations across the field of view. These seven locations are the red lines pictured in Fig (a) for each run. The x/c locations of the seven lines that are used for measurements are as follows: 0.089, 0.155, 0.245, 0.340, 0.434, 0.538, These profiles were then plotted on top of each other to show how different activation cases affected the flow differently. This comparison can be seen in Fig The black line in this figure represents the separated case or the baseline. The red solid line represents steady activation of the LE actuator and the dashed red line is the same actuator but with pulsed activation with F + = 1. The blue lines are the aft actuator cases represented in the same fashion as the LE actuator cases. The green lines represent the cases where both actuators were activated, with the solid line being the steady activation and the dashed line being the pulsed, F + = 1, in-phase case. It can be observed that in four of the six activation cases presented in Fig reattachment was achieved. In the two cases where the aft actuator was active, reattachment was not achieved and the flow remained separated. The lower part of Fig also illustrates the effect of the actuator activation on the vertical flow within the flow field. Along the first measurement line it can be seen that the LE actuator was actually pulling the flow inward toward the surface of the airfoil. Examining the third measurement location it is seen that with the activation of the aft actuator, even in the cases where reattachment was not achieved, the actuator was actively drawing in the air flow. A further comparison was performed with the cases of in-phase and out-of-phase activation of the actuators. This comparison can be seen in Fig The black line here is the separated case, as it was before, the red line is the in-phase case, and the blue line is the out-of-phase case. As can be observed in this figure, both the in-phase and out-of-phase cases reattached the flow. Further observation reveals that the out-of-phase case had a more favorable effect on the separated flow than the 72

85 in-phase case did at almost every location. Figure 4.38: Velocity Profiles for 50,000 Reynolds Number Cases: Solid Black, No Control; Solid red, LE Steady; Dashed Red, LE Pulsed F + = 1; Solid Blue, Aft Steady; Dashed Blue, Aft Pulsed F + = 1; Solid Green, Both Steady; Dashed Green, Both Pulsed F + = 1 in-phase 73

86 Figure 4.39: Velocity Profile Comparison of In-Phase and Out-of-Phase Actuator Activation at 50,000 Reynolds Number: Solid Black, No Control; Solid Red, Both Pulsed F + = 1 in-phasae; Solid Blue, Both Pulsed F + = 1 out-of-phase 74

87 A similar set of eight runs were performed for a Reynolds number of 75,000. Several things changed when than Reynolds number increased to 75,000. First, the angle of attack had to be increased to degrees to achieve the same type of deep stall condition that was present when the Reynolds number was 50,000. In this set of experiments, the aft actuator was not run alone, because when the aft actuator was activated without the LE actuator in the previous runs, it had no effect on the separated flow. Also a couple of different forcing frequencies had to be tested because through some trial and error an F + = 1 no longer reattached the flow as it did with a Reynolds number of 50,000. The first run performed at 75,000 Reynolds number was the separated case for all comparisons. Figs and 4.41 illustrate the separated flow. Further trials with the standard actuator configuration demonstrated that these plasma actuators had no effect on the separated flow, so a small alteration was made to the setup. We connected a second transformer to the actuator so we had two power leads connected to the same actuator, one hot lead to the exposed electrode and another hot lead to the embedded electrode to effectively double the power input to the actuator to 100 kv. The signal was then altered so that the two power leads were 180 degrees out of phase from each other. This change allowed us to have a better control authority over the actuator. The next set of runs completed were with just the LE actuator active. The LE actuator was tested with a steady activation and with a pulsed activation. The constant activation of the LE actuator reattached the separated flow and this can be seen in Figs and Fig (a) demonstrates that the flow was reattached during this activation and is confirmed by examining the RFP seen in Fig Further examination of Fig (c) and (d) illustrates where the actuator is injecting energy into the flow. During the pulsed activation cases, the actuator was pulsed at three different forcing frequencies. The first frequency tested corresponded to F + = 1. 75

88 Results for this test case are seen in Figs and The flow separated again under this condition, so it was decided to test some lower forcing frequencies to see if pulsing the plasma had any effect at this Reynolds number. It was also observed that the flow separated from the same location when the actuator was pulsed at F + = 1 and when no actuator was on at all. Varying the forcing frequencies by 5-10 Hz starting at 10 Hz, it was seen that a forcing frequency of 10 Hz and 15 Hz reattached the flow, which correspond to F + = and respectively. Forcing frequencies above 15 Hz showed no effect on the flow which remained separated. The results with F + = can be seen in Figs and Fig (a) shows that the flow was reattached and was confirmed in Fig F + = also showed favorable results and was tested to compare to the case with F + = The results from the F + = tests are illustrated in Fig and Fig Very similar results were seen when comparing the two different F + cases. Three runs were performed using both the LE and aft actuator. One run was performed using a constant activation of both actuators and the other two runs were pulsed activation. The two pulsed cases had synchronous activation at F + of and F + = 1 was not tested in this case because it proved ineffective when tested with the LE. The results from these three runs can be seen in Figs through The steady activation showed similar results to when the two actuators were activated at a Reynolds number of 50,000. The largest difference was seen in Fig. 4.51, since the actuator had more control authority there was a lower RFP for the Reynolds number flow of 75,000. For the case where both actuators were active with F + = 0.198, the flow was reattached and is demonstrated in Figs and Fig (a) illustrates that reattachment was achieved, and the RFP in Fig supports this. The results for F + = when both actuators were on is pictured in Figs and F + = with both actuators on had similar results to that of F + = with both actuators on. 76

89 Figure 4.40: Actuators Off (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE Figure 4.41: Actuators Off Reverse Flow Probability within the Flow Field 77

90 Figure 4.42: Leading edge Actuator, Constant Activation (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE 78

91 Figure 4.43: Leading edge Actuator, Constant Activation Reverse Flow Probability within the Flow Field Figure 4.44: Leading edge Actuator, Pulsed Activation with an F + = 1 (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 79

92 Figure 4.45: Leading edge Actuator, Pulsed Activation with an F + = 1 Reverse Flow Probability within the Flow Field Figure 4.46: Leading edge Actuator, Pulsed Activation with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 80

93 Figure 4.47: Leading edge Actuator, Pulsed Activation with F + = Reverse Flow Probability within the Flow Field Figure 4.48: Leading edge Actuator, Pulsed Activation with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 81

94 Figure 4.49: Leading edge Actuator, Pulsed Activation with F + = Reverse Flow Probability within the Flow Field Figure 4.50: Steady Activation on the LE and Aft Actuators (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 82

95 Figure 4.51: Steady Activation on the LE and Aft Actuators Reverse Flow Probability within the Flow Field 83

96 Figure 4.52: Pulsed Activation on the LE and Aft Actuators with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE 84

97 Figure 4.53: Pulsed Activation on the LE and Aft Actuators with F + = Reverse Flow Probability within the Flow Field Figure 4.54: Pulsed Activation on the LE and Aft Actuators with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 85

98 Figure 4.55: Pulsed Activation on the LE and Aft Actuators with F + = Reverse Flow Probability within the Flow Field 86

99 Profiles for each of the runs performed for Reynolds number 75,000 are shown in Fig Each one of the profiles were plotted together to show comparisons between all the runs. The black line represents the baseline separated case. The four runs performed on the LE actuator are shown by the red line. The solid red line is the constant activation case, the dashed line is the pulsed case at an F + = 1, the dash-dot line is the pulsed case with F + = and the dotted line is the case with F + = The case when both the LE and aft actuators were activated together are shown by the blue lines. The case with constant activation is the solid line, the dashed line is F + = 0.198, and the dash-dot line is the case with F + = Fig 4.56 shows that every case except one reattached the flow. The only case that did not reattach the flow was the case where the LE actuator was activated with F + = 1. The case that provided the best results seemed to be when both actuators were activated with constant activation. The bottom part of Fig 4.56 shows that when the LE actuator was active that it pulled flow inward towards the actuator from the flow above the actuator. 87

100 Figure 4.56: Velocity Profiles for 75,000 Reynolds Number Cases: Solid Black, No Control; Solid Red, LE Steady; Dashed Red, LE Pulsed F + = 1; Dash-Dot Red, LE Pulsed F + = 0.198; Dotted Red, LE Pulsed F + = 0.297; Solid Blue, Both Steady; Dashed Blue, Both Pulsed F + = 0.198; Dash-Dot Blue, Both Pulsed F + =

101 Eight more runs were performed at a Reynolds number of 100,000. The angle of attack that was needed to maintain the deep stall condition that was desirable was about 22 degrees. The runs that were performed here were the same runs that were performed when the Reynolds number was 75,000. The first run was with no actuators active. This was the separated flow case that was used for comparison. The separated flow can be seen in Figs and These figures demonstrate that the flow is separated from the airfoil, as is seen with the share layer in Fig (a) through (d), and with the high RFP seen in Fig The next four tests were performed using the LE actuator. The first test was to activate the LE actuator with a steady activation. The results are seen in Figs and Reattachment was achieved in this configuration and these two graph illustrate this. Fig (a) shows that the flow had been reattached to the surface of the airfoil while (c) and (d) both show where the LE actuator injected energy into the flow. Further proof of reattachment in this configuration is seen in Fig because the high RFP seen in the previous case is gone. The next configuration run was to pulse the LE actuator with F + = 1 to see if the flow would reattach. Figs and 4.62 show that the flow was not reattached when the LE actuator was pulsed with F + = 1. The last two runs that were performed on the LE actuator was to pulse the actuator with a forcing frequency of 10 Hz and 15 Hz or F + = and F + = respectively. The results for F + = are seen in Figs and For this case it was observed that the flow was reattached along the surface of the airfoil. For the last test performed on the LE actuator with F + = 0.222, the results are presented in Figs and It can be seen that when the LE actuator was pulsed with F + = 0.222, that the results were very similar to the case when the LE actuator was pulsed with F + = The last set of runs performed at this Reynolds number were to have both actuators active. Three runs were completed for this configuration. The first run was to have 89

102 the actuators active with a steady activation. The results from this run are pictured in Figs and Just as was observed for the LE case, this configuration also reattached the flow to the surface of the airfoil. It can be observed that the RFP in Fig along the surface of this case was lower than that of the case when just the LE actuator was used as pictured in Fig The next test was to run both actuators pulsed with F + = Results are seen in Figs and These figures illustrate that the flow was reattached for this configuration as was recorded before with just the LE actuator active. It can be observed that the aft actuator is active by examining the RFP in Fig The last test run at this Reynolds number was to pulse both actuators at F + = The results from this test are presented in Figs and The flow reattached for this case as can be observed in these two figures. Figure 4.57: Actuators Off (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE 90

103 Figure 4.58: Actuators Off Reverse Flow Probability within the Flow Field Figure 4.59: Leading edge Actuator, Constant Activation (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE 91

104 Figure 4.60: Leading edge Actuator, Constant Activation Reverse Flow Probability within the Flow Field Figure 4.61: Leading edge Actuator, Pulsed Activation with an F + = 1 (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 92

105 Figure 4.62: Leading edge Actuator, Pulsed Activation with an F + = 1 Reverse Flow Probability within the Flow Field Figure 4.63: Leading edge Actuator, Pulsed Activation with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 93

106 Figure 4.64: Leading edge Actuator, Pulsed Activation with F + = Reverse Flow Probability within the Flow Field Figure 4.65: Leading edge Actuator, Pulsed Activation with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 94

107 Figure 4.66: Leading edge Actuator, Pulsed Activation with F + = Reverse Flow Probability within the Flow Field Figure 4.67: Steady Activation on the LE and Aft Actuators (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 95

108 Figure 4.68: Steady Activation on the LE and Aft Actuators Reverse Flow Probability within the Flow Field Figure 4.69: Pulsed Activation on the LE and Aft Actuators with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity 96

109 Figure 4.70: Pulsed Activation on the LE and Aft Actuators with F + = Reverse Flow Probability within the Flow Field 97

110 Figure 4.71: Pulsed Activation on the LE and Aft Actuators with F + = (a) Flow Field Velocity Vectors, (b) Flow Field Vorticity, (c) Flow Field Urms, (d) Flow Field TKE 98

111 Figure 4.72: Pulsed Activation on the LE and Aft Actuators with F + = Reverse Flow Probability within the Flow Field 99

112 Profiles for each of the runs performed for Reynolds number 100,000 are shown in Fig Each one of the profiles was plotted together to show comparisons between all the runs. The black line represents the baseline separated case. The four runs performed on the LE actuator are shown by the red line. The solid red line is the constant activation case, the dashed line is the pulsed case at F + = 1, the dashdot line is the pulsed case with F + = and the dotted line is the case with F + = The cases when both the LE and aft actuators were activated together are represented by the blue lines. The case with constant activation is the solid line, the dashed line is F + = 0.148, and the dash-dot line is the case with F + = Fig 4.73 shows that every case except one reattached the flow. The only case that did not reattach the flow was the case where the LE actuator was activated with F + = 1. The bottom part of Fig 4.73 shows that when the LE actuator was active that it pulled flow inward towards the actuator from the flow above the actuator. The vertical fluctuation for this Reynolds number was less evident due to the increase in freestream velocity. 100

113 Figure 4.73: Velocity Profiles for 100,000 Reynolds Number Cases: Solid Black, No Control; Solid Red, LE Steady; Dashed Red, LE Pulsed F + = 1; Dash-Dot Red, LE Pulsed F + = 0.148; Dotted Red, LE Pulsed F + = 0.222; Solid Blue, Both Steady; Dashed Blue, Both Pulsed F + = 0.148; Dash-Dot Blue, Both Pulsed F + =

114 CHAPTER 5 Discussion and Conclusions 5.1 Discussion We have investigated the use of plasma actuators for airfoil separation flow control. We first examined the impact of actuator configuration on the jet velocity and momentum, namely actuator material or dielectric constant, for a variety of input parameters including plasma frequency, modulation frequency, and voltage difference. Once an optimum configuration for maximum jet velocity was determined, this was applied to controlling separation over a La203a airfoil at low Reynolds numbers. From the bench-top tests, it was observed that depending on what task is being performed, that different dielectric materials are better for particular situations. In the case that a material s physical properties do not matter as much as achieving the strongest plasma jet possible, the alumina dielectric material was the best choice for the job. If a material is needed to perform at a wider range of input voltages without the material failing, then acrylic is the best material for a mid-range plasma jet generation. As was required for this investigation, the material s ability to be formed to a surface was the most desirable parameter. It would have been desirable if the alumina or acrylic could have been used for this task, but both materials lacked the flexibility or manufacturability that the Teflon could provide. So for the task of affixing a plasma actuator to the surface of an airfoil, Teflon provided the needed flexibility, so to write. While investigating the La203a airfoil, it was seen that it has a tendency to stall at the trailing edge and with the separation point moving forward as the angle of 102

115 attack is increased, which is the typical trend for a fat airfoil. Two strategies were investigated for separation flow control. By placing a plasma actuator near the leading edge the plasma jet can affect the leading edge separation bubble. By placing an actuator at the 40% chord the plasma jet is adding momentum close to the separation point at higher angles of attack and impacting the incipient separation. The added momentum being added to the flow in these key locations provides the best chance to maintain flow attachment. From the wind tunnel test data, boundary layer profiles were graphed for each of the runs. The boundary layer profiles provides critical information about the flow characteristics over a surface, such as if the flow is laminar, turbulent or even if the flow is separated. Two additional boundary layer parameters that were calculated from the profile data for all the tests include δ and θ, the displacement and momentum thicknesses respectively. δ and θ measures the mass and momentum flux within the flow. These equations are very useful all the way until separation occurs, once the flow is separated these two parameters become ill defined. used in conjunction with the profile or skin friction data. Thus, these should be As an example of this, Fig. 5.1 illustrates the case where no flow control was active for a Reynolds number of 50,000 at an angle of attack of approximately 20 degrees. If the boundary layer was just showing typical growth behavior the slope should not be negative anywhere. Similar observations were seen for many of the cases where the flow was separated or unaffected by the plasma actuators, such as seen in Fig For a case where the plasma actuators were used to control separation, it was observed that the two parameters had a negative slope. This negative slope demonstrates that the boundary layer growth is being reversed and that the boundary layer is shrinking. The test case where the LE actuator was run under constant activation at a Reynolds number of 50,000 is pictured in Fig Cases where both actuators were used had results very similar to those pictured in Figs. 5.4 and

116 Figure 5.1: δ and θ Vs. x/c for 50,000 Reynolds Number, Actuators Off Figure 5.2: δ and θ Vs. x/c for 50,000 Reynolds Number, Aft Actuator, Constant Activation 104

117 Figure 5.3: δ and θ Vs. x/c for 50,000 Reynolds Number, Leading Edge Actuator, Constant Activation 105

118 Figure 5.4: δ and θ Vs. x/c for 75,000 Reynolds Number, Constant Activation on the LE and Aft Actuators 106

119 Figure 5.5: δ and θ Vs. x/c for 100,000 Reynolds Number, Pulsed Activation on the LE and Aft Actuators with F + =

120 Specific configurations of SDBD plasma actuators for separation control depend on the particular application. For a situation where the only goal is to control separation at the LE and power consumption is not a factor, then using just a LE actuator is a good solution, since separation was controlled across the range of Reynolds numbers investigated. Another option would be to use an array of actuators starting at the LE and moving aft to control separation over a much wider range of the airfoil, as was seen when both the LE and aft actuators were used together. When power consumption is a concern, using an actuator in a pulsed fashion as was done in this investigation, the actuator uses 50% less power because it is only on for half the time. Using two pulsed actuators at a duty cycle of 50% consumes the same amount of power as one actuator alone, so in cases where separation is an issue in multiple locations, multiple pulsed actuators would be a better choice. 5.2 Conclusions This thesis had several objectives. The first objective was to examine how different geometries in a SDBD plasma actuator affect each of the three dielectrics in bench-top testing. The three dielectrics tested were Teflon, acrylic, and alumina. The results from the bench-top testing were then implemented in wind tunnel tests. The second objective was to attempt to control separation on an airfoil at different Reynolds numbers using different plasma actuator configurations at post-stall angles of attack. Bench-top testing demonstrated that the alumina produced the strongest plasma jet of the three dielectric materials, with a maximum jet velocity of 161 cm/s at a plasma frequency of 8,000 Hz. The alumina also proved to be a very brittle material and would not be the ideal material for any surface with curvature, unless it was machined that way from the start. The alumina was relatively unaffected by changes in plasma frequency. The acrylic demonstrated that it could withstand a much wider 108

121 range of input voltages than that of the other two dielectrics, but when at some of the lower voltages, the plasma jet was very weak and unusable. The acrylic had a maximum input voltage of 80 kv and produced a jet velocity of 146 cm/s at a plasma frequency of 7,000 Hz. The acrylic was more sensitive to plasma frequency than the alumina. The acrylic proved to be less brittle than the alumina but was still not malleable enough to conform to a surface unless designed to do so. The Teflon was shown to produce a plasma jet velocity of 110 cm/s at a plasma frequency of 15,000 Hz, and was affected greatly by varying the plasma frequency. This jet velocity was achieved with an input voltage of 50 kv. Teflon is a very flexible material, therefore making it an ideal material for a highly curved surface like an airfoil. Even though the Teflon did not produce the strongest jet velocity it was chosen as the dielectric material to be used on the wing test model in the wind tunnel for several reasons: One it produced a moderately strong plasma jet velocity, also Teflon actuators were easy and very cheap to make, and finally, Teflon provided the flexibility that was needed to attach an actuator to the LE of an airfoil. Wind tunnel testing demonstrated that plasma actuators were effective for Reynolds numbers up to 150,000 at the power levels used herein. For tests completed at 50,000 Reynolds number, all cases tested were able to reattach a separated flow over the La203a wing model except when the aft actuator was run alone. This was the case because when the wing was placed at a 20 degree angle of attack, it was in a deep stall configuration and the actuator was just not able to draw the air flow back to the surface. The wind tunnel test matrix was adjusted upon observation of these tests and can be seen in Fig As it is seen in the test matrix a reduced frequency of F + = 1 was used for the pulsed cases at this Reynolds number, and separation control was achieved as pictured in Fig No other reduced frequencies were investigated at a Reynolds number of 50,000 because separation control was achieved with F + = 1. For Reynolds number of 75,000 similar results were recorded for many of the 109

122 test cases. Separation control was achieved for all test cases except when the LE actuator was pulsed at F + = 1. An array of reduced frequencies was tested to examine if reattachment could be achieved for a pulsed configuration. F + was examined from < F + < It was observed that F + = O (1) and F + > 1 separation control did not work. While testing the lower frequencies, a modulation frequency of 10 and 15 Hz was observed to reattach the flow, corresponding to reduced frequencies of F + = and F + = 0.297, respectively. Frequencies higher than this had no effect on the separated flow and were also not considered for further testing as is seen in the test matrix. The same set of tests were run at 100,000 Reynolds number. Separation control was achieved for test runs in this Reynolds number. For pulsed cases F + = 1 was tested again and still had no effect, so the lower frequencies found for 75,000 were examined. Both modulation frequencies of 10 and 15 Hz, corresponding to reduced frequencies of F + = and F + = 0.222, reattached the separated flow. During the initial testing of 150,000 Reynolds number, no separation control was achieved for any cases including both steady and pulsed activation case with F + corresponding to F + = 1, and This is due to the limited control authority of the plasma actuator. Fig. 5.6 summarizes the separation control demonstrated during the pulsed activation cases. According to the literature F + > 1 should be a more effective control strategy, but on the contrary it was observed that F + < 1 improved flow control performance that F + = O (1) did not demonstrate. In conclusion, it was demonstrated that separation control can be achieved using plasma actuators while at lower Reynolds numbers. 110

123 Figure 5.6: F + Vs. Reynold Number for Pulsed Activation Tests 111

124 5.3 Recommendations In this thesis bench-top testing of three common dielectric materials was performed. The state of the art of plasma actuators and their materials has been constantly evolving and will continue to evolve. There are several tests that need to be performed. First, this research should return to the bench-top to evolve the science along with the new technology. A list of all materials needs to be created, then systematically tested for a variety of parameters. Test all materials in the same configuration to examine their performance. Then test other parameters such as; material thickness, electrode configuration, and power to examine their effects. Develop some trends that illustrate how increasing material thickness effects the velocity produced by the plasma jet. Then examine how the electrode configuration affects the jet velocity by testing several configurations such as; have the two electrodes with different widths, have a gap (or overlap) between the trailing edge of the exposed electrode and the leading edge of the embedded electrode, test how different shapes of the electrode effect like a chevron or serpentine configurations. Finally test each material for an operational voltage range, so to develop an upper limit for a particular material. With that upper limit documented, that actuator can then be operated for a longer period of time without burning out. A second set of tests that would need to be run on this new technology would be to take the best configuration from the bench-top and apply it to a series of wind tunnel tests. The testing should start on a flat plate to expand the science of these actuators to incorporate the effect of a fluid flow over the actuator. By developing this freestream test model, a change in velocity, ( V ), can be demonstrated. With a known V for a particular actuator an actuator can be chosen for a specific situation. Once the specifics for a particular actuator is known, test parameters like duty cycle, 112

125 how multiple actuators in an array, and phasing multiple actuators effect that change in velocity. Now with the V is known for a vast amount of actuator designs, create a test case that has a constant pressure gradient or known pressure gradient. With these test cases develop a method for comparing the V for a known actuator to a dimensionless pressure gradient coefficient, such as Pohlhausen or Thwaites parameter [26]. With this relationship it will be known if a particular plasma actuator can control a specific pressure gradient. Separation control has been demonstrated in this paper, but not other aerodynamic performance effects. Activation of the plasma actuators created a vortex and body force that effected the flow. It was not examined in this thesis whether these entities improved or reduced the aerodynamic performance of an airfoil. According to the literature, both an increase and decrease to lift has been seen. Lift, drag and moment need to be measured to examine the aerodynamic effects of these actuators. The setup illustrated above needs to be placed on a lift balance and run with the same settings as were used in this investigation to get the lift and drag data. A test wing with static pressure ports needs to be created and tested to gather the pressure distribution and moment calculations. This investigation concentrated on orienting the plasma actuators such that the plasma jet was generated in the direction of the flow. With the jet positioned in this manner the momentum is injected into boundary layer in the flow direction, thereby energizing the downstream flow. A further investigation needs to be performed into what effect a plasma actuator positioned to generate a counter-flow plasma jet. 113

126 APPENDIX A High Re Testing The appendix contains results from testing a larger La203a airfoil in the OSU subsonic wind tunnel at high Reynolds numbers. Due to time constraints and limited control authority found in the small scale testing, plasma actuator flow control tests were not performed. However, the baseline results are included here for completeness and to serve as a reference for future experiments. A.1 Wind Tunnel Setup In a joint effort between the University of Kentucky (UK) and Oklahoma State University (OSU) a larger Liebeck La203a wing was made using a SLA rapid prototyper. This larger wing had a span of 36 inches and a chord length of 8 inches. This wing fit snugly into the OSU large wind tunnel. This wing was designed with a series of pressure ports designed to measure the coefficient of pressure along the top and bottom side of the wing. A total of 25 pressure ports were designed into the wing, 12 on the top, 12 on the bottom and one at the leading edge. A 3D CAD model for this wing is shown in Fig. A.1 and a finalized SLA wing is illustrated in Fig. A

127 Figure A.1: La203a Large Wind Tunnel Wing CAD Model 115

128 Figure A.2: La203a Large Wind Tunnel Wing 116

129 The OSU low-speed subsonic wind tunnel will serve as the main test facility for the laboratory tests. A schematic of the OSU wind tunnel is illustrated in Fig. A.3. The low turbulence wind tunnel is an open loop wind tunnel with 1:16 contraction and has a clear test section with a cross-section of 1x1m and 2 m long with swappable test sections. A 125 hp centrifugal fan powers the tunnel and has a top speed of 70 m/s. Tunnel speed is controlled with a feedback control mechanism and monitored by multiple Pitot-static probes. The tunnel is instrumented with wall mounted lift and drag balance and a traversing Pitot-static probe. The layout of the lift and drag balance and Pitot-static probe can be seen in Fig. A.4 and Fig. A.5. Two wall mounted balances are placed on both sides of the test section, mounted on an external bracket. Each balance consists of two Transducer Technologies strain gages that be can tailored to the specific tests load expectations. The strain gages are conditioned using a model 2120B Vishay Strain Gage Conditioner. Each load cell is calibrated separately using calibration weights prior to each wind tunnel test. Labview and a NI USB-6158 DAQ unit is used to monitor and record the data at typically 1 khz. The pyramidal balance is a 6 component Aerolab model with load limits for lift, drag, and side forces of 275, 85 and 95 lbs, respectively, and 720 in-lbs for pitching, yawing, and rolling moments. High fidelity velocity data are obtained using either a hot-wire or PIV system (discussed above). Multiple Dantec MiniCTA hot-wire systems are available to measure velocity and velocity fluctuations at multiple points in the tunnel simultaneously. Both single and two component hot-wires are available. Pressure measurements were taken using a bank of water manometers. This bank of manometers consists of 50 individual water filled manometers, which are all linked to an adjustable reservoir. The manometer has a 44 inch range, with 22 inches for positive pressures and 22 inches for negative pressures. Each one of the manometer 117

130 tubes can then be hooked to individual ports to allow for a pressure profiles. The manometer bank is illustrated in Fig. A.6. Figure A.3: OSU Large Low-Speed Wind Tunnel Figure A.4: OSU Wind Tunnel Test Section with Lift/Drag Balances and Pitot-Static Tube for Wake Surveys 118

131 Figure A.5: OSU Wind Tunnel Test Section with Lift/Drag Balances and Pitot-Static Tube for Horizontal Sweeps 119

132 Figure A.6: Manometer Bank 120