DEVELOPMENT AND CHARACTERIZATION OF EFFICIENT PLASMA ACTUATORS By PENGFEI ZHAO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2015
Â© 2015 Pengfei Zhao
To my Mom , parents in law and my wife, for all the trust and support on me
4 ACKNOWLEDGMENTS First and foremost, I would like to thank to my supervisor and committee chair, Dr. Subrata Roy for providing me a precious opportunity to do my graduate research in Applied Physics Research Group (APRG) over last five years. His willingness to motivate me contributed tremendously to this project. I also would like to thank him for his abundantl y help and invaluable assistance, support and guidance. Then, I am thankful to all my committee members for their support. I deeply appreciate the opportunity to work and consult with them. Specially, I would like to thank Dr. Lawrence Ukeiley for his hel p on the PIV analysis and Dr. Robert Moore for his help and kindness. Furthermore, I also would like to thank colleagues in APRG for their help ful discussion and kindness during past years . Thank you, Dr. Ankush Bhatia, Ariel Blanco, Arnob Das Gupta, Moses Divaker , Dr. R yan Durshcer , Tomas Houba, Dr. Navya Mastanaiah, Sherlie Portugal , Dr. Mark Riherd and Dr. Chin Cheng Wang. Lastly, I would like to thank my family. I would like to thank my mom, Peixian Liu, and my parents in law, Xinxia Shi and Jianxi n Lu, for all the love and support they gave me. I would like to thank my wife, Lu Lu, for her love, trust and support behind this work .
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ........................... 11 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 2 PHYSICS OF EFFICIENT PLASMA ACTUATORS ................................ ................ 17 2.1 Plasma Actuators ................................ ................................ .............................. 18 2.1.1 Definition of Plasma Actuators ................................ ................................ 18 2.1.2 Classification of Plasma Actuators ................................ .......................... 19 2.2 Corona Discharge Based Plasma Actuators ................................ ..................... 26 2.2.1 Physics of Corona Discharge and Electric Wind ................................ ..... 26 2.2.2 Previous Study on Corona Discharge Based Plasma Actuators ............. 33 3 DESIGN AND CHARACTERIZATION OF EFFICIENT PLASMA ACTUATORS .... 35 3.1 Experiment Setup ................................ ................................ ............................. 35 3.1.1 Electrodes Configuration ................................ ................................ ......... 35 3.1.2 Experimental Method ................................ ................................ ............... 37 3.2 Result and Discussion ................................ ................................ ...................... 39 3. 2.1 PIV Study ................................ ................................ ................................ 39 3.2.2 Power Consumption Study ................................ ................................ ...... 44 3.3 Conclusion ................................ ................................ ................................ ........ 46 4 SURFACE COOLING USING EFFICIENT PLASMA ACTUATORS ....................... 48 4.1 Experiment Setup ................................ ................................ ............................. 48 4.1.1 Electrodes Configuration Design ................................ ............................. 50 4.1.2 Heat Source ................................ ................................ ............................ 54 4.1.3 Electrical Supply System ................................ ................................ ......... 55 4.1.4 Thermal Measurement ................................ ................................ ............ 56 4.1.5 Velocity Measurement ................................ ................................ ............. 58 4.2 Result and Discussion ................................ ................................ ...................... 58 4.2.1 Voltage and Polarity Study ................................ ................................ ...... 61 4.2.2 Needle Height Study ................................ ................................ ................ 72
6 4.2.3 Radius of Curvature Study ................................ ................................ ...... 77 4.3 Summary of Surface Cooling Study ................................ ................................ .. 79 5 VORTICES GENERATION USING EFFICIENT PLASMA ACTUATORS ............... 81 5.1 Experiment Setup ................................ ................................ ............................. 81 5.1.1 Electrode Configuration Design ................................ ............................... 81 5.1.2 Testing Environment ................................ ................................ ................ 83 5.2 Result and Discussion ................................ ................................ ...................... 83 5. 3 Application on Surface Cooling ................................ ................................ ......... 84 5.3.1 Experiment Setup ................................ ................................ .................... 84 5.3.2 Results and Discussion ................................ ................................ ........... 86 5.4 Summary of Vortices Generation Study ................................ ............................ 91 6 SUMMAR Y ................................ ................................ ................................ ............. 93 6.1 Development on Efficient Plasma Actuators ................................ ..................... 93 6.2 Surface Cooling Performance Study ................................ ................................ . 94 6.3 Vortices Generation and Applications ................................ ............................... 95 APPENDIX : PIV UNCERTAINTY ANALYSIS ................................ ............................... 97 LIST OF REFERENCES ................................ ................................ ............................. 102 BIOGRAPH ICAL SKETCH ................................ ................................ .......................... 107
7 LIST OF TABLES Table page 2 1 Coeffiecents, A and B , used to calculate the breakdown voltage of air based on Pas . ................................ ................................ ............................. 22 2 2 Main characteristics of LTE and non LTE plasma . ................................ ............. 28 A 1 Mean PIV velocity uncertainty for specific parameters used in this work . ......... 101
8 LIST OF FIGURES Figure page 2 1 Titanium dioxide smoke jet illustrates that paraelectric neutral gas flow is induced by a low pressure region associated a DBD pl asma actuator . .............. 20 2 2 Voltage and current characteristic for a DC discharge in a low pressure discharge tube. ................................ ................................ ................................ ... 21 2 3 Different electrode geometrical configurations found in the literature as dc plasma actuat ors. ................................ ................................ ............................... 25 2 4 Schematic of a point to plate corona discharge configuration. ........................... 30 2 5 Schematic of corona discharges. ................................ ................................ ........ 31 2 6 Discharge model schematic. ................................ ................................ ............... 34 3 1 Channel electrodes configuration of needle actuation device . ............................ 36 3 2 PIV velocity profiles . ................................ ................................ ........................... 40 3 3 XY plane velocity fields. ................................ ................................ ..................... 41 3 4 Y axis velocity line profiles at downstream of the exit (X = 2 mm) of the needle actuation channel. ................................ ................................ ................... 43 3 5 Z axis line velocity profile at downstream of the exit (X = 2 mm) of the needle actuation channel for positive and negative 12 kV DC signal . ............................ 44 3 6 Electrical power consumption and maximum inducing velocity of the needle actuation channel for various applied positive and negative vol tages. ............... 45 3 7 Energy conversion efficiency of the needle channel for various applied positive and negative voltages. ................................ ................................ ........... 46 4 1 Experimental setup inside the transparent testing chamber (Not to scale). ........ 49 4 2 Side view of the electrodes configuration for surface cooling effect experiment with dimensions. ................................ ................................ .............. 51 4 3 Vertical acrylic holder for needle electrodes (SolidWorks Drawing). ................... 51 4 4 Example of the estimation process on radius of curvature of a thin needle tip. .. 53 4 5 Design of the cooling performance test section including needle actuator and semi insulated rubber heater (SolidWorks Drawing). ................................ ......... 54
9 4 6 Top view of the thermocouple locations on the rubber heater for the first method of thermal measurement (not to scale). ................................ ................. 56 4 7 IR camera setup above the rubber heater for the second thermal measurement method of the surface cooling test (not to scale). ........................ 57 4 8 Cooling effect of needle actuation device setup (total power consumed by all three needles shown in legend). ................................ ................................ ......... 60 4 9 XZ plane X direction velocity profile (Y = 10 mm) of a single needle actuator. ... 62 4 10 XZ plane Z direction velocity profile (Y = 10 mm) of a single needle actuator. ... 63 4 11 X direction Velocity profile with negative DC signal powered on needle. ............ 64 4 12 X direction Velocity profile with Positiv e DC signal powered on needle. ............. 65 4 13 Comparison of the X direction velocity profile between positi ve and negative DC signal powered on the needle . ................................ ................................ ...... 67 4 14 XY plane X direction velocity profile (Z = 2 mm) of a single needle actua tor. ..... 68 4 15 X direction Velocity profile with Positive DC signal powered on needle. ............. 69 4 16 Distribution of the enhancement factor of the local heat convection coefficient over the rubber heater with different voltage magnitude. ................................ .... 70 4 17 Average ehancement factor as function of applied voltage and polarity. ............ 71 4 18 Comparison of the X direction velocity profile between different height of the needle. ................................ ................................ ................................ ................ 72 4 19 Distribution of the enhancement factor of the local heat convection coefficient over the rubber heater at 6kV with different needle height. ................................ 73 4 20 Distribution of the enhancement factor of the local heat convection coefficient over the rubber heater at 8kV with different needle hei ght. ................................ 74 4 21 Distribution of the enhancement factor of the local heat convection coefficient over the rubber heater at 10kV with different needle height. .............................. 75 4 22 Average ehancement factor as function of needle height and applied volatge magnitude . ................................ ................................ ................................ .......... 76 4 2 3 Radius of curvature estimation with microscopic pictures. ................................ . 77 4 2 4 Comparison of the X direction velocity profile between different radius of curvature of the needle. ................................ ................................ ...................... 78
10 4 2 5 Average ehancement factor as function of radius of curvature of the needle tip. ................................ ................................ ................................ ....................... 79 5 1 Electrodes configuration for surface cooling effect. ................................ ............ 82 5 2 Controllable vortices generated by two needle actuator at X = 10 mm. ............. 85 5 3 Needle configurations for number of needle test. ................................ ............... 86 5 4 XY plane X direction velocity profile (Z = 2 mm) of needle actuators with different number of needles. ................................ ................................ ............... 87 5 5 X direction Velocity profile with Positive DC signal powered on different number of needles. ................................ ................................ ............................. 89 5 6 Distribution of the enhancement factor of the local heat convection coefficient over the rubber heater at 10kV with different number of needles. ...................... 90 5 7 Average heat convection enhancement factor as a function of number of needles. ................................ ................................ ................................ .............. 91 A 1 Standard deviation as a function of the number of samples used . ................... 100
11 LIST OF ABBREVIATIONS CPU Central P rocessing U nit DBD Dielectric Barrier Discharge DC / AC Direct Current / Alternating Current eV Electronvolt (unit of energy, equal to ) HV High Voltage IC I ntegrated C ircuit IR I nfrared LED Light E mitting D iode LTE Local Thermodynamic (or Thermal) Equilibrium PIV Particle I mage V elocimetry
12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosoph y DEVELOPMENT AND CHARACTERIZATION OF EFFICIENT PLASMA ACTUATORS By Pen gfei Zhao December 2015 Chair: Subrata Roy Major: A erospace Engineering The objective of presented work is to conduct an in depth study for efficient low power plasma actuators for various applications. Needle corona discharge within a channel electrodes configuration was employed to design the efficient low power plasma actuators . Compared to traditional dielectric barrier discharge ( DBD ) plasma actuat ors, corona discharge based efficient plasma actuators can increase the energy conversion efficiency from 0.7% to 2.8 %. Studies on flow structure and power consumption have been done by analyzing data from PIV and electrical parameter measurement. Invest igation on the input signals with different voltage and polarity shows an influence on the actuators, with positive polarity yielding better energy conversion efficiency . Then the mechanism of this influence is also studied through comparing velocity profi les between cases with different polarity. Capabilities of efficient cooling and vortices generation have also been observed. Furthermore, parameters of needle actuator configuration, including needle height, tip size and number of needles, were investigat ed on its surface cooling performance . A Glauert like wall jet is generated by the needle plate actuator to
13 enhance the forced convection on the hot surface. The wall jet velocity increases with the increasing voltage and/or the decreasing radius of curvat ure. And the needle location also plays an important role in the flow generation. Vortices can also be generated with two or more needle actuator s promoting mix ing of air near a surface.
14 CHAPTER 1 INTRODUCTION Atmospheric plasma driven active flow control devices ha ve been extensively studied in recent years . Applications of these devices range from the control of the laminar to turbulent transition (Riherd & Roy, 2013) to the drag reduction (Jukes, 2007) . The advantages o f dielectric barrier discharge ( DBD ) driven plasma actuators include lack of moving parts, fast response, small scale, surface compliance, and ease of construction and application. Major disadvantage of these devices is that the ir energy conversion efficie ncies are extremely low , only one tenth of a percent (Moreau, 2007; Riherd & Roy, 2012) . This is because of a large amount of energy loss in light emiss ion as well as heating of both dielectric material and plasma. Also the momentum injection to th e neighboring flow is largely limited due to high viscous losses encountered in traditional DBD wall jets. As compared to DBD, direct current ( DC ) corona discharge only generates minimal glow near the surface of power ed electrode without using an y dielectr ic material. Thus, the energy loss mentioned above can be avoided . However, the energy conversion efficiencies of DC corona devices remain less than 1% (Colas, Ferret, Pai, Lacoste, & Laux, 2010; Mor eau, 2007) . This is again due to excessive viscous loss from the thin wall shear layer or/and discharge loss near the wire surface. In contrast, the channel configuration (Campbell & Roy, 2014) with needle electrode can be employed to min imize the ionization loss and the viscous loss from flow induced by plasma body fo rce . Typica lly, atmospheric DC corona discharge devices consist of a powered electrode and a grounding electrode separated by air . The p owered electrode usually ha s a sharp tip or round edge with a minute radius of curvature. The ground ing
15 electrode often ha s a large smooth surface. Application of high voltage DC signal to the powered electrode forms a surface corona discharge around the sharp tip or the round edge where the maximum electric field is generated . The ground ing electrode i s placed some distan ce away to generate desired electric field to induce the ions formed by impact ionization . Then, collisional momentum transfer between ions and neutral particles occurs in the space between the electrodes. Both positive and negative DC voltages can be used to induce flow with slightly different mechanisms (Moreau, 2007) . For the positive corona, ionization region is generated by anode electron avalanche near the electrode. Then, the Coulom b force repels positive ions toward the cathode. In the case of negative corona, negative ions are created by electron attachment and then rep ell ed toward the anode . T his study experimentally explores a class of DC corona needle actuation device s that has been developed to generate moderate directed airflow with extremely low power consumption , and thus improving the energy conversion efficiency by an order of magnitude than its DBD counterpart . Particular electrode arrangements for these devices showed gre at benefits for cooling and vortex generation applications. The goal of this study is to exploring the capability of corona discharge plasma actuators and their performance on different conditions. The contents of this document can be broken into several s ections. In chapter 2, fundamental concepts of plasma actuators including definition and classification a re discussed. The desired type of plasma actuators is specified and compared with others. Furthermore, comprehensive physical characteristics on corona
16 discharge plasma and electric wind induced by corona discharge based actuators are described . Experimental investigation is conducted to analyze the performance of the needle actuation devices under many different conditions in chapter 3 . Electr ical properties and velocity field of the airflow induced are obtained from the experiment . R esults are presented demonstrating huge increase in energy conversion efficiency . In chapter 4 and 5 , experimental analyses on applications of needle actuation devices are addressed . The surface cooling performance and vortices generation using newly designed actuators will be investigated . Finally, a summary of highlights in this the study is presente d in chapter 6, along with the recommendations for the future work .
17 CHAPTER 2 PHYSICS OF EFFICIENT PLASMA ACTUATORS An actuator , by definition, is a type of motor that is responsible for moving or controlling a mechanism or system. Particularly, actuators mentioned in this work are referring to the devices which can move the working gas fluid in desir ed manor. Generally , researchers develop these actuators to manipulat e fl ow to affect desired change in thermal and/or fluid systems. For example, efficient actuators could modify the lami nar tur bulent transition (Riherd & Roy, 2012) inside the boundary layer to reduce the drag and to enhance the lift of airfoils . Al so these actuators could be used to adding or removing heat flux from any surface (Go, Maturana, Fisher, & Garimella, 2008) . This is of larg e technological importance for industries where thermal and/or fluid system are involved, and more specifically for mechanical and aerospace engineering. Among all the actuation methods, a new and original technology using non thermal plasmas is in full expansion. Indeed, although mechanical devices may be effective, they have many dra wbacks. In particular, they are complicated, add significant amount of weight, have volume and are sources of noise and vibration. Moreover, they are composed of mechanical parts that wear away and tha t may break down. Consequently, plasma actuators are now in ful l expansion because they do not exhibit all these drawbacks. The main advantage of plasma actuator is that it directly converts electric energy into k inetic energy w ithout involving moving mechanical part s. Thus it may be considered to be a very simple MEMS. Secondly, its response time is very short and enables a real time control at high frequency.
18 This chapter reviews the research status of the plasma actuators . The emphasis in this chapter is on the definition, classification and physical description of the plasma actuators , especially ones with corona discharge . 2.1 Plasma Actuators Generally , the plasma actuator is composed of at least two electrodes mounted near the wall between which a high voltage is applied, resulting in a non thermal plasma region . In most of the cases , plasma actuators can generate the discharge induced gas flow jet to modify the properties of boundary layer and then to actively manipulate th e airflow. The gas flow jet induced by discharged gas in electric field also can be called as electric wind. Fig ure 2 1 shows a manipulated airflow along a flat plate with a plasma actuator . In this section, definition and classification of plasma actuators will be described and discussed. Different types of plasma actuators will be compared. Then, this comparison will be used to design more efficient plasma actuators. 2.1.1 Definition of Plas ma Actuators The plasma actuators discussed in this section r efer to the plasma based device s. Such devices can generate secondary air flow, usually called electric wind or ionic wind, in atmospheric condition. In order to induce the secondary airflow in am bient air, plasma actuators should create both an atmospheric pressure plasma and a strong electric field through the plasma region. Usually, plasma is also generated by the strong electric field. Consequently, all charge particles inside plasma will be ac celerated by the strong electric field. This process can also be described as the effect of body force, which can be expressed approximately as:
19 (2 1) where is the electric field generated by plasma actuators and represents the net total charge of the particles in the electric field . Then, due to the collision between neutral particles and accelerated charged particles, momentum will be transferred from fast moving charged particles to slow moving neutral particles. In the momentum transfer process, since the mass of electrons can be n eglected in comparison with ions, the acceleration time of electrons is much shorter than that of ions which is at least 2000 times heavier. Thus, majority momentum gained by neutral particles comes from ions, both positive and negative. Eventually, airflo w generated consists of both neutral particles and charged particles at approximately the same velocity. 2.1.2 Classification of Plasma Actuators Generally, the plasma actuators work in atmospheric condition because most application needs this. Thus, class ification is done on this kind of actuators. Many different types of p lasma actuators working in atmospheric pressure condition have been developed and tested in past years . These actuators can be classified by their discharge mechanism and geometry design . Based on discharge physics , plasma actuators can be divided into three groups, DBD plasma actuator, corona plasma actuator and other plasma actuator. In each group, plasma discharge along with electrical body force can be generated near the surface or in space depending on the geometry design of these actuators. According to the statement above, t he m ajority of plasma actuator s studied is based on atmospheric pressure plasma . Over the past few decades, atmospheric pressure plasmas have been studied for nu merous industrial applications such as
20 ozone generation, pollutant removal and surface treatment. Several papers and books give a complete review of this type of discharge (J. Reece Roth, 1995; Schutze et al., 1998) . Due to the density of air at atmospheric pressure, the atmospheric pressure plasma is non thermal, which means that the ion temperature is not as high as the electron temperature. These non thermal plasmas may be produced by a variety of electrical discharges and have very low energy cost in comparison to thermal plasma. This is due to the fact that th e majority of the input electrical energy goes into the production of energetic electrons, instead of heating the surrounding gas. Figure 2 1. Titani um dioxide smoke jet illustrat es that paraelectric neutral gas flow is induced by a low pressure region associ ated a DBD plasma actuator (J. R. Roth, 2003) .
21 Figure 2 2. Voltage and current characteristic for a DC discharge in a low pressure discharge tube. Taken from (J. Reece Roth, 1995) . Generally, the gas discharge may be classified into three main types based on delivered current: dark discharge, glow discharge, and arc discharge. Figure 2 2 displays the three regimes in a DC voltage current (VI) plot. The dark discharge, also called Tow nsend discharge, is characterized by very little light emission. In this regime, the electron generated from other source (cosmic ray or photoionization , (Howatson, 1976) ) is accelerated by the electric field. For a constant electric field, the energetic electron can ionize neutral par ticles to multiple the electron number density. This phenomenon is called electron avalanche (Raizer, 1991) . Thus, the discharge is self sustained th rough electron avalanche. After the current start s to increase , the breakdown of the gas occurs . D u ring this process, some neutral particles get excited by the energetic electrons and emits photons. Thus, the plasma starts to become luminous. The breakdown voltage of the gas is found to have a function of the ambient
22 working gas pressure ( ) and discharge gap ( T he breakdown voltage, , of a working gas can be calculated by (A. A. Fridman, 2008) (2 2) where and are gas dependent constants and is the secondary electron coefficient for the cathode. Table 2 1 provides A and B values for air. Table 2 1. Coeffiecents, A and B , used to calculate the breakdown voltage of air based (A. A. Fridman, 2008) . Gas Air 15 365 After this breakdown happens, the plasma is said to enter the glow discharge regime. In this regime, positive ions also gain a high speed towards cathode . When the heavy ions collide with cathode, it can give electrons on the cathode surface enough energy to escape. This phenomenon is called secondary electron emission, which is the primarily sustain ing mechanism for the glow discharge (Howatson, 1976) . If the current through plasma is allowed to increase furthermore , the glow to arc transition eventually occurs with a significant drop in voltage . The voltage drop is due to the reduction in resistance inside the plasma region. G enerally , hot cathode and hot plasma are accompanie d with h igh current arc discharge . These phenomen a consume large amount of energy and may also causes damage. B riefly, the formation of steady discharge under atmospheric pressure is mainly based on the electron avalanche or secondary electron emission . The discharge mechanisms used for plasma actuators are usually atmospheric pressure corona discharges (dark discharge) and dielectric barrier discharges ( special form of glow
23 discharge). Their characteristics are typically as follows: high voltage of a few kV to several tens of kV with DC or AC excitation, with a frequency from 50 Hz to 50 kHz (only for AC or pu lsed DC), electrical current from a few n A to a few mA. In these conditions, the density of charged species is between 10 9 and 10 13 cm 3 and the electron temperature is a few eV. The first studies of plasma actuators generating airflow by electrical discha rges dealt with DC corona discharges . For example, it is demonstrated that the transition point on a flat plate could be changed by the application of an electric field (Velkoff & Godfrey, 1979; Velkoff & Ketcham, 1968) . In this application, the plasma actuator with four wire electrodes is placed above the surface of a flat plate and powered with HV DC signal . N o grounded electrode is used in this application . Thereafter, corona discharges are widely implemented into many different plasma actuators for different applications . There are two main advantages of the corona discharge based plasma actuators. First, DC input signal maintains a constant strong electric field to accelerate charged particles. Second, very small amount of power is needed to maintain the corona discharge. On the other hand, its disadvantage is the corona to arc transition when current is high. In order to prevent corona to arc transition, a dielectric barrier can be placed in the discharge gap, which stops the electric current between electrodes and prevents arc discharge formation. Such discharge is called DBD. Because of the charge accumulation on the dielectric barrier, HV DC signal canno t be used to generate plasma . Then DBDs are usually excited by AC or pulsed DC signals , with frequencies between 50 Hz and 500 kHz (A. Fridman, Chirokov, & Gutsol, 2005) . T he plasma is
24 generated in the electrode gap with consecutive micro discharges . These micro discharges , also called streamers, are distributed in time and space in a random manner . Generally, t hese streamers have a diameter of about 100 lifetime of a few nano seconds (with velocity of these streamer s in the range of 10 7 10 8 ms 1 ). I developed a novel glow discharge actuator worki ng at atmospheric pressure . It was a surface DBD actuator. This kind of actuator consists of at least two electrodes mounted asymmetrically on the two side s of a dielectric plate . It is shown that a secondary airflow of several ms 1 tangentially to the wal l is driven by the discharge . Meanwhile , when the applied voltage reaches 30 kV pp , the reaction force of the airflow can be as high as 11 mN . It is also demonstrate that this secondary airflow can modify the boundary layer of the free stream , with a drag modification as the result . After this original work, DBD based plasma actuators are extensive ly studied. The main advantage of DBD based plasma actuators is that they can generate strong steady glow discharge. Its disadvantages include joule heati ng in plasma (Tirumala et al., 2014) and saturation effect (Durscher, Stanfield, & Roy, 2012) . For both corona discharge and DBD mechanism, electrod es configuration can be design ed to generate surface plasma or volumetric plasma. For example, Fig ure 2 3 shows different corona discharge plasma actuators examined in literatures. Actuators displayed in Fig ure 2 3 A 2 3 D generate volumetric plasma between electrodes while surface plasma sheet are created by actuators in Fig ure 2 3 E 2 3 H . The airflow generated by volumetric and surface plasma source shows different characteristics. Since airflow induced by surface plasma actuators are close to the surface, viscous shear stress will play an important role in this kind of configuration. On the other hand,
25 both volumetric and surface plasma actuators are designed according to the needs of applications. Figure 2 3 . Differe nt electrode geometrical configurations found in the lite rature as dc plasma actuators: A) Volume tric multiple wire. B) V olume tric wire to plate in plane configuration. C ) V olume tric wire to plate in cylindrical configuration. D ) V olume tric multi hole plat e to plate configuration . E) Surface plate to plate. F) Surface wire to plate. G) Surface wire to wire. H) Surface wire to plate or point to plate in cy lindrical configuration (Moreau, 2007) . In summary, atmospheric pressure plasma actuators can be classified by their discharge mechanism as well as the geometry of their electrodes configuration. In
26 comparison with DBD plasma actuators, corona discharge plasma actuators only generate plasma near the sharp electrodes. And volumetric plasma actuators can avoid the viscous shear stress near the wall. In order to design a more efficient plasma actuator, corona discharge and volumetric electrodes configuratio n will be investigated further in the following section. 2.2 Corona Discharge Based Plasma Actuators As discussed in the precious section, corona discharge based plasma actuators only generate plasma corona near the sharp electrodes while maintaining a con stant strong electric field between electrodes . In the process of inducing airflow, energy used to heat and light is negligible. These characteristics make this type of plasma actuator great potential to improve their efficiency and to be used in industria l applications. In this section, the focus is on the plasma actuators based on corona discharge. In order to further understand the mechanism and characteristics of this type of plasma actuators, physics of corona discharge and previous work related to thi s type of plasma actuators will be reviewed. 2.2.1 Physics of Corona Discharge and E lectric Wind Plasma is a fully or partially ionized gas. It is defined as the fourth state of matter and constitutes more than 99% of the universe. E lectrons, ions and neut ral particle s in fundamental and excited states are the particles in the plasma . From point of view a t macroscopic level , plasma is electrically neutral in a certain volume . However, plasma also con tains free charge carriers . Thus, it is electrically condu ctive. A plasma is created by applying energy to a gas (Conrads & Schmidt, 2000) in o rder to break the electronic structure of the neutral particle s (atoms, molecules) and to
27 generate electrons, excited neutral particles and ions. This energy can be thermal, or carried by either an electric current or electromagnetic radiations. The atmosp heric corona discharge plasmas described in this study are generated by the energy from electrical current . The electric field gives energy to the free electrons (which have the highest mobility among the charged particle s) in gas . This energy carried by electrons can be then transferred to the neutral particles through collisions. These collisions (Braithwaite, 2000) follow probabilistic rules and can be divided in two distinguishable forms. The first form is elastic collisions. This kind of collisions only slightly r a ise their kinetic energy but do es not change the internal energy of the neutral species . On the other side , when energy carried by electron is high enough, the collisions change the electronic structure of the target particles . Then, the excited particles or ions (if energy carried by electron is even higher) can be generated . This kind of collisions is called inelastic collisions. Generally, a nother distinction of plasma can be made between plasmas at local thermodynamic (or the rmal) equilibrium (LTE) and at Non local thermodynamic equilibrium (non LTE). They are also known as thermal plasmas (LTE plasmas) and cold plasmas (non LTE plasmas). The primary property t hat distinguish es them is whether the electron temperature is the s ame as the heavy particle temperature or not. Detailed properties of LTE and non LTE plasmas are listed in Table 2 2. The LTE notion (Griem, 1962) is really important, especially for a spectroscopic study of the plasma, since the determination of the plasma parameters (electron, excitation, vibration temperatures and so on ) is based on whether the plasma is in LTE or not.
28 Table 2 2 . Main characteristics of LTE and non LTE plasma (Claire, et al, 2006) LTE plasmas Non LTE plasmas Current name Thermal plasmas Cold plasmas Properties High electron density: Inelastic collisions between electrons and heavy particles create the plasma reactive species whereas elastic collisions heat the heavy particles (the electrons energy is thus consumed) Lower electron density: Inelastic collisions between electrons and heavy particles induce the plasma chem istry. Heavy particles are slightly heated by a few elastic collisions (that is why the electrons energy remains very high) Examples Arc plasma (core) Glow discharges Those rules for LTE are very strict. Thus , most of the plasmas are not in LTE, especially all types of low density plasma in laboratories. Apparently, the corona discharge plasmas discussed in this work is non LTE plasma with a heavy particle temperature close to room temperature. In the non LTE plasma, e lectrons with much higher temperature move much fast er than heavy particles which can be considered static . Thus, the majority of collisions and transitions phenomena is governed by energetic electrons . As a non LTE plasma source, corona discharges are relatively low power electrical discharges that take place at or near atmospheric pressure. The corona is invariably generated by strong electric fields associated with small radius of curvature, such as small diameter wires, needles , or sharp edges of the electrode. The corona discharge looks like many filamentary discharge s radiating outward from the powered electrode with small radius of curvature . Because the corona is relatively easy to create
29 in atmospheric condition , it has alr eady been applied in a variety of industry and research areas . Corona discharge is a non LTE discharge with low current density and relatively high voltage . D epending on polarity of the input signal , c orona discharge could be classified as a Townsend disch ar ge or a negative glow discharge (Schutze, et al., 1998) . Figure 2 4 shows a design of the standard point to plane corona discharge setup , which can create a strong electric field thus a n ionization zone around the point w here the high voltage is applied. For different polarity of the input signal, t he mechanism is different to sustaining the continuous ionization near the high voltage electrode. If the point electrode is powered with negative HV signal , it is called a nega tive corona. If the point electrode is powered with positive HV signal , then it is a positive corona. The HV input signal can be continuous or pulsed DC signal or even AC signal . Typically, for the positive corona discharge operating with this point to pla ne configuration, the power and input DC voltage follow a relationship (Moreau, 2007) : (2 3) where is the time average electrical power consumed, the potential difference between electrodes, the ignition voltage of the corona discharge and a constant depending 2 in air (Z ouzou, Moreau, & Touchard, 2006) ).
30 F igure 2 4. Schematic of a point to plate corona discharge configuration (Moreau, 2007) . Corona discharges can be classified in several forms, mainly depending on the polarity of the field and the electrode geometrical configurations. For positive corona in the needle to plate configuration, discharges start with burst pulse corona and proceed to the streamer corona, glow corona, and spark discha rge while the applied voltage increases (Figure 2 5). For negative corona in the same geometry, the initial form will be the Trichel pulse corona, which is followed by a pulseless corona and then a spark discharge as the applied voltage increases. For a wi re pipe or wire plate electrode configuration, corona generated at a positive wire electrode may appear as a tight ionization sheath around the powered electrode or as a streamer moving outward from the powered wire electrode. Corona generated at negative electrodes may take the form of a general, rapidly moving glow or it may be concentrated into small active spots. Negative corona generally propagates by impact ionization of the gas molecules. While p ositive corona depends more on photoionization for its propagation .
31 A B Figure 2 5. Schematic of corona discharges (Chang, Lawless, & Yamamoto, 1991) . The positive corona form is also (Hermstein, 1960) and is similar to the low pressure discharge in a Geiger tube. It is characterized by a steady current at a fixed voltage, quiet operati on, and almost no sparking. The negative corona is made up of individual electron avalanches. Thus, the current of the discharge consists of many tiny spikes. However, the total current is relatively st eady . The corona discharge is usually space charge limited in magnitude, since the plasma emits ions of one polarity that accumulate in the inter electrode space apart from ionization sheath (several Debye lengths thick , , where is permittivity of free space , is th e Boltzmann constant, is the temperatures of the electrons, is the density of electrons , is the charge of an electron ) . Th us, the corona discharge has a positive resistance characteristic: increase in current need s
32 higher voltage to drive. However, when the total current keeps increase to a certain level , other charged particle s will be generated to carry current. Thus arc discharges will be formed. The arc discharge is characterized by a neg ative resistance characteristic. For flow jet actu ation purposes, different polarity will also introduce different mechanisms. In the case of a positive corona, it is assumed that electrons are produced by photo ionization in the electrode gap. These electrons are accelerated towards the point anode, resu lting in an avalanche. An ionization zone is created around the point anode, where the number of positive ions and electrons is equal. This corresponds to a plasma zone of a few tenths of a millimeter. Then ions are repulsed towards the cathode by Coulombi an forces, and constitute the drift region, where one can assume that there is no recombination because the electric field is not sufficiently large. In the case of the negative corona, Goldman and Sigmond (Goldman & Sigmond, 1982) assume that positive ions created by detachment in the ionization zone go rapidly towards the cathode. Then negative ions created by attachment drift towards the grounded plane electrode. As explained by Robinson in his famous paper (Myron Robinson, 1961) phenomenon variously known as the electric wind, corona wind and electric aura refers to the movement of gas induced by the repulsion o f ions from the vicinity of a high Hauksbee and the first explanation was given by Faraday in 1838. A history of the electric wind is given in (M. Robinson, 1962) . Many recent works are still dealing with electric wind.
33 Indeed, the electric wind is due to the momentum transfer between the ions accelerated by electric field and the neutral particles in the electrode gap region. However, altho ugh the electron velocity is much higher than the ion one, one can neglect the role of electrons because their mass is very low compared with that of ion s . Thus only small amount of momentum can be carried by electrons. 2.2.2 Previous Study on Corona Disch arge Based Plasma Actuators Corona discharge has been used for industrial applications for a few decades now (e.g., electrostatic precipitators). Thus, large amount of research work has been done for the development of applications using corona discharge . Among these applications, the plasma actuator i s the most representative one . The investigation of the corona plasma actuator can be classified into two parts: analyzing the characteristics of flow jet generated in various electrode arrangements, and verif ying its performance in various applications like cooling, active flow control and so on . Study on the characteristics in various electrode arrangements has been cases such as plate to plate (McKinney, Davidson, & Leone, 1992) , parallel plates (Asano, Hishinuma, & Yatsuzuka, 2000) , wire to cylinder (Ze bboudj & Ikene, 2000) , wire to plate (Zebboudj & Ikene, 2000) , rod to plate (Toyota, Zama, Akamine, Matsuoka, & Hidaka, 2002) , sphere to sphere (Toyota, et al., 2002) , and needle to plate (Kalman & Sher, 2001) , ha ve been studied so far. Especially, (Drews, Cademartiri, Whitesides, & Bishop, 2013) showed th at the ion cloud generated by corona discharge reduces its size when the signal frequency increases (Figure 2 6). Study on the verification of the application s of electric wind has recently been performed widely. Kalman (Kalman & Sher, 2001) , Molki (Molki & Bhamidipati, 2004) ,
34 Go (Go, et al., 2008) , and S heu (Sheu, Hsiao, & Wang, 2013) , proved the cooling performance of the corona plasma actuator for wire to plate, wire to cylinder, wire to flat plate, and point to flat plate electro de arrangements. Kasayapanand used the computational fluid dynamics technique to analyze the cooling performance of electric wind for a vertical fin array, and Chen (Chen, Guo, Yang, & Wang, 2013) verified the cooling performance of the corona plasma actuator for LED devices by measuring the thermal resistivi ty of the LEDs. Figure 2 6 . Discharge m odel schematic (Drews, et al., 2013) . A ) At DC to low frequencies , the ion cloud extends a distance L to the counter electrode where ions are consumed . B ) At high frequencies, the ion cloud extends only a distance x before begin consumed by ion ion recombination. C ) At even higher frequencies, the ion cloud is increasingly localized to within a distance r near the tip of the point electrode .
35 CHAPTER 3 DESIGN AND CHARACTER IZATION OF EFFICIENT PLASMA ACTUATORS In this section, corona discharge and volumetric geometry are chosen to design an efficient plasma actuator working under atmospheric pressure condition. Specifically , channel electrodes configuration is used to avoid viscous shear stress near the surface. Sharp needles are used as powering electrodes to enhance the volumetric corona plasma near the tip. C orona discharge based needle actuation devices with low energy loss are tested and shown to generate airflow up to while consuming only milliwatts level power. In a specific configuration, t he needle actuation device in side a channel show approxim ately 3000% and 300% increase in energy conversion efficiency as compared to conventional surface and channel dielectric barrier discharge actuators, respectively. The polarity of the needle is shown to have an effect on actuator performance, with positive polarity yielding better efficiency. 3.1 Experiment Setup An experiment was designed to develop and test the needle actuation devices. Channel electrodes configuration was used to develop an efficient corona discharge plasma actuator. Then, PIV and electr ical parameter measurement were employed to examine the performance of this type of plasma actuator. 3. 1 .1 Electrodes Configuration The electrodes configuration used to inducing a channel flow is shown in Fi gure 3 1 . Here, a steel needle powered with high voltage is placed upstream between two plate electrodes to induce a jet at the core of the channel cross section. With this configuration, direct momentum transfer into bulk flow happens near the center of the
36 channel height and should be affected by signi ficantly less wall shear in comparison to the wall jet configuration where both electrodes are at the wall. Figure 3 1. Channel electrodes configuration of n eedle actuation device. Detailed description of dimensions and experiment setup are shown in Fig ure 3 1 and described below. Powered needles and plate electrodes are separated by gap g in x direction and height h in z direction. And t he width of plate electrode denotes as w in the configuration. In all test cases, y direction (span wise) length of plate electrodes is set to be 10 cm. For channel configuration, only one needle is mounted in the configuration while multiple needles evenly separated in y direction (span wise) a re employed in cooling and vortex generation setup. The needle s are powered with high voltage DC supply (TREK Model 30/20A High Voltage Power Amplifier) and the plate electrodes are grounded. The DC signal used here is continuous with constant amplitude. A ll needle s used in this letter have a nominal radius of curvature , and all plate electrode s are 60 Âµ m thick and wide ( ). Also in all cases .
37 3.1.2 Experiment al Method The performances of the needle actuation device are tes ted by measuring its electrical parameters and velocity of the airflow induced. Then, the electrical properties and aerodynamic performance of the needle actuation devices can be analyzed. Electrical parameters were measured using an oscilloscope (Tektroni x DPO2014) and a digital multi meter (BK Precision 5491A). DC Voltage was measured using a passive probe (Tektronix Model P6015A) with an attenuation of 1000 Ã— through the oscilloscope. The oscilloscope has a maximum sampling rate of at a bandwidth of . The range of voltages used in the power and efficiency study is from to . The digital multi meter of 1 mV resolution was used to measure the potential difference across connected in series with the actuator . This poten tial difference can be converted to current with an equivalent resolution of 1 nA. V oltage and current were obtained from the measurement to calculate average electrical power using V I method, (3 1) where and are voltage and current measured, respectively . The integration is performed over a T = 10 s period. LaVision PIV system was used to collect the velocity profile due to needle actuation. L aser sheet used to illuminate Ondina oil seeding particles with diameter of 0.8 m was created by a Nd:YAG (New Wave Research Model Solo PIV II 30) laser generator fitted with a divergent cylindrical lens . Images of illuminated seeding particles were captured by a Phantom 7.3 high speed camera which has a resolution of
38 pixel s on a window size . LaVision s Davis 7.2 software was used to control both laser generator and high speed camera by generating external trigger through Model PTU 9, programmable timing unit (synchronization resolution of 10 ns with <1 ns j itter) . In this study, PIV system was employed to capture XY plane and XZ plane velocity profile for 3 D velocity visualization of the needle actuation channel and YZ plane velocity profile to show vortex generation. According to a recent study (R. Durscher & S. Roy, 2012) , statistical convergence for the plasma jet is within 300 samples (image pairs for correlation). Nonetheless, we took 1000 samples to eliminate the deviation. The uncertainty of the PIV results is less than 3.9 % without velocity gradient and is less than 13 % with maximum gradient in the experiment. Detailed uncertainty analysis is shown in Appendix. According to the velocity profile measured from the PIV system, electro mechanical energy conversion efficiency was obtained to evaluate the performance of the needle actuation device. In mechanical engineering, the induced mechanical power of the airflow inside a duct is given by ( Moreau , 2007 ) (3 2) where is the air density, is the velocity at X direction and the 2 D integration covers all the channel cross section area in YZ plane . Under its definition, electro mechanical energy conversion efficiency can be calculated as the ratio of the induced mechanical power of the airflow to the electrical power consumed in the circuit , (3 3)
39 3.2 Result and Discussion After gathering the data from the experiment, analyzing processing is demonstrated in this section. PIV results give the aerodynamic properties of the needle actuation devices. Electrical properties are obtained from the electrical parameters me asured. Then a discussion on the results is performed to further understand the performance of the device. 3.2.1 PIV Study Fig ure 3 2 shows the PIV data for the velocity field induced by the needle actuation channel powered with a positive 12 kV DC signal. This was the highest positive voltage that could be applied to the channel while maintaining the steady glow and generating a maximum velocity of 2.6 m/s. PIV results shown on the XZ plane (Fig ure 3 2 A ) and the XY plane (Fig ure 3 2 B ) demonstrate the 3 D f low structure generated by a single needle actuat ed channel flow . Fig ure 3 2 A shows strong jet flow cover ing the entire channel exit in Z direction with distinct entrainment effects from both sides of the channel. The induced airflow covers about 2 cm wide region at the channel exit in Y direction. N ote that a single needle was powered to create this velocity field. A uniform flow field can be generated by using an array of carefully positioned needles. A lso, a strong airflow penetration effect into quiesce nt surroundings can be noticed in both XY and XZ plane . These results suggest that this DC needle channel configuration has great potential in flow control and propulsion applications .
40 A B Figure 3 2. PIV velocity profiles. A ) XZ plane velocity field for needle actuation channel operated at positive 12kV DC . B ) XY plane velocity field for needle actuation channel operated at positive 12kV DC.
41 A D B E C F G Figure 3 3. XY plane velocity field s. A ) 6 kV . B ) 9 kV . C ) 12 kV and positive . D ) 6 kV . E ) 9 kV . F ) 12 kV input signal . G ) Color scheme for x component velocity .
42 Fig ure 3 3 shows the velocity field for different input signals of both polarities. Both polarities applied can induce airflow from 6 kV to 12 kV. At 6 kV and 9 kV, the maximum velocity (in the core region of the jet) induced by positive input signal is larger than that induced by negative input signal. However, Negative input signal with 12 kV create the max velocity ( ) of the entire study. Fig ure 3 4 display s the velocity line profile at downstream of the channel exit (X = 2 mm) under different operating signals. As shown in Fig ure 3 4 A, B , velocity profile generated by the negative DC signal is adhering to the plate electrodes and has a nearly flat middle region. This indicates that the momentum transfer from negative ions to neutral particles mainly happens near the wall. This is evidence for electron attachment occurr ing outside the ionization zone and around the needle tip. On the ot her hand, velocity profile of positive signal shows a sharp peak at the middle of the channel as displayed in Fig ure 3 4 A, B , which suggests the momentum transfer between positive ions and neutral particles mainly happens near the ionization zone around th e needle tip. In addition, this phenomenon also suggests that applying positive signal on the channel configuration will have less wall shear in comparison to applying negative signal. For the span wise (Y axis) velocity distribution shown in Fig ure 3 5 , b oth positive and negative signals generate similar effective region width at 9kV as the wall effects are minimal for this direction . In addition , induced mechanical power was calculated by integrati ng measured velocity profiles along the Y and Z axes.
43 A B Figure 3 4. Y axis velocity line profile s at downstream of the exit (X = 2 mm) of the needle actuation channel. A ) Positive DC signals. B ) N egative DC signal s.
44 Figure 3 5. Z axis line velocity profile at downstream of the exit (X = 2 mm) of the needle actuation cha nnel for positive and negative 12 kV DC signal 3.2.2 Power Consumption S tudy Fig ure 3 6 shows the channel electrical power consumptions and maximum velocities as a function of applied voltage for both positive and negative signals in log scale. Interestingly, both power profiles can be characterized by using linear regression with coefficient of determination . The power consumption by the negat ive DC signal increases faster ( = 9.4) than that by the positive DC signal ( = 5.6). This difference is mainly due to the different mechanism of generating working ions. For positive signal, majority positive ions are generated inside the ionization zone near the needle tip. Contrarily for negative signal, majority negative ions are generated by electron att achment outside the ionization zone. When absolute signal amplitude increases, positive ion generation becomes limited by the size of the ionization zone
45 while much more electrons can be used to generate negative ions outside the ionization zone. This mani fests as a faster growth in peak velocity with voltage for the negative DC input signal. Not e that the power consumed by the needle actuation channel is only several milliwatts , which is extremely low compare d to other plasma actuated flow s . Figure 3 6. Electrical power consumption and maximum inducing velocity of the needle actuation channel for various applied positive and negative voltages . Calculated electro mechanical energy conversion efficiencies for both positive and negative signals are displaye d in Figure 3 7 . Note t he maximum efficiencies for the needle actuation channel were 2.8% and 1.5% for positive and negative DC signals, respectively. Thes e efficiencies are one order of magnitude larger than the reported efficiency of the surface corona a ctuator (Moreau, 2007) and are about twice of the reported efficiency of wire to cylinder configuration (Colas, et al., 2010) . Interestingly, efficiencies for both polarities were attained at a magnitude of 7.5 kV DC input . For both
46 positive and negative input voltages, the efficiency increases rapidly from 6 kV to 7.5 kV and then gradually decrease s from 7.5 kV to 12 kV. This suggests that more energy is wasted at higher voltages . One can also notice that the efficiency of p ositive signal is about twice that of the efficiency of the negative signal for the same voltage magnitude. This might be explained by the loss in mechanical power output from wall shear due to different flow structure mentioned above. Figure 3 7. Energ y conversion efficiency of the needle channel for various applied positive and negative voltages . 3.3 Conclusion In summary, the performance and potential applications of the low power DC needle actuation devices with channel or plate configuration have be en demonstrated. For the channel configuration, momentum was directly injected by the needle to plate electrodes configuration into the bulk airflow of the channel minimizing the viscous penalty from induced wall shear stress (as compared to a wall jet con figuration of
47 standard DBD?) . Both positive and negative high voltage DC signals were used to generate strong ionic flow in the same direction. However, positive ionic flow with a stronger peak near the center causes less viscous penalty than negative ioni c flow adhering to the plate electrodes resulting in higher energy conversion efficiency for positive signal. Plasma jet from a single needle actuator can cover 2 cm range in span wise direction (Y axis) with a strong core flow jet. In combination of these characters, the electro mechanical energy conversion efficiency of such device can reach up to 2.8 %, almost equal to 30 times that of standard DBD channel (Riherd & Roy, 2012) or 4 times that of wire DBD channel (Campbell & Roy, 2014) . Strong entrainment effect was also observed during the study. Efficient corona discharge based plasma actuators are developed and examined. A maximum 2.8 % energy conversion efficiency is achieved . Only milliwatts level energy i s consumed to generate strong channel airflow jet with maximum velocity up to .
48 CHAPTER 4 SURFACE COOLING USING EFFICIENT PLASMA ACTUATORS The use of electric winds in the presence of a wall jet flow to modulate an external boundary layer has been an area of growing interest in the aerospace community, but has received little attention for heat transfer enhancement. Soetomo (1992) , and more recently, LÃ©ger et al. (2002) and Artana et al. (2002) , demonstrated the ability of corona discharges to reduce drag by modulating the boundary layer on a flat plate. Using electrodes perpendicular to the flow and either flush or in contact with a flat pla te, they demonstrated a near wall electric wind that accelerates the local boundary layer, promoting drag reduction. In a similar way, electric wind s can also be used to enhance convective heat transfer on a hot surface. Heat transfer enhancement by electr ic winds in the presence of external flows were reported by Velkoff and Godfrey (1979) , who used an array of corona wires aligned with the flow and extended above a flat plate. With the flat plate acting as the collectin g electrode, they demonstrated that heat transfer is enhanced in the low velocity regime but that the ionic wind is swamped by the bulk flow, and thus ineffective, as the bulk velocity increases. In this chapter, the performance of surface cooling using developed efficient plasma actuators is examined with variety of parameters. The results are reported along with the discussion. 4.1 Experiment Setup In order to evaluate the cooling performance of the developed efficient plasma actuators, an experimental system was setup to simulate the normal cooling process near a CPU chip inside an electronic device. The experimental system includes a designed efficient plasma actuator for surface cooling and the DC high voltage powering
49 system, electrical parameter measurement devices, and a semi insulated rubber heater simulates the hot CPU chip, a closeable transparent acrylic testing chamber, environmental condition monitor device, temperature measurement devi ces and the PIV analysis system. According to the function of these components, the entire system can be divided into two sections: testing section and measuring section. The testing section includes the chamber, the actuator and the monitor device. On the other hand, electrical and thermal measurement devices and the PIV system belong to the measuring section. Figure 4 1. Experimental setup inside the transparent testing chamber (Not to scale). The overview of the testing section for the experiment system is displayed in Figure 4.1. Adjacent to the semi insulated rubber heater, the efficient plasma actuator
50 for surface cooling was placed in the center of the transparent testing chamber which has the dimension of . IR camera was placed right above the actuator to measure the surface temperature. In order to measure the upstream air condition, the probe of the environmental monitor device was attached on the wall of the chamber to the upstream direct of the actuator. DC high voltage powering system and measuring section (excluding the IR camera) were located outside the chamber to avoid perturbing the flow while running the experimen t. 4.1.1 Electrode s Configuration Design For t he purpose of enhancing convection cooling effect on a surface, a wall jet need to be created near the hot surface. So the channel electrodes configuration, which create the channel jet flow, is not the best solution here. A wall jet must be generated by a body force towards the target wall. Thus, only one ground electrode attached on the target surface is needed to drive the electrical field towards to that surface. After this, this kind of electrodes configuration is called single plate configuration. The side view of the single plate configuration i s shown in Fig ure 4 2. In the configuration , one single needle or a n array of high voltage steel needle s is located in the upstream side of the grounded single plate electrode on one surface . After applying wit h either positive or negative high voltage DC signal on the needle, s uch arrangement will create a corona discharge driven by an electric field from needle tip to the grounded single plate electrode on the surface. Then, this discharge should drive the air flow from upstream of the needle towards the surface after grounded electrode . This is the first time such configuration is used to explore convective cooling effects over a surface .
51 Figure 4 2 . Side view of the electrode s configuration for surface cooling effect experiment with dimensions . Figure 4 3 . Vertical acrylic holder for needle electrodes (SolidWorks Drawing). Detailed description of dimensions and experiment setup are shown in Fig ure . 4 2 and described below. P owered needle ( s ) and grounded plate e lectrode are separated by gap g in x direction and height difference h in z direction. And the width of plate electrode denotes as w in the configuration. In all test cases, y direction (span wise) length of plate elect rodes is set to be . All plate electrode s are thick and
52 width ( ) and all gaps in the experiments of this chapter is set to be . However, the height difference between needle tip and ground electrodes, h, c , will be varied during the experiment to study their effects on the cooling performance of the designed actuators. As shown in Figure 4 3, thin s traight vertical acrylic pillar s w ere used to hold the needles during the experiment. Since the acrylic is a dielectric material and the thin design of the pillar, the influences on the electrical properties of the system and the flow field driven are cons idered negligible. On the acrylic pillar, 5 holes with the same diameter of the needle ( ) are located with different hole height from 2 mm to 10 from the center of the hole to the plate surface. By placing needle in different holes, the height difference, h, can be varied to study its effects on cooling performance. The pillar was design in Solidworks and cut by a 150 W CO 2 laser cutter (UNIVERSAL laser system PLS6.1 50D ) in the UF Infinity Fab Lab. Steel needles were used as powered electrode in the designed corona discharge actuators. These tips of needles have varies radius of curvature, c , which is the most important property that affects performance of the actuat ors. As mentioned in M s work (Moreau, Benard, Lan Sun Luk, & Chabriat, 2013) , wire electrodes with smaller radius of curvature can generate stronger air flow jet. The reasons are listed below. Electrical potential is constant everyplace on the surface of a charged conductor , ju st like the needles used in this experiment . Also, on the surface of an irregularly shaped conductor (non sphere) , the charge density is high in convex regions with small radius of curvature (especially, for example, at sharp points), and low in regions of large radius
53 of curvature, and therefore, by Gauss's Law, the field intensity must be high just above a sharp point, and relatively low just above a gently curved region. In this test, the effects of radius of curvature on cooling performance were also e xamined by using needles with different radius of curvature. The radius of curvature of a single needle was determined using a high m agnification microscope ( Omano OM136C ) with a camera (ESPA systems co. ltd D55EB digital LCD camera with maximum 10 Mega Pixel) instead of the ocular len ( s ). An example of how to measure the radius of curvature is displayed in Fig ure 4 4. Picture of the needle tip with 60X magnification was taken by the camera on the microscope. The focus plane of the picture was on the outline of the needle tip. Then a circle was fitted on the picture to match tip curve. The radius of the circle was measured according to the reference ruler as the radius of curvature of the needle tip. Three different radii of curvatures ( ) were used during the test. Figure 4 4 . Example of the estimation process on radius of curvature of a thin needle tip .
54 4.1.2 Heat Source In order to simulate a CPU chip as the heat source, a rubber heater with rated voltage 24 V and rated power 12 W was used in the experiment. The rubber heater is a resistive heater using joule heating mechanism with rubber layer coated outside the resistor array. The rubber heater has a top flat rubber surface with 2 mm in thickness. The flat s urface is used as a hot surface during the test. In order to estimate the heat flux from the fat surface, all other surfaces of the rubber heater were insulated by a novel heat insulation material, . Airloy X103 with density cl ass M is the first commercially available mechanically robust aerogels with the strength to stand up to real world applications. Although the Airloy is heavier and stronger than classic silica aerogels, it is still a thermally super insulating material wit h a heat conductivity less than and a maximum operating temperature ~80 Â°C. Figure 4 5 . Design of the cooling performance test section including needle actuator and semi insulated rubber heater (SolidWorks Drawing).
55 As displayed in Figure 4 5, rubber heater embedded in a piece of super insulating material, Airloy, was placed adjacent to the downstream of the grounded plate electrode. Such arrangement will allow wall jet airflow driven by the corona plasma actuator directly go through the hot surface of the rubber heater. 4.1.3 Electrical Supply System The needles are powered with high voltage DC signal supply system ( including a TREK Model 30/20A High Voltage Power Amplifier and a Tektronix Model AFG3022B arbitrary waveform generator ) and th e plate electrodes are grounded. The DC signal used here is continuous with constant amplitude. Electrical parameters were measured using an oscilloscope (Tektronix DPO2014) and a digital multi meter (BK Precision 5491A). DC Voltage was measured using a pa ssive probe (Tektronix Model P6015A) with an attenuation of 1000Ã— through the oscilloscope. The frequency response of the probe is DC to 75 MHz and this probe can be used to measure DC voltages up to 20 kV or peak AC voltages of 40 kV with a rated accuracy of 3%. The input impedance characteristics of the probe, as rated, are 100 M and 3 pF (when compensated properly). The oscilloscope has a maximum sampling rate of 1 GSa/s at a bandwidth of 100 MHz. The range of voltages used in the power and efficienc y study is from 6 kV to 12 kV. The digital multi meter of 1 mV resolution was used to measure the potential difference can be converted to current with an equivalent re solution of 1 nA.
56 4.1.4 Thermal Measurement Before the thermal measurements, experimental conditions, including ambient temperature and relative humidity, were monitored continuously using a humidity and temperature recorder (EXTECH Instruments RH 520). Th e probe of the recorder Two kinds of methods were used to measure the surface temperature of the rubber heater. In the first method thermocouples were attached on the surface of the rubber heater to give a brief understanding of cooling performance of the designed actuators. On the other hand, the second method employs an IR camera to obtain s patial temperature data on the simul ated CPU chip (rubber heater) adjacent to the grounded plate electrode . Figure 4 6 . Top view of the thermocouple locations on t he rubber heater for the first method of thermal measurement (not to scale).
57 For the first method , measurements of thermocouples were monitored by a LabVIEW code and were showed lively on the computer screen. As shown in Fig ure 4 6, three needles separated by a span wise distance of were placed at height with stream wise gap g . At room temperature , one rubber coated resistant heater (24V, 12W with surface area ) placed adjacent to the plate electrode was used as a heat ing source simulating an CPU chip during the test. Four T type thermo couples were placed on the corners of a inner square beginning from 1 cm downstream of the plate electrode. Time varying temperature data were obtained from the average of all four thermocouples represent ing the average surface temperature. Figure 4 7 . IR camera setup above the rubber heater for the second thermal measurement method of the surface cooling test (not to scale) .
58 To further understand the cooling performance of the d esigned actuators, s patial temperature distribution on the simulated CPU chip (rubber heater) will be obtained from IR camera. Fig ure 4 7 displays the design of the IR temperature measurement setup. IR camera (FLIR A325) is located 20 cm a bove the rubber h eater to collect the IR emission of the heater. Meanwhile, t he flat surface of the rubber was painted with Krylon Ultra Flat Black Spray 1602, which has a known emissivity of 0.96 (National Aeronautics and Space Administration, 2003) , to ensure accurate IR readings. Then, t he accuracy of the camera was assessed to be Â±1 Â° C over the range of 25 80 Â° C during calibration process using thermocouples . Th us , s tabilized surface temperature distribution can be obtained from this measurement to study the cooling performance at different locations. With the surface temperature data from IR camera, the distribution of convection coefficient on the surface and the re lationship between convection coefficient and velocity at the same location also can be investigated . 4.1. 5 Velocity Measurement The same PIV system used in Chapter three was also employed in the cooling performance study. PIV system was employed to examin e the flow field in the cooling region over the rubber heater . By obtaining the flow velocity distribution near the hot surface, detailed study on the forced convection was performed . XY plane and YZ plane velocity field was captured because it can give in formation on the cooling area that covered by the actuator as well as whether the 2 cm needle space can sustain uniform flow in the area between two adjacent needles. 4.2 Result and Discussion In the cooling performance test, several important parameters were chosen to study their influence. The location of the needle was adjusted to get the optimal cooling
59 performance. High voltage DC signals with different voltages and both polarity were applied to the needle. Various radius of curvature of the steel needle tip was also investigated on its influence on both flow velocity field and cooling performance. U n like other plasma based f low inducement , the needle actuated channels can generate sufficient amount of flow with very low power consumption. This suggests that much less power is wasted in heat generation. So, this makes it a perfect ly suitable as a convective cooling device. We demonstrate this cooling capability using a p late needle configuration with three needles . A rubber heater was powered by a 8 V DC signal placed adjacent to the plate electrode with a surface heat flux of . The upstream ambient air temperature was maintained as with relative humidity of 64 % to 66 % during the and the average surface temperature under free convection was . Then the free convection coefficient can be calculated as by using equation : ( 4 1 ) Time varying temperature curve for cooling effects are shown in Fig ure 4 8 . The convection coefficients under needle actuation were with the applied voltage of 6, 9, 12 kV . At positive 12 kV, needle actuation can cool a area from to within . In the meantime, the total power of three needles for each voltage applied shown in the legend of Fig ure 4 8 remains at miliwatts level which demonstrate great potential of needle actuation device for cooling application.
60 Figure 4 8 . Cooling effect of needle actuation device setup (total power consumed by all three needles shown in legend). In order to further understand the cooling performance with surface temperature images captured by IR camera, local temperature drop and local enhancement factor of heat con vection co efficient b etween natural convection and forced convection with needle actuator were both used. Since the input power of rubber heater was maintained during the experiment, the relation between natural convection confident and forced convection confident can be write as (4 2) Thus, one can further calculated the enhancement factor of the local heat convection coefficient from Equation 4 2,
61 (4 3) Similarly, average heat convection enhancement factor can be calculated to represent the overall cooling performance of the actuator for each case. (4 4 ) 4.2.1 Voltage and P olarity S tudy In this section, effects of voltage and polarity were investigated on the cooling performance of the single needle actuators . The needle was placed in the upstream of the ground electrode with gap, g = 15 mm, and height, h = 3 mm. Inside the testing chamber, the ambient temperature w as maintained as with relative humidity from 64 % to 66 %. With both positive and negative polarity of the input signal, the voltages were set as 6 kV, 8kV, 10 kV and 12 kV. PIV study on the XY plane and XZ plane as well as the IR surface temperat ure study were performed. XZ plane PIV images on varying positive voltage magnitudes with stream wise direction (positive X direction) velocity profiles are shown in Figure 4 9. XZ plane PIV profiles were recorded in the center plane (Y = 0 mm) of the need le actuator. These images suggest that the needle actuator can generate higher flow velocity with higher magnitude of voltage. However, with voltage higher than 8 kV, velocity profiles show hump shape before the plate ground electrode, which starts at X = 1 5 mm and ends at X = 2 0 mm. This hump shape velocity profile lifts the wall jet driven from the needle tip and then hits the surface at the beginning of the rubber heater.
62 A B C D Figure 4 9. XZ plane X direction velocity profile (Y = 10 mm) of a single needle actuator. A) Positive 6 kV. B) Positive 8 kV. C ) Positive 10 kV. D ) Positive 12 kV.
63 A B C Figure 4 10 . XZ plane Z direction velocity profile (Y = 10 mm) of a single needle actuator. A) Positive 6 kV. B) Positive 8 kV. C ) Positive 10 kV.
64 A B Figure 4 1 1 . X direction Velocity profile with negative DC signal powered on needle. A) At the edge of the rubber heater (X = 2 0 mm). B) At middle of the rubber heater (X = 4 0 mm) . To further analyze the hump, Z direction velocity profiles were also shown in Figure 4 10. The red region represents upward flow and blue region represents downward flow. There were clear signs that beginning of the grounded electrode (X = 5 mm) is the sepa ration interface of opposite direction flow. One possible reason for this phenomenon is that after the corona discharge ignites, ions are driven by the electric field between the needle tip and the grounded electrode. During this process, some ions will be accumulated on the surface the dielectric material before reach the grounded electrodes. Then, these accumulated ions will repulse the ions came from the needle to
65 form a hump. Meanwhile, the grounded electrode maintains at zero potential attracting ions towards its surface. To further analyze the effect of the voltage magnitude and polarity of the high voltage DC signal, Fig ure 4 1 1 and Fig ure 4 1 2 show the X direction velocity profiles of the surface cooling system at X = 0 mm and X = 20 mm. In each individual figure, velocity profiles are driven by 4 different voltages from 6 kV to 12 kV. These velocity profiles are typical Glauert wall jet profiles. A B Figure 4 1 2 . X dire ction Velocity profile with Positive DC signal powered on needle. A) At the edge of the rubber heater (X = 2 0 mm). B) At middle of the rubber heater (X = 4 0 mm) .
66 Due to the existence of the hump in velocity profile, at the beginning of the rubber heater (X = 2 0 mm), the wall jet profile is barely attach to the surface of the heater with about 1 mm zero velocity space above the heater (Figure 4 11A, Figure 4 12A). However, after developing through the heater surface, the wall jet profiles finally attached wi th the heater surface at the middle of the heater (X = 4 0 mm) as displayed in Figure 4 11B and Figure 4 12B. Furthermore, in both positive and negative cases , needle actuator show s increasing velocity w ith increasing voltage magnitude . Similar to the chann el needle actuator described in c hapter 3, under negative signal, the velocity profile increase faster than that under positive signal. This may be due to the different discharge mechanisms of positive or negative corona discharge. For negative signal, whi le voltage magnitude increases, the number of electrons emitted from the discharge region near the needle tip increase much faster than the increment of the positive ions under positive signal. This fast growth in electron emission will lead to the growth of negative ions through electron attachment process. Comparison of the Glauert wall jet profiles between positive and negative 10 kV cases are displayed in Fig ure 4 1 3 . At both locations, the profiles from two cases are almost identical. Unlike the channel needle actuator described in chapter 3, in the plate needle actuator, electric field is generated between the needle and the flat grounded electrode. Subsequentially, the ions generated under either positive signal or negative signal wi ll be driven towards the surface of the grounded electrode. So, no difference was shown in this comparison. Thus, no polarity analyze is need in the following studies.
67 A B Figure 4 1 3 . Comparison of the X direction velocity profile between positive and negative DC signal powered on the needle. A) At the edge of the rubber heater (X = 2 0 mm). B) At middle of the rubber heater (X = 4 0 mm) . While the velocity is increasing with voltage ma gnitude increment , the cooling performance should be better with higher magnitude of voltage. To examine this, XY plane PIV images were also captured to analyze the velocity distribution on the heater surface. The PIV plane was set 1 to 2 mm above the heat er surface. And the images were only covers the rubber heater, which is the region of interest on cooling performance test.
68 A B C D Figure 4 14 . X Y plane X direction velocity profile (Z = 2 mm) of a single needle actuator. A) Positive 6 kV. B) Positive 8 kV. C ) Positive 10 kV. D ) Positive 12 kV.
69 A B Figure 4 15 . X direction Velocity profile with Positive DC signal powered on needle. A) At the edge of the rubber heater (X = 2 0 mm). B) At middle of the rubber heater (X = 4 0 mm) .
70 A B C D Figure 4 16 . Distribution of the enhancement factor of the local heat convection coefficient over the rubber heater with different voltage magnitude . A) Positive 6 kV. B) Positive 8 kV. C ) Positive 10 kV. D ) Positiv e 12 kV.
71 The average heat convection enhancement factor s calculated using Eq. (4.4) for different voltage s and polarit ies are displayed in Figure 4 17. With the voltage magnitude larger than 10 kV , actuator with negative signal showed slight better cooling performance . However the power consumption of the negative corona actuator is about t wice of that for positive corona actuator . In both polarity cases, the maximum speed reaches about 1.2 m/s between X = 5 mm and X = 1 0 mm, which indicates that the flow i s keep accelerating after the grounded flat electrode. Figure 4 1 7 . Average ehancement factor as function of applied voltage and polarity.
72 Figure 4 1 8 . Comparison of the X direction velocity profile between different height of the needle. 4.2.2 Needle H eight S tudy Height of a single needle in the actuator was varied from 3 mm to 1 1 mm above the heated surface. Velocity profiles at the edge of the heater for selected needle height are shown in Figure 4 18. All case were operated under positi ve 10 kV DC input signal. The results suggest that highest velocity is generated with needle height equals to 7 mm.
73 A B C D Figure 4 19 . Distribution of the enhancement factor of the local heat convection coefficient over the rubber heater at 6kV with different needle height . A) h = 3mm . B) h = 5mm . C ) h = 7mm . D ) h = 9mm .
74 A B C D E Figure 4 20 . Distribution of the enhancement factor of the local heat convection coefficient over the rubber heater at 8kV with different needle height . A) h = 3mm . B) h = 5m m . C ) h = 7mm . D ) h = 9mm . E) h = 11 mm.
75 A B C D E Figure 4 2 1 . Distribution of the enhancement factor of the local heat convection coefficient over the rubber heater at 10kV with different needle height . A) h = 3mm . B) h = 5mm . C ) h = 7mm . D ) h = 9mm . E) h = 11 mm.
76 Distributions of local heat convection enhancement factor are shown in Fig ure 4 1 9 to Figure 4 21 . Result s suggest that when needle is placed at 7 mm, the cooling effects of the needle actuator reaches maximum for voltage from 6 kV to 10 kV. In the cases that needle placed in higher or lower locations, maximum flow velocity was limited by the geometry. As shown in the enha ncement fa ctor distribution figures (Figure 4 19 to 4 21), t he hump effect also affected the cooling performance of the actuator . With the lower needle height, the hump effect starts with lower applied voltage. The possible reason is that when the needle i s closer to the dielectric surface, it is easier to accumulate the charge on the surface. Figure 4 22 . Average ehancement factor as function of needle height and applied volatge magnitude .
77 Figure 4 22 shows the entire average heat convection enhancement factor s in the needle height study. It clearly shows that for the cases studied 7 mm is the best needle height for cooling us ing needle plate actuators. A B C Figure 4 2 3 . Radius of curvature estimation with microscopic pictures. A) ~25 m needle tip. B) ~45 m needle tip. C) ~60 m needle tip. 4.2.3 Radius of C urvature S tudy Radius of curvature of needle tip was tested with three different size of the As shown in Fig ure 4 2 3 , radius of curvature of needle tips
78 were estimated using microscopic pictures and the reference length. In all three cases, the applied DC signal was maintained at 10 kV. Figure 4 2 4 . Comparison of the X direction velocity profile between different radius of curvature of the needle. As displayed in Fig ure 4 2 5 , temperature drop r esult indicates smaller radius of curvature leads to stronger cooling effect as well as the power consum ption. Similar trend was observed by Leonov s group on electrode with small sharp tip (Leonov, Opaits, Miles, & Soloviev, 2010) . Th is is due to the fact that high potential conductor with a smaller radius of curvature will have stronger E field. With stronger E field, more ions will be generated. Thus higher electric body force will be provide to the bulk ions, resulting in a faster wall jet.
79 Figure 4 2 5 . Average ehancement factor as function of radius of curvature of the needle tip. 4. 3 Summary of Surface Cooling Study The cooling effect using multi needle plate configuration was investigated. The preliminary results on cooling effects shows that needle actuation device can reduces the temperature of a hot area by using only miliwat ts level energy with in 80 s. This suggest s that the needle actuation can efficiently enhance heat transfer near hot surface. Voltage magnitude and polarity study suggests that the needle actuator can generates higher flow velocity with higher magnitude of voltage. Subsequently, the cooling performance is better with higher magnitude of voltage. With the same voltage magnitude, actuator with negative signal showed slight better cooling performance as well as twice of the power consumption.
80 Height of a sing le needle in the actuator was varied from 2 mm to 12 mm above the heated surface. Result suggests when needle is placed at 6 mm, the cooling effects of the actuator reaches maximum for voltage from 6 kV to 10 kV. In the cases that needle placed in higher o r lower locations, flow velocity was limited by the geometry. Radius of curvature of needle tip was tested with 3 different size of the needle cooling effect as well as the power consumption.
8 1 CHAPTER 5 VORTICES GENERATION USING EFFICIENT PLASMA ACTUATORS As a method of introducing perturbations for low speed flow control, plasma actuators present themselves as a useful tool (Corke, Enloe, & Wilkinson, 2010) . This type of plasma actuator has shown itself to be effective for boundary layer control (J. R. Roth, Sherman, & Wilkinson, 2000; Visbal, 2010) and for flow reattachment at low to moder ate Reynolds numbers (Rizzetta & Visbal, 2010; Schatzman & Thomas, 2010) , especially when used in a pulsed manner. Plasma actuators with novel geometries, including the serpentine geometry with DBD actuators , have been effective in a number of n umerical studies (Rizzetta & Visbal, 2011; Wang, Durscher, & Roy, 2011) as well as some experimental work (R. J. Durscher & S. Roy, 2012; Santhanakrishnan & Jacob, 2007) . Specifically, the serpentine geometry actuators are able to introduce streamwise, as well as spanwise and surface normal vorticity to and promote increased mixing in the flow. The manner that three dimensional perturbations can be introduced to the flow allows for an additional freedom when designing and implementing control. This additional degr ee of freedom may allow for greater control authority. Due to the benefit of generating vorticities on active flow control, efficient corona based plasma actuator is further developed to generate vorticies in this section. The primary goal of this study is to generate controllable vo r ticies using mil l iwatt level power. 5.1 Experiment Setup 5.1.1 Electrode Configuration Design The configuration for the plate configuration is shown in Fig ure 5 1. In this, an array of high voltage steel needles is placed in th e upstream side of the plate electrode.
82 Such arrangement should allow airflow driven by the corona discharge from the needle tip to the plate with either positive or negative DC signals. S ince this configuration will generate 3D flow structures, a particul ar arrangement of multiple needles is then employed to generate three dimensional vortical structures. A B Figure 5 1. Electrodes configuration for surface cooling effect. A ) Side view . B ) Top view . Detailed description of dimensions and experiment setup are shown in Fig ure 5 1 and described below. Powered needles and plate electrodes are separated by gap g in x direction and height h in z direction. And the width of plate electrode denotes as w in the configuration. In all test cases, y direction (s pan wise) length of plate electrodes is
83 set to be . All needle s used in this letter have a nominal radius of curvature , and all plate electrode s are 60 Âµ m thick and wide ( ). Also in all cases h = 2.5 mm. 5.1.2 Testing Environme nt The needles are powered with high voltage DC supply (TREK Model 30/20A High Voltage Power Amplifier) and the plate electrodes are grounded. The DC signal used here is continuous with constant amplitude. As previously mentioned, we used m ultiple configurations of needle plate actuation to investigate the ir performances for various applications . Controllable vortical structure is created using plate configuration (Fig ure 5 1) with two needles. In vortex generation case, two needles at span wise dis tance of 3 cm were placed at the same height with gap . In order to control the direction of the vortex, the applied DC signals of these two needles are kept slightly different (described in section 5.2 ). 5.2 Result and Discussion Needle actuation device may also be used to induce flow vortices near the surface. Two cases were studied b y applying slight ly different signals on two span0wise displaced needle electrodes . We managed to generate a counter rotating vortex pair between the two n eedles for which the strength and direction of the majority vortex can be controlled . As displayed in the inlay in Figure 5 2A , needle 1 powered with 10 kV DC signal and needle 2 powered with 8 kV DC signal were separated by a distance . After powe ring both needles, a counter clockwise vortex with maximum X vorticity greater than 300 s 1 was captured using PIV system at YZ plane with X = 10 mm . Note that the airflow generated by needle 1 is stronger near the wall so that it
84 pushes the airflow generated by needle 2 and then forms a counter clockwise vortex near the wall. F or case 2 shown in Figure 5 2B , the needle 2 at Y = 15 mm was powered with 10kV DC signal which is higher than the 8 kV DC signal applied on needle 1 at Y = 15 mm. Stronger airflow near surface was generated near the needle 2 side. Then, the major vortex generated was clockwise under this condition . Furthermore , the power used to generate th e vortex pair was only 12 mW in total . Such low power vortical flow p erturbation may be quite beneficial for drag reduction, flow mixing and many other flow control applications (Riherd & Roy, 2013) . 5. 3 Application on Surface Cooling As demonstrated in the previous section, v ort ices can be generate d between two needles with different voltage . These vortices will increase the mixture of the cold air on top and hot air on the heater. Thus, it may also be used in enhancing surface cooling applications. In this section, two parameters were investigated: number of needles used in the actuation device and voltage difference between 2 needl es. 5.3.1 Experiment S etup Fig ure 5 3 shows the needle configuration of the needle number study. With single needle, the needle was placed in the center position. In the Double needle case, two needles were separated by 20 mm. Triple needle case were the c ombination of the single needle case and double needle case. The voltage difference study was performed on double needle configuration as shown in Fig ure 5 3B. Three cases were tested with voltage pairs (8 kV, 8 kV), (7 kV, 9 kV) and (6 kV, 10 kV).
85 A B Figure 5 2 . Controllable vortices generated by two needle actuator at X = 10 mm . A ) Anti clockwise vortices . B ) C lockwise vortices.
86 A B C Figure 5 3. Needle configurations for number of needle test. A) Single needle case. B) Double needle case (separated by 20 mm) . C) Tripe needle case (separated by 10 mm between adjacent needles) . 5.3.2 Results and D iscussion Even with the same voltage magnitude, the two needle actuator still generates vortices between the needles. The strength of the vort ices is slight weaker than that generated by the different voltage actuator.
87 A B C Figure 5 4 . X Y plane X direction velocity profile (Z = 2 mm) of needle actuator s with different number of needles . A) Single needle . B) Double needle . C ) Triple needle. Figure 5 4 and Figure 5 5 display the X direction velocity profile with in an XY plane 2 mm above the heater surface. In all three cases, needle actuators were operate
88 with positive 12 kV DC signal. The maximum velocity increases when number of needles increases. For double needle actuator, the area between two needles show s almost zero velocity, which indicates that the vortices was formed. However, without downstream velocity, air cir culating in this region was heated by the heater below. Thus, no significant is observed in the area between the needles. On the other hand, triple needle actuators shows nearly uniform velocity distribution over the heater area. Hump effect can be observed on the velocity profiles of single nee dle and double needle actuators. Meanwhile, triple needle actuator shows strong air flow without hump effect. As shown i n Fig ure 5 6 , cooling effects are represented by local convection enhancement factor distribution for number of n eedles study. Similar as the velocity profiles, w ith increasing number of needles, the cooling effect of the needle actuator becomes stronger. H ump effect shows negative influence on the cooling performance of single needle and double needle actuators. Mea nwhile, strong uniform cooling effect was observed in triple needle case. One possible reason for this phenomenon is that with needle closer to each other, bo dy forces generated by the needle array cannot propagate to the side direction, which allows more ambient cool air to go through the hot surface. Average heat convection enhancement factors are also plotted in Figure 5 7. With increasing number of needles on the actuator, better cooling performance was observed.
89 A B Figure 5 5 . X direction Velocity profile with Positive DC signal powered on different number of needle s . A) At the edge of the rubber heater (X = 2 0 mm). B) At middle of the rubber heater (X = 4 0 mm) .
90 A B C Figure 5 6 . Distribution of the enhancement factor of the local heat convection coefficient over the rubber heater at 10kV with different number of needles . A) Single needle . B) Double needle . C ) Triple needle.
91 Figure 5 7. Average heat convection enhancement factor as a function of number of needles. 5.4 Summary of Vortic es Generation Study The vortex generation using multi needle plate configuration was investigated. Controllable vort ices generate d between 2 needles with different voltage were demonstrated in th is preliminary study . Based on these studies, this kind of lo w power actuation device s show great potential to be developed for many prospective applications. Vortices with controllable rotation direction were generated by a two needle actuator with different high voltage DC signal applied on two needles. Even with the same voltage magnitude , the two needle actuator still generates vortices in between needles. T he strength of the vortices is slight weaker than that generated by the different voltage actuator.
92 With increasing number of needles, the cooling effect of the needle actuator becomes stronger. Hump effect gives negative influence on the cooling performan ce of single needle and double needle actuators. And, the vortices generated by the double needle actuator circulate the hot air on the surface area between needles, which also reduces the cooling performance. Meanwhile, triple needle actuator generates a strong uniform flow, which achieves the best cooling performance in this work.
93 CHAPTER 6 SUMMAR Y AND FUTURE WORK As mentioned in Chapter 2, plasma actuators have been applied to various aerodynamic and thermostatic applications and show great potential in these fields. In order to meet the needs of the real world applications, however, a huge increase on e lectro me chanical energy conversion efficiency as well as a reduction in joule heating generation is necessary. T o accomplish this, the development of plasma actuator needs a significant improvement in its performance . The objective of this work was to explore and characterize novel plasma actuators in an effort to improve the performance. Results of this work including development and characterization of the novel efficient plasma actuator, cooling performance and controllable vortex generation using the newly deve loped actuators were presented in the previous chapters. The following sections highlight and summarize the primary findings of these investigations, as well as provide recommendations for future work in these areas. 6.1 Development on Efficient Plasma Act uators As shown in Chapter 3, development and characterization on novel efficient plasma actuators indicates significant improvement in energy conversion efficiency as well as reduction in heat generation. Newly designed actuator uses corona discharge with volumetric geometry. Air flow jet can be induced by the discharge and the electric field generated between the needle and grounded flat electrode. Highlights of this topic are list below. 1. Air flow jet driven by a single needle actuator can cover 2 cm rang e in span wise direction (Y axis) with a strong core flow jet . Its e lectro mechanical energy conversion efficiency can reach up to 2.8 %, almost equal to 30 times that of standard DBD channel or 4 times that of wire DBD channel. Strong entrainment effect was also observed during the study.
94 2. For the channel configuration , momentum was directly injected by the needle to plate electrodes configuration into the bulk airflow of the channel minimizing the viscous penalty from wall shear stress. Both positive and negative high voltage DC signals were used to generate strong ionic flow in the same direction. 3. P ositive ionic flow with a s tronger peak near the center causes less visc ous penalty than negative ionic flow adhering to the plate electrode s resulting in higher energy conversion efficiency for positive signal . These highlights show that this novel plasma actuator accomplish the original goal of the work. Furthermore the sma ll size and easy implementation also indicate that this kind of actuator has great potential for wide range of applications in thermal and fluid areas. However, energy conversion efficiency of 2.8 % is not huge enough number to fully satisfy the real worl d , further research on improving the plasma actuator efficiency is advised. 6. 2 Surface Cooling Performance Study As shown in Chapter 4, newly developed plasma actuator was used for surface cooling over a semi insulated rubber heater simulating the heat g eneration CPU chip. Glauert wall jet generated by the needle actuator enhances the heat convection process near the hot surface. Various parameters such as the location of needle, radius of curvature of the needle tip and voltage and polarity of the high v oltage DC signal were studied. Both thermocouples and IR camera were employed to measure the surface temperature. PIV system was also used to capture the flow field over the hot surface. Flow field and surface temperature distribution were combined togethe r to analyze the cooling performance of the actuator. Highlights of this topic are listed below. 4. The study on cooling effects shows that needle actuation device can reduce the temperature of a hot surface area by using only mW energy with in 80 s. 5. Vo ltage magnitude and polarity study suggests that the needle actuator can generates higher flow velocity with higher magnitude of voltage. Subsequent ly, the cooling performance is better with higher magnitude of voltage. With the
95 same voltage magnitude , act uator with negative signal showed slight better cooling performance as well as twice of the power consumption. 6. Height of a single needle in the actuator was varied from 2 mm to 12 mm above the heated surface. R esult suggests when needle is placed at 6 mm, the cooling effects of the actuator reaches maximum for voltage from 6 kV to 10 kV. In the cases that needle placed in higher or lower locations, flow velocity was limited by the geometry. 7. Radius of curvature of needle tip was tested with 3 different size of the needle from 25 m to 60 m. Result indicates smaller radius of curvature leads to stronger cooling effect as well as the power consumption. These highlights give us a brief understan ding of how the needle actuator s perform when applying them on surface cooling. With a ppropriate location and smaller radius of curvature of the needle electrode, the actuator can reach optimal condition for surface cooling. 6. 3 Vortices Generation and App lications As shown in Chapter 5, c ontrollable vort ices generate d between 2 needles with different voltage were also demonstrated in the study . Due to the asymmetric assigned high voltage DC signal, body forces generated by two needles stir the flow in bet ween and then create controllable vortices along with the streamline. With the controllable vortices, this technology can be applied to many aerodynamic problems. Furthermore, these vortices will increase the mixture of the flow. Thus, it may also be used in surface cooling applications. In this section, two parameters were investigated: number of needles used in the actuation device and voltage difference between 2 needles. Highlights with this topic are listed below. 8. Vortices with controllable rotation di rection were generated by a two needle actuator with different high voltage DC signal applied on two needles. 9. Even with the same voltage magnitude , the two needle actuator still generates vortices in between needles. The strength of the vortices is slight weaker than that generated by the different voltage actuator.
96 10. With increasing number of needles, the cooling effect of the needle actuator becomes stronger. 11. The vertices generated by two needle actuator don t move downstream as fast as the air jet. Thus, the heat convection enhancement is weaker between the needles in this case. However, three needle actuator shows uniform heat convection enhancement over the hot surface. 6.4 Future Work In this study, efficient needle corona plasma actuators have been designed and tested. Such actuators ha ve great potential for many applications such as active flow control, gas mixture, enhanced surface cooling and so on. Potential future work after this study includes: 1. Parameter of plasma can be measured to further und erstand the status of the plasma. Thus, this information can be used to optimize the performance of actuators. 2. Applying this kind of actuator on applications such as flow control or gas mixture. 3. A compact high voltage low power DC supply can be designed to develop useful products with the actuators.
97 APPENDIX PIV UNCERTAINTY ANALYSIS In recent years, PIV is a noninvasive quantitative flow visualization tool that can give researchers spatiotemporal velocity field associated with complex flows. Through the use of a high speed camera, micro sized flow seeding can be captured in a series of images. Generally, these images are captured in pairs. Since the time ga p between each pair is set by the researcher, only the displacement of the flow seeding is calculated from the image pairs. To find the displacement of the seeding, standard cross correlation or auto correlation method is employed. According the method of PIV described above, the main error sources of PIV comes from the seeding in the flow and the algorism in PIV software. The main concern with flow seeding is whether the seeding particle can accurately capture the fluid dynamics without altering the fluid properties. T he non dimensional Stokes number can be used to estimat e the accuracy in which a seeding particle follows the flow. The Stokes number is defined as (A 1) where part is the relaxation time of the particle in the flow , u ref is the characteristic reference velocity (typically the freestream velocity), and L ref is the characteristic reference length. A Stokes number 1 indicates the particle will not fellow the flow accurately or even change the flow. Mean wh ile a Stokes number approaching zero (St 1) implies a perfect seeding for tracking the flow . For a constant accelerating Stokes flow , the seeding particle relaxation time is defined as
98 (A 2) w here part is the particle density, d part and fl uid is the the dynamic viscosity of the fluid. G eneral ly , the dynamics of a suspended particle in a flow are much more complicated (see Adrian, (1997) ) . But , Equation A 2 provides a reasonable estimation of the particles . For the experiment setup used in this work, vaporized Ondina oil (Shell 917) was used as the sending particles to fill the test chamber. The oil was vaporized us ing a TSI atomizer (Model 9302) . W hen the atomizer is pressurized at 25 psi, it can produce a droplet with a mean diameter of 0.8 Âµm ( d part 0. 8 Âµm) (TSI, 2000). Since air w as th e working fluid , the was substituted into Equation A 2 (along with part = 854 kg / m 3 , and fl uid = 1.846E 5 kg/m s ) gives a relaxation time of 1.6 4 E 6 s. Substituting the characteristic velocity of 3 m / s and a referenc e length of 1 mm into Equation A 2 gives Stokes number of 4.9 E 3 . This Stokes number can be treated as St 1 . Thus, the error from seeding in the flow can be neglected under such a small Stokes number. The other major error source come s from the algorism in the software used. In this experiment, DaVis 7.2 software was used to capture the image pairs from high speed camera and to post process the acquired data. To resolve the error from the PIV algorism, a n uncertainty analysis is propose d by (Timmins, Wilson, Smith, & Vlachos, 2012) and (Wilson & Smith, 2013b) . They estimate d the PIV algorism uncertaint ies by examining the distribution of errors from many measurements tha t have identical inputs. In keeping with ASME recommendations , 95 % confidence intervals were used on all uncertainties (ASME 2009).
99 For simplicity and clarity, a 1D equation for calculating the velocity in PIV measurements is used to begin the analysis. T he velocity is calculated from the approximation that (A 3) Thus, the uncertainty of u will come from both the displacement and the time gap t . A ssuming that the uncertainty of and is not correlated , the Taylor series method can be used for uncertainty propagation . T he n, the total uncertainty for the velocity can be estimated as (A 4) where U and U are the uncertainty estimates of and . Since the time gap t , which is in the range of several hundred micro second, is controlled by hardware with nano second accuracy (2.6 GHz CPU). The uncertainty of can be assumed negligible . Thus the uncertainty of u becomes proportional to the computed displacement uncerta inty, U . Equation A 4 simplifies to (A 5) Thus, for all the calculation below, velocity can be replaced by displacement for simplicity . There are several major parameters that influence the accuracy of PIV, including particle image densi ty, particle image diameter, particle displacement, and velocity gradient (also can be called shear). Thus, for every experimental set up, the uncertainty of need s to be estimated with the specific parameters mentioned above.
100 Figure A 1. Standard deviation as a function of the number of samples used . In order to accurately estimate the uncertainty of the PIV setup used in this work , a pressurized air flow was measured by both the PIV setup and a n OMEGA HH USD hot wire anemometer . The uncertainty of this anemometer is 5 %. The time varying velocity obtained from hot wire anemometer is used to estimate the instantaneous bias, b ui , of the mean velocity measured from PIV. Then the uncertainty on the mean velocity measured by PIV is (Wilson & Smith, 2013a) (A 6) where is the precision uncertainty on the mean value calculated from sample standard deviation, N is total number of measurement obtained in each cases. Sample standard deviation was calculated from the measured data for the specific parameters
101 mentioned above. The number of data t o be used was determined after a convergence study was performed. Figure A 1 displays the results. It suggests that 10000 samples for each case were sufficient for statistical convergenc e on standard deviation , s u . Then the precision uncertainty on the mean value can be expressed as (A 7) Table A 1 shows the calculated uncertainty with the same parameters used in this experimental work. For different velocity, time gap, , was changed to maintained to render the same particle displacement. And 0.1 pixels/pixel is the maximum gradient found in the results. Thus, Table A 1 covers all the possible regions of the uncertainty surface defined by particle image density, particle image diameter, particle displacement, and velocity gradient . Total uncertainty shown in percentage including the hot wi re anemometer uncertainty and PIV uncertainty. Table A 1. Mean PIV velocity uncertainty for specific parameters used in this work . Particle Diameter (pixels) Particle Density (part/pixel 2 ) Displac e ment (pixels) Gradient (pixels/pixel) Time Gap ( s ) Total Uncertainty 2.1 0.048 3.2 0 .0 250 6 .1 % 2.1 0.048 2.5 0.1 250 13.5 % 2.1 0.052 3.0 0 .0 80 5.5 % 2.1 0.052 2.4 0.1 80 12. 7 %
102 LIST OF REFERENCES Adrian, R. J. (1997). Dynamic ranges of velocity and spatial resolution of particle image velocimetry. [Article]. Measurement Science & Technology, 8 (12), 1393 1398. Artana, G., D'Adamo, J., Leger, L., Moreau, E., & Touchard, G. (2002). Flow control with electrohydrodynamic actuators. [Article]. Aiaa Journal, 40 (9), 1773 1779. Asano, K., Hishinuma, R., & Yatsuzuka, K. (2000, 2000). Bipolar corona discharge from a floating filamentary particle between parallel plate electrodes. Paper pres ented at the Industry Applications Conference, 2000. Conference Record of the 2000 IEEE. Braithwaite, N. S. J. (2000). Introduction to gas discharges. [Article]. Plasma Sources Science & Technology, 9 (4), 517 527. Campbell, N. S., & Roy, S. (2014). Plasma channel flows: Electro fluid dynamic jets. [Article]. Applied Physics Letters, 105 (13), 4. Chang, J. S., Lawless, P. A., & Yamamoto, T. (1991). CORONA DISCHARGE PROCESSES. [Article]. Ieee Transactions on Plasma Science, 19 (6), 1152 1166. Chen, I. Y., Guo, M. Z., Yang, K. S., & Wang, C. C. (2013). Enhanced cooling for LED lighting using ionic wind. [Article]. International Journal of Heat and Mass Transfer, 57 (1), 285 291. Colas, D. F., Ferret, A., Pai, D. Z., Lacoste, D. A., & Laux, C. O. (2010). Ionic wind generation by a wire cylinder plate corona discharge in air at atmospheric pressure. [Article]. Journal of Applied Physics, 108 (10), 6. Conrads, H., & Schmidt, M. (2000). Plasma generation and plasma sources. [Article]. Pla sma Sources Science & Technology, 9 (4), 441 454. Corke, T. C., Enloe, C. L., & Wilkinson, S. P. (2010). Dielectric Barrier Discharge Plasma Actuators for Flow Control Annual Review of Fluid Mechanics (Vol. 42, pp. 505 529). Palo Alto: Annual Reviews. Dre ws, A. M., Cademartiri, L., Whitesides, G. M., & Bishop, K. J. M. (2013). Electric winds driven by time oscillating corona discharges. [Article]. Journal of Applied Physics, 114 (14), 7. Durscher, R., & Roy, S. (2012). Evaluation of thrust measurement tech niques for dielectric barrier discharge actuators. [Article]. Experiments in Fluids, 53 (4), 1165 1176.
103 Durscher, R., Stanfield, S., & Roy, S. (2012). Characterization and manipulation of the "saturation" effect by changing the surface temperature of a die lectric barrier discharge actuator. [Article]. Applied Physics Letters, 101 (25), 4. Durscher, R. J., & Roy, S. (2012). Three dimensional flow measurements induced from serpentine plasma actuators in quiescent air. [Article]. Journal of Physics D Applied P hysics, 45 (3), 9. Fridman, A., Chirokov, A., & Gutsol, A. (2005). Non thermal atmospheric pressure discharges. [Review]. Journal of Physics D Applied Physics, 38 (2), R1 R24. Fridman, A. A. (2008). Plasma chemistry . Cambridge: New York. Go, D. B., Matura na, R. A., Fisher, T. S., & Garimella, S. V. (2008). Enhancement of external forced convection by ionic wind. [Article]. International Journal of Heat and Mass Transfer, 51 (25 26), 6047 6053. Goldman, M., & Sigmond, R. S. (1982). CORONA AND INSULATION. [A rticle]. Ieee Transactions on Electrical Insulation, 17 (2), 90 105. Griem, H. R. (1962). HIGH DENSITY CORRECTIONS IN PLASMA SPECTROSCOPY. [Article]. Physical Review, 128 (3), 997 &. Hermstein, W. (1960). Die Stromfaden Entladung und ihr Ãœbergang in das Glimmen. Archiv fÃ¼r Elektrotechnik, 45 (3), 209 224. Howatson, A. M. (1976). An Introduction to gas discharges (2d ed.). Oxford: New York. Jukes, T. N. (2007). Turbulent drag reduction using surface plasma. Unpublished Ph.D.; University of Nottingham. Kalman, H., & Sher, E. (2001). Enhancement of heat transfer by means of a corona wind created by a wire electrode and confined wings assembly. [Article]. Applied Thermal Engineering, 21 (3 ), 265 282. Leger, L., Moreau, E., & Touchard, G. G. (2002). Effect of a DC corona electrical discharge on the airflow along a flat plate. [Article; Proceedings Paper]. Ieee Transactions on Industry Applications, 38 (6), 1478 1485. Leonov, S., Opaits, D., Miles, R., & Soloviev, V. (2010). Time resolved measurements of plasma induced momentum in air and nitrogen under dielectric barrier discharge actuation. [Article]. Physics of Plasmas, 17 (11), 8. McKinney, P. J., Davidson, J. H., & Leone, D. M. (1992). C URRENT DISTRIBUTIONS FOR BARBED PLATE TO PLANE CORONAS. [Article]. Ieee Transactions on Industry Applications, 28 (6), 1424 1431.
104 Molki, M., & Bhamidipati, K. L. (2004). Enhancement of convective heat transfer in the developing region of circular tubes usin g corona wind. [Article]. International Journal of Heat and Mass Transfer, 47 (19 20), 4301 4314. Moreau, E. (2007). Airflow control by non thermal plasma actuators. [Review]. Journal of Physics D Applied Physics, 40 (3), 605 636. Moreau, E., Benard, N., Lan Sun Luk, J. D., & Chabriat, J. P. (2013). Electrohydrodynamic force produced by a wire to cylinder dc corona discharge in air at atmospheric pressure. [Article]. Journal of Physics D Applied Physics, 46 (47), 14. National Aerona utics and Space Administration. (2003). Paints. Retrieved August 3, 2015, from http://masterweb.jpl.nasa.gov/reference/paints.htm Raizer, I. P. (1991). Gas discharge physics Retrieved fr om WorldCat database Available from http://www.gbv.de/dms/bowker/toc/9780387534459.pdf Available from Materials specified: Table of contents http://www.gbv.de/dms/bowker/toc/9780387534459.pdf Riherd, M., & Roy, S. (2012). Measurements and simulations of a channel flow powered by plasma actuators. [Article]. Journal of Applied Physics, 112 (5), 9. Riherd, M., & Roy, S. (2013). Serpentine geometry plasma actuators for flow control. [Article]. Journal of Applied Physics, 114 (8), 13. Rizzetta, D. P., & Visbal, M. R. (2010). Large Eddy Simulation of Plasma Based Turbulent Boundary Layer Separation Control. [Article; Proceedings Paper]. Aiaa Journal, 48 (12), 2793 2810. Rizzetta, D. P., & Visbal, M. R. (2011). Numerical Investigation of Plasma Based Contr ol for Low Reynolds Number Airfoil Flows. [Article; Proceedings Paper]. Aiaa Journal, 49 (2), 411 425. Robinson, M. (1961). Movement of air in the electric wind of the corona discharge. American Institute of Electrical Engineers, Part I: Communication and Electronics, Transactions of the, 80 (2), 143 150. Robinson, M. (1962). A HISTORY OF ELECTRIC WIND. [Article]. American Journal of Physics, 30 (5), 366 &. Roth, J. R. (1995). Industrial plasma engineering. Vol. 1 . Bristol: Institute of Physics Pub.
105 Roth, J. R. (2003). Aerodynamic flow acceleration using paraelectric and peristaltic electrohydrodynamic effects of a One Atmosphere Uniform Glow Discharge Plasma. [Article; Proceedings Paper]. Physics of Plasmas, 10 (5), 2117 2126. Roth, J. R., Sherman, D. M., & Wilkinson, S. P. (2000). Electrohydrodynamic flow control with a glow discharge surface plasma. [Article; Proceedings Paper]. Aiaa Journal, 38 (7), 1166 1172. Santhanakrishnan, A., & Jacob, J. D. (2007). Flow control with plasma synthetic jet actuators. [Article]. Journal of Physics D Applied Physics, 40 (3), 637 651. Schatzman, D. M., & Thomas, F. O. (2010). Turbulent Boundary Layer Separation Control with Single Dielectric Barrier Discharge Plasma Actuators. [Article; Proceedings Paper]. Aiaa Journal, 4 8 (8), 1620 1634. Schutze, A., Jeong, J. Y., Babayan, S. E., Park, J., Selwyn, G. S., & Hicks, R. F. (1998). The atmospheric pressure plasma jet: A review and comparison to other plasma sources. [Article]. Ieee Transactions on Plasma Science, 26 (6), 1685 1 694. Sheu, W. J., Hsiao, J. J., & Wang, C. C. (2013). Effect of oscillatory EHD on the heat transfer performance of a flat plate. [Article]. International Journal of Heat and Mass Transfer, 61 , 419 424. Soetomo, F. (1992). The influence of high voltage d ischarge on flat plate drag at low Reynolds number air flow. Unpublished M.S.; Iowa State University. Timmins, B. H., Wilson, B. W., Smith, B. L., & Vlachos, P. P. (2012). A method for automatic estimation of instantaneous local uncertainty in particle im age velocimetry measurements. Experiments in Fluids, 53 (4), 1133 1147. Tirumala, R., Benard, N., Moreau, E., Fenot, M., Lalizel, G., & Dorignac, E. (2014). Temperature characterization of dielectric barrier discharge actuators: influence of electrical and geometric parameters. [Article]. Journal of Physics D Applied Physics, 47 (25), 12. Toyota, H., Zama, S., Akamine, Y., Matsuoka, S., & Hidaka, K. (2002). Gaseous electrical discharge characteristics in air and nitrogen at cryogenic temperature. [Article]. Ieee Transactions on Dielectrics and Electrical Insulation, 9 (6), 891 898. Velkoff, H. R., & Godfrey, R. (1979). LOW VELOCITY HEAT TR ANSFER TO A FLAT PLATE IN THE PRESENCE OF A CORONA DISCHARGE IN AIR. [Article]. Journal of Heat Transfer Transactions of the Asme, 101 (1), 157 163. Velkoff, H. R., & Ketcham, J. (1968). EFFECT OF AN ELECTROSTATIC FIELD ON BOUNDARY LAYER TRANSITION. [Note] . Aiaa Journal, 6 (7), 1381 &.
106 Visbal, M. R. (2010). Strategies for control of transitional and turbulent flows using plasma based actuators. [Article]. International Journal of Computational Fluid Dynamics, 24 (7), 237 258. Wang, C. C., Durscher, R., & Roy , S. (2011). Three dimensional effects of curved plasma actuators in quiescent air. [Article]. Journal of Applied Physics, 109 (8), 9. Wilson, B. M., & Smith, B. L. (2013a). Taylor series and Monte Carlo method uncertainty estimation of the width of a prob ability distribution based on varying bias and random error. [Article]. Measurement Science & Technology, 24 (3), 11. Wilson, B. M., & Smith, B. L. (2013b). Uncertainty on PIV mean and fluctuating velocity due to bias and random errors. [Article]. Measurem ent Science & Technology, 24 (3), 15. Zebboudj, Y., & Ikene, R. (2000). Positive corona inception in HVDC configurations under variable air density and humidity conditions. [Article]. European Physical Journal Applied Physics, 10 (3), 211 218. Zouzou, N., Moreau, E., & Touchard, G. (2006). Electrostatic precipitation in a point plan configuration. [Article; Proceedings Paper]. Journal of Electrostatics, 64 (7 9), 537 542.
107 BIOGRAPHICAL SKETCH Pengfei Zhao wa s born in 1988 in the city of Xi an, China . In 2010, he received his B.S. degree of applied physics from University of Science and Technology of China, Hefei, China. Then, he joined Applied Physics Research Group (APRG) in the Department of Mechanical and Aerospace Engineering at the University of Florida. His research interests include plasma physics, plasma actuator and flow control. He received his M.S. of aerospace engineering from the University of Florida in the spring of 2014. He also received his Ph.D. of aerospace engineering with a minor in statistics in the fall of 2015.