N AS A / TM- 1999-209838

Active Control of Flow Separation Over anAirfoil

S. S. Ravindran

Langley Research Center, Hampton, Virginia

National Aeronautics and

Space Administration

Langley Research CenterHampton, Virginia 23681-2199

December 1999

Available from:

NASA Center for AeroSpace Information (CASI)7121 Standard Drive

Hanover, MD 21076-1320

(301) 621-0390

National Technical Information Service (NTIS)

5285 Port Royal Road

Springfield, VA 22161-2171(703) 605-6000

Active Control of Flow Separation Over An Airfoil

S. S. Ravindran *t

NASA Langley Research Center

Hampton, Virginia

Abstract

Designing an aircraft without conventional control

surfaces is of interest to aerospace community. In

this direction, smart actuator devices such as syn-thetic jets have been proposed to provide aircraft

maneuverability instead of control surfaces. In this

article, a numerical study is performed to inves-

tigate the effects of unsteady suction and blow-ing oil airfoils. The unsteady suction and blow-

ing is introduced at the leading edge of tile airfoilin the form of tangential jet.. Numerical solutions

are obtained using Reynolds-averaged viscous com-

pressible Navier-Stokes equations. Unsteady suc-tion and blowing is investigated as a means of sep-aration control to obtain lift on airfoils. The ef-

fect of blowing coefficients oil lift. and drag is in-vestigated. The numerical simulations are com-

pared with experiments from tile Tel-Aviv Univer-

sity (TAU). These results indicate that unsteady

suction and blowing can be used as a means of sep-

aration control to generate lift. on airfoils.

Nomenclature

A,B

CL

c.

<c,>

flux Jacobianlift coefficient

unsteady momentum blowing coefficient,

steady momentum blowing coefficient,

ie<

*NRC Research Fellow, Flow Physics and Control

Branch, NASA Langley Research Center, Mail Stop 170,

Hampton, VA 23681

tCurrently assistant professor, Department of Math-

ematics, University of Alabama, Huntsville, AL 35899

(ravindra@ult ra .math. uah. edu).

CD

c_F,G

F,,,G,,

f +

H

I

M

M

QPr

Re

St

T

U,V

UjetU_a

c

e

f

p

etr

t

U, V

< Ujet >x, yAt

(t

"7A

P

P

7ij(,_,_

drag coefficientpressure coefficient

fluxes of mass, momentum and energyviscous terms of the Navier-Stokesfornullation

non-diluensional actuator frequency, fc/U_

width of the orifice, m

identity matrixfree-streanl Math number

transformation matrix from conserved

variables to primitive variables OQ/Oqconservation variables

Prandtl nuinber

Reynolds mlmber, P_"_D

actuator Strouhal number, fH/Votemperature, °Ccontravariant velocities

spatial mean jet. velocity from actuator

free-stream velocity, m/s

speed of sound, m/schord length, m

total energy

actuator forcing frequency, Hzpressure, nondimensionalized by 7P_

primitive variables

distance outward from body

time, nondimensionalized by c/ao¢

Cartesian velocities in x and y direction,

respectively, nondimensioimlized by ao¢RMS velocity of the actuator jet oscillations

Cartesian coordinates

time step (nondimensional)

angle of attack (degrees)

difference operator

ratio of specific heats

coefficient of bulk viscosity

viscosity

densityviscous stress tensor

curvilinear coordinate directions

Subscripts

max maximummin nfininmmx, y denotesdifferentiationinx andy directions,

respectivelyJ],_ denotesdifferentiationin 7/and

( directions, respectively

() _. free-stream condition

Superscripts

n denotes time level

0 denotes quantities in generalized coordinates+- denotes positive and negative flux

conditions

I. Introduction

The experimental and computational investiga-

tions of active control of flow past airfoils at high

angles of attack is an area of active research as ex-

tending the usable angles of attack has many im-

portant applications. There are a number of arti-

cles showing the effectiveness of flow control for air-

foils. For example, in [10] leading edge suction was

investigated for transition delay, in [23] jet flapswere employed for lift increase and in [6] surface

suction/blowing was used to rapidly change lift and

drag on rotary wing aircraft. However, most of the

control techniques considered in the past required

relatively more power input or involved weight and

comt)lexity penalties.

In [11, [2] and [18], an innovative method for ac-tive control has been experimentally demonstrated

using the so-called synthetic jet. The synthetic jet

actuator produces a high-frequency jet from the

surrounding fluid with zero net mass input. A

novel feature of this actuator is that it requires

only electrical power. In [1] azld [18], a synthetic jet

was used to produce a larger jet and in [2] a pair

of actuators were used to show significant lift on

cylinders. In another experimental work, Seifert

et al [15] have investigated unsteady suction andblowing on a syinmetric airfoil to increase post-

stall lift. They show by introducing an unsteady

jet near the leading edge tangential to the surface

of a NACA0015 airfoil significant increase in lift

can be obtained with relatively low momentum in-

put. They also observed re-at, tachment of the flow

and elimination of large wake region above the at-

tached region. Later Seifert et. al. [16] and Seifert

and Pack [17], investigated the dependence of ac-

tuator location, momentum coefficient of the jet

and frequency of the oscillation, and performance

in higher Reynolds number.

Some of the recent numerical simulations re-

ported in [21, 6, 5] also support these findings.

In [21], a periodic blowing and suction normal tothe surface was used at 2.5% chord from the lead-

ing edge of a NACA0012 airfoil. They showed

that lift was increased for angles of attack be-

tween 18°-35 o . They used tile Reynolds average

Navier-Stokes (RANS) approach with the Baldwin-

Lomax algebraic turbulence model. Hassan et. al.

[6] have also used RANS approach with Baldwin-Lomax turbulence model. In their work zero net

mass suction/blowing was placed at 13% chord and

they found out that for certain oscillation frequency

and peak amplitude, the lift can be increased albeit

with high momentum input. The comparison with

experiments were not given in both of these works.

Finally, Donovan et. al. [5] reported numerical in-

vestigations of both steady and unsteady jet oil air-

foils. They used an unsteady RANS incompressible

flow solver with Spalart-Allmaras [19] turbulencemodel and compared their results with the experi-

ments of [15]. Performance improvements were ob-tained by placing the actuator near the airfoil lead-

ing edge. A significant lift increase of about 297c

was obtained using synthetic jet actuators in the

post-stall regime. However, their sinmlations as-

sumed incomt)ressibility which is not not strictly

valid as the Mach number in tim experiments of

Seifert [15] was M = 0.15.

In this article, we present a numerical investiga-

tion of unsteady suction and blowing on an airfoil

using a compressible flow solver CFL3D [9]. We

build upon the previous work of Donovan et al [5]

to validate CFL3D for active flow control applica-tions and to address some of the anomalies in the

previous reports.

The plan of the paper is as follows. In §II, we will

present the governing equations. In §III, we will

present the computational methods used to solve

them. In §IV, we will present numerical results for

the baseline case. In §V, we will present numeri-

cal results for the active control case. In §VI, we

conclude the paper with a summary.

II. The Governing Equations

Tile governing equations considered are the time-

dependent, viscous compressible Navier-Stokesequations. The nonodimensional form of these

equations in Cartesian coordinates for an ideal gasare

O__G_x_G_0_0__Q+ OF 0G 1 (_, + (1)Ot _ q- Oy -- Re Oy ]

where Q, F and G are the flux vectors given by

Q

P

pupv , F =

C

p_U

pu 2 + P

flUV

(e + p)ll

,G=

pY

puvpv 2 + P

(e + p)v

The viscous flux vectors F,, and G,, are defined as

F1, z

0

7-xy ,

A

0

7-xv

ru_g4

where

2

rxx = _,(2u_ -,,,_) r_ = _(u_ -- vx),

2

7=u_= 5#(2v_ - Ux),

p Oa"-f4 = ?tTa.x + Vrxy + _-_r(7 -- 1) -l Ox

P 1)-I Oa ').q4 = u_ + vT-y_ + Fr(" Y- Oy

Here the Prandtl number, Pr, is

Pr = IRcpI_oc

where cp is the specific heat at constant pressure,

and _ is the coefficient of thermal conductivity.

The pressure is defined by the equation of state

for an ideal gas:

P(_ + v'_)], (2)p= ('y-1)[e-

where 7 = _ and has a value of 1.4 for air. TheCv _

speed of sound, a, is defined as a 2 = "y_. The

Reynolds number is defined as Re = p__e__, where

c refers to a chord length and the subscript, oc,

refers to free stream quantities. The dynamic vis-

cosity, p, is approxinmted by Sutherland formula

and equation (1) is closed by Stokes hypothesis for

bulk viscosity (,_ + 2p/3).

A. The Coordinate Transformations

In order to apply the numerical algorithm and

boundary conditions easily, the governing equa-tions which are derived in the Cartesian coordi-

nates, (x,y), must be transformed to the compu-

tational domain or generalized coordinates, (_, q).The transformation from Cartesian coordinates to

general curvilinear coordinates for two-dimensionsare

_-= t, _ = _(z,._, t), ,j = ,j(x, y, t).

These coordinates conform to the surface of the

body and maps the original Cartesian space (x,y)or physical domain onto a computational doinain

(_, _1), which is rectangular with a uniform mesh. Interms of these curvilinear coordinates the Navier-

Stokes equations t)ecome

0,,- e,, + ), (3)

where

t

pv, _,=d-1

pU

puU + PG

pvU + PG(e + p)U - &p

--1pu V + WI:_

pvV + 1hi._

v(e + p) - 71fp

with

U = & + Gu + Gv, I" = _lt + 7j_u + rl_v,

where U and V are the contravariant velocities. In

curvilinear coordinates the viscous flux terms are

given by

Fv = g-'(_F,,+_uav) and G,, = J-a(rlxF,,+,luG,,),

where J = o(_.,s) Similarly, the stress terms inO(x,y,t) •Fv and G,, are also transformed.

III. Computational Algorithm

In this study tile algorithnl used to solve the

Navier-Stokes equations is the CFL3D code re-

ported ill [9]. CFL3D solves the time-dependentconservation law form of the Reynolds-averaged

Navier-Stokes equations. The spatial diseretization

involves a semi-discrete finite-volume approach.

Upwind-biasing is used for the convective and pres-sure terms, while central differencing is used forthe shear stress and heat transfer terms. Time

advancenIent is implicit with the ability to solve

steady or unsteady flows [13], [14], [20], [12]. Multi-

grid and mesh sequencing are available for conver-

gence acceleration.

A. Time Differencing

The Navier-Stokes equations (3) are discritized

in time using the backward Euler implicit scheme

and then the resultin}_ nonlinear system is lin-earized in tilne about Q" to obtain

I[_ + 6_A" + 6,y%xO" = R(O"), (4)

where

0_ _ 0 1R = -[ (Y - F,) + -

0 _ _e _5Q" = (_"+' - O", A = _--_(F - F,,)Q

alia

o(G_ 1B =

Approximate factorization

The central difference discretization to Equa-

tion (3) results in a large banded square matrix

which is sparse but would be computationally ex-

pensive. To overcome this problem an approximatefactorization is introduced which converts the two-

dimensional operator into one-dimensional opera-tors:

M

[._7 + _A*]Aq'

M

= R(q n)

51/ ,= 7At&q

and

qn+l = qn + ,.._qn,

where q = (p, u, v, p),

oOQ A* = (F - _ ,,),M = 0--_-' q

and

(5)

B'= 1G

This factorization yields two block tridiagonal in-

versions for each sweep. Prior to the execution

of Equation (5), the corrections are constrained in

order to maintain tim positivity of tile therlnody-

naniic scalars p and p. For examt)le, the update to

pressure is taken as

pn+l = pn + A[1 + ¢_(ac + I A---_P])]-1pn

whenever _ _< ac, where a = -0.2 and 0_ = 2.0.

B. Spatial Differencing

The spatial differencing is done using a second-order-accurate upwind-t)iased scheme. The fluxes

F, G representing pressure and convective terms

are differenced using up-winding and a flux-vector-

splitting method. For example, the flux difference

in the _ direction is

0j = + +

The flux vector splitting is due to Van Leer. Theflux F, for example, is split according to the con-

travariant Math mimber in the ( direction, see [9].

The split-flux differences are implemented as a fluxbalance across the ith cell holding spatial indices j

and k constant as

= [F+(Q -) + F-(Q+)]i+I/2-[?+(Q-) + ?-(Q+)]i-1/2,

where F+(Q-)i+I/2 denotes a forward flux evalu-

ated using the metric terms at the cell interface

(i + 1/2), and state variables are obtained by fully

upwind second-order interpolation of cell-centeredvariables

Qi-+1/2 = (3/2)Qi - (1/2)Qi_l,

Q++l/2 = (3/2)Qi+l - (1/2)Qi+_.

Thediffusiontermsaredifferencedusingasecondordercentraldifferencing:

C. Convergence Acceleration

A sequence of grids Go, Gl, G2,. ...... ,G_ is

defined, where GN denotes tile finest grid and

coarser grids are formed by successively deleting ev-ery other line in all coordinate directions. The fine

grid serves to damp the high-frequency errors while

the coarser grids damp the low-frequency errors.

The coarse grids are solved with a forcing function

on the right-hand side, arising from restricting the

residual from the finer grids. The forcing function

is the relative truncation error between the grids,such that the solution on the coarser meshes are

driven by the fine grid residual. A fixed cyclingstrategy (W-cycle) is used for the results presented.

The solution were smoothed oil each grid through

five steps before switching to the next mesh.

D. Turbulence Models

In order to predict turbulent flows by solving the

Reynolds averaged Navier-Stokes equations, clo-

sure assumptions must be made about the turbu-

lent stress and heat-flux quantities. For separated

unsteady flow computations using Reynolds aver-

aged Navier-Stokes equations, choosing the appro-

priate turbulence models is not trivial. Among the

available models, the Spalart-Alhnaras (SA) model

has been shown to be effective for a variety of flows

including 2-D separated airfoil flows, see [5] and[8]. The Spalart-Allmaras model is a one equation

model which solves a single transport equations for

a modified turbulent viscosity.

VI. Baseline Simulations

The airfoil configuration used here is the one used

at Tel-Aviv University (TAU) for low-speed wind

tunnel test. This airfoil, we call it TAU0015, has

a 0.4% chord notch at the leading edge and a 3%

chord thick trailing edge, otherwise is a NACA0015

airfoil. In order to lay a single-block structured grid

around the airfoil, the original airfoil was modified

to smooth out the square corners at the leading

edge actuator. Figure 1 shows this modification in

comparison with the original TAU0015 airfoil withthe leading edge actuator. The baseline simulations

given in this section uses no actuation. The follow-

ing parameters were used in the simulations which

are the same as in the experimental conditions: the

Math number of 0.15 and a chord Reynolds num-

ber of 1.2 × 106. The computations were performed

using a 417X129 C-grid shown in Figure 2 with aminimum normal spacing of 0.0000015c.

The boundary conditions applied are no-slip with

no normal velocity at the body surface. Along the

far-field upstream and the circumferential bound-

aries, a quasi-one-dimensional characteristic analy-

sis is used to determine the boundary data, assum-

ing free-stream conditions exterior to the boundary.

Along the downstream boundary, first-order ex-trapolation of the conserved variables is used. The

outer boundary is sufficiently flu' away, rmax =

12c, from the airfoil, thereby minimizing the effects

of the outer boundary on the flow over the airfoil.

Figure 3 shows tile enlarged view around leading-edge showing the grid near the actuator. Figure 4

shows the computed lift coefficients versus angleof attack computed with experimental data. The

results show very good agreement before stall, but

deviate fi'om the experimental data at stall, similar

to the results of [5]. A drag polar is shown in Figure

4 for the baseline case which also shows good agree-

ment with experiment at lower angles of attack but

deviate from the experimental values for post-stall

angles of attack. Similar observations were made

in [5] and [7] using the same airfoil and turbulencemodel but with different numerical methods and

models.

In [7] an unstructured grid was employed to

study the effects of smoothing the square corners

at the leading-edge actuator. Both the original

TAU0015 and the modified TAU0015; see Fig-

ure 1, were studied for the baseline case. Theyfound remarkable agreement between modified and

original TAU0015 airfoil computations but origi-nal TAU0015 model stalled before the modified

TAU0015 model resulting in some disagreement in

the post-stall region. They concluded that the dis-

crepancies between computed and experimental re-

sults was due to the way aerodynamic properties

were computed and thus a proper interpretation

of the experimental results is important. For ex-ample, experimental measurements used discrete

pressuretapsandhencethelift coefficientconsistsof thepressurecontributiononlyandleadingedgeactuatorregionwasnot includedin thecomputa-tion. Thiswasverifiedin [7]byrecomputingthelift withoutincludingtheactuatorportionat theleadingedge.

Wenoteherethat wedidnotgobackandrecom-putelift anddrag;ratherbecausethecomparisonsinFigures4 5aresufficientlygoodttlat theremain-derof tilepaperwill focusonthisconditions.

V. Control Simulations

In this section, we present some numerical simu-

lations of synthetic jet control and compare them

with experiments. The experiments that we se-

lected here for conlparison used a leading-edge un-

steady jet tangential to the surface.

A. Actuator Boundary Conditions

In all the calculations the synthetic jet actuator

is modified using suction and blowing type bound-

ary condition by prescribing velocity at the surface.

The velocity boundary condition is given by

u(x, 0, t) = A sin(a_t)f(x),

where the amplitude A = ._2_U_v/2 < C u >,the prescribed frequency of oscillation w =

(F+U_c/27re), the spatial distribution f(x) = 1,

F + is the non-dimensional frequency and the os-

cillatory momentum blowing coefficient < C_, > isdefined as

< C. >= 2(H/c)(< Uje t >/U_) 2.

Tile saine C-grid used in tile baseline simulations

was used here. The jet at the slot was resolved

using a fine grid consisting of twenty grid points.

B. Numerical Results

The parameters Mach number and Reynoldsnumber were taken to be the same as in the base-

line case. The experiments included various blow-

ing coefficients < C o >, with F + = 0.58. Toassess the effectiveness of synthetic jets, various

blowing coefficients were tested at fixed post-stall

angle-of-attacks a = 22 °, a = 240 and a range of

< C, > were used in the simulations. The turbu-lence model used was again Spalart-Allmaras with

tile time step At = 0.00005. In order to reduce the

computational time required to converge to a so-

lution with blowing, the baseline case solution was

used as the initial flow conditions for the blowing

computations.

The variation of ACt with blowing coefficients

< C_, > for _ = 22o is shown in Figure 6. Theincremental lift increases as the momentum blow-

ing coefficient increase for both the experimental

and computational data in much better agreement

than reported in [5]. Tile computational resultsover/under-predict lift. This may be due to in-

ability of the turbulence model's to predict sepa,

rated flows, inconsistency in the actuator geome-

try or the lack of grid resolution. Figure 7 shows

the analogous variation in ACI with blowing coef-

ficients < G, > for (_ = 24 °. The computed ACI

increase smoothly with increasing < C, > unlike

tile (_ = 22 o case. For the uncontrolled flow, a large

region of separated flow wa_ seen with two coherent

structures on the suction surface. The applicationof control makes the flow more attached which is

consistent with the works of [4] and [5].

Figure 8 shows surface pressure distribution for

tile baseline and controlled cases. Figure 9 and

Figure 10 show tile lift coefficient time histories

with synthetic jet control as a function of non-

dimensional time (L_) for a= 220 and a = 24°,

respectively. Note that approximately 40 nondi-mentional times were necessary to obtain a statis-

tically stationary solution.

VII. Summary

A computational investigation of tangential un-

steady suction and blowing for separation control

on an airfoil has been performed. The effects of

zero net mass suction and blowing on lift increasewere studied. The computed results were comparedwith available wind tunnel test results to determine

tile accuracy of the computational results. The

numerical solutions were obtained by solving the

Reynolds averaged Navier-Stokes equations. A grid

resolution study was conducted using baseline (un-

controlled) case to determine the appropriate grid

density. The computed baseline results agreed rea-

sonably with experimental results. For the active

controlcase,variousblowingcoefficientswerein-vestigatedat twoanglesof attack,a = 220 and

a = 24°. The computed results were comparedwith the experiments from TAU. The results show

reasonable agreement with tile trends observed in

the experiment.

In general, the computations showed that the lift

increased as tile blowing coefficient increased. In

all the control simulations, tile grid resolution in

general and time step-size in particular was foundto be critical.

Computational simulations shows that tangen-

tial unsteady suction and blowing on airfoil can be

used as a means of separation control to generatelift.

Acknowledgments

The author would like to thank Dr. Avi Seifert for

providing the experimental data and for many use-

ful technical discussions. This work was supported

in part by the National Aeronautics and Space Ad-ministration under NASA Contract NASW-8154

while the author was an NRC resident research as-

sociate under the supervision of Dr. Ronald Joslin.

The CRAY time was provided by the Numerical

Aerodynamic Simulation facility at NASA AmesResearch Center.

References

[1] M.G. ALLEN AND A. GLEZER, Jet VectoringUsing Zero Mass Flux Control .lets, AFOSR

Contractor and Grantee Meeting on T_rbu-

lence and Internal Flows, Wright Patterson

AFB, May 1995.

[2] _I. AMITAY, A. HONOHAN, _'1. TRAUTMAN

AND A. GLEZER, Modification of Lifting Body

Aerodynamics Using Synthetic .let Actuators,AIAA Paper 98-0209.

[3] R.C. CHANG, F.B. HSIAO AND R.N. SHYU,

Forcing Level Effects of Internal Acoustic Ex-

citation on the Improvement of Airfoil Perfor-

mance, Journal of Aircraft, 29(5) 1992.

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

A. DARABI AND I. V_rYGNANSKI, On the

Mechanism Responsible for Forced Flow Reat-

tachment Proceedings of the EUROMECH

Colloquium: Active Control of TurbulentShear Flows, Tech. Uni. Berlin, March 1997.

J.F. DONOVAN, LINDA D. KRAL AND A.W.

CARY, Active Control Applied to an AirJoil,

AIAA Paper 98-0210, 1998.

A.A. HASSAN AND R.D. JANAKIRAM, Effects

of Zero-mass Synthetic jets on the Aerodynam-

ics of the NACA-O012 Airfoil, AIAA 97-2326,1997.

R.D. JOSLIN, L.G. HORTA AND F.J. CHEN,

Transitioning Active Flow Control to Applica-tions, AIAA Paper 99-3575, 1999.

S. Ko AND _V.J. MCCROSKEY, Compu-

tations of Unsteady Separating Flows Over

an Oscillating Airfoil, AIAA Paper 95 0312_1997.

S.L. KRIST, R.T. B]EDRON AND C.L. Rt:M-

SEY, CFL3D User's Manual, NASA TM 1998-208444.

D. V. N'IADDALON, F. S. COLLIER, F. S.

NIONTOYA AND C. K. LAND, Transition

Flight Experiments on a Swept Wing with Suc-

tion, AIAA Paper 89-1893.

V..l. MODI, F. ]_'IOKHTARIAN, _,'I. FER-

NANDO AND T. YOKOMIZO, Moving Surface

Boundary-Layer Control as Applied to Two-

Dimensional Airfoils, ,1. Aircraft, Vol. 28,

1991, pp. 104 112.

G. OSWALD, K. GHIA AND U. GHIA, An b_-

plicit Time Marching Method for Studying Un-

steady Flow with Massiw_ Separation, AIAA

Paper 85-1489, 1985

C. RUMSEY, Time-Dependent Navier-Stokes

Computations of Separated Flows Over Air-

foils, AIAA Paper 85-1684, 1985.

C. RUMSEY, J.L. THOMAS, G.P. WARRENAND G.C. LIU, Upwind Navier-Stokes Solu-

tions for Periodic Flows, AIAA Paper 86-0247,1986.

A. SElrEn_r, T. BACHAR, D. Koss, M.

SIt[_PSHELOVICH AND I. _TYGNANSKI, Oscil-

latoryBlowing:A Toolto DelayBoundary-LayerSeparution,AIAA Journal, 31(11) 1993.

[16] A. SEIFERT, A. DARABI AND I. V_TYGANSKI,

Delay of Airfoil Stall by Periodic Excitation,Journal of Aircraft, 33(4) 1996.

[17] A. SEIFERT AND L. G. PACK, Oscillatory

Control of Separation at High Reynolds Num-

bers, AAIA Paper 98-0214.

[18] B. L. SMITIt AND A. GLEZER, Vectoring andSmall Scale Motions Effected in Free Shear

Flows Using Synthetic .let Actuators, AIAA

Paper 97-0213.

[19] P.R. SPALART AND S.R. ALLMARAS, A OneEquation Turbulence Model for Aerodynamie

Flows, AIAA Paper 92-0439, 1992.

[20] J. THOMAS AND R. _VALTERS, Upwind Relax-ation Algm_thms for the Navier-Stokes Equa-

tions, AIAA Paper 85-1501, 1985.

[21] J.M. Wu, X.Y. Ll_, A.G. DENNEY AND ]_4.FAN, Post-Stall Lift Enhancement on an Air-

foil by Local Unsteady Control, Part I. Lift,

Drag and Pressure Characteristics, AIAA 97-2063.

[22] J.M. Wu, X.Y. Lu AND J.Z. Wu, Post-Stall Lift Enhancement on an Airfoil by Local

Unsteady Corttrol, Part H. Mode Competition

and Vortex Dynamics, AIAA 97-2064.

[23] H. YOSHIIIARA AND D. ZONARS, The Tran-

sonic Jet Flap- A Review of Recent Results,

Proceedings, SAE National Aerospace & Man-

ufactoring Meeting, November 1975.

0.05

0.025

0

-0.025

-0.05I

0 0.05X/C

FIG. 1 Leading edge regions of TAU0015and modified TAUO015 airfoils.

FIG. 2 TAU0015 airfoil 417 x 129 C-grid,

Armi n = 0 0000015c, rmax = 12c.

FIG. 3 Blowup of the TAU0015 airfoil near

the leading edge.

2

1,8

1.6

1.4

1.2

1

0.8

0.6

0.4_ CFD, S-A

5 10 15 20o(

FIG. 4 Baseline lift coefficient for TAU0015

airfoil computed with experiments.

1,5

1,4

1.3

1.2

1.1

1

jO.90

0.8

0.7

0.6

0,5

0.4

0.3

0

-- ExpCFD, S-A

i i L L I i , , , I , , , I j ,0.1 0.2 0,3

Cv

FIG. 5 Baseline drag polar for TAU0015 air-

foil compared with experiments.

0,25

0.2

0.15

0.1

0.05CFD, S-A

0,0001 0.0002 0,0003 0,0004 0.0005 0,0006 0.0(]07

<C_>

FIG. 6 ACL versus < C_, > for the TAU0015

airfoil at a : 220 compared with experiments.

0.6

0.5

0.4

0.3

0.2

0.1

ol/

0.0001 0,0002 0.0003 0,0004 0.0005 0.0006<C _.>

1.05

1

095

0,9

0.85

0.8

0.75

0.7

0.65

0,6

0.55

0.5

0.45

0.4

0.35

[ ........ CocL

!i !'!j_'ili r t r t. t _ , t ' ' _ _ _ ' ' _ t ': t t[_J" .i _ J'!i /!iiiii! ! !i !i (! ii /i i! i! i! !! i_! i i

ijij:iji_)i ii;i ii,:iiiiiitiLiiiiiliiiO LiiFi_i

• I I I I [ i h I I I , , , ,1O0 200 300

t/.O01

FIG. 7 ACL versus < C, > for the TAU0015airfoil at _t = 24 °.

o"

Controlled

3 ,_ Uncontroled

2

O _.j _.ac,cc.c, cc._-c.c.c_ccc=ccc

0 0.25 0.5 0.75x/c

FIG. 8 Pressure distribution on TAU0015 air-

foil at Re = 1.2 x 106 , M=0.15, _ = 240 ,

F + = 0.58 and < C, >= 0.0006.

FIG. 9 Lift and Drag history of time accuratecalculations of TAU0015 airfoil for M=0.15,

Re = 1.2 × 106 , o = 220 , f+ = 0.58 and < C, >=

0.0003.

1.4

1.3

1.2

1.1

1

0.9

0.8

0.7

0.6 p\. / " /

I'\\\ /// \. /0.5 '\-. /

k /

25O

- c,]L_7...... col,

t\ I_'\ 1 4\

\

f

"\ / .!

\ / \ r

, I ,"{ , , I ,_J , J500 750 1000

tLO0005

FIG. 10 Lift and Drag history of time accurate

calculations of TAU0015 airfoil for M=0.15,

Re = 1.2 x 10 ", _ = 240 , f+ = 0.58 and < C, >=0.0005.

10

REPORT DOCUMENTATION PAGE For_App,ovedOMB No 0704-0188

Public reporting burden for this collection of information is estimated to average t hour per response, including the time for reviewing instructions, searching existing datasources, gathering and maintaining the data needed, and completing and reviewing the collection of information Send comments regarding this burden estimate or any otheraspect of this collection of reformation, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations andReports, 1215 Jeff_son DavisH_ghway`Suite12_4_Adingt_n_vA222_2-43_2`andt_the_ice_fManagementandBudget`Paperw_rkReductionPr_ect(_7_4-_88)_Washir_ton, DC 205031. AGENCY USE ONLY (Leave blank) 2. REPORT DATE

December 1999

4. TITLE AND SUBTITLE

Active Control of Flow Separation Over an Airfoil

6. AUTHOR(S)

S. S. Ravindran

7. PERFORMINGORGANIZATIONNAME(S)ANDADDRESS(ES)

NASA i.angley Research Center

Hampton, VA 23681-2199

9. SPONSORINCVMONrrORINGAGENCYNAME(S)ANDADDRESSEES)

National Aeronautics and Space Administration

Washington, DC 20546-0001

3. REPORT TYPE AND DATES COVEREDTechnical Memorandum

5. FUNDING NUMBERS

WU 706-32-21-02

8. PERFORMING ORGANIZATIONREPORT NUMBER

1,-17932

10. SPONSORING/MONITORINGAGENCY REPORT NUMBER

NASA/TM- 1999-209838

11. SUPPLEMENTARY NOTES

Ravindran: NRC-Resident Research Associate, Langley, Research Center, Hampton, VA

12a. DISTRIBUTIOWAVAILABILITY STATEMENT

I Inclassified-Unlimited

Subject ('ategory 02 Distribution: Nonstandard

Availability: NASA CASI (301)621-0390

12b. DISTRIBUTION CODE

13. ABSTRACT (Maximum 200 words)

Designing an aircraft without conventional control surfaces is of interest to aerospace community. In thisdirection, smart actuator devices such as synthetic jets have been proposed to provide aircraft maneuverability

instead of control surfaces. In this article, a numerical study' is performed to investigate the effects of unsteadysuction and blowing on airfoils. The unsteady suction and blowing is introduced at the leading edge of the

airfoil in the form of tangential jet. Numerical solutions are obtained using Reynolds-Averaged viscous

compressible Navier-Stokes equations. Unsteady suction and blowing is investigated as a means of separation

control to obtain lift on airfoils. The effect of blowing coefficients on lift and drag is investigated. The

numerical simulations arc compared with experiments from the Tel-Aviv University (TAU). These results

indicate that unsteady suction and blowing can be used as a means of separation control to generate lift onairfoils.

14. SUBJECT TERMS

Active Flow Control; separation control; time-accurate CFD;

Oscillatory control; synthetic jets

17. SECURITY CLASSIFICATION 1B, ,_PUUHITY CLASSlFK;ATIUN 19. _I::(.;UHIIY CLA_SIPICATIUNOF REPORT OF THIS PAGE OF ABSTRACT

Unclassified Unclassified Unclassified

N,_N 7540-O1-280-5,_U0

15. NUMBER OF PAGES

1516. PRICE CODE

A0320. LIMHATIUN

OF ABSTRACT

UL

:standard Porto 298 (Flev. 2-89)Prescribed by ANSI Std. Z-39-18298-102