Sunteți pe pagina 1din 109

Foundations of Engineering Mechanics

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

The Rotating Beam Problem


in Helicopter Dynamics

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

ISSN 1612-1384 ISSN 1860-6237 (electronic)


Foundations of Engineering Mechanics
ISBN 978-981-10-6097-7 ISBN 978-981-10-6098-4 (eBook)
https://doi.org/10.1007/978-981-10-6098-4
Library of Congress Control Number: 2017953797

© Springer Nature Singapore Pte Ltd. 2018


This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature


The registered company is Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore
Preface

The helicopter is a complicated dynamic system. A key aspect of the helicopter is


the main rotor which provides lift, propulsive thrust, and control capacity.
Modeling of helicopter rotor dynamics is complicated by highly flexible blades and
aerodynamics. Comprehensive books on helicopter dynamics have been written.
However, they typically derive and discuss the rigid blade equations and provide a
cursory treatment to the elastic blade equations, aerodynamic modeling, and
solution method for blade response and stability. Thus, a large pedagogical gap
exists between the helicopter dynamic books and the theory manuals of the com-
prehensive rotor codes. This book tries to fill this gap.
We explain the basics of helicopter dynamics which are needed for solving
helicopter problems. We take a detailed approach toward the problem of elastic
rotating blade, a problem which involves a partial differential equation in space and
time and requires a numerical solution. Finite element method plays an important
role here, first in space and later in time.
Chapter 1 provides a background on the vibration of discrete and continuous
system. It introduces the momentum theory and blade element theory and presents
the rigid and elastic flapping blade. The elastic blade equation is derived at the end
of Chap. 1. This is a partial differential equation with periodic coefficients and
forcing terms. Chapter 2 discusses the finite element method and its use to discretize
the rotor blade equation in the spatial domain. Methods to calculate the rotating
beam natural frequencies and mode shapes are presented. The aerodynamic forces
are also formulated on this chapter. Chapter 3 discusses the finite element method in
time, a technique which is ideal for calculating the steady response for periodic
systems. This method is illustrated for some periodic differential equations and then
used for elastic rotor blade problem. A p-version of the finite element in time is
introduced in this chapter. Chapter 4 presents the stability analysis of the elastic
blade equation. The constant coefficient approach is illustrated for the rotor in
hover. Then, Floquet theory is developed for periodic systems in forward flight.

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.

Bangalore, India Ranjan Ganguli


2017 Vijay Panchore
Contents

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

2.11 Shape Function Formulation for Two Elements . . . . . . . . . .... 52


2.12 FEM Formulation of Rotating Beam with Only One Shape
Function (for Free Vibration) . . . . . . . . . . . . . . . . . . . . . . . .... 54
2.13 Calculation of Mode Shapes and Frequencies . . . . . . . . . . . .... 57
2.14 FEM Formulation of Aerodynamic Force for Rotor
Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 57
3 Finite Element in Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.2 Selection of Shape Function in Time . . . . . . . . . . . . . . . . . . . . . 62
3.3 Finite Element in Time Example . . . . . . . . . . . . . . . . . . . . . . . . 64
3.4 Solution of Coupled Differential Equations with Finite
Element in Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 68
3.5 Enforcing Periodicity in the System . . . . . . . . . . . . . . . . . . . . .. 70
3.6 Advantage of Choosing an Element from (0 to 2p), p-Version
of Finite Element in Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.7 Selection of Number of Nodes . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.8 Effect of Forcing Term in Finite Element in Time . . . . . . . . . . . 72
3.9 Finite Difference Method (Runge–Kutta Fourth Order) . . . . . . . . 77
4 Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 83
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 83
4.2 Stability Analysis of Equations with Constant Coefficients . . ... 83
4.3 Stability Analysis of a Coupled Differential Equations with
Constant Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 84
4.4 Stability Analysis of the Equation with Periodic Coefficients,
Floquet Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.5 Analytical Solution with the Floquet Theory . . . . . . . . . . . . . . . 88
4.6 Numerical Method to Evaluate a Transition Matrix . . . . . . . . . . . 89
4.7 Stability Analysis for Rotor Problem . . . . . . . . . . . . . . . . . . . . . 90
5 Helicopter Rotor Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 93
5.1 Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 93
5.2 Result 1 (Mode Shapes and Frequencies of the Rotating
Beam) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 93
5.3 Result 2 (Response of the Rotor Blade with the Uniform
Inflow Model, Using Three Different Cases) . . . . . . . . . . ...... 95
5.4 Result 3 (Response of the Rotor Blade with the Linear
Inflow Model, Using Three Different Cases) . . . . . . . . . . . . . . . . 95
5.5 Result 4 (Stability in Hover Condition) . . . . . . . . . . . . . . . . . . . 96
5.6 Result 5 (Stability in Forward Flight) . . . . . . . . . . . . . . . . . . . . . 97
5.7 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
About the Authors

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

15 Ph.D. and 35 postgraduate students. He has written books on “Engineering


Optimization” and “Gas Turbine Diagnostics,” both published by CRC Press, New
York, and books titled “Structural Damage Detection using Genetic Fuzzy
Systems” and “Smart Helicopter Rotors,” published by Springer.
Vijay Panchore obtained his B.Tech. in Industrial Engineering from the National
Institute of Technology, Jalandhar, in 2009, M.Des. from the Indian Institute of
Science, Bangalore, in 2011, and Ph.D. in Aerospace Engineering from the Indian
Institute of Science, Bangalore, in 2017. He works in the area of finite element
method, meshless methods, and helicopter dynamics. He has published two inter-
national journal papers, where finite element method and meshless methods were
used to solve the rotating beam problems.
List of Figures

Fig. 1.1 Spring–mass system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 2


Fig. 1.2 Displacement-timegraph (spring–mass system) . . . . . . . . . . . . .. 3
Fig. 1.3 Damped vibration system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 3
Fig. 1.4 Graphical representation of an over-damped system
(a), a critically damped system (b), and an under-damped
system (c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 5
Fig. 1.5 Forced vibration of the spring–mass system . . . . . . . . . . . . . . .. 6
Fig. 1.6 Forced vibration of the mass–spring–damper system . . . . . . . . .. 7
Fig. 1.7 MATLAB plot of “dynamic amplification factor versus
frequency ratio” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 9
Fig. 1.8 a Two-degrees-of-freedom system. b Free-body diagram of
two-degrees-of-freedom spring–mass–damper system . . . . . . . .. 9
Fig. 1.9 Simply-supported beam with boundary conditions . . . . . . . . . . .. 12
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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 14
Fig. 1.11 Fixed-free beam with boundary conditions . . . . . . . . . . . . . . . .. 14
Fig. 1.12 a First mode shape of fixed-free beam. b Second mode shape
of fixed-free beam. c Third mode shape of fixed-free beam . . . .. 16
Fig. 1.13 Simply-supported-free beam with boundary conditions . . . . . . .. 17
Fig. 1.14 a First mode shape of simply-supported-free beam. b Second
mode shape of simply-supported-free beam. c Third mode
shape of simply-supported-free beam . . . . . . . . . . . . . . . . . . . . .. 19
Fig. 1.15 Simple pendulum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 19
Fig. 1.16 Air flow through the control volume in momentum theory for
axial flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 23
Fig. 1.17 Glauert flow model for momentum analysis of a rotor in
forward flight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 26
Fig. 1.18 a Blade element theory. b Blade element theory. c Blade
element theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 29

xi
xii List of Figures

Fig. 1.19 Tangential and perpendicular components of the


flow velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Fig. 1.20 Forces acting on a small element of a rigid rotor blade . . . . . . . . 31
Fig. 1.21 a Deflection of an elastic rotor blade. b Force diagram . . . . . . . . 36
Fig. 2.1 Elastic bar subjected to uniform load . . . . . . . . . . . . . . . . . . . . . . 42
Fig. 2.2 Bar element for shape function formulation . . . . . . . . . . . . . . . . . 44
Fig. 2.3 Beam element for shape function formulation . . . . . . . . . . . . . . . 44
Fig. 2.4 Two elements in a bar for FEM in space . . . . . . . . . . . . . . . . . . . 47
Fig. 2.5 Bar element (shape function properties) . . . . . . . . . . . . . . . . . . . . 48
Fig. 2.6 Centrifugal force on the rotating beam. . . . . . . . . . . . . . . . . . . . . 52
Fig. 2.7 Shape function formulation of two elements . . . . . . . . . . . . . . . . 53
Fig. 2.8 FEM formulation of rotating beam with the shape function
of one element (for free vibration) . . . . . . . . . . . . . . . . . . . . . . .. 54
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) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 67
Fig. 3.2 Finite element in time for coupled differential equations . . . . . .. 71
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) . . . . . . . .. 73
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) . . . . . . . . .. 76
Fig. 3.5 Runge–Kutta fourth-order result. . . . . . . . . . . . . . . . . . . . . . . . .. 80
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) . . . . . . . . . . . . . . . . . . . 85
Fig. 5.1 Campbell diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Fig. 5.2 Response with uniform inflow . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Fig. 5.3 Response with linear inflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
Fig. 5.4 Stability in hover condition with varying Lock number . . . . . . . . 96
Fig. 5.5 Stability in forward flight condition . . . . . . . . . . . . . . . . . . . . . . . 97
List of Tables

Table 5.1 Inputs for the elastic rotor problem . . . . . . . . . . . . . . . . . . . . . . . 94


Table 5.2 Non-dimensional rotating frequencies . . . . . . . . . . . . . . . . . . . . . 94

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.

© Springer Nature Singapore Pte Ltd. 2018 1


R. Ganguli and V. Panchore, The Rotating Beam Problem in Helicopter Dynamics,
Foundations of Engineering Mechanics, https://doi.org/10.1007/978-981-10-6098-4_1
2 1 Introduction

To understand the material in this book, a good understanding of structural


dynamics and helicopter aerodynamics is required. From Sects. 1.1–1.9, we review
the basics of structural dynamics. Since the elastic rotor blade is a continuous
system, we start with a single-degree-of-freedom system and end with a continuous
system. The basics of helicopter aerodynamics including momentum theory and
blade element theory are covered later in this chapter. We also discuss the rigid
rotor blade flap equation and end this chapter with the derivation of the elastic rotor
blade flap equation, including the forcing term on the right-hand side.

1.1 Free Vibration of a Single-Degree-of-Freedom System

Consider Fig. 1.1, which is a spring–mass system. The mass m is suspended by a


spring of stiffness k. The mass is then pulled by a distance x and released. The mass
then vibrates about an equilibrium position.
The governing differential equation is given by

m€x þ kx ¼ 0 ð1:2Þ

Assume the solution x ¼ C1 cosðxn tÞ þ C2 sinðxn tÞ


At t ¼ 0; xð0Þ ¼ C1 ; x_ ð0Þ=xn ¼ C2
Substituting C1 and C2 in the assumed solution, we get the system response as

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.

Fig. 1.1 Spring–mass system


1.1 Free Vibration of a Single-Degree-of-Freedom System 3

Fig. 1.2 Displacement-time


graph (spring–mass system)

We can write Eq. (1.3) as

x ¼ C sinðxn t þ /Þ; ð1:4Þ


qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi  
where C ¼ xð0Þ2 þ x_ ð0Þ x 2 and / ¼ tan 1 x_ ð0Þ
xð0Þxn are the amplitude and phase of
n

the response.

1.2 Free Vibration of a Damped,


Single-Degree-of-Freedom System

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

m€x þ c_x þ kx ¼ 0 ð1:5Þ

Assume the solution x ¼ Aest and substitute it into Eq. (1.5) yields

ðms2 þ cs þ kÞAest ¼ 0 ð1:6Þ

Fig. 1.3 Damped vibration


system
4 1 Introduction

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

Here, we define the damping ratio f ¼ ccc ¼ 2mx


c
n
; where cc is the critical
damping.
We write Eq. (1.7) as
qffiffiffiffiffiffiffiffiffiffiffiffiffi
s ¼ fxn  xn f2  1 ð1:8Þ

There are three possible situations depending on the value of f.


 c 2
If 2m  x2n [ 0; we get two real roots. In more compact form, if f [ 1; we get
two real roots. This is an over-damped system.
 c 2
If 2m  x2n \0; we get two complex roots. In more compact form, if f\1; we
get two complex roots. This is an under-damped system.
 c 2
If 2m  x2n ¼ 0; we get one repeated real root. In more compact form, if f ¼ 1;
we get one repeated real root. This is a critically damped system.
For an over-damped case, solution is given by
 ffi
pffiffiffiffiffiffiffi  ffi
pffiffiffiffiffiffiffi
fxn þ xn f 1 t
2
fxn xn f 1 t
2
x ¼ C1 e þ C2 e

or
 pffiffiffiffiffiffiffi
ffi pffiffiffiffiffiffiffi
ffi
x ¼ efxn t C1 exn t f 1 þ C2 exn t f 1 :
2 2
ð1:9Þ

This is also called the system response of the damped system.


For an under-damped case, solution is given by
 pffiffiffiffiffiffiffi2ffi pffiffiffiffiffiffiffi2ffi 
x ¼ efxn t C1 eixn t 1f þ C2 eixn t 1f

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

Fig. 1.4 Graphical


representation of an
over-damped system (a),
a critically damped system
(b), and an under-damped
system (c)

For a critically damped system, solution is given by

x ¼ efxn t ðC1 þ C2 tÞ ð1:11Þ

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.

1.3 Forced Vibration of a Single-Degree-of-Freedom


System

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

 P0 is the amplitude of the sinusoidal forcing and x


Here, f ðtÞ ¼ P0 sinðxtÞ;  is its
frequency.
We rewrite the governing equation as


m€x þ kx ¼ P0 sinðxtÞ ð1:12Þ
6 1 Introduction

Fig. 1.5 Forced vibration of


the spring–mass system

Solution to Eq. (1.12) is given by

x ¼ xh þ xp

where xh is the homogenous solution and xp is the particular solution.


The homogenous solution is identical to the solution to the system with no
forcing term:

xh ¼ C1 cosðxn tÞ þ C2 sinðxn tÞ ð1:13Þ

For the particular solution ðxp Þ; we assume the solution to be

 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

The complete solution is given by

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.

1.4 Forced Vibration of a Damped,


Single-Degree-of-Freedom System

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

The solution to Eq. (1.17) is given by

x ¼ xh þ xp

where xh is the homogenous solution and xp is the particular solution.


The homogenous solution was derived in Sect. 1.2.
In a real-life situation, we typically deal with an under-damped vibration.
Therefore, we take an under-damped vibration case in this section:
 qffiffiffiffiffiffiffiffiffiffiffiffiffi   qffiffiffiffiffiffiffiffiffiffiffiffiffi 
xh ¼ efxn t C1 sinðxn 1  f2 t þ C2 cos xn 1  f2 t ð1:18Þ

For the particular solution ðxp Þ; we assume the solution to be

Fig. 1.6 Forced vibration of


the mass–spring–damper
system
8 1 Introduction

 þ A2 sinðxtÞ
xp ¼ A1 cosðxtÞ  ð1:19Þ

Substituting xp in Eq. (1.17), we get


! !
P0 1  b2 P0 2fb
A2 ¼  2 and A1 ¼   2 :
k 1  b2 þ ð2fbÞ2 k 1  b2 þ ð2fbÞ2

The complete solution is given by

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.

1.5 Two-Degrees-of-Freedom System

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”

Fig. 1.8 a Two-degrees-of-freedom system. b Free-body diagram of two-degrees-of-freedom


spring–mass–damper system

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






m1 0 €x1 c þ c2 c2 x_ 1 k þ k2 k2 x1 P1


þ 1 þ 1 ¼ :
0 m2 €x2 c2 c2 x_ 2 k2 k2 x2 P2
ð1:22Þ
10 1 Introduction

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

Assume the solution



x1 C1 sinðxt  /Þ
¼ :
x2 C2 sinðxt  /Þ

Substituting the assumed solution into Eq. (1.23), we get




m1 x2 þ k1 þ k2 k2 C1 sinðxt  /Þ 0


¼ ð1:24Þ
k2 m2 x2 þ k2 C2 sinðxt  /Þ 0

For a non-trivial solution



m1 x2 þ k1 þ k2 k2
¼0
k2 m2 x2 þ k2

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

From Eq. (1.25), we can get two natural frequencies x1 and x2 :


From the homogenous Eq. (1.24), we can get the ratio of C1 and C2 :

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

The two-degrees-of-freedom system has two modes of vibration corresponding


to two natural frequencies x1 and x2 ; respectively, and can be expressed as
" # " #
ð1Þ ð1Þ
C1 C1
½C ð1Þ  ¼ ð1Þ ¼ ð1Þ ð1:28Þ
C2 rC1
" # " #
ð2Þ ð2Þ
ð2Þ C1 C1
½C  ¼ ð2Þ ¼ ð2Þ ð1:29Þ
C2 rC1

where ½Cð1Þ  and ½C ð2Þ  are the two modes of vibration.


Free vibration response of the two-degrees-of-freedom system is given by

" # " #
ð1Þ ð2Þ
x1 C1 C1
¼ ð1Þ sinðxt  /Þ þ ð2Þ sinðxt  /Þ ð1:30Þ
x2 rC1 rC1

The concepts and derivation for the two-degrees-of-freedom system can be


extended to a multi-degree-of-freedom system.

1.6 Free Vibration of a Continuous System

In a continuous system, we consider continuous distribution of mass, damping, and


elasticity. Rods, beams, and plates are typical examples of such systems. Helicopter
rotor blades are typically modeled as beams and so we discuss beams as an example
of a continuous system.
Governing differential equation of a beam is given by
 
@2 @2w
EI þ m€
w¼0 ð1:31Þ
@x2 @x2

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

This equation is an ordinary differential equation in space:

 ðxÞ mx2
d4 w
  ðxÞ ¼ 0
w ð1:34Þ
dx4 EI

 ðxÞ ¼ epx ; we get


Assume the solution w

mx2 px mx2
p4 epx  e ¼ 0 ) p4 ¼
EI EI

or

p4 ¼ k4
 1=4
mx2
where k ¼ EI

p ¼ k; k; ik; ik

The solution is given by

 ðxÞ ¼ G1 ekx þ G2 ekx þ G3 eikx þ G4 eikx


w

 ðxÞ ¼ C1 sin hðkxÞ þ C2 cos hðkxÞ þ C3 sinðkxÞ þ C4 cosðkxÞ


w ð1:35Þ

where G1 ; G2 ; G3 ; G4 ; C1 ; C2 ; C3 ; and C4 are constants. Here, we need four


boundary conditions to find the deflection.
Consider a simply-supported beam, and boundary conditions are shown in
Fig. 1.9. This type of boundary condition gives simple mode shapes and fre-
quencies and is given for illustration.
Here, boundary conditions are

Fig. 1.9 Simply-supported


beam with boundary
conditions
1.6 Free Vibration of a Continuous System 13


@ 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

We apply the boundary conditions to Eq. (1.35) and get

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

sin hðkLÞ  C1 þ cos hðkLÞ  C2 þ sinðkLÞ  C3 þ cosðkLÞ  C4 ¼ 0

sin hðkLÞ  C1 þ cos hðkLÞ  C2  sinðkLÞ  C3  cosðkLÞ  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

The first three natural frequencies are


rffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffi
EI EI EI
x1 ¼ p 2
4
;x2 ¼ 4p2
4
; and x3 ¼ 9p 2
mL mL mL4

The first three mode shapes are


px    
2px 3px
 1 ðxÞ ¼ C3 sin
w  2 ðxÞ ¼ C3 sin
;w  3 ðxÞ ¼ C3 sin
; and w :
L L 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

Fig. 1.11 Fixed-free beam


with boundary conditions

rpx
wðx; tÞ ¼ sin eixt : ð1:38Þ
L

The rotor blade is typically a fixed-free beam (hingeless rotor) or a


simply-supported-free beam (articulated rotor). We discuss these two types of
boundary conditions next.
Fixed-free beam
Boundary conditions for a fixed-free beam are (Fig. 1.11)
1.6 Free Vibration of a Continuous System 15


@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:

We rewrite assumed solution (1.35) as

 ðxÞ ¼ C1 fsin hðkxÞ  sinðkxÞg þ C2 fcos hðkxÞ  cosðkxÞg


w ð1:39Þ

wðxÞ @ 3 wðxÞ
Applying boundary conditions EI@ @x
2
2 ðx ¼ LÞ ¼ 0 and EI @x3 ðx ¼ LÞ ¼ 0 to

Eq. (1.39), we get




sin hðkLÞ þ sinðkLÞ cos hðkLÞ þ cosðkLÞ C1 0


¼
cos hðkLÞ þ cosðkLÞ sin hðkLÞ  sinðkLÞ C2 0

For a non-trivial solution



sin hðkLÞ þ sinðkLÞ cos hðkLÞ þ cosðkLÞ

cos hðkLÞ þ cosðkLÞ sin hðkLÞ  sinðkLÞ ¼ 0

or

1 þ cos hðkLÞ cosðkLÞ ¼ 0 ð1:40Þ

We solve Eq. (1.40) numerically to get the first three solutions

kL ¼ 1:875104; 4:694091; and 7:48547



wðxÞ
Using boundary condition EI@ @x
3
3 ðx ¼ LÞ ¼ 0; we get

cos hðkLÞ þ cosðkLÞ


C2 ¼ C1
sin hðkLÞ  sinðkLÞ

We rewrite Eq. (1.39) as

 ðxÞ ¼C1 f½sin hðkxÞ  sinðkxÞ


w

cos hðkLÞ þ cosðkLÞ
 ½cos hðkxÞ  cosðkxÞ ð1:41Þ
sin hðkLÞ  sinðkLÞ

The first three natural frequencies are given by


16 1 Introduction

rffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffi


EI2 2 EI 2 EI
x1 ¼ ð1:8751Þ ;x2 ¼ ð4:6940Þ ; and x3 ¼ ð7:8547Þ
mL4 mL4 mL4

The first three mode shapes are given by


nh  x  x i
 1 ðxÞ ¼ C1
w sin h 1:8751  sin 1:8751
L L
cos hð1:8751Þ þ cosð1:8751Þ h  x  x i
 cos h 1:8751  cos 1:8751 ;
sin hð1:8751Þ  sinð1:8751Þ L L
nh  x  x i
 2 ðxÞ ¼ C1 sin h 4:6940
w  sin 4:6940
L L
cos hð4:6940Þ þ cosð4:6940Þ h  x  x i
 cos h 4:6940  cos 4:6940 ;
sin hð4:6940Þ  sinð4:6940Þ L L

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.

Fig. 1.12 a First mode shape


of fixed-free beam. b Second
mode shape of fixed-free
beam. c Third mode shape of
fixed-free beam
1.6 Free Vibration of a Continuous System 17

Fig. 1.13 Simply-supported-free beam with boundary conditions

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:

We rewrite assumed solution (1.35) as

 ðxÞ ¼ C1 sin hðkxÞ þ C3 sinðkxÞ


w ð1:42Þ

wðxÞ @ 3 wðxÞ
Applying boundary conditions EI@ @x
2
2 ðx ¼ LÞ ¼ 0 and EI @x3 ðx ¼ LÞ ¼ 0 to
Eq. (1.42), we get


sin hðkLÞ sinðkLÞ C1 0


¼
cos hðkLÞ cosðkLÞ C3 0

For a non-trivial solution



sin hðkLÞ sinðkLÞ
¼0 ð1:43Þ
cos hðkLÞ cosðkL

We solve Eq. (1.43) numerically to get the first three solutions

kL ¼ 3:9266; 7:0685; and 10:2101


18 1 Introduction

wðxÞ
Using boundary condition EI@ @x
3
3 ðx ¼ LÞ ¼ 0; we get

cos hðkLÞ
C3 ¼ C1
cosðkLÞ

We rewrite Eq. (1.42) as



cos hðkLÞ
 ðxÞ ¼ C1 sin hðkxÞ þ
w sinðkxÞ ð1:44Þ
cosðkLÞ

The first three natural frequencies are given by


rffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffi
2EI 2 EI 2 EI
x1 ¼ ð3:9266Þ ;x2 ¼ ð7:0685Þ ; and x3 ¼ ð10:2101Þ
mL4 mL4 mL4

The first three mode shapes are given by



x cos hð3:9266Þ  x
 1 ðxÞ ¼ C1 sin h 3:9266 þ
w sin 3:9266 ;
L cosð3:9266Þ L

x cos hð7:0685Þ  x
 2 ðxÞ ¼ C1 sin h 7:0685 þ
w sin 7:0685 ;
L cosð7:0685Þ L

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).

1.7 Hamilton’s Principle

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

Fig. 1.14 a First mode shape


of simply-supported-free
beam. b Second mode shape
of simply-supported-free
beam. c Third mode shape of
simply-supported-free beam

Fig. 1.15 Simple pendulum

where T is the kinetic energy and U is the potential energy.


We can write Eq. (1.45) as
Z
d Ldt ¼ 0 ð1:46Þ

where L is the Lagrangian of the system.


The equation of motion is derived using Lagrange’s equation
 
d @L @L
 ¼0 ði ¼ 1; 2; 3Þ ð1:47Þ
dt @ q_ i @qi

Consider Fig. 1.15, the case of a simple pendulum.


Here, T ¼ 12 mL2 h_ 2 and U ¼ mgLð1  cosðhÞÞ
20 1 Introduction

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

1.8 Diagonalization of a Symmetric Matrix

In a multi-degree-of-freedom system, the governing equation is given by

€ þ ½C½X
½M½X _ þ ½K½X ¼ ½FðtÞ ð1:49Þ

Here, ½M and ½K are symmetric matrices.


It would be easier to solve Eq. (1.51) than Eq. (1.50):
2 32 3 2 32 3 2 3
m11 0 0 €x1 k11 x1 f1 ðtÞ
4 0 m22 0 54 €x2 5 þ 4 k21 k22 54 x2 5 ¼ 4 f2 ðtÞ 5 ð1:50Þ
0 0 m33 €x3 k31 k32 k33 x3 f3 ðtÞ
2 32 3 2 32 3 2 3
m1 0 0 €f1 k1 0 0 f1 f1 ðtÞ
4 0 m2 0 54 €f2 5 þ 4 0 k2 0 54 f2 5 ¼ 4 f2 ðtÞ 5 ð1:51Þ
0 0 m3 €f3 0 0 k3 f3 f3 ðtÞ

We can diagonalize a symmetric matrix ½ A with a matrix ½P such that


1
½P ½ A½P is a diagonal matrix.
 
½Pnxn ¼ ½v1 nx1 ½v2 nx1 ½v3 nx1 . . .½vn nx1 ð1:52Þ

where ½v1 nx1 ; ½v2 nx1 . . .; &½vn nx1 are the eigenvectors of the matrix ½ A.
We know

½ Anxn ½v1 nx1 ¼ k1 ½v1 nx1 ð1:53Þ

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

We see that ½P1


nxn ½ Anxn ½Pnxn is a diagonal matrix.
The eigenvectors of a real symmetric matrix corresponding to distinct eigen-
values are orthogonal.
We rewrite Eq. (1.41)

½ Anxn ½v1 nx1 ¼ k1 ½v1 nx1

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

ðk2  k1 Þ½v1 Tnx1 ½v2 nx1 ¼ 0 ðk2 6¼ k1 Þ

or

½v1 Tnx1 ½v2 nx1 ¼ 0 ð1:56Þ

Equation (1.56) shows that eigenvectors corresponding to distinct eigenvalues


are orthogonal.
From Eq. (1.56), we write

½PTnxn ½Pnxn ¼ ½I nxn

or

½P1 T
nxn ¼ ½Pnxn

We rewrite Eq. (1.55) as


22 1 Introduction

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

1.9 Transformation of Coordinates

We write a multi-degree-of-freedom equation

€ þ ½K½X ¼ ½FðtÞ
½M½X ð1:58Þ

Typically, mass and stiffness matrices are fully populated.


We write ½XðtÞ ¼ ½/½fðtÞ, where ½XðtÞ are the general coordinates, ½/ is the
modal transformation matrix, and ½fðtÞ is the principal coordinate.
 
Here, ½unxn ¼ ½u1 nx1 ½u2 nx1 . . .½un nx1 and ½fðtÞnx1 ¼ ½f1 ðtÞf2 ðtÞ. . .fn ðtÞT
After the coordinate transformation, we get Eq. (1.59):

½M½/½€fðtÞ þ ½K½/½fðtÞ ¼ ½FðtÞ ð1:59Þ

or

½/T ½M½/½€fðtÞ þ ½/T ½K½/½fðtÞ ¼ ½/T ½FðtÞ ð1:60Þ

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

Equation (1.61) is a set decoupled equation and is easier to solve than


Eq. (1.58). The modal transformation converts a multi-degree-of-freedom system
into a series of single-degree-of-freedom systems.
1.10 Momentum Theory for Axial Flight 23

1.10 Momentum Theory for Axial Flight

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

Fig. 1.16 Air flow through


the control volume in
momentum theory for axial
flight
24 1 Introduction

We write Eq. (1.63) for flow from area A2 to area A3 ; to get

qðV1 þ Vu ÞA2 ¼ qðV1 þ Vl ÞA3

Since A2 ¼ A3 ; we get

ðV1 þ Vu Þ ¼ ðV1 þ Vl Þ ) Vu ¼ Vl ; and there is no jump in the velocity across


the rotor.
We write the momentum conservation equation as
ZZZ ZZ
@  
T¼ ~
Vqdv þ q~
V ~
V:~
n dA ð1:64Þ
@t CS
CV

 @ 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 Eq. (1.65) for flow from area A1 to area A4 to get

_ f ¼ qA1 V1 Vf ¼ qA4 ðV1 þ Vf ÞVf


T ¼ mV ð1:66Þ

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 do not use Bernoulli equation between 1 and 4 because energy and


momentum are added by the rotor disk. Since P4 ¼ P1 from Eqs. (1.67) and (1.68),
we get
q 2 
P3  P2 ¼ Vf þ 2Vf V1 ð1:69Þ
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

From Eqs. (1.66) and (1.70), we get


1.10 Momentum Theory for Axial Flight 25

q  2   
A2 Vf þ 2Vf V1 ¼ qA4 V1 þ Vf Vf ð1:71Þ
2

From the equation of continuity, we write

qA2 ðV1 þ Vu Þ ¼ qA4 ðV1 þ Vf Þ ð1:72Þ

From Eqs. (1.59) and (1.60), we get

Vf ¼ 2Vu ð1:73Þ

or

w ¼ 2vi

where vi ¼ Vu ¼ Vl (induced velocity at the rotor plane) and w ¼ Vf (induced


velocity at the slipstream). These velocities are induced by the rotor as it tries to
create lift.
Induced velocity at the slipstream is two times the induced velocity at the rotor
plane.
Substituting V1 ¼ 0 in Eq. (1.70), we get the induced velocity for the hover
case

T  
v2h ¼ A2 ¼ A3 ¼ A ¼ pR2 ¼ Rotor disk area ð1:74Þ
2qA

where R is the rotor radius.

1.11 Momentum Theory for Forward Flight

It is possible to extend the momentum theory concept to forward flight by making


some assumption. Consider Fig. 1.17, which shows the forces acting on the rotor
plane in forward flight.
Here, T is the thrust, P is the propulsive force, D is the drag, W is the weight of
the helicopter, L is the lift, and w is the slipstream velocity.
We write the conservation of momentum equation in forward flight as
ZZ
 
T¼ q~
V ~ n dA ) T ¼ m_ ðVout  Vin Þ
V:~ ð1:75Þ
CS
26 1 Introduction

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

From Eq. (2.62), we get

v2h
vi ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1:77Þ
½V1 cosðaÞ2 þ ½vi þ V1 sinðaÞ2

Here, we introduce the terms inflow and advance ratio.


1.11 Momentum Theory for Forward Flight 27

Inflow ðkÞ is the non-dimensional coefficient of the velocity perpendicular to the


plane of rotor:
vi
kh ¼ ðinflow in case of hover conditionÞ ð1:78Þ
XR
V1 sinðaÞ þ vi
k¼ ðinflow in forward flight conditionÞ ð1:79Þ
XR

where V1 is the forward velocity, vi is the induced velocity, X is the angular


velocity, and a is the angle of attack.
Typically, non-dimensional form of equations is often used. Advance ratio ðlÞ is
non-dimensional coefficient of the velocity parallel to the plane of the rotor:

V1 cosðaÞ
l¼ ð1:80Þ
XR

From Eqs. (1.77), (1.79), and (1.80), we get


qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k ¼ l tanðaÞ þ CT =2 l2 þ k2 ð1:81Þ

where CT is the non-dimensional coefficient of thrust:

T
CT ¼
qAX2 R2

The model in (1.81) assumes that inflow is uniform. In reality, inflow is


non-uniform and models have been developed to account for this reality. We can
write an equation for inflow which will vary over the length of the blade and over
the azimuthal angle:
 
CT =2 Kx x cosðwÞ Ky x sinðwÞ
k ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi 1 þ þ ðlinear inflow modelÞ ð1:82Þ
l2 þ k R R

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

1.12 Newton–Raphson Method

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

Assume the initial value of inflow to be k0

f ðk0 þ hÞ ¼ 0

We write the Taylor series as

h2 00
f ðk0 Þ þ hf 0 ðk0 Þ þ f ðk0 Þ þ Higher Order Terms ¼ 0
2

We assume the solution up to the first order

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.

1.13 Blade Element Theory

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

We write the drag produced by the section dy

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

dFZ ¼ dL cosð/Þ  dD sinð/Þ ð1:86Þ


30 1 Introduction

Fig. 1.19 Tangential and


perpendicular components of
the flow velocity

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.

UT ¼ Xy þ V1 cosðaÞ sinðwÞ ð1:89Þ

where Xy is the result of angular velocity of rotor and V1 cosðaÞ sinðwÞ is the result
of flow velocity:

UP ¼ V1 sinðaÞ þ vi þ yb_ þ V1 sinðbÞ cosðwÞ cosðaÞ ð1:90Þ

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

1.14 Derivation of Equation of Motion of Flapping


Rigid Blade

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

(b) Centrifugal Force

Force on the small segment having mass dm ¼ ðXyÞy dm ¼ X2 ydm


2

Associated moment at the hinge ¼ X2 ydmðy sinðbÞÞ ¼ X2 y2 dmb

where, ðy sinðbÞ  ybÞ

Fig. 1.20 Forces acting on a


small element of a rigid rotor
blade
32 1 Introduction

ZR
Integrating all the small segments; moment due to full rotor ¼ X2 y2 bdm ¼ IX2 b
0
ð1:94Þ

(c) Aerodynamic Force


Force on the small segment having mass dm ¼ Ldy

Associated moment at the hinge ¼ Ldyðy cosðbÞÞ ¼ Lydy

where ðy cosðbÞ  yÞ

ZR
Integrating all the small segments; moment due to full rotor ¼ Lydy: ð1:95Þ
0

We write moment equation about the hinge to get

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

We write Eq. (1.84) as

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

where aerodynamic force per unit length is given by


 
1 UP
L ¼ qUT2 cCLa h  ð1:98Þ
2 UT

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

Substituting lift term in right-hand side of Eq. (1.97), we get

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

Here, h ¼ h1 þ htw Ry ; where h1 ¼ h0 þ h1s sinðwÞ þ h1c cosðwÞ;


The helicopter is controlled through h0 ; h1s ; and h1c which are inputs by the pilot
to the main rotor via the swashplate. Here, h0 is called the collective pitch, h1c the
lateral cyclic, and h1s the longitudinal cyclic. Also, htw represents the built-in twist
in the rotor blade.
34 1 Introduction

From Eqs. (1.91) and (1.101), we get

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

qcCLa R4 1 l2 sin2 ðwÞ l sinðwÞ


Term1 ¼ h1 þ þ
I 8 4 3


1 l sin ðwÞ l sinðwÞ
2 2
þ htw þ þ
10 6 4
4
Here, we define a term, Lock number c ¼ qcCILa R ; which represents the ratio of
aerodynamic forcing to inertial forcing:

1 l2 sin2 ðwÞ l sinðwÞ


Term1 ¼c h1 þ þ
8 4 3


1 l sin ðwÞ l sinðwÞ
2 2
þ htw þ þ ð1:102Þ
10 6 4

We write

ZR
1 1
Term2 ¼ 2 qcCLa UP UT ydy ð1:103Þ
IX 2
0

From Eqs. (1.91), (1.92), and (1.103), we get

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

qcCLa R4 1 l sinðwÞ _ 1 l sinðwÞ


Term2 ¼ k þ þb þ
I 6 4 8X 6X


1 l sinðwÞ
þ bl cosðwÞ þ
6 4

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

From Eqs. (1.102), (1.104), and (1.100), we get

ZR

1 1 l2 sin2 ðwÞ l sinðwÞ


Lydy ¼ c h1 þ þ
IX2 8 4 3
0


1 l2 sin2 ðwÞ l sinðwÞ
þ htw þ þ
10 6 4

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

From Eqs. (1.105) and (1.97), we get

d2 b b
þ b ¼ cM ð1:106Þ
dw2

If we put c ¼ 0; the above equation simulates a flapping blade in a vacuum.


Solution to Eq. (1.106) is given by Fourier series

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.

1.15 Derivation of Elastic Rotor Blade Equation

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

Fig. 1.21 a Deflection of an elastic rotor blade. b Force diagram


1.15 Derivation of Elastic Rotor Blade Equation 37

@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

From Euler–Bernoulli beam theory, we can write

@2w
M ¼ EI ð1:111Þ
@x2

From Eqs. (1.109) and (1.110), we write


 
@2M @2w @ @w
¼ m 2 þ G ð1:112Þ
@x2 @t @x @x

From Eqs. (1.111) and (1.112), we write


   
@2 @2w @2w @ @w
EI 2 ¼ m 2 þ G ð1:113Þ
@x2 @x @t @x @x

From Eqs. (1.108) and (1.113), we write


00 R 1 1
  Z
@2 @2w @2w @ @@ @w
EI 2 ¼ m 2 þ mX2 xdxA A ð1:114Þ
@x 2 @x @t @x @x
x

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

Forced vibration equation is given by


2 3
ZR
00
€ þ X2 4mxw0  w00
ðEIw00 Þ þ mX2 w mxdx5 ¼ Fðx; wÞ ð1:117Þ
x

where ðw_ ¼ @w=@w; w0 ¼ @w=@xÞ


Forcing term is given by
qac
Fðx; wÞ ¼  ðUT2 h  UP UT Þ ð1:118Þ
2
     
Here, UT ¼ XR Rx þ l sinðwÞ ; UP ¼ XR wR_ þ k þ lw0 cosðwÞ ,
and h ¼ h1 þ htw Rx ; where h1 ¼ h0 þ h1s sinðwÞ þ h1c cosðwÞ;
 
[Note that w_ ¼ @w
@w ]
Substituting values of UP ; UT , and h in Eq. (1.118), we get

h1 2h1 l sinðwÞ
Fðx; wÞ ¼ Const x2 2 þ x
R R
k   htw htw l2 sin2 ðwÞ
 x þ h1 l2 sin2 ðwÞ  kl sinðwÞ þ x3 3 þ x
R R R
2 2htw l sinðwÞ 1 l sinðwÞ @w
þx þ Const x 2 þ
R2 R R @w

l cosðwÞ l2 sinð2wÞ @w
þ Const x þ ð1:119Þ
R 2 @x

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

where Const ¼ qac


2 X R ;
2 2

a1x ¼ Const  x2 ; a2x ¼ Const  x; a3x ¼ Const  x; a4x ¼ Const;


a5x ¼ Const  x3 ; a6x ¼ Const  x; a7x ¼ Const  x2 ;
a1w ¼ h1 =R2 ; a2w ¼ 2h1 l sinðwÞ=R; a3w ¼ ki =R; a4w ¼ h1 l2 sin2 ðwÞ  ki l sinðwÞ;
a5w ¼ htw =R3 ; a6w ¼ l2 htw sin2 ðwÞ=R; a7w ¼ 2lhtw sinðwÞ=R2 ;
b1x ¼ Const  x; b2x ¼ Const; b1w ¼ 1=R2 ; b2w ¼ l sinðwÞ=R;
c1x ¼ Const  x; c2x ¼ Const; c1w ¼ l cosðwÞ=R; and c2w ¼ l2 sinð2wÞ=2

Equation (1.120) shows the presence of motion-dependent forces. For example,


forces are a function of w_ and w0 : Such terms need to be taken to the left-hand side
of Eq. (1.117), where they will influence the damping and stiffness of the rotor
system. The presence of motion-dependent forces makes the rotor dynamic problem
an aeroelastic problem.
Chapter 2
Finite Element Analysis in Space

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.

2.2 Finite Element in Space

Finite element method is a numerical method for getting approximate solution of


differential equations. Analytical solution of the equation gives us the exact solution
at any point in that domain, while finite element method gives us the approximate
solution at discrete number of points in that domain. In finite element method, we
divide the full domain into a number of elements, which are connected through nodal
points. Then, we write equations for each element and combine them to get the
solution. Finite element approach is a weak formulation of the physical problem.

2.3 Strong Form of the Equation

Consider Fig. 2.1, a case of an elastic bar subjected to uniform load.


The governing differential equation (2.1) along with boundary conditions (2.2)
and (2.3) gives the strong form of the problem

d2 u
EA ¼ f0 ð2:1Þ
dx2

© Springer Nature Singapore Pte Ltd. 2018 41


R. Ganguli and V. Panchore, The Rotating Beam Problem in Helicopter Dynamics,
Foundations of Engineering Mechanics, https://doi.org/10.1007/978-981-10-6098-4_2
42 2 Finite Element Analysis in Space

Fig. 2.1 Elastic bar


subjected to uniform load

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.

2.4 Weak Form of the Equation

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.

2.5 Galerkin’s Method

In Galerkin’s method, we find the solution u ¼ ~u, such that

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.

2.6 Shape Function in 1 Dimension

In finite element, we get the solution at nodal points of an element. Interpolation


within the element is achieved by shape function.
Here, we take a bar element for illustration. Governing equation of linear-elastic
bar element is
 
d du
AE ¼0 ð2:9Þ
dx dx

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

Fig. 2.2 Bar element for


shape function formulation

Linear displacement along the x-axis is assumed as

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

From (2.11) and (2.12), we get a0 ¼ q1 ; a1 ¼ q2 q


L
1

We write  the  linear displacement


  within the finite element as
uðxÞ ¼ q1 þ q2 L q1 x ) uðxÞ ¼ q1 1  Lx þ q2 Lx ) uðxÞ ¼ H1 q1 þ H2 q2
or
 
q1
uðxÞ ¼ ½ H1 H2  ð2:13Þ
q2
 
where H1 ¼ 1  Lx and H2 ¼ Lx are shape functions for the bar elements.
Typically, polynomials are used as shape functions in finite element methods.

2.7 Shape Function Formulation for Beam Element

The static governing differential equation of an Euler–Bernoulli beam is given by


 
d2 d2 v
EI 2 ¼ 0 ð2:14Þ
dx2 dx

Here, in Fig. 2.3, we consider a beam element


Each element has two nodes; each node has vertical displacement and rotation.

Fig. 2.3 Beam element for


shape function formulation
2.7 Shape Function Formulation for Beam Element 45

Total degree of freedom per element is 4. Assume the transverse displacement


vðxÞ to be

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

From Eqs. (2.16)–(2.19), we write


2 3 2 32 3
q1 1 0 0 0 a0
6 q2 7 6 0 1 0 0 7 6 a1 7
6 7¼6 76 7 ð2:20Þ
4 q3 5 4 1 L L2 L3 54 a2 5
q4 0 1 2L 3L2 a3

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

From Eq. (2.15), we write


2 3
a0

6 a1 7
vðxÞ ¼ 1 x x2 x3 6 7
4 a2 5 ð2:22Þ
a3

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

2.8 Properties of Shape Function in 1D

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

Fig. 2.4 Two elements in a


bar for FEM in space

For formulation of shape function, we follow Eqs. (2.10)–(2.13).


h We substitute
i
ð1Þ ð1Þ
x ¼ 0 and x ¼ L=2 and get the shape function for first element H1 H2 , and
hwe substitutei x ¼ L=2 and x ¼ L and get the shape function for second element
ð2Þ ð2Þ
H1 H2 .
 
ð1Þ ð1Þ 2x 2x
uðxÞ ¼ H1 q1 þ H2 q2 ¼ 1 q1 þ q2 ð2:26Þ
L L

For second element, we write


   
ð2Þ ð2Þ 2x 2x
uðxÞ ¼ H1 q2 þ H2 q3 ¼ 2 q2 þ  1 q3 ð2:27Þ
L L

Put x ¼ L2 into Eqs. (2.26) and (2.27) to get

uðxÞ ¼ q2 :

3. Completeness
(a) Rigid body mode

H1 þ H2 ¼ 1

If the element moves by an unit displacement (q1 ; q2 ¼ 1), displacement at any


point in the element should be one.

uðxÞ ¼ H1 q1 þ H2 q2 ¼ 1 ðfor q1 ; q2 ¼ 1Þ

(b) Constant strain state


Consider Fig. 2.5, if q1 ¼ L=2 and q2 ¼ L, then

q1  q2 L  L2
eðxÞ ¼ L ¼ L ¼1
2 2
48 2 Finite Element Analysis in Space

Fig. 2.5 Bar element (shape


function properties)

Check of strain state with displacement approximation


 
2x 2x
uðxÞ ¼ H1 q1 þ H2 q2 ¼ 1  q1 þ q2
L L
 
2x L 2x
uðxÞ ¼ 1  þ L
L 2 L

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.

2.9 Finite Element Formulation of Rotating Beam

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

Kinetic energy is given by

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

From Eq. (2.25), we have shape function of the beam element


2 3
q1
6 q2 7
wðxÞ ¼ ½ H1 H2 H3 H4 6 7
4 q3 5 ¼ ½H ½q
q4

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

We write Eq. (2.28) as

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

We write Eq. (2.29) as

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

T ¼ ½q_ T14 ½M 44 ½q_ 41 ð2:35Þ

Rewriting the Lagrange’s equation


 
d @L @L
 ¼0 ð1:47Þ
dt @ q_ i @qi

where L ¼ T  V. From Eqs. (2.32), (2.35), and (1.35), we get the free vibration
problem

½M ½€q þ ½K ½q ¼ 0 ð2:36Þ

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

½K  ¼ ½K1  þ ½K2  ð2:39Þ


2.9 Finite Element Formulation of Rotating Beam 51

Assuming EI and m constant over the length of the blade:


where
2 3
RR 00 2 RR 00 00 RR 00 00 RR 00 00
6 ðH1 Þ dx H1 H2 dx H1 H3 dx H1 H4 dx 7
60 0 0 0 7
6 RR RR RR 7
6 00 00 00 00 00 7
6 ðH2 Þ2 dx H2 H3 dx H2 H4 dx 7
6 7
½K1  ¼ EI 6
6
0 0 0 7;
6 RR RR 00 00 7
H3 H4 dx 7
00
6 ðH3 Þ2 dx 7
6 7
6 0 0
7
4 R 00 2 5
R
ðH4 Þ dx
0
2 3
RR 2 0 RR 0 0 RR 0 0 RR 0 0
6 GðH1 Þ dx GH1 H2 dx GH1 H3 dx GH1 H4 dx 7
60 0 0 0 7
6 RR RR RR 7
6 0 0 0 0 0 7
6 GðH2 Þ dx 2
GH2 H3 dx GH2 H4 dx 7
6 7
½K2  ¼ 6
6
0 0 0 7;
7
6 RR RR
GH3 H4 dx 7
0 0 0
6 GðH3 Þ2 dx 7
6 7
6 0 0
7
4 RR 0 2 5
GðH4 Þ dx
0
2 3
RR 2 RR RR RR
6 ðH1 Þ dx H1 H 2 dx H1 H4 dx 7
H1 H3 dx
60 0 0 7
0
6 RR RR 7
RR
6 7
6 ðH2 Þ2 dx H2 H3 dx H2 H4 dx 7
6 7
½M  ¼ m6
6
0 0 0 7
7
6 RR RR
6
2
ðH3 Þ dx H3 H4 dx 77
6 7
6 0 0
7
4 RR 5
ðH4 Þ2 dx
0

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.

2.10 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

Fig. 2.6 Centrifugal force on


the rotating beam

ZR
G¼ mX2 xdx ð2:40Þ
x

For constant mass per unit length


 
R2 x2
G ¼ mX2  : ð2:41Þ
2 2

2.11 Shape Function Formulation for Two Elements

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

We integrate shape functions over the domain


R R
½H2T dx
L=2 L
Element 1 0 ½HT1 dx Element 2 L=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

Fig. 2.7 Shape function


formulation of two elements

We see that

ZL=2 ZL
½H T1 dx ¼ ½H T2 dx ð2:42Þ
0 L=2

We write ½K1  matrix

ZR
½H 00  ½H 00 dx
T
½K1  ¼ EI
0

We notice that

½K1 1st(element) ¼ ½K1 2nd(element)

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

We write ½K2  matrix

ZR
G½H 0  ½H 0 dx
T
½K2  ¼
0

We notice that

½K2 1st(element) 6¼ ½K2 2nd(element)

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.

2.12 FEM Formulation of Rotating Beam with Only One


Shape Function (for Free Vibration)

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.

Fig. 2.8 FEM formulation of


rotating beam with the shape
function of one element (for
free vibration)
2.12 FEM Formulation of Rotating Beam with Only One … 55

EI is stiffness, and mi is mass of particular element.


Rewriting ½K2  matrix

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

GðxÞ ¼ mX qdq ) GðxÞ ¼


2
mX qdq 
2
mX2 qdq
xi þ x xi xi

or

GðxÞ ¼ Term1  Term2 ð2:46Þ

ZR N Z
X
xj þ 1

Term1 ¼ mX qdq ) Term1 ¼


2
mX2 xdx
j¼i
xi xj

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

Term2 ¼ mi2X ð2xxi þ x2 Þ (It will be a varying term because of x)


2

We write Eq. (2.46) as

X2 Ai mi X2
GðxÞ ¼  ð2xxi þ x2 Þ ð2:47Þ
2 2
56 2 Finite Element Analysis in Space

Now we rewrite ½K2 

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

where i ¼ 1; 2; 3 (for different elements).


From this formulation, we get
2 3
13l 11l2 9l 13l2
6 35 210 70 420
7
6 l3 13l2 l3 7
6 7
½ M  ¼ m6 105 420 140
7;
6 13l 11l2 7
4 35 210 5
l3
105
2 12 6 12 6 3
l3 l2 l3 l2
6 4 6 2 7
6 l 7
½K1  ¼ EI 6 l l2
6 7
4 12
l3 l2
5
4
l

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

When X ¼ 0; ½K2  ¼ ½0 and the formulation reduces to the well-known


non-rotating beam element found in [15]. When X 6¼ 0; the centrifugal stiffening
causes a spatially varying stiffness matrix ½K2 .
2.13 Calculation of Mode Shapes and Frequencies 57

2.13 Calculation of Mode Shapes and Frequencies

We rewrite Eq. (2.36)

½M ½€q þ ½K ½q ¼ 0 ðFree vibration problemÞ

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

V ¼ eigvecð½ AÞ ð2:51Þ

Non-dimensional rotating frequency g and non-dimensional rotating speed s are


given by
rffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffi
mR4 mR4
s¼X ;g ¼ f
EI EI

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:

2.14 FEM Formulation of Aerodynamic Force for Rotor


Problem

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

We rewrite Eq. (1.108)

Fðx; wÞ ¼ a1x a1w


þ a2x a2w þ a3x a3w þ a4x a4w þ a5x a5w þ a6x a6w þ a7x a7w
þ ðb1x b1w þ b2x b2w Þw_ þ ðc1x c1w þ c2x c2w Þw0 ð1:120Þ

Element load vector is given by

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

Here, w ¼ ½H½q and w_ ¼ @w


@w :
We write Eq. (2.54) as

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

From Eqs. (2.52) and (2.55), we write

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

_ on the right-hand side of


We can see the presence of term involving ½q and ½q
this equation. These are motion-dependent forces. The presence of three forces
changes the problem from a structural dynamic problem to an aeroelastic problem.

_ 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 

½C ¼ b1w ½Ca1  þ b2w ½Ca2 

½D ¼ c1w ½Da1  þ c2w ½Da2 

After transformation of coordinate, we write Eq. (2.57) as

X2 ½M½/½€f þ ½K½/½f ¼ ½Q þ ½C½/½f


_ þ ½D½/½f ð2:58Þ

where ½q ¼ ½/½f, and ½/ being the eigenvectors. We can then write

X2 ½/T ½M½/½€f þ ½/T ½K½/½f ¼ ½/T ½Q þ ½/T ½C½/½f


_ þ ½/T ½D½/½f ð2:59Þ

or

X2 ½M1 ½€f þ ½K1 ½f ¼ ½Q1  þ ½C1 ½f


_ þ ½D1 ½f ð2:60Þ

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

X2 ½M1 1 ½€f þ ½K1 ½f ¼ ½Q1  þ ½C1 ½f


_ þ ½D1 ½f ð2:60Þ

Second-order Eq. (2.60) can be written as two first-order Eqs. (3.2) and (3.3)

_  ½C1 ½P þ ð½K1   ½D1  Þ½f ¼ ½Q1 


X2 ½M1 ½P ð3:2Þ
1

_
½P ¼ ½f ð3:3Þ

Weak formulation of Eqs. (3.2) and (3.3) in time gives Eqs. (3.4) and (3.5)

© Springer Nature Singapore Pte Ltd. 2018 61


R. Ganguli and V. Panchore, The Rotating Beam Problem in Helicopter Dynamics,
Foundations of Engineering Mechanics, https://doi.org/10.1007/978-981-10-6098-4_3
62 3 Finite Element in Time

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.

3.2 Selection of Shape Function in Time

Assume the displacement and the velocity as a function of w (azimuthal angle). If


the number of nodes is n, then degree of polynomial will be n  1:
Example Take a case of 4 nodes ð0; 2p=3; 4p=3; 2pÞ; where degree of polynomial
is 3.
Here, u is the approximation for the displacement and v is the approximation for
the velocity

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

Values of the displacements at the different nodal points are given by

uð0Þ ¼ a1

uð2p=3Þ ¼ a1 þ 2pa2 =3 þ 4p2 a3 =9 þ 8p3 a4 =27

uð4p=3Þ ¼ a1 þ 4pa2 =3 þ 16p2 a3 =9 þ 64p3 a4

uð2pÞ ¼ a1 þ 2pa2 þ 4p2 a3 þ 8p3 a4

Above equations can be written in a matrix form


2 3 2 32 3
uð0Þ 1 0 0 0 a1
6 uð2p=3Þ 7 6 1 2p=3 4p2 =9 8p3 =27 7 6 a2 7
6 7 6 76 7
4 uð4p=3Þ 5 ¼ 4 1 4p=3 16p2 =9 64p3 =27 54 a3 5
ð3:8Þ
uð2pÞ 1 2p 4p2 8p3 a4
3.2 Selection of Shape Function in Time 63

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

u ¼ N1 uð0Þ þ N2 uð2p=3Þ þ N3 uð4p=3Þ þ N4 uð2pÞ ð3:11Þ

Similarly, formulation for the velocity is given by

v ¼ N1 vð0Þ þ N2 vð2p=3Þ þ N3 vð4p=3Þ þ N4 vð2pÞ ð3:12Þ

where N1 ; N2 ; N3 ; and N4 are the shape functions in time.

N1 ¼ 9w2 =4p2  11w=4p  9w3 =16p3 þ 1

N2 ¼ 9w=2p  45w2 =8p2 þ 27w3 =16p3

N3 ¼ 9w2 =2p2  9w=4p  27w3 =16p3

N4 ¼ w=2p  9w2 =8p2 þ 9w3 =16p3


2 3
u1
6 u2 7
u ¼ ½ N1 N2 N3 N4 6 7
4 u3 5 ð3:13Þ
u4

or

u ¼ ½N½u ð3:14Þ
64 3 Finite Element in Time

3.3 Finite Element in Time Example

To illustrate the FEM in time formulation and concepts, we consider an example of


an ordinary differential equation with periodic forcing.

€u þ 2u_ þ 3u ¼ f ðwÞ ð3:15Þ

where

f ðwÞ ¼ sinðwÞ þ cosðwÞ þ sinð2wÞ þ cosð2wÞ þ sinð3wÞ þ cosð3wÞ


þ sinð4wÞ þ cosð4wÞ þ sinð5wÞ þ cosð5wÞ

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

up ¼ a1 sinðwÞ þ a2 cosðwÞ þ a3 sinð2wÞ þ a4 cosð2wÞ þ a5 sinð3wÞ


ð3:16Þ
þ a6 cosð3wÞ þ a7 sinð4wÞ þ a8 cosð4wÞ þ a9 sinð5wÞ þ a10 cosð5wÞ

Substituting the assumed particular solution in Eq. (3.15), we get

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

The particular solution of Eq. (3.15) is given by

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

The homogenous part of Eq. (3.15) is given by

€u þ 2u_ þ 3u ¼ 0 ð3:19Þ

We assume the homogenous solution to be

uh ¼ Aesw ð3:20Þ

Substituting the assumed solution in Eq. (3.19), we get


pffiffiffi
s ¼ 1  2i

Substituting value of s in the assumed homogenous solution, we get


pffiffi pffiffi
uh ¼ A1 eð1 þ 2iÞw
þ A2 eð1 2iÞw
ð3:21Þ
pffiffiffi pffiffi pffiffiffi pffiffi
vh ¼ ð1 þ 2iÞA1 eð1 þ 2iÞw þ ð1  2iÞA2 eð1 2iÞw ð3:22Þ

Complete solution of Eq. (3.15) is given by

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

We write Eqs. (3.25) and (3.26) in weak form

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

In a matrix form, we can write Eqs. (3.29) and (3.30) as


2 3
R2p R2p 2 2p 3
6
T _ þ 2½NT ½NÞdw
ð½N ½N ð3½N ½NÞdw 7 
T
R
6 7 ½u T
6 0 0
7 ¼ 4 0 ½N f ðwÞdw 5
4 R2p R2p 5 ½v
ð½NT ½NÞdw _
 ð½NT ½NÞdw 0
0 0
ð3:31Þ

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

Fig. 3.1 (continued)

3.4 Solution of Coupled Differential Equations with 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

where f ðwÞ ¼ 1 þ sinðwÞ þ cosðwÞ þ sinð2wÞ þ cosð2wÞ.


3.4 Solution of Coupled Differential Equations with Finite Element in Time 69

Analytical solution of coupled differential Eq. (3.32) is given by

11 21 577 467 225


f1 ¼ þ cosðwÞ  sinðwÞ  cosð2wÞ  sinð2wÞ ð3:33Þ
52 3145 3145 6554 6554
7 46 54 117 505
f2 ¼  cosðwÞ  sinðwÞ þ cosð2wÞ  sinð2wÞ ð3:34Þ
52 629 629 6554 6554

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.

3€f1  f_ 1  2f_ 2 þ f1  9f2 ¼ f ðwÞ ð3:35Þ

4€f1  3f_ 1  4f_ 2  6f1 þ 2f2 ¼ f ðwÞ ð3:36Þ

3p_ 1  p1  2p2 þ f1  9f2 ¼ f ðwÞ ð3:37Þ

4p_ 2  3p1  4p2  6f1 þ 2f2 ¼ f ðwÞ ð3:38Þ

p1 ¼ f_ 1 ð3:39Þ

p2 ¼ f_ 2 ð3:40Þ

Weak formulation gives us

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

Writing, p1 ¼ ½N½p1 ; p2 ¼ ½N½p2 ; f1 ¼ ½N½f1 ; f2 ¼ ½N½f2 ; dW1 ¼ ½dW1 T


½NT ; dW2 ¼ ½dW2 T ½NT ; dW3 ¼ ½dW3 T ½NT ; and dW4 ¼ ½dW4 T ½NT and writing
Eqs. (3.41)–(3.44) in a matrix form, we get
70 3 Finite Element in Time

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.

3.5 Enforcing Periodicity in the System

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

Enforcing periodicity will convert Eq. (3.46) to Eq. (3.47).


3.5 Enforcing Periodicity in the System 71

Fig. 3.2 Finite element in time for coupled differential equations

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

The p-version finite element in time permits straightforward application of the


periodicity boundary condition.

3.6 Advantage of Choosing an Element from (0 to 2p),


p-Version of Finite Element in Time

Here, we take two time elements: first is 0 to 2p and second is 2p to 4p:


Let ½N ¼ ½N1 ; N2 ; N3 ; N4  (Shape functions for first element 0 to 2p)
72 3 Finite Element in Time

Also ½M ¼ ½M1 ; M2 ; M3 ; M4  (Shape function for second element 2p to 4p)


We notice that
Z2p Z4p
0 T
AðwÞ½N  ½Ndw ¼ AðwÞ½M 0 T ½Mdw ð3:48Þ
0 2p

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:

3.7 Selection of Number of Nodes

Here, we take a differential equation with the periodic forcing.

€u þ 2u_ þ 3u ¼ sinðwÞ þ cosðwÞ þ sinð2wÞ þ cosð2wÞ ð3:49Þ

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)

3.8 Effect of Forcing Term in Finite Element in Time

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.

€u þ 2u_ þ 3u ¼ sinðwÞ þ cosðwÞ ð3:51Þ

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)

Particular solution of Eq. (3.51) is given by

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.

€u þ 2u_ þ 3u ¼ sinðwÞ þ cosðwÞ þ sinð2wÞ þ cosð2wÞ ð3:54Þ

where u_ ¼ v ¼ ddwu :
74 3 Finite Element in Time

Fig. 3.3 (continued)

Particular solution of Eq. (3.54) is given by

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

Fig. 3.3 (continued)

€u þ 2u_ þ 3u ¼ sinðwÞ þ cosðwÞ þ sinð2wÞ þ cosð2wÞ þ sinð3wÞ þ cosð3wÞ


ð3:57Þ

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)

Now, we solve Eq. (3.41), with 17 node, 1 time element in 0 to 2p.


We can see in Fig. 3.4b–d, Eq. (3.54) can be solved with lesser number of
nodes, but for Eq. (3.57), higher number of nodes are required within the same time
element. For a typical problem, the analytical solution is not known. Thus, it is a
good idea to perform a convergence study for a given forcing function. The number
of nodes is increased and the value of nodes fixed at the point where the response
does not change due to an additional node.
3.9 Finite Difference Method (Runge–Kutta Fourth Order) 77

Fig. 3.4 (continued)

3.9 Finite Difference Method (Runge–Kutta Fourth


Order)

We solve a periodic differential equation using Runge–Kutta fourth-order method.


Such a method helps to verify the finite element in time results.

€u þ 2u_ þ 3u ¼ f ðwÞ ð3:15Þ

where

f ðwÞ ¼ sinðwÞ þ cosðwÞ þ sinð2wÞ þ cosð2wÞ þ sinð3wÞ þ cosð3wÞ


þ sinð4wÞ þ cosð4wÞ þ sinð5wÞ þ cosð5wÞ

We write the second-order Eq. (3.15) as two first-order differential equations


78 3 Finite Element in Time

Fig. 3.4 (continued)

v_ þ 2v þ 3u ¼ f ðwÞ ) v_ ¼ 2v  3u þ f ðwÞ ð3:60Þ

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

Fig. 3.4 (continued)

ui þ 1 ¼ ui þ 16ðku1 þ 2ku2 þ 2ku3 þ ku4 Þh ð3:64Þ

vi þ 1 ¼ vi þ 16ðkv1 þ 2kv2 þ 2kv3 þ kv4 Þh ð3:65Þ

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

ku4 ¼ f2 ðui þ ku3 h; vi þ kv3 h; w þ hÞ

kv4 ¼ f1 ðui þ ku3 h; vi þ kv3 h; w þ hÞ

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).

Fig. 3.5 Runge–Kutta fourth-order result


3.9 Finite Difference Method (Runge–Kutta Fourth Order) 81

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

Stability analysis of a system includes calculation of the eigenvalues of the matrix


containing the physics of the governing differential equation. If a differential
equation has constant coefficients, eigenvalues can be calculated analytically. In
rotor problem, we get a differential equation, which has periodic coefficients, and a
numerical solution is possible. Here, Floquet theory is used to find out the eigen-
values of a periodic system. Equations (4.1) and (4.2) represent a differential
equation with non-periodic coefficients and a differential equation with periodic
coefficients, respectively. We will use these examples to illustrate the method used
in stability analysis.

€u þ 2u_ þ 3u ¼ sinðwÞ þ cosðwÞ ð4:1Þ

sinðwÞ€u þ 2 cosðwÞu_ þ 3fsinðwÞ þ cosðwÞgu ¼ sinðwÞ þ cosðwÞ ð4:2Þ

4.2 Stability Analysis of Equations with Constant


Coefficients

Here, we take two differential Eqs. (4.3) and (4.4) with constant coefficients.

€u þ 2u_ þ 3u ¼ sinðwÞ þ cosðwÞ ð4:3Þ

€u  2u_ þ 3u ¼ sinðwÞ þ cosðwÞ ð4:4Þ

For stability analysis, we write state-space representation of differential Eqs. (4.3)


and (4.4)

© Springer Nature Singapore Pte Ltd. 2018 83


R. Ganguli and V. Panchore, The Rotating Beam Problem in Helicopter Dynamics,
Foundations of Engineering Mechanics, https://doi.org/10.1007/978-981-10-6098-4_4
84 4 Stability Analysis

_
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.

4.3 Stability Analysis of a Coupled Differential Equations


with Constant Coefficients

We take a coupled differential equations with constant coefficients

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

Fig. 4.1 (continued)

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

4.4 Stability Analysis of the Equation with Periodic


Coefficients, Floquet Theory

Consider the equation

AðwÞ€f þ BðwÞf_ þ CðwÞf ¼ DðwÞ ð4:12Þ

To find stability solution for Eq. (4.12), we should use Floquet theory.
We have the state-space representation.
 
_
XðtÞ ¼ ½AðtÞ½XðtÞ ð4:13Þ

Solution of Eq. (4.13) is given by [9]


 
½XðtÞ ¼ ½/ðt  t0 Þ XðtÞ0 ð4:14Þ

where ½Xðt0 Þ is initial value at t ¼ 0 and ½/ðt  t0 Þ is a transition matrix.


Our goal is to find out the eigenvalues of the matrix ½AðtÞ. If we know the
eigenvalues of a transition matrix ½/ðt  t0 Þ; we can find the eigenvalues of the
matrix ½AðtÞ.
Suppose we know that one of the eigenvalues of the transition matrix ½/ðt  t0 Þ
is H and we want to find out what will be the eigenvalues ðkÞ of the matrix ½AðtÞ
after time period T; relation between the two eigenvalues is given by
   
1 1 1 ImðHÞ
k ¼ lnðHÞ þ i tan ð4:15Þ
T T ReðHÞ

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

4.5 Analytical Solution with the Floquet Theory

Here, we solve a stability problem which includes periodic functions


2 3
dx     
4 dw 5 ¼ sinðwÞ 0 x 0
þ ð4:17Þ
dy ecosðwÞ 0 y F sinðwÞ
dw

For stability analysis, we substitute F ¼ 0: We get the equation in the form of


 
_
XðtÞ ¼ ½AðtÞ½XðtÞ
2 3
dx   
4 dw 5 ¼ sinðwÞ 0 x
ð4:18Þ
dy ecosðwÞ 0 y
dw
 
xð0Þ
Time period T ¼ 2p: We take two sets of initial conditions ½Xðw0 Þ ¼ ¼
yð0Þ
     
1 xð0Þ 0
and ½Xðw0 Þ ¼ ¼ to find out the transition matrix /ðtÞ:
0 yð0Þ 1
Solution of Eq. (4.18) is
x ¼ c1 e cosðwÞ ð4:19Þ

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

Equations (4.21) and (4.22) are in state-space representation form

½Xð2pÞ ¼ ½/½Xð0Þ ð4:23Þ

From Eqs. (4.21), (4.22), and (4.23), we get


 
1 0
½/ ¼ ð4:24Þ
2pe 1

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.

4.6 Numerical Method to Evaluate a Transition Matrix

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

0 ¼ w0 \w1 \w2 . . .\wk1 \wk ¼ T;

where kth interval is ðwk1 ; wk Þ

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

This concept is easier to understand through an example.


Example Take state-space representation (4.27)
    
x_ 1 ðtÞ sinðtÞ 2 cosðtÞ þ 3 x1 ðtÞ
¼ ð4:27Þ
x_ 2 ðtÞ cos3 ðtÞ 4 x2 ðtÞ

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

Transition matrix /ð2p; 0Þ is given by

½/ð2p; 0Þ ¼ expðD32 ½C32 Þ  expðD31 ½C31 Þ. . . expðD1 ½C1 Þ ð4:28Þ

For example (4.17), we get


     
  1 0 1 0 1 0
/Analytical ¼ ¼ and ½/Numerical  ¼
2pe 1 17:0795 1 17:0802 1

198 equal intervals are taken while solving numerically.

4.7 Stability Analysis for Rotor Problem

For rotor problem, we get Eq. (2.60) after the finite element in space

X2 ½M1 ½€f þ ½K1 ½f ¼ ½Q1  þ ½C1 ½f


_ þ ½D1 ½f ð2:60Þ

For stability analysis, we solve Eq. (4.29)

X2 ½M1 ½€f þ ½K1 ½f  ½C1 ½f


_  ½D1 ½f ¼ 0 ð4:29Þ

We write Eq. (4.29) into two first-order differential equations

_  ½C1 ½p þ ð½K1   ½D1 Þ½f ¼ 0


X2 ½M1 ½p ð4:30Þ

_
½P ¼ ½f ð4:31Þ

We write Eq. (4.30) as

_ þ ½C2 ½P þ ½K2 f ¼ 0


½M2 ½P ð4:32Þ

where ½M2  ¼ X2 ½M1 ; ½K2  ¼ ½K1  D1 ; and ½C2  ¼ ½C1 :


4.7 Stability Analysis for Rotor Problem 91

We write Eqs. (4.31) and (4.32) in matrix form as


    
_
½f ½0 ½I  ½f
_ ¼ ½M2 1 ½K2 
½P ½M2 1 ½C2  ½P
ð4:33Þ

_
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.

5.2 Result 1 (Mode Shapes and Frequencies


of the Rotating Beam)

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.

© Springer Nature Singapore Pte Ltd. 2018 93


R. Ganguli and V. Panchore, The Rotating Beam Problem in Helicopter Dynamics,
Foundations of Engineering Mechanics, https://doi.org/10.1007/978-981-10-6098-4_5
94 5 Helicopter rotor results

Table 5.1 Inputs for the elastic rotor problem


Radius R 4.9378 m
Chord/Radius c=R 0.08
Rotor solidity r 0.1
Rotating speed X 40.12 rad/s
Thrust coefficient over solidity CT 0.1
r
Number of blades N 4
Lift curve slope a 5.73
Blade linear twist htw −8°
Mass per unit length m 6.4636 kg/m
Lock number c 6.35
Sectional flap stiffness EI 0.008345
mX2 R4

Table 5.2 Non-dimensional rotating frequencies


Mode (s = 1) (s = 2) (s = 3) (s = 4) (s = 5)
Frequencies Frequencies Frequencies Frequencies Frequencies
1 3.6816 4.1373 4.7972 5.5850 6.4495
2 22.1810 22.6149 23.3202 24.2733 25.4460
3 61.8417 62.2731 62.9849 63.9667 65.2050
4 121.0509 121.4966 122.2355 123.2614 124.5664
5 200.0115 200.4669 201.2232 202.2767 203.6220

Fig. 5.1 Campbell diagram


5.3 Result 2 (Response of the Rotor Blade with the Uniform … 95

5.3 Result 2 (Response of the Rotor Blade


with the Uniform Inflow Model, Using
Three Different Cases)

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].

5.4 Result 3 (Response of the Rotor Blade with the Linear


Inflow Model, Using Three Different Cases)

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.

Fig. 5.2 Response with uniform inflow


96 5 Helicopter rotor results

Fig. 5.3 Response with linear inflow

5.5 Result 4 (Stability in Hover Condition)

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).

Fig. 5.4 Stability in hover


condition with varying Lock
number
5.6 Result 5 (Stability in Forward Flight) 97

Fig. 5.5 Stability in forward


flight condition

5.6 Result 5 (Stability in Forward Flight)

Floquet theory is used to obtain the stability results in forward flight.


Non-dimensional frequency ðx=XÞ is taken as 1.189, advance ratio is varied from
0 to 0.5, and the Lock number is kept at 6. Result is similar to [5].

5.7 Summary and Conclusions

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)

© Springer Nature Singapore Pte Ltd. 2018 99


R. Ganguli and V. Panchore, The Rotating Beam Problem in Helicopter Dynamics,
Foundations of Engineering Mechanics, https://doi.org/10.1007/978-981-10-6098-4