52.80.Hc MARIAH hypersonic wind tunnel and the AJAX scramjet engine concepts have led to a new phase of

ospace industry, with many interdisciplinary applications. Compared with

propulsion systems and channel ow accelerators, EMFC concepts applied to control surface

aerodynamics have not seen the same level of advancement that may eventually produce a device

. . . . . .

n and

ues . .

w expe

ackagi

ts . . . .

. . . . .

that an electromagnetic eld could be coupled with an ionized

Withame.

was the ability to solve the complex magnetohydrodynamic

ARTICLE IN PRESS

Contents lists availab

ls

Progress in Aero

Progress in Aerospace Sciences 45 (2009) 3049requirement of high strength magnets has been achieved to someat the 39th AIAA Plasmadynamics and Lasers Conference, Seattle, Washington,

June 2326, 2008.gas ow to accelerate or decelerate it, delay boundary layer equations, which has eased tremendously stemming from thedevelopment of powerful computing hardware and numericalmethodologies. Next, Resler and Sears mentioned that powerfulmagnets would be needed for ionized uid ow control. This

extent. Electromagnets can produce elds of several tesla,and superconducting magnets can reach tens of tesla. However,

$ An abbreviated version of this paper was presented as AIAA Paper 2008-3788

Corresponding author. Tel.: +18172725120; fax: +18172725124.

E-mail address: eric.braun@mavs.uta.edu (E.M. Braun).0376-04

doi:10.1Fifty years ago, an article was written describing the prospectsfor Magneto-Aerodynamics [1]. In it, Resler and Sears stated

The authors also discussed several advancements critical to theprogress of electromagnetic ow control (EMFC). Among them1. Introduction separation, or to control skin friction and heat transfer.several additions since that time, these goals remain the s3.5. Overall feasibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4. Flow control by glow discharge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

5. Flow control by DBD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

6. Conclusions and future outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47Contents

1. Introduction . . . . . . . . . . . . . . . .

2. Magnetohydrodynamic interactio

3. EMFC surface actuator design iss

3.1. Channel ow and open o

3.2. Power consumption and p

3.3. Selection of EMFC magne

3.4. Conductivity . . . . . . . . . . .21/$ - see front matter & 2008 Elsevier Ltd. A

016/j.paerosci.2008.10.003that can be integrated with an aircraft or missile. The purpose of this paper is to review the overall

feasibility of the different electric and EMFC concepts. Emphasis is placed on EMFC with high voltage

ionization sources and experimental work.

& 2008 Elsevier Ltd. All rights reserved.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

rimentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

ng. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3952.80.Pi MHD research in the aerExperimental research in aerodynamic control with electric andelectromagnetic elds$

E.M. Braun , F.K. Lu, D.R. Wilson

Department of Mechanical and Aerospace Engineering, Aerodynamics Research Center, University of Texas at Arlington, Arlington, TX 76019, USA

a r t i c l e i n f o

Available online 7 January 2009

PACS:

47.65.d47.85.lb

47.85.ld

a b s t r a c t

Fifty years ago, publications began to discuss the possibilities of electromagnetic ow control (EMFC) to

improve aerodynamic performance. This led to an era of research that focused on coupling the

fundamentals of magnetohydrodynamics (MHD) with propulsion, control, and power generation

systems. Unfortunately, very few designs made it past an exploratory phase as, among other issues,

power consumption was unreasonably high. Recent proposed advancements in technology like thejournal homepage: www.ell rights reserved.le at ScienceDirect

evier.com/locate/paerosci

space Sciences

ARTICLE IN PRESS

erosthe size of these magnets makes their integration into anaerospace vehicle problematic. Also, the magnets are dependentupon large power supplies. Similarly, research into rare-earthmaterials has progressed considerably since 1958, with inexpen-sive neodymium-based magnets currently available with max-imum surface elds in the 0.51.0 T range. However, their use foraerodynamic control is limited since their magnetic elds arereduced as temperature is increased, making their incorporationinto applications like scramjet inlets difcult if not impractical.

Nomenclature

b induced magnetic eld, TB magnetic eld, Tcf skin friction coefcientDBD dielectric barrier dischargeE electric eld, V/me electron charge, 1:602 1019 CEFC electrohydrodynamic ow controlEMFC electromagnetic ow controlF force, NFL Lorentz force, N (single particle), N=m3 (ionized

particles per unit volume)I current, AIBL sB2L=ru1cf =21=2, boundary layer magnetic inter-

action parameterIEM BEsL=ru2, electromagnetic interaction parameterIM sB2L=ru, magnetic interaction parameterJ current eld, A=m2

L characteristic length, mn number density, 1=m3

p static pressure, Paq electric charge, C

E.M. Braun et al. / Progress in AIn addition to the strength of the magnetic eld, EMFC is alsodependent upon the conductivity of the ionized airow. Reslerand Sears believed that articial seeding of the airow to createhigher plasma conductivities would need development. At thetime, plasma jet sources were capable of creating high valuesof conductivity for ground testing. As an example, consider alinear Lorentz force accelerator developed in the 1960s [2]. Theaccelerator had a square cross-section of 2.54 cm sides at the inletthat diverged to about 2.54 5 cm at the exit with an overalllength of 76 cm. The 60 electrode pairs in the accelerator werepowered by a warehouse of 1700, 12V automotive batteries.The current draw of the accelerator electromagnets reached up to900A at 80V. Finally, the plasma generator operated with a10MW power supply, and could create a ow with a conductivityof up to 500

A

=m (with seeding). Note that units of conductivityare labeled using

A

=m which is equal to O1=m. Historically,conductivity has been described using mhos per meter althoughthe unit name siemens (S) has been designated for O1 to make

A

=m equal to S/m. In order to reduce the power requirement,seeding the plasma jet with low ionization energy potassium andcesium compounds was explored, which resulted in a tremendousincrease in conductivity relative to the unseeded gas. For instance,a hypersonic vehicle ying at an altitude of 30 km at Mach 16would ionize the air after a bow shock to s 0:05 A=m. Adding0.1% potassium by weight could boost the conductivity to roughly1

A

=m [3,4], a 20-fold increase. However, the vehicle-scaledpower requirement of an air-breathing engine incorporatingthermal ionization and thrust generation by an electromagneticaccelerator currently may only be met by a system such as an on-board nuclear reactor. Research in this eld waned by 1970.For control surface aerodynamics, thermal ionization, whetheraugmented by seeding or not, may not be feasible or evendesirable. Its benets and drawbacks have been discussed forconcepts like the MARIAH hypersonic wind tunnel facility [5] andthe AJAX scramjet power generator [68]. In particular, seedingmay contaminate the ow of the MARIAH wind tunnel. At speedsbelow that which result in signicant shock-induced ionization,EMFC may have serious limitations compared to its overallbenets since a separate non-thermal ionization system must be

ReM m0suL, magnetic Reynolds numberr distance, mS cross-sectional area, m2

u ow speed, m/sV voltage, VWIG weakly ionized gasx streamwise Cartesian coordinatey spanwise Cartesian coordinatez transverse Cartesian coordinateZ number of net charges on a particle0 permittivity of vacuum, 8:854 1012 F=mm0 permeability of vacuum, 1:257 106 N=A2r density, kg=m3re charge density, C=m3s conductivity,

A

=m

Subscripts

BL boundary layerp particle1 freestream

pace Sciences 45 (2009) 3049 31utilized. A few situations do exist in which ionization is currentlyexperienced by an aerospace vehicle. For instance, the SpaceShuttle interacts with ionized particles while in low Earth orbit [9]and during re-entry, where the re-entry environment has resultedin numerous studies (e.g., Refs. [10,11]) of how to improve currentvehicle designs with the addition of electromagnetic elds.However, with the Space Shuttles impending retirement and nonew prospects for the incorporation of actuators into full-scalespace vehicles (assuming the Orion design is nearly nalized), lessgrandiose platforms like hypersonic missiles may currently be thebest near-term candidate for EMFC systems. With that in mind,the ight Mach number may be limited to below about 15, forwhich articial creation of an adequate amount of ow con-ductivity is necessary. Recently, generation of a conductive gas,also known as a weakly ionized gas (WIG), has been accomplishedusing high voltage elds, laser beams, or perhaps directedmicrowaves [12,13]. However, these ionization methods producea far lower level of conductivity when compared with results from50 years ago. Experimentally, the maximum realized values of s inair are currently in the 0:121

A

=m range with high voltage elds.Raising the gas conductivity and minimizing power consumptionare obviously priorities if practical aerospace systems are to berealized.

Also of signicant note has been the development of owcontrol systems utilizing only an electric eld to create plasma.The design of an EFC device with only a high voltage eld is muchless complex since the eld will ionize the air itself. Electrohy-drodynamic ow control techniques can be divided into twocategories: glow discharges and dielectric barrier discharges(DBDs). The physics for the two categories is similar. An air gap

F2r2 eZ2 14p02 1

jr2 r1j3, (2)

ARTICLE IN PRESS

erosexists between the anode and cathode region of a glow discharge,while a much thinner dielectric material barrier is used forDBD systems which limits arcing. Paschens law states that theelectrical breakdown voltage is based on gap distance andpressure. Because the anode and cathode of a DBD are separatedby a thin dielectric gap, the operating pressure is higher andadditionally the high electric eld signicantly raises the outputCoulomb body force. Because of a larger air gap distance, mostglow discharge research has occurred with low pressure and thecontrol mechanism is thought to be more of a thermal effect.Although both systems solve the conductivity generation problemby ionizing the air without a separate system, the value of s isvery low, even 1052107

A

=m for some DBD actuators. Bothsystems also operate often using low power requirements thatmay be met by current on-board generators.

One may assume the magnitude of the force generated byelectric or electromagnetic elds is naturally a reection of theamount of power consumed. Considering the potential use of eachin the aerospace industry, there is a tendency to associateelectromagnetic elds with systems consuming a large amountof power. Systems based solely on the Coulomb force have beenproposed and used for applications with relatively less power(i.e., electrostatic ion thrusters and ion lifters). Therefore, afundamental issue to address for the future of ow control usingthese elds is dening how much power is needed for anappreciable control force. Put another way, while an electro-magnetic eld is generally associated with larger scale aerospaceapplications and may be more robust, an electric eld may be allthat is necessary to generate a satisfactory control force. Theanswer to this question will likely cause one to eventually be farmore appropriate for use over the other.

Recent experimental results have not entirely addressed thisissue, although some progress has been made. Aerospace researchinvolving electromagnetic elds has focused on hypersonic ows[14], particularly for the augmentation of scramjet propulsionsystems. Most work involving MHD and scramjets has beencomputational, which is clearly understandable due to the costand complexity of testing. However, this situation has led tosignicant differences between experimental and computationalwork. Experimental values of pressure, conductivity and magneticeld strength are usually below what is assumed analytically or incomputational simulations. For example, Bruno et al. assumed amagnetic eld between 7 and 17T as part of a rst-orderelectromagnetic hypersonic propulsion system [15]. In anotherdesign, Park et al. computed values of B 11:28T, s 35:87 A=m(with seeding) and p 1:25MPa at the entrance to the scramjetsMHD accelerator [16]. Recent experimental environments havebeen generally limited to 0.54.0 T, at most a few

A

=m, andrareeld pressures. Low pressure testing has been a method toincrease IM understandably as such conditions facilitate ioniza-tion. Little discussion has been articulated on the subject matterof selecting appropriate values of these parameters for aero-dynamic control surfaces. Interestingly, these experimental valuesof B, s and p might be better suited for control surfaces ratherthan for propulsion systems. Practical values of crucial scalingparameters must be established so as to dene what value rangesshould be associated with larger, propulsion-associated systemsand smaller control surface systems.

Conversely, electrohydrodynamic ow control surfaces havemostly been experimentally demonstrated with low freestreamspeeds and Reynolds numbers [17]. These demonstrationsrepresent a signicant departure from the high-speed ightregime where electric and electromagnetic elds were rst

E.M. Braun et al. / Progress in A32considered for use. This situation is especially true for DBDs.High-speed experimentation has appeared with glow discharges,but the studies have historically centered upon bluff bodies morewhere the bracketed term represent the electric eld. Similarly,the magnetic force law states that a force is developed betweentwo current carrying wires which is dependent on distance and,additionally, the orientation of the wires. Over a length of wire dlthis force is written as

dFr m04p

I I0

jr r0j3 dl dl0 r r0, (3)

where the prime is used to denote the properties of one wire fromanother. Invoking the BiotSavart law leads to an expression forthe magnetic eld at r, namely,than aerodynamic surfaces. Again, static pressure is often in therange of a few torr to facilitate the creation of a diffuse electricaldischarge. Although the high-speed control effect of glowdischarges appears to depend more on an increase in of the localspeed of sound caused by Joule heating rather than the presenceof an electrohydrodynamic force, rapid control actuation isdesirable. Increasing the Reynolds number while still providingeffective control with these systems is crucial. For ow environ-ments besides very low speeds where DBDs appear best suited,glow discharge and EMFC concepts are competitive for a widerange of control possibilities.

2. Magnetohydrodynamic interaction and scaling

The central difference between electrohydrodynamics (EHD)and magnetohydrodynamics (MHD) is the force produced duringthe interaction of ionized particles with the electric or electro-magnetic elds, respectively. For EHD, it is the Coulomb forcewhile for MHD it is the Lorentz force. These interactions are oftensummed up in one equation written as

F qE u B. (1)With research in electric and EMFC beginning to focus mainlywithin the boundary layer where E is high and u is low, there is atendency to observe Eq. (1) and conclude that the presence of amagnetic eld has little effect on the magnitude of the body forcefrom a simple order-of-magnitude comparison of the two termson the right-hand side. This conclusion is incorrect since both RHSterms of Eq. (1) actually contain an electric eld component. Notethat the u B product is actually an electric eld, usually referredto as the internal-induced electric eld in channel ow applica-tions. The single E term is referred to as the applied or externalelectric eld for accelerators and generators, respectively. Inter-action of only an electric eld with ionized particles will producea body force, but its relative magnitude with respect to the forceproduced by an electromagnetic eld cannot easily be determinedby Eq. (1). A strong current eld interacting with ionized particlescan create an induced magnetic eld, but that eld and any bodyforce generated from that interaction is likely negligible for EMFCas will be shown. The concepts above have been well establishedin the literature [18,19].

In order to provide insight to MHD interaction and scaling, aderivation of the EHD and MHD forces is useful. Coulombs lawstates that two charged particles exert a mutual force in adirection parallel to the line connecting each particle. If one ofthese particles is held stationary in the reference frame of theother one, it creates an electric eld and exerts a force on the otherparticle, written as

eZ r r

pace Sciences 45 (2009) 3049Br m04p

Zl0

I0r0dl0 r r0jr r0j3 . (4)

term and velocity proles using the inner-law variables should beuseful in comparing effects over different test facilities.

As an example of the importance of these parameters, considera at plate, EMFC actuator with ve surface electrodes eachseparated by 1.59 cm of dielectric material as shown in Fig. 1. Theelectrodes alternate with embedded magnets. The gure is acomputer rendition of an actuator constructed and tested [25],with the dielectric material shown as transparent. Such actuatorgeometry may be typical for surface EMFC as the eld progresses.Flat plate actuators with alternating magnets and electrodes havebeen previously demonstrated with salt water environments ascan be seen in [22] along with a discussion of related research. Themiddle and outer electrodes are grounded, while the potential ofthe other two is 500V. The length and width of the electrodes are1.27 and 0.51 cm, respectively. The permanent, NdFeB magnets,1.27 cm square by 2.54 cm long, are embedded about 1mm below

ARTICLE IN PRESS

erosNext, using Amperes law, Eq. (3) can be simplied to

dFr Irdlr B. (5)If S denotes the cross-sectional area of the wire, n denotes thenumber of particles and the subscript p denotes a particle alongthe wire, then the force is

dF npeZpSdlup B. (6)Furthermore, the force on one particle is

Fp eZpup B. (7)Therefore, the combined electric and magnetic forces on a particleare

Fp eZpE up B. (8)Neglecting polarization and magnetization effects, the body forcecomponents on an ionized gas or liquid are shown in Eq. (9),which also has omitted the Hall effect (which can be signicant)and ion slip:

F reE J B. (9)Next, the order of magnitude of these electric and electromagneticLorentz force components can be approximated, allowing for acomparison between the two [18]

reE 0E

2

L, (10)

FL j J Bj sE uBB. (11)The relative magnitude of the two terms approximating FL inEq. (11) is critical for EMFC characterization. From Eqs. (10)and (11), the ratio of the electrohydrodynamic force to themagnetohydrodynamic force is

FEHDFMHD

reEj J Bj 0E

2

sLE uBB. (12)

With the potential difference between electrodes on MHDaccelerators and recent EMFC actuators usually on the order of1000V or less, it is apparent that the electrohydrodynamic forcewill be negligible, unless the characteristic length is very small.Aerodynamic control systems will operate in far larger environ-ments.

For EMFC characterization, it appears that much progressis needed to support experimental research. Historically, themagnetic interaction parameter IM has been used to dene theratio of magnetic body force to the inertia of the uid:

IM EM forces

Inertia forces sB

2L

ru. (13)

The prospects of reaching IM 1 were discussed as a performancebenchmark in literature decades ago (e.g., Ref. [20]) and has seenuse again with many recent EMFC publications. However, theserecent EMFC environments usually demonstrate IM51 and ittakes a combination of very high ow speed and low density toreach unity. Despite having a low value, these same experimentalresults still demonstrate appreciable changes to the ow. AsElsasser remarked, dimensional relations in MHD are often muchlarger or much smaller than unity [21]. Perhaps other dimension-less numbers are more suitable for characterizing and scaling theeffects of EMFC. With pressure changes often measured to conrmthe effect of the Lorentz force, variables to consider include B, E, r,p, u, L, and s. Table 1 shows several resulting dimensionlessnumbers derived using the Buckingham Pi theorem.

The most common term developed from Table 1 is E=Bu,

E.M. Braun et al. / Progress in Areferred to as the MHD loading parameter. It can be seen as a ratioof the total power per unit volume added to the ow to thedirected kinetic energy [2]. As such, the parameter should beminimized to raise the conversion efciency of energy used toaccelerate the ow (rather than letting it contribute to Jouleheating). From the last combination of variables, E=Bu and IM areformed. As is common with Buckingham Pi theorem results, twodimensionless numbers can be multiplied to form furtherdimensionless parameters. Usually only a few will have signi-cance, and one to highlight from that set is distinguished as theelectromagnetic interaction parameter:

IEM BEsLru2

. (14)

This parameter, like IM , is a ratio of the magnetic body force to theuid inertia. It is interesting to note their individual similarity toone of the respective Lorentz force components shown in Eq. (11).Although E Bu has been assumed in the derivations of manypublications (e.g., Refs. [18,21]), this assumption is not necessarilyvalid for small-scale environments like boundary layer EMFCwhere E is high and u is low. In fact, it appears that IEM and IM areuseful over separate design spaces depending on where EbuB orE5uB, respectively. The term B3sL=rE has not seen use, and isanother product of the MHD loading factor and the interactionparameter. The term developed in the third row is interestingsince it relates a magnetic force to static pressure, which mayprove to be useful with further experimental testing. Anothermodied interaction parameter that has appeared in the literatureis [12,2224]

IBL sB2L

ru1cf =2

p . (15)This term denes u as the friction velocity, whereby u u1

cf =2

p. As research into boundary layer EMFC increases, this

Table 1Resulting dimensionless numbers for several EMFC variable combinations.

Variable combinations Dimensionless numbers

B; E; s; u; p EBu

B; E; s; u; r EBu

B; E; s; L; p BEsLp

B; E; s; L; r B3sLEr

B; E; s; L; p; u EBu

;BEsLp

B; E; s; L; r; u EBu

;sB2Lru !

BEsLru2

pace Sciences 45 (2009) 3049 33the surface. The measured magnetic eld across the surface of theactuator B is shown in Fig. 2. Next, assume this device is operatingunder a freestream airow of 1000m/s. The conductivity of the air

ARTICLE IN PRESS

Fig. 3. The common logarithm of the Lorentz force (N=m2) across a spanwise slice

erosFig. 1. Image of the ve electrode, four magnet actuator plate with dielectricmaterial shown as transparent.

E.M. Braun et al. / Progress in A34is arbitrarily set to be 1

A

=m, produced by a separate ionizationsystem. The boundary layer thickness has been xed at 1.0 cmand the velocity prole follows a typical turbulent shape. Theboundary layer velocity as a function of height off of the at plateis denoted as uBL uz.

The magnetic eld components (Bx and By), actuator geometry,and electrode potentials were used as boundary conditions in acomputational MHD code to determine the Lorentz force andinteraction parameters. The magnetic eld generated showedreasonable agreement with experimental magnetic eld measure-ments that were taken up to z 1:27cm.

Fig. 3 shows the common logarithm of the local Lorentz forcefor a two-dimensional spanwise slice over the actuator atx 1:8 cm. The locations of the magnets and electrodes arelabeled at the bottom. The Lorentz force is most heavilyconcentrated over the electrodes and especially at their edgeswhere electric charge builds. The magnitude of the Lorentz forceproduced in part with the NdFeB magnets drops signicantly withheight, but remains viable for boundary layer applications. TheLorentz force is not as high over the outer electrodes because theyare grounded. In this spanwise arrangement, it is important tokeep the outer electrodes grounded to prevent a counteractingbody force from developing past the edges of the actuator.

Fig. 2. Total magnetic eld located on the surface of the at plate over the10:8 3:2 cm area.pace Sciences 45 (2009) 3049Figs. 4 and 5 show the common logarithm of the localinteraction parameters IM and IEM , respectively. Logarithms areused for two reasons, rst, because the localized values range overseveral orders of magnitude. Second, the similarities anddifferences between the gures are more important than theactual values. For the calculation of the parameters, r wasapproximated as 1kg=m3 throughout the boundary layer.(In actuality it will increase or decrease towards the surfacedepending on the wall conditions.) Additionally, L was removedfrom the parameters since it is not meaningful for the two-dimensional slice of interest. Fig. 4, as would be expected, showsIM distinctly centered upon the embedded NdFeB magnets. If IM istruly a good choice as a non-dimensional representation of theLorentz force for this actuator, then the localized contours of Figs.3 and 4 would be similar. An evaluation of the data used for thisexample indeed shows that EbuB across the boundary layer.Fig. 5 shows that the Lorentz force more appropriately follows thesame contours as IEM . In Fig. 5, the maximum values of IEM arecentered on the electrodes. The only difference between thegures is IEM is more uniform near to the actuator surface becauseof the use of u2BL.

over the actuator at x 1:8 cm.

Fig. 4. The common logarithm of IM=L across a spanwise slice over the actuator atx 1:8cm.

ARTICLE IN PRESS

erosThe consequences of selecting the correct interaction para-meter for an actuator will have a larger impact than thecomparison of these gures. Most important, the performancewill scale differently with E, B, and u depending on the value of theMHD loading parameter. On a case-by-case basis, the average ofthe localized MHD loading parameter in the boundary layershould be known before the actuator and its operating conditionsare characterized by IM or IEM. Although IM has been more widelyused than IEM , the example shows that the design space overwhich IEM is applicable is signicant. Also, where E Bu, neitherterm may be appropriate for correctly scaling the Lorentz forceeffects.

Going back to Fig. 3, the Lorentz force is noticeably non-uniform across the span of the actuator. This may be unavoidablefor surface actuators since the electric and magnetic elds haveinherently large gradients. The simplest strategy for creating someuniformity is to match the maximum B eld points with theminimum E eld points and vice versa along the actuator surface.Earlier MHD studies have shown that the geometry of segmentedelectrodes has a large impact on the distribution of the electric

Fig. 5. The common logarithm of IEM=L across a spanwise slice over the actuator atx 1:8 cm.

E.M. Braun et al. / Progress in Aeld [26], and that observation certainly applies to this exampleactuator. As the MHD loading parameter decreases, the maximumLorentz force locations will gradually shift from over the surface ofthe electrodes to over the magnetic poles.

Computing these local parameters is only practical in acomputational environment, but doing so may be necessarybefore accurately scaling the performance of the actuator with anyparticular interaction parameter. Establishing u as the frictionvelocity in the manner of IBL for comparison purposes acrossdifferent facilities is recommended.

Continuing on, the magnetic Reynolds number is a measure ofthe ease with which an ionized gas moves through a magneticeld, and is dened as

ReM m0suL. (16)

The number can also be seen as a ratio between convection andmagnetic diffusion effects, where diffusion (and therefore B)dominates the system for ReM51 [27]. When ReM41, the motionof charged particles in a current eld can create an inducedmagnetic eld b, responsible for many astrophysical phenomena.A relationship between the magnetic Reynolds number, thecurrent eld, and the induced magnetic eld is

ReMJ r b. (17)

20 torr. The channel itself is small enough to be surrounded with

an electromagnet that can reach B 2T, while a NdFeB magnetconguration (B 0:4T) was also demonstrated in earlier works[12,30]. A decrease was measured in the pressure uctuation dueto the Lorentz force in Ref. [12], but no direct measurements ofpressure were taken at the time. The ow conductivity from theionization system has risen with more recent publications, and isusually on the order of 0:1

A

=m. Further studies with electro-magnets have shown large pressure differences created by theWith the aforementioned ranges of s, u, and L, the magneticReynolds number is far lower than unity within the boundarylayer for this example actuator. This result should be true for mostif not all EMFC actuators, as the creation of an induced magneticeld for a ight vehicle needs a combination of very high speedand a current eld likely too large to be supplied by an on-boardgenerator.

3. EMFC surface actuator design issues

For electromagnetic elds to be successfully applied into acontrol surface actuator, several issues must be considered. First,while channel ow setups are ideal for understanding the physicsof EMFC, open ow experiments must be considered in which theEMFC actuator is contained in a at plate or airfoil. Powerconsumption and packaging are important issues to address, withthe selection of magnets and the method of ionization key tosuccess. The selection of EMFC magnets is a signicant mattersince rare-earth materials would be ideal for placement inside athin control surface, except for the major problem in which theireld strength is relatively low and adversely affected by heat.Finally, unlike EFC, a sufciently high value of conductivity mustbe created by an additional system for applications in whichthermal ionization is not encountered in the ight regime chosen.

3.1. Channel ow and open ow experimentation

While one of the focal points of EFC has been for controlsurfaces, the same cannot necessarily be said for EMFC systemsthus far. EMFC experiments applied to aerospace systems havetypically been for scramjet-like systems and have taken place in achannel ow environment. Analytical approaches have also beenwell established for MHD channel ows while open owmodeling requires more sophistication. The design highlightsand operating conditions of several recent facilities are discussedbelow. Ionization systems for each facility will be elaborated uponfurther into the text.

One recent channel ow facility has been developed with theintent to place an accelerating or retarding Lorentz force on ahigh-speed ow (Mach 34) of air or another mixture of gases[12,23,2837]. The walls of the test section each have electrodesmounted into them. Two electrodes mounted opposite of eachother create a WIG in the test section using high voltage, low dutycycle pulsing while the other two are connected to a DC currentsystem. That system provides the energy for the Lorentz force solong as the level of conductivity provided by the ionizationelectrodes is enough for current to cross the electrode gap. TheLorentz force may be applied with or against the ow dependingon the electrode and magnet polarity. Experimental data collectedhave included ow visualization, ow uctuation measurements,Lorentz force-induced pressure changes, and the output of theionization and Lorentz force systems.

The test section static pressure of this facility ranges from 5 to

pace Sciences 45 (2009) 3049 35Lorentz force. For instance, Nishihara et al. [29] have shown astatic pressure increase of 1720% for a retarding force and 57%for an accelerating force of the same magnitude. Similar data of

ARTICLE IN PRESS

erosgreater magnitude from [35] are shown in Fig. 6. The gure showsthe normalized pressure difference for dry air between unalteredow, ow with a retarding force, and ow with an acceleratingforce. The retarding force is more effective than the acceleratingforce because it works with Joule heating to create the rise inpressure, while the accelerating force works against Joule heating.The magnitude of the pressure rise also appears to be dependent

Fig. 6. Normalized static pressure traces downstream of an EMFC actuator forM 3 dry air for four electromagnetic arrangements (from [35]).

E.M. Braun et al. / Progress in A36upon Lorentz force polarity, where it is suggested that the testsection Mach number and pressure are affected by the electro-magnetic force interaction.

Another channel ow test section has been constructed toexplore the effects of a constricted plasma column operatingunder the presence of a magnetic eld [38]. In the test section,two tapered electrodes were placed on the side of one ofthe tunnel walls. After actuating a high voltage DC circuit, aconstricted plasma column forms between the electrodes andpropagates downstream due to the tapering. A helium-cooledsuperconducting ring magnet surrounds the channel and cangenerate a B eld up to 7 T, which increases the velocity at whichthe plasma column travels [39]. Since EFC systems are based onmomentum transfer due to collisions between the ions and aneutral ow, this system can be seen as a novel method toenhance the momentum transfer using magnetic elds. Instead oftwo separate power supplies for ionization and Lorentz forcegeneration, a single 20 kV, current regulated power supply is used.The electric eld generated by the 20 kV potential ionizes the gasto the point of breakdown, and the resulting arc draws up to aspecied current limit. Once the current limit is reached, thepower supply voltage drops signicantly, so the power input intothe ow is considerably less than its maximum value of 20 kW.Therefore, the initial 20 kV potential before breakdown acts likean ignition system for the EMFC actuator.

The test section has a Mach number of 2.8 and static pressureof 28 torr. Experimental data collected consisted of static pressuremeasurements in addition to ow visualization. A wedge wasplaced inside the test section to study the effects of the systemon shock/boundary layer interaction [39]. Results show that theactuator is able to move the separation bubble induced by theshock interaction, and the static pressure across the wedge isaffected.

A third facility has been developed with a at plate securedinside a free jet test section designed for basic research in MHD[40]. An electromagnet with a maximum eld of 3.5 T surroundsthe entire channel, and a NdFeB conguration has also beentested. The Mach number in this facility is about 5.0, and the testsection static pressure is designed to simulate an altitude between30 and 50km (approximately 0.67 torr). Besides covering thealtitude range mentioned, the low test section pressure has alsobeen designed with consideration to raising the value of IM , whichreaches approximately 1.5 per meter. The at plate, shown inFig. 13, has two embedded electrodes which use high voltage DC,RF, or a combination of both elds to ionize the air up to 2:5

A

=m.The gure also shows the rareed air pressure in the test sectionallows for a relatively low voltage glow discharge to transmit asubstantial amount of power to support the Lorentz force.

Before incorporating a at plate into the facility, tests wereconducted with blunt body congurations at Mach 5.8 whereexperimental results included plasma diagnostics and aerody-namic force measurement [41]. For the at plate conguration,surface pressure measurements indicated that a Lorentz forcedirected out of the plate has more of an effect than directing it intothe plate, again due to ow coupling and Joule heating issues [42].Both push the luminous region of the glow discharge onto or offthe actuator surface. Although a glow discharge raised lift by up to18% in one set of experiments, applying a magnetic eld cannegatively affect the discharge and void the change in lift [43].A more recent study has moved to testing rectangular andcylindrical inlets supported by computational modeling [44].

EMFC publications have increased in the past few years andother facilities are likely to join those above. The ionizationsystems and overall electrode design for the generation of a WIGor plasma column and interaction with a magnetic eld toproduce the Lorentz force appear to be feasible for high-speed,boundary layer EMFC based on results from these facilities.However, the value of conductivity generated (0:122:5

A

=m) bythe high-voltage systems as well as the test section pressure areseveral orders of magnitude below those that may be necessaryfor the AJAX engine concept. It would be very interesting tomodify the geometry and examine the performance of thesefacilities under an open ow, at plate environment withpressures closer to what may be encountered by a wing or nduring high-speed ight. Magnets for such applications shouldalso be embedded in the surface. Control of slender wings andns, and perhaps the initial stage of an inlet compression system,are likely the best applications for these systems. If changing thegeometry and increasing p are not formidable obstacles, perhapsthese types of systems could be placed on a high-speed missile forcontrol purposes.

With electromagnetic elds, experimental measurements canbe difcult because of signal interference. Typically, one or moretransducer ports are placed downstream of the electromagneticarrangement to capture the change in static pressure resultingfrom the Lorentz force. These data, along with ow visualization,power input, and plasma diagnostics results provide the meansto understand the basic physics of EMFC. The measurement ofaerodynamic forces, conducted in a few studies, will need tobecome more widespread as the actuator designs become morerepresentative of control surfaces. Since most EMFC studies havebeen primarily focused on boundary layer control, it is desirable touse more rened techniques to analyze changes to the boundarylayer prole. The inherent non-uniformity of the Lorentz force

pace Sciences 45 (2009) 3049eld likely adds unusual effects that must be measured as afunction of height above the plate as well as in the spanwisedirection. Fig. 7 shows an example of the change in boundary layer

discuss, increased interest in low Reynolds number aerodynamiccontrol for micro air vehicles (MAVs) has brought the use ofDBDs into consideration [54]. Although current computationaland experimental research appears promising, scaling down tosmaller, low-speed vehicles that t into the useful design space for

ARTICLE IN PRESS

Fig. 9. Awater-cooled electromagnet surrounds an EMFC free jet test section (from[43]).

erosprole for a low-speed salt water freestream ow of about 18mm/s as measured with a particle image velocimetry (PIV)system. Since salt water is naturally conductive (a few

A

=m), atplate Lorentz force actuators have been much easier to build, test,and characterize [22,45]. PIV imaging directly over an EMFCactuator for gas ow may be difcult because of the luminosity ofthe ionized gas and Lorentz force energy addition.

Although the ow speed in Fig. 7 is very low, other researchwith salt water has been conducted at higher speeds. The conceptof EMFC and propulsion for naval applications has existed just aslong as it has for aircraft [46]. The concept has also been provenwith subscale submarines and ships [47,48]. A collection of papersin this eld concerning electromagnetic drag reduction can befound in Ref. [49]. Studies of MHD propulsion have concluded thatit is feasible and desirable because of stealth [50], but effects frombubble formation at the electrodes in salt water and thegeneration of hydrogen and chlorine gas will need mitigation.MHD propulsion for a full-scale submarine will require signicantpower and new developments in efciency for on-board nuclearreactors.

3.2. Power consumption and packaging

For EFC systems, power consumption and packaging arerelatively simple issues. Glow discharges require high voltage,but they are generally low power phenomena. DBDs also requirehigh voltage, but the low operating current again leads to low

Fig. 7. Boundary layer velocity prole downstream of a at plate EMFC actuator forsalt water ow (from [45]).

E.M. Braun et al. / Progress in Apower consumption. Corke and Post report a power level ofapproximately 6.5130W per spanwise linear meter for DBDactuators [51]. Fig. 8 shows the ease at which these DBD actuatorscan be placed onto a surface as long as the material they areembedded in is dielectric [52]. Since the actuators are thin in thestreamwise direction, the power requirement is anywherebetween several hundred watts and a few kilowatts per squaremeter of a hypothetical control surface built from placing theminto rows. In the gure, the separation between DBD strips is1.0 cm, and the RF signal input is 4.5 kV RMS at a frequency of3.3 kHz. At these conditions, a ow speed of 6 m/s is induced atthe edge of the panel. The thrust from these DBD arrangementsrises with dissipated power [53]. Assuming the induced owspeed for ight-ready control surface actuators will need to behigher than a few meters per second, they will eventually havesignicantly higher power consumption than what was reportedin Ref. [51].

Since glow discharge and DBD systems often use these thinsets of electrodes, packaging into control surfaces is alsostraightforward. The largest components of these systems mayindeed be the high voltage circuit elements. As Jayaraman et al.Fig. 8. Smoke visualization of a DBD control surface composed of rows of actuatorscreating an electrostatic force that acts from left to right (from [52]).

pace Sciences 45 (2009) 3049 37these actuators may pose a problem due to the mass and volumeof high voltage circuit components. Studies on the integration ofthese components to small-scale aircraft appear limited and it isrecommended that future research efforts cover this topic.

Concerning EMFC systems, the packaging issue is morecomplex and dependent not only on the power requirement butthe choice of magnets. Fig. 9 shows a large electromagnetsurrounding a hypersonic test section. As used in Ref. [43], itgenerated a magnetic eld of 0.9 T. Although electromagnets usedfor experimental channel ow EMFC are useful in that they canprovide a steady magnetic eld inside the test section (whereas aeld from NdFeB magnets will vary as shown in Fig. 2), theirability to be integrated into ight vehicles is questionable due totheir mass and weight. The use of superconducting magnets onlyexacerbates the problem. The use of rare-earth magnets will leadto the most compact EMFC actuator that may be placed on thesurface of a wing or at the beginning of an inlet compressionsystem. Clearly, the drawback of efcient packaging withembedded permanent magnets is the relative reduction inmagnetic eld strength across the control surface. However, goingback to the discussion of the interaction parameters, if the

to increase the magnetic interaction parameter. However, some of

[6063] and apparently much less work carried out withneodymium alloys. Literally tens of thousands of other alloyshave been tested to determine if they produce a stronger magneticeld, with no signicant ndings. Another method that has beenused in an effort to improve rare-earth magnets is to useexchange-coupling to effectively combine the benets of so-calledhard and soft phases of different magnetic materials [64]. A hardmagnetic phase resists permanent demagnetization (i.e., B eldreduction by heating), and a soft phase exhibits a high level ofmagnetization. This concept has been experimentally proven toraise the energy product of rare-earth magnets where the phasesare combined in nanoparticle assemblies [65]. One of the keyissues to resolve in this eld is uniform nanoparticle dispersionsof the hard and soft phases.

As far as the surface eld is concerned, one can conclude thatneodymium magnets are the better choice for applications withtemperatures ranging up to 350K. However, the typical decline ofthe neodymium surface eld strength as shown in Fig. 12indicates that the performance of samariumcobalt magnets

ARTICLE IN PRESS

erosthese electromagnets and superconducting magnets have massesup to hundreds of kilograms (e.g., Ref. [39]) which will make ightapplications problematic.

The advancements in several new rare-earth magnetic alloysbetween 1970 and 2000 [56] have made permanent magnets acompetitive choice for aerospace applications of MHD. Prelimin-ary demonstrations of their capabilities have been seen in severalpublications [12,22,25,30,40,45]. Permanent magnets, wherepossible, should be considered instead of electromagnets sincethey consume no power and demonstrate much higher values ofenergy density making their strength-to-weight ratio relativelysuperior. Unfortunately, one major drawback of using permanentmagnets for aerospace applications is the fact that high tempera-tures drastically weaken their overall surface eld strength, whichis already low when compared to electromagnets. Permanentmagnets lose their magnetic properties at a specied point calledthe Curie temperature. Prior to that point there is anothertemperature called the maximum operating point, after which amagnet will experience permanent losses to its original strength[57]. For AJAX-style scramjet engines, it is unlikely that perma-nent magnets could be used in the high-temperature environ-ments even with robust active cooling systems.

Fig. 10 shows the maximum operational temperatures ofactuator performance is not wholly dependent on B2, a reductionin B is easier to offset with one of the other variables.

Assuming EMFC surfaces themselves can be made compactenough for ight, there is still a problem with the possibility thatthe Lorentz force power requirements will exceed the output of areasonable on-board generator. As discussed, ionization systemsmay also require a separate power supply. Lorentz force powerrequirement estimates similar to the one presented above for DBDactuators is still relatively unknown. The pressure increase of1720% in Ref. [29] was created with about 2 kW of power wherethe electrodes enclosed a volume of approximately 40 cm3. Thisgure does not appear unreasonably high as many aerospacesystems currently require power of that scale. The power supplyrequirements are also heavily based upon the time of use of theactuator. If the purpose is long-term drag reduction or liftenhancement, an innovative generator is probably required.Perhaps a short duration missile (with EMFC actuators for nal,rapid course corrections) can operate based off a thermal batterytypical of current technology. The fact that many other potentialsources (MEMS microturbine generators, fuel cells, ywheels,capacitors, etc.) of compact on-board power supplies are underdevelopment is encouraging [55].

3.3. Selection of EMFC magnets

Concerning the viability of EMFC, an inquiry must be made tounderstand exactly what range of magnetic surface eld values isneeded. This inquiry leads back to the reason why magnets areneeded for EMFC in the rst place, namely, that the cross productof magnetic eld and the electric eld produces the Lorentz force.The presence of a magnetic eld acts as a facilitator for energyaddition into the uid ow from the electrodes. The energyaddition is then split into Joule heating and the kinetic energy(rate of work done by the Lorentz force) of the uid. The selectionof appropriate magnets for EMFC systems is a current topic ofdebate. As was mentioned previously, many recent experimentalEMFC facilities have used powerful electromagnets capable ofsurrounding a test section since it is a straightforward way

E.M. Braun et al. / Progress in A38samariumcobalt and neodymium magnets charted along withtypical post-shock temperature curves as a function of Machnumber for different wedge angles with an incoming stream at220K. Neodymium magnets are operationally limited to tem-peratures just over 400K, while some samariumcobalt alloys canbe used at temperatures exceeding 800K. While these tempera-tures are still far below the requirements of implementation into,for instance, a multi-shock scramjet inlet, these magnets could beused for slender control surfaces to some extent on supersonicand hypersonic vehicles. Moreover, while it would appear thatsamariumcobalt alloys are superior to neodymium for high-speed aerodynamic control because of the higher operating point,Fig. 11 shows that the high-temperature alloys typically have lessoverall magnetic eld strength [58]. Note that the magnetic uxdensity is measured using teslas, but the values from Fig. 11 arenot representative of the maximummagnetic eld (also measuredin teslas) that will be present on the surface of the magnets.Neodymium and samariumcobalt magnets are widely available,but they rarely demonstrate maximum surface elds over 0.5 T.Fig. 12 shows that it is common for permanent magnets to lose thebulk of their surface eld before reaching their maximumoperational temperatures [59]. Typically, these magnets will seea slight linear decline in surface eld for a limited temperaturerange before reaching a point of rapid decline extending to themaximum service temperature. One of the focal points of currentresearch in magnet development has been to broaden thetemperature range in which only a slight linear decline is present,with signicant improvements made to samariumcobalt alloys

Fig. 10. Temperature versus Mach number for lines of constant wedge angle (11, 51,101, 201) after an oblique shock wave (based on an initial temperature of 220K)along with neodymium and samariumcobalt maximum operating temperatures.

pace Sciences 45 (2009) 3049becomes comparable for higher temperatures. It appears thatmany, if not all, high-speed aerodynamic control conceptsinvolving permanent magnets will require active cooling. This

ARTICLE IN PRESS

erosFig. 11. Magnetic ux density charted as a function of the maximum operatingtemperature for several neodymium and samariumcobalt alloys [58].

E.M. Braun et al. / Progress in Aneed for active cooling is not solely associated with the use ofrare-earth magnets. As exemplied by Fig. 9, some electromagnetsalso require cooling just to operate.

3.4. Conductivity

In 1968, Garrison stated that the performance of MHDaccelerators depends directly upon the magnitude of the electricalconductivity of the seeded working gas [66]. This statementremains true for aerodynamic control, only with more emphasisplaced on improving low-temperature, WIGs. Before then, theconcept of propulsion using electric and magnetic elds hadappeared in the literature for several decades. Jahn presented ashort review of early literature in electric propulsion [67]. Effortsat experimentation began in the late 1950s beginning with theimplementation of plasma jets for propulsion systems [68].Plasma jets were certainly capable of generating highly conductivegases through thermal ionization, but the temperature and powerrequirements were too high for viable aerospace applications atthe time. Alkali salt seeding was introduced into the plasma jet inorder to achieve the same level of conductivity at a considerablylower temperature [3]. Extensive experimentation with differentlow ionization energy seed materials (potassium and cesium)with air or noble gases (argon and helium) appeared in theliterature through the end of the 1960s [6977]. Generating a bulkow conductivity on the order of 1000

A

=m was achievable. Theexperimental gas pressures reported were usually on the order of

Fig. 12. A typical plot of the surface eld decline versus temperature for aneodymium magnet with a maximum operating temperature of 423K [59].one atmosphere. At higher pressures around 10 atmospheres,electron attachment by positive oxygen ions signicantly reducess [78].

Like many other elds, research in MHD was affected by thedirection of the Apollo program. It appears that engineers mayhave assumed that megawatts of power produced by an on-boardnuclear reactor would be available for future MHD accelerator-based propulsion systems, but the nuclear prospect nevermaterialized with the exception of the Project Pluto enginetesting program [79]. Although nuclear-powered ramjet grounddemonstrators were built and successfully tested during ProjectPluto, the environmental concerns outweighed their strategicadvantage. Additionally, the success of controlled ablationreduced the need for further research into EMFC for use on re-entry capsules [80]. Arc jets were then applied to ground testingsystems with many integrated into wind tunnels as a source ofhigh enthalpy, high velocity ow [6]. Seeding is still a viablemethod for increasing the performance of those wind tunnels, butit can lead to undesirable contamination of the ow. Simmonset al. concluded that discrepancies in the air chemistry caused byseeded MHD accelerator concepts for the MARIAH hypersonicwind tunnel made non-MHD options more appealing [81]. Thelack of post-Apollo funding and interest in on-board nuclearpower effectively halted the prospects of MHD systems foraerospace vehicles. Non-nuclear power generation became thenew focus of MHD research [82,83]. Despite the fact that electricpropulsion engines make use of a comparatively weaker force,concepts [84] developed simultaneously with those of MHDpropulsion and eventually ourished with help of a low powerrequirement capable of matching with radioisotope thermo-electric generator technology then emerging [85].

The past decade has certainly seen a reemergence of MHDresearch applied to aerodynamics. The history above shows thatgenerating and controlling a ow with s4100

A

=m is difcultbecause of the power requirements. Creating ionization from athermal source such as a plasma jet is not desirable, and is notpossible for aerodynamic control surfaces. Fortunately, ionizationcan also be achieved through high voltage elds, laser beams,microwaves, and radiation (any method of transferring energy tocause molecular excitation of the gas). Of the EMFC facilitiesmentioned thus far, all have at minimum employed high voltageelds. However, a difference exists between each on how the highvoltage elds are applied.

The easiest method of creating plasma is to apply an electriceld with a large potential difference between two electrodes.Based on factors like separation distance, potential difference,geometry, and the gap medium, plasma will develop between theelectrodes. The current is based on the effective resistance of thegap. When plasma lls the gap, the resistance is immediately andsignicantly lowered. Normally, the end result is an arc dischargewhere the gap becomes a short circuit and draws maximumpower from the potential source. Because of their difculty tocontrol and destructive nature, arc discharges are most often seenas detrimental for engineering applications. Thus for aerodynamiccontrol, ionization with electric elds has focused on producingdischarges of a more diffuse nature, known as glow or coronadischarges. However, Zaidi et al. have in fact used a plasmacolumn control concept operating in a constant current, variablevoltage mode [38]. The power supply potential was 20kVDC, andthe maximum current was 1.0A. Fixing the current and activatingthe power supply causes a high voltage eld to be applied untilelectrical breakdown occurs and a plasma column forms. Once thecolumn forms, the voltage required to maintain the set point

pace Sciences 45 (2009) 3049 39current can be very low. When a plasma column forms betweenthe path of least resistance where the tapered electrodes areclosest to each other, it travels downstream where the gap

between the electrodes gradually increases, visually similar to aJacobs ladder. The boundary layer control is provided by themomentum transfer from the plasma column propagatingdownstream.

During the wind tunnel experiments, the plasma column thatforms between the electrodes was found to be periodic with afrequency of 110kHz depending on the current and magneticeld strength [86]. While applying a 1.7 kV eld at 35mA withno magnetic eld, the plasma column travels downstream at360m/s. When a magnetic eld of B 2:0T is applied, the columnspeed increased to 2000m/s [87] to generate the control resultspreviously discussed. Thus the electric and magnetic eldscombine to allow for greater momentum addition to the ow. Itappears that the performance of this facility can be increasedsimply by raising the set point current. Unfortunately, Jouleheating has worked against the plasma column in the case of

DC discharge ionization system allows for energy addition in theboundary layer of high-speed air ow. As a result, the energyaddition allows for an EMFC actuator to signicantly affect thesurface pressure distribution. Direct current discharges do noteasily remain diffuse as pressure rises [89], but this system shouldbe operational for high-altitude ight conditions.

Research has shown that the degree of ionization produced byan electric eld may be higher for pulsed discharges rather thanfor a steady DC discharge of similar potential difference.Discharges with periodic high voltage pulses simply can with-stand a higher applied electric eld before a transition to arcingoccurs, so long as they are sufciently short enough to sustainstreamers before transitioning to sparks [90]. This property allowsmore power to be transmitted during the pulse, which willincrease s. Very short duration, low duty cycle pulses canmaintain a conductive path between electrodes as long as thefrequency is additionally high enough to counteract the WIGdecay. However, the applied Lorentz force should be continuousand therefore generated with a DC power supply. Palm et al. [12]

ARTICLE IN PRESS

E.M. Braun et al. / Progress in Aerospace Sciences 45 (2009) 304940moving a shock wave-induced separation bubble downstream andwill begin to suppress results while current is increased beyond acertain point [39]. To conduct further studies with higher plasmacolumn power, an assembly consisting of a sapphire base plateand high-temperature arc corrosion-resistant electrodes wasconstructed [87].

The DC ionization system rst presented by Shang et al. [41] in2002 was part of an EMFC actuator designed to affect the shockwave structure around a blunt body. A diffuse WIG is createdinstead of a plasma column due to a lower applied potentialdifference and a lower static pressure. The blunt body study wasfollowed by a at plate model constructed of a ceramic base andtwo embedded copper electrodes [42]. Fig. 13 shows theseelectrodes, with the upstream cathode experiencing a moreintense glow [44]. This diffuse glow discharge begins to constrictand transition to an arc when the current surpasses 100mA orwhen B is greater than 0.2 T [42]. In addition to using only a DCdischarge, pulsing the discharge with a frequency between 5Hzand 10kHz was also used to explore the response of the Mach 5ow to plasma actuation with a magnetic eld present [88].Volumetric heating of the air by the plasma was found to occurfaster than the 3ms response time of the pitot probe. In anothercase, RF radiation was added to augment the ionization created bythe DC glow discharge, resulting in a reduction in the impedanceacross the electrode gap [40]. Accounting for all of the ionizationmethods, the maximum power requirement remained a fewkilowatts or less and can result in a conductivity of a few

A

=m inthe static pressure environment of 0.67 torr.

The research presented in the various publications for thisfacility certainly shows that a low-temperature WIG created by a

Fig. 13. A DC voltage discharge between two electrodes at freestream conditions of

M 5:15 and p 0:59 torr. The applied voltage is 880920V at a current of 50mA.The addition of a magnetic eld signicantly affects the plasma and creates a

virtual hypersonic leading edge strake (from [44]).addressed these issues by creating an EMFC channel ow facilitywith RF WIG generation and simultaneous DC Lorentz forceapplication using the electrode conguration previously dis-cussed. The facility generated a diffuse WIG for Mach 24 oworiginally by using a 13.56MHz, 600W RF power supply to createthe conductive path for the Lorentz force energy addition [30].Conductivity (0:121

A

=m) scales with the power draw of thesystem. Since that time, a more complex ionization system hasbeen constructed to raise the attainable level of conductivitywithout raising the power draw. Nishihara et al. reported usingthis system to attain s 0:1 A=m in a Mach 3 ow bycompressing a 500V, 1ms pulse into a high frequency (up to50kHz), high peak voltage (20 kV), short duration pulse (1020ns)[29]. During the peak voltage application, the current may reach100A, but the short duty cycle results in a reasonable overallpower consumption of, for instance, 4080W in Ref. [29]. Theionization power requirement is therefore much less than what isused to apply the Lorentz force. According to Ref. [34], raising thefrequency of the system from 40 to 50kHz increases the owconductivity along with lowering the ballast resistance. Thefrequencies in this range match reasonably well with the WIGdecay and provide fairly steady ow conductivity.

The life of the WIG can be observed by measuring the currentdraw from the DC Lorentz force power supply. Fig. 14 shows fourFig. 14. Voltage oscillogram for 40 kHz pulsed ionization of a Mach 3 nitrogenow. The test section pressure is 8.4 torr and B 1:5T (from [36]).

pulses measured from the ionization system operating at 40 kHz.Fig. 15 shows two current oscillograms measured from the DCLorentz force circuit for the same conditions as Fig. 14, createdwith constant electrode potentials of 2 kV (one at each polarity).The current rises at the initiation of each ionization pulse, andthen falls with the WIG decay. With slightly different test sectionconditions, a current oscillogram appears in Ref. [37] for anionization frequency of 100 kHz, which results in a more constantLorentz force over time because the WIG decays less. As one cansee from Fig. 15, the Lorentz force system has an average powerconsumption of about 2 kW and it is capable of creating thepressure changes shown in Fig. 6.

3.5. Overall feasibility

A few examples of recent EMFC facilities have been presentedin this section. These facilities have emphasized the use ofelectromagnetic forces within the boundary layer and they have

shown that EMFC can have a considerable effect at supersonic andhypersonic ow speeds. However, the pressures at which theseexperiments have been conducted, as well as the magnitudes of sand B, are far below what may be necessary for a hypotheticalAJAX scramjet engine. Consequently, these systems are moreapplicable for boundary layer control of an aerodynamic surfaceor inlet system. Several steps must be taken to transition to afeasible electromagnetic virtual control surface. Experimentalfacilities have demonstrated success using low pressure core owsoften surrounded by large magnets, and it is time to considermore compact congurations that can simulate environments likeexternal ow over a wing or the beginning of an inlet compressionsystem. In these environments, the magnets can be embeddedbelow the surface and between the electrodes. Novel coolingmethods must be developed for permanent magnets to survivethe high-temperature environment. EMFC actuators may beplaced in regions where the surface is actively cooled or isrelatively cool. Considering the maximum operating temperatureof rare-earth magnets, cooling to a temperature as low as 350Kmay be necessary. Test section pressures must be increased, notnecessarily to atmospheric, but perhaps to simulate the pressure

ARTICLE IN PRESS

Fig. 15. Current oscillogram for Lorentz force power supply with the testconditions of Fig. 14 and different 2 kV electrode polarities (from [36]).

M

N

El

N

[12] 2002 4 0.050.1 210 0.45 N

El

N

El

El

El

El

El

El

El

El

Su

El

El

El

Su

E.M. Braun et al. / Progress in Aerospace Sciences 45 (2009) 3049 41[32] 2003 3 0.020.12 1215 01.5

[45] 2003 Low-speed 3 760 0.4[33] 2004 3.4 0.080.18 720 01.75

[43] 2005 5.1 1 0.6 0.10.2[34] 2005 3.4 0.140.23 720 01.5

[88] 2005 5.1 1 0.8 0.2[40] 2005 5 2 0.67 01

[23] 2005 3 0.1 720 01.5[28] 2005 3 0.07 720 01.5[35] 2006 3 0.1 720 01.5[38] 2006 2.8 N/A 28 02

[36] 2006 3 0.1 720 01.5[42] 2006 5.3 0.06 0.6 01[29] 2006 3 0.1 720 01.5[86,87] 2007 2.8 N/A 28 04.5Table 2EMFC experimental environment summary.

Ref. Year Mach num. s A=m p (torr) B (T)

[22] 1995 Low-speed 3.2 1300 0.4[41] 2002 5.8 12.5 1.22 02

[30] 2002 4 0.01 68 0.4[37] 2007 4 0.1 4.8 01.63 El[39] 2008 2.6 N/A 28 04.5 Su

[44] 2008 5.15 1 0.6 00.2 Elafter a shock over a thin wedge. The Mach 5 EMFC facility testsection pressure reported by Shang et al. is meant to simulate analtitude from 30 to 50km (0.67 torr) [40], but accounting for areal ight vehicle with a leading shock leads to much higherpressures in that altitude range (i.e., a 101 wedge at that speedwould lead to a static pressure of 220 torr).

Although it appears challenging, experimental generation of aLorentz body force is not particularly difcult for EMFC actuators.The key problem is creating non-thermal ionization to supply aconductive working uid for the actuator [24]. Under rareedconditions, DC ionization systems are capable of creating a diffuseWIG for which s can reach a few

A

=m at high speeds. As pressurerises, the voltage required to sustain a glow discharge rises. Thisrelationship makes arcing with DC discharges more probable forEMFC as charge builds up on the electrodes. High voltage, high-frequency pulsed ionization sources are another available methodfor creating the same value of conductivity. These discharges withhigh energy transfer during low duty cycle pulses can be appliedin systems to produce non-thermal ionization, and s is sustainedby using a frequency fast enough to counteract the WIG decay.

agnet type Ionization source Medium Conguration

dFeB N/A Salt water Flat plate

ectromagnet High voltage DC Air Blunt body

dFeB High voltage RF He, N2 Channel

dFeB High voltage RF He Channel

ectromagnet High voltage RF N2 Channel

dFeB N/A Salt water Flat plate

ectromagnet High voltage RF N2, air Channel

ectromagnet High voltage DC Air Flat plate

ectromagnet High voltage RF N2, air Channel

ectromagnet HV, 10Hz-RF Air Flat plate

ectromagnet, NdFeB HV DC, RF rad. Air Flat plate

ectromagnet High voltage RF N2, He Channel

ectromagnet High voltage RF N2, He, air Channel

ectromagnet High voltage RF N2, air Channel

perconducting Plasma column Air Channel

ectromagnet HV RF, seeding N2, air Channel

ectromagnet High voltage DC Air Flat plate

ectromagnet High voltage RF N2, air Channel

perconducting Plasma column Air Wedgeectromagnet High voltage RF N2 Channel

perconducting Plasma column Air Wedge

ectromagnet High voltage DC Air Flat plate, inlet

to ow control by electromagnetic elds, it must be noted that

properties of a glow discharge may be used to improve the off-design performance of a high-speed inlet compression system bymanipulating the shock wave structure. These features of glowdischarges are signicant for the future development of re-entryvehicles and hypersonic airbreathing propulsion. Although someof the early shock tube literature makes a case for electrohy-drodynamic effects as the reason behind some of the shock wavealteration [9597], the general consensus is that most of theeffects seen are a product of the heating from the plasma [92].Computational studies also indicate this result [98].

If the bulk of the glow discharge control effect is from heating,then the next logical step in the process of estimating itsfeasibility is to determine if there are benets of plasma asopposed to other heating sources. One benet of heating by glowdischarges when compared to a typical heating element is rapidactuation. This may be a large enough benet to continueexperimentation with surface glow discharges for aerodynamiccontrol. For instance, Shin et al. measured a glow dischargeactuation time of less than 220ms using pin electrodes on a atplate in a Mach 2.85 ow environment [99]. This at plate plasmaactuator is capable of creating a weak shock wave over theactuator when the plasma is diffuse. A more constricted plasmaformation in that environment, although produced with higherpower, does not have the same shock wave control effect. Thedifference between plasma heating and surface resistanceheating is noticeable, whereby the plasma has more of avolumetric effect and is not exclusively characterized as typical

ARTICLE IN PRESS

Fig. 16. Split image of a bow shock around a sphere with and without plasma for aow velocity on the order of 1600m/s (from [93]).

erosconsiderable interest for ow control with only plasma or electricelds has developed. Techniques for aerodynamic ow control byelectric elds can be categorized into glow discharges and DBDs,covered in the next two sections, respectively. A glow discharge isformed across a gap of air or another gas between two electrodeswith a difference in electric potential. The presence of a glowdischarge is based on factors such as electrode geometry, ambientpressure, the gap medium, and the voltage. The glow dischargeessentially means that the gap is lled with free radicals andelectrons traveling between the electrodes. As such, the currentincreases rapidly after initial formation. Increasing the voltageafter the glow discharge is formed eventually leads to electricalbreakdown and arcing. A diffuse discharge is desirable since itindicates that the WIG effects will be uniform throughout theglow region. Often, experimentation with this phenomenon hasoccurred in low pressure environments where it is easier to createa diffuse discharge with a relatively low voltage. High pressureglows are also possible, and applying the potential difference withan increasing frequency has shown that the same current ismaintained with a lower voltage [91].

Bletzinger et al. have provided a review of plasmas related tohigh-speed aerodynamics, containing a short history of thedevelopment of experimentation with glow discharges in recentdecades [92]. In summary, initial shock tube experiments wereconducted by measuring the drag and shock wave structure ofobjects (often blunt) while recording the differences with andwithout the actuation of a plasma source. As shown in Fig. 16,plasma ow as opposed to typical ow can drastically change thestandoff distance of a bow shock around a blunt body [93]. SimilarIonization sources for EMFC actuators should trend towards usingRF or square wave signals, as long as the packaging of the pulsingcircuit elements does not lead to size and weight requirementsmuch greater than a DC system. Fridman et al. postulated thatvoltage pulse durations less than 100ns/cm of anode and cathodeseparation can sustain streamers without transformation into arcs[90]. Rening that estimate and determining the plasma decayrate between pulses will allow for the optimization of theionization source and will minimize uctuations to the owconductivity. The constant development of power semiconductorsshould make high-frequency pulsing systems smaller and morecost effective. Separate ionization and Lorentz force powersupplies can be combined over the same at plate electrodeswith the use of rectiers or diodes. The largest difference inpressure is obtained when the Lorentz force and Joule heatingeffectively work together. Since Joule heating thickens theboundary layer and decreases the local speed of sound, a Lorentzforce applied to retard the freestream ow or direct it off of thesurface has the greatest effect.

With these issues properly addressed, EMFC could potentiallybe used in place of traditional control surfaces at high altitudes. Itshould not be questioned if the Lorentz force is powerful enoughto provide high-speed aerodynamic control. The main issue is thedetermination of whether or not the ionization and Lorentz forcepower requirements of an actuator result in a system compactenough to be implemented into a ight vehicle. Table 2 provides asummary of literature in this section with experimental environ-ments highlighted.

4. Flow control by glow discharge

Although much of the previous discussion has been dedicated

E.M. Braun et al. / Progress in A42results were demonstrated as early as 1959 by Ziemer [94]. Thechange in shock wave geometry is important because a change instandoff distance can reduce heating and drag. Furthermore, thepace Sciences 45 (2009) 3049surface heating [100].Flat plate experimentation with glow discharge plasma

actuators for aerodynamic control has yielded promising results.

Concerning the use of plasma ow control systems as part of

ARTICLE IN PRESS

erosExperimental results from Ref. [100] show the plasma and surfaceheater both cause a 50% change in pitot pressure in Mach 5 owover the cathode, with the glow discharge heater acting an orderof magnitude faster. As was previously mentioned, the EMFCactuator in Ref. [43] produced an 18% change in lift while usingonly a DC glow discharge. Many other studies show similarresults. Although efciency improvements have been made, theperformance of these plasma control systems is still uncertainacross wide ranges of ight speeds and vehicle congurations[92]. If indeed the plasma effect is thermal and increases the localspeed of sound, than it can intuitively be expected that the effectwill be lessened as speeds increase and the post-shock airtemperature and enthalpy increase. Hence, a at plate actuatorimmersed in a low-temperature wind tunnel ow is a best casescenario for demonstrating glow discharge control. The power ofthe plasma source can be raised in order to compensate for a owof higher enthalpy, but the efciency of control may be drasticallyreduced. Experimental studies using plasma discharges withwedges and blunt bodies are discussed next to elaborate on thisissue.

Although the presence of plasma can change the structure of ashock wave, it appears that most literature involving inletcompression systems also contains magnetic elds for the fullLorentz force effect. A detailed literature review appears in Ref.[92]. Some research with only plasma has been reported. For aMach number of two and a ow mixture of nitrogen and helium, aglow discharge yielded a signicant change in the oblique shockangle over a wedge [101]. The change in the shock angle indicatesa change in the Mach number from 2 to 1.8 due to the plasmaheating. The WIG source was located on the walls of the windtunnel. Since the initial ramps of an inlet compression system areoften not surrounded by an outer wall (e.g., X-43A design), itwould be benecial to test if this effect could be duplicated with aWIG source located entirely on the surface of the wedge. Placing adiffuse plasma source on the tip of a wedge and creating an effecton the oblique shock angle is a logical direction to move in todetermine if these systems can be placed on a vehicle. Such adesign was recently attempted by Gnemmi et al., where plasmadischarges on a conical tip of a projectile were used to disturb ashock wave at freestream conditions of Mach 4.57 and 54kPastatic pressure [102]. More research is needed to determine if suchsystems can produce an appreciable aerodynamic control force,but it appears that ramp and inlet concepts of low to moderateturning angles are just as viable as at plate concepts. Closelyrelated to that design is the concept of a virtual cowl that can becreated by plasma heating [103,104]. The plasma source is morelikely to be high-energy electron beams or microwaves ratherthan glow discharges. The heat addition specically can alter theupstream oweld in order to reduce the inlet spillage. Thisconcept will require a considerable amount of power to operate,but one must consider that any system with plasma heating maybe used only during a (presumably short) transition process byacceleration to the design Mach number. Although most of thesesystem designs are analytical models, some preliminary experi-mental studies have demonstrated the concept [105].

Concerning blunt bodies, many studies focusing on dragreduction have appeared in the literature. One initial studyshowed that the drag coefcient for a sphere in the presence ofa WIG was signicantly reduced for subsonic ow [106]. The sameexperiment for supersonic speeds showed that the drag coef-cient was higher using a WIG than with typical airow, attributedto an increase in the pressure integral on the front of the model atcertain conditions. Other plasma sources constructed for drag

E.M. Braun et al. / Progress in Areduction have proven to be more effective since then. Ganievet al. reported a reduction in the drag coefcient of about 50%from Mach 0.59 to 4 using a plasma jet placed at the tip of arealizable ight vehicles, some appear more feasible than others.In order to overcome problems including but not limited to powerconsumption, scaling, and hypersonic interaction at true ightconditions, the next step for plasma control for aerodynamics is atransition into realistic systems. It is understandable that some ofthe models of full-scale hypersonic systems have not beenconstructed due to the cost, but plasma control needs to be betterproven experimentally as part of more ight-ready systemsinstead of basic shapes. Manipulating the bow shock wave aroundblunt bodies with plasma has been experimented with for 50years, but no concrete applications are yet practical. It appearsthat ground testing of aerodynamic surfaces and inlet systems ismoving forward, with the rapid plasma heating effect showingpromise for control applications. The main challenge is producingsystems that make use of current technology while maintainingpower and packaging considerations.

5. Flow control by DBD

Considering the physics involved, a DBD is similar to a glowdischarge. Where a glow discharge has an air gap, a DBD containsa gap of dielectric material between the anode and cathode.Typical materials like glass, polymers, and ceramics have a muchhigher resistivity than air, allowing for the electrodes to be placedcloser to one another. Closer placement increases the electric eldaround the electrodes and ultimately raises the Coulomb force inEq. (10) without the occurrence of electrical breakdown. Thedielectric barrier is self-limiting as it prevents charge accumula-tion over the barrier material to prevent arcing. DBDs have beenrecognized since the mid-19th century, with the their rstapplication being the production of ozone [111]. Since that time,research has continued to grow and now applications includesurface treatment, reduction of pollutants, lasers, and plasmadisplay panels. Systems using glow discharges often use lowpressure, but these discharges were stabilized across the barrier atatmospheric pressure beginning in the 1980s [112].

DBDs constructed specically for aerodynamic ow controlapplications appeared in the literature near the end of the 1990s[113,114]. In the decade since those reports, research intosomewhat blunt body [107]. The reduction in the drag coefcientwas found to depend on the stagnation temperature of thecounterow jet. Plasma jets appear to be inefcient for stream-lined shapes [108]. At the time of Ref. [107], other publicationsalso described drag reduction with plasma jets and other forms offocused energy addition. A thorough list of these can be found inRef. [109]. However, the large drag reduction by the plasma jetinjection appears to be more directly related to the counterowjet instead of the thermal effects of the plasma. Fomin et al.experimentally determined that uid dynamics instead of plasmais the dominant effect using the jet for moderate supersonic Machnumbers [110]. Those experiments were conducted using atruncated cone cylinder at Mach 2, 2.5, and 4. As was discussed,the use of plasma jets was eventually deemed unrealistic for MHDight applications in the 1960s because of the power requirement.Many of these current systems have been met with enthusiasm,but again scaling the power requirements to ight vehicles ormissiles may pose insurmountable problems with currenttechnology. Although new publications continue to emerge withdifferent plasma sources and test geometries, very little of it ispredominantly different from what was carried out at thebeginning of this decade.

pace Sciences 45 (2009) 3049 43aerodynamic ow control with DBDs has rapidly increased bothexperimentally and computationally. A number of reviews havebeen written [51,115117], which probably indicates a variety of

ARTICLE IN PRESS

erosE.M. Braun et al. / Progress in A44opinions on their applicability. Most conclude the DBD controleffect is applicable for low Reynolds number ows below thegeneral aviation range, and improvements to their strength willhave to be made for them to be applied to ight aerodynamics. Atlow speeds, DBD actuators have a signicant effect on boundarylayer ow. Fig. 17 shows a notable image by Roth et al. of owreattachment made possible by an array of DBD actuators [118].This actuator systemworks at atmospheric pressure, and has beennamed the One Atmosphere Uniform Glow Discharge Plasma(OAUGDPTM). The ionization is created with a high voltage RFsignal and the barrier material is Kapton. An RF signal is usedrather than a DC signal because it creates a cycle of chargeexchange between the electrodes that increases the control effect.The waveform shape and frequency along with the dielectricmaterial choice may be optimized to some extent, with manydifferent congurations reported. Several studies with the systemin Fig. 17 have resulted in successfully increasing or decreasingdrag on a at surface, adding momentum to the boundary layerow, reducing the boundary layer thickness, and inducing a ow(also known as the ionic wind) of up to 6m/s [119]. Fig. 17 raisesthe immediate question about the ability to apply a DBD system tohigh-speed ow where ow reattachment, drag reduction, andturbulence suppression are all major concerns. Although DBDactuators are studied by several institutions, the spanwiseelectrode geometry is always fairly similar and is depicted in

Fig. 17. Smoke visualization shows ow reattachment on a NACA 0015 airfoil at a121 angle of attack by an array of EFC actuators. The freestream ow speed is 2.6m/s (from [118]).Fig. 18 [120]. This linear arrangement also can be modied into anannular jet source (known as a plasma synthetic jet actuator),where pulsed operation can generate vortex rings [121]. Themaximum jet velocity for that study was on the order of 1m/s.The electrodes also may be wrapped around the internal diameterof an axisymmetric jet. Benard et al. demonstrated that thisconguration may be used for jet mixing enhancement, whereexperiments increased mixing in a ow up to 30m/s on a modelwith an exhaust diameter of 72.5mm [122]. No modication wasseen at a jet speed of 40m/s. The self-limiting DBD allowed about10W of power to be transferred into the ow, and it was notedthat this value must be increased for the DBD to have more effecton the jet ow.

The relative strength of current systems can be compared bytheir ability to induce a certain ow speed of air passing over theactuator. The ion wind speed measured in most recent surfaceDBD actuators is only a few meters per second, and efcientcontrol results are obtained when u1 is less than 30m/s [115].However, some experiments have been conducted using higher

Fig. 18. Typical spanwise cross-section geometry of a dielectric barrier dischargeactuator for aerodynamics applications (from [120]).

pace Sciences 45 (2009) 3049freestream speeds. Opaits et al. [123] investigated DBD control ofa NACA 0015 airfoil with freestream speeds of 2075m/s atatmospheric pressure. The stall angle was raised with the DBDactuators at u1 75m=s, and a change in pressure distributionwas also recorded. Similarly, Roupassov et al. [124] measuredchanges in the pressure distribution for a NACA 0015 airfoil atspeeds up to 110m/s. In this case, the electrodes were placedparallel to the ow, and it appears that the pressure distributionincurs a greater change with the DBD actuator when the airfoil isclose it its stall angle. One attempt was made recently to mount aDBD actuator on the leading edge of the wing of a Jantar StandardSZD-48-3 sailplane [125]. It appears that the DBD systemwas ableto affect the separation and lift characteristics of the wing surface,but the data collected were not particularly reliable and renedtests are needed. A study by Corke et al. stated a DBD actuator wasable to excite three-dimensional boundary layer instabilities on asharp cone at Mach 3.5, but no similar results have appeared since[126]. For current DBD actuators, it is clear that a speed of 30m/srepresents the freestream ow limit in which a noticeable controleffect is demonstrated. For freestream speeds over 30m/s, DBDactuators may slightly delay stall, assuming they are placed on thelocation where the ow will normally begin to separate.

In order to maximize DBD actuator performance for high-speed ow control, one may initially assume that the anode andcathode should have minimal size and be placed as close as

thermal ionization. However, the radiation makes their imple-

ARTICLE IN PRESS

erospossible to each other and separated by a very thin layer ofdielectric material. This geometry would maximize the electriceld, where the Coulomb force grow