AN EXPERIMENTAL EVALUATION OF WIND TUNNEL WALL CORRECTIONMETHODS FOR HELICOPTER PERFORMANCE

Hans-Juergen Langer

Deutsche Forschungsanstalt für Luft- und Raumfahrt e.V.

Institut für Flugmechanik

Braunschweig, Germany

Randall L. Peterson Thomas H. Maier

NASA Ames Research Center Aeroflightdynamics Directorate

Moffett Field, CA Aviation Research, Development and Engineering Center

U.S. Army Aviation and Troop Command

Ames Research Center

Moffett Field, CA

ABSTRACT

Accurate measurements of rotorcraft performance asmeasured in a wind tunnel are strongly influenced by thetest section configuration, whether it be closed or openjet. The influence of wind tunnel walls on the inducedvelocity of lifting bodies has been studied by manyresearchers over the years. Methods have been developedto adjust the angle-of-attack and dynamic pressure for fixedwing aircraft in a wind tunnel to approximate free flightconditions. These methods have largely been adopted bythe rotorcraft community with very little testing to verifythe applicability of these methods to helicopter rotors andflight test measurements. Recent tests conducted by theDeutsche Forschungsanstalt für Luft- und Raumfahrt e.V.(DLR) in the Duits-Nederlandse Wind Tunnel (DNW)have provided data suitable for the evaluation of thesemethods. A 40% scale model Bo105 rotor was tested infive different wind tunnel test sections: 1) 6x6m closed, 2)8x6m closed, 3) 8x6m open slots, 4) 9.5x9.5m closed,and 5) the 8x6m open jet. These data along with full-scaledata from a NASA Ames 40- by 80-Foot Wind Tunneltest and a DLR flight test program provide a means toevaluate wind tunnel wall correction methods specificallyfor helicopter rotors. Good correlation of rotor power overa range of advance ratios for these three data sets has beenshown using wall correction methods after accounting fortrim deficiencies between the data sets.

Presented at the American Helicopter Society 52nd AnnualForum, Washington, D.C., June 4-6, 1996. Copyright ©1996 by the American Helicopter Society, Inc. All rightsreserved.

NOMENCLATURE

a two-dimensional lift-curve slope

A area, m2

c blade chord, m

cL non-dimensional lift coefficient, windaxis

cP, CP non-dimensional rotor power coefficient

cT non-dimensional thrust coefficient, shaftaxis

C rotor control vector

D rotor diameter, m

D derivative matrix

F correction factor, Eq. (5)

FX x-force, N

FZ z-force, N

F hub load vector (e.g., FX ….... MZ)

L lift, N

MX rolling moment, Nm

MY pitching moment, Nm

MZ yaw moment or torque, Nm

p per rev

R rotor radius, m

s rotor model-scale factor ( = 2.456)

T thrust, N

vtip rotor blade tip speed, m/s

V tunnel speed or flight speed, m/s

W test section width, m

∆cP0profile power loss

∆cd0profile drag loss

∆α rotor correction angle due to wallinterference, rad or deg

α s, α shaft rotor mast incidence, deg

δW wall or boundary correction factor

θ0.7 collective pitch angle, r/R = 0.7

θc lateral cyclic pitch angle

θs longitudinal cyclic pitch angle

µ advance ratio, V/vtip

ρ air density, Ns2/m4

σ solidity

Abbreviations, Superscripts and Subscripts:

FS full-scale

FT flight test or free flight

Fus fuselage

mo model rotor

Re Reynolds number

Ro rotor

TPP tip-path-plane

TS test section

WT wind tunnel

INTRODUCTION

The use of wind tunnel test measurements, flight testmeasurements, and analytical prediction plays a key rolein the development of new rotor systems. Such tests aretypically performed using a range of rotor system sizesand wind tunnel test facilities. To assure the accuracy ofwind tunnel testing methodology, a validation study is inprogress using test results from model- and full-scale testsin comparison with flight test data. This study is beingconducted under the auspices of the U.S. Army/GermanMemorandum of Understanding on Cooperative Researchin the Field of Helicopter Aeromechanics. Thiscomparison will allow for a determination of the abilityto accurately predict helicopter flight behavior from windtunnel experiments and the influence of the test facility onthese results. Experimental data from a series of windtunnel tests, including both model- and full-scaleexperiments, have been studied to determine the extent towhich wind tunnel test results can be used to predict flightbehavior.

This paper presents the results of a recentlycompleted model-scale test of a Bo105 hingeless rotor inthe DNW. A 40% scale model Bo105 rotor (R = 2.0m)was tested in five different test sections of the DNW windtunnel with the intent to evaluate and identify theinfluence of wind tunnel walls on measured rotorperformance. The five different wind tunnel test sectionsused in this series of tests included the: 1) 6x6m closed,2) 8x6m closed, 3) 8x6m open slots, 4) 9.5x9.5m closed,and 5) the 8x6m open jet. The influence of wind tunnelwalls and the flow breakdown phenomenon has beenstudied and reported by many researchers over the years(Refs. 1-13). From these studies, a number of methodshave been developed to account for tunnel wall inducedeffects in order to approximate free-flight conditions in thewind tunnel. Results from the DNW model-scale test wereused to evaluate the applicability of two of these methodsas they apply to rotorcraft testing. Additionally, the DNWdata are compared with full-scale (R = 4.912m) data froma NASA Ames 40- by 80-Foot Wind Tunnel test (Ref.14) and a DLR flight test program (Ref. 15) to furtherevaluate the wind tunnel wall correction methods.

a) flight test

b) full-scale rotor

c) model-scale rotor

Figure 1. Test programs; a) flight test of Bo105Helicopter b) full-scale tests in the 40- by 80-Foot WindTunnel and c) model-scale tests in the DNW.

TEST PROGRAM

Many wind tunnel tests are conducted with scaledmodels with the intention of determining thecharacteristics of full-scale flight aircraft. This is done toreduce costs and to obtain experimental measurementswithin a reasonable amount of time. There is, however, agreat leap between a small-scale wind tunnel test and afull-scale flying vehicle. Differences in structure, rotortrim, rotor/body interaction, and the like may cause thescale model test to not be representative of the full-scaleaircraft. Investigations into improving wind tunnel testingmethodologies and evaluation of the suitability of scaledmodel testing to determine full-scale flight characteristicsare the main goals for the rotor data correlation taskwithin the Memorandum of Understanding (MOU)between U.S. Army/NASA and the Institute of FlightMechanics of "Deutsche Forschungsanstalt für Luft- undRaumfahrt" (DLR). Correlation efforts have beenconducted, based on flight tests, NASA Ames 40- by 80-Foot Wind Tunnel tests with a full-scale Bo105 rotor anda test program with a scaled Bo105 rotor/fuselage in theGerman-Dutch wind tunnel (DNW). Representativephotos of each of the test programs are shown in Fig. 1.

Flight Test Program

The primary task of the flight test program was toprovide the basis for follow-on wind tunnel tests of a full-scale rotor in the NASA Ames 40- by 80-Foot WindTunnel and a model-scale rotor in the German-Dutch windtunnel. For the data correlation task, flight data acquired ata density altitude of approximately 762 m (2500 ft) wasused. Steady-state flights were performed between hoverand the maximum speed of the helicopter, with a stepsizein speed of approximately 10 knots.

The primary task of the test pilot was to establish asteady-state condition with minimum climb/descent,sideslip or pitch rate for each forward speed. Once thepilot was 'on condition' data was acquired with hands offthe controls. Unfortunately, when gathering data with thepilots hands off the controls the aircraft maintains a steadylevel flight condition for a very short period of time dueto the aircraft's instability. For this reason only the firstthree rotor revolutions were processed to form the flighttest database. Each speed sweep was repeated three timesto assess data scatter.

Rotor thrust could not be measured directly in theflight test program, therefore, the aircraft weight was usedas an approximate measure. Weight was also not a directmeasure, therefore the helicopter was weighed before andafter each flight, and it was assumed that fuel

consumption was linear with time as shown in Fig. 2. Tominimize the influence of this approximation on aircraftweight, the flight tests were performed as quickly aspossible. It was found from the analysis of the data thatthe flight test data presented in this paper is valid for acT = 0.005 ± 0.0001.

Since the data correlation program in the variouswind tunnels is primarily based on flight test data,emphasis was placed on the creation of a reliable database. This data base consists of different sensors signals,some of which are used just to confirm the validity ofother sensors. For example, the rotor mast torque wasused to determine rotor power, while the power indicationfrom the cockpit was used to check the mast torque asshown in Fig. 3.

100

150

200

250

300

350

400

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07

Rem

aini

ng F

uel,

kg

Stopwatch (decimal), 24 Hours = 1.00 unit

maximum fuel (main tank) 382 kg

all values fromcockpit indicator

Figure 2. Flight test fuel consumption as a function oftime.

200

250

300

350

400

450

500

550

0.05 0.10 0.15 0.20 0.25 0.30 0.35

Rotor MastTotal (cockpit)

Pow

er, k

W

Advance Ratio, µ

Figure 3. Accuracy check of rotor mast power sensorusing the total power indicator from cockpit.

The total power data shown in Fig. 3 includes the tailrotor power, the gearbox efficiency, and the generatorpower. The power measurements shown in Fig. 3 in thelow speed region, for µ < 0.1, suggest that the speedindicator does not provide an accurate measure in thisregion. Both measures of power track with one anotherwell, but the curve with speed is not smooth. In this lowspeed region the airspeed sensor is probably adverselyinfluenced by the rotor downwash.

-12

-10

-8

-6

-4

-2

0

0.05 0.10 0.15 0.20 0.25 0.30 0.35

Rot

or S

haft

Ang

le, d

eg

Advance Ratio, µ

Figure 4. Shaft angle versus advance ratio for three testruns at constant density altitude.

-2000

-1000

0

1000

2000

3000

0.05 0.10 0.15 0.20 0.25 0.30 0.35

1p S

haft

Ben

ding

Mom

ent,

Nm

Advance Ratio, µ

1p Cosine

1p Sine

Figure 5. Mast bending moment versus advance ratio forthree test runs at constant density altitude. The 1p cosinemoment is the pitching moment and the 1p sine momentis the rolling moment.

Other important sensors for the correlation of flightand wind tunnel data are the rotor shaft angle (or fuselageattitude) and the mast bending. Figure 4 shows the shaftangle measurement versus advance ratio for three different

speed sweeps. Repeatability of the shaft anglemeasurement as compared with the least-squares curve-fitshows acceptable data scatter. Hub pitch and roll momentswere determined from the rotating shaft bending gauges. Aonce per revolution spike over the tail of the aircraftprovided a phase reference. The shaft bending momentswere harmonically analyzed and the 1p cosine and 1p sinevalues were taken as the steady pitch and steady rollmoments, respectively. These measurements may be seenin Fig. 5 for three speed sweeps. Again, the repeatabilityis acceptable.

Figures 4-5 provided the basis for trim settings in thewind tunnel test programs in the NASA Ames 40- by 80-Foot Wind Tunnel and in the DNW.

Other important parameters that were measuredinclude rotor rpm, temperature, and pressure. In addition, asingle blade was highly instrumented with 16 straingauges. The flapwise strain gauges, 11 in all, provide ameans to assess the elastic bending portion of the rotortip path-plane. The large number of sensors along theblade span allows for the determination of the higher bladebending modes and allows for the evaluation of thelocation and number of sensors necessary to find thesemodes. The rotor was also instrumented with four lead-lagand one torsion sensor.

NASA Ames 40- by 80-Foot Wind TunnelTests

From the results of the flight test program, windtunnel tests were conducted in the NASA Ames 40- by80-Foot Wind Tunnel test section with a full-scale Bo105Rotor installed on the NASA Ames Rotor Test Apparatus(RTA) as shown in Fig. 1b. Reference 15 describes thistest program and presents the correlation of these resultswith flight test.

The RTA is a special-purpose drive and supportsystem for operating helicopter rotors in the 40- by 80-and 80- by 120-Foot Wind Tunnels. The RTA houses twoelectric drive motors, the hydraulic servo-actuators of theprimary control-system, and a dynamic control systemcapable of introducing dynamic perturbations to the non-rotating swashplate (collective and tilt) at frequencies upto 40 Hz. Installed on the RTA is a five-componentsteady/dynamic rotor balance to determine rotor loads atthe hub moment center. The balance was designed andfabricated to measure both the steady and vibratory rotornormal, axial and side forces, together with rotor pitchingand rolling moments up to rotor thrust levels of 98,000 N(22,000 lb). An instrumented flex-coupling measuresrotor torque and the residual normal force.

Instrumentation for the 40- by 80-Foot Wind Tunneltest included the five-component rotor balance andinstrumented flex-coupling, thirty-seven blade bending andtorsional moment strain gauge measurements (distributedamongst the four blades), one rotating pitch-linkmeasurement, one blade root pitch angle measurement,three stationary control system measurements and standardwind tunnel operating condition measurements.

German-Dutch Wind Tunnel (DNW) Tests

The wind tunnel test program in the DNW wasperformed with the DLR's Modular Wind tunnel Model(MWM). Reference 16 describes the capabilities of theMWM in detail.

The complete wind tunnel model consisted of a 40%scaled rotor and fuselage. Although a tail rotor was notinstalled, the drag of the tail rotor hub and shaft wereroughly simulated by a simple cylinder. Both the rotorand the fuselage were each equipped with a 6-componentbalance. Rotor torque was measured by a torque meter andby the rotor balance. Since the rotor model allowed for themeasurement of mast bending in the rotating axis frameand the rolling and pitching moment in the fixed axisframe, correlation between these signals was important.Therefore, one requirement of the test program in theDNW had been to trim the rotor to the 1p mast moments(sine and cosine) and to the steady rotor balance roll andpitch moments. The influence on rotor performance (i.e.,lift, drag and power) was not significant for these differenttrim procedures.

The rotor blades were equipped with flap, lead-lag,and torsion sensors as shown in Table 1.

Table 1. Model-scale rotor blade instrumentation.

Blade No. Flap Lag Torsion

reference 14 12 8

2 2 1 -

3 4 2 1

4 4 2 1

The wind tunnel program in the DNW was tailored tofive different tasks:

1) Correlation with flight and 40- by 80-Foot WindTunnel tests;

Sections tested: 6x6m closed, 8x6m closed, 8x6mopen, 9.5x9.5m closed

2) Rotor trim to power from flight test;

Sections tested: 6x6m closed, 8x6m closed, 8x6m12% slotted, 8x6m open, 9.5x9.5m closed

3) Minimized flap bending moment trim for zero αshaft,correlation with 40- by 80-Foot Wind Tunnel tests;

Sections tested: 6x6m closed, 8x6m closed, 8x6m12% slotted, 8x6m open, 9.5x9.5m closed

4) Hover tests, correlation with flight tests and 40- by80-Foot Wind Tunnel tests;

Sections tested: 6x6m closed, 8x6m closed, 8x6mopen

5) Derivatives, correlation with 40- by 80-Foot WindTunnel tests;

Sections tested: 6x6m closed, 8x6m closed, 8x6mopen, 9.5x9.5m closed (one speed only )

Tests with minimized flap bending trim at zero shaftangle (task 3) were conducted to identify trim differencesbetween the RTA in the Ames 40- by 80-Foot WindTunnel test section and the scaled model in the DNW.Since the model control is well defined, data correlationbetween both configurations requires no interpolation.

The hover tests (task 4) were performed at -20° rotorshaft angle in the closed test sections and at 0° in the opentest section.

The derivative measurements in task 5 are anessential tool if interpolation is necessary betweenmeasured results. Since the derivative elements (e.g.,∆cT/∆α) are assumed to be linear in a small α-range only,the use of derivatives for extrapolation is often notaccurate enough.

The most important parameters (e.g., thrust, rotorspeed, etc.) were controlled in non-dimensional form sothat the density influence was considered.

In all but the open test section, DNW personnelacquired wall pressure measurements using 92 pressuresensors. The sensors were installed along the floor (3rows), along the side walls (2 rows each), and along theceiling (3 rows). Preliminary signal analysis of thepressure sensors shows that the flow has strong gradientsand has no symmetry. An in-depth signal analysis has notbeen performed yet.

CORRECTION METHODS

From early rotor investigations in wind tunnels it isknown that wind tunnel measurements cannot directly beapplied to free-flight conditions. Rotor reactions in thewind tunnel depend on various parameters such as:

• type of the test section (open, closed, slotted orclosed on bottom only)

• shape of the test section (rectangular, square, elliptic,etc.)

• dimensions of the wind tunnel test section withrespect to the rotor size

• position of the model rotor regarding distance to thewall (eccentricity)

• rotor disk loading, dynamic pressure, and wake skewangle

Additionally, the effects of tunnel blockage due to therotor and support system and flow breakdown due to theimpingement of the rotor downwash on the tunnel floorin the low speed regime also affect rotor loads andperformance (e.g., power) as can be seen in Figs. 6-7.

0.00000

0.00010

0.00020

0.00030

0.00040

0.00050

0.00060

0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009

9.5x9.5m closed8x6m open8x6m closed6x6m closed8x6m 12% slotted40x80ft

Rot

or P

ower

Coe

ffici

ent,

CP

Rotor Thrust Coefficient, CT

Figure 6. Rotor power as a function of rotor thrust withsmooth flow conditions, µ = 0.07, αs = 0°.

Figure 6 shows a clear difference in rotor powerbetween the open and closed test sections where smoothflow in the tunnel exists. This difference becomes smallor even vanishes for conditions where flow breakdownexists as shown in Fig. 7. Therefore, wall correctionmethods are not applicable for conditions of flowbreakdown.

0.00010

0.00020

0.00030

0.00040

0.00050

0.00060

0.00070

0.00080

0.00090

0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009

9.5x9.5m closed8x6m open8x6m closed6x6m closed8x6m 12% slotted40x80ft

Rot

or P

ower

Coe

ffici

ent,

CP

Rotor Thrust Coefficient, CT

Figure 7. Rotor power as a function of rotor thrust withflow breakdown conditions, µ = 0.023, αs = 0°.

Corrections For Wall Interference

Wind tunnel wall induced interference can partly beeliminated by applying angle-of-attack corrections to therotor shaft. Open test sections also require corrections, butof opposite sign to that of closed test sections.

To determine the direction of the angle-of-attackcorrection, one must imagine that the rotor flow isdeflected by the test section ceiling which sharply turnsthe inflow. The test section floor changes the direction ofthe downwash, while the side walls have an impact on therotation of the rotor flow. All can influence the localangle-of-attack due to inflow along the blade span.

Due to the change in inflow caused by the testsection ceiling the rotor needs a more negative incidenceto compensate for this effect. This is opposite to that foran open test section.

Glauert Correction Methodology

Glauert was the first to investigate in detail windtunnel wall induced effects on wings. The Glauertcorrection is considered to be the classical or conventionalwall correction method for fixed wing testing in a windtunnel. The average downwash or induced angle correctionis in the form

∆α =

δw Awing

ATS

cL

(1)

where Awing is the wing area, ATS is the test section areaand δW is the boundary correction factor. The boundarycorrection factor, δW is dependent on the test sectionshape, the ratio of the wing span to tunnel width and theposition of the wing in the test section. A comprehensivecollection of boundary correction factors for various testsection shapes can be found in Ref. 2.

Assuming lift Lwing is equivalent to the rotor thrustT, but

Lwing = ƒ(V) and T = ƒ(vtip),

cL of Eq. (1) can be replaced by cT using

cL

≡ cT

vtip2

V2 (2)

With

µ = V

vtip

Eq. (1) becomes

∆α = 2 δw c

T Aro

µ2 ATS

180π [deg] (3)

and

∆α = αFT

− αWT

(4)

Note that Eq. (3) has a factor of 2 due to the fact that δWfor wings refers to the wing span and thus 2R for rotors.

The boundary correction factors used in this paper arefound in Table 2. Also shown in Table 2 are thecorresponding values for rotor diameter to wind tunnel testsection width. The boundary correction factors for theDNW test sections were determined from the figures foundin Ref. 2. The boundary correction factor for the 40- by80-Foot Wind Tunnel was determined from a figure foundin Ref. 17.

In addition, δW can also be found by using morecomprehensive flow correction theories or by experimentas will be discussed later.

Table 2. Wall correction factors δW from Ref. 2 and 17.

Test Section D/W δW

DNW 6x6m closed 0.667 0.160

DNW 8x6m closed 0.500 0.119

DNW 9.5x9.5m closed 0.421 0.145

DNW 8x6m open 0.500 -0.158

Ames 40x80ft 0.403 0.112

For completeness it should be mentioned that asimilar closed form equation can be found in Ref. 10 asshown below

∆α = F tan-1

( cT

cos αFT

2 µ2 - cT

sin αFT) (5)

Without substantial lack on accuracy the aboveequation can be re-written

∆α = F cT

2 µ2 (6)

The factor 2 in Eq. (6) comes from the air densitythat is often given as ρ instead of ρ/2 when thrust iswritten in the non-dimensional form.

From Eqs. (3) and (6) one gets the correction factor F

F = 4 δw AroA

TS (7)

Heyson and Brooks Correction Methodologies

To determine boundary correction factors (δW) or forlocal angle-of-attack corrections along the blade span,Heyson's approach is widely used. The Heyson approachis based on 'potential theory' assumptions (Refs. 3-6).This method can be applied to rectangular test sectionsthat can be closed, open, or closed on the bottom only.The resulting angle-of-attack and dynamic pressurecorrections are dependent on various parameters such asrotor to test section width, width to height ratio, tunnelspeed, rotor radius, hub eccentricity, rotor rotational speed

and thrust. The FORTRAN programs of Heyson are foundin Ref. 4.

When a disagreement was found between calculatedcorrection factors as determined by the Langley modifiedversion of the Heyson program, and the boundarycorrection factors found in Ref. 2, it was determined thatthe Langley version contained a coding error. As a resultof this disagreement and the desire to determine a detailmapping of wind tunnel corrections over the region of therotor disk, Brooks derived a new correction method (Refs.12-13). Results from this code have been validated bycomparing correction values with those from a variety ofpublished benchmark correction cases.

The approach used by Brooks is similar to that usedby Heyson, however it is based on vortical rather thandipole wake distribution modeling. The Brooks codecontains streamline curvature effects due to a lifting rotorin rectangular test sections. This code gives a spatialdistribution of the correction, while the Heyson codegives only the correction in the rotor disk plane. Solid andwake blockage effects are not included in the Brooks codesince it is assumed that in an open test section the streamis free to expand and in a closed test section the modelsize is much smaller than the cross-sectional area.

The Brooks code has been applied to all DNW testsections with the exception of the slotted walls. Becausethis paper deals only with rotor performance rather thanlocal blade loads, results from Brooks code are used solelyfor the calculation of the global angle-of-attack correction,∆α . Angle-of-attack corrections (∆α 's) as calculated bythe Brooks code are compared with the ∆α 's from Eq. (3)and δW's from Table 2. Comparisons of the Glauertequation results with the results from the Brooks code areshown in Figs. 8-10. More detailed comparisons of eachmethod are presented in the results section of the paper.

Figure 8 is a comparison of the angle-of-attackcorrections for the Glauert equation and the Brooks codecalculations as a function of rotor thrust at a fixed advanceratio (µ = 0.072). For the 6x6m closed test section, theGlauert equation calculates a larger ∆α-correction as afunction of rotor thrust than the Brooks code calculations.Also, the difference in the ∆α-corrections between the twomethods increases with increasing thrust. Figure 9 is acomparison of the angle-of-attack corrections as a functionof advance ratio for the 6x6m closed test section. Thecalculations shown in Fig. 9 are for a fixed value of rotorthrust (cT = 0.005). The difference in the ∆α-correctionsbetween the two methods is greatest at the lowest speedsand the difference decreases as advance ratio is increased.

2

3

4

5

6

7

8

9

0.002 0.003 0.004 0.005 0.006 0.007 0.008

Brooks Codeδw = 0.16, Ref. 2

Ang

le-O

f-A

ttack

Cor

rect

ion

∆α

, deg

Rotor Thrust Coefficient, CT

Figure 8. Correction angles as a function of rotor thrustfrom the Brooks code and Eq. (3), µ = 0.072, 6x6m closedtest section.

0

1

2

3

4

0.05 0.10 0.15 0.20 0.25 0.30 0.35

Brooks Codeδw = 0.16, Ref. 2

Ang

le-O

f-A

ttack

Cor

rect

ion

∆α

, deg

Advance Ratio, µ

Figure 9. Correction angles as a function of advance ratiofrom the Brooks code and Eq. (3), cT = 0.005, 6x6mclosed test section.

Figure 10 is a comparison of the angle-of-attackcorrections as a function of advance ratio for the 8x6mclosed test section. The calculations shown in Fig. 10 arefor a fixed value of rotor thrust (cT = 0.005) similar tothose shown in Fig. 9. Unlike Fig. 9, the difference inthe ∆α-corrections between the two methods is negligibleover the whole speed range.

Figures 8-10 show that there can be differences in thecalculated ∆α-corrections between the Glauert formula andthe Brooks code. At this point it is not known whichmethod is more accurate. It may be that the boundarycorrection factor, δW for the 6x6m closed test section hasto be re-determined. A new value for δW can bedetermined by feeding the ∆α value from the Brooks code

into Eq. (3) as will be discussed in the next section of thepaper.

0

1

2

3

0.05 0.10 0.15 0.20 0.25 0.30 0.35

Brooks Codeδw = 0.119, Ref. 2

Ang

le-O

f-A

ttack

Cor

rect

ion

∆α

, deg

Advance Ratio, µ

Figure 10. Correction angles as a function of advanceratio from the Brooks code and Eq. (3), cT = 0.005, 8x6mclosed test section.

Correction Factors Derived From The BrooksCode

The Brooks code ∆α-results were fed back into Eq.(3) resulting in new boundary correction factors, δW asgiven in Table 3. The boundary correction factors in Table3 can be applied to all cT's and all µ's. This is importantto know because it allows on-line global angle-of-attackcorrection within a test program. Using δW from Table 3,the differences between ∆α from the Brooks code and theGlauert formula are negligible. Also shown in Table 3 arethe corresponding values for rotor diameter to wind tunneltest section width.

Table 3. Boundary correction factors (δW) as determinedby the Brooks code.

Test Section D/W δW

DNW 6x6m closed 0.667 0.1353

DNW 8x6m closed 0.500 0.1163

DNW 9.5x9.5m closed 0.421 0.1345

DNW 8x6m open 0.500 -0.1775

DNW 8x6m 12% slotted* 0.500 -0.0081

*δW determined from experiment

The boundary correction factor for the open testsection given in Table 3 is related to an 8x6m effectivetest section size. Due to flow contraction, the effectivetunnel dimensions may be somewhat smaller (e.g.,

7x5m). An exact value can hardly be given as it dependson the rotor condition (e.g., thrust) and its position withrespect to the tunnel flow.

The wall correction factor for the 8x6m slotted walltest section is of high interest as it indicates that this testsection needs only negligible ∆α -corrections for rotortesting. This has been confirmed by fixed wingmeasurements. In addition, fixed wing measurements haveindicated that α -corrections are necessary along thelongitudinal axis of the model in the flow direction due toa longitudinal velocity gradient (Ref. 7). The negativesign of boundary correction factor (δW) shows that theslotted test section behaves more like an open sectionthan a closed section.

Although the Brooks code is not applicable for the8x6m slotted wall configuration, the boundary correctionfactor can be determined based on the measured resultsfrom the other test sections. Since it will be shown in asubsequent section of the paper that corrections fordifferent test sections will collapse the power data onto asingle curve, one can use this data to extrapolate to a testsection that cannot be represented by the Brooks code ifone assumes that these corrections are accurate. From thepower difference between the corrected and uncorrected dataone can determine the ∆α, and thus the δW with the helpof Eq. (3). Strictly speaking, this procedure may only beused for identical test configurations.

Correction By Experiment

A prerequisite for the application of this method isthat data from flight tests (e.g., rotor power and shaftangle) or from calculations are available and reliable. Anapplicable procedure is to trim the rotor in the windtunnel to the measured rotor power from flight test byadjusting the shaft angle while maintaining constant cL.This is the so-called trim to torque method (TtoT). Thismethod allows for the determination of the differencebetween α FT and α WT and thus the angle-of-attackcorrection (∆α ) due to wall interference, assuming allother differences (i.e., scale, inflow, Reynolds No., ...) arenegligible. The TtoT procedure is described in more detailin a later section of the paper.

Figures 11-13 show the variation in rotor power as afunction of rotor shaft angle (αshaft) for three differentadvance ratios. A dashed horizontal and vertical linerepresent the regression curve-fits of the rotor mast powerand the flight test shaft angle (see also Fig. 4). Theintersection between the rotor shaft angle in the windtunnel and power from flight test is the correspondingfree-flight condition. These figures show that with

0.00025

0.00030

0.00035

0.00040

0.00045

-25 -20 -15 -10 -5 0 5 10 15

9.5x9.5m closed8x6m open8x6m closed6x6m closed8x6m 12% slotted

Rot

or P

ower

Coe

ffici

ent,

CP

Rotor Shaft Angle, deg

From Flight TestRegression Curves

Figure 11. Rotor shaft power versus rotor shaft angle,µ = 0.056, cL = cweight = 0.005.

0.00020

0.00022

0.00024

0.00026

0.00028

0.00030

-7 -6 -5 -4 -3 -2

9.5x9.5m closed8x6m open8x6m closed6x6m closed8x6m 12% slotted

Rot

or P

ower

Coe

ffici

ent,

CP

Rotor Shaft Angle, deg

From Flight TestRegression Curves

Figure 12. Rotor shaft power versus rotor shaft angle,µ = 0.172, cL = cweight = 0.005.

0.00040

0.00045

0.00050

0.00055

-11 -10 -9 -8 -7

8x6m open8x6m closed6x6m closed

Rot

or P

ower

Coe

ffici

ent,

CP

Rotor Shaft Angle, deg

From Flight TestRegression Curves

Figure 13. Rotor shaft power versus rotor shaft angle,µ = 0.32, cL = cweight = 0.005.

increasing speed, the shaft angle difference between theintersection of the linear curve-fits and the flight testregression curve data becomes smaller between thedifferent test sections. This is consistent with the wallcorrection theories where the wall induced angle-of-attackcorrections reduce with increasing advance ratio, µ.

Corrections For Scaling Effects

Rotor loads and performance testing at model-scalerequires careful design and manufacturing to simulate full-scale rotor behavior. In particular, the tip Mach numbermust be very close or equal to that of the full-scale rotorto capture the influence of compressibility on bladesection loading. Blade per rev natural frequencies and Locknumber must also be very close to provide similar coningand flapping. However, the Reynolds number for the scalemodel rotor will not match the full-scale values unless themodel is run at higher pressures, either in a pressurizedtunnel or by changing the working fluid of the tunnel.

Table 4. Comparison of full-scale (flight and wind tunnel)and model-scale rotors.

model Bo105 Factor*

Rotor Diameter [m] 4.0 9.82 s-1

Rotor Speed [rpm] 1040 424 s

Blade Chord [m] 0.121 0.27 0.91s

Blade Twist [deg] -8 -8 1

Tip Speed [m/s] 218 218 1

Solidity, σ 0.077 0.07 0.91s

Hub Precone Angle [deg] 2.5 2.5 1

Tip Mach No. 0.64 0.64 1

Re-No. at r/R = 0.7, x10-6 1.26 2.82 2.24

Blade Profile NACA 23012 mod.

*Scale Factor s = 2.456

The 40% scale model Bo105 rotor tested in air at theDNW was designed according to the criteria stated above.Following conventional Mach scaling rules with bothrotors running at a tip Mach number of 0.64, the full-scale and model-scale Reynolds numbers at 70% radius are2.82 and 1.13 million respectively. Wind tunnelmeasurements conducted by Messerschmitt-Bölkow-Blohm GmbH (MBB) on a NACA 23012 airfoil haveshown that this reduction in Reynolds number causes adecrease in cL,max from 1.58 to 1.35, or about 15%, and adecrease in the lift-curve slope of about 7.5%. Therefore,the airloads for the model blade would be low with respectto the inertial and elastic forces when compared to the

full-scale rotor. To compensate for this relative decrease inairloads at model-scale, the blade chord was increased by10%. The influence of this increase in the model-scalechord is assessed below. Important rotor parameters andthe resulting scale factors for the full-scale and model-scale rotors are given in Table 4. The influence ofReynolds number on rotor performance for model-scalerotor testing was addressed in Ref. 18. In Ref. 18 it wasshown that the non-dimensional power consumption for aconventionally Mach scaled model rotor was higher than afull-scale rotor under the same conditions. From Ref. 18,

∆cP

0

= cP

0,FS

- cP

0,mo

(8)

∆cP

0

=

σ ∆cd

0

8 (1+ 4.6 µ

2)

(9)

cd

0,FS

cd

0,mo

= ( ReFS

Remo)

5

(10)

Since Remo < ReFS and from Eq. (10) is

cd

0,mo

> cd

0,FS

From (8) and (9)

∆cd

0

< 0 and ∆cp0 < 0

Therefore

cP

O,FS

< cP

O,mo

This was shown in Ref. 18, as an offset in the powercoefficient versus thrust coefficient curves for a full-scaleand a one-sixth scale CH-47D rotor in hover. A similarcomparison for the rotors discussed here shows no offsetin the power coefficient versus thrust coefficient curvesbetween the full-scale Bo105 and the 40% scale modelBo105 with 10% increase in chord length. This may beseen in Fig. 14, where the full-scale rotor was tested in

the NASA Ames 40- by 80-Foot Wind Tunnel and thescale model rotor was tested in the DNW 8x6m open jettest section at the same tip Mach number. The goodagreement in hover performance between the full- andmodel-scale rotors indicate that the increase in chordlength for the model-scale rotor was an effective means ofcompensating for the Reynolds number difference,although differences due to profile roughness,compressibility effects, and the body or test stand werenot investigated.

0.00000

0.00020

0.00040

0.00060

0.00080

0.000 0.002 0.004 0.006 0.008

8x6m open40x80ft

Rot

or P

ower

Coe

ffici

ent,

CP

Rotor Thrust Coefficient, CT

Figure 14. Comparison of full- and model-scale rotorpower as a function of rotor thrust in hover.

ROTOR TRIM PROCEDURES

The goals of the 40% scale model DNW test were tomeasure rotor performance and blade loads on a model-scale rotor for the same conditions as were measured on aflight vehicle by the DLR, and to assess the wallcorrections of the different test sections used at the DNW.To acquire wind tunnel data suitable for the first goal onemust determine what is required to establish the samecondition in the wind tunnel that existed for a particularflight point. Several different methods of adjusting therotor control to achieve comparable trim conditions withthe flight data were investigated.

During the full-scale Bo105 rotor test in the 40- by80-Foot Wind Tunnel, three different trim conditions wereestablished to compare with flight; minimized flappingtrim, prescribed hub moment trim, and prescribed cycliccontrol angle trim (Ref. 15). All of these methods had incommon the same rotor speed, shaft angle, rotor thrust,and tunnel speed. The prescribed hub moment trimrequires the rotor operator to adjust the cyclic stick controluntil the steady hub pitch and roll moments agree withvalues measured from the flight aircraft. Minimized

flapping trim requires the rotor operator to adjust thecyclic stick control until the 1p flap moment in theflexure portion of the blade is a minimum (i.e., nearzero). Cyclic control angle trim requires the rotor operatorto adjust the cyclic stick control until the longitudinal andlateral cyclic displays agree with values that weremeasured on the flight aircraft. These different trimprocedures cause the rotor tip-path-plane to deviate fromone another resulting in different hub loads. Ideally, theprescribed hub moment trim and cyclic control angle trimmethods would result in very similar conditions, however,the measurements shown in Ref. 15 show that theresulting hub loads were very different. Hub loads forminimized flapping trim were expected to be differentfrom the other two trim procedures and this was the case.

0.00020

0.00024

0.00028

0.00032

0.10 0.15 0.20 0.25 0.30

HM trim; 8x6m closedMF trim; 8x6m closedHM trim; 8x6m openMF trim; 8x6m open

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 15. Comparison of rotor power as a function ofadvance ratio for minimized flapping (MF) and hubmoment (HM) trim, cT = 0.005.

Prescribed hub moment and minimized flapping trimwere again used in the model-scale DNW test. Figure 15compares the rotor power coefficient versus advance ratiofor these two trim methods in the 8x6m closed and 8x6mopen jet test sections in the DNW. Both the trim methodand the test section are seen to have a strong influence onthe rotor power. This shows the importance of choosingan accurate trim procedure. It further shows thesignificance of the test section used in the wind tunnel anddemonstrates that some type of correction is required. Thelarge differences shown in Fig. 15 along with thecomparison with flight data shown in Ref. 15 indicatethat minimized flapping trim is not suitable forcomparisons with flight data. It does, however, provide awell-defined condition that will certainly be used in windtunnel testing for a long time. The cyclic control angletrim has been investigated by the DLR and NASA andwas found to produce large differences in hub loads when

compared to flight data, so it was not used in the DNWtest.

The experimental test set-up in the DNW allowedadditional methods of achieving the flight trim conditions.Because the rotor and the model-scale fuselage weremounted independently of each other on their own balancemeasuring device, the rotor and fuselage forces andmoments could be measured independently. Independentmeasurements of rotor lift and fuselage lift provided thecapability to trim the rotor lift to match the sum of theflight aircraft weight and the vertical force on the fuselageas defined by Eq. (11).

- cWeight = cL,Ro + cL,Fus = cL (11)

This method removes the inaccuracy of assuming thatrotor thrust is equal to aircraft weight, which is often usedbecause cL,Ro and cL,Fus from flight are normally notknown. At low speed the rotor induces a download on thefuselage producing a negative value of cL,Fus and at highspeed the fuselage aerodynamics are likely to producesignificant negative value of cL,Fus. Trimming to lift asdescribed above, assumes instead that the rotor wake andthe model-scale fuselage behave in a similar manneraerodynamically to the flight aircraft.

Until now it had been assumed that the shaft anglemeasured in flight was accurate enough to establishcomparable flight test trim conditions in the wind tunnel.Figure 15 clearly indicates that there is a significantinfluence on rotor power due to the wind tunnel testsection. Therefore, a means of correcting for wall effectsis required. An alternative approach to achieving the rotortrim of flight was evaluated for the first time in theDNW. The rotor was trimmed to rotor speed, rotor liftcoefficient as defined by Eq. (11), tunnel speed, and hubpitch and roll moment coefficient. However, instead oftrimming to the flight shaft angles, the shaft angle wasadjusted until the measured shaft torque coefficient agreedwith the flight measurement. Results for this trim totorque method (TtoT) are shown in Figs. 11-13. Sinceflight test data normally show a certain amount of scatter,three shaft angles were tested which bracket the flightpower coefficient. From these tests, linear relationships ofthe change in the rotor power coefficient as a function ofrotor shaft angle was established at each forward speedtested. These relationships and how they were used arediscussed in a subsequent section of the paper.

Another approach to acquiring relationships similarto the trim to torque method, is to acquire data with smallpositive and negative perturbations in all the trim controls(e.g., V, α , θ0.7, θc, θs) around a baseline condition.

This derivative approach provides the information tocorrect the measured performance data for any parameterthat turns out to be off the target trim value. However, itdoes require a significant investment in tunnel occupancytime to gather all the necessary derivatives. For the DNWtest, the derivatives were taken after trimming to rotorspeed, shaft angle, rotor thrust coefficient, tunnel speed,and hub pitch and roll moment coefficient. The results aremore or less linear relationships between the controlvector elements and the hub load vectors.

Since only the differences are of interest one can write

∆F = D * ∆C ,

where each element of matrix D gives the slope of loadvector F and control vector C.

The differences between the closed and open testsection are clearly seen in the derivative data. Thederivatives appear linear with advance ratio within thescatter of the data.

Representative results are shown in Figs. 16-18,where the change in thrust due to the change in threecontrol vector elements are shown.

The derivative method is more general than the trimto torque method and was therefore applied for angle-of-attack corrections within the minimized flapping, zeroshaft tilt test series. This will be shown in a later sectionof the paper.

600

700

800

900

1000

1100

0.10 0.15 0.20 0.25 0.30

8x6m closed8x6m open6x6m closed

∆F

z/∆

θ 0.7

, N/d

eg

Advance Ratio, µ

Figure 16. Collective derivative as a function of advanceratio for the 40% scale model rotor, hub moment trim.

0

100

200

300

400

500

0.10 0.15 0.20 0.25 0.30

8x6m closed8x6m open6x6m closed

∆F z

/ ∆θ s

, N/d

eg

Advance Ratio, µ

Figure 17. Longitudinal cyclic derivative as a function ofadvance ratio for the 40% scale model rotor, hub momenttrim.

50

100

150

200

250

300

350

400

450

0.10 0.15 0.20 0.25 0.30

8x6m closed8x6m open6x6m closed

∆Fz/∆

α s, N

/deg

Advance Ratio, µ

Figure 18. Shaft angle derivative as a function of advanceratio for the 40% scale model rotor, hub moment trim.

RESULTS

Trim Comparisons

Once the trim procedure has been established,experimental consideration determines whether thesetargets are simultaneously achieved. In this section, theability to match hub moments and a comparison of theresulting 1p blade flapping moments are shown. Theflight test trim target was established as steady level flightwith minimum sideslip and no control input during datagathering. Steady hub moments were derived fromharmonic analysis of the strain gauge signals on therotating shaft.

In the last section several wind tunnel trim procedureswere described. For the data shown in this section a singletrim procedure was used for each wind tunnel. The full-scale test in the 40- by 80- Foot Wind Tunnel definedtrim as matching rotor speed, shaft angle, tunnel velocity,rotor thrust equal to aircraft weight, hub pitch moment,and hub roll moment matching flight measurements. Thehub moments were derived from balance measurementsmade below the rotor hub yet transferred to the hub center.The 40% scale-model test in the DNW defined trim asmatching rotor speed, shaft angle, tunnel velocity, rotorlift equal to the sum of the aircraft weight and the verticalforce on the scale model fuselage, 1p cosine shaftbending, and 1p sine shaft bending matching flightmeasurements.

-2000

-1500

-1000

-500

0

0.05 0.10 0.15 0.20 0.25 0.30 0.35

Flight Test40x80ftDNW, all TS

Rol

ling

Mom

ent,

Nm

Advance Ratio, µ

Figure 19. Comparison of measured hub roll moment as afunction of advance ratio, cL = 0.005.

0

500

1000

1500

2000

2500

3000

0.05 0.10 0.15 0.20 0.25 0.30 0.35

Flight Test40x80ftDNW, all TS

Pitc

hing

Mom

ent,

Nm

Advance Ratio, µ

Figure 20. Comparison of measured hub pitch moment asa function of advance ratio, cL = 0.005.

-300

-200

-100

0

100

0.0 0.2 0.4 0.6 0.8 1.0

Flight Test40x80ft8x6m closed

1p C

osin

e B

lade

Fla

p M

omen

t, N

m

Blade Span r/R

Figure 21. Longitudinal 1p flap moment distributionversus span, cL = 0.005, µ = 0.32.

-100

-50

0

50

100

150

0.0 0.2 0.4 0.6 0.8 1.0

Flight Test40x80ft8x6m closed

1p S

ine

Bla

de F

lap

Mom

ent,

Nm

Blade Span r/R

Figure 22. Lateral 1p flap moment distribution versusspan, cL = 0.005, µ = 0.32.

The ability to match the flight test targets may beseen in Figs. 19-20, where the 40% scale model rotor datahave been scaled up for comparison. Figure 19 comparesthe rolling moment for these three tests. The flight datashow relatively little scatter except at low forward speedwhere the speed indication is least accurate. In general thecorrelation is good. Figure 20 compares the pitchingmoment for these three tests. The pitching moment is ofparticular interest because it has a direct influence on thelongitudinal tip-path-plane tilt, and since typical windtunnel wall corrections impose an angle-of-attackcorrection, factors affecting tip-path-plane angle-of-attackare clearly important. Flight data shows more scatter forpitch moment than was seen for roll moment. This ismost likely due to basic aircraft stability, atmospheric

unsteadiness, and slight deviations from steady level flightequilibrium. The 40- by 80- Foot Wind Tunnel data isslightly off the flight test target values at low speed andmore significantly for the highest speed point whereunsteadiness in the tunnel flow made it difficult tomaintain a steady hub pitching moment. Unfortunately,there was only time for a single data point at each speed.It is expected that had more time been made available forgathering this data the agreement with pitch momentswould be better.

Another way to assess the ability to which windtunnel trim agrees with flight trim is to look at theresulting flapping moment along the span of the blades.Figure 21 shows the cosine component of the 1p flappingmoment for the comparable highest speed points. Themoment distributions look very similar except in the rootsection for the 40- by 80- Foot Wind Tunnel data. Thereare two things that can possibly explain this difference.Blade-to-blade dissimilarities as was shown in Ref. 15 canbe one explanation for these differences. The otherexplanation to these differences is that the pitch momentdoes not agree well with the flight target value at thisspeed. Figure 22 shows the sine component of the 1pflapping moment with span for the comparable highestspeed points. Better agreement is seen here over the entirespan due to the good agreement in hub roll moments.

Wall Induced Corrections

The primary focus of this paper is the presentation ofthe results from a series of systematic tests conducted inthe DNW to identify the influence of tunnel walls onmeasured rotor performance and to evaluate the ability ofexisting wall correction methodologies to minimizefacility dependent effects. To quantify the influence on themeasured rotor performance of the various DNW testsection configurations, two different approaches for rotortrim were used throughout this test program. Minimizedflap bending moment trim was used to investigate theinfluence of wall induced effects on measured rotorperformance as a function of rotor thrust and advanceratio. Prescribed hub moment trim, where the cycliccontrols were adjusted until rotor hub moments matchedvalues measured during previous flight testing were usedto make comparisons with free-flight rotor performance inaddition to investigating wall induced effects.

To quantify the influence of the tunnel walls on themeasured rotor performance both as a function of forwardspeed and rotor thrust, a series of thrust sweeps utilizing aminimized flap moment trim was conducted at discreteforward speeds in each of the five different test sectionconfigurations. Due to forward speed limitations in some

0.00010

0.00015

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.002 0.003 0.004 0.005 0.006 0.007 0.008

9.5x9.5m closed8x6m open8x6m closed6x6m closed8x6m 12% slotted

Rot

or P

ower

Coe

ffici

ent,

CP

Rotor Thrust Coefficient, CT

Figure 23. Comparison of rotor power as a function ofrotor thrust for five different DNW test sections,minimized flap bending trim, µ = 0.106, αs = 0°.

0.00010

0.00015

0.00020

0.00025

0.00030

0.002 0.003 0.004 0.005 0.006 0.007 0.008

8x6m open8x6m closed6x6m closed

Rot

or P

ower

Coe

ffici

ent,

CP

Rotor Thrust Coefficient, CT

Figure 24. Comparison of rotor power as a function ofrotor thrust for three different DNW test sections,minimized flap bending trim, µ = 0.251, αs = 0°.

of the test section configurations, not all test conditionswere repeated. The influence of the tunnel test section onrotor power with minimized flap moment trim is clearlyseen in Figs. 23-24. Figure 23 shows the rotor power as afunction of rotor thrust at low speed (µ = 0.106) for thefive different test section configurations. Figure 24 showsthe rotor power as a function of rotor thrust at high speed(µ = 0.251) for the three test sections for which data isavailable. The data presented in Figs. 23-24 have not beencorrected for the wall induced effects. The data shown inthese two figures quantify the influence of the test sectionsize on rotor power as measured in this test program. Thetrends with rotor thrust are very clear in both of these

figures. The data demonstrate that the influence of thetunnel walls increases as rotor thrust increases. It alsoshows that the measured power is lower with relativelysmaller test sections. Both of these findings are consistentwith existing wind tunnel wall correction methodologies.Differences between the 6x6m and the 8x6m closed testsections are much greater in Fig. 23 at low speed thanthey are in Fig. 24 at high speed. This trend is alsoconsistent with existing wall correction theories.

The influence of the wall induced effects on rotorpower as a function of tunnel speed is more clearly seenin Fig. 25. This figure shows rotor power as a function oftunnel speed with rotor thrust and hub moments adjustedto match flight test measurements. Data is presented forfive different test section configurations. The datapresented in Fig. 25 has not been corrected for the wallinduced effects. The influence of the tunnel walls reducesas the tunnel speed increases. This is consistent with themeasurements shown in Figs. 23-24 for minimized flapmoment trim. These uncorrected measured data usingprescribed hub moment trim in the different test sectionswill be referred to as 'baseline' measurements.Measurements with corrections for the influence of thetunnel walls or corrections to provide equivalent trimconditions will be compared to these baselinemeasurements.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.05 0.10 0.15 0.20 0.25 0.30 0.35

9.5x9.5m closed8x6m open8x6m closed6x6m closed8x6m 12% slotted

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 25. Comparison of rotor power as a function ofadvance ratio without wall corrections for five differentDNW test sections with identical rotor trim conditions.

To quantify the influence of the tunnel walls on themeasured rotor performance and to evaluate the existingwall correction methodologies upon completion of theDNW test program, a series of tests (trim to torque) wasdeveloped and utilized in each of the test sectionconfigurations. To allow for interpolation of the change

-10

-5

0

5

10

0.05 0.10 0.15 0.20 0.25 0.30 0.35

8x6m open, BaselineBrooks CodeGlauert Eq.

Rot

or S

haft

Ang

le, d

eg

Advance Ratio, µ

Figure 26. Comparison of baseline and wall corrected rotorshaft angle as a function of advance ratio in the 8x6mopen jet test section.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.05 0.10 0.15 0.20 0.25 0.30 0.35

8x6m open, BaselineBrooks CodeGlauert Eq.

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 27. Comparison of baseline and wall corrected rotorpower coefficient as a function of advance ratio in the8x6m open jet test section.

in rotor power as a function of rotor shaft angle, the rotorwas trimmed with rotor thrust and hub moments adjustedto match flight test measurements at several rotor shaftangles around the measured flight test shaft angle. Fromthese tests, a linear relationship of the change in the rotorpower coefficient as a function of rotor shaft angle wasestablished at each forward speed tested.

Two wall correction methodologies were selected forevaluation. The wall correction methodologies selected arethe classical wall-correction method of Glauert (GlauertEq.) and the Brooks computer code (Brooks Code). These

methods are presented and discussed in a previous sectionof the paper. The results of these evaluations are presentedin Figs. 26-31.

Comparison of baseline and corrected rotor shaftangle and rotor power coefficient data as a result of thetwo wall correction methodologies for the 8x6m open jettest section are shown in Figs. 26-27. Figure 26 presentsa comparison of the wall corrected shaft angles (includes∆α-correction) with the baseline shaft angle as a functionof advance ratio for the two wall correctionmethodologies. The rotor shaft angle requirements asdetermined by the Glauert equation and the Brookscomputer code are very similar. For advance ratios greaterthan 0.15, the shaft angles are essentially identical. Smalldifferences (less than one deg) in rotor shaft anglerequirements were determined for advance ratios less than0.15. The corresponding comparison of rotor powercoefficient for each of the correction methodologies withthe baseline as a function of advance ratio is shown inFig. 27. Using the ∆α -correction's from Fig. 26, bothmethodologies result in nearly the same rotor power overthe entire speed range with negligible differences at thelowest speeds.

Results for the 8x6m closed test section are shown inFigs. 28-29. Comparisons of the baseline wind tunnelshaft angle as a function of advance ratio with the wallcorrected shaft angles are shown in Fig. 28. The Glauertequation and the Brooks computer code calculate nearlyidentical values for shaft angle over the entire speed range.Comparisons of the corresponding rotor powercoefficients with advance ratio are shown in Fig. 29.Based on the nearly identical corrected shaft angle resultsshown in Fig. 28, it is not surprising that the rotor poweris nearly identical over the entire speed range as shown inFig. 29.

Comparison of the results for each of the four testsections using the same correction methodology in each isshown in Figs. 30-31. Results using the Brookscomputer code correction method are shown in Fig. 30.Figure 31 presents the results for each test section usingthe Glauert equation correction method. The results for the8x6m open and closed test sections shown in Figs. 30-31are the same as those presented in Figs. 26-29 comparingthe different correction methodologies in each test section.Although direct comparisons of the two correctionmethodologies for the 9.5x9.5m and 6x6m closed testsections are not shown in the paper, the shaft angle androtor power coefficient results were found to be verysimilar to those shown in Figs. 26-29.

-10

-8

-6

-4

-2

0

0.05 0.10 0.15 0.20 0.25 0.30 0.35

8x6m closed, BaselineBrooks CodeGlauert Eq.

Rot

or S

haft

Ang

le, d

eg

Advance Ratio, µ

Figure 28. Comparison of baseline and wall corrected rotorshaft angle as a function of advance ratio in the 8x6mclosed test section.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.05 0.10 0.15 0.20 0.25 0.30 0.35

8x6m closed, BaselineBrooks CodeGlauert Eq.

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 29. Comparison of baseline and wall corrected rotorpower coefficient as a function of advance ratio in the8x6m closed test section.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.05 0.10 0.15 0.20 0.25 0.30 0.35

9.5x9.5m closed8x6m open8x6m closed6x6m closed

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 30. Comparison of wall corrected rotor powercoefficient as a function of advance ratio using the Brookscode correction in four different DNW test sections.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.05 0.10 0.15 0.20 0.25 0.30 0.35

9.5x9.5m closed8x6m open8x6m closed6x6m closed

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 31. Comparison of wall corrected rotor powercoefficient as a function of advance ratio using the Glauertequation correction in four different DNW test sections.

Results using the Brooks code are shown in Fig. 30for each of the test sections. As seen in Fig. 30, thecorrections based on the Brooks computer code havecollapsed the data from the 9.5x9.5m, 8x6m and 6x6mclosed test sections onto a single curve. For comparisonpurposes, refer to Fig. 25 that shows the uncorrected dataand the distinct differences in the measured rotor power foreach test section. The results for the 8x6m open jet testsection indicate that perhaps not enough correction wasmade to this set of data relative to the other three testsections. Further review of this data set indicated that theresultant propulsive force was greater in the 8x6m openjet test section than the other test sections despiteattempts to maintain consistent test conditions

throughout. The greater propulsive force in the 8x6mopen jet test section resulted in higher measured rotorpower.

The results of corrected rotor power using the Glauertequation correction method for each of the test sections arecompared in Fig. 31. The comparisons shown in Fig. 31are very similar to those shown in Fig. 30. This is to beexpected based on the comparisons of correctionmethodologies shown previously where the Brookscomputer code calculations and the Glauert equationcorrections were nearly the same. As seen in Fig. 31, theGlauert equation correction has collapsed the data from the9.5x9.5m, 8x6m and 6x6m closed test sections onto the

same curve. The results for the 8x6m open jet test sectionare very similar to that shown in Fig. 30 as is to beexpected based on the comparisons shown in Fig. 27.

Propulsive Force Trim

The testing approach used to account for wall inducedcorrections as discussed in the previous section alsoresulted in linear relationships identifying the change inpropulsive force with rotor shaft angle at each of theforward speeds tested. These relationships were determinedso that a consistent propulsive force equilibrium for therotor could be maintained during the evaluation of each ofthe wall correction methodologies. These relationshipswere used to correct the baseline values from each of theDNW test sections to the correct propulsive forcerepresenting the equivalent flat plate area of the Bo105helicopter and subsequently the influence of the tunnelwalls as defined by an angle-of-attack correction from eachof the wall correction methodologies under evaluation.For the Bo105 helicopter (including rotor hub and shaft),the equivalent flat plate area is 1.33 m2 (14.32 ft2) asprovided by Eurocopter Deutschland (ECD).

0

1

2

3

4

5

0.05 0.10 0.15 0.20 0.25 0.30 0.35

9.5x9.5 closed8x6 open8x6 closed6x6 closed

Equ

ival

ent F

lat P

late

Are

a, m

2

Advance Ratio, µ

1.33 m2

Figure 32. Comparison of equivalent flat plate areas asmeasured in each of the four DNW test sections with theECD provided value of 1.33 m2.

Figure 32 presents a comparison of the equivalent flatplate areas as determined from data measured in each of thefour DNW test sections as a result of trim conditionsbased on the shaft angle measurements from flight testand Eq. (11). Also shown on this figure is a horizontalreference line representing the equivalent flat plate areavalue of 1.33 m2 as provided by ECD. For all advanceratios greater than 0.10, the equivalent flat plate area asdetermined in the wind tunnel is less than this value. Thedifferences identified in this figure are the basis for the

propulsive force correction to the wind tunnel data usingthe linear relationships of the change in power andpropulsive force with rotor shaft angle.

The first step in the analysis of this data was tocorrect the baseline wind tunnel data from each of theDNW test sections, in this case the rotor propulsive forceto that which is determined by the equivalent flat platearea. By moving up or down the linear relationship ofpropulsive force as a function of rotor shaft angle at eachforward speed, a corrected value of rotor shaft angle wasestablished by matching the propulsive force to theprescribed flat plate area of 1.33 m2. This corrected valuefor rotor shaft angle primarily addresses the uncertaintyand scatter of the flight test measurements. The otheruncertainty that is not addressed directly in this analysis isthe uncertainty in the measurement of the aircraft speed.

-12

-10

-8

-6

-4

-2

0

0.05 0.10 0.15 0.20 0.25 0.30 0.35

8x6m closed, Baseline8x6m closed, Propulsive Force TrimR

otor

Sha

ft A

ngle

, deg

Advance Ratio, µ

Figure 33. Comparison of baseline shaft angle with shaftangle for propulsive force trim in the wind tunnel as afunction of advance ratio.

Figure 33 shows a representative comparison of thebaseline wind tunnel rotor shaft angle (without wallinduced corrections) as determined from flight test withthe rotor shaft angle due to propulsive force corrections ofthe wind tunnel data as a function of advance ratio. Thiscorrected rotor shaft angle value at each forward speed wasthen used to determine the corrected rotor powercoefficient using the linear relationship of the change inrotor power coefficient as a function of rotor shaft angle.A comparison of corrected rotor power coefficient basedon propulsive force trim (without wall inducedcorrections) with measured rotor power in the 8x6mclosed test section based on rotor shaft angle from flighttest as a function of advance ratio is shown in Fig. 34.Also shown on Fig. 34 is the measured rotor power datafrom flight test for three different speed sweeps as

discussed previously in this paper. The importance ofpropulsive force trim corrections to the measured windtunnel data is evident in Fig. 34 at the higher advanceratios. Evaluations with different equivalent flat plateareas were also conducted. From these evaluations itappears that the value provided by ECD was indeed themost appropriate or representative value for thesecomparisons.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.00055

0.05 0.10 0.15 0.20 0.25 0.30 0.35

Flight Test8x6m closed, Baseline8x6m closed, Propulsive Force Trim

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 34. Comparison of baseline and propulsive forcetrim corrected rotor power coefficient in the wind tunnelas a function of advance ratio with flight testmeasurements.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.00055

0.00060

0.05 0.10 0.15 0.20 0.25 0.30 0.35

PFT ReferenceBrooks CodeGlauert Eq.

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 35. Comparison of propulsive force trim (PFT)reference and wall corrected rotor power coefficient as afunction of advance ratio in the 8x6m open jet testsection.

Results using the propulsive force trim analysis withand without wall corrections applied for the 8x6m openand closed test sections are shown in Figs. 35-36. Figure

37 compares the four different DNW test sections usingthe Brooks computer code correction method. All datapresented have been corrected to an equivalent flat platearea of 1.33 m2 (14.32 ft2) before application ofcorrections for wall induced effects.

Comparison of propulsive force trim reference andcorrected rotor power coefficient data as a result of the twowall correction methodologies for the 8x6m open jet testsection is shown in Fig. 35. This figure may be comparedwith Fig. 27 to show that propulsive force trim has littleinfluence on the final corrected rotor power coefficientresults for µ < 0.25. However, the wind tunnel wallcorrected rotor power coefficient result for propulsive forcetrim is 15% greater at µ = 0.32. As is to be expected fromthe results previously shown in Fig. 27, the correctedrotor power coefficients for the Glauert equation and theBrooks computer code methodologies are nearly identical.

Rotor power coefficient results for the 8x6m closedtest section are shown in Fig. 36. Comparison of thepropulsive force trim reference rotor power coefficientwith the corrected rotor power coefficients based on thetwo wall correction methodologies as a function ofadvance ratio as shown in Fig. 36 is very similar to thatshown in Fig. 29. As was shown in Fig. 29 and again inFig. 36, the results using either wall correction methodare virtually identical.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.00055

0.00060

0.05 0.10 0.15 0.20 0.25 0.30 0.35

PFT ReferenceBrooks CodeGlauert Eq.

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 36. Comparison of propulsive force trim (PFT)reference and wall corrected rotor power coefficient as afunction of advance ratio in the 8x6m closed test section.

Comparisons of the propulsive trim results for eachof the four test sections using the Brooks computer codecalculations are shown in Fig. 37. The comparisons arevery similar to those shown in Fig. 30, except that theresults for the 8x6m open jet test section have collapsed

onto essentially the same curve as that for the 8x6m and6x6m closed test sections at the higher speeds. Thisimprovement is largely due to the requirement forlongitudinal force equilibrium between different testsection data sets. At low speeds there still remain somedifferences as the influence of propulsive force on rotorpower is small or negligible up to advance ratios of 0.15as shown in Fig. 34. Remaining differences, althoughsmall, may indicate that not enough correction wasidentified by the correction methodology or there may beaccuracy and repeatability limitations within the differentdata sets that has not yet been identified.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.00055

0.00060

0.05 0.10 0.15 0.20 0.25 0.30 0.35

9.5x9.5m closed8x6m open8x6m closed6x6m closed

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 37. Comparison of wall corrected rotor powercoefficient as a function of advance ratio using the Brookscode correction in four different DNW test sections,propulsive force trim.

Model-Scale/Full-Scale Flight Correlation

In addition to the test series developed to evaluateexisting wall correction methodologies, another series oftests was conducted for direct comparison with dataacquired in the NASA Ames Research Center 40- by 80-Foot Wind Tunnel. These data points were acquired, forthe most part, at speeds other than those described in theprevious series of tests and did not include variations inrotor shaft angle about a baseline value to determine therotor power coefficient and propulsive force relationships.To apply the rotor power coefficient and propulsive forcecorrections to these data, relationships of the change inrotor power coefficient and propulsive force with rotorshaft angle as a function of advance ratio were determined.The relationships of rotor power coefficient andpropulsive force with rotor shaft angle as functions ofrotor thrust and advance ratio for the full-scale 40- by 80-Foot Wind Tunnel were determined from thecomprehensive data base reported in Ref. 14. These

relationships were then used to correct the correspondingdata for wall induced effects so that direct comparisons ofmodel- and full-scale wind tunnel test and flight test datacould be made as shown in Figs. 38-40. The wind tunneldata presented in Figs. 39-40 have been corrected to anequivalent flat plate area of 1.33 m2 (14.32 ft2).

Model- and full-scale results with and withoutcorrections for propulsive force and wall induced effectsfor this series of tests are presented in Figs. 38-40.Although this series of tests was conducted in each of thefour test sections, only the results for the 8x6m closedtest section are shown for clarity. The results for the othertest sections are very similar to the results shownpreviously in Figs. 26-31 and Figs. 35-37. Also shownin Figs. 38-40 is the corresponding flight test datapreviously shown in Fig. 34.

Comparisons of model- (8x6m closed) and full-scale(40x80ft) wind tunnel data without corrections for eitherpropulsive force trim or wall induced effects are shown inFig. 38. Measurements of rotor power from flight are alsoshown in Fig. 38. As seen in Fig. 38, the model- andfull-scale rotor power data as measured in the wind tunnelfor identical trim conditions compare reasonably well foradvance ratios greater than 0.15. It is clear from Fig. 38that both the model- and full-scale data do not comparewell with the flight test data. Comparisons (not shown)of the baseline equivalent flat plate areas as measured inthe wind tunnel for both the model- and full-scale rotorswere very similar to the data shown previously in Fig.32.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.00055

0.05 0.10 0.15 0.20 0.25 0.30 0.35

Flight Test8x6m closed40x80ft

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 38. Comparison of model- and full-scale rotorpower as a function of advance ratio without propulsiveforce trim or wall induced corrections with flight testmeasurements.

Model- and full-scale results that account for thepropulsive force trim to an equivalent flat plate area of1.33 m2 (without wall induced corrections) are shown inFig. 39 along with flight test measurements. Applicationof propulsive force corrections improved the correlation ofthe model- and full-scale wind tunnel data for advanceratios above 0.15, along with improvements in thecorrelation with flight test. Figure 39 clearly shows theimportance of the propulsive force trim corrections in thecorrelation of these wind tunnel and flight test results.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.00055

0.05 0.10 0.15 0.20 0.25 0.30 0.35

Flight Test8x6m closed40x80ft

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 39. Comparison of model- and full-scale rotorpower with flight test measurements as a function ofadvance ratio for propulsive force trim.

Rotor power results that include corrections for bothpropulsive force and wall induced effects from the 8x6mclosed test section and the 40- by 80-Foot Wind Tunnelare compared with flight test measurements in Fig. 40.Both the model- and full-scale wind tunnel data werecorrected using the Glauert equation correctionmethodology. The reason for selecting the Glauertequation correction method is that the Brooks codecalculations are really only applicable for rectangular testsection shapes. Because the 40- by 80-Foot Wind Tunnelis not rectangular, small errors in the analysis might beintroduced with the use of the Brooks code to determinethe angle-of-attack corrections. For advance ratios greaterthan 0.20, both the model- and full-scale results comparereasonably well with flight test. At the lower speeds, themodel-scale results tend to follow more closely to thetrends of the flight test, however there is sufficient scatterin the flight test data to be misleading in making thisstatement. Difficulties in maintaining a trimmedrepeatable condition in flight at these low speeds makes itdifficult to make any definitive statements regarding thecorrelation of either the model- or full-scale data withflight test measurements. Overall, the correlation of

model- and full-scale wind tunnel test data with flight testresults is reasonably good.

0.00020

0.00025

0.00030

0.00035

0.00040

0.00045

0.00050

0.00055

0.05 0.10 0.15 0.20 0.25 0.30 0.35

Flight Test8x6m closed, Glauert Eq.40x80ft, Glauert Eq.

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 40. Comparison of model- and full-scale rotorpower as a function of advance ratio with propulsive forcetrim and wall induced corrections with flight testmeasurements.

One possible explanation for the differences shown inFigs. 38-40 between the model- and full-scale results foradvance ratios less than 0.15 is that the geometry of thetwo test sections is sufficiently different so as to affect therotor downwash at the lower speeds. The 8x6m closed testsection is a rectangular tunnel, while the 40- by 80-FootWind Tunnel is a closed test section with semicircularsides of 20 ft radius. As suggested in Ref. 9, the additionof fillets or in this case the influence of the semicircularsides may be reducing the allowable downwash or lift ofthe rotor. This is offered only as a possible explanationfor these differences.

Model-Scale/Full-Scale Minimized FlappingTrim Correlation

As shown in the previous section, the propulsiveforce trim approach showed good correlation betweenflight and wind tunnel test rotor power. It is also possibleto avoid the propulsive force issue and just look atcomparing the measured performance of the two windtunnel tests. Data gathered with minimized flapping trimand αs = 0°, while not suitable for comparison with flighttest, are especially suited for wind tunnel to wind tunnelcomparisons. Figures 40-41 present a comparisonbetween the full-scale rotor in the Ames 40- by 80-FootWind Tunnel and the 40% scaled model rotor in theDNW. Since the rotor trim is well-defined with thisapproach, results can be directly compared.

0.00010

0.00015

0.00020

0.00025

0.00030

0.00035

0.05 0.10 0.15 0.20 0.25 0.30

8x6m closed8x6m open40x80ft

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 41. Comparison of model- and full-scale rotorpower as a function of advance ratio for minimizedflapping trim, αs = 0°, cT = 0.005.

Figure 41 shows full-scale rotor power in the 40- by80-Foot Wind Tunnel with rotor power for the 40% scaledmodel rotor in the open and closed DNW test sections.The α TPP is held constant as a result of the trimprocedure for all advance ratios. Full- and model-scale datafor the closed test sections are very similar indicating thatthe wall induced angle-of-attack corrections are of thesame magnitude. The differences between the open andclosed test sections also reduce as the tunnel speedincreases. This is consistent with the results presentedpreviously, indicating that the results are independent ofαTPP and in agreement with wall correction theory.

0.00010

0.00015

0.00020

0.00025

0.00030

0.00035

0.05 0.10 0.15 0.20 0.25 0.30

8x6m closed8x6 open7x5m open, effective40x80ft

Rot

or P

ower

Coe

ffici

ent,

CP

Advance Ratio, µ

Figure 42. Comparison of model- and full-scale wallcorrected rotor power as a function of advance ratio usingthe derivative correction method for minimized flappingtrim, αs = 0°, cT = 0.005.

Comparison of the power curves from the 40- by 80-Foot Wind Tunnel and the 8x6m closed test section asshown in Fig. 41, indicate that the different modelsupports (RTA and 40% model-scale fuselage on theDNW sting) appear to have negligible influence on rotorpower.

Figure 42 presents a comparison of the wall-correctedrotor power for the two wind tunnel tests. The angle-of-attack corrections were determined using the Brooks code,and the corresponding changes in power were determinedusing the derivative method. Since the speed derivatives(e.g., ∂cP/∂V) are non-linear, the derivative method isonly accurate for speeds where derivatives were measured.Unfortunately, the experimental derivatives were notobtained at the advance ratio values shown in Fig. 41.Therefore it was necessary to use the least-squares curvefit in Fig. 41 to provide uncorrected values of rotor powerat the advance ratio values for which derivatives wereavailable. Linearity can be assumed for the α-derivativefor small perturbations thus providing the powercorrections required for the calculated ∆α's.

Due to the uncertainty of the effective test sectionsize of the 8x6m open test section (as previouslydiscussed), the wall induced angle-of-attack correctionswere also calculated for a 7x5m open test section. Thesmaller open test section size data collapses much betterwith the data from the 40- by 80-Foot Wind Tunnel andthe 8x6m closed test sections.

CONCLUDING REMARKS

The influence of wind tunnel test section physicalcharacteristics on measured rotor performance both as afunction of rotor thrust and advance ratio were clearlyshown.

The influence of wind tunnel wall induced interferenceon performance measurements can be compensated for bymeans of a global angle-of-attack correction. Both theGlauert equation and the Brooks computer code appear toprovide adequate angle-of-attack corrections for rotorperformance.

It was shown, that with proper trim procedures (i.e.,hub moment and propulsive force trim) and corrections forwall induced interference effects, that it is possible toacquire model- and full-scale performance data in the windtunnel that agrees well with flight measurements.

Differences in the model- and full-scale rotor supportsused in the two wind tunnel tests reported here appears tohave a negligible influence on the measured rotor power.

At nominal thrust values with the 40% scale modelBo105 rotor, the 8x6m slotted test section requires thesmallest angle-of-attack corrections as compared to theother available DNW test sections.

REFERENCES

1 Glauert, H., "The Interference on the Characteristics ofan Airfoil in a Wind Tunnel of Rectangular Section," R& M 1459, 1932.

2 Pope, Alan, and Harper, John J., Low-Speed WindTunnel Testing, John Wiley & Sons, Inc., 1966.

3 Heyson, H. H., "Use of Superposition in DigitalComputers To Obtain Wind Tunnel Interference Factorsfor Arbitrary Configurations, With Particular Reference toV/STOL Models," NASA TR R-302, 1969.

4 Heyson, H. H., "FORTRAN Programs For CalculatingWind-Tunnel Boundary Interference," NASA TM X-1740,1969.

5 Heyson, H. H., "Rapid Estimation Of Wind TunnelCorrections With Applications To Wind Tunnel AndModel Design," NASA TN D-6416.

6 Heyson, H. H., "Linearized Theory of Wind-Tunnel Jet-Boundary Corrections and Ground Effect for VTOL-STOLAircraft," NASA TR R-124, 1962.

7 Garner, H. C., Rogers, E. W., Acum, W. E. A., andMaskell, E. C., "Subsonic Wind Tunnel WallCorrections," AGARDograph 109, October 1966.

8 Rae, W. H. Jr., "Limits on Minimum-Speed V/STOLWind Tunnel Tests," Journal of Aircraft, Vol. 4, (3),May-June 1967, pp. 249-254.

9 Rae, W. H. Jr., and Shindo, S., "An ExperimentalInvestigation of Wind Tunnel Wall Corrections and TestLimits for V/STOL Wind-Tunnel Tests," U.S. ArmyGrant No. DA-ARO-31-124-G-809, Project No. 4506-E,AD-764 255, Dept. of Aeronautics and Astronautics,Univ. of Washington, July 1973.

10 Brooks, T. F., Jolly, J. R., and Marcolini, M. A.,"Helicopter Main Rotor Noise," NASA TP 2825, August1988.

11 Shinoda, P. M., "Wall Interaction Effects For A Full-Scale Helicopter Rotor In The NASA Ames 80- By 120-Foot Wind Tunnel," Paper Nº 20, AGARD 73rd FluidDynamics Panel Meeting and Symposium on Wall

Interference, Support Interference and Flow FieldMeasurements, Brussels, Belgium, October 1993.

12 Brooks, T. F., and C. L. Burley, C. L., "A WindTunnel Wall Correction Model for Helicopters in Open,Closed, and Partially Open Rectangular Test Sections,"NASA TM to be published August 1996.

13 Beaumier, P., Tung, C., Kube, R., Brooks, T. F. etal., "Effect of Higher Harmonic Control on HelicopterRotor Blade-Vortex Interaction Noise: Prediction andInitial Validation," AGARD Symposium on Aerodynamicand Aeroacoustic of Rotorcraft; Berlin, Germany; October11-13, 1994.

14 Peterson, R. L., "Full-Scale Hingeless RotorPerformance and Loads," NASA TM 110356, June 1995.

15 Peterson, R. L., Maier, T., Langer, H. J., andTränapp, N., "Correlation of Wind Tunnel and Flight TestResults of a Full-Scale Hingeless Rotor," Proceedings ofthe American Helicopter Society AeromechanicsSpecialist Conference, San Francisco, CA, January 1994.

16 Stephan, M., Klöppel, V., and Langer, H. -J., "A NewTest Rig For Helicopter Testing," Proceedings of theFourteenth European Rotorcraft Forum, Milan, Italy,1988.

17 Johnson, W., and Silva, F., Rotor Data ReductionSystem User's Manual for the National Full-ScaleAerodynamics Complex NASA Ames Research Center,1986.

18 Keys, C. N., Mc Veigh, M. A., Dadone, L., and McHugh, F. J., "Considerations In The Estimation Of Full-Scale Rotor Performance From Model Rotor Test Data,"Proceedings of the 39th Annual Forum of the AmericanHelicopter Society, St. Louis, MO, May 1983.