See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/23887899

Aeroelastic stability of rotor blades using finite element analysis

Article · September 1982

Source: NTRS

CITATIONS READS

19 408

2 authors, including:

Inderjit Chopra
University of Maryland, College Park
532 PUBLICATIONS 12,883 CITATIONS

SEE PROFILE

All content following this page was uploaded by Inderjit Chopra on 10 November 2015.

The user has requested enhancement of the downloaded file.

 i

f
"^ V
NASA CONTRACTOR REPORT 166389
f

^!« .;,v . -CR-166389)

A,ERQNLA$TIC STABILITY
BLADES USING FINITE FLEMENT ANALY
(Stanford Univ.) 119 p HC ln6/MF A01
C S (,.

I
` Aeroelastic Stability of Rotor Blades
Using Finite Element Analysis

Inderjit Chopra and Nithiam Ti Sivaneri

E

r+
NASA Cooperative Agreement NCC 2-13
August 1982

.x

NASA
^Y
Page intentionally left blank
NASA CONTRACTOR REPORT 166389

Aeroelastic Stability of Rotor Blades

Using Finite Element Analysis

Inderjit Chopra and Nithiam Ti Sivaneri

Joint Institute forAeronautics and Astronautics
Stanford University, California

Prepared for Ames Research Center

under NASA Cooperative Agreement NCC 2-13

PJASA
National AeronaWICS and
Spx(l Admins1ration
Ames Research Center
Moffett Field, California 94035
 TABLE OF CONTENTS

List of Tables . . . . . . . , . . . . . . . . . . . . . . . . v
Listof Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
Abstract. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , ix
Nomenclature. . . . . . . . . . . . . . . . . . . . . . .. . . . . x
Chapter 1 Introduction . . . . . . . . . . . . . . . . . . . . . 1
Chapter 2 Hamilton's Principe Formulation . . . . . . . . . . . . ,. . 8
2.1 Coordinate Systems , . . . . . . . . . . . . . . . . . . . . . , . . 8
2.2 Hamilton's Principle ;Formulation . . . . . . . . . . . . . . . . . . 9
2,3 Elimination of the Axial Displacement . . . .. . . . . .. . . . . 13
2.4 Nondimensionalisation . . . . . . . . . . . . . . . . . 14
Chapter 3 Aerodynamic Loads . . . . . . . . . . . . . . . . 17
3.1 Circulatory Forces . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2 Noncirculatory Forces . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 Combined Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Chapter 4 Finite Element Discretization . . . . . . . . . . . . . . . . 26
4.1 Equations of Motion in Terms of Nodal Degrees of Freedom . . . . . . 26
4.2 Boundary Coariitions . . . . . . . . . . . . . . . . . . . . . . 30
Chapter 5 Solution Procedure . . . . . . . . . . . . . . . . . . . . . . 31
5.1 'rim Solution . . . . . . . . . . . . . . . . . . . . . . . 31
5.2 Flutter Stabili(,y . . . . . . . . . . . . . . . . . . . . . . . . 32
Chapter 6 Application to Single-Load-Path blades . . . . . . . . . . . . . 35
6.1 Hingeless Rotor Blades . . . . . . . . . . . . . . . . . . . . . . . 35
6.2 Articulated Rotor Blades . . . . . . . . . . . . . . . . . . . 38
Chapter 7 Modifications for Analysis of Multiple-Load-Path Blades .. . . . .. 40
7.1 Inclusion of Axial Degree of Freedom . . . . . . . . . . . . . . . 40
7.2 Refined Finite Element . . . .. . . . . . . . . . . . . . . . . . . . 40
7.3 Displacement Compatibility Conditions at the Clevis . . . . . . . . 41
7.4 Modifications in solution Procedure . . . . . . . . . . . . . . . . 42
Chapter 8 Application to Multiple-Load-Path Blades . . . . . . . . . . 43
8.1 Twinbeam Model with Zero Inboard Pitch . . . . . . ... . 43
8.2 Twinbeam Model with Same Pitch For Inboard and Outboard . . . . 45
8.3 Bearingless Model with Torque tube . . . , . . . . . . . . . 46
v Chapter 9 General Conclusions . . . . . . . . .. . . . . . . . . .. . 47
Appendix A. Nondimensional Quantities . . ... . . . . . . . . . . . 49
Appendix B. Steady-State Element Matrices . . . . . . . . . . . . . . 50
` Appendix C. Perturbation Element Matrices . . . .. . . . . . . . . . . 59
Appendix D. Equivalent-Beam Properties . . . . . . . ....... 64
References . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Tables.. . . . . . .. . . . . . . . . . . . . . . . . . . . . . . 69
Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . 81

PRECEDING PAGE BLANK NOT FILMED

w
 LIST OF TABLES

Table 1 Values of parameters of uniform blade for numerical results.

Table 2 Steady tip deflections of a hingeless rotor blade; CT /o —

0.1, wv — 1.552, ww — 1.1512, wo — 2.5i1, 'Y = 5.0, flp =
0.05 rad.

Table 3 Fundamental coupled frequencies of a hingeless rotor blade;

= = =
CT/Q 0.1, wy = 1.551, ww 1.1552, wo 2.552, y = 5.0,
V.
Op = 0.05 rad.

Table 4 Real Parts of the flutter eigenvalues of a hingeless rotor

blade; CT Ia — 0.1, w„ = 1.552, ww = 1.15n, wo — 2.51'2,
-1 — 5.0, flp = 0.05 rad.

Table 5 Steady tip deflections of an articulated rotor blade; CT /a

0.1, hinge offset=0.06R, wo — 2.512, 7 — 5.0, #p = 0.05 rad.

Table 6 Fundamental coupled natural frequencies of an articu-

lated rotor blade; CT/v — 0.1, hinge offset=0.06R, wo
2.551, -y = 5.0, #p = 0.05 rad. {

Table 7 Real parts of the flutter eigenvalues of an articulated

blade; CT/v = 0.1, hinge offset=0.06R, wo — 2.50, y
5.0, flp = 0.05 rad,

Table 8 Structural properties of a bearingless blade.

Table 0 Steady tip deflections of a twinbeam model (zero inboard

pitch); CT/v -s 0.1, wv — 1.8752, ww = 1.1512, wo 2.9111,
ry — 5.0, /9p 0.05 rad.
i
Table 10 Fundamental Natural frequencies of a twinbeam model r
(zero inboard pitch); CT /Q = 0.1, w„ = 1.8711, ww =
1.1511 1 wo = 2.9111, ry = 5.0, #p = 0.05 rad.
h
{
y

Table 11 Real parts of the flutter eigenvalues of a twinbeam model

(zero inboard pitch); CT /o = 0.1, w„ = 1.8711, ww =
4

=
1.1511, wo 2.9152, ry — 5.0, 9p = 0.05 rad.

Table 12 Fundamental natural frequencies at various thrust levels

(zero inborad pitch).

v
1
 LIST OF FIGURES

Fig. 1 Schematic of an articulated rotor hub and root.

Fig. 2 Bearingless Main Rotor (BMR)

Fig. 3 Flexure of BMR

Fig. 4 Blade coordinate systems and deflections

Fig. 5 Deformed-blade-section aerodnamics

Fig. 6 A finite element showing nodal degrees of freedom

Fig. 7 (a) Flap bending (b) lead-lag bending (c) torsion. Sta
tip deflections of a stiff- inplane hingeless rotor blade; w„
=
1.512, w w 1.15f2, wo = 5.0f2, y — 5.0, flp — 0.05 rad.

Fig. 8 (a) Flap bending (b) lead-lag bending (c) torsion. Steady
tip deflections of a soft-inplane hingeless rotor blade; w„
=
0.71l, ww — 1.15f2, wo — 5.00, y 5.0, flp — 0.05 rad.

Fig. 9 Root locus plot for a hingeless rotor blade; w„ — 1.513,

=
ww = 1.15f3, wo 2.5f2, y = 5.0, #p — 3.05 rad.

Fig. 10 Stability boundaries for a hingeless rotor blade; ww =

1.150, wo = 2.5f2, y = 5,,0,
,Bp — 0.05 rad.

Fig. 11 Comparison of hingelesss rotor blade results with experimen-

tal results.

Fig. 12 a) Flap bending, b) lead-lag bending c) torsion. Steady x

tip deflections of an articulated rotor blade; hinge offset=0.06R,
wo = 2.50 1 7 5.0, flp = 0.05 rad. i
a

i
Fig. 13 Root locus plot for an articulated rotor blade; hinge ofl'set=0.06R,
wo = 2.5f2, y 5.0, #p = 0.05 rad.

Fie. 14 Analytical model of a bearinLr less blade.

vi _
S
 Fig. 15 Refined finite element for multiple-load-path blades.

Fig. 16 Pitch distribution of a bearingless rotor blade.

Fig. 17 Twinbeam multiple-load-path blade model.

"z !
t
{

Fig. 18 Root locus plot of lead-lag mode of stiif4nplane, soft-in-

torsion twinbeam model (zero inboard pitch); w„ = 1.87n,
=
ww 1.150, wo = 2.910, y = 5.0, flp = 0.05 rad.

Fig. 19 Root locus plot of lead-lag mode of stiff inplane, soft-

in- torsion twinbeam model (zero inboad pitch; coupling
structural stiffness terms of nonlinear origin suppressed
in flutter analysis); wv — 1.87n, ww = 1.151, wo — 2.91(1,
ry — 5.0, #p = 0.05 rod.

Fig. 20 Root locus plot of lead-lag mode of soft4nplane, soft-in-

torsion twinbeam model (zero inboard pitch); w„ = 0.90n,
WW = 1.150, wo — 2.910, -y — 5.0 ,Bp — 0.05 rad.
1

Fig. 21 Root locus plot of lead-lag mode of soft-inplane, soft-

in-torsion twinbeam model (zero inboard pitch; coupling
structural stiffness terms of nonlinear origin suppressed
in flutter analysis); w„ 0.900, ww — 1.1512, wo — 2.910,
ry — 5.0, Op = 0.05 rad.

Fig. 22 Root locus plot of lead-lag mode of stiff-inplane, stiff-in-

torsion twinbeam model (zero inboard pitch); w,, — 1.870,
WW = 1.150, wo = 5.380, 'Y = 5.0, flp = 0.05 rad.
a

Fig. 23 Root locus plot of lead-lag mode of stiff=inplane, stiff-

in-torsion twinbeam model (zero inboard pitch; coupling
structural stiffness terms of nonlinear origin suppressed
in flutter analysis); w„ = 1.870, ww = 1.150, wo = 5.380,
ry = 5.0, flp 0.05 rad.

Fig. 2¢ Root locus plot of lead-lag mode of stiff-inplane, soft-

in-torsion twinbeam model (same pitch for inboard and
outboard); w„ — 1.870, ww — 1.150, wo — 2.910, 7 — 5.01
,Bp == 0.05 rad.
 Fig. 25 Root locus plot of lead-lag mode of soft-inplanc, soft-
in-torsion twinbeam model (same pitch for inboard and
outboard); wv - 0.0052, ww = 1.15f2, w# — 2.91n, '1i — 5.0,
Op : 0.05 rad.

Fig. 20 Root locus plot of lead-lag ;node of stif -inplane, stiff-

in-torsion twinbeam model (same pitch for inboard and
outboard); w„ — 1.870, ww —1,15(2, wo s 5,3851, ry — 5.01
Op — 0.05 rad, 4

Fig. 27 Root to us blot of a bearingless blade with two identical

flexbeams and a torque tube; w„ — 1.87(1, ww = 1.150,
wo = 3.1012, ry — 5.0, Op — 0.05 rad.

i
1

H i

}
 (:1

A 3STRA.CT

The flutter stability of flap bending, lead-lag bending, and torsion of helicopter
rotor blades in hover is investigated using a finite element formulation based on
Hamilton's principle. The blade is divided into a number of finite elements. Quasi-
V steady strip theory is used to evalua lw the aerodynamic leads. The nonlinear
equations of motion are solved for steady,.state blade deflections through an iterative
procedure. The equations of motion are linearized assuming blade motion to be a
small perturbation about the steady deflected shape. The normal-mode method
based on the, coupled rotating natural modes is used to reduce the number of
equations in the flutter eigenanalysis. First the formulation is applied to single-
load-path blades, for example, articulated and hingeless blades. Numerical results
show very good agreement with existing results obtained using a modal approach.
The second part of the application concerns multiple-load-path blades, namely
bearingless blades. The flexure of a bearingless blade consists of multiple beams
(flexbeams and torque tube) leading to redundancy. The formulation is modified so
that the multibeams of the flexure could be modelled individually. Numerical results
are presented for several analytical models of the bearingless blade. Results are
also obtained using an equivalent beam appraoch (the common approach) wherein
a bearingless blade is modelled as a single beam with equivalent properties. The

Y comparison between the two sets of results show that the equivalent beam modelling
is inaccurate.

i
 NOMENCLATUR

a lift-curve slope
4 A blade cross-section area

Bt, B2 blade cross-section integrals

C blade chord

I CL blade-section lift coefficient

CMac blade-section pitching moment coefficient

CT rotor thrust coefficient

C I, C'2 blade cross-section integrals

ea tension-center offset from elastic axis, positive forward

ed aerodynamic-center offset from elastic axis, positive aft

ey center-of-mass offset from elastic axis, positive forward

E Young's modulus

F centrifugal force

GJ blade elastic torsion stiffness i

Iy, Iz blade cross-section moments of inertia in the flap and lead -lag direc-
tions, respectively l

kA polar radius of gyration of blade cross section, Iy + Iz TA

km mass radius of gyratioa of blade cross section, (k., + km2)

km 1, km2 principal mass radii of gyration of blade cross section

l length of ith element

Lu, Lv, Lw aerodynamic forces per unit length in u, v, w directions, respectively

m mass per unit length of blade

6

M aerodynamic moment per unit length about elastic axis

G
z

u_
 01

n number of elements

{q} vector of global degrees of freedom

R rotor radius

f time

u, v, w elastic displacements in x, y, x directions, respectively

Up, UT blade cross-section air velocity components in negative n and s direc-

tions, respectively I#

Vi induced inflow

V blade cross-section resultb nt air velocity

X) y, r, undeformed blade coordinates

local axial coordinate of the ith element_

a blade cross-section angle of attack

OP blade precone angle

7 Lock number, 3pacR/m

bT, bU variation of kinetic and strain energies, respectively

bW virtual work done due to aerodynamic loads

8 blade prepitch

X rotor inflow ratio, VilOR

f, deformed-blade coordinates r t

F
solidity ratio

elastic twist about elastic axis

geometric twist about elastic axis

dimensionless time, fit

WV , WW' WO fundamental coupled rotating lead-lag, flap, torsion natural requencies,

E
respectively
f
xf
n rotor blade angular

(Y aOlax
() a(^lao

t>pa

xii
1

1. INTRODUCTION

Helicopter dynamics is a complex problem involving nonlinear structural, in-

ertial, and aerodynamic forces, For efficient blade designs, there is an increasing
trend toward nonuniformity in structural and aerodynamic properties, The root
N.
geometry is complex.

Most of the helicopters in service employ either an articulated or a teetering

rotor. In recent years, there has been considerable interest in alternate designs with
an eye toward mechanical simplicity and increased maintainability. The hingeless
and bearingless rotors are the outgrowth of such efforts. A conventional articulated
rotor blade has mechanical flap and lag hinges, a lag damper and a pitch bearing,
Figure 1 shows the schematic of an articulated blade hub and root. A hingeless
blade, as the name implies, lacks the hinges and is cantilevered at the hub, It does
have a pitch bearing for pitch control. A bearingless design eliminates the hinges
and the pitch bearing as well. Only some hingeless and bearingless designs include
lag dampers. The bearingless blade has an elastic flexure consisting of flexbeams
and a torque tube to facilitate pitch changes. An example (Fig. 2) that has been I
1
tested in flight and wind dunnel is the Boeing- Vertol designed Bearingless Main
Rotor, BMR (Dixon and Bishop (1080); and Warmbrodt and McCloud (1981)]. The
BMR flexure (Fig, 3) extends from the hub to about 25%blade radius where it
is connected to the outboard blade by a rigid clevis. The flexure consists of two I

parallel, torsionally-soft, open-section beams and a torsionally-stiff (flexurally-soft)

torque tube. The inboard end of the torque tube is connected to a pitch link via q
a pitch horn at the hub; and the outboard end is attached to the clevis. The a
pitch control to the blade is applied through the torque tube by rotating the blade-
root clevis which elastically twists the flexbeams. The 'blade outboard of clevis
is of airfoil section and is similar to a conventional blade.: The articulated and

L x^
i
hingeless blades are referred to as single-load-path blades and the bearingless blade
as multiple-load-path blade, This is because the blade loads can be transmitted to
the hub by redundant paths of the flexure in the case of a bearingless blade.

To help in the design of helicopter rotor blades, it is necessary to solve for trim
deflections and to determine the aeroelastic stability in hover and forward flight.
The aeroelastic stability of articulated blades is well understood, 1n the case of
hingeless and bearingless blades, the absence of hinge motion modifies the dynamic
R
characteristics of the blade and may significantly influence the aeroelastic stability.
Not enough is known about the parameters affecting the aeroelastic behavior of
bearingless blades. The influence of multilple load paths on aeroelastic stability is
not fully understood,

The simplest form of roptor blade representation is the rigid blade _model with
spring restrained hinges (Ormiston and Hodges ( 1972) and Kaza and Kvaternik
(1076)J. For this model, the trim solution is generally obtained from coupled linear
or nonlinear flap-lag equations and the Butter stability is determined assuming linear
perturbation motion about the trim solution. This type of modelling is satisfac-
tory for simple blade configurations such as articulated. A more appropriate repre-
sentation is to treat the rotor blade as an elastic beam. Houbolt and Brooms (.1958)
have derived the differential equations of motion for the combined bending and
torsion of an elastic blade. Several authors (Hodges and Dowell (1974), Kvaternik
and Kaza (1976), and Rosen and Friedmann (1078)j have modified. these equations
to consistently include nonlinear structural and inertial terms fnr moderately large
deflections. A number of researchers have mialysed the aeroelastic stability of hinge-
less blades with elastic-beam modelling. Examples are Hodges and Ormiston (1676)
and Johnson. (1977). The common approach for calculating the trim deflections,
as well as the aeroelastic stuff .ii ^; is the modal method (for example, the Galerkin
method) using coupled natural modes. In general, the trim deflections are assumed
I

4_
 to be large and obtained by solving nonlinear steady-state equations; and the flutter
equations of motion are linearized about the trim state. This linearization simplifies
the stability analysis. There have been some attempts at determining the stability
w
using a complete nonlinear analysis such as the limit cycle analysis. These have
been for simple blade models. For example, Friedmann and Tong (1973) consider a
flap-lag blade; and the analysis of Chopra and Dugundji (1979) inclVes the torsion
degree of fraedom but the blade is modelled as rigid with spring restrained hinges.

The theoretical analysis of a bearie less blade is more involved than that of a
hingeless blade primarily due to the redundancy of the flexure. Several authors have
examined the aeroelastic stabililty of bearingless blades. Bielawa (1976) has given a
theoretical analysis for composite bearingless rotor blades and experimental results
from dynamically scaler' models. The formulation is made for a blade with one
flexbeam and a torque tube. Hodges (1979) has analyzed bearingless configurations
with single flexbeam. The bending stiffness of the torque tube is neglected. The
outboard blade is treated as rigid. Hodges has used an equivalent-beam approach
to model the twinbeam root flexure of the BMR blade. Harvey (1976) has modelled
a single-flexbeam bearingless blade by lumped massed connected by elastic beam
elements. If any of the analyses mentiond above are used to study the BMR blade,
then the twinbeams of the the flexure has to be modelled as single flexbeam with
equivalent properties. The equivalent properties can be obtained by matching the
fundamental frequencies of the equivalent-beam model with that of the multibeam
blade.

With the modal approach, it becomes increasingly difficult to handle geometric

coomplexities. For example, it is difficult to electively model the multibeam flexure
of a bearingless blade. The finite element method [Gallagher (1975)] has been
extensively used in linear structural analysis. The structure is divided into a number
of finite elements and the application of energy principles, or the method of weighted

_g_
0

residuals, yields approximate expressions for forces (inertial, elastic, etc.) over
each element, The global equations of motion are obtained by assembling the
elements. Nonuniform properties can be easily accommodated. The finite element
method is very flexible and the formulation can be adapted to different rotor blade
configurations with a few modifications. Multibeams of a bearingless blade can be
I
modelled individually.

The finite element method has been applied for the determiniation of free
vibration characteristics of rotating one dimensional (flap bending) beams by many
authors.. A few examples are Nagaraj and Shanthakumar (1975), Putter and Manor
(1978), and Hodges and Rutkowski (1981). It is common practice to use simple beam
elements (represented by end nodes) satisfying the continuity of displacement and
slope between elements. Refined elements with internal nodes may be used instead.
One can attain a desired accuracy by using either simple or refined beam elements,
but, if simple elements are used, a greater number of them may be necessary.
However, the assembly of simple elements results in narrowly banded matrices; the
assembled matrices for refined elemnts are more populated. Murty and Raman
(1980) have used the finite element method for computing the nonlinear response
of a rotating beam under prescribed forces.

Yasue (1978) has obtained the natural-vibration characteristics of a hingeless

blade in combined flap bending, lag bending, and torsion using a finite element
formulation.. The normal mode method based on these natural modes is utilized to i
a
calculate the trim solution and the response of the blade under gust loading. Yasue's
finite elements exhibit the interelement continuity of the torsional deflection as well
as its slope; and hence a blade with jumps in torsional rigidity cannot be handled.
Friedmann and Straub (1980) and Straub and Friedmann and Straub (1981) have
applied a finite element formulation based on the method of weighted residuals to
9

calculate the free-vibration characteristics of a flap-lag hingeless blade. Then, the

_,t_
r

trim state, flutter stability, and response of the blade in hover and forward flight
have been computed using a modal approach. These three rdeferences use the finite
element method just to obtain the natural modes; the trim solution is based on a
modal approach.

In the present research, a finite element formulation is developed to investigate

Y

the flutter stability of flap bending, lead-lag bending, and torsion of rotor blades in
hover. The formulation is made for a nonuniform blade with pretwist and precone,
and having chordwise offsets of the center of mass, aerodynamic center, and tension
center from the elastic axis. The deflections considered a,re the axial, lead-lag, and
flap deflections of the elastic axis and a torsional deflection about the elastic axis.
The spatial dependence of the deflections is made discrete by dividing the blade
into a number of elements. Approximate expressions for element forces are obtained
applying Hamilton's principle; and the assembly of elements results in the global
equations of motion in terms of the nodal degrees of freedom. The aerodynamic
loads are based on quasi-steady strip theory. The trim solution is computed from
nonlinear steady state equations without making a modal transformation. The
coupled rotating natural-vibration characteristics are calculated about the trim
condition. The flutter motion is assumed to be a small perturbation about the
trim state. The normal mode method based on the rotating natural modes is used
to solve the linearized flutter equations as an eigenvalue problem. Flutter stability
is inferred by studying the flutter eigenvalues.

n Here it should be pointed out that the emphasis of the present study is on the
feasibility of the application of the finite-element method to rotor blade aeroelas-
ticity; and hence, importance is not given to certain aerodynamic aspects such as
compressibility effects and the effects of preceding and returning wakes. A brief
review of the research in these areas follows. It is popular to use a quasisteady
strip theory approximation without wake effects for estimating the unstemdy loads

- 5-
 a

on a rotor, one of the main reasons being the difficulties involved in accuractely
representing the Theodorsen function c(x). The difficulties arise because, unlike
that of fixed wings, the wake of one rotor blade interacts with that of the others.
The first significant attempts at unsteady aerodynamic theories for rotary wings in
incompressible flow are due to Loewy (1957), Timman and Van De Vooren (1957), 4

and Jones (1958). A theory to study the compressiblity effects including wake is .^ I

presented by Jones and Rao (1970); aerodynamic coefficients for certain Mach num-
bers and frequencies are given in this paper. In another study, on the effects of tip
vortices on rotor blades in hovering flight, Jones and Rao (1971) colclude that the
use of two-dimensional strip theory would not result in serious error provided the
blades oscillate at several cycles per rotation. Murthy and Pierce (1976) formulate a
technique for the prediction of flutter of a helicopter rotor in hover and axial flight
in such a manner that various unsteady aerodynamic theories and various types of
coupling between the degrees of freedom could be accommodated. Kato and Yamane
calculate the rotor impedence of articulated-rotor helicopters in hover (1979a) and
forward flight (1979b) using a quasisteady approximation that neglects the effects of
preceding and returning wakes. In trying to correlate their theoretical results with
experiments, Kato and Yamane (1981) find that the quasisteady theory does not
predict the amplitude peaks and clefts which appear near the multiples of the blade
passage frequency bfl (where b is the number of blades) whereas an analysis with
Loewy's function agrees fairly well with the experiment. Various dynamic inflow
models are included in a flap-lag stability analysis of a rigid blade in forward flight
by Peters and Gaonkar (1980). Their study concludes that unsteady perturbations
in the induced flow have a significant effect on both the flap and lead-lag damping,
particularly for the regressing mode.

The application of the present method consists of two parts. First, numerical
results are obtained for uniform single-load-path blades. Results are compared

—6 —
with that of a previous research based on a modal method. Comparisons with
experimental results are also made. The second part of the application concerns
multiple-load-path blades. The formulation made for single-load-path blades is
modified here to include the multibeams of the root flexure of bearingless blades.
Each of the flexbeams and the torque tube are modelled as individual beams. The
displacement compatibility conditions, at the clevis, between the inboard beams and
the outboard beam are used in the modelling of the bearingless blade. Refined finite
elements with internal nodes are used, Numerical results are obtained for several
analytical models of the bearingless blade in hover. To determine the accuracy of
the conventional equivalent-beam approaches, results are obtained for a blade with
equivalent properties, Comparisons are made between the results of the present and
equivalent-beam models. i

This is the first attempt to solve the nonlinear trim equations of a flap-lag-
torsion rotor blade using the finite element analysis directly. This approach enables
the coupled natural rotaing modes to be calculated about the steady deflected
positon of the blade. This is the first research to apply the finite element method
to bearingless blades; and the first one to model t^° ^^^^^^^^ ^^^° ^^ *k-
individually.
t
-707
a

x i

2. HAMILTON'S PRINCIPLE FORMULATION

2.1 Coordinate Systems

The coordinate systems used are shown in Fig. 4. The rotor blade is treated as
an elastic beam rotating at constant angular velocity n. be rectangular coordinate
system z, y, z is attached to the undeformed blade which is at a precone angle of Pp.
The origin is at the root of the blade, the x axis coincides with the elastic axis, and t

the y axis is in the plane of rotation pointed towards the leading edge. The deformed
position of the'Aade is defined as follows. A point P on the undeformed elastic axis
undergoes displreements u, v, w in the x, y, z directions, respectively, and occupies
the position P1 on the deformed elastic axis; u is the axial deflection, v the lead-lag
deflection, and w the flap deflection. Then the blade cross section containing P1
undergoes a rotation 9 1 about the deformed elastic axis. The orientation of the
deformed-beam cross section with respect to the undeformed-beam cross section is
described by a sequence of three rotations. This analysis uses the lag-flap-pitch
sequence of rotations defined in Hodges, Ormiston and Peters (1980). The thrid
angle in the sequence., $I , is written as

ei =e+ (2.1)

3
and ' to second order is

t
x
frw' da (2.2)
J0 v

where 8 is the pretwist, ^ is the geometric twist, and 0 is the elastic twist due to
torsion. Th_e orthogonal coordinate system, q, s is attached to the deforemed blade

—8—
 ^j

such that the F axis is tangential to the deflected elastic axis and the rl and axes
are the principal axes of the cross section. The transformation between the two
coordinate systems is given as

f^ r

(T) ,y (2.3)

where t , Iq l IV and iz j iy, a', are the unit vector systems in the two coordinate frames, k
F
R
The transformation matrix is defined in Hodges et al (1080 ) and is reproduced ir.

a ppendix A. ^'

2.2 Hamilton's Principle

The formulation is based on Hamilton's principle:

A
t2
1 (au — aT — awl d^ o (2.4)
f

f
where 6U, 6T, and bW are, respectively, the variation of strain energy, the variation
[' of kinetic energy, and the virtual work done by external forces. Hodges and Dowell
4

` (1974)
. give expressions for W, and bT. The expression for the variation of the strain z

energy is

_9_^ V
 1 a

R OF POOR QUALITY
bU = f (F` bu' + v'bv' + wlbwl)
` ` n
r
+ I GJO' + EAkq(6 + ^Y(u' + v 12 + 2 w1"^
2

++ EB16 i — EB261(v N cos 6 1 + wn sin 81)J6¢^

+ [EC10N+ EC2(w"cos 81 —vNsin61)jbQN

+ ff EIz (v ►1 cos 6 1 + w ►► sin S1 — EAca(u l + 12 + W ,2 — EB2 0101 1 cos 01 M
^l \ l \ 2v 2 J J
+ [EIV V II sin 8 1 — wN cos 6 1 )—EC2 0N1 sin B 1 (6v N + w116¢)
(

+I Elz w 11 cos 9 1 + war sin 0 1 ^ — EAea u ? + 1+ v12 2 wp — EB264 01 sin 81

J
N
— [Eiv (V sin 0 1 — wN cos B1)—EC20 J cos 91)(bwN — 04)) dx (2.5)

where the axial force F is defined by

F = EAI+.j + 'v'2 + + wr
v kg6r 0' — ea(vN cos 61 +w" sin 61), (2.8)
2

and the section constvmts are

 ORIGINAL PAG
OF POOR QUA

A = f fA dq ds

Aea s f fA q dq ds

Iy : r 12 drl
J fA ds
2
iz M f fA q dry dS

Akj = f fA (11 2 + dq ds
s2)
111Y r

!
fA [ 11— a
j f
( s +\s+an^

Bl
m f fA(V2 +f 2) dq ds

f f+2 + s2) dq
B2 do

CI M a2 dgds
f fA
C2 M f fA sa dq ds

The blade cross-section area effective in carrying tension is A. IV and I are,

respectively, the Aapwise and chordwise area moments of inertia of the cross section;
Ak, is the polar moment of inertia of the cross section; J is the torsional constant
i
including cross-section warping; ea is the offset of the tension axis from the elastic
axis, positive forward; Bt , B2i C1 , and C2 are other section constants; X is the warp
a
function which is assumed to be antisymmetric about the chord. The cross section M

is assumed to be symmetric about the chord.

The variation of the kinetic energy for the elastic beam is

11 —

u
 ORIGINA.1L P AG E 19
OF POOR QUALI TY
R
bT (jj2X + 2W 4)6u
(M

+ [M(12(v + ey cos 0 1 ) + 2mn (ppty — u)

+ 2megrl 61 Cos 0 1 + u? sin 0 1 ) — mii + meg sin 01 J6v

— [Mpp(jj2X + 2(16) + mw + mega cos 0 1 bw

1

— [mk 2 " + m02(k2 r Cos 01 sin 01

k2m1!
+ Meg 0 2 x(w f cos 0 1 — v 1 sin 0 1 ) + m egn2V Sin 01
+ Megn2 ppx Cos 91— me 9 (ii sin 9 1 — ti cos 01)J b^

— Meg 0 2 Zcos 0 1 + 2W cos 0 1 )6V'

— Meg 02 x sin 01 + 2N sin 01)b W') dx (2.8)

Where tine section integrals are defined by

M Z= f fA' p dq d5
Meg 39 f fA p n dry dS

mk,2„ 1 M f fA 05 2 do d S (2.9)

mkt Mf fA 092 do ds
k2,„ =kM1+ km2

> I

The blade mass per unit length is m; e9 is the center of mass offset from the
elastic axis, positive forward; and km 1 and k,,,2 are the flapwise and chordwise mass
moments of inertia per unit length.

The virtual work bW of the external forces may be expressed as

— 12 —

f
 6W - (Lu6u + L„6v+ Lw6w + M¢60) dz (2.10)
ZOR

where Lu, Lv, Lw, and lt10 are the external loads distributed along the length of the
blade in the axial, lead-lag, flap, and torsion directions respectively and the virtual
rotation 60 as defined in Hodges et al (0, 80) is

60 P_-^ 6¢ + w16VI(2,11)

Since 6U, 6T, and 6W, as given in Eqs. (2.5), (2.8), and (2.10), are independent of
the time derivatives of 6u, bv, 6w, and 6 ¢, Eq. (2,4) can be written as

0=6U-6T-6W =0 (2.12)

2.3 Elimination of the Axial Displacement

For single load path blades, the axial displacement u can be eliminated in terms
of the other deflections v, w, and 0 and the centrifugal force F. This is possible
only for single-load-path blades since the force equilibrium equation in the axial
direction can be solved a priori,

3
Equation (2.12) would result in four equations of motion representing 6u, 6v,
6w; and 6o, respectively. The 6u equation is
 ORIGINAL PAGE 13
Integrating by parts and noting that OF POOR QUALITY

6u=0 at z-0

and

0
Fe0 at x—_R,

the above equation can be written as

R
(F'+Mjj2X+2m(16+Lu)6udx'M0 (2.14)
f0

Since this equation is true for arbitrary values of 6u,

F1 + mf2 2 x + 2mf16 + Lu — o (2.15)

The external loads are of aerodynamic origin and the radial drag force L„ is negli-
gible in hover. So neglecting Lu and integrating Eq. (2.15) results in

F(a) — fzR +2 x + 202 ) dx (2.18)

The displacement u is eliminated from 6U, 6T, and 6W using Eqs. (2.6) and (2.16).

2 .4 Non dimensionalisatio q

Alter eliminating u, the 6U, 6T, and 6W are nondimensionalised by dividing the

f
expressions by m 0 fl 2R3 , where mo is _a reference mass per unit length (for example,
 O RIGINAL PAGE 13
OF POOR QUALITY

the value of rn at 4,54 This is done to see the i

each term. The nondimensional displacements v/I

be of order t where a is a small nondimensional quantity such that e" « 1. The
other nondimensional quanities and their assumed orders of magnitude are given in
Appendix B. The lowest order (i.e, first order) terms in 6U and 6T are of order E2,
Terms of order e2 and 0 are retained while neglecting all terms of order e4 (i.e., third
order) and higher, excepting some linear third order terms which are important for
the torsion equation. The following simplified nondimensional expressions for 6U,
and 6T, and 6W result. The symbols are retained after nondimensionalisation.

6U _ ft w► 6v► + Ow►)
(F
m0 f U

+ G#6^ ► + ^'Ov" + ^► v ►► 6w ► + VV601)

+ [Fkq(8► + ^►►) — ,,Akg9►r O► + EAeakg@

► (v►► co s 0 1 + w" s n Al)

►2
+ EBl 0 ► — EB20(v ►► cos 01 + w ►► sin 91) 6 ►
1
+ LECI ^" + EC2(w" cos 91 v" sin 01)J6^"

+ Pt. cos 0 1 + (EIz cos t (B + j) + Ely sin g (B ,+, j)j V ►

+ 2( EIz—EIy)w ►► sin2B1 EAea! v ►► cos`B1 + 2w" sin 211^

(.pAe,,k2 — E82)B► ► cos$1 —EC2^"sinh]60

+ f —Fea sin 61 + EIz sine (B + ¢) + EIy cost (e + j)jw►►

l ^ ^
2 (EIz — Ety)v" sin 291 — EAea^w►► sin"'$,+!" sin 281
1r

(EAeakA2 -- E62)8►0 sin B l —E C2 * of cos P116w►►

+ [Fea (V II sin $l — w ►► cos B1 f

+ (EIz .- EIy)i 2 w"` — v►►` ) sin20 j +v11w#f cos20j 6^ dx (2.17)

—1s-
A

ORIGINAL PAGE 13
OF POOR QUALITY

^,^R^
06 2 ,^ ••^
Jot
m
{Cu + co cos S1 + 2flp w + 2^ (uv^ + w"'1 ^) dx
+ 2co(O' cos 01 + iv y An 0 1 ) -- ii + cf^ sin 1116u

[^p(x + 26) + w + io^ cos 01]6w

[ k2M ^ + 2 ( k2 2 — k2 1 ) sin 201 + cox(w1 cos 01 — vo sin 0 1)
+ cou sin 01 +cg#px Cos / l --co(u sin 01 -ib Cos
01)14

— 9 0 (x cos 01 + 26 cos 01)6v?

— to(x sin 01 +V sin / 1)6W,)dx (2.18►

and

F
6W 1
(L„ 6v + Lw6w + M^60) dx (2.19)
2s
n'o^ R
The singly underlined terms are nonlinear and the doubly underlined terms are
third order (E4) linear torsion terms.
 3. AERODYNAMIC LOADS

The exter al loads L,,, Lw, and M4, of Bq. (2,19) are of aerodynamic origin,
The expressions for these aerodynamic loads in hover, distributed along the length
of the blade, are obtained based on a quasisteady strip-theory approximation. The
flow is assumed to be incompressible and inviscid, The induced inflow is assumed
to be uniform and steady. Virtual inertia effects are included. The variation in free
stream velocity due to flap, lead-lag, and torsion motions are taken into account.
Blade angle-of-attack stall is neglected.

3.1 Circulatory Forces

The forces L, and L,u are in the y and z directions respectively (i.e., in the
undeformed coordinate frame). In the present analysis the strip theory is based on
sections in the deformed frame. As a result, forces obtained in the deformed frame
have to be transformed to the undeformed frame. The magnitude and direction
of the relative wind over the blade section is obtained by vectorially adding the
negative of the blade velocity to the induced inflow.:
1

V' — — Vb+V (3.I)

where Y is the relative wind, Yb the blade velocity, and V; the inflow velicity. Hodges
and Dowell (1970 present the expression for Vb in terms of the unit vectors in the
undeformed frame.

Vb a (ii — fly( Cos #P )1z ♦ wl + (zl Cos #P = zl sin #P)j,Y

+ (l + flyl sin#P ) 1 t (3.2)

t,
— 17 —
h

ORIGINAL PAGE IS
with OF POOR QUALITY

X1 x+u X ^'— V(rl cost j — S sin/1)

Y1 v + n cos 01 — i! ain Al
«1 = m + 9 sin S l + s COS 01

The following approximations are made.

(a) 1^ is obtained at a reference point ( tlr,0) o

(6) $in,$p r ,Bp; cos Op s:W 1.

(c) Terrnq of ordfr d and higher are neglecter..

Then Eq. ( 3.1) is written as

V=Uz z+Ug y +uzIZ (14)

where

Uz — --u + 6 1 flr cos 0 1 + lAr ain 01 + 11( v + h r cos 0 1) v i#p

Uy — --6 + ^ q r ain 0 1 — O(x + u — v 'nr cos Bl — Wi er sin $ 1 ) + f1 flp ( w + Fir sin 01) (3,5)
A
uz — —tu + O q r cos 0 1 n#p(v + fir CUs e1) - vi

"We are interested in expressing V in the deformed frame since the airfoil section
v., considered in that frame. So if is written as

V - UR: f — UT Iq Upis (3.6)

Then
k
-18--

rw
 ORIGINAL l
OF POOR QUALITY

—CIRUX

— UZ, — (Tj Uy

—UP) l uz i

The axial velocity UR is negligible in hover; the expressions for UT and Ut

UT — flxK 1 — j r f cos 01 v?wI sin 81

2v
+ O [u cos 0 I + (vv' — rpp) cos $i + (w' + pp)v sin 01
+u Cos $ I +(iu +v,) sin 01 (3.8)

and

r n.
Up —rlxl (1 — 2 v^r ^ sin $I + v owt Cos Bl^
+ tlPu sin 9i — (vv' — w#p)sin 0 j + (w , + Q11, + a cos 11)]
—u sin $1+(w+ Vi) Cos 0l+q,^ (3.9)

According to the quasisteady approximation of the thin airfoil theory, the

circulatory forces are based on the angle of attack at the three-quarter-chord point
(e.g., see Fung (1969) pp. 1931. Therefore UT and Up are evaluated at the three
quarter chord point; that is, n, is the n coordinate of the three-quarter-chord point.

Figure 5 shows the aerodynamic environment in the deformed frame. Tile

circulatory forces Z„,, Zw,, and ht^, in the n, s, and ^ directions, respectively, can
be written as

_19_
 ORIGINAL. PAM- I'- {
OF POOR QUALITY

sin a - D cos a
Zuc — L
4C — — Leosa — Dsina (3.10)
Moe " Mae -` LW'cd

where L, D, and Mae are, respectively, the lift, drag and pitching moment (about the
aerodynamic center) per unit length, a is the angle of attack at the three-quarter-
chord point, and ed is the aerodynamic center offset from the elastic axis. The
expressions for L, D, and Mae are

L-CLjpV2c

D _ CD pV'e (3.11)
2
Y
Mae CMae
2 pV

with

v'-' -• Uf UP
a

and

a
tan oUP
UT

where CL, CD and CMae are the section 'lift, drag, and pitching moment coefficients,
and c is the section chord. Substituting Eqs. (3.11) in Eqs. (3,10) yields 41

-20`
!M
F

t
ORIGINAL PAQ'
OF POOR QUAI

Eve ° 2 Pe(CLUPV — CD
Tnuc — 2Pe(CLUT V +GDvpv l
3

M 0 2PC2CM4ev2 — ed Lwc

With the assumptions

t

CL= co +cla
CD — do+dia+d2a'
a9Wein a (3.13)
VMWUT
Yp

Eqs. (3,12) can be written as

Zoo 2 pckdoUll + (co — d1)UT Up + (c, — d2)UP]

Zeuo ° —2pe[`oUT +( el + do)UTUp + diU^ (3.14)

J
1 n n n
MOc '^ 2 Pc.cmac(4 + 4) — tdT,wc

T
The assumed order of magnitude is unity for cl and d2 ; E for co, d i , Cm.,, and a;
10P for ed; and e2 for do. The circulatory forces Lo, and L, ,1e in the undefo;
frame are obtained from Zoc and Lw e using the transformation matrix.

Luc p

L„ c (TIT hoc

Lwc 11we
i'

ORIGINAL PAGE IS
OF POOH QUALITY
Thus, the circulatory aerodynamic force components to order E2 are

Ly c = pc cos 91 doil2x2 cost B1 + (co — d l )(-2 n2X2 sin 29 1 + t1xv; cos 201
2 ( I—
+ (c1 d2+(1x sin 9 1 — v; cos 91)2 — 2n2 xnrgp sin 91^^

+ sin 9 2 [C 0 (fj 2 x 2 cost 9 1 + nxvi sin 291) 41

+ ( c l + do )I nx cos 9 1 (—nx sin 9 1 + v; coa 0 1 + nt^r ,Bp) — nxv sine 91 + v sin 2011

+d1(n 2 x 2 sine 91 — nxv; sin 29 1 ) }

1
— un2 x ( do + d2 ) sin 291 sin 01
— vn 2 xflp((c1 — d2) sin 20 1 cos 0 1 + (e1 + do) sin 0 1 Cos 2011
+ w02x#p(do + d2) A in 291 sin 91
+ w 1 tl 2 X11r(—c l + do + 2d2) sin 91 cos 01
i
+6 (Cos 911 -2nxdo cos t 91 M ( co — d1) sin 291 + (cl — d2)(2nx sing 91 — v; sin 201)
1
+ sin 9 1 12n xco cos t 91 — (e1 + doXnx sin 291 — v, cos 291) + 2 11xd l si u2
8 11}
+ ti{ cos 91I — nxdo sin 291 + nx ( co — dl) cos 201 + (c1 — d2) ( — nx sin 291 + 2v* cost 91A

+ sin 9 1 0=0 sin 29 1 + (c1 + do)(nx Cos 29 1 + v; sin 201)—nxdl sin 29

1 11 111}
+ 0 i1x nr( —c 1 + do + 2d2) sin 9 1 cos 01
vv 1 n 2 x(d0 + d2) sin 9 1 sin 291
vw t n 2 x( — ( c1 — d2 ) sin 291 cos 01 + (c1 + do) sin 91 cos 20i1
— ur n2 x2 2(do + d2) sing 9 1 cos 91 A

— v fw f n 2 x 2 [(c1 - d?,)(sin201 cos 01— sin 3 91)

+ (c 1 + do)(— sin 201 cos 91 + sin 3 91)^- (3.18)

-22-
 ORIGINAL RAG - F:V
OF POOR QUALITY

Gw c = 2 pe( sin 01
I—
dofl 2 x2 c 05 2 B1 + ( co dl )
( —$
n 2 x 2 sin 20 1 + flxv; cos 201
1
+ (cl — d2)I(flx sin 01 — v; cos 01)2 — 2112X?lr#p sin 01J^

— cos 01 (co ( 112 X2 cost e l + flxv; sin 201)

+ (c 1 + do)I nx cos 01(—Qx sin 01 + v; cos 01 + flgr flp) — flxv, sing Il + 2n v si 201
1
+dlPX 2 sin` of — nxvi sin 201) }

u 2n2xl( c l -- d 2 )sin 3 0 1 + ( e l + do)sin 0 1 cos` 01

— vfl 2 X#p(( c l — d2) sin 01 sin 20 1 + (cl + do) cos 0 1 coos 2011
— w2f2x#p1 (cl d2 )sin3 01+(c1+ do) sin01cos"01^

wffl2xnr12(c1 — d2)sin2 01 + (cl + do) CO52

011
+ v sin 01 I —2fZzdo cos3 . 01 _ flx(co -= dl) sin 20 1 + ( cl — d2) 2flx sin e 0 1 vi sin 201)
1
— cos 0 1 [2nxco cos' 01 — (Cl + doKflx sin 20 1 — v; cos 201) + 2f]xd l sin g 01 1

+ w{ sin 0 1 I —ilxdo sin 20 1 + flx ( co — d Z ) cos 201 + ( c1 — d 2 ) ( — flx sin 201 + 2v; cos' Ol)^
111 `
1
cos O 1 flxco sin 201 + ( cl + do)(flz cos 201 + v; sin 20 1 ) — flxdl sin 2011}
l J i
— ilz q^2(e
r l —d.,. ,) si n 1 +(1
B e+ d0) ros 2 0 1
1
— vvifl`x12(cl — d2) sin3 Ol + (cl + do) sin 20l cos 011
— vw^n2 x((cl — d2 )sin 0 l sin 20 1 + (cl + do)co5 01 cos 20 1 1 j

v^^fl 2 x 2 l( c1 — d2)s n3 of +(cl + do) sin 0 1 cos t 0 1 a

l 1
+ v^w^fl ` x2 [(cl d2) sin 01 sin 20l + ( cl + do) cos 01 cos 2011

— writ 2
x^ 2. Ir(cl — d2 ) sin3 e + do) sin 01 cost B 1,
^ 01 + (t (^--
"'' ))

3i

x
1
r
 t ORIGINAL PACE I'
OF POOR QUALITY
//
— e d l (e j + do) sin / 1 cos /1 — d i sin ce 11
Moe — 2 pc(il'x' CCMacr
( l 111

— Oxed( e l + do) 6 sin 211 — w cos 201) (3.18)

The singly underlined terms are nonlinear and the doubly underlined terms are
third order damping terms.

3.2 Noncirculatory Forces

The noncirculatory components are obtained from unsteady thin airfoil theory
(Fung (1969), pp 209-2101. To order e2, they are

Lw ne — xpa, 1 — w + ilx^ + ( 4t -1- td/"I

`! l (3.19)
Mont — 4 7rpc "L4\ e + ad iv — ( I e +'d Inx^^

The expressins for the complete aerodynamic forces in the lead-lag, flap, and
torsion directions are given by

L„ — L '
LW — Lwe +
Lwne (3.20)
M —
O
Moe+Mono
I

In the derivation of the above aerodynamic forces, nonlinear rate product terms
such as u2 , w` , vw are neglected while retaining other nonlinear terms up to order r

E". This is necessary for a linearized perturbation analysis about th e steady state.

a
Some damping terms of order c o , which may be important for the stability analysis,

are retained. The uniform induced velocity is based on the momentum theory.
The pitet, of the blade is obtained by combining the momentum and blade clement

theories.

F(3.21)
0,75R OCT
ad
+
1 5xi

where k h is an empircal factor, CT the thrust coefficient, v the solidity ratio of the

rotor, and 0,75R the blade pitch at three-quarter span.

t
F i

4, FINITE ELEMENT DISCRETIZATION

4,1 Equations of Motion in Terms of Nodal Degrees of Freedom

The finite element analysis is used to discretize the spatial dependence of the
equations of motion, The blade is divided into a number of beam elements, Every
element consists of two nodes, denoted by mode-1 and node (Fig. 6), with five
degrees of freedom namely v, V, w, wl , and at each node. The reason for choosing
over ¢ as the torsional degree of freedom is as follows. The aerodynamic loads
depend explicitly on ^. If 0 is chosen as the torsional degree of freedom, then
the presence of the integral term jo v"w1 dx [Eq. (2,2)] makes the global matrices
nonbanded because of the coupling of the degrees of freedom of one element with
that of the other Elements. The choice of yi eliminates this integral and thus
preserves the banded structure of the global matrices.

Hamilton's principle, Eq. (2.12), is discretized as

n
OMEaj-0 (4.1)
im l

with

A j mbUi — bTi — Mi(4.2)

where Wi, bTi , and Wi are respectively the strain energy, kinetic energy, and the
virtual work contributions of the ith element and n is the total number of elements.
The distribution of the deflections v, w, ^ over an element is represented in terms
of the nodal displacements using shape functions. For the ith element (Fig. 6),

-26—
 ORIGINAL PAGfi` IS
OF POOR QUALITY
V

where the shape function matrix [771 is

Ir

0
Hi H2 Hs H4 0 0 0 0 0

1111 — 0 0 0 0 H1 H2 H3 H.1 0 0 14.4)

0 0 0 0 0 0 0 0 H^ 'Ho"

and the vector of element degrees of freedom (q;) is defined as

02j (4.5)
(qs} ^.- L v l vl v2 u2 wl w1 w2 w2 01

The nodal degrees of freedom at node-1 of the element are v i , vi, wt, WO ^, and j

those at node-2 are v 2 , v2, w 2, WI The shape functions in Eq. (4.4) are the

Hermite polynomials defined as ti

Hl(xs) ti + 1 i
C ^^ )3 — 3Cx
d 2
Zs ^ — 2 xti + Xi
HAZO Ii

x; 3 3 z;1

(xi 3
l rx; 2

r,

H02(xi) a I
9

— 27 —
 I

where l; is the length of the ith element and xi is the local axial coordinate for
the ith element, measured from the left end of the element ('Fig, 6). Similar to
Eq, (4.3), the distribution of the virtual displacements 6v, 6er, and 6^ over the ith
element is assumed to be

6v

6W 1771 094 (4.7)

6

The substitution of Eq. (4.3) and Eq. (4.7) in Eq. (4.2) results in

i
D ^ f(4ii 4i: gip 6qi) (4.6)

which is nonlinear in qi. This function can be conveniently written as

A i' M ( 6gi) T jM%jgi)]( gi) + (6gi)T[Ci(gi))(4i)

(4.9)
+ ( 6gi ) TjKi(gi)j (gi) ' (6gi)T(Qi)
i

where [Mi(gi)], (C,{qi)), and [ (qi)), represent the element inertia, damping and
stiffness matrtees, respectively and (Qi) is the element load vector for the ith
{ element. The global matrices are obtained by the assembly of the element matrices n
3

-28-
 ORIGINAL PAGE 13
OF POOR QUALI'T'Y

The global degree of freedom vector is denoted by (q) and the global virtual ='
z

displacement vector by (bq). The global load vector (Q) is formed by combining y
the element load vectors (Q;). The assembly of element inertia matrices (M;(q;))
results in global inertia matrix (M(q)) as shown below. i

RNs-sal

IM
MZ]

[M(q)]
M3]

0 CM-1

The matrix [M] is a square matrix of order 5(n + 1) it is also banded with the
semibandwidth N. equal to 10. The global damping and stifness matrices )C(q)) and i
{K(q)) are obtained in the same manner. The matrix )K] also is banded, whereas
A

the matrix )C] is nonbanded because of the presence of the double integral term
0

t^ x
JO ) (6wV + V.f) dx dxi

in the expression for 6T; ]Eq. (2.18)]. Matrix )M) is symmetric; the presence of
terms in [C] and [K] due to the nonconservative aerodynamic forces makes them
asymmetric. The process of assembling element matrices to obtain global matrices
can be mathematically indicated by substituting Eq. (4,9) in Eq. (4. 1),
4

29
i

A (6q)TjM(q))(q) + (6q) T (C,(q))(q)

(4,1Q)
+ (6q)T[K(q))(q) " (6q)T (Q) - p

Since the virtual dispalcements 6q are arbitrary, Eq. (4,10) leads to the equations
of motion

IM(q))(4) + IC(q ))(4) + (K(q))(q) — (Q) (431)

These equations are nonlinear in q. The bandedness of (M) and (KJ helps reduce
the storage space needed in computer programs since the zero elements aB the band
need not be stored,

4.2 Boundary Conditions

The formulation of the problem is based on energy prieniples; and hence, the
force boundary conditions, which are imbedded in the formulation, do not have to
be considered separately. The nature of the displacement boundary conditions at
the root determines the configuration of the rotor. For example, v, V I , a,, w, and 0
are zero at the root for a hingeless rotor blade; and for an articulated blade, v, w,
and are zero at the hinge.
r

b. SOLUTION PROCEDURE

.5.1 Trim Solutim

The steady trim condition of the rotor in hover is determined first since the
flutter stability equations are linearized about the trim state. The nonlinear steady-
state equations are obtained by dropping all time-dependent terms from Eq. (4.11).

(Ko(go)l{ qo} _ (Qo) (5.1)

The matrix (Kol is asymmetric, banded, and a function of the steady displacement
qo. The contribution to the matrices (Kol and (Qo) is from both structural and
aerodynamic forces. The expressions for the terms of the steady-state element
matrices are given in Appendix C. These matrices , are evaluated numerically us-
ing Gauss quadrature formulas. The numerical solution of the nonlinear steady-
state equations is evaluated iteratively using Brown's algorithm [Brown and Dennis
(1972)].. This algorithm is a modified form of the Newton-Raphson method and the
solution is estimated such that the sum of the squares of the errors in the equations
is a minimum. The linear solution of Eqs. (5.1) is used as the initial estimate in
the solution procedure. The nonlinear stiffness matrix (Ko(go)l is updated at every
V
step of the iteration using the estimate from the previous step. The bandedness of
(Kol is useful in two ways. The storage space needed in the computer program is
less since only the elements within the band are stored. The number of computer
operations is reduced considerably since the matrix operations corresponding to the
zero entries of (Ko', outside the band, are skipped.

For a given level of thrust, the collective pitch required at three-quarter span

( 0 ,758) is calculated from Eqs. (3.18). The steady elastic twist modifies the pitch

- 31

3
 f

distribution of the blade. The now pitch at three-quarter span is 8 ,758 + 0,7511 and
corresponding to this pitch vallue, a new thrust level is calculated.

5.2 Flutter Stability

The flutter equations of motion are linearized about the steady-state solution.
The displacement q is written as the sum of a steady component (90) and an unsteady
perturbation (q).

q—qo+4 (5.2)

Substituting Eq. (5.2) into Eqs.. (4.11), subtracting Eqs. (5.1), and keeping only
linear perturbation terms, the flutter equations are obtained as

[if(go)]{9} + [&(go)] I q- ) + [k(o)](4)

* _ ( 0) (5.3)

The inertia, damping, and stiffness matrices are functions of the steady deflection
qa. The expressions for the terms of the perturbation element matrices are given in

Appendix D.
The flutter stability is determined by studying the eivenvalues of Eqs. (5.3).
The normal mode method (which considerably reduces the number of equations) is
used to calculate the flutter eivenvalues; and hence the natural modes are needed,
The coupled rotating vibration characteristics about the equilibrium position are
evaluated by dropping the damping matrix and removing all aerndvnamie terms
from Egs. (5.3). The resulting equations are

r
IMO
I {91 + [K.J( g) _ ( 0)
-82-
The inertia and stiffness matrices [M.] and (K,1 are sym
(frequencies) of Eqs, (5,^) are real.
The flutter equations of motion, Eqs. (5.3), are transformed to the modal space
by writing

(q) °'- (I l(p) (5,5)

where 1(P1 is the matrix of the first N eigenvectors ( colulmnwise) and (p) is the
vector of N generalized coordinates in the modal space. Substituting Eqs, (5.5)
into Eqs. ( 5.3) and premultiplying by 14^ 1 T yields the modal-space equations

(M"1(0) `1' (C '1(0) + (K`l(p) — ( 0) (5.8) 1

{

The matrices (M'1, (C'l, and [K'] are asymmetric and of order N. These
equations can be written as a first order system of 2N equations.

[A](r) — [BI (r) (0 ) (5.7)

where

IMO ) (C'1
(Al s
(01 V]

3n

(ol (K' l`3^t

[B]
[V) 101
!3

,.
and
^F
i

f(p)

By writing r as

r — feXt

Eqs. (5.7) becomes

[A] ( F) _ X(B) (r)

These equations are solved as an algebraic eigenvalue problem. The e _o__._._^

are complex.

t l^
 8. APPLICATION TO SINGLE-LOAD-PATH BLADES

To show the fesibility of the finite element formulation, the method is first
applied to single-load-path blades; a hingeless blade and an articulated blade with
a hinge offset of B%are considered. Convergence studies with respect to increasing
k number of finite efiements and natural modes are presented. For the hingeless blade,
comparisons are made with previous theoretical results obtained by modal methods
and with some experimental results.

The numerical results obtained are for blades with uniform spanwise poperties.
The chordwise offsets of the center of mass, aerodynamic center, and tension center
from the elastic axis are considered to be zero. The section constatnts EB I , EB2,
and the warping constants F.,CI, EC2r are taken to be zero. A precone (#p) of 0.05
rad' (2.0 deg), Lock number (-I) of 5, and solidity ratio (o) of 0.1 are used.

6.1 Hingeless Rotor Blades

The uniform blade properties selected for the stability analysis of a hingeless
rotor blade are given in Table 1. The stiffnesses EIy, EIz, GJ, and the inertial
parameters k,,, l , k,,,2 , kA are chosen such that the rotating frequencies corresponding
to given values. The rotating Hap frequency of the blade is taken to be 1.15f2. Two
different lead-lag frequencies are considered; a frequency of 0.70 represents a soft-
inplane blade whereas 1.50 represents a stiff: inplane blade. Similarly a torsional
rotating frequency of 2.5n represents a torsionally - soft blade and 5f) a torsionally-
stiff blade.

The convergence of the steady state deflections using different numbers of finite
elements is presented first. Table 2 shows the steady tip deflections va itp wo, p
(nondimensionalized with respect to the rotor length) and ^0,,p (in rad) as the total
number of finite elements is varied from 2 to 8. The results are for a thrust level,

-85-
 I
F;

CT/v, of 0,1, The results show, for the case coi

sufficient for good convergence ( three to four digit a,......,,.
the stiffness matrix reduces both the storage space and computation time required.
For example, if eight elements are used, the full storage mode would require 2025
storage locations whereas the band storage mode needs only 855 locations for the
stiffness matrix. Furthermore, the number of arithmetic operations involving the
stiffness matrix is reduced by about 60 %. The use of the linear solution as the
starting vector makes the nonlinear iterative solution converge rapidly; only three
or four iterations are needed for a converged solution. As a result, the computation
time required for the trim solution is small.

Figure 7 shows the steady tip deflections of a stiff - inplane hingeless blade as the
j
thrust CT Jv, (and hence the pitch 8) is varied. The torsion frequency corresponds
to a torsionally-stiff blade for these results. The magnitue of the flap and lead-lag
tip deflections increase with thrust. The magnitude of the torsional tip deflection
increases first and at higher thrust levels decreases. The torsional deflection is
primarily due to nonlinear flap-torsion and lag-torsion couplings. The steady-state
aerodynamic torsional moment is zero since CM,, and ed are assumed to be zero
[see Eq. (3.18)]. At low values of CT/o, both the flap and lead-lag deflections are
negative. At high values, the flap deflection is positive while the lead-lag deflection
is Negative. The results of Hodges and Ormiston (1976) are also shown in this figure.
Hodges and Ormiston use a modal method with five nonrotating beam modes for
each one of the deflections vo, wo, and Oo . The agreement between the two results
K
is excellent except at high trust levels where a slight deviation appears. This may
be due to the assumption by Hodges and Ormiston that sin B P^e B and cos B mze 1 (not
assumed here) for the claculation of aerodynamic loads, which may not be very
accurate at high values of B. The steady tip deflections with respect to thrust for
F

a soft-inplane blade are shown in Fig. 8. The flap, lead-lag, and torsion curves ' x
36 —
 are similar to that of the previous case. The flap deflections are almost identical
to that of the stiff,-inplane blade. The lead-lad deflections are higher in magnitude;,

as expected, since the lead-lag stif fness is lower. The torsional deflections also are

higher in magnitude than for the still inplane blade, which can be explained as due

to the coupling between lead-lag and torsion modes. Corresponding results from
ro
Hodges and Ormiston (1076) are included in Fig. S. Comparisons show that, as

ro
before, the agreement between the two results is very good at low thrust levels and
the deviation increases with the thrust,

The rotating coupled natural frequencies of the stiff inplane blade about its
steady deflected position are investigated for convergence as the number of finite

elements is varied. Table 3 presents the fundamental lead-lag, flap, and torsion

frequencies with increasing number of elements, It is seen that six elements are
sufficient for four digit accuracy. The bandedness of the inertia and stiffness
matrices is used effectively to mimimize computation time.

The solution of the flutter equations results in complex eigenvalues. The normal'

mode method is examined for convergence as the number of modes is varied, keeping
the number of elements fixed at six. The real parts of the first three eigenvalues for

different numbers of modes is given in Table 4. This table shows that five modes
result in a well-converged solution (five-digit accuracy). Figure 0 shows the root
locus plot of the fundamental lead-lag, flap, aad torsion modes as CT/O' is varied
from 0 to 0.3. Even though blade angle-of-attack stall has been neglected, such

high thrust levels are included for comparision with other theoretical results. Six
finite elements and five normal modes are used in obtaining these results. The flap
3

and torsion modes are stable over the entire range of CT /o considered, whereas the

lead-lag mode is unstable for CT /o between 0.01 and 0.05 and above 0.17. The

complete solution using sic elements and five normal modes at a particular thrust
level requires about 7 s of computation time on a CDC 7600 machine. Figure

- 37 -

w-
10 shows the flutter stability boundaries as the lead-lag frequency is varied. The
results of Hodges and ormiston (1076) are also shown for comparison in this figure.
In this reference, the stability results are calculated using the normal mode method
with five coupled rotating modes. There is general agreement between the two
results, The differences that appear between the two results may be due to changes
in aerodynamic forces by including all second-order terms (order 0) in the present
analysis.

Figure 11 shows the root locus of the lead-lag mode of a stiff-inplane hingeless
blade when the blade pitch is varied from -10 deg to 12 deg, The blade precone
is zero for this case. Experimental results from Sharpe (1985) are also shown in
this figure. In the theoretical analysis structural damping is included such that the
results match that of the experiment at zero pitch. It is seen that the agreement
between the present analysis and experiment is good at low pitch settings and not so
good at higher pitch settings. Other theoretical analyses show similar discrepencies.

The same blade properties as given in Table 1 are used for the stability analysis
of an articulated blade with a hinge offset of 6%. The convergence characteristics
for the nonlinear steady solution, natural frequencies, and flutter eigenvalues are
3
presented in Tables 5-7, respectively. These results confirm the conclusions reached
for the hingeless blade that about eight elements and five normal modes are adequate
for four-digit accuracy. As before, only a few iterations result in a converged
nonlinear trim solution. The equilibrium tip deflections are plotted with respect
to thrust in Fig. 12. The lead-lag tip deflections are much higher than that for
the corresponding Wngeless rotor blade because the lead-lag stiffness is lower when
the blade is hinged. The root locus of the flutter eigenvalues obtained using six

- 88 -
 }

elements and five normal modes is shown in Fig, 13, All three modes are
over the range of CT le, shown,
 7. MODIFICATIONS FOR ANALYSIS OF MULTIPLE-LOAD-PATH BLADES

Having shown that the finite element formulation can be successfully applied for
the determination of the aeroelastic stability of single-load-path blades, the analysis
is extened to include the more complex multiple-toad-path blades. As indicated in
the Introduction, the flexure of a bearingless blade contains one or two flexbeams
and a torque tube. The analytical model of a bearingless blade is shown in Fig. 14.
Each of the flexbeams and the torque tube are modelled as individual beams, For 4
this purpose several modification have to be made to the formulation presented in
the previous chapters.

7.1 Inclusion of Axial, Degree of Freedom

In the analysis for single-load-path blades, the axial deflection u has been
eliminated in terms of the other deflections and the expression for the centrifugal
force. The force equilibrium equation in the axial direction cannot be solved a priori
for multiple load-path blades; and hence the u deflection has to be included as a
degree of freedom.

7.2 Refined Finite Element

The beam element considered for discretizing multiple-load-path blades is shown

in Fig, 15. This element is more refined than the simple beam element used for
single-load-path blades. The element includes the axial degrees of freedom. Each
element consists of two end nodes and three internal nodes with a total of fifteen
degree of freedom. Each of the end nodes (1 and 2) have six degrees of freedom,
namely u, v, v1 1 w, ud, and i^ . There are two internal nodes for u and one inter-
nal node for ^. Two internal nodes are needed for the axial degree of freedom
to accurately represent the centrifugal force distribution over the element. With

-{0-
E
r

two internal nodes, the distribution of u over the element is cubic. Hence, the axial
force distributiion is quadratic and of the same order as that of the centrifugal fore.

The internal node for the torsional deflection assures that the torque approxima-

tion over the element is of the same order as the flap and lead-lag bending moment
approximations.

7.3 Displacement Compatibility Conditions at the Clevis

The element matrices are assembled, by considering displacement compatibility

conditions at interelement boundaries, to form global matrices. Except at the clevis,
the displacement compatibility conditions generally imply the continuity of u, V, V,
W ? WI , and across element boundaries. The root element of the outboard blade is
connected to the outermost elements of the flexbeams and the torque tube through
the clevis which is assumed to be rigid (Fig. 14). The inboard beams are numbered
1,2,3, etc. The displacement compatibility conditions at the clevis can be written
i
as

tt ;
3
4

Ui — u niVI
Vi V
V' V,
(?.1)
A
Vi '. W + Qi (Sae + )
W WO

and

Ol e ae + 0 (for flezbeame)
Oi : 0 (fortorquetube) (7.2)

_ 41
 l

f
where ui , u„ wi, and O; are the deflections at the clevis for the ith inboard 'beam; and
u, , w, and 0 are the deflections at the clevis for the outboard beam, The n; is the
t, ,

elastic axis oflset of the ith inboard beam from that of the outboard beam (positive
r forward) and 0., the difference between the collective pitch and the pretwist at the
clevis (Fig. 16). Equation (7.2) for the flexbeam indicates that a rigid body pitch
'r .rotation of the outboard blade is transferred to the flexbeam as an elastic twist.

7.4 Modifications in Solution Procedure

In addition to the differences noted above, there are two major differences in
tk - -wl ;ion procedure for nonlinear trim solution. Some of the terms of the matrix
[Kol depend on the centrifugal force F. The centrifugal force distribution is known
a priori only over the ouboard blade; it is unknown over the flexure because of
the multiple load paths involved. In obtaining the linear estimate of the steady
equations, it is assumed that the centrifugal force in each of the inboard beam is
in the ratio of its tensile stiffness EA. Then, at successive iterations the centrifugal
force distrubution is updated using Eq. (2.6).
The lead-lag and flap stiQnesses depend on the orientation (pitch) of the cross
section. The pitch distribution is known a priori for the outboard blade. For the
flexbeams, the resultant pitch is the sum of the pretwist Bpe and the elastic twist due
to Bay at the clevis (Fig. 16). For the first iteration of the nonlinear trim solution,
the pitch distribution along the flexbeams is assumed to be linear. This distribution
is updated at successive iterations based on the torsional deflection. a

s
F
t

8. APPLICATION TO MULTIPLE-LOAD-PATH BLADES

The formulation modified for multiple-load-path blades is applied to several

analytical models of bearingless rotor blades in hover. The outboard blade, the
flexbeams, and the torque tube are considered to be uniform. The chordwise offsets
of the tension center, mass center, and aerodynamic center from the elastic axis are
assumed to be zero. The section constants EB I , EB2, and the warping constants
ECI, BC2 are taken to be zero. The aerodynamic forces over the flexure is neglected.
A precone (pp) of 0.05 rad (2,9 deg), Lock number (-i) of 5, and solidity ratio (v) of
0,1 are used. The offset of the elastic axis of the flexbeams (Fig. 14) from that of the
outboard blade (raj a —113) is taken to be 0.4c, where c is the chord of the outboarad
blade; the torque tube offset (pz) is zero. The length of the flexure is 0.25R. The
blade structural poperties are given in Table 8; the ac, ►odynamic properties are the
same as given in Table 1. Stiff and soft-inplane and torsionally-stiff and soft blades
are considered.

8.1 Twinbeam Model with Zero Inboard Pitch

-
To understand the behavior of multiple-load path blades, a simple analytical
l
model is considered first. This mode has an inboard flexure cosisting of two identical
(
flexbeams; the torque tube is not included Fig. 17). The pitch of the flexure is
fixed at zero while the pitch of the outboard blade is varied. This is equivalent to
a physical model with a pitch bearing at the clevis. First, a convergence study is
made to determine the number of finite elements needed to obtain the nonlinear
trim solution to a desired accuracy. Table 9 shows the steady -state tip deflections
tr
uo s , Vo ttip , wo, lp ( nondimensionalized with respect to the rotor length), and ^0, p (in
rad) as the number of both the flexbeam and the outboard elements is varied. The
blade considered is stiff inplane and soft in torsion. A thrust level, CT /o, of o.1 is
used. The results show that two elements each for the flexbeams and two elements

J
 for the outboard blade are sufficient for five digit accuracy, The nonlinear iterative
solution converges rapidly since the linear solution its used as the initial estimate;
only four or five iterations are needed for a converged solution.

The rotating coupled natural frequencies of the blade about its steady deflectd
position are invelitigated for convergence as the number of finite elements is varied,
Table 10 presents the fundamental lead-lag, flap, and torsion frequencies with k
increasing number of elements. Again, it is seen that a total of six elements, two
g
for the outboard blade and two each for the flexbeams, are sufficient for Ave digit
accuracy.

The solution of the flutter equations results in complex eigenvalues, The normal
mode method is examined for convergence as the number of modes is varied, keeping
the total number of elements fixed at six. The real parts of the first three eigenvalues r

for different numbers of modes is presented in Table 11. This table shows that five ^ a
modes result in a converged solution ( five digit accuracy).

Figure 18 shows the root locus plot of the fundamental lead-lag mode as CT /v
is varied from 0 to 0.2. It is seen that the lead-lag mode is unstable for CT/O' higher 3 {
than about 0.05. In this figure, the results from an equivalent - beam model are also r;

shown. In general, the properties of the equivalent - beam model are obtained by
matching the fundamental frequencies of the equivalent - beam model with that of
the multibeam blade. For flexbeams with uniform spanwise poperties, analytical
expressions for equivalent properties can be obtained. These are given in Appendix
D. The two results shown in Fig. 18 differ from each other, particularly at low
and high levels of thrust. The results at low thrust levels are of significance to tail w
rotors. A careful study has been made to determine the cause of the differences
between the two results. The equivalent-beam model does not accurately simulate
the cross-coupling stiffness terms which are nonlinear in nature and depend on the
N
steady deflections. Results have been computed by suppressing, in the perturbation
t
^44_
 analysis, nonlinear cross-coupling terms of nonlinear origin from the structural
stiffness matrix of the flexure, for the twin- beam as well as the equivalent beam
model, These are shown in Fig, 19.: Both sets of results are identical.

For the same model but considering a soft-inplane, torsionally-stiff blade, the
root locus of the lead-lag mode is plotted in Fig. 20, The behaviur of this blade
is very different from that of the still inplane blade. The coressponding equivalent-
beam model results are also included in this figure. The discrepency between the
results of the two models are more pronounced, particularly at high thrust levels,
than that of the sti&inplane blade. Again the removal, in the perturbation analysis,
of the cross-coupling terms of nonlinear origin from the structural stiffness matrix
of the flexure makes the results identical (Fig. 21). Similar conclusions are reached
from the root locus plots of the lead-lag mode of a torsionally-stiff blade (Figs. 22
and 23).

8.2 Twinbeam Model with Same Pitch For Inboard and Outboard Segments a

This blade model also has a flexure consisting of two identical flexbeams; and
i
there is no torque tube. Both the inboard and the ouboard segments are maintained
at the same uniform pitch distribution. This represents a physical model with pitch a

bearings at the hub end of the flexbeams. The root loci of the lead-lag mode with
4
r :^
respect to increasing level of thrust are shown in Fig. 24 for both the twinbeam and
the equivalent-beam models. The results are for a stifl=inplane and torsionally-soft
blade. The two results are in major disagreement with each other. This disparity
is explained as follows. The fundamental frequencies of the equivalent-beam model
are matched with that of the twinbeam model at a chosen pitch setting (zero in the
i
present example). As the pitch is changed, the effective flap and lead-lag stiffnesses
vary because of flap-lag coupling. This modifies the frequencies of the blade; and
^i

{5
t^

z
the frequency match no longer exists. This phenomenon is clearlyl seen from Table
12 where the fundamental frequencies are presented for various thjrust levels. As
the thrust (and hence the pitch) increases, the difference between the frequencies
of the two model widens. Thus, the equivalent-beam simulation is not accurate
with changing pitch. The results of Fig. 25 are for a soft-inplane blade. Again the
discrepency between the twin-beam and the equivalent-beam models is big. Similar
effects see observed from the results of a torsionally-stiff blade (Fig. 26). .

8.3 Bearingless Model with Torque Tube

This blade has a root flexure consisting of two identical flexbeams and a
torque tube in the middle. This model is a more appropriate representation of
the bearingless blade, Pitch control of the blade is achieved via the torque tube
by rotating the clevis. Due to the piltch application at the clevis, there is an
elastic pitch distribution over the flexbeam and this is determined iteratively. As
a result, the number of iterations required Ivr convergence of the nonlinear trim

solution is increased. Now it takes about ten iterations for five digit accuracy,
whereas, four or five itereations have been adequate for the two previous models.
The convergence characteristics are similar to the one presented for the previous
models in tables 9-11. The results have indicated that a total of eight elements -
two elements for the outboard blade, two for the torque tube, and two each for
the flexbeams - are adequate for five digit accuracy. The root loci of the flutter
eigenvalues (fundamental lead-lag, flap, and torsion modes) are shown in Fig. 27 s

as CT/ o is varied from to 0.15. The flap and torsion modes are stable while the
lead-lag mode becomes unstable at high levels of thrust (CT /of greater than about
0.13). This is a weak instability and the introduction of a small amount of structural
(
damping would stabilize the mode.
1i

i
c

- 46 - 4

f}l

A
 9. GENERAL CONCLUSIONS

The finite element method has been successfully applied to determine the
nonlinear trim deflections and the, flutter stability of sing.) ,^- and multiple-load-path
blades in flap bending, lead-lag bending and torsion. The formulation is based on
Hamilton's principle, The spatial dependence of the equations of motion is made
discrete by dividing the blade into a number of beam elements. Nonlinear trim
deflections are evaluated iteratively, solving the complete set of of global equations,
without making a modal transformation. The solution procedure is made efficient
using the bandedness of the stiffness matrix and the linear solution as the starting
vector. The coupled rotating natural modes are calculated about the trim condition
of the blade. The normal mode method based or., these modes is used to solve the
linearized flutter equations as an eigenvalue problem.

The formulation is first applied to uniform single-load-path blades, namely

hingeless and articulated. Six finite elements are sufficient to determine the trim
deflections, and the free vibration characteristics. Five coupled rotating modes
are adequate for the normal mode solution of the flutter equations. The steady
tip deflections agree very well with that of a previous research which used modal
method based on a total of fifteen nonrotating uncoupled modes; the flutter stability
boundaries also show excellent agreement with that previous research based on a
modal approach. The root-locus plots of flutter eigenvalues as a function of the
pitch are generally in good agreement with experimental results.

Suitable modifications have been made in the formulation so as to consider

multiple-load-path blades. Numerical results are presented for several baringless
fJJ

models. Results of the redundantbeam analysis are compared with that of the
1

equivalent-beam modelling; and the following important conclusions are reached.

The equivalent beam does not simulate the nonlinear structural coupling terms ap-
}

_ 47 , i
prapriately. Also, the properties of the equivalent beam are estimated by matching
frequencies with the bearingless blade at a particular pitch setting. At other pitch

settings, the frequencies change and the match no longer holds. As a result, the

stability results obtained using the equivalent-beam model are inaccurate.

Future extensions may include forward Right, ground and air resonance, and

consideration of composite blades.

f
 APPENDIX A. NONDIMENSIONAL QUANTITIES

The nondimensional quantities in 6U, 6T, and 6W (Eqs. (2.17),

(2.18), and (2.10)) are defined here. These symbols are retained
after nondimensionalisation.

Assumed order
Old quantity New quantity of magnitude
x/R x 1
0/ n ^) 1
v/R v E
w/F? w E
^ ^ E
e e i
F/mo11 2 R 2 F 1
EIV /mof2 2 R 4 EIS, 1
Elx/mo f] 2 R4 El, 1
GJ/mofl ? R' GJ i
EA/mon o R 2 EA 1 /E2
EB l /mof2 2 R° EBr to
EB2 /mof3 2 R a EB2 E

EC, /mof2 2 R6 ECl E2

EC2 /mof2 2 R6 EC2 E

k A IR kA e
km, /R km, E

km a/ R k's 6
e, /R e. E3/2
e0 /R e0 E3/2
#p I #p E

p +

r
tx ;
 i
ti

APPENDIX B. STEADY-STATE ELEMENT MATRICES

The expressions for steady-state element stiffness matrix and load vector are
derived from the steady part of Eq. (4.2). M

r
1

Y.

AOi 6Uoi — 6Tps —'Wpi(B.1)

The strain energy, kinetic evergy, and the virtual work contributions of the ith
element in steady state are obtained from E,j3. (2.17), (2.18), and (2.19) by dropping
all time dependent terms and writing

$in (B + ¢) sin e + ^ cos 8

(B.2)
cos (B + 0) CO3 0 — ^ sine r

r l

a
 v

ORIGINAL PAGE IS
OF POOR QUALITY
-
o p
-- 1 ^Fo(,, 0 -1- wpdw)

+ Gj(i ,+ ^pw I bVI, + 0pupdw, + upwp6

[Fokj(01 + p — EAkgBA of + EAeakg9'(vp cos 0 + wg sin 0)

+ EB1 0' Oo — EB20'(up cos 0 + woo sin 0)J

) d

(ECI^lof + EC2(w0 cos 9 — vp sin B)J d0

+ [ —Foe. cos 0 + (EIS cost 0 + EIy sine 0)up

+ (Eh — EIy)^ O(wp cos 20 — vp sin 20)
+ 2 (EIz — EIy)w p sin 20 — EAea(vp cost A -}^ 2 wp sin 28)
a
-r (EAe,, kA — EB 2 )0'00 cos 0 — EC24 sin 0 16V

• Poea sin 0 + (EI, sine 0 + EIy cost 0)A

• (EIz — EIy)0p(vp cos 20 -#- w oo - iin 20)
• 2 (Er, — EIy)vo sin 20 — EAta1 w"
p sine0 + 2 vo sin 20^

• (EAc,,k,2, — EP)0 1 ^ 1 sin 0 — EC2^p cos 0 dw"

• lFbea (vpins 0 — wp cos 0)

• (EI, — EIy){
1
21 wp2 — vp ain 26 + vpwp ccys 21 } d^ dz; (B.3)

i
9

- 51 '-
 ORIGINAL ME IS
OF POOR QUALITY

^z
r 1
o ^^ ` to [vo + cg cos B — ^o ain o 6v
Jp
moil
.
_ #px6w
[ 2 ( m2 k im)(2 sin 2000 cos 2I^ + cpx(W0 cos B — vp sin /)

+ epvo sin B + e D #px cos 91b^ x

— esx(cos_0 — ¢o sin 0)6v#

egx(x sin I + ^o cos 6)6w') dxs (B.4)

and

1
64Yo k
--
mo
0 i = f (Lvo6v + L,6w + M0060) dx
0
(B.5)

a
-d
A
where the steady-state aerodynamic forces are 1
i
ra
{
it 3

T
u
—52—
t
ILI
_.,
 ORIGINAL PAGE IS
OF POOR QUALITY

Leo _ cos $ — dox 2 cos t 0 + ( co — dl)(— 2 x 2 sin 2e + AicoA

C 0)
+ (cI — d2)1 (x sin a 03 0)2 -- 2xqr flp sin e1
}
+ sin o co X2 cos` a +-xXisin 20)
f (
+ (c l + do)I x cos 6(—x sine + xi cos e + grAp) xXi sin g e + 22 a; sin 201

+dl(x2 sing 9 — Ai sin 29))

— uox(do + d2) sin 29 sin 9 111
voxOp(( cI —d 2 ) sin 29 cos0+ (cl + do) sin 0 cos 201
+ woxflp(do + d2) si n 29 sin e
+ wp2xqr( — cl + do + 2d2 ) sin a cos e
+ 0o cos 9[—(co — dl)x 2 cos 29 + (cl — d 2 )(x2 sin 29 — 2A cos 20)]

— sin e1(co — dl )x2 sin 2e + (el + do)(x2 cos 2e + 2xX i sin 29)]

sin 0 dox 2 cos t 9 + (co — d l - 1 x 2 sin 29 + xXicoe2ii^

1— )(
+ (cl — d2 )1(x sin _0 — Xi cos 0)2 — 2xqr#p Sin
01
+ cos e j co(x 2 cost a + Aioin2e)

+ (c l + do)i x cos 9(—x sin e + X i cos 9 + gf^Bp) — Aj sin`+

e Z nX si 29
1
+di(x 2 sine 9 — A i sin 2B)1
110
— vovox(do + d2) sine sin 2
— yowox(—(cI — d2) sin 20 cos 0 + (c l + do) sin 0 cos 201
— vp x 2 2 (do +d2 ) sin2$cos9

- vow f 2 (cl — d2)(sin 20 cos e — sin 3 e)

a

+ (cI + do)(— sin 2e cos e + sin3 9)] (B.B)

hr..
 i

ORIGINAL PAGE IS
OF POOR QUALITY

Lwo = 2 sin 0 + xX icoe20

9 eiu o` —dox2 cost o + (co — di)` — 2x2
!
+ (c 1 — d2)1(z sin 0 — Xi cos 0) 2 — 2x17 r#p sin 01 }
— cos 0 1 eo(x 2 coat 0 + xX isin20)
+ (c1 + do )I x cos 0( — x sin 0 + Xi COS 0 + pr#o) xXi sing 0 + 2 X sin 20

+d i ( a2 sine 0 — xXi sin 20))

— uo 2xf d2)sin3 0 +(cl + do)siuOcos 2 01
vox^p((c1 — d2)'sin0 sin 20 + (c 1 + do)cos Bcos 20J
- wo 2x#pl ( c 1 — d2) sin 3 0 + ( Cl + do) sin 0 cos t 0^

— w 0'2xgr12(c 1 — d2 )sin 2 0 + (c1 + do)cos 2 0

J

+0 sin 0 —(co — d 1 )x 2 cos 20 + (c1 + do) X 2 sin 20 — 2zXi cos 2B)

l
+ cos 0( ( co d1 )x 2 sin 20(c1 + do)(x2 cos 20 + 2xX,, sin 20)
J

+ cos 0 dox 2 cos t 0 + (co di )^— 2x 2 sin 20 + aX icoW )

I—
+ (c1 — d2)1( x sin 0 — X , cos 0) 2 — 2xtlrgp sin 011

+ sin 0{ co(x2 cost 0 + xXein20)

+ (c1 + do) i x cos B(—x sin 0 + Xi cos 0 + rir^?p)— X,\, ain2 0 + 2g

X sin 21
1
+d i(x2 sin g 0 xXi sin 20)^l

vovpxl 2(c 1 — d2) sin B + (Cl + do) sin 20 Cos 0l

— yowpxl(cI — d2) sin 0 sin 20 + (c1 + do) cos 0 cols 201

— vp x2 [( c1 — d2) sin 3 0 + ( c 1 + do) sin 0 cost Bl

+ yowox2 ((c 1 — d2) sin 0 sin 20 + (c1 + do) cos 0 cos 201
i
— wp x2 2 ^(c1 — d2) sin3 0 + (c 1 + do) sin 0 cost 00 (B,7)

{
9
I

u..
 ORIGINAL PAGE 15
OF POOR QUALITY

Nfoo . ax` j cC Mac — c d[(c t + do)ainl cos!

The ©o,. of Eq. (B.1) can also be written as

ooi no (6q ) (Kilo ( gs)o + (6%)

By comparing Eqs. (B.1) and (B.0) the element stiffness matrix and the load vector
can be written. The contribution from Mo i and 67oi to the element stifTenss matrix
and the load vector is given as follows (the contribution from Mo i can be easily
inferred from Lvo, Lwo, and Moo).

[Kvvl JKvwl IC„

[Kilo [Kwvl [Kwwl h'wm (B.10)
l[Kov] [ Kow] Koo

and

(Qv)
M10 NO

where

(HI) ( EI. cost B Ely sin` 8 — EAea cos 6' {H"} {H"}T
l
flKv
Fo (HI) T+ +
L
— m(H)(H) T dxi
I

EI, — Ety EAta) sin 28{H"} {H"} T dzi

lKvu,l
f 2I

-55—
 ORIGINAL PAGE r
OF POOR QUALITY
T
(Kvol fo IjI EA ca kA
2 EB2 )01 cos*( Ho ) H^
(
+ Foca sin 0(HM) (Ho)T
T
EC'2 ain 4(X11 )
( Hll I
+ Meg sin $(H) {Ho }T
me9 z sin 6 {H} {Ho} T
(El; — aly)sin 29{H//}{H" }T {uu}(Ho)T
+ (Elt — Ely) cos 20 {H/} {H" } T ( w o) (Ho )T
H^) Tj
+ GJ(H"}{ H l } T {tio}I dz,

ol

(xwvl (Kvwl

l; 1
(Kwwl — [ Fo(HI) {H }T + (EI, sing 6 + Ely cost I — EAca sing I){H ll } {H"}T dz;
J
a
r
i

ii
4
1^
[Kwo] = I (EAedkq — EB2 )B'sin A{H"} ( H^) '

— Foe, cos 0(19") {HO }T

T
+ E02 cos O(Ho) HlN
1
+ Metz cos 9{Hl} {HO }T
+ (EIZ — EIy) cos 2B(H"){H"}T (up) {Ho}T
+ (EIz — EIy)sin20(H"){H" } T (wo){H } T
rr T
+ GJ{Hl } {H"}T (v0}{ Hl } d.—

-56—
 ORIGINAL PAGE 19
OF POOR QUALITY
^; f
[Kov) " [^EAoak^ —EB2)0' cos P(H) (H") T

+ Foo d sin O (Ho) (H" )T4

—EC2einOiH^^(H")T

+ mcs sin 0(H0)!{H}T

mcozsin0 (HO)(H' )T
— 2 (EIx —EIy)sin20(H0)(uo ) T (H")(H")T

+ (Elz — Ely) cos 20(H0 )(wp) T (H")(H" ) TI dz;

j 1
[(EAe,k 2 — FB2 )0'sin 0{ H?, }
(K^W ^ .. ..(Hot

Fpca cos I (H0 ) (H") T` J

+ E02 cos 0(H;)(H")T

+ meez cos 0(HO ) (H'}T

)T
-- 2( Eh — EIy} sin 20(HO)(wo) T (H")(H "
)T
+ CJ(H j(V0*)T(Hjj) (H' I dz;

s rr rr T
4 ? + EB I Ce2 #) H )
[Koo] " f ' [(GJ + Fgkp — EAkA
``
rr T
+ ECI{H'J(H )

+ m(k 2M2 km,) cos 20(H^} {H^} TI dz;

r;
NO — Jo (Focu cos 0(H") + mep cos 0(H) — megz cos O(H)} dz;

(Qw } ^if qca sln 0(H") + m^BP x{H} — mryz sin 0(H')} dxi
fo
Y

fo (-Fok 9'{ H• ^ 2 k o.
t /
— km ) sin 20{H0 } mcQ flpx cos 0 H dx;
1
I
57
_a

L
 The column vectors {H}, (H O), and

iH}T ^ L ZII "z n7 JAU

{HO} T LHO I H02 1

p}
{tJ'
T .ri lUpl U0 IV0 2 Vp 2j
A similar definition holds good for (WOV , }

- 58

i ^

Ld
 APPENDIX C. PERTURBATION ELEMENT MATRICES

The strain and kinetic energy contributions to the perturbation inertia, damp-
ing, and stiffness element matrices are given here, The contribution of the aerodynamic
forces is easily inferred from Egs.(3.16)-(3.18). Since these matrices are linearized
. about the steady deflection, the inertia and stiffness matrices are symmetric whereas
the damping matrix is antisymmetric; but the noncanservative aerodynamic terms
will make the damping and stiffness matrices asymmetric.
Inertia Matrix
The element inertia matrix for the ith element is written as

IV
[ IVVI Awl [Advol

IMi1 [Mwv1 IMww[ [Mwm1 (C.1)

[ MOV1 [ Mow] [M401
where

[MVVI s —
m(H)(H) T dzi
a

[MVW1-[01

[ MU01 _ f Meg sin 00(H){H) T dzi

!i
[Mww] -^ m(H)(H) T dzi

59 -
 ORIGINAL PaOE 1'
OF POOR QUALIT

t;
[MWm1 met cos do {H} {H

t
(?IS^^^ km{H}{H} T
fo m

Damping il^tatI
The element damping matrix for the ith element is written as

[ CVVI [CVWI [CVOI

[C,I — C
[ W VI [ CWWI [CWO1 (C.2)

[ C OVI [ COW] [COO]

whw.2

[CVVI' = Jots 2mey cos Io [(H l ) {H} — {H} {HI}T dz;

J

[CVW1= 1-2m#p(H)(H)T— 2megsineo - (H){HI }TI dx

[CV01 a 101

[CW W I - [o]

-60-
 ORIGINAL P,
OF POOR QI

[Cl

[C

Stiff m Mntrix

I•-vii I- - „w, &--vyj

(C.3)
[Ki l (ICwvl [ Kwwl [Kwol

[K¢v] [Kowl [Kmml

where

[Kvv j — f Fp{H l }{HI } T + (Eh coat Bp + EIy Sint 60 ,— E.la cod Ap

o [

m{H}{ H }T da;
1

'i rr T
[Kvw l = 1 2 (EIZ —EIy —EAca)sin28p{H^^}{Hn} a
f
T
+ GJ{H N } Hj ►0 {H` } T dzi

k j
1

- 61 -

^i
 OF
ooa QUALITY

[K„ol — Jo [(EA,,,k2 — EB2)B0 cos Ap{

+ Fpea sin 9p{H"} {HO} T

T
— E02 sin Bp {H"} F^^
1
+ meg sin 00 (H) {HO }T
— Megx sin 00(H,){ HO}T r

— (Elz —Ely) sin 20o (H") {H" }T {vo} {H0}T

+(Eh — Ely) Cos 2Bp{H"}{H^^ }T {wo}{H0}T
r T
+ GJ{HO}{H' }T dx;
{wo}(H, I

is T
JH'
{Kw^^ = J (EAeak' —EB2)8°p sin 0o(1r)'I

— Foe, cos Bp {H"} {HO}T

T
+ EC2 cos Bp {H"} '^I
JH

+ MegxCos 8p {H}{HO}T
+ (El.,— Ely) Cos 28o (H" {HN}T {u o} {HO }T
+ (El, — Ely) sin 2Bp{H M }{H" } T {w0*}{ H0}T
IT
r
+ G3'{H}{H"} T {uo
}(Hl10 dz{

i 4 $ +EBIfoXH0}{HO
2 — EAkA
I T
[Koo] _J ^ (GJ+FpkA T

• ECl
H" H^1
• Mk;,, — k,2„1 ) cos 28p{ HO) {HO} T 1 dxs

The column vectors {H}, {HO}, and { up} are defined as

— 62
r,
 T — LHI
{H} H2 Hs Hd

{HO }T _ LH+1 H#21

F f

li k

{vo}T •- l vo, vo i X02 vo2i

•
A similar definition ' and
holds good for {wo} Also so is defined as
to0)'

i1

ti

i
 ORIGINAL. PAGE IS s
OF POOR QUALITY

APPENDIX D. EQUIVALENT-BEAM PROPERTIES

4
G

Mathematical expressions of equivalent properties for modelling twin beams

t
(Fig. 17) as an equivalent beam are presented here.

l (EA)e — (EA) 1 + (EA)2

( Ely)t — (EIV)1 + (EIV)2

12 (+
( EIz)e ( El=), + ( EIz)2 + ni( EA + 172( EA )2 t
(GJ) e (GJ) 1 + (GJ) 2 +
l1 — n2)" (EIy)1(EIy)2
L j (EI0 1 + (EI1)2
l l
1
+ rii A l + 2A2
(Ak' )1 + (` k2A 2
(mkM i (mkM
(AkAI a

Me —
2

MI + m2
+ (mk2m)2 + q I + 92m2
F

The subscripts ej, and 2 refer, respectively, to the equivalent beam, nexbeam one,
b
and flexbeam two; and L f is the length of the flexure. These expressions become
F
simpler for the present case where the two beams are identical.

^` a
 ^^

REFERENCES

Bielawa, R.L., "Aeroelastic Characteristics of Composite Bearingless Blades,"32nd

Annual Forum of the American Helicopter Society, Paper no. 1032, May 1976.

Brown, K.M. and Dennis, J.E., "Derivative Free Analogues of the Levenberg-
Marquardt and Gauss Algorithms for Nonlinear Least Squares Approximations,"
Numerische Mathematik, 18, 1972, pp. 289-297.

Chopra, I. and Dugundji, J., "Nonlinear Dynamic Response of a Wind Turbine

Blade," Journal of Sound and Vibration, Vol. 63(2), 1979, pp, 265-236.

Dixon, P.G.C. and Bishop, H.E., "The Bearingless Main Rotor," Journal of the
American Helicopter Society, Vol. 25, July 1980, pp, 15-21.

Friedmann, P. and Tong, P., "Nonlinear Flap-Lag Dynamics of Hingeless Helicopter

Blades in Hover and Forward Flight," Journal of Sound and Vibration, Vol. 30(1),
1973, pp.9-31.

Fridemann, P. and Straub, F.K., "Application of the Finite Element Method to

Rotary Wing Aeroelasticity," Journal of the American Helicopter Society, Vol.25,
Ja.a. 1980, pp. 36-44.

Fung, Y.C., An Introduction to the Theory of Aeroelasticity, Dover Publications,

Inc., New York, 1969.

Gallagher, R.H., Finite Element Analysis Fundamentals, Prentice Hall, Inc.,

` Englewood Cliffs, New Jersey, 1975.

Harvey, K.W., Aeroelastic Analysis of a Bearingless Rotor, American Helicopter

i
Society Mideast Region Symposium on Rotor Technology, Aug. 1976.

Hodges, D.H. and Dowell, E.H., "Nonlinear Equations of Motion for the Elasti,-
Bending and Torsion of Twisted Nonuniform Blades," NASA TN D-7818, Dec. 1974.

- 65
 Hodges, D.H. and Ormiston, R.A. "Stability of Elastic Bending and Torsion

of Uniform Cantilever Rotor Blades in Hover with Variable Structural Coupling,"

NASA TN D-8192, Apr. 1976.

Hodges, D.H., "An Aeromechanical Stability Analysis for Bearingless Rotor

Helicopters," Journal of American Helicopter Society, Vol. 24, Jan. 1970, pp. 2-0.

Hodges, D.H., Ormiston, R.A., and Peters, D.A., "On the Nonlinear Deformation

Geometry of Euler-Bernoulli Beams," NASA TP 1566, Apr. 1980, 8

Hodges, D.H. and Rutkowski, M,J:, "Free-Vibration Analysis of Rotating Beams

by a Variable-Order Finite-Element Method," AIAA Journal, Vol. 19, Nov, 1981,

pp. 1459-1466.

Houbolt, J.C. and Brooks, G,W., "Differential Equations of Motion for Combined
Flapwise Bending, Chorwase Bending, and Torsion of Twisted, Nonuniform Rotor

Blades," NACA Rep. 1346, 1958.

Johnson, W,, "Flap-Lag-Torsion Dynamics of a Uniform, Cantilever Rotor

Blade in Hover," NASA TM 73248, May 1977.

Jones, J,P., "The Influence of the Wake on the Flutter and Vibration of Rotor

Blades," Aeronautical Quarterly, Vol. IX, 1958, 258-286.

Jones, W.P. and Rao, B.M., "Compressibility Effects on Oscillating Rotor

Blades in Hovering Flight," AIAA Journal, Vol. 8, Feb. 1970, pp. 321-329.

Jones, W,P. and Rao, B.M., "Tip Vortex Effects on Oscillating Rotor Blades

in Hovering Flight," AIAA Journal, Vol. 0, Jan. 1971, pp. 106-113.

Kato, K. and Yamane, T., "A Calculation of Rotor Impedance for Hovering

Articulated-Rotor Helicopters," Journal of Aircraft, Vol. 16, Jan. 1979, pp. 15-22.

Kato, K. and Yamane, T., "Calculation of Rotor Impedance for Articulated-

Rotor Helicopters in Forward Flight," Journal of Aircraft, Vol. 16, July 1879, pp,

470-476.
- 66 -
 A ^

Kato, K. and Yamane, T., "Experimental Substantiation for Hovering Rotor

Vertical Impedance Calculations," Journal of Aircraft, Vol. 18, June 1481, pp.
445-450,

Kaza, K.R.V. and Kvaternik, R.G., "A Critical Examination of The Flap-Lagi
Dynamics of Helicopter Rotor Blades in Hover and Forward Flight," 32nd Annual
Forum of the American Helicopter Society, Paper No. 1034, May 1976.
Kvaternik R.G. and Kaza, K.R.V., "Nonlinear Curvature Expressions for fi
4 A

Combined F'lapwise Bending, Chordwise Bending, Torsion, and Extension of Twisted

Rotor Blades," NASA TM X-73 1 997, 1976. 11p
F

Loewy, R.G., "A Two-Dimensional Approximation to the Unsteady Aerodynamics

of Rotary Wings," Journal of the Aeronautical Sciences, Vol. 24, Feb. 1957, pp.
81-92.

Murthy, V.R. and Pierce, G,A., "Effect of Phase Angle on Multibladed Rotor
Flutter, "Journal
Flutter J l of Sound and Vibration Vol. 48s No. 2 r1976 , PP• 221-234.

Murty, A.V.K. and Raman, A., "Nonl%near Dynamic Analysis of Rotors by

Finite Element Method," Journal of Sound and Vibration, Vol. 69, No. 4, 1980,
pp. 559-568.

Nagaraj.V.T. and Shanthakumar P., "Rotor Blade Vibrations by the Galerkin

Finite Element Method ^, " Journal of Sound and Vibration
^ Vol. 43
a No. 3 ► 1975 f

. PP• 575-577.

Ormistion, R.A. and Hodges, D.H., "Linear Flap-Lag Dynamics of Hingeless

Helicopter Rotor Blades in Hover," Journal of the American Helicopter Society,
Vol 17, Apr. 1972, pp. 2-14.

Peters, D.A. and Gaonkar, G.H., "Theoretical Flap-Lag Damping With Various
dynamic Inflow Models, Journal of the American Helicopter society, Vol 25, July
1080, pp. 29-36.

— 67 —
 I Rip RN7

Futter, S. and Manor, H., "Natural Frequencies of Radial Rotating Beams,"

Journal of Sound and Vibration, Vol. 56, No.2, 1078, pp. 175-185,
Rosen, A. and Friedmann, P., "Nonlinear Equations of Equilibrium for Elastic
Hellicopter or Wind Turbine Blades Undergoing Moderate Deformations," NASA
CR-159478, Dec, 1978.
Sharpe, D.L., "An Experimental Investigation of the Flap-Lag-Torsion Aeroelastic
Stability of a Small-Scale Hingeless Helicopter Rotor in Hover," NASA TP (to be
published in 1983).
Straub, F.K. and Friedmann, F.P., "A Galerkin Type Finite Element for
Rotary-Wing Aeroelasticity in Hover and Forward Flight," Vertica, Vol 5, 1981,
pp, 75-98.
Timman, R. and Van de Vorren, A.I., "Flutter of a Helicopter Rotor Rotating
in Its Own Wake," Journal of the Aeronautical Sciences, Vol. 24, Sep. 1957, pp,
694-702.

Warmbrodt, W. and McCloud III, J.L., "A Full-Scale Wind Tunnel Investigation
of a Helicopter Bearingless Main Rotor," NASA TM 81321, Aug. 1981.

Yasue, M., "Gust Response and Its Alleviation for a Hingeless Helicopter Rotor
in Cruising Flight," ASRL-TR- 189-1, Dept. of Aeronautics, Massachusetts Institute
of Technology, Cambridge, Mass., Sep. 1978.

- 68
 Table 1. Values of parameters of uniform blade for numerical results

Parameter Numerical Value

EIe /mn2 R4 0.014450(w„ +• 1.15)

Elr/mf22R4 0.026655 (ww 0,7)

0.166008(wv 1,5)
i
Y

CJ/mf22R4 0.000025(w+ = 2.5)

0.006661(w# •• 5)

km, /R 0

king lR 0.025

kA/km 1.5

c/R x/40

4 0.05 rad

a 0,1

ry 5

a 6

ca 0

Cl 6

do 0.0035

dl 0

Ctif.. 0

ki, 1.15

-1—
 Table 2. Steady tip deflections of a hingeless rotor blade

Number of uog,,/R W011P It

elements

2 -0.00350 0,00487 -0.04388

3 -0.00342 0.00469 -0.04334

4 -0.00338 0.00446 -0.04315

5 -0.00337 0,00440 -0.04307

6 -0,00330 O.^IW436 -0.04302

7 -0,00335 0,00434 -0.04299

8 -0,00335 0.00433 -0.04297

(CT10 m 0,1, Op — 0.06rad, y 5.0, w„ 1.512, w. m 1.15n, w1 m 2.512)

_1,
Tabi p Fundamental Coupled natural frequencies of a hingeless rotor blade

Number of W, /f2 w„ /A WO/11

elements

2 1.5196 1.1251 2.6106

3 1.6184 1,1221 2,4872

4 1.5182 1.1214 2.4787

5 1.6181 1,1212 2.4748

0 1.6181 1.1211 2.4724

7 1.5181 1.1211 2.4710

8 1.5180 1.1210 2.4702

(CT /o -0.1, Qi - 0.05rad, y-5.0,w„- IM,w W - 1.1512, w#" 2.512)

 Table 4. Real prxts of tht

Number of Lead-lab Flap Torsion

modes

3 -0.03074 -0131488 -0.35148

4 -0103049 -0,31383 -0.35049

b -0,03034 -0,31443 -0,35207

6 -0.03034 -0,31442 .0.35206

7 -0,03034 -0,31448 -0.35209

(CT /o — 0.1, Pp — 0.05 rad, I m 5,0, wy 1.5A, w„ — 1.151), w# = 2,5f2)

Lilt
 'x
ii^

Table 5. Steady tip deflections of an articulated rotor blade

Number of v004 /R w014P /R

elements

2 -0.06040 0.01274 -0.04476

3 -0.06001 0,01238 - 0.04425

r
4 -0.0.5987 0.01224 -0.04405

b -0.05980 0.01217 -0.04396

6 -0.05977 0.01212 -0.04391

7 -0.05975 0.01210 -0.04388

8 -0.05973 0.01208 -0,04386

(CT /o s 0. 1, Pp — 0.06 rad, y = 5.0, Hinge offset s 0.06R, wo 2XII)

"I,,

L_
 4

t
i

'table 6. Fundamental Coupled natural frequencies of an articulated rotor blade

'f(
1
Number of w./n ww/A wO/n
elen"ats

2 0.2999 1.0440 2.6285

3 0.2999 1.0440 2.6049

4 0.2909 1.0440 2.4963

,5 0,2999 1.0440 2.4923

6 0.2999 1.1440 2.4901

7 0.2999 1.0440 2.4889

8 0.2999 1.0440 2,4878

(C;r,'o Lai 0. 1, Ap 0.05 rad, 7 5.0, Hinge offset 0.06R, ur#2.5(2)

 0

Table 7. Real parts of the Butter eigenvalues of an articulated rotor blade

Number of Lead-lag Flap Torsion

modes

3 -0.00055 -0.33926 -0.39160

4 -0,00951 -0,34108 -0.39338

5 -0.00957 -0.34498 -0.39471

6 -0.00953 -0.34232 -0,39448

7 -0.00953 -0.34230 -0.39449

(CT Ia = 0. 1, Pr - 0.05rud, q = 5.0, Hinge offset = 0.06R, wa = 2.511)

r
{

1 1

1
{
 a }

Table 8. Structural properties of a bearingless blade

Property Outboard blade Each eexbeam Torque tube

Ely/mon 2 R4 0.026855 0.013328 01001

0.014486 0.007243 0.001

Els/moA2R4 0.166908 0.083454 0.001

GJ/m0 2 0.000925 0.000463 0.02'

0.005661 0.002830 0.02

k,,,, /R 0 0 0

km s /R 0.025 0.025 0.01

k A /km 1.5 1.5 1

m /mo 1 0.5 0.1

 Table 9. Steady tip deflections of a twinbeam model
(Zero inboard pitch)

Number of elements
r^
F lexbeam Flexbeam Outboard uo,,,/R vo,r,/R wo,,,/R ^olrp
one two blade

J
1 1 2 0.001613 -0.003750 0,011214 .0.030734
t'
2 2 1 0.001614 -0.003794 0.011289 -0.030634

2 2 2 0.001613 -0.003750 0.011213 -0.030735

2 2 3 0.001613 -0.003747 0.011206 -0.030732

l
3 3 2 0.001613 -0.003750 0.011213 -0.030735

(Cr10=0.1,pp-0.05rad , ry=6A,w„ = 1.870, w, —1.1512, we=2.9112)

y^

1
 r

Table 10. Fundamental natural frequencies of a twinbeam model

(Zero inboud pitcb)

Number of elements

Flexbeam Flexbeam Outboard w„ p ww/fl WO/fl

one two blade

1 1 2 1.77973 1.15018 2.88149

2 2 1 1.78825 1.15027 2.89491

2 2 2 1.77962 1.14977 2.88144

2 2 3 1.77877 1.14971 2,88094

3 3 2 1.77961 1.14974 2;88141

(CT/o= 0.1, pp = 0.05rad, 7 = 5.0, w, = 1.870, ww = 1.15(2, w# = 2.91(2)

Table 11. Real parts of the flutter eigenvalues of a twinbeam model
(Zero inboard pitch)

Number of Lead-lag Flap Torsion

males
i

3 0.00506 -0.35029 -0.38113

4 0.00450 -0.35228 -0.38094

5 0.00445 -0.35267 -0.38119

0 0,00445 -0,35267 -0.38119

( CT Io 0.1, Ar — 0.05 rad, y - 5.0, w, — 1.87n, ww = 1.15n, w# — 2.91n)

a
i
av
a

9
9
i
 w

{I

Table 12. Fundamental natural frequencies at various thrust 'levels

(Zero inboard pitch)

Lead -lag Flap Torsion

Twin Egvlt. Twin Egvlt. Twin Egvlt.

0 1.870 1.866 1.149 1.149 2.910 2.910

0.0,E 1.864 1.886 1.152 1.142 2.921 2.400

0.08 1.851 1.896 1.157 1.129 2.930 2.887

0.12 1.830 1.894 1.161 1.112 2.934 2.875

0.16 1.805 1.878 1.163 1.001 2.930 2.864

0.2 1.775 1.849 1,162 1.066 2.916 2.854

_j

j
z
Y
 C"ONALPA
OF pOOR'QU

LAG HINGE

HUB
PITCH
BEARING

00"o
OA e 9.

ROTOR
SHAFT ROTOR
TO CONTROL BLADE
SYSTEM

Fig. I Schematic of an articulated rotor hub and root.

L.J
BLACK AND WHITF, PHOTOGRAPIR

Fig. 2 Bearingless Main Rotor (BMR)

 E

ORIGINAL PAGE IS
OF POOR QUALITY

*I- BEAM FLEXURE

TRAILING
E
EDGE BEAM

i CLEVIS

Fig. 3 Flexure of BMR.

}

1
 ORIGINAL PAGE IS
OF POOL QUALITY

AXIS OIL
ROTATION
n 17

z^

i
."
.07 x

Fig. 4 Blade coordinate systems and deflections

41

E
 ORIGINAL PAG6 I
or POOR QUALIT

t
IZWc
V
^.^. M c ^ a P

Mac ,^ T
L c

Fig.5 Deformed-blade-sectii
 a
i^.

is

OR1G114At" R^A
OF POO R i.ITY

Y.

ith ELEMENT

V1 v7
v'4 0 Xi
v2
W 1 —► NnDE 1 , NODE 2 W2
W W2
1 2 rl

Fig. 6 A finite element showing nodal degrees of freedom

J
 ORIGINAL PAGE i$
OF POOR QUALITY

.12
.10 --- PRESENT
• HODGES AND
.08 ORMISTON (1976)

.06
a.
04
9 .
.02

-.02 (a)

-.04

d -.02
P -.OAS
(b)
-. 06

-.01
— tc)
-.020
.1 .2 .3 .4 .5
0, red
0 .02 .06 .10 .15 .20 .25'
CT/a

Fig. 7 (a) Flap bending (b) lead-lag bending (c) torsion. Steady
tip defections of a stiff-inplane hingeless rotor blade; w„
1.513, ww = LIM, W O 5.00, ry ,5-0, Op = 0.05 rad. x

a
a
 ORIGINAL 'AGE 15
OF POOR QUA LITY.

PRESENT
HODGES AND ORMISTON (1976)'
.16

.12

IL .06

.04

-.04

IL -.04

-.08
(b)
-.12

'U.-.01

-.02
^ 64
-.03 0 .4 .5
.1 .2 .3
B, red

0 .02 .06 .1 .15 2 .25 .3 Y ^

CT/O
6

r 1

Fig. 8 (a) Flap bending (b) lead-lag bending (c) torsion. Steady
tip deflections of a soft-inplane hingeless rotor blade; w„ —
0.7n, ww a LIM, wo = 5.0 2, 7 5.0, flp =
0,05 rQd. s
`F
 a •,,
yk^^rre^:,1^i,
^^ Poc),

Ually

2.6

2.2 TORSION. LAG

CT/a

E 1.4 0 -1-2 ^8

1.0
F LAP

.6 i

-.8 -.6 -A -.2 0 .2 .4

R9 X
a
9
i

Fig. 9 )soot locus plot for a ping

ww = 1.15n, w^ _ 2.512, y
 .30 .5
.25
.4
,2Q

,15 ,3

,10
.2
.06
.1
.02

0 0
.I
WV

Fir. 10 Stabilitv boundaries for a hin¢eless rotor blade: w,,, _

9
 O'kCINN", 'k. ill
I

rnwativ i
0 EXPERIMENT
-0

-5

-4

,-Z -3
E
-2

-1

0
-1
0. d"

Fig. 11 Comparison of hingelesss rotor bla(

tal results.
 ORIGINAL PALM 19
OF POOR QUALITY

110-.

LL .05

0
-.05

wr

-.20-

-.25

a if
f
i

-.05
(C)
0 .04 .08 .12 .16 .20 .24
CT/v

Fig. 12 a) Flap bending, b) lead-lag bending c) torsion. Steady

tip deflect-ions of an articulated rotor blade; hinge offset=0.06R,
2.5f2, 5.0t flp — 0.05 rad.

A
 _ ^ v

v,

I^

O Rt'Go'NA
Poon , p g *^ t
UALITY
t

2,'

2.

2.

2.i

1.J

1.1

1.1
0
E
1.9

^.0

.s

.6

.4

.2

0
- •5 -.5 -4 -3 -.2 -.1 0
RI X

1
Fig. 13 Root locus plot for an articulated rotor blade; hinge offset=0.06R,
w^ = 2.512, a 5.0, #p 0.05 rod.
i
 ORIGINAL MOO 1
OF POOR QUALITY

TORQUE TUBE
-
1-12

FLEXBEAkS
CLEVI!

Fig, 14 Analytical model of a bo

 OF 1pE^
'fin 4

I
2
2
2
r2

wi wZ
U3 04 u b
;2 J

r t",
;^ t

Fig. 15 Refined finite element for multiple-load-path blades. 3

a

P {

T
 ORIGINAL PAGI" IS
OF POOR QUALITY

e,C
PITCH DISTRIBUTION
AFTER COLLECTIVE
go e
00400

T
elm
B

CLEVIS PRE-TWIST
DISTRIBUTION
1
0 .25 .50 .75 1
x/R

Fig. 16 Pitch distribution of a bearingless rotor blade.

 n -----^^__-- ---

441'.0
OF PoOR
QfVALITY

^ |
; r71
`
BEAM I \
'

Lf

BEAM 2

Fig. 17 Twinbcam multiple-load-path blade model.

`
 ORIGINAL PAGE 13
OF POOR QUALITY

1.9

0 .1 .2
1.8

EQUIVALENT - BEAM MODEL

04
E 1.7

1.6 k 4. TWIN - BEAM MODEL

1.5
I
-.01 L
0 .01 .02 .03 .04
RR x

Fig. 18 Root locus plot of lead-lag mode of stiff-inplane, soft-in-

torsion twinbearn model (zero inboard pitch); w,, 1.8712,
WW — 1,1513 t wo — 2.9111 p5.0, Op 0.05 ►ad.
 O RIC'NAL q
i' '
OF POOR QUALITY

19

0 .1 .2
1.8 CT/a

E 1.7

1.6
—°-- TWIN-BEAM MODEL
♦ EOUIVALENT • 6EAM 'MODEL

1.6
–,01 0 .01 .02 .03
RR

i
r
Fig. In Root locus plot of lead-lag mode of stiff-nplane, soft-
in-torsion twinbeam model (zero inboard pitch; coupling
structural stiffness terms of nonlinear origin suppressed
in flutter analysis); wv s 1 " 87f2, wtu l .15f1, ,wd = 2,91 t2,
ry = 5,0, /9p ,ri d 5 r'ad.
i
rt

i
1

C ^
 ORIGINAL P"k3S I""
OF POOR QUALITY

,,a I

1.1
CT/O

0 .1 .2

1.0

^c EQUIVALENT-BEAM MODEL
E

.9
TWIN-BEAM MODEL

.9L -.02 -.01 0

-.04 -.03
RR X

a
k

Fig. 20 Root locus plot of lead-lag mode of soft-inplane, soft-in-

torsion twinbeam model (zero inboard pitch); wy 0.90il,
ww = 1.1512, wo : 2.91 t1, '1 5.0, #p = 0.05 rad.

L
r

lb- `._,. . _.__.^_^...^.y...^mW_<..^...._.,..._d:w.._e.^...._.__.^Y. .._. Vic.. ,. J...«:.._.__ _.....__e.....^_.^

 1.1

1.0

rc
E

.4

.8
—.t

M'

Fig. 21
 ORIGINAL PAGE 13
OF POOR QUALITY

ti

1.9

CT/U
1.8
t
0- OP

0 .2

1.7
EQUIVALENT-BEAM
MODEL
TWIN-BEAM MODEL \
1.6
-.02 -.01 0 .01
R2 X

Fig. 22 Root locus plot of lead-lag mode of stiff-inplane, stiff-in-

torsion twinbeam model (zero inboard pitch); wv — 1.870,
ww — 1. 150 ? wo — 5,38f2, 5.0, Op — 0.05 rad.
 ORIGINAL RASE 13
OF POOR QUALITY

^.
t

TWIN-BEAM MODEL
A EQUIVALENT-BEAM MODEL
1.9

1.8
:T/°
E
.1 .2
1.7

1
-.01 0 .01
R9 ).

Fig. 23 Root locus plot of lead-lag mode of stiff-inplane, stiff-in-

torsion twinbcam model (zero inboard pitch); coupling
structural stiffness terms of nonlinear origin suppressed
in Rutter analysis); w,, =1..$712, ww _ 1.1512, wo 5.3812,
7 5.0, Op = 0.05 rad.
 2.2

2J

2.1

2.1

1.1

i.

Fig.
 1.

A
E .9

.8
-.w _.U4 -.02 0
RQ
i

Fig. 25 Root locus plot of lead-lag mode of soft-inplane, soft-

in- torsion twinbeam model (same pitch for inboard
and
outboard); w„ = 0.90f2, ww = 1.15n, wo = 2.91f3, 7 = 5.0
_#p = 0.05 rad.

is
1
}
t
 ORIGINAL PAGE 1
OF POOR QUALITY

20

\t3U1VALENT-BEAM MODEL

J1.91 CT/Cr
0 .1 .Z
TWIN-BEAM MODEL

1.8
-.08 -.06 --.04 -.03 -.02 -.01 0
RA X

' 7

s {
Fig. 26 Root locus plot of lead-lag mode of stiff-inplane, stiff-
in-torsion twinbeam model (same pitch for inboard and
outboard); w„ = 1.8712, ww = 1.1512, w#5.3812, ry,= 5.0,
PP 0.05 rad. i

yy
 3.1

3.0

2.9

1.9

E 1.E

1.1

1.S

1.

1:

Fig. 27

View publication stats