Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Dynamic Response analysis code, ADAMS, to analyze one of these new

of a Variable-Speed Rotor designs undergoing conversion from rotor-borne to


wing-borne Sight.
During Rapid Sbaft Tilt
Typically, helicopter rotor dynamics analysis
Andrew S. EBiott codes sre based on equations of motion which in-
Senior Development Engineer corporate many implicit assumptions, mast notably
Mechanical Dynamics, Inc. constant rotor speed, emsfi to moderate hub snd
2301 Commonwealth Blvd. blade dieplscements, restricted binge sequencing snd
Ann Arbor, Michigan 48105 periodic response. Many of the advanced concept
rotors violate some or sB of these sesumptione, ee-
pecisfiy during conversion Born their VTOL Sight
mode to the high speed Sight mode. Because of this,
Abstract s new analytical tool for rotor dynamics, based on
The dynamic responses of five-bladed, articulated the general-purpose, large displacement, multibody
rotor undergoing translational accelerations, rapid systems analysis (MSA) code ADAMS, wae devel-
changes in rotor speed snd large pylon tilts sre oped se s joint project between the McDonnell Dou-
examined snalyticsfiy using sn extended general- glas Helicopter Co. snd Mechanical Dynamics, Inc.
purpcee mechanical system simulation code. AB Verification of this code'e sbihty to handle mtor-
nonlinear geometric snd hinemstic coupling efiects Bhe problems hse been presented by EBiott and
are retained in the analysis. Aerodynamic forces McConvine[1]. Due to its basis in ADAMS, this new
on the rigid rotor blsdee sre generated using two- code requires no dynamic or hinemstic assumptions
dimeneionsl, quasi-eteady strip theory snd enhanced and is restricted, st least in structural terms, mainly
table loohup. An autopilot-lite automatic controfier by the limits of computational fiosting-point mathe-
is used to obtain commanded thrust snd hub mo- matics. Derivation of appmpriste sirloade for these
ments, and to maintain desired mtor speeds. The unusual configurations is s separate issue which re-
necessary controfier equations sre appended directly mains s subject for extensive future research.
to the structural equations and integrated simulta- The particular rotor design considered in this
neously. The combined effects of forward velocity study is derived firom s configuration called the Trail
snd acceleration, rotor epeed snd deceleration, tilt Rotor Convertiplane (TRC) which hss been under
and tilting rate on the oversB system response have review st McDonnell Douglas. Similar to the tiltro-
been computed over a range of Sight conditions for s tor, the TRC is s conventional airplane configuratio
lightly loaded rotor. Selected results are presented, with tilting, pylon-mounted rotors st the wing tips.
demonstrating the advantages snd problems of this However, the TRC concept includes separate for-
computational approach to rotor system analysis. ward propulsion units eo that, at conversion speed,
with the wings supporting the aircraft weight, the
Introduction pylone csn tilt efl while the rotors spin dowa and
finsfiy fold to s streamlined position for high-speed
Many of the advanced, high-speed VTOL and forward Sight. Full conversion could occur in se little
STOVL concepts under consideration for future ad- ss fifteen seconds, during which time rotor response
vanced miTitsry snd civiTisn needs include various snd loads sre criticaL
combinations of stopped rotors, tilting rotors, tilt- An almost overwhelmingly large amount of out-
ing wings, jets snd propefiers. A deeiga goal for put data is generated when the ADAMS model is
some of t, here concepts is s giy% speed increase over run through s convennon scenario, especisfiy when
current V-22 capabilities while maintaining compa- compared to more typical rotor analyses where re-
rable payload fractions. These new configurations sponeee sre given in terms of s few lower harmonics
present very difficult analytical challenges in both
of the rotor speed. In fact, analysis of these data
structural dynamics snd aerodynamics. This pa- seems more similar to that of Sight test data, but
per presents the results of a successful attempt to where every parameter the analyst could desire hss
uee sn extended general-purpose multibody systems
been instrumented snd sampled at a very high rate
Presented at ihe 4T'" Anneal Forum of the American (the integration output step cise). This type of anrd-
Behcepier Soaety, Phoenix, Ariiosa, May, last. ysis might best be thought of as a "numerical experi-

1029
Qo
Aero Control Points
~

Figure Ii ADAMS Model of Tilting Rotor Figure 2i Rotor Blade Assembly

ment. " In order to concentrate on the most interest- respectively. The blades have a '23.33% root cutout,
ing (to the author) transient efl'sets discovered, only and sre serodynsmicsfly twisted st — 204. Nominal
s very smaB portion of the total available data will hover tip speed is 760 ft/sec.
be presented snd discussed. Additional analyses of The rigid rotor shaft transmits power to the rotor
these data sre invited. itself, snd is hinged at the bottom to afiow for rotor
tilting using a prescribed motion. Since the purpose
Structural Modelling of this study wss epecificafiy to examine transient
rotor response, no eifort wss made to model shaft or
Rn4or - The 5-bladed rotor structural model used pylon/wing dyuamics, although adding these to the
in this study is very similar to the 1-bladed model model would be straightforward. To achieve forward
presented by EBiott snd McConvifie[1]. It consists of velocity, the base of the pylon is translated through
a 30-foot diameter, fully articulated rotor, mounted the "inertial" sir, also using a prescribed motion. As
at the end of s 9.5 foot long pylon which can tilt will be seen below, this approach is especisfiy con-
fmm s vertical to fufly sft horisontsl position (see venient for the ADAMS-based rotor model, because
figure I). The general physical parameters of this ro- net incident velocities st aerodynamic cont, rol points
tor were chosen for sn approximately 30,000 pound sre the same as the inertial velocities which the code
maximum hover Bft capacity. uses to define system motions.
The mtor blade assemblies consist of three rigid
parts, connected to each other snd to the hub us-
iag revolute joints to give the unusual flap-torsion-
Power Tkafn - The very simple "engine" model
incorporated in the analysis consists of s single-
lsg hinge sequence, designed to permit blade fold-
stste rprn smoother, equation (I), snd s single-
ing about the fisp hinge (see figure '2). Construc-
tion of this model within the ADAMS framework
state integistor, equation (2), combined into s sim-
ple proportioasl-integral controfler for rotor torque.
is quite easy, since the pmgrsm will automaticsfly
take care of the nonlinear geometric and kinematic The time constants snd gains are chosen to give res
effects. The analyst uses the real, physical pmp- sonsbly realistic response chsracteristica
erties of the blades snd places the hinges st their
actual design locations. Adding blade ilexibiTity to
ii = 10(II —xi) (I)
this model is not diflicult, but computer run times ii = 100/x(IIc —xt) (2)
become extremely long. q = 25000(II, —xi) + 375xr (3)
Each 16-foot long blade assembly weighs 164
pounds. The blade chord is 1.542 feet. Flap, tor- where Ac is the command rotor speed in radi-
sion snd lag hinge offsets sre 1.25, 2.375 snd 3.5 feet ans/second snd Q is the shaft torque in foot-pounds.

1030
It is important to note that the drive system tions made it unrealistic to include true 3-D calcu-
described in equation (3) csn snd does provide lations. For these reasons, a simple, fixed induced
both positive (motor) and negative (braking) torque inflow distribution was used snd only one (coflec-
equsfly well. While this type of drive technology is tive) dynamic inflow state wae considered. Blade
not currently available, such convertible engines snd aerodynamics were restricted to 2-D strip theory for
bidirectional power transmission systems are under the same reasons.
development and will be essential to making conver- General-purpcse MSA codes do not notmafly in-
tiplsnes like the TRC feasible. clude aerodynamic forces as part of their catalog of
standard parts snd connections, but the facilities
Contrctls - An elflcient, autopilot-like automatic needed to add them do exist. Steady, inviecid 2-
controller is used to trim the rotor to the desired D aemdynsmic forces csn be written directly into
thrust levels snd hub moments in real time. This the ADAMS dstaset, while more complicated un-
simple, six-state controfler is based on s pmto- steady and/or nonlinear models can be added via
type trim controller developed by Peters, IGm snd user-written subroutines which are linked directly
Chen['2] snd later enhanced by Chouchane[3]. Minor into the executable code. Because the inertial veloc-
modifications were made to limit tbe advance ratio ities (i.e. with respect to the air mass) of sny point
dependency and to reduce the gains as the shaft tilts on the structure csn be requested and resolved into
sft and total rotor inflow increases to very large val- sny convenient reference frame, the needed inputs
ues. The fully coupled controfler attempts to adjust to any aerodynamic forcing model are easy to gen-
the collective pitch to achieve commanded thrust erate. Two aerodynamic control points were placed
level (10% of nominal rotor thrust for sfl cases in at Gaussian integration points in each of the four
this paper) snd to adjust the cyclic pitch to give semdynamic "strips" on each blade as shown ia fi-
sero hub moments in afl flight conditions. Details of gur 2.
the controfler formulation can be found in the abave At MDHC, s fully nonlinear, unsteady two-
refemnce. dimensionsl aerodynamics model, including viscous
The controflet outputs, fib, etc snd 8rs, ste snd compressibility eifects, was implemented with
used to create tesctionless pitching moments which force snd differential equation submutines linked
are s function of the diiference between the actual directly into the main ADAMS executable. This
snd the commanded blade pitches. In the current model, based on the approach suggested by Leish-
model, these moments sre linearly dependent on the msn and Beddoes[4], has been extensively validated
blade pitch enure, with an equivalent control sys- for a few airfoil sections snd used in various other
tem stilfness st the pitch bearing of about 50,000 rotor snalysee[5]. For the results presented in this
ft-lbs/radian. The functional dependencies sre eas- paper, however, a NACA0012 section wss used and
ily defined in ADAMS and could be used to simulate blade airlosds were computed using s simpler, qussi-
mechanical or hydraulic connections if the actual ro- steady strip theory approach, with viscous snd com-
tor control system was defined. Similarly, the actual pressibility elfecte induded via table lookup. One
autopilot controller equations could be used if they advantage to the quasi-steady technique is that data
were available. for advanced sections can be easily incorporated.
However, because of the extreme angles of attack
and very rapid changes in local flow conditions as-
Aerodynamic Modelling sociated with the high advance ratios snd negative
Adding reasonably accurate aerodynamic forces to shaft angles, evea this formulation caused some com-
the rotor structure and controls wae probably the putational difficulties (see Computational Consider-
mmt chaflenging part of the overall model coastruc- ations section below).
tion. Even more than for etsadard rotorcraft config-
urations, wake elfects, rotor/body interactioas, vis- Section Loads - The strip theory model is a
cous snd compressibility effects, and spanwiee snd slightly modified version of the classical approach to
tip flow effects are undoubtedly very important. Un- 2-D unsteady aerodynamics found in Bisplinghoif,
fortunately, very little work hss been done on rotor Ashley snd Hoffman[6]. As is customary for qussi-
wake modelling for atypical flight conditions such ss steady formulations, the reduced frequency is as-
these of interest here. Also, computational limita- sumed to be 0 and the lift deficiency factor, C(k), is

1031
set identically to 1. When rewritten in terms of sec- so large that the wake effects are barely noticeable
tion velocities, the inviscid, incompressible lift snd at the disk plane.
moment expressions sre:

L = rrplP nit + 2ripU brut (4) Computational Considerations


I
= -xpfPui— The main reason that most dedicated hehcopter ro-
Mtf, (5)
tor dynamics codes, for example CAMRAD]y] and
xpfP (Ua/2+ ba/8) COPTER[8], do make the assumptions mentioned
above, i.e. nearly constant rotor speed, moderate
where w is the upwssh velocity at the indicated displacements, restricted geometries and periodic re-
chordwiee station, a is the pitching velocity snd b sponse, is to reduce the number, complexity snd cou-
is the blade semichord. pling of the equations of motion that must be solved.
It csn be seen that the unsteady lift, equation 4, These swmmptions sre entirely appropriate for the
is composed of noncirculatory snd circulatory por- major portion of rotor analyses snd they usually al-
tions, snd that the unsteady moment, equation 5, is low solutions to be found in relatively short order
composed of one term due to the offset of the noncir-
(a few hours), even on older generation computer
culatory lifl fmm the I/4-chord snd one due purely hardware.
to pitching motions. Since the MSA-based approach to rotor analysis
This formulation is inadequate for rotor problems, makes none of these assumptions, it is to be expected
which often exceed the limits of linear theory, snd that computational requirements will be higher. Al-
is improved by incorporating wind tunnel data for though the more intuitive modefling process snd the
the 0012 section in place of the circulatory portion automatic generation of the equations of motion by
of the lift. Tabular data is also added to the mo- the code tate some load oif the analyst, they trans-
ment expression and used entirely for drag. To ac- fer it directly to the computer. In fact, the pro-
count for the diminishing influence of noncirculatory cess somewhat parallels the development of compu-
efl'ects with increasing Mach, those portions of the
tational Suid dynamics, where now almost sll eifort
model sre reduced by the empirical factor (I —M ),
goes into solution methodology snd grid generation
yielding these final expressions for section lift, drag snd very little is spent on formulation. Fortunately,
and moment: hardware speed, capability, availability snd slford-
abiTity have risen dramatically in the psst few years
(I —M )xpb rot + Liame (8)
snd it is now passible to run what used to be consid-
I
(I —M )[-ripPrbt—
2
(7) ered very large problems on workstations and even
on PC's.
xpfp(Ua/2+ ba/8)] + Mrus, Even so, the computational structural dynamics
Dura (8) problem posed here is quite complex. As the system
was assembled by ADAMS, there were 850 highly
Inflaw - Incorporating the effect of the induced coupled, nonlinear equations to be solved. Further-
inflow is less intuitive in the ADAMS model, since more, ss stated above, the high advance ratios snd
it is not available sss part velocity. The approach negative shaft angles considered produced extreme
used wss to first compute the total thrust by sum- angles of attack snd very rapid changes in local Sow
ming the net force along the shaft sxie over sll the conditions. Finally, because the entire rotor tilts
control points. This thrust is used in s standard through 90', it wss not possible to define the model
iterative solution for the induced inflow, which is so as to completely avoid Euler angle singularities,
then delayed by passing it through the differentia which force ADAMS to rotate part axes internally
equation representing s simple dynamic inflow filter. snd often lead to s complete regeneration of the sys-
The resultant inflow is resolved into components at tem equations by the code. The net result of these
each control point and added to the inertial veloci- difflculties is that the integrator is forced to take
ties before computing the incident velocity snd angle smaller steps snd runs slower. The esses selected for
of attack for the table lookup. It should be noted, this study were run on a Sun SPARCstation I+, snd
however, that soon after the rotor begins to tilt ait averaged about I CPU hour per second of simulation
during conversion, the non-induced inflow becomes time, depending, of course, on the flight conditions

1032
Figure 3: Original Data from C81 Tables Figure 4: "Improved" Mach-Dependent Data

involved. solution. Total run time for the full eet, not includ-
Various changes were made to the force subrou- ing pmt procesain, wse over 350 CPU hours on the
tines snd to the integrstor parameters snd step sixes SPARCststion I+. As might be expected, the tilt-
to improve code performance. As experience with ing cases st higher airspeed were the most difficult
the model grew, it wss found that the convergence to complete. For all cases, the pitch controfler was
of the aerodynamic forces wss taking sn unexpect- active snd attempted to maintain 3000 pounds rotor
edly large amount of CPU time when the control sec- thrust snd sero hub pitching and rofling moments.
tions were experiencing significant amounts of stall As wss mentioned previously, even though only s
or reversed Sow. This problem was traced to two very small subset of the information available from
sources, the table lookup methodology in ADAMS ADAMS wss requested, s very large amount of data
and the tabular data itself. It wss found that the was generated for each of these runs. The datsset
code uses diferent orders of interpolation (linear snd included 46 computed or derived output requests.
cubic) along the two axes of the table, eo that the Typically, in order to improve integrator perfor-
user must, be sure to orient his data appropriately mance, the integration output step vise wss between
to avoid spline buckling. Also, it wae found that the one snd Sve mifliseconds. The output data were
CSI tables being used for the NACA 0012 section later resampled during poetproceesing at a more res
were extremely sparse outside of the "normal" angle sonable 100 Hs rate to reduce the file sixes. A com-
of attack range, snd that outside this same range, no plete listing of the output requests appears in ts
compressibility corrections st sll were being made. ble 1.
MDHC wss able to pmvide new tsbhw with Mach The hub reactions were found by measuring the
effects included and with many more grid points in forces across the stifl' bushing which separates the
the angle of attack range 3IF & rr & 3W', which hub hearing from the static mast. Rotor thrust is
greatly improved integrator performance. An exam- the shaft axis component of the summed aerody-
ple of the table improvement is shown in figures 3 namic forces for all five blades. Tip deilections are
and 4, which are surface plots of 0012 pitching mo- measured from the original, undefiected tip position
ment vs. angle of attack vs. Mach number, covering and are resolved in s plane perpendicular to the
the ranges -1804 & cr & 18iP and 0 & Af & 1. undeflected blade axis. Aerodynamic normal snd
chordwtse forces for blade 182 sre resolved into the
local section reference frames; pitching moments are
about the I/4-chord.
A complete summary of the test cases csn be found It is clearly not possible to present sll of these re-
in table 2. In order to begin the acceleration cases, sults in this paper. In fact, due to the large quantity
the simulation was first allowed to reach a fairly of data, it wss diflicult to find s method to present it
steady hover condition. About two seconds of sim- here in any coherent fashion. The author has chosen
ulation time were added to each end of each of the to borrow from the experimental world snd use the
other runs to sfiow the model to settle into a periodic strip chart foriuat to present "traces" of the most in-

1033
S' al Units Si al Units 3150
z OHub lbf Ftv OSta. I lbf
lbf
Fy
Fz OHub
OHub Ibf
lbf
" CISta.
Sta.
2
3 lbf 3100
" rn
Mz OHub ft-Ibf Sta. 4 lbf
My OHub ft-Ibf
" OSta. 5 lbf
ft-Ibf
" OSta. 6 Ibf
Mz OHub " OSta. 7 w 3050
Flap (blade 2) deg lbf
Lsg (blade 2) deg
" OSta. 8 lbf
Pitch (blade 2) deg Fc
" OSta. I lbf 3000
deg Ogta. 2 lbf
ffts deg
" OSts. 3 Ibf
Stc
" OSta. 4 Ibf
deg
" Sta. 5 2950
Ingow ft/sec Ibf
Pawer hp
" OSta. 6 Ibf 7 7.1 7.2 7.3
" OSta.
Thrust ibf
" OSta. 7 lbf Time (secs)
Yiisfti ft 8 lbf
Hiirdtl ft Me OSts. I ft lbf
Airspeed ft/sec
" OSts. 2 ft-lbf
Rotation Hs
" @Sta. 3 ft-lbf Figure 5: Vertical Hub Load
Advance Ratio
" OSta. 4 ft-lbf
deg OSfa. 5 ft Ibf bilities of the ADAMS-based analysis methodology
Shaft Tilt deg
" OSts. 6 R-lbf
" OSts. and, because the Sight profile is more usual, to con-
7 ft-lbf firm the basic response of the modeL From the
OSta. 8 ft-Ibf thmst snd power curves in figure 7, you can see that
the rotor system was not quite stsbilised m hover
Table I: Complete Signal List
st the beginning af the acceleration. In spite of the
steep acceleration, the pitch controBer is able to keep
teresting signals for three particular test cases, f55, the thrust deviation belaw 10% snd the mean hub
pitching moment below 70 R-lbs. With slower accel-
tgll snd 4117. These sre sn acceleration to 200 fps,
snd a spin-down and shaft tilt at 200fps. These cases erations, these deviations are reduced accordingly.
csn be though of ss joining to form a partial hypo- As expected, the power requited drops slightly at
thetical conversion. Differences between these cases moderate speeds snd then climbs again; whBe the
snd similar cases, but with differing rates or st dif- rotation rate is nearly constant at the commanded
fering sitspeeds, will be discussed. 50 radians/sec (=7.96 Hs). As the advance ratio
.
increases (st 200 fps, ft = 31), the level of hub vi-
bration also increases. Figures 5 snd 6, which show
Level Aoaderation - Figures 7 and 8 show re- the hnb response on an expanded time scale, clearly
sults from test case f55, a level acceleration from reveal that the major component is at 5/mv, ss ex-
hover to 200 fps (118.5 kts) over s six-second interval pected for this 5-bladed rotor. Higher harmonics
from 1.5 to 8.0 seconds Himulation time. Prior to the have been removed by the 100 Hs sample rate.
beginning of this interval, the rotor has rapidly gone Figure 8 concentrates on blade response. The Sap
from gat pitch to a 3000 pound thrust condition. response shows the expected steady increase with
Afterwards, the system was allowed about 5 seconds airspeed. The lsg motion, however, exhibits a "bal-
to reach s steadily periodic response, for which the looning" during the acceleration. This occurred dur-
controls snd blade response were compared to re- ing all acceleration prafile, with the effect becoming
sults for that Sight condition computed at MDHC more pronounced with acceleration rate. The thtee
using both DART snd CAMRAD JA. With small al- .
plots of section aerodynamic loads at 94II plainly
lawances fot differences in aerodynamic models, the show the effect of the thick 0012 section close to the
three codes compared very well. tip, with sevete shock st, all on the advancing side,
These results serve to highlight the overall caps leading to large increases in pitching moment snd

1034
the system numerically, which leads, in this case,
to subhsrmonic hub vibration immediately on inte-
gration stsrtup. The same numerically inducedm-
verberation csn occur afier s too-large commanded
50 change in step siss, ss happens st 14 seconds simu-
lation time and which is damped out fairly quickly
by the aerodynamics.
The thrust snd moment curves also demonstrate
a cmious bsflooning in hub vibration levels between
about 15 snd 18 seconds, foflowed by s period of
-50 very low levels before retreating blade stall begins
to dominate the response. This sho occurs in the
in-plane hub forces snd in the rolling moment (not
-100 shown). The sim of this blimp in the curves increase
with airspeed, but does not appear to be related to
7 7.1 7.2 7.3 deceleration rate. The author has no clear expla-
nation for this effect st this time, but notes that it
Time (secs) seems to parallel increases in chordwise force am-
plitudes st aerodynamic stations closer to the hub
Figure 8: Hub Pitching Moment (also not shown). Note that overall performance of
the pitch controller is very good; the deviations oC
mean thmst and moments sre very small during the
swings in chordwise force. The normal force shows a
deceleration.
much smaller reaction. As might be expected with
this section snd the — 20v twist, there is no notice-
.
Figure 10 shows the transition st 94R from the
compressibility-dominated flows associated with the
able retreating blade stall at this radial station.
thick section to the viscosity-dominated iowa asso-
ciated with an advance ratio above 0.6. Also note
Rotor ~tion - Figures 9 snd 10 show re-
sults from test csee Sill, a smooth deceleration of
that, despite the very high control system stiifness,
the extreme moment excursions due to the shocks
the rotor from 50 to 25 radians/sec (=3.98 Hs) over a are refiected in the total blade pitch seen in figure 9.
ten-second interval from 23 to 33 seconds simulation
time. The rotor stays in level Sight at 200 fps air- Shnik Tilt - Figures 11 snd 12 show the computed
speed, with 3000 pounds command thrust and sero results from test csee 4117, s smooth, aft tilting of
command hub moments. the mtor shaft from vertical (0') to horisontsl (90v)
The hub thrust snd pitch moment curves, and to over s ten-second interval from 23 to 33 seconds
s lesser extent the power curve, in figure 9 demon- simulation time. As before, the Sight path remains
strate one of the numerical pitCalls of trying to in- level st 200 fps airspeed. Rotor speed is 25 rsd/sec,
tegrate a large set of numerically stifF, differential- with 3000 pounds command thrust and sero com-
algebrsic equation like those generated by ADAMS mand hub moments. To appreciate the challenge
(see reference [9]). Typically, when exiting the code of this maneuver, note that the advance ratio drops
after a partial simulation, the integrator only saves fmm 0.6 to sero snd that the total inflow ratio sim-
the state vector snd its fimt derivative, even if s ilarly rises to -0.6, with s collective pitch reduction
higher order of integration is being used at the time. of about 40'.
When attempting to restart the simulation in s sub- This study wss originally conceived at McDonnell
sequent run, the integrator often needs to complete Douglas to determine if such s Sight maneuver would
enough steps to reach the same integration order be- be feasible in terms of overall system response, given
Core continuing, which is done by taking very small some assumed future structural and aerodynamic
steps. This csn lead, temporarily, to a poorly condi- technology level. It is important, when consider-
tioned system Jacobian snd give results with accel- ing the curves in figures 11 snd 12, to keep in mind
erations and forces slightly inconsistent with the last that a single, particular system hss been modelled,
step of the previous run. The net effect is to "ping" including the choices of rotor, engine, controls and

1035
aerodynamic representations. The data presented qualitative than quantitative. Also, it is important
include the coupled etfecto of afl these choices, mak- to note that with the shaft tilted aft snd the net in-
ing it somewhat rkdty to generalize to a larger class Sow up thmugh the rotor, the blade twist is sctuafly
of problems. reversed from that needed to equalize the radial load
First, note that what appear to be "dmpoutz" in distribution along the blade, further invalidating the
the hub thrust and pitching moment tracer are, in inflow model. In fact, the computed vibratory loads
fact, numerical artifacts caused by integrstor step st inboard blade stations sre much more severe than
size snd order changes. Series of such changes oc- st the outboard station shown here. This drastic ra-
cur twice during the simulated tilt, between approx- dial loading distribution would probably cause high
imately 45' snd 55' shaft angle and again between steady snd vibratory streszez in real, Sexible blsdez
754 snd 854, as particular parte of each blade assem- snd would be an important design consideration.
bly come close to Euler angle singularities. When
this happens, ADAMS internsfly mtstes part axes Conclusions
to avoid the singularity. Often this change in the in-
ternal representation of the model causes the choice 1. It is possible to uze sn analysis based on a
of pivots being used by ADAMS'e solver to become general-purpose multibody systems analysis (MSA)
unsuitable snd leads to a complete regeneration snd code to predict the response of standard and non-
refactorizstion of the system equations. As men- staad srd rotor eyetenw in transient Sight conditions
tioned in the description of the rotor deceleration which csn not be simulated with more conventional
csee, the lms of higher-order derivatives snd accu- rotor analysis codes. This modelling approach is ex-
mulation of numerical error can produce slightly in- ceptionsfly flexible.
consistent accelerationz snd the pinging effects seen
here in force responses. Because the displacemente 2. This type of time domain analysis produces
and velocities are etfected only indirectly, as seen in great quantities of output which are very similar to
figure 12, the averaged response should still be valid. test data, eo that validation against real, physical
The hub reactions shown in figure 11 show that tests should be straightforward. Impmved methods
the simple autopilot is, not surprisingly, incapable of sre needed for digesting this voluminous data to gen-
maintaining the commanded forces throughout the erate meaningful insights for design decisions.
tilt sequence. A 2000 pound excursion in shaft axis
force and 8000 ft-lb excursion in pitch moment would
cause significant Sight path deviations in a real air-
8. It appears possible, given advanced technology
power transmission systems snd high authority Sight
craft. Because of their very low frequency nature,
controls, to maintain positive rotor control with very
however, s human pilot, or an autopilot with more
light mtor loadingz snd kiw rotation speeds, even
sophisticated contml laws, should have no trouble
st high airepeede during demanding maneuvering.
eliminating these snomaliez. As shown in the to-
More accurate prediction of blade snd hub loads for
tal blade pitch trace, a larger than normal cyclic
Sight profiles like those presented here will require
pitch control authority is needed in addition to the
sn improved (snd validated) inflow model.
extreme coflective range, probably implying a non-
standard (i.e. not via ewszhplate) actuation mecha-
References
The flap and lsg hinge rotations presented in fig-
ure I'2 sre fairly benign, although their character fl] Efliott, A.S. and McConvifle, J.B., "Applica-
changes continuously throughout the tilt. The re. tios of a General-Purpose Mechanical Systems
maining periodic fag snd lag motions with shaft hor- Analysis Code to Rotorcrsft Dynamics Prob-
izontal sre due to the sizable gravity moment com- "
lems, Pivcecdisyo of thc 7989 ASS Nettosel
bined with the large pitch angle and unusual hinge Specialists' Meotisy on Rotorcruft Dyassitc,
sequencing. Ft. Worth, Texas, October, 1989.
The sections loads st .94R are also rather benign
throughout the tilt, but change character continu- [2] Peters, D.A. , Kim, B.S. snd Chen, H.S., "Cal-
ously. Due to the lack of detail in the induced inflow culations of Trim Settings for s Helicopter Ro-
model, however, the loads should be considered more tor by sn Optimized Automatic Controfler,
"

1036
Josrnsl of Gsidsnce, Control ssd Dynamics, [6] Bisplinghotf, R.L., Aahley, H. and Halfman,
Vol. 7, No. I, Jan. -Feb. 1984. R.L., Acroctsaticity, Addison-Wesley, Cam-
bridge, Massachusetts, 1955.
[3] Chouchsne, M. , sabaosm litic, Doctoral Disser-
tation, Georgia Institute of Technology, 1989. [I] Johnson, W. , "A Comprehensive Analytical
Model of Rotorcrsft Aerodynamics and Dynam-
[4] Leishman, J.G. snd Beddoes, T.S., "A Gener- "
ics, Part I: Analysis Development, NASA TM
slised Model (or Airfoil Unsteady Aerodynamic 81182, June 1980.
Behaviour and Dynamic Stall Using the lndicsl
Method, " Proccedinya of ttic 42nd Anssel Fo- [8] Comigsn, J.J., Schillings, J.J., Yin, S.K.
ram of tbe America Ifcticoptcr Society, Wash- snd Hsieh, P.Y., "Developments in Dynamics
"
Methodology st Bell Helicopter Textro, Pro-
ington, D.C., June, 1986.
ccedinya af the 44th Aasse/ Foram of the Amer-
[5] Torok, M. S. snd Choprs, I., "Rotor Loads
Pre-
icas Bclicoptcr Society, June 1988.
diction Utilising s Coupled Aeroelsstic Analysis
with Refined Aerodynamic Modeling,
" Journal [9] Brensn, K.E., Campbell, S.L. snd Pet
of the Amcrices Bclieopter Society, Vol. 36, No. sold, L.R., Numerica Solstice of Initial-Valse
I, January 1991. Problems in Differentia-Atycbmic Efsstiosa,
Elevier Science Publishing Co. , New York, New
York, 1989.

Run Type of Run Airspeed Shaft Angle Rotor Speed Duration


Number fps de rad sec aces
I Level Accel. 0 «150 0 50 10
'2 Level Accel. 0 «200 0 60 10
3 Level Accel. 0 «250 0 50 10
Level Accel. 0 «150 0 50 6
5 Level Accel. 0 «200 0 50 6
6 LevelAcceL 0 «250 0 50 6
7 Rotor Decel. 150 0 50 «25 6
8 Rotor DeceL 200 0 50 «25 6
9 Rotor Decel. 250 0 50 «25 6
10 Rotor Decel. 150 0 50 «25 10
11 Rotor Decel. 200 0 50 «25 10
12 Rotor Decel. 250 0 50 «25 10
13 Rotor Decal. 150 0 50 «'25 15
14 Rotor Decel. 200 0 50 «25 15
15 Rotor Decel. 250 0 50 «25 15
16 Shalt Tih 150 0 «90 25 10
17 Shaft Tilt '200 0 «90 25 10
18 Shalt Tilt 250 0«90 '25 10
19 Shaft Tilt 150 0«90 25 15
20 Shaft Tilt 200 0 «90 25 15
21 ShaR Tilt 250 0 «90 25 15

Table 2: Summary of Test Cases

1037
7.98
7.96
7.%
7.92
7.9
6
N
0

— e,
8rs
C
0
-- e1C
O 0
-2
100

-100
500

r 450

0 400
CL
350

3200
N 3'l00
r
3000
I—
2900
2800
200
0
4)
I)
aN
i

0
4 5 7 8
Tlffl6

Figure 7: Results from Level A eleratim, Case 196


1038
-100
-200
-300
z -400
-500
20
CL 10
O)
8
-10
-20
90
70
or& 50
50
10
-10
1

0.5

-0.5

0.5
0
W.5

150
rt)
n. 100
r
50
0
1 4 5 7 8
Time
Figure 8: Results from Level Accrderation, Csee gg
1039
12

8
4
0
10
— e0

0 =- e1c

-10
500
300
100
-100
-300
-500
125
75
25
-25
-75
-125
3100

3050
3000
2950
2900
8.5
7.5
6.5
5.5
4.5
3.5
13 15 19 21 23
Time

Fignre 9: Reenlts from Rotor Deceleration, Case 4 1I


1040
-100

-300

-500
15

CL

8
O

-15
90
70

10
-10
1

0.7
0.4
0.1

-0.5
3

-1
8.5
7.5
6.5
5.5
4.5
3.5
15 17 2'1 23
Time
Figure 10: Results from Rotor Deceieratiou, Case
gill
1041
20

-20

-40
15

V)
5
0
C
-5
-15

—e,
0 e
C3 -25 -- eiC

V 2000
'6.
'1000

0
i
-200
0 -400
CL
-600

N
4500

I- 3500

0
23 25 27
Time

Fiyue ll: Reeelle Sem Shalt Tilt, Ceee 1617

1042
0

-100

-200
15

10

-5
5

-1

-5
4

-1
5
4
3
2
I
0
-1
-2
90

0
23 25 27
TII718

FISme 12: Results &am Shalt Tilt, Case IS17

1043

You might also like