novel passive and active flow control for high lift - icas is

NOVEL PASSIVE AND ACTIVE FLOW CONTROL FOR HIGHLIFT

L.L.M. Veldhuis∗ , D.P. Jansen∗ , J. El Haddar∗ , G. Correale∗∗Delft University of Technology

Keywords: High Lift, Flow Control

Abstract

In this paper we discuss the separation controlcapabilities of various passive and active tech-niques. The goal of this research is to furtherquantify the possibilities of enhanced high liftsystems for a multi-element wing that consists ofa main wing plus fowler flap. To study the effectof the control techniques, experimental investi-gations where performed in different low speedwindtunnels on a single element airfoil and wing-flap model. The results show that a lift enhance-ment of 18% can be obtained on single elementairfoils using nano-pulsed plasma actuators. Anattempt was made to enhance the high lift perfor-mance on a wing flap model through increasedflap deflection combined with separation controlby applying several passive and active techniqueson the flap leading edge. It was shown that the ap-plication of a circular cylinder in front of the flapleads to a noticeable increase in the lift coeffi-cient in cases where the flap boundary layer tendsto exhibit trailing edge stall. The other appliedcontrol techniques had a limited effect which waseither due to limited momentum input or the oc-currence of wake bursting.

1 Introduction

Over the past decades several innovative tech-niques to control flow separation over airfoilshave been proposed. In recent years especiallythe research on active flow control methods hasreceived interest. The application of active blow-ing, suction and the utilization of zero mass trans-

fer synthetic jets (SJA) [1, 2, 3, 4] have shown tobe capable of postponing the separation to someextend. However, the practical implementationof these active methods, in general, leads to anunwanted increase in mass, degradation of thewing’s structural integrity (SJA) and increasedpower demands to drive pumps (suction or blow-ing) or excite piezo-electrical systems and asso-ciated hardware.

Although attractive from the point of furtherimprovement opportunities, many active systemshave not matured to be incorporated in moderntransport aircraft currently developed. This ispartly due to the need to adapt the wing or flapinternal structure for most of these systems. Inthis respect it is attractive to revisit the possiblepassive control methods and compare them withactive methods that require only a limited adapta-tion of the wings structure, enabling applicationof the particular technique as a retrofit.

At Delft University a research program is un-dertaken which aims at developing novel passiveand active control techniques to enhance the highlift capabilities of wings. Especially the effect ofmodern techniques like plasma actuators is inves-tigated. In this paper the experimental researchon passive on and off-surface control elements aswell as active techniques in the form of Nano-pulsed actuators will be addressed. The goal ofthis part of the project is to get further insightin the possibilities to employ extended (Fowler)flap deflection which under normal, uncontrolled,conditions would lead to serious flow separationover the flap surface. This separation is preventedby applying three different control techniques:

1

L.L.M. VELDHUIS, D. JANSEN, J. EL HADDAR & G. CORREALE

Fig. 1 Flow field around a circular cylinder withrotation. The circulation strength is chosen suchthat the 2 stagnation points coincide at the lowerside.

plasma actuators, drooped spoiler and periodicflow excitation based on off-surface positionedcylinders.

2 Background on high lift enhancement

To better understand how the parameters deter-mine the high lift and flow separation character-istics of the high lift system, it is necessary toaccurately identify the interaction effects of mul-tiple elements. A very good overview is givenby Smith [9] and later by van Dam [10] and thereader is referred to these references for a com-plete overview of high lift aerodynamics in gen-eral. An interesting point that is related to thecurrent research is question as to what maximumlift may be obtained.

When comparing the lifting capabilities of asystem equipped with slat and flap (either singleof multiple flaps) one may argue (see [9]) that thetheoretical lift coefficient that can be obtained isthe one that is found for inviscid flow over a cir-cular cylinder airfoil with both stagnation pointsat the lower side (fig. 1). The lift coefficient is inthis case

Cl = 4π

This is of course an extreme case which is farfrom the practical layout of a wing with high liftelements deployed. However, a circular arc air-foil may be quite close to the optimum flow fieldthat is found for a real high lift model (fig. 2). Inthis case the flow is extremely curved and the (in-viscid) theoretical maximum lift coefficient will

(a)

(b)

Fig. 2 Typical flow pattern over a three elementhigh lift airfoil (a) and a circular arc airfoil (b) ininviscid flow. The streamline pattern shows closeresemblance between the two.

beClmax = 4π/

√2≈ 8.8

Even this value is way beyond the typical val-ues that are found for modern transport aircraft.In fact one of the main limitations for reachingvalues close to this inviscid value is the flow sep-aration on either the main wing and/or the flap inreal viscous flows. Hence, in case the maximumlift coefficient for a given layout is to be maxi-mized the occurrence of flow separation needs tobe minimized. If such an approach would be suc-cessful, the high lift system would benefit fromvery large flap deflections. In this research flapdeflections angle larger than 40o are discussedin conjunction with possible new passive and ac-tive flow control techniques. As the current CFD(RANS) approaches have difficulty in accuratelypredicting flow with severe separation, this re-search is performed through windtunnel experi-ments.

3 Experimental setup and test approach

3.1 Windtunnel and models

3.1.1 Windtunnels

Experimental investigations were performed ina 2 different low speed windtunnels: the M/W-tunnel (Model 1) and the Low Turbulence Tun-

2

Novel passive and active flow control for high lift

NACA 63618

c = 200 mm

b = 400 mm

400 mm

400 mm

V

M\W-tunnel

Fig. 3 Model 1: Straight NACA 63618 wingmodel dimensions and geometry.

nel (LTT) (Model 2). Both windtunnels are lo-cated in the Aerodynamics Laboratory of the Fac-ulty of Aerospace Engineering of Delft Univer-sity of Technology. The M/W-tunnel consists ofan open circuit which produces a rectangular jetof 400× 400mm. The turbulence level in thiswindtunnel is typically in the order of 0.1%.

The LTT is closed circuit windtunnel withan octagonal test section of 1.80m × 1.20m.The maximum speed in this facility is Vmax =120m/s, while the turbulence level is very low,in the order of 0.01%− 0.2%. The pressurerecording in this facility is done with a 200 portelectronic transducer (PSI-DMT, “Initium sys-tem”). Boundary layer transition is recorded withan electronically amplified stethoscope. Further-more, wake and boundary layer velocity profileswere measured with a dedicated BL-probe.

3.1.2 Model 1, Single element airfoil

The first wing model that was investigated isbased on a span of b= 400mm and constant chordof c = 150mm having a NACA-63618 airfoil (fig.3). The model is positioned between walls toobtain 2-dimensional flow. This setup was cho-sen to check the effect of Nano-pulsed actuatorcontrol [23] of leading edge separation at differ-ent wind speeds (Reynolds numbers) in a rangethat can not be attained with standard Di-electricbarrier discharge (DBD) actuators [22, 15]. Themain aerodynamic coefficients (lift, drag andpitching moment) were measured with an exter-nal 6-component balance.

NLF-22 mod-B

0.5 1.0

600 mm

(a) Airfoil geometry.

ESWIRP Workshop, Cologne, 2010

Flow Separation Control Research at TUD

Traversing wake rake

BL suction

NLF 22 mod (Extra 400)

600 mm

Fowler Flap

Vmax=120 m/s Tu=0.01-0.02%

(b) Wing flap model (Model 2) in Low Speedwindtunnel.

Fig. 4 Model 2: NLF-22-mod-B wing modelwith extendable flap (X400).

3.1.3 Model 2, Multi-element airfoil

The second model that was investigated consistsof wing-flap model mounted vertically and span-ning the windtunnel over the full height (fig. 4).This modified NLF22 airfoil is in fact used onthe Extra-400 aircraft1. It was used in an earliertest campaign in which the flap overlap and gapwere optimized[7]. The flap could be deflectedover a range from δ f = 0o to 60o to produce ei-ther attached, partly separated or fully separatedflow over the flap upper surface. Especially thelarger flap deflections are investigated with re-spect to the control capabilities of passive andactive techniques. The basic characteristics wereobtained through surface pressure and wake rakemeasurements. To prevent unwanted flow sep-aration on the windtunnel upper and lower wall(wall-model juncture) the boundary layer was re-moved through suction holes connected to a highcapacity pump.

3.2 Flow control devices

1German built 6 seater, all composite turboprop aircraft

3


3.2.1 Plasma Generators

Standard DBD actuators The potential ofplasma actuators for flow control has been rec-ognized by a number of researchers [17, 18, 19,21, 22, 14, 15, 13, 12]. Most of the existingwork has focused on the development of stan-dard dielectric barrier discharge (SDBD) actua-tors. Their working principle is based on apply-ing a high voltage (5-30 kV) signal to two stag-gered electrodes separated by dielectric materialin order to create a biased flow of ions and elec-trons which adds momentum to the main flow(see fig. 5a). Using this technique, near-walljets on the order of 1-2 mm thick have been cre-ated, with jet speeds approaching 8m/s. Thesehave been used to successfully control low-speedseparated flows The effectiveness of SDBD ac-tuators has been found to be sensitive to a num-ber of parameters. It has shown that relativelythick dielectrics tend to produce larger changesin momentum, while Enlo et al [16] have shownthat when using non-conventional thin wire elec-trodes, decreasing the wire diameter increases thechange of momentum at a given input power. Theform of the input signal has also been shown tobe significant. More recently Kotsonis [17] hasshown that much larger changes in momentumcan be obtained using combinations of sinusoidalexcitation and signals with steep rises and falls.

However, one of the limitations of SDBDactuators is their rather low momentum inputwhich makes them suited for low Reynolds num-ber flows only [12]. Typical Reynolds numbersat which a beneficial separation suppressing ef-fect was found are in the range Re = 20,000−50,000, which is too low for general applicationson aircraft.

Nano-pulsed actuators Over the last threeyears a new class of plasma flow actuators hasbeen developed which work on a completelydifferent principle [23]. These so-called nano-pulse dielectric barrier discharge (NPDBD) actu-ators have geometries similar their SDBD coun-terparts, but use a high-energy (20-50 mJ) short-duration (10-100 nsec) pulses to produce local-

ized heating in the order of 200−300K (fig, 5b).This process delivers very little direct momen-tum input, but generates shock waves which radi-ate from the actuator that interact with the outerflow [24, 20]. Recent experiments carried outby the TUD Aerodynamics group have demon-strated that NPDBD actuators can be used to re-attach airfoil flows with leading-edge separation,leading to increases in maximum lift of up to 20%at wind speeds up to 40 m/s [20]. This may bethe result of a number of mechanisms, includingthe periodic forcing of laminar-turbulent transi-tion, shock-shear layer interactions, or the cre-ation of large vortical structures. The latter havebeen observed in Schlieren imaging [20]. How-ever, the parameters that influence the effective-ness of NPDBD separation control have yet tobe fully explored. Furthermore, the basic mecha-nisms of the flow phenomena that occur is a pointof continued research. Nevertheless, since bene-ficial results were obtained over a stalled NACA-63618 airfoil is decided to apply the NPDBD ac-tuator on the flap of Model 2 as well.

3.2.2 Cylinder

In the selection of a flow control device the struc-tural simplicity of application as well as the min-imization of power usage is crucial. For this rea-son a novel passive technique was tried in whicha cylinder is placed directly upstream of the flapleading edge of Model 2. The beneficial effectof such a device was already discussed by Veld-huis and van der Steen [6], who applied a like-wise setup on a simplified flap model. The work-ing principle of the device is based on the gener-ation of von Karman vortex street that performstwo actions: a) the introduction of additional vor-ticity which enhances the mixing process in theboundary layer and b) the unsteady pulsing of theflow due to the periodic shedding of the vorticesbehind the cylinder (fig. 6b). The latter effecthas been described also by Nishri et al [26] whofound a beneficial effect of pulsed excitation ofthe flow.

Since the system is passive its structural lay-out is very simple and in a practical application

4



DBD and Nano-pulsed Plasma actuators (1)

Flow

~ O(1-10 kHz) O(3-20 kV)

Upper electrode

Lower electrode

Insulator (Kapton)

O(10mm)

O(0.2-1mm)

E

Plasma

Principle of standard Di-electric Barrier Discharge Actuator

V

Fb

y

- +

u

Wall jet

t=1-2 mm

Umax=O(8m/s)

AC

a) SDBD actuator


DBD and Nano-pulsed Plasma actuators (2)

Flow

O(1-10 kHz) O(3-20 kV)

Upper electrode

Lower electrode

Insulator (Kapton)

Principle of Nano-Pulsed Actuator

V

Nanosecond-pulse

Local heating 200-300K

Shock Wave

Prodcution of vorticity

b) NPDBD actuator

Fig. 5 Working principle of Standard DBD (a)and Nano-pulsed DBD (b) actuator. The SDBDactuator produces a body force, Fb, whereas theNano-pulsed actuator is based on thermal heatingwhich leads to the production of shockwaves thatin turn produce vortical structures.

VG

(a)

(b)

Fig. 6 Layout of the model with VG’s applied onthe main element lower surface (a) and a cylinderin front of the flap leading edge (b).

the cylinder may be stowed in the cove cavity.Ref. [27] has shown that the presence of sucha cavity has no detrimental effect on the overallflap performance as the flow always separates atthe sharp corner on the lower side of the slot.

3.2.3 Vortex generators

Based on the positive results that were obtainedearlier with off-surface vortex generators (VG)[6], the lower side of the main element wasequipped with VG’s at the trailing edge. Againthe flow separation control mechanism is basedon increased mixing in the flap upper surfaceboundary layer (fig. 6a). The optimum VG posi-tion and dimension was determined in earlier testcampaigns on a simplified wing-flap model.

3.2.4 Drooped spoiler

To improve the high lift characteristics newspoiler and flap kinematics are currently inves-tigated in various research programs like Clean-Sky [8]. The so-called drooped spoiler panel wasdiscussed by Reckzeh [11] and more recently by

5


Drooped Spoiler Panel (DSP)

θDSP

Fig. 7 Drooped spoiler panel (DSP) layout as ap-plied in this research.

Wang and Wang[25].In order to increase the low speed aerody-

namic performance and reduce the complex flapkinematics and weight, the spoiler panel, whichis directly positioned in front of the flap, is al-lowed to have a downward deflection for par-ticular (high) flap deflections. In our study thedrooped spoiler effect was simulated by adding asmall additional panel at the main wing trailingedge. Its main purpose is the relieve of the largepositive pressure gradient after the flap nose suc-tion peak to prevent premature flow separation atlarge flap deflection angles (fig. 7).

3.3 Measurements

Balance measurements Since Model 1 wasnot equipped with pressure taps the overall modelforces were determined using an external 6-component balance.

Pressure measurements Surface pressuremeasurements were performed were performedon Model 2 only. A total of 84 surface pressureswere measured (56 on the main element and28 on the flap). The wake of this model wasmeasured using a dedicated wake rake that waspositioned at approx. 1 m behind the model(fig. 4b). Moreover, at several stations thetotal pressure distribution in the wake and theboundary was determined by using a traversing aboundary layer probe.

Shadow graph technique (Schlieren Method)In case of the nano-pulsed plasma controller the

flow in the vicinity of the actuator was visualizedusing the so-called Schlieren technique. Thistechnique was essential in the detailed recogni-tion of shock waves that were emitted from theactuator.

Surface flow visualization In addition to pres-sure measurements, tuft and flow visualizationbased on a fluorescent oil technique were car-ried out. These data were used to check the in-terpretation of the measured pressure data thatwere sometimes inconclusive with respect to theamount of separated flow over the surface. Be-cause the tuft is extremely flexible, it will pointin the direction of the flow.

4 Experimental results

4.1 Model 1

The flow control capabilities of nano-pulsedplasma actuators was first tested on a single el-ement airfoil (Model 1)[5, 20]. The model wastested in wind speeds between 5m/s and 30m/s(Rec = 0.5−3.5×105). Earlier investigations in-dicated that the standard DBD actuators showedlittle effect at wind speeds beyond V = 15m/sdue to their limited momentum input. As theworking principle of the NP-DBD actuator isquite different (the exact working principle iscurrently under investigation) significant effectswere obtained for higher wind speeds. Fig. 8shows the effect of an NP-DBD actuator appliedon the nose of Model 1. In the stall regime asignificant increase in the lift coefficient was ob-tained.

Although the exact working principle of theNP-DBD actuator is unclear to date the produc-tion of spanwise vortical structures is evidencedby the Schlieren images that were taken (fig. 9).The images clearly show the occurrence of vor-tices shortly after the pulse which in turn leadto flow reattachment. The vorticity productionmechanism might either be the result of shockwave shear layer interaction or high speed heat-ing (up to T = 300K). To get further insightin these mechanisms currently an additional re-

6


NACA63618 airfoil model with a doubleNP-DBD actuator applied close to the LE; Wind

speed isV = 30m/s.

10 12 14 16 18 20 22 24 26 280.9

0.95

1

1.05

1.1

1.15

1.2

1.25

1.3

1.35

1.4

α (deg)

CL

ac tuator off

ac tuator on

Fig. 8 Lift polar of Model 1 showing stall delayat high angle of attack. The angle of attack,α,was not corrected for windtunnel wall effects.

search program has been initiated at Delft Uni-versity.

Furthermore, with these results in mind boththe standard DBD actuator and the Nano-pulseactuator were tested on the flap of Model 2.

4.2 Model 2

4.2.1 Basic characteristics

To determine the reference characteristics forlater application of flow control devices, first theoptimal position of the flap with respect to themain element (the gap and overlap, fig. 10) wasdetermined through a series of windtunnel tests.

Fig. 11 shows the effect of changing the gapfor a fixed overlap of 8% in case of a flap de-flection of δ f = 15o. As long as the gap is wellover 1% the effect of gap changes are relativelysmall. In case of a 1% gap the flow through thecove loses to much energy (total pressure) whichresults in a premature separation over the flap up-per surface which in turn reduces the circulationover the airfoil.

An overview of gap and overlap effects fordifferent flap settings is presented in figs. 11 and12. As can be seen for a gap range of 2% to 4%and and overlap range of 0% to 1% of the chord,rather flat curves are obtained. Based on thesedata the gap and the overlap for the higher flapdeflections (beyond δ f = 40o were selected 3%and 1% respectively. Typical pressure distribu-tions for different angles of attack and a flap set-ting angle of δ f = 50o, are presented in fig. 13.

It was found that for low angles of attack theflow over the flap is attached up to δ f = 30o.However for larger flap deflections flap trailingedge flow separation starts to occur. The dent inthe flap pressure distribution close to the nose (ar-row in fig. 13) is the result of an interaction be-tween the main element and the flap. It pushes thelow pressure peak down to a lower level whichin turn leads to a limited pressure rise behindthe (2nd) low pressure peak. This interaction be-tween main element and flap is thus beneficialin the postponement of flow separation over theflap.

7


Δt = 8μs,V = 30m/s, α= 26o, Re = 4×105

Actuation phase at time t0 (left) and flowreattachment phase at (right) ; Carrier

frequency fc = 200Hz,V = 30m/s, α= 260.

Fig. 9 Schlieren images taken over Model 1,showing the presence of Nano-plasma actuatorinduced shock waves with subsequent vorticalstructures and flow reattachment.

overlap

gap

airfoil chord line

δf

Fig. 10 Definition of gap and overlap for theX400 wing-flap model.

−5 0 5 10 15 20−0.5

0

0.5

1

1.5

2

2.5

3

α(deg)

Cl

Baseline, gap = 1%

Baseline, gap = 2%

Baseline, gap = 3%

Baseline, gap = 4%

Fig. 11 Effect of the gap on the lift polar for δ f =15o and an overlap of 8%.

0 1 2 3 4 52.2

2.3

2.4

2.5

2.6

2.7

gap%

Cl m

ax

Baseline, δf = 15o

(a) δ f = 15o; overlap = 8%; Re = 2.0×106

2 2.5 3 3.5 43

3.05

3.1

3.15

3.2

3.25

3.3

gap%

Cl m

ax

Baseline, overlap = 0.0%



(b) δ f = 30o; Re = 1.7×106

Fig. 12 Overview of gap and overlap effect onmaximum lift coefficient.

8


0 0.2 0.4 0.6 0.8 1 1.2

−8

−7

−6

−5

−4

−3

−2

−1

0

1

x/c

Cp

α = 0o

α = 6o

α = 12o

α = 13o

Fig. 13 Typical pressure distributions over thebaseline configuration with large flap deflection;Re = 1.7×106 ; δ f = 50o.

4.2.2 Effect of lower side vortex generators

Model 2 was tested at a constant wind speed ofV = 50m/s (Rec = 2× 106). To prevent laminarseparation bubbles zigzag type transition stripswere applied at appropriate position at the winglower side. The vortex generator consisted ofDelta-type skewed elements as presented in fig.14.

Fig. 14 Vortex generator geometrical definitions.

For this research only the counter rotatingVG’s were be tested at several sizes height. Thegeometrical definitions of the VG’s, include:

• λ the mutual spacing of the vortex genera-tors

• h, the height of the vortex generator

• d, for the distance between the VG’s andthe bottom trailing edge of the Model 2.

The vortex generators are all tested for differentvalues for these parameters. Based on this fixed

values were chosen of β = 18o and and s = h re-spectively, for which Veldhuis and van der Steen[6] found best results on a simplified windtunnelmodel.

Fig. 15 shows the lift performance of two setsof multiple VG configurations. It can be seenfrom the figures that the VG configurations allshow an increase in lift at lower angles of attackbut generally have a decrease in lift near the stallangle. However, it should be noted that in ourcase the main area of interest is the flow separa-tion control at low α and not at Clmax . This is be-cause the model is tested with a trailing edge flaponly which lacks a leading edge slat that gives abetter performance at higher angle of attack.

From fig. 15 it can be seen that VG’s withsmaller and larger heights (h∗ = 0.010 and h∗ =0.070) show only minor effect in lift improve-ments. In this case the VG height is made di-mensionless with the wing chord (600mm), h∗ =h/c. The smaller height, h∗ = 0.010, results intoo small vortices that are insufficient of mixingenough high energy flow to the boundary layer atthe flap. The larger VG’s, h∗ = 0.070, are proba-bly causing adverse vortex interaction since theylead to very fluctuating lift values also visible inthe lift polar. The spacing shows an effect wherevalues should neither be too large nor too smallfor an optimal vortex performance. VG’s at aspacing of λ/h = 6.5 and λ/h = 4.5 do not resultin a significant lift improvement. This is prob-ably due to a too large spacing where an insuf-ficient amount of healthy air is pumped into theboundary layer.

Only the VG configuration with spacingλ/h = 4.0 and height h∗ = 0.023 shows a signifi-cant performance gain over the whole range of α,this setup is found to be the optimum VG config-uration. In this case a gain of ∆Cl = 0.11 is foundat α = 0o,which is an improvement of 6% com-pared to the baseline. At high angles improve-ments are obtained of ∆Clmax = 0.064 an improve-ment of 2.2%. It is clear for all configurationsthat stall is not delayed. This is not surprisingsince the stall for this airfoil flap model is mainlydetermined by the main element stall characteris-tics and not the performance of the flap.

9


−10 −5 0 5 10 151

1.5

2

2.5

3

3.5

α (deg)

Cl

Base l in e

VG, λ/h = 6 .5

VG, λ/h = 4 .5

VG, λ/h = 4 .0

Set 1, h∗ = 0.023, d∗ = 0

−10 −5 0 5 10 151

1.5

2

2.5

3

3.5

α(deg)

Cl

Baseline

VG, λ/h= 6.5, h∗ = 0.010, d∗ = 0

VG, λ/h= 4.5, h∗ = 0.010, d∗ = 0

VG, λ/h= 4.0, h∗ = 0.070, d∗ = 0.5

Set 2

Fig. 15 Effect of height, spacing and edge dis-tance on the lift polars for multiple VG configu-rations at δ f = 55o

4.2.3 Effects of cylinders

Although the effect of cylinder was tested for dif-ferent flap angles, only the results for δ f = 55o

will discussed here. Fig. 17 shows the ef-fect of on the lift performance of several cylin-der configurations. It can be seen that multi-ple cylinder configurations show significant im-provement over the complete α-range. Cylinderswith varying diameters all show a performancegain when positioned properly, in the range ofr∗ = 0.06− 0.09 and θ = 30o − 37o (fig. 16).Here r∗ and θ respectively denote the distancefrom a reference point and angle with respect tothe horizontal. The positioning seems to be moreimportant for the lift improvement than the par-ticular cylinder diameter.

The cylinder performs best with diameterD∗ = 0.0167, but hardly loses efficiency forlarger diameters. The optimum configuration is

r∗θ

Fig. 16 Definition of cylinder position parameters.

found for D∗ = 0.0167, r∗ = 0.090 and θ = 37o

The optimum cylinder diameter of D∗ = 0.0167corresponds to a reduced frequency of F+ = 3.6.For this configuration the largest gain in lift isfound near α = 0o where lift increases up to∆Cl = 0.32. This is an astonishing improvementof 18% compared to the baseline. At high an-gles improvements are obtained of ∆Clmax = 0.24.Similar to the case of the vortex generators thecylinders do not delay the stall of the airfoil flapmodel.

Fig. 18 shows the pressure distribution forthe optimum cylinder configuration compared tothe baseline model at α = 0o and 12o. It shows atboth angles of attack that the low pressure peak atthe leading edge of the upper flap surface is sig-nificantly lower with the cylinder installed. Thepressure rise is also found further towards the flaptrailing edge, which suggests that flow separationis postponed. It further seems that the pressurerecovery is smoother for the cylinder configura-tion. The enhanced flap lift reduces pressures atthe upper airfoil which is achieved through an in-creased circulation effect of the flap on the mainelement. It should be noted that apparently asmall downstream displacement of the separationpoint may lead to noticeable effects on the overalllift of the model.

Fig. 19 shows fluorescent oil visualizationpictures of the flap for the baseline and cylinderconfiguration at α = 0o and α = 10o respectively.Observation of fig.19a shows that the fluorescentoil is streaked downstream (from right to left) uptill the dividing vertical line. This is the pointwhere laminar to turbulent transition takes place.At this point, inside the bubble, there is a verylow flow speed and oil is clumped up resulting in

10


−10 −5 0 5 10 151

1.5

2

2.5

3

3.5

α(deg)

Cl

Baseline

Cylinder, r∗ = 0.060, θ = 37◦

Cylinder, r∗ = 0.090, θ = 30◦

Cylinder, r∗ = 0.090, θ = 37◦

(a) Cylinders set 1, D∗ = 0.0167

−10 −5 0 5 10 151

1.5

2

2.5

3

3.5

α(deg)

Cl

Baseline

Cylinder, r∗ = 0.0167, θ = 55◦

Cylinder, r∗ = 0.060, θ = 18◦

Cylinder, r∗ = 0.090, θ = 37◦

(b) Cylinders set 2

−10 −5 0 5 10 151

1.5

2

2.5

3

3.5

α(deg)

Cl

Baseline

Cylinder, r∗ = 0.090, θ = 30◦

Cylinder, r∗ = 0.105, θ = 30◦

Cylinder, r∗ = 0.045, θ = 42◦

(c) Cylinders set 3, D∗ = 0.025

Fig. 17 Effect of size and positioning on the liftfor different cylinder configurations at δ f = 55o.

0 0.2 0.4 0.6 0.8 1 1.2

−8

−7

−6

−5

−4

−3

−2

−1

0

1

x/c

Cp

Baseline

Cylinder, D∗ = 0.0167, r∗ = 0.090, θ = 37o

(a) α = 0o

0 0.2 0.4 0.6 0.8 1 1.2

−8

−7

−6

−5

−4

−3

−2

−1

0

1

x/c

Cp

Baseline

Cylinder, D∗ = 0.0167, r∗ = 0.090, θ = 37o

(a) α = 12o

Fig. 18 Pressure distributions for optimum cylin-der and baseline configuration at low and high α;D∗ = 0.0167, r∗ = 0.090, θ = 37o, δ f = 55o.

11


the clear dividing line. After the bubble the flowreattaches and becomes turbulent. In this area allthe oil moves downstream leaving a rather cleansurface. Further downstream the turbulent flowis unable to cope with the adverse pressure gradi-ent and separates. The separation point is clearlyseen by a second dividing line where the flow hascome to rest. After this point the flow is non-uniform and circulates and oil is smeared in dif-ferent directions.

These oil flow patterns are seen in all the pic-tures, although clear differences are also visible.Most noticeable is the clear shift downstream ofthe second vertical line for the cylinder config-uration model with respect to the baseline con-figuration. This confirms the observations fromthe pressure distributions that flow separation hasbeen postponed. Again this proves that the cylin-der vortices indeed have a positive effect on de-laying flow separation. Comparing the oil pat-terns between the baseline and cylinder configu-rations also shows that the flap with the cylinderin front exhibits a turbulent layer directly fromthe leading edge. Hence no laminar separationbubble is present in this case.

4.2.4 Plasma actuators

Standard plasma actuator To control the flowover the flap both standard DBD and NP-DBDactuator were utilized (fig. 20). The AC DBDactuator consist of 60μm thin rectangular cop-per electrodes made out of self-adhesive coppertape. A dielectric layer (Kapton) separates theupper electrode from the lower one. Tests arecarried out with one to three Kapton tape lay-ers. To prevent the actuators discharging withthe model itself, the surface is locally insulatedby one Kapton layer before placing the actuator.The thickness of one such layer measures 2mil(1mil = 25.4μm). The upper electrode is con-nected to the HV output cable of a TREK 20/20CHV amplifier (± 20kV, ± 20mA, 1000W) whilethe lower electrode is connected to a groundedcable.

Separate measurements in which the lengthof surface roughness was changed over the span

Baseline, α = 0o(left), α = 10o(right)

Cylinders, α = 0o(left), α = 10o(right)

Fig. 19 Oil flow visualization on the flap for thebaseline and optimum cylinder configuration atδ f = 55o. The main flow direction is from theright.

Fig. 20 Standard DBD actuator positioned closeto the LE of the flap. At the dark rectangular areamultiple layers of dielectric were placed to pre-vent discharge with the metal orifice.

12


−2 0 2 4 6 8 10 12 14 16 180.5

1

1.5

2

2.5

3

α (deg)

Cl

AC DBD off

AC DBD on

Fig. 21 Effects of the standard (AC) DBD plasmaactuator placed on the LE flap on the lift polar;δ f = 45o, Re = 1.7×106; width upper electrodewu = 10mm; applied voltage V = 28kVpp; carrierfrequency fac = 700Hz

showed that an actuator should span at least 80%of the total flap span to obtain reasonable quasi-2D effects. A longer actuator length would leadto limited power input per meter resulting in re-duced effects on the flow.

The range of parameters that were changedduring the experiments include

• actuator length 20−100cm

• applied voltage 20−30kV

• carrier frequency 600−800Hz

• actuator position w.r.t. separation line

• tunnel speed 10−40m/s

The possible effects in these experiments werefirst quantified by applying tufts visualization onthe flap’s upper surface. As more or less ex-pected, in none of these cases AC DBD plasmaactuator succeeded in showing any separationpostponement effects during the tuft visualiza-tion. Even not at very low velocities up to 10m/s.This can be seen in fig. 21 where the lift polar ispresented for the actuator on and off test.

Nano-pulsed actuator Following a similar ap-proach as for the AC actuator the NP-DBD ac-tuator was also tested for different configurations

−2 0 2 4 6 8 10 12 14 16 180.5

1

1.5

2

2.5

3

α (deg)

Cl

NS DBD on

C le an

Fig. 22 Typical example of the effect of theNS DBD plasma actuator placed on the LE flapon the lift polar. ; δ f = 45o, Re = 1.7× 106,carrier frequency fac = 700Hz, burst frequencyfb = 3.3kHz.

while maintaining the proposed minimum actua-tor length and the energy discharge density. Theactuators were all driven by the maximum volt-age the generator can deliver i.e. 10kV . First, theeffect of the actuator is investigated using tuft vi-sualization by varying the following parameters:

• actuator length 20−100cm

• actuator position w.r.t. separation line

• tunnel speed 10−40m/s

Since no visible change in the separation line lo-cation on the flap could be observed, the amountof parameters to investigate were limited. Fig. 22compares the actuator-on case with the clean case(actuator completely removed). This is a typicalresult obtained for most of the NP DBD actua-tors in this research. The lift coefficient is lowerin the case of the active actuator. Apparently theflow situation for the single element airfoil in sec-tion 4.1 is quite different here and no beneficialeffect of the actuator is noticed. Furthermore, thesomewhat lower value of the lift coefficient com-pared to the clean case is very likely a result ofthe thickening effect of due to the actuator itselfwhich leads to earlier flow separation.

13


LETE

BL transition

attached flow over flap

DSP

V

δf = 55o

Fig. 23 Flow visualization over the upper surfaceof Model 2, indicating fully attached flow overthe flap when the DSP is applied; α = 0o, δ f =55o.

4.2.5 Drooped spoiler

The drooped spoiler (DSP) was investigated at aflap angle of δ f = 55o since the main goal of itsapplication was to reattach the flow in the area ofa large adverse pressure gradient. The expectedeffect is an increased effective camber of the air-foil with associated increase in the lift coefficient.Before taking pressure data surface flow over themodel was checked using the fluorescent oil tech-nique. Fig. 23 indicates that the DSP with alength of approximately 30mm ( 5% chord) anda deflection angle of θDSP = 15o now leads to afully attached flow. This is a considerable im-provement over the other techniques that only ledto a limited delay of flap stall.

However, when looking at the lift polar ofModel 2 equipped with DSP (fig. 24) a disap-pointing effect is noticed. For small angles ofattack the lift has increased slightly but beyondα = 4o the value drops considerably below thebaseline result. Although the flow is fully at-tached over the full angle of attack range and theflap introduces a stronger downwash directly be-hind its trailing edge, the adverse pressure gradi-

0 5 10 151

1.5

2

2.5

3

3.5

α (deg)

Cl

Base l in e

DSP, θ = 15o

Fig. 24 Comparison of the lift polar of the base-line model and the model equipped with DSP(θDSP = 15o).

ent is now so strong that “wake bursting” occurs.This phenomenon, that received only limited

attention in literature, is not very well understoodso-far. The overall circulation around the modelis considerably reduced due to the wide wake inwhich flow reversal takes place. To understandthe behaviour of the model equipped with DSPthe pressure distributions for α = 0o and 8o areconsidered in fig. 25.

The higher lift due to the DSP at low an-gles of attack (fig. 25a) seems to be a direct re-sult of the forced flow attachment (see flap TE)that leads to increased circulation on the mainwing. In this case the boundary layer thicknesson the wing upper side combined with the mod-erate positive pressure gradient does not yet leadto significant wake bursting. However, the situ-ation changes completely for the larger angle ofattack (fig. 25b). The DSP again forced fullyattached flow over the flap but the thicker upperside boundary layer over the model now leads toa strong wake bursting effect. The lift on boththe main wing element and the flap is much lowerthan found for the baseline configuration.

Wake bursting is associated with flow rever-sal in the wake (of the main element) and as suchis hard to detect through surface pressure mea-surements and surface flow visualization. Flowreversal was indeed noticed in our experiment byscanning the flow field behind the model withDSP using a Pitot tube (fig. 26).

14


0 0.2 0.4 0.6 0.8 1 1.2

−8

−7

−6

−5

−4

−3

−2

−1

0

1

x/c

Cp

Base l in e

DSP, θ = 15o

a) α = 0o

0 0.2 0.4 0.6 0.8 1 1.2

−8

−7

−6

−5

−4

−3

−2

−1

0

1

x/c

Cp

Base l in e

DSP, θ = 15o

b) α = 8o

Fig. 25 Pressure distributions of the modelequipped with DSP (θDSP = 15o) for α = 0o and8o.

−600 −400 −200 0 200 400 600 800 1000 12000

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

gauge pre ssure (Pa)

ytrav/c

Fig. 26 Wake traverse result of a Pitot tubeat short distance behind the model showing off-surface flow reversal, in the range ytrav/c =0.022− 0.062, indicating the presence of “wakebursting”; α = 8o.

With the wake bursting occurring as a re-sult of the very powerful flow control effect ofthe DSP one might argue that apparently pas-sive methods for the increase of maximum lift forthis particular model are quickly exhausted. Thisalso suggests that any active/passive flow separa-tion control method for flaps in a high lift sys-tem might have limited applicability when strongpressure gradients exist in the wake flow area.

5 Conclusions

Experimental analysis of several passive and ac-tive control techniques to control flow separationover a highly deflected flap have been performedsuccessfully. Among the techniques that wereused are: VG’s on the main element lower sur-face, cylinder in front of the flap leading edge,plasma actuators close to the wing leading edgeand a drooped spoiler panel. From the exper-imental results the following main conclusionscan be drawn.

• The application of NPDBD actuatorsshowed significant suppression of flowseparation over a single element airfoil atspeeds up to 30-40 m/s. However, the ef-fects of both SDBD and NPDBD actuatorson the flap of a multi-element airfoil werenegligible. The mechanism by which theNPDBD actuator suppresses flow separa-tion under certain circumstances is still un-clear.

• The application of VG’s on the wing lowerside as well as a cylinder in front of theflap, leads to a noticeable increase of thelift at flap deflections were serious trail-ing edge separation occurs. The work-ing principle of these passive techniques isbased on enhancement of increased mix-ing in the flap boundary layer due to eitherstreamwise vortices (VG’s) or lateral ori-ented vortices produced by the cylinder.

• A drooped spoiler panel (DSP) positionedat the back of the main wing (simulat-ing a downward deflected spoiler) led to a

15


fully attached flow over the flap at flap an-gles up to δ f = 55o. However, for mod-erate to high angles of attack the wake ofthe main element experiences a very largepositive pressure gradient which leads towake bursting. The occurrence of this phe-nomenon very likely limits the positive ef-fect of passive and active flow control at theflap of multi-element airfoils.

References

[1] Gad-el Hak, M. Flow Control; Pas-sive, Active, and Reactive Flow Man-agement. Cambridge University Press,2000.

[2] Gilarranz, J., Traub, L. and Redini-otis, O, Characterization of a Com-pact, High-Power Synthetic Jet Ac-tuator for Flow Separation Control.40th AIAA Aerospace Sciences Meet-ing and Exhibit, 14-17 January 2002.AIAA 2002-0127.

[3] Greenblatt, D. and Wygnanski, I., Thecontrol of flow separation by periodicexcitation. Progress in Aerospace Sci-ences, 36:487-545, 2000.

[4] M. Bauer, I. Peltzer, W. Nitsche andB. Gölling, Active flow control onan industry-relevant civil aircraft halfmodel, Notes on Numerical FluidMechanics and Multidisciplinary De-sign, 2010, Volume 108/2010, 95-107,DOI: 10.1007/978-3-642-11735-0_7

[5] Correale, G., Popov, I., Ratikin, A.,Starikovskii, A. and Veldhuis, L.L.M.,Flow Separation Control on Airfoilwith Pulsed Nanosecond DischargeActuator, 63rd Annual Gaseous Elec-tronics Conference and the 7th Inter-national Conference on Reactive Plas-mas, Vol. 55, Number 7.

[6] Veldhuis, L.L.M. and van der Steen,M., Flow separation control by off sur-face elements, AIAA-2010-4684, 28thAIAA Applied Aerodynamics Con-ference, Chicago, Illinois, June 28-1,2010

[7] Boermans, L.M.M. and Rutten, P.B.,Two-dimensional aerodynamic char-acteristics of airfoil NLF-MOD22with fowler flap, Internal Report LSW-95-3, Delft University of Technology,1995

[8] Cleansky, http://www.cleansky.eu[9] Smith, A.M.O., High lift aerodynam-

ics, Journal of Aircraft, Vol. 12, No. 6,1975

[10] van Dam, C.P., The aerodynamic de-sign of multi-element high-lift sys-tems for transport airplanes, Progressin Aerospace Sciences 38 (2002)101–144

[11] Reckzeh, D., The Role of High-LiftAerodynamics . KATnet II conference(pp. 1-18). Bremen: Airbus, 2009

[12] Göksel, B. , D. Greenblatt, D.,Rechenberg, I., Singh, Y, Nayeri, C.N.and Paschereit, C.O., Pulsed PlasmaActuators for Separation Flow ControlB, Conference on Turbulence and In-teractions TI2006, May 29 – June 2,2006, Porquerolles, France

[13] Roth, J.R. Physics and phenomenol-ogy of plasma actuators for controlof aeronautical flows. s.l. : J., Phys.D:Appl. Phys. 40, 2010.

[14] Post, M. L. and Corke, T.C. Separationcontrol on high angle of attack airfoilusing plasma actuators. s.l. : AIAAJournal, vol. 42, no11, pp. 2177-2184,2004.

[15] Post, M.L. and Corke, T.C. Separationcontrol using plasma actuators: dy-namic stall vortex control on oscillat-ing airfoil. s.l. : AIAA Jornal, 44 3125,2006

[16] Enloe, et al. Mechanisms and Re-sponses of a Single Dielectric BarrierPlasma Actuator: Plasma Morphol-ogy. AIAA Journal, v. 42, n. 3 : s.n.,2004. pp. pp. 589-594.

[17] Kotsonis, M. , Dielectric Barrier Dis-charge Actuators for Flow Control,PhD dissertation, Delft University ofTechnology, 2012

16


[18] Kotsonis, M., Ghaemi, S., Veldhuis,L., & Scarano, F., Measurement of thebody force field of plasma actuators.Journal of Physics D: Applied Physics,44(4), 2011.

[19] Kotsonis, M. and Veldhuis, L., Ex-perimental study on dielectric barrierdischarge actuators operating in pulsemode, J. Appl. Phys. 108, 113304,2010.

[20] Correale, G., Popov I., Rakitin A.,Starikovskii A., Hulshoff, S. ,Veld-huis L., Flow Separation Controlon Airfoil With Pulsed NanosecondDischarge Actuator, AIAA-2011-107949th AIAA Aerospace Sciences Meet-ing, Orlando, Florida, Jan. 4-7, 2011.

[21] Moreau, E. Airflow control by non-thermal plasma actuators. s.l. : J. Phys.D: Appl. Phys. 40 605, 2007.

[22] Okita Y, Jukes T N, Choi K-S andNakamura K. Flow reattachment overan airfoil using surface plasma actu-ator. s.l. : AIAA 2008-4203, 2008,2008.

[23] Starikovskii, A.Y. Nikipelov, A.A.,Nudnova, M.M.M and Roupassov,D.V. SDBD plasma actuator withnanosecond pulse-periodic discharge.s.l. : Plasma Sources Sci. Technol. 18.

[24] Little, J., Takashima, K., Nishihara,M. , Adamovich, I andSamimy, M.High Lift Airfoil Leading Edge Sepa-ration Control with Nanosecond PulseDriven DBD Plasma Actuators. s.l. :AIAA 2010-4256, 2010.

[25] Xiao-liang Wang and Fu-xin Wang,Aerodynamic Characteristics of High-Lift Devices with Downward Deflec-tion of Spoiler, Journal oif AircraftVol. 48, No. 2, March–April 2011

[26] Nishri, B. and Wygnanski, I., ” Ef-fects of periodic excitation on turbu-lent flow separation on a flap. AIAAJournal, 36(4):547-556, April 1998

[27] Potter, R.C. and van Dam, C.P., Vis-cous flow analysis of a subsonic trans-port aircraft high-lift system and cor-

relation with flight data, NASA-CR-199610, 1995

Copyright Statement

The authors confirm that they, and/or their company ororganization, hold copyright on all of the original ma-terial included in this paper. The authors also confirmthat they have obtained permission, from the copy-right holder of any third party material included in thispaper, to publish it as part of their paper. The authorsconfirm that they give permission, or have obtainedpermission from the copyright holder of this paper, forthe publication and distribution of this paper as part ofthe ICAS2012 proceedings or as individual off-printsfrom the proceedings.

17

novel passive and active flow control for high lift - icas is

Documents