Wall-interference assessment and corrections for transonic NACA 0012 airfoil data from various wind tunnels

NASA

Technical

Paper

- April 1991

Wall-Interference Assessment

and Corrections for Transonic

NACA 0012 Airfoil Data From

Various Wind Tunnels

Lawrence L. Green and Perry A. Newman ........ lr_f _L

NASA

Technical

Paper

Wall-Interference Assessment

and Corrections for Transonic

NACA 0012 Airfoil Data From

Various Wind Tunnels

Lawrence L. Green and Perry A. Newman Langley Research Center Hampton, Virginia National Aeronautics and Space Administration Office of Management Scientific and Technical Information Division Summary been somewhat more difficult and time-consuming than initially expected from similar previous expe- A nonlinear, four-wall, post-test wall-interference rience with WIAC applications to slotted-wall data.

assessment/correction (WIAC) code has been devel- This is apparently due to a sensitivity within the cor- oped for transonic airfoil data from solid-wall wind rection code to the details of the pressure distribu- tunnels with flexibly adaptive top and bottom walls.

tions for the high blockage-ratio tests considered in The WIAC code has been applied over a broad range the adaptive-wall tunnel.

of test conditions to four sets of NACA 0012 air- Introduction foil data from two different adaptive-wall wind tun- nels. The data include many test points for fully Wind-tunnel wall interference for two- adapted walls as well as numerous partially adapted dimensional-airfoil data arises from a variety of and unadapted test points, which together represent sources. First, the top and bottom tunnel walls many different model/tunnel configurations and pos- confine the vertical flow and may induce blockage, sible wall-interference effects.

flow-angularity, blockage-gradient, and streamline- Small corrections to the measured Mach num- curvature interferences as described classically in ref- bers and angles of attack are obtained from the erences 1 and 2. Second, the boundary layers on all WIAC code even for the fully adapted data. Larger four tunnel walls interact with the model pressure corrections to the Mach number and angle of at- field to influence the blockage interference and add tack axe generally obtained for partially adapted and three-dimensionality to the flow. Third, shock waves unadapted data. The corrections are applied to de- spanning the test section in transonic flow may inter- termine new free-air flow conditions that should be act with the tunnel sidewall boundary layer to induce associated with the measured data; the new Mach flow separation in the model/sidewall junction region number is also used to renormalize the measured independent of any top and bottom wall interaction.

pressure coefficient data for comparisons with other Fourth, turbulence, noise, and temperature fluctua- corrected or interference-free data. For most of tions within the wind-tunnel circuit may cause pre- the cases investigated, the corrections improve the mature flow transition or the separation of model or correlation among the various sets of airfoil data tunnel-wall boundary layers.

and simultaneously improve the correlation of the All these factors adversely affect airfoil data mea- data with calculations from a two-dimensional, free- sured in wind tunnels and point to the need for a re- air Navier-Stokes code. For several cases, however, liable and robust means to eliminate and/or correct three-dimensional effects, possibly due to flow sepa- wind-tunnel interference. Certainly, ventilated- and ration on the airfoil or the tunnel walls, undermine adaptive-wall experimental test sections significantly the success of the correction scheme and can lead to reduce some aspects of transonic wall interference poorer correlation among data sets.

compared with that encountered with solid, fixed- The WIAC corrections for fully adapted data geometry wind-tunnel walls. Simultaneously, many from two differently sized models in the same useful computational schemes have been developed adaptive-wall tunnel are somewhat different in mag- to assess and correct some aspects of wall interfer- nitude, but the corrections generally improve the cor- ence in wind-tunnel data. It should not be expected, relation between the data sets. This demonstrates however, that all wind-tunnel wail interference can the effectiveness of the sidewall boundary-layer ap- be removed from airfoil data solely through the proximation used in the corrections, which includes use of ventilated- or adaptive-wall test sections, or the model aspect ratio. The WIAC corrections for solely through the use of wall-interference correction fully adapted data from two similar-sized models schemes. Instead, the combined use of ventilated- in two different adaptive-wall tunnels are shown to or adaptive-wall test sections and wall-interference be quite small. This shows the effectiveness of the assessment/correction (WIAC) schemes may be the most practical route for obtaining airfoil data that adaptive-wall tunnels in reducing wall interference for properly sized models. The WIAC corrections for axe nearly free of interference.

airfoil data taken in test sections with fully adapted Such a combined experimental/computational ef- walls are shown to be significantly smaller than those fort toward obtaining interference-free transonic air- for comparable airfoil data from test sections with foil data appeared in reference 3, which described straight, slotted walls. This indicates, as expected, a the nonlinear, two-wall, post-test WIAC program lesser degree of wall interference in the adaptive-wall TWINTAN (applicable to airfoil data from wind tunnels relative to the slotted-wall tunnels. Despite tunnels with ventilated top and bottom walls) and which utilized measured airfoil and wind-tunnel data the successes shown in correcting a broad range of data, application of the WIAC code to these data has in the construction of its boundary conditions. In references 4, 5_and 6, this scheme wasimproved airfoil data, evendata from fully adaptedtunnels, to accountfor four-wallinterference by including and (4) the effectiveness of the TWNTN4AWIAC a simplemodel for the tunnel sidewallboundary codein eliminatingdifferentkindsandseverities of

layer(SWBL)based on reference 7. The improved wall interference. The TWNTN4AWIAC program

is usedto correlateairfoil data from up to seven

program,calledTWINTN4, wasappliedto a lim-

ited amountof data from the Langley0.3-Meter differentsources of variousquality levelsacrossa Transonic Cryogenic Tunnel/slotted-wall testsection broad rangeof test conditionsfor differentmodel

(0.3-mTCT/SWTS), and comparisons weremade sizesandfromdifferenttunneltypesandsizes.The

with free-air calculations fromtheGRUMFOIL aero-

TWNTN4AWIAC program is alsoapplied to a rea-

sonable amount ofdatawith certainwall-interference

dynamic analysis codedescribed in reference 8. The

TWINTN4 programwasthen incorporated within effects peculiarto an adaptive-wall tunnel.

a semiautomatic procedure described in reference 9, Symbols and Abbreviations

whichincludeddata preprocessing, the TWINTN4

WIAC program,a collection of corrected data into

AWTS adaptive-wall test section

a database,andcomparisons with the GRUMFOIL

BC boundary condition

code. This WIAC procedure wasthen appliedto

a muchlargersampling of 0.3-mTCT/SWTSdata

B-S SWBL Barnwell-Sewall SWBL

in reference 10, whichshowed that the WIAC cor-

approximation

rections significantly improved the correlation of the

model span or test section various setsof lift-curvedata.

width

The TWINTN4 programof the WIAC proce-

durewasmodifiedin reference 11to includean im-

bit model aspect ratio

provedsidewallboundary-layer modelbasedupon

pressure coefficient

cp

reference 12, whichincludes the modelaspect-ratio

c model chord

effect; the improved program wasapplied to selected

data fromthe 0.3-mTCT/SWTSfor two different-

Cd drag coefficient

sizedmodels.It wasshown in reference 11that the

lift coefficient cl

model aspect ratioplayed animportantrolein corre-

latingthe drag-rise datafor the different-sized mod- lift-curve slope

Clc,

els,but that the quality of the corrections available

normal-force coefficient an

from the TWINTN4 codequantifiably deteriorates

EDM as the lift coefficient is increased in transonic flow.

empirically correlated Davis- Moore correction

Theprogram wasfurthermodified in reference 13to

account for wall interference within a tunnelhaving

H tunnel-empty SWBL shape

flexiblyadaptable top andbottom walls;the mod-

factor at model location

ified program,calledTWNTN4A, wasappliedto

h/c ratio of tunnel half-height to

limited datafromboth simulated andrealadaptive

model chord

andventilatedwallsfor an NACA0012airfoil. The

TWNTN4Aprogram wasapplied in reference 14to a

k2 Murthy aspect-ratio factor

verylargesampling ofNACA0012 dataovera broad

defined by equation (4)

range oftestconditions andpossible wall-interference

AM Mach number correction

effects; comparisons weremadein reference 14with

the free-aircalculations for flow aroundan NACA

tunnel Mach number MT

0012airfoilcomputed using the finite-volume, multi-

free-air Mach number Moc grid Navier-Stokes solver described in reference 15.

MSWBL Murthy SWBL approximation

Usingthedataofreferences 13and14,thepresent

paperis intended to showclearlythe followingfour

N-S Navier-Stokes

pointspertinentto airfoil data from adaptive-wall

magnitude of total velocity q

wind tunnels:(1) t'heeffectiveness of fully adapted

walls in eliminatingmost wall-interference effects Rc

Reynolds number based on model chord compared with partiallyadapted or unadapted walls,

(2) the effectiveness of fully adapted wallsin elimi-

RMS root mean square

natingmostwall-interference effects compared with

S SWBL coefficient defined by

typical transonicslotted-walltunnels,(3) the im-

equation (3)

portanceof TWNTN4A WIAC corrections for all

imposed: measured wind-tunnel data for the in-

SWBL sidewall boundary layer

tunnel solution, and unbounded flow and source/ slotted-wall test section

SWTS

doublet distributions for two different free-air solu- tions. The WIAC code includes two options to ac-

TCT

transonic cryogenic tunnel count for the influences of the sidewall boundary

TPT

transonic pressure tunnel layer in the in-tunnel solution.

transonic small-disturbance

TSDE The lift and drag coefficients are constraints

within this procedure; values of the lift and drag coef- equation ficients are changed only because of renormalization u,U components of total velocity effects. The Mach number and angle-of-attack cor- in x-direction (see eqs. (5) rections, AM and Aa, respectively, that result from and (2), respectively) the TWNTN4A WIAC code are added to the experi- mental values to obtain the "interference-free" condi- v component of total velocity in tions for the measured airfoil data. Of course, for the y-direction corrections to be considered valid, all calculated solu- WIAC wall-interference assessment/ tions must be sufficiently well-converged and several correction measures of correctability must be below specified tolerances, as discussed later. Under these circum- x streamwise direction of TSDE stances, data from different-sized models and differ- y normal (vertical) direction in ent tunnel configurations should collapse to a com- TSDE mon curve, or a family of curves, at the corrected flow conditions. Not all airfoil data will be correctable by a angle of attack this means alone; the interference may be too large or Aa angle-of-attack correction not adequately modeled (i.e., boundary-layer separa- tion, vortex interactions, and extreme Mach number aT tunnel angle of attack or downwash gradients through the test section).

It is important to realize that the TWNTN4A 7 ratio of specific heats WIAC program represents a significant improvement 5" tunnel-empty SWBL displace- over classical wind-tunnel correction methods such as ment thickness those summarized in reference 1. Classical correction methods generally give reasonable trends for sub- flow angle with respect to critical flows at low-to-moderate lift coefficient (cl) x-direction levels, but these methods tend to fail if the flow is _W tunnel wall inclination with transonic or at high c t. Also, most classical meth- respect to x-direction ods are linear and use linear homogeneous boundary conditions (such as ref. 2, for example), whereas the A coefficient in TSDE defined by TWNTN4A program solves for nonlinear transonic equation (2) effects and includes a higher order term to enhance ¢ dimensionless disturbance the transonic modeling. Classical correction meth- velocity potential ods may rely on empirical correlations and cannot correct for angle-of-attack (a) biases that are due to 2-D, 3-D two- and three-dimensional, misalignment of the airfoil and tunnel reference lines; respectively on the other hand, measured data are used in both the exterior and interior boundary conditions of the WIAC Procedure TWNTN4A program. Classical correction methods do not account for sidewall boundary-layer (SWBL) Overview effects, whereas the TWNTN4A program has two op- tions to approximately account for some influence of The TWNTN4A program is a nonlinear, four- the test section flow-field interaction between side- wall, post-test VIIAC code applicable to transonic airfoil data obtained in wind tunnels where the top wall boundary layer and model pressure. Some essen- and bottom walls can be slotted, porous, or shaped tially classical correction methods use measured data in the boundary conditions of linear-flow solvers; to reduce interference. The program solves sev- such a method was used during the acquisition of erai boundary-value problems subject to the two- some of the data shown herein to determine the dimensional (2-D) transonic small-disturbance equa- magnitude of residual corrections as wall adaptation tion (TSDE) but with differing boundary conditions proceeded. Thesecorrection methods, however, do a model aspect ratio of 1.0. The data presented in not accurately represent the nonlinear behaviors of this paper include cases with both free transition and

the tunnelflowfieldandviscous interactions within

transition fixed at 5 percent of the model chord; the the test section.

data are shown, however, with different symbols for free and fixed transition.

Wind-Tunnel Data

The models were instrumented in the streamwise The TWNTN4A WIAC code for transonic, direction with pressure taps on both the upper and

adaptive-wall airfoiltunnels requires asinputthefol-

lower surfaces. In addition, the upper surface of each

lowingtwo dataarrays: (1) the coordinates of the

model was instrumented with three spanwise rows of

pressure tapsandthe corresponding values of pres-

pressure taps. The top and bottom walls of the test

surecoefficient Cp along the two bounding outer sur-

section contained a streamwise row of pressure taps faces (generally the upper and lower walls of the along the centerline of each wall. For the small-chord test section) extending fore and aft of the model, model (6.5-in.), these pressure taps extended from and (2) the coordinates of the pressure taps and the 4.3 chords ahead of the model to 5.1 chords behind corresponding values of Cp on the model upper and it. For the large-chord model (13-in.), pressure taps lower surfaces. The WIAC code also requires as in- extended from 1.9 chords ahead of the model to put the values of M T, a T, cl, Cd, Rc, and the tunnel- 2.3 chords behind it. Figure 1 shows a streamwise empty SWBL parameters _i* and H at the model vertical plane through the test section appropriate to location. Also required as inputs are measurements the large-chord model (hie ._ 0.5) with walls adapted of the vertical (upwash) velocity component along at M T _ 0.5 and a T ._. 4 °.

the forward face of the test section near the upper and lower bounding surfaces; in practice, these veloc- The TWNTN4A program was also applied to a ity components are rarely measured, which compli- limited amount of NACA 0012 data reported in ref- cates the correction process by introducing a global erence 20 from the ONERA/CERT T2 adaptive-wall iteration to deduce these velocity components and tunnel in Toulouse, France. The 15.75-in-wide by establish the proper computational inflow boundary 15-in-high wind tunnel was instrumented with pres- condition.

sure taps on the centerline of the top and bottom Corrections using TWINTN4 for a limited walls that extended 4.7 chords ahead of the model amount of data taken on a 6-in-chord NACA 0012 and 2.4 chords behind the model. The 5.9-in-chord airfoil in the 8-in-wide by 24-in-high 0.3-m model was instrumented with a row of pressure taps TCT/SWTS (h/c = 2.0) have been previously re- in the streamwise direction along the model center- ported in reference 10; the entire uncorrected data line; no report was given, however, of any spanwise set has been published in references 16 and 17. Two rows of pressure taps on the model. All data from the NACA 0012 airfoils with chords of 6.5 and 13 in.

ONERA/CERT T2 test were with transition fixed at have since been tested in the 0.3-m TCT with its 5 percent of the model chord. The ONERA/CERT nominally 13- by 13-in. adaptive-wall test section.

T2 model had h/c = 1.27 with a model aspect ratio (These unpublished data were provided by E. J. Ray, of 2.7.

R. E. Mineck, S. W. D. Wolf, W. G: Johnson, Jr., The data taken in the 0.3-m TCT are written on a and A. S. Hill, all of the Langley Research Center.)

Timeliness and ease of access to these data, plus the file that must be preprocessed (ref. 9) in order to se- experience gained through prior application of the lect and format the proper inputs for the TWNTN4A WIAC code to the slotted-wall data for this airfoil, WIAC code. The preprocessor plots the model and led to the NACA 0012 airfoil being selected for this wall pressure coefficient data, as well as the spanwisc study of transonic wall-interference assessment and drag-rake measurements. Some examples of these corrections.

preprocessor plots are shown in this report to allow In the 0.3-m TCT/AWTS, the top and bottom for qualitative assessments of the 2-D character of the tunnel flow field as no flow-visualization techniques walls are deformable using motor-driven jacks un- were employed in any of these tests. Although these der the control of one (refs. 18 and 19) of several preprocessor plots are useful, specific features of the possible adaptation strategies. The straight, solid sidewalls include turntables that rotate the model, in flow field, such as separation bubbles on the model or tunnel walls, cannot be detected with this lim- this case, about the 50-percent-chord location. The 6.5-in-chord NACA 0012 model installed in the 13- by ited amount of data. Data from the ONERA/CERT T2 test were manually digitized from the tables and 13-in. 0.3-m TCT/AWTS has h/c = 1.0 with a model charts of reference 20 and put into proper format for aspect ratio (b/c) of 2.0. The 13-in-chord NACA 0012 model in the same facility has h/c -- 0.5 with input to the WIAC code.

The TWNTN4A WIAC Code globallyiterative(multiple-pass) correction scheme, asdiscussed in references 9-11,13,and14.

As described in reference 5, a dimensionless dis-

turbance velocitypotential¢ is usedto allowfordif-

In the TWNTN4A WIAC code,the 2-D tran-

ferentfar-fieldvelocities. The 2-D, transonic small-

sonicsmall-disturbance equation is solvedfor three

disturbance equation is written in the form

distinct setsof boundaryconditions.A simplified

description of the solutionmethodology follows, but

morecomplete detailsaregivenin reference 5. The ACxx + Cyv = 0 (1)

first boundary-value problem is aninverse oneusing

where

the measured wind-tunneldata (mentioned earlier)

as boundary conditions;it attemptsto numerically

modelthe in-tunnelflow field. If no normalveloc-

A=I-M2-(7+I)M_-_-_¢x 1+2-_¢z +S

ity components havebeen measured at theupstream

(2)

faceofthe testsection, theyareapproximated from

The quantity U R is the velocity at MT, whereas Uoc

the wall shape andusedin theinflowboundary con-

is the velocity at Mc¢. It should be noted that the

dition. This first solutionresultsin the determina-

classical TSDE includes only the first three terms of

tion of anequivalent inviscidbody,assensed by the

equation (2); the resulting ¢2 x term is included to

tunnelflow field,consisting of the actualmodelge-

enhance the TSDE modeling of the sonic condition.

ometryandintegrated effects of the modelandwall

The last term in equation (2) is introduced to model boundary layersandtheir interaction with shocks.

the effect of the sidewall boundary layer. This term,

The secondsolutionis actually a sequence of

called the SWBL influence coefficient, includes three

converging boundary-value problemsto determine

options when defined as

the free-airflowfieldaroundthe equivalent inviscid

body.Duringthis process, thefree-stream boundary

conditions areincrementally changed in an attempt

2_f* (2 + 1M_)[sink_k2)] (3)

s=-v \ E

to minimizethe root-mean-square (RMS)difference

between the velocitydistributioncomputed with the

where b is the model span and 6" and H are the

free-air modelandthe velocitydistributiondeduced

tunnel-empty sidewall boundary-layer (SWBL) dis-

fromthepressure coefficient datameasured with the

tunnelmodel.

placement thickness and shape factor, respectively.

The Murthy SWBL aspect-ratio factor k2 for the ap-

The third solutionis requiredin orderto deter-

proximation model (ref. 12) is

minethe "classical-type" wall-induced perturbation

velocityfield. The solutionfor the tunnelflow per-

turbationdueto the modelusesthe corrected free-

k2 = r(1 - M2)b (4)

air boundary conditions determined fromthe second

c

(previous) solutionandanairfoilboundary condition

where c is the model chord. Murthy obtained the

that matches thein-tunneldoubletstrengthandvor-

ticity distributions.This third solutionis thensub- form of S (see eq. (3)) by considering the subsonic flow past a wavy sidewall. In equation (4) the wavy

tractedfromthein-tunnel solution(thefirstsolution)

sidewall wavelength has been set to c (the model

to determine the wall-induced perturbation velocity

chord) in k2, which approximates the SWBL response

fieldin regions of interest, suchasalongthe coordi-

to the model pressure field at the model. Note that nateline containing the modelslit.

if k2 is set to zero (i.e., b/c = 0 or MT = 1), the

Thecomputed, equivalent, inviscid-body camber

ratio of k2/sinh (k2) is unity and S is the Barnwell-

linemaydistort duringthis process to accommodate

Sewall sidewall factor as given in reference 7. Also,

the removal of the wail-induced perturbation.If no

if 6" is taken to be zero, the sidewall term drops

normalvelocitycomponents havebeenmeasured at

out and only a two-wall (top and bottom) correction

the upstream faceof the test section, it is assumed

is applied. For four-wall corrections, the sidewall

that the camber line of the equivalent inviscidbody

approximations can be applied either sequentially

shouldalignoverthe forwardpart of the airfoilwith

(2-D/2-D) or unified (3-D/2-D). (See ref. 10.) For

that oftheactualmodel geometry. This,in turn,may

results in the present paper, the unified procedure

requiresubsequent correction passes usingupdated

has been used. Only corrected results using the

normalvelocitycomponents in the TWNTN4Ain-

Murthy sidewall boundary-layer (MSWBL) approxi-

flowboundary condition to alignthe computed body

mation will be shown in this paper, based on conclu- shape as well aspossible with the modelgeometry.

sions detailed in references 11 and 14 which show

The threesolutions arethenrepeated, comprising a

the Murthy approximation to be superiorto the

ences 9-11, 13, and 14 that several global iterations Barnwell-Sewall or two-wallapproximations.

or passes through the correction code were required The grid for all three processes is generally using sequentially updated values of v at the inflow corners.

stretched Cartesian, although the airfoil boundary

condition is applied in a region of uniform stream- These successive passes, in general, improve the wise spacing. Outer boundary conditions are applied overall agreement between the actual tunnel flow on rectangular contours surrounding the airfoil slit.

and the TWNTN4A-computed tunnel flow by better Placement of the outer boundary conditions and the matching some average value of v along the inflow information used to construct the outer boundary face of the computational tunnel domain. Changes to conditions are different for each of the three small- these inflow corner values of v are made using a flow- disturbance solutions previously discussed. Figure 2 alignment criterion over the forward part of the airfoil shows a schematic diagram of the TWNTN4A grid, and assuming that the magnitude of the velocity but only the upper half-plane is shown.

component v at a point upstream of the airfoil is For the in-tunnel solution with shaped, solid linearly related to the lift coefficient (valid for low- walls, the effective outer boundary condition is lift cases). For the high-lift cases, or those cases with transonic flows where shock waves lie on the forward = q(Cp) cosO (5) part of the airfoil and distort the equivalent body, these updated normal velocity components may be where u is the x-component of velocity (also equal extrapolated from lower lift data on the same polar to 1 + Cz), q is the total velocity determined from plot. More discussion of this iteration process is given the measured pressure coefficient, and 0 is the flow in references 10, 11, 13, and 14.

angle. It is assumed herein that the flows on the The downstream boundary condition for the in- top and bottom test section walls are attached and tunnel solution is the same as that described in ref- that 0 _ 0w. The TW_TN4A WIAC program spline erence 5. The reference and far-field Mach numbers fits the input ordinates of the top and bottom tun- are set to the measured tunnel Mach number for the nel walls and then interpolates at streamwise grid in-tunnel solution; therefore, UR ---- Uc_ = UT, which points. The wall tangent angles are found and used simplifies equation (2). The airfoil boundary condi- tion for the in-tunnel solution uses measured wind- in equation (5) to modify Kemp's original in-tunnel tunnel data as described in reference 5. An effective wall boundary conditions (ref. 5). Since the wall adaptation process may result in the top and bot- inviscid body is deduced from the in-tunnel solution tom walls being unequally spaced above and be- that includes the real airfoil geometry plus integrated low the airfoil, the TWNTN4A program searches for effects of the model boundary layers, wall boundary the grid locations in the normal direction that most layers between the top and bottom walls, and any closely approximate average upper and lower wall po- shock interaction with the boundary layers.

sitions. A Dirichlet boundary condition is applied The effective inviscid body is then used in a along straight lines at these locations; it is obtained free-air search (constrained optimization as described by integrating the x-derivative of the disturbance po- in ref. 5) in which the free-stream Mach number tential given by and angle of attack are varied until the computed velocity distribution at the body surface best fits ¢5 = q(Cp) cos Ow - 1 (6) the distribution deduced from the measured Cp data.

Note that this measured Cp must be re-reduced (i.e., The upstream boundary condition for the in- shifted and renormalized) using the WIAC-computed tunnel solution is a fourth-degree polynomial for ¢ free-stream Mach number and static and dynamic in terms of the normal coordinate y. It is uniquely pressures. The RMS error in this matching is taken determined with the five conditions: (a) ¢ -- 0 at as a measure of the data correctability. The Mach number and angle-of-attack corrections that result the upstream lower wall, (b) and (c) Cy -- v at the from this free-air search are added to the measured upstream top and bottom walls, respectively, as an conditions, and then the results are taken to be the input quantity assumed to be measured in the wind corrected flow conditions for the re-reduced (shifted tunnel, and (d) and (e) Cyy at the upstream top and and renormalized) airfoil data.

bottom walls, respectively, as determined from the governing equation (1) by expressing Czx in terms WIAC Validation of streamwise differences in the measured upstream wall pressure coefficients. As no measurements of v For the TWNTN4A corrections to be assumed were made at the inflow face of the test sections for valid, several criteria must be met. First, all cal- the data considered to date, it was found in refer- culated results must be sufficiently well-converged.

Second, theaccuracy of thein-tunnelflowfieldmod-

corrections. A second measure of goodness for the eling mustbe suchthat the globalpropertiesare data correctability is the flow-alignment criterion at well-represented. Third, several measures of good- the airfoil, as mentioned earlier. This criterion is

ness for thedatacorrectability mustbewithin spec-

used in adjusting the upstream in-tunnel boundary ifiedtolerances.

condition, as discussed previously. This correlation With respectto the convergence, both the in- generally holds for low lift coefficients and cases for

tunnelandthe free-airsearch solutions mustbecon-

which no shock is found on the forward part of the airfoil.

sidered. Input parameters setthe convergence crite-

rion to be met. At eachgrid point, the change in

There are two further expectations of valid cor- the disturbance potentialper iterationwaslessthan rections. The first of these is that data sets from dif-

0.5x 10 -5 forthe present results.In addition,forthe

ferent model/tunnel configurations, but at the same nominal tunnel Mach number, angle of attack, and

free-airsolution,the change in free-stream velocity

periterationwasspecified to belessthan0.1× 10 -°.

Reynolds number, should collapse (when corrected)

Usinga vertical-line over-relaxation scheme to solve

to a common curve or family of curves. The sec- ond is similar to the first: that corrected data should

equation (1),mostofthein-tunnelsolutions converge

in under100iterations, but some cases require 200or agree with interference-free data at the same flow 300iterations to converge. Generally, cases that are conditions. Thus, corrected or interference-free data for the same airfoil should have the same lift coeffi-

not correctable eitherwill not converge thein-tunnel

solutionat all or they maytake manyhundreds of cient versus angle-of-attack behavior and the same iterationsto converge. Thefree-airsearch generally drag divergence behavior, independent of whether

takesanywhere from 300to 700iterationsto con-

the data were measured in a straight solid-wall tun-

verge,althoughsome cases take morethan 2000it-

nel, a slotted-wall tunnel, an adaptive-wall tunnel,

erations to converge. Stopping the calculations be-

in free air, or were calculated numerically. That is, foretheyreach convergence is generally foundto give there is only one interference-free flow field for a given

questionable corrections. It shouldbenotedthat no

airfoil at a given Mach number, angle of attack, and

attemptwasmadeto optimizethe WIAC codeexe-

Reynolds number, assuming that the flow is steady cution;significant improvements in code performance and not separated. The WIAC correction is consid- ered valid if it eliminates different kinds and different beyond that statedmaybepossible.

With respectto the accuracy of the in-tunnel severities of wall interference. The problem for tran- solution,it is not possible to resolve finedetailsof sonic flow comes in finding "interference-free" data the flowfield. Certainly, by current computational to compare with the WIAC-produced corrections, standards for fluid dynamics, the WIAC resolution of since both numerical and experimental techniques in- the flow field is very coarse. Typically, the transonic volve, to some degree, different kinds of boundary interference.

small-disturbance equation is solved on a free-air grid having 73 × 44 points; fewer points are generally used For the data presented in this paper, compar- to model the in-tunnel flow field depending on where isons are made with the Swanson/Turkel 2-D Navier- the wall and inflow/outflow boundaries occur with Stokes code described in reference 15; this code is respect to the airfoil.

taken to represent the state-of-the-art in interference- The RMS difference between the free-air surface free viscous numerical solutions for the airfoil prob- lem. Previous comparisons in references 6 and 10 velocities (at the corrected Mach number and angle of attack) and the surface velocities deduced from were made with the GRUMFOIL code (ref. 8); that the measured pressure coefficient distribution, re- program uses a conservative full-potential solver for reduced at the corrected free-stream conditions, is the inviscid calculations that has been shown by taken as one measure of the goodness for data cor- Salas and Gumbert (ref. 21) to admit nonphysical rectability. Another criterion could be used that solutions when shocks are present in the flow field.

would place less emphasis on the rapid gradients at A similar problem occurs in all conservative full- the leading and trailing edges and near shocks. Nev- potential programs including the flow solver used ertheless, some quantifiable parameter consistently in references 13 and 14 to simulate the inviscid, 2-D airfoil data and its "interference-free" reference produced by the WIAC code for a wide variety of data. Even the TWNTN4A WIAC code itself uses cases is most useful in assessing the converged cor- the conservative formulation for the transonic small- rections. Typically, this measure of goodness would be better for low-lift cases and becomes poorer as the disturbance equation which may exhibit similar be- lift levels increase as shown in reference 11. However, havior. Usage has been restricted to low transonic no quantitative upper threshold has been established Mach numbers and lift levels where some errors may to clearly distinguish valid corrections from invalid exist, but where the magnitude of this problem is not Shown first in this paper are data from models of

severe. The resultsfromthe WIAC codeappearto

bereasonable corrections wellintothe transonic flow two different sizes in the same adaptive-wall tunnel at chord Reynolds numbers of 9 × 106 and 15 × 106;

regime.To the authors'knowledge, thesenonphys-

ical solutions havenot occurred in the TWNTN4A several of these figures include special data obtained to allow for the study of specific kinds of wall in-

program,whichusesthe measured pressure coeffi-

terference peculiar to an adaptive-wall tunnel. Fol-

cientdataas boundary conditions for the in-tunnel

lowing this are data from two different adaptive-wall

solutionand then the derivedequivalent inviscid

tunnels at a chord Reynolds number of 3 × 106. Com- shape asa boundary condition in the free-air search.

parisons are then made between data from adaptive-

Useof suchphysicalboundarydata shouldrestrict

and slotted-wall tunnels at a chord Reynolds num-

the shocklocationandstrengthto reasonable values

ber of 9 x 106. It should be kept in mind, however,

in thepotentialsolution.Some of these pointsareil-

lustratedin thelift-curveresultsfor theNACA0012 that although the chord Reynolds number has been matched for these results, the unit Reynolds number, airfoil at M T _ 0.80 presented in reference 10.

which influences the growth of the wall boundary lay- Other comparisons are made with airfoil data ers, is different between the various data sets. Also, from the 0.3-m TCT/SWTS and WIAC corrections the model aspect ratios that influence the streamwise to these data from reference 10. The data of Harris gradients within the tunnel are different between the (ref. 22) from the Langley 8-Foot Transonic Pres- various data sets. These differences result in differ- sure Tunnel (8-ft TPT) are also used in comparisons ent SWBL behavior and in different wall-interference with the 0.3-m TCT adaptive-wall data. TWNTN4A characteristics between the various data sets.

WIAC corrections could not be made to these 8-ft TPT data since wall pressure signatures were not AWTS Data for Different Model Sizes available. However, estimates of classical block- The data and corrections presented in this section age corrections to these data (based on refs. 1, 2, illustrate the effects of model size and include a wide and 23) and to those due to sidewall effects are dis- variety of unadapted and partially and fully adapted cussed when the data are presented.

points for both the 6.5- and 13-in-chord models in the 0.3-m TCT/AWTS. The data at a chord Reynolds Results and Discussion number of 9 × 106 are the most extensive sets available and include both free and fixed (5 percent chord) Presentation Format transition points. The Murthy sidewall boundary- layer (MSWBL) approximation has been used for all The uncorrected and TWNTN4A WIAC- TWNTN4A results shown here.

corrected airfoil data and results will be presented Two conventions, based on conclusions drawn in two types of plots. Plots of drag coefficient ver- from reference 13, have been followed in correct- sus Mach number (drag curves) for a given nominal ing this large amount of data. First, all data on chord Reynolds number and lift coefficient will be the 13-in-chord model have received three correc- shown to describe Mach number corrections. Plots tion passes; this is generally the minimum number of lift coefficient versus angle of attack (lift curves) of passes required to adequately correct at least the for a given nominal tunnel Mach number and chord lifting cases in this data set because of the large dis- Reynolds number will be shown to describe angle-of- turbances present in the flow field. Second, most attack corrections. Table I presents a symbol key of the data for the 6.5-in-chord model have received for these figures in which the shape of each sym- only two correction passes. Some lifting data on the bol distinguishes the test data by its ratio of tun- 6.5-in-chord model previously shown in reference 13, nel half-height to chord (h/c). Also, the shading however, have received as many as three correction of each symbol distinguishes the number of global passes, whereas other data points (both lifting and correction passes applied to the data. Recall that nonlifting) from the same reference were deemed to each pass consists of three distinct TSDE solutions; be corrected after only one correction pass.

as part of each pass, upwash velocity components of Rc _ 9 x 106. Uncorrected, fully adapted, zero- the inflow boundary condition are iteratively approx- lift drag curves at Rc _ 9 x 106 are compared in imated, which serves to align the effective inviscid figure 3(a) with Navier-Stokes free-air results. The body with the actual model geometry. The lift and experimental drag coefficients are in reasonable drag coefficients are constrained to the measured val- agreement with each other up to a Mach number ues during this correction process, except for effects due to renormalization at the corrected Mach num- of 0.7, but all are above the Navier-Stokes curve.

The large-model data (circle symbols) are generally ber; the constraint on drag, however, is not strongly higher than the rest up to a Mach number of 0.74 enforceable in the TSDE approximation.

rection passes, and it may be uncorrectable because (thehighest Machnumber for thelarge-model data); of three-dimensional effects and possible separation

for Machnumbers above 0.70,thereis considerable

within the test section. Preprocessor plots for this

spread in the data, especially in the free-transition

point are shown in figure 4. The spanwise drag co-

data. In figure3(a)the bestagreement overallwith

efficient distribution from the wake-rake survey and

theNavier-Stokes curve is found in the first test entry

the pressure coefficient distribution of the spanwise of the small model (square symbols) which, though model (oblique view) are seen in figure 4(a). A large at a slightly higher drag level than the Navier-Stokes calculation, closely follows the theoretical curve even spanwise variation in the wake-rake drag coefficient is observed, although the spanwise pressure coefficient into the drag rise.

distribution varies only slightly.

WIAC corrections to this fully adapted data with the MSWBL approximation are shown in figure 3(b).

The streamwise pressure coefficient distributions At first glance it would appear that the correlation of for the model and wall are shown in figure 4(b). The the WIAC-corrected data in figure 3(b) is worse than airfoil is shown in its proper streamwise position rel- that of the uncorrected data shown in figure 3(a).

ative to the wall pressure coefficient; note, however, It is important to realize, however, that the WIAC that the vertical scales of the two pressure coeffi- corrections in this figure primarily change the Mach cient plots are different. Noticeably different mini- number associated with a particular data point and mum wall pressure coefficients are observed for the that changes to the drag coefficient occur only be- upper and lower walls and model surfaces; moreover, cause of a renormalization of the data. Thus, it the wall Cp curves are shifted considerably away from is expected that the corrections will act to improve the expected upstream asymptotic value of zero, pos- the correlation of the drag-rise Mach number of the sibly because of massive flow separation within the various data sets without significantly improving the test section. The sonic pressure coefficient for this correlation of the different levels of drag associated case is about -0.5; thus the forward sonic point oc- with the various data sets. Within these expectations curs near the leading edge in a region where the there is considerable improvement in figure 3(b) to SWBL may be thinning. The spanwise variation the large-model data (circle symbols), both in agree- in the wake-rake drag coefficient and the downward ment with the rest of the experimental data and in shift of the wall pressure coefficients from the asymp- agreement with the Navier-Stokes curve, through ap- totic value of zero are common to those second-entry, plication of Mach number corrections up to about free-transition points with M T > 0.76. The example 0.025.

shown in figure 4(b) has the largest downward shift The first-entry data of the small model (square in the wall Cp curves; however, its spanwise varia- symbols) are also in better agreement with the tion is not the worst of those points. The observed Navier-Stokes curve; Mach number corrections in this three-dimensional effect and the possible separation case are much smaller than those of the large model, within the test section are not properly accounted for indicating that this data set has much less wall inter- within the WIAC code, which assumes 2-D flow. No ference than the large-model data set. (See fig. 3(b).)

flow visualization, however, is available in the 0.3-m Also, notice that for a range of M T from approxi- TCT/AWTS to determine if separation regions are mately 0.60 to 0.70 the data points have been some- present.

what rearranged along a family of lines parallel to the theoretical drag curve. Some of the second-entry To be more specific, the Murthy SWBL approx- imation used in these calculations is based upon a data for the small model (diamond symbols) in this figure are also in better correlation with the other subsonic wavy-wall problem; the approximation can- data; however, some points of this data set have re- not properly model the 3-D supercritical flow that ceived corrections that are too large, or even in the exists when the forward sonic point occurs where the wrong direction. Generally, such points are at or SWBL is thinning (near the model leading edge), above the drag-rise Mach number where strong shock such as at high Mach numbers and high lift coeffi- waves and their interaction with the tunnel SWBL cient. Unfortunately, an analogous sonic wavy-wall are present. approximation is probably not practical and would One such free-transition, second-entry point for be of little use anyway. Those cases where the for- ward sonic point is located in a region where the the 6.5-in-chord model (diamond symbol marked SWBL is thinning result in correspondingly strong with an arrow) has received a particularly large Mach aft shocks downstream, which will generally cause number correction (and, surprisingly, for this nomi- nally zero-lift case, a large a-correction as well) that the SWBL to separate and introduce even more 3-D effects into the flow field. In these cases, it is expected throws it out of the line with the rest of the points.

The point is still badly corrected even after two cor- that only a full 3-D viscous solution will improve the

modeling ofthetunnelflowfieldsufficiently andlead

bols) appear to have a slightly lower lift-curve slope to animproved correction.

ct_ than the other data sets. The a corrections shown A varietyof special,uncorrected zero-lift drag in figure 6(b) are generally small but range up to 0.35 ° for the small model at a _ 4 °. The corrected

dataareshown in figure5(a).These datapointswere

obtained by operating the 0.3-mTCT/AWTSin un- data lie mostly on a line with nearly the same slope usualwaysto allowfor the studyof specific kindsof as, but shifted to the left of, the Navier-Stokes cal-

wall-interference effects peculiarto an adaptive-wall

culations. WIAC corrections for the highest a data tunnel. The unusual operations, discussed in more shown in figure 6(a) did not converge and the point detail in references 13and 14,includedthe follow- is not shown in figure 6(b).

ing: (1) adaptation sequences (the wallsbeingitera-

The special, uncorrected data shown in figure 6(c)

tivelyadjusted fromnominally straightcontours with

for the same Mach number and Reynolds number

datatakenat eachwallshaping step)to compare re-

include three adaptation sequences (both circles and sultsforunadapted, partiallyandfully adapted walls, squares at a_2 ° , and squares only at a_ 0°),

(2) runsin whichsomewall jacksweremadeinop-

with each displaying some variation in c I at the

erativeat the entrance andexit of the test section

same nominal angle of attack because of changes in to simulate tunnel truncationeffects, and (3) runs the wall-interference effects during the iterative wall in whichthe centerline of the tunnelwasartificially shaping. WIAC corrections to these data are shown

rotatedup 0.5 ° in the adaptation software to simu-

in figure 6(d). Most of the variation in lift coefficient at c_ _ 0 ° and 2 ° has been corrected into a variation

late an errorin the angleof attackmeasured in the

test section.The data pointsof figure5(a) include along a family of curves with the same slope as the threeadaptation sequences forthelarge model(circle Navier-Stokes lift curve, but shifted slightly to the symbols, 1_[ T .._ 0.60, 0.65, and 0.70) and one adap- left of the theoretical curve. The overall agreement tation sequence for the small model (square symbols, among the data sets and with the Navier-Stokes M T ,._ 0.60). These sequences are characterized by a curve is better than that of the uncorrected data group of like data points at the same nominal Mach shown in figure 6(c); however correlations for the number, for which the drag coefficient varies signif- small-model data are better than those for the large- icantly due to the changing wall-interference effects model data. The largest a correction is about 0.5 ° for during the iterative wall shaping. one large-model point. It is interesting to note that Also included in figure 5(a) are: (1) a group the Mach number corrections shown in figure 5(b) for the large-model points at nominally zero lift improve of fixed-transition, small-model data points (flagged diamond symbols) for which the walls were kept in a the correlation, whereas the a corrections to the nominally straight position as the Mach number was same data shown in figure 6(d) do not improve, and possibly worsen, the correlation.

varied from 0.6 to 0.74; (2) several free-transition, small-model data points (unflagged diamond symbols Lift-curve data are shown in figure 7 for M T ,._ at 2/_T _ 0.7 and 0.76) with simulated tunnel trun- 0.65 and Rc .-_ 9 x 106. The uncorrected, fully cation effects, which exhibit a small variation in adapted data in figure 7(a) are all in good agree- ment with each other and with the Navier-Stokes re- the drag coefficient at these Mach numbers; and (3) three free-transition, small-model data points sults, although the small-model data (diamonds) ap- (unfiagged diamond symbols, one at M T ,._ 0.7 and pear to have a slightly lower lift-curve slope than the Navier-Stokes data. WIAC corrections to these data two at M T ,_ 0.76) with simulated angle-of-attack errors with a drag coefficient of approximately the shown in figure 7(b) do not correlate as well as the mean value in each of these groups. The WIAC uncorrected data because the large-model points (cir- corrections to these unadapted or partially adapted cles), indicated with arrows, have not met the body- zero-lift drag data are shown in figure 5(b). The alignment criterion even after three correction passes.

various sets of drag data are all made to correlate However, the lift-curve slope of the small-model data much better with each other and with the Navier- points is in better agreement with the Navier-Stokes Stokes curve through the application of Mach number curve despite those points being shifted to the left of the Navier-Stokes curve.

corrections as large as 0.1 for the large-model data, and as much as 0.05 for the small-model data.

The uncorrected data shown in figure 7(c) include Uncorrected, fully adapted lift-curve data over a large- and small-model adaptation sequence (circles a range of angles of attack at M T,._ 0.6 and and squares, respectively) at a _ 2 ° and also one (cir- Rc _ 9 x 106 are shown in figure 6(a). The data cles) at a --_ 0 °. Corrections to the data are shown sets are in reasonably good agreement with each in figure 7(d). The correlation is considerably bet- other and with the Navier-Stokes results, although ter between the large- and small-model data (circles the first-entry data for the small model (square sym- and squares), except for one point (indicated with an corrected Mach numbers significantly higher than

arrow)that hasreceived a Machnumber correction

0.70. The Navier-Stokes results are shown in fig-

of about0.1.ThelargeMachnumber correction may

ures 8(b) and 8(d) at the nominal tunnel Mach num- indicatea largeseparation regionwithin the tunnel.

ber, whereas the corrected data should properly be

It is againnotedthat the Machnumber corrections

compared with free-air results at the corrected Mach

shown in figures3(b) and5(b) for the large-model

numbers for each test data point. The spread of the

datapointsat nominallyzerolift improve the cor-

individual corrected Mach numbers (from 0.692 to

relation,whereas the a corrections to the same data

0.760 in this case) was much broader than the spread shown in figures 7(b) and 7(d) worsen the correlation.

of uncorrected Mach numbers (from 0.697 to 0.711, It is possible that a separation region in the tunnel, also for this case) over the range of angle of attack at when considered part of the effective inviscid body as the same nominal Reynolds number. Additionally, in TWNTN4A, may produce a correctable increase the range of corrected Mach numbers was typically in the blockage while altering the the effective body skewed to higher mean values than the nominal un- cambering that undermines the a corrections. No corrected free-air Mach number. In order to more visualization techniques were employed in the 0.3-m fairly assess the corrections in this case, the data from TCT/AWTS that could validate or dispel any spec- figures 8(b) and 8(d) are repeated in figures 8(e) and ulation about flow separation within the tunnel.

8(f), respectively, with an additional Navier-Stokes Uncorrected, fully adapted lift-curve data at curve at a Mach number of 0.76 added to the plots.

M T _, 0.7 and Rc _ 9 x 10_are shown in figure 8(a).

The large-model corrected data are seen to correlate There is reasonably good correlation among the dif- better with the Navier-Stokes curve at a Mach num- ferent data sets, although most of the small-model ber of 0.76 than at 0.70, whereas the small-model data (squares and diamonds) appear to have a data correlate better with the lower Mach number slightly lower lift-curve slope than the Navier-Stokes curve, at least for low lift.

curve. Corrections to the data from the TWNTN4A Lift-curve data at M T ,_ 0.72 and Rc _ 9 x 106 WIAC program are shown in figure 8(b). Most of are shown in figure 9. Uncorrected, fully adapted the data in the range of angle of attack from -2 ° data are shown in figure 9(a) that agree quite well to 2 ° lie on one or more lines with slopes nearly the with the Navier-Stokes results. The corrected results same as the Navier-Stokes curve, but shifted to the left of that curve. In this instance, the first-entry in figure 9(b) seem to have a lift-curve slope that is too large, although there are not enough data points small-model data point (square) and the large-model to see a trend for either model. Two unadapted data data point (circle), indicated with arrows, have not points shown in figure 9(c) are made to correlate bet- met the body-alignment criterion.

ter (fig. 9(d)) with the Navier-Stokes curve through A wide variety of unadapted and partially the WIAC corrections, although the corrected Mach adapted data for the same nominal Mach num- numbers of these points are significantly higher than ber and Reynolds number are shown in figure 8(c).

the measured value.

The data include large- and small-model adaptation Uncorrected, fully adapted lift-curve data sequences (circles at a _ 1° and 2 ° and squares at M T _ 0.74 and Rc _ 9 × 106 are shown in fig- at a _ 2 ° and 4 °, respectively), points with simu- ure 10(a) that correlate well with each other and lated truncation effects (clustered diamonds on the with the theoretical curve. Corrections to the data Navier-Stokes curve), and points for which the test in figure 10(b) have slightly more scatter at a _ 0 ° section centerline was rotated up (diamonds above than the uncorrected data, whereas there is a slight the Navier-Stokes curve) in the adaptation software.

WIAC corrections to the data are shown in fig- improvement in correlation with the Navier-Stokes results at a ,-_ 2 °. Two unadapted, uncorrected data ure 8(d). The corrected small-model data (squares points at the same Mach number and Reynolds num- and diamonds) correlate better with each other along ber are shown in figure 10(c). The WIAC correction a line with nearly the same slope as the Navier-Stokes for only one of these data points is shown in fig- curve, but once again shifted to the left of it. The ure 10(d); the correction for the point at o_ T _ 2 ° did large-model data (circles) have received a corrections not converge and it is not shown in the figure. The as large as 0.8 ° , and these points do not correlate well corrected Mach number of the data point shown in with the other data or with the Navier-Stokes curve.

Most of the large-model data points shown in fig- figure 10(d) is significantly higher than the nominal tunnel Mach number.

ure 8(d), however, failed to meet the body-alignment Uncorrected lift-curve data for one adapt- criterion, and corrections for two of these points, which did not converge, are not shown in figure 8(d). ation sequence of the small model at MT _ 0.75 and R¢,-_ 9x 106 are shown in figure ll(a).

It is important to note, however, that several of Corrections to the data with the MSWBL the data points shown in figures 8(b) and 8(d) have approximation are shown in figure l l(b). Agree- correlate slightly better with the Navier-Stokes drag rise than do the uncorrected data.

ment with the Navier-Stokes curve is not as good as with the uncorrected data; however, most of these It is interesting that a significant improvement data points have corrected Mach numbers somewhat in the correlation of these interpolated drag curves above 0.75, but not as high, relatively, as those in the (circles and squares) is observed when improvement preceding corrected figures.

in lift-curve correlations for these fully adapted data Uncorrected, fully adapted lift-curve data at was not always clearly observed. It was hoped that 151 T _ 0.76 and Rc _ 9 × 106 over a range of an- with the larger amount of transonic data (includ- gle of attack are shown in figure 12(a). The two ing unadapted and partially adapted data) given in small-model data sets agree well with each other and this report (compared with that in ref. 13), the im- with the Navier-Stokes curve. Corrections to these provement in drag-rise correlation at lift could be data are shown in figure 12(b); correlation of the two even more easily seen. However, the spreads in cor- data sets with each other and with the theoretical rected Mach numbers for these diverse data, which curve is not as good as the uncorrected data. The were greater than the spreads in the uncorrected corrected data for the first entry (squares) lie on a Mach numbers, made drag-rise interpolation rather line with nearly the same slope as the Navier-Stokes uncertain, and the resulting curves showed no bet- results, but shifted to the left of it; corrected data ter correlation than those presented in figure 14(b).

from the second entry have a slightly larger lift-curve Additional drag-rise curves at lift are shown in references 13 and 14.

slope than the other data. Figure 12(c) shows many unadapted or partially adapted, uncorrected data, In general, small angle-of-attack and Mach num- including several data points with simulated test sec- ber corrections were calculated with the TWNTN4A tion truncation and with the tunnel centerline arti- WIAC code for the fully adapted data with Rc ficially rotated up 0.5 ° . Corrections to the data are 9 × 106, indicating that little wall interference is shown in figure 12(d). The correlation among the present in the fully adapted data. As expected, larger data sets and with the Navier-Stokes results is some- corrections were calculated for the many partially what improved at low lift levels; however, the points adapted and unadapted data points that were in- at _ -_ 2 ° do not correlate as well as the uncorrected cluded here to study different kinds and severities data.

of wall interference. The corrections shown in the Two uncorrected, fully adapted lift-curve data preceding figures were typically larger for the large- points at .SI T _ 0.8 are shown in figure 13(a). The chord model. Large-model data generally required corrected data shown in figure 13(b) correlate better at least three global correction passes. Small-model with the Navier-Stokes curve than the uncorrected data were generally easier to correct, and usually two data. The uncorrected lift-curve data from one passes were sufficient. The shape of the drag curves !

| small-model adaptation sequence at M T .._ 0.8 and and the point of drag divergence were made to corre- Rc _ 9 x 106 are shown in figure 13(c). WIAC cor- late better through application of the WIAC correc- tions; the level of drag measured in the tunnel was rections to tile data are shown in figure 13(d). Cor- relation with the Navier-Stokes data is improved, al- relatively unchanged through the corrections.

though there is again a fairly large spread in the Agreement among the various lift-curve slopes corrected Mach numbers.

was generally improved with corrections. Most of Uncorrected, fully adapted drag-curve data for the 0.3-m TCT/AWTS data, however, were cor- large- and small-chord models interpolated from fig- rected slightly to the left of the Navier-Stokes results.

The effects of free or fixed transition on the WIAC ures 6(a), 7(a), 8(a), and 12(a) at a lift coefficient of 0.2 are shown in figure 14(a). The large-chord corrections are not clear in these examples. The model, as observed previously, has higher drag lev- corrections tended to improve correlation among the els than the small-chord model, but it is impossible various sets of data and simultaneously improve cor- to determine its drag-divergence Mach number since relation with the Navier-Stokes free-air calculations, no drag data with a lift coefficient greater than zero unless severe three-dimensional effects or possible were obtained above M T _ 0.7 for the large-chord flow separation were present in the uncorrected data.

model The corrected, interpolated drag curve taken The 3-D effects, however, are not properly modeled from figures 6(b), 7(b), 8(b), and 12(b) at a lift coef- bythe Murthy SWBL approximation in supercritical flow, thus resulting in corrections that are either too ficient of 0.2 are shown in figure 14(b). It is clear large or, perhaps, even in the wrong direction.

that two corrected data sets (circles and squares) correlate better with each other than with the un- A surprising number of points, however, es- corrected data, but the third data set shows little pecially for the large model, either did not con- improvement. The limited corrected data appear to verge or did not meet the body-alignment criterion to which these disturbances propagate upstream in

afterseveral globalcorrection passes. This behavior

the test section. These strong disturbances may vio-

wasneitherencountered asmuchin the slotted-wall

late the assumptions in the WIAC code and the as-

datapreviously reported in reference 9 for whichthe

sumptions of linear flow exterior to the walls that are

wallpressure signatures weremuchmildernorin the

inherent in the adaptation scheme (refs. 18 and 19)

smoothlyvaryingsimulated, inviscid,adaptive-wall

used to shape the walls. Moreover, the computed

datapresented in references 13and 14. Thebehav-

equivalent inviscid body shapes resulting from the

ior suggests that the TWNTN4AWIAC program is

TWNTN4A code for these poorly corrected cases are

very sensitive to the details(includingrandomness

two to three times the actual thickness of the models.

dueto viscous effects andinstrument imperfections)

These extremely thick, effective inviscid bodies

of the pressure distributionsusedin the boundary

may indicate substantial viscous interaction and/or

conditions.Despite this sensitivity, the WIAC code

massive flow separation since the model and wall

seems capable of detecting andcorrecting evensmall

amounts of the differentkindsor severities of wall boundary layers must be considered as part of the effective inviscid body shape within this approxima-

interference, or, on the otherhand,of givingsome

indicationthat it is not correctable. tion. Once again, the forward sonic point occurs in

a region where the SWBL is probably thinning, and Rc .._ 15 × 106. The following figures show data the Murthy SWBL approximation cannot properly from the large- and small-chord NACA 0012 model model the resulting supercritical 3-D flow. None of tests of the 0.3-m TCT/AWTS at a chord Reynolds these effects is properly accounted for within the cor- number of 15 × 106. The figures include almost every rection code. Data from one adaptation sequence for data point available for the 6.5- and 13-in. models the small model are shown in figure 17(a). Correc- at this Reynolds number. The data again include tions to these data shown in figure 17(b) significantly a variety of wall-interference conditions with several improve the correlation by spreading the points along points having partially adapted or unadapted test a curve parallel to the Navier-Stokes curve.

sections; the data also include a mix of free and fixed Uncorrected, fully adapted lift-curve data at transition points.

MT _ 0.6 and Rc _ 15 × 106 are shown in fig- Uncorrected, fully adapted drag-curve data at ure 18(a). Some disagreement is observed between the three data sets. The data appear to have a zero lift are shown in figure 15(a). There is more dis- agreement among these data sets at the lower Mach slightly lower lift-curve slope than the Navier-Stokes solution. WIAC corrections to the data are shown in numbers than among the comparable data sets at Rc _ 9 × 106 shown in figure 3(a). One large-model figure 18(b). The corrected data agree better with each other and appear to have the same slope as the data point (indicated by an arrow) is obviously out of line with the rest of the data, probably due to tun- Navier-Stokes curve in the angle-of-attack range from -2 ° to 2 °, although displaced somewhat to the left nel choking. Corrections to these fully adapted data of the Navier-Stokes curve. Corrections for the high- shown in figure 15(b) generally improve the correla- est angle-of-attack cases for the small model did not tion among the various data sets up to a Mach num- ber of about 0.78; the corrections primarily change converge and are not shown in the figure. The cor- the Mach number associated with particular points, rection for one small-model point, indicated by an which tends to make them lie along curves that paral- arrow, did not meet the body-alignment criterion.

lel the Navier-Stokes curve. Although the corrected Uncorrected, fully adapted lift-curve data at points shown in figure 15(b) are all converged and M T .._0.7 and Rc _ 15x 106 are shown in fig- have met the body-alignment criterion, three small- ure 19(a). The data are in good agreement with each other and with the Navier-Stokes results. Cor- model data points and the large-model point previ- rections to these data are shown in figure 19(b); the ously noted have received large Mach number cor- rections in the wrong direction. These points are data have, in general, been moved from one side of the Navier-Stokes curve to the other side of it with indicated in figure 15(b) by arrows and exhibit a no- ticeable spanwise variation in the pressure coefficient a resulting slight increase in the lift-curve slope. It is difficult to conclude that the corrections have im- and wake-rake drag survey.

proved the correlation. The partially adapted data Preprocessor plots for the large-model data point shown in figure 19(c) include several adaptation se- are shown in figure 16. Considerable spanwise varia- quences for the small model. The corrected data tion in the wake-rake drag coefficient is observed in shown in figure 19(d) correlate better with each other figure 16(a); the spanwise variation in the measured and with the Navier-Stokes data up to stall. Correc- drag is typical of all the data points that have been tions for three small-model data points, indicated by badly corrected. In figure 16(b) note the large dis- arrows, failed to meet the body-alignment criterion.

turbances impressed upon the walls and the extent

Uncorrected,fully adaptedlift-curve data at

from the 0.3-m TCT/AWTS (unflagged squares), M T _0.76and Rc _ 15×i06 over a range of an- and some fixed-transition drag data obtained from gle of attack are shown in figure 20(a); noticeable dis- the ONERA/CERT T2 tunnel obtained by pres- agreement is observed among the small-model data sure integration (flagged, inverted triangles). Navier- at angles of attack of 1 ° and 2 °. Corrections to the Stokes results with the transition fixed are shown as data shown in figure 20(b) display an improved corre- the solid line for comparison purposes. The WIAC lation in lift-curve slope, although the corrected data corrections to these data with the MSWBL approx- points are displaced to the left of the Navier-Stokes imation are shown in figure 21(b). As expected, the curve.

fixed-transition wake-rake data agree best with the For the data presented at this Reynolds number Navier-Stokes results, but the drag-rise data point of of 15 x 106, the WIAC code again appears to gener- all the corrected curves is aligned well despite the no- ally improve the correlation in lift-curve slope among ticeable differences in the levels of the various drag the various data sets (except for those data points curves. Sample preprocessor plots for highest Mach as noted), although the corrected experimental data number data (marked with an arrow) from the 0.3-m were generally displaced to the left of the Navier- TCT are shown in figure 22. The data again display Stokes data. Corrections also generally improved the a large spanwise variation in the drag coefficient that correlation among the various drag data sets. Several is not accounted for in the WIAC code and that re- data points corrected at this Reynolds number failed sults in rather poor correlation at the highest Mach numbers.

to meet the body-alignment criterion, as noted pre- viously with the data at a chord Reynolds number of Uncorrected, fully adapted lift-curve data at 9 x 106.

M T ,_ 0.6 and Rc _ 3 × 106 are shown in fig- It should be noted from the preceding examples ure 23(a). Some disagreement is observed between that the large-chord-model test envelope was signif- the two data sets, but overall the agreement is very icantly smaller than that of the small-chord model good. The data appear to have a slightly lower lift- because of two effects. First, limitations on wall dis- curve slope than that of the Navier-Stokes solution.

The WIAC corrections to these data are shown in placement and curvature (wall safety criteria) prohib- ited testing at high angles of attack. Second, another figure 23(b). The corrected data appear to have limitation was imposed then by an earlier onset of the same slope as the Navier-Stokes curve, but the choked flow conditions within the test section for high 0.3-m TCT data are displaced slightly to the left of Mach numbers at zero lift. The data shown herein the curve and the ONERA/CERT T2 data fall nearly on the Navier-Stokes curve.

include almost every point available for the 13-in- chord model, and also most of those points available The uncorrected, fully adapted lift-curve data from both tunnel entries of the 6.5-in-chord model.

at M T _0.7 and Rc _ 3× 106 are shown in fig- Considerably more discussion about the limitations ure 24(a). Agreement is good between the data of the large-chord test envelope and their impact on sets, although the cl_ again appears to be slightly this study is given in reference 14.

less than that of the Navier-Stokes curve. Correc- tions to the adaptive-wall data with the MSWBL AWTS Data From Different Tunnels approximation are shown in figure 24(b). The cor- The following figures show data for the small- rected ONERA/CERT T2 data again lie almost on top of the Navier-Stokes curve, whereas the 0.3-m chord NACA 0012 model test from the 0.3-m TCT/ AWTS and data for a similar-sized model in the TCT data are corrected to slightly left of the Navier- ONERA/CERT T2 AWTS at a nominal chord Stokes curve with possibly a slightly higher lift-curve Reynolds number of 3 × 106. The data shown in- slope.

clude almost every data point available for the two The uncorrected, fully adapted lift-curve data at models at this Reynolds number. Only fully adapted M T ,_ 0.75 and Rc _ 3 × 10 U over a range of angle data are shown.

of attack are shown in figure 25(a). Corrections to The uncorrected zero-lift drag divergence at a the data are shown in figure 25(b). Both sets of data nominal chord Reynolds number of 3 x 106 for the have the same lift-curve slope as the Navier-Stokes ONERA/CERT T2 data and the fixed-transition curve, but the 0.3-m TCT data are again displaced to the left.

6.5-in-chord model data from the 0.3-m TCT/AWTS are shown in figure 21(a). Three distinct data The Mach number and angle-of-attack corrections sets are shown in figure 21; these include normal for Rc _ 3 × 106 were small in all cases, indicat- fixed-transition drag data obtained with a wake rake ing that little wall interference was present in these (flagged triangles, squares, and the diamond), some fully adapted cases. Since the ONERA/CERT T2 free-transition drag data obtained with a wake rake tunnel is similar to, but slightly larger than, the a lift-curve slope much nearer to the TWNTN4A-

0.3-m TCT (h/c _ 1.27 compared with 1.0; b/c _ 2.7

corrected adaptive-wall data and the Navier-Stokes compared with 2.0), it was expected that corrections solution. TWlNTN4 corrections also remove the zero would be about the same size as, or less than, those shift from the slotted-wall data. The overall corre- of the 0.3-m TCT/AWTS that were already shown to lation among the data sets is dramatically improved be small. Angle-of-attack corrections, though small, over the uncorrected data.

generally tended to improve the lift-curve-slope cor- The empirically correlated Davis-Moore correc- relation among the data sets. The displacement of the corrected 0.3-m TCT data was not observed with tions (fig. 26(c)) to the 0.3-m TCT slotted-wall data are shown in order to aid in the analysis of upcoming the ONERA/CERT T2 data and may indicate some angle-of-attack bias in the 0.3-m TCT/AWTS, such figures with slotted-wall data from the 8-ft TPT. It as a misalignment between the tunnel and the model is important to notice that the empirically correlated zero-angle reference lines. Davis-Moore corrections have nearly the same cl_ as the TWINTN4-corrected slotted-wall data, but they AWTS and $WTS Data do not remove the zero shift in the data. The fol- lowing three factors, taken together about this com- The figures discussed in this section repeat se- parison, give clear evidence that the fully adapted lected fully adapted data and WIAC corrections from data have significantly less wall interference than the the large- and small-chord NACA 0012 model.tests of slotted-wall data: (1) the uncorrected data indicate the 0.3-m TCT/AWTS at a chord Reynolds number that corrections for the slotted-wall data should be of 9× 106 taken from figures 3-14. Adaptive-wall data much larger than those for the adaptive-wall data to are compared with slotted-wall data from two dif- improve the correlation of the three data sets, (2) the ferent sources. Uncorrected and corrected adaptive- size of corrections applied to the slotted-wall data are wall data are first compared with slotted-wall data much larger than those applied to the fully adapted from the 0.3-m TCT/SWTS for which WIAC cor- wall data, and (3) the correlation of the three data rections by Gumbert and Newman were presented sets after corrections are applied is improved.

in reference 10 using an earlier version (TWINTN4) Comparisons similar to those of figure 26 are of the present TWNTN4A WIAC program. The shown in figure 27 at a nominal tunnel Mach number present TWNTN4A program includes all the previ- of 0.76. In figure 27(a) it is easily seen that the ous capability and can be used to correct data from uncorrected adaptive-wall data and the uncorrected fixed-geometry, solid- or slotted-wall tunnels or from slotted-wall data do not correlate well either with adaptable tunnels with variable porosity or flexible each other or with the Navier-Stokes curve. The walls. Empirical corrections to these slotted-wall TWNTN4A-corrected adaptive-wall data (fig. 27(b)) data also shown in reference 10 are likewise compared correlate very well with the TWINTN4-corrected with the adaptive-wall data to aid in the analysis of slotted-wall data, but both disagree somewhat with upcoming figures. The adaptive-wall data are then the Navier-Stokes results. The Navier-Stokes results compared with earlier slotted-wall data from the 8-ft tend to lie between the uncorrected adaptive-wall TPT tunnel that were reported and empirically cor- data and the corrected data sets. The empirically rected by Harris in reference 22. No WIAC correc- correlated Davis-Moore corrections (fig. 27(c)) in this tions for these slotted-wall data were possible because case lie below the Navier-Stokes curve, similar to of the lack of wall pressure coefficient data, but use- the uncorrected adaptive-wall data, but with a small ful comparisons are made with adaptive-wall data by zero-lift shift.

building upon the analysis of the 0.3-m TCT/SWTS data. Figure 28(a) shows the uncorrected drag diver- gence at lift (c I = 0.2) for large and small models Figure 26(a) shows a comparison of uncorrected lift curves for slotted-wall and fully adapted NACA with walls fully adapted (i.e., fig. 14(a)) with the 6-in- 0012 data from the 0.3-m TCT. Data for M T _ 0.7 chord-model slotted-wall data superimposed. These and R¢ _ 9 x 106 are shown. The adaptive-wall uncorrected slotted-wail data appear to agree well with the Navier-Stokes results. The drag divergence data are for only the first entry of the small-chord of the small model from the adaptive-wall tunnel model; the slotted-wall data are for a 6-in-chord model in the 8-in-wide by 24-in-high test section at may also agree as well, but it is difficult to see this the same nominal tunnel conditions. It is clear that clearly with the limited amount of data. Data for the 13-in-chord model in the adaptive-wall test sec- the uncorrected slotted-wall data have a different cl, _ tion have higher drag levels than the 6.5-in-chord- than the adaptive-wall or Navier-Stokes data. Also, the slotted-wall data have a zero-lift coefficient at model data, the 6-in-chord slotted-wall data, and the Navier-Stokes results. The TWINTN4- and an angle of attack of about 0.3 °. TWINTN4 cor- TWNTN4A-corrected drag divergence for the same rections (fig. 26(b)) to the slotted-wall data have tunnel Mach number of 0.76. Once again there is a

threetestsisshown in figure 28(b).It isclearthatthe

WIAC corrections makeall the data correlate well

noticeable difference in cto in these data sets. Good correlation is shown in figure 30(b) among the empir-

with eachotherand agree better with the Navier-

Stokes results.Theslotted-wall datainfigures 26,27, ically correlated, Davis-Moore-corrected slotted-wall data, the TWNTN4A-corrected adaptive-wall data,

and28werecorrected (ref. 10)using the B-SSWBL

and the Navier-Stokes data.

approximation, whereas the adaptive-wall datawere

corrected usingthe MSWBLapproximation. Refer- In figures 26, 27, 29, and 30 it is important to no-

ences 11and14demonstrate that thisdifference may

tice that the uncorrected, fully adapted data receive much smaller c_ corrections than either set of uncor-

have a noticeable effectonthedrag-curve corrections

rected slotted-wall data. It is worth recalling that

(with MSWBLbeingthe better of the two options

for improvingdragcorrelations), but the difference

the slotted-wall tunnels have h/c ratios of 2.0 (0.3-m should have almostnoeffectonlift-curvecorrections.

TCT/SWTS) and 1.7 (8-ft TPT); the adaptive-wall A comparison of fully adapted wall,small-model tunnel, by contrast, has an h/c ratio of 1.0 for these

lift-curvedatawith data fromthe 8-ft TPT seenin

comparisons. Simulated, inviscid 2-D results from figures29 and 30 showsthe sameformat as that references 13 and 14 for straight solid-wall tunnels in figures26 and 27. In figure29(a)a compar- show the opposite effect of h/c on the corrections.

isonis madeat M T _ 0.7 with uncorrected slotted- This clearly demonstrates that the adaptive-wM1 tun- nel can eliminate most of the wall interference present wall data from the 8-ft TPT as reported in refer- in typical transonic slotted-wall test sections.

ence 22. Noticeable disagreement in the lift-curve slope is observed in this figure among the sets of data. In figure 31(a) uncorrected drag-divergence re- TWNTN4A corrections are not possible for this set sults are shown for the data interpolated at a lift of slotted-wall data because of the lack of measured coefficient of 0.2. Agreement between the uncor- rected 8-ft TPT data and the Navier-Stokes data wail pressure signatures. The empirically correlated, Davis-Moore angle-of-attack correction for slotted- is quite good. Uncorrected 0.3-m TCT/AWTS data wall test sections assumes (among other things) con- are taken from figure 14(a) and do not agree as well with the 8-ft TPT data or with the Navier-Stokes stant openness-ratio slots of infinite length. Slot shapes in the 8-ft TPT vary a good bit over their fi- results. The disagreement between the small-model nite length so that several (perhaps) meaningful aver- data from the 0.3-m TCT/AWTS and the 8-ft TPT ages can be obtained. Both the original Davis-Moore data is not very great and both sets of data appear correction (ref. 2) and the empirically correlated, to have nearly the same drag-rise point. Corrections Davis-Moore angle-of-attack correction (ref. 23) re- to adaptive-wall data from figure 14(b) for this case quire a value for the geometric openness ratio of clearly improve the correlation among all data sets the slotted wall. For the data of reference 22, the as shown in figure 31(b). TWNTN4A corrections are 25-in-chord airfoil was placed in the 8-ft TPT test not possible for the 8-ft TPT data because of the lack section so that several average openness ratios and of measured wall pressure distributions; however, we estimate that Mach number corrections to these data their corresponding empirically correlated, Davis- Moore angle-of-attack corrections were as given in from the TWNTN4A procedure would be small, as chart A. The empirically correlated Davis-Moore re- shown in the following discussion.

sults from the 8-ft TPT for the largest and smallest The range of openness ratios discussed above angle-of-attack corrections are shown in figure 29(b).

in regard to the angle-of-attack corrections (fig- These data agree reasonably well with Navier-Stokes ure 29(b)) also gives rise to an uncertainty in the calculations and TWNTN4A-corrected adaptive-wall classical-model solid-blockage correction for Mach data.

number. All openness ratios quoted in chart A give test sections more closed than that for minimum Chart A (2-D) blockage according to the classical theory outlined in references 1, 2, and 23. The solid- blockage correction of the classical 2-D model for Openness (from Mach number varies from 0°003 to 0.001 with open- ratio ref. 22) Region of slot width averaged ness ratio at M T = 0.7, whereas at MT = 0.82 0.051 Over 25-in. chord ......... -1.55cn it varies from 0.007 to 0.002. These are compen- 0.062 - 1.67cn Effective test section length ....

sated by the sidewall boundary-layer Mach number 0.085 --1.82Cn Total slot length .........

correction that is of opposite sign. The maximum Mach number changes for the Barnwell-SewaI1 and Murthy sidewall approximations can be estimated In figure 30(a) uncorrected lift-curve results sim- from the known displacement thickness and model ilar to those in figure 29(a) are shown for a nominal found for larger ratios of tunnel half-height to model

aspectratio fcr the slotted-wall data. Thesemax-

chord (h/c), as evidenced by the large-chord model

imum Mach numberchanges are -0.008 for the

frequently requiring three or more correction passes,

Barnwell-Sewall approximation and -0.0002or less

although two were generally sufficient for the small-

for the Murthy sidewallapproximation.Sincethe

chord model. Only two correction passes were re-

magnitudes of sidewallMach numbercorrections

were aboutthesame sizeasthesolid-blockage correc- quired in previous research for all slotted-wall cases.

The considerable number of data points that

tions,but of opposite signandsomewhat uncertain

either did not converge or for which the body-

anyway, wedid not assign anyMachnumber correc-

tionsto the 8-ft TPT slotted-wall data.

alignment criterion was not met indicates that the

The NACA 0012slotted-walldata for the 8-ft

TWNTN4A WIAC program is sensitive to the de- tails of the pressure distributions used in formulat-

TPT givenin reference 22discussed above werealso

investigated by McCroskey (ref. 24) and foundto ing the boundary conditions. Hence, the correction procedure was found to be more difficult and time-

be among the besttransonic dataavailable for this

airfoil if suitableangle-of-attack corrections, suchas consuming to apply for some adaptive-wall cases the empiricallycorrelatedDavis-Moore correction, compared with the slotted-wall cases. In almost wereapplied.McCroskey alsofoundthat thesedata every case the 0.3-m TCT data were corrected to the left of the Navier-Stokes curves. In some cases,

wouldrequireonly verysmall,if any,Machnumber

corrections and that the data set wasa probable this may be partly due to the corrected Mach num- candidate forthe bestoverall setofNACA0012 data.

bers being significantly higher than the Mach number used for the Navier-Stokes calculations. The Navier-

This finding is consistent with the higher aspect

Stokes calculations are shown at the nominal tunnel

ratio of the modelcompared with thoseratios in

Mach numbers, whereas the corrected data should

the adaptive-wall testsection andwith the resultant

reduced SWBLeffect. The agreement between the properly be compared with curves at higher Mach numbers with greater slopes. In other cases, the dis-

small-chord data from the 0.3-mTCT/AWTS and

theHarrisdatashown in figure31supports theabove crepancy between the WIAC results and the Navier- conclusion ofreference 24. McCroskey alsolookedat Stokes calculations may be due to several factors in- the limitedamountofdatapresented in reference 13 cluding problems with the Navier-Stokes solution, an inability of the WIAC sidewall boundary-layer ap-

andfoundit (theTWNTN4A-corrected NACA0012

data of the 0.3-mTCT/AWTS) to be amongthe proximation to correctly model three-dimensional su-

bestdatasetsavailable for this airfoil for the Mach

percritical flows, or an angle-of-attack bias within the

number rangetested. 0.3-m TCT/AWTS facility due to misalignment of

the tunnel and model reference lines. Certainly, im- Concluding Remarks provements made in any of these factors may influ- ence the data correlations presented. The difficulty in This application of the wall-interference assess- correcting some of the data, particularly for the large- ment/correction (WIAC) modified code TWNTN4A chord model, indicates that some flow-visualization to a large amount of transonic wind-tunnel data, in- techniques should be available to determine if sepa- cluding a broad range of model/tunnel configurations ration regions exist within the test section.

and possible wall-interference effects for the NACA Data from the ONERA/CERT T2 adaptive-wall 0012 airfoil, has led to the formulation of several tunnel were found to correct easily and correlated conclusions. With respect to corrected data from well with the Navier-Stokes data and with a lim- the Langley 0.3-Meter Transonic Cryogenic Tunnel ited amount of 0.3-m TCT/AWTS small-chord data.

(0.3-m TCT) having an adaptive-wall test section Corrections were very small in all the cases inves- (AWTS), good correlations were shown in lift and tigated in this study. A comparison of these cor- drag curves at low-to-moderate lift coefficients for rected data with those of the small model from the unadapted and partially to fully adapted walls over a 0.3-m TCT/AWTS supported a possible angle-of- wide range of Reynolds numbers at transonic Mach attack bias in the latter facility.

numbers using the Murthy sidewall boundary-layer It was also found that the uncorrected, fully approximation for the four-wall correction in the adapted lift-curve data from the 0.3-m TCT were TWNTN4A WIAC code. As expected, corrections much better than the uncorrected slotted-wall data for the fully adapted data were much smaller than in correlating with the Navier-Stokes free-air data.

for the partially adapted or unadapted data; this in- TWNTN4A-corrected adaptive-wall data and dicates that the fully adapted data have very little TWINTN4-corrected slotted-wall data correlated wall interference, although the corrections for fully much better with each other and with the Navier- adapted data do tend to improve the data correla- Stokes calculations in the limited number of tions. Smaller, more easily obtained corrections were 10. Gumbert, Clyde R.; and Newman, Perry A.: Validation comparisons shown. Comparisons between 0.3-m of a Wall Interference Assessment/Correction Procedure TCT/AWTS data and slotted-wall data from the for Airfoil Tests in the Langley 0.3-m Transonic Cryogenic Langley 8-Foot Transonic Pressure Tunnel (8-ft Tunnel. AIAA-84-2151, Aug. 1984.

TPT) were similar to the previous comparisons in 11. Gumbert, Clyde R.: Wall Intelference Assessment/ that the uncorrected adaptive-wall data correlated Correction of Data Prom Tests of a CAST IO-2/DOA 2 much better with the Navier-Stokes data than did Airfoil in the Langley 0.3-m Transonic Cryogenic Tunnel.

the uncorrected slotted-wall data. TWNTN4A- M.S. Thesis, George Washington Univ., May 1988.

corrected, adaptive-wall lift-curve data and empiri- 12. Murthy, A. V.: Effects of Aspect Ratio on Sidewall cally corrected, slotted-wall lift-curve data from the Boundary-Layer Influence in Two-Dimensional Airfoil 8-ff TPT correlated reasonably well also. Correla- Testing. NASA CR-4008, 1986.

tion in the drag-curve data between the corrected 13. Green, Lawrence L.; and Newman, Perry A.: Transonic Wall Interference Assessment and Corrections for Airfoil adaptive-wall data and the 8-ft TPT data was very Data From the 0.3-Meter TCT Adaptive Wall Test Sec- good despite some noticeable differences in the drag tion. AIAA-87-1431, June 1987.

levels.

14. Green, Lawrence Lee Richard: Wail Interference Assess- NASA Langley Research Center ment and Corrections for Transonic Adaptive Wail Airfoil Hampton, VA 23665-5225 Data. M.S. Thesis, George Washington Univ., Apr. 1988.

February 5, 1991 15. Swanson, R. C.; and Turkel, Eli: A Multistage Time- Stepping Scheme for the Navier-Stokes Equations. AIAA- References 85-0035, Jan. 1985.

16. Ladson, Charles L.; Hill, Acquilla S.; and Johnson, 1. Pindzola_ M.; and Lo, C. F.: Boundary Interference at William G., Jr.: Pressure Distributions From High Subsonic Speeds m Wind Tunnels With Ventilated Walls.

Reynolds Number Transonic Tests of an NACA 0012 Air- AEDC-TR-69-47, U.S. Air Force, May 1969. (Available foil in the Langley 0.3-Meter Transonic Cryogenic Tunnel.

from DTIC as AD 687 440.)

NASA TM-100526, 1987.

2. Davis, Don D., Jr.; and Moore, Dewey: Analytical Study 17. Ladson, Charles L.; and Hill, Acquilla S.: High Reynolds of Blockage- and Lift-Interference Corrections for Slotted Number Transonic Tests of an NACA 0012 Airfoil in the T_nnels Obtained by the Substitution of an Equivalent Langley 0.3-Meter Transonic Cryogenic Tunnel. NASA Homogeneous Boundary for the Discrete Slots. NACA TM-100527, 1987.

RM L53E07b, 1953.

18. Judd, M.; Wolf, S. W. D.; and Goodyer, M. J.: Ana- 3. Kemp, William B., Jr.: TWINTAN: A Program for Tran- lytical Work in Support of the Design and Operation of sonic Wall Interference Assessment in Two-Dimensional Two Dimensional Self Streamlining Test Sections. NASA Wind Tunnels. NASA TM-81819, 1980.

CR-145019, 1976.

4. Kemp, William B., Jr.; and Adcock, Jerry B.: Combined 19. Goodyer, M. J.; and Wolf, S. W. D.: The Development Four-Wall Interference Assessment in Two-Dimensional of a Self-Streamlining Flexible Walled Transonic Test Airfoil Tests. AIAA J., vol. 21, no. 10, Oct. 1983, Section. A Collection of Technical Papers--AIAA 11th pp. 1353-1359.

Aerodynamic Testing Conference, Mar. 1980, pp. 325-335.

5. Kemp, William B., Jr.: TWINTN4: A Program for (Available as AIAA-80-0440.)

Transonic Four-WaU Interference Assessment in Two- 20. Archambaud, J. P.; Seraudie, A.; and Gobert, J. L.: Dimensional Wind Tunnels. NASA CR-3777, 1984.

Rapport d'Essais sur le Profit NACA 0012 de 150 mm 6. Gumbert, Clyde R.; Newman, Perry A.; Kemp, William de Corde en Presence de Parois Adaptables a la Soufflerie B., Jr.; and Adcock, Jerry B.: Adaptation of a Four- T2 de I'ONERA/CERT. Rapp. Teeh. OA 22/3075 AND, Wall Interference Assessment/Correction Procedure for O.N.E.R.A., Juillet 1981.

Airfoil Tests in the 0.3-m TCT. Wind Tunnel Wail Inter- 21. Salas, M. D.; and Gumbert, Clyde R.: Breakdown of the ference Assessment/Correction 1983, Perry A. Newman i Conservative Potential Equation. NASA TP-2539, 1986.

and Richard W. Barnwelt, eds., NASA CP-2319, 1984, 22. Harris, Charles D.: Two-Dimensional Aerodynamic Char- pp. 393 411.

acteristics of the NACA 0012 Airfoil in the Langley 8-Foot 7. Barnwetl, Richard W.; and Sewall, William G.: Similarity Transonic Pressure Tunnel. NASA TM-81927, 1981.

Rules for Effects of Sidewall Boundary Layer in Two- 23. Barnwell, Richard W.: Design and Performance Evalua- Dimensional Wind Tunnels. Wall Interference in Wind tion of Slotted Walls for Two-Dimensional Wind Tunnels.

Tunnels, AGARD-CP-335, Sept. 1982, pp. 3-1 3-10.

NASA TM-78648, 1978.

8. Melnik, R. E.; Mead, H. R.; and Jameson, A.: A Multi- Grid Method for the Computation of Viscid/Inviscid In- 24. McCroskey, W. J.: A Critical Assessment of Wind Tun- nel Results for the NACA 0012 Airfoil. Aerodynamic teractions on Airfoils. AIAA-83-0234, Jan. 1983.

Data Accuracy and Quality: Requirements and Capabili- 9. Gumbert, Clyde R.: User Manual for 0.3-M TCT Wall- ties in Wind Tunnel Testing, AGARD-CP-429, July 1988, Interference Assessment/Correction Procedure: 8- by 24- pp. 1-i-1-21.

Inch Airfoil Test Section. NASA TM-87582, 1985.

Table I. Key to Data Symbols Used in Figures (a) Data symbols depicting ratio of tunnel half-height to chord (h/c) Data source h/c Data symbol 0.5 Adaptive wall; test 208 in 0.3-m TCT 1.0 Adaptive wall; test 201

[3

in 0.3-m TCT 1.0 Adaptive wall; test 209 in 0.3-m TCT

/x

1.27 Adaptive wall; ONERA/CERT tunnel (ref. 20)

X?

Slotted wall; 8-fl TPT 1.7 h, (ref. 22) i[_ 2.0 Slotted wall; test 119 in 0.3-m TCT (ref. 10) ¢,O Free air; Navier-Stokes code (ref. 15) (b) Definition of data symbols Visual Definition Data symbol description Uncorrected data Open

o

Right filled First pass; WIAC-corrected

[]

Left filled Second pass; WIAC-corrected Third pass; WIAC-corrected O • Filled AI_ /'_ Diagonally Empirically corrected filled [_ _ _ i[_ Flagged Fixed transition (5 percent) J Solid line Free-air calculation Arrow Special features Figure 1. Streamwise section view of NACA 0012 airfoil in the Langley 0.3-m TCT with walls adapted for h/c _ 0.5, M T _ 0.5, and aT _ 4°- Far-field BC

Far-[WT I Z Far-

field 1- 1- _-1-t-1 t -ttlqqlllfl_t1_ l_t _l-----r fil_l[_

Airfoil BC Figure 2. Schematic diagram of TWNTN4A computational grid (upper half-plane only).

2O .02 m Cd .01

I I I i

.6 .7 .8 MT (a) Uncorrected, fully adapted data.

.02 Cd .01 m

i 1

I

.7 .8 .6 MT (b) TWNTN4A-corrected, fully adapted data.

Figure 3. Drag curves for NACA 0012 airfoil at Rc _ 9 × 106 and c I _ 0. See table I for symbol key.

180 - 0 0 Drag counts 40 vVall (a) Spanwise drag-rake survey and model pressure coefficient distribution.

-1.20 [] Upper wall Lower surface O Lower wall F [] Upper surface ".2 -.80 %1 -.40 Cp Cp 0 0 .40 -1.z -3 -2 -1 0 1 2 3 4 x/c .80

C -

I I 0 1 x/c (b) Model and wall pressure coefficient distributions.

Figure 4. Preprocessor plots for NACA 0012 airfoil at Rc ,_ 9 x 106, kl T _ 0.79, and a T _ 0 °.

.O2

¢f

.01

Cd

m

I I I _ I

.6 .7 .8 MT (a) Uncorrected, unadapted, or partially adapted data.

.02 -- .01 Cd

i I J 1

.6 .7 .8 MT (b) TWNTN4A-corrected, unadapted, or partially adapted data.

Figure 5. Drag curves for NACA 0012 airfoil at Rc _ 9 x 106 and c l _ O. See table I for symbol key.

.8 m .4

Cl

I I

-.4 -Z 0 4 oq deg (a) Uncorrected, fully adapted data.

.8 B .4 Cl

1 I

-,4 0 4 a, deg (b) TWNTN4A-corrected, fully adapted data.

Figure 6. Lift-curve data for NACA 0012 airfoil at Rc ,_ 9 x 106 and M T _ 0.6. See table I for symbol key.

.8--

.4 m

Cl

0 m

I I

-.4 0 4 a, deg (c) Uncorrected, unadapted, or partially adapted data.

.8m .4 m Cl 0--

r

I I ]

-.4 0 4 8 a, deg (d) TWNTN4A-corrected, unadapted, or partially adapted data.

Figure 6. Concluded.

.8--

Cl

1 I

-.4

0 4 c_,deg (a) Uncorrected, fully adapted data.

.8

.4 Cl __

1 1

-.4 0 4 _,deg (b) TWNTN4A-corrccted, fully adapted data.

Figure 7. Lift-curve data for NACA 0012 airfoil at Rc _ 9 x 106 and MT _ 0.65. Sce table I for symbol key.

.8--

.4--

Cl

0 m

-.4 -Z 0 4 oc,deg (c) Uncorrected, unadapted, or partially adapted data.

.8--

/

.4 ct 0 --

I I 1

-.4 -Z 0 4 8 a, deg (d) TWNTN4A-corrected, unadapted, or partially adapted data.

Figure 7. Concluded.

.8--

C_

.4

c!

-.4 0 4 a, deg (a) Uncorrected, fully adapted data.

.8-- .4-- Cl

lit

I I I

-.4 0 4 8 -z (:z,deg (b) TWNTN4A-corrected, fully adapted data.

Figure 8. Lift-curve data for NACA 0012 airfoil at Rc _ 9 x 106 and M T _ 0.7. See table I for symbol key.

.8--

.4--

Cl

0 m

1 I

-.4 -Z 0 4 8 a, deg (c) Uncorrected, unadapted, or partially adapted data.

.8B .4 m Cl 0--

l

I

-.4 0 4 -d _, deg (d) TWNTN4A-corrected, unadapted, or partially adapted data.

Figure 8. Continued.

.8--

M T = 0.76

.4

M T = 0.7

Cl

I t

-.4

0 4 8 a, deg (e) TWNTN4A-corrccted, fully adapted data with two N-S curves.

.8-- .4 -- M T = 0.7 M T = 0.76 c!

0q

-.4i I I J

-4 0 4 8 a, deg (f) TWNTN4A-correctcd, unadapted, or partially adapted data With two N-S curves.

Figure 8. Concluded.

3O

.8m

.4--

Cl

0--

-.4 0 4 a, deg (a) Uncorrected, fully adapted data.

.8m .4 Cl

I

-.4 4 8 a, deg (b) TWNTN4A-corrected, fully adapted data.

Figure 9. Lift-curve data for NACA 0012 airfoil at Rc _ 9 x 106 and M T _ 0.72. See table I for symbol key.

.8 m

.4

Cl

l

-.4

0 4 8 co,deg (c) Uncorrected, unadapted, or partially adapted data.

.8 m Cl

I I

-.4 4 8 a, deg (d) TWNTN4A-corrected, unadapted, or partially adapted data.

Figure 9. Concluded.

.8--

.4

c!

-.4

-Z 0 4 a, deg (a) Uncorrected, fully adapted data.

.8 .4 Cl

I T

-.4 4 8 a, deg (b) TWNTN4A-corrected, fully adapted data.

Figure 10. Lift-curve data for NACA 0012 airfoil at Rc .w. 9 x 106 and MT _ 0.74. See table I for symbol key.

.8 --

.4

c!

I

-.4

-dr 8 0 4 a, deg (c) Uncorrected, unadapted, or partially adapted data.

.8 m .4 m ct 0 m

I J

-.4 -Z 0 4 8 a, deg (d) TWNTN4A-corrected, unadapted, or partially adapted data.

Figure 10. Concluded.

.8

.4 B

Cl

I

-.4

a, deg (a) Uncorrected, unadapted, or partially adapted data.

.8-- .4 cl

I 1

-.4 4 8 (b) TWNTN4A-corrected, unadapted, or partially adapted data.

Figure 11. Lift-curve data for NACA 0012 airfoil at Rc ,_ 9 x 106 and M T ,_ 0.75. See table I for symbol key.

.8m

©

.4 m

ci

0 m

-.4 -Z 0 4 8 a, deg (a) Uncorrected, fully adapted data.

.8w .4 c/ -.4 -Z 0 4 a, deg (b) TWNTN4A-corrected, fully adapted data.

Figu_c 12. Lift-curve data for NACA 0012 airfoil at Rc _ 9 x 106 and hi T ,._ 0.76. Sce table I for symbol key.

.8 --

O

.4 Cl

(>

-.4 -L 0 4 a, deg (c) Uncorrected, unadapted, or partially adapted data.

.8w

¢

.4 cl ¢,

I

-.4 I

-4 8 0 4 a, deg (d) TWNTN4A-corrected, unadapted, or partially adapted data.

Figure 12. Concluded.

.8m

.4

c!

I I

-.4

0 4 a, deg (a) Uncorrected, fully adapted data.

.8-- cl

J I

-.4 4 8 cq deg (b) TWNTN4A-corrected, fully adapted data.

Figure 13. Lift-curve data for NACA 0012 airfoil at Rc _ 9 x 106 and M T _ 0.8. See table I for symbol key.

.8--

.4

d

c/

-.4

8 4 a, deg (c) Uncorrected, unadapted, or partially adapted data.

.8 -- .4 Cl

I l I

-.4 0 4 8 a, deg (d) TWNTN4A-corrected, unadapted, or partially adapted data.

Figure 13. Concluded.

.02 --

cY O .01

Cd

1 j I i

.8 .6 .7 MT (a) Uncorrected, fully adapted data.

.O2 Cd .01

1 I I l I

.6 .7 .8 MT (b) TWNTN4A-corrected, fully adapted data.

Figure 14. Drag curves for NACA 0012 airfoil at Rc _ 9 x 106 and cI _ 0.2. See table I for symbol key+ 4O

.02 w

O"

Cd .01 m

O;

%

0m

t

.6 .7 .8 MT (a) Uncorrected, fully adapted data.

.O2 .01 Cd

%

0 --

I I 1 l

.6 .7 .8 MT (b) TWNTN4A-corrected, fully adapted data.

Figure 15. Drag curves for NACA 0012 airfoil at Rc _ 15 × 106 and cl ,._ O. See table I for symbol key.

180 - 0 0 0 0 iiYlll,,i j,_l

Drag _,_,, ,,1!111; _

counts i i I_3..,_t-,._ J,.,d_L._¼ ' / 100 I III 4O m Wall (a) Spanwise drag-rake survey and model pressure coefficient distribution.

-,4 [] Upper wall 0 Lower wall [] Upper surface Lower surface I 0 1 .1 -1 0 1 2 x/c -2 x/c (b) Model and wall pressure coefficient distributions.

Figure 16. Preprocessor plots for NACA 0012 airfoil at Rc ,_ 15 x 106, hiT ,_ 0.76, and a T _ 0 °.

.02 --

.01

Cd

O

O

O m

I

1 l I [

.8 .6 .7 MT (a) Uncorrected, unadapted, or partially adapted data.

.O2 .01 m Cd

41>

I l I i I

.6 .7 .8 MT (b) TWNTN4A-corrected, unadapted, or partially adapted data.

Figure 17. Drag curves for NACA 0012 airfoil at Rc _ 15 x 106 and cl _ 0. See table I for symbol key.

.8 .4 c!

I

-.4 .t 0 4 oc,deg (a) Uncorrected, fully adapted data.

Cl

1 I l

-.4 0 4 8 oc,deg (b) TWNTN4A-corrected , fully adapted data.

Figure 18. Lift-curve data for NACA 0012 airfoil at Rc _ 15 x 106 and MT ._ 0.6. See table I for symbol key.

.8 m

.4 B

cl

0 m

I I

-.4 -L 0 4 a, deg (a) Uncorrected, fully adapted data.

.8-- .4

/"

Cl

l I

-.4 0 4 s a, deg (b) TWNTN4A-corrected, fully adapted data.

Figure 19. Lift-curve data for NACA 0012 airfoil at Rc ,-_ 15 × 106 and hi T _ 0.7. See table I for symbol key.

.8w

<>

<_

.4 c!

-.4 -L 0 4 c¢, deg (c) Uncorrected, unadapted, or partially adapted data.

.8-- .4 c!

I I

-.4 0 4 a, deg (d) TWNTN4A-corrccted, unadapted, or partially adapted data.

Figure 19. Concluded.

.8--

m

.4

c!

-.4

a, deg (a) Uncorrected, fully adapted data.

.8-- [] .4 Cl

I

-.4 4 8 a, deg (b) TWNTN4A-corrected, fully adapted data.

Figure 20. Lift-curve data for NACA 0012 airfoil at Rc .._ 15 × 106 and M T _ 0.76. See table I for sym_bol key.

.02 m [3 [] [] Cd

.01 --_

[] [] --W 0w

I i I

.7 .8 .6 MT (a) Uncorrected, fully adapted data.

.02 -- Cd .01 [] 1313 [] [] m

I i 1 I I

.6 .7 .8 MT (b) TWNTN4A-corrected, fully adapted data.

Figure 21. Drag curves for NACA 0012 airfoil at Rc _ 3 x 106 and ct _ 0. See table I for symbol key.

180 - ._Y, ',,', i i I I?

_Y II i' I I .,__L._'. o (_ O TI II I 1 i , - Drag O cou nts O 100 - O T I i i /,_TI -;-o i

,', ,_ i iL_ ,_-_/

,'_/£ ',,±-.?:_ / O 4(3- Wall (a) Spanwise drag-rake survey and model pressure coefficient distribution.

-1.20 - [] Upper surface 0 Lower surface -.2 - [] Upper wall 0 Lower wall ",1 Cp Cp I I I _ I I I -L -3 -2 -1 0 1 2 3 4 x/c .80 I x/c (b) Model and wall pressure coefficient distributions.

Figure 22. Preprocessor plots for NACA 0012 airfoil at Rc --_ 3 × 106, MT _ 0.79 and a T _ 0 °.

8m gi¢

.4

cl

I I I

-.4

-d 0 4 8 _, deg (a) Uncorrected, fully adapted data.

.8m .4 Cl

I I

I

-.4 -Z 0 4 o_, deg (b) TWNTN4A-corrected, fully adapted data.

Figure 23. Lift-curve data for NACA 0012 airfoil at Rc _ 3 x 106 and MT _ 0.6. See table I for symbol key.

5O .8 m

_t

.4-- Cl 0--

I

t

-,4 -Z 8 0 4 a, deg (a) Uncorrected, fully adapted data.

.8 m .4 cl

I J

-.4 8 0 4 a, deg (b) TWNTN4A-corrected, fully adapted data.

Figure 24. Lift-curve data for NACA 0012 airfoil at Rc _ 3 x 106 and M T ,_ 0.7. See table I for symbol key.

.8--

.4--

Cl

I J

-.4

8 0 4 a, deg (a) Uncorrected, fully adapted data.

.8-- Cl

1 I

-.4 -Z 0 4 8 oc,deg (b) TWNTN4A-corrected, fully adapted data.

Figure 25. Lift-curve data for NACA 0012 airfoil at Rc _ 3 × 106 and M T _ 0.75. See table I for symbol key.

.8m

.4

ct

m

I I

-.4 0 4 -Z a, deg (a) Uncorrected, fully adapted data and slotted-wall data.

.8F

/

.4--

/

Cl

/

0--

1 1

-.4 0 4 8 -Z a, deg (b) TWNTN4A-corrected, fully adapted data and TWINTN4-corrected slotted-wall data.

Figure 26. Lift-curve data for NACA 0012 airfoil at Rc _ 9 x 106 and M T "_ 0.7. See table I for symbol key.

.8--

.4

ct

I I

-.4

0 4 8 a, deg (c) TWNTN4A-corrected, fully adapted data and EDM-corrected slotted-wall data.

Figure 26. Concluded.

- 54

.8m

.4--

Cl

I I I

-.4

0 4 8 -L a, deg (a) Uncorrected, fully adapted data and slotted-wall data.

.8-- .4 m c!

0 m

1 I

-.4 0 4 8 a, deg (b) TWNTN4A-corrected, fully adapted data and TWINTN4-corrected slotted-wall data.

Figure 27. Lift-curve data for NACA 0012 airfoil at Rc _ 9 × 106 and hit _ 0.76. See table I for symbol key.

.8m

.4

c/

] I I

-.4

-L 0 4 8 o_,deg (c) TWNTN4A-corrected, fully adapted data and EDM-corrected slotted-wall data.

Figure 27. Concluded.

.02 m

d c;

Cd .01

.6 .7 .8 MT (a) Uncorrected, fully adapted data and slotted-wall data.

.O2 .01 Cd .c If

I i I i I

.6 .7 .8 MT (b) TWNTN4A-corrected, fully adapted data and TWINTN4-corrected slotted-wall data.

Figure 28. Drag curves for NACA 0012 airfoil at Rc _ 9 × 106 and c I _ 0.2. See table I for symbol key.

.8m

r_ .4--

Cl

0 m

I 1 I

-.4 ..L 0 4 8 a, deg (a) Uncorrected, fully adapted data and slotted-wall data.

.8-- .4 m c/ 0--

I 1 I

-.4 0 4 8 a, deg (b) TWNTN4A-corrected, fully adapted data and EDM-corrected slotted-wall data.

Lift-curve data for NACA 0012 airfoil at Rc _ 9 × 106 and MT _ 0.7. See table I for symbol key.

Figure29.

.8--

.4

Cl

/

I I

-.4

0 4 8 -d a, deg (a) Uncorrected, fully adapted data and slotted-wall data.

.8-- .4 m cl 0 m

I I I

-.4 0 4 8 a, deg (b) TWNTN4A-corrected, fully adapted data and EDM-corrected slotted-wall data.

Figure 30. Lift-curve data for NACA 0012 airfoil at Rc _ 9 × 106 and M T _ 0.76. See table I for symbol key.

.02 --

Cd .01

b_

1 l t

.6 .7 .8 MT (a) Uncorrected, fully adapted data and slotted-wall data.

.02 -- Cd .01

I l J I

.6 .7 .8 MT (b) TWNTN4A-corrected, fully adapted data and uncorrected slotted-wall data.

Figure 31.

Drag curves for NACA 0012 airfoil at Rc _ 9 x 106 and c l _ 0.2. See table I for symbol key.

6O Report Documentation Page Natior'_at Aeronautics and Soace AOminislrat_on 1. Report No, | 2, Government Accession No. 3. Reciplent's Catalog No.

NASA TP-3070

J

4. Title and Subtitle 5. Report Date Wall-Interference Assessment and Corrections for Transonic April 1991 NACA 0012 Airfoil Data From Various Wind Tunnels 6. Performing Organization Code 7. Author(s) 8. Performing Organization Report No, Lawrence L. Green and Perry A. Newman L-16721 10. Work Unit No, Performing Organization Name and Address 505-61-01-04 NASA Langley Research Center 11. Contract or Grant No.

Hampton, VA 23665-5225 13. Type of Report and Period Covered 12.

Sponsoring Agency Name and Address Technical Paper National Aeronautics and Space Administration 14. Sponsoring Agency Code Washington, DC 20546-0001 15.

Supplementary Notes Presented in part as AIAA 87-1431, June 8 10, 1987, in Honolulu, Hawaii; condensed from thesis given in partial fulfillment of the requirements for the Degree of Master of Science, George Washington University, Washington, D.C., April 1988.

16. Abstract A nonlinear, four-wall, post-test wall-interference assessment/correction (WIAC) code has been developed for transonic airfoil data from solid-wall wind tunnels with flexibly adaptive top and bottom walls. The WIAC code has been applied over a broad range of test conditions to four sets of NACA 0012 data from two different adaptive-wall wind tunnels. The data include many test points for fully adapted walls as well as numerous partially adapted and unadapted test points, which together represent many different model/tunnel configurations and possible wall- interference effects. Small corrections to the measured Mach numbers and angles of attack are obtained from the WIAC code even for the fully adapted data; these corrections generally improve the correlation among the various sets of airfoil data and simultaneously improve the correlation of the data with calculations from a two-dimensional, free-air Navier-Stokes code. The WIAC corrections for airfoil data taken in test sections with fully adapted walls are shown to be significantly smaller than those for comparable airfoil data from test sections with straight, slotted walls. This indicates, as expected, a lesser degree of wall interference in the adaptive-wall tunnels relative to the slotted-wall tunnels. Application of the WIAC code to these data has, however, been somewhat more difficult and time-consuming than initially expected from similar previous experience with WIAC applications to slotted-wall data.

18. Distribution Statement 17. Key Words (Suggested by Author(s)) Unclassified--Unlimited Wall-interference assessment/correction (WIAC) Transonic flow Adaptive-wall tunnel Sidewall boundary-layer (SWBL) effects Subject Category 02 19. Security Classif. (of this report)unclassified 120" Security Classif' (°f this page)Unclassified t 21" N°' °f Pages 22' Price 61 A04 NASA FORM 1626 OCT 8s NASA-Langley, I99I For sale by the National Technical Information Service,Springfield, Virginia 22161-2171