-
1
Rotor-Fuselage Interaction:Analysis and Validation with
Experiment
John BerryAeroflightdynamics Directorate, US Army ATCOM
Langley Research Center, Virginia
Nicolas BettschartONERA – Office National D’Étude et de
Recherches Aérospatiales
Châtillon, France
ABSTRACT1
The problem of rotor-fuselageaerodynamic interaction has to
beconsidered in industry applications fromvarious aspects. First,
in order to increasehelicopter speed and reduce operational
costs,rotorcraft tend to be more and more compact,with a main rotor
closer to the fuselagesurface. This creates
significantperturbations both on the main rotor and onthe fuselage,
including steady and unsteadyeffects due to blade and wake passage
andperturbed inflow at the rotor disk. Furthermore, the main rotor
wake affects thetail boom, empennage and anti-torquesystem. This
has important consequencesfor helicopter control and vibrations at
lowspeeds and also on tail rotor acoustics (mainrotor wake-tail
rotor interactions). Thispaper describes the cooperative work on
thisproblem from both the theoretical andexperimental aspects.
Using experimental3D velocity field and fuselage surfacepressure
measurements, three codes thatmodel the interactions of a
helicopter rotorwith a fuselage are compared. Thesecomparisons
demonstrate some of thestrengths and weaknesses of current
modelsfor the combined rotor-fuselage analysis.
INTRODUCTIONThe aerodynamic environment of
rotorcraft configurations is complex due tothe nature of the
airflow of both the rotatingblade and wake systems as well
asrelatively bluff fuselage shapes Presented at the American
Helicopter Society 53rdAnnual Forum, Virginia Beach, VA, April
29-May 1,1997.
characteristic of rotorcraft. Until recently,analysis of the
coupled effects of the rotorand the fuselage have been
addressedthrough linear superposition and empiricalcorrections.
Recently methods have beendeveloped to model the
non-linearaerodynamic interaction of rotor and wakewith the
fuselage.
Two principal effects of the rotor-fuselage interaction are
important in thedesign (or redesign) of rotorcraft. The firsteffect
is that of the fuselage on the rotor. Thechange in onset flow to
the rotor due to thefuselage is important in loading and
wakestrength. These differences, in turn, changethe system
vibratory excitation and response,operating conditions and trim,
and also theradiation of acoustic pressures from therotor.
The second principal effect of theaerodynamic interaction of the
rotor and thefuselage is that of the rotor on the fuselage.The
effect of the rotor on the fuselage is toproduce both a change in
the steadyaerodynamic load as well as an unsteadyloading on the
surfaces of the fuselage. It i swell known (reference 1) that the
wake of themain rotor vastly changes the loading of thetail rotor
in certain flight conditions. Closerotor-canopy spacing required in
militarydesigns for transportability can produce veryhigh unsteady
pressures on canopy surfacesthat lead to poor fatigue life of
thesecomponents. Fuselage trim also changes dueto wake effects on
the empennage.
At the Aeroflightdynamics Directorate(AFDD, US Army ATCOM) two
theoreticalmethods were used to simulate
rotor-fuselageinteractions. The Rotor-Wake-Fuselage(RWF) code is an
in-house developed code
-
2
for exploring methods of computing thecombined rotor-fuselage
problem. R W Fcouples a time-stepping vortex lattice methodfor the
rotor blade and wake system with asource panel fuselage model. The
other UScode is a recent version of the ContinuumDynamics, Inc.
Computation of RotorAirloads in Forward flighT/AeroacousticAnalysis
(RotorCRAFT) code. This codecouples a doublet-panel representation
of thefuselage with the Constant Vorticity Contour(CVC) wake model.
At ONERA, the PEIRF(Programme d’Etude d’InteractionRotor/Fuselage)
code was also developed tosimulate this problem. This
quasi-steadyapproach couples a fuselage code to a rotorcode which
assumes a periodic in timesolution. The fuselage is simulated by a
low-order panel method (source and doubletdistribution). A module
was specificallydeveloped to compute the unsteady pressurecomponent
on the fuselage surface, which ismainly due to the blades and wake.
Therotor is modeled by a lifting-line with a fullfree-wake
analysis, using 2D airfoil tables tocompute the compressible loads
occurring onthe blades. Coupling between these twosimulations is
performed until a periodicsolution is achieved.
Experimental data from poweredhelicopter models were shared
during thiscooperation for comparing the analyticalmethods.
Comparison with field velocity datais a means of assessing the
accuracy of thefree-wake methods embedded in these codes.A
realistic 1/7.7 scaled Dauphin poweredmodel was tested in the ONERA
S2Ch windtunnel (figure 1). From this model steadyand unsteady
pressure data on the fuselagewere acquired. Also, 3D laser
velocimeter(LV) data were acquired in two verticalplanes. The first
plane was taken at adownstream location through the hub and
thesecond plane was taken at 0.42 radiusdownstream from the hub.
Theseexperimental data provide calibration pointsfor the analytical
models. Comparisons ofcode results to determine relative effects
ofrotor-on-fuselage interaction are shown i nthe Results section.
Comparison of codeswith experimental data allows theassessment of
the accuracy of the codes i nmodeling the physics of the
interactionaleffects. The comparison with experimentaldata also
helps in establishing the relativeimportance of un-modeled effects
such asregions of flow separation from the fuselage.A code-to-code
comparison has also been
made of the predicted geometry of the tipvortex. Comparisons of
the surfacepressures, both steady and unsteady, show thelocal
impact of the effect of the rotor wake onthe fuselage.
NOTATIONThe coordinates used for this study are
defined relative the wind tunnel axis systemfor the velocity
field measurements and witha body-local coordinate system to
identifypressure measurements on the fuselage.
a Angle of attack, deg.
aS Shaft angle of attack, deg.
b Blade flapping angle, relative tothe hub plane, deg.
j Velocity potential, m2/s
m Advance ratio, U
RW, 0.20
nominal
q Blade pitch angle at 3/4 radius,degrees
r Density, 1.225 nominal, kg/m3
s Rotor area solidity, bc
Rp, 0.0849
W Rotational speed, 133 radians/snominal
y Rotor azimuth, relative todownstream aligned with thefuselage,
degrees
b Number of blades, 4
c Blade chord, 0.05 m
CP Pressure coefficient, P P
q
- ¥
¥
P Pressure, Pa
P¥ Static pressure, Pa
q¥ Dynamic pressure, 1
22rU , 245
Pa nominal, Pa
R Radius of the blade, 0.75 m
u, v, w Components of local velocity,downstream , right, and
uppositive, m/s
-
3
U Onset velocity, 20 m/s nominal
V Local velocity magnitude, m/s
x, y, z Location components,downstream , right, and uppositive,
m
PSID Pressure rating of transducerin lbf/in2
ANALYSISThe three analyses used in this study
are based on Green’s theorem that allows thetransformation of 3D
field distributions to 2Dsingularity distributions on the
boundariesof the field. These singularities arenormally seen as
source, doublet, or vortexsingularities. This is the
underlyingprinciple in panel and lattice methods. TwoUS codes, RWF
and RotorCRAFT/AA, willbe described. A French code, PEIRF, is
alsoused in this study and is described here.
RWF CodeThe RWF code was developed during
graduate studies at the Georgia Institute ofTechnology supported
by the US Army(reference 2 and 3). This singularity methodcoupled a
source panel method (reference 4and 5) and a vortex lattice method
wake(reference 6). The vortex lattice thatrepresents the rotor
blade and wake wasimproved from the previous methods toinclude the
effects of cyclic pitch, multipleblades, and a far-wake downwash
model.The coupling of the paneled fuselage with thevortex lattice
wake was done by using animpulsively started wake with no
presumedgeometry. The wake was developed with thefull interaction
of the fuselage source panelsas the solution was incremented in
time.Although not a production code, RWF hasbeen used as a test bed
to study improvementsfor interactional singularity methods.
For this study, azimuthal step sizes ofboth 4 and 8 degrees were
computed.Although no significant change in thecharacter of the
predicted velocities was seen,the 4 degree solution is used for
comparison.A distribution of 3 chordwise panels and 12spanwise
panels was used for the rotor tocompute the bound circulation
strengths. Therotor tip-path plane was defined based onmeasured
flapping angles and shaft angle.Cyclic pitch was set from the
measuredcontrol angles (shaft axis) and transformed
to the tip-path plane. The code set theminimum core effective
size to be 30 percentof the panel diagonal dimension for both
thebound and free lattice cells. This numberwas used based on
unpublished studies on theeffects of this radius on velocities
computedon planes adjacent to the lattice. The codewas run for
sufficient iterations to allow therotor disk to pass the starting
vortex. Thiscriteria can be expressed as:
ymp
=180x
R
where the forward velocity is related to theazimuthal increment
needed to move thedisk forward x in distance. In this study
adistance of 2 R gives a minimum of 573degrees to meet this
criteria.
RotorCRAFT CodeThe RotorCRAFT/AA (Mod 1.0) code is
a computer program that determines theoverall performance,
aerodynamic loadingand aeroacoustic pressure signature of arotor in
steady forward flight given the rotorgeometry and its flight
condition (reference7). The code also supplies other
informationincluding the circulation distribution on therotor
blades, the wake structure downstreamof the rotor blades (as
modeled by a full-span,Constant Vorticity Contour (CVC) free wakeof
Basic Curved Vortex Elements, seereference 8), and the load
distribution acrossthe blade span at each azimuth
location(reference 9). The code can also perform astructural
analysis of the blade to determineits mode shapes and natural
frequencies andcan be used to determine the far field soundpressure
level and spectrum at user-specifiedlocations.
In RotorCRAFT/AA the blade ismodeled as a vortex lattice
withcompressibility correction using airfoiltables. The wake is
modeled using a uniquemethod based on curves of constant
vorticityshed by the changes in bound circulation onthe rotor
blades. These curves are distributedin space and can form closed
loops as themaxima and minima of the boundcirculation shed by the
blades change. Thecurves of vorticity are convected as a free-wake
using local and global inducedvelocities. The fuselage is modeled
as acollection of vortex panels. Similar to abound vortex lattice,
the influence of thevortex panels is computed on all aspects ofthe
flow. A unique feature of this analysis
-
4
code is the implementation of a Fast Vortextechnique that
reduces the size of theinfluence matrices (reference 10),
allowingfaster processing of complex geometries.
For this study, azimuthal step sizeswere varied from 15 degrees
(24steps/revolution) to 7.5 degrees (48steps/revolution). The blade
is representedby 1 chordwise panel and 30 spanwisepanels. The code
assumes an undistortedform for the initial wake and relaxes
thisgeometry during iterations as the rotor i sturned through 2
revolutions. Nominalselections for wake core size, distribution
oftip and sheet filaments, and other controlvariables were set as
recommended in thecode documentation.
PEIRF CodeThe PEIRF code is based on an
iterative coupling between two singularitymethods, one modeling
the fuselage, the otherone the rotor and its wake.
The fuselage code is a low order panelmethod (constant source
and doubletdistribution), developed at ONERA(reference 11). The
sources intensities areexplicitly given by the slip condition on
thefuselage surface and they define the righthand side of a linear
system of which theunknowns are the doublet strength.
The rotor code is a lifting line methodwith a vortex wake model.
Initiallydeveloped by Eurocopter France (reference12) and known as
METAR (Modele d’ETudede l’Aérodynamique du Rotor) with aprescribed
rotor wake, this code has beenimproved by ONERA which developed
andvalidated the MESIR code (Mise en Equilibredu Sillage Rotor)
with a full free wakeapproach (reference 13). In these codes,
theblades are replaced by 25% chord liftinglines which take the
actual geometry of theblade such as spanwise variation of
chord,twist, sweep, anhedral, etc. into account. Therotor wake is
modeled by lattices of spanwiseand tangential vortices of constant
strength;therefore, it is equivalent to a constantdoublet
distribution. The rotor aerodynamicsolution is carried out by an
iterative processinitialized by a mean Meijer-Drees
inducedvelocity; the lift is obtained through 2-Dairfoil tables
with the computed local Machnumber and incidence and the
circulation iscalculated from the Joukowski law. Thevelocities
induced by the new rotor wakestrength can be computed by means of
the
Biot-Savart law. This iterative process i sstopped when the
variation of the inducedvelocities is less than a defined
value(typically 0.001 m/s) from one iteration toanother. The free
wake computationimplemented in the MESIR code is alsobased on a
quasi-steady azimuthal marchingprocess: the vortices forming the
rotor wakeare moved from one azimuthal position to thenext one by
taking into account the inducedvelocities from the rotor wake, the
blades andthe freestream. Because the rotor wakegeometry changes
from one revolution toanother, the influence coefficients
matrixshould be reevaluated. In order to acceleratethe process, the
influence coefficients matrixis computed again only every
threerevolutions but the induced velocities and thesingularity
strengths carried by the rotorwake are computed, as described
above, ateach azimuth. In most applications, ninerotor revolutions
are sufficient to convergethe process based on the mean
displacementof the wake from one iteration to another.The blade
angles (flapping and pitch) arespecified as input to the code.
The PEIRF code couples the two codesdescribed above by an
azimuthal marchingtechnique (reference 14). Two overlappedloops are
started on an initial configurationwhich could be a fuselage and a
prescribedrotor wake (METAR type), a free rotor wake(MESIR type),
or even a partially convergedresult of the PEIRF code. The internal
loopdistorts the rotor wake geometry consideringthe effects of the
rotor, its wake, and thefuselage. The velocities induced by
theblades and the rotor wake are computedusing the MESIR module.
The velocitiesinduced by the source and doubletdistributions on the
fuselage are evaluatedusing the Hess and Smith formulation. Thenthe
velocity and the new singularitydistribution on the fuselage are
evaluatedwith this new rotor wake geometry and theinner loop is
repeated for each azimuth. Theouter loop computes the
circulationdistribution on the lifting lines due to thenew wake
geometry using the iterativeMETAR/MESIR procedure and the
influencecoefficient matrix is computed again. A nacceleration
technique based on the farfield/near field approach has been
developedand has reduced the computational time byone third up to
as much as one half(reference 14).
Finally, because the PEIRF code is
-
5
based on a quasi-steady hypothesis, a specificalgorithm has been
developed in order tocompute the fuselage unsteady pressure. It i
sbased on the development of the unsteadyterm in the Bernoulli
equation. Details canbe found in references 14 and 15.
Unsteadyfuselage pressures are computed using theunsteady Bernoulli
equation:
CV
U U tP= - æ
èöø
-122
2
¶j¶
.
The first two terms are referred to as the“quasi-steady”
pressure. The last term is the“unsteady” term of the equation.
Thevelocity potential unsteady contribution onthe surface pressures
comes from changesin potential from all components of the
flow.Unsteady potential components include thatfrom the bound
circulation on the blade as itmoves relative to the fuselage
surface, thatfrom the motion of the wake vorticity as it i
sconvected relative to the fuselage, and termsfrom other
singularities that change i nstrength with time. All of the codes
includethe “quasi-steady” contribution to unsteadyfuselage
pressures. The PEIRF codeincludes three unsteady
potentialcontributions to the fuselage unsteadypressures: bound
circulation, wake vorticity,and unsteady fuselage doublet
strength.
Code Summary A summary of the significant
characteristics of the codes is given in thetable below. These
characteristics identifythe model used to capture the
significantinteractional features of the aerodynamics
ofrotor-fuselage configurations.
EXPERIMENT
Model and Apparatus
Powered Dauphin (365N) Model:The powered model consists of
three
components: a fiberglass fuselage shell, theinternal rotor drive
and control system andthe model rotor system. The fiberglass
shellis a 1/7.7 scale of the 365N model Dauphinhelicopter. No
simulation of secondary flows(engine inlet and exhaust or tail fan,
forexample) is attempted for this study. Theinternal drive and
control system consists ofa drive motor and electric control
actuatorsfor the blade pitch control via a swashplate.Two drive
motors were used. During thevelocity field survey and static
pressuremeasurements, an electric motor was used,but to measure the
unsteady pressures, ahydraulic motor was installed to minimizethe
electrical noise in the pressuretransducer signal. The rotor system
consistsof a hub and four elastic blades. Flap, lag,and pitch are
allowed about a singlespherical bearing that retains the blade.
Theblades are rectangular planform withconstant OA 209 airfoil
section and lineartwist of -8.8 degrees from the root cutout
at27.5% radius to the tip.
S2Ch Wind Tunnel:The ONERA S2Ch wind tunnel is a 3
m diameter test section, open return windtunnel. The test
section is reduced by a flatfloor section of 1.65 m width. The
tunnel i scapable of speeds up to 120 m/s with anaverage of 0.2%
freestream turbulence levels.At the nominal speed used for this
study, thetunnel is known to have a turbulence level of0.27% . Good
optical access in the test sectionallows implementation of laser
velocity
Table of Code Characteristics
Characteristic R W F RotorCRAFT PEIRF
Fuselage Source Panel Vortex Panel Source &Doublet
Blade Vortex Lattice Vortex Lattice Lifting Line
Compressibility Prantl-Glauert Airfoil Table Airfoil Table
Wake Vortex Lattice CurvedVortex
Vortex Lattice
Tip Vortex Fixed CoreRadius
Fat CoreModel
Fixed CoreRadius
-
6
measurement systems. Figure 1 is aphotograph of the powered
Dauphin model i nthe S2Ch wind tunnel.
3D Laser Velocimeter:For this study the ONERA has
implemented a three component LaserVelocimeter (LV). This
velocimeter useslight from 2 argon lasers (9 Watt), one for aviolet
beam (476.5 nm) at approximately 3Watts, and another that is split
into 2principal colors (green at 514.5 nm and blueat 488 nm) with
approximately 6 Watts in al llines. Once each color is split it is
processedby sending one of the beams through a Braggcell to apply a
frequency shift. Thisfrequency shift allows determination of
thedirection of travel for a particle passingthrough the fringes in
the sample volume.The monochromatic beams are directed toone
focused position in the test section at afocal length of
approximately 2 m. Thesample volume at this focused position
isapproximately spherical of diameter 0.4 mm.In this sample volume
interference fringesare produced for each of the colors.
Incenseparticles that seed the flow pass through thesample volume
and scatter light withvariation in amplitude corresponding to
thelight fringes. This scattered light i scaptured by two
Cassegrain telescopes i nbackward scatter mode. The receive
opticsseparate the three colors into pseudo-
components and convert the optical signalinto electrical signal
with photo-detectors.Signals from the photo-detectors areprocessed
by counters for frequency content.Velocities are derived from the
frequency ofthe fringe modulation of the light signals.Pseudo
components are converted to actual u,v, and w components using
trigonometricrelations of the physical characteristics of theoptic
system. The entire system is mountedon a three axis rigid table
that is translatedto move the measurement location in thetunnel up
to 600 mm in each axis with0.01mm accuracy.
Test ProceduresFor this study only a single flight
condition was evaluated. This flightcondition is characteristic
of moderateforward flight speed. At this condition therotor wake is
not expected to impact directlyon the fuselage, but even so, its
influence isexpected to be significant. At this testcondition,
collective and cyclic blade pitch isset to a trim condition defined
below.
The conventions during this study forthe harmonic coefficients
of blade pitch andblade flap are:
q q q y q y= + -0 1 1C Scos sin
b b b y b y= - +0 1 1C Scos sin .
Significant Test Parameters
Parameter Value Units
Thrust Coefficient, CT 0.0062
Advance Ratio, m 0.20
Tip Speed, WR 100 m/s
Shaft Angle of Attack, aS -7.0 degrees
Fuselage Angle of Attack, a -3.0 degrees
Collective Blade Pitch, q0 6.17 degrees
Longitudinal Blade Pitch, q1S -2.98 degrees
Lateral Blade Pitch, q1C -2.60 degrees
Blade Coning, b0 2.63 degrees
Longitudinal Flapping, b1C 5.45 degrees
Lateral Flapping, b1S -0.28 degrees
-
7
The data for this study were acquiredduring several entries in
the S2Ch windtunnel. Changes in the model andinstrumentation
between entries will bedescribed here.
Model measurements:Model instrumentation consisted of a
total balance that responds to both fuselageand rotor
aerodynamic loading. Two forcebalances were used: during the first
entrythe S1S2 balance was used, during the secondentry the D91
balance was used. Principalcharacteristics of these balances are
given inthe table of balance characteristics.
Field Velocity Measurement:LV measurements were taken in two
planes on the advancing side of thehelicopter model. Optical
access andsignificant effort in realigning the LVsystem would have
been involved i nmeasuring the retreating side. In the firstplane,
locations are distributed above andbelow the rotor. In the second
plane, al lmeasurements are below the rotor i nlocations to capture
the wake velocitydistributions.
Rotor azimuth was determined from ashaft encoder with 360 steps
per revolution.Two measurement techniques were usedduring this
study. For the first method, onlythe velocity field of one blade is
captured bylimiting the acquisition of signals to anazimuth range
of 45 < y < 135 degrees. Ateach measurement location,
themeasurement consisted of 100 particles forevery increment of 2
degrees of azimuth,resulting in 45 increments of azimuth. Inthe
second measurement technique, the
azimuth window was open for the entirerevolution, capturing the
velocity fieldinformation for all four blades. In this case,4
degree azimuth increments were used,resulting in 90 increments of
azimuth.
Accuracy for the LV system used i nthis configuration is stated
to have a relativeerror of 0.3% of the measured velocity.
Theprecision of the measurement is on the orderof 0.2 m/s.
Static Surface PressureMeasurement:
A distribution of 238 pressure tapswere made on the surface of
the model shell.The locations for these taps is shown i nfigure
2-1. Tubes from these taps wereconnected to a 6 head scanning valve
with aDruck 1 PSID pressure transducer. During apressure
measurement, each port wassampled for 0.5 seconds after settling
for 2seconds. Pressure measurement errors areintroduced by accuracy
of the individualtransducer, the reference pressure system,and also
by the processing of the signal fromthe transducer. The stated
precision of thestatic pressure measurements is 10 Pa.
Unsteady Pressure Measurement:Another test entry in the S2Ch
tunnel
was made with another fuselage shell thatincluded 44 dynamic
pressure transducersdirectly on the surface of the fuselage
shell.Locations for these transducers are shown i nfigure 2-2. The
range of the transducers usedwas 2 PSID. During measurement
ofunsteady pressure, the signal from thetransducer was sampled at
64 times per rotorrevolution. Unsteady pressures can containerrors
due to the accuracy and frequency
Table of Balance Characteristics
Balance S1S2 D91
Range Accuracy Range Accuracy
X Force, N 1800 ±40 250 ±0.50
Y Force, N 4500 ±4.5 200 ±0.40
Z Force, N 2100 ±3.0 2000 ±4.00
L Moment, m N 250 ±0.25 48 ±0.10
M Moment, m N 80 ±0.10 60 ±0.12
N Moment, m N 140 ±0.15 26 ±0.05
-
8
response of the individual transducers, thereference pressure,
timing of the samplesrelative to the blade azimuth, as well as
thedynamic characteristics of filtering andanalog to digital
conversion during theacquisition process. The stated precision
ofthe unsteady pressures is 10 Pa.
RESULTSComparisons between experimentally
obtained data and predicted results fromrotor aerodynamic codes
are shown here. Insome instances where experimental data arenot
available, the comparison is made onlybetween the predictive
methods.
Field VelocitiesTen locations have been chosen for
detailed comparison of the availablemethods. In the forward
velocitymeasurement plane, four locations on eachside of the model
have been chosen. Twolocations are 7% radius above the hub plane(at
75% and 107% radius) and two locationsare 4% radius below the hub
plane. On theretreating blade side of the model, these
fourlocations are mirrored. In the plane located42% radius behind
the hub, one location oneach side of the fuselage, relatively close
tothe fuselage, has been chosen.
The comparisons of azimuth dependentvelocity are shown in
figures 3-1 through 3-10. The first four figures include
theexperimental velocities. This set ofvelocities was taken with
full azimuth of therotor with a resolution of 4 degrees ofazimuthal
resolution. In the next fourfigures (3-5 through 3-8), there are
noexperimental data for comparison(retreating side of the model).
Figures 3-9and 3-10 are on the advancing side andretreating side of
the tail in the 42% radiusplane. Again, experimental data are
onlyavailable on the advancing side.
The predictive methods are shown i nall of the locations and
represent the state-of-the-art in singularity methods for
rotorcraftanalysis. Unfortunately, the predictivemethods did not
use the same increments ofazimuthal resolution. The RWF code wasrun
with the highest resolution (4 degrees),while the PEIRF code was
run with the lowest(15 degrees) of the three. Running the R W Fcode
at such a high resolution preventedattainment of true periodicity.
In the figures
that follow, subsequent blade passages shownfor the RWF code
predictions are muchcloser to a representation of periodic
solution.
In figures 3-1 and 3-5 the flow aboveand outboard the rotor disk
is observed.Little variation in downstream component i sseen with
the mean value very close to thefreestream value (20 m/s). Most of
thevariation is seen in the w , (vertical)component of velocity
with a 4 per revolutionvariation of approximately 1 m/s
peak-to-peak. None of the codes predict thismagnitude of variation,
or even the meanvalue of this component. Even the smallmagnitude of
the cross-flow component ofvelocity is missed in sign by the
methods.This location may be very sensitive to thelocation of the
tip vortices from the previousblades where even a small change i
nrelative position can significantly effect thismeasurement.
In figure 3-2 and 3-6, the velocity abovethe lifting portion of
the rotor blade/disk isobserved. In contrast to the
observedvelocities outboard of the tip, there is a strongperiodic
content in the u and w componentsof velocity, indicating the
passage of a vortexor circulation oriented principally in the
ydirection. The impulsive acceleration in theu component indicates
that the boundcirculation on the blade passed under themeasurement
location. The “up-down” spikein the w component also confirms
passage ofbound circulation. The RWF and CDI codescapture, but
under predict, the magnitude ofthe downstream velocity spike,
indicatingthat the location of the blade to themeasurement location
may be closer than thegeometry seems to indicate. The PEIRF codemay
not have enough azimuthal resolution topredict this impulsive
behavior. For thelateral component of velocity, theexperimental
measurement indicates verylittle periodic content, while the PEIRF
andRWF models predict significant variation.This may be due to the
lattice model used forthe “inboard sheet” of vorticity. The
passageof lines of vorticity that represent the sheetvorticity can
produce these periodicvariations that are non-physical. Theinboard
sheet model of the CDI code seems tomodel this particular region to
a much betterdegree. This localized effect of a vortexlattice sheet
may also be an explanation forsome of the velocity perturbation in
verticalflow predicted by RWF here. On theretreating side of the
disk, figure 3-6, the u
-
9
component of velocity sees a spike of theopposite polarity
indicating the change i nsign of the bound circulation passing
themeasurement point.
In figures 3-3, 3-4, 3-7, and 3-8 similarcharacteristics can be
observed. Asurprising correlation of all of the methodswith the u
component of velocity is seen i nfigure 3-4, while the other two
componentsdemonstrate either mean value or phasediscrepancies.
In figures 3-9 and 3-10, there is littleevidence that the
measurement locationexperiences any close vortex passage. Eachof
the codes seem to miss the measured meanvalue of at least one
component of velocity byapproximately 1 m/s.
Velocity field prediction methods arestill very sensitive to
wake geometryaccuracy. Although all of the methods use afree-wake
model, the actual geometry of thewake will significantly affect the
predictedvelocity field in the neighborhood of thewake.
Wake GeometryDue to the discrepancies noted in the
velocity field predictions, a comparison ofthe wake geometry
predictions is warranted.Unfortunately, the actual wake geometry
andthe RWF code wake data are not available.A comparison with the
predicted tip vortexgeometry predicted by CDI and PEIRF codesis of
some value.
Predicted wake shapes of a tip filamentfrom one blade with the
rotor stopped at fourazimuths are shown in figures 4-1 through 4-4.
In these figures three views of thegeometry are shown along with a
plot ofvertical deflection versus wake age.
Figure 4-1 shows the tip filament whenthe rotor is stopped at
azimuth of 0. The threesubsequent figures, 4-2, 4-3, and
4-4progressively show the rotor tip filamentswith the rotor at
azimuths of 90, 180, and 270degrees, respectively. In each figure
thereare four plots, counterclockwise from the top-left plot they
are: top view (Y versus X), sideview (Z versus X), back view (Z
versus Y),and wake age (Z versus azimuth sincereleased from the
blade). In the two lowerplots, the Z scale is multiplied by 2 to
expandthe vertical distortion.
One potentially significant observationis the starting location
for the filaments. In
the PEIRF model for the wake, the wake istrailed from the 1/4
chord of the blade tip.The RotorCRAFT code, however, begins
thetrailed filaments at the trailing edge of theblade at radial
locations governed by thegradients of circulation on the blade.
Thedependence of the wake evolution is stronglydependent on the
starting locations of thesetip vortex filaments. Once these
filamentsare unbound from the blade, their localconvection is
dominated by localinteractions with the tip vortices shed
fromprevious blades.
The total age of the two methods arealso different. This
specific run ofRotorCRAFT includes only two revolutionsof converged
wake, while PEIRF includesthree revolutions. The difference i
nazimuthal step size can also be seen in themore abrupt changes in
the CDI verticaldisplacement.
Vertical displacement of the wake dueto the influence of the
fuselage is also seenin these figures. This is seen in the rear (
Yversus Z) view and wake age plot. In thewake age plot, a small
effect of the fuselageis noticeable at age 180 degrees and
moreprominent vertical displacement occurs atage of 360, 540, and
720 degrees for figures 4-1 and 4-3. The similar effect in figures
4-2and 4-4 is seen at 270, 450, and 630 degrees.
Also shown in figures 4-1 through 4-4 isa tip vortex geometry
that is void ofperturbation due to wake or fuselage. Thistrajectory
is from the METAR model used toinitiate the PEIRF code. The only
verticaldisplacement is due to the component offreestream normal to
the rotor disk and auniform distribution of thrust inducedvelocity
from the Meijer-Drees inflowmodel.
Without verification fromexperimental wake geometry, the
accuracyin the geometric models must be assessedsubjectively from
information such as thevelocity field or unsteady pressure
datacomparison.
Surface PressuresSurface pressures for a powered model
fuselage are characterized by twocomponents, the steady part and
the unsteadypart. Several reasons can be given forpresenting these
parts separately. First, themeasurement of these data were
completed inseveral wind tunnel entries using different
-
10
pressure instrumentation. Second, thedynamic range of the
unsteady part i ssignificantly smaller than the meancomponent.
Plotting scales may mask theunsteady component by the magnitude of
thesteady offset.
In figure 5-1 the experimental pressurecoefficients along the
dorsal line of thefuselage is compared with the predictionfrom the
CDI and PEIRF programs. Theexperimental pressure values were
takenfrom both the test using static pressure portsand the average
of the data from theunsteady pressure test. Both codes used thesame
mesh of approximately 3000 panels forthe fuselage. Over the nose
the comparisonbetween the codes and both experimentalvalues are
reasonable. At a station justahead of the 500 mm location, the
codespredict a significant difference in CP . TheCDI code predicts
that the flow will stagnateat this body juncture while the PEIRF
codedoes not predict any deceleration of the flow.There is a single
data point with a CP valueof approximately 0.0 just behind the
hublocation where separation is expected but notincluded in the
panel model used by thecodes. The panel methods also predict
ahigher acceleration in the region of the hub.Without a model for
the hub and itsseparation, this acceleration of the flow isexpected
to be different from the measuredvalues.
In the region downstream of the hubthere is some discrepancy
between the steadypressure values shown in figure 5-1 and
theaverage of the unsteady pressure data. Inthis region some flow
separation is expectedand may account for some of themeasurement
differences. The prediction ofthe codes is also in some
disagreement withthe data due to this separation that the codesare
not accounting for. Over the tail boomthere clearly is a
discrepancy between thetwo codes in the region of strongest
influenceof the rotor wake. Although the CDI valuesfor pressure are
nominally closest to theexperimental value, the effect of the
hubregion separation on this comparison has notbeen determined. In
general, both codes do areasonable job of predicting the
steadycomponent of pressure with the effects of arotor at this
speed.
Unsteady pressure comparisons areshown in figures 5-2 to 5-6. In
each of thesefigures the perturbation pressure (mean
removed) is compared as a function of rotorazimuth. Figure 5-2
shows the unsteadypressures at five locations at the section cutA-A
(140 mm from the nose of the model).Both codes agree with each
other i namplitude and phase of the unsteadypressures at the sides
of this section cut.There is a 180 degree phase disagreementwith
the experimental values on thestarboard side (retreating blade
side) of thissection. Additionally, both codes are i ndisagreement
with the amplitude of theexperimental pressures over the top of
thissection.
In figures 5-3 to 5-6 unsteady pressuresare shown at selected
locations from thesection cuts at B-B (240 mm), D-D (735 mm),E-E
(885 mm), and F-F (1035 mm). From allof these locations only 5
locations (19, 20, 33,34, and 35) from two cuts (D-D and F-F)
wereshown to have acceptable correlation with thePEIRF code in both
phase and amplitude.The locations at section F-F have the
mostdirect physical relation between the section ofthe rotor blade
with the highest loading andtransducer location. Here the
unsteady
potential term, ¶j¶t
, due to close passage of
the bound circulation (local lift) with veryhigh relative
velocity, is expected todominate the unsteady pressure.
Thislocation is also downstream of the hubregion, where the
presence of the hub coulddiffuse any structure of the strong tip
vortexfrom the leading half of the rotor disk. Atother locations
strong contributions areexpected from both passage of the
boundcirculation on the blade and the convection ofthe strong tip
vortex at lower speed(approximately freestream) but with
closerspacing to the transducers. Accuratelypredicting the phase
and amplitude of thesetwo sources is still in question.
CONCLUSIONSComparisons have been made with
three analytical methods and a unique set ofexperimental data.
The Rotor-Wake-Fuselage (RWF) code is an AFDD in-housedeveloped
code for exploring methods ofcomputing the combined
rotor-fuselageproblem. . The other US code is a recentversion of
the Continuum Dynamics, Inc.Computation of Rotor Airloads in
ForwardflighT/Aeroacoustic Analysis (RotorCRAFT)
-
11
code. At ONERA, the PEIRF (Programmed’Etude d’Interaction
Rotor/Fuselage) codewas also developed to simulate this problem.The
effects of the rotor wake on the flowfieldof a helicopter have been
assessed usingexperimental data and the predictions ofthese codes.
The significant observationsare:
1. From the field velocitycomparisons, locations where blade
boundcirculation and tip vortex come close givegood indications to
the relative location andstrength of the vorticity. However,
latticemodels for the wake used by RWF andPEIRF codes produced
unexpected periodicvelocities when the filaments of the sheet
areconvected close to the measurement location.
2. The geometry of the tip vortexpredicted by the codes has only
beencompared between codes. The wakeevolution is strongly dependent
on thestarting locations of these tip vortexfilaments. The larger
azimuthal step sizeresults in the more abrupt changes i nvertical
displacement. Comparison withexperimental wake geometry data is
neededto resolve additional differences between thecode
methods.
3. In general, the CDI and PEIRF codesdo a reasonable job of
predicting the steadycomponent of pressure with the effects of
arotor at this speed. Without a model for thehub and its
separation, the flow predicted bythe codes is expected to be
different from themeasured values in the hub region.
4. Over the tail boom there is clearly adiscrepancy in steady
pressure valuesbetween the two codes in the region ofstrongest
influence of the rotor wake.Although the CDI values for pressure
arenominally closest to the experimental value,the effect of the
hub region separation on thiscomparison has not been
determined.
5. The PEIRF code matchedexperimental values of unsteady
pressurevery closely at only 5 of the points examinedin this study
while the CDI code did not showeven this level of correlation. The
locationson the tailboom top, where this correlation isbest, have
the most direct physical relationbetween the section of the rotor
blade with thehighest loading and transducer location.Here the
unsteady potential term due to closepassage of the lifting sections
of the blade isexpected to dominate the unsteady pressure.At other
locations contributions to the
measured unsteady pressure due to passageof the strong tip
vortex and the boundcirculation of the blade cannot be
separated.
ACKNOWLEDGMENTSThe authors wish to acknowledge the
significant contributions of the S2 Chalais-Meudon wind tunnel
team and EurocopterFrance for technical support of theexperiments.
This study was partiallysupported by the Service Technique
desProgrammes Aeronautiques (STPA) andDirection des Recherches,
Etudes etTechniques (DRET). The study was alsosupported in part by
the National Aeronauticsand Space Administration, LangleyResearch
Center.
REFERENCES1. Prouty, R. W.: Helicopter
Aerodynamics , PJS Publications,1985.
2. Berry, J.: “A Method of Computing theAerodynamic Interactions
of a Rotor-Fuselage Configuration in ForwardFlight”, PhD
Dissertation, GeorgiaInstitute of Technology, May 1990.
3. Berry, J.: “A Multi-Element VortexLattice Method for
Calculating theGeometry and Effects of a HelicopterRotor in Forward
Flight”, 26th AIAAApplied Sciences Meeting, Reno, NV,1988. (AIAA
88-86-0336)
4. Hess, J. L. and Smith, A. M. O.:“Calculations of Non-lifting
PotentialFlow About Arbitrary Three-Dimensional Bodies”,
DouglasAirdraft Company Report E. S. 40622,March 1962.
5. Hess, J., and Smith, A. M. O.:“Calculation of Potential Flow
AboutArbitrary Bodies”, Vol. 8 of Progressin Aeronautical Sciences,
PergamonPress, Oxford and New York, 1966
6. Crispin, Y.: “Computing the Wake of aRotor in Forward
Flight”, AIAApaper 82-8000, 1982.
7. Wachspress, D. A., Quackenbush, T. R.,Boschitsch, A. H. and
Lam, C-M. G.:“RotorCRAFT/AA (Mod 1.0) User’sManual”, Continuum
Dynamics, Inc.
-
12
Technical Note No. 95-23, February1996.
8. Bliss, D. B., Teske, M. E., andQuackenbush, T. R.: “A
NewMethodology for Free Wake AnalysisUsing Curved Vortex
Elements”,NASA Contractor Report 3958,December 1987 (also
ContinuumDynamics, Inc. Report No. 84-6, May1984).
9. Quackenbush, T. R., Bliss, D. B.,Wachpress, D. A.,
Boschitsch, A. H.,and Chua, K. C.: “Computation ofRotor Aerodynamic
Loading i nForward Flight using a Full-SpanFree Wake Analysis”,
NASAContractor Report 177611, October 1990(also Continuum Dynamics,
Inc.Report No. 90-05, Dec. 1990).
10. Chua, K. and Quackenbush, T. R.:“Fast Vortex Technology (FVT
Mod1), Theory Documentation, SoftwareUser’s Manual,
Programmers’Manual”, Continuum Dynamics,Inc. Report No. 91-06P,
December1991.
11. Ryan, J., Falempin, G., Le TH.: “RotorPlane Velocities
Induced by a
Helicopter Fuselage”, Presented at theSecond Helicopter Basic
ResearchConference, Army Research Office,College Park, MD,
1988.
12. Dehondt, A., Toulmay, F.: “Influence ofFuselage on Rotor
InflowPerformance and Trim”, Presentedat the 15th European
RotorcraftForum, Amsterdam, 1989.
13. Michea, B., Desopper, A., Costes, M.:“Aerodynamic Rotor
LoadsPrediction Method with Free WakeAnalysis for Low Speed
DescentFlights”, Presented at the 18thEuropean Rotorcraft
Forum,Avignon, 1992.
14. Bettschart, N., Gasser, D.: “Analysis ofHelicopter
Rotor-FuselageInteraction”, Presented at the 20thEuropean
Rotorcraft Forum, 1994.
15. Gasser, D., Bettschart, N., Drouin, B.:“Theoretical and
ExperimentalStudies on Unsteady HelicopterInteractional
Aerodynamics”,Presented at the American HelicopterSociety, Vertical
Lift Aircraft DesignConference, San Francisco, CA, 1995.
-
13
Figure 1: Dauphine Model in S2Ch Wind Tunnel
Figure 2-1: Locations of Pressure Ports
Figure 2-2: Locations of Unsteady Pressure Taps
-
14
1516171819202122232425
u, m/s
-5-4-3-2-1012345
v, m/s
0 90 180 270 360
-5
0
5
Azimuth, deg.
w, m/s
Experiment
RWF
CDI.
PEIRF
Figure 3-1: Velocity at x =0.0 R; y=-1.07 R; z=0.07 R
1516171819202122232425
-5-4-3-2-1012345
0 90 180 270 360
-5
0
5
Azimuth, deg.
u, m/s
v, m/s
w, m/s
Experiment
RWF
CDI.
PEIRF
Figure 3-2: Velocity at x =0.0 R; y=-0.75 R; z=0.07 R
-
15
1516171819202122232425
-5-4-3-2-1012345
0 90 180 270 360
-5
0
5
Azimuth, deg.
u, m/s
v, m/s
w, m/s
Experiment
RWF
CDI.
PEIRF
Figure 3-3: Velocity at x =0.0 R; y=-1.07 R; z=-0.04 R
1516171819202122232425
-5-4-3-2-1012345
0 90 180 270 360
-5
0
5
Azimuth, deg.
u, m/s
v, m/s
w, m/s
Experiment
RWF
CDI.
PEIRF
Figure 3-4: Velocity at x =0.0 R; y=-0.75 R; z=-0.04 R
-
16
1516171819202122232425
-5-4-3-2-1012345
0 90 180 270 360
-5
0
5
Azimuth, deg.
u, m/s
v, m/s
w, m/s
Experiment
RWF
CDI.
PEIRF
Figure 3-5: Velocity at x =0.0 R; y=1.07 R; z=0.07 R
1516171819202122232425
-5-4-3-2-1012345
0 90 180 270 360
-5
0
5
Azimuth, deg.
u, m/s
v, m/s
w, m/s
Experiment
RWF
CDI.
PEIRF
Figure 3-6: Velocity at x =0.0 R; y=0.75 R; z=0.07 R
-
17
1516171819202122232425
-5-4-3-2-1012345
0 90 180 270 360
-5
0
5
Azimuth, deg.
u, m/s
v, m/s
w, m/s
Experiment
RWF
CDI.
PEIRF
Figure 3-7: Velocity at x =0.0 R; y=1.07 R; z=-0.04 R
1516171819202122232425
-5-4-3-2-1012345
0 90 180 270 360
-5
0
5
Azimuth, deg.
u, m/s
v, m/s
w, m/s
Experiment
RWF
CDI.
PEIRF
Figure 3-8: Velocity at x =0.0 R; y=0.75 R; z=-0.04 R
-
18
1516171819202122232425
-5-4-3-2-1012345
0 90 180 270 360
-5
0
5
Azimuth, deg.
u, m/s
v, m/s
w, m/s
Experiment
RWF
CDI.
PEIRF
Figure 3-9: Velocity at x =0.42 R; y=-0.19 R; z=-0.35 R
1516171819202122232425
-5-4-3-2-1012345
0 90 180 270 360
-5
0
5
Azimuth, deg.
u, m/s
v, m/s
w, m/s
Experiment
RWF
CDI.
PEIRF
Figure 3-10: Velocity at x =0.42 R; y=0.19 R; z=-0.35 R
-
19
-1.0 -0.5 0.0 0.5 1.0
-0.4
-0.2
0.0
0.2
Y/R
0 90 180 270 360 450 540 630 720-0.50
-0.25
0.00
0.25
Z/R
Wake Age, deg.
0.0 1.0 2.0 3.0 4.0
-0.4
-0.2
0.0
0.2
Z/R
X/R
(Z/R mag x2)
0.0 1.0 2.0 3.0 4.0
-1.0
-0.5
0.0
0.5
1.0
Y/R
Code
PEIRF y = 0°
CDI y = 0°
METAR y=0°
Figure 4-1: Wake geometry with rotor at y=0.0
-
20
-1.0 -0.5 0.0 0.5 1.0
-0.4
-0.2
0.0
0.2
Y/R
0 90 180 270 360 450 540 630 720-0.50
-0.25
0.00
0.25
Z/R
Wake Age, deg.
0.0 1.0 2.0 3.0 4.0
-0.4
-0.2
0.0
0.2
Z/R
X/R
(Z/R mag x2)
0.0 1.0 2.0 3.0 4.0
-1.0
-0.5
0.0
0.5
1.0
Y/R
Code
PEIRF y = 90°
CDI y = 90°
METAR y=90°
Figure 4-2: Wake geometry with rotor at y=90.0
-
21
-1.0 -0.5 0.0 0.5 1.0
-0.4
-0.2
0.0
0.2
Y/R
0 90 180 270 360 450 540 630 720-0.50
-0.25
0.00
0.25
Z/R
Wake Age, deg.
0.0 1.0 2.0 3.0 4.0
-0.4
-0.2
0.0
0.2
Z/R
X/R
(Z/R mag x2)
0.0 1.0 2.0 3.0 4.0
-1.0
-0.5
0.0
0.5
1.0
Y/R
Code
PEIRF y = 180°
CDI y=180°
METAR y=180°
Figure 4-3: Wake geometry with rotor at y=180.0
-
22
-1.0 -0.5 0.0 0.5 1.0
-0.4
-0.2
0.0
0.2
Y/R
0 90 180 270 360 450 540 630 720-0.50
-0.25
0.00
0.25
Z/R
Wake Age, deg.
0.0 1.0 2.0 3.0 4.0
-0.4
-0.2
0.0
0.2
Z/R
X/R
(Z/R mag x2)
0.0 1.0 2.0 3.0 4.0
-1.0
-0.5
0.0
0.5
1.0
Y/R
Code
PEIRF y = 270°
CDI y=270°
METAR y=270°
Figure 4-4: Wake geometry with rotor at y=270.0
-
23
0 500 1000 1500
-1.0
-0.5
0.0
0.5
1.0
1.5
CP
X station, mm
Experience
Exp Uns
CDI
PEIRF-30
Figure 5-1: Mean pressure along dorsal line of fuselage
Unsteady Pressures
Section Cut AA
Location 5 Location 4 Location 3
Location 1Location 7
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ExperimentCDIPEIRF
1
2
34
5
6
7
Z
Y
Figure 5-2: Unsteady pressures at fuselage station AA
-
24
Unsteady Pressures
Location 14
Location 12 Location 11 Location 10
Location 8
Section Cut BB
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ExperimentCDIPEIRF
8
9
1011
12
13
14 Z
Y
Figure 5-3: Unsteady pressures at fuselage station BB
Unsteady Pressures
Section Cut DD
Location 17
Location 19Location 20Location 21
Location 23
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ExperimentCDIPEIRF
17
18
1921
22
23Z
Y
20
Figure 5-4: Unsteady pressures at fuselage station DD
-
25
Unsteady Pressures
Location 30
Location 28 Location 27 Location 26
Location 24
Section Cut EE
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0 90 180 270 360-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0 90 180 270 360-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ExperimentCDIPEIRF
24
252628
29
30Z
Y
27
Figure 5-5: Unsteady pressures at fuselage station EE
Unsteady Pressures
Section Cut FF
Location 31
Location 33Location 34Location 35
Location 37
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
0 90 180 270 360-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0 90 180 270 360-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0 90 180 270 360-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0 90 180 270 360-0.10
-0.05
0.00
0.05
0.10
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ψ, deg.
CP
ExperimentCDIPEIRF
31
323335
3637
Z
Y
34
Figure 5-6: Unsteady pressures at fuselage station FF
