Correlation of flight, tunnel and prediction data on a helicopter main rotor

(1)

ERF91-67

CORRELATION OF FLIGHT, TUNNEL AND PREDICTION DATA ON A

HELICOPTER. MAIN ROTOR

G.PAGNANO, F.NANNONI, M.SIMONI

AGUSTA S.p.A., Cascina Costa - ITALY

H.J.LANGER

· D.L.R., Braunschweig - GERMANY

Abstract:

'Ihis paper presents the results of a detailed analysis perfonred on the available data of the wirrl tunnel testing perf onred at DNW on an isolated articulated 4-bladed main rotor model (refs. 1 an::i 2].

'!he correlation with prediction methods an::i flight tests data is

discussed in term.s of global data, i.e. power level, rotor forces, control an;;les an::i control loads.

Different prediction methods are applied, ran;;ing from energy

methods an::i sinplified trim algorithm to a blade element code; the codes are described in term.s of characteristics, input data an::i solution procedures, an::i level of confidence already gained with flight test data comparison.

A general discussion then follows on the effects of some simulation parameters, both in calculation methods an::i in wind tunnel modeling,

like the blade . dynamics representation and the rotor system configuration.

'!he differences in flight an::i tunnel measurement techniques an::i the

reduction procedures applied in the comparison of data from different sources are analyzed an::i discussed.

'!he conclusions state the level of confidence achieved in tunnel simulation and model · testing, an::i in the prediction of rotor characteristics (perfonnance an::i loads); further inprovements required

an::i future work in the inprovement of all techniques ( calculation an::i

(2)

1. IN.Imll:11_ICN

. '!he validation of prediction methods for perfonnance and loads of conventional helicopters rotor remains an important subject for the helicopter designer, asking repeated effort at any significant design ch.an3e or new technology application.

Both aerodynamic aspects, like full three dimensional flCM co:rrlitions at the blade tips and the interactions between wakes and

blades; and dynamics issues, like the aeroelastic tailoring of the blade design or the need to exterx:1 the flight envelope of conventional helicopters; all require an extensive experimental work (either win:i tunnel or flight) to produce firstly a valid data base and then a correlation activity to validate the prediction codes.

'Ibis paper deals with this aspect, based on data available from both tunnel and flight on the same rotor configuration.

We refer to the DNW testing corxiucted in the framework of the collaborative programme I.AH, on a mcx:iular model of a main rotor (ref.l and 2).

'!he test programme included a series of test points based on flight co:rrlitions; the scope was the correlation with flight tests, for the successive evaluation of blade m:xiification effects (both tip geometry

and twist distribltion).

'!he scope of this study is to understand the problems associated with this kin:i of comparison, starting with the global parameters like perfonnance and forces, using the results obtainable from prediction methods; also the control loads have been included, as provided as preliminary results from the applied codes.

'!he philosophy is that described as an integrated approach in a previous paper (ref .1) : the approach follCMed is therefore a starting point for the appreciation of features of tunnel model testing, with respect to the application to conventional rotor design, using available computer codes and flight test data.

2. DESCRIPI'ICN OF DATA AND 'IOOI.S

In the follCMing a short description of tests data and of the computational methods used for correlation iB provided.

Soire of the important differences due to configuration,

assumptions, scaling etc. are already mentioned below, while their effects are explained in the comparison part.

Table 1 in appendix presents the basic list of symbols used in the paper.

(3)

2 .1 Flight test data

2 .1.1. Data aa;iu.isiticn and pr:ocessirg

'!he helicopter configuration is the basic Al29 E. I. (Italian Anny} •

'!he prototype instrumentation is the standard arrangement far a flight developrent program.re; a digital (PCM} data acquisition system are used for global data of flight (OAT, pressure altitude, airspeed, etc) and a FM system far loads and paraneters of rotating parts data acx;iuisition.

'!he prototype was instrumented with nose l:xx:xn for speed

measurement, the outp.It. being directly in CAS: the accuracy obtainable is about 2 %.

'!he rotor torque can be measured by two sensors nounted at 90° on the mast, while a direct measure of thrust is not possible: the thrust is obtained by global data from flight with trimming methods; however, to reduce data from flight, instead of using the torque at M.R. , the total po.,1er output P__±_n±: (torque delivered by the engines) is used, after subtracting the 'Y.R po.,1er and account made for accessory power ( 40 HP} and M.G.Box efficiency (0.96): these simple fo:rnn..ll.as are applied, whose fair correlation has been dem::>nstrated with all flight test data

on the Al29 helicopter: ·

P _M.R.

= [ (

ptot - 40} x 0.96) / 1.10 P _{M.R . .}

= [ (

ptot - 40)

x

0.96) / 1.05

(hover CGE)

(µ, > 0.1)

'!he blade pitch angles are obtained from control position - blade pitch relations for each pilot canunarrl. '!he procedure used is the following: a longitudinal input is checked for blade pitch variation at 90° and 270°, with accuracy of ± 20' on 20°. ·

'!he pitch link loads are converted from the time domain to frequency domain applying a FFT for rotor revolution; to enable the

correlation with wirrl tunnel data the anplitude in the frequency damain are nonnalized with the maximum value in the range of 1 to 8 rev.

For a complete comparison of data, the time histories from flight

and wind tunnel tests are also plotted. ·

'!he blade instrumentation accounts for strain gages to nonitor both

berrling and torsion: table 3 in appendix shows the stations of the

full-scale blade compared with the positions on the nodel blades.

2.2 'l\mnel data

'!he testing was con:iucted at DNW, in the closed test section 8 x 6

m2 , in septernber 1989; the tunnel rotor support, DIR ROI'ESI' nounted on

DNW sting (including drive system, 110 kW hydraulic m:>tor; instrumentation, data acx;iuisition and processing) and related personnel, was provided by DIR Braunschweig, Institut filr Flugrnec.hanik. See fig.1. ·

(4)

·2. 2. 1. !b3el sebJp

am

dlaracteristics

'!he nxxiel rotor is Mach an::l dynamically-scaled; the rotor hub is geanetrically scaled with respect to the full scale configuration.

'!he· nxxiel scale is 29.4 % (1/3.4). '!he nxxiel diameter is therefore 3.5 m, while the main blade chord is 0.115 m. '!he solidity of l:oth.nrd.el rotor arxi full-scale rotor is

a

=

O. 084. ·

'!he

nost

significant differences fran full-scale are:

a. Isolated nxxiel rotor vs cc:rrplete helicopter

(Havever the presence of the

support

.fairing and of the hub fairi.I'Y; have also to be acc:ounted for)

b. Al::sence of lead-lag danpers in the nxxiel rotor

c. Tests in the tunnel are conducted at I _{zero flapping' con::lition,}

while the flight are tri.rrared with non zero flapping.

lateral tr.imn.in;J is not awlied in the tunnel, whereas it is

significant in flight.

d. Blade dynamics is only scaled up to best matching of first 5 frequencies (3 flap; 1 torsion; 1 chord)

Test parameters are:

Advance ratioµ.; shaft tiltin';J an:;Jle o:shaft;

Vertical force F; Propulsive force

FX'

. an::l related coefficients, based on: Cf= f ( (n R2 ~ V2 _tip]

'!he test corrlitions were based on three different full-scale flat plate areas arxi on three values of thrust coefficients (see table 2).

1.·.

Fig. 1 - Rotar :nrde] in 00W

(5)

2. 2. 2. Data acquisition and pcooessing

Data a~sition is based on sensor signals fran the rotating. system an:i fran the fixed system.

a. Rotating system

Four data a ~ i t i o n units (rotor PCM) are nounted on the rotor hub: each consists of 16 analog input channels, A/D converters, 240 Hz

filters an:i amplifiers. Amplification of each channel can be set remotely fran grour:rl.

'!he digital signals are multiplexed an:i sent to grour:rl via four slip rir:gs (one each PCM unit). 'lhree additional slip rings are used for pa,,er supply, grour:rl an:i reference signal.

'!he reference signal has a of saw tooth shape yielding the azimuth position of the reference rotor blade. '!he azimuth signal is sent fran a sensor in the fixed system to the PCM an:i then to the grour:rl. '!his sensor is coupled with the rotor shaft: all signal from the rotating system have therefore a negligible phase shift with respect

to

the rotor azimuth angle. '!his is important for online analysis in the time domain

an:i in the frequency domain.

'!he PCM signals are decoded on grour:rl, so that all sensor signals are available in analog form via a crossbar distributor.

b. Fixed system

'Ihese signal are from the the rotor balance an:i from the wirrl tunnel (terrperature, pressure an:i speed). Similar signal corrlitionin;r applies.

'!he data stream from both systems are fed to a computer via grour:rl

PCM, whereas the PCM signals are recorded on a magnetic tape: this recorder stores all signals continuously, as a 'flight recorder'.

Data processing is perfonned in two steps:

a. after assembling the test data in computer RAM, all data are converted from the time domain to frequency domain applying a FFT for one rotor revolution.

A printout provides the data in engineering units; the conplete calibration path (i.e. sensors, cables, filters) is considered so that the results can directly be used for interpretation.

b. All data in time domain are stored as raw data on tape for off. line analysis: this data are gathered for 20 rotor revolution (1 secorrl), with the possibility t o ~ this data frame if needed.

Each time signal has 2 data points, i.e. 10 bit resolution; it follC1.YS that a 8th order hanrv:)nic still consists in 6 data points.

'!he raw data are transfonned into frequency domain an:i stored on hard disk; this procedure reduces data by 98 % without significant loss of information.

'!he time to frequency domain transformation considers only the rotor hanrv:)nics up to the 8th order; therefore peaks between the rotor hanrv:)nics are suppressed. For each rotor revolution an:i for each signal, a FFT is perfonned yielding the mean values of static an:i hanrv:)nic contribution.

(6)

~

For exanple fig. 2 (a,b,c,d,) shows data analysis of the flap

nate1t

sensor(@ 18% blade radial station), in time and frequency domain for both raw data arrl reduced data.

'!he time signals differ by

Jilase,

whilst the frequency curves differ by anplitude;

this

is due

to

calibration consideri.n; the transfer function of the whole data. path, inclusive of the filter characteristics. '!he time signal of the reduced data was b.l.ilt by a harnonic synthesis fran the 1St

to

the 8th hannonic: therefore the signal looks sm:x:,ther than the raw data signal.

'lb.is ~ l e irrl.icates that the reduction of the data fran time danain to frequency danain does not lead to a significant loss of infonnation urxier nonnal rotor c:peration, as the dynamic content of the sensor signal consists of rotor harm:>nics only.

'!his may d'lan;;Je if rotor arrl/ or blade instability oc::o.rrs

am

frequency other than rotor harm:>nics are important. Ha.vever tunnel

experience shows that even

stron:.:J

flc:M separation on a rotor blade does not cause peaks between the rotor harm:>nics in the frequency spectrum, for stable rotors; this is also confirmed by flight experience at Agusta

up

to operational limits of conventional helicopters.

·---· - - ~ ~ - ~ - , : , r T W l J •' 111.fi I iiQ' I C,Vi,

).m, t lif"I I et. W.2'7 I NI I )All• I • • ( WI l •1 l:D..1.1 I Nl I

,.._.-J.Y, I ... t e\ 11,94 ( tll' \ 11 . . Y 8.U I ll't I •• 1m.r, f nl l I .... Tlt'E DOnAIN CHANNEL 51 FLAPI-TE I 111 l SPECTRAL ANALTSIS CHANN!:L 51 FLAPl-TE l 111 l

J.5 29

"

!i ₁₈

...

_:, ;; .i:

•

·'5 -11

i.e I.I I .2 1 .3 1.• 1.!I 1.6 1.7 1,8 1,9 2.11

rOt.c,r"'rPwOlut.lon

·+-

NIGH IIJSI l'EllfORIWCt $El'$ ff•Fn

I n \ - : K1)A311 29·11:f·ll!l "~1'17 111,12

TltDlilll'I ••••• l ~ 1

-

UN, t ••-"'I• I IMN'UU .. t •J.~ • . . . . , l

CO..LICtl'fl ,

..

,

...

, L,Af C:1'C ,

..

,

...

,

L-C'TC -1.a r UHl9U >

••

ltW'f. I • J

..

-.

... ltlf\) ftT PlfOI ~. ,

..

,

T lr'E OOMAI N tHANNtl 51 FLAPI-Tt { "'1 J

1'5 II !i

'

-

';"

•

~

_-,

·-_,,

1911 158 281 258 Az J o,ulh I DEGREE l

+-

NIGH 1t1S1 1'81fo,,wcf MEl'S FF·Fn

f n \ - 1 K13A311 2!-U·19 .,,,1111 111,12

...

... r;;;.,

-

.... .... ""1., ~,

...

,

....

,.-,

CALKtl . . ,

...

,

...

, Ut Cfl:

....

,

...

,

·-....

···-,

••

'1W'f.r•t . . IO.J.

...

, "'"''°' )4, C WI) 3.e Ii

2.,

~ 2.1 ~ 1.,

,

..

! .,

I

l

,8 J=~~.::=:...::::,~~e,d,~L,h'.1~""°4~

• 21 41 5-e ea 111111 1211 1•• 15-e 1&11 ,ae

Fr•CIU"'C•I I HZ l

-:.L .. ·1;;jg;j°rlSI----;..;--, !M~- . .

l

lf•l"ft

- ~ - I N l - - ' - - -MlllllW_ 1'_ll.l'·f'j. ~-~"~- . '~'12

. . , . . . " · " ' " " ' , " " tlfl. C .... ""'. I ...,nnt ., .• I . . . )

ca1ac11• ., . . & Clllllllllt J ,. .. , C't'C , • • I . . . . ) -...e Ct( - · - -I M..U a

PI 119', I • J , . IQ,.&. 11, I . . l ,_ PIJ'CM _ ? 4 ~ .

- - - - , ; . - 1 "lK r . r , r : a - r ~ ··· ··,:w-t·...,-.r 11:M" 1.JT"Ylllf .• -1.1, I IN'l ! •t. Jl,11 I NI) 1•KY 1,tl l _, l M fll,11 I ttl J I"'

• • • I N'1 ) • \ te.• C llllf ) ,,,_.., t.'J'l I .,. ) a\ 111.• t Ml I tl'NY

• ..1.Jl 11" l..ilJ\A...1.111.J. . .J'"'·---· -·'··"-J.1r1.1.•1.t211,G.l.Jll..l..t·O\t

SPECTRA\. ANALTSIS CHANNEL 51 fLAPI-TE { N'I l

4,, 4,0

-:a.,

! , .•

... 2.,

s

2.0

_,.,

JI.I

lJ

= ., ' I .o -1--"'---'---'--~--'-- -...- -~ I 29 48 5e It 1118 l:.>8 1•1 IMt 1811 211@ Fr•QtAnc:11 [ HZ l NI ... 111St . ~ Sll&P$ A'•Fn l,sl - t IIT:ll.311 21-\IEP-tS 19~:17 11Vl2

...

'"")

·----·-·

'1W'f, I • I

-1.Af CTC

...

.-, ,

...

, •la. ( W I ) ....,,t&\t -a.- Cm ... J ..-cTC .. ,.c...-> "'•110. ,.._ ... >

Fig. 2 - Data analysis fer flap JIODent sensor

(7)

2. 3 P.redictiat methods 2.3.1. RJVER

Hovering perfonnance and wake distrib..ltion are evaluated. by c::c:xrpu-ter program HOVER by Analytical Methcx:ls Inc., ref. 3; a lifting surface ccx:ie allowing both prescribed and free wake nodels. Different blade sections and platfonn.s can be studied; output consists of global

param-eters, blade load distrib..ltion and imuced velocities at off-body points.

A c:x::s:rparison has been co:rrlucted on the effect of the elastic blade scheme in the ccx:ie.

2.3.2. CDSMIC

'!his ccx:ie is the Agusta version of ccx:ie of ref~ 4, distril::uted by

CX>SMIC U.S.A., for helicopter flight dynamics and aeroelasticity predi-ction.

Based on a blade element approach, it has the feature of

calcu-lation the blade frequencies, 'whereas output consists of frequencies and nodes,

trimmed

conditions, global forces.

'Ihe upjating consisted. of porting to F77, inprovements in input and output capabilities (including different blade sections along span and a general file for graphical processing); it is coupled with a rotor stress prediction ccx:ie, and CFD ccx:ies for inproved aerodynamic analysis and interactional aerodynamics.

Validation was co:rrlucted in terms of power and trinuning prediction of the A109 helicopter.

It is planned to sul:stitute this ccx:ie with CAMRAD.

D.le to problems with nodel blade dynamics explained belav, the application of this ccx:ie is still in progress and the relevant results might be published in another paper.

2. 3. 3. Prcprietary Software padcages

'Ihree different kirrls of methodologies have been used for compa-rison purpose.

'Ihe ccx:ies are used ma.inly in the preliminary design phase of new

helicopters and are therefore aimed at the prediction of global quali-ties, at the evaluation of design data for subsystems and at the

preli-minary prediction of rotor loads and dynamic behavior.

Refer to a paper presented. in previous Fonnns (ref.5).

a. NFCNI'L ccx:ie (blade element)

'!his is the last release of the ccx:ie NFCl'LL presented. in ref. 5:

NFCl'LL was a blade element ccx:ie that can evaluate, k:noiling the control

angles or the desired forces in the shaft reference system, all the rotor quantities: power, flapping and lagging motion; for any rotor attitude in space.

'Ihe program is particularly dedicated. to the prediction of the torsional loads at the blade root, to provide an inportant indication for a correct dimensioning of the flight control system already in earlier design stage.

(8)

Aerodynamic characteristics of the airfoils distril:uted on the blade are provided in tal:ular fonn as coefficients vs Mach rnnnber an:i an;Jle of attack (up to 5 different airfoil along the blade).

'llle ccxrplex Man;Jler an:i Squire model for irrluced velocity is used arrl a prcx::edure fran Ericsson theory accounts for the unsteady effects.

'Ibis ccxie was extensively tested with Agusta flight tests data, with positive results, despite its relative simplicity (for exanple, rigid blade scheme).

b. NF1'RlM ccxie (simplified trim procedure)

'llle NF1'RlM ccxie is based on a simplified trim procedure able to evaluate, at given aircraft speed an:i a'bnospheric.'-flight conditions, the forces generated by the rotor, the control an;Jles, the power required

an:i the fuselage pitch attitude. ·

All equations calculating rotor forces and the differential equat-ions representing flapping m::>tion are solved in closed fonn. A rigid blade with constant chord is considered, and a constant lift curve slope is assumed; stall, ccxrpressibility an:i reverse flow effects are ignored in force calculation, whereas account is made in power estimation. An

original mathematical model developed at Agusta is incorporated for the evaluation of an average rotor Cd at every operating conditions. Irrluced velocity is considered constant: on the rotor disc, the average value being obtained by a ccxrplete fonnulation.

Fuselage aerodynamic loads are obtained from wi.rxi tunnel tests on tail-off configurations; separate models are used for horizontal arrl vertical tail surfaces.

'llle influence of main rotor wake on horizontal tailplane is consi-dered: the stall is accounted for an:i the program calculates the condi-tions of wake impingrnent in terms of thrust, speed, climb an;Jle, pitch attitude arrl flapping an;Jle.

'llle ccxie is extensively used in the Preliminary Design Phase an:i has proven its reliability up to stall limits for conventional rotors.

c. POI.ARIII ccxie (Energy rrethod)

'Ibis is a classical, simple an:i flexible energy rrethod used for a first quick estimation of power required and perfonnance of new helic-opters.

'llle rrethod is based on narentmn theory for the estimation of irrluced power, an:i makes use of the classic breakdown of power in: irrluced; parasite; profile; tail rotor power.

Simple f onnulas are used for the estimation of rotor thrust an:i rrean value of irrluced velocity, an:i a K, factor accounts for the non unifonn irrluced velocity distrirution. 1

'llle profile power is calculated using the Bennet theory, while a m::>re ccxrplex fonnulation is used for the evaluation of the rotor

ed.

Cc:.lrrpressibility an:i stall effects are considered; the variation of fuselage

%

with an;Jle of attack is provided by a quadratic parabola law.

Tip losses effects are included.

'Ibis rrethod, due to the easy arrl quick use, is largely applied at Agusta, providing reliable data up to rotor limits, on

the

basis of sets of coefficients obtained by the available flight test data base.

(9)

'I\mnel test data is corrected for the influence of the hub fairing: this is

estimated

at 0.07 m2 _{in wind axis (confirmed}_by_{previous tunnel}

measurement with blade-off configuration), and subtracted fram ha.lance measurements.

Refer to table 2 in ai;pe.rrlix for all conditions fram both flight

and tunnel used for the ccmparison.

3 .1 Presentation of results

camparison is made between tunnel tests, computer codes results and flight tests data, for the following parameters:

3.2.1

- Power vs thrust (hover) - Power vs speed

- Code prediction vs tunnel measurement [at one test point, without interpolation] - Flight data vs tunnel as nonnalized

S,

vs µ. - Flight control angles

- Collective angle vs µ - longitudinal angle vs µ - lateral angle vs µ - Pitch link load vs azimuth

Hanronic analysis of first 8 hannonic of pitch link loads from

flight, at the 3 selected test conditions, compared with

correspon::ling tunnel tests.

All calculations are cx:;E.

a) pc&er comparison

'!he HOVER code was applied to tunnel test # SI'Hl58W; 5 iterations of prescribed wake followed by 5 with free wake, produce an average~ value at 0.00516 (T=300 Kg) for a collective angle of 7.9

°

and a

powet

C

=

0.000395 (P=49.6 kW); corresponding tunnel values at about the same ~.005186, give a collective 9.84 °, whereas measured

c

₀ i:5 0.000385.

'!he blade is elastic; however no exact scheme for lnadel blade is made (like blade tip joint mass).

'!he higher collective values in the tests could be explained by the test conditions: the p:rrking hall may detennine some recirculation effect, with some inflow at the disc, thus requiring an higher collective: and naturally by the elasticity of the blade and the control linkage kinematics.

All other three codes can be applied providing outputs.in tenns of

power; the two blade element codes also provide coning angle and colle-ctive angle: see fig. 3 (a, band c) below.

(10)

... 0 ...--. er> Q) 0 ..__.,. 0:: ~ 1/) r-... I u I-a:: w > i= u w _J _J 0 u ,.--.... "O (I)

-~

0 E _L. 0 C ,.__,, a::: w ~ 0 a_ 17 1S 15 14 1J 12 11 10 9 8 7 8 5 4 J 2 0 1.100 1.000 0.900 0.800 0.700 O.IIOO 0.500 0.400 0.300 0.200 0.100 0.000

W.T. DATA COMPARED WITH SEVERAL CALCULATION METHODS HOVER CONDITION

COLLECTNE PITCH at 75% of R

9 - Ill WH) TUfffl. M'PI

0 100 200 JOO 400 SOO 600

THRUST (Kg)

W.T. DATA COMPARED WITH SEVERAL CALCULATION METHODS HOVER CONDITION

POWER REQUIRED

e - 9 WH) TUfffl. Mll\

0 · · <> Illa IUMOO ccac (1'ClM ~

t,, t,, All'U'D - """""""'(Pow _ ,

.,. • .,. EJIDOIT/ ICIHQO C-l'OIMI)

0 100 200 400 SOO 600 THRUST (Kg) ,.--.... "O Q)

.~

0 E L. 0 C ... a. u 10 9 8 ,... CJ'> (I) 7 0 ' - / w 6 ...J <.:> ₅ z <{ <.!) 4 z

z

3 0 u 2

W.T. DATA COMPARED WI.TH SEVERAL CALCULATION METHODS HOVER CONDITION

CONING ANGLE

9 - e 'tlKD 1l.lffl. WA

<> . . <> IM.U ~ cao( <-tmlll)

t,, - · t,, IN'U.-.:D lllH l'l!OC(l)Ul'E (l'<W IFTIIII)

0 ... ~-.... - -..._~,.____._,__..._~ ... _,__..._~ ... _,__..._~ ... _,__..._~ ... _,__..._~ 0 1.100 1.000 0.900 0.800 0.700 0600 0 500 0.400 0.300 0.200 100 200 300 400 500 600 THRUST (Kg)

W.T. DATA COMPARED WITH A 129 HOVER FLIGHT TEST DATA

(MAIN ROTOR ONLY)

0 - 0 -'1H _ . IIOltlll HIMJI fUllll Tt'IT !WA y

,.,·

.,.

0.300 0.400 . 0.500 0.600 0.700 0.800 0.1100 1.000 1.100

Ct (normalized)

(11)

PcMer (in fig. 3c and 3d given as normalized with respect to maxinu.nn value measured in wind tunnel) agrees well with blade element code, including high thrust corxtitions; the other two simpler methods are very good for nonnal blade loadings, and still acceptable (within 10%) at high blade loadin;]s.

'!he collective shows a difference of 2. 5 °; either blade flexi-bility (twistin:J due to inertial characteristics and aerodynamic

loadi-ng) or control linkage deformation can be the cause: this difference increases with thrust, and seems to confirm the influence of aerodynamic effects.

'!he fla., corrlitions in the hall (fla., recirculation) could also contrirute to the explaination of the difference.

'!he caning angle on the m:xiel, read as average flappin:J, is not zero at zero thrust; this can be due to sensor calibration and blade trackin:J (whose procedure in tunnel is also based on an optical methcx:l, applied ha.,ever on the reference blade as datt.nn).

'!he variation with thrust is good: the two codes applied provide comparable slope predictions.

Fig. 3d shows the comparison between wind tunnel data and A129 flight test data; the main rotor pa.,er in hover flight is obtained usin:J the formula in par.2.1.1. '!he correlation appears very good at normal

Ciri

a small discrepancy exists at very high disc loading.

'!he ~ of full-scale A129 was corrected to take into account the fuselage d~oad, this value deriving from Agusta experience validated by aerodynamic calculations.

b) blade loads

·Blade loading distrirution from tunnel and flight will be compared with HOVER code prediction, using the elastic blade model, after a check

on inst.rl..noontation calibration and the validation of the m:xiel blade dynamics used in the code; the ca:nparison looks already acceptable with flight tests data. Final data may be sha.,n in another paper.

3.2.2 Forward Flight

. Fig. 4 to 7, sha., the results from 4 selected tunnel test points campared with predictions, in terms of power, control angles and pitch link loads, for all 3 codes; as a result of the elaboration with NFCI'LL

(12)

!50 45 40 35 ~ l JO ::,( ..._,, 0:: 25 w 3:: 0 lO Q. 15 10 s 0 \0 f-.J O'I -J f-.J N 110 IIO 70 z .._,, !50 w 0 ::, JO I:: _J a. ~ 10 -10 -JO -50

W.T. DATA COMPARED WITH SEVERAL CALCULATION METHODS

W.T.T. N. ST3A17 ct-0.0055 Cx=-0.00001 V=21.9 m/s

POWER COMPARISON

• TEST RESULT nJ NFCNTl CODE

ID NfTffllol COO(

i;z;;J POLAAIII COO£

W.T ~TA

'

\ \ \ \ '\ '\ BlADt: UDIENT CODE: 17 ', / / / / /

i

V '.j , , , /

~

SMPUfl[D TRIM

PROC[OVRE ENERGY MCTHOO

W.T. DATA COMPARED WITH BLADE ELEMENT CODE W.T.T. N. ST3A17 Ct=0.0055 Cx=-0.00001 V=21.9 m/s

PITCH LINK LOADS COMPARISON - ROTOR HARMONIC's AMPLITUDE

CONSWfl TERM

- WINO TUNNEL TEST

QJ NFCNTL COO[ g 8 7 5

~

5 0:: &:l 0 w (/) < I D.. J 2 0 160 120 80 40 0 -40 -eo -120 -160

W.T.T. N. ST3A17 Ct=0.0055 Cx=-0.00001 V=21.9 m/s FLIGHT CONTROL ANGLES COMPARISON

- COLJ.ECTM: PITCH 7511 or R OJ LATERAL CYCUC PITCH

SZs:I -1 • LONGllVOIIW. CYCLIC PITCH

W.T. CV.TA BI.ADE ELEWENT CODE

(IRHTL CODC)

SIMPUFlEO TRIM P!lOCtOURE

(IEl!IM COO()

W.T. DATA COMPARED WITH BLADE ELEMENT CODE W.T.T. N. ST3A17 Ct=0.0055 Cx=-0.00001 V=21.9 m/s

PITCH LINK LOADS COMPARISON - ROTOR HARMONIC's PHASE

1 • REV 2 • REV.

- WINO lUr-,.a(L TESl

i3J NfCNfi COOE

3 • REV. 4 • REV.

(13)

....

w Cl:'. w 3-'. 0 0.. z w 0 ::, t:: _J 0... ~

W.T.T. N. ST3A11 Ct-0.00593 Cx"'0.000239 V=44.1 m/s POWER COMPARISON

~,---,

JO 25 lO 15 10 - lEST RESULT cs:J NFCNTl COOE en NFTRM COOE

li2SI POlAIIII OOO(:

r,: \ \ \ \ '\ \ \ ' ' v V V V ~ V ~ I I V I I ~ _\ ~ _V \ I ) 0 ~ - - ~ - - - - ~ ~ " ~ - - - - ~ / ~ - - - ~ - - ~ 90 70 ~ JO 10 -10

W.T. Cl'.lA BWJE ELEMENl CODE

SIMPLIFIED TR1M

PROCEOUR( El-l:RGY MflHOO

W.T. DATA COMPARED WITH BLADE ELEMENT CODE W.T.T. N. ST3A11 Ct=0.00593 Cx=0.000239 V=44.1 m/s

- WINO lUHHEL TEST

'31 HFCHTL COO( ~ c,, Q) 0 ,.__.. w VJ < I 0.. 9 !I ll 3 2 0 200 160 120 80 40 0 -40 -80 -120

W.T. DATA COMPARED WITH SEVERAL CALCULATION METHODS W.T.T. N. ST3A11 Ct=0.00593 Cx=0.000239 V=44.1 m/s

FLIGHT CONTROL ANGLES COMPARISON

- COLLECTM: PITCH 7511 or R cs:J lAlERAI. cm.IC PITCH

~ -1 • LONCffiJDltW. CYCLIC PITCH

W.T. CV<lA BWJE ELEMENT CODE

~LCOOl)

SIMPLIFIED TR1M PRQC(l)URE

(""""' COO()

W.T. DATA COMPARED WITH BLADE ELEMENT CODE W.T.T. N. ST3A11 Cl=0.00593 Cx=0.000239 V=44.1 m/s·

1 • REV. 2•REV

- WlHD 1Ut-l'l[L lEST

lwJ NFCHTL CODE

(14)

\0 I-' 100 IIO 80 70 'i' 60 ::.'.'. ' - ' a::: 50 w ~ 0 40 ll. JO 20 10 0 90 70 -_z. 00 ... w 0 JO ::> t:: _J a.. 10

l

-10 -JO

W.T.T. N. ST3A21 Ct-0.0055 Cx=0.000395 Y=66.1 m/s

POWER COMPARISON

m TEST RESULT

l::3J NF"CNTl CODE IL! NITTM COOE

12Sa POLARII CCOE

W.T. Dl.TA BlADE ELEMENT

CODE 7 I / / I / I / / / SINPUFIED TR1M

PROCEDURE H£RCY METl-lOD

W.T. DATA COMPARED WITH BLADE ELEMENT CODE

W.T.T. N. ST3A21 Ct=0.0055 Cx=0.000395 V=66.1 m/s

- WINO TUHtlEI. TEST

l::3J Nf'CNTL COO[ CONSTANT TERM !{I w 0:: c., w 0

-

c,, (I) 0 ... w Vl < I Cl. 1~

14-,~

12 11 10 9 II 7 e 5 4 3 2 0 200 160 120 80 40 0 -40 -110 -120 -160 -200

W.T.T. N. ST3A21 ct-0.0055 Cx=0.000395 V=66.1 m/s

FLIGHT CONTROL ANGLES COMPARISON m COLLECTIVE PITCH 7511 or R

l::3J LATERAL CYCUC PITCH

IZiJ -1 • l0NCIT1JDIPW. CYCLIC PITCH

W.T. D'ITA BlADE ELEMENT CODE

(IRNll. COO[)

S111Pl1F1ED TR1M PROCEDURE

(IETRM COO(J

W.T. DATA COMPARED WITH BLADE ELEMENT CODE

W.T.T. N. ST3A21 Ct=0.0055 Cx=0.000395 V=66.1 m/s

;I

1 • REV.

- WINO TUl>NEl TEST

l::3J NFCHTL COOE

(15)

I-' U1 0:: w 3: 0 n. w 0 :::> t: _J n. ~ Q() eo 70 60 40 JO 20 10

W.T. T. N. ST3A 14A Ct-0.0059 Cx=0.000506 V= 75.1 M/S POWER COMPARISON

m 1EST RESULT C'lJ NFCNTl COOE !L! NFTRIM COOE

l2SI POIMII COO[

ro; \ \ \ \ \ \ \ \ 7 / I I .) I I / le > 0 .__ _ _ ... _ _ _ _ _ ...,_ _ _ _ _ __...c,__ _ _ _ _ -'-"'. _ _ __. 110 IX) 70 JO 10 -10 W.T. [)t.l" 0lJ'DE ElEMENT

CODE SIMP\.IFIEO TRIM PROC£DUR[ Ei'£RGY M[IH()O

W.T. DATA COMPARED WITH BLADE ELEMENT CODE W.T.T. N. ST3A14ACt=0.0059 Cx=0.000506V=75.1 m/s PITCH LINK LOADS COMPARISON - ROTOR HARMONIC's AMPLITUDE

11111 WlNO TUNNEL TEST

i;:;sJ NrCNTL CODE 4 • REV. 1:i 14 1:1 12 11 10 9 I{! 8 w 0:: 7

8

6 0 5

•

3 2 0 200 160 120 80 01 40 Q) 0 ....__,, ₀ w VJ -40 < I Q. -eo -120

W.T.T. N. ST3A 14A Ct•0.0059 Cx=0.000506 V= 75.1 rn/s FLIGHT CONTROL ANGLES COMPARISON

m COUEC'IIVE PITCH 7511 OF R

C'lJ LAlERAL CYCUC PITCH

liZ!J -1 • LONCITUOINAI. CYCLIC PITCH

w.T. a...1" 0lJ'DE ELEMENT CODE

(NrCNll~ .

SIIIPUFIED TRIM PROC(OURE

(Hr11114 COO(J

W.T. DATA COMPARED WITH BLADE ELEMENT CODE W.T.T. N. ST3A14A Ct=0.0059 Cx=0.000506 V=75.1 m/s PITCH LINK LOADS COMPARISON - ROTOR HARMONIC's PHASE

m WIND 1U""1CL l(Sl

(16)

0.9 0.8 : 0.7 .,.., ; 0.6 0.5 0.4

th--.

0.3

_.

0.2 . 0.1 a) power CX'J'J"p.lrison

A

<=n

vs

µ, far both turmel

am

flight, reduced to the

same

Crt, ard

c

values, ~th oormalized ordinates (as

<=n /

max

c;,

in flight) is c)iven

IB

fig. 8: the best fit of all available 'flight test data en the caxplete heliccpter is reduced to isolated_ rotor pc:,.,er arrl directly caxpared with

tunnel measurement.

'1be plots are

provided

far 0.15 µ,

s

0.35.

Tunnel data lll.lSt be corrected far lateral trim ( side farce

influ-ence) arrl speed accuracy (as measured

in

flight) at l<:7w' speeds (µ, <

0.1), as oc,.Jot₁at arrl ocptoVx; at high speeds (µ > o.3) for al'Y:Jle of attadc, as

fcptoa.

An attarpt was made to ai:ply a derivative procedure to tunnel data

in order to take into accamt these factors: the results axai.ned shcM

that a better correlaticn can be adtleved. Hov,1ever the sensitivity of

the methcx:i to sane of the relevant inpJt

paranvanters

am

the difficulty

to accurately measure the lateral forces in flight do

not

yet all<:7w' to ai:ply the procedure

with

a reasonable level of confidence.

W.T. Data compared with A129 forward flight lest data

I

I I

_I

I

_..

,

.

II _W.T._{lxlto (Ct=0.0059)} _{- I}

.

/

.

!

.

W.T. Interpolated doto (Cl=0.0059) ! j

· ' /

- - - flight lxlto (Ct=0.0059)

..

· .

:

I

.

. · · ~

'

~

_I' ~" _I

-·

_.,.,

~

_... ....

---

...

---

___

... ~-·-- _! I _! I ' ' ; j

I

j .

.

. I

!

i i

j

I 0.15 0.2 0.25 0.3 0.35 Advance Ratio

Fig.

a -

<=p

vs µ. fer flicpt

am

tmntl

(17)

b) Control angles

'llle energ'j method is not awlicable. Collective predictions agree

well, with inproved oorrelatial at higher speed:. this can be explained· by the wake modeling in the c::xx:ie beirq -nore

awr

op1. iate far high speed

flight, than far low speeds~ the wake is more carplex. I..crgitudinal angles are all very well predicted.

. Far lateral control angles CX1Ip'lrisal, the tunnel cxn:litial is set

at zero flap (lR) at referenc:e blade: the flap sensor is therefore used, instead of static mast :nanent readin; in the rotating system ar balance

nanent. ·

'llle rodes are usai :inposing the balance readalt: lateral angles

b?cane umerestimate1, an:l the final results are thus affected by the

wake modelling, very sensitive al rotor lateral trimning.

It looks

as

if

the sinpler codes (based al sinple wake model, like

1st harnauc of Mangler-Squire distrib.Itial), o:xipare better with tunnel

data.

c)

Pitdl link loads

Only the blade element cede can l:e awlied, even if the blade nniel

remains sinplified, not interrled for detailed blade load predictia,s.

'!he load waveform prediction is beyorxi the cede scope; nevertheless

the results obtained are incouraging in terns of peak-to-peak values (fig. 9), whidl are the. inportant paraneter in preliminary design of flight controls. ,-... z

--

0 <( 0 _J

COMPARISON BETWEEN MEASURED AND PREDICTED PITCH LINK LOADS FOR DIFFERENT WIND TUNNEL TESTS AT DIFFERENT Ct, Cx and SPEED 200 180 100 ao 120 100 80 60 40 20 0 -20 -40 -60 -80 -100 11111 W.T. SfATIC W.ut a:, W .T. I /'l PW' WOi..Ul'. t::l OOU:ULAT[l) SfllTIC VALUE

C::C CALCULATED 1 /2 PIDP VJ.L.1.£

(18)

'!be sane cxxie awlied to flight test data provide nudl

t:etter

carrelaticn (ref.5); this also ~ that the rotor systen dynamics (t.arsiaial behaviar; elastaneric characteristics; a:uttul systelll

stiff-ness)

is sanewhat different fran

full-scale blade:

this

would require a

m::>re

detailed krx::lwleaJe

of the

m:xiel

rotor

system,

'Whidl

was beyaxl the scx:pe of the testin;J an:i of this paper. ·

'!he followin;1 figures 10 and 11, show a carparisal of flight

test

data an:i tunnel test data, in terms of tine histories;

a oormalizaticn

factor is

cq:plied

(the highest

load

is p.It equal to

1,

the mean value beirg rem::M:d), an:i the flight force is scaled to tunnel

cy-

scale factor squared.

Frequency danain plots are shown in fig.12: the static value differs both in sign and anplitude. 'lhis could be explained by a

different

blade

dlord ex; positicnin; alOI'X1 the span, between JIWJdel and

full-scale blade.

'lhis

seems

to oc:nfirm how

challen;Jin;J

is the task to realize a caipletely similar dynamic IOOdel.

It can be noted that the harnaric content of the tunnel signal at the higher frequencies is m::>re jnportant then that in flight: this confi.nn.s the hypothesis previuosly made on the effects of the different blade dynamics. I

-a:, C: s.. 0 z

Pitch link load

1.5 - - - , - - - . . - - - , : 1 ' I ; 05~ I I

I

ol

I ,/\

~

--- Flight test ( 05) -. f'.. ~ , ; \ ., I ',

·:h /_

wr

(ST3Al1) \ / --.._J

-1---~---~---'---~---'

0 50 100 150 200 250 300 350 400 Azimuth (deg)

Fig. 10 - Flicjrt: and TUmel pitdl link loads j.F0.2

(19)

Pitch link· load

I:~

0.5 I

-

~ --- Flight test ~ 0 -1

\

/

\J

- WT (ST3Al3) -1.5 Azimuth (deg)

Fig. 11 - F1igbt and Tunnel pitdl

link

loads µ=:0.25-;-0.3

4. DL.qysc:;JCH

'llle stu:iy corrlucted evidenced sare problems in ccrrparin:J the different data: this is a basic list of difficulties met durin:J this

work:

- reliability of measured data (ex. strain gages on the tunnel m:x:lel

blade)

- explanation of the differences between data, whether due to accuracy

of measurement, sinplifyin:J assurrption.s in the . prediction cx:rles, real

differences between m:xiels arrl full-scale.

- interpolation of the global parameters for direct cc:rcparison, or their extrapolation

- thoro.Jgh knowledge of all gecrcetry, mass distrirution arrl dynamics of the m:xiel testec'i in tunnel, as an essential basis for cc:rrparison

On the other end, the lessons le.anled from this experience which

seem of general value are:

- the effect of lead-lag danper reIOved on m::xiel is only affectin:] the stability of the rotor system arrl the blade loadi.n:J at the rcx:,t, rut

seems to have a negligible effect oo rotor performance arrl certain

loads.

- the test c:x:>rrectian.s on tunnel data can be applie::i to .inprove the

correlation, for exarrple at low speed con:li.tions (where flight test data

need c:x:>rrection for speed accuracy, arrl pmnel data for · miss in:] the

lateral trinmirg) arrl also at high spee::is, for attitude difference

(20)

°'

-.J IV ·o r-.. 1J II N

~

_... 0 C ' - ' w 0 :::> t-:J a. ~ < 1.100 1.000 0.1100 0.800 0.700 .0.800 0.500 0.400 0.300 0.200 0.100 0.000 -0.100 -0.200 -0.300 -0.400 -o.:ioo

W.I. UAIA CUMl-'ARl:.U WIIH AIZ~ 1-UGHI It.SI UAIA

W.T.T. N. ST3A1 OA ADVANCE RATIO = 0.15

W.I. UAIA CUMl-'Akl:.U WIIH AIL~ 1-LIGHI IL::il UAIA

W.T.T. N. ST3A11 AfJVANCE RATIO = 0.2

PITCH LINK LOADS COMPARISON - ROTOR HARMONIC's AMPLITUDE PITCH LINK LOADS COMPARISON - ROTOR HARMONIC's AMPLITUDE

CONSTANT TEl'IM 1 • REV. 2 • REV.

1.100 ~ - - - ,

- WtO rulNEl TEST 1.000 m WIND TUNNEL TEST

,-., 1J <P N 0 E

...

0 C ' - ' w 0 :::> t-::i a. ~ <( 0 !XlO ,..-... 0.800 "O <P ~ 0 E .... 0 C '-" w 0700 0.600 0.!,00 O.JOO 0 0 :IOO :::J t-- 0.100 ::J 0.. 0000 ~ -0.100 -0.200 -0.JOO -0.400

CSJ A 129 FUGHT TEST DATA

-0 500 - ' - - - '

3 • REV. 4 • REV. CONSTANl TERM

W.T. DATA COMPARED WITH A129 FLIGHT TEST DATA W.T.T. N. ST.3A13 ADVANCE RATIO= 0.3

I• REV.

1.100

1.000 _- _{WlND 1V~tl TEST}

0.900 _rn

A 179 FLIGHT TEST MTA

0.1!00 0.700 0.800 0.500 0.400 0.300 0.200 0.100 0.000 -Q.100 -0.200 -0.300 -0.400 -0.600

CONSTANT TERM 1 • REV. 2 • REV. J • REV. , • REV.

2 • REV. 3 • REV. 4 • REV.

(21)

5.

Tunnel model testi.r'g provides reliable data on rotor glota.l loads

and power, as derronstrated by comparison with prediction methods validated with flight test data.

'Ihe correlation on control loads is fairly good at low speeds and

still acceptable in terms of peak-to-peak values at high speeds: improverrents could be obtained by a more accurate blade dynamic scheme,

base:l ooth on a direct dynamic characterization of the model blade and

on the inclusion of a more sc::pristicated modeli.r'g in the ccx:ies.

still to be validated are the tunnel results on blade vibratory loads, due to the need. of a better sinn.llation of model dynamic characteristics, which may be the subject for future work.

Based on this experience, it can be stated that the application of

tunnel model testi.r'g in rotor design requires a thorough kncMledge of the model characteristics and a careful design of test con:litions.

Also, as the direct comparison with flight tests seems to require in any case the use of prediction methods for the evaluation of missi.r'g data and for interpolation/extrapolation, it seems preferable to con:luct the comparison of experimental data versus prediction data only, due to the difficulty in the direct comparison between tunnel data and flight test data.

continuous effort will be spent in the future for improvi.r'g this correlation, both on the existi.r'g data and with further experimental activities.

Adcnowle:lce1e1t

'Ihe authors wish to thank the I.AH partners and especially NIR for the kin::i pennission of usi.r'g the tunnel data for this paper.

1. G. PAGNANO; An Integrated Approach to Rotorcraft Aerodynamic Design

and Develop:nent; 15th :Enropean Rotarc:raft Fonmi, AMSI'ERDAM NL,

Sept. 89, paper 10

2. J.W.G. Van NUNIN, C.HERMANS; H.J. I.ANGER; I.AH Main Rotor Model Test

at the DNW; 16th :Enropean Rotarc:raft Farum, GIASGCM U.K., Sept.90,

paper II.8.2.1

3. Analytical Methods, Inc; Evaluation of blade tip planform effects on hover performance; A.M.I. Rep.n.t 7908, Nov. 79

4. W. JOHNSON; Aeroelastic analysis for rotorcraft in flight or in a wW tunnel; NASA. '1N D-8515, July 1977

5. F.NANNONI, A.STABELLINI; Agusta methcxiology for pitch link loads prediction in preliminary design phase; 14th :Enropean Rotarc:raft

(22)

Appendix: Tables

Table 1 - List of Symbols

Advance ratioµ.;

shaft tiltim an:;Jle o:shaft;

Vertical force F ; .

z Propulsive force FX' Coefficients: Cf= f ( [~ R2

y

_V2tip]

vt. _ip

=

220 m/s

Table 2 - Selected conditions Hover:

tunnel flight

STH158W 496 (04)

STH150W -> STH164W Forward Flight:

tunnel (speed m/s) µ

_<;,

flight (speed Kts)

ST3A17 21.9 .10 .0054 498 (04) 56 ST3All 44.1 .20 .0059 498 (05) 82 ST3A21 66.1 .30 .0055 498 (06) 112 ST3Al4A 75.1 ·.341 .0059 ST3A10A 32.77 .149 .0059 ST3Al3 66.19 .301 .0059

Table 3 - Blade sensor po~itions (r/R in%)

F.S. A129 Pl P3 Model Blue Yellow (36-45) (46-63) 1.7 no 9.4 10 18 23.3 27 35 41.6 53.9 50 62 72.1 74 85 81 91 - 67.22 µ

_<;,

.131 .0061 .192 .0061 .262 .0061