Professional Documents
Culture Documents
Ranjan Ganguli
Vijay Panchore
The Rotating
Beam Problem
in Helicopter
Dynamics
Foundations of Engineering Mechanics
Series editors
V.I. Babitsky, Loughborough, Leicestershire, UK
Jens Wittenburg, Karlsruhe, Germany
More information about this series at http://www.springer.com/series/3582
Ranjan Ganguli Vijay Panchore
•
123
Ranjan Ganguli Vijay Panchore
Department of Aerospace Engineering Department of Aerospace Engineering
Indian Institute of Science Indian Institute of Science
Bangalore, Karnataka Bangalore, Karnataka
India India
v
vi Preface
Finally, the last chapter gives some numerical results for a typical helicopter rotor
blade. Frequency analysis, blade response calculation, and stability analysis are
presented.
This book will help graduate students and researchers to understand the typical
derivations and solution methods used in aeroelastic analysis of helicopter blades.
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 1
1.1 Free Vibration of a Single-Degree-of-Freedom System . . . . .... 2
1.2 Free Vibration of a Damped, Single-Degree-of-Freedom
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 3
1.3 Forced Vibration of a Single-Degree-of-Freedom System . . .... 5
1.4 Forced Vibration of a Damped, Single-Degree-of-Freedom
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 Two-Degrees-of-Freedom System . . . . . . . . . . . . . . . . . . . . . . . 8
1.6 Free Vibration of a Continuous System . . . . . . . . . . . . . . . . . . . 11
1.7 Hamilton’s Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.8 Diagonalization of a Symmetric Matrix . . . . . . . . . . . . . . . . . . . 20
1.9 Transformation of Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.10 Momentum Theory for Axial Flight . . . . . . . . . . . . . . . . . . . . . . 23
1.11 Momentum Theory for Forward Flight . . . . . . . . . . . . . . . . . . . . 25
1.12 Newton–Raphson Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.13 Blade Element Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.14 Derivation of Equation of Motion of Flapping Rigid Blade . . . . . 31
1.15 Derivation of Elastic Rotor Blade Equation . . . . . . . . . . . . . . . . 36
2 Finite Element Analysis in Space . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.2 Finite Element in Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.3 Strong Form of the Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.4 Weak Form of the Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.5 Galerkin’s Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.6 Shape Function in 1 Dimension . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.7 Shape Function Formulation for Beam Element . . . . . . . . . . . . . 44
2.8 Properties of Shape Function in 1D . . . . . . . . . . . . . . . . . . . . . . 46
2.9 Finite Element Formulation of Rotating Beam . . . . . . . . . . . . . . 48
2.10 Centrifugal Force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
vii
viii Contents
Prof. Ranjan Ganguli obtained his M.S. and Ph.D. in Aerospace Engineering
from the University of Maryland, College Park, in 1991 and 1994, respectively, and
his B.Tech. in Aerospace Engineering from the Indian Institute of Technology,
Kharagpur, in 1989. Following his Ph.D., he worked at the Alfred Gessow
Rotorcraft Center of the University of Maryland as Assistant Research Scientist
until 1997 on projects on rotorcraft health monitoring and vibratory load validation
for the Naval Surface Warfare Center and United Technology Research Center,
respectively. He also worked at the GE Research Laboratory in Schenectady, New
York, and at Pratt and Whitney, East Hartford, Connecticut, from 1997 to 2000. He
joined the Aerospace Engineering Department of the Indian Institute of Science,
Bangalore, as Assistant Professor in July 2000. He was promoted to Associate
Professor in 2005 and to Full Professor in 2009. He is currently the Satish Dhawan
Chair Professor at the Indian Institute of Science, Bangalore. He has held visiting
positions at TU Braunschweig, University of Ulm, and Max Planck Institute of
Metal Research, Stuttgart, Germany; University Paul Sabatier and Institute of
Mathematics, Toulouse, France; Konkuk University, South Korea; the University of
Michigan, Ann Arbor, USA; and the Nanyang Technological University,
Singapore. Professor Ganguli’s research interests are in helicopter aeromechanics,
aeroelasticity, structural dynamics, composite and smart structures, design opti-
mization, finite element methods, and health monitoring. He has published 178
articles in refereed journals and over 100 conference papers. He received the
American Society of Mechanical Engineers (ASME) best paper award in 2001, the
Golden Jubilee award of the Aeronautical Society of India in 2002, the Alexander
von Humboldt fellowship in 2007, and the Fulbright Senior Research fellowship in
2010. Professor Ganguli is a Fellow of ASME; a Fellow of the Royal Aeronautical
Society, UK; a Fellow of the Indian National Academy of Engineering; a Fellow
of the Aeronautical Society of India; an Associate Fellow of the American Institute
of Aeronautics and Astronautics; and a Senior Member of the Institute of Electrical
and Electronics Engineers (IEEE). He has taught courses on flight and space
mechanics, engineering optimization, helicopter dynamics, aircraft structures,
structural mechanics, aeroelasticity, and navigation. He has supervised the thesis of
ix
x About the Authors
xi
xii List of Figures
xiii
Chapter 1
Introduction
Helicopters are important flight vehicles which are prone to high vibration caused
by an unsteady aerodynamic environment and highly flexible rotating blades. Such
high vibrations can result in damage to structural and avionics components, pas-
senger discomfort, and high maintenance costs. The prediction of vibration is
therefore a major problem in helicopter engineering. The rotating beam is the
fundamental mathematical model for the helicopter rotor blade. Rotor aeroelastic
analysis codes typically use rotating beam models, and knowledge of these struc-
tures is very necessary for the helicopter dynamics researchers. Typically, calcu-
lation of rotating frequencies, blade response, and aeroelastic stability are the key
components of helicopter aeroelastic analysis. Thus, numerical methods for solving
these problems are a key tool. Specifically, the rotating beam equation is an
important model for helicopter dynamics and is a major pedagogical tool. In this
chapter, background material for the development of the rotating beam equation is
presented. Vibration terminology which is used in this book is explained, and the
aerodynamic forces acting on the rotor blade are investigated. The rotor blade
equations for a flapping blade and the equation for a rotating beam are then derived.
These equations are key to a sound understanding of helicopter rotor dynamics.
The governing partial differential Eq. (1.1) of an elastic rotor blade in flap
bending does not have an analytical solution. The numerical solution involves
discretization in space and time. We will spend a large part of this book in deriving,
analyzing, and solving this equation:
2 3
2 Z
R
@2 @2w @2w @w @ w
EI 2 þ m 2 þ X2 4mx mxdx5 ¼ Fðx; tÞ ð1:1Þ
@x2 @x @t @x @x2
x
Here, EI is the flexural stiffness of the blade, m is the mass per unit length, X is
the rotating speed of the blade, w is the flap bending deflection, F is the external
force, R is the blade radius, and x and t are the spatial and time coordinates.
m€x þ kx ¼ 0 ð1:2Þ
x_ ð0Þ
x ¼ xð0Þ cosðxn tÞ þ sinðxn tÞ ð1:3Þ
xn
qffiffiffi
where xn ¼ k
m rad/s is the natural frequency of the spring–mass system. The
response of the spring–mass system is given in Fig. 1.2.
the response.
Consider Fig. 1.3, which shows the damped vibration case. Here, c represents a
viscous damper.
Governing equation for the spring–mass–damper system is given by
Assume the solution x ¼ Aest and substitute it into Eq. (1.5) yields
or
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c 2 kffi
c c2 4mk c
s¼ )s¼
2m 2m 2m m
or
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c 2
c
s¼ x2n ð1:7Þ
2m 2m
or
pffiffiffiffiffiffiffi
ffi pffiffiffiffiffiffiffi
ffi
x ¼ efxn t C1 exn t f 1 þ C2 exn t f 1 :
2 2
ð1:9Þ
or
qffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffi
fxn t
x¼e C1 sinðxn 1 f tÞ þ C2 cosðxn 1 f2 t
2
ð1:10Þ
1.2 Free Vibration of a Damped, Single-Degree-of-Freedom System 5
Consider Fig. 1.4, where curve (a) shows an over-damped system, curve
(b) shows a critically damped system, and curve (c) shows an under-damped sys-
tem. Damping ensures that the system response dissipates after some time. Adding
damping to a system is useful for reducing vibration.
Consider Fig. 1.5, which shows the forced vibration case. Here, an external force
f ðtÞ is applied to the mass-spring system.
The governing equation is given by
m€x þ kx ¼ f ðtÞ
m€x þ kx ¼ P0 sinðxtÞ ð1:12Þ
6 1 Introduction
x ¼ xh þ xp
þ A2 sinðxtÞ
xp ¼ A1 cosðxtÞ ð1:14Þ
P0
Substituting xp in Eq. (1.12), we get A1 ¼ 0 and A2 ¼ k
:
1x 2
2
xn
x ¼ xh þ xp
or
P0
x ¼ C1 cosðxn tÞ þ C2 sinðxn tÞ þ sinðxtÞ
ð1:15Þ
2
k 1x x 2
n
or
P0
x ¼ C1 cosðxn tÞ þ C2 sinðxn tÞ þ
sinðxtÞ ð1:16Þ
kð1 b2 Þ
1.3 Forced Vibration of a Single-Degree-of-Freedom System 7
where b ¼ xxn (frequency ratio). Recall that xn is the natural frequency of the
spring–mass system and x is the forcing frequency. The natural frequency of the
system is its intrinsic property which depends on stiffness (k) and mass (m) only.
From Eq. (1.16), we see that the response becomes infinite when b ¼ 1 or x ¼ xn :
This condition is known as resonance.
Consider Fig. 1.6, which shows the forced vibration of the mass–spring–damper
is applied to the system.
system. Here, an external force f ðtÞ ¼ P0 sin xt
The governing equation is given by
m€x þ c_x þ kx ¼ PO sinðxtÞ ð1:17Þ
x ¼ xh þ xp
þ A2 sinðxtÞ
xp ¼ A1 cosðxtÞ ð1:19Þ
x ¼ xh þ xp
or
qffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffi
x ¼ efxn t C1 sinðxn 1 f2 tÞ þ C2 cosðxn 1 f2 t
P0 ð1:20Þ
þ 1 b2 sinðxtÞ
2fb cosðxtÞ
2 2 2
k 1 b þ ð2fbÞ
qffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffi
P0
x ¼ efxn t C1 sinðxn 1 f2 t þ C2 cos xn 1 f2 t þ D uÞ
sinðxt
k
ð1:21Þ
where D ¼ 1
1=2 and / ¼ tan1 : 2fb
1b2
ð1b Þ þ ð2fbÞ
2 2 2
Here, D is the dynamic amplification factor. We can see that when f ! 0 and
b ! 1; D ! 1, a condition known as resonance where the amplitude of the
system increases in an unbounded manner.
Figure 1.7 shows how the dynamic amplification factor varies with the fre-
quency ratio.
We see in Fig. 1.7, as the damping factor increases, the amplitude of the
dynamic amplification factor decreases. Adding damping to a system results in
amelioration of the resonance peak. Also, it is advisable to keep the frequency ratio
away from one. Considerable effort is spent in the development of dampers and for
frequency placement for vibration reduction.
Consider Fig. 1.8a, a two-degrees-of-freedom system. There are two masses m1 and
m2 ; two springs k1 and k2 ; and two dampers c1 and c2 : A force P1 is applied to mass
m1 , and a force P2 is applied to mass m2: The degrees of freedom are the coordinates
1.5 Two-Degrees-of-Freedom System 9
Fig. 1.7 MATLAB plot of “dynamic amplification factor versus frequency ratio”
x1 and x2 which represent the motion of the two masses. In Fig. 1.8b, we draw a
free-body diagram for each mass and then write the equations of motion.
We write the governing equation as
We write the free vibration equation by ignoring the damping term and the force
vector
m1 0 €x1 k þ k2 k2 x1 0
þ 1 ¼ : ð1:23Þ
0 m2 €x2 k2 k2 x2 0
x1 C1 sinðxt /Þ
¼ :
x2 C2 sinðxt /Þ
or
k1 þ k2 k2 k1 k2
x4 þ x2 þ ¼0
m1 m2 m1 m2
or
( )1=2
1 k1 þ k2 k2 1 k1 þ k2 k2 2 k1 k2
k¼x ¼ 2
þ þ 4 : ð1:25Þ
2 m1 m2 2 m1 m2 m1 m2
ð1Þ
C1 K2 m2 x21 þ K2
r1 ¼ ð1Þ
¼ ¼ ð1:26Þ
C2 m1 x21 þ K1 þ K2 K2
1.5 Two-Degrees-of-Freedom System 11
ð2Þ
C1 K2 m2 x22 þ K2
r2 ¼ ð2Þ
¼ ¼ : ð1:27Þ
C2 m1 x22 þ K1 þ K2 K2
where EI is the flexural stiffness, m is the mass per unit length, and w is the
deflection in the transverse direction.
For constant EI; we get
@4w
EI þ m€
w¼0 ð1:32Þ
@x4
ðxÞeixt ; we get
Assume the solution wðx; tÞ ¼ w
12 1 Introduction
ðxÞ ixt
d4 w
EI e w2 m
wðxÞeixt ¼ 0
dx4
ðxÞ
d4 w
EI x2 m
wðxÞ ¼ 0 ð1:33Þ
dx4
ðxÞ mx2
d4 w
ðxÞ ¼ 0
w ð1:34Þ
dx4 EI
mx2 px mx2
p4 epx e ¼ 0 ) p4 ¼
EI EI
or
p4 ¼ k4
1=4
mx2
where k ¼ EI
@ 2 wðxÞ @ 2 wðxÞ
wð0Þ ¼ 0; EI ðx ¼ 0Þ ¼ 0; wðLÞ ¼ 0; and EI ðx ¼ LÞ ¼ 0:
@x2 @x2
0 C1 þ 1 C2 þ 0 C3 þ 1 C4 ¼ 0
0 C1 þ 1 C2 þ 0 C3 1 C4 ¼ 0
Solving, we get
C3 sinðkLÞ ¼ 0
or
2 1=4 rffiffiffiffiffiffiffiffiffi
mx EI
kL ¼ np ) L ¼ np ) x ¼ n2 p2
EI mL4
Substituting r ¼ n, we get
rffiffiffiffiffiffiffiffiffi
EI
xr ¼ r p
2 2
ð1:36Þ
L4 m
where x1 ; x2 . . .xr are the natural frequencies of the system and the respective
mode shapes are given by
rx
r ðxÞ ¼ C3 sin
w ð1:37Þ
L
Figure 1.10 shows the plots of the first three mode shapes.
The complete solution is given by
14 1 Introduction
Fig. 1.10 a First mode shape of simply-supported beam. b Second mode shape of
simply-supported beam. c Third mode shape of simply-supported beam
rpx
wðx; tÞ ¼ sin eixt : ð1:38Þ
L
@wðxÞ @ 2 wðxÞ @ 3 wðxÞ
wð0Þ ¼ 0; ðx ¼ 0Þ ¼ 0; EI ðx ¼ LÞ ¼ 0; and EI ðx ¼ LÞ ¼ 0:
@x @x2 @x3
Applying boundary conditions wð0Þ ¼ 0 and @wðxÞ
@x ðx ¼ 0Þ ¼ 0 to Eq. (1.35), we
get
C2 þ C4 ¼ 0 and C1 þ C3 ¼ 0:
or
and
nh x x i
3 ðxÞ ¼ C1
w sin h 7:8547 sin 7:8547
L L
cos hð7:8547Þ þ cosð7:8547Þ h x x i
cos h 7:8547 cos 7:8547
sin hð7:8547Þ sinð7:8547Þ L L
Figure 1.12 shows the plots of the first three mode shapes. These mode shapes
are also known as beam function for a cantilever beam and are used as basis
functions for approximate methods.
Simply-supported-free beam
Boundary conditions for a simply-supported-free beam are (Fig. 1.13).
@ 2 wðxÞ @ 2 wðxÞ @ 3 wðxÞ
wð0Þ ¼ 0; EI ðx ¼ 0Þ ¼ 0; EI ðx ¼ LÞ ¼ 0; and EI ðx ¼ LÞ
@x2 @x2 @x3
¼ 0:
wðxÞ
Applying boundary conditions wð0Þ ¼ 0 and EI @ @x
2
2 ðx ¼ 0Þ ¼ 0 to Eq. (1.35),
we get
C4 ¼ 0 and C2 ¼ 0:
cos hðkLÞ
C3 ¼ C1
cosðkLÞ
and
x cos hð10:2101Þ x
3 ðxÞ ¼ C1 sin h 10:2101 þ
w sin 10:2101
L cosð10:2101Þ L
Figure 1.10 shows the plots of the first three mode shapes. These are elastic
mode shapes (Fig. 1.14).
Hamilton’s principle is often used to derive the equation of motion for helicopter
rotor blades undergoing motion in several directions such as flap and lag bending,
axial, and torsion. It states that of all of the possible paths of a mechanical system,
the path actually followed is the one that minimizes the time integral of the dif-
ference between the kinetic and potential energies. That is, the actual path chosen
by the system is the one that makes the variation of the following integral vanish:
Z
d ðT UÞdt ¼ 0 ð1:45Þ
1.7 Hamilton’s Principle 19
Using the Lagrange’s Eq. (1.47), we get the equation of motion of a simple
pendulum
€h þ g sinðhÞ ¼ 0: ð1:48Þ
L
€ þ ½C½X
½M½X _ þ ½K½X ¼ ½FðtÞ ð1:49Þ
where ½v1 nx1 ; ½v2 nx1 . . .; &½vn nx1 are the eigenvectors of the matrix ½ A.
We know
or
2 3
k1 0 0 0 0
60 k2 0 0 07
6 7
½ Anxn ½Pnxn ¼ ½Pnxn 6
60 0 : 0 077 ð1:54Þ
40 0 0 : 05
0 0 0 0 kn nxn
1.8 Diagonalization of a Symmetric Matrix 21
or
2 3
k1 0 0 0 0
60 k2 0 0 07
6 7
½P1 6 07
nxn ½ Anxn ½Pnxn ¼ 6 0 0 : 0 7 ð1:55Þ
40 0 0 : 05
0 0 0 0 kn nxn
or
T
½ Anxn ½v1 nx1 ¼ k1 ½v1 Tnx1 ) ½v1 Tnx1 ½ ATnx1 ¼ k1 ½v1 Tnx1
or
½v1 Tnx1 ½ Anxn ½v2 nx1 ¼ k1 ½v1 Tnx1 ½v2 nx1 ) k2 ½v1 Tnx1 ½v2 nx1 ¼ k1 ½v1 Tnx1 ½v2 nx1
or
or
or
½P1 T
nxn ¼ ½Pnxn
2 3
k1 0 0 0 0
60 k2 0 0 07
6 7
½PTnxn ½ Anxn ½Pnxn ¼ 6
60 0 : 0 077 : ð1:57Þ
40 0 0 : 05
0 0 0 0 kn nxn
€ þ ½K½X ¼ ½FðtÞ
½M½X ð1:58Þ
or
Here,
Mi [ 0ði ¼ jÞ x2i Mi [ 0ði ¼ jÞ
½/i Tnx1 ½Mnxn ½/j nx1 ¼ ; ½/i Tnx1 ½Knxn ½/j nx1 ¼ ;
0ði 6¼ jÞ 0ði 6¼ jÞ
or
2 32 3 2 32 3 2 3
m1 0 0 0 0 €f1 ðtÞ k1 0 0 0 0 f1 ðtÞ f1 ðtÞ
6 0 0 7 6 €f ðtÞ 7 6 0 07 6 7 6 7
6 m2 0 0 76 2 7 6 k2 0 0 76 f2 ðtÞ 7 6 f2 ðtÞ 7
6 0 : 0 7 6 7 6 76
0 76 : 7 ¼ 6 : 7
7 6
6 0 0 76 : 7 þ 6 0 0 : 0 7
4 0 0 0 : 0 54 : 5 40 0 0 : 0 54 : 5 4 : 5
0 0 0 0 mn €fn ðtÞ 0 0 0 0 kn fn ðtÞ fn ðtÞ
ð1:61Þ
The forcing function vector for the helicopter rotor blade equation comes from
aerodynamics. We review basic helicopter aerodynamics in the next few sections.
Momentum theory is a simple approach which can predict rotor inflow and power.
In this theory, the helicopter rotor is modeled as an actuator disk. Momentum theory
uses the principle of linear momentum conservation and assumes incompressible,
irrotational, and steady flow.
Consider Fig. 1.16, where we take a control volume over the plane of the rotor.
Here, V1 is the flow velocity, Vu and Vl are the induced velocities at upper and
lower side of the rotor plane, respectively, Vf is the induced velocity at the slip-
stream, A1 ; A2 ; A3 , and A4 are the areas of the four stations 1, 2, 3, and 4, respec-
tively, and P1 ; P2 ; P3 , and P4 are the pressures at the four stations 1, 2, 3, and 4,
respectively.
We write the mass conservation equation as
ZZZ ZZ
@
qdv þ q~ ndA ¼ 0
V:~ ð1:62Þ
@t CS
CV
@ RRR
Since flow is steady @t CV qdv ¼ 0 ; we get the equation
ZZZ
q~ ndA ¼ 0
V:~ ð1:63Þ
CS
Since A2 ¼ A3 ; we get
@ RRR
Since the flow is steady ~
Vqdv ¼ 0 ; we get
@t CV
ZZ
T¼ q~
V ~ n dA ) T ¼ m_ ðVout Vin Þ
V:~ ð1:65Þ
CS
We write the Bernoulli equation between areas A1 and A2 and between areas A3
and A4 :
1 2 1
P1 þ qV1 ¼ P2 þ qðV1 þ Vu Þ2 ð1:67Þ
2 2
1 2 1
P4 þ q V1 þ Vf ¼ P3 þ qðV1 þ Vl Þ2 : ð1:68Þ
2 2
We write the thrust ðTÞ in terms of the pressure difference across the disk
q
T ¼ ðP3 P2 ÞA2 ) T ¼ A2 Vf2 þ 2Vf V1 ð1:70Þ
2
q 2
A2 Vf þ 2Vf V1 ¼ qA4 V1 þ Vf Vf ð1:71Þ
2
Vf ¼ 2Vu ð1:73Þ
or
w ¼ 2vi
T
v2h ¼ A2 ¼ A3 ¼ A ¼ pR2 ¼ Rotor disk area ð1:74Þ
2qA
Fig. 1.17 Glauert flow model for momentum analysis of a rotor in forward flight
or
T ¼ m_ ½w þ V1 sinðaÞ V1 sinðaÞ
or
_ ) T ¼ 2mv
T ¼ mw _ i ð1:76Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Here, m_ ¼ qAU and U ¼ ½V1 cosðaÞ2 þ ½vi þ V1 sinðaÞ2
We get
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
T ¼ 2qAvi ½V1 cosðaÞ2 þ ½vi þ V1 sinðaÞ2
or
T
vi ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2qA ½V1 cosðaÞ þ ½vi þ V1 sinðaÞ2
2
v2h
vi ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1:77Þ
½V1 cosðaÞ2 þ ½vi þ V1 sinðaÞ2
V1 cosðaÞ
l¼ ð1:80Þ
XR
T
CT ¼
qAX2 R2
where
2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3
2ffi
4 4 k k
Kx ¼ 1 1:8l2 1 þ 5
3 l l
Ky ¼ 2l:
The uniform inflow model is appropriate for hover and the linear inflow model
for forward flight. In reality, the wake is highly non-uniform, i.e., k ¼ kðr; wÞ, and
free wake models are needed for its accurate prediction.
28 1 Introduction
The uniform inflow model for forward flight requires a numerical solution
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
f ðkÞ ¼ k l tanðaÞ þ CT =2 l2 þ k2 ¼ 0
f ðk0 þ hÞ ¼ 0
h2 00
f ðk0 Þ þ hf 0 ðk0 Þ þ f ðk0 Þ þ Higher Order Terms ¼ 0
2
f ðk0 Þ
f ðk0 Þ þ hf 0 ðk0 Þ ¼ 0 ) h ¼
f 0 ðk0 Þ
Thus,
f ðk0 Þ
k1 ¼ k0 þ h ) k1 ¼ k0
f 0 ðk0 Þ
or
kn l tanðaÞ þ pffiffiffiffiffiffiffiffiffiffi
CT ffi
f ðkn Þ 2 22
l þ kn
kn þ 1 ¼ kn 0 ¼ kn ð1:83Þ
f ðkn Þ 1 CT k n
2ðl2 þ k2n Þ
3=2
qffiffiffiffiffi
CT
The hover value 2 can be used as the initial guess.
Momentum theory considers the rotor to be an actuator disk and is not able to
directly relate the blade section properties to the rotor thrust, power, etc. Blade
element theory is useful for deriving equations which can guide blade design.
In this theory, thrust is calculated for a small section of the blade and then
integrated over the length of the rotor blade to get the thrust produced by one blade.
1.13 Blade Element Theory 29
Fig. 1.18 a Blade element theory. b Blade element theory. c Blade element theory
Consider Fig. 1.18a, which shows the pitch angle ðhÞ; inflow angle ðuÞ; and
effective angle of attack ða ¼ h uÞ; where UT and UP are tangential and per-
pendicular components of flow velocity, respectively. Figure 1.18b shows an
infinitesimal section dy; dT is the thrust produced by this section. Figure 1.18c
shows the forces acting on the blade section. dL and dD are the lift and drag
produced by the section and are perpendicular and parallel to the resultant velocity
U; respectively. dFZ and dFX are the components of the force perpendicular and
parallel to UT :
We write the lift produced by the section dy
1 1
dL ¼ qU 2 cCl dy ) dL ¼ q UT2 þ UP2 cCl dy ð1:84Þ
2 2
1 1
dD ¼ qU 2 cCd dy ) dD ¼ q UT2 þ UP2 cCd dy ð1:85Þ
2 2
We write the forces dFZ and dFX in terms of lift and drag
or
1 1
dFZ ¼ q UT2 þ UP2 cCl dy cosð/Þ q UT2 þ UP2 cCd dy sinð/Þ ð1:87Þ
2 2
dFX ¼ dL sinð/Þ þ dD cosð/Þ
or
1 1
dFX ¼ q UT2 þ UP2 cCl dy sinð/Þ þ q UT2 þ UP2 cCd dy cosð/Þ ð1:88Þ
2 2
Here, dT ¼ dFZ .
Consider Fig. 1.19, where tangential and perpendicular components of the flow
velocity are shown.
where Xy is the result of angular velocity of rotor and V1 cosðaÞ sinðwÞ is the result
of flow velocity:
where V1 sinðaÞ is the result of flow velocity, vi is the induced velocity, yb_ is the
result of flap motion, and V1 sinðbÞ cosðwÞ cosðaÞ is the result of anhedral effect.
We write Eqs. (1.89) and (1.90) as
y
UT ¼ XR þ l sinðwÞ ð1:91Þ
R
!
yb_
UP ¼ XR k þ þ lb cosðwÞ : ð1:92Þ
XR
These are the tangential and perpendicular velocities in forward flight. The
expressions for velocities will be useful for deriving the helicopter rotor blade
equations.
1.14 Derivation of Equation of Motion of Flapping Rigid Blade 31
Consider Fig. 1.20, which shows a rigid rotor blade, hinged at the root. The blade
flaps up and down under the aerodynamic forces, centrifugal forces, and inertial
forces. A small element having mass dm is taken at a distance y from the center.
Here, b is the flapping angle, dL is the elemental lift force, X is the angular velocity,
dFZ is the force acting perpendicular to the rotor disk plane, and dCF is the
centrifugal force.
Each force is analyzed physically to form the equation of motion of the flapping
rigid blade.
(a) Inertia Force
€
Force on the small segment having mass dm ¼ ðybÞdm
€
Associated moment at the hinge ¼ yðybÞdm
€
Integrating all the small segments; moment due to full rotor ¼ I b ð1:93Þ
RR
where I ¼ dmy2 ; is the mass moment of inertia.
0
ZR
Integrating all the small segments; moment due to full rotor ¼ X2 y2 bdm ¼ IX2 b
0
ð1:94Þ
where ðy cosðbÞ yÞ
ZR
Integrating all the small segments; moment due to full rotor ¼ Lydy: ð1:95Þ
0
ZR
€ þ IX2 b ¼
Ib Lydy ð1:96Þ
0
Since w ¼ Xt
db db dw db
b_ ¼ ¼ ¼X
dt dw dt dw
€ db_ db_ dw db_ d db d2 b
b¼ ¼ ¼X ¼X X ¼ X2 2
dt dw dt dw dw dw dw
ZR
d2 b
IX2
2
þ IX2 b ¼ Lydy
dw
0
or
ZR
d2 b 1
2
þb ¼ 2 Lydy ð1:97Þ
dw IX
0
1.14 Derivation of Equation of Motion of Flapping Rigid Blade 33
Here, h is the pitch angle, c is the chord length, CLa is the lift curve slope, and UT
and UP are tangential and perpendicular components of the flow velocity.
1
L ¼ qcCLa hUT2 UP UT
2
ZR ZR
1 1 1
Lydy ¼ 2 qcCLa hUT2 UP UT ydy ð1:99Þ
IX2 IX 2
0 0
or
ZR ZR ZR
1 1 1 1 1
Lydy ¼ 2 qcCLa hUT2 ydy 2 qcCLa UP UT ydy
IX2 IX 2 IX 2
0 0 0
or
ZR
1
Lydy ¼ Term1 Term2 ð1:100Þ
IX2
0
RR 1 RR 1
Here, Term1 ¼ IX1 2 2 qcCLa hUT ydy and Term2 ¼ IX2
2 1
2 qcCLa UP UT ydy
0 0
We write
ZR
1 1
Term1 ¼ 2 qcCLa hUT2 ydy ð1:101Þ
IX 2
0
ZR 3
1 1 2 2 y 2y2 l sinðwÞ
Term1 ¼ 2 qcCLa X R h1 2 þ yl2 sin2 ðwÞ þ
IX 2 R R
0
4
y y2 l2 sin2 ðwÞ 2y3 l sinðwÞ
þ X2 R2 htw 3 þ þ dy
R R R2
or
We write
ZR
1 1
Term2 ¼ 2 qcCLa UP UT ydy ð1:103Þ
IX 2
0
ZR 2
1 1 y
Term2 ¼ 2 qcCLa X R k 2 2
þ yl sinðwÞ
IX 2 R
0
3
2
_ y y2 l sinðwÞ y l cosðwÞ
þb þ þb þ l y sinðwÞ cosðwÞ dy
2
XR2 XR R
1.14 Derivation of Equation of Motion of Flapping Rigid Blade 35
or
or
1 l sinðwÞ db 1 l sinðwÞ
Term2 ¼c k þ þ þ
6 4 dw 8 6
1 l sinðwÞ
þ bl cosðwÞ þ ð1:104Þ
6 4
ZR
1 l sinðwÞ db 1 l sinðwÞ
c k þ þ þ
6 4 dw 8 6
1 l sinðwÞ
þ bl cosðwÞ þ
6 4
or
ZR
1 b
Lydy ¼ cM ð1:105Þ
IX2
0
where
2
b ¼ h1 1 þ l sinðwÞ þ l sin2 ðwÞ
M
8 3 4
1 l sin ðwÞ l sinðwÞ
2 2
þ htw þ þ
10 6 4
1 l db 1 l
k þ sinðwÞ þ sinðwÞ
6 4 dw 8 6
1 l
bl cosðwÞ þ sinðwÞ
6 4
36 1 Introduction
d2 b b
þ b ¼ cM ð1:106Þ
dw2
X
N
bðwÞ ¼ b0 þ ðbnc cosðnwÞ þ bns sinðnwÞÞ ð1:107Þ
n¼1
Fourier series expansion for a given N is put into both sides of the equation, and
the harmonic coefficients are equated. This approach is called the harmonic balance
method.
Here, we derive the equation for free vibration of the rotor blade.
Consider Fig. 1.21a, which shows the deflection of an elastic blade.
Figure 1.21b shows the forces acting on a small section of the elastic blade. Here,
M is the bending moment, G is the centrifugal force, and S is the shear force acting
on the blade section.
We write the force and moment balance equations, considering the centrifugal
force and inertial force:
ZR
dG þ mX xdx ¼ 0 ) G ¼
2
mX2 xdx ð1:108Þ
x
@2w @S @2w
dS þ mdx ¼ 0 ) ¼ m ð1:109Þ
@t2 @x @t2
@M @w @ 2 M @S @ @w
Gdw þ Sdx dM ¼ 0 ) ¼ SþG ) ¼ þ G ð1:110Þ
@x @x @x2 @x @x @x
@2w
M ¼ EI ð1:111Þ
@x2
We write
00 R 1 1 0 1
Z 2 Z
R
@ @@ @w @w @ w
mX2 xdxA A ¼ X2 @mx þ 2 mxdxA ð1:115Þ
@x @x wx @x
x x
From Eqs. (1.114) and (1.115), we write the equation where we have shifted
from t to w for the time coordinate:
2 3
ZR
€ þ X2 4mxw0 w00
ðEIw00 Þ00 þ mX2 w mxdx5 ¼ 0 ð1:116Þ
x
38 1 Introduction
We can see that the forcing term for the elastic blade equation has periodic terms
such as sinðwÞ; sin2 ðwÞ and sinð2wÞ: These terms induce vibratory response and
loads in the rotor system.
For the uniform inflow model, we can write
Fðx; wÞ ¼ a1x a1w þ a2x a2w þ a3x a3w þ a4x a4w þ a5x a5w þ a6x a6w
ð1:120Þ
þ a7x a7w þ b1x b1w þ b2x b2w w_ þ c1x c1w þ c2x c2w w0
1.15 Derivation of Elastic Rotor Blade Equation 39
2.1 Introduction
In this chapter, finite element in space is discussed in detail, and it is typically the
first step in the solution of the elastic rotor problem as it yields the rotating natural
frequencies. Bar, beam, and rotating beam finite element formulation are explained.
d2 u
EA ¼ f0 ð2:1Þ
dx2
uð0Þ ¼ 0 ð2:2Þ
du
EA ¼0 ð2:3Þ
dx ðx¼LÞ
where u is the axial displacement; Young’s modulus ðEÞ and cross sectional area
ðAÞ are constant over the length of the bar.
The weak form is a variational statement of the equation, where we multiply the
differential equation by a test function ðvÞ and integrate it over the domain.
ZL
d2 u
v EA 2 f0 dx ¼ 0 ð2:4Þ
dx
0
We choose test function ðvÞ such that it satisfies the homogeneous boundary
conditions. Equation (2.4) can be written after integration by parts as
ZL ZL
du dv du L
EA þ EAv ¼ f0 vdx
dx dx dx 0
0 0
or
ZL ZL
du dv du du
EA þ EAvðLÞ EAvð0Þ ¼ f0 vdx
dx dx dx x¼L dx x¼0
0 0
We apply boundary condition (2.2), (2.3), and vð0Þ ¼ 0, to get the weak form of
the equation
2.4 Weak Form of the Equation 43
ZL ZL
du dv
EA ¼ f0 vdx ð2:5Þ
dx dx
0 0
The order of the derivatives in the equation is reduced in the weak form.
ZL
d2 ~u
v EA 2 f0 dx ¼ 0 ð2:6Þ
dx
0
~uð0Þ ¼ 0 ð2:7Þ
d2 ~u
EA ¼ f0 ð2:8Þ
dx2
P P
where we choose ~uðxÞ ¼ Nj¼1 cj /j ðxÞ and vðxÞ ¼ Ni¼1 bj /j ðxÞ. Here, cj is
unknown and bj is arbitrarily chosen. The interpolation or basis function /j ðxÞ must
satisfy all the boundary conditions for the problem. A good solution is obtained by
taking many terms of the series. While Galerkin’s method uses global interpolation
function, the key idea in finite element method is to interpolate locally.
Consider Fig. 2.2, two-node bar element. Here q1 and q2 are the displacements
at the two nodes and are called degrees of freedom.
44 2 Finite Element Analysis in Space
uðxÞ ¼ a0 þ a1 x ð2:10Þ
At x ¼ 0
uð0Þ ¼ a0 ) q1 ¼ a0 ð2:11Þ
At x ¼ L
uðLÞ ¼ a0 þ a1 L ) q2 ¼ a0 þ a1 L ð2:12Þ
vðxÞ ¼ a0 þ a1 x þ a2 x2 þ a3 x3 ð2:15Þ
At x ¼ 0
vð0Þ ¼ a0 ) q1 ¼ a0 ð2:16Þ
dvð0Þ
¼ a1 ) q 2 ¼ a 1 ð2:17Þ
dx
At x ¼ L
vðLÞ ¼ a0 þ a1 L þ a2 L2 þ a3 L3 ) q3 ¼ a0 þ a1 L þ a2 L2 þ a3 L3 ð2:18Þ
dvðLÞ
¼ a1 þ 2a2 L þ 3a3 L2 ) q4 ¼ a1 þ 2a2 L þ 3a3 L2 ð2:19Þ
dx
or
2 3 2 32 3
a0 1 0 0 0 q1
6 a1 7 6 0 1 0 07 6 q1 7
6 7 ¼ 6 3 76 7 ð2:21Þ
4 a2 5 4 2 2 3 1 54
q3 5
L L L2 L
2 1 2 1
a3 L3 L2 L3 L2 q4
or
2 32 3
1 0 0 0 q1
6
3 6 0 1 0 07 6 q1 7
vðxÞ ¼ 1 x x2 x 4 3 76 7
1 54 ð2:23Þ
2 L
2
L
3
L2 L q3 5
2 1 2 1
L3 L2 L3 L2 q4
46 2 Finite Element Analysis in Space
or
2 3
q1
h x 2 x 3 i6 7
vðxÞ ¼ 2 Lx 3 3 Lx 2 þ 1 x3 2
x3 x2 6 q1 7 ð2:24Þ
L2 2 xL þ x 3 L 2 L L2 L 4 q 3 5
q4
or
2 3
q1
6 q2 7
vðxÞ ¼ ½ H1 H2 H3 H4 6 7
4 q3 5 ¼ ½H ½q ð2:25Þ
q4
where H1 ; H2 ; H3 and H4 are shape functions for the beam finite element.
x 3 x 2 x3 x2 x 2 x 3
H1 ¼ 2 3 þ 1; H2 ¼ 2 þ x; H3 ¼ 3 2 ; H4
L L L2 L L L
x3 x2
¼ 2 :
L L
In the rotor blade problem, we focus on 1D structures. The properties of the shape
functions are discussed next.
1. Kronecker delta property
Shape function of a node has value equal to one on that node and zero at all the
other nodes.
Consider Fig. 2.2.
Node 1 ðx ¼ 0Þ
x x
H1 ¼ 1 ) H1 ¼ 1 and H2 ¼ ) H2 ¼ 0
L L
Node 2 ðx ¼ LÞ
x x
H2 ¼ ) H2 ¼ 1 and H1 ¼ 1 ) H1 ¼ 0:
L L
2. Compatibility condition
Displacement approximation is continuous across element boundaries.
Consider Fig. 2.4, where two elements are taken in a bar.
2.8 Properties of Shape Function in 1D 47
uðxÞ ¼ q2 :
3. Completeness
(a) Rigid body mode
H1 þ H2 ¼ 1
uðxÞ ¼ H1 q1 þ H2 q2 ¼ 1 ðfor q1 ; q2 ¼ 1Þ
q1 q2 L L2
eðxÞ ¼ L ¼ L ¼1
2 2
48 2 Finite Element Analysis in Space
or
L
uðxÞ ¼ x þ
2
or
eðxÞ ¼ 1
A brief outline of finite element has been provided. We are now ready to apply
the finite element method for the rotating beam problem.
Finite element formulation in space for the rotating beam is done using Hamilton’s
energy principle. (Complete derivation is given in [14].)
Potential energy is given by
ZR 2 ZR 2
1 @2w 1 @w
V¼ EI dx þ G dx ð2:28Þ
2 @x 2 2 @x
0 0
ZR 2
1 @w
T¼ m dx ð2:29Þ
2 @t
0
where w is the transverse displacement, G is the centrifugal force, and m is the mass
per unit length.
2.9 Finite Element Formulation of Rotating Beam 49
or
2 3
q1
@w 6 q2 7
¼ ½ H10 H20 H30 H40 6 7 0
4 q3 5 ¼ ½H ½q
@x
q4
or
2 3
2 q1
@ w 6 q2 7
¼ ½ H100 H200 H300 H400 6 7 00
4 q3 5 ¼ ½H ½q
@x2
q4
2
@w
¼ ½ q T ½ H 0 ½ H 0 ½ q
T
@x
ZR ZR
1 00 T 00 1
G½qT ½H 0 ½H 0 ½qdx
T T
V¼ EI ½q ½H ½H ½qdx þ ð2:30Þ
2 2
0 0
or
0 1
ZR ZR
1
V ¼ ½ q T @ EI ½H 00 ½H 00 dx þ G½H 0 ½H 0 dxA½q
T T
ð2:31Þ
2
0 0
or
1
V ¼ ½qT14 ½K 44 ½q41 ð2:32Þ
2
50 2 Finite Element Analysis in Space
ZR
1
T¼ m½q_ T ½H T ½H ½qdx ð2:33Þ
2
0
or
0 1
ZR
1
T ¼ ½q_ T @ m½H T ½H dxA½q_ ð2:34Þ
2
0
or
where L ¼ T V. From Eqs. (2.32), (2.35), and (1.35), we get the free vibration
problem
where
ZR
½M ¼ m½H T ½H dx ð2:37Þ
0
0 1
ZR ZR
½K ¼ @ EI ½H 00 ½H 00 dx þ G½H 0 ½H 0 dxA
T T
ð2:38Þ
0 0
or
We solve the free vibration problem with the above matrices. The formulation is
valid for any general axial force GðxÞ. For a rotating beam, we are interested in the
centrifugal force.
Centrifugal force is the additional term to beam equation in rotating beam equation.
In the formulation, it is ½K2 (Fig. 2.6).
52 2 Finite Element Analysis in Space
ZR
G¼ mX2 xdx ð2:40Þ
x
Consider Fig. 2.7, shape function for both the elements will be different.
For formulation of shape function, we follow Eqs. (2.15)–(2.25). For first ele-
ment, we evaluate value of x at 0 and L=2. For second element, we evaluate value of
x at L=2 and L:
Shape function for element 1 ð½H1 Þ Shape function for element 2 ð½H2 Þ
3 2 2 3
16 Lx 3 12 Lx 2 þ 1; 24 Lx 36 Lx 2 þ 16 Lx 3 4;
2 3 2 3
x 4 xL þ 4 Lx 2 ; 8x 2L 10 xL þ 4 Lx 2 ;
2 3 2 3
12 Lx 2 16 Lx 3 ; 36 Lx 2 24 Lx 16 Lx 3 þ 5;
3 2 2 3
4 Lx 2 2 xL 5x L 8 xL þ 4 Lx 2
4;
L L
4
L2 L2
48 ; 48
L L
4 4
2 2
L48 48
L
2.11 Shape Function Formulation for Two Elements 53
We see that
ZL=2 ZL
½H T1 dx ¼ ½H T2 dx ð2:42Þ
0 L=2
ZR
½H 00 ½H 00 dx
T
½K1 ¼ EI
0
We notice that
or
ZL=2 ZL
½H 00 1 ½H 00 1 dx ½H 00 2 ½H 00 2 dx
T T
EI ¼ EI ð2:43Þ
0 L=2
So, we should calculate the stiffness matrix ½K1 for only one element, and it will
be the same for all the elements.
R L=2 RL
Now we integrate 0 ½H T1 f ðxÞdx and L=2 ½H T2 f ðxÞdx
where f ðxÞ ¼ x
R R
L=2 L
Element 1 0 ½HT1 xdx Element 2 L=2 ½HT2 xdx
3L2 13L2
80 80
L3 7L3
240 480
7L2 17L2
80 80
3 3
160
L
L60
54 2 Finite Element Analysis in Space
We see that
ZL=2 ZL
½H T1 xdx 6¼ ½H T2 xdx ð2:44Þ
0 L=2
ZR
G½H 0 ½H 0 dx
T
½K2 ¼
0
We notice that
or
ZL=2 ZL
GðxÞ½H 0 1 ½H 0 1 dx GðxÞ½H 0 2 ½H 0 2 dx
T T
6¼ ð2:45Þ
0 L=2
So, we should calculate the stiffness matrix ½K2 for each element.
Formulation of ½K2 matrix with the shape function of the first element
Considering Fig. 2.8, we write the FEM formulation for third element using the
shape function of one element. Here, L is the length of each element, R is the radius
of rotor, x is the length along the element, xi is the distance of element from the
starting point and will depend on the element we have taken, X is the angular
velocity, N is the number of elements, q is the global coordinate system, and x is the
local coordinate system.
Zl
GðxÞ½H 0 1 ½H 0 1 dx
T
½K2 ¼
0
From Fig. 3.9, we write the centrifugal force for the element as
ZR ZR xZþ xi
or
ZR N Z
X
xj þ 1
or
X
N ðx2j þ 1 x2j Þ Ai
Term1 ¼ mj X2 ) Term1 ¼ X2
j¼i
2 2
P ðx2 x2 Þ
where Ai ¼ Nj¼i mj j þ 12 j (It will be a constant term)
We write Term2
xZþ xi
mi X2
Term2 ¼ mX2 qdq ) Term2 ¼ fðxi þ xÞ2 x2i g
2
xi
or
X2 Ai mi X2
GðxÞ ¼ ð2xxi þ x2 Þ ð2:47Þ
2 2
56 2 Finite Element Analysis in Space
Zl
GðxÞ½H 0 1 ½H 0 1 dx
T
½K2 ¼
0
Zl
X2 Ai mi X2
ð2xxi þ x2 Þ ½H 0 1 ½H 0 1 dx
T
½K2 i ¼ ð2:48Þ
2 2
0
26 1 6 1 3
5l 10 5l 10
6 1 l 7
X2 Ai 6 30 7
2l
½K2 i ¼ 6 25 10 7
2 6 4 6
5l
1 7
10 5
2l
2 3xi 15
l2 3xi l2
3
5 þ 6l
35 10 þ
xi l
28 5 35
6l
70
6 l2 xi l3 lxi l2 l2 xi l3 7
6
mi X2 6 30 þ 10 28 60 70 7
105
7
4 5
2
5 þ 35
3xi 6l l
70
l2 xi 3l2
10 þ 70
The natural frequency f and the respective mode shape V of a rotating beam can
be obtained from the Jacobian matrix ½ A.
½ A ¼ ½M 1 ½K ð2:49Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
f ¼ eigvalð½ AÞ ð2:50Þ
We have solved the free vibration problem, so we consider the case of forced
vibration.
ZL
½M½€q þ ½K½q ¼ ½HT Fðx; wÞdx ð2:52Þ
0
where q_ ¼ dq=dt:
We have done finite element formulation for the free vibration. We develop the
finite element formulation for the forced vibration in this section.
Finite element formulation of the right-hand side of Eq. (1.105) yields
ZL
½Q ¼ ½H T Fðx; wÞdx ðElement load vector Þ ð2:53Þ
0
58 2 Finite Element Analysis in Space
ZL
½QF ¼ ½HT fa1x a1w þ a2x a2w þ a3x a3w þ a4x a4w þ a5x a5w þ a6x a6w þ a7x a7w
0
þ ðb1x b1w þ b2x b2w Þw_ þ ðc1x c1w þ c2x c2w Þw0 gdx
ð2:54Þ
½QF ¼ a1w ½Qa1 þ a2w ½Qa2 þ a3w ½Qa3 þ a4w ½Qa4 þ a5w ½Qa5 þ a6w ½Qa6 þ a7w ½Qa7
_ þ ðc1w ½Da1 þ c2w ½Da2 Þ½q
þ ðb1w ½Ca1 þ b2w ½Ca2 Þ½q
ð2:55Þ
where
ZL ZL ZL
T T
½Qa1 ¼ a1x ½H dx; ½Qa2 ¼ a2x ½H dx; ½Qa3 ¼ a3x ½H T dx; ½Qa4
0 0 0
ZL
¼ a4x ½H T dx;
0
ZL ZL ZL
T T
½Qa5 ¼ a5x ½H dx; ½Qa6 ¼ a6x ½H dx; ½Qa7 ¼ a7x ½H T dx;
0 0 0
ZL ZL
T
½Ca1 ¼ b1x ½H ½Hdx; ½Ca2 ¼ b2x ½HT ½Hdx;
0 0
ZL ZL
0
½Da1 ¼ T
c1x ½H ½H dx; and ½Da2 ¼ c2x ½HT ½H 0 dx:
0 0
2.14 FEM Formulation of Aerodynamic Force for Rotor Problem 59
X2 ½M½€q þ ½K½q ¼ a1w ½Qa1 þ a2w ½Qa2 þ a3w ½Qa3 þ a4w ½Qa4 þ a5w ½Qa5
_ þ ðc1w ½Da1
þ a6w ½Qa6 þ a7w ½Qa7 þ ðb1w ½Ca1 þ b2w ½Ca2 Þ½q
þ c2w ½Da2 Þ½q
ð2:56Þ
_ þ ½D½q
X2 ½M½€q þ ½K½q ¼ ½Q þ ½C½q ð2:57Þ
where
½Q ¼ a1w ½Qa1 þ a2w ½Qa2 þ a3w ½Qa3 þ a4w ½Qa4 þ a5w ½Qa5 þ a6w ½Qa6 þ a7w ½Qa7
where ½q ¼ ½/½f, and ½/ being the eigenvectors. We can then write
or
where ½M1 ¼ ½/T ½M½/; ½K1 ¼ ½/T ½K½/; ½Q1 ¼ ½/T ½Q; ½C1 ¼ ½/T ½C½/;
and ½D1 ¼ ½/T ½D½/: We notice that Eq. (2.60) is an ordinary equation having
periodic coefficients and motion-dependent forcing.
We write Eq. (2.60) as
½AðwÞ½€f þ ½BðwÞ½f
_ þ ½CðwÞ½f ¼ ½DðwÞ ð2:61Þ
where ½A; ½B; ½C; and ½D contain periodic functions. Thus, all the motion-
dependent forces are moved to the left-hand side. This is important for solving the
equation. The motion-dependent forces change the stiffness and damping terms and
thus the behavior of the system. At this point, the spatial coordinate has been
60 2 Finite Element Analysis in Space
removed from the equation. The resulting set of ordinary differential equations now
need to be solved. Chapter 3 will address the solutions of the rotor dynamics
problem for the time response.
Chapter 3
Finite Element in Time
3.1 Introduction
In this chapter, finite element in time is explained with the help of examples and
coupled differential equations are solved with the periodic conditions. Finite dif-
ference method (Runge–Kutta fourth order) is explained as well. Note that the
helicopter blade equations are periodic differential equations.
Finite element in time is based on the weak form of Hamilton’s principle
Ztf Ztf
dLdt þ dqT Qdt ¼ dqT pjttfii ð3:1Þ
ti ti
where L is the Lagrangian of the system, and p is the set of generalized momenta.
These concepts are introduced in [1].
Here, p-version of finite element in time is formulated using the continuous
Galerkin’s method (velocity and displacement both are continuous on the nodal
boundaries).
We write Eq. (2.60) from the previous chapter
Second-order Eq. (2.60) can be written as two first-order Eqs. (3.2) and (3.3)
_
½P ¼ ½f ð3:3Þ
Weak formulation of Eqs. (3.2) and (3.3) in time gives Eqs. (3.4) and (3.5)
Z
_ ½C1 ½P þ ð½K1 ½D1 Þ½f ½Q1 gdw ¼ 0
dW1 fX2 ½M1 ½P ð3:4Þ
Z
_
dW2 f½P ½fgdw ¼0 ð3:5Þ
Note: Equations (3.4) and (3.5) are the matrix form of the equations which are
coupled; it contains a number of equations, so each equation should be formulated
as weak form. For better understanding, see the example of coupled differential
equations given later in this chapter.
u ¼ a1 þ a2 w þ a3 w2 þ a4 w3 ð3:6Þ
v ¼ b1 þ b2 w þ b3 w2 þ b4 w3 ð3:7Þ
uð0Þ ¼ a1
or
2 3 2 32 3
a1 1 0 0 0 uð0Þ
6 a2 7 6 11=4p 9=2p 9=4p 1=2p 7 6 7
6 7¼6 76 uð2p=3Þ 7 ð3:9Þ
4 a3 5 4 9=4p2 45=8p2 9=2p2 9=8p 2 54
uð4p=3Þ 5
a4 9=16p3 27=16p3 27=16p3 9=16p3
uð2p
or
2 32 3
1 0 0 0 uð0Þ
6 11=4p 9=2p 9=4p 1=2p 7 6 7
u¼ 1 w w2 w3 6 76 uð2p=3Þ 7
4 9=4p2 45=8p2 9=2p2 9=8p2 54 uð4p=3Þ 5
9=16p3 27=16p3 27=16p3 9=16p3 uð2pÞ
ð3:10Þ
or
or
u ¼ ½N½u ð3:14Þ
64 3 Finite Element in Time
where
and u_ ¼ v ¼ ddwu :
Here, we are interested in finding the steady-state part of solution, so we assume
the particular solution as
1 3 5 1
a1 ¼ ; a2 ¼ 0; a3 ¼ ; a4 ¼ ; a5 ¼ 0; a6 ¼ ;
2 17 17 6
5 21 3 4
a7 ¼ ; a8 ¼ ; a9 ¼ ; and a10 ¼ :
233 233 146 73
1 3 5 1 5 21
up ¼ sinðwÞ þ sinð2wÞ þ cosð2wÞ þ cosð3wÞ þ sinð4wÞ þ cosð4wÞ
2 17 17 6 233 233
3 4
þ sinð5wÞ þ cosð5wÞ
146 73
ð3:17Þ
1 6 10 1 20 84
vp ¼ cosðwÞ þ cosð2wÞ þ sinð2wÞ þ sinð3wÞ þ cosð4wÞ þ sinð4wÞ
2 17 17 2 233 233
15 20
þ cosð5wÞ þ sinð5wÞ
146 73
ð3:18Þ
3.3 Finite Element in Time Example 65
€u þ 2u_ þ 3u ¼ 0 ð3:19Þ
uh ¼ Aesw ð3:20Þ
u ¼ uh þ u p
pffiffi 1 pffiffi
u ¼ A1 eð1 þ 2iÞw
þ A2 eð1
sinðwÞ2iÞw
2
3 5 1
þ sinð2wÞ þ cosð2wÞ þ cosð3wÞ ð3:23Þ
17 17 6
5 21 3 4
þ sinð4wÞ þ cosð4wÞ þ sinð5wÞ þ cosð5wÞ
233 233 146 73
v ¼ vh þ vp
pffiffiffi pffiffi pffiffiffi pffiffi
v ¼ ð1 þ 2iÞA1 eð1 þ 2iÞw þ ð1 2iÞA2 eð1 2iÞw
1 6 10 1
þ cosðwÞ þ cosð2wÞ þ sinð2wÞ þ sinð3wÞ ð3:24Þ
2 17 17 2
20 84 15 20
þ cosð4wÞ þ sinð4wÞ þ cosð5wÞ þ sinð5wÞ
233 233 146 73
For finite element in time formulation, we write Eq. (3.15) as two first-order
equations
v_ þ 2v þ 3u ¼ f ðwÞ ð3:25Þ
v ¼ u_ ð3:26Þ
66 3 Finite Element in Time
Z2p
dW1 ð_v þ 2v þ 3u f ðwÞÞdw ¼ 0 ð3:27Þ
0
Z2p
_
dW2 ðv uÞdw ¼0 ð3:28Þ
0
Writing u ¼ ½N½u; v ¼ ½N½v; dW1 ¼ ½dW1 T ½NT and dW2 ¼ ½dW2 T ½NT ; we
get equations
Z2p
_
½dW1 T ½NT ð½N½v þ 2½N½v þ 3½N½u f ðwÞÞdw ¼ 0 ð3:29Þ
0
Z2p
_
½dW2 T ½N T ð½N½v ½N½uÞdw ¼0 ð3:30Þ
0
From Eq. (3.31), we get the displacement and velocity at nodal points. Here, we
can get the solution with periodic conditions as well as with initial conditions.
Periodic conditions give us the steady-state solution, and with initial conditions, we
get the transient part as well.
Here, periodic conditions are uð0Þ ¼ uð2pÞ and vð0Þ ¼ vð2pÞ, and initial con-
ditions are uð0Þ ¼ 0 and vð0Þ ¼ 0:
Figure 3.1a, b shows the solution using periodic conditions. Here, element length
is p=8; number of elements are 16, and number of nodes within the element are 6.
Figure 3.1c, d shows the solution using initial conditions. Here, element length is
p=4; number of elements are 40, and number of nodes within the element are 6.
3.3 Finite Element in Time Example 67
Fig. 3.1 a Finite element in time with periodic conditions (displacement), b finite element in time
with periodic conditions (velocity), c finite element in time with initial conditions (displacement),
d finite element in time with initial conditions (velocity)
68 3 Finite Element in Time
In case of coupled differential equations, solution may not converge with initial
conditions so we use periodic conditions.
Equations (3.4) and (3.5) are coupled differential equations. We consider an
example of coupled equations to illustrate the method.
3 0 €f1 5 0 f1 1 2 f_ 1 4 9 f1 f ðwÞ
0 4 €f2 þ 0 9 f2
¼
3 4 f_ 2
þ
6 7 f2
þ
f ðwÞ
ð3:32Þ
For finite element formulation, we write two coupled differential Eqs. (3.35) and
(3.36) of second order into four differential Eqs. (3.37)–(3.40) of first order. First,
we move the motion-dependent term in Eq. (3.32) to the left-hand side.
p1 ¼ f_ 1 ð3:39Þ
p2 ¼ f_ 2 ð3:40Þ
Z2p
dW1 ð3p_ 1 p1 2p2 þ f1 9f2 f ðwÞÞdw ¼ 0 ð3:41Þ
0
Z2p
dW2 ð4p_ 2 3p1 4p2 6f1 þ 2f2 f ðwÞÞdw ¼ 0 ð3:42Þ
0
Z2p
dW3 ðp1 f_ 1 Þdw ¼ 0 ð3:43Þ
0
Z2p
dW4 ðp2 f_ 2 Þdw ð3:44Þ
0
2 2p 3
R T _ T R2p R2p R2p
6 ð3½N N ½N ½N Þdw 2½N T ½N dw ½N T ½N dw 9½N T ½N dw 7
60 0 0 0 7
6 72 3
6 R2p R2p T R2p R2p 7 ½p1
6 3½N T ½N dw ð4½N N_ 4½N ½N Þdw T
6½N T ½N dw 2½N ½N dw 7
T
6 76 ½p 7
6 0 0 0 0 76 2 7
6 76 7
6 R2p T R2p 74 ½f1 5
6 ½N ½N dw 0 ½N T N_ dw 0 7
6 7 ½f
6 0 0 7 2
6 7
4 R2p R2p 5
0 ½N T ½N dw 0 ½N T N_ dw
0 0
2 2p 3
R T
6 ½ N f ðwÞdw 7
60 7
6 7
6 R2p T 7
¼ 6 ½N f ðwÞdw 7
6
7
60 7
6 7
4 0 5
0
ð3:45Þ
Here, each entry in the matrix is also a matrix, and matrix size depends on the
number of nodes selected. For example, for 2 nodes, each entry will be a 2 2
matrix and ½p1 ; ½p2 ; ½f1 ; and ½f2 will be a 2 1 vector.
Here, periodic conditions are used, element length is 2p number of element is 1,
and number of nodes within the element are 17. Results match well with the
analytical solutions as can be seen from Fig. 3.2a, b.
Suppose, we take only one time element with four nodes 0; 2p=3; 4p=3; and 2p:
Here,u1 ; u2 ; u3 ; and u4 are the displacements at the four nodes, and v1 ; v2 ; v3 ; and v4
are the velocities at the four nodes, respectively.
We know uð0Þ ¼ uð2pÞ ) u1 ¼ u4 & vð0Þ ¼ vð2pÞ ) v1 ¼ v4 and vð0Þ ¼
vð2pÞ ) v1 ¼ v4 :
2 32 3 2 3
a1 a2 a3 a4 b1 b2 b3 b4 v1 f1
6 a5 a6 a7 a8 b5 b6 b7 b8 7 6 v2 7 6 f 2 7
6 76 7 6 7
6 a9 a10 a11 a12 b9 b10 b11 b12 7 6 7 6 7
6 7 6 v3 7 6 f 3 7
6 a13 a14 a15 a16 b13 b14 b15 7
b16 76 7 6 7
6 6 v4 7 ¼ 6 f 4 7 ð3:46Þ
6 c1 c2 c3 c4 d1 d2 d3 d4 7 6 u1 7
7 6 6 7
6 7 6 f5 7
6 c5 c6 c7 c8 d5 d6 d7 7
d8 7 6 u2 7 6
6 7 7
6 6 f6 7
4 c9 c10 c11 c12 d9 d10 d11 d12 5 4 u3 5 4 f7 5
c13 c14 c15 c16 d13 d14 d15 d16 u4 f8
2 32 3 2 3
a2 þ a14 a3 þ a15 a1 þ a4 þ a13 þ a16 b2 þ b14 b3 þ b15 b1 þ b4 þ b13 þ b16 v2 f1 þ f4
6 a6 a7 a5 þ a8 b6 b7 b5 þ b8 76 v3 7 6 f2 7
6 76 7 6 7
6 a10 a11 a9 þ a12 b10 b11 b9 þ b12 76 v4 7 6 f3 7
6 76 7 ¼ 6 7
6 c2 þ c14 c3 þ c15 c1 þ c4 þ c13 þ c16 d2 þ d14 d3 þ d15 d1 þ d4 þ d13 þ d16 76 u2 7 6 f5 þ f8 7
7 6 7 6
6 7
4 c6 c7 c5 þ c8 d6 d7 d5 þ d8 5 4 u3 5 4 f6 5
c10 c11 c9 þ c12 d10 d11 d9 þ d12 u4 f7
ð3:47Þ
AðwÞ is a periodic function with a period of 2p: We can use shape function of
one element for all the elements if the length of the each element is 2p:
where u_ ¼ v ¼ ddwu :
Particular solution of Eq. (3.49) is given by
1 3 5
up ¼ sinðwÞ þ sinð2wÞ þ cosð2wÞ ð3:50Þ
2 17 17
1 6 10
vp ¼ cosðwÞ þ cosð2wÞ þ sinð2wÞ ð3:51Þ
2 17 17
Equation (3.49) is solved using finite element in time with only one time ele-
ment; length of the element is 0 to 2p: Periodic conditions are used to get
steady-state solution. Three cases are considered here with different number of
nodes. The results in Fig. 3.3a–c show excellent agreement with analytical solution.
Case1—(1 element, 6 nodes)
Case2—(1 element, 11 nodes)
Case3—(1 element, 17 nodes)
Here, we compare three Eqs. (3.51), (3.54) and (3.57), with 11 node, 1 time ele-
ment in 0 to 2p: These equations model an increase in the order of forcing.
where u_ ¼ v ¼ ddwu :
3.8 Effect of Forcing Term in Finite Element in Time 73
Fig. 3.3 a Selection of number of nodes (1 element, 6 nodes), b selection of number of nodes (1
element, 11 nodes), c selection of number of nodes (1 element, 17 nodes)
1
up ¼ sinðwÞ ð3:52Þ
2
1
vp ¼ cosðwÞ ð3:53Þ
2
In Fig. 3.4a, we see that the FEM correlates well with the analytical results.
where u_ ¼ v ¼ ddwu :
74 3 Finite Element in Time
1 3 5
up ¼ sinðwÞ þ sinð2wÞ þ cosð2wÞ ð3:55Þ
2 17 17
1 6 10
vp ¼ cosðwÞ þ cosð2wÞ þ sinð2wÞ ð3:56Þ
2 17 17
Figure 3.4b also shows good agreement. However, Fig. 3.4c shows a deterio-
ration in performance as the polynomial discretization is insufficient for the high
degree of forcing.
3.8 Effect of Forcing Term in Finite Element in Time 75
where u_ ¼ v ¼ ddwu :
Particular solution of Eq. (3.57) is given by
1 3 5 1
up ¼ sinðwÞ þ sinð2wÞ þ cosð2wÞ þ cosð3wÞ ð3:58Þ
2 17 17 6
1 6 10 1
vp ¼ cosðwÞ þ cosð2wÞ þ sinð2wÞ þ sinð3wÞ ð3:59Þ
2 17 17 2
Hence, we can see how the forcing affects our results. When there is a higher
harmonic content in the forcing, a lower number of nodes are insufficient.
76 3 Finite Element in Time
Fig. 3.4 a Effect of forcing (1 element, 11 nodes, f ðwÞ ¼ sinðwÞ þ cosðwÞ), b effect of forcing (1
element, 11 nodes, f ðwÞ ¼ sinðwÞ þ cosðwÞ þ sinð2wÞ þ cosð2wÞ), c effect of forcing (1 element,
11 nodes, f ðwÞ ¼ sinðwÞ þ cosðwÞ þ sinð2wÞ þ cosð2wÞ þ sinð3wÞ þ cosð3wÞ), d effect of
forcing (1 element, 11 nodes, f ðwÞ ¼ sinðwÞ þ cosðwÞ þ sinð2wÞ þ cosð2wÞ þ sinð3wÞ þ
cosð3wÞ)
where
v ¼ u_ ) u_ ¼ v ð3:61Þ
We write Eqs. (3.42) and (3.43) in the form of Eqs. (3.44) and (3.45) with initial
conditions uð0Þ ¼ 0; and vð0Þ ¼ 0:
dv
¼ f1 ðu; v; wÞ; vð0Þ ¼ 0 ð3:62Þ
dw
du
¼ f2 ðu; v; wÞ; uð0Þ ¼ 0 ð3:63Þ
dw
We solve Eqs. (3.44) and (3.45) using Runge–Kutta method. We March in time
using following equations
3.9 Finite Difference Method (Runge–Kutta Fourth Order) 79
where
ku1 ¼ f2 ðui ; vi ; wi Þ
kv1 ¼ f1 ðui ; vi ; wi Þ
h h h
ku2 ¼ f2 ui þ ku1 ; vi þ kv1 ; wi þ
2 2 2
80 3 Finite Element in Time
h h h
kv2 ¼ f1 ui þ ku1 ; vi þ kv1 ; wi þ
2 2 2
h h h
ku3 ¼ f2 ui þ ku2 ; vi þ kv2 ; wi þ
2 2 2
h h h
kv3 ¼ f1 ui þ ku2 ; vi þ kv2 ; wi þ
2 2 2
Result of Runge–Kutta method with a time step is p=32 is shown in Fig. 3.5a.
A very similar result was obtained with finite element in time in the Sect. (3.3)
for Eq. (3.15).
Thus, helicopter rotor problems can be solved using the finite element in time or
Runge–Kutta method. Finite element in time is useful to get the periodic response
efficiently. In most comprehensive aeroelastic analysis, response, loads, and sta-
bility are calculated for steady-state condition. Therefore, finite element in time is
often used. Since the problem is 1D in time, a p-version time finite element works
very well and is illustrated in this book.
Chapter 4
Stability Analysis
4.1 Introduction
Here, we take two differential Eqs. (4.3) and (4.4) with constant coefficients.
_
½XðtÞ n1 ¼ ½Ann ½XðtÞn1 þ ½Bnn ½uðtÞn1 ð4:5Þ
For a system to be stable, the real part of all the eigenvalues of the matrix ½ A has
to be negative. Equations (4.6) and (4.7) are state-space representation of differ-
ential Eqs. (4.3) and (4.4) respectively.
" #
du
dt ¼ 0 1 u 0
þ ð4:6Þ
dv 3 2 v sinðwÞ þ cosðwÞ
dt
" #
du
dt ¼ 0 1 u þ 0
ð4:7Þ
dv 3 2 v sinðwÞ þ cosðwÞ
dt
pffiffiffi
Eigenvalues of the matrix ½A of differential Eq. (4.6) are 1 2i and
pffiffiffi
eigenvalues of the matrix ½A of differential Eq. (4.7) are 1 2i: For a system to
be stable, real part of the eigenvalues has to be negative Therefore, Eq. (4.6) is a
stable system, and Eq. (4.7) is an unstable system. Such systems are sometimes
displayed in a root locus plot as shown in Fig. 4.1a, b.
If all of the eigenvalues lie in the left-half plane, the system is stable. If any of
the eigenvalues lie on the right-half plane, the system is unstable. Eigenvalues on
the y-axis show neutral stability.
If we solve Eqs. (4.6) and (4.7), we get the result shown in Fig. 4.1c, d,
respectively. The problem of instability is seen here in the response of the system.
Note that stability is an intrinsic characteristic of the system. Equations (4.6) and
(4.7) have the same forcing but different coefficient matrices. But the coefficient
matrix can completely change the nature of the system.
½A½€f þ ½B½f
_ þ ½C½f ¼ ½D ð4:8Þ
or
3 0 €f1 5 0 f_ 1 1 2 f1 sinðwÞ
0 4 €f2 þ 0 9 _f2 3 4 f2
¼
cosðwÞ
ð4:9Þ
4.3 Stability Analysis of a Coupled Differential Equations … 85
Fig. 4.1 a Stable system, root locus plot of differential Eq. (4.3), b unstable system, root locus
plot of differential Eq. (4.4), c stable system, solution of differential Eq. (4.3), d unstable system,
solution of differential Eq. (4.4)
3 0 5 0 1 2 sinðwÞ
where ½ A ¼ ; ½B ¼ ; ½C ¼ ; and ½D ¼
0 4 0 9 3 4 cosðwÞ
We can write state-space representation of Eq. (4.8)
"h i #
f_ ½022 ½I 22 f ½022
21 ¼ þ ð4:10Þ
P_ ½ A1
22 ½C 22 ½ A1
22 ½B22 P ½ A1
22 ½D22
21
h i
P1
where ½P ¼ f_ ¼
P2
86 4 Stability Analysis
or
2 3 2 32 3 2 3
f_ 1 0 0 1 0 f1 0
6 f_ 7 6 0 1 7 6 7 6 7
6 7¼62 0 0 7 6 f2 7 þ 6 0 7 ð4:11Þ
4 P_ 1 5 4 0:33 0:66 1:66 0 5 4 P1 5 4 sinðwÞ=3 5
P_ 2 0:75 1 0 2:25 P2 cosðwÞ=4
Eigenvalues of the system are 0.5537, −0.0649, −2.6956, and −1.7033. Real
part of all the eigenvalues is not negative, so system is unstable. In hover condition,
the helicopter rotor equation has constant coefficients, and the methods discussed
above can be used.
4.4 Stability Analysis of the Equation … 87
To find stability solution for Eq. (4.12), we should use Floquet theory.
We have the state-space representation.
_
XðtÞ ¼ ½AðtÞ½XðtÞ ð4:13Þ
or
k ¼ nP þ iwP ð4:16Þ
h i
ImðHÞ repre-
where nP ¼ T1 lnðHÞ represents damping rates and xP ¼ T1 tan1 Re ðHÞ
h i
1 ImðHÞ
sents frequencies. Since tan ReðHÞ is multi-valued, there can be ambiguous
answers for frequencies.
88 4 Stability Analysis
y ¼ c1 w þ c2 ð4:20Þ
We apply the 1st set of initial condition to Eqs. (4.19) and (4.20) to get c1 ¼ e
and c2 ¼ 0:
We apply the 2nd set of the initial condition to Eqs. (4.19) and (4.20) to get
c1 ¼ 0 and c2 ¼ 1:
Suppose we want to know solution after T ¼ 2p: Equations. (4.19) and (4.20)
with the 1st set of initial conditions, after time period T ¼ 2p; give us
x ¼ 1; y ¼ 2pe
Equations (4.19) and (4.20) with the 2nd set of initial conditions, after time
period T ¼ 2p; give us
x ¼ 0; y ¼ 1
We relate the solution at T ¼ 2p; with solution at T ¼ 0; in Eqs. (4.21) and (4.22)
1 1 0 1
¼ ð4:21Þ
2pe 2pe 1 0
0 1 0 0
¼ ð4:22Þ
1 2pe 1 1
4.5 Analytical Solution with the Floquet Theory 89
Eigenvalues of this transition matrix are (1, 1). From Eq. (4.15), we get the
eigenvalues (0, 0) of the matrix ½Að2pÞ: The system is neutrally stable.
Here, we get a transition matrix numerically. Time period T is divided into k equal
intervals, and in each interval, the matrix ½AðtÞ is replaced by the matrix ½Ck with
constant coefficients.
We take increment in time as
ZwK
1
½Ck ¼ A½fdf ð4:25Þ
Dk
wK1
where Dk ¼ wk wk1 .
Transition matrix after time period T is given by Eq. (4.26).
½/ðT; 0Þ ¼ expðDk ½ck Þ expðDk1 ½ck1 Þ expðDk2 ½ck2 Þ. . . expðD1 ½c1 Þ
ð4:26Þ
Suppose we want to find stability after a period of 2p: We divide the period T ¼ 2p
into 32 divisions, each division having an interval of p=32:
90 4 Stability Analysis
Here,
p 2p p 3p 2p 32p 31p
Dk ¼ 0¼ ¼ ... ¼ ;
16 16 16 16 16 16 16
D1 ¼ D2 ¼ D3 . . . D32 ;
p
Z16 Z16
2p
Z2p
1 1 1
½C1 ¼ p AðtÞdt; ½C2 ¼ 2p p AðtÞdt; . . .½C32 ¼ AðtÞdt
16 16 16 2p 31p
16
0 p 31p
16 16
½/ð2p; 0Þ ¼ expðD32 ½C32 Þ expðD31 ½C31 Þ. . . expðD1 ½C1 Þ ð4:28Þ
For rotor problem, we get Eq. (2.60) after the finite element in space
_
½P ¼ ½f ð4:31Þ
_
Equation (4.33) is in state-space representation form ½XðtÞ ¼ ½AðtÞ½XðtÞ: We
can find out the eigenvalues of the matrix ½AðtÞ numerically using the Floquet
theory. Chapter 5 illustrates the numerical results for the helicopter rotor problem.
Chapter 5
Helicopter Rotor Results
The theory discussed in previous chapters is now applied to a helicopter rotor. The
results are matched with published literature. A MATLAB code is given along with
this book which can be used to generate results given in this chapter.
5.1 Inputs
Table 5.1 shows the inputs [3], used for the elastic rotor problem. These are typical
of a four-bladed hingeless rotor. In this book, we assume a flapping elastic blade
with cantilever boundary conditions.
We have discussed the mode shape and the frequency calculation in Sect. 2.13. The
non-dimensional rotating frequencies are shown in Table 5.2 and matched with [6],
for the given values of non-dimensional rotating speed s. The beam is divided into
200 finite elements with four degrees of freedom for each element. The rotation
leads to an increase in the beam stiffness, and therefore in the natural frequency. We
observe this stiffening effect for the first mode. For the second and higher modes,
stiffening comes largely from flexure, and the effect of rotation is much less.
We draw a Campbell diagram (Fig. 5.1) and observe the stiffening effect in the
rotating beam with the increase in the rotating speed, which will result increased
natural frequencies. Here x1 ; x2 ; x3 ; x4 ; and x5 are the first five natural
frequencies of the rotating beam, respectively.
Three sets of vehicle control and attitude are considered at three different flight
speeds which range from slow speed ðl ¼ 0:15Þ to high speed ðl ¼ 0:35Þ: The
results are shown in Fig. 5.2. The response increases at higher forward speeds.
Case 1 ðh1s ¼ 2:4 ; h1c ¼ 1:9 ; h0 ¼ 6:75 ; htw ¼ 8 ; as ¼ 1:2 ; l ¼ 0:15Þ
Case 2 ðh1s ¼ 5:4 ; h1c ¼ 1:8 ; h0 ¼ 8:5 ; htw ¼ 8 ; as ¼ 3:9 ; l ¼ 0:30Þ
Case 3 ðh1s ¼ 6:8 ; h1c ¼ 1:8 ; h0 ¼ 9:3 ; htw ¼ 8 ; as ¼ 4:2 ; l ¼ 0:35Þ
The results obtained by present formulation are similar to those given in [11].
The same three cases are considered with a linear inflow model. Finite element in
time is used to obtain the results in Figs. 5.2 and 5.3 with 1 element and 17 nodes.
Advance ratio ðlÞ goes to zero in hover condition. The stability graph is plotted
with the non-dimensional natural frequency ðx=XÞ 1.189 and with varying Lock
number. Here, the constant coefficient method discussed in Chap. 4 is used.
Initially, as the Lock number is zero, there is no aerodynamic force, and we will get
the complex conjugate roots 1.189i and −1.189i. Our results are similar to [5]
(Figs. 5.4 and 5.5).
This book has outlined an unaddressed area of helicopter dynamics which involves
the rotating beam problem. The rotating beam in pure flap bending represents a sound
pedagogical model to explain concepts of natural frequency, solution of the gov-
erning differential equation, and analysis of the stability of the differential equations.
In this book, the rotating beam problem has been systematically developed and solved
for helicopter dynamics. The basics of vibration theory are presented, followed by the
derivation of the rotating beam governing differential equation including aerody-
namic loads. The partial differential equation has spatial and temporal coordinates.
The spatial variable is handled through the finite element in space domain, and the
time variable is handled using the finite element in time domain. The use of time finite
element allows us to enforce the periodic boundary condition for the time domain
problem. Once the response of the rotating beam to aerodynamic forcing has been
calculated, the stability analysis is performed. The Floquet method of stability
calculation is explained for the periodic system. Finally, some solutions generated by
the model developed in this book are compared with published literature. The topics
developed in this book arms the reader with the necessary tools for understanding and
working with sophisticated helicopter rotor dynamics software packages where
rotating beams with multiple deformations are typically used.
References
1. Borri, M.: Helicopter rotor dynamics by finite element time approximation. Comput. Math.
Appl. 2(1), part A, 149–160 (1986)
2. Gudla, P.K., Ganguli, R.: Discontinuous Galerkin finite element in time for solving periodic
differential equations. Comput. Methods Appl. Mech. Eng. 196, 682–696 (2006)
3. Zhang, J.: Active-passive hybrid optimization of rotor blades with trailing edge flaps. In: A
Thesis in Aerospace Engineering. The Pennsylvania State University (2001)
4. Lim, I.G., Lee, I.: Aeroelastic analysis of bearingless rotors with composite flexbeam in
hover and forward flight. In: 16th International Conference on Composite Materials, July
2007
5. Johnson, W.: Helicopter Theory. Princeton University Press, Princeton (1980)
6. Sandilya, K., Ganguli, R., Mani, V.: Non-rotating beams isospectral to a given rotating
uniform beam. Int. J. Mech. Sci. 66, 12–21 (2013)
7. Sheng, G., Fung, T.C., Fan, S.C.: Parametrized formulations of Hamilton’s law for numerical
solutions of dynamic problems: Part 1 and 2. Time finite element approximation. Comput.
Mech. 21, 441–460 (1998)
8. Friedmann, P.P.: Numerical methods for determining the stability and response of periodoc
systems with application to helicopter rotor dynamics and aeroelasticity. Comput. Math.
Appl. 12A, 131–148 (1986)
9. Bauchau, O.A., Nikishkov, Y.G.: An implicit Floquet analysis for rotorcraft stability
evolution. J. Am. Helicopter Soc. 46, 200–209 (2001)
10. Leishman, J.G.: Principle of Helicopter Aerodynamics. Cambridge University Press, New
York (2002)
11. Hu, X.Y., Han, J.L., Yu, M.: Nonlinear aeroelastic coupled trim and stability analysis of
rotor-fuselage. Appl. Math. Mech. Engl. 31(2), 237–246 (2010)
12. Lee, I., Jeong, M.S., Yoo, S.J.: Aeroelastic analysis for helicopter rotor blades in hover and
forward flight. In: 28th International Congress of the Aeronautical Sciences, 25 Sept 2012
13. Sinha, S.C.: Stability analysis of systems with periodic coefficient: An approximate
approach. J. Sound Vib. 64(4), 515–527 (1979)
14. Meirovitch, L.: Fundamental of vibration. McGraw-Hill Higher Education, New York (2002)
15. Hutton, D.V.: Fundamentals of finite element analysis. McGraw-Hill Higher Education, New
York (2005)
16. Rao, S.S.: Mechanical Vibrations. Pearson Education, Upper Saddle River (2004)