Felix L. Chernousko · Leonid D.

 Akulenko
Dmytro D. Leshchenko

Evolution of
Motions of a
Rigid Body
About its
Center of Mass
Evolution of Motions of a Rigid Body About
its Center of Mass
Felix L. Chernousko • Leonid D. Akulenko •
Dmytro D. Leshchenko

Evolution of Motions
of a Rigid Body About
its Center of Mass
Felix L. Chernousko Leonid D. Akulenko
Institute for Problems in Mechanics Institute for Problems in Mechanics
Russian Academy of Sciences Russian Academy of Sciences
Moscow, Russia Moscow, Russia

Dmytro D. Leshchenko
Odessa State Academy of Civil
Engineering and Architecture
Odessa, Ukraine

ISBN 978-3-319-53927-0 ISBN 978-3-319-53928-7 (eBook)

DOI 10.1007/978-3-319-53928-7

Library of Congress Control Number: 2017938017

© Springer International Publishing AG 2017

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 International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

The problem of motion of a rigid body about a fixed point is one of the classical
problems of mechanics, the study of which in the eighteenth to nineteenth centuries
is connected with the names of Euler, Lagrange, and Kovalevskaya. These scien-
tists discovered three cases in which a complete integration of the motion equations
of a rigid body is possible; these cases are named after their discoverers. The
solution in Euler’s case describes the motion of a free rigid body. Lagrange’s case
corresponds to a heavy rigid body having a dynamic symmetry, with the center of
gravity of the body and the fixed point lying on the dynamic symmetry axis.
Kovalevskaya’s case occurs when there is a special relation between the inertia
moments. Later on, the research in this field was mainly aimed to finding first inte-
grals and particular solutions in the problem of the rigid body dynamics.
The interest to the problems of the rigid body dynamics has significantly
increased in the second half of the twentieth century in connection with the devel-
opment of rocket and space technologies, the increasing speed and maneuverability
of aircrafts, and the creation of gyroscopic systems. The study of the motion of
satellites and space vehicles about the center of mass is important for creating the
systems of orientation control, stabilization of motion, and, ultimately, solving the
practical problems of astronautics.
A satellite or a spacecraft in its motion about the center of mass is affected by the
moments of forces of various physical nature. It is influenced by the gravitational,
aerodynamic, electromagnetic torques, the torques due to the light pressure, as well
as the torques due to the motion of some masses inside the body. These motions
may have various causes: the presence of fluid in the cavities in the body (e.g.,
liquid fuel or oxidizer in the tanks of a rocket), structural flexibility (elastic flexi-
bility) of the flying vehicle, the presence in the body of rotating masses (rotors,
gyroscopes, gyrodines), the complex internal structure (in the case of natural cele-
stial bodies), and also the movements of the crew members (in the case of a
crew vehicle).

v
vi Preface

The above-mentioned torques, acting on the body, are often relatively small and
can be considered as perturbations. Therefore, it is natural to use the asymptotic
methods of small parameter or the perturbation techniques to analyze the dynamics
of rigid body under the action of the applied torques.
This monograph presents the results of the authors’ research on the dynamics of
rigid body motion about its center of mass. The authors consider the evolution of
these motions under the influence of various perturbation torques. The basic method
applied in the studies is the Krylov–Bogolubov asymptotic averaging method.
The main content of the book is preceded by the literature review, where the
studies with the topics close to the subject of the monograph are briefly described.
The book consists of 11 chapters.
In Chap. 1, the fundamentals of the rigid body dynamics are briefly presented.
The basic concepts are introduced; the fundamental kinematic and dynamic equa-
tions are formulated.
Chapter 2 is devoted to the inertial motion of a rigid body, that is, to Euler’s case.
This motion serves as a reference or generating one for the majority of perturbed
motions considered in the book, because, with the action of small perturbation
torques, the motion turns out to be close to Euler’s motion over short intervals of
time. Some information about Euler’s motion necessary for further consideration is
given.
Lagrange’s case is described in Chap. 3. This case is used as a reference one in
Chap. 11. The necessary relations for this motion are presented, including its parti-
cular cases: regular precession and fast rotation.
Chapter 4 contains the basics of the averaging method, widely used in the book.
The notions of a system in a standard form and a system with rapidly rotating phase
are introduced; some accuracy estimates for the method are indicated. The appli-
cation of the averaging method to the equations of the perturbed motion that is close
to the motion in Euler’s case is discussed. It is the procedure of averaging over the
motion in Euler’ case that allows studying the evolution of the satellite motions in
the case of various perturbations. Next, the equations of the perturbed motion that is
close to Lagrange’s case are also considered.
Chapter 5 is dedicated to the description of various perturbation torques, acting
on a rigid body. We present some relations, necessary for further reasoning, for the
gravitational torques acting on a satellite, for the torques of resistance forces of the
external medium, for the torques of the light pressure forces, as well as for the
torques due to the presence of a viscous fluid in a cavity of the rigid body. Different
cases are considered of the influence of internal masses on the body motion: the
presence of elastic and dissipative elements and the viscoelastic properties of the
moving body.
In the subsequent Chaps. 6–10, we study the perturbed motions of a rigid body
under the influence of various external and internal torques. The corresponding
expressions for the perturbation torques are taken from Chap. 5. As a reference
Preface vii

(generating) motion in Chaps. 6 10, we use the motion in Euler’s case, and the
averaging over this motion is performed according to the asymptotic procedure
explained in Chap. 4.
Chapter 6 is devoted to the satellite motion on an elliptic orbit about the center of
mass under the action of gravitational torques. We investigate the cases of a satellite
with the close moments of inertia and the fast rotations of a satellite with arbitrary
moments of inertia. The planar oscillatory and rotational motions of a satellite about
its center of mass on an elliptic orbit are considered separately.
In Chap. 7, we consider the motions of a rigid body with a cavity filled with a
viscous fluid. As a result of the conducted asymptotic analysis, a solution is
obtained which describes, in a nonlinear setting, the evolution over a significant
time interval of the motion of a body having a cavity with a high-viscosity fluid. It is
demonstrated that the dynamics of a body with a cavity containing a viscous fluid is
equivalent to the dynamics of a gyrostat, carrying the rotors interacting with the
body through the linear resistance forces.
Chapter 8 is devoted to the motion of a rigid body in a resistant medium that acts
on the body by the torques depending on its angular velocity. We consider also the
case of simultaneous influence of the moments of gravitational and resistance
forces. The motion of a satellite about its center of mass under the action of
gravitational torque and the torque of resistance forces is studied.
In Chap. 9, various cases are studied of the motion of a rigid body having internal
degrees of freedom. We consider the motions of a body that contains linear elastic
and dissipative elements. In particular, the motions are investigated of a body
carrying masses which are attached to it by means of elastic forces with linear or
quadratic damping. This situation simulates the presence of loosely fixed compo-
nents on a spacecraft, having a significant influence on its movement relative to the
center of mass.
The influence of the torque of the light pressure forces on the motion of a
satellite of the Sun about the center of mass is considered in Chap. 10. We study
the evolution of angular motions in a number of cases.
Chapter 11 is dedicated to the perturbed motions of a rigid body that are close to
Lagrange’s case. As a generating motion, we use the motion described in Chap. 3.
By means of averaging over the motion in Lagrange’s case, we analyze the
perturbed motions under the action of linear dissipative torques.
For all cases of motion considered in the book, we present and analyze the basic
equations of motion, perform an averaging procedure, and obtain the averaged
equations, which, being significantly simpler than the original ones, describe the
motion over a large time interval. We present the accuracy estimates for the
asymptotic procedure. As a result of analysis and solution of the obtained equations,
we establish some quantitative and qualitative specific features of the motions and
provide a description of the evolution of the body motion. The presentation is
illustrated by numerous examples.
viii Preface

The authors hope that the book will be of interest for the scientists working in the
field of mechanics and applied mathematics, engineers, postgraduate students, and
students of the corresponding specialties.
The work was partly supported by the Russian Science Foundation, project
no. 14-11-00298.

Moscow, Russia Felix L. Chernousko

Moscow, Russia Leonid D. Akulenko
Odessa, Ukraine Dmytro D. Leshchenko
Survey of Literature

The problem of the motion of a rigid body about a fixed point is one of the classical
problems of mechanics. In 1758, L. Euler [1] obtained a solution of this problem for
the case of a free rigid body when the center of mass coincides with the fixed point.
In the year 1788, J. Lagrange [2] investigated the motion of a heavy rigid body in
the case when the ellipsoid of inertia about the fixed point is an ellipsoid of
revolution, whereas the center of mass of the rigid body lies on the axis of symmetry
of this ellipsoid.
After Lagrange, the research of rotation of a rigid body about a fixed point
continued, but it was only in 1889 when S.V. Kovalevskaya discovered one more
case [3], in which the solution can be obtained for arbitrary initial conditions.
Besides, several cases were discovered, for which some particular solutions of the
motion equations were obtained. Among them are the cases of W. Hess,
D.K. Bobylev, V.A. Steklov, D.N. Goryachev, S.A. Chaplygin, G. Grioli, and
others.
In the second half of the twentieth century, new forms of equations of a rigid
body were obtained; the methods of their investigation were developed. As a result,
new solutions for the problem of the rigid body motion about a fixed point were
constructed. The main results in this area were obtained by P.V. Kharlamov,
E.I. Kharlamova, and other researchers, first of all, the representatives of the
Donetsk school of mechanics.
A review of the obtained results, their classification, and detailed bibliography
can be found in the works by P.V. Kharlamov [4]; I.N. Gashenenko, G.V. Gorr, and
A.M. Kovalev [5]; and A.V. Borisov and I.S. Mamaev [6].
Some general issues of the rigid body dynamics are considered in the mono-
graphs [7, 8].
Practical problems require studying more complex motions of a rigid body. A
number of objects in nature and technology can be simulated by a single rigid body.
These include flight vehicles, aircrafts, spaceships, and submarines, celestial bodies
with complex internal structure, and the disturbances acting on them.

ix
x Survey of Literature

This calls for studying rotational motions of the rigid and quasi-rigid (close to
rigid) bodies about a fixed point under the action of external and internal pertur-
bation torques of various physical nature. The theory and calculation methods are
developed for the motion of a body containing a viscous fluid and elastic and
viscoelastic elements. Such questions arise in modern problems of dynamics,
orientation, control, and stabilization of the natural and artificial celestial bodies,
technical objects, gyroscopy, and other fields of mechanics.
Different physical nature of the perturbation torques and their rather complex
dependence on the generalized coordinates result in the equations of motion, for
which precise analytical solution is hardly possible. On the other hand, the numer-
ical solution, characterizing the particular cases in detail, does not allow tracing the
general properties of the motion and its evolution. Therefore, the role of various
asymptotic methods is great, which allow, if correctly applied, identifying the main
features of the motion already in the first approximation.
Application of the Poincare method of small parameter for the construction of
the solution of the problem, concerning the rigid body motion about a fixed point, is
described in the books [9, 10]. In [11], one can find a survey of the methods for
integrating the motion equations of mechanical systems.
The main purpose of the present book is to study the evolution of motion of a
rigid body about its center of mass under the action of various perturbation torques.
In this context, the main attention is paid to the averaging method [12 14]. This
method has been long used in the celestial mechanics, though without proper
substantiation. For the first time, it was strictly formulated and justified in the
works of N.M. Krylov and N.N. Bogolubov [12].
Presently, there are many papers, dedicated to the substantiation and application
of asymptotic methods. Exposition of these methods, as well as a detailed biblio-
graphy on this subject, is contained in the books by N.N. Bogolubov and Yu.A.
Mitropolsky [12], V.M. Volosov and B.I. Morgunov [13], Yu.A. Mitropolsky [14],
N.N. Moiseev [15], and V.I. Arnold, V.V. Kozlov, and A.I. Neishtadt [11, 16].
For the first time, the averaging technique was applied to studying the perturbed
motions of a satellite about its center of mass in the papers of V.V. Beletsky
[17] and F.L. Chernousko [18]. In [17], a satellite having a dynamic symmetry
was considered. In the work [18], an averaging procedure was constructed for a
satellite with an arbitrary triaxial ellipsoid of inertia, i.e., the averaging was
performed over the Euler Poinsot motion. Besides, in the paper [18], an averaging
procedure was proposed for a triaxial satellite with close to one another moments of
inertia. In both cases, the motion of the satellite is composed of the Euler Poinsot
motion around the vector of angular momentum and the motion of this vector itself
in the space.
Let us briefly consider the works which are devoted to the research of perturbed
motions of a rigid body about its center of mass and are close to the subject of the
present book. The monograph [17] is dedicated to the methods of research and the
main effects of the motion of an artificial satellite about its center of mass under the
action of the gravitational, magnetic, aerodynamic torques and the torques due to
the light pressure. In the book [19], the theory of relative motion of a satellite in the
Survey of Literature xi

gravitational field is presented in detail, while the main attention is paid to nonlinear
resonance effects. The effects of the satellite motion, described in [17, 18], were
also investigated in the paper [20].
V.V. Beletsky and A.M. Yanshin [21] studied the influence of aerodynamic
forces on the rotational motion of artificial satellites. The work [22] is devoted to
researching the influence of decelerating aerodynamic torques on the rotational
motion of satellites of various shapes. In the paper [23], the effects connected with
the perturbed motion of an asymmetric artificial Earth satellite (AES) about the
center of mass under the action of the forces of aerodynamic dissipation are
investigated. In the work [24], the evolution of the satellite rotation is investigated
using the complete formula for the dissipative aerodynamic torque.
The main research directions concerning the motion of space vehicles and
simulation of the external forces acting on satellites are described in the reviews
by V.M. Morozov [25], V.A. Sarychev [26], and S.K. Shrivastava and
V.J. Modi [27].
The paper [28] is dedicated to the analysis of resonance effects in the rotational
motion of a satellite with unequal inertia moments in the gravitational field. In the
work [29], the nonresonance and resonance perturbed rotations of a triaxial satellite
in the gravitational field are studied.
In the paper [30], the fast rotations of a triaxial satellite perturbed by the gravi-
tational torque are considered. The solution expressed by elliptic functions and
Jacobi integrals is valid for nonresonance rotations, provided the rotation velocity is
much larger than the angular velocities of the orbital movement and precession.
Among the systems of stabilization for artificial satellites by means of the
external torques, the most widespread are the systems of gravitational stabilization.
The first model of gravitational stabilization of artificial satellites and the study of
the dynamics of this system were presented in the work by D.Ye. Okhotsimsky and
V.A. Sarychev [31]. Detailed information about the systems of gravitational stabili-
zation can be found in the survey [26]. In the paper [32], a review is provided of the
problems and works, connected with the development of passive systems of satel-
lite orientation. A bibliography of the Russian and foreign research on the passive
systems of orientation of satellites and space vehicles is presented also in the books
[33 35]. In the work [36], a review is given of the basic results obtained in the
applied celestial mechanics and the spacecraft motion control.
A model of dynamically nonsymmetric satellite with the moments of inertia
close to each other, moving in the central gravitational field under the action of the
moment of aerodynamic resistance forces, is investigated in [37]. The paper [38] is
devoted to the question of evolution of the rapid motion of a satellite under the
action of the gravitational and aerodynamic torques.
In the work [39], a mathematical model is proposed for the rotational motion of
the “Photon” satellite. In the paper [40], the issues are considered of simulating the
moments of aerodynamic forces acting on a satellite with a gravitational stabili-
zation system.
xii Survey of Literature

An important field of practical applications of the rigid body dynamics is the

mechanics of gyroscopic systems. Most comprehensively, the results on the
mechanics of gyroscopes are reflected in the books by A. Yu. Ishlinsky [41] and
K. Magnus [42]. Some cases of integration of the gyroscope motion equations in a
resistant medium are considered in [6, 42–54]. The method of averaging is applied
to studying the gyroscope dynamics. In the work by D.M. Klimov, G. N.
Kosmodemyanskaya, and F.L. Chernousko [55], a fast motion of a heavy rigid
body about a fixed point or the equivalent motion of a gyroscope with noncontact
suspension is researched. Using the method of averaging, Yu. G. Martynenko
continued studying the motions of gyroscopes of various types with noncontact
suspension. In his book [56], he considered the motion of a conducting rigid body in
the electrical and magnetic fields.
In the studies of G.G. Denisov and Yu. M. Urman [57–59], an analysis is carried
out of the precession motions of a rigid body with a fixed point under the action of
torques having a force function. The motion of a gyroscope with a noncontact
suspension under the influence of nonconservative moments is considered in [60].
A number of studies are devoted to the dynamics of a rigid body in a resistant
medium. In the works of L.D. Akulenko, D.D. Leshchenko, F.L. Chernousko, and
A.L. Rachinskaya [61–64], fast rotation of a nonsymmetric heavy rigid body about
a fixed point in a resistant medium is considered. The motion of the body is
composed of the Euler–Poinsot motion about the vector of angular momentum
(with the slowly decreasing values of angular momentum and kinetic energy) and
the motion in the space of the angular momentum vector itself. As a result of the
application of the averaging method, an autonomous equation is obtained, describ-
ing the motion of the vector of angular momentum. An analysis of this equation
allows discovering the quasi-stationary motions, for which the motion as a whole
dies out (the angular momentum and kinetic energy tend to zero), but the character
of the body motion about the angular momentum vector remains invariable.
The papers of A.I. Neishtadt [65] and M.L. Pivovarov [66] study the motion
about the center of mass of a nonsymmetric rigid body influenced by two small
perturbation torques: a constant one in the body-fixed axes and a linear dissipative
one or, alternatively, a constant one and a torque involving the terms quadratically
depending on the angular velocity.
In the work [67], the perturbed motion of a rotating spacecraft on a circular orbit
is considered under the action of a small aerodynamic torque proportional to the
angular velocity of the body.
In the works [68–70], some analytical approximate solutions are obtained for the
problem of the motion of a rigid body close to symmetrical one, as well as of a body
with arbitrary inertia characteristics influenced by a moment which is constant in
the body-fixed axes.
The works [71, 72] consider the problems on the motion of a heavy rigid body
about a fixed point under the action of a dissipative torque, including the terms
which are linear and quadratic with respect to the angular velocity.
In the paper [73] and the book [56], the motion about the center of mass is
studied of a symmetric rigid body in the presence of resistance of the medium and
Survey of Literature xiii

an active rotating torque providing the constancy of the angular rotation of the
rotor.
The work [74] investigates the stability of rotations of a body about the center of
mass in a linearly resisting medium under the presence of a torque directed along
one of its principal axes.
In the books by B. Ya. Lokshin, V.A. Privalov, V.N. Rubanovsky,
V.A. Samsonov, and M.V. Shamolin [75 77], the problem of the motion of a
rigid body in a resistant medium is studied. Quasi-stationary models of the medium
resistance, spatial movement in a resistant medium, as well as the motion of axially
symmetric bodies with a fixed point in the flow of medium are considered.
In the study [78], the Euler equations have been integrated for a symmetric
gyrostat, which is considered to be a rigid body with a symmetric rotor placed
inside, taking into account the external dissipative torques. The paper [79] considers
a problem of the motion of an asymmetric rigid body about its center of mass in a
resistant medium. A qualitative description of the phase trajectories is given; some
of their characteristics and quantitative estimates are presented. In the work [80],
the conditions are obtained for the global asymptotic stability of the stationary
rotations of a nonsymmetric rigid body about its center of mass under the action of a
constant external torque and a dissipative torque.
In the paper [81], free rotational movement is considered of a rigid body under
the action of a linear viscous torque. The work [82] is devoted to the construction of
an exact solution for the problem of free rotation of an axisymmetric rigid body,
taking into account the viscous friction torque linearly dependent on the angular
velocity of the body. In the paper [83], the evolution of the rigid body rotation under
the influence of the sum of constant and dissipative moments is studied by means of
numerical methods. In [84], some cases of the rigid body rotation, similar to Euler’s
case, under various damping torques are investigated.
The problem of rotational motion of a spacecraft under the action of light pres-
sure forces is one of the most important parts of the dynamics of rotational motion
of a rigid body about its center of mass. First, satellites and space vehicles equipped
with extended solar panels or reflective antennas were studied. Then, the problems
were considered of attitude control using the light pressure forces. The literature on
these matters can be found in the reviews [26, 85] and the books [86 88].
In the monograph [17], rotation of a dynamically symmetric satellite on a
heliocentric orbit under the action of the light pressure torque is investigated. In
[89, 90], integral characteristics of the force action of the light flow on the frame,
as well as the formulas for the light pressure torque, acting on a body bounded by a
surface of revolution, are obtained.
One can identify the main areas of studying the influence of light pressure on the
rotational motions of celestial objects. The first area is the analysis of the use of
light pressure for the spacecraft orientation. In the late 1980s of the twentieth
century, the studies were conducted in the Soviet Union on the astrometric project
“Regatta–Astro,” in the framework of which it was supposed to bring a space
vehicle, oriented toward the Sun by the light radiation pressure, to a heliocentric
xiv Survey of Literature

orbit. Various aspects of the dynamics of such spacecraft were considered in the
papers [91 93].
The second direction concerns the influence of light pressure on the rotational–
translational motion of asteroids. Since the beginning of the 1990s, the threat of the
Earth collision with a large asteroid has been actively discussed. Predicting such
events requires the construction of a precise theory of the asteroid movement. In the
light flow, a complex geometry of real asteroids leads to the appearance of
perturbing torques, changing the orientation of the rotation axis and, as a result,
changing the value of the total light pressure force, perturbing the orbital motion.
As an example of this kind of studies, we mention the papers [94 96].
The third direction is the study of the so-called Yarkovsky effect. Sunlight,
striking on an object, heats it up; as a result, thermal radiation appears. The impact
of this effect on the movement of the center of mass was considered in the papers
[97 99].
The specific features of the light pressure influence on the orientation regime and
stabilization of a spacecraft with solar sails or reflective panels are studied in the
book [100]. The paper [101] is devoted to calculating the principal vector and
principal moment of the light pressure forces acting on a spacecraft with a solar sail.
In the works of L.D. Akulenko, D.D. Leshchenko, and A.S. Shamaev
[102 104], the rotational motion is researched of a dynamically asymmetric satel-
lite with an axially symmetric surface about its center of mass under the action of
the light pressure torque. With the help of the averaging method, the evolution is
studied of the rotations of a triaxial satellite, close to a dynamically spherical one,
under the action of the light pressure torque in the case when the spacecraft is a
body of revolution. In addition, the coefficient of the light pressure torque is
approximated by trigonometric polynomials of an arbitrary order with respect to
the orientation angle. A first integral is discovered for the system of averaged equa-
tions of the first approximation for the angles of nutation and proper rotation. The
numerical and qualitative analysis of the phase plane is conducted; new qualitative
effects of the satellite rotation are discovered.
The papers [105, 106] study the evolution of rotations of the Sun satellite,
moving along an elliptical orbit with an arbitrary eccentricity under the influence
of the torques of the forces of gravity and light pressure.
Let us consider the influence of the moments of internal dissipation forces on the
rotation of a rigid body. The problems of the dynamics of bodies with the fluid-
containing cavities are among the classical problems of mechanics. A fundamental
study of the rotational motion of a rigid body having a cavity, filled with a homo-
geneous ideal fluid, was carried out in general formulation by N. E. Zhukovsky
[107].
A great interest to the problems of rotation of rigid bodies with the fluid-
containing cavities has arisen in connection with the development of the rocket
and space technology. The presentation of the results on the dynamics and stability
of motion about the center of mass of a body with the fluid-containing cavities is
given in books by N.N. Moiseev and V.V. Rumyantsev [108], G.N. Mikishev and
B.I. Rabinovich [109], and G.S. Narimanov, L.V. Dokuchaev, and I. A. Lukovsky
Survey of Literature xv

[110]. In the survey articles [111, 112], the formulations are presented for the
problems of the stability theory and oscillations of rigid bodies with the fluid-
filled cavities, various forms of the rotational motion equations and their first
integrals are considered, and a systematic description is given for the results of
research of the gyrostat motion.
The problems of dynamics of a rigid body with cavities, containing a viscous
fluid, are significantly more difficult than in the case of ideal fluid. An important
contribution to the solution of these problems has been made by the works of
F.L. Chernousko [113 120], summarized in the monograph [121]. Its translation is
given in [122]. These studies showed that solving the problems of dynamics of a
body with a homogeneous viscous fluid can be subdivided, under some natural
assumptions, into two parts—the hydrodynamic and dynamic ones—which can
greatly simplify the initial problem. In the paper [113] and the first chapter of the
monograph [121, 122], the results of which are used in this book, the motion is
considered about the center of mass of a rigid body with a cavity filled with a fluid
of high viscosity (for low Reynolds numbers). A system of ordinary differential
equations is constructed there, which approximately (in the quasi-stationary
approximation) describes the rotational movement of the rigid body with a fluid
outside a small initial time interval, when the flow in the cavity is significantly
unsteady. The influence of the fluid on the body movement is characterized, in the
quasi-stationary approximation, by a tensor, which is determined only by the cavity
shape. As an example, the problem is considered of the spatial motion of a free rigid
body with a cavity filled with a viscous fluid.
In the works of A. I. Kobrin [123, 124], an initial period of rotation of a body
with a cavity containing a high-viscosity fluid is investigated with the help of the
boundary layer method, and the initial conditions for the system of equations
proposed in [113, 121, 122] are specified. The controlled motion about the center
of mass for a body having a cavity filled with a viscous fluid is studied.
The paper [125] is devoted to studying the stabilizing effect of a viscous fluid in
a cavity on the rotation of a top around the given axis. There, on the basis of the
equations obtained by F.L. Chernousko, a characteristic time of stabilization and
the best orientation of the cavity relative to the rigid body are found. In [126], the
oscillations on an elliptic orbit of a rigid body with a spherical cavity entirely filled
with a viscous fluid are studied at low Reynolds numbers. In the papers [127, 128],
the motion in a resistant medium of a rigid body with cavities filled with a fluid of
high viscosity about a fixed point is considered. Fast rotational motion about the
center of mass for a dynamically symmetric satellite with a cavity filled with a
viscous fluid under the action of gravity torque and resistance of the medium is
studied. In the work [129], a possibility is considered of damping the nutation
oscillations by means of a viscous fluid filling a cavity in the rotor or within the
gyroscope.
The paper [130] is devoted to the study of oscillations of a rigid body with a
toroidal cavity filled with a viscous fluid. It is precisely the toroidal cavities with
fluid that are used in certain systems of damping the spacecraft oscillations about
the center of mass. In the work [131], the rotation is studied of a satellite with a
xvi Survey of Literature

permanent magnet in the plane of a polar elliptical orbit. The damping is investi-
gated with the help of a viscous fluid which entirely fills the cavity of an arbitrary
shape, under low Reynolds numbers. In the papers [132, 133] and the book [134], an
asymptotic method is used to studying the inertial motion of a rigid body and the
rotational motion of a symmetric satellite with a spherical or ellipsoidal cavity filled
with a viscous fluid. Secular effects in the rotation motion of a planet, caused by the
dissipation of energy in the matter of the core, are studied in [135]. According to
[113, 121, 122], it is assumed that the influence of the fluid core is equivalent to the
action on the “frozen” planet of a nonconservative moment of a special kind. In the
works [136 138], the fast rotational motions about the center of mass of a
dynamically asymmetric satellite with a cavity completely filled with a high-
viscosity fluid under the influence of the gravitational and light torques are studied.
The paper [139] presents the analytical and numerical results for the rigid body
with a cavity filled with a viscous fluid. The chaotic motions of a rigid body and a
satellite with a cavity filled with fluid are investigated in [140 142].
In the papers [143, 144], the movement is considered relative to the center of
mass for a satellite gyrostat with a fluid-containing cavity under low Reynolds
numbers. A control based on the feedback principle is constructed, which stabilizes
the stationary motions of the gyrostat. In the book [145], some control problems for
the rotating rigid bodies with cavities filled with an ideal or a viscous fluid are
considered.
A large number of works are devoted to the study of rotation of a rigid body with
a movable internal mass, with elastic and dissipative elements. A survey of the
studies on the mechanics of the systems of connected rigid bodies is presented in
[146]. A survey of the research, published prior to the year 1980, on nonlinear
dynamics of the elastic spacecraft or satellite with deformable elements is given in
[147]. The works in this direction are described also in the surveys on the spacecraft
dynamics [26, 27, 148, 149].
The need to consider the rotational dynamics of a system of bodies arose in
connection with the development of practical astronautics. On one hand, we note
the works associated with the study of the movements of satellite gyrostats,
containing rotating masses, and, on the other, the works in which the elastic
properties of the satellites and their components are taken into account. A number
of problems in the indicated fields and a bibliography on these issues are presented
in the monographs [7, 150, 151]. In the book of B.V. Rauschenbach and E.N. Tokar
[152], the equations are presented for the angular motion of the carrier of a
spacecraft, which contains movable masses.
The paper [153] (R.E. Roberson) considers the disturbance torques acting on a
satellite, which are generated by the relative motion of the bodies inside the
satellite. The work [154] (W.R. Haseltine) is devoted to the study of damping of
the nutation motion of a rotating artificial Earth satellite with the help of an internal
passive device. In the paper [155] (G. Colombo) and the book [156]
(W.T. Thomson), the influence of the inner elasticity and dissipation on the motion
of the satellite relative to its center of mass is studied.
Survey of Literature xvii

In the space flight, there arises sometimes a necessity to suppress the chaotic
rotation that occurs for one reason or another. To this end, the relative displace-
ments of movable masses are used [157 161].
A significant number of works are devoted to the analysis of various problems of
the dynamics of space vehicles containing internal movable masses. The issues of
stability and instability, resonance phenomena, and the problems of control and
stabilization of motions have been studied. In this regard, we can mention the works
[162 180].
In the papers by F.L. Chernousko [181 183], some cases are considered of the
motion of a rigid body containing movable internal masses connected to the body
by means of elastic and dissipative elements. The angular motion of a body
containing a mass of a continuous viscoelastic medium is investigated. A number
of problems on the motion of a rigid body containing elastic and dissipative
elements are examined in [184 189]. Some issues of the dynamics and stability
of rotations of a rigid body containing elastic and dissipative elements were
considered in the paper [190] and the books [191 194].
The monograph [195] considers the issues of the aircraft motion with large
rotation angles when the deformable elements like rods, plates, or fluid masses
perform oscillatory displacements under the action of inertial forces. In the book
[134], stationary motions of mechanical systems with elastic elements and their
stability are investigated. The monograph [196] studies the dynamics of multipiece
orbiting space systems consisting of rigid and elastic deformable bodies. In the
book [197], the issues connected with the movement of elastic space structures
relative to the center of mass under the action of the gravitational field torques are
considered. In the work [198], the transient processes related to the oscillations of
an elastic satellite in its moving about its center of mass under the action of the
control torque are investigated. In the paper [199], some quantitative estimates are
found for the transient process, which leads a viscoelastic solid body of a spherical
shape in a noncontact suspension to rotate steadily around the axis of maximum
moment of inertia. The work [200] considers the free movement of a linearly elastic
solid body about its center of mass.
In the papers [201, 202], a research is carried out concerning the effect of the
elastic and viscous properties of bodies in their free angular motions.
The works [203, 204] study the rotational motion of a rigid body carrying
viscoelastic inextensible rods. The paper [205] studies the evolution of motion of
a satellite with viscoelastic flexible rods on a circular orbit. In the book [206],
nonlinear oscillations of a rigid body, elastically connected with a point mass, in a
central force field are studied. The research [207] is devoted to the dynamics and
stability of a rigid body with internal dissipation.
The problem of evolution of the rigid body rotations about a fixed point
continues to attract the attention of researchers. In the aspect of applications, the
analysis of rotational motions of bodies about a fixed point is important for solving
the problems of astronautics, the problems of the entry of flying vehicles into the
atmosphere, and the movement of a rotating projectile and gyroscopy. Moreover, in
many cases, the motion in Lagrange’s case can be regarded as a generating
xviii Survey of Literature

(reference) motion of the rigid body, which takes into account the main torques
acting on the body. Recall that in this case the body is assumed to have a fixed point
and to be in the gravitational field, with the center of mass of the body and the fixed
point both lying on the dynamic symmetry axis of the body. A restoring torque,
analogous to the moment of the gravity forces, is created by the aerodynamic forces
acting on the body in the gas flow. Therefore, the movements, close to Lagrange’s
case, have been investigated in a number of works on the aircraft dynamics, where
various perturbation torques were taken into account in addition to the restoring
torque. Note the papers by G.E. Kuzmak [208, 209] and V.A. Yaroshevsky [210], in
which the movements relative to the center of mass of the flying vehicles entering
the atmosphere at high speed were studied. The method of reference equations and
the method of averaging were used.
In V.S. Aslanov’s monograph [211], the motion is studied of a rotating rigid
body in the atmosphere under the action of a time-dependent sinusoidal or
biharmonic restoring torque and small perturbation torques.
The book [212] considers the influence on the gyroscope motion of the equato-
rial and axial braking torques, playing a significant role in the study of the rotational
motion of artillery shells. The monograph by N.N. Moiseev [15] investigates the
Lagrange problem on the motion of an axially symmetric top under the action of
overturning moment, directed perpendicular to the plane passing through the sym-
metry axis of the top. In the paper [213], the impact on the Lagrange gyroscope
motion of the direction change of the force that creates an overturning or restoring
torque is considered. The work [214] investigates the influence of the dissipative
forces on the stability of the permanent motions of the Lagrange gyroscope.
In the paper [215], the effect of viscous friction on the Lyapunov stability of the
rotation of a heavy rigid body about a fixed point is investigated. A number of
works are devoted to the study of the movement of a “sleeping” Lagrange top. In the
paper [216], the sufficient conditions are found for the stability of the vertical rota-
tion of the Lagrange top in the presence of a damping torque. In the work [217],
the sufficient conditions for the asymptotic stability of a “sleeping” top in a resistant
medium are obtained.
In the paper of A.M. Kovalev [218], the movement of a body, which differs little
from the Lagrange gyroscope, is studied with the help of the Kolmogorov–Arnold
theorem. In the work [219], the problem concerning the existence of periodic solu-
tions for the equations of the rigid body motion about a fixed point with the
distribution of mass close to Lagrange’s case is considered. The book [10] describes
the application to the rigid body dynamics of the averaging techniques of the Gauss
type or the methods due to Fatou, N.D. Moiseev, I. Delone, and G. Hill, introduced
in the celestial mechanics. The periodic motions of the Lagrange top under small
displacement of its center of gravity or small deviation from its axial dynamic
symmetry are studied. In the work [220], using the method of Hori, the movement is
studied of a heavy rigid body with a fixed point, the mass distribution in which
differs little from the case of Lagrange, whereas the center of gravity is located
Survey of Literature xix

sufficiently close to this point. The paper [221] studies the motion of a heavy rigid
body with a fixed point in Lagrange’s case with an asymmetry due to inequality of
the equatorial moments of inertia. The work [222] considers the origination of
chaotic motions of a rigid body with a small shift of the center of mass from the
dynamic symmetry axis.
In the papers [223–225], an analogy is considered between the perturbed prob-
lem of the motion of the Lagrange gyroscope in the case of potential perturbations
and the problem of rotation of a satellite, the center of mass of which moves along a
circular orbit in the equatorial plane, taking into account the influence of the
Earth magnetic field.
In the work [226], the stationary motions of a rigid body in Lagrange’s case
under the action of dissipative forces and the thrust imbalance creating a restoring
torque are investigated. The domains of fulfilling the conditions of stability of
uniform rotations are found. The stability of the motion of symmetrical heavy rigid
body in the presence of resistance forces and the engine torque with respect to the
axis of symmetry, defined as a function of time, was considered earlier in [227].
In the work by V.F. Zhuravlev [228] and in the monograph [229], the problem is
considered on the behavior of the Lagrange top in the case when the suspension
point performs harmonic oscillations in the horizontal plane. The articles by
V.N. Koshlyakov [230, 231] consider the problem of stabilization, by means of
vertical vibration, of a symmetric rigid body rotating about a fixed point.
A number of other studies are dedicated to the problems of dynamics of a rigid
body with a vibrating suspension point. Thus, the movement of a rapidly rotating
symmetrical or close to symmetrical gyroscope under the vertical vibrations of the
suspension point is investigated in the works [232, 233]. In the paper [234], the
rotation of a viscoelastic solid with the movable base is considered. The work [235]
investigates the perturbed angular motions of the Lagrange top under random oscil-
lations of the fulcrum. The paper [236] considers the movement of the Lagrange
top, the suspension point of which performs vertical harmonic oscillations of
high frequency and small amplitude.
The work [237] studies the rotation of the Lagrange gyroscope, along the axis of
dynamic symmetry of which a point mass moves under the influence of the gravity
force and elastic force. In the paper [238], the influence is estimated of the moving
point masses (linear oscillators), performing oscillations along the symmetry axis of
the top or along the axes orthogonal to the axis of symmetry, on the stability of
uniform rotation of the Lagrange top. In [239], the problem of passive stabilization
of the rotations around the vertical of the Lagrange gyroscope with two degrees of
freedom damper of the “swing” type is considered.
In the paper [240], the motion of a dynamically symmetric heavy rigid body with
a fixed point under the action of constant and dissipative moments is considered.
The stationary regimes of the system are determined and their stability is investi-
gated. In the work [241], the effect is estimated of the dissipative and permanent
torques on the stability of uniform rotation of the Lagrange top with an arbitrary
xx Survey of Literature

axisymmetric cavity completely filled with an ideal fluid. The paper [242] is
devoted to studying the motion of a symmetrical top with a cavity filled with a
viscous fluid in the gravitational field, when the axis of the top is deflected from the
vertical. In the work [243], the motion is under consideration of a symmetrical
heavy rigid body with a fixed point under the action of the friction forces caused by
the surrounding dissipative medium.
The paper [244] investigates the evolution of regular precessions of a rigid body
close to Lagrange’s case. In the works [245, 246], the asymptotic behavior is
studied of the Lagrange gyroscope motions, close to regular precessions, under
the influence of a small perturbation torque.
The work [247] offers an overview of the results obtained prior to the year 1998
on the problem of evolution of the rigid body rotations, close to Lagrange’s case.
The perturbed motions of a rigid body motion, close to Lagrange’s case, are
investigated with the help of the averaging method in the paper of L.D. Akulenko,
D.D. Leshchenko, and F.L. Chernousko [248]. It describes the conditions for the
possibility of averaging the equations of motion with respect to the nutation angle;
an averaged system of equations is obtained. The body motion in the medium with
linear dissipation is considered. In [249], the perturbed motion of the Lagrange top
under the action of the slowly time-varying linear dissipative moment is considered.
In the papers [250, 251], the perturbed fast rotations of a rigid body, close to
regular precession, are considered. An averaged system of the motion equations in
the first and second approximations is obtained and investigated. The works
[252–254] describe the evolution of rotations in a more general case, when the
value of the restoring torque depends on the nutation angle. Some examples are
studied which correspond to the constant and linear external torques. The solutions
of the averaged systems of equations of the first and second approximations are
determined.
The paper [189] investigates the perturbed motions of a rigid body, close to the
regular precession in Lagrange’s case, under the influence of a slowly time-varying
perturbation torque and a restoring torque depending on the nutation angle. In
[255–257], the evolution is studied of the rigid body rotations, close to regular
precession, under the action of a restoring torque, depending on slow time and
nutation angle, as well as a perturbation torque, slowly varying in time.
In the works [258, 259], the perturbed rotational motions are considered of a
rigid body, close to the regular precession in Lagrange’s case, under the action of a
restoring torque, depending on the angles of nutation and precession.
The presented brief survey does not purport to be complete and can be signifi-
cantly expanded. However, it is clear already from this survey that there is an
extensive literature on the dynamics of a rigid body, moving about its center of
mass under the influence of perturbation torques of various physical nature. The
research in this area is in demand in connection with the problems of motion of
flying vehicles (satellites, spacecraft, aircraft, unmanned aerial vehicles), celestial
bodies (planets, comets), gyroscopes, and other objects of modern technology.
Survey of Literature xxi

References

1. Euler, L.: Du mouvement de rotation des corpes solides author d’un axe variable. Historie de
l’ Academie Royale des Sciences 4, 154–193, Berlin (1758–1765)
2. Lagrange, J.-L.: Mechanique Analytique. Chez La Veuve Desaint, Paris (1788)
3. Kowalevski, S.: Sur le probleme de la rotation d’ un corpes solide autor d’un point fixe.
Acta Math. 12, 177–232 (1889)
4. Kharlamov, P.V.: Modern state and perspectives of development of classical problems of
dynamics of a rigid body. Mekh. Tverd. Tela. 30, 1–12 (2000) in Russian
5. Gashenenko, I.N., Gorr, G.V., Kovalev, A.M.: Classical Problems of Dynamics of a
Rigid Body. Naukova Dumka, Kiev (2012) in Russian
6. Borisov, A.V., Mamaev, I.S.: Dynamics of a Rigid Body. “Regular and Chaotic Dynamics”,
Izhevsk (2001) in Russian
7. Lur’e, A.I.: Analytical Mechanics. Fizmatgiz, Moscow (1961) in Russian
8. Wittenburg, J.: Dynamics of Multibody Systems. Springer, Berlin (2008)
9. Kozlov, V.V.: Methods of Qualitative Analysis in the Dynamics of a Rigid Body. “Regular and
Chaotic Dynamics”, Izhevsk (2000) in Russian
10. Demin, V.G., Konkina, L.I.: New Methods in Dynamics of a Rigid Body. Ilim, Frunze
(1989) in Russian
11. Arnold, V.I., Kozlov, V.V., Neishtadt, A.I.: Mathematical Aspects of Classical and Celestial
Mechanics. Springer, Berlin (2007)
12. Bogolubov, N.N., Mitropolsky, Y.A.: Asymptotic Methods in the Theory of Nonlinear Oscil-
lations. Gordon and Breach Science, New York, NY (1961)
13. Volosov, V.M., Morgunov, B.I.: The Averaging Method in the Theory of Non-linear Oscil-
latory Systems. Moscow State Univ., Moscow (1971) in Russian
14. Mitropolsky, Y.A.: The Method of Averaging in Nonlinear Mechanics. Naukova Dumka,
Kiev (1971) in Russian
15. Moiseev, N.N.: Asymptotic Methods of Nonlinear Mechanics. Nauka, Moscow (1981) in
Russian
16. Arnold, V.I.: Geometrical Methods in the Theory of Ordinary Differential Equations.
Springer, London (2012)
17. Beletsky, V.V.: Motion of an Artificial Satellite about its Center of Mass. Israel Program for
Scientific Translation, Jerusalem (1966)
18. Chernousko, F.L.: On the motion of a satellite about its center of mass under the action of
gravitational moments. J. Appl. Math. Mech. 27(3), 708–722 (1963)
19. Beletsky, V.V.: Spacecraft Attitude Motion in Gravity Field. Moscow State Univ., Moscow
(1975) in Russian
20. Holland, R.L., Sperling, H.J.: A first order theory for the rotational motion of a triaxial rigid
body orbiting and oblate primary. Astron. J. 74(3), 490–496 (1969)
21. Beletsky, V.V., Yanshin, A.M.: The Influence of Aerodynamic Forces on Spacecraft Rota-
tion. Naukova Dumka, Kiev (1984) in Russian
22. Po, N.T.: On the rotational motion of a spherical satellite under the action of retarding aero-
dynamical moments. Bull. ITA. 10(5), 84–91 (1965) in Russian
23. Fedorova, L.I.: Influence of dissipative moment of the aerodynamic friction forces on the
rotary motion of a non-symmetric artificial Earth satellite. Prob. Mech. Control. Motion.
Perm. 7, 119–121 (1975) in Russian
24. Beletsky, V.V., Grushevsky, A.V.: The evolution of the rotational motion of a satellite under
the action of a dissipative aerodynamic moment. J. Appl. Math. Mech. 58(1), 11–19 (1994)
25. Morozov, V.M.: Stability of motion of spacecrafts. Advances in Science: General Mecha-
nics, pp. 5–83. VINITI, Moscow (1971) in Russian
xxii Survey of Literature

26. Sarychev, V.A.: Problems of artificial satellites orientation. Advances in Science,

Space Research. VINITI, Moscow (1978) in Russian
27. Shrivastava, S.K., Modi, V.J.: Satellite attitude dynamics and control in the presence of
environmental torques – a brief survey. J. Guid. Contr. Dyn. 6(6), 461–471 (1983)
28. Torzhevskii, A.P.: Fast rotation of an artificial satellite around its center of mass in a
resonance regime. Kosm. Issl. 6(1), 58–70 (1968) in Russian
29. Hitzl, D.L., Breakwell, J.V.: Resonant and non-resonant gravity-gradient perturbations of a
tumbling triaxial satellite. Celest. Mech. 3(3), 346–383 (1971)
30. Lara, M., Ferrer, S.: Closed form perturbation solution of a fast rotating triaxial satellite
under gravity-gradient torque. Cosmic Res. 51(4), 289–303 (2013)
31. Okhotsimsky, D.Y., Sarychev, V.A.: Gravitational stabilization system for artificial satel-
lites. Artif. Earth Sat. 16, 5–9 (1963) in Russian
32. Sarychev, V.A.: D.Ye. Okhotsimsky and its role in creating the systems of passive orient-
ation of satellites. Applied Celestial Mechanics and Motion Control. On the 90th anniversary
of D. Ye. Okhotsimsky, pp. 223–271. Compilers: T.M. Eneev, M.Yu. Ovchinnikov,
A.R. Golikov. M.V. Keldysh Institute of Appl. Math., Moscow (2010) in Russian
33. Rumyantsev, V.V.: On the Stability of Steady Motions of Satellities. Computing Center AN
SSSR, Moscow (1967) in Russian
34. Rauschenbakh, B.V., Ovchinnikov, M.Y., McKenna-Lawlor, S.: Essential Space-Flight
Dynamics and Magnetospherics. Kluwer Academic, New York, NY (2003)
35. Markeev, A.P.: Linear Hamiltonian Systems and Some Problems of Stability of Satellite’s
Motion Relative to its Center of Mass. “Regular and Chaotic Dynamics”, Moscow (2009) in
Russian
36. Okhotsimsky, D.Y., Eneev, T.M., Akim, E.L., Sarychev, V.A.: Applied celestial mechanics
and motion control. Applied Celestial Mechanics and Motion Control. On the 90th anniver-
sary of D. Ye. Okhotsimsky, pp. 328–367. Compilers: T.M. Eneev, M.Yu. Ovchinnikov,
A.R. Golikov M.V. Keldysh Institute of Appl. Math., Moscow (2010) in Russian
37. Pupyshev, Y.A.: The effect of gravitational and aerodynamic perturbations on the rotational
motion around the mass centre of an asymmetrical solid body. Vestn. Leningr. Un-ta.
Matem. Mekh. Astr. 7(2), 129–134 (1971) in Russian
38. Kuznetsova, E.Y., Sazonov, V.V., Chebukov, S.Y.: Evolution of the satellite rapid rotation
under the action of gravitational and aerodynamic torques. Mech. Solids 35(2), 1–10 (2000)
39. Sazonov, V.V., Sazonov, V.V.: The use of refined model in the problem of reconstruction of
the Foton satellite rotational motion. Cosmic Res. 49(2), 110–120 (2011)
40. Maslova, A.I., Pirozhenko, A.V.: Modeling of the aerodynamic moment acting upon a satel-
lite. Cosmic Res. 48(4), 362–370 (2010)
41. Ishlinsky, A.Y.: Orientation, Gyros and Inertial Navigation. Nauka, Moscow (1976) in
Russian
42. Magnus, K.: Kreisel. Theorie und Anvendungen. Springer, Berlin (1971)
43. Klein, F., Sommerfeld, A.: The Theory of the Top. Perturbations, Astronomical and Geo-
physical Applications, vol. 3. Birkhauser, Boston, MA (2012)
44. Krylov, A.N., Krutkov, Y.A.: General Theory of Gyroscopes and some Technical Appli-
cations. AN SSSR Press, Moscow (1932) in Russian
45. Svetlov, A.V.: On the gyroscope rotation in a resistive medium. Prikl. Mat. Mekh. 1(3),
371–376 (1938) in Russian
46. Koshlyakov, V.N.: On some particular cases of integration the dynamic Euler equations as
applied to the motion of a gyroscope in a resistive medium. Prikl. Mat. Mekh. 17(2),
137–148 (1953) in Russian
47. Koshlyakov, V.N.: Problems in Dynamics of Solid Bodies and in Applied Gyroscope Theory:
Analytical Methods. Nauka, Moscow (1985) in Russian
Survey of Literature xxiii

48. Bulgakov, B.V.: Applied Theory of Gyroscopes, 3rd edn. Moscow State Univ., Moscow
(1976) in Russian
49. Leimanis, E.: The General Problem of the Motion of Coupled Rigid Bodies About a
Fixed Point. Springer, Berlin (1965)
50. Gray, A.: A Treatise on Gyrostatics and Rotational Motion. Theory and Applications. Dover,
New York, NY (1959)
51. Routh, E.J.: Advanced Dynamics of a System of Rigid Bodies. Dover, New York, NY (2005)
52. Appel, P.: Traite de Mechanique Rationnelle. Gauthier–Villars, Paris (1953)
53. Macmillan, W.D.: Theoretical Mechanics. Dynamics of Rigid Bodies. McGraw-Hill,
New York, NY (1936)
54. Grammel, R.: Der Kreisel. Seine Theorie und Seine Anwendungen. Erster Band. Izd-vo
Inostr. Lit-ry, Moscow (1952) Russian translation from German
55. Klimov, D.M., Kosmodem’yanskaya, G.V., Chernousko, F.L.: Motion of a gyroscope with
contactless suspension. Izv. Akad. Nauk SSSR Mekh. Tverd. Tela. 2, 3–8 (1972) in Russian
56. Martynenko, Y.G.: Motion of a Rigid Body in Electric and Magnetic Fields. Nauka, Moscow
(1988) in Russian
57. Denisov, G.G., Urman, Y.M.: Precessional motions of a rigid body acted on by moments that
have a force function. Mech. Solids 10(6), 1–9 (1975)
58. Urman, Y.M.: Irreducible tensors and their applications in problems of dynamics of solids.
Mech. Solids 42(6), 883–896 (2007)
59. Urman, Y.M.: The Theory of Symmetry in Classical Systems: Study Guide. NSPU, Nizhny
Novgorod (2009) in Russian
60. Denisov, G.G., Komarov, V.N.: Nonconcervative moments and their effect on precession of
a noncontact gyroscope. Mech. Solids 14(3), 12–19 (1979)
61. Leshchenko, D.D.: Motion of a ponderous rigid body with a fixed point in a mildly resisting
medium. Soviet Appl. Mech. 11(3), 299–303 (1975)
62. Akulenko, L.D., Leshchenko, D.D., Chernousko, F.L.: Fast motion of a heavy rigid body
about a fixed point in a resistive medium. Mech. Solids 17(3), 1–8 (1982)
63. Akulenko, L.D., Leshchenko, D.D., Rachinskaya, A.L.: Evolution of the satellite fast rota-
tion due to the gravitational torque in a dragging medium. Mech. Solids 43(2), 173–184
(2008)
64. Akulenko, L.D., Leshchenko, D.D., Rachinskaya, A.L.: Evolution of the dynamically sym-
metric satellite fast rotation due to the gravitational torque in a dragging medium.
Mekh. Tverd. Tela. 36, 58–63 (2006) in Russian
65. Neishtadt, A.I.: Evolution of rotation of a solid acted upon by the sum of a constant and a
dissipative perturbing moments. Mech. Solids 15(6), 21–27 (1980)
66. Pivovarov, M.L.: The motion of a gyroscope with low self-excitation. Izv. Akad. Nauk SSSR
Mekh. Tverd. Tela. 6, 23–27 (1985) in Russian
67. I~narrea, M., Lanchares, V.: Chaotic pitch motion of an asymmetric non-rigid spacecraft with
viscous drag in circular orbit. Int. J. Non-linear Mech. 41(1), 86–100 (2006)
68. Van der Ha, J.C.: Perturbation solution of attitude motion under body-fixed torques.
Acta Astronaut. 12(10), 861–869 (1985)
69. Kane, T.R., Levinson, D.A.: Approximate description of attitude motion of a torque-free,
nearly axisymmetric rigid body. J. Astronaut. Sci. 35(4), 435–446 (1987)
70. Ayobi, M.A., Longuski, J.M.: Analytical solution for translational motion of spinning-up
rigid bodies subject to constant body-fixed forces and moments. Trans. ASME J. Appl.
Mech. 75(1), 011004/1–011004/8 (2008)
71. Medvedev, A.V.: The motion of a rapidly spun gyroscope under the influence of con-
stant torque in a resistive medium. Izv. Akad. Nauk SSSR Mekh. Tverd. Tela. 2, 21–24
(1989) in Russian
xxiv Survey of Literature

72. Kuryakov, V.A.: Rapid motion around a fixed point of a heavy solid in a medium with a
square law of resistance. Soviet Appl. Mech. 24(10), 1024–1033 (1988)
73. Kudin, S.F., Martynenko, Y.G.: Acceleration of contactless gyroscope in resistive medium.
Izv. Akad. Nauk SSSR Mekh. Tverd. Tela. 6, 14–22 (1985) in Russian
74. Denisov, G.G.: On the rotation of a rigid body in a resistive medium. Izv. Akad. Nauk SSSR
Mekh. Tverd. Tela. 4, 37–43 (1989) in Russian
75. Lokshin, B.Y., Privalov, V.A., Samsonov, V.A.: An Introduction to the Problem of Motion
of a Body in a Resistive Medium. Moscow State Univ., Moscow (1986) in Russian
76. Rubanovsky, V.N., Samsonov, V.A.: The Stability of Steady Motion in Examples and Prob-
lems. Nauka, Moscow (1988) in Russian
77. Shamolin, M.V.: Methods for Analysis of Variable Dissipation Dynamical Systems in
Rigid Body Dynamics. Ekzamen Press, Moscow (2007) in Russian
78. Kane, T.R.: Motion of a symmetric gyrostat in a viscous medium. AIAA J. 8(10), 1786–1789
(1970)
79. Puzyrev, V.E., Suikov, A.S.: On the motion of a rigid body about its center of mass under
partial dissipation of energy. Mekh. Tverd. Tela. 39, 157–166 (2009) in Russian
80. Leonov, G.A., Morozov, A.V.: The global stability of the steady rotations of a solid. J. Appl.
Math. Mech. 56(6), 897–901 (1992)
81. Krivtsov, A.M.: Description of motion of an axisymmetric rigid body in a linearly viscous
medium in terms of quasicoordinates. Mech. Solids 35(4), 18–23 (2000)
82. Ivanova, E.A.: Exact solution of a problem of rotation of an axisymmetric rigid body in a
linear viscous medium. Mech. Solids 36(6), 11–24 (2001)
83. Tronin, K.G.: Numerical analysis of rotation of a rigid body subject to the sum of a constant and
dissipative moment. Russian J. Nonlinear Dyn. 1(2), 209–213 (2005) in Russian
84. Ge, Z.M., Wu, M.H.: The stability of motion of rigid body about a fixed point in the case of
Euler with various damping torques. Trans. Can. Soc. Mech. Eng. 12(3), 165–171 (1988)
85. Modi, V.J.: On the semi-passive attitude control and propulsion of space vehicles using
solar radiation pressure. Acta Astronautica. 35(2–3), 231–246 (1995)
86. Polyakhova, E.N.: Space Flight with a Solar Sail. Librokom Book House, Moscow (2011) in
Russian
87. Popov, V.I.: Systems of Orientation and Stabilization of Spacecraft. Mashinostroyeniye,
Moscow (1986) in Russian
88. Kargu, L.I.: Systems of Angular Stabilization of Spacecrafts. Mashinostroyeniye, Moscow
(1980) in Russian
89. Karymov, A.A.: Determination of forces and moments due to light pressure acting on a body
in motion in cosmic space. Prikl. Math. Mekh. 26(5), 867–876 (1962) in Russian
90. Karymov, A.A.: Stability of rotational motion of a geometrically symmetric artificial satel-
lite of the Sun in the field of light pressure forces. Prikl. Math. Mekh. 28(5), 923–930 (1964)
in Russian
91. Beletsky, V.V., Grushevsky, A.V., Starostin, E.L.: Controlling the rotation of a spacecraft by
means of solar radiation pressure forces. J. Comput. Syst. Sci. Int. 32(3), 70–76 (1994)
92. Sidorenko, V.V.: Rotational motion of a spacecraft with solar stabilizer. Kosm. Issl. 30(6),
780–790 (1992) in Russian
93. Kogan, A.Y., Kirsanova, T.S.: Rotation of a spun spacecraft in the light flow. Kosm. Issl. 32
(3), 74–87 (1994) in Russian
94. Komarov, M.M., Sazonov, V.V.: Calculation of the light pressure forces and torques acting on
asteroid of an arbitrary shape. Astr. Vestnik. 28(1), 21–30 (1994) in Russian
95. Sazonov, V.V.: Motion of an asteroid relative to the centre of mass under the action of the
moment of light pressure forces. Astr. Vestnik. 28(2), 95–107 (1994) in Russian
96. Vokrouhlicky, D., Milani, A.: Direct solar radiation pressure on the orbits of small near-earth
asteroids: observable effects? Astr. Astrophys. 362, 746–755 (2000)
Survey of Literature xxv

97. Rubincam, D.P.: Radiative spin-up and spin-down of small asteroids. ICARUS 148, 2–11
(2000)
98. Scheeres, D.J., Mirrahimi, S.: Rotational dynamics of a solar system body under solar radi-
ation torques. Celest. Mech. Dyn. Astr. 101, 69–103 (2008)
99. Cicalo, S., Scheeres, D.J.: Averaged rotational dynamics of an asteroid in tumbling rotation
under the YORP torque. Celest. Mech. Dyn. Astr. 106, 301–337 (2010)
100. Vasiliev, L.A.: Determination of light pressure on the spacecraft. Mashinostroyeniye, Moscow
(1985) in Russian
101. Sazonov, V.V., Sazonov, V.V.: Calculation of resultant vector of light pressure forces acting
on the spacecraft with a solar sail. Cosmic. Res. 49(1), 56–64 (2011)
102. Leshchenko, D.D., Shamaev, A.S.: Motion of a satellite relative to the center of mass under
the action of moments of light-pressure forces. Mech. Solids 20(1), 11–18 (1985)
103. Akulenko, L.D., Leshchenko, D.D.: Evolution of rotation of a nearly dynamically spherical
triaxial satellite under the action of light pressure forces. Mech. Solids 31(2), 1–10 (1996)
104. Leshchenko, D.D.: Evolution of rotation of a triaxial body under the action of the torque
due to light pressure. Mech. Solids 32(6), 12–20 (1997)
105. Akulenko, L.D., Leshchenko, D.D., Suksova, S.G., Timoshenko, I.A.: Motion of a satel-
lite relative to the center of mass under the action of gravitational and light torques.
Mekh. Tverd. Tela. 34, 95–105 (2004)
106. Akulenko, L.D., Leshchenko, D.D., Suksova, S.G., Timoshenko, I.A.: Evolution of rotation
of a nearly dynamically spherical triaxial satellite under the action of gravitational and light
torques. Mech. Solids 41(4), 73–82 (2006)
107. Zhukovsky, N.E.: On the Motion of a Rigid Body with Cavities Filled with a Homogeneous
Liquid in Drop, vol. 1, pp. 31–152. Selected Works. Gostekhizdat, Moscow-Leningrad
(1948) in Russian
108. Moiseev, N.N., Rumyantsev, V.V.: Dynamics and Stability of Bodies Containing Fluid.
Springer, New York, NY (1968)
109. Mikishev, G.N., Rabinovich, B.I.: Dynamics of a Rigid Body Containing Partially Liquid-
Filled Cavities. Mashinostroyeniye, Moscow (1968) in Russian
110. Narimanov, G.S., Dokuchaev, L.V., Lukovsky, I.A.: Nonlinear Dynamics of a Flight Vehicle
with Fluid. Mashinostroyeniye, Moscow (1977) in Russian
111. Rumyantsev, V.V., Rubanovsky, V.N., Stepanov, S.Y.: Vibrations and Stability of
Rigid Bodies with Fluid-Filled Cavities. Vibrations in Engineering, vol. 2, pp. 280–306.
Mashinostroyeniye, Moscow (1979) in Russian
112. Anchev, A., Rumyantsev, V.V.: On the dynamics and stability of gyrostats. Usp. Mekh. 2(3),
3–45 (1979) in Russian
113. Chernousko, F.L.: Motion of a rigid body with cavities filled with viscous fluid at
small Reynolds number. USSR Comput. Math. Math. Phys. 5(6), 99–127 (1965)
114. Chernousko, F.L.: Motion of a body with a cavity filled with a viscous fluid at large Reynolds
number. J. Appl. Math. Mech. 30(3), 568–589 (1966)
115. Chernousko, F.L.: On free oscillations of a viscous fluid in a vessel. J. Appl. Math. Mech. 30
(5), 990–1003 (1966)
116. Chernousko, F.L.: The motion of a body with a cavity partly filled with a viscous liquid.
J. Appl. Math. Mech. 30(6), 1167–1184 (1966)
117. Chernousko, F.L.: Oscillations of a vessel containing a viscous fluid. Fluid Dyn. 2(1), 39–43
(1967)
118. Chernousko, F.L.: Oscillations of a rigid body with a cavity filled with a viscous fluid.
Inzh. Zh. Mekh. Tverd. Tela. 1, 3–14 (1967) in Russian
119. Chernousko, F.L.: Rotational motions of a solid body with a cavity filled with fluid. J. Appl.
Math. Mech. 31(3), 416–432 (1967)
xxvi Survey of Literature

120. Chernousko, F.L.: Motion of a solidcontaininga spherical damper. J. Appl. Mech. Techn.
Phys. 9(1), 45–48 (1968)
121. Chernousko, F.L.: Motion of a Rigid Body with Viscous-Fluid-Filled Cavities. Computing
Center AN SSSR, Moscow (1968) in Russian
122. Chernousko, F.L.: The Movement of a Rigid Body with Cavities Containing a Viscous Fluid.
NASA, Washington, DC (1972)
123. Kobrin, A.I.: On the motion of a hollow body filled with viscous liquid about its center of
mass in a potential body force field. J. Appl. Math. Mech. 33(3), 418–427 (1969)
124. Kobrin, A.I.: Asymptotic solution for the problem of influence of the liquid fill on the motion
of a controlled rigid body at low Reynolds numbers. Proc. Inst. Mech. Moscow Univ. 28,
65–78 (1973) in Russian
125. Smirnova, E.P.: Stabilization of free rotation of an asymmetric top with cavities completely
filled with fluid. J. Appl. Math. Mech. 38(6), 931–935 (1974)
126. Osipov, V.Z., Sulikashvili, R.S.: On vibrations of a solid body with spherical cavity
completely filled with viscous liquid in elliptical orbit. Tr. Tbilis. Mat. Inst. Akad. Nauk
Gruz. SSR. 58, 175–186 (1978) in Russian
127. Akulenko, L.D., Leshchenko, D.D.: Rapid rotation of a heavy gyrostat about a fixed point in
a resisting medium. Soviet Appl. Mech. 18(7), 660–665 (1982)
128. Leshchenko, D.D., Suksova, S.G.: About motion of non-symmetric gyrostat in a resisting
medium. J. Intern. Feder. Nonlin. Anal.-Acad. Nonlin. Sci. “Probl. Nelin. Anal. Inzhen. Sist.”
9(2)(18), 83–89 (2003)
129. Ivashchenko, B.P.: Motion of a gyroscope with a cavity filled with a viscous liquid.
Soviet Appl. Mech. 14(8), 872–876 (1978)
130. Pivovarov, M.L., Chernousko, F.L.: Oscillations of a rigid body with a toroidal cavity filled with
a viscous liquid. J. Appl. Math. Mech. 54(2), 164–168 (1990)
131. Pivovarov, M.L.: Fluid damping of the oscillations of a satellite with a large magnetic moment.
Kosm. Issl. 28(6), 865–873 (1990) in Russian
132. Vil’ke, V.G., Shatina, A.V.: Evolution of rotation of a satellite with cavity filled with
viscous liquid. Cosmic Res. 31(6), 609–617 (1993)
133. Baranova, E.Y., Vil’ke, V.G.: Evolution of motion of a rigid body with a fixed point and an
ellipsoidal cavity filled with a viscous fluid. Moscow Univ. Mech. Bull. 68(1), 15–20 (2013)
134. Vil’ke, V.G.: Analytical Mechanics of Systems with an Infinite Number of Degrees of Freedom.
Parts 1, 2. Mech.-Math. Dept. Moscow State Univ., Moscow (1997) in Russian
135. Sidorenko, V.V.: Evolution of the rotational motion of a planet with a liquid core. Solar Syst.
Res. 27(2), 201–208 (1993)
136. Akulenko, L.D., Leshchenko, D.D., Rachinskaya, A.L.: Evolution of rotations of a satellite with
cavity filled with viscous liquid. Mekh. Tverd. Tela. 37, 126–139 (2007) in Russian
137. Akulenko, L.D., Leshchenko, D.D., Rachinskaya, A.L.: Rotations of a satellite with cavity
filled with viscous liquid under the action of moment of light pressure forces. Mekh. Tverd.
Tela. 38, 95–110 (2008) in Russian
138. Akulenko, L.D., Zinkevich, Ya.S., Leshchenko, D.D., Rachinskaya, A.L.: Rapid rotations of
a satellite with cavity filled with viscous fluid under the action of moments of gravity and
light pressure forces. Cosmic Res. 49(5), 440–451 (2011)
139. Disser, K., Galdi, G.P., Mazzone, G., Zunino, P.: Inertial motions of a rigid body with a
cavity filled with a viscous liquid. Arch. Rational Mech. Anal. 221, 487–526 (2016)
140. Kuang, J.L., Leung, A.Y.T., Tan, S.: Chaotic attitude oscillations of a satellite filled with a
rotating ellipsoidal mass of liquid subject to gravity-gradient torques. Chaos. 14(1), 111–117
(2004)
141. Leung, A.Y.T., Kuang, J.L.: Chaotic rotations of a liquid – filled solid. J. Sound Vib. 302(3),
540–563 (2007)
Survey of Literature xxvii

142. Baozeng, Y., Xie, J.: Chaotic attitude maneuvers in spacecraft with completely liquid-filled
cavity. J. Sound Vib. 302(4–5), 643–656 (2007)
143. Alekseev, A.V.: Movement of the satellite gyrostat containing a cavity with a liquid of the
large viscosity. Izv. SNC RAN 9(3), 671–676 (2007) in Russian
144. Alekseev, A.V., Bezglasnyi, S.P., Krasnikov, V.S.: Building stabilizing control for stationary
motions of a gyrostat with a cavity with viscous fluid. Izv. SNC RAN 15(6), 563–567 (2013)
in Russian
145. Gurchenkov, A.A., Nosov, M.V., Tsurkov, V.I.: Control of Fluid- Containing Rotating
Rigid Bodies. CRC Press, Boca Raton, FL (2013)
146. Litvin-Sedoi, M.Z.: Mechanics of the Systems of Connected Rigid Bodies. Advances in
Science: General Mechanics, vol. 5, pp. 3–61. VINITI, Moscow (1982) in Russian
147. Dokuchaev, L.V.: Nonlinear Dynamics of an Elastic Aircraft. Advances in Sciences:
General Mechanics, vol. 5, pp. 135–197, VINITI, Moscow (1982) in Russian
148. Modi, V.J.: Dynamics of satellites with flexible appendages – a brief review. J. Spacecr.
Rockets 11(11), 743–751 (1974)
149. Roberson, R.E.: Two decades of spacecraft attitude control. J. Guid. Control. 2(1), 3–8 (1979)
150. Lilov, L.K.: Simulation of the Systems of Connected Rigid Bodies. Nauka, Moscow (1993)
in Russian
151. Gorr, G.V., Maznev, A.V.: Dynamics of a Gyrostat having a Fixed Point. Donetsk National
University, Donetsk (2010) in Russian
152. Rauschenbach, B.V., Tokar, E.N.: Control of the Spacecraft Orientation. Nauka, Moscow
(1974) in Russian
153. Roberson, R.E.: Torques on a satellite vehicle from internal moving parts. J. Appl. Mech. 25
(2), 196–200, 287–288 (1958)
154. Haseltine, W.R.: Passive damping of wobbling satellites: general stability theory and exam-
ple. J. Aerosp. Sci. 29(5), 543–549, 557 (1962)
155. Colombo, G.: The motion of satellite 1958 Epsilon around its center of mass. Smithson.
Contrib. Astrophys. 6, 149–163 (1963)
156. Thomson, W.T.: Introduction to Space Dynamics. Dower, New York, NY (1986)
157. Letov, A.M.: Fligt Dynamics and Control. Nauka, Moscow (1969) in Russian
158. Kachurina, N.M., Krementulo, V.V.: Stabilization of the rotational motion of a solid with
movable masses. Izv. Akad. Nauk SSSR Mekh. Tverd. Tela. 3, 96–101 (1981) in Russian
159. Kunciw, B.G., Kaplan, M.H.: Optimal space station detumbling by internal mass motion.
Automatica 12(5), 417–425 (1976)
160. Kravets, V.V.: Stabilization of a material system by moving the carried mass. Space Res.
Ukraine 10, 48–50 (1977) in Russian
161. Pankova, N.V., Rubanovsky, V.N.: Stability and bifurcation of steady rotations of a free rigid
body and a point mass elastically connected to it. Izv. Akad. Nauk SSSR Mekh. Tverd. Tela.
4, 14–18 (1976) in Russian
162. Kane, T.R., Levinson, D.A.: Stability, instability and terminal attitude motion of a
spinning dissipative spacecraft. AIAA J. 14(1), 39–42 (1976)
163. Cloutier, G.J.: Resonances of a two-dof system on a spin-stabilized spacecraft. AIAA J. 14
(1), 107–109 (1976)
164. Kane, T.R., Sher, M.P.: A method of active attitude control based on energy considerations.
J. Spacecr. Rockets 6, 633–636 (1969)
165. Cochran, J.E., Speakman, N.O.: Rotational motion of a free body induced by mass redistri-
bution. J. Spacecr. Rockets 12(2), 89–95 (1975)
166. Halmser, D.M., Mingori, D.I.: Nutational stability and passive control of spinning rocket with
internal mass motion. J. Guid. Contr. Dyn. 18(5), 1197–1203 (1995)
167. Janssens, F.L., Van der Ha, J.C.: Stability of spinning satellite under axial thrust and
internal mass motion. Acta Astronautica 94(1), 502–514 (2014)
xxviii Survey of Literature

168. Rajan, M., Bainum, P.M.: Influence of gravitational torques on dynamics of spinning
satellite with movable appendages. Cosmic Res. 16(4), 497–504 (1978) in Russian
169. Ananiev, V.V.: Bifurcation set in the problem of motion of a heavy rigid body and a point
mass elastically connected with it. Vestn. Lening. Un-ta. Mat. Mekh. Astron. 1, 54–60
(1982) in Russian
170. Month, L.A., Rand, R.H.: Stability of a rigid body with an oscillating particle. An application of
MACSYMA. Trans. ASME J. Appl. Mech. 52, 686–692 (1985)
171. Salimov, G.R.: On the influence of the astronaut motion on the spatial position of the
spacecraft. Izv. Akad. Nauk SSSR Mekh. Tverd. Tela. 2, 20–26 (1987) in Russian
172. Burov, A.A.: On the Motion of a Rigid Body Carrying a Movable Mass on a Spring. The
Problems of Studying the Stability and Stabilization of Motion, pp. 3–12, Computing Center
AN SSSR, Moscow (1987) in Russian
173. Loseva, N.N.: Stability of uniform rotations of a rigid body with a movable point mass. Dokl.
Akad. Nauk. Ukr. SSR Ser. A 6, 18–21 (1988) in Russian
174. Coppola, V.T.: The method of averaging for Euler’s equations of rigid body motion.
Nonlinear Dyn. 14(4), 295–308 (1997)
175. El-Gohary, A.: A global stability of rigid body containing moving masses. Int. J. Non-linear
Mech. 36(4), 663–669 (2001)
176. El-Gohary, A.: On the orientation of a rigid body using point masses. Appl. Mech. Comput.
151(1), 163–179 (2004)
177. Gray, G.L., Kammer, D.C., Dobson, I., Miller, A.J.: Heteroclinic bifurcations in rigid bodies
containing moving parts and a viscous damper. Trans. ASME J. Appl. Mech. 66, 720–728
(1999)
178. Miller, A.J., Gray, G.L., Mazzoleni, A.P.: Nonlinear spacecraft dynamics wih a flexible
appendage, damping and moving submasses. J. Guid. Contr. Dyn. 24(3), 605–615 (2001)
179. Markeyev, A.P.: Approximate equations of rotational motion of a rigid body carrying a
movable point mass. J. Appl. Math. Mech. 77(2), 137–144 (2013)
180. Baozeng, Y., Dandan, Y.: Study on the global chaotic dynamics and control of liquid filled
spacecraft with flexible appendages. Acta Mech. 209(1–2), 11–25 (2010)
181. Chernousko, F.L.: On the motion of rigid body with moving internal masses. Izv. Akad.
Nauk SSSR Mekh. Tverd. Tela. 4, 33–44 (1973) in Russian
182. Chernousko, F.L.: On the motion of solid body with elastic and dissipative elements. J. Appl.
Math. Mech. 42(1), 32–41 (1978)
183. Chernousko, F.L.: Motion of a viscoelastic solid relative to the center of mass. Mech. Solids
15(1), 17–21 (1980)
184. Chernousko, F.L., Shamaev, A.S.: The asymptotic behavior of singular perturbations in the
problem of the dynamics of a rigid body with elastic and dissipative elements. Izv. Akad.
Nauk SSSR Mekh. Tverd. Tela. 3, 33–42 (1983) in Russian
185. Leshchenko, D.D.: Motion of a rigid body with movable point mass. Mech. Solids 11(3),
33–36 (1976)
186. Akulenko, L.D., Leshchenko, D.D.: Some problems of the motion of a solid with a
moving mass. Mech. Solids 13(5), 24–28 (1978)
187. Leshchenko, D.D., Sallam, S.N.: Some problems on the motion of a rigid body with
internal degrees of freedom. Intern. Appl. Mech. 28(8), 524–528 (1992)
188. Kushpil, T.A., Leshchenko, D.D., Timoshenko, I.A.: Some problems of evolution of rota-
tions of a solid under the action of perturbation torques. Mekh. Tverd. Tela. 30, 119–125
(2000) in Russian
189. Akulenko, L., Leshchenko, D., Kushpil, T., Timoshenko, I.: Problems of evolution of rota-
tions of a rigid body under the action of perturbing moment. Multibody Syst. Dyn. 6(1), 3–16
(2001)
Survey of Literature xxix

190. Morozov, V.M., Rubanovsky, V.N., Rumyantsev, V.V., Samsonov, V.A.: On the bifurcation
and stability of the steady-state motions of complex mechanical systems. J. Appl. Math.
Mech. 37(3), 371–382 (1973)
191. Ganiev, R.F., Kononenko, V.O.: Oscillations of Solid Bodies. Nauka, Moscow (1976) in Russian
192. Veretennikov, V.G., Karpov, I.I., Markov, Y.G.: Oscillation Processes in the Mechanical
Systems with Elastic and Dissipative Elements: Study Guide. Moscow Aviation Institute,
Moscow (1998) in Russian
193. Ganiev, R.F., Koval’chuk, P.S.: The Dynamics of Systems of Rigid and Elastic Bodies.
Mashinostroyeniye, Moscow (1980) in Russian
194. Savchenko, A.Y., Bolgrabskaya, I.A., Kononykhin, G.A.: Stability of Motion of
Coupled Rigid Bodies. Naukova Dumka, Kiev (1991) in Russian
195. Dokuchaev, L.V.: The Nonlinear Dynamics of Aircraft with Deformable Elements.
Mashinostroyeniye, Moscow (1987) in Russian
196. Nabiullin, M.K.: Stationary Motions and the Stability of Elastic Satellites. Nauka, Novosi-
birsk (1990) in Russian
197. Banichuk, N.V., Karpov, I.I., Klimov, D.M., et al.: Mechanics of the Large Space Structures.
Factorial Publishing House, Moscow (1997) in Russian
198. Markov, Y.G., Minyaev, I.S.: On the dynamics of a spacecraft with elastic oscillating masses.
Kosm. Issl. 29(5), 684–694 (1991) in Russian
199. Martynenko, Y.G., Podalkov, V.V.: On the nutation of a rigid body in a non-contact
suspension. Izv. Akad. Nauk SSSR Mekh. Tverd. Tela. 2, 26–31 (1995) in Russian
200. Egarmin, N.E.: Influence of elastic deformations on the inertia tensor of a solid. Izv. Akad.
Nauk SSSR Mekh. Tverd. Tela. 6, 43–48 (1980) in Russian
201. Denisov, G.G., Novikov, V.V.: Free motions of a quasi-spherical deformable rigid body. Izv.
Akad. Nauk SSSR Mekh. Tverd. Tela. 3, 43–50 (1983) in Russian
202. Novikov, V.V.: Free motion of an anisotropically elastic ball. Prikl. Mat. Mekh. 51(5),
767–774 (1987) in Russian
203. Sidorenko, V.V.: Motion of a rigid body with flexible rods that admits a symmetry group.
Mech. Solids 30(1), 1–9 (1995)
204. Sidorenko, V.V.: The dynamic evolution of a mechanical system with a very rigid linear
damper. J. Appl. Math. Mech. 59(4), 533–539 (1995)
205. Shatina, A.V.: Evolution of the rotational motion of a symmetric satellite with flexible visco-
elastic rods. Kosm. Issl. 40(2), 178–192 (2002) in Russian
206. Gulyaev, V.I., Lizunov, P.P.: Oscillations of the Systems of Rigid and Deformable Bodies
during Complex Motion. Vishcha Shkola, Kiev (1989) in Russian
207. Smolnikov, B.A., Belyaev, A.K.: Evolutional dynamics and stability of dissipative solids.
Acta Mech. 195, 365–377 (2008)
208. Kuzmak, G.E.: The Dynamics of the Uncontrolled Motion of a Vehicle during Atmospheric
Re-Entry. Nauka, Moscow (1970) in Russian
209. Kuzmak, G.E.: Motion of an axially symmetric rigid body about a fixed point under the
influence of torques slowly changing in time. Izv. Akad. Nauk SSSR OTN. Mekh.
Mashinostr. 4, 65–78 (1961) in Russian
210. Yaroshevsky, V.A.: Motion of an Uncontrolled Body in the Atmosphere. Mashino-
stroyeniye, Moscow (1978) in Russian
211. Aslanov, V.S.: Spatial Motion of a Body at Descent in the Atmosphere. Fizmatlit, Moscow
(2004) in Russian
212. Okunev, B.N.: The Free Motion of a Gyroscope. GITTL, Moscow (1951) in Russian
213. Pogosyan, T.I., Savchenko, A.Y.: On the motion of the Lagrange gyroscope in a field of
forces variable in direction. Mekh. Tverd. Tela. 12, 85–90 (1980) in Russian
214. Uzbek, E.K.: On the stability of permanent rotations of a rigid body in a field of forces
variable in direction. Mekh. Tverd. Tela. 26(1), 40–46 (1994) in Russian
xxx Survey of Literature

215. Mc Gill, D.J., Long, L.S.: The effect of viscous damping on spin stability of a rigid body with
a fixed point. Trans. ASME J. Appl. Mech. 44(2), 349–352 (1977)
216. Ge, Z.M., Wu, M.H.: The stability of a sleeping top with damping torque. Int. J. Eng. Sci. 27
(3), 285–288 (1989)
217. Puzyrev, V.E.: Stability of nonuniform rotations of Lagrange’s gyroscope about a main axis
in the presence of medium resistance forces. Mekh. Tverd. Tela. 17, 66–70 (1985) in Russian
218. Kovalev, A.M.: Motion of a body similar to Lagrange gyroscope. Mekh. Tverd. Tela. 3,
25–27 (1971) in Russian
219. Sergeev, V.S.: Periodic motions of a heavy, rigid, dynamically almost symmetric body with
a fixed point. J. Appl. Math. Mech. 47(1), 130–133 (1983)
220. Motorina, N.N.: Application of the Hori method to the study of a perturbed Lagrange top.
Izv. Akad. Nauk SSSR Mekh. Tverd. Tela. 1, 102–104 (1988) in Russian
221. Holmes, P.J., Marsden, J.E.: Horseshoes and Arnold diffusion for Hamiltonian systems on
Lie groups. Indiana Univ. Math. J. 32(2), 273–308 (1983)
222. Losyakova, D.A.: Chaotic motion of a top with the offset center of gravity. Izv. Sarat. Univ.
Nov. Ser. Math. Mech. Informatics 12(2), 90–95 (2012) in Russian
223. Aksenenkova, I.M.: On the influence of geomagnetic field on periodic motion of a satellite
relative to its center of mass. Kosm. Issl. 29(1), 145–148 (1991) in Russian
224. Konkina, L.I.: Conditionally periodic solutions for the problem of rotation of a magnetized
satellite in magnetic field. Kosm. Issl. 34(4), 442–444 (1996) in Russian
225. Sidorenko, V.V.: One class of motions for a satellite carrying a strong magnet. Cosmic Res.
40(2), 133–141 (2002)
226. Savchenko, A.Y., Bezruchenko, V.S.: Study of stationary motions of the Lagrange gyro-
scope in the presence of dissipation and unbalance of thrust. Mekh. Tverd. Tela. 25, 75–80
(1993) in Russian
227. Matrosov, V.M.: On the stability of motion. Prikl. Mat. Mech. 26(5), 885–895 (1962) in Russian
228. Zhuravlev, V.F.: One form of equations of the motion of symmetric body. Izv. Akad. Nauk
SSSP Mekh. Tverd. Tela. 3, 5–11 (1986) in Russian
229. Zhuravlev, V.F., Klimov, D.M.: Applied Methods in the Theory of Oscillations. Nauka,
Moscow (1988) in Russian
230. Koshlyakov, V.N.: On stability of motion of a symmetric body placed on a vibrating base.
Ukr. Math. J. 47(12), 1898–1904 (1995)
231. Koshlyakov, V.N.: On the structural changes of dynamical systems with gyroscopic forces.
Prikl. Mat. Mekh. 61(5), 774–780 (1997) in Russian
232. Fedorchenko, A.M.: The motion of a heavy symmetrical gyroscope with a vibrating point of
support. Ukr. Mat. Zh. 10(2), 209–218 (1958) in Russian
233. Fedorchenko, A.M.: On a dynamical method for increasing stability of a rapidly rotating
symmetrical gyroscope. Prikl. Mat. Mekh. 25(5), 938–940 (1961) in Russian
234. Markov, Y.G.: On the motion of a viscoelastic solid with a vibrating suspension point.
Izv. Akad. Nauk SSSR Mekh. Tverd. Tela. 6, 16–23 (1991) in Russian
235. Kovaleva, A.S.: Multi-frequency systems under stationary random perturbation. Part 1. Izv.
Akad. Nauk SSSR Mekh. Tverd. Tela. 3, 44–52 (1994) in Russian
236. Kholostova, O.V.: The dynamics of a Lagrange top with a vibrating suspension point.
J. Appl. Math. Mech. 63(5), 741–750 (1999)
237. Shilo, A.P.: Stability of stationary motions of the Lagrange gyroscope with a movable point
mass. Mekh. Tverd. Tela. 13, 97–101 (1981) in Russian
238. Kononov, Y.N.: Spin stability of a Lagrange top containing linear oscillators. J. Math. Sci.
103(1), 38–42 (2001)
239. Pozdnyakovich, A.E., Puzyrev, V.E.: Stability of permanent rotations around the principal
axes of the Lagrange’s gyroscope with dampers balancers. Mekh. Tverd. Tela. 42, 153–162
(2012) in Russian
Survey of Literature xxxi

240. Karapetyan, A.V., Lagutina, I.S.: The influence of dissipative and constant torques on the
form and stability of steady-state motions of Lagrange’s top. Izv. Akad. Nauk Mekh. Tverd.
Tela. 5, 29–33 (1998) in Russian
241. Kononov, Y.N., Kiseleva, N.V.: Influence of dissipative and constant torques on the stability
of uniform rotation in the resistive medium of the Lagrange top with ideal fluid. Visnyk
V.N. Karazin Kharkivsk. Nat. Univ. Ser. Math., Appl. Math. Mech. 850, 52–56 (2009) in
Russian
242. Ivashchenko, B.P.: On the motion of a symmetric gyroscope with a cavity filled with
viscous fluid. Dokl. Akad. Nauk Ukr. SSR Ser. A 9, 794–797 (1976) in Russian
243. Simpson, H.C., Gunzburger, M.D.: A two time scale analysis of gyroscopic motion with
friction. J. Appl. Math. Phys. 37(6), 867–894 (1986)
244. Markhashov, L.M.: On the evolution of regular precessions of a rigid body close to a
Lagrange top. Izv. Akad. Nauk SSSR Mekh. Tverd. Tela. 3, 8–12 (1980) in Russian
245. Sazonov, V.V., Sidorenko, V.V.: The perturbed motions of a solid close to regular Lagrang-
ian precessions. J. Appl. Math. Mech. 54(6), 781–787 (1990)
246. Sidorenko, V.V.: Capture and escape from resonance in the dynamics of the rigid body in
viscous medium. J. Nonlinear Sci. 4, 35–57 (1994)
247. Leshchenko, D.D.: The evolution of the rigid body motions close to Lagrange case. Top.
Probl. Aviat. Aerosp. Syst. Processes Models Exp. 2(6), 32–37 (1998) in Russian
248. Akulenko, L.D., Leshchenko, D.D., Chernousko, F.L.: Perturbed motions of a rigid body,
close to the Lagrange case. J. Appl. Math. Mech. 43(5), 829–837 (1979)
249. Kozachenko, T.A., Leshchenko, D.D., Rachinskaya, A.L.: Perturbed rotation of Lagrange
top under the action of nonstationary dissipative torques. Vestn. Odes. Nats. Univ. Mat.
Mekh. 16(16), 152–157 (2011) in Russian
250. Akulenko, L.D., Leshchenko, D.D., Chernousko, F.L.: Perturbed motions of a rigid body that
are close to regular precession. Mech. Solids 21(5), 1–8 (1986)
251. Leshchenko, D.D., Shamaev, A.S.: Perturbed rotational motions of a rigid body that are
close to regular precession in the Lagrange case. Mech. Solids 22(6), 6–15 (1987)
252. Leshchenko, D.D., Sallam, S.N.: Perturbed rotational motions of a rigid body similar to
regular precession. J. Appl. Math. Mech. 54(2), 183–190 (1990)
253. Leshchenko, D.D., Sallam, S.N.: Perturbed rotation of a rigid body relative to fixed point.
Mech. Solids 25(5), 15–23 (1990)
254. Leshchenko, D.D.: Perturbed rotational motion of a rigid body. In: Borne P. and Matrosov V.
(ed.) The Lyapunov Functions Method and Applications, pp. 227–232. Baltzer, J.C. AG,
Scientific Publishing, IMACS, Basel (1990)
255. Akulenko, L.D., Kozachenko, T.A., Leshchenko, D.D.: Rotations of a rigid body under the
action of unsteady restoring and perturbation torques. Mech. Solids 38(2), 1–7 (2003)
256. Akulenko, L.D., Kozachenko, T.A., Leshchenko, D.D.: Evolution of rotations of a rigid body
under the action of restoring and control moments. J. Comput. Syst. Sci. Int. 41(5), 868–874
(2002)
257. Akulenko, L.D., Kozachenko, T.A., Leshchenko, D.D.: Perturbed rotational motions of a
rigid body under the action of nonstationary restoring moment. Mekh. Tverd. Tela. 32,
77–84 (2002) in Russian
258. Akulenko, L.D., Kozachenko, T.A., Leshchenko, D.D.: Perturbed rotational motions of a
rigid body under the action of restoring moment depending on the angles of precession and
nutation. Mekh. Tverd. Tela. 35, 97–102 (2005) in Russian
259. Kozachenko, T.A., Leshchenko, D.D.: Perturbed rotational motions of Lagrange top.
Mekh. Tverd. Tela. 39, 62–68 (2009) in Russian
About this Book

This monograph presents the results of the authors’ research on the dynamics of the
rigid body motions about its center of mass. The authors consider the evolution of
these motions under the influence of various perturbation torques. The basic method
applied in the studies is the Krylov–Bogolubov asymptotic averaging method.
These questions arise in modern problems of dynamics, orientation and stabiliza-
tion of natural and artificial celestial bodies, gyroscopy and other areas of
mechanics.
For all cases of motion considered in the book, we present and analyze basic
equations, perform the averaging procedure and obtain the averaged equations,
which, being significantly simpler than the original ones, describe the motion over a
large time interval. We present the accuracy estimates for the asymptotic procedure.
As a result of analysis and solution of the obtained equations, we establish some
quantitative and qualitative specific features of the motions, provide a description
of the evolution of the body motion. The presentation is illustrated by numerous
examples.

Felix L. Chernousko
Leonid D. Akulenko
Dmytro D. Leshchenko

xxxiii
Contents

1 The Foundations of Dynamics of a Rigid Body with a Fixed Point . . . 1

1.1 Orientation of a Body: The Euler Angles . . . . . . . . . . . . . . . . . 1
1.2 Geometry of Mass: Moments of Inertia . . . . . . . . . . . . . . . . . . 1
1.3 Theorem of Change of Angular Momentum . . . . . . . . . . . . . . . 5
1.4 Dynamic Euler’s Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.5 Kinematic Euler’s Equations: Direction Cosines . . . . . . . . . . . . 8
1.6 Equations of Motion of a Heavy Rigid Body About a Fixed
Point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2 Motion of a Rigid Body by Inertia. Euler’s Case . . . . . . . . . . . . . . . 13
2.1 First Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Basic Formulas for the Jacobian Elliptic Functions . . . . . . . . . . 16
2.3 Integration of Dynamic Euler’s Equations: Analysis
of the Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4 Particular Cases (Regular Precession, Permanent Rotations) . . . 22
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3 Lagrange’s Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.1 Integration of the Equations of Motion and Analysis
of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2 Regular Precession . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3 Fast Spinning Top . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4 Equations of Perturbed Motion of a Rigid Body About Its Center of
Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1 The Concept of a Perturbed Motion . . . . . . . . . . . . . . . . . . . . . 41
4.2 Basic Concepts of the Averaging Method: Systems in a Standard
Form and Systems with Fast Rotating Phase . . . . . . . . . . . . . . . 42
4.3 Systems Containing Slow and Fast Motions . . . . . . . . . . . . . . . 46

xxxv
xxxvi Contents

4.4 Higher-Order Averaging in Systems with Fast and Slow

Phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.5 Equations of Perturbed Motion of a Rigid Body Close to
Euler’s Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.6 Equations of Perturbed Motion of a Satellite About Its Center
of Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.7 Procedure of Averaging for a Body with Moments of Inertia
Close to One Another . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.8 Equations of Perturbed Rotational Motion of a Rigid Body
Close to Lagrange’s Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.8.1 General Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.8.2 The Case Where the Projections of the Perturbation
Torque Vector Are of Different Orders of Smallness . . . 65
4.8.3 Perturbation Torques Are Small Compared to the
Restoring Ones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5 Perturbation Torques Acting upon a Rigid Body . . . . . . . . . . . . . . 73
5.1 Gravitational Torques Acting upon a Satellite . . . . . . . . . . . . . . 73
5.2 Rigid Body in a Resistant Medium . . . . . . . . . . . . . . . . . . . . . . 78
5.3 Rigid Body with a Cavity Filled with the Fluid of High
Viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.4 Case of Moving Masses Connected to the Body by Elastic
Couplings with Viscous Friction . . . . . . . . . . . . . . . . . . . . . . . 82
5.5 Body with Elastic and Dissipative Elements . . . . . . . . . . . . . . . 87
5.6 Viscoelastic Solid Body . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.7 Influence of a Moving Mass Connected to the Body
by an Elastic Coupling with Quadratic Friction . . . . . . . . . . . . . 101
5.8 Torque Due to the Solar Pressure . . . . . . . . . . . . . . . . . . . . . . . 103
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
6 Motion of a Satellite About Its Center of Mass Under the Action
of Gravitational Torque . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
6.1 Motion of a Triaxial Satellite with Moments of Inertia Close
to One Another . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
6.2 Fast Rotations of a Satellite with a Triaxial Ellipsoid
of Inertia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
6.3 Resonance Phenomena in the Planar Motion of a Satellite
About Its Center of Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
7 Motion of a Rigid Body with a Cavity Filled with a Viscous
Fluid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
7.1 Equations of Motion of a Body with a Viscous Fluid
in a Cavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
7.2 Planar Motion of a Pendulum with a Viscous Fluid . . . . . . . . . . 141
Contents xxxvii

7.3 Free Three-Dimensional Motion of a Body with a Viscous

Fluid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
7.4 On the Motion of a Rigid Body Containing a Damper . . . . . . . . 153
7.5 Stability of Motion of a Rigid Body with a Damper . . . . . . . . . 159
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
8 Evolution of Rotations of a Rigid Body in a Medium . . . . . . . . . . . 165
8.1 Fast Motion of a Heavy Rigid Body About a Fixed Point . . . . . 165
8.2 Rotation of a Heavy Rigid Body About a Fixed Point
in a Viscous Medium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
8.2.1 Statement of the Problem and the Averaging
Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
8.2.2 Analysis of the Equation for k2 . . . . . . . . . . . . . . . . . . 172
8.2.3 Qualitative Investigation of the Specific Cases
of the Rigid Body Motion . . . . . . . . . . . . . . . . . . . . . . 175
8.2.4 Investigation of the Stability of Quasi-stationary
Motions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
8.2.5 The Case of Dynamic Symmetry . . . . . . . . . . . . . . . . . 180
8.3 Fast Rotation of a Satellite About Its Center of Mass Under
the Action of Gravitational Torque in a Resistive Medium . . . . 181
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
9 Motion of a Rigid Body with Internal Degrees of Freedom . . . . . . . 191
9.1 Dynamics of a Rigid Body with a Movable Internal Mass . . . . . 191
9.1.1 The Case of Complete Body Symmetry . . . . . . . . . . . . 191
9.1.2 Motion of Dynamically Symmetric Rigid Body with
a Movable Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
9.2 Motion of a Rigid Body with a Movable Mass Connected
to the Body by an Elastic Coupling with Quadratic Friction . . . . 200
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
10 Influence of the Torque Due to the Solar Pressure upon
the Motion of a Sun Satellite Relative to Its Center of Mass . . . . . . 205
10.1 Equations of Spacecraft Rotation Under the Action of the
Torque Due to the Solar Pressure . . . . . . . . . . . . . . . . . . . . . . . 205
10.2 Evolution of Rotation of a Spacecraft with Moments of Inertia
Close to Each Other . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
10.2.1 Basic Assumptions and Statement of the Problem . . . . 207
10.2.2 Transformation of the Expression for the Force Function:
Averaging Procedure and Construction of the First
Approximation System . . . . . . . . . . . . . . . . . . . . . . . . 209
10.2.3 Investigation of Equations for the Angles of Nutation
and Proper Rotation . . . . . . . . . . . . . . . . . . . . . . . . . . 212
10.2.4 Taking into Account the Zero and First Harmonics
in the Approximation of the Solar Pressure Torque . . . . 212
xxxviii Contents

10.3 Taking into Account the Third and Even Harmonics in the
Approximation of the Solar Pressure Torque . . . . . . . . . . . . . . . 221
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
11 Perturbed Motions of a Rigid Body Close to Lagrange’s Case . . . . 227
11.1 General Properties of the Averaging Procedure
over the Lagrange Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
11.2 Perturbed Motion of a Body Under Linear Dissipative
Torques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
11.3 Evolution of Rotation of a Rigid Body Under Various
Assumptions for the Perturbation Torque . . . . . . . . . . . . . . . . . 233
11.3.1 General Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
11.3.2 Influence of External Dissipative Torques . . . . . . . . . . 236
11.3.3 Action of a Small Constant Torque, Applied Along
the Symmetry Axis . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
11.3.4 The Case of a Body Close to the Dynamically
Symmetric One . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
Chapter 1
The Foundations of Dynamics of a Rigid
Body with a Fixed Point

1.1 Orientation of a Body: The Euler Angles

Consider a rigid body moving about a fixed point O. The position of the rigid body
in space is determined at each moment of time by the position of the moving
coordinate system Oxyz connected with the body relative to the fixed coordinate
system Ox1y1z1. To determine the body position, three independent (among them-
selves) parameters corresponding to the number of degrees of freedom of the body
are introduced. Let us consider one of the most common ways of determining
orientation of a rigid body using the Euler angles [1–4]. The plane Oxy intersects
with plane Ox1y1 along the line ON, called the line of nodes. Figure 1.1 shows the
angle ψ of precession as the angle between the axis Ox1 and the line of nodes ON,
the angle of nutation θ as the angle between the axes Oz and Oz1, and the angle φ of
proper rotation as the angle between the line of nodes ON and the axis Ox.
It is usually assumed that 0
ψ < 2π, 0
θ < π, 0
φ < 2π. The transition from
the coordinate system Ox1y1z1 to the system Oxyz is carried out by means of three
successive rotations: by angle ψ about the axis Oz1, by angle θ about the line of
nodes ON, and by angle φ about the axis Oz.
The cases θ ¼ 0 and θ ¼ π are special. In these cases, the line of nodes ON and
angles φ and ψ are not defined; only their sum φ + ψ is defined.

1.2 Geometry of Mass: Moments of Inertia

The moment of inertia of a mechanical system consisting of n material points with

respect to an axis u is defined as the sum of the products of the point masses and the
squared distances hi from the points to the axis:

© Springer International Publishing AG 2017 1

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_1
2 1 The Foundations of Dynamics of a Rigid Body with a Fixed Point

Fig. 1.1 Euler angles

X
n
Ju ¼ mi h2i : ð1:1Þ
i¼1

In the case of a rigid body, we obtain by the corresponding limiting process

X
n ð
J u ¼ lim h2i Δmi ¼ h2 dm: ð1:2Þ
Δ mi !0
i¼1
n!1 ðMÞ

Here, the integral is defined for the mass of the body in such a way that
each element of the mass is multiplied by the square of its distance to the axis.
For a continuous perfectly rigid body, the mass dm of an elementary part of the
body equals dm ¼ ρdτ, where ρ is the body density and dτ is the volume element.
Consider a rigid body in the Cartesian rectangular coordinate system Oxyz. If the
body is nonhomogeneous, then its density ρ changes from point to point and is a
function of coordinates ρ ¼ ρ(x, y, z). The mass M of the body can be calculated by
the triple integral
ð ððð
M ¼ ρdτ ¼ ρdxdydz: ð1:3Þ

The integration in (1.3) is performed over the volume of the body.

If we denote by h the distance from an infinitesimal element of mass dm to the
axis u, then, according to (1.2), the moment of inertia Ju of the body with respect to
the axis u is determined by the integral
1.2 Geometry of Mass: Moments of Inertia 3

ððð
Ju ¼ ρh2 dxdydz: ð1:4Þ

In view of (1.4), the moments of inertia of a rigid body with respect to the
axes Ox, Oy, and Oz have the form
ÐÐÐ
J x ¼ ÐÐÐ ρðy2 þ z2 Þdxdydz,
J y ¼ ÐÐÐ ρðx2 þ z2 Þdxdydz, ð1:5Þ
Jz ¼ ρðx2 þ y2 Þdxdydz:

The center of mass (center of inertia) of a mechanical system is defined as a

geometric point C, the radius vector of which is defined by the formula

P
n
m i ri
rC ¼ i¼1 , ð1:6Þ
M

P
n
where M ¼ mi is the mass of the mechanical system, and ri are the radius vectors
i¼1
of the points of this mechanical system.
In the case of a continuous rigid body, the coordinates of the center of mass are
determined as follows:
ððð
1
xC ¼ ρxdxdydz,
Mððð
1
yC ¼ ρydxdydz, ð1:7Þ
Mððð
1
zC ¼ ρzdxdydz:
M

Here, M is calculated by formula (1.3).

According to the Huygens–Steiner theorem [1, 3, 4], the moment of inertia of a
mechanical system with respect to some axis is equal to the sum of the moment of
inertia with respect to the central axis parallel to the given one and the product of
the system mass by the squared distance between the axes:

J z ¼ J z0 þ Md 2 : ð1:8Þ

Here, the axis z0 passing through the center of mass is called central, M is the
mass of the system, and d is the distance between the axes.
The products of inertia are defined by the equalities
4 1 The Foundations of Dynamics of a Rigid Body with a Fixed Point

ÐÐÐ
J xy ¼ ÐÐÐ ρxydxdydz,
J xz ¼ ÐÐÐ ρxzdxdydz, ð1:9Þ
J yz ¼ ρyzdxdydz:

The symmetric matrix composed of the axial Jx, Jy, Jz and the opposites of the
products of inertia Jxy, Jzy, Jzx is called the tensor of inertia of the body at the
given point O:
 
 Jx J xy J xz 
 
J¼  J yx Jy J yz 
 ð1:10Þ
 J zx J zy Jz 

Let us draw three mutually perpendicular axes x, y, and z through the point O of
the body and an arbitrarily directed axis u which form the angles α, β, and γ,
respectively, with the above axes. The moment of inertia of the rigid body relative
to the axis u is expressed by the formula

J u ¼ Je  e,

where e is the unit vector (basis vector) directed along the axis u. In the scalar form,
we have

J u ¼ J x cos 2 α þ J y cos 2 β þ J z cos 2 γ  2J xy cos α cos β  2J yz cos β cos γ

 2J zx cos γ cos α: ð1:11Þ

Consider the change of the moment of inertia Ju taking place under changing the
direction of the axis u, i.e., under changing the angles α, β, γ. To visualize this
pffiffiffiffiffi
change, let us draw a segment OL of length 1= J u from the point O along the axis u.
If we exclude the bodies having the form of an infinitely thin rod, for which the
moment of inertia relative to the rod axis equals zero, then the point L describes a
closed surface under changing of the direction of the axis u. The equation of this
surface is written as follows:

J x x2 þ J y y2 þ J z z2  2J xy xy  2J yz yz  2J zx zx ¼ 1: ð1:12Þ

This surface is an ellipsoid and is called the ellipsoid of inertia of the body for the
point O. Three axes of symmetry of the ellipsoid of inertia are called principal axes
of inertia of the body for this point.
If we consider principal axes of inertia as the coordinate axes Ox0 y0 z0 , then the
tensor of inertia is diagonal and matrix (1.10) has the form
1.3 Theorem of Change of Angular Momentum 5

 
 A1 0 0 
 
J¼
 0 A2 0 : ð1:13Þ
 0 0 A3 

The moments of inertia A1, A2, A3 are called principal moments of inertia of the
body for the point O. The principal axes of inertia for the center of mass of the body
are called principal central axes of inertia.

1.3 Theorem of Change of Angular Momentum

The angular momentum of a material point relative to a pole O is defined as the

vector product of the radius vector r of the material point, drawn from this pole,
by its momentum mv:

m0 ðmvÞ ¼ r  mv: ð1:14Þ

The angular momentum of a mechanical system relative to the pole O is

understood as the vector sum of the angular momenta of the points of the system
relative to the same pole:

X
n X
n
G¼ m 0 ð m i vi Þ ¼ ri  mi vi : ð1:15Þ
i¼1 i¼1

Here, mivi is the momentum of the i-th point, and ri is the radius vector
connecting the pole O with the i-th point of the system.
The angular momentum of a rigid body with respect to the pole O in the case of
continuous distribution of mass in the volume τ is defined as follows:
ððð ððð
G¼ ρr  v dτ ¼ ρr  v dxdy dz, ð1:16Þ

where ρ is the density, v is the velocity vector of the infinitesimal element dm of the
body, and dτ is the volume element. The integration in (1.16) is carried out over the
entire volume.
If the body rotates about a fixed pole O, which is taken as the origin of the
coordinate system Oxyz rigidly connected with the rigid body, then we will get for
the projections of angular momentum on the axes of this coordinate system
6 1 The Foundations of Dynamics of a Rigid Body with a Fixed Point

Gx ¼ J x p  J xy q  J xz r,
Gy ¼ J yx p þ J y q  J yz r, ð1:17Þ
Gz ¼ J zx p  J zy q þ J z r:

Here, p, q, r are the projections of the angular velocity vector ω of the body on
the axes of the selected coordinate system Oxyz.
Using the notation J from (1.10) for the tensor of inertia of the body for the
pole O, we can represent (1.17) in the form

G ¼ Jω: ð1:18Þ

If we take as moving axes the principal axes of inertia of the body for a motion-
less pole, then formulas (1.17) are simplified and can be written as follows:

Gx ¼ A1 p, Gy ¼ A2 q, Gz ¼ A3 r: ð1:19Þ

Theorem of change of angular momentum relative to the fixed pole has the form

dG
¼ L0e : ð1:20Þ
dt

Thus, the time derivative of the angular momentum of the system relative to the
fixed pole equals to the principal moment L0e of external forces of the system rela-
tive to this pole.
If L0e ¼ 0, then, according to (1.20), the angular momentum G of the system is a
constant quantity:

G ¼ const: ð1:21Þ

Projecting both sides of equality (1.21) on the coordinate axes, we obtain

three scalar integrals:

Gx ¼ C1 , Gy ¼ C2 , Gz ¼ C3 : ð1:22Þ

1.4 Dynamic Euler’s Equations

Consider a rigid body moving about a fixed point O. To describe the motion of the
body, we introduce a fixed coordinate system Ox1y1z1.
A moving coordinate system Oxyz is connected with the body, and its axes are
directed along the principal axes of inertia of the body for the point O. We apply the
theorem of change of angular momentum relative to the fixed point (1.20)
(Fig. 1.2).
1.4 Dynamic Euler’s Equations 7

Fig. 1.2 Coordinate

systems

dG
The absolute derivative relative to the fixed coordinate system Ox1y1z1 is
dt
e
connected with the local derivative ddtG with respect to the moving coordinate system
Oxyz by the formula

dG dGe
¼ þ ω  G: ð1:23Þ
dt dt

Then equation (1.20) can be written in the form

e
dG
þ ω  G ¼ L0e : ð1:24Þ
dt

Projecting both sides of equality (1.24) onto the axes of the moving coordinate
system Oxyz, we get

dGx
þ qGz  rGy ¼ Lxe ,
dt
dGy
þ rGx  pGz ¼ Lye , ð1:25Þ
dt
dGz
þ pGy  qGx ¼ Lze :
dt

Here, p, q, r are the projections of the angular velocity ω of the body on the
axes of the coordinate system Oxyz.
If the axes of the moving coordinate system Oxyz are directed along the principal
axes of inertia of the body for the point O, then

Gx ¼ A1 p, Gy ¼ A2 q, Gz ¼ A3 r, ð1:26Þ

where A1, A2, A3 is the principal moments of inertia of the body for the point O.
As a result, equations (1.25) are written as follows:
8 1 The Foundations of Dynamics of a Rigid Body with a Fixed Point

A1 p_ þ ðA3  A2 Þqr ¼ Lxe ,

A2 q_ þ ðA1  A3 Þrp ¼ Lye , ð1:27Þ
A3 r_ þ ðA2  A1 Þpq ¼ Lze :

Here and elsewhere the point denotes the derivatives with respect to time t.
Equations (1.27) are called dynamic Euler’s equations.
If Lxe , Lye , Lze are functions of p, q, r, t, then equations (1.27) form a closed system
of equations, integration of which yields the dependence of the quantities p, q, r on
time t and initial conditions p0, q0, r0.

1.5 Kinematic Euler’s Equations: Direction Cosines

Consider a rigid body moving about a fixed point O. Denote the fixed coordinate
system by Ox1y1z1, whereas the moving coordinate system invariably connected
with the rigid body, by Oxyz. The body position at the given moment of time is
determined by the position of the moving coordinate system relative to the
fixed one, which is given by the Euler angles ψ, θ, φ (Fig. 1.3).
The angular velocity ω of the rigid body is equal to the sum of angular velocities of
three rotations: with angular velocity ψ_ about the Oz1 axis, with angular velocity θ_
about the line of nodes ON, and with angular velocity φ_ about the Oz axis (Fig. 1.3).
Kinematic Euler’s equations, expressing the projections p, q, r of angular
velocity ω onto the axes of the moving coordinate system Oxyz through the Euler
angles and their derivatives, can be obtained from Fig. 1.3, in which an auxiliary
line OM lies in the plane Oxy and is perpendicular to the line of nodes. We have

p ¼ ψ_ sin θ sin φ þ θ_ cos φ,

q ¼ ψ_ sin θ cos φ  θ_ sin φ, ð1:28Þ
r ¼ ψ_ cos θ þ φ_ :

The relations expressing ψ_ , θ_ , φ_ by p, q, r have the form

1
ψ_ ¼ ðp sin φ þ q cos φÞ,
sin θ ð1:29Þ
θ_ ¼ p cos φ  q sin φ,
φ_ ¼ r  ðp sin φ þ q cos φÞθ:

Expressions (1.29) have singularities for small angles of nutation.

Let us express the directional cosines of the axes Oxyz in terms of the Euler
angles [5]. Denote the unit vector of the line of nodes by n, the unit vectors of the
axes Oz1 and Oz, by k1 and k, respectively. Construct two auxiliary coordinate
0
trihedra: n, n , k1 and n, n1, k, oriented as right-handed coordinate systems
0
(Fig. 1.4), where vector n lies in the plane Ox1y1 and vector n1, in the plane Oxy.
1.5 Kinematic Euler’s Equations: Direction Cosines 9

Fig. 1.3 Coordinate

systems and Euler angles

Fig. 1.4 Unit vectors and

Euler angles

Then the unit vectors of the axes Ox1, Oy1, Ox, Oy can be represented as

i1 ¼ n cos ψ  n0 sin ψ, j1 ¼ n sin ψ þ n0 cos ψ,

ð1:30Þ
i ¼ n cos φ þ n1 sin φ, j ¼ n sin φ þ n1 cos φ:

Note the following relations

n  n ¼ 1, n  n0 ¼ n  n1 ¼ 0, n0  n1 ¼ cos θ,
n  k ¼ n  k1 ¼ 0, n0  k ¼  sin θ:
10 1 The Foundations of Dynamics of a Rigid Body with a Fixed Point

Let us find nine directional cosines of the axes Oxyz, which are equal to the
scalar products of the respective unit vectors of the fixed and moving coordinate
systems. We have

α11 ¼ cos ðxd

1 , x Þ ¼ cos ψ cos φ  sin ψ sin φ cos θ,
α12 ¼ cos ðxd
1 , y Þ ¼  cos ψ sin φ  sin ψ cos φ cos θ,
α13 ¼ cos ðxd
1 , z Þ ¼ sin ψ sin θ,
α21 ¼ cos ðyd
1 , x Þ ¼ sin ψ cos φ þ cos ψ sin φ cos θ,
α22 ¼ cos ðyd
1 , y Þ ¼  sin ψ sin φ þ cos ψ cos φ cos θ, ð1:31Þ
α23 ¼ cos ðyd
1 , z Þ ¼  cos ψ sin θ,
α31 ¼ cos ðzd
1 , x Þ ¼ sin φ sin θ,
α32 ¼ cos ðzd
1 , y Þ ¼ cos φ sin θ,
α33 ¼ cos ðzd
1 , z Þ ¼ cos θ:

The mutual orientation of the coordinate systems Oxyz and Ox1y1z1 is deter-
mined with the help of the matrix Q of directional cosines given by the table

x y z
x1 α11 α12 α13
: ð1:32Þ
y1 α21 α22 α23
z1 α31 α32 α33

Its elements αij are expressed in terms of the Euler angles by formulas (1.31).

1.6 Equations of Motion of a Heavy Rigid Body About

a Fixed Point

Consider the motion of a rigid body about a fixed point under the action of gravi-
tational force P. Let us direct upward vertically the Oz1 axis of the fixed coordinate
system (Fig. 1.5).
Let us direct the axes of the moving coordinate system Oxyz along the principal
axes of inertia of the body for the fixed point O. Denote by xC, yC, zC the coordinates
of the center of gravity C in the coordinate system Oxyz. The body orientation with
respect to the fixed coordinate system is determined with the help of the
Euler angles (Fig. 1.5).
Denote by γ 1, γ 2, γ 3 the projections of the unit vector k1 of the vertical axis Oz1
on the moving axes. We have

k1 ¼ γ 1 i þ γ 2 j þ γ 3 k,

where i, j, k are unit vectors of the moving axes connected with the body. By virtue
of equations (1.31), we obtain
1.6 Equations of Motion of a Heavy Rigid Body About a Fixed Point 11

Fig. 1.5 Coordinate

systems for heavy
rigid body

γ1 ¼ α31 ¼ sin φ sin θ,

γ2 ¼ α32 ¼ cos φ sin θ, ð1:33Þ
γ3 ¼ α33 ¼ cos θ:

Vector k1 is constant in the fixed coordinate system; therefore, its absolute deri-
vative equals zero: dk dt ¼ 0. Taking into account the connection between the
1

absolute and local derivatives of the vector, we have

e 1
dk
þ ω  k1 ¼ 0:
dt

Hence,
 
 i j k 
e 1
dk 
¼ ω  k1 ¼  p q r : ð1:34Þ
dt γ
1 γ2 γ3 

Projecting both sides of this equality on the moving axes, we arrive at the equa-
tions called the Poisson equations:

dγ 1 dγ 2 dγ 3
¼ rγ 2  qγ 3 , ¼ pγ 3  rγ 1 , ¼ qγ 1  pγ 2 : ð1:35Þ
dt dt dt

Let us derive the dynamic Euler’s equations for this problem. The external forces
acting on the body are the force of gravity and the reaction of the fixed point
O which does not create any torque with respect to this point. The force of gravity
P is directed opposite to the z1axis, thus P ¼  Pk1. The gravitational torque rela-
tive to the point O equals
12 1 The Foundations of Dynamics of a Rigid Body with a Fixed Point

 
 i j k 

LO ¼ rC  P ¼ PrC  k1 ¼ P γ 1 γ2 γ 3 :
 xC yC zC 

Thus, the dynamic Euler’s equations (1.27) for this case have the following form

A1 p_ þ ðA3  A2 Þqr ¼ Pðγ 2 zC  γ 3 yC Þ,

A2 q_ þ ðA1  A3 Þrp ¼ Pðγ 3 xC  γ 1 zC Þ, ð1:36Þ
A3 r_ þ ðA2  A1 Þpq ¼ Pðγ 1 yC  γ 2 xC Þ:

Equations (1.35) and (1.36) form a system of six nonlinear differential equations
of the first order which describes the motion of a heavy rigid body about a
fixed point.
If p, q, r, γ 1, γ 2, γ 3 are obtained from systems (1.35) and (1.36) as functions of
time, then functions θ(t), φ(t) can be found from (1.33), whereas to determine ψ(t),
one has to use the first of the Euler equations (1.29).
Thus, the basic problem is to integrate the system of equations (1.35) and (1.36).

References

1. Bukhol’ts, N.N.: Fundamental Course of Theoretical Mechanics, vol. 2. Nauka, Moscow (1969)
in Russian
2. Goldstein, H.: Classical Mechanics. Addison-Wesley, Cambridge, MA (1950)
3. Loytsyansky, L.G., Lur’e, A.I.: Course in Theoretical Mechanics, vols. 1, 2. Nauka, Moscow
(1982) in Russian
4. Markeev, A.P.: Theoretical Mechanics. Nauka, Moscow (1990) in Russian
5. Lur’e, A.I.: Analytical Mechanics. Fizmatgiz, Moscow (1961) in Russian
Chapter 2
Motion of a Rigid Body by Inertia. Euler’s
Case

2.1 First Integrals

In Euler’s case, the principal moment of external forces acting on a rigid body
relative to a fixed point is equal to zero L0e ¼ 0. The dynamic Euler’s equations
(1.27) assume the form:

A1 p_ þ ðA3  A2 Þqr ¼ 0,
A2 q_ þ ðA1  A3 Þrp ¼ 0, ð2:1Þ
A3 r_ þ ðA2  A1 Þpq ¼ 0:

In this chapter, we present information about this case of motion of a rigid body
which will be needed later [1–7].
Since the principal moment of external forces L0e relative to the fixed point
O equals zero, it follows from the theorem of change of angular momentum (see
(1.21)) that:

G ¼ const, ð2:2Þ

i.e., the angular momentum of the body relative to the point O is invariable in the
fixed coordinate system Ox1y1z1.
Let us obtain the first integral (2.2) in a different way. For this purpose, we
multiply the motion equations (2.1), respectively, by A1p, A2q, A3r and add them
term by term; we get:

dp dq dr
A21 p þ A22 q þ A23 r ¼ 0:
dt dt dt

Integrating, we have:

© Springer International Publishing AG 2017 13

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_2
14 2 Motion of a Rigid Body by Inertia. Euler’s Case

G2 ¼ A21 p2 þ A22 q2 þ A23 r 2 ¼ const: ð2:3Þ

Here, A1p, A2q, A3r are the projections of vector G on the principal axes of
inertia of the body Ox, Oy, Oz, whereas G2 is the squared modulus of vector G.
Before obtaining another first integral of equations (2.1), let us recall the notions
of kinetic energy of a mechanical system and kinetic energy of a rigid body.
The kinetic energy of a mechanical system is the quantity T defined by the
formula:

1X n
T¼ mi v2i : ð2:4Þ
2 i¼1

Here, mi is the mass of the i-th point, vi, its velocity.

For a rigid body, in the case of continuous distribution of mass in the volume τ,
the expression for the kinetic energy is as follows:
ððð ððð
1 1
T¼ ρv2 dτ ¼ ρv2 dx dy dz, ð2:5Þ
2 2

where ρ is the density, v is the velocity vector of the infinitesimal element dm of the
body, and dτ is the volume element. The integration in (2.5) is performed over the
entire volume of the body.
If the rigid body moves about a fixed point O, the origin of the coordinate system
Oxyz rigidly connected with the body, then the kinetic energy of the body is
determined by the equality:

1 2 
T¼ J x p þ J y q2 þ J z r 2  J xy pq  J xz pr  J yz qr: ð2:6Þ
2

Here Jx, Jy, Jz and Jxy, Jxz, Jyz are axial and centrifugal moments of inertia of the
body for the point O. Equality (2.6) can be represented in the form:

1
T ¼ Jω  ω,
2

where J is the tensor of inertia of the body for the point O (see (1.10)).
If axes Ox, Oy, Oz are the principal axes of inertia of the body for the point O,
then equality (2.6) assumes the form:

1 
T¼ A1 p2 þ A2 q2 þ A3 r 2 , ð2:7Þ
2

where A1, A2, A3 are the principal moments of inertia for this point.
Theorem of change of kinetic energy of a mechanical system can be written in
the differential form as [4, 5]:
2.1 First Integrals 15

X
n X
n
dT ¼ Fie dri þ Fii dri : ð2:8Þ
i¼1 i¼1

The differential of the kinetic energy of the system is equal to the sum of
elementary works of all external and internal forces acting on the system denoted
by the upper indices e and i, respectively.
The equality:

X
n X
n
T  T0 ¼ Aie þ Aii ð2:9Þ
i¼1 i¼1

expresses the theorem of change of kinetic energy in the integral form: the change
of the kinetic energy of the system during its displacement from the initial position
to the given one equals the sum of works on this displacement done by all external
and internal forces applied to the system.
For a rigid body, the work of internal forces equals zero, and equality (2.8)
assumes the form:

X
n
dT ¼ Fie dri : ð2:10Þ
i¼1

The work of external forces applied to the rigid body is calculated by the
formula:

X
n
Fie dri ¼ Re vO dt þ L0e ω dt, ð2:11Þ
i¼1

where Re is the principal vector and L0e is the principal moment of external forces
with respect to the pole O, vO is the pole velocity, ω is the angular velocity of the
body, and dt is the infinitesimal interval of time corresponding to infinitesimal
displacements of the points of the system on which the work is calculated.
For the motion of a rigid body in Euler’s case, we have:

dT ¼ Re vO dt þ L0e ω dt ¼ 0,

since vO ¼ 0, L0e ¼ 0.
Thus, there is one more first integral of equations (2.1):

1 
T¼ A1 p2 þ A2 q2 þ A3 r 2 ¼ const: ð2:12Þ
2

This first integral can also be obtained by multiplying equations (2.1), respec-
tively, by p, q, r and adding them term by term. We have:
16 2 Motion of a Rigid Body by Inertia. Euler’s Case

dp dq dr
A1 p þ A2 q þ A3 r ¼ 0:
dt dt dt

Integrating and denoting by h an arbitrary constant, we find:

A1 p2 þ A2 q2 þ A3 r 2 ¼ h: ð2:13Þ

This is the energy integral, because the left-hand side of equality (2.13) is the
doubled kinetic energy 2T.

2.2 Basic Formulas for the Jacobian Elliptic Functions

We present below some information from the theory of elliptic functions necessary
for further considerations.
The integral:

ðφ
dx
u ¼ Fðφ; kÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2:14Þ
1  k2 sin 2 x
0

is called the elliptic integral of the first kind [8–10]. The number k is called the
modulus of the elliptic integral and satisfies the inequalities 0  k < 1.
In the particular case φ ¼ π2, this integral is called the complete elliptic integral of
the first kind:

π  ð
π=2
dx
K ðk Þ ¼ F ; k ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð2:15Þ
2 1  k2 sin 2 x
0

This integral can be expanded in a power series of k (for k2  1):

 
π 1 2
K ðk Þ ¼ 1 þ k þ ... : ð2:16Þ
2 4

The integral u (2.14) is a function of the upper limit φ. The inverse function φ of
u is called the amplitude and denoted as:

φ ¼ am u: ð2:17Þ

Jacobi introduced the functions:

2.2 Basic Formulas for the Jacobian Elliptic Functions 17

sn u ¼ sn ðu; kÞ ¼ sin φ ¼ sin am u, cn u ¼ cn ðu; kÞ ¼ cos φffi ¼ cos am u,

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dφ
dn u ¼ dn ðu; kÞ ¼ 1  k sin φ ¼ Δφ ¼
2 2 ¼ 1  k2 sn2 ðu; kÞ,
du
ð2:18Þ

which are, respectively, called elliptic sine, elliptic cosine, and the delta amplitude.
The functions sn u, cn u have real period equal to 4K(k), whereas the function dn
u, real period 2K(k).
If the modulus k ¼ 0, then the elliptic functions degenerate into sinu and cosu.
Note that sn 0 ¼ 0, cn 0 ¼ 1, and dn 0 ¼ 1.
There are the following relations between the functions sn u, cn u, dn u:

sn2 u þ cn2 u ¼ 1, dn2 u þ k2 sn2 u ¼ 1: ð2:19Þ

The differentiation formulas for elliptic functions are written as follows:

d
sn u ¼ cn u dn u,
du
d
cn u ¼ sn u dn u, ð2:20Þ
du
d
dn u ¼ k2 sn u cn u:
du

The integral:

ðφ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Eðφ; kÞ ¼ 1  k2 sin 2 x dx ð2:21Þ
0

is called the elliptic integral of the second kind [8–10].

The complete elliptic integral of the second kind in the case φ ¼ π2 is defined as
an integral of the form:

π  ð pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
π=2

EðkÞ ¼ E ; k ¼ 1  k2 sin 2 x dx: ð2:22Þ

2
0

This integral can be expanded in a power series with respect to k (for small k2):
 
π 1 2
Eð k Þ ¼ 1  k þ ... : ð2:23Þ
2 4
18 2 Motion of a Rigid Body by Inertia. Euler’s Case

2.3 Integration of Dynamic Euler’s Equations: Analysis

of the Motion

Let us obtain an analytical solution of equations (2.1). We assume for definiteness

that A1 > A2 > A3.
Figure 2.1 presents the trajectories of vector G in Euler’s case. The arrows
indicate the direction of increasing time, and the letters A1, A2, A3 denote the
principal central axes with the corresponding moments of inertia. The trajectories
of the vector are closed, whereas the steady-state motions (singular points) are the
rotations about the principal central axes of inertia. The rotations about the axes
with the moments of inertia A1 and A3 are stable (singular points of center type),
whereas the ones about the intermediate axis with the moment of inertia A2 are
unstable (singular point of saddle type, A1 > A2 > A3). Separatrices demarcate the
trajectories surrounding the axis A1, for which the inequality 2TA2 < G2 holds, from
the trajectories surrounding the axis A3 (for which 2TA2 > G2). Let us also note that
there holds the inequalities 2TA1  G2  2TA3, which follow from (2.3), (2.12) for
A1 > A2 > A3 and which turn into equalities for the rotations of the body about the
axes with the moments of inertia A1, A3.
Express the quantities p2 and r2 by q2, A1, A2, A3 and the constants T, G from the
first integrals (2.3), (2.12):

p2 ¼ 2TA3  G2  A2 ðA3  A2 Þq2 ,

A 1 ð A3  A1 Þ
1 2
ð2:24Þ
r2 ¼ G  2TA1  A2 ðA2  A1 Þq2 :
A 3 ð A3  A1 Þ

Substituting the obtained values of p and r into the second of equations (2.1), we
get:

dq 1 

¼  pffiffiffiffiffiffiffiffiffiffi 2TA3  G2  A2 ðA3  A2 Þq2 

dt A2 A1 A3


 G2  2TA1  A2 ðA2  A1 Þq2 1=2 : ð2:25Þ

If this equation with separable variables is integrated, then p and q are deter-
mined from expressions (2.24).
Consider three cases corresponding to various relations between T and G.
1. Let the inequalities 2TA2 > G2  2TA3 be fulfilled which correspond to the
trajectories (see Fig. 2.1) surrounding axis A3. In this case, the value of r is always
different from zero. To integrate equation (2.25), we introduce new variables:
2.3 Integration of Dynamic Euler’s Equations: Analysis of the Motion 19

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
G2  2TA3 ðA2  A3 Þ 2TA1  G2
q¼ sin λ, τ¼ ðt  t0 Þ,
A2 ðA2  A3 Þ A 1 A2 A3

where t0 is an arbitrary constant, as well as a positive parameter k2 < 1 according to

the formula:
 
ðA1  A2 Þ G2  2TA3
k ¼ 2
 : ð2:26Þ
ðA2  A3 Þ 2TA1  G2

Equation (2.25) has the following form in the new variables:

dλ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ 1  k2 sin 2 λ: ð2:27Þ
dt

Let q ¼ 0 at t ¼ 0. Then, according to Sect. 2.2, we find from (2.27) λ ¼ am τ. In

this case, the solution of Euler’s equations (2.1) is expressed through the Jacobian
elliptic functions as follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
G2  2TA3
p ¼ cn τ,
A1 ð A 1  A3 Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
G2  2TA3
q ¼ sn τ, ð2:28Þ
A2 ð A 2  A3 Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2TA1  G2
r ¼ dn τ:
A3 ð A1  A3 Þ

Either both upper or both lower signs are taken here.

2. Consider the motion under the condition 2TA1  G2 > 2TA2 that corresponds
to the trajectories in Fig. 2.1 encompassing axis A1. In this case, the quantity p is
different from zero during the entire motion.

Fig. 2.1 Trajectories of

angular momentum vector
in Euler’s case
20 2 Motion of a Rigid Body by Inertia. Euler’s Case

The change of variables has the form:

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
2TA1  G2 ðA1  A2 Þ G2  2TA3
q¼ sin λ, τ¼ ðt  t0 Þ:
A2 ð A 1  A2 Þ A1 A2 A 3

Let us introduce the notation for the squared modulus k2of elliptic functions:
 
ðA2  A3 Þ 2TA1  G2
k ¼ 2
 : ð2:29Þ
ðA1  A2 Þ G2  2TA3

Then equation (2.25) assumes the form (2.27), and, under the condition q ¼ 0 for
t ¼ t0, the solution of equations (2.1) is written in the form:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
G2  2TA3
p ¼ dn τ,
A ð A1  A3 Þ
s1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2TA1  G2
q ¼ sn τ, ð2:30Þ
A 2 ð A1  A2 Þ
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi
2TA1  G2
r ¼ cn τ:
A 3 ð A1  A3 Þ

3. If G2 ¼ 2TA2, then equalities (2.24) have the form:

ð A2  A 3 Þ   ðA1  A2 Þ  
p2 ¼ 2T  A2 q2 , r2 ¼ 2T  A2 q2 : ð2:31Þ
A1 ð A 1  A3 Þ A3 ð A1  A3 Þ

It follows that A1(A1  A2)p2 ¼ A3(A2  A3)r2.

If we set:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2T ðA1  A2 ÞðA2  A3 Þ
τ¼ ðt  t0 Þ,
A1 A2 A 3

then in this case equation (2.25) assumes the form:

dq 1  
¼ pffiffiffiffiffiffiffiffiffiffi 2T  A2 q2 : ð2:32Þ
dτ 2TA2

We take the plus sign in the right-hand side of (2.32). Let q ¼ 0 at t ¼ t0. Then
from (2.32) and (2.31), we find the solution of Euler’s equations (2.1) expressed
through the hyperbolic functions:
2.3 Integration of Dynamic Euler’s Equations: Analysis of the Motion 21

Fig. 2.2 Coordinate

systems associated with
angular momentum vector

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2T ðA2  A3 Þ 1 2T 2T ðA1  A2 Þ 1
p¼ , q¼ thτ, r¼ : ð2:33Þ
A1 ðA1  A3 Þ chτ A2 A3 ðA1  A3 Þ chτ

In this case, the trajectories of motion of vector G on the unit sphere are
separatrices which separate the periodic motions encircling axes A1 and A3.
If the quantities p, q, r are defined as functions of time, then the Euler angles ψ,
θ, φ can be found from the kinematic Euler’s equations (1.28). However, the
calculations can be simplified, if we make use of the constancy of the angular
momentum vector G and choose a fixed coordinate system Ox1y1z1 (Fig. 2.2) in
such a way that the direction of axis Oz1 coincides with the invariable direction of
vector G.
With this selection of the coordinate system, the projections A1p, A2q, A3r of
vector G on the axes of the connected with the body coordinate system Ox, Oy, Oz
formed by the principal axes of inertia are calculated by the formulas according to
Fig. 2.2:

A1 p ¼ G sin θ sin φ, A2 q ¼ G sin θ cos φ, A3 r ¼ G cos θ: ð2:34Þ

From here we can immediately determine two Euler angles:

A3 r A1 p
cos θ ¼ , φ¼ : ð2:35Þ
G A2 q

To find angle ψ, we make use of expression (1.29) for ψ_ :

1
ψ_ ¼ ðp sin φ þ q cos φÞ:
sin θ

If we substitute into this expression the quantities sinφ and cosφ obtained from
the first two equalities (2.34) and use the third equality of (2.34), we get:
22 2 Motion of a Rigid Body by Inertia. Euler’s Case

2T  A3 r 2
ψ_ ¼ G : ð2:36Þ
G2  A23 r 2

If the dynamic Euler’s equations are integrated, then the right-hand sides of
equalities (2.35), (2.36) are known functions of time t.
The motions of the body different from the permanent rotations and motions
corresponding to separatrices will be two-frequency motions. One frequency is the
frequency ω1 of periodic oscillations of angles θ and φ determined from (2.35),
(2.28) or from (2.35), (2.30). The second frequency ω2 is the time-average value of
the right-hand side of equation (2.36).
If frequencies ω1 and ω2 are commensurable (the ratio ω1/ω2 is a rational
number), then a resonance case takes place. In this case, the motion of the body
in space is periodic. If frequencies ω1 and ω2 are incommensurable (nonresonance
case), then the motion of the body in space is conditionally periodic.
Let us note a property of the constructed solution which will be used later (see
also [7]). For all motions, except for the motions along the separatrices, variables p,
q, r are periodic with respect to t with the period τ depending on constants G and T.
The angles θ and φ defined by relations (2.35) are also either periodic for these
motions with the period τ(G, T ) or increase by 2π during the period. The angle ψ
determined from equation (2.36) possesses the following property for the consid-
ered motions: it can be represented in the form:

2π ðt þ t0 Þ
ψ ðtÞ ¼ ψ 1 ðtÞ þ ψ 2 ðtÞ, ψ2 ¼ , t0 ¼ const,
τ0

where τ0 is the period, whereas the functions ψ 1(t) have the same periodicity
properties with period τ as the functions θ(t) and φ(t). We suppose below that the
periods τ0 ¼ τ0 (G, T ) and τ are incommensurable.

2.4 Particular Cases (Regular Precession, Permanent

Rotations)

In the particular case when the inertia ellipsoid relative to a fixed point O is an
ellipsoid of rotation, the integration of the motion equations in Euler’s case can be
performed in elementary functions. In this case, the body is called dynamically
symmetric, and two of its principal moments of inertia for the point O are equal:
A1 ¼ A2. In this case, dynamic Euler’s equations (2.1) assume the form:
2.4 Particular Cases (Regular Precession, Permanent Rotations) 23

A1 p_ þ ðA3  A1 Þqr ¼ 0,
A1 q_ þ ðA1  A3 Þrp ¼ 0, ð2:37Þ
A3 r_ ¼ 0:

It follows from the last equation of system (2.37) that:

r ¼ r 0 ¼ const, ð2:38Þ

i.e., the projection of angular velocity of the body on its axis of dynamic symmetry
is constant.
The solution of the first two equations of system (2.37) has the form:

p ¼ p0 cos λt þ q0 sin λt,

ð2:39Þ
q ¼ p0 sin λt þ q0 cos λt:

Here λ ¼ A1AA1
3
r 0 , whereas the subscript 0 indicates the initial values of the
corresponding quantities.
We select the fixed coordinate system Ox1y1z1 as in Sect. 2.3, having directed
axis Oz1 along vector G which is constant in Euler’s case (Fig. 2.2). Then the
projections of angular momentum on the moving axes Oxyz are determined by
equalities (2.34).
It follows from the last equation of (2.34) that:

A3 r 0
cos θ ¼ ¼ const, ð2:40Þ
G

i.e., the angle of nutation is constant: θ ¼ θ0 ¼ const.

The kinematic Euler’s equations (1.28) of the first chapter yield for θ ¼ θ0,
r ¼ r 0:

p ¼ ψ_ sin θ0 sin φ, q ¼ ψ_ sin θ0 cos φ, r ¼ ψ_ cos θ0 þ φ_ : ð2:41Þ

Substituting the expression for p from (2.41) into the first of equations (2.34), we
get:

G
ψ_ ¼ ¼ ω2 ¼ const: ð2:42Þ
A1

The quantity ω2 is called the angular velocity of precession.

Using formulas (2.40), (2.42) and the last equality of (2.41), we find φ_ :

G A3 A1  A3
φ_ ¼ r 0  ψ_ cos θ0 ¼ r 0  cos θ0 ¼ r 0  r 0 ¼ r0 ¼
A1 A1 A1
¼ λ ¼ ω1 ¼ const, ð2:43Þ
24 2 Motion of a Rigid Body by Inertia. Euler’s Case

where λ is introduced in formulas (2.39), whereas ω1 is called the angular velocity

of proper rotation.
The motion of a rigid body about a fixed point, consisting of its rotation about an
axis connected with the body and the rotation of its axis about an axis which
intersects it and is immovable in the considered reference system, is called preces-
sion. The precession is called regular, if the rotation of the body about the axis
connected with it and the rotation of the axis itself have angular velocities with
constant magnitudes.
Thus, in Euler’s case, the dynamically symmetric body performs regular pre-
cession. Moreover, the symmetry axis of the body describes a circular cone with
axis G, whereas the rotation of the symmetry axis about G takes place with a
constant angular velocity ω2. Simultaneously, the body rotates about the symmetry
axis with a constant angular velocity ω1.
Equations (2.1) admit partial solutions corresponding to rotations of the rigid
body with a constant angular velocity about an axis passing through the fixed point
O. Such motions are called permanent rotations. In this case, the quantities p, q,
r are constant. Then it follows from (2.1) that:

ðA3  A2 Þqr ¼ 0, ðA1  A3 Þrp ¼ 0, ðA2  A1 Þpq ¼ 0: ð2:44Þ

In the case of different principal moments of inertia, equalities (2.44) are

possible, only if two of the three quantities p, q, r equal zero. Thus, in the general
case, permanent rotation of the body can occur only about one of the principal axes
of inertia of the body for the point O, while the magnitude of the body angular
velocity can be an arbitrary constant.
For a dynamically symmetric body (A1 ¼ A2 6¼ A3), equalities (2.44) are fulfilled
either for p ¼ q ¼ 0 or for r ¼ 0. In this case, permanent rotations are possible either
about the axis of dynamic symmetry ( p ¼ q ¼ 0, r ¼ r0) or about any of the axes
orthogonal to the axis of dynamic symmetry.
On the other hand, if all three principal moments of inertia are equal to one
another (A1 ¼ A2 ¼ A3), then equations (2.44) are satisfied for any invariable p ¼ p0,
q ¼ q0, r ¼ r0. In this case, the body uniformly rotates about some axis fixed in the
space, the direction of which is determined by initial data and can be taken
arbitrarily. Under this condition (A1 ¼ A2 ¼ A3), the ellipsoid of inertia for the
point O is a sphere; thus, any axis passing through the point O is a principal axis
of inertia of the body.

References

1. Magnus, K.: Kreisel. Theorie und Anwendungen. Springer, Berlin (1971)

2. Appel, P.: Traite de Mechanique Rationnelle. Gauthier-Villars, Paris (1953)
3. Grammel, R.: Der Kreisel. Seine Theorie und Seine Anwendungen. Erster Band. Izd-vo Inostr.
Lit-ry, Moscow (1952) Russian translation from German
References 25

4. Bukhol’ts, N.N.: Fundamental Course of Theoretical Mechanics, vol. 2. Nauka, Moscow

(1969) in Russian
5. Markeev, A.P.: Theoretical Mechanics. Nauka, Moscow (1990) in Russian
6. Zhukovsky, N.E.: Mechanics of a System. Rigid Body Dynamics. Oborongiz, Moscow (1939)
in Russian
7. Landau, L.D., Lifshitz, E.M.: Course of Theoretical Physics, Mechanics, vol. 1. Pergamon
Press, Oxford (1976)
8. Zhuravsky, A.M.: Handbook of Elliptical Functions. Academy of Science Press, Moscow
(1941) in Russian
9. Jahnke, E., Emde, F., Losch, F.: Tables of Higher Functions. McGraw-Hill, New York, NY
(1960)
10. Gradshtein, I.S., Ryzhik, I.M.: Tables of Integrals, Sums, Series and Products. Academic
Press, San Diego, CA (2000)
Chapter 3
Lagrange’s Case

3.1 Integration of the Equations of Motion and Analysis

of Motion

Suppose that the ellipsoid of inertia of a rigid body relative to a fixed point O is an
ellipsoid of revolution, i.e., A1 ¼ A2, whereas the center of gravity of the body lies
on the axis of dynamic symmetry of the body. Let us introduce the moving and
fixed systems of coordinates Oxyz and Ox1y1z1 in the following way (Fig. 3.1):
A common origin of these two systems of coordinates is chosen at the fixed point
O. The axis Oz1 of the fixed system of coordinates is vertical. The axis Oz of the
moving coordinate system, invariably connected with the body, is directed along
the axis of dynamic symmetry. Denote the radius vector of the center of gravity
C with respect to the point O by l(0, 0, l ), whereas the direction cosines of the
vertical axis Oz1 relative to the moving axes, by γ 1, γ 2, γ 3. Their expressions in
terms of the Euler angles are determined by formulas (1.33). The projections of
the gravity force of the body P ¼ mg on the moving axes have the form P(mgγ 1,
mgγ 2, mgγ 3). Let us draw up the dynamic Euler’s equations. The moment of the
gravity force relative to the fixed point O equals




i j k




L ¼ l  P ¼ P

0 0 l



γ γ γ

1 2 3

or

Lx ¼ mglγ 2 ¼ mgl sin θ cos φ,

Ly ¼ mglγ 1 ¼ mgl sin θ sin φ, ð3:1Þ
Lz ¼ 0:

© Springer International Publishing AG 2017 27

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_3
28 3 Lagrange’s Case

Fig. 3.1 Heavy rigid body

The dynamic Euler’s equations for the considered Lagrange’s case assume the
form

A1 p_ þ ðA3  A1 Þqr ¼ mgl sin θ cos φ,

A1 q_ þ ðA1  A3 Þpr ¼ mgl sin θ sin φ, ð3:2Þ
A3 r_ ¼ 0:

We add to these equations the kinematic Euler’s equations (1.28):

p ¼ ψ_ sin θ sin φ þ θ_ cos φ,

q ¼ ψ_ sin θ cos φ  θ_ sin φ, ð3:3Þ
r ¼ φ_ þ ψ_ cos θ:

Let us find the first integrals of motion. The unit vector ξ0 directed along the
fixed vertical axis Oz1 can be expressed in terms through the projections on the
moving axes as follows:

ξ0 ¼ γ 1 i þ γ 2 j þ γ 3 k: ð3:4Þ

Squaring both sides of equation (3.4), we obtain a trivial integral in the form

γ 21 þ γ 22 þ γ 23 ¼ 1: ð3:5Þ

To obtain another integral, we apply the theorem of change of angular momen-

tum with respect to the axis Oz1. Since the gravity torque relative to this axis equals
zero, we have

dG1
¼ 0:
dt

It follows that G1 ¼ Gz1 ¼ const.

For the considered case, the angular momentum in terms of projections on the
moving axes is determined by the formula
3.1 Integration of the Equations of Motion and Analysis of Motion 29

G ¼ A1 pi þ A1 qj þ A3 rk: ð3:6Þ

Multiplying scalarly the vectors (3.6) and (3.4), we find the first integral, the
projection of the angular momentum vector on the vertical Oz1:

G1 ¼ G  ξ0 ¼ A1 ðpγ 1 þ qγ 2 Þ þ A3 rγ 3 ¼
¼ A1 sin θðp sin φ þ q cos φÞ þ A3 r cos θ ¼ C1 ¼ const: ð3:7Þ

Yet another first integral will be constructed from the theorem of the total energy
conservation. The potential energy of the body is equal to

Π ¼ mgz1C , ð3:8Þ

where z1C ¼ lγ 3 ¼ l cos θ is the coordinate of the center of gravity C in the

fixed reference system.
Taking into account the expression for the kinetic energy

1  2  
T¼ A1 p þ q 2 þ A3 r 2 ,
2

we write the energy integral in the form

1  2  
H ¼TþΠ¼ A1 p þ q2 þ A3 r 2 þ mgl cos θ ¼ C2 ¼ const: ð3:9Þ
2

Here, H is the total energy of the body.

The last equation of system (3.2) shows that

r ¼ C3 ¼ const: ð3:10Þ

Consider some information on the rigid body motion in Lagrange’s case which
will be needed in the sequel [1–10]. Using the kinematic Euler’s equations (1.28),
the first integral (3.7) can be represented in the form
 
G1 ¼ A1 ψ_ sin θ sin φ þ θ_ cos φ sin θ sin φþ
  
þ ψ_ sin θ cos φ  θ_ sin φ sin θ cos φ þ A3 r cos θ ¼ C1 :

After simplifications we get

G1 ¼ A1 ψ_ sin 2 θ þ A3 r cos θ ¼ C1 : ð3:11Þ

The energy integral (3.9) can be expressed after similar substitutions in the
following way:
30 3 Lagrange’s Case

1  2 2  
H¼ A1 ψ_ sin θ þ θ_ 2 þ A3 r 2 þ mgl cos θ ¼ C2 : ð3:12Þ
2

Engaging kinematic Euler’s equations, integral (3.10) can be written as

r ¼ ψ_ cos θ þ φ_ ¼ C3 : ð3:13Þ

Denote

cos θ ¼ u ð3:14Þ

and express ψ_ and φ_ as functions of u. From relation (3.11) we find in view of (3.14)

G1  A3 ru
ψ_ ¼ : ð3:15Þ
A1 ð1  u2 Þ

From equality (3.13) we get

ðG1  A3 ruÞu
φ_ ¼ r  : ð3:16Þ
A1 ð 1  u 2 Þ

Expression (3.12) assumes the form

1   ðG1  A3 ruÞ2
θ_ 2 ¼ 2H  A3 r 2  2mglu  2 : ð3:17Þ
A1 A1 ð 1  u 2 Þ

We have on the basis of (3.14)

u_
θ ¼ arccosu, θ_ ¼  pffiffiffiffiffiffiffiffiffiffiffiffiffi :
1  u2

Thus, equation (3.17) can be written as follows:

1 h    i
u_ 2 ¼ QðuÞ ¼ 2
A1 2H  A3 r 2  2mglu 1  u2  ðG1  A3 ruÞ2 : ð3:18Þ
A1

For any motion of the body, there hold the conditions u ¼ cos θ 2 [1, 1] and
u_ 2 ¼ QðuÞ  0. Consequently, there exists a value u0 2 [1, 1], for which Q(u0) >
0. Besides, by virtue of (3.18) we have

Qð1Þ < 0, Qð1Þ < 0, Qðþ1Þ < 0, Qðþ1Þ > 0: ð3:19Þ

Therefore, the cubic polynomial Q(u) has three real roots u1, u2, u3 in the
intervals
3.1 Integration of the Equations of Motion and Analysis of Motion 31

1  u1  u0  u2  þ1 < u3 < þ1: ð3:20Þ

Represent the polynomial Q(u) (3.18) in the form

2mgl
QðuÞ ¼ ðu  u1 Þðu  u2 Þðu  u3 Þ: ð3:21Þ
A1

It follows from (3.18) and (3.21) that in order for the derivative u_ to have a real
value, the quantity u must satisfy the inequality

u1  u  u2 :

Introduce a new variable γ by the relation

u ¼ u1 cos 2 γ þ u2 sin 2 γ ¼ u1 þ ðu2  u1 Þsin 2 γ: ð3:22Þ

We find

u  u1 ¼ ðu2  u1 Þsin 2 γ,
u  u2 ¼ ðu2  u1 Þcos
 γ,
2
ð3:23Þ
u2  u1
u  u3 ¼  ð u3  u1 Þ 1  sin γ :
2
u3  u1

Introduce the notation

u2  u1
k2 ¼ , ð3:24Þ
u3  u1

where by virtue of (3.20) we have 0  k2 < 1. Differentiating relation (3.22), we

obtain

u_ ¼ 2ðu2  u1 Þγ_ sin γ cos γ: ð3:25Þ

Substituting (3.23) and (3.25) into equality u_ 2 ¼ QðuÞ and using relation (3.21),
we find
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
γ_ ¼ α 1  k2 sin 2 γ , α ¼ ½mglðu3  u1 Þ=ð2A1 Þ1=2 : ð3:26Þ

When u increases from u1 to u2, the sign in front of the square root will be
positive; when it decreases from u2 to u1, negative. When γ changes from 0 to π/2,
the sign at the square root in the right-hand side of (3.26) is positive, whereas in the
increase of γ from π/2 to π, negative.
For the time interval corresponding to the change of γ from 0 to π/2, we find
32 3 Lagrange’s Case

ðγ
dγ
αt þ β ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð3:27Þ
1  k2 sin 2 γ
0

Here β is the constant of integration. Using the properties of elliptic functions

(see Sect. 2.2), we find with the help of the Jacobi amplitude

γ ¼ amðαt þ βÞ: ð3:28Þ

It follows, in view of (3.22), the solution of equation (3.18) is expressed in terms

of elliptic sine as

u ¼ cos θ ¼ u1 þ ðu2  u1 Þsn2 ðαt þ βÞ,

ð3:29Þ
snðαt þ βÞ ¼ sin amðαt þ β; kÞ:

The variables ψ and φ are obtained by quadratures from equations (3.15) and
(3.16).
The function u defined by relation (3.29) depends on four integration constants
G1, H, r, and β. The subsequent integration of equations (3.15) and (3.16) yields two
more arbitrary constants.
The general solution (the expressions for angular velocities, direction cosines, or
the Euler angles) in various forms is presented in [3, 10–13].
To visualize the properties of motion of a rigid body in Lagrange’s case, we use a
geometric method. Let us construct a unit sphere about the fixed point and consider
the curve which is described on the sphere surface by the end of the unit vector ON
directed along the axis of dynamic symmetry Oz (Fig. 3.1). The end of this vector is
called apex. The apex movement along the sphere represents the motion of the axis
Oz, i.e., the precession and nutation of gyroscope.
Since u ¼ cos θ, from the conclusion that for the motion u is within the limits
u1  u  u2, it follows that

θ 1  θ  θ2 , ð3:30Þ

where cosθ1 ¼ u1, cosθ2 ¼ u2.

The nutation angle θ, according to the law of its changing (3.29), is a
periodic function of time with the period

2K ðkÞ
T¼ ,
α

where K is the complete elliptic integral of the first kind. It also follows from (3.29)
that the increase of the angle θ from the value θ2 to θ1 and its decrease from the
value θ1 to θ2 occur monotonically during the time interval KαðkÞ. In this case, the
point N describes on the sphere a wavy line enclosed between two parallels formed
by the intersection of the sphere and the cones of angles 2θ2 and 2θ1 (Fig. 3.2).
3.1 Integration of the Equations of Motion and Analysis of Motion 33

Fig. 3.2 Unidirectional

precession

At regular time intervals KαðkÞ, this point alternatively falls on the upper and lower
parallels.
The shape of the apex trajectory between the limiting parallels depends on the
character of precession motion. The angular velocity of precession is determined by
formula (3.15). The denominator of expression (3.15) is a positive quantity; thus,
the precession direction depends on the value of the constant G1, i.e., on the
initial conditions. The following cases are possible:
1. For G1 > A3r cos θ2 (or G1 < A3r cos θ1), the inequalities cosθ1  cos θ  cos θ2
are fulfilled during the entire motion; in this case ψ_ has a constant sign and does
not turn to zero at θ ¼ θ1 or θ ¼ θ2 when θ_ ¼ 0. It follows that the curve traced by
the point N is tangent to the parallels depicted in Fig. 3.2.
2. If the equality G1 ¼ A3r cos θ2 is true, then at the value θ2 of the angle of nuta-
tion, not only the value of θ_ but also the value of ψ_ turns to zero. The apex
trajectory has the form of a spherical cycloid depicted in Fig. 3.3.
3. Let us analyze the case of fulfillment of the inequalities

A3 r cos θ1 < G1 < A3 r cos θ2 :

For the angles θ satisfying the condition cos θ1  cos θ < AG31r, the angular
velocity ψ_ of precession is positive, whereas for AG31r < cos θ  cos θ2 , it is negative.
Hence, at the angle θ∗ satisfying the equation cos θ∗ ¼ AG31r, the precession direction
changes. In this case, the apex trajectory will be a self-crossing curve (Fig. 3.4).
34 3 Lagrange’s Case

Fig. 3.3 Precession with

cusps

Fig. 3.4 Precession with

loops

3.2 Regular Precession

One particular type of the rigid body motion in Lagrange’s case is characterized by
a constant value of the angle of nutation θ ¼ θ0. Let us eliminate the variables p and
q in the motion equations (3.2) with the help of kinematic Euler’s equations (3.3).
Substituting into (3.2) and (3.3) r ¼ r0 и θ ¼ θ0 (θ_ ¼ 0), we find

A1 p_ þ ðA3  A1 Þqr 0 ¼ mgl sin θ0 cos φ,

A1 q_ þ ðA1  A3 Þpr 0 ¼ mgl sin θ0 sin φ,
p ¼ ψ_ sin θ0 sin φ,
q ¼ ψ_ sin θ0 cos φ:

After the elimination of p and q from these equations, we obtain

A1 ðψ€ sin φ þ ψ_ φ_ cos φÞ þ ðA3  A1 Þr 0 ψ_ cos φ ¼ mgl cos φ,

ð3:31Þ
A1 ðψ€ cos φ  ψ_ φ_ sin φÞ þ ðA1  A3 Þr 0 ψ_ sin φ ¼ mgl sin φ:
3.2 Regular Precession 35

Multiplying the first equation in (3.31) by sinφ and the second by cosφ and
adding them, we find

A1 ψ€ ¼ 0, ψ_ ¼ ψ_ 0 ¼ const: ð3:32Þ

It follows from the third equation in (3.3) in view of (3.10) and (3.32) that

φ_ ¼ r 0  ψ_ 0 cos θ0 ¼ φ_ 0 ¼ const: ð3:33Þ

Thus, if the condition θ ¼ θ0 holds, then we have constancy of the angular

velocities ψ_ and φ_ . This motion is called regular precession.
The substitution ψ€ ¼ 0 into the second equation in (3.31) yields the relation

A1 ψ_ φ_  ðA1  A3 Þr 0 ψ_ ¼ mgl:

Substituting (3.33) into it, we obtain a quadratic equation with respect to ψ_ :

A1 cos θ0 ψ_ 2  A3 r 0 ψ_ þ mgl ¼ 0, ð3:34Þ

the roots of which are

" sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi#
A3 r 0 4A1 mgl cos θ0
ψ_ 1, 2 ¼ 1 1 : ð3:35Þ
2A1 cos θ0 A23 r 20

The curves of change of these roots in their dependence on r0 for various incli-
nation angle θ0 are represented in [1, 14].
Note the following cases:
1. 0  θ0 < π2, cosθ0 > 0: this is a «standing» rigid body;
cos θ0
for r 20 < 4A1 mgl
A2
¼ r ∗2
0 there are no real roots (3.35);
3

for r 20 ¼ r ∗2
0 there are two equal roots, ψ_ 1 ¼ ψ_ 2 ;
for r 20 > r ∗2
0 there are two real roots of the same sign as r0.
2. θ0 ¼ π2, cosθ0 ¼ 0; in this case, the axis of the body is horizontal, and equation
(3.34) has a real root

mgl
ψ_ 2 ¼ : ð3:36Þ
A3 r 0

π
3. < θ0  π, cosθ0 < 0: this is a «hanging» rigid body; in this case, two real roots
2
have opposite signs.





According to the adopted notations, we have always
ψ_ 1

ψ_ 2
. Therefore, the
motion with the angular velocity ψ_ 1 is called fast precession; whereas the motion
with the angular velocity ψ_ 2 , slow precession.
36 3 Lagrange’s Case

The angular velocity of the fast regular precession for g ¼ 0 or l ¼ 0 coincides

with the angular velocity of precession of a symmetric rigid body which is free from
any torques. In fact, for l ¼ 0 the roots of equation (3.34) are
A3 r 0
ψ_ 1 ¼ , ψ_ 2 ¼ 0: ð3:37Þ
A1 cos θ0

Consider another approach to the definition of regular precession. Suppose that

equation Q(u) ¼ 0 has a multiple root u1 ¼ u2 ¼ u0. Then it follows from (3.29) that
u ¼ u1 ¼ u0, and the angle θ is constant: θ ¼ θ0. Note that in this case k2 ¼ 0, as it
follows from (3.24). According to formulas (3.15) and (3.16), the angular velocities
ψ_ and φ_ are also constant, and the body performs regular precession.

3.3 Fast Spinning Top

If the roots u1 and u2 of cubic polynomial (3.18) are not equal to each other, but
differ a little, then the parallels in Figs. 3.2–3.4 almost merge, whereas the point
N describes a spherical curve between them. In this case, the body performs the
so-called pseudoregular precession, i.e. a motion close to regular precession.
Let us derive the relations for pseudoregular precession. We present relation
(3.21) in the form

2mgl  3 
Q ð uÞ ¼ u þ a2 u2 þ a1 u þ a0 , ð3:38Þ
A1

where we introduce notations for the coefficients of cubic polynomial in terms of its
roots

a2 ¼ ðu1 þ u2 þ u3 Þ, a1 ¼ u1 u2 þ u1 u3 þ u2 u3 , a0 ¼ u1 u2 u3 : ð3:39Þ

On the other hand, from relations (3.18) for Q(u), we obtain the following
expressions for coefficients (3.39) in terms of constants m, g, l, A1, and A3 and
the integration constants r, H, and G1:

A3 ðA1  A3 Þr 2  2HA1
a2 ¼ ,
2mglA1
G 1 A3 r
a1 ¼  1, ð3:40Þ
mglA1
2HA1  A1 A3 r 2  G21
a0 ¼ :
2mglA1
3.3 Fast Spinning Top 37

Suppose that angular velocity r is sufficiently large, so that the following

condition holds

A3 r 2 mgl: ð3:41Þ

Under condition (3.41), one can introduce a dimensionless small parameter

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
ε¼ mgl=A3 r
1: ð3:42Þ

If the body performs fast rotation around the axis of symmetry, then the potential
energy of the body is small in comparison with its kinetic energy, and we have in
the first approximation

1
G1  A3 r, H  T  A3 r 2 : ð3:43Þ
2

From relation (3.42) we obtain

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
r¼ mgl=A3 ε , ð3:44Þ

whereas for constants G1 and H, according to (3.43) and (3.44), the following
expansions hold
pffiffiffiffiffiffiffiffiffiffiffiffiffi 1
G1 ¼ A3 mgl ε ð1 þ εg1 þ ε2 g2 þ . . .Þ,
mgl 2   ð3:45Þ
H¼ ε 1 þ εh1 þ ε2 h2 þ . . . :
2

Here gi and hi are yet unknown constants.

Substitute formula (3.44) and expansions (3.45) into relations (3.40) for coeffi-
cients a2, a1, a0. Performing expansions with respect to powers of the small para-
meter and keeping the first two terms of expansions, we get after transformations
and simplifications

a2 ¼ χε2  ε1 ðh1 =2Þ  h2 =2 þ OðεÞ,

a1 ¼ 2χε2 þ 2χε1 g1 þ 2χg2  1 þ OðεÞ, ð3:46Þ
a0 ¼ χε2 þ ε1 ðh1 =2  2χg1 Þ þ h2 =2  2χg2  χg21 þ OðεÞ:

Here we introduce the notation

χ ¼ A3 =ð2A1 Þ  1: ð3:47Þ

Substituting coefficients (3.46) into polynomial (3.38) and omitting small quan-
tities of higher orders, we represent it in the form
38 3 Lagrange’s Case


A1 QðuÞ 2 2 1 h1
¼ u  χε ðu  1Þ þ ε ðu  1Þ 2χg1  ðu þ 1Þ 
3
2mgl 2
h2  2 
 u  1 þ 2χg2 ðu  1Þ  u  χg21 : ð3:48Þ
2

Let us find the roots of polynomial (3.48) in the form of expansions in powers of
ε, taking into account inequalities (3.20). We are to find the largest root u3 in the
form
 
u3 ¼ ε2 u03 þ εu13 þ . . . :

Substituting this expansion into equation Q(u) ¼ 0, where Q(u) is defined by

formula (3.48), we get u03 ¼ χ. Thus, we have up to the higher order terms

u3 ¼ ε2 χ: ð3:49Þ

We represent two other roots u1 and u2 in the form

u ¼ 1  εx, ð3:50Þ

where x  0 by inequalities (3.20).

Substituting expression (3.50) into polynomial (3.48) and equating it to zero, we
obtain, up to small quantities of higher orders, the following quadratic equation for
the quantity x:

χx2 þ ð2χg1  h1 Þx þ χg21 ¼ 0: ð3:51Þ

The roots of equation (3.51) are represented in the form

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h1  2χg1 h1 ðh1  4χg1 Þ
x1 ¼ x0 þ Δ, x2 ¼ x0  Δ, x0 ¼ , Δ¼ : ð3:52Þ
2χ 2χ

Analyze the obtained roots. Note that equation (3.51) can be written in the form

χ ðx þ g1 Þ2 ¼ h1 x:

Since both roots are real and positive, it follows that h1 > 0, whereas the
expressions (3.52) for Δ and x0 yield h1 > 4χg1 and x0 > 0. Thus, the roots u1 and
u2 of polynomial Q(u) are defined in the form

u1 ¼ 1  εx1 , u2 ¼ 1  εx2 , u1 < u2 < 1, x2 < x0 < x1 , ð3:53Þ

where x0, x1, and x2 are given by relations (3.52).

3.3 Fast Spinning Top 39

Recall that the quantity u is connected with the angle of nutation θ by relation
(3.14). Define an angle θ0 by the relation

cos θ0 ¼ u0 ¼ 1  εx0

and put

θ ¼ θ0 þ η, ð3:54Þ

where η is a small quantity of the order ε. We also set

u1 ¼ 1  εx0  εΔ ¼ cos ðθ0 þ η1 Þ,

ð3:55Þ
u2 ¼ 1  εx0 þ εΔ ¼ cos ðθ0 þ η2 Þ:

Here η1 and η2 are small constant quantities which are also of the order of ε. Note
that, by virtue of relations (3.53) and (3.55), we have η1 > 0 > η2. Therefore, up to
small quantities of higher order, the following equalities are satisfied which follow
from relations (3.14), (3.54), and (3.55):

u  u1 ¼ cos ðθ0 þ ηÞ  cos ðθ0 þ η1 Þ ¼  sin θ0 ðη  η1 Þ,

u  u2 ¼ cos ðθ0 þ ηÞ  cos ðθ0 þ η2 Þ ¼  sin θ0 ðη  η2 Þ:

Let us substitute these equalities and relation (3.49) for u3 into expression (3.21).
We get

2mglsin 2 θ0 χ
QðuÞ ¼ ðη  η1 Þðη  η2 Þ:
A 1 ε2

Substitute the obtained expression for Q(u) into differential equation (3.18), and
proceed from variable u to variable η using the equality

u_ ¼  sin θ0 η_ ,

which follows from relations (3.14) and (3.54). We find

2mglχ
η_ 2 ¼ ðη  ηÞðη  η2 Þ:
A1 ε 2 1

Substitute expression (3.42) for ε and formula (3.47) for χ into the obtained
equation. As a result, we get

A23 r 2
η_ 2 ¼ ðη1  ηÞðη  η2 Þ: ð3:56Þ
A21
40 3 Lagrange’s Case

Fig. 3.5 Phase portrait of

system (3.56)

Differential equation (3.56) describes small oscillations of the nutation angle

θ ¼ θ0 + η around its average value θ0. The phase portrait of system (3.56) is repre-
sented in Fig. 3.5.
Note that the only close curve on the phase plane ðη; η_ Þ corresponds to the
fixed values of constants η1 and η2. By virtue of relations (3.55), (3.53), and (3.52),
these constants can be expressed in terms of constants h1 and g1, i.e., through the
constant values of the motion integrals H and G1.

References

1. Magnus, K.: Kreisel. Theorie und Anwendungen. Springer, Berlin (1971)

2. Appel, P.: Traite de Mechanique Rationnelle. Gauthier–Villars, Paris (1953)
3. Macmillan, W.D.: Theoretical Mechanics. Dynamics of Rigid Bodies. McGraw-Hill,
New York, NY (1936)
4. Grammel, R.: Der Kreisel. Seine Theorie und Seine Anwendungen. Erster Band. Izd-vo Inostr.
Lit-ry, Moscow (1952) Russian translation from German
5. Okunev, B.N.: The Free Motion of a Gyroscope. GITTL, Moscow (1951) in Russian
6. Bukhol’ts, N.N.: Fundamental Course of Theoretical Mechanics, vol. 2. Nauka, Moscow
(1969) in Russian
7. Suslov, G.K.: Theoretical Mechanics. Gostekhizdat, Moscow (1946) in Russian
8. Kil’chevskii, N.A.: Course in Theoretical Mechanics, vol. 2. Nauka, Moscow (1977) in
Russian
9. Golubev, V.V.: Lectures on Integration of the Equations of Motion of a Rigid Body about a
Fixed Point. Gostekhizdat, Moscow (1953) in Russian
10. Klein, F., Sommerfeld, A.: The Theory of the Top, Development of the Theory in the Case of
Heavy Symmetric Top, vol. 2. Birkhauser, Boston, MA (2010)
11. Sikorsky, Y.S.: Elements of the Theory of Elliptic Functions: With Applications in Mechanics,
2nd edn. rev. KomKniga, Moscow (2006) in Russian
12. Whittaker, E.T.: A Treatise on the Analytical Dynamics of Particles and Rigid Bodies, 4th edn.
Cambridge University Press, Cambridge, MA (1959)
13. Greenhill, A.G.: On the motion of a top and allied problems in dynamics. Quart. J. 11, 176–194
(1877)
14. Wittenburg, J.: Dynamics of Multibody Systems. Springer, Berlin (2008)
Chapter 4
Equations of Perturbed Motion of a Rigid
Body About Its Center of Mass

4.1 The Concept of a Perturbed Motion

The subject of this book is the investigation of perturbed motions of a rigid body
about its center of mass under the action of torques of various physical nature. If the
body is not acted upon by the internal or external torques, then it performs a
certain motion which is called unperturbed. As an unperturbed motion, one usually
considers the motion in the case of Euler or Lagrange. In real conditions, the body is
acted upon by the perturbation moments of internal or external forces, in particular,
gravitation forces, the forces of the medium resistance and the internal dissipative
forces.
In the case when the perturbation torques acting on the rigid body are small in
the sense that the current value of kinetic energy T of the body’s rotational motion
significantly exceeds the work A of perturbation forces on a certain time interval,
i.e., T
A, then the motion on this time interval will be close to the unperturbed
motion. However, on a large interval of time, the action of small perturbation
torques may lead to a gradual evolution of motion. Such motion is called a
perturbed motion.
The main purpose of the book is to study the evolution of a perturbed motion. In
this study, good results can be achieved by the application of the averaging method
[1–3]. To use the method of averaging, the equations of motion of a rigid body
should be reduced to a standard form of systems with one or several rotating phases.
The success in studying a perturbed motion depends, to a large extent, on the
choice of variables in writing the equation of a perturbed motion. In the general
case, the exact equations of a perturbed motion cannot be integrated; therefore,
the approximate methods of the study are used.
In celestial mechanics, the method of osculating elements is applied to study the
equations of a perturbed motion. The osculating elements are the characteristics of
the orbit that remain constant in the unperturbed motion and change in time in the
perturbed one. The convenience of the equations in terms of osculating elements

© Springer International Publishing AG 2017 41

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_4
42 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

consists of approximate solution that can be conducted with the help of asymptotic
methods.
In the problems of perturbed motion of a rigid body about its center of mass, one
can find the variables to be analogous to the osculating elements. The variables
describing the motion should be chosen in such way that the equations in these
variables for a perturbed motion have a form suitable for the application of asym-
ptotic methods. The slow and fast variables are identified, whereas in the
unperturbed motion (the motion in the case of Euler or Largange), the slow vari-
ables are constant. The variables satisfying these conditions are called evolutionary.
They can be selected in different ways, but it is desirable that they have a
simple mechanical or geometric meaning.
Various methods are known of introducing the evolutionary variables in the
study of the rotational motion of celestial bodies: Kepler’s osculating elements;
canonical elements of Jacobi, Delaunay, and Poincare; and the elements of Poisson,
Andoyer, Charlier, Deprit, and others [4]. B. V. Bulgakov introduced phase coordi-
nates close to the Andoyer–Deprit variables to describe the motion of a symmetric
rigid body [5]. V.V. Beletsky used osculating elements which differ by additive
constants from the Andoyer–Deprit variables to solve a number of problems of the
dynamics of rotational motion of dynamically symmetric satellites [6, 7]. The
further development of the method of osculating elements was carried out by
F.L. Chernousko, who investigated the movement of a satellite with a triaxial ellip-
soid of inertia [8].

4.2 Basic Concepts of the Averaging Method: Systems

in a Standard Form and Systems with Fast Rotating
Phase

In the subsequent Sects. 4.2–4.4, we consider the mathematical apparatus which is

widely used in this book for studying the problems of mechanics: the method of
averaging [1–3]. This is connected with the fact that usually in these problems, one
can promptly identify the fast and slow changing of variables. The method of aver-
aging has been long used in celestial mechanics, though without proper justifi-
cation. For the first time, it was rigorously formulated and substantiated in the
works of N. M. Krylov and N. N. Bogolubov.
There are presently many works devoted to the justification and application of
asymptotic methods. An exposition of these methods and a detailed bibliography on
this issue are contained in the books of N. N. Bogolubov, Yu. A. Mitropolsky [1, 3],
V. M. Volosov, B. I. Morgunov [2], N. N. Moiseev [9], and others.
N. N. Bogolubov applied the method of averaging to the standard systems of the
form (see, e.g., [1])
4.2 Basic Concepts of the Averaging Method: Systems in a Standard Form and. . . 43

x_ ¼ εXðx; t; εÞ ¼ εX1 ðx; tÞ þ ε2 X2 ðx; tÞ þ    ð4:1Þ

Here x, X are real n-dimensional vectors; ε > 0 is a small parameter.

It is assumed that the function X is sufficiently smooth with respect to variables
x and t and has the time average. We look for a change of variables:

x ¼ x þ εu1 ðx; tÞ þ ε2 u2 ðx; tÞ þ . . . , ð4:2Þ

which allows excluding t from the right-hand sides of the system and leads to an
averaged system of the form:

x_ ¼ εAðx; εÞ ¼ εA1 ðxÞ þ ε2 A2 ðxÞ þ . . . ð4:3Þ

This system does not explicitly contain the time t and can be integrated simpler
than (4.1). The solutions x allow approximating the solutions of system (4.1) on a
large interval of time t ~ ε1 with arbitrary accuracy. An algorithm for computing
the coefficients uk, Ak has been developed. In the first approximation, the system
assumes the form

ðT
x_ ¼ εX1 ðxÞ ¼ ε lim X1 ðx; tÞdt ¼ εA1 ðxÞ: ð4:4Þ
T!1
0

The method of averaging the systems (4.1) was generalized by N. N. Bogolubov

and D. N. Zubarev to the systems with a fast-rotating phase of the form [1]

x_ ¼ εXðx; y; εÞ ¼ εX1 ðx; yÞ þ ε2 X2 ðx; yÞ þ . . . ,

ð4:5Þ
y_ ¼ ωðxÞ þ εY ðx; y; εÞ ¼ ωðxÞ þ εY 1 ðx; yÞ þ . . .

Here x is a vector of dimension n, y is a scalar phase, and ε > 0 is a small para-

meter. It is assumed that X and Y are periodic functions of the variable y with the
period 2π. A formal scheme of averaging the system (4.5) consists of the following.
In system (4.5), the variables x change slowly, whereas the phase y changes rapidly.
In averaging the system (4.5), it is required to obtain a system in which the
slow variables x and the fast phase y will be separated. In this case, the fast phase
should not be included in the right-hand sides of the averaged equations.
To obtain the averaged system, we perform in system (4.5) a change of variables:

x ¼ x þ εu1 ðx; yÞ þ ε2 u2 ðx; yÞ þ . . . ,

ð4:6Þ
y ¼ y þ εv1 ðx; yÞ þ . . . ,

where x, y are new variables; ui ðx; yÞ, vi ðx; yÞ are some yet unknown functions which
are periodic with respect to y with the period 2π. Note that for ω0 ¼ dω dx 6 0, the
44 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

phase y is determined with a smaller (by one) degree of accuracy with respect to ε
than x.
We look for an averaged system in the form

x ¼ εA1 ðxÞ þ ε2 A2 ðxÞ þ . . . ,

ð4:7Þ
y ¼ ωðxÞ þ εB1 ðxÞ þ . . .

Here, Ai ðxÞ, Bi ðxÞ are some yet unknown functions.

The averaged system (4.7) is simpler than system (4.5), because in (4.7), the
equation x ¼ εA1 ðxÞ þ . . . is integrated independently of y and allows introducing
the slow time τ ¼ εt, changing on a bounded (as ε ! 0) interval of time.
After finding xðt; εÞ, the variable y is determined by a quadrature.
The process of obtaining transformation (4.6) consists in determining functions
ui ðx; yÞ, vi ðx; yÞ, Ai ðxÞ, and Bi ðxÞ. Functions X, Y, and ω are considered to be differ-
entiable functions of their variables as many times as the number of terms in (4.6) to
be calculated.
Let us differentiate the change of variable formulas (4.6), taking into account
system (4.7). Substitute in the obtained expression the derivatives x, y by the right-
hand sides of equations (4.7), expand all the functions with respect to powers of
parameter ε, and compare the coefficients at the same powers of the indicated para-
meter. We obtain a system of equations determining the sought-for functions [2, 3]:

∂u1
ωðxÞ ¼ X1 ðx; yÞ  A1 ðxÞ ¼ g1 ðx; yÞ  A1 ðxÞ,
∂y
∂v1 ∂ωðxÞ
ωðxÞ ¼ u1 þ Y 1 ðx; yÞ  B1 ðxÞ ¼ h1 ðx; yÞ  B1 ðxÞ,
∂
y ∂x
ð4:8Þ
∂u2 ∂X1 ðx; yÞ ∂X1 ðx; yÞ
ωðxÞ ¼ u1 þ v1 þ X2 ðx; yÞ
∂y ∂x ∂y
∂u1 ∂u1
 A1  B1  A2 ðxÞ ¼ g2 ðx; yÞ  A2 ðxÞ
∂x ∂y

and so on.
From the obtained relations, one can determine the coefficients u1, v1, u2 of
transformation (4.6) and functions A1, B1, A2 included in the right-hand sides of the
averaged system (4.7).
Consider the first equation in (4.8). Integrating it, we find

ðy
1
u1 ðx; yÞ ¼ ½g1 ðx; yÞ  A1 ðxÞdy þ φðxÞ: ð4:9Þ
ωðxÞ
y0

Here φðxÞ is an arbitrary function of x. We seek a solution for the functions
satisfying the boundedness condition
4.2 Basic Concepts of the Averaging Method: Systems in a Standard Form and. . . 45

lim jui ðx; yÞj < 1:

y!1

In (4.9), a periodic function of y with the period 2π with respect to y is under the
integral sign, since g1 ðx; yÞ is a periodic function of y, and A1 does not depend on y.
Suppose that the average value over the period for the integrand in (4.9) is not
equal to zero:

y0 þ2π
ð
1
g1  A 1 ¼ ½g1 ðx; yÞ  A1 ðxÞdy ¼ cðxÞ 6¼ 0:
2π
y0

In this case,

1
lim u1 ðx; y0 þ 2πkÞ ¼ lim 2πcðxÞk ¼ 1:
k!1 ωðxÞ k!1

The choice of sign depends on the sign of cðxÞ. Hence, for the boundedness of u1,
it is necessary and sufficient that

cðxÞ ¼ 0 ð4:10Þ

for any x.

Condition (4.10) is sufficient for a unique determination of the unknown func-
tion A1 ðxÞ in the integral in the equality (4.9). It follows from (4.10) that the
average value of function g1 is determined by the relation

2ðπ
1
g1 ðxÞ ¼ g1 ðx; yÞdy ¼ A1 ðxÞ:
2π
0

Consequently, in view of (4.8), we have

A1 ðxÞ ¼ g1 ðxÞ ¼ X1 ðxÞ:

Thus, the system of the first approximation has the form

2ðπ
1
x ¼ εX1 ðxÞ ¼ ε X1 ðx; yÞdy: ð4:11Þ
2π
0

Note that in many applied problems, it is enough to limit oneself to the first
approximation and stop at this stage of calculations. Of the main interest in appli-
cations is the behavior of slow variables. For approximate description of their
change over the times of the order of ε1, the system of equations (4.5) is
46 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

substituted by the averaged system (4.11). The obtained system for the description
of slow motion is simpler than the original one and is widely used in practice.

4.3 Systems Containing Slow and Fast Motions

Consider a more general scheme of averaging for the systems of differential equa-
tions containing multidimensional slow and fast motions of the form [2]

x_ ¼ εXðx; y; t; εÞ ¼ εX1 ðx; y; tÞ þ ε2 X2 ðx; y; tÞ þ . . . ,

ð4:12Þ
y_ ¼ Y ðx; y; t; εÞ ¼ Y 0 ðx; y; tÞ þ εY 1 ðx; y; tÞ þ . . .

Here x, X are n-dimensional, whereas y, Y are m-dimensional vector functions; ε

is a small positive parameter. Variables x are slow, since x_  ε, whereas variables
y are fast (relative to x), since y_  1.
Simultaneously with system (4.12), we will consider the corresponding singular
or the so-called unperturbed system obtained from (4.12) for ε ¼ 0:

y_ ¼ Y 0 ðx; y; tÞ ¼ Y ðx; y; t; 0Þ, x_ ¼ 0: ð4:13Þ

It is assumed that in the considered domain of the values of variables x, y, the

general solution y ¼ φ(t) of system (4.13) is known. This solution depends on para-
meters x and the choice of initial values yt¼t0 ¼ y0 or on other arbitrary constants.
Suppose that there exist along the integral curves of system (4.13) the average
values for the right-hand sides of system (4.12) and some other functions. If Φ(x, y, t)
is some function, then we take the following limit as its average value:

t0ð
þT
1
Φ ¼ lim Φðx; φ; tÞdt, ð4:14Þ
T!1 T
t0


where y ¼ φ(t) is a solution of (4.13) for yt¼t0 ¼ y0 . In the general case, function Φ
depends on x and the initial values of t0, y0; however, as shown in [2], the general
case of change is reduced to an important particular case, when the dependence of
Φ on the initial values is absent, and we assume it. Thus, we suppose that the
average values (4.14) depend only on x. If the average Φ depends on t0, y0, then the
averaging procedure becomes significantly more complicated. We construct the
change of variables

x ¼ x þ εu1 ðt; x; yÞ þ ε2 u2 ðt; x; yÞ þ . . . ,

ð4:15Þ
y ¼ y þ εv1 ðt; x; yÞ þ . . . ,

which transforms system (4.12) to the averaged form

4.3 Systems Containing Slow and Fast Motions 47

x_ ¼ εAðx; εÞ ¼ εA1 ðxÞ þ ε2 A2 ðxÞ þ . . . ,

y_ ¼ Y 0 ðx; y; tÞ þ εBðx; εÞ ¼ Y 0 ðx; y; tÞ þ εB1 ðxÞ þ . . . ð4:16Þ

In system (4.16), the slow motions of x are integrated independently of y;

therefore, system (4.16) is significantly simpler than the original system (4.12). If
Y0 does not depend on y, t (as, e.g, in equations (4.5), where Y0 ¼ ω(x)), then fast
variables y and time t are eliminated from the right-hand sides of the system (4.16).
Integrating the first equation in (4.16), it is possible to determine x with arbitrary
accuracy on the interval t ~ ε1 and then calculate the approximations for y. After
that, by formulas (4.15), we calculate the approximations for solutions (4.12).
Moreover, as shown in [2], in the general case, slow variables are calculated with
each approximation by an order of magnitude more accurately (with respect to ε)
than the rapid motions.
For the calculations by the presented scheme, one needs to know the coefficients
uk, vk  1, Ak, Bk  1. The rules for determination of these quantities are developed
and justified in [2].
For the calculations, it suffices to know the solution of the generating system
(4.13), with the help of which the average values (4.14) are obtained. The coeffi-
cient A1 is found to be X1; therefore, the system of the first approximation for x has
the form

t0ð
þT
1
x_ ¼ εX1 ðxÞ ¼ ε lim X1 ðx; φ; tÞdt: ð4:17Þ
T!1 T
t0

The averaging scheme for the system with a rotating phase is a particular case of
the presented general averaging scheme. In system (4.5), the generating equation of
the type (4.13) is written as: y_ ¼ ω ; its solution has the form y ¼ ω(t  t0) + y0.
Therefore, the average values (4.14) for such systems coincide with the average
with respect to y over the period

t0ð
þT
1
Φðx; yÞ ¼ lim Φðx; y0 þ ωðt  t0 ÞÞdt ¼
T!1 T
t0
ωðt0ðþT Þ 2ðπ
1 1
¼ lim Φðx; yÞdy ¼ Φðx; yÞdy:
T!1 Tω 2π
ωt0 0

Thus, formula (4.11) and other equations of the method of averaging for system
(4.5) follow a particular case from formulas (4.17) and other relations of the
method of averaging [2].
48 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

4.4 Higher-Order Averaging in Systems with Fast

and Slow Phases

In nonlinear oscillatory systems, the situations often occur when the evolution of
osculating variables takes place with different average velocities with respect to the
powers of some natural small parameter. Such mechanical systems describe a
number of problems of the theory of oscillations of mechanical systems (oscillators
and pendulums) and the dynamics of rigid bodies and gyroscopes, orbital motions,
and rotations of natural and artificial celestial bodies. In theoretical and applied
aspects, of significant interest is the study of evolution of a system on a sufficiently
large interval of time, which leads to a substantial variation of osculating variables,
including the slowest ones.
It turns out that in many important cases, one can apply and substantiate a modi-
fied scheme of the Krylov–Bogolubov averaging method and separation of motions
(change of variables [1, 2, 8, 10, 11]) on relatively large intervals of time [12].
We consider the standard in the sense of N.N. Bogolubov system [1, 2] for
two slow vectors x, y of arbitrary dimensions, and we assume that the following
requirements for the averages with respect to t are fulfilled:

x_ ¼ εXðt; x; yÞ, x ð 0Þ ¼ x 0 , X0 ðx; yÞ ¼ Mt fXg  0,

ð4:18Þ
y_ ¼ εY ðt; x; yÞ, y ð 0Þ ¼ y 0 , Y 0 ðx; yÞ ¼ Mt fY g 6¼ 0:

Functions X, Y are assumed to be piecewise continuous, 2π-periodic in t, and

sufficiently smooth for (x, y) 2 Dx Dy, where Dx, Dy are closed-bounded sets.
Here Mt means averaging with respect to argument (fast phase) t:
2Ðπ
Mt fXg ¼ 2π1
Xðt; x; yÞdt. In the first approximation with respect to ε, the average
0
rate of change of x equals to zero, i.e., |x  x0| ¼ O(ε), whereas y is of order of O(ε),
i.e., |y  y0| ¼ O(1), t ~ ε1. Of interest for applications is the substantial evolution
for t ~ ε2 of the slower variable x which characterizes the basic parameters of the
oscillatory system (energy, amplitude). The faster variable y is usually associated
with the evolution of phase or angular variable and can significantly influence the
change of vector x.
In the general case of the system of the form (4.18), application and justification
of the standard procedure of the averaging method on the interval t ~ ε2 are
difficult. Therefore, we consider the frequently encountered situation when the
averaged system for y under the constant x ¼ ξ admits a complete set of rotational
oscillatory single-frequency motions [1, 2, 10, 11]:

x ¼ ξ 2 Dx , y ¼ η0 ðφ; ζ; ξÞ 2 Dy , φ_ ¼ εωðζ; ξÞ: ð4:19Þ

Here φ is the slow phase ( φ_  ε); ζ is the slow variable ( ζ_  ε2 ); the total
dimension of vectors ζ, φ (mod2π) coincides with the dimension of y.
4.4 Higher-Order Averaging in Systems with Fast and Slow Phases 49

By the change (x, y) ! (ξ, η), which is close to the identical one, system (4.18) is
rearranged into the form

ξ_ ¼ ε2 Ξðt; ξ; η; εÞ, ξð0Þ ¼ x0 , ξ 2 Dx ,

η_ ¼ εY 0 ðξ; ηÞ þ ε2 H ðt; ξ; η; εÞ, ηð0Þ ¼ y0 , η 2 Dy ,
ðt ðt ð4:20Þ
x ¼ ξ þ ε Xðs; ξ; ηÞds, y ¼ η þ ε ½Y ðs; ξ; ηÞ  Y 0 ðξ; ηÞds:
0 0

Functions Ξ, H satisfy the required conditions of smoothness and periodicity.

Dropping the summands O(ε2) in (4.20) leads to expressions (4.19) for ξ, η.
Note that, due to change in (4.19), there holds the identity
 
∂η0
ωðξ; ζ Þ  Y 0 ðξ; η0 Þ ð4:21Þ
∂φ

with respect to variables φ, ζ, ξ. The change y ¼ η ! (ζ, φ), according to (4.19) and
taking into account the quantities O(ε2) and identity (4.21), results in a system with
the fast t and slow φ phases of the form
 T
α_ ¼ ε2 Aðt; α; φ; εÞ, αð0Þ ¼ α0 , α ¼ ξT ; ζ T ,
φ_ ¼ εωðαÞ þ ε2 Φðt; α; φ; εÞ, φð0Þ ¼ φ0 ðmod2π Þ,
    ð4:22Þ
 T  T ∂η0 ∂η0 1 ∂η0
A ¼ ΞT ; Z T , Φ; ZT ¼  ;
 ∂φ ∂ζ 
 H  Ξ :
∂ξ

Functions A, ω, Φ are sufficiently smooth with respect to α, φ, ε, piecewise

continuous with respect to t and 2π-periodic and with respect to t and φ. The initial
values α0, φ0 are determined by the change y ! (ζ, φ) (4.19). System (4.22) is
subject to further study on the interval 0
t
εL2 , on which the slow variable α,
generally speaking, may have an increment δα ~ 1. In this case, with a given degree
of accuracy and with respect to ε, the separation is carried out of the fast phase, the
argument t, whereas the relatively slow phase φ and the variable α are connected.
The «averaged» system admits introducing the argument τ ¼ εt, 0
τ
Lε and after
that can be subjected to the standard asymptotic analysis [1, 2, 8, 10, 11, 13, 14]. In
the case of scalar phase φ, the averaging procedure is applied [1, 2] developed for
the systems with rapidly rotating phase. As in the classical Krylov–Bogolubov
method, the proposed averaging scheme of the higher orders is based on the
requirement that the asymptotic expansions do not contain singular terms of the
type (ε2t)k on the expanded interval 0
t
εL2 . A particular situation when y is
absent was investigated earlier in [15].
In the separation of the fast phase t, the change (α, φ) ! (β, ψ) is used such that
the equations do not include t with the required degree of accuracy with respect to ε:
50 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

α ¼ β þ ε2 Πðt; β; ψ; εÞ, φ ¼ ψ þ ε2 Γðt; β; ψ; εÞ,

ð4:23Þ
β_ ¼ ε2 Bðβ; ψ; εÞ, ψ_ ¼ εωðβÞ þ ε2 Ψðβ; ψ; εÞ:

The unknown 2π-periodic with respect to t, ψ functions of change Π, Γ and not

containing t (averaged) functions B, Ψ in the right-hand sides of system (4.23) can
be determined approximately by the asymptotic expansions or successive approx-
imation with respect to the powers of ε of the solutions of the partial differential
equations:
 
I þ ε2 Πβ0 B þ Πψ 0 ðεω þ ε2 ΨÞ ¼ Aðt; β þ ε2 Π; ψ þ ε2 Γ; εÞ  Πt0 ,
ε2 Γβ0 B þ 1 þ ε2 Γψ 0 ðωðβÞ þ εψ Þ ¼ ð4:24Þ
¼ ωðβ þ ε2 ΠÞ þ εΦðt; β þ ε2 Π; ψ þ ε2 Γ; εÞ  εΓt0 :

In particular, the first coefficients of the expansions determining the significant

evolution of variables equal

ðt
B0 ¼ Mt fAðt; β; ψ; 0Þg, Π0 ðt; β; ψ Þ ¼ ððAÞ  Mt fðAÞgÞds,
0
ðt ð4:25Þ

ψ 0 ¼ Mt fΦðt; β; ψ; 0Þg, Γ0 ðt; β; ψ Þ ¼ ðΦÞ  Mt fðΦÞgds,
 0
B1 ¼ Mt fðAε0 Þg  ωðβÞMt Π0φ0 , ...

Here, Mt means averaging with respect to the explicitly involved argument t,

while the expressions of the type (A), (Φ) correspond to the values α ¼ β, φ ¼ ψ,
ε ¼ 0. The subsequent coefficients Bj, Πj  1, Ψj  1, Γj  1, j  2, are calculated
recursively. By analogy with the classical scheme of averaging, in constructing
the j-th approximation on the interval t ~ ε2, it is required to determine the func-
tions up to Bj  1, Πj  2, Ψj  2, Γj  3. In particular, the initial value problem of the
first approximation has the form

β_ ¼ ε2 A0 ðβ; ψ Þ, βð0Þ ¼ α0 ;
ð4:26Þ
ψ_ ¼ εωðβÞ, ψ ð0Þ ¼ φ0 ; 0
t
Lε2 :

System (4.26) is subject to further analytic or numerical study. It is significantly

simpler than the original system (4.22) and admits the introduction of slow time
τ ¼ εt and a representation in the standard form with the «fast phase» ψ [1, 2, 10, 11,
13, 14]. If the function ω(β) is separated from zero, then the phase ψ is rotating, and
in the first approximation, the method of averaging with respect to variable ψ can be
applied to the system on the interval of time 0
τ
Lε , i.e., t ~ ε2.
4.5 Equations of Perturbed Motion of a Rigid Body Close to Euler’s Case 51

4.5 Equations of Perturbed Motion of a Rigid Body Close

to Euler’s Case

Consider the influence of small perturbations on the motion of a rigid body close to
Euler’s case. Let us turn to the choice of evolutionary variables in this case.
Note that, in the motion of a free rigid body, its kinetic energy is preserved:

1  1
T¼ A1 p2 þ A2 q2 þ A3 r 2 ¼ ðG; ωÞ, ð4:27Þ
2 2

as well as the modulus of angular momentum G

G2 ¼ A21 p2 þ A22 q2 þ A23 r 2 ¼ ðJωÞ2 ð4:28Þ

Here, ω ¼ ( p, q, r) is the vector of angular velocity; p, q, r are its projections on

the axes of the moving coordinate system Oxyz, directed along the principal axes of
inertia of the body for a fixed point O; G is the vector of angular momentum of the
body relative to the point O; J ¼ diag(A1, A2, A3) is the tensor of inertia which has a
diagonal form in the body-fixed axes. Without loss of generality, we assume that
A1 > A 2 > A 3.
If we consider the motion of the angular momentum vector in the axes connected
with the body, then the end of vector G moves along the line of intersection of the
ellipsoid and sphere; the equations of which are obtained from equalities (4.27) and
(4.28), expressed by the components G1, G2, G3 of vector G:

G21 A1 2 1 2 1
1 þ G2 A2 þ G3 A3 ¼ 2T, ð4:29Þ
G21 þ G22 þ G23 ¼G :
2
ð4:30Þ

The intersection of the ellipsoid and sphere is provided by the inequalities

2TA1  G2  2TA3 : ð4:31Þ

In studying the motion of vector G in the body-fixed axes, the trajectories of its
end on the unit sphere are investigated (see [16–19]). In this case, to describe the
position of the angular momentum vector on the sphere, one introduces a positive
parameter k2 ¼ k2(T, G2), 0
k2
1 (see Sect. 2.3). For example, in the domain
2TA1  G2 > 2TA2, the parameter k2 is represented in the form (2.29):
 
ðA2  A3 Þ 2TA1  G2
k2 ¼  2  , 0
k2
1: ð4:32Þ
ðA1  A2 Þ G  2TA3

The value k2 ¼ 0 corresponds to the rotation of a rigid body about the axis with
the moment of inertia A1, whereas k2 ¼ 1 to the motion along the separatrix (see
52 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

Fig. 2.1). In the passage to the domain 2TA2 > G2  2TA3, one needs to interchange
A1 and A3 in the expression for k2. Here k has the meaning of the modulus of
elliptic functions of the Euler–Poinsot motion.
In the presence of small perturbations, the motion of a rigid body is described by
the perturbed dynamic Euler’s equations:

G_ þ ω G ¼ εL, ωðt0 Þ ¼ ω0 , ε  1: ð4:33Þ

It is assumed that L is the perturbation torque, whereas the motion is considered

on an asymptotically large interval of time t  t0 ~ ε1.
In the absence of perturbations (ε ¼ 0), the rigid body performs the Euler–
Poinsot motion. In this case, the quantities T, G, and k2 are preserved, whereas
the angular velocities of the body rotation are expressed by elliptic functions
(see Sect. 2.3).
For example, in the domain 2TA1  G2 > 2TA2, the solution is expressed through
the Jacobian elliptic functions by formulas (2.30). In addition,

2π
η¼ ðt  t0 Þ þ η0 : ð4:34Þ
N

Here η is the phase; the period N of the change of the angular velocities of the
Euler–Poinsot motion with respect to time has the form
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
A1 A2 A3
N ¼ 4K ðkÞ  , ð4:35Þ
ðA1  A2 Þ G2  2TA3

K(k) is the complete elliptic integral of the first kind.

In the domain 2TA2 > G2  2TA3, in all the formulas (2.30), and, besides, in
(4.32), one has to interchange A1 and A3.
To study the evolution of motion of a rigid body by the method of averaging, it is
natural to proceed from variables p, q, r to the evolutionary variables G2, T, η or G2,
k2, η, where G2, T, k2 are slow variables, whereas the phase η is the fast variable.
Due to equation (4.33), the equations for slow variables have the form
 
G2 ¼ 2εðGLÞ, T_ ¼ εðωLÞ,
 
 2  ∂k2 ∂k2 ð4:36Þ
k ¼ε 2 Jω þ ω; L
∂G2 ∂T

The equation for the phase has the form

2π
η_ ¼ þ OðεÞ: ð4:37Þ
N
4.5 Equations of Perturbed Motion of a Rigid Body Close to Euler’s Case 53

The specific form of the function O(ε) is not given here because of its awkward-
ness and because it is not used for the construction of the first approximation
solution. The right-hand sides of the equations for slow variables are periodic
with respect to η with the period 2π.
Since G2, T, k2 are slowly changing the functions of time, then, in the first
approximation, it is possible to substitute, instead of p, q, r, their values (2.30) from
the unperturbed Euler–Poinsot motion into the right-hand sides of equations (4.36).
Next, by averaging with respect to the fast phase η and assuming the slow variables
to be constant, we obtain in the general form
2 Þ ¼ 2εMη fðG0 ; L0 Þg, T ¼ εMη fðω0 ; L0 Þg,
ðG
 
2  ∂k2 ∂k2 ð4:38Þ
ðk Þ ¼ εMη 2 2 Jω0 þ  ω0 ; L0
∂G ∂T

Here L0, ω0 are known functions of argument η for zero approximation (upon the
substitution of solution in Euler’s case).
The specific form of the right-hand sides of equations (4.38) depends on the form
of the perturbation torque εL. The bar quantities above denote the corresponding
averaged variables G 2 , T2 , k2 . The averaging of any function F in (4.38) with respect
k
to η is carried out by the scheme

2ðπ
1
M η f Fg  Fdη: ð4:39Þ
2π
0

Let us discuss the introduction of a system of coordinates for the description of a

perturbed motion of a rigid body. Consider first a fast motion of a heavy rigid body
about a fixed point, when the role of perturbation is played by the gravity
torque [20].
The origin of all systems of coordinates O is superposed with the fixed point of
the rigid body. The axis Ox3 of the fixed coordinate system Ox1x2x3 is directed
vertically upward. The axis Oy3 of the system Oy1y2y3 is directed along the vector
of angular momentum G of the body. The axis Oy1 is perpendicular to Oy3, whereas
the axis Oy2 is perpendicular to Oy1 and Oy3 (Fig. 4.1).
The position of the vector of angular momentum G in the coordinate system
Ox1x2x3 is determined by the angles δ and λ, where δ is the angle between the axis
Ox3 and vector G, whereas λ is the angle between the axis Ox1 and the projection of
vector G on the plane Ox1x2.
The principal central axes of inertia of the body Oz1z2z3 are introduced. The
relative position of the axes Oz1z2z3 and Oy1y2y3 is defined by the Euler angles
(Fig. 4.2).
In the unperturbed motion, the vector of angular momentum G of the body is
constant in modulus and direction. Therefore, the quantities G, δ, and λ are constant
in the unperturbed motion and are variable in the perturbed motion.
54 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

Fig. 4.1 Angles of

orientation of the angular
momentum vector

Fig. 4.2 Euler angles for

principal central axes of
inertia

As the variables describing the perturbed motion, we choose G, δ, λ, θ, φ, ψ. The

differential equations of the perturbed motion are written for the value of angular
momentum G and the angles δ, λ, θ, φ, ψ. These variables are convenient that in the
unperturbed Euler–Poinsot motion G, λ, δ are constant, whereas the Euler angles θ,
φ, ψ describe the motions by the known formulas.

4.6 Equations of Perturbed Motion of a Satellite About Its

Center of Mass

Consider a rigid body (a satellite) performing an arbitrary motion in space. Three

right-handed Cartesian systems of coordinates are introduced, the origin of which,
in distinction from the ones introduced in Sect. 4.5, is superposed with the center of
inertia of the satellite. The system of coordinates Ox1x2x3 moves translationally
together with the center of inertia of the satellite. The axis Ox1 is parallel to the
radius vector of the orbit’s perigee, the axis Ox2 is parallel to the velocity vector of
the satellite’s center of mass in the perigee, and the axis Ox3 is directed along the
normal to the orbital plane.
The axis Oy3 of the system of coordinates Oy1y2y3 is directed along the vector of
angular momentum G of the satellite relative to the center of inertia; the axis Oy1 is
4.6 Equations of Perturbed Motion of a Satellite About Its Center of Mass 55

perpendicular to Oy3 and lies in the plane Ox3y3, whereas the axis Oy2 is perpen-
dicular to Oy1 and Oy3 and, hence, lies in the orbital plane Ox1x2 (Fig. 4.1). The
transition from the system of coordinates Ox1x2x3 to the system Oy1y2y3 is realized
by two rotations: by the angle λ about the axis Ox3 and by the angle δ about the axis
Oy2. The angles λ and δ determine the orientation of vector G in the fixed space.
The axes of the body-fixed system of coordinates Oz1z2z3 are superposed with
the principal central axes of inertia of the satellite. Their orientation relative to the
system of coordinates Oy1y2y3 is determined by the Euler angles θ, φ, ψ (Fig. 4.2)
and the direction cosines αik ¼ yizk, i , k ¼ 1, 2,3. Here, yi is the unit vector of the
system Oy1y2y3 and zk is the unit vector of the system Oz1z2z3. The relations
between the direction cosines and the Euler angles are given by the expressions
(1.31):

α11 ¼ cos φ cos ψ  cos θ sin φ sin ψ,

α12 ¼  cos ψ sin φ  sin ψ cos φ cos θ, α13 ¼ sin ψ sin θ,
α21 ¼ sin ψ cos φ þ cos ψ sin φ cos θ,
ð4:40Þ
α22 ¼  sin ψ sin φ þ cos ψ cos φ cos θ,
α23 ¼  cos ψ sin θ,
α31 ¼ sin φ sin θ, α32 ¼ cos φ sin θ, α33 ¼ cos θ:

Let us form the equations of motion of the satellite relative to the center of mass,
taking six desired functions as the value of the angular momentum G and the angles
δ, λ, θ, φ, ψ. The theorem of change of angular momentum has the following vector
form (see (1.20)):

dG
¼ L0 , ð4:41Þ
dt

where L0 is the perturbation torque relative to the center of inertia of the satellite.
Projecting the vector equation (4.41) on the axes of the system of coordinates
Oy1y2y3, we arrive at

dG dδ L1 dλ L2
¼ L3 ¼ , ¼ , ð4:42Þ
dt dt G dt G sin δ

where Li is the projection of the perturbation torque relative to the center of inertia
on the axes Oyi.
Let us derive the equations for the angles θ, φ, ψ. The vector ω of the absolute
angular velocity of the satellite rotation relative to the system of coordinates
Ox1x2x3 is composed of five angular velocities corresponding to the rotations by
the angles δ, λ, θ, φ, ψ. Taking into account the directions of these components
(Figs. 4.1 and 4.2), we find
56 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

 
ω ¼ δ_ y2 þ λ_ cos δy3  sin δy1 þ θ_ ð cos φz1  sin φz2 Þ þ ψ_ y3 þ φ_ z3 : ð4:43Þ

Let us project equality (4.43) on the axes Oz1z2z3 and take into account that
αik ¼ yizk. Then

ω ¼ pz1 þ qz2 þ rz3 ,

p ¼ δ_ α21 þ λ_ ðα31 cos δ  α11 sin δÞ þ θ_ cos φ þ ψ_ α31 ,
ð4:44Þ
q ¼ δ_ α22 þ λ_ ðα32 cos δ  α12 sin δÞ  θ_ sin φ þ ψ_ α32 ,
r ¼ δ_ α33 þ λ_ ðα33 cos δ  α13 sin δÞ þ φ_ þ ψ_ α33 ,

On the other hand, projecting the angular momentum vector G on the axes of the
body-fixed coordinate system Oz1z2z3 yields

G1 ¼ A1 p ¼ G sin θ sin φ,
G2 ¼ A2 q ¼ G sin θ cos φ, ð4:45Þ
G3 ¼ A3 r ¼ G cos θ,

where A1, A2, A3 are the principal central moments of inertia of the satellite relative
to the axes Oz1, Oz2, Oz3, respectively.
Let us substitute into equations (4.44) p, q, r from (4.45), δ_ , λ_ from (4.42), and αij
from (4.40) and solve them with respect to the derivatives of the Euler angles θ, φ,
ψ:
 
1 1 L2 cos ψ  L1 sin ψ
θ_ ¼ G sin θ sin φ cos φ  þ ,
 A 1 A 2  G
1 sin 2 φ cos 2 φ L1 cos ψ þ L2 sin ψ
φ_ ¼ G cos θ   þ , ð4:46Þ
 2 A3 A1  A2 G sin θ
sin φ cos 2 φ L1 cos ψ þ L2 sin ψ L2
ψ_ ¼ G þ  θ  δ:
A1 A2 G G

Equations (4.42), (4.46) form a system of equations of the perturbed motion in

the form convenient for the application of asymptotic methods. Equations (4.42)
describe the change of the angular momentum vector, whereas in equations (4.46),
the satellite motion relative to this vector.
Consider a satellite, the moments of inertia of which A1  A2  A3, A1
A2 + A3
are arbitrary. Suppose that the angular velocity ω of the satellite motion relative to
the center of mass is significantly larger than the angular velocity of the orbi-
tal movement ω0 (in other words, the applied torques are small) and put
 
ω0 A 1 ω0 L G
ε¼   1,  μ ω  : ð4:47Þ
ω G A 1 ω2 A1

Thus, the velocity of the satellite motion relative to the center of mass is charac-
terized by two independent small parameters. The first of them is ε as the ratio of the
4.6 Equations of Perturbed Motion of a Satellite About Its Center of Mass 57

angular velocity of the orbital motion to the angular velocity of the relative motion.
The second small parameter μ equals the ratio of the work of the applied torques
during a characteristic time of relative motion (the rotation time of the satellite
about its center of mass) to the average kinetic energy of the relative motion.
Suppose that the unit of measurement of time and the period of relative motion
2π
ω are of the same order, then ω0 ~ ε. For the gravity torques, these small parameters
are connected by the relation μ ¼ ε2 [8] and the perturbation torques Li ~ ε2.
In the general case, parameters μ, ε may be connected differently, and prior to
the analysis of the fast motions of the satellite under the action of perturbation
torques of a specific type, one has to set or evaluate the relative magnitude of the
small parameters in order to restrict oneself to a necessary accuracy in the con-
struction of an asymptotic solution.
In the unperturbed motion (ε ¼ 0), the perturbation torques turn to zero, and this
motion is the Euler–Poinsot motion. The quantities G, δ, λ and the kinetic energy
T of the satellite motion relative to the center of mass are constant. We have
 2  
1  G2 sin φ cos 2 φ cos 2 θ
T ¼ A1 p2 þ A 2 q2 þ A3 r 2 ¼ þ sin θ þ
2
: ð4:48Þ
2 2 A1 A2 A3

For ε ¼ 0, the Euler angles θ, φ, ψ are variable, while the function ψ ¼ ψ(t) can
be represented in the form ψ ¼ ψ 1(t) + ψ 2(t). At that, as it was noted in the end of
Sect. 2.3, the functions θ(t), φ(t), ψ 1(t) are either periodic in t with period τ (the
period τ ¼ τ(G, T) is the time of motion of the vector G along a closed trajectory) or
have the increment of 2π during the time τ. The function ψ 2(t) equals

2π
ψ 2 ðtÞ ¼ ðt þ t0 Þ, t0 ¼ const:
τ0

Generally speaking, the periods τ0 ¼ τ0 (G, T ) and τ are incommensurable

[21, 22].
We introduce two variables (phases) by the relations

y1 ¼ ω1 ðt þ t1 Þ, y2 ¼ ω2 ðt þ t2 Þ, ð4:49Þ

where t1, t2 are arbitrary constants, while

2π 2π
ω1 ðG; T Þ ¼ , ω2 ðG; T Þ ¼ : ð4:50Þ
τ τ0

Then, in the Euler–Poinsot motion, the angles θ, φ, ψ will be some functions of

these variables (phases), as well as the quantities G and T (they do not depend on
angles δ and λ due to isotropy of different directions in the space for
L1 ¼ L2 ¼ L3 ¼ 0).
Using the equality ψ 2 ¼ y2, we write these functions in the form
58 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

θ ¼ θðG; T; y1 Þ, φ ¼ φðG; T; y1 Þ, ψ ¼ ψ 1 ðG; T; y1 Þ þ y2 : ð4:51Þ

In the increasing of y1 by 2π, functions θ, φ, ψ 1 in (4.51) either do not change or

get an increment of 2π. Note that the increments of 2π during the period are of no
consequence, since there are the trigonometric functions of the angles in our equa-
tions everywhere.
In the perturbed motion (ε 6¼ 0), the slow variables are G, δ, and λ, while the
fast ones are θ, φ, and ψ.
Consider a motion of the satellite relative to its center of mass under the action of
perturbation torques. Up to the quantities of the magnitude order of the squared
ratio of the linear dimensions of the satellite to the orbit sizes, we can assume that
the motion of the satellite relative to its center of mass does not affect the motion of
the center of mass itself. The center of mass moves along the Kepler ellipse with
eccentricity e and orbital period Q0. The dependence of true anomaly (polar angle)
v on time t is given by the relation [6–8]

dv ω0 ð1 þ e cos vÞ2 2π
¼ , vðt þ Q0 Þ ¼ vðtÞ þ 2π, ω0 ¼ : ð4:52Þ
dt ð1  e2 Þ3=2 Q0

The equations of motion (4.42), (4.46), and equation (4.52) under the conditions
ω0 ¼ O(ε), Li ¼ O(ε2) assume the form

x_ ¼ ε2 Xðx; y; vÞ, y_ ¼ Y 0 ðx; yÞ þ ε2 Y 1 ðx; y; θÞ, v_ ¼ εf ðvÞ, ð4:53Þ

where f(v) is the right-hand side of equation (4.52).

Here, x is the vector of slow variables which includes variables G, δ, λ, whereas
y is the vector of fast variables containing variables θ, φ, ψ.
System (4.53) can be simplified by performing the change of variables by
formulas (4.51). Instead of three fast variables θ, φ, ψ, we introduce two fast vari-
ables y1 and y2 together with one slow variable T (the constant in the Euler–Poinsot
motion). Phases y1, y2 in the perturbed motion are no longer determined by formulas
(4.49) but are new desired functions, whereas, in the first approximation, their
rates of change equal ω1, ω2, respectively. Therefore, the system of equations of
motion can be written in the form

x_ ¼ ε2 Xðx; y1 ; y2 ; vÞ, v_ ¼ εf ðvÞ,

ð4:54Þ
y_ 1 ¼ ω1 ðxÞ þ ε2 Z1 ðx; y1 ; y2 Þ, y_ 2 ¼ ω2 ðxÞ þ ε2 Z 2 ðx; y1 ; y2 Þ:

Here x, X are four-dimensional vector functions corresponding to variables G, T,

δ, λ, while other functions are scalar.
Functions X, Z1, Z2 are periodic in y1, y2 with the periods of 2π.
Thus, the equations of the perturbed motion of a satellite are reduced to a
system with two rotating phases.
4.7 Procedure of Averaging for a Body with Moments of Inertia Close to One Another 59

Let us describe now a scheme of averaging proposed in [8] by F.L. Chernousko

for studying the motion of a nonsymmetric satellite under the action of gravity
torque when conditions (4.47) are satisfied. It is shown in this paper [8] that
averaging of a periodic function F(θ, φ, ψ) with respect to t, taking into account
the dependencies of the variables θ ¼ θ(t), φ ¼ φ(t), ψ ¼ ψ 1(t) + ψ 2(t), can be
subdivided into two independent stages: averaging with respect to variable ψ and
averaging with respect to time t taking into account the dependencies of the vari-
ables θ(t), φ(t) on t.
0
Indeed, by virtue of incommensurability of the periods τ and τ , we have

ðτ ðτ0  
1 2πt0
Mt fFðθ; φ; ψ Þg ¼ 0 F θðtÞ; φðtÞ; ψ 1 ðtÞ þ 0 dt0 dt ¼
ττ τ
00 8 9
ðτ 2ðπ ðτ < 2ðπ =
1 1 1
¼ FðθðtÞ; φðtÞ; ψ Þdψdt ¼ Fðθ; φ; ψ Þdψ dt ¼
2πτ τ :2π ;
0 0 0 0
¼ M1 Mψ ½Fðθ; φ; ψ Þ :
ð4:55Þ

Here Mψ means averaging over ψ, while M1 over θ and φ connected by relation

(4.48), the averaging being performed along the closed trajectories of the
angular momentum vector in the Euler–Poinsot motion (Fig. 2.1).
The error of the averaged solution for slow variables has the magnitude order of
ε on the time interval, during which the body performs ε1 rotations. The reso-
nance case when τ and τ0 are commensurable requires additional consideration.
System (4.53) is a nonlinear oscillatory system containing a fast and a relatively
slow phase [12]. In [12] and in Sect. 4.4, a modified averaging method is proposed
for the situation when the averaged variables do not change. A procedure of separ-
ation of variables is described and substantiated, which is performed on sub-
stantially larger time intervals, on which a significant evolution of all variables
takes places.

4.7 Procedure of Averaging for a Body with Moments

of Inertia Close to One Another

Let us consider, following [8], the case when the principal central moments of
inertia of the satellite are close to one another and can be represented in the form

A1 ¼ J 0 þ εA01 , A2 ¼ J 0 þ εA02 , A3 ¼ J 0 þ εA03 , ð4:56Þ

60 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

where 0 < ε  1 is a small parameter. We apply the asymptotic method of averag-

ing of the systems containing slow and fast motions to the equations of relative
motion of the satellite (4.42), (4.46), and (4.52).
For ε ¼ 0, these motions describe the motion of a spherically symmetric satellite.
In this case, in the unperturbed motion, we have L1 ¼ L2 ¼ L3 ¼ 0, and from the
system (4.42), (4.46), we find that G, δ, λ, θ, and φ are constant, whereas

ψ ¼ GJ 1
0 t þ ψ 0, ð4:57Þ

i.e., the satellite rotates uniformly about the translationally moving vector of
angular momentum.
For ε 6¼ 0, the role of slow variables in the system of seven equations (4.42),
(4.46) is played by G, δ, λ, θ, φ, while the role of the fast ones by ψ and v. To obtain
the solution in the first approximation, it suffices just to average the right-hand sides
of equations (4.42), (4.46) by substituting into them v from the solution of equa-
tion (4.52) and ψ from (4.57).
Note that the right-hand sides of equations for slow variables are sums of the
terms of the type sinψ, cosψ, sinv, cosv. It is assumed that no resonance relation
holds for the frequencies ω0 and GJ 1 1
0 , i.e., mω0 þ nGJ 0 6¼ 0, where m and n are
any integers.
Under this assumption, the averaging of the right-hand sides of equations (4.42),
(4.46) with respect to time can be substituted by independent averaging over
variables ψ and v(t). Let f(x, v, ψ) be a function of the variables x{G, δ, λ, θ, φ}, v,
and ψ, which periodically depend on v and ψ, where the variables x, v, and ψ are
solutions of the unperturbed system, i.e., G ¼ const, δ ¼ const, λ ¼ const, θ ¼ const,
φ ¼ const, v ¼ v(t), and ψ ¼ GJ 10 t þ ψ 0 . Function f{x, v, (t), ψ} has the period ω0 in
2π

t and the period 2π in ψ; it can be expanded into Fourier series with respect to
arguments t and ψ.
Thus, under the fixed values of slow variables, the right-hand sides of the
equations to be averaged will be sums of the terms of the form f1(ψ)f2(v), where
functions f1, f2 are periodic with respect to their arguments with the periods of 2π.
Besides, the expansion of f1(ψ) into Fourier series contains the harmonics not
higher than the third ones. Therefore, the expansion of the right-hand sides of
equations (4.42) and (4.46) into the double Fourier series (with respect to ψ and
v), after the substitution of ψ and v as functions of time, will be a sum of terms of the
form
  
cmn cos m LJ 1
0 t þ ψ0 cos nω0 ψ ðm ¼ 0; 1; 2; 3; n ¼ 0; 1; 2; . . .Þ

and the similar ones in which one or both cosines are substituted by sines.
Suppose that for all natural n, none of the following equalities hold
4.7 Procedure of Averaging for a Body with Moments of Inertia Close to One Another 61

1 1
G ¼ nJ 0 ω0 , G ¼ nJ 0 ω0 , G ¼ nJ 0 ω0 : ð4:58Þ
2 3

In this case, time integration of all terms of the double Fourier series, except for
the term with m ¼ n ¼ 0, yields trigonometric functions of t. Because of their
boundedness, after a passage to the limit by formula (4.14), we get zero limit for
these series, while for the term with m ¼ n ¼ 0, the limit equals c00. Consequently,
the result of averaging is written as Mt{f1(ψ)f2(v)} ¼ c00, and it does not depend on
ψ 0. On the other hand, the Fourier coefficient c00 can be obtained by independent
averaging of functions f1 and f2 considered as functions of time. The averaging of
f1(ψ) with respect to time is obviously reduced to averaging with respect to ψ,
because in the unperturbed motion, ψ is a linear function of time (4.57).
Thus, in the case when none of equalities (4.58) is true, the averaging of the
right-hand sides of the motion equations with respect to time can be substituted by
independent averaging over ψ and over v, as functions of t; moreover, the result of
averaging does not depend on the initial value ψ 0.
If at least one of equalities (4.58) holds, then the resonance effects take place
which are not considered here.
The time averaging of the functions depending on v is reduced, by virtue of
(4.52), to averaging over v in the following way:

ð0
Q 2ðπ 3=2
1 1 ð1  e2 Þ FðvÞdv
M t fFð v Þ g ¼ FðvðtÞÞdt ¼ ¼
Q0 2π ð1 þ e cos vÞ2
0 ( 0 ) ð4:59Þ
2 3=2 Fð v Þ
¼ ð1  e Þ M v :
ð1 þ e cos vÞ2

The equations (4.42) and (4.46) of the perturbed motion of a satellite have a
general character. However, the problems are often encountered, in which the
acting torques have a force function

U ¼ Uðλ; δ; θ; φ; ψ; tÞ:

Here λ, δ are angles which determine the orientation of the angular momentum
vector G in the fixed space (see Fig. 4.1); θ, φ, ψ are the Euler angles.
Then equations (4.42) and (4.46) are transformed as follows [6, 7]:
62 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

1 ∂U 1 ∂U 1 ∂U
λ_ ¼ , δ_ ¼  þ ctg δ ,
G sin δ ∂δ G sin δ ∂λ G ∂ψ
∂U   1 ∂U 1 ∂U
G_ ¼ , θ_ ¼ G sin θ sin φ cos φ A1
1  A2
1
 þ ctg θ ,
∂ψ G sin θ ∂φ G ∂ψ
  1 ∂U
φ_ ¼ G cos θ A1 1 1
3  A1 sin φ  A2 cos φ þ
2 2
,
 G sin θ ∂θ 
  1 ∂U ∂U
ψ_ ¼ G A1 2 1
1 sin φ þ A2 cos φ 
2
ctg δ þ ctg θ :
G ∂δ ∂θ
ð4:60Þ

The averaging schemes for studying the motion of a rapidly rotating symmetric
or triaxial satellite for the perturbations having a force functions are suggested in
[6, 7].
In the case of rapidly rotating satellite with a triaxial ellipsoid of inertia, there are
three fast frequencies: the orbital one ω0, the one of precession ωψ , and the fre-
quency of periodic motion along the polhodes ωθ. Therefore, the force function
U should be averaged three times: over v(t), over ψ, and along the polhode of the
unperturbed motion. Just note that near the separatrix (Fig. 2.1), vector G moves
slowly, and therefore it is possible to average the motion along the polhode every-
where except a certain neighborhood of the separatrix.

4.8 Equations of Perturbed Rotational Motion of a Rigid

Body Close to Lagrange’s Case

4.8.1 General Case

We consider a perturbed motion relative to a fixed point for a dynamically sym-

metric heavy rigid body in the case of perturbations of arbitrary nature. In contract
to Sect. 4.5, the gravity torque is not considered here as a perturbation torque but is
related to unperturbed motion, which is a motion in Lagrange’s case (see Sect. 3.1).
The equations of motion have the form

A1 p_ þ ðA3  A1 Þqr ¼ mgl sin θ cos φ þ εL1 ,

A1 q_ þ ðA1  A3 Þpr ¼ mgl sin θ sin φ þ εL2 ,
A3 r_ ¼ εL3 , Li ¼ Li ðp; q; r; φ; θ; ψ Þ, i ¼ 1, 2, 3,
ð4:61Þ
φ_ ¼ r  ðp sin φ þ q cos φÞcotanθ,
θ_ ¼ p cos φ  q sin φ,
ψ_ ¼ ðp sin φ þ q cos φÞcosecθ:

The dynamic equations in (4.61) written on the basis of general equations (1.27)
are composed in terms of projections on the principal axes of inertia of the body.
Here, p, q, r are the projections of the angular velocity vector on these axes; εLi,
4.8 Equations of Perturbed Rotational Motion of a Rigid Body Close to. . . 63

i ¼ 1 , 2 , 3, are the projections of the vector of perturbation torques on the same

axes; ψ, θ, φ are the Euler angles; ε is a small parameter characterizing the pertur-
bation magnitude. In particular, for ε ¼ 0 system, (4.61) describes a motion in
Lagrange’s case (see also (3.2)). Here, m is the mass of the body, g is the gravi-
tational acceleration, l is the distance from the fixed point O to the center of gravity
of the body, and A1 is the equatorial and A3 the axial moments of inertia of the
body relative to the fixed point O.
The problem is formulated of studying the behavior of the solution of system
(4.61) for the values of small parameter ε different from zero on a sufficiently large
interval of time t ~ ε1.
The necessary relations for the unperturbed motion, i.e., for ε ¼ 0, are given in
Sect. 3.1. The first integrals of the equations of motion for the unperturbed system
(4.61) are the following: G1 is the projection of the vector of angular momentum on
the vertical Oz1, H is the total energy of the body, and r is the projection of the
angular velocity vector on the axis of dynamic symmetry. They are represented by
formulas (3.7), (3.9), and (3.10).
The expression for the angle of nutation θ in the unperturbed motion as a func-
tion of time t, integrals of motion, and arbitrary phase constant β is known from
(3.29). The formula for the dependence of the squared modulus of elliptic functions
k2 on the real roots u1, u2, u3 of the cubic polynomial Q(u) (see (3.18)) is introduced
according to (3.24).
The relations between the roots of polynomial Q(u) and the first integrals can be
written, according to Vieta’s theorem, in the following way:

H A3 r 2 A2 r 2
u1 þ u 2 þ u3 ¼  þ 3 ,
mgl 2mgl 2A1 mgl
G1 A3 r
u1 u2 þ u1 u3 þ u2 u3 ¼  1, ð4:62Þ
A1 mgl
H A3 r 2 G21
u1 u2 u3 ¼  þ :
mgl 2mgl 2A1 mgl

Let us reduce the equations of the perturbed motion (4.61) to the form allowing
application of the averaging method. To this end, we identify slow and fast vari-
ables. In the considered problem, the first integrals (3.7), (3.9), and (3.10) will be
slow variables for the perturbed motion (4.61). The fast variables are the angles of
proper rotation φ, nutation θ, and precession ψ.
Using relations (3.7), (3.9), and (3.10) as the transformation formulas from vari-
ables ( p, q, r, φ, θ, ψ) to variables (G1, H, r, φ, θ, ψ), we reduce the first three equa-
tions (4.61) to the form
64 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

G_ 1 ¼ ε½ðL1 sin φ þ L2 cos φÞ sin θ þ L3 cos θ,

H_ ¼ εðL1 p þ L2 q þ L3 r Þ, ð4:63Þ
r_ ¼ εA1 3 L3 , Li ¼ Li ðp; q; r; φ; θ; ψ Þ, i ¼ 1, 2, 3:

Here and in the three last (kinematic) equations (4.61), it is assumed that vari-
ables p, q, r are expressed, using relations (3.7), (3.9), and (3.10) as the functions of
G1, H, r, φ, θ, ψ and substituted into (4.61) and (4.63). The initial values of
slow variables G1, H, r can be calculated with the help of (3.7), (3.9), and (3.10).
The right-hand sides of equations (4.63) contain three fast variables, which
create difficulties for the application of the averaging method connected with the
possibility of appearance of nonlinear resonances. To eliminate these difficulties,
we require that the right-hand sides of equations (4.63) for slow variables actually
depend just on a single fast variable, the angle of nutation θ, and are periodic func-
tions of θ with the period 2π. Then equations (4.63) can be averaged with respect to
θ, and the equations of the first approximation can be obtained. It turns out that a
number of applied problems possess the indicated property and allow averaging
with respect to one variable, the angle of nutation θ.
Let us present some sufficient conditions for the possibility of averaging the
equations (4.63) only with respect to the angle of nutation θ. Under the fixed values
of slow variables, the right-hand sides of equations (4.63), which are to be averaged,
contain the combinations of the following type:

L1 sin φ þ L2 cos φ, L1 p þ L2 q, L3 ,

where Li do not depend on t.

We require that, by means of identical transformations, these combinations can
be represented as functions of slow variables and the angle of nutation θ be periodic
in θ with the period 2π, i.e.,

L1 sin φ þ L2 cos φ ¼ L∗ 1 ðG1 ; H; r; θ Þ,

L1 p þ L2 q ¼ L∗ 2 ð G 1 ; H; r; θ Þ, ð4:64Þ
L3 ¼ L∗3 ð G 1 ; H; r; θÞ:

Note that relations (3.7) and (3.9) lead to the equalities

G1  A3 r cos θ
p sin φ þ q cos φ ¼ ,
A1 sin θ
ð4:65Þ
2ðH  mgl cos θÞ  A3 r 2
p2 þ q2 ¼ :
A1

Therefore, the combinations of the type (4.65) are expressed only through
slow variables and the angle of nutation θ, and, in addition, they are periodic in θ
with the period 2π. Therefore, combinations (4.65) are reduced to the form (4.64).
Comparing relations (4.64) and (4.65), we see that if the perturbation torques Li
satisfy the conditions
4.8 Equations of Perturbed Rotational Motion of a Rigid Body Close to. . . 65

L1 ¼ pf , L2 ¼ qf , L3 ¼ L∗
3 ð4:66Þ

or the conditions

L1 ¼ F sin φ, L2 ¼ F cos φ, L3 ¼ L∗
3 ð4:67Þ

where the arbitrary functions f, F, L∗

3 have the form

f ¼ f ðG1 ; H; r; θÞ,
F ¼ FðG1 ; H; r; θÞ, ð4:68Þ
L∗
3 ¼ L∗ 3 ðG1 ; H; r; θÞ

and are periodic in θ with the period 2π, then the imposed requirements (4.64) are
satisfied.
Thus, the sufficient conditions for the possibility of averaging of slow variables
of system (4.63) with respect to the angle of nutation θ for the perturbed Lagrange’s
motion are the requirements (4.66) or (4.67), imposed on the applied torques. In
what follows, we assume the general (necessary and sufficient) conditions (4.64) or,
in particular, the sufficient conditions (4.66) or (4.67) (together with (4.68)) are
satisfied, which ensure that relations (4.64) hold. Then the system (4.63) of equa-
tions of the perturbed motion of a rigid body close to Lagrange’s case can be
represented in the form

G_ 1 ¼ εF1 ðG1 ; H; r; θÞ, F1 ¼ L∗ ∗

1 sin θ þ L3 cos θ,
H_ ¼ εF2 ðG1 ; H; r; θÞ, F2 ∗ ∗
¼ L2 þ L3 r, ð4:69Þ
r_ ¼ εF3 ðG1 ; H; r; θÞ F3 ¼ A1 ∗
3 L3 :

Here F1, F2, F3 are 2π-periodic functions of θ.

The further study of this system will be conducted in Chap. 11.

4.8.2 The Case Where the Projections of the Perturbation

Torque Vector Are of Different Orders of Smallness

Consider another possible variant of application of the averaging method for the
perturbed motion in the case close to Lagrange’s case. The equations of motion of a
dynamically symmetric rigid body about a fixed point O under the action of
restoring and perturbation torques (dynamic and kinematic Euler’s equations)
have the form
66 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

A1 p_ þ ðA3  A1 Þqr ¼ k sin θ cos φ þ L1 ,

A1 q_ þ ðA1  A3 Þpr ¼ k sin θ sin φ þ L2 ,
A3 r_ ¼ L3 , Li ¼ Li ðp; q; r; φ; θ; ψ Þ, ði ¼ 1; 2; 3Þ,
ð4:70Þ
ψ_ ¼ ðp sin φ þ q cos φÞcosecθ,
θ_ ¼ p cos φ  q sin φ,
φ_ ¼ r  ðp sin φ þ q cos φÞcotanθ:

Dynamic equations (4.70) are written in the projections on the principal axes of
inertia of the body passing through the point O. Here p, q, r are the projections of the
angular velocity vector of the body on these axes; Li, i ¼ 1, 2, 3 are the projections
of the perturbation torque vector on the same axes, which are periodic functions of
the Euler angles ψ, θ, φ with the periods 2π; ψ is the angle of precession; θ is the
angle of nutation; φ is the angle of proper rotation; and A1 is the equatorial and A3
the axial moments of inertia of the body relative to the point O, A1 6¼ A3. It is
assumed that the body is acted upon by the restoring torque, the maximum value of
which is equal to k and which is created by a constant in magnitude and direction
force applied at a fixed point of the axis of dynamic symmetry. In the case of a
heavy top, we have
k ¼ mgl: ð4:71Þ

Here m is the mass of the body, g is the gravitational acceleration, and l is the
distance from the fixed point O to the center of gravity of the body. The restoring
torque can also be caused by aerodynamic forces.
The perturbation torques Li in (4.70) are assumed to be given functions of their
arguments. For Li ¼ 0, i ¼ 1, 2, 3, equations (4.70) correspond to Lagrange’s case.
Let us make the following basic assumptions:
 
p2 þ q2  r 2 , A3 r 2
k, Li   k, i ¼ 1, 2, L3  k, ð4:72Þ

which means that the direction of the angular velocity of the body is close to the
axis of dynamic symmetry; the angular velocity is high enough, so that the kinetic
energy of the body is much greater than the potential energy due to the restoring
torque; the two projections of the perturbation torque vector on the principal axes of
inertia of the body are small compared with the restoring torque, whereas the
third one is of the same order of magnitude. Inequalities (4.72) allow introducing
a small parameter ε and putting
p ¼ εP, q ¼ εQ, k ¼ εK, ε  1,
Li ¼ ε2 L∗
i ðP; Q; r; ψ; θ; φÞ, i ¼ 1, 2, ð4:73Þ
L3 ¼ εL∗ 3 ð P; Q; r; ψ; θ; φ Þ:

New variables P, Q, as well as the variables and constants r, ψ, θ, φ, K, A1, A3,

L∗
i , i ¼ 1, 2, 3, are assumed to be bounded quantities of the order of unity as ε ! 0.
4.8 Equations of Perturbed Rotational Motion of a Rigid Body Close to. . . 67

The problem is formulated to study the asymptotic behavior of the solution of

system (4.70) with a small ε, provided that the conditions (4.72) and (4.73) are
fulfilled. The method of averaging will be used. The set of the simplifying assump-
tions (4.72) or (4.73) made allows obtaining a relatively simple, in general, aver-
aging scheme and exploring a number of examples.
Let us perform in system (4.70) the change of variables (4.73). Canceling both
sides of the first two equations (4.70) by ε, we get

A1 P_ þ ðA3  A1 ÞQr ¼ K sin θ cos φ þ εL1 ∗ ,

A1 Q_ þ ðA1  A3 ÞPr ¼ K sin θ sin φ þ εL2 ∗ ,
A3 r_ ¼ εL3 ∗ ,
ð4:74Þ
ψ_ ¼ εðP sin φ þ Q cos φÞcosecθ,
θ_ ¼ εðP cos φ  Q sin φÞ,
φ_ ¼ r  εðP sin φ þ Q cos φÞcotanθ:

Consider the system of zero-order approximation and put ε ¼ 0 in (4.74).

Then from the last four equations (4.74), we obtain

r ¼ r0 , ψ ¼ ψ 0, θ ¼ θ0 , φ ¼ r 0 t þ φ0 : ð4:75Þ

Here, r0, ψ 0, θ0, φ0 are constants equal to the initial values of the corresponding
variables for t ¼ 0. Substitute equalities (4.75) into the first two equations of system
(4.74) for ε ¼ 0 and integrate the resulting system of two linear equations for P, Q.
The solution can be represented in the form

P ¼ a cos γ 0 þ b sin γ 0 þ KA1 1

3 r 0 sin θ 0 sin ðr 0 t þ φ0 Þ,
1 1
Q ¼ a sin γ 0 þ b cos γ 0 þ KA3 r 0 sin θ0 cos ðr 0 t þ φ0 Þ,
a ¼ P0  KA1 1
3 r 0 sin θ 0 sin φ0 , ð4:76Þ
1 1
b ¼ Q0 þ KA3 r 0 sin θ0 cosφ0 , 
γ 0 ¼ n0 t, n0 ¼ ðA3  A1 ÞA1 n0 =r 0 
1:
1 r 0 6¼ 0,

Here, P0, Q0 are the initial values of new variables P, Q, introduced in accord-
ance with (4.73), whereas the variable γ ¼ γ 0 has the meaning of oscillation phase.
System (4.74) is strongly nonlinear (the natural oscillation frequency of the vari-
ables P, Q depends on the slow variable r), so further we introduce an additional
variable γ defined by the equation

γ_ ¼ n, γ ð0Þ ¼ 0, n ¼ ðA3  A1 ÞA1

1 r: ð4:77Þ

For ε ¼ 0, we have γ ¼ γ 0 ¼ n0t in accordance with (4.76). Equalities (4.75) and

(4.76) determine the general solution of the system (4.74) and (4.77) for ε ¼ 0.
Taking into account (4.75) and eliminating the constants, the first two relations
(4.76) can be rewritten in the equivalent form
68 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

P ¼ a cos γ þ b sin γ þ KA1

3 r
1
sin θ sin φ,
1 1 ð4:78Þ
Q ¼ a sin γ  b cos γ þ KA3 r sin θ cos φ

and solved with respect to a,b:

a ¼ P cos γ þ Q sin γ  KA1

3 r
1
sin θ sin ðγ þ φÞ,
1 1 ð4:79Þ
b ¼ P sin γ  Q cos γ þ KA3 r sin θ cos ðγ þ φÞ:

Relations (4.78) and (4.79) can be considered as the formulas of the variable
change (containing the variable γ), determining the transition from the variables P,
Q to the variables of the Van der Pol-type [1] a, b and conversely. Using these
formulas, we make a transition in the systems (4.74) and (4.77) from variables P, Q,
r, ψ, θ, φ, γ to new variables a, b, r, ψ, θ, α, γ, where

α ¼ γ þ φ: ð4:80Þ

After the transformations, we obtain a more convenient for the further study
system of seven equations (instead of six (4.74)):
 
a_ ¼ εA1 1 L∘1 cos γ þ L∘2 sin γ εKA3 1 r 1
cos θ b  KA3 1 r 1 sin θ cos α þ εKA3 2 r 2 L∘3 sin θ sin α,
 
b_ ¼ εA1 1 L∘1 sin γ þ L∘2 cos γ þ εKA3 1 r 1
cos θ a þ KA3 1 r 1 sin θ sin α  εKA3 2 r 2 L∘3 sin θ cos α, ð4:81Þ
r_ ¼ εA3 1 L∘3 , θ_ ¼ εða cos α þ b sin αÞ,
ψ_ ¼ εcosecθða sin α  b cos αÞ þ εKA3 1 r 1 ,
α_ ¼ A3 A1 1 r  εcotanθða sin α  b cos αÞ  εKA3 1 r 1 cos θ,
γ_ ¼ ðA3  A1 ÞA1 1 r:

Here, by L0i , we denote the functions obtained from L∗

i (see (4.73)) as a result of
the performed substitution (4.78)–(4.80), i.e.,

L0i ða; b; r; ψ; θ; α; γ Þ ¼ L∗
i ðP; Q; r; ψ; θ; φÞ, i ¼ 1, 2, 3: ð4:82Þ

Note that the transition from two variables P, Q to three variables a, b, γ is

motivated by the considerations of convenience: for ε ¼ 0, the system for P, Q has a
linear form, whereas the substitution (4.78) is nonsingular for all a,b.
Introduce a vector x, the components of which are the slow variables a, b, r, ψ, θ
of system (4.81). Then this system can be written in the form

x_ ¼ εXðx; α; γ Þ, α_ ¼ A3 A1
1 r þ εY ðx; αÞ, ð4:83Þ
γ_ ¼ ðA3  A1 ÞA1
1 r, xð0Þ ¼ x0 , αð0Þ ¼ α0 , γ ð0Þ ¼ 0:
4.8 Equations of Perturbed Rotational Motion of a Rigid Body Close to. . . 69

Here, the vector function X and scalar function Y are determined by the right-
hand sides of equations (4.81); the initial conditions are obtained according to
(4.75)–(4.77).
Consider the system (4.81) or (4.83) from the point of view of application of the
averaging method [1–3]. The system (4.81) contains slow variables a, b, r, ψ, θ and
two fast variables, the phases α, γ, while γ appears only in the first three equations
(4.81). The system is nonlinear, and the direct application of the averaging method
is very difficult [23].
The solution of this system will be carried out in Chap. 11.

4.8.3 Perturbation Torques Are Small Compared

to the Restoring Ones

As in the previous Sect. 4.8.2, we consider the motion of a dynamically symmetric

rigid body about a fixed point O under the action of restoring and perturbation
torques. The equations of motion have the form (4.70). All the notations of
Sect. 4.8.2 are kept. In this subsection, the following assumptions are made:
 
p2 þ q2  r 2 , A3 r 2
k, Li   k, i ¼ 1, 2, 3, ð4:84Þ

which mean that (a) the direction of the angular velocity of the body is close to the
axis of dynamic symmetry; (b) the angular velocity is large enough, so that the
kinetic energy of the body is much greater than the potential energy due to restoring
torque; and (c) perturbation torques are small in comparison with the restoring ones.
Unlike Sect. 4.8.2 (see (4.72)), here we assume that all (not just the first two)
projections of the perturbation torque vector on the principal axes of inertia of the
body are small compared to the restoring torque.
Inequalities (4.84) allow introducing a small parameter ε  1 and set

p ¼ εP, q ¼ εQ, k ¼ εK,

ð4:85Þ
Li ¼ ε2 L∗
i ðP; Q; r; ψ; θ; φÞ, i ¼ 1, 2, 3:

As in Sect. 4.8.2, new variables P, Q and the variables and constants r, ψ, θ, φ, K,

A1, A3, L∗i are assumed to be bounded by the quantities of the order of unity as
ε ! 0. Our goal is to study the asymptotic behavior of system (4.70) with a small ε,
provided the conditions (4.84) and (4.85) are satisfied. We will use the method of
averaging on the time interval of the order of ε1.
Let us perform the change of variables (4.85) in system (4.70). By reducing
both sides of the first two equations (4.70) by ε, we obtain
70 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

A1 P_ þ ðA3  A1 ÞQr ¼ K sin θ cos φ þ εL1 ∗ ,

A1 Q_ þ ðA1  A3 ÞPr ¼ K sin θ sin φ þ εL2 ∗ ,
A3 r_ ¼ ε2 L3 ∗ ,
ð4:86Þ
ψ_ ¼ εðP sin φ þ Q cos φÞcosecθ,
θ_ ¼ εðP cos φ  Q sin φÞ,
φ_ ¼ r  εðP sin φ þ Q cos φÞcotanθ:

All equations of the systems (4.86) and (4.74), besides the third one, coincide.
Consider first the zero-order approximation system and set ε ¼ 0. In this case,
from the last four equations (4.86), we obtain formulas (4.75). Substitute expression
(4.75) into the first two equations of system (4.86) for ε ¼ 0 and integrate the
resulting system of equations for P, Q. The solution can be represented in the
form (4.76). We introduce again an additional variable γ, defined by equation
(4.77).
For ε ¼ 0, it follows from (4.76) that γ ¼ γ 0 ¼ n0t. The general solution of system
(4.86) and (4.77) with ε ¼ 0 is determined by equalities (4.75) and (4.76). Elimi-
nating the constants and taking into account (4.75), the first two expressions (4.76)
can be rewritten in the form (4.78). Solving equalities (4.78) with respect to a and b,
we obtain formulas (4.79).
We introduce a new variable δ as follows:

r ¼ r 0 þ εδ: ð4:87Þ

Now, let us turn to system (4.86) for ε 6¼ 0 and consider relations (4.78), (4.79),
and (4.87) as the formulas of the variable change, determining the transition from
the variables P, Q, r to the variables a, b, δ (in addition, a new variable γ is included
in these formulas). Using these formulas, we make a transition in the systems (4.86)
and (4.77) from variables P, Q, r, ψ, θ, φ, γ to new variables α, b, δ, ψ, θ, α, γ,
where α is determined according to (4.80).
After the transformations, we obtain a system of seven equations (instead of
six equations (4.86)) which is more convenient for studying:
 
a_ ¼ εA1
1 L1 cos γ þ L  2 sin γ  
εKA3 r 0 cos θ b KA1
1 1 1
3 r 0 sin θ cos α þ 
þε2 KA1 2 1 1
3 r 0 δ cos θ b  2KA3 r 0 sin θ cos α þ
2 2
þε KA
2
 3 r 0 L3 sin θ sin α,

b_ ¼ εA1
1 L1 sin γ  L2 cos γ þ 
þεKA1 1 1 1
3 r 0 cos θ aþ KA3 r 0 sin θ sin α  
ε2 KA1 2 1 1
3 r 0 δ cos θ a þ 2KA3 r 0 sin θ sin α 
2 2
ε KA3 r 0 L3 sin θ cos α,
2
References 71

δ_ ¼ εA3 1 L∘3 , θ_ ¼ εða cos α þ b sin αÞ,

ψ_ ¼ εcosecθða sin α  b cos αÞ þ εKA3 1 r 0 1  ε2 KA3 1 r 0 2 δ,
α_ ¼ A3 A1 1 r 0 þ εA3 A1 1 δ  εcotanθða sin α  b cos αÞ ð4:88Þ
εKA3 1 r 0 1 cos θ þ ε2 KA3 1 r 0 2 δ cos θ,
γ_ ¼ n0 þ εðA3  A1 ÞA1 1 δ:

Here, by Li , we denote the functions obtained from L∗
i (see (4.85)) as a result of
substitution of (4.78)–(4.80) and (4.87):

Li ða; b; δ; ψ; θ; α; γ Þ ¼ L∗
i ðP; Q; r; ψ; θ; φÞ, i ¼ 1, 2, 3: ð4:89Þ

The considered system of equations (4.88) can be reduced to the form

x_ ¼ εF1 ðx; yÞ þ ε2 F2 ðx; yÞ, x ð 0Þ ¼ x0 ,

y_ 1 ¼ ω1 þ εg1 ðx; yÞ þ ε2 g2 ðx; yÞ, y 1 ð 0Þ ¼ y10 , ð4:90Þ
y_ 2 ¼ ω2 þ εh1 ðx; yÞ þ ε2 h2 ðx; yÞ, y 2 ð 0Þ ¼ y20 :

Here, the vector-function x ¼ (x1, . . . , x5) is composed of slow variables a, b, δ,

ψ, θ; by y1 and y2, we denote fast variables (phases) α, γ; ω1, ω2 are constant fre-
quencies equal to A3 A1 1
1 r 0 and ðA3  A1 ÞA1 r 0 , respectively. Vector functions Fi,
gi, hi, i ¼ 1, 2 are determined by the right-hand sides of equations (4.88).
Further study of systems (4.88) and (4.90) is carried out in [24].

References

1. Bogolubov, N.N., Mitropolsky, Y.A.: Asymptotic Methods in the Theory of Nonlinear Oscil-
lations. Gordon and Breach Science Publishers, New York (1961)
2. Volosov, V.M., Morgunov, B.I.: The Averaging Method in the Theory of Non-linear Oscil-
latory Systems. Moscow State Univ., Moscow (1971) in Russian
3. Mitropolsky, Yu.A.: The Method of Averaging in Nonlinear Mechanics. Naukova Dumka,
Kiev (1971) in Russian
4. Demin, V.G., Konkina, L.I.: New Methods in Dynamics of a Rigid Body. Ilim, Frunze (1989)
in Russian
5. Bulgakov, B.V.: Applied Theory of Gyroscopes, 3rd edn. Moscow State Univ., Moscow
(1976) in Russian
6. Beletsky, V.V.: Motion of an Artificial Satellite about its Center of Mass. Israel Program for
Scientific Translation, Jerusalem (1966)
7. Beletsky, V.V.: Spacecraft Attitude Motion in Gravity Field. Moscow State Univ., Moscow
(1975) in Russian
8. Chernousko, F.L.: On the motion of a satellite about its center of mass under the action of
gravitational moments. J. Appl. Math. Mech. 27(3), 708–722 (1963)
9. Moiseev, N.N.: Asymptotic Methods of Nonlinear Mechanics. Nauka, Moscow (1981) in
Russian
10. Akulenko, L.D.: Asymptotic Methods of Optimal Control. Nauka, Moscow (1987) in Russian
11. Akulenko, L.D.: Problems and Methods of Optimal Control. Kluwer, Dordrecht (1994)
72 4 Equations of Perturbed Motion of a Rigid Body About Its Center of Mass

12. Akulenko, L.D.: Higher-order averaging schemes in systems with fast and slow phases.
J. Appl. Math. Mech. 66(2), 153–163 (2002)
13. Arnold, V.I., Kozlov, V.V., Neishtadt, A.I.: Mathematical Aspects of Classical and Celestial
Mechanics. Springer, Berlin (2007)
14. Neishtadt, A.I.: On the separation of motions in the systems with the fast-rotating phase.
Prikl. Mat. Mekh. 48(2), 197–204 (1984) in Russian
15. Akulenko, L.D.: Higher-order averaging schemes in the theory of non-linear oscillations.
Prikl. Mat. Mekh. 65(5), 843–853 (2001) in Russian
16. Chernousko, F.L.: Motion of a rigid body with cavities filled with viscous fluid at
small Reynolds number. USSR Comput. Math. Math. Phys. 5(6), 99–127 (1965)
17. Chernousko, F.L.: Motion of a Rigid Body with Viscous-Fluid-Filled Cavities. Computing Center
AN SSSR, Moscow (1968) in Russian
18. Chernousko, F.L.: The Movement of a Rigid Body with Cavities Containing a Viscous Fluid.
NASA, Washington (1972)
19. Lamy, P., Burns, J.: Geometrical approach to torque free motion of a rigid body having
internal energy dissipation. Am. J. Phys. 40(3), 441–445 (1972)
20. Klimov, D.M., Kosmodem’yanskaya, G.V., Chernousko, F.L.: Motion of a gyroscope with
contactless suspension. Izv. Akad. Nauk SSSR. Mekh. Tverd. Tela. 2, 3–8 (1972) in Russian
21. Appel, P.: Traite de Mechanique Rationnelle. Gauthier – Villars, Paris (1953)
22. Landau, L.D., Lifshitz, E.M.: Course of Theoretical Physics, Mechanics, vol. 1. Pergamon Press,
Oxford (1976)
23. Arnold, V.I.: Geometrical Methods in the Theory of Ordinary Differential Equations. Springer,
London (2012)
24. Leshchenko, D.D., Shamaev, A.S.: Perturbed rotational motions of a rigid body that are
close to regular precession in the Lagrange case. Mech. Solids. 22(6), 6–15 (1987)
Chapter 5
Perturbation Torques Acting upon
a Rigid Body

This chapter examines the internal and external torques of various physical natures,
which can act as perturbation torques during the motion of a rigid body relative to
the center of mass. The dynamics of rigid bodies under the influence of these
torques will be investigated in subsequent chapters.

5.1 Gravitational Torques Acting upon a Satellite

Consider the motion of an artificial Earth satellite relative to the center of mass
under the action of gravitational torque. Let R be the distance from the center of
gravitation (center of mass of the Earth) to the center of inertia of the satellite, and
let l be a characteristic size of the satellite. Suppose that l
R and neglect the
quantities of the orders of magnitude higher than (l/R)2. Then the motion of the
satellite relative to the center of mass virtually does not affect the motion of the
center of mass. The center of mass of the satellite moves along an elliptical orbit,
the equation of which has the form:
p0
R¼ , ð5:1Þ
1 þ e cos v

where, as shown in Fig. 5.1, p0 is the focal parameter, e is the eccentricity of the
orbit, R and v are the polar coordinates describing the position of the center of mass
of the satellite relative to the center of gravitation which is situated at the focus F of
the ellipse, and O is the center of mass of the satellite. In celestial mechanics, the
angle v is called the true anomaly.
The time dependence of the true anomaly v is given by the relation (see (4.52)):

© Springer International Publishing AG 2017 73

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_5
74 5 Perturbation Torques Acting upon a Rigid Body

Fig. 5.1 Elliptic orbit of

the satellite

" #
3 1=2
dv ω0 ð1 þ e cos vÞ2 2π μð1  e2 Þ
¼ , ω0 ¼ : ð5:2Þ
dt ð1  e2 Þ3=2 Q0 p30

Here, ω0 is the average angular velocity of motion of the center of mass along the
elliptical orbit, Q0 is the period of revolution of the satellite, and μ is the product of
the gravitational constant and the mass of the Earth.
We introduce a coordinate system Oxyz connected with the satellite (Fig. 5.2):
the axis Oz is directed along the current radius vector of the orbit, the axis Oy is
parallel to the normal to the orbital plane, and the axis Ox is perpendicular to Oy and
Oz.
The element of the satellite with the mass dm and coordinates x, y, z is acted upon
by the Newton force dF, directed to the center of gravitation:

μdm
dF ¼  r0 : ð5:3Þ
x2 þ y2 þ ð z þ RÞ 2

Here, the unit vector r0 is determined by the direction cosines:

x y zþR
cos ðx,
d r0 Þ ¼ , cos ðy,
d r0 Þ ¼ , cos ðz, r0 Þ ¼
d ,
r r
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi r ð5:4Þ
r ¼ x2 þ y2 þ ðz þ RÞ2 :

Since the satellite dimensions are small compared to the distance R to the center
of gravitation, we can assume that x, y, z are small compared with R. Then up to the
terms of the second order of smallness, the projections of the force dF on the
coordinate axes Ox, Oy, Oz are expressed as follows:

μdm x
dFx ¼  2  ,
R R
μdm y
dFy ¼  2  , ð5:5Þ
R R
μdm 2μdm z
dFz ¼  2 þ 2  :
R R R
5.1 Gravitational Torques Acting upon a Satellite 75

Fig. 5.2 Coordinate system

connected with the satellite

The expression μdmR2

in dFz from (5.5) equals the elementary force applied to the
element dm, which causes the motion of the center of mass of the satellite along the
elliptical orbit. The force:
 
μxdm μydm 2μzdm
dF1 ¼  ; 3 ;
R3 R R3

creates the torque dL0 relative to the center of mass O of the satellite. Let us
calculate this torque by the formula dL0 ¼ ρ  dF1, where ρ{x, y, z} is the radius
vector of the element dm relative to the point O. Using formulas (5.5), we get:
 
dL0 ¼ dLx ; dLy ; dLz ,
μdm μdm ð5:6Þ
dLx ¼ 3yz 3 , dLy ¼ 3xz 3 , dLz ¼ 0:
R R

To find gravitational torque L0 acting on the satellite, it is necessary to integrate

expression (5.6) over the volume of the satellite. To do this, we first go to coordinate
system Oz1z2z3 whose axes are directed along the principal axes of inertia of the
satellite [1, 2]. We describe the relative position of coordinate systems Oxyz and
Oz1z2z3 by the table of direction cosines:

z1 z2 z3
x α α0 α00
00 : ð5:7Þ
y β β0 β
z γ γ0 γ 00

Note the following property of the matrix of direction cosines (5.7): each
00
element of the matrix equals its algebraic complement [3, 4]. For instance, α0 ¼ β
γ  βγ 00 .
Let us find the projections of gravitational torque dL0 on axes Ozi, i ¼ 1 , 2 , 3.
The projection on axis Oz1 in view of (5.6) and (5.7) equals:
76 5 Perturbation Torques Acting upon a Rigid Body

3μdm 3μdm
dLz1 ¼ ðyzα  xzβÞ ¼ zðyα  xβÞ:
R3 R3

Note that:
 00 
yα  xβ ¼ z1 β þ z2 β0 þ z3 β α  ðz1 α þ z2 α0 þ z3 α00 Þβ ¼
 00 
¼ z2 ðβ0 α  αβ0 Þ  z3 α00 β  β α ¼ z2 γ 00  z3 γ 0

according to the mentioned property of the matrix of direction cosines. Thus:

3μdm
dLz1 ¼ ðz1 γ þ z2 γ 0 þ z3 γ 00 Þðz2 γ 00  z3 γ 0 Þ:
R3

Integrating over the volume V of the satellite, we obtain:

ð
3μ  00 2

z1 z2 γγ 00 þ z22 γ 0 γ 00 þ z2 z3 γ  z1 z3 γ 0  z2 z3 γ 0  z23 γ 0 γ 00 dm:
2
Lz1 ¼ 3 ð5:8Þ
R
V

Since the axes connected with the rigid body are directed along the principal
central axes of inertia, we have:
ð ð ð
z1 z2 dm ¼ z2 z3 dm ¼ z1 z3 dm ¼ 0: ð5:9Þ
V V V

The principal central moments of inertia equal:

ð ð ð
     
A1 ¼ z22 þ z23 dm, A2 ¼ z21 þ z23 dm, A3 ¼ z21 þ z22 dm: ð5:10Þ
V V V

Substituting relations (5.9), (5.10) into integral (5.8), we get:

μ
Lz1 ¼ 3 ðA3  A2 Þγ 0 γ 00 : ð5:11Þ
R3

Similar calculations lead to the following expressions for Lz2 , Lz3 [1]:
μ μ
Lz2 ¼ 3 ðA1  A3 Þγ 00 γ, Lz3 ¼ 3 ðA2  A1 Þγγ 0 : ð5:12Þ
R3 R3

Introducing the basis vectors z1, z2, z3 of coordinate system Oz1z2z3, we can
rewrite (5.11), (5.12) as follows:
5.1 Gravitational Torques Acting upon a Satellite 77

L0 ¼ Lz1 z1 þ Lz2 z2 þ Lz3 z3

3μ
¼ ½ðA3  A2 Þγ 0 γ 00 z1 þ ðA1  A3 Þγ 00 γz2 þ ðA2  A1 Þγγ 0 z3 : ð5:13Þ
R3

Here γ, γ 0 , γ 00 are direction cosines of the radius vector R of the satellite’s center
of inertia, drawn from the fixed center of gravitation, with respect to the principal
central axes of inertia Ozi, i ¼ 1, 2, 3. Introduce an auxiliary vector:

I ¼ A1 γz1 þ A2 γ 0 z2 þ A3 γ 00 z3 : ð5:14Þ

Then equality (5.13) can be written in the form:

L0 ¼ 3μR4 ðR  IÞ: ð5:15Þ

In Chap. 4, a system of coordinates Oy1y2y3 (Fig. 4.1) was introduced, and also
equations (4.42), (4.46) of motion of the satellite were written relative to the
center of inertia in the projections on the axes of coordinate system Oy1y2y3.
We will project the vector equality (5.15) on axes Oyi. Denote by σ i the direction
cosines of the radius vector R with respect to axes Oyi. The following relations
hold:

γ ¼ α11 σ 1 þ α21 σ 2 þ α31 σ 3 ,

γ 0 ¼ α12 σ 1 þ α22 σ 2 þ α32 σ 3 ,
γ 00 ¼ α13 σ 1 þ α23 σ 2 þ α33 σ 3 :

Here, αij are the direction cosines determining orientation of the coordinate
system Oz1z2z3 connected to the satellite relative to the system of coordinates
Oy1y2y3 (see, e.g., expressions (1.31)).
The projections of vector I from (5.14) on axes Oyi equal:

I i ¼ Iyi ¼ si1 σ 1 þ si2 σ 2 þ si3 σ 3 , i ¼ 1, 2, 3, ð5:16Þ

where we introduce the notation:

sij ¼ A1 αi1 αj1 þ A2 αi2 αj2 þ A3 αi3 αj3 : ð5:17Þ

Expanding the vector product (5.15) with the help of (5.17) and eliminating
R and μ taking into account equality (5.1) and the second expression from (5.2),
we finally obtain:
78 5 Perturbation Torques Acting upon a Rigid Body

3ω20 ð1 þ e cos vÞ3 X

3  
L1 ¼ σ 2 σ j s3j  σ 3 σ j s2j ,
ð1  e 2 Þ3 j¼1
3ω20 ð1 þ e cos vÞ3 X
3  
L2 ¼ σ 3 σ j s1j  σ 1 σ j s3j , ð5:18Þ
ð1  e 2 Þ3 j¼1
3X
3ω20 ð1 þ e cos vÞ 3  
L3 ¼ σ 1 σ j s2j  σ 2 σ j s1j :
ð1  e 2 Þ3 j¼1

To calculate the direction cosines σ i of the radius vector with axes Oyi, we note
that the radus vector lies in the orbital plane Ox1x2 and form the angle of true ano-
maly v with axis Ox1. It immediately follows that (see Fig. 4.1):

σ 1 ¼ cos δ cos ðv  λÞ, σ 2 ¼ sin ðv  λÞ, σ 3 ¼ sin δ cos ðv  λÞ: ð5:19Þ

5.2 Rigid Body in a Resistant Medium

The presence of the velocity of proper rotation of the body leads to the appearance
of dissipative torques causing the braking of the body rotation. These torques
depend on the properties of resistant medium in which the motions occur, on the
body shape, on the properties of the surface and the distribution of mass in the body,
but also on the characteristics of the body motion. Therefore, the dependence of the
resistance torque Lr on the orientation of the body and its angular velocity can be
quite complicated and, in the general case, requires consideration of the motion of
the medium around the body.
We confine ourselves here to some simple relations that can qualitatively
describe the resistance to rotation at small angular velocities and are used in the
literature [5–17].
The dependence of dissipative resistance torque Lr on the angular velocity
vector ω ¼ ( p, q, r) is often taken as linear:
0 1 0 1
D11 D12 D13 p
Lr ¼ D  ω, D ¼ @ D21 D22 D23 A, ω ¼ @ q A: ð5:20Þ
D31 D32 D33 r

Here Dij are constant coefficients of resistance torque.

Many authors take tensor D in (5.20) to be diagonal:

D ¼ diagðλ1 ; λ2 ; λ3 Þ:
5.3 Rigid Body with a Cavity Filled with the Fluid of High Viscosity 79

Here λ1, λ2, λ3 are constant coefficients of proportionality. In this case, we have,
according to (5.20), in the projections on the axes of the moving system of coordi-
nates Oxyz (where O is the fixed point):

Lx ¼ λ1 p, Ly ¼ λ2 q, Lz ¼ λ3 r:

In some studies, it is assumed that matrix D defining resistance torques in (5.20)

has the form:
0 1
μ1 0 0
D ¼ @ 0 μ2 0 AJ,
0 0 μ3

where μ1, μ2, μ3 are constant coefficients, while J is the tensor of inertia of the
rigid body. Thus, in this case, the components of resistance torque (5.20) equal:

Lxr ¼ μ1 A1 p, Lyr ¼ μ2 A2 q, Lzr ¼ μ3 A3 r, ð5:21Þ

where A1, A2, A3 are the principal central moments of inertia of the rigid body.

5.3 Rigid Body with a Cavity Filled with the Fluid of High
Viscosity

In [18–20], the motion of a rigid body with a cavity filled with the viscous fluid is
considered. The Reynolds number is assumed small: R ¼ l2 T 1 ∗ v
1

1. Here l is a
characteristic linear dimension of the cavity; T∗ is a characteristic time scale of
relative motion, inversely proportional to the characteristic angular velocity ω,
whereas v is the kinematic coefficient of viscosity of the fluid. If l and T∗ are
taken as the units of measurement of the length and time, then the kinematic
coefficient of viscosity is a large parameter, i.e., v ¼ 1/R  1. The velocity u of
the fluid in the system of coordinates connected with the rigid body is sought in the
form:

u ¼ w þ W; wðr; tÞ ¼ w0 ðr; tÞ þ v1 w1 ðr; tÞ þ v2 . . . ,

Wðr; τÞ ¼ W0 ðr; τÞ þ v1 W1 ðr; τÞ þ v2 . . . , ð5:22Þ
τ ¼ vt:

The pressure is sought in the form of a similar series. Here, r is the radius vector
measured from a point connected with the rigid body, t is time, and τ ¼ vt is the
“fast” time, whereas the superscripts indicate the approximation number. As a
result of substituting the series of the form (5.22) into the Navier–Stokes equations,
the boundary value problems for the coefficients of these series are obtained. The
examination shows that identically w0 ¼ 0 and everywhere, except for small initial
80 5 Perturbation Torques Acting upon a Rigid Body

interval of time, one can take u ¼ v1w1 + O(v1). Determination of function w1(r,
t) is reduced to solving three stationary boundary value problems:

Δzi ¼ ∇Si þ ei  r; div zi ¼ 0 in D zi ¼ 0 on Sði ¼ 1; 2; 3Þ ð5:23Þ

Here ei are the unit vectors of coordinate axes, D is the domain of the cavity, S is
its boundary, and zi and si are the desired functions depending only on the shape of
the cavity. The velocity and pressure of the fluid are expressed by these functions.
Angular momentum G of the body with fluid relative to the center of inertia of the
system can be represented in the form:

ρ dω
G¼J  ω P  ,
ð v dt
ð5:24Þ
Pij ¼  ej ðr  zi Þdv, i, j ¼ 1, 2, 3:
D

Here J is the inertia tensor of the system relative to the center of inertia; ω is the
vector of absolute angular velocity of the rigid body; ρ is the density of the fluid; the
dot denotes the scalar product of a tensor and a vector. The constant tensor
P depends only on the shape of the cavity and characterizes the energy dissipation
due to viscosity of the fluid, and Pij are its components in the coordinate system
connected with the body. It is shown in [18–20] that tensor P is symmetric and
corresponds to a positive-definite quadratic form. For the cavities having the shape
of a sphere, ellipsoid, cylinder, and some others, the works [18–20] provide a
solution of the boundary value problems (5.23), and the components of tensor
P are calculated.
Consider the problem of the influence of the viscous fluid in the cavity on the
dynamics of the rigid body. Tensor P is assumed to be known; the mass forces
acting on the fluid are assumed potential. The principal moment of external forces
L may depend on the coordinates and velocities of the rigid body as well as on time
and does not depend on the internal motions of the fluid.
Let h be a vector function of any dimension n, the components of which are the
kinematic parameters characterizing orientation of the rigid body (e.g., the Euler
angles or direction cosines), time t, and also the coordinates and velocities of the
center of inertia of the system, if the rigid body is not fixed.
In [18–20], the theorem of change of angular momentum G of the system is
reduced, after a number of transformations, to the form:

J  ε þ ω  ðJ  ωÞ  Lðω; hÞ ¼ m,
m ¼ ρv1 ½P  ε_ þ ω  ðP  εÞ, ω_ ¼ ε, ð5:25Þ
h_ ¼ f ðh; ωÞ:

Here ε is the angular acceleration of the body, and f is an n-dimensional vector

function; the last equation (5.25) includes all kinematic relations, the equations of
5.3 Rigid Body with a Cavity Filled with the Fluid of High Viscosity 81

motion of the center of inertia of the system and other equations which are not
associated with a cavity filled with the fluid.
For v1 ¼ 0, m ¼ 0, equations (5.25) describe the motion of the system under the
condition that the fluid in the cavity has solidified. These are the usual equations for
the coordinates and velocities, i.e., for the variables ω and h.
For v1 6¼ 0, system (5.25) contains a small parameter v1 at the derivatives ε_ . It
follows from the system that ε_  v; thus after a time of the order of unity, we get
jε_ j  v  1. However, the analysis of hydrodynamic equations, carried out in [18–
20], is based on the assumption that the angular acceleration ε and its derivatives ε_ ,
€ε are the quantities of the order of 1. Therefore, it makes sense to consider only
those solutions of system (5.25) that satisfy this assumption. In this case, we
disregard the fast processes of settling that are taking place during a small initial
interval of time and which have a character of the boundary layer with respect
to time.
Having in mind only the “slow” processes, in which ε  ε_  €ε are bounded
(of the order of 1) as v ! 1 and provided the inverse tensor J1 exists, we obtain
from the first equation of (5.25):

ε ¼ J1  ½Lðω; hÞ  ω  ðJ  ωÞ þ . . . ð5:26Þ

The omitted terms and their derivatives are of the order of v1. Differentiating
equality (5.26) and substituting ω_ and h_ from (5.25), we get:

dL dL
ε_ ¼ J1  εþ  f  ε  ðJ  ω Þ  ω  ðJ  ε Þ þ . . . ð5:27Þ
dω dh

dh are matrices of partial derivatives of the dimensions 3  3 and

dL
Here dω and dL
3  n, respectively, which are multiplied by the vectors of the corresponding
dimensions. In this case, one can substitute ε from (5.26) to (5.27) without decreas-
ing the order of accuracy.
We substitute equalities (5.26), (5.27) instead of ε, ε_ into equation (5.25) for m.
In so doing, we introduce an error of the order of v2 into the expression for m. As a
result, applying equalities (5.26), (5.27), system (5.25) can be substituted by the
following one:
J  ε þ ω  ðJ  ωÞ  Lðω; hÞ ¼ m,
m ¼ ρv1 ½P  b þ ω  ðP  aÞ,
1
a ¼ J  ½Lðω; hÞ  ω  ðJ  ωÞ,
ð5:28Þ
dL dL
b ¼ J1  aþ  f  a  ðJ  ωÞ  ω  ðJ  aÞ ,
dω dh
ω_ ¼ ε, h_ ¼ f ðh; ωÞ:

The resulting system has the same order as the equations of dynamics in the case
of solidification of the fluid. In it, m is a small perturbation torque depending on the
82 5 Perturbation Torques Acting upon a Rigid Body

coordinates and velocities ω, h. System (5.28), as well as an approximate solution of

the hydrodynamic problem [18–20], is not applicable on a small (of the order of v1)
initial interval of time, when the fluid motion in the cavity is essentially
nonstationary. After the initial interval of time, the motion is described by system
(5.28) with an accuracy of the order of v2 on the time interval of the order of unity.
On a large interval of the order of v, the error will be of the order of v1.
In [21], a rigorous justification of the presented method is carried out and
error estimates are given which correspond to the above ones.
Consider a particular case of a free rigid body and put L ¼ 0 in (5.28). Then from
(5.28) we get that vector a will be a homogeneous quadratic function of the compo-
nents of vector ω, while vector b and, hence, vector m third-degree homogeneous
functions of these components. We have:

X
3
m ¼ ρv1 mijk ωi ωj ωk : ð5:29Þ
i, j, k¼1

Here mijk are the constant vector coefficients depending on the constant tensors
J and P; ωi are the components of vector ω for the axes of the coordinate system
connected with the body. Thus, in the simplest case of a free rigid body containing a
viscous fluid at low Reynolds numbers (high viscosity), the additional perturbation
torque acting on the body turns out to be a polynomial of the third degree of the
angular velocity of the body. The motion of the body under the action of such torque
is investigated in Sect. 7.3.
Let us present the one obtained in [18–20] expression for tensor P in the simplest
case of a spherical cavity of the radius a:

8πa7
Pij ¼ P0 δij , P0 ¼ , i, j ¼ 1, 2, 3: ð5:30Þ
525

Here, δij is the Kronecker symbol.

System (5.28) can be studied by analytical or numerical methods of the theory of
ordinary differential equations. It contains a small parameter v1, so the averaging
method can be applied to solve it. Moreover, it suffices to consider the first approx-
imation ensuring the accuracy of the order of v1 on the time interval of the order of v.

5.4 Case of Moving Masses Connected to the Body by

Elastic Couplings with Viscous Friction

1. Following [22], we consider the motion of a system consisting of a rigid body

D of the mass m, relative to which n point particles Pi with masses mi, i ¼ 1 , . . . , n
can move. Denote by O the center of inertia of the rigid body and by Oi, some points
5.4 Case of Moving Masses Connected to the Body by Elastic Couplings with. . . 83

fixed relative to the body and representing “average” positions of the points Pi. The
position of the points O, Oi will be characterized by their radius vectors R0, Ri
relative to the fixed pole, while the positions of the points Pi by their displacements
ri ¼ OiPi relative to the points Oi. The absolute velocities of the points O, Oi will be
denoted by V0, Vi; the absolute angular velocity of the body, by ω. Then the
momentum (impulse) of the entire system can be written as:
X
n
Q ¼ mV0 þ mi ðVi þ ω  ri þ r_ i Þ: ð5:31Þ
i¼1

In (5.31), the dot denotes differentiation with respect to time t in the coordinate
system connected with the body. Let us consider a system D∗ consisting of a rigid
body D and points Pi, which are all the time at the positions Oi when ri ¼ 0,
i ¼ 1 , . . . , n. Obviously, D∗ is a rigid body with the mass:

X
n
M¼mþ mi , ð5:32Þ
i¼1

whereas its center of inertia C has the radius vector RC relative to the fixed pole
which is defined by the relation:

X
n
MRC ¼ mR0 þ mi Ri : ð5:33Þ
i¼1

Relation (5.31) can be rewritten in the form:

X
n
Q ¼ MVC þ q, q¼ mi ðω  ri þ r_ i Þ, ð5:34Þ
i¼1

where VC is the absolute velocity of the point C. Now write the angular momentum
of the entire system relative to the point C:

X
n
GC ¼ mðR0  RC Þ  VC þ JC  ω þ m i ð ρi þ r i Þ
i¼1

 ðVi þ ω  ri þ r_ i Þ: ð5:35Þ

Here, JC is the tensor of inertia of rigid body D relative to point C, while

ρi ¼ Ri  RC. The first two terms represent the angular momentum of body
D relative to point C. Representing Vi in the form Vi ¼ VC + ω  ρi and
transforming expression (5.35) taking into account equalities (5.32), (5.33),
we arrive at:
84 5 Perturbation Torques Acting upon a Rigid Body

GC ¼ G þ g,
G ¼ J∗
C  ω,
X
n ð5:36Þ
g¼ mi fri  ½VC þ ω  ðρi þ ri Þ þ r_ i  þ ρi  ðω  ri þ r_ i Þg:
i¼1

We denote by G the angular momentum of rigid body D∗ with respect to point

C and by J∗ C its tensor of inertia relative to C. At that, the terms q, g in relations
(5.34), (5.36) turn to zero, if the relative motions of the points are absent and ri ¼ 0,
i ¼ 1 , . . . , n.
The theorems of change of impulse and angular momentum of the system with
respect to point Pi have the form:

dQ dGC
¼ F, ¼ LC þ Q  VC : ð5:37Þ
dt dt

Here, F is the principal vector of all external forces, and LC is the principal
moment of these forces relative to point C. We will assume that F and LC do not
depend on the motion of points Pi. After substitution of formulas (5.34), (5.36) into
relations (5.37), we get:

dVC dGC
M ¼ F þ f, ¼ L C þ lC ,
dt dt ð5:38Þ
dq dg
f ¼  , lC ¼ q  VC  :
dt dt

Equations (5.38), obtained in [22], can be considered as the equations of motion

of rigid body D∗; then the quantities f and lC are the perturbation force and
perturbation torque due to internal motions. The expression for f и lC in the
expanded form will be obtained by substituting into (5.38) the expression for
q from (5.34) and for g from (5.36) and using the equality:

da
¼ a_ þ ω  a, ð5:39Þ
dt

which holds for any vector a. Similar equations in another form are obtained in
book [23]. If the internal motions are given, i.e., the dependencies ri(t) are known,
then equations (5.38), (5.34), (5.36) together with the kinematic relations form the
system of motion equations of body D∗. Otherwise, these equations should be
supplemented by the equations of motion of points Pi.
2. Following [22], let us consider the motion of a free rigid body, to which one
point P with the mass m is attached at the point O1 by means of elastic coupling.
The motion of this point is not given and must be found. We use the notations from
subsection 1 and drop the subscript i, so that ρ is the radius vector of point O1
fixedly connected with the body, while r is the radius vector of point P relative to
5.4 Case of Moving Masses Connected to the Body by Elastic Couplings with. . . 85

O1. Since the body is free (F ¼ 0, LC ¼ 0), an inertial coordinate system can be
chosen in such a way that Q ¼ const ¼ 0 in it. Then, according to (5.34), we get:
m
VC ¼  ðω  r þ r_ Þ: ð5:40Þ
M

After substituting equalities (5.40) into relations (5.36), (5.37), we find:

n GC ¼ G þ g ¼ const,
h G ¼ J∗
C  ω,
m io
ð5:41Þ
g ¼ m ρ  ðω  r þ r_ Þ þ r  ω  ρ þ 1  ðω  r þ r_ Þ :
M

Let the moving point P be connected with point O1 of the rigid body by
elastic coupling with stiffness c and coefficient of viscous friction δ. In this case,
the equation of the relative motion of point P assumes the form:
 
dVC
m€rþδr_ þcr ¼ m þ ω  ½ω  ðρ þ rÞ þ ω_  ðρ þ rÞ þ 2ω  r_ :
dt
ð5:42Þ

Substituting into equation (5.42) velocity VC from (5.40) and expanding the
total derivative VC according to (5.39), we reduce equation (5.42) to the form:

λr_ þ Ω2 r ¼  ω  ðω  ρÞ þ ω_
 m o
 ρþ 1  ½ω  ðω  rÞ þ ω_  r þ 2ω  r_ þ €r : ð5:43Þ
M

Here, the quantities Ω and λ, having the dimension t1, are defined by the
equalities:

c δ
Ω2 ¼ , λ¼ ð5:44Þ
m m

and characterize the frequency and decay time of free oscillations.

Equations (5.42), (5.43) are considered in the coordinate system connected with
the body, ρ is a constant vector, and ω is a function of time. The solution of equation
(5.43), linear with respect to r, can be represented as a sum of free and forced
motions. If the condition λ  ω is fulfilled, free oscillations of the point P die down
faster than the body makes a turn. Therefore, in studying the evolution of the rigid
body motion, free oscillations of point P can be neglected and only its forced
motions relative to the body taken into account. If an additional condition Ω2  ωλ
is satisfied, then the forced solution of equation (5.43) can be sought in the form of
expansion in powers of Ω2. We can assume that the unit for measuring time is
selected to be of the order of the body’s revolution period, so that ω  1, and then
the above conditions take the form:
86 5 Perturbation Torques Acting upon a Rigid Body

Ω2  λ  1: ð5:45Þ

Calculating the first terms of the expansion for the forced solution of equation
(5.43) with respect to powers of Ω2 provided that conditions (5.45) are fulfilled,
we obtain:
 
r ¼ Ω2 a þ λΩ4 a_ þ O Ω4 ; λ2 Ω6 : ð5:46Þ

Here:

a ¼ ω  ðω  ρÞ þ ω_  ρ, ð5:47Þ

where the dot everywhere denotes the derivatives in the system of coordinates
connected with the body. In view of relations (5.45), (5.46), expression (5.41) for
g assumes the form:
 
g ¼ m½ρ  ðω  r þ r_ Þ þ r  ðω  ρÞ þ O Ω4 ; λ2 Ω6 : ð5:48Þ

Condition (5.41) of the constancy of vector GC is written in the form:

 
J∗ _ þ ω  J∗
C ω _ þ ω  gÞ:
C  ω ¼ ðg ð5:49Þ

Equation (5.49) can be transformed as follows:

   4 2 6 
J∗ _ þ ω  J∗
C ω C  ω ¼ ΦðωÞ þ O Ω ; λ Ω , ð5:50Þ

where vector function Φ includes the terms of the orders of Ω2 and λΩ4. To
calculate this function explicitly, we need to substitute expression (5.47) into
(5.46), then (5.46) into (5.48), and (5.48) into the right-hand side of (5.49). The
derivatives ω_ , ω
€ and so on, which will be included into the right-hand side of
equation (5.49), should be expressed through ω, using the equality:
 1     
ω_ ¼  J∗
C  ω  J∗
C ω þ O Ω2 , ð5:51Þ

which follows from (5.49) and the above estimates. As a result, we obtain vector
equation (5.50) which describes the change of vector ω in the system of coordinates
connected with the body. Function Φ(ω) is a polynomial containing the fourth and
fifth powers of ω; because of its awkwardness, this expression is not presented here.
In [18–20], a similar equation is obtained, when Φ(ω) is a third-degree polynomial
of ω (see Sect. 5.3). Equation (5.50) can be solved by the method of averaging.
5.5 Body with Elastic and Dissipative Elements 87

5.5 Body with Elastic and Dissipative Elements

In this section, we consider a more general than in Sect. 5.4 case of a system
consisting of a rigid body and material points connected with the body by
elastic and dissipative elements.
1. Following [24], consider the motion of a system S, consisting of a rigid body
D with the mass m and N particles Pi with the masses mi, i ¼ 1 , . . . , N. Particles
(further referred to as points) Pi are connected with the body D and with one another
by means of ideal connections and linear elastic connections with linear damping.
Let Oi be the equilibrium positions of points Pi in the case when the system is at rest
and there are no external forces.
We introduce three Cartesian systems of coordinates: a fixed system O0 X0 1 X02 X03 ,
a system Ox1x2x3 rigidly connected with the body, and a system OX1X2X3 whose
origin O is connected with the rigid body, while the axes move translationally and
parallel to the axes of system O0 X0 1 X02 X03 . Introduce the notations:

R0 ¼ O0 O, ρi ¼ OOi , r i ¼ O i Pi ,
0
ð5:52Þ
Ri ¼ O Pi ¼ R0 þρi þ ri , i ¼ 1, . . . , N:

The time derivatives of scalar quantities will be denoted by dots. For an arbitrary
0
three-dimensional vector a, we denote by a and a_ its derivatives with respect to
time in coordinate systems Ox1x2x3 and O0 X0 1 X02 X03 , respectively. We have:

a_ ¼ a0 þ ω  a, ð5:53Þ

where ω is the vector of absolute angular velocity of body D, i.e., of coordinate

system Ox1x2x3. Note that since points O and Oi are rigidly connected to body D,
then ρ0i ¼ 0 for i ¼ 1 , . . . , N. We obviously have ω_ ¼ ω0 .
Let us construct the equations of motion of system S. First, we consider the
motion of points Pi relative to body D. Suppose that the set of points Pi has
n degrees of freedom relative to body D, and its position in coordinate system
Ox1x2x3 can be characterized by an n-dimensional vector q of generalized coordi-
nates q1, ..., qn. We assume that, in the case of small oscillations, the displacement
vectors ri are linearly expressed in terms of generalized coordinates qj:

X
n
ri ¼ Hij qj , i ¼ 1, . . . , N: ð5:54Þ
j¼1

Here Hij are vectors which are constant in the system of coordinates Ox1x2x3. In
view of equations (5.54), the kinetic energy of motion of points Pi with respect to
coordinate system Ox1x2x3 has the form:
88 5 Perturbation Torques Acting upon a Rigid Body

1X n  2 1 X n X
N
mi r0i ¼ ajk q_ j q_ k , ajk ¼ mi H ij H ik , j, k ¼ 1, . . . , n: ð5:55Þ
2 i¼1 2 j, k¼1 i¼1

We write the equations of small oscillations of points Pi relative to body D in the

form of Lagrange equations:

A€q þ Bq_ þ Cq ¼ Q: ð5:56Þ

Here A, B, C are constant symmetric matrices of dimension n  n (we assume

them to be positive definite). Matrix A ¼ k ajkk characterizes the kinetic energy (see
(5.55).); matrix B ¼ k bjkk, dissipation; and C ¼ k cjkk, the elastic potential energy.
In equation (5.56), Q denotes n-dimensional vector of generalized forces Q1, . . .,
Qn, which are due to the action of external forces Fi on points Pi and of inertia
forces in the coordinate system Ox1x2x3. They are equal to:

X
N   
Qj ¼ Hij  Fi  mi R€ 0 þ ω  ðω  ðρi þ ri ÞÞ þ ðω0  ðρi þ ri ÞÞ þ 2ω  r0 ,
i
i¼1
j ¼ 1, . . . , n:
ð5:57Þ

External forces Fi acting on each point Pi are assumed to be independent of its

absolute position, velocity, and time:
   
Fi ¼ Fi Ri ; R_ i ; t ¼ Fi R0 þ ρi þ ri ; R_ 0 þ ω  ðρi þ ri Þ þ r0i ; t ,
ð5:58Þ
i ¼ 1, . . . , n:

Here we have used relations (5.52), (5.53) and the equalities ρ0i ¼ 0, i ¼ 1 , . . . ,
n.
Let us write the equation of moments for the entire system S relative to pole O,
while considering its motion in coordinate system OX1X2X3. We have:

G_ ¼ L þ L1 : ð5:59Þ

Here G is the angular momentum of system S relative to pole O in its motion

with respect to coordinate system OX1X2X3; L and L1 are, respectively, principal
moments relative to pole O of all external forces and inertia forces acting on
system S. By definition, the indicated quantities equal:
5.5 Body with Elastic and Dissipative Elements 89

X
G¼ mα ðρα þ rα Þ  ðω  ρα þ r_ α Þ,
α X
L¼ ð ρα þ r α Þ  F α , ð5:60Þ
Xα
L1 ¼  mα ðρα þ rα Þ  R € 0:
α

Here we use formulas (5.52), (5.53), and the summation extends to all points Rα
of system S. For the points of rigid body D in (5.60), we should take rα 0.
For further consideration, it is convenient to introduce an auxiliary system S∗
consisting of rigid body D and points Pi, rigidly fixed at their equilibrium positions
Oi. Thus, S∗ is a rigid body with the mass:

X
N
m∗ ¼ m þ mi ð5:61Þ
i¼1

and for it all ri ¼ 0. The center of inertia S∗ of the body is denoted by C, whereas its
tensor of inertia relative to point O by J. Obviously, tensor J is constant in coordi-
nate system Ox1x2x3. The quantities (5.60) can be represented in the form:

X
N
G¼Jωþ mi ½ri  ðω  ρi þ r_ i Þ þ ρi  r_ i ,
i¼1
X N   
L ¼ L∗ þ r i  F i þ ρi  F i  F ∗
i , ð5:62Þ
i¼1
X
N
L1 ¼ L∗ € 0:
1  m i ri  R
i¼1

From now on, the asterisk means that the corresponding quantities are calculated
for rigid body S∗, i.e., for ri 0, i ¼ 1 , . . . , N.
Torques L∗, L∗ 1 equal:
  X X
L∗ R0 ; R_ 0 ; ω; σ; t ¼ ρα  F ∗
α, L∗
1 ¼ 
€ 0:
mα ρα  R ð5:63Þ
α α

Torque L∗ may depend on the variables which characterize the motion of

rigid body S∗, namely, on R0, R_ 0 , ω, t and the vector parameter σ defining the
orientation of system Ox1x2x3 relative to system OX1X2X3. As the components of
vector σ, there can be taken, for example, the Euler angles of the direction cosines
of system Ox1x2x3 relative to OX1X2X3. Vector σ satisfies the usual kinematic equa-
tions for a rigid body (for instance, kinematic Euler’s equations), which we write in
the form:

σ_ ¼ f ðσ; ωÞ: ð5:64Þ

90 5 Perturbation Torques Acting upon a Rigid Body

Let the motion of point O is given, i.e., R0(t) is a known function of time. For
example, this is true when body D has a fixed point. Then the motion of system S is
completely described by equations (5.56), (5.59), (5.64) and relations (5.52)–(5.55),
(5.57), (5.58), (5.62), (5.63). In this case, the moment L∗
1 of the inertia forces from
(5.63) depends only on the orientation of σ and on t by means of R € 0 ðtÞ.
If the motion of point O is not given, then the equation of impulses for system
S should be added to the indicated equations. In this case, the center of inertia C of
body S∗ can be conveniently chosen as the pole O.
Then, taking into account the second equality of (5.63), we have:
X
mα ρα 0, L∗ 1 ¼ 0: ð5:65Þ
α

The equation of the impulse change assumes the form:

X
N  
€ 0 ¼ F∗ þ
m∗ R Fi  F∗
i  mi €
ri : ð5:66Þ
i¼1

 
Here m∗ is given by formula (5.61), whereas F∗ ¼ F∗ R0 ; R_ 0 ; ω; σ; t is the
principal vector of all external forces acting on rigid body S∗, i.e., under the
condition ri 0, i ¼ 1 , . . . , N. The obtained equations of motion will be studied
and simplified on the basis of assumptions which are made below.
2. Let us introduce three characteristic time scales: the characteristic period T1 of
free oscillations of points Pi relative to body D, the characteristic decay time T2 of
these oscillations, and the characteristic time T3 of the motion of the system as a
whole. For example, we can assume that T3  ω1. We assume that the introduced
time scales are connected by the inequalities:
T1
T2
T3: ð5:67Þ

If conditions (5.67) are satisfied, then the free elastic oscillations die down
during the time T2 which is much smaller than the time T3 of the body’s revolution
about its center of mass. Therefore, in studying the evolution of the system’s motion
during the time intervals T3 (and larger), the free oscillations can be neglected, and
only the forced movement of points Pi under the action of external forces and
inertial forces may be taken into account. To fulfill the condition (5.67), as will be
shown below, it suffices to put in equations (5.57):

C ¼ ε2 C0 , B ¼ δε1 B0 : ð5:68Þ

Here C0, B0 are matrices with bounded elements, whereas ε and δ are dimension-
less small parameters which are subject to conditions:
5.5 Body with Elastic and Dissipative Elements 91

0 < ε
δ
1: ð5:69Þ

In the limiting case ε ! 0, corresponding to the infinitely large rigidity of elastic

couplings, equalities (5.56), (5.68) yield q 0. From relations (5.54) we obtain that
ri 0 for i ¼ 1 , . . . , N. Hence, equations (5.59), (5.62), (5.66) in view of (5.53)
yield for ε ¼ 0:

J  ω0 þ ω  ðJ  ωÞ ¼ L∗ þ L∗ € 0 ¼ F∗ :
m∗ R ð5:70Þ
1,

Equations (5.70) are the equations of motion of rigid body S∗, into which the
system S is transformed as ε ! 0.
In the case of small positive ε, δ, equation (5.56) under conditions (5.68) takes
the form:

ε2 A€q þ δεB0 q_ þ C0 q ¼ ε2 Q: ð5:71Þ

An approximate solution of equation (5.71) with small parameters at the deriv-

atives in view of (5.69) can be constructed using the asymptotic methods and
consists of a regular part and a solution of the type of boundary layer, rapidly
fading out in moving away from the initial time moment.
Consider first the free elastic oscillations described by homogeneous system
(5.71) for Q ¼ 0. The corresponding characteristic equation has the form:
 
det ε2 λ2 A þ δελB0 þ C0 ¼ 0: ð5:72Þ

We proceed from the generalized coordinates q to normal coordinates in which

two positive definite matrices A and C0 are simultaneously reduced to diagonal
form [25]. Under this transformation, matrix A becomes the identity matrix I, and
matrix C0 becomes a diagonal matrix C01 with positive diagonal elements, while
matrix B0 turns into some positive definite matrix B01 . The characteristic equation
(5.72) takes the form:
 
det Λ2 I þ δΛB01 þ C01 ¼ 0, Λ ¼ ελ  1: ð5:73Þ

The roots of equation (5.73) will be obtained in the form of expansions with
respect to small parameter δ. We get:

 1=2 1    
Λj ¼
i C01 jj  δ B01 jj þ O δ2 , j ¼ 1, . . . , n: ð5:74Þ
2

Here the subscripts jj indicate the diagonal elements of matrices; these elements
are positive. Coming back to variable λ from (5.73), we obtain from (5.74):
92 5 Perturbation Torques Acting upon a Rigid Body

1      1=2
λj ¼
iΩj  δε1 B01 jj þ O δε1 , Ωj ¼ ε1 C01 jj , j ¼ 1, . . . , n: ð5:75Þ
2

The quantities Ωj are the fundamental frequencies of oscillations of the conser-

vative system, into which system (5.56) is transformed for B ¼ 0. Equalities (5.75)
yield the estimates:
 1   1 
T 1
1 ¼ O ε , T 1
2 ¼ O ε δ , T 3 ¼ Oð1Þ: ð5:76Þ

The last estimate (5.76) follows from the independence of T3 of parameters ε, δ.

It follows from (5.76), (5.69) that inequalities (5.67) are satisfied. Thus, the made
assumptions (5.68) are justified. Free oscillations, corresponding to eigenvalues
(5.75), are a rapidly decaying part of the solution of the boundary layer type. These
oscillations can be neglected far from the initial moment of time, i.e., for the
times of the order of T3 and larger.
The part of solution of equation (5.71), regular with respect to ε, δ, will be
obtained in the form of expansion in powers of quantities ε2 and εδ. In view of
inequality (5.69), we write:
 
q ¼ ε2 qð0Þ þ ε3 δqð1Þ þ O ε4 : ð5:77Þ

Substituting expansion (5.77) into equation (5.71) and equating the coefficients
at the powers of parameters ε, δ, we obtain:
 1  1
qð0Þ ¼ C0 Q∗ , qð1Þ ¼  C0 B0 q_ ð0Þ : ð5:78Þ

Here Q∗ is the vector of generalized forces, where one should put q 0.

Therefore, the forces Q∗ ∗
j correspond to rigid body S and are calculated by formulas
(5.57), (5.58) for ri 0, i ¼ 1 , . . . , N. We have from these formulas:

X
N   
Q∗ H ij F∗ € 0
j ¼ i  mi R0 þ ω  ðω  ρi Þ þ ω  ρi ,
i¼1
  ð5:79Þ
F∗ _
i ¼ Fi R 0 þ ρi ; R 0 þ ω  ρi ; t ,
i ¼ 1, . . . , N, j ¼ 1, . . . , n:

Coming back to the original notations (5.68), we write solutions (5.77), (5.78) in
the form:
  
q ¼ C1 Q∗  BC1 Q_ ∗ þ O ε4 : ð5:80Þ

This solution describes small forced motions of points Pi relative to body D.

3. To obtain simplified equations of motion of system S under these assumptions,
one needs to substitute solution (5.80) into relations (5.59), (5.62), (5.66). Note that
5.5 Body with Elastic and Dissipative Elements 93

according to (5.68), vector q from (5.80) is of the order of ε2. The vectors ri defined
by formula (5.54) are of the same order too. Taking this into account, we reduce
equations (5.59), (5.62), (5.66) to the form:

J  ω0 þ ω  ðJ  ωÞ ¼ L∗ þ L∗ m∗ R € 0 ¼ F∗ þ η,
1 þ μ, ð5:81Þ
N   ∗ 
X ∂Fi ∂F∗
μ¼ ri  F ∗ þ ρ  r þ i
_
r
∂R_ i
i i
i¼1
i i
∂Ri
   
€ 0  mi ½ri  ðω  ρi Þ þ r_ i  g þ O ε4 ,
mi ri  R

N  
X ∂F∗ ∂F∗  
η¼ i
ri þ i r_ i  mi€ri þ O ε4 : ð5:82Þ
i¼1
∂Ri ∂R_ i

∂F∗ ∂F∗
Here i
∂Ri
and i
∂R_ i
are the matrices of partial derivatives of functions F∗
i from
(5.79) with respect to components of their vector arguments.
Equations (5.81) are similar to the equations of motion (5.70) of rigid body S∗
and differ from them by terms μ and η. These terms can be interpreted, respectively,
as the principal moment relative to point O and the principal vector of forces acting
on rigid body S∗ which are due to the presence of elastic and dissipative elements.
Vectors μ and η in (5.82) linearly depend on vectors ri and their derivatives and
contain terms of the orders of ε2 and ε3δ. The terms O(ε2) correspond to the internal
elastic forces, while the terms O(ε3δ) to the dissipative forces.
Let us show that vectors μ, η can be expressed up to quantities of the order of ε4
only in terms of variables R0, R_ 0 , σ, ω, t, characterizing the motion of rigid body S∗
. To do this, first, we replace in (5.82) the derivatives r_ i , €ri using formula (5.53) and
00
then express the resulting vectors ri, r0i , ri in terms of q, q_ , €q by means of formula
(5.54). In differentiating (5.54) one should note that vectors Hij are constant in
coordinate system Ox1x2x3. After that, vectors μ, η will depend on vector q and its
derivatives, which we eliminate by means of formulas (5.80) and (5.79). As a result,
we obtain vectors μ, η as functions of vectors R0, ω and their derivatives, as well as
vectors Hij, ρi, which are known and constant in coordinate system Ox1x2x3. The
expressions for μ, η will include higher derivatives of vectors R0, ω so that
equations (5.81) will formally be of a higher order than the usual equations of the
rigid body dynamics. However, with no harm made to the accuracy, these higher
derivatives can be excluded.
To this end, we note that the equations (5.70) of motion of rigid body S∗ can
always be solved for higher derivatives, i.e., with respect to ω and R
0
€ 0 , obtaining the
dependences:
   
ω0 ¼ f 1 R0 ; R_ 0 ; σ; ω; t , € 0 ¼ f 2 R0 ; R_ 0 ; σ; ω; t :
R ð5:83Þ
94 5 Perturbation Torques Acting upon a Rigid Body

Differentiating equalities (5.83) taking into account (5.64), we can also obtain
00 ...
the expressions for higher derivatives ω , R0 and so on in terms of the same
variables R0, R_ 0 , σ, ω, t. Since the obtained equations of motion of the deformable
body (5.81) differ from the equations of rigid body motion (5.70) by terms μ, η of
the order ε2, expression (5.83) and their derivatives are also valid for equations
(5.81) with an error O(ε2). Therefore, the substitution of expressions (5.83) and
their derivatives into formulas (5.82) for quantities μ, η, which themselves are of the
order of ε2, will result in an error O(ε4), which is within the accuracy of equalities
(5.82).
...
0
Thus, eliminating derivatives ω , ω , R
00
€ 0 , R and so on with the help of equalities
0
(5.83) and their derivatives, we obtain the desired dependencies of the form:
   
μ ¼ μ R0 ; R_ 0 ; σ; ω; t , η ¼ η R0 ; R_ 0 ; σ; ω; t ð5:84Þ

with an error O(ε4). Functions (5.84), similar to torque L∗ from (5.63) and force F∗
from (5.66), depend only on the parameters of motion of rigid body S∗. Therefore,
equations (5.81) together with (5.84), (5.64) form a closed system, similar to the
equations of motion of a rigid body. These equations describe the motion of a
deformable system S on the intervals of time larger than the decay time of elastic
natural oscillations. Here quantities μ, η play a role of small perturbations, so to
integrate the resulting system, one can use various methods of small parameter, in
particular, the method of averaging.
The explicit form of dependencies (5.84) is not given here due to its awkward-
ness, but the transformation procedure described above, using relations (5.82),
(5.53), (5.54), (5.79), (5.80), (5.70), (5.83), allows uniquely constructing the func-
tions (5.84).
We restrict ourselves to two examples, in which we establish the structure of
functions (5.84). It is assumed in both examples that torque L∗ as well as all
external forces Fi acting on points Pi equal zero. Then, according to (5.79), (5.82),
we have:

X
N 
Q∗ € 0 þ ω  ðω  ρi Þ þ ω0  ρi ,
j ¼ Hij mi R j ¼ 1, . . . , n,
i¼1

X
N   X
N
μ¼ € 0 þ ½ri  ðω  ρi Þ þ ρi  r_ i  ,
ri  R φ¼ mi€ri : ð5:85Þ
i¼1 i¼1

In the first example, we assume, in addition, that point O is fixed, R0 0, and

then, according to (5.63), we have, L∗
1 ¼ 0.
In the second example, the whole system is assumed to be free from external
forces, F∗ ¼ 0. Choosing here the center of inertia C of system S∗ as the pole O,
we have, according to (5.65), that L∗
1 ¼ 0.
∗ ∗
Thus, in both examples L ¼ L1 ¼ 0, and the first equation (5.81) yields:
5.5 Body with Elastic and Dissipative Elements 95

 
J  ω0 þ ω  ðJ  ωÞ ¼ μ ¼ O ε2 : ð5:86Þ

The first equality in (5.83) assumes the form:

 
ω0 ¼ J1  ðω  J  ωÞ þ O ε2 : ð5:87Þ

Differentiating equality (5.87), we conclude that the k-th derivative ω(k) will be,
up to the terms of the order of ε2, a homogeneous polynomial of degree k + 1 of the
components of vector ω, where k ¼ 0 , 1 , . . ..
In the first example (R0 0), we conclude from the first equality (5.85) that Q∗j
are homogeneous polynomials of ω of the second degree having the order of m0lω2.
Here, m0 is the characteristic mass of points Pi, and l is the characteristic linear
dimension of the order of ρi.
Then it follows from (5.80) that q is the sum of homogeneous polynomials of ω
of the second and third degrees having the order of m0lc1ω2 and m0lc2bω3,
respectively. Here, c is the characteristic stiffness of elastic couplings (of the
order of elements of matrix C), and b is the characteristic dissipation coefficient
(of the order of elements of matrix B). The vectors ri, i ¼ 1 , . . . , N, defined by
formula (5.54), have the same structure. Note that, according to (5.87), each differ-
entiation increases the degrees of polynomials of ω by one. Therefore, vector μ
from (5.85) for R0 0 is the sum of homogeneous polynomials of the fourth and
fifth degrees of components ωj of vector ω, namely:

μ ¼ μ4 ðωÞ þ μ5 ðωÞ þ Oðε4 Þ,

X  2 4
3
m0 l ω  
μ4 ðωÞ ¼ Djklm ωj ωk ωl ωm ¼ O ¼ O ε2 ,
j, k, l, m¼1
c ð5:88Þ
X 3  
m0 lbω3  
μ5 ðωÞ ¼ Ejklmn ωj ωk ωl ωm ωn ¼ O 2
¼ O ε3 δ :
j, k, l, m, n¼1
c

The magnitude orders in (5.88) are specified in accordance with the orders in
(5.68). Coefficients Djklm, Ejklmn are constant in the coordinate system connected
with the rigid body and expressed in terms of constants mj, J, C0, B0, Hij, ρi.
Polynomial μ4 is the moment of elastic forces, while μ5 is the moment of dissipative
forces.
Now turn to the second example (F∗ ¼ 0). Here we obtain from the second
equation (5.81) that R€ 0 ¼ Oðε2 Þ. Therefore, quantities Q∗ , μ from (5.85), with the
j
accepted accuracy of O(ε4), are of the same form as in the first example. Pertur-
bation torque μ is again represented by formulas (5.88), whereas perturbation force
φ from (5.85) by similar relations as the sum of homogeneous polynomials of ω of
the fourth and fifth degree.
Equations (5.86), (5.88) for the case of a single point Pi on the elastic coupling
(N ¼ 1) were first obtained in [22] (see Sect. 5.4).
96 5 Perturbation Torques Acting upon a Rigid Body

For some special cases (rigid body S∗ possesses a symmetry), expressions (5.88)
are explicitly computed, and equation (5.86) with torque (5.88) is integrated in [22].
Papers [26, 27] consider the problem of dynamics of a rigid body which carries
elastic and dissipative elements in the formulation proposed in [22, 24] (Sects. 5.4
and 5.5). The coefficients of stiffness and damping are assumed to be large enough.
On the basis of the method of singular perturbations, a rigorous mathematical sub-
stantiation and error estimates are given for the asymptotic approach proposed in
[22, 24].

5.6 Viscoelastic Solid Body

1. Following [28], we consider the motion of a mechanical system S consisting of

two parts: a perfectly rigid body D0 and a deformable body D, which are rigidly
coupled. First, we assume that the body has a fixed point O, connected with D0, and
it is not acted upon by external forces and torques (except for the reaction of
coupling at point O). Denote by r the current radius vector measured from point
O and by u(r, t) the displacement vector of the points of system S in a coordinate
system Ox1x2x3 associated with the body D0. We introduce an auxiliary system S0:
a perfectly rigid body in which there are no deformations (u ¼ 0). Its inertia tensor
with respect to point O is denoted by J. Tensor J in coordinate system Ox1x2x3 is
constant.
Let us assume that the deformable body D is subject to the Kelvin–Voigt
equations of linear viscoelastic medium [29]:

σ ¼ G0 ½σ 0 ðuÞ þ τσ 00 ðu_ Þ,

σ 0ij ¼ 2eij þ 2μð1  2μÞ1 δij div u, i, j ¼ 1, 2, 3,

2
σ ij 00 ¼ 2e_ ij  δij div u_ ,
3
 
1 ∂ui ∂uj
eij ¼ þ : ð5:89Þ
2 ∂xj ∂xi

Here, σ is the stress tensor which is the sum of elastic and viscous stresses, G0 is
the shear modulus, μ is Poisson’s ratio, and τ is a constant with the dimension of
time (relaxation time in shear). From now on, the dot in (5.89) denotes the partial
derivative with respect to t in coordinate system Ox1x2x3. We write equation (5.89)
of the medium motion in the domain D in the moving coordinate system Ox1x2x3
taking into account the forces of inertia:
5.6 Viscoelastic Solid Body 97

ρG1 € þ ω  ðω  rÞ þ ω_  r þ 2ω  u_ 
0 ½u
 
1
¼ Δu þ ð1  2μÞ1 div u þ τ Δu_ þ ∇div u_ : ð5:90Þ
3

Here, ω is the vector of angular velocity of system Ox1x2x3, and ρ is the density
of the medium in the non-deformed state. The boundary conditions are of the form:
u ¼ 0 on Γ1 , σ  n ¼ 0 on Γ2 , ð5:91Þ

where Γ1 is the boundary between bodies D and D0, Γ2 is the free part of the
boundary of body D, and n is a normal to D.
Denote by OX1X2X3 a fixed coordinate system whose origin is at point O and
the axes move progressively. Angular momentum G of system S relative to
pole O equals:
ð
G¼ r  ðω  r þ u_ Þρ0 dv ¼ J  ω þ g,
D0 þD0
ð ð ð5:92Þ
g¼ r  ðω  r þ u_ Þρ0 dv  r  ðω  rÞρ dv:
D0 D

Here, ρ, D are the density of the medium and the domain occupied by it in the
non-deformed state; the primes refer to the deformed state.
0
In the considered case of small deformations, we have ρ ¼ ρ(1  div u), and
expression (5.92) for g is transformed to the form:
ð I
g ¼ ½r  u_  rðω  rÞdiv uρ dv þ r  ðω  rÞðunÞρdS,
D Γ

where n is the unit vector of the outer normal to the boundary Γ of domain D.
Replacing integration over Γ by integration over D, we obtain after transformations:
ð
g ¼ ½r  u_ þ ðu  ∇Þ  ðr  ðω  rÞÞρ dv
D
ð
¼ ½r  u_ þ r  ðω  uÞ þ u  ðω  rÞρdv: ð5:93Þ
D

Angular momentum G of system S is preserved; therefore, the following equa-

tion holds in the moving system Ox1x2x3:

J  ω_ þ ω  ðJ  ωÞ ¼ L, L ¼ g_  ω  g: ð5:94Þ
98 5 Perturbation Torques Acting upon a Rigid Body

Here, g is defined by equality (5.93). Quantity L in (5.94) has the meaning of

perturbation torque acting on the auxiliary rigid body S0 and is due to proper elasti-
city and dissipation.
2. Let us construct equations similar to (5.90), (5.94) in another case of motion
when body S is free: it is not acted upon by external forces and torques. As a point
O connected with body D0, we take the center of inertia of auxiliary body S0,
so that:
ð
rρdv ¼ 0: ð5:95Þ
D0 þD

The origin of coordinate systems OX1X2X3 and Ox1x2x3 is now moving with the
acceleration w0 of point O, so one term is added into equation (5.90), and this
equation assumes the form:

ρG1 € þ ω  ðω  rÞ þ ω_  r þ 2ω  u_ 
0 ½w 0 þ u
 
1 1
¼ Δu þ ð1  2μÞ ∇div u þ τ Δu_ þ ∇div u_ : ð5:96Þ
3

Angular momentum G has the same form (5.92), but now it is the angular
momentum of system S relative to pole O in the coordinate system OX1X2X3
whose axes are moving progressively. Therefore, to the right-hand side of the
equation of torques (5.94), there should be added the principal moment relative to
point O of inertia forces due to the motion of system OX1X2X3. This moment equals:
ð
L1 ¼ w0  rρ0 dv ¼ w0  rc , ð5:97Þ
D0 þD0

where rc is the radius vector of the center of inertia of system S relative to point O.
With the help of transformations similar to those used in the derivation of
formula (5.93) for g, it is easy to get:
ð ð
rc ¼ rρ0 dv ¼ uρdv: ð5:98Þ
D0 þD0 D

Here, we have used condition (5.95). In the considered case of motion, the
absolute acceleration of the center of inertia is equal to zero. Representing it as
an acceleration of complex motion, we obtain the equality for w0:

w0 ¼ ω  ðω  rc Þ  ω_  rc  2ω  r_ c  €rc : ð5:99Þ

Thus, the additional term w0 in equation (5.96) is expressed through u via rela-
tions (5.98), (5.99). Within small deformations, vector u and its derivatives with
5.6 Viscoelastic Solid Body 99

respect to r, t are considered to be small quantities of the first order, while the
second-order small quantities should be ignored. But from (5.98), (5.99) it follows
that rc, w0 are small quantities of the first order, so quantity L1 in (5.97) is a small
quantity of the second order. Consequently, no additional terms appear in equation
(5.94), and it is applicable in the case under consideration. In what follows, both
cases of motion (the motion about a fixed point O and free movement) are treated in
parallel.
3. We assume that the following relations hold:

G ¼ ε2 G0 , τ ¼ εδτ0 , ω1  τ0 , ε
δ
1, ð5:100Þ

where G0, τ0 are bounded quantities, while ε, δ are small dimensionless parameters.
It follows from (5.100) that the largest period T1 of elastic natural oscillations, the
characteristic time T2 of their damping, and the characteristic time T3 of motion of
the system as a whole are related as follows:

T 1  ετ0 , T 2  εδ1 τ0 , T 3  ω1  τ0 , T1
T2
T3: ð5:101Þ

These conditions are analogous to conditions (5.67), (5.76).

Conditions (5.101) can be written in the form c  lT 1 2  V, where c is the
characteristic velocity of elastic waves, V is the linear velocity of rotation of the
body, and l is its linear dimension. Note that conditions (5.100), (5.101) exclude the
possibility of resonance between elastic oscillations of the body and its rotations as
a whole.
Under conditions (5.100), (5.101), the free elastic oscillations of system S are
rapidly damped and can be neglected far away from the initial moment of time t0 for
t  t0  T3. The solution of system (5.90) or (5.96), (5.98), (5.99) with boundary
conditions (5.91) under the given ω(t) is reduced to forced movements, which can
be presented, on the basis of (5.100), in the form:
   0  4
uðr; tÞ ¼ ε2 u0 þ ε3 δu1 þ O ε4 ¼ ρG1 0 u ðr; tÞ þ τu ðr; tÞ þ O ε : ð5:102Þ
1

Substitute expansion (5.102) into equations (5.90) (or (5.96), (5.98), (5.99)) and
boundary conditions (5.91). We find that functions u0, u1 satisfy the quasi-static
problems of the theory of elasticity:

Δu0 þ ð1  2μÞ1 ∇div u0 ¼ ω  ðω  rÞ þ ω_  r,

ð5:103Þ
u0 ¼ 0 on Γ1 , σ 0 ðu0 Þ  n ¼ 0 on Γ2 ,
1
Δu1 þ ð1  2μÞ1 ∇div u1 ¼ Δu_ 0  ∇div u_ 0 ,
3 ð5:104Þ
00
u1 ¼ 0 on Γ1 , σ 0 ðu1 Þ  n ¼ σ ðu_ 0 Þ  n on Γ2 :

Here, time plays a role of a parameter. Note that the summands u €, 2ω  u_ , w0 in

the left-hand sides of equations (5.90), (5.96) do not affect the first two terms of
100 5 Perturbation Torques Acting upon a Rigid Body

expansions (5.102), (5.104). Therefore, we do not differentiate further between the

two cases.
Having solved problems (5.103), (5.104), we substitute solution (5.102) into
relations (5.93), (5.94). We find that g  ε2, L  ε2; therefore, according to (5.94):
 
ω_ ¼ J1  ½ω  ðJ  ωÞ þ O ε2 : ð5:105Þ

Using equality (5.105), we can express the derivatives ω_ , ω

€ and so on in terms of
ω with the error ε2 far away from the initial moment of time. Then the right-hand
side of equation (5.103) is a homogeneous quadratic form of ω; and therefore,
solution u0 of problem (5.103) will have the same structure. Similar arguments
show (in view of (5.105)) that u1 is a homogeneous polynomial of the third degree
of ω, and vector g from (5.93) contains polynomials of the third and fourth degree
of ω. Perturbation torque L in (5.94) is represented in the form:

L ¼ L4 ðωÞ þ L5 ðωÞ þ Oðε2 Þ,

X 3  
L4 ðωÞ ¼ ρG1
0 Aijkl ωi ωj ωk ωl ¼ O ε2 ,
i, j, k, l¼1 ð5:106Þ
X 3  
L5 ðωÞ ¼ τρG10 Bijklm ωi ωj ωk ωl ωm ¼ O ε3 δ :
i, j, k, l, m¼1

Quantities Aijkl, Bijklm are constant in the coordinate system connected with the
body. To calculate them, one needs to solve the static problems of elasticity theories
(5.103), (5.104) and then calculate g, L as described above. Therefore, Aijkl, Bijklm
are similar to the apparent additional masses in the case of ideal fluid and the
viscous added masses introduced in [18–20]; they depend on the shape of domain
D, on the boundaries Γ0, Γ1 in (5.91), as well as on J and μ.
Note for comparison that in the case of a body with a cavity containing fluid at
high Reynolds numbers [18–20] (see also Sect. 5.3), the corresponding perturbation
torque is a polynomial of the third degree of ω. For some domains, for example,
ellipsoid [30], as well as for rods and plates, the solutions of problems (5.103),
(5.104) are known or can be constructed explicitly.
After calculating Aijkl, Bijklm, the problem is reduced to integration of equations
(5.94) of the motion of rigid body S0 about the fixed point in the presence of pertur-
bation torques (5.106). The solution of these equations was constructed in [22], where
the motion of a body with a finite number of internal degrees of freedom was studied.
As a result of energy dissipation, the body movement is rearranged tending toward
the only stable motion: rotation about the axis of the largest moment of inertia. The
characteristic time of this transition is estimated in [22] as T  Ω4ω4λ1, where Ω is
the lowest frequency of elastic natural oscillations, λ ¼ T 1 2 with T2 being the
characteristic time of their damping. In the notation of (5.100), (5.101), we obtain
the following equivalent estimates of the transition process time:
5.7 Influence of a Moving Mass Connected to the Body by an Elastic Coupling. . . 101

T 43 T 2 τ0 Ω2
T   : ð5:107Þ
T 41 ε 3 δ ω4 τ

It is assumed here that all the principal moments of inertia and the difference
between them are of the same order; otherwise, formula (5.107) includes an
extra coefficient (see [22]).
The motion of a viscoelastic rigid body under the action of forces and torques of
the general form can be considered similarly. As a result, a system of ordinary
differential equations is obtained, which describes the motion of an auxiliary rigid
body S0 under the presence of perturbation forces and torques. The dependence of
these forces and torques has a complex character. This conclusion for the case of a
body with a finite number of degrees of freedom is given in [22].

5.7 Influence of a Moving Mass Connected to the Body by

an Elastic Coupling with Quadratic Friction

We consider, following [31], the free movement of a rigid body, to which a point
mass m is affixed at a certain point O1 connected to the body. It is assumed that,
during the relative movement, the point m is acted upon by a restoring elastic force
with the stiffness coefficient c, as well as a resistance force proportional to the
squared velocity—the quadratic friction with the coefficient μ. Then, according to
the method of [22] (see Sect. 5.4), the vector equation of relative motion can be
written as:
 
λ1 r_ r_ þ Ω2 r ¼
fω  ðω  ρÞ þ ω_  ρ þ ð1  m=MÞ½ω  ðω  rÞ þ ω_  r þ 2ω  r_ þ€rg:
ð5:108Þ

Here Ω2 ¼ c/m, λ1 ¼ μ/m, ρ is the radius vector of point O1, r is the radius vector
of the point m relative to O1, ω is the absolute angular velocity of the body, M is the
total mass of the rigid body and the moving mass, and the dot denotes the derivative
with respect to time t in the coordinate system connected with the body; then ρ is a
constant vector, and ω is some yet unknown function of time.
Our problem is to study the motion of the system, i.e., to determine the vectors
r and ω as functions of time for the arbitrary given initial conditions. In the general
case, the solution of the problem cannot be constructed. However, if we assume that
the coefficients of connection λ1 and Ω are such that the “free” motions of the point
m caused by initial deflections decay much faster than the body makes a revolution,
and then the motion of the body will be closed to the Euler–Poinsot motion, while
the relative oscillations of the point forced by this movement will be small.
If we take:
102 5 Perturbation Torques Acting upon a Rigid Body

λ1 ¼ ΛΩ3 , Ωω ðω ¼ jωjÞ,

then the “forced” motion of system (5.108) can be approximately written as the
expansion:
   
r ¼ Ω2 a þ ΛΩ3 a_ a_ þ O Ω4 ,
ð5:109Þ
a ¼ ω  ðω  ρÞ þ ω_  ρ:

As was noted, the dot means differentiation in the coordinate system associated
with the body. It is assumed in the sequel that the origin of this system is at the point
C, the center of inertia of the body and the mass m. Then the equation defining the
desired vector ω(t) is obtained from the condition of constancy of the angular
momentum of the system relative to point O and can be represented in the form
[22]:
 
J∗
Cω_ þ ω  J∗ _ þ ω  gÞ:
C ω ¼  ðg ð5:110Þ

Here J∗C is the inertia tensor of the rigid body together with the mass m located at
point O1, with respect to the center of inertia C. The quantity g can be convention-
ally called the vector of angular momentum of the moving mass m: it vanishes, if
the internal motions are absent, i.e., when r_ 0, r 0.
In the general case, in view of (5.109), vector g equals:
 
g ¼ m½ρ  ðω  r þ r_ Þ þ r  ðω  ρÞ þ O Ω4 : ð5:111Þ

Here, quantity r is calculated approximately according to (5.109), while the

derivative r_ , which is expressed in terms of ω_ , is found from the relation which
follows immediately from (5.110):
 1     
ω_ ¼  J∗
C ω  J∗
C ω þ O Ω2 : ð5:112Þ

Thus, quantity r_ is determined as a function of ω with the needed degree of

accuracy. The subsequent derivatives €r, ω€ are obtained similarly. As a result, to
determine the angular velocity vector ω from (5.110) on the basis of (5.111),
(5.112), we obtain the desired equation of the form [22]:
    
J∗
Cω _ þ ω  J∗ C ω ¼ ΦðωÞ þ O Ω4 : ð5:113Þ

Here, the function Φ(ω) is a polynomial containing the fourth and eighth powers
of vector ω and consists of summands of the order Ω2 and Ω3. In the general
case, function Φ is rather lengthy, and its expression is not given here.
Subsequently, for a specific problem, we will present an expression for pertur-
bation torque Φ(ω) and construct the solution of the system of motion equations.
5.8 Torque Due to the Solar Pressure 103

5.8 Torque Due to the Solar Pressure

In the motion of an artificial cosmic body along the orbit around the Earth and
especially around the Sun, it is possible that the force of light pressure exerts a
significant influence on its movement. In book [32], the moments of various forces
acting on a satellite are evaluated. For a satellite of the Earth with the characteristics
like that of the third Soviet satellite, it turns out that the torques due to light pressure
are greater than the gravitational ones at the altitudes exceeding 35,000–40,000 km.
For a spacecraft moving along the orbit around the Sun, the light pressure torques
are larger by several orders of magnitude than the gravitational ones.
Following [32–34], we present here the expressions for the light pressure torques
acting upon the surface of a satellite. The value of light pressure pc at the distance
R from the center of the Sun is determined by the formula:
 
E0 R 0 2
pc ¼ , ð5:114Þ
c R

where c is the velocity of light and E0 is the flow of the light pressure energy at the
distance R0 from the center of the Sun. If R0 is the radius of the Earth orbit, then
pc ¼ Ec0 ¼ 4, 64  106 mN2 .
For a satellite which is a body of revolution, the torque due to light pressure
can be approximately calculated by the formula [32, 34]:

ac ðεs ÞR20
L¼ e r  k0 : ð5:115Þ
R2
0
Here, er is the unit vector in the direction of the radius vector of the satellite; k is
the unit vector of the axis of symmetry of the satellite; εs is the angle between
0 0
vectors er and k , so that |er  k | ¼ sin εs; R is the current distance from the Sun’s
center to the center of mass of the satellite (a satellite of Sun is considered); R0 is
some fixed value of R; and ac(εs) is the coefficient of the light pressure torque.
We assume that ac ¼ ac(cosεs) and approximate ac by polynomials in powers of
cosεs. Specific expressions for ac(cosεs) have been considered in [32, 34] for a
satellite in the form of a body of revolution and in more general cases, in [35–37].
For the case of complete absorption, we have:

R20
ac ð εs Þ ¼ pc Sðεs Þz00 ðεs Þ: ð5:116Þ
R2

Here, S is the area of the “shadow” on the plane normal to the flow; z00 is the
distance from the center of mass to the center of pressure. The dimensions of ac and
L are the same.
Relation (5.115) gives an approximate analytical expression for the light pres-
sure torque. A more detailed account of these forces requires a description of the
104 5 Perturbation Torques Acting upon a Rigid Body

vehicle geometry, the properties of its surface, and the lighting conditions.
Such studies have been performed in a number of published works [38–40].

References

1. Beletsky, V.V.: Spacecraft Attitude Motion in Gravity Field. Moscow State Univ., Moscow
(1975) in Russian
2. Landau, L.D., Lifshitz, E.M.: Course of Theoretical Physics, Mechanics, vol. 1. Pergamon Press,
Oxford (1976)
3. Kurosh, A.G.: Course of Higher Algebra. Nauka, Moscow (1965) in Russian
4. Shilov, G.E.: Introduction to the Theory of Linear Spaces. Gostekhizdat, Moscow (1952) in
Russian
5. Koshlyakov, V.N.: On some particular cases of integration the dynamic Euler equations as
applied to the motion of a gyroscope in a resistive medium. Prikl. Mat. Mekh. 17(2), 137–148
(1953) in Russian
6. Koshlyakov, V.N.: Problems in Dynamics of Solid Bodies and in Applied Gyroscope Theory:
Analytical Methods. Nauka, Moscow (1985) in Russian
7. Leimanis, E.: The General Problem of the Motion of Coupled Rigid Bodies about a Fixed Point.
Springer, Berlin (1965)
8. Routh, E.J.: Advanced Dynamics of a System of Rigid Bodies. Dover, New York, NY (2005)
9. Akulenko, L.D., Leshchenko, D.D., Chernousko, F.L.: Fast motion of a heavy rigid body about
a fixed point in a resistive medium. Mech. Solids. 17(3), 1–8 (1982)
10. Neishtadt, A.I.: Evolution of rotation of a solid acted upon by the sum of a constant and a
dissipative perturbing moments. Mech. Solids. 15(6), 21–27 (1980)
11. I~narrea, M., Lanchares, V.: Chaotic pitch motion of an asymmetric non-rigid spacecraft with
viscous drag in circular orbit. Int. J. Nonlinear Mech. 41(1), 86–100 (2006)
12. Denisov, G.G.: On the rotation of a rigid body in a resistive medium. Izv. Akad. Nauk SSSR.
Mekh. Tverd. Tela. 4, 37–43 (1989) in Russian
13. Koshlyakov, V.N.: On stability of motion of a symmetric body placed on a vibrating base.
Ukr. Math. J. 47(12), 1898–1904 (1995)
14. Simpson, H.C., Gunzburger, M.D.: A two time scale analysis of gyroscopic motion with
friction. J. Appl. Math. Phys. 37(6), 867–894 (1986)
15. Sidorenko, V.V.: Capture and escape from resonance in the dynamics of the rigid body in
viscous medium. J. Nonlinear Sci. 4, 35–57 (1994)
16. Akulenko, L.D., Leshchenko, D.D., Chernousko, F.L.: Perturbed motions of a rigid body,
close to the Lagrange case. J. Appl. Math. Mech. 43(5), 829–837 (1979)
17. Aslanov, V.S.: The motion of rotating body in a resistive medium. Izv. Ros. Akad. Nauk.
Mekh. Tverd. Tela. 2, 27–39 (2005) in Russian
18. Chernousko, F.L.: Motion of a rigid body with cavities filled with viscous fluid at
small Reynolds number. USSR Comput. Math. Math. Phys. 5(6), 99–127 (1965)
19. Chernousko, F.L.: Motion of a Rigid Body with Viscous-Fluid-Filled Cavities. Computing Center
AN SSSR, Moscow (1968) in Russian
20. Chernousko, F.L.: The Movement of a Rigid Body with Cavities Containing a Viscous Fluid.
NASA, Washington, DC (1972)
21. Kobrin, A.I.: On the motion of a hollow body filled with viscous liquid about its center of
mass in a potential body force field. J. Appl. Math. Mech. 33(3), 418–427 (1969)
22. Chernousko, F.L.: On the motion of rigid body with moving internal masses. Izv. Akad. Nauk
SSSR Mekh. Tverd. Tela. 4, 33–44 (1973) in Russian
23. Lur’e, A.I.: Analytical Mechanics. Fizmatgiz, Moscow (1961) in Russian
References 105

24. Chernousko, F.L.: On the motion of solid body with elastic and dissipative elements. J. Appl.
Math. Mech. 42(1), 32–41 (1978)
25. Gantmakher, F.R.: Lectures on Analytical Mechanics. Chelsea, New York, NY (1974)
26. Chernousko, F.L., Shamaev, A.S.: The asymptotic behavior of singular perturbations in the
problem of the dynamics of a rigid body with elastic and dissipative elements. Izv. Akad. Nauk
SSSR Mekh. Tverd. Tela. 3, 33–42 (1983) in Russian
27. Chernousko, F.L., Shamaev, A.S.: Evolution equations for slow variables in the theory of
singularly perturbed system. Soviet Phys. Doklady. 29, 541–544 (1984)
28. Chernousko, F.L.: Motion of a viscoelastic solid relative to the center of mass. Mech. Solids.
15(1), 17–21 (1980)
29. Freudenthal, A., Geiringer, H.: Mathematical Theories of an Inelastic Continuum. Fizmatgiz,
Moscow (1962) in Russian
30. Lur’e, A.I.: Theory of Elasticity. Nauka, Moscow (1970) in Russian
31. Akulenko, L.D., Leshchenko, D.D.: Some problems of the motion of a solid with a
moving mass. Mech. Solids. 13(5), 24–28 (1978)
32. Beletsky, V.V.: Motion of an Artificial Satellite about its Center of Mass. Israel Program for
Scientific Translation, Jerusalem (1966)
33. Karymov, A.A.: Determination of forces and moments due to light pressure acting on a body in
motion in cosmic space. Prikl. Math. Mekh. 26(5), 867–876 (1962) in Russian
34. Karymov, A.A.: Stability of rotational motion of a geometrically symmetric artificial satellite
of the Sun in the field of light pressure forces. Prikl. Math. Mekh. 28(5), 923–930 (1964) in
Russian
35. Leshchenko, D.D., Shamaev, A.S.: Motion of a satellite relative to the center of mass under the
action of moments of light-pressure forces. Mech. Solids. 20(1), 11–18 (1985)
36. Akulenko, L.D., Leshchenko, D.D.: Evolution of rotation of a nearly dynamically spherical
triaxial satellite under the action of light pressure forces. Mech. Solids. 31(2), 1–10 (1996)
37. Leshchenko, D.D.: Evolution of rotation of a triaxial body under the action of the torque due to
light pressure. Mech. Solids. 32(6), 12–20 (1997)
38. Polyakhova, E.N.: Space Flight with a Solar Sail. Librokom Book House, Moscow (2011) in
Russian
39. Sazonov, V.V.: Motion of an asteroid relative to the centre of mass under the action of the
moment of light pressure forces. Astr. Vestnik. 28(2), 95–107 (1994) in Russian
40. Sazonov, V.V., Sazonov, V.V.: Calculation of resultant vector of light pressure forces acting
on the spacecraft with a solar sail. Cosmic. Res. 49(1), 56–64 (2011)
Chapter 6
Motion of a Satellite About Its Center of Mass
Under the Action of Gravitational Torque

The work [1] considers two cases of motion of a satellite under the action of
gravitational torques when the presence of a small parameter allows applying the
averaging method and obtaining asymptotic solutions.
In the first case, presented in Sect. 6.1, the values of three principal moments of
inertia of the satellite are assumed to be close to one another, though different; the
eccentricity of the orbit and the angular velocity of the satellite rotation are
arbitrary.
In Sect. 6.2, we explore the rapid motion of the satellite relative to its center of
mass under the action of gravitational torques in the case when no restrictions are
imposed on the eccentricity of the orbit and the moments of inertia.
In Sect. 6.3, based on paper [2], we discuss some resonance phenomena in the
dynamics of relative motion of the satellite in the orbital plane.

6.1 Motion of a Triaxial Satellite with Moments of Inertia

Close to One Another

In Sect. 4.6, three right-hand coordinate system were introduced, the origin of
which was placed at the satellite’s center of inertia (Figs. 4.1 and 4.2), and the
equations of motion of the satellite relative to the center of inertia were composed
(see (4.42), (4.46)). In the notation adopted in Sect. 4.6, these equations have the
form:
dG dδ L1 dλ L2
¼ L3 , ¼ , ¼ , ð6:1Þ
dt dt G dt G sin δ

© Springer International Publishing AG 2017 107

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_6
108 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .


1 1 L2 cos ψ  L1 sin ψ
θ_ ¼ G sin θ sin φ cos φ  þ ,

A 1 A 2  G
1 sin φ cos φ
2 2
L1 cos ψ þ L2 sin ψ
φ_ ¼ G cos θ   þ , ð6:2Þ

2 A3 A1  A2 G sin θ
sin φ cos 2 φ L1 cos ψ þ L2 sin ψ L2
ψ_ ¼G þ  ctg θ  ctg δ:
A1 A2 G G

Suppose that the principal central moments of inertia of the satellite are close to
one another, i.e., they are represented in the form (4.56):
A1 ¼ J 0 þ εA10 , A2 ¼ J 0 þ εA20 , A3 ¼ J 0 þ εA30 , ð6:3Þ

where 0 < ε  1 is a small parameter.

For ε ¼ 0, it follows from (5.17), (6.3) that sij ¼ J0δij (δij is the Kronecker
symbol), and then we obtain from (5.18) that L1 ¼ L2 ¼ L3 ¼ 0.
In fact, we have, for example, for L1:

3ω20 ð1 þ e cos vÞ3 X

3  
L1 ¼ σ 2 σ j s3j  σ 3 σ j s2j ¼
ð1  e2 Þ3 j¼1
3ω20 ð1 þ e cos vÞ3 X
3  
¼ J0 σ 2 σ j δ3j  σ 3 σ j δ2j ¼
ð1  e2 Þ3 j¼1
3
3ω20 ð1 þ e cos vÞ
¼ J 0 ðσ 2 σ 3  σ 3 σ 2 Þ ¼ 0:
ð1  e2 Þ3

In this case, we obtain from systems (6.1), (6.2) that G, δ, λ, θ, and φ are
constant, whereas:

ψ ¼ GJ 1
0 t þ ψ 0, ð6:4Þ

i.e., the satellite rotates uniformly about the translationally moving axis of angular
momentum.
The averaging procedure in this case is briefly described in Sect. 4.7.
To average the right-hand sides of (6.1), (6.2), we first calculate the average with
respect to ψ values for the functions sij, sij cos ψ, sij sin ψ, i , j ¼ 1 , 2 , 3, which will
be needed in the future. Denote the operation of averaging with respect to ψ by Mψ :

2ðπ
1
M ψ f Fð ψ Þ g ¼ Fðψ Þdψ: ð6:5Þ
2π
0

Using formulas (5.17) and (1.31), we find:

6.1 Motion of a Triaxial Satellite with Moments of Inertia Close to One Another 109

1  
Mψ ðs11 Þ ¼ Mψ ðs22 Þ ¼ A1 cos 2 φ þ cos 2 θsin 2 φ þ
2 
þ A2 ðsin 2 φ þ cos 2 θcos 2 φÞ þ A3 sin 2 θ ,
Mψ ðs33Þ ¼ A1 sin 2 θsin 2 φ þ A2 sin 2 θcos 2 φ þ A3 cos 2 θ,
Mψ sij ¼ 0, i 6¼ j,
Mψ ðsii cos ψ Þ ¼ Mψ ðsii sin ψ Þ ¼ Mψ ðs12 cos ψ Þ ¼ ð6:6Þ
¼ Mψ ðs12 sin ψ Þ ¼ 0, i ¼ 1, 2, 3,
1
Mψ ðs13 cos ψ Þ ¼ Mψ ðs23 sin ψ Þ ¼ ðA1  A2 Þ sin θ cos φ sin φ,
2
Mψ ðs23 cos ψ Þ ¼ Mψ ðs13 sin ψ Þ ¼
1  
¼ sin θ cos θ A1 sin 2 φ þ A2 cos 2 φ  A3 :
2

Now, substituting (6.6) into (5.18) and taking into account that all σ i do not
depend on ψ according to (5.19), we express the average with respect to ψ pro-
jections of external torque Li and their combinations included in the equation of
motion:

3ω20 ð1 þ e cos vÞ3

Mψ ðL1 Þ ¼ σ 2 σ 3 Mψ ðs33  s22 Þ,
ð1  e 2 Þ3
3ω0 ð1 þ e cos vÞ3
2
Mψ ðL2 Þ ¼ σ 1 σ 3 Mψ ðs11  s33 Þ,
ð1  e 2 Þ3
Mψ ðL3 Þ ¼ 0,
Mψ ðL1 cos ψ þ L2 sin ψ Þ ¼ ð6:7Þ
3ω2 ð1 þ e cos vÞ3  2 
¼ 0 3
σ 1 þ σ 22  2σ 23 Mψ ðs23 cos ψ Þ,
ð1  e Þ
2
Mψ ðL2 cos ψ  L1 sin ψ Þ ¼
3ω2 ð1 þ e cos vÞ3  2 
¼ 0 3
2σ 3  σ 21  σ 22 Mψ ðs13 cos ψ Þ:
ð1  e Þ
2

The time averaging of functions depending on the true anomaly v is reduced to

averaging over v according to (4.59).
Averaging the right-hand sides of equations (6.1), (6.2) in view of (5.17), (5.18),
and (4.40) first over ψ and then over v according to (4.59), we obtain the equations
of the first approximation in the form:
110 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

G_ ¼ 0, δ_ ¼ 0, ψ_ ¼ GJ 10 þ OðεÞ,
3ω20 cos δ 
λ_ ¼ 3=2
A1 þ A 2 þ A3 
4Gð1  e2 Þ
 3½ðA1 sin 2 φ þ A
2 cos 2 φÞsin
 θ þ A3 cos θ ,
2 2

1 1
θ_ ¼ G sin θ sin φ cos φ  þ
A1 A2
ð6:8Þ
3ω2 ð1  3cos 2 δÞ
þ 0 ðA1  A2 Þ sin θ sin φ cos φ,
4G
ð1  e2 Þ3=2 
1 sin 2 φ cos 2 φ
φ_ ¼ G cos θ   
A3 A1 A2
3ω2 ð1  3cos 2 δÞ  
 0 3=2
cos θ A1 sin 2 φ þ A2 cos 2 φ  A3 :
4Gð1  e Þ 2

To further simplify these equations, we note that, up to the quantities of the order
J2
of ε2, we have by (6.3) A11 ¼ J20  AJ21 , A1 ¼ 2J 0  A01 and similarly for the moments of
0
inertia A2, A3.
Using these approximate equalities, we transform equations (6.8) without
diminishing their order of accuracy with respect to ε:

3ω20 Φ cos δ
G_ ¼ 0, δ_ ¼ 0, ψ_ ¼ ω þ OðεÞ, λ_ ¼ ,
4G ð 1  e 2 Þ3=2


1 1
θ_ ¼ G sin θ sin φ cos φ  D ¼ OðεÞ, ð6:9Þ

A1 A2 
1 sin 2 φ cos 2 φ
φ_ ¼ G cos θ   D ¼ OðεÞ:
A3 A1 A2

Here:


6TJ 20 2 1 1 1
Φ ¼ 2  J0 þ þ  ε,
G A1 A 2 A3
ð6:10Þ
3ω ð1  3cos δÞ
2 2
G
D¼1 0 3=2
, ω ¼  1,
4ω2 ð1  e2 Þ J0

where the kinetic energy T of motion of the satellite relative to the center of mass is
defined by formula (4.48).
The solution of system (6.9), (6.10) approximates the exact solution of system
(6.1) (6.2) on the time interval of the order of T0ε1, which is much larger than the
period of rotation of the satellite, with an accuracy of the order of ε for slow
variables G, δ, λ, θ, φ and with an accuracy of the order of 1 for ψ. The time
dependence of the true anomaly v is determined by equation (4.52), which can be
easily integrated.
6.1 Motion of a Triaxial Satellite with Moments of Inertia Close to One Another 111

The relative motion of the satellite, described by equations (6.9), is subdivided

into three parts: rapid motion (variable ψ) and two slow motions (variables θ, φ and
G, δ, λ).
Rapid motion is a rotation of the satellite about the angular momentum vector
with the angular velocity ω ¼ GJ 1 0 which is constant due to the first equation of
(6.9).
The equations for variables θ, φ describe the motion of the angular momentum
vector relative to the satellite. The projections Gi of this vector on the principal
central axes of inertia Ozi are given by formulas (4.45).
The considered equations (the last two in system (6.9)) differ from the equations
for θ, φ in the case of inertial motion of a body (the first two equations (6.2) for
L1 ¼ L2 ¼ L3 ¼ 0, the Euler–Poinsot case) only by the factor D from (6.10) which is
constant in this approximation. Therefore, the equations for θ, φ admit the energy
integral T ¼ const, as can be easily seen also by the direct averaging of the right-
hand side of the equation for the derivative of kinetic energy obtained from (4.48)
with the help of (6.1), (6.2):


2 
_ 2T sin φ cos 2 φ 1
T ¼ L3 þ G sin θ cos θ þ  ðL2 cos ψ  L1 sin ψ Þþ
G A1 A2 A3

 
1 1
þ sin φ cos φ  ðL1 cos ψ þ L2 sin ψ Þ : ð6:11Þ
A1 A2

Thus, the action of gravitational torques only changes (by a constant number
D of times) the velocity of movement of the angular momentum vector G along the
trajectories of its motion in the Euler–Poinsot case. These trajectories are defined by
the relations following from (4.45) and (4.48) (see also (4.29) (4.30)):

G21 þ G22 þ G23 ¼ G2 , G21 A1 2 1 2 1

1 þ G2 A2 þ G3 A3 ¼ 2T,

where G and T are constant. A number of such trajectories at the fixed T and various
G are shown in Fig. 2.1, where the arrows indicate the direction of motion in the
EulerPoinsot case and we suppose A1 > A2 > A3. The quantity D in (6.10) can
vary within wide limits, taking both positive and negative values, which corre-
sponds to a change of direction in Fig. 2.1. For ωω0 ! 1, we have D ! 1, i.e., during
relatively fast rotation relative to the center of inertia, the motion of the satellite is
transformed into the EulerPoinsot motion. Note that, as in the EulerPoinsot
case, the permanent rotation axes are the principal axes of inertia, whereas the
rotation about axis Oz2 is unstable, and about Oz1, Oz3, stable (see Fig. 2.1; this is
true both for D > 0 and for D < 0).
112 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

However, if the condition:

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
G 3ð1  3cos 2 δÞ
ω¼ ¼ ω0 ð6:12Þ
J0 4ð1  e2 Þ3=2

holds, we get D ¼ 0 in (6.10), and the rotation of the satellite is stationary for any
orientation of the axis of rotation relative to the satellite. In this case, the centrifugal
and gravitational torques are balanced in the first approximation, and the angular
momentum vector G does not move relative to the body. Condition (6.12) shows
that it can be achieved for any orientation of vector G in space by the choice of the
magnitude of the angular momentum G ¼ J0ω, provided the angle π2  δ between the
angular momentum vector and the orbital plane satisfies the inequality:
π rffiffiffi
2
 δ  arcsin  54∘ 440 :
2 3

The movement of the angular momentum vector in space is described by the first
three equations (6.9). By virtue of constancy of G, δ, T and Φ from (6.10), this
movement is a uniform rotation of the vector of the constant magnitude G about the
normal to the orbital plane at a constant angular distance δ from it. The angular
velocity λ_ is a quantity of the order of εω20 ω1 (since Φ  εJ0), while its sign
depends on the character of motion of G relative to the satellite. For example, if
δ < π/2, then in the case of the satellite rotating about the axis of the largest moment
of inertia Oz1 (G2 ¼ 2TA1), we get from (6.10) Φ < 0, λ_ < 0 (rotation of G in the
direction opposite to the orbital motion), whereas for G2 ¼ 2TA3 (rotation about
Oz3), we have λ_ > 0. If the angular momentum vector lies in the plane of the orbit
(δ ¼ π/2), then it remains fixed in space (λ ¼ const).
Let us emphasize that the analysis carried out in this section is based on a single
assumption (6.3) of the proximity to one another of the principal central moments of
inertia of the satellite. The magnitude ω of angular velocity can be larger, smaller,
or of the same order as the angular velocity ω0 of the orbital motion. The eccen-
tricity of the orbit is also arbitrary (0  e < 1).

6.2 Fast Rotations of a Satellite with a Triaxial Ellipsoid

of Inertia

Consider now a satellite, whose moments of inertia are arbitrary (A1 > A2 > A3),
assuming that the angular velocity ω of the satellite’s motion relative to the center
of mass is substantially greater than the angular velocity of the orbital motion ω0,
i.e., we assume that ε ¼ ωω0  A1Gω0  1. In this case, the kinetic energy of rotation of
6.2 Fast Rotations of a Satellite with a Triaxial Ellipsoid of Inertia 113

the satellite is large in comparison with the work of the gravitational torques during
the satellite’s revolution about the center of mass.
Assume that the unit for measuring time and the period of relative motion 2π ω are
of the same order, then ω0  ε and Li ¼ O(ε2), i ¼ 1 , 2 , 3.
The unperturbed motion (ε ¼ 0) will be the Euler–Poinsot motion, while the
quantities G, δ, λ, and T are constant.
The slow variables (x) (in the terminology of Chap. 4) in the perturbed motion
(ε 6¼ 0) will be G, δ, λ and T, while the fast ones ( y) will be φ, θ, ψ.
As shown in Chap. 4, the equations of perturbed motion of a satellite can be
reduced to the system (4.53) with two rotating phases of the form (4.54). If v were a
fixed constant in the system (4.54), then ε2 could be taken as a small parameter.
Then, in the first approximation in ε2, the solution for slow variables would be
determined, according to (4.16), (4.17), by the system:

x_ ¼ ε2 Mt fXðx; ω1 ðt þ t1 Þ; ω2 ðt þ t2 Þ; vÞg: ð6:13Þ

Here Mt is determined according to (4.14); instead of y1, y2, we substitute their

values (4.49) for the unperturbed motion. It is expected that the result of averaging
in (6.13) does not depend on the arbitrary constants t1, t2, and then the error of the
first approximation is of the order of ε2 on the interval of the order of ε2.
Let us show that system (6.13) approximates the equations for the slow variables
in (4.54) with the same accuracy also in the case when v is not constant but
determined by (4.54). The solution for variables x, y1, y2 of system (4.54) is sought
for in the form (4.15):

x ¼ x þ εu1 ðx; y1 ; y2 ; vÞ þ ε2 u2 ðx; y1 ; y2 ; vÞ þ . . . ,

y1 ¼ y1 þ εv1 ðx; y1 ; y2 ; vÞ þ . . . , ð6:14Þ
y2 ¼ y2 þ εw1 ðx; y1 ; y2 ; vÞ þ . . .

Functions ui, vi , wi are assumed to be periodic in y1 , y2 with the periods of 2π.
Variable v is not expanded into a series. New variables x, y1 , y2 (the first of them
is vectorial) must satisfy the system of equations (analogously to (4.16)):

x_ ¼ εA1 ðx; vÞ þ ε2 A2 ðx; vÞ þ ε3 A3 ðx; vÞ þ . . . ,

y_ 1 ¼ ω1 ðxÞ þ εB1 ðx; vÞ þ . . . , ð6:15Þ
y_ 2 ¼ ω2 ðxÞ þ εC1 ðx; vÞ þ . . .

Substituting (6.14) into (4.54), we express the derivatives of x, y1 , y2 according to
(6.15) (and also take into account the equation for v in (4.54)) and then equate the
coefficients at the successive powers of ε in the obtained equations. Equating the
coefficients at εn, we get the relations:
114 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

∂un ∂un
ω1 þ ω2 ¼ Pn ðx; y1 ; y2 ; vÞ  An ðx; vÞ,
∂
y1 ∂y2
∂vn ∂vn
ω1 þ ω2 ¼ Qn ðx; y1 ; y2 ; vÞ  Bn ðx; vÞ, ð6:16Þ
∂
y1 ∂
y2
∂wn ∂wn
ω1 þ ω2 ¼ Rn ðx; y1 ; y2 ; vÞ  Cn ðx; vÞ:
∂
y1 ∂y2

Here the functions Pn, Qn, Rn depend on f, Ak, Bk, Ck, uk, vk , wk, k ¼ 1 , 2 , . . . ,
n  1, and their derivatives. Furthermore, Pn depends on function X and its deriv-
atives, whereas Qn, Rn depends on Z1, Z2, w1, w2 and their derivatives, as well as on
un. Thus, at each step, it is first necessary to determine An from system (6.16) and un
from the first (vector) equation and then Bn, vn , Cn, wn.
All equations (6.16) have the form of the following scalar equation:

∂u ∂u
ω1 þ ω2 ¼ Pðy1 ; y2 Þ  A, ð6:17Þ
∂
y1 ∂
y2

where P is periodic in its arguments with the period 2π. We omit the dependence on
x and v, since these variables are included in equations (6.16) as parameters. We
expand function P and the desired function u, which is sought as a periodic function
of y1 , y2 with the periods 2π, in a double Fourier series in the complex form:
X
P ¼ Pms expðim
y1 þ is
y2 Þ,
Xm, s
ð6:18Þ
u ¼ ums expðim
y1 þ is
y2 Þ:
m, s

The summation in (6.18) is performed over all integers m, s. Substituting (6.18)

into (6.17) and equating the coefficients at the same harmonics, we find:
2ðπ 2ðπ
1
A ¼ P00 ¼ 2 Pðy1 ; y2 Þ d
y1 d
y2 , ð6:19Þ
4π
0 0
 
iðω1 m þ ω2 sÞums ¼ Pms m2 þ s2 6¼ 0 : ð6:20Þ

It is seen from (6.20) that if the frequencies ω1, ω2 are incommensurable, i. e.:

mω1 þ sω2 6¼ 0 ð6:21Þ

for all integers m, s, not both equal to zero, then the Fourier coefficients of function
u can be determined, and this function itself can be found up to an arbitrary
constant. Formula (6.19) determines the value of A.
6.2 Fast Rotations of a Satellite with a Triaxial Ellipsoid of Inertia 115

Thus, under condition (6.21), functions An, Bn, Cn, un, vn , wn can be determined
from (6.16), whereas the asymptotic solutions (6.14), (4.51) can be obtained in any
approximation. The functions of x and v, appearing at each step (similar to arbitrary
constant in solving (6.17)), can be chosen arbitrary; they will not affect the final
result [3].
Since frequencies ω1, ω2 depend on x, the condition of their incommensurability
(6.21) may be violated (resonance phenomena). However, for the large (in absolute
value) numbers m, s, the resonances are manifested weakly. It is quite obvious from
physical considerations and due to the fact that the amplitudes of the higher
harmonics of the Fourier series for the right-hand sides of equations of motion
are small. We assume that condition (6.21) holds for small m, s in the absolute value
at the initial moment of time with a certain margin, i.e., the right-hand side of (6.21)
is not close to zero. We demonstrate below that variables G and T, on which
frequencies ω1, ω2 depend, are constant in the first approximation. Then condition
(6.21) for small m, s will be fulfilled on the entire interval of motion. Thus, here, as
in Sect. 6.1, the resonance phenomena are not considered. Note that such approach
(ignoring the higher resonances) is not very strict, but it is justified physically.
We write the first equation (6.16) for n ¼ 1. For this equation, we get after the
performed transformations P1 ¼ 0, and the equation assumes the form:

∂u1 ∂u1
ω1 þ ω2 ¼ A1 ðx; vÞ:
∂
y1 ∂
y2

It can be seen from this that we can put:

A1 ¼ 0, u1 ¼ 0: ð6:22Þ

The second and third equations (6.16) for n ¼ 1 yield in view of (6.22):

∂v1 ∂v1
ω1 þ ω2 ¼ B1 ðx; vÞ,
∂
y1 ∂
y2
∂w1 ∂w1
ω1 þ ω2 ¼ C1 ðx; vÞ,
∂
y1 ∂
y2

from which it follows:

B1 ¼ C1 ¼ 0, v1 ¼ w1 ¼ 0: ð6:23Þ

For n ¼ 2 the first equation (6.16) assumes the form (in view of (6.22), (6.23)):

∂u2 ∂u2
ω2 þ ω2 ¼ Xðx; y1 ; y2 ; vÞ  A2 ðx; vÞ:
∂
y1 ∂
y2

We obtain from this, in accordance to (6.19):

116 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

2ðπ 2ðπ
1
A2 ðx; vÞ ¼ Xðx; y1 ; y2 ; vÞ d
y1 d
y2 , ð6:24Þ
4π 2
0 0

while the arbitrary function, included in u2, is chosen in such a way that the
condition:

2ðπ 2ðπ

u2 ðx; y1 ; y2 ; vÞd y2 ¼ 0

y1 d ð6:25Þ
0 0

holds identically for x, v. We also write the first equation (6.16) for n ¼ 3 in view of
(6.22), (6.23):

∂u3 ∂u3 ∂u2

ω1 þ ω2 ¼ f ðvÞ  A3 ðx; vÞ:
∂
y1 ∂
y2 ∂v

From this we determine A3 according to (6.19):

2ðπ 2ðπ
1 ∂
A3 ðx; vÞ ¼  2 f ðvÞ u2 ðx; y1 ; y2 ; vÞd y2 :
y1 d
4π ∂v
0 0

Taking the differentiation out of the integral sign and taking into account
equality (6.25), we obtain:
A3 ¼ 0: ð6:26Þ

A solution for x will be sought in the form:

x ¼ x, x ¼ ε2 A2 ðx; vÞ: ð6:27Þ

In the right-hand side of the first equation (6.15), we drop the terms of order ε4
(since A1 ¼ A3 ¼ 0), and so the error in the determination of x will not exceed ε2 on
the interval of the order of ε2. It can be seen from expansion (6.14) that x
approximates x with an error of the order of ε2 (since u1 ¼ 0). However, under
condition (6.24), system (6.27) coincides with system (6.13). So, the solution of
system (6.13) or (6.27), indeed, approximates x with an error of the order of ε2 on
the time interval of the order of ε2 that corresponds to the number of revolutions of
the satellite along the orbit of the order of ε1 (because the rate of change of angle v
is of the order of ε).
To construct the averaged system (6.13), we need to average the right-hand sides
of equations of motion (for fixed slow variables and v) over phases y1, y2 as the
6.2 Fast Rotations of a Satellite with a Triaxial Ellipsoid of Inertia 117

independent variables (see (6.24)). In other words, we need to calculate integrals of

the type:

ð 2π
2π ð
1
Mt fFðθ; φ; ψ Þg ¼ 2 Fðθðy1 Þ; φðy1 Þ; ψ 1 ðy1 Þ þ y2 Þdy1 dy2 ,
4π
0 0

where the dependence of the Euler angles on y1, y2 is given by the formulas of the
EulerPoinsot motion (4.51), whereas the dependence on the slow variables is
omitted. Substituting the variables of integration (y2 by ψ and then y1 by t according
to (4.49)), we obtain (see (4.55)):
ð 2π
2π ð
1
Mt fFðθ; φ; ψ Þg ¼ 2 Fðθðy1 Þ; φðy1 Þ; ψ Þdy1 dψ ¼
4π
0 0

ðτ 2ðπ
1 
¼ FðθðtÞ; φðtÞ; ψ Þ dψ dt ¼ M1 Mψ ½Fðθ; φ; ψ Þ : ð6:28Þ
2πτ
0 0

Here, Mψ means averaging over ψ according to (6.5), while M1, over θ and φ
connected by relation (4.48), carried out along closed trajectories of the angular
momentum vector in the EulerPoinsot motion for fixed G, T(Fig. 2.1).
Let us average the right-hand sides of equations (6.1), (6.11) according to the
scheme (6.28), using formulas (6.6), (6.7), in which the necessary averaging over ψ
has been already carried out. We obtain an averaged system of the form:

3ω2 ð1 þ e cos vÞ3  2 

G_ ¼ 0, T_ ¼ 0 3
2σ 3  σ 21  σ 22 
2ð 1  e 2 Þ
ð A1  A2 Þ ð A2  A 3 Þ ð A3  A1 Þ
 M1 fG1 G2 G3 g,
A1 A2 A3 G2
ð6:29Þ
3ω2 ð1 þ e cos vÞ3
δ_ ¼  0 σ 2 σ 3 N,
2Gð1  e2 Þ3
3ω2 ð1 þ e cos vÞ3
λ_ ¼ 0 σ 1 σ 3 N:
2Gð1  e2 Þ3 sin δ

Here, Gi are the projections of the angular momentum on axes Ozi (4.45), while:

N ¼ 2M1 Mψ ðs11  s33 Þ ¼ M1 A1 þ A2 þ A3 
  
3 A1 sin 2 φ þ A2 cos 2 φ sin 2 θ þ A3 cos 2 θ : ð6:30Þ

Because of the symmetry of the trajectories of vector G with respect to the

coordinate planes of coordinate system Oz1z2z3, we have for the Euler–Poinsot
motion:
118 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

M1 fG1 G2 G3 g ¼ 0, T ¼ const:

The quantity N from (6.30) for the moments of inertia A1, A2, A3, close to one
another (condition (6.3)), coincides with Φ from (6.10). In the general case, using
(4.45) and (4.48), it can be transformed to the form:

N ¼ A1 þ A2 þ A3  3G2 M1 A31 p2 þ A32 q2 þ A33 r 2 ¼ A2  2A1 
 
2A3 þ 6A1 A3 TG 2 þ 3A2 ðA1  A2 ÞðA2  A3 ÞG2 M1 q2 : ð6:31Þ

We substitute into equation (6.31) the quantity q(t) from (2.30) for the
EulerPoinsot motion and average function q2 over its period. Finally, for the
trajectories (Fig. 2.1) of the vector G encircling axis Oz1 (2TA1 G2>2TA2), we
have:
N ¼ A2 þ A3  2A1 þ



2TA1 K ð k Þ  Eð k Þ
þ3  1 A 3 þ ð A 2  A 3 Þ : ð6:32Þ
G2 k 2 K ðk Þ

Here, K(k), E(k) are the complete elliptic integrals of the first and second kind,
respectively, and:
 
ðA2  A3 Þ 2TA1  G2
k ¼
2
 : ð6:33Þ
ðA1  A2 Þ G2  2TA3

For the trajectories of vector G encircling axis Oz3 (2TA2 > G2 2TA3), one just
needs to swap A1 and A3 in formulas (6.32), (6.33).
The quantity N depends on the moments of inertia of the satellite and on the ratio
G2/T which determines the trajectory in Fig. 2.1; this quantity is constant in the
considered approximation. For the satellite rotation about the axes Oz1 (G2 ¼ 2TA1)
and Oz3 (G2 ¼ 2TA3), we, respectively, obtain from (6.32):

N ¼ A2 þ A3  2A1 < 0, N ¼ A2 þ A1  2A3 > 0:

In the case of dynamic symmetry (A1 ¼ A2), formulas (6.31) and (4.48) yield:

N ¼ 6A1 A3 TG 2  A1  2A3 ¼ ðA1  A3 Þð2  3sin 2 θÞ,

ð6:34Þ
θ ¼ const:

In the approximation under consideration, the relative motion of the satellite

consists of the EulerPoinsot motion about the angular momentum vector G (with
constant G and T ) and the movement of the vector G itself in space described by
equation (6.29) for δ, λ. Let us study these equations, taking as an independent
variable the true anomaly v. Using (5.19), (4.52), and (6.29), we write the equations
for δ, λ as follows:
6.2 Fast Rotations of a Satellite with a Triaxial Ellipsoid of Inertia 119

dδ
¼ χ ð1 þ e cos vÞ sin δ sin ðλ  vÞ cos ðλ  vÞ,
dv
dλ
¼ χ ð1 þ e cos vÞ cos δcos 2 ðλ  vÞ, ð6:35Þ
dv
3ω0 N
χ ¼ :
2ð1  e2 Þ3=2 G

Obviously, the introduced dimensionless constant quantity χ is of the order of ε

and is constant in the considered approximation (due to the constancy of G, T and N ).
In the case of a circular orbit (e ¼ 0), equations (6.35) have the first integral:

1
cos δ þ χsin 2 δcos 2 ðλ  vÞ ¼ const,
2

and their integration is reduced to quadrature.

However, an easier way to solve (6.35) is to use again the asymptotic methods.
As it was mentioned above, the derived averaged equations determine the quantities
δ, λ with an error of the order of ε2 (or χ 2) on the interval Δv  ε1  χ 1. Therefore,
it suffices to solve equations (6.35) with this accuracy, and to this end, one needs to
find the asymptotic solution in the second approximation with respect to χ.
System (6.35) is written as:

dδ
¼ χXðv; δ; λÞ,
dv
dλ
¼ χY ðv; δ; λÞ, χ  1,
dv

where X, Y are periodic in v with the period 2π and is a system in the standard form
[3]. Its solution in the second approximation is sought in the form:

δ ¼ ξ þ χuðv; ξ; ηÞ, λ ¼ η þ χwðv; ξ; ηÞ, ð6:26Þ

where u, w satisfy the equations:

du
¼ Xðv; ξ; ηÞ  Mv fXðv; ξ; ηÞg,
dv
dw
¼ Y ðv; ξ; ηÞ  Mv fY ðv; ξ; ηÞg,
dv
Mv fuðv; ξ; ηÞg ¼ Mv fwðv; ξ; ηÞg ¼ 0,

where Mv means the averaging with respect to v over the period 2π. Functions ξ, η
satisfy the system of the second approximation:
120 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .


dξ dX dX
¼ χMv ðXÞ þ χ Mv u þ w
2
,
dv
dξ dη
dη dY dY
¼ χMv ðY Þ þ χ 2 Mv u þ w :
dv dξ dη

We get for system (6.35):

1
Mv ðXÞ ¼ 0, M v ðY Þ ¼ cos ξ,
2
sin ξ
u¼ ½3 cos ð2v  2ηÞ þ 3e cos ðv  2ηÞ þ e cos ð3v  2ηÞ, ð6:37Þ
12
cos ξ
w¼ ½3 sin ð2v  2ηÞ þ 6e sin v þ 3e sin ðv  2ηÞ þ e sin ð3v  2ηÞ,
12

whereas the equations of the second approximation take the form:

dξ 1
¼ χ 2 e2 sin ξ cos ξ sin 2η,
dv 8
 ð6:38Þ
dη 1 1   2
¼ χ cos ξ þ χ 2 3cos 2 ξ  1 1 þ e2 þ e2 cos 2η :
dv 2 16 3

Dividing the second equation (6.38) by the first, we obtain a linear equation for
cos2η as a function of ξ. Integrating it, we find the first integral of the system (6.38):


 
1 2
sin 2 ξ 1 þ χ 1 þ e2 þ e2 cos 2η cos ξ ¼ const,
4 3

and then integration of the system is reduced to quadrature.

However, the solution of system (6.38) with the required accuracy (an error of
the order of χ 2 on the interval of the order of χ 1) can be found using the following
general technique.
Let it be required to find the solution of the system of second approximation:
dz
¼ χA1 ðzÞ þ χ 2 A2 ðzÞ, ð6:39Þ
dt

where z, A1, A2 are vector functions and χ  1 is a small parameter, under the initial
condition z(0) ¼ z0, whereas the solution error should not exceed χ 2 in its magnitude
order on the time interval of the order of χ 1. The general solution of the system of
first approximation:

dz
¼ χA1 ðzÞ ð6:40Þ
dt
6.2 Fast Rotations of a Satellite with a Triaxial Ellipsoid of Inertia 121

is assumed to be known. We denote by z1(t) its particular solution satisfying the

condition z1(0) ¼ z0. Since the right-hand sides of systems (6.39) and (6.40) differ
by the quantities of the order of χ 2, then the solutions of these systems coinciding
for t ¼ 0 will differ by a quantity of the order of χ on the time interval of the order of
χ 1, that is, z  z1 ¼ O(χ). Taking into account this estimate, system (6.39) can be
rewritten in the form:

dz
¼ χA1 ðzÞ þ χA01 ðz1 Þðz  z1 Þ þ χ 2 A2 ðz1 Þ, ð6:41Þ
dt

where we omit the terms of order of χ 3 in the right-hand side, which will introduce
an error into the solution not exceeding the required limit. The term A01 ðz1 Þðz  z1 Þ
here denotes the product of the matrix of partial derivatives and the vector. System
(6.41) is simpler than the original one (6.39), since it is linear and its solution can be
found by the method of variation of arbitrary constants, if the general solution of the
homogeneous system is known. But the homogeneous system corresponding to
system (6.41) is a system in variations for the equations of the first approximation
(6.40), while its fundamental system of solutions is obtained by differentiating the
general solution of system (6.40) with respect to the initial data (see, e.g., [4]).
Thus, solving the system of the second approximation with the specified accu-
racy is reduced to solving the system of the first approximation and quadratures.
Similar simplifications take place also for the systems of higher approximations.
Coming back to system (6.38), we first find that the solution of the system of the
first approximation, satisfying arbitrary initial data:

ξ ð 0Þ ¼ ξ 0 , ηð0Þ ¼ η0 , ð6:42Þ

has the form:

1
ξ ¼ ξ0 , η ¼ η0 þ χv cos ξ0 : ð6:43Þ
2

We simplify system (6.38) with the help of (6.43) according to the scheme
(6.41):

dξ 1
¼ χ 2 e2 sin ξ0 cos ξ0 sin ð2η0 þ χv cos ξ0 Þ,
dv 8
dη 1 1
¼ χ cos ξ0  χ ðξ  ξ0 Þ sin ξ0 þ ð6:44Þ
dv 2 2

1   2
þ χ 2 3cos 2 ξ0  1 1 þ e2 þ e2 cos ð2η0 þ χv cos ξ0 Þ :
16 3

The solution of system (6.44) satisfying the formulated initial conditions (6.42)
can be obtained by two quadratures and has the form:
122 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

1
ξ ¼ ξ0 þ χ 2 e2 sin ξ0 ½ cos 2η0  cos ð2η0 þ χv cos ξ0 Þ,
8
1
η ¼ η0 þ χv cos ξ0 þ
2

  ð6:45Þ
1 2   2
þ χ v 3cos 2 ξ0  1 1 þ e2  e2 sin 2 ξ0 cos 2η0 þ
16 3
1 2
þ χe cos ξ0 ½ sin ð2η0 þ χv cos ξ0 Þ  sin 2η0 :
8

Substituting (6.37) and (6.45) into (6.26), we obtain the desired expressions for
δ, λ (an asymptotic solution of system (6.35)) in the form:

1 
δ ¼ ξ0 þ χ sin ξ0 6 cos ½ð2  χ cos ξ0 Þv  2η0 þ
24
þ 6e cos ½ð1  χ cos ξ0 Þv  2η0  þ 2e cos ½ð3  χ cos ξ0 Þv  2η0 þ
þ 3e2 cos 2η0  3e2 cos ðχv cos ξ0 þ 2η0 Þ ,
1
λ ¼ η0 þ χv cos ξ0 þ
2

  ð6:46Þ
1 2   2
þ χ v 3cos 2 ξ0  1 1 þ e2  e2 sin 2 ξ0 cos 2η0 þ
16 3
1 
þ χ cos ξ0 6 sin ½ð2  χ cos ξ0 Þv  2η0  þ 12e sin vþ
24
þ 6e sin ½ð1  χ cos ξ0 Þv  2η0  þ 2e sin ½ð3  χ cos ξ0 Þv  2η0 þ
þ 3e2 sin ðχv cos ξ0 þ 2η0 Þ  3e2 sin 2η0 :

This solution differs from the exact one by the quantities of the order of χ 2 on the
interval Δv  χ 1, where ξ0 and η0 are arbitrary constants. They are connected with
the values of δ, λ for v ¼ 0 (which we denote by δ0, λ0) by the formulas resulting
from (6.46):
1  
ξ0 ¼ δ0  χ sin δ0 ð6 cos 2λ0 þ 6e cos 2λ0 þ 2e cos 2λ0 Þ þ O χ 2 ,
24
1  
η0 ¼ λ0 þ χ cos δ0 ð6 sin 2λ0 þ 6e sin 2λ0 þ 2e sin 2λ0 Þ þ O χ 2 :
24

However, if we limit ourselves by the accuracy of the order of χ, then solution

(6.46) will be simplified and will take the form:

1
δ ¼ ξ0 , λ ¼ η0 þ χv cos ξ0 : ð6:47Þ
2

This solution of the first approximation describes the rotation of the angular
momentum vector under a constant δ about the normal to the orbital plane with the
velocity which is equal, according to (6.35), to:
6.2 Fast Rotations of a Satellite with a Triaxial Ellipsoid of Inertia 123

3ω20 N cos δ
λ_ ¼ : ð6:48Þ
4Gð1  e2 Þ3=2

In particular, for A1 ¼ A2 and e ¼ 0, using expression (6.34) for N, we arrive at

the formula derived in [5].
The trajectories of the trace of the angular momentum vector on the unit sphere,
fixed in coordinate system Ox1x2x3 in the first approximation are the circles
δ ¼ const. In the second approximation, there appear oscillations of the angles δ
and λ, whereas with the increase in the average value of δ (i.e., ξ0) from 0 to π/2, the
oscillation amplitude of δ increases, while the amplitude of oscillations of λ, as well
as the average angular velocity of rotation (6.48), decreases. The total rate λ_ does
not change sign for δ 6¼ π/2 and vanishes only at isolated points where δ_ ¼ 0 as well
(see (6.35)). For δ  π/2 the changes of λ are the quantities of the second order of
smallness in comparison with the changes of δ. The trajectories of vector G on the
unit sphere, fixed relative to Ox1x2x3, are depicted in Fig. 6.1 taking into account the
indicated properties, where it is assumed that χ > 0 (for χ < 0 only the direction of
motion along these trajectories changes; it is indicated by arrows).
On the circular orbit (e ¼ 0), the oscillations of δ and λ are nearly sinusoidal with
angular frequency equal to twice the angular velocity of the orbital motion, while
the curves in Fig. 1.1 are close to cycloids subjected to contraction or dilatation
along the coordinate axes. In the case of elliptical orbit, the oscillations of δ and λ
get complicated: the first and third harmonics appear, as well as a substantial
dependence of the oscillation mode on the given initial η0; however, the basic
properties of the trajectories in Fig. 6.1 (in particular, the presence of cusps directed
toward the poles) are preserved.
The applicability regions of the asymptotic solutions from Sects. 6.1 and 6.2
clearly overlap: for ω
ω0 (rapid relative motion) and the close to one another A1,
A2, A3, the results of Sect. 6.1 coincide with the first approximation from Sect. 6.2.

Fig. 6.1 Trajectories of the

angular momentum vector
on the unit sphere
124 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

6.3 Resonance Phenomena in the Planar Motion

of a Satellite About Its Center of Mass

In the motion of a satellite along an orbit which is far enough from the Earth’s
surface, the major impact on it is exerted by the gravitational forces and torques.
With a high degree of accuracy (up to the quantities of the order of squared ratio of
the satellite’s linear dimensions to the orbit size), it can be assumed that the
movement about the center of inertia does not affect the movement of the center
of inertia itself which moves along a Keplerian orbit (ellipse).
The gravitational torque relative to the center of mass, caused by the gravita-
tional field inhomogeneity, depends both on the orientation of the satellite and its
position on the orbit.
The relative equilibrium and libration of a satellite moving along a circular orbit
have been studied by V.V. Beletsky [6]. He also noted the possibility of resonance
in the oscillations of a satellite on an elliptical orbit.
Note that the solutions below are the first terms of asymptotic expansions with
respect to a small parameter. In principle, the subsequent terms of these expansions
can also be obtained. The oscillations amplitude is not assumed to be small. The
applied method allows considering the essentially nonlinear oscillations and rota-
tions of the satellite.
Consider a planar motion of the satellite relative to the center of inertia which
moves along an elliptical orbit in a central gravitational field. Let the principal
central axis of inertia of the satellite, the moment of inertia with respect to which
equals A2, be perpendicular to the orbital plane at all times. The moments of inertia
with respect to the other two principal central axes of inertia are denoted by A1, A3
(A1 A3).
Up to the quantities of the order of the ratio of satellite’s dimensions to the orbit
size, the equation of motion of the satellite relative to the center of mass has the
form [6]:

d2 δ dδ
ð1 þ e cos vÞ 2
 2e sin v þ 3a2 sin δ ¼ 4e sin v: ð6:49Þ
dv dv

Here, δ ¼ 2θ is twice the angle between the radius vector of the center of mass
and the axis of inertia, the moment of inertia with respect to which is A3, a2
¼ (A1  A3)/A2, e is the orbit eccentricity, and v is the angular distance between
the radius vector and the perigee of the orbit (Fig. 6.2). Note that the well-known
inequality for the moments of inertia A1  A2 + A3 yields a  1.
Equation (6.49) is a nonlinear differential equation of the second order
containing two numerical parameters a, e in the interval [0, 1]. In the case of a
circular orbit (e ¼ 0), equation (6.49) is reduced to the equation of motion of a
pendulum and can be integrated in elliptic functions. In the case of a ¼ 0 (the
satellite is a body with dynamic symmetry, A1 ¼ A3, the axis of symmetry of which
is perpendicular to the orbit plane), equation (6.49) can be integrated in elementary
6.3 Resonance Phenomena in the Planar Motion of a Satellite About Its Center of Mass 125

Fig. 6.2 Planar motion of

the satellite

functions. Moreover, the gravitational torque turns to zero, and the motion of the
satellite is a uniform rotation about the axis of dynamic symmetry.
Below, following [2], we investigate three cases for which an asymptotic
solution is constructed:
1. e  1, the orbit is nearly circular.
2. a  1, the satellite is close to a body with dynamic symmetry.
3. |dδ/dv|
1, the angular velocity of rotation of the satellite is much greater than
the angular velocity of the radius vector of its center of inertia.
The third case (fast rotation of the satellite), unlike the first two, is not a
resonance case.
1. In the case of a small e, we rewrite (6.49) in the form:


d2 δ dδ  
þ 3a sin δ ¼ e 4 sin v þ 2 sin v þ 3a cos v sin δ þ O e2 :
2 2
ð6:50Þ
dv2 dv

The consideration of this paragraph follows the scheme from [7].

For e ¼ 0, the general solution of equation (6.50) describes the oscillatory,
rotational, or aperiodic motions and depends on two arbitrary constants c, v0, one
of which, v0, is additive to the independent variable v. Excluding from the consid-
eration aperiodic motions and denoting the angular frequency of oscillations or
rotations by ω(c), we write the known general solution of equation (6.50) for e ¼ 0
in the form:

δ ¼ Qðφ; cÞ, ð6:51Þ

φ ¼ ωðcÞðv þ v0 Þ: ð6:52Þ

For any c function Q has the property:

Qðφ þ 2π; cÞ ¼ Qðφ; cÞ þ γ, ð6:53Þ

126 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

where in the case of oscillations γ ¼ 0 and in the case of rotations γ ¼ 2π. Partial
derivatives of Q with respect to φ, c are periodic in phase φ with the period of 2π.
Substituting (6.51) and (6.52) into (6.50), we find that function Q(φ, c) satisfies the
equation:

d2 Q
ω2 ðcÞ þ 3a2 sin Q ¼ 0 ð6:54Þ
dφ2

identically for φ, c.
We also introduce the integral of action for equation (6.50):
2ðπ
ωð c Þ
J ðc Þ ¼ Q2φ ðφ; cÞdφ: ð6:55Þ
2π
0

Here, the subscript denotes partial derivative.

One can prove [7] that, in this case, the derivative of the integral of action J(c)
equals:
 
0 ∂ Q; ωQφ
J ðcÞ ¼ , ð6:56Þ
∂ðφ; cÞ

whereas the last expression does not depend on φ.

We look for the solution of equation (6.50) for e 6¼ 0 in the form (6.51), assuming
additionally that:

dδ
¼ ωQφ ðφ; cÞ: ð6:57Þ
dv

After the substitution of (6.51) into (6.50), (6.57), we obtain a system of ordinary
differential equations for the new unknown functions φ, c. Solving this system for
the derivatives of the unknown functions and using identities (6.54), (6.56), we
obtain:

dc e
¼ 0 Fðθ; φ; cÞQφ ðφ; cÞ,
dv J ðcÞ
dφ e ð6:58Þ
¼ ωðcÞ  0 Fðθ; φ; cÞQc ðφ; cÞ:
dv J ðcÞ

Here, we denote by F the coefficient at e in the right-hand side of equation

(6.50), in which δ and dδ/dv are replaced by their expressions according to (6.51),
(6.57). The terms O(e2) in equation (6.50) are neglected.
The right-hand sides of equations (6.58) are periodic in φ and v with the period
2π; c is a slowly varying quantity.
6.3 Resonance Phenomena in the Planar Motion of a Satellite About Its Center of Mass 127

Consider first the nonresonance case: the frequency ω(c) is not close to 1 (or to
another rational number r/s, where r, s are small relatively prime numbers). In this
case, to obtain the solution in the first approximation, one has to average the right-
hand sides of system (6.58) over v and φ as independent variables. However, it is
easy to see that the averaging of F over v yields zero (see equation (6.50)).
Therefore, in the nonresonance case, the perturbation torque due to ellipticity of
the orbit does not change amplitude c and frequency ω(c).
Of greatest interest is the resonance case: ω(c) is close to r/s (r, s are relatively
prime natural numbers), or more precisely:
r
ωðcÞ  ¼ OðeÞ: ð6:59Þ
s

We introduce a new variable α (phase shift) by the relation:

r
φ ¼ ðv þ αÞ: ð6:60Þ
s

It follows from (6.58), (6.60) subject to condition (6.59) that α is a slowly

changing variable: dα/dv ¼ O(e). We solve equations (6.58) with respect to deriv-
atives of c and α using (6.60) and average their right-hand sides, assuming c and α
constant, over the period of fast motion which is equal to 2πs for the variable v
(or 2πr for the variable φ).
Denote by c0 the value of c for which the exact resonance takes place:
r
ωð c 0 Þ ¼ : ð6:61Þ
s

Then condition (6.59) is equivalent to the condition c  c0 ¼ O(e) (in the essen-
tially nonlinear system ω0 6¼ 0 everywhere except, perhaps, some individual points).
Therefore, without loss of the order of accuracy with respect to e, the averaged
system can be written as follows (the integration interval can be changed by another
of the length 2πr):

ð
2πr
sφ 
dc e
¼ F  α; φ; c0 Qφ ðφ; c0 Þdφ,
dv 2πrJ 0 ðc0 Þ r
0
ð6:62Þ
ð
2πr
sφ 
dα s 0 cs
¼ ω ðc0 Þðc  c0 Þ  F  α; φ; c0 Qc ðφ; c0 Þdφ:
dv r 2πr J 0 ðc0 Þ
2 r
0

As usual, the solution of the averaged system differs from the solution of the
exact equation by a quantity of the order of e on the interval of argument changing
of the order of 1/e, if the condition c  c0 ¼ O(e) is satisfied on this interval. This
condition can be checked on the basis of the solution of the averaged system.
Proceeding to finding explicitly the functions included in (6.62), we write the
general solution of equation (6.50) for e ¼ 0 (see [7, 8]):
128 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

 pffiffiffi 
δ1 ¼ 2arcsin k1 sn 3aðv þ v0 Þ; k1 , 0  k1  1,

pffiffiffi 
3a ð6:63Þ
δ2 ¼ 2 am ðv þ v0 Þ; k2 , 0  k2  1:
k2

Index 1 refers to the oscillatory motions (|δ1|  π), index 2, to the rotational
motions (when δ2 monotonically increases), while the role of constant c is played by
the elliptic function moduli k1, k2. For k1 ¼ k2 ¼ 1, solutions (6.63) coincide and
describe an aperiodic motion. The frequencies of oscillations and rotations are
respectively equal to (see [7, 8]):
pffiffiffi pffiffiffi
π 3a π 3a
ω1 ð k 1 Þ ¼ , ω2 ð k 2 Þ ¼ , ð6:64Þ
2K ðk1 Þ k 2 K ðk 2 Þ

where K(k) is the complete elliptic integral of the first kind.

Let us write functions Q from equality (6.51):


2K ðk1 Þ
Q1 ðφ; k1 Þ ¼ 2arcsin k1 sn φ; k1 ,

π ð6:65Þ
K ðk 2 Þ
Q2 ðφ; k2 Þ ¼ 2 am φ; k2 :
π

We calculate their derivatives with respect to phase φ according to the rules of

differentiation of elliptic functions:


4k1 K ðk1 Þ 2K ðk1 Þ
Q1φ ðφ; k1 Þ ¼ cn φ; k1 ,
π
π  ð6:66Þ
2K ðk2 Þ K ðk 2 Þ
Q2φ ðφ; k2 Þ ¼ dn φ; k2 :
π π
0
Let us find functions ω0 , J , included in system (6.62). To this end, we use
relations (6.56), (6.64)–(6.66), and the formulas for the derivatives of complete
elliptic integrals with respect to modulus (see [8]). After some transformations, we
obtain, using the properties of elliptic functions:
pffiffiffi   
πa 3 E1  1  k21 K 1
ω01 ðk1 Þ ¼   ,
2k1 1  k21 K 21
pffiffiffi
πa 3E2
ω02 ðk2 Þ ¼  2   ,
ð6:67Þ
k 1  k22 K 22
pffiffiffi 2 pffiffiffi
0 8a 3 0 4a 3K 2
J 1 ðk 1 Þ ¼ k 1 K 1 , J 2 ðk 2 Þ ¼  ,
π πk22
K i ¼ K ðki Þ, Ei ¼ Eðki Þ, i ¼ 1, 2,
6.3 Resonance Phenomena in the Planar Motion of a Satellite About Its Center of Mass 129

where E(k) is the complete elliptic integral of the second kind.

The integrals in (6.62), after simple transformations and integration by parts,
taking into account evenness and periodicity of the integrands, can be reduced to
the form:
πðr
sφ
cos ΦðφÞdφ,
r
πr

where Φ(φ) is a certain even function of φ. Expanding it into a Fourier cosine series
and assuming φ ¼ rβ, we get under the integral sign a sum of expressions of the
form an cos sβ cos nrβ (n is integer). Therefore, the whole integral is different from
zero, unless s ¼ nr for some integer n.
Since s, r are relatively prime natural numbers, it is possible only if r ¼ 1. So, if
r 6¼ 1, then the resonance terms in equations (6.62) are absent (due to the fact that
perturbation force F is a purely sinusoidal function of v). Below we consider only
the case r ¼ 1.
So, after these transformations, using the first integral of equation (6.54), that is,
the expression ω2 Q2φ  6a2 cos Q ¼ const, system (6.62) can be written as:

dc e sin α
¼ Ms ðc0 Þ,
dv J 0 ðc0 Þ
dα e cos α ð6:68Þ
¼ sω0 ðc0 Þðc  c0 Þ  N s ðc0 Þ,
dv J 0 ðc0 Þ

where we introduce the notation

ðπ

1 3  
M s ðcÞ ¼  cos sφ 8Qφ ðφ; cÞ þ Q2φ φ; c dφ,
2π s
0
ðπ ð6:69Þ
∂Ms ω0 ðcÞ
N s ðcÞ ¼ þ cos sφQ2φ ðφ; cÞdφ:
∂c π
0

The integrals in formulas (6.69) can be calculated using the known expansions of
elliptic functions in Fourier series [8]. Let us present the result in the case of the
main resonance s ¼ 1; the second subscript i ¼ 1 , 2 in Msi, Nsi refers, respectively,
to oscillations and rotations:
130 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

pffiffiffiffiffi pffiffiffiffiffi
16 q1 4π 2 q1 ð1  q1 Þ
M11 ðk1 Þ ¼  , N 11 ðk1 Þ ¼   ,
1 þ q1 k1 1  k21 K 21 ð1 þ q1 Þ2
14q2 X1
q2nþ1
M12 ðk2 Þ ¼   12  2 ,
1 þ q22 1 þ q 2n
1 þ q2nþ2
(  n¼1  2 2
4π 2 q2 3 1  q22
N 12 ¼   2  2 þ
k2 1  k2 K 2
2
1 þ q22 ð6:70Þ
    )
X 1 2n
q2 3q2n 2 1  q2 þ ð2n þ 1Þ 1  q2
2 4nþ2
þ  2  2 ,
n¼1 1 þ
q2n 1 þ q2nþ2
2 2
qffiffiffiffiffiffiffiffiffiffiffiffiffi 32

πK 1  k21
6 7
qi ¼ qðki Þ ¼ exp6 4
7, i ¼ 1, 2:
5
K ðk i Þ

Numerical calculations can be performed using the available tables of the

complete elliptic integrals K, E and functions q [9].
Now we turn to the analysis of averaged system (6.68). Stationary resonant
modes (α∗, c∗) correspond to singular points of the system and are determined by
equating its right-hand sides to zero:

eð1Þm N s ðc0 Þ
α∗ ¼ mπ, c∗  c0 ¼ , m ¼ 0,  1,  2, . . . ð6:71Þ
sω0 ðc0 ÞJ 0 ðc0 Þ

For fixed natural s and a 2 [0, 1], we find c0 from (6.61), (we set r ¼ 1); the graph
of dependence c0(a) will be called the backbone curve. After that, we find from
(6.71) two different stationary values of amplitude c∗; the curve c∗(a) is called a
resonance curve. It is seen from (6.71) that the resonance curve is split into two
branches, which lie on either side of the backbone curve and approach it as e ! 0.
Above we considered as c the moduli of elliptic functions ki; now it is more
convenient to introduce a variable Ω, setting Ω ¼ k1 in the domain of oscillations,
and Ω ¼ k12 in the domain of rotations. Then, the domain Ω < 1 in the plane (a, Ω)
corresponds to oscillations, the region Ω > 1, to rotations.
The backbone curve consists of two branches with asymptotes a ¼ 0 and Ω ¼ 1;
 pffiffiffi
for a ¼ 1= s 3 , the branch corresponding to oscillations intersects the axis a at a
right angle. The qualitative course of the backbone and resonance curves in the case
of the main resonance (s ¼ 1) is shown in Fig. 6.3. For s 6¼ 1, the backbone curves
are obtained from the curve in Fig. 6.3 by contraction by s times along the a axis;
the resonance curves in the case s 6¼ 1 are tangent to the backbone one for Ω ! 0,
while in other respects their form remains the same.
The behavior of the resonance curves for the oscillations at the main resonance
pffiffiffi
near the point a ¼ 1= 3, Ω ¼ 0 is determined by the formula:
6.3 Resonance Phenomena in the Planar Motion of a Satellite About Its Center of Mass 131

Fig. 6.3 Backbone and

resonance curves

1
3a2  1 ¼ Ω2  2eΩ1 ,
2

and here Ω has the meaning of the oscillations amplitude of angle θ ¼ δ/2. The
resonance modes corresponding to the upper sign of this formula (the right branch
of the curve in Fig. 6.3) exist only if:
3
3a2  1 ð2eÞ3=2 :
2

To study non-stationary resonance modes, one has to integrate the averaged

system (6.68) for various initial data. It is easy to find the first integral of this
system; after that, obtaining its general solution is reduced to a quadrature. Let us
divide the second equation of (6.68) by the first; we obtain a linear equation with
respect to cosα as a function of c (c0 is considered as a parameter). The first integral
has the form:


sω0 ðc0 ÞJ 0 ðc0 ÞMs ðc0 Þ N s ðc0 Þ
cos α ¼ 1þ ðc  c0 Þ þ
eN 2s ðc0 Þ M s ðc 0 Þ


N s ðc 0 Þ
þconst  exp ðc  c0 Þ : ð6:72Þ
Ms ðc0 Þ

The nature of singular points (α∗, c∗) from (6.71) can be established
using (6.72). It is easier to linearize system (6.68) in the neighborhood of singular
point (α∗, c∗):
132 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

dðc  c∗ Þ eMs ðc0 Þ

¼ 0 ð1Þm ðα  α∗ Þ,
dv J ðc0 Þ ð6:73Þ
d ðα  α ∗ Þ
¼ sω0 ðc0 Þðc  c∗ Þ:
dv

Let us form the characteristic equation of system (6.73):

seMs ðc0 Þð1Þm ω0 ðc0 Þ
λ2 ¼ : ð6:74Þ
J 0 ðc0 Þ

It can be immediately seen that, in the case of oscillations, singular points with
even m (Ω < Ω0, the lower part of the resonance curve) are saddles (unstable mode),
whereas the points with odd m (Ω > Ω0, the upper part of the resonance curve) are
centers (stable mode). Here Ω0(a) corresponds to the backbone curve. In the case of
rotations, the unstable modes (saddle) corresponds to the upper part of the reso-
nance curve (m is odd, Ω > Ω0), whereas the stable ones (center), to the lower part
(m is even, Ω < Ω0). Besides, it follows from (6.74) that the frequency of oscilla-
tions about the center and the rate of exponential growth of the solution near the
pffiffiffi pffiffiffi
saddle are of the order of e ðjλj  eÞ. The phase plane α, c is periodic with the
period of 2π.
Thus, in the case of small e, stationary resonant modes of oscillations and
rotations of the satellite are found, and their stability is investigated.
2. Let us turn to the case a  1 (the satellite is close to a body with dynamic
symmetry); the eccentricity of the orbit is arbitrary 0  e < 1.
It is convenient to perform a change of independent variable and unknown
function in equation (6.49). As an independent variable, we take the time τ,
measured from perigee and calculated as a ratio to the satellite’s revolution period
divided by 2π:

rffiffiffiffiffiffiffiffiffiffiffi  pffiffiffiffiffiffiffiffiffiffiffiffiffi
1e v e 1  e2 sin v
τ ¼ 2Arctan tan  , ð6:75Þ
1þe 2 1 þ e cos v
τðv þ 2π Þ ¼ τðvÞ þ 2π:

As a new unknown function, we take angle x between the central axis of inertia
(the moment of inertia with respect to which equals A3) and the radius vector of the
perigee (Fig. 6.2):

δ
x¼θþv¼vþ : ð6:76Þ
2

After the change of variables, the equation of motion takes the form:
6.3 Resonance Phenomena in the Planar Motion of a Satellite About Its Center of Mass 133

d2 x 3a2 ð1 þ e cos vÞ3

þ sin 2ðx  vÞ ¼ 0, ð6:77Þ
dτ2 2 ð1  e2 Þ3

where v is considered as the function of τ defined by equality (6.75).

It is clear from equation (6.77) that a body with dynamic symmetry (a ¼ 0)
performs a uniform rotation about the center of symmetry (x ¼ C1τ + C2). For small
a, the motion of the satellite will be close to a uniform rotation.
We replace equation (6.77) by the equivalent system:

dx dy 3a2 ð1 þ e cos vÞ3

¼ y, ¼ sin 2ðx  vÞ ð6:78Þ
dτ dτ 2 ð1  e 2 Þ3

and seek its solution in the form

x ¼ Ωτ þ φ, y ¼ Ω þ az, ð6:79Þ

where Ω is a constant and φ, z are new unknown functions introduced instead of

x, y. Substitution of (6.79) into (6.78) leads to a system:

dφ dz 3a ð1 þ e cos vÞ3
¼ az, ¼ sin 2ðΩτ þ φ  vÞ: ð6:80Þ
dτ dτ 2 ð1  e 2 Þ3

Since the right-hand sides of equations (6.80) include the small parameter a,
thus, φ and z are slowly changing variables. Therefore, in the first approximation,
one can average the right-hand sides of equations (6.80) assuming φ and z constant.
The right-hand side of the second equation (6.80) has the form:
f ðτÞ ¼ f 1 ðτÞ sin 2Ωτ þ f 2 ðτÞ cos 2Ωτ, ð6:81Þ

where f1, f2 are periodic functions of τ of the period 2π. We expand them in Fourier
series. These series converge absolutely and uniformly due to continuous differen-
tiability of f1(τ), f2(τ) [10]. Substitute the obtained series in (6.81) and integrate f(τ)
with respect to τ from τ0 to T. If 2Ω is not an integer, then the series obtained as a
result of the integration will contain only the trigonometric functions of T and
therefore will be bounded for T ! 1 (its absolute convergence follows from the
absolute convergence of the Fourier series for f1(τ) and f2(τ)). Then we have for the
averaged value:
ðT
1
Mτ ff g ¼ lim f ðτÞdτ ¼ 0;
T!1 T
τ0
134 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

thus, in the first approximation, it follows that z ¼ const, x ¼ C1τ + C2. In this case,
gravitational perturbation torque has no effect on the uniform movement of the
satellite in the first approximation.
Consider a more interesting case 2Ω ¼ m, where m is an integer (for m ¼ 2 we
have the principal resonance: the satellite’s rotation period is close to the period of
its revolution along the orbit). In addition, the right-hand side of the second
equation (6.80) is periodic with respect to τ with the period 2π, and the averaging
should be carried out over this period. After the averaging, system (6.80) assumes
the form:

dφ dz 3a
¼ az, ¼  Φm ðeÞ sin 2φ, ð6:82Þ
dτ dτ 2

where we introduce the notation:

ðπ
1 ð1 þ e cos vÞ3
Φ m ðeÞ ¼ cos ðmτ  2vÞdτ: ð6:83Þ
2π ð1  e2 Þ3
π

From system (6.82), we obtain the equation for φ:

d2 ð2φÞ
þ 3a2 Φm ðeÞ sin 2φ ¼ 0, ð6:84Þ
dτ2

which coincides with the equation of the pendulum motion. Its solution is well
known and is expressed in terms of elliptic functions (see solution (6.63) of
equation (6.50) for e ¼ 0).
The asymptotic solution of equation (6.77) is determined by the first of the
formulas (6.79), into which one should substitute Ω ¼ m/2 and φ from the solution
of equation (6.84). It describes the rotation with angular velocity Ω, on which slow
rotations or oscillations are superimposed. This asymptotic solution is valid up to
quantities of the order of a on the time interval of the order of 1/a (much larger than
the orbital period of the satellite).
The equilibrium position for equation (6.84) are the points φ ¼ nπ 2,
n ¼ 0 ,  1 ,  2 , . . .. As seen from the relationship between x and φ (6.79), they
correspond to the rotation with a constant angular velocity. Note that at the time of
passage through the perigee (τ ¼ 2kπ), the value of φ is equal to the angle between
the principal axis of inertia with the moment A3 and the radius vector of the perigee.
Thus, the rotation of the satellite with a constant angular velocity Ω ¼ m/2 is only
possible provided that at the perigee one of the principal central axes of inertia,
lying in the orbital plane, is directed along the radius vector.
The stability of equilibrium positions depends on the sign of Φm(e). If Φm(e) > 0,
then the positions φ ¼ nπ (n ¼ 0 ,  1 ,  2 , . . .) are stable, while all other equilib-
rium positions are unstable. Conversely, when Φm(e) < 0, then the positions φ ¼ nπ
are unstable, while the other ones are stable. Therefore, in the case of Φm(e) > 0, at
6.3 Resonance Phenomena in the Planar Motion of a Satellite About Its Center of Mass 135

the stable stationary rotation, the axis of the smallest moment of inertia A3, which
lies in the orbit plane, is directed at the perigee along the radius vector, while in the
case of Φm(e) < 0, tangentially to the orbit (for v ¼ 0 we have θ ¼ 0 or θ ¼ π/2,
respectively; see Fig. 6.2). The frequency of small oscillations relative to the stable
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

mode of rotation equals a 3 Φm ðeÞ .
Let us turn to the study of function Φm(e), the sign of which determines the
stability of the rotation modes.
Taking v as the integration variable in (6.83), we obtain:
ðπ
1
Φm ðeÞ ¼ ð1 þ e cos vÞ cos ½mτðvÞ  2vdv, ð6:85Þ
π ð1  e2 Þ3=2
0

where τ(v) is defined by formula (6.75). Expanding the integrand in powers of e (for
small e), we find:

1   5  
Φ1 ðeÞ ¼  e þ O e2 , Φ2 ðeÞ ¼ 1  e2 þ O e2 ,
2 2
7   17   ð6:86Þ
Φ3 ðeÞ ¼ e þ O e2 , Φ4 ðeÞ ¼ e2 þ O e2 ,
2 2
Φm ðeÞ ¼ Oðe2 Þ, m ¼6 1, 2, 3, 4:

Besides, Φ0(e) 0. It follows from formulas (6.86) that for the orbits close to
circular, the most significant is the principal resonance (m ¼ 2); in other cases, the
rotation of the satellite will be almost uniform since Φm(e) is small. In the case of a
circular orbit and m ¼ 2 (e ¼ 0, v ¼ τ, φ ¼ θ ¼ δ/2), the averaged equation (6.84)
coincides with the exact equation of the relative motion (e.g., with (6.49) for e ¼ 0).
The function Φ2(e) was calculated numerically for various e 2 [0, 1]. It turned
out that it monotonically decreases and passes through zero at e0  0.682 (see
Fig. 6.4). Thus, the steady mode of the satellite rotation with the rotation period
equal to the period of its revolution along the orbit is stable, if at the perigee the axis
of the smallest moment of inertia is directed:
For e < e0, along the radius vector
For e > e0, tangentially to the orbit
Note that in the case of a circular orbit or an orbit close to circular, this result
coincides with the already known one [6].
In [11, 12], the results are presented of studying the periodic and generalized
periodic solutions of equation (6.49), which describes the planar oscillations and
rotations of the satellite about the center of mass moving along an elliptical orbit.
3. Consider the case of fast rotation of the satellite in the orbital plane: the
rotation period is much less than the satellite’s revolution period along the orbit;
e and a are arbitrary. We start from equation (6.77), which for brevity can be written
as:
136 6 Motion of a Satellite About Its Center of Mass Under the Action of. . .

Fig. 6.4 Graph of function Ф2(e)

d2 x
þ gðx; τÞ ¼ 0: ð6:87Þ
dτ2

Function g(x, τ) is periodic in x with the period π and has zero average value over
the period for any τ.
Introduce the notation:

ðπ ðξ
1
G1 ðτÞ ¼ gðη; τÞdη dξ,
π
00
ð6:88Þ
ðλ ðξ
G2 ðλ; τÞ ¼ gðη; τÞdη dξ:
00

Then the asymptotic representation for the general solution of equation (6.87)
for fast rotations can be represented in the form [13]:


1 1
x ¼ Ωðτ þ τ0 Þ þ 2 fΩðτ þ τ0 ÞG1 ðτÞ  G2 ½Ωðτ þ τ0 Þ; τg þ O 3 : ð6:89Þ
Ω Ω

Here Ω and τ0 are arbitrary constants (Ω

1), and expansion (6.89) is valid on
the interval Δτ  1 (on the time interval of the order of the satellite’s revolution
period along the orbit).
Calculating integrals G1 and G2 from (6.88) (function g(x, τ) is determined by
equation (6.77)), we finally obtain:
References 137

3a2 ð1 þ e cos vÞ3

x ¼ Ωðτ þ τ0 Þ þ fsin 2½Ωðτ þ τ0 Þ  v  sin 2vg
8Ω2 ð1  e2 Þ3


1
þO 3 : ð6:90Þ
Ω

The connection between v and τ is still given by relation (6.75).

The asymptotic solution (6.90) consists of three summands. The first describes
the fast rotation with the frequency Ω
1; the third (the second term in the braces)
describes the slow oscillations with the frequency 2; and the second (the first term in
the braces), the short-periodic oscillations with the frequency 2Ω, the amplitude and
phase of which is slowly changing with the frequencies 1 and 2, respectively. The
amplitude of the fast periodic oscillations takes its maximum value at the perigee
(v ¼ 0), and the minimum one, in the apogee of the orbit, while the ratio of the
maximum to minimum amplitude equals (1 + e)3(1  e)3.

References

1. Chernousko, F.L.: On the motion of a satellite about its center of mass under the action of
gravitational moments. J. Appl. Math. Mech. 27(3), 708–722 (1963)
2. Chernousko, F.L.: Resonance phenomena in the motion of a satellite relative to its mass centre.
USSR Comput. Math. Math. Phys. 3(3), 699–713 (1963)
3. Volosov, V.M., Morgunov, B.I.: The Averaging Method in the Theory of Non-linear Oscilla-
tory Systems. Moscow State Univ., Moscow (1971) in Russian
4. Pontryagin, L.S.: Ordinary Differential Equations. Nauka, Moscow (1965) in Russian
5. Beletsky, V.V.: Motion of an Artificial Satellite about its Center of Mass. Israel Program for
Scientific Translation, Jerusalem (1966)
6. Beletsky, V.V.: Libration of a satellite. Artif. Earth Sat. 3, 13–32 (1959) in Russian
7. Chernousko, F.L.: On resonance in an essentially non-linear system. USSR Comput. Math.
Math. Phys. 3(1), 168–185 (1963)
8. Zhuravsky, A.M.: Handbook of Elliptical Functions. Academy of Sciences Press, Moscow
(1941) in Russian
9. Jahnke, E., Emde, F., Losch, F.: Tables of Higher Functions. McGraw-Hill, New York, NY
(1960)
10. Fikhtengol’ts, G.M.: Course of Differential and Integral Calculus, vol. 3. Fizmatgiz, Moscow
(1970) in Russian
11. Bruno, A.D.: Families of periodic solutions of the Beletskii equation. Kosm. Issl. 40(3),
295–316 (2002) in Russian
12. Sadov, S.Y.: On the stability of resonant rotation of a satellite about its center of mass in the
orbital plane. Kosm. Issl. 44(2), 170–181 (2006) in Russian
13. Moiseev, N.N.: The asymptote of rapid rotations. USSR Comput. Math. Math. Phys. 3(1),
186–203 (1963)
Chapter 7
Motion of a Rigid Body with a Cavity Filled
with a Viscous Fluid

In this chapter, we consider the motion about the center of mass of a rigid body with
a cavity filled with a viscous fluid. It is assumed that the viscosity of the fluid is
sufficiently high, so the corresponding Reynolds number is small. The torques
acting on the body by the viscous fluid in the cavity are determined by the method
developed in [1–3] and described in Sect. 5.3.
In Sect. 7.1, we present and discuss the general equations of motion of a rigid
body with a cavity filled with a viscous fluid at low Reynolds numbers. In Sect. 7.2,
we consider the planar motion of a pendulum containing a viscous fluid in the
cavity. In Sect. 7.3, we examine the free spatial movement of a body with a cavity
containing a viscous fluid. Section 7.4 is devoted to the motion of a rigid body
containing one or more dampers, that is, the bodies situated inside the supporting
body and interacting with it by means of the viscous friction forces. These forces
are generated by a lubricant layer which is in a narrow band between the carrier
body and dampers. It turns out that such mechanical systems are described, under
certain assumptions, by equations similar to the equations for a body with a viscous
fluid in the cavity. In Sect. 7.5, the stability of motion of a rigid body with dampers
is studied.

7.1 Equations of Motion of a Body with a Viscous Fluid

in a Cavity

The equations of motion of a body with a cavity filled with a viscous fluid at low
Reynolds numbers have the form (5.28). All notations in this system are described
in detail in Sect. 5.3. System (5.28) can be studied and solved by the usual analytical
and numerical methods of the theory of ordinary differential equations. It contains a
small parameter v
1, where v ¼ R1 is the dimensionless viscosity inverse to the
Reynolds number, v  1. To solve system (5.28), we can apply the method of

© Springer International Publishing AG 2017 139

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_7
140 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

averaging, and it suffices to consider just the first approximation which ensures the
accuracy of the order of v
1 on the time interval of the order of v.
Note one specific feature of system (5.28). The perturbation torque m due to the
influence of the viscous fluid also depends on the moment L.
Let us calculate, based on equations (5.28), the power of the torque m:

m  ω ¼ ρv
1 ½ðP  bÞ  ω:

Relations (5.26)–(5.28) yield the equalities



ε ¼ a þ O v
1 , ε_ ¼ b þ O v
1 ,

by means of which the formula for m  ω can be transformed, up to the terms of the
order v
2, into the form
 
d
m  ω ¼ ρv
1 ½ðP  bÞ  ω
ðP  εÞ  ε ¼
 dt  ð7:1Þ

1 d
¼ ρv ½ ð P  aÞ  ω 
ð P  aÞ  a :
dt

Since the quadratic form (P  a)  a is positive definite, then the second term in
formula (7.1) is always negative and corresponds to the energy dissipation due to
viscosity. The first term may have any sign and is associated with the transfer of
kinetic energy from the body to the fluid and back. Let the unperturbed motion, i.e.,
the motion of the body with the hardened fluid, be periodic (the motion is oscilla-
tions or rotations). Then, vectors ω, L, a are periodic in the unperturbed motion, and
the average value over the period of the first term in (5.29) is zero.
Consider an important special case when the unperturbed system (the system
subject to the condition of hardening of the fluid, i.e., for v ¼ 1, m ¼ 0) is
conservative. Then, the total energy of the unperturbed system

1
H ¼ ðJ  ωÞ  ω þ ΠðhÞ, ð7:2Þ
2

where Π is the potential energy, is preserved. For m 6¼ 0, the equation dH dt ¼ m  ω

holds, while in the case of perturbed motion, function H from (7.2) is no longer the
total energy of the system (H is the energy under the condition of hardening of the
fluid). Suppose that, in the unperturbed motion, the system performs a periodic
motion (oscillations or rotations) and ω, L, a are periodic. In the perturbed motion,
H is a slowly changing variable, and, according to the averaging method, in the first
approximation we can average the rate of change of H over the period of the
unperturbed motion. Averaging of equation (7.2) yields
7.2 Planar Motion of a Pendulum with a Viscous Fluid 141

dH
¼ Mt fm  ωg ¼
ρv
1 Mt fðP  aÞ  ag < 0: ð7:3Þ
dt

Here, Mt is the operation of averaging over the period along the motions of the
unperturbed system, where, in the process of averaging, function H and other first
integrals of the unperturbed system are considered to be constant. Equation (7.3) is
obtained in the first approximation of the averaging method and has the accuracy of
the order of v
1 on the time interval of the order of v. Relation (7.3) demonstrates
the dissipative nature of the influence of the viscous fluid in the cavity on the
dynamics of the rigid body.
The resulting system (5.28) can be used to investigate various cases of motion of
bodies with fluid. We will consider below some specific problems.

7.2 Planar Motion of a Pendulum with a Viscous Fluid

Consider the planar motion about a fixed axis of a body with a cavity filled with a
viscous fluid. Let axis O1y3 be the axis of the fixed coordinate system O1y1y2y3,
about which the body rotates. The system Ox1x2x3, associated with the body, is
chosen so that axis Ox3 is parallel to the rotation axis O1y3. As the only variable y,
we take the angle φ of rotation of the body about the axis, measured from a fixed
direction. Denote by J the moment of inertia of the body with fluid with respect to
the rotation axis O1y3 and by P ¼ P33 > 0 the component of tensor P corresponding
to axis Ox3. In this case, vector equations (5.28) are reduced to a scalar equation of
the second order:
 
ρP ∂L L ∂L
J€φ
Lðφ; φ_ Þ ¼ þ φ_ : ð7:4Þ
vJ ∂φ_ J ∂φ

Here, Lðφ; φ_ Þ is a given moment of external forces with respect to axis O1y3; the
dot denotes the time derivative. If the unperturbed system is conservative (L does not
depend on φ_ ), then equation (7.4), equality (7.2), as well as the averaged equation
(7.3) are converted to the form
ρP dL
φ
LðφÞ ¼
J€ _
φ,
vJ dφ
ðφ
1
H ¼ J φ_ 2 þ ΠðφÞ, ΠðφÞ ¼
LðφÞdφ, ð7:5Þ
2
φ0
ρP 
H_ ¼
2 Mt L2 ðφÞ < 0:
vJ
142 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

Here, Π(φ) is the potential energy. Suppose that the system is in a uniform
gravitational field, i.e., it is a physical pendulum with a viscous fluid, where φ is the
angle of pendulum’s deviation from the lower equilibrium position. Then, equali-
ties (7.5) yield
1
L ¼
G0 l0 sin φ, H ¼ J φ_ 2
G0 l0 ð1 þ cos φÞ,
2 ð7:6Þ
G 0 l0 ρPG0 l0
€þ
φ sin φ ¼
cos φφ_ :
J vJ 2

Here, G0 is the weight of the system; l0 is the distance from the suspension axis
to the system’s center of inertia. An arbitrary constant of energy is chosen so that
Π ¼ 0 at the upper position of equilibrium. Then, for 0 > H >
2G0l0, the
unperturbed system (i.e., the system with v ¼ 1) performs oscillations and, for
H > 0, rotations; the equality H ¼
2G0l0 corresponds to the lower position of
equilibrium and the equality H ¼ 0 to aperiodic motion of the pendulum.
The solution of the unperturbed (v ¼ 1) equation (7.6) (the equation of motion
of the pendulum) is expressed in terms of elliptic functions [4]
rffiffiffiffiffiffiffiffiffi
G 0 l0
φ1 ¼ 2arcsinfk1 sn½cðt þ t0 Þ; k1 g, c¼ ,
  J ð7:7Þ
cðt þ t0 Þ
φ2 ¼ 2am ; k2 , 0  ki  1:
k2

Here, t0 and ki are arbitrary constants; the subscript 1 always corresponds to

oscillations and the subscript 2 to rotations, whereas for k1 ¼ k2 ¼ 1 both solutions
describe aperiodic motion. Let us quote the known formulas [4] which connect the
moduli ki of elliptic functions with energy (7.6), the periods of oscillations and
rotations, as well as the oscillation amplitude of angle φ0 and the maximum angular
velocity ω0 during rotation:
H1 4K ðk1 Þ
1þ ¼ k21 , T 1 ¼ , φ0 ¼ 2arcsink1 ,
2G0 l0 c
H2 1 2k2 K ðk2 Þ 2c ð7:8Þ
1þ ¼ , T2 ¼ , ω0 ¼ :
2G0 l0 k22 c k2

Here, K(k) is the complete elliptic integral of the first kind.

Substitute solution (7.7) into expression (7.6) for L and perform averaging
according to (7.5) over the period of the unperturbed motion:

ði
T

1
Mt L ðφi ðtÞÞ ¼
2
L2 ðφi ðtÞÞdt, i ¼ 1, 2:
Ti
0
7.2 Planar Motion of a Pendulum with a Viscous Fluid 143

In the calculation of integrals, we use the well-known formulas [5] for the
integrals of elliptic functions, where we consider ki, which are uniquely associated
with energy Hi, to be constant in the averaging. Then, in the obtained averaged
equations, we express H in terms of ki according to (7.8) and finally obtain the
following equations for ki:

dk21

 Eð k 1 Þ
¼
1
k21 þ 1
2k21 ,
dξ K ðk1 Þ
dk22

 Eð k 2 Þ
¼
2 1
k22 þ 2
k22 , ð7:9Þ
dξ K ðk2 Þ
t
t∗ 3vJ 2
ξ¼ , T0 ¼ > 0:
T0 2ρPG0 l0

The arbitrary constant t∗ is chosen in such a way that the time moment t ¼ t∗,
ξ ¼ 0 corresponds to the transition from rotations to oscillations, i.e., when t < t∗,
ξ < 0, rotations occur in the system and, when t > t∗, ξ > 0, oscillations. Then, as
the initial conditions for system (7.9), we should take k21 ð0Þ ¼ k22 ð0Þ ¼ 1, because
the values k1 ¼ k2 ¼ 1 correspond to the boundary between oscillations and rota-
tions. Clearly, equations (7.9) can be integrated in quadratures. These equations
have been integrated numerically under the initial conditions k21 ð0Þ ¼ k22 ð0Þ ¼ 1,
corresponding to the transition from rotations to oscillations. Function k21 ðξÞ was
defined only for ξ  0 and k22 ðξÞ for ξ  0. When |ξ| increases from 0 to 1, both
functions decrease monotonically from 1 to 0, while k01 ð0Þ ¼ k02 ð0Þ ¼ 0 which
follows directly from equations (7.9). Figure 7.1 shows the graphs of the functions:
0
k21 ðξÞ is represented by curve 1, k22 ð
ξÞ by curve 2, and k22 ð
10ξÞ by curve 2 .
Using the decomposition of complete elliptic integrals into series with respect to
k2 for small k (see [5]), we can simplify the right-hand sides of equations (7.9), after
which they can be easily integrated. Here are the obtained asymptotic solutions
corresponding to small k2 or large |ξ|:

8
k21 ¼ C1 expð
3ξ=2Þ for ξ ! þ1, k22 ¼
for ξ !
1: ð7:10Þ
3ξ

The numerical solution (see Fig. 7.1) has confirmed the asymptotic formulas
(7.10). In addition, from the comparison with the numerical solution, we have
determined the constant C1 ¼ 3.57.
The asymptotic solutions (7.10) describe damping of small oscillations and fast
rotations. Substituting equations (7.10) into relations (7.8) and using notations (7.9)
for ξ, T0, we obtain the laws of attenuation of energy and amplitude of small
oscillations for t ! + 1, as well as the laws of attenuation of energy and the
maximum angular velocity in fast rotations for t !
1:
144 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

Fig. 7.1 Graphs of the

squared modulus of
elliptic functions for a
pendulum

 
ðt
t∗ ÞρPG0 l0
H1 ¼
2G0 l0 þ 2G0 l0 C1 exp
2
,
  vJ
pffiffiffiffiffiffi ðt
t∗ ÞρPG0 l0
φ0 ¼ 2 C1 exp
, t ! þ1,
2vJ 2
3G0 l0 ξ ðt
t∗ ÞρPG20 l20 ð7:11Þ
H2 ¼
¼ ,
4 s 2vJ 2
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3G0 l0 ξ ðt∗
tÞρPG20 l20
ω0 ¼
¼ , t !
1:
2J vJ 3

The law (7.11) of attenuation of amplitude φ0 of small oscillations can also be

obtained by directly solving the linearized equation (7.6).
The obtained functions k21 ðξÞ, k22 ðξÞ, shown in Fig. 7.1, are universal. Using them
and performing simple conversion by formulas (7.9) for ξ, T0, and (7.8), one can
find the law of time variation of the amplitude and energy of motion in each specific
case. This conversion is carried out as follows. Suppose that at the time t ¼ t1 it is
given that the pendulum performs rotations (or oscillations) with a given energy or
amplitude. Then, according to formulas (7.8), we find the initial value of k2 (or k1).
Using the graphs of k22 ðξÞ (or k21 ðξÞ ), we determine the initial value of ξ
corresponding to the given k2 (or k1). Knowing ξ and the constant T0, we can find
t∗ by formula (7.9). In the case of rotations, we have t∗ > t1, while in the case of
oscillations, we have t∗ < t1. The further development of the process, including the
transition from rotations to oscillations, is described by functions k21 ðξÞ, k22 ðξÞ and
formulas (7.8). All the constants in these formulas, including t∗, are already found.
At the moment of time t ¼ t∗, ξ ¼ 0, it is necessary to switch from the formulas with
the subscript i ¼ 2 to the formulas with the subscript i ¼ 1. Note that, according to
the general theorems of the averaging method [6, 7], the error of solutions for
energy or amplitude is of the order of v
1 on the time interval of the order of v. The
asymptotic solution for angle φ is determined in the first approximation by formulas
(7.7), in which the slowly changing variables ki are to be substituted. The obtained
relations describe the entire course of change of the motion from rotations to small
oscillations, i.e., the decrease of energy H from 1 to
2G0l0. The characteristic
damping time of the motion is equal to T0.
7.3 Free Three-Dimensional Motion of a Body with a Viscous Fluid 145

7.3 Free Three-Dimensional Motion of a Body

with a Viscous Fluid

Consider a free (L ¼ 0) three-dimensional motion of a rigid body with a fluid rela-

tive to the center of inertia. Let O1y1y2y3 be a translationally moving coordinate
system whose origin is connected with the center of inertia of the system. For
simplicity, tensor P is given in the form Pij ¼ P0δij, where δij is the Kronecker
symbol, and P0 > 0. Tensor P has this form, for example, in the case of a spherical
cavity, for which we have according to (5.30) P0 ¼ 8πa7/525, where a is the radius
of the cavity. Denote by A1, A2, A3 the principal central moments of inertia of the
system and by p, q, r the projections of the absolute angular velocity ω on the
corresponding principal central axes of inertia. Equations (5.28) for L ¼ 0 are
written in the projections on the principal central axes of inertia (the dot denotes
the derivative with respect to time):
ρP0
A1 p_ þ ðA3
A2 Þqr ¼ p A3 ðA1
A3 ÞðA1 þ A3
A2 Þr 2 þ
vA1 A2 A3

þA2 ðA1
A2 ÞðA1 þ A2
A3 Þq2 : ð7:12Þ

The remaining equations are obtained from (7.12) by cyclic permutation of

letters A1, A2, A3 and p, q, r. The kinematic relations are not written, because
equations (7.12) form a closed system.
Multiplying equation (7.12) by A1p and the remaining equations of motion by
A2q and A3r and then adding them, we verify that these equations have the first
integral
 
G2 ¼ A21 p2 þ A22 q2 þ A23 r 2 ¼ const, G ¼ G, ð7:13Þ

which expresses the constancy of angular momentum G of the body with the
hardened fluid.
Calculating the time derivative of the kinetic energy

1

T¼ A1 p2 þ A2 q2 þ A3 r 2 , ð7:14Þ
2

due to the motion equations (7.12), we get

ρP0
T_ ¼
ðA1 þ A3
A2 ÞðA1
A3 Þ2 p2 r 2 þ
vA1 A2 A3
þðA1 þ A2
A3 ÞðA1
A2 Þ2 p2 q2 þ

þðA2 þ A3
A1 ÞðA2
A3 Þ2 q2 r 2  0, ð7:15Þ

i.e., the kinetic energy decreases, as it should be expected.

146 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

Let us express the projections of angular momentum G of the body with

hardened fluid on the principal central axes of inertia:

A1 p ¼ G sin θ sin φ,
A2 q ¼ G sin θ cos φ, ð7:16Þ
A3 r ¼ G cos θ:

Here, G ¼ |G|, θ is the nutation angle, and φ is the angle of proper rotation; these
angles determine the orientation of vector G relative to the rigid body.
Let us consider first the case of dynamic symmetry A1 ¼ A2 6¼ A3, and let us turn
in (7.12) to variables G, θ, φ according to formulas (7.16). We obtain the system

ρP0 ðA1
A3 ÞG2 sin θ cos θ A1
A3

G_ ¼ 0, θ_ ¼ , φ_ ¼ G cos θ:
vA31 A3 A1 A3

This system can be easily integrated, and it yields

 
ρP0 ðA1
A3 ÞG2 t
tan θ ¼ tan θ0 exp ,
vA31 A3
ð
t!1
ðA1
A3 ÞG dt ð7:17Þ
φ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi þ φ0 ,
A1 A 3 2 
ρP 0 ð A 1
A 3 ÞG t
t0 1 þ tan 2 θ0 exp
vA31 A3

where G, θ0, φ0 are arbitrary constants. Here, it is assumed without loss of general-
ity that θ lies in the first quarter; in the contrary case, we can change the directions
of axes to the opposite ones.
It can be seen from formulas (7.17) that for A1 > A3 angle θ increases and tends
to π2 as t ! 1, whereas angle φ also grows and for t ! 1 exponentially approaches
a certain constant depending on the initial conditions. The final motion will be a
rotation about an axis perpendicular to the axis of dynamic symmetry.
In the case of A1 < A3, angle θ decreases, and for t ! 1 we have θ ! 0,
φ ! (A1
A3)G/(A1A3). The final movement here will be a rotation about the axis
of dynamic symmetry. In terms of the order of magnitude, the characteristic time of


the transition process in both cases equals vA31 A3 = ρP0 G2 jA1
A3 j .
Suppose now that the moments of inertia are different. Without loss of general-
ity, we assume that A1 > A2 > A3. Performing in (7.12) the transition to variables G,
θ, φ according to (7.16), we obtain a system of the form

G_ ¼ 0, θ_ ¼ f 1 ðG; θ; φÞ, φ_ ¼ f 2 ðG; θ; φÞ: ð7:18Þ

The autonomous second-order system for θ, φ can be easily studied by the usual
methods of the qualitative theory of ordinary differential equations. We can find the
singular points of the system corresponding to the steady motions, investigate the
7.3 Free Three-Dimensional Motion of a Body with a Viscous Fluid 147

Fig. 7.2 Trajectories of the

angular momentum vector
for a body with viscous fluid

nature of the singular points (in particular, their stability), and construct a quali-
tative picture of the phase trajectories. The results are conveniently represented in
the form of the trace of the angular momentum vector G on the unit sphere,
connected with the rigid body, the center of which is placed at the center of inertia
O1 of the system.
In Fig. 2.1, the trajectories are shown of vector G in Euler’s case (a free rigid
body without fluid; see, e.g., [8]), whereas in Fig. 7.2 these trajectories are given for
the considered case of a body with fluid (equations (7.18) or (7.12)). The arrows in
these figures indicate the direction of time increasing, while the letters A1, A2, A3
designate the principal central axis with the respective moments of inertia.
Let us consider the motion of a body with a viscous fluid, temporarily
abandoning the condition A1 > A2 > A3. In this case, the stationary motions are
again rotations about the principal axes of inertia; this can be seen from equations
(7.12).
Consider the motion in the vicinity of the stationary rotation about axis A3,
assuming that r ¼ r0 + x and r0 > 0 is constant, while quantities p, q, x are small of
the first order. In the case of stationary rotation, we have p ¼ q ¼ x ¼ 0. Then,
equations (7.12), linearized in the neighborhood of solution p ¼ q ¼ x ¼ 0,
take the form
A1 p_ þ ðA3
A2 Þr 0 q ¼ αp, A2 q_ þ ðA1
A3 Þr 0 p ¼ βq, x_ ¼ 0,
ρP0
α ¼ ðA1
A3 ÞðA1 þ A3
A2 Þr 20 ,
vA1 A2 ð7:19Þ
ρP0
β ¼ ðA2
A3 ÞðA2 þ A3
A1 Þr 20 :
vA1 A2

We write the characteristic equation for the first pair of equations (7.19):

A1 A2 λ2
ðαA2 þ βA1 Þλ þ αβ þ ðA2
A3 ÞðA1
A3 Þr 20 ¼ 0:

In this equation, α and β are quantities of the order of v
1(see (7.19)). Solving the
characteristic equation taking into account the values of α, β from (7.19), we obtain,
up to the terms of the order of v
2,
148 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

ρP0 r 20 ½A2 ðA1
A3 ÞðA1 þ A3
A2 Þ þ A1 ðA2
A3 ÞðA2 þ A3
A1 Þ

λ1, 2 ¼ 2 2

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2vA1 A2 ð7:20Þ
ðA1
A3 ÞðA2
A3 Þ
ir 0 , λ3 ¼ 0:
A1 A2

The root λ3 ¼ 0 corresponds to the third equation of (7.19). It is seen from (7.20)
that for A1 > A3, A2 > A3 we have Reλ1 , 2 > 0, Imλ1 , 2 6¼ 0, which corresponds to
unstable singular point of the focus type.
For A1 > A3, A2 < A3 or A1 < A3, A2 > A3, the roots λ1 and λ2 are real and of
opposite signs, which corresponds to an unstable singular point of the saddle type.
For A1 < A3, A2 < A3, we obtain from (7.20) that Reλ1 , 2 < 0, Imλ1 , 2 6¼ 0, which
corresponds to a stable focus for variables p, q. However, due to the presence of
zero root λ3 ¼ 0, this does not imply stability with respect to variables p, q, r.
To prove stability, we form the Lyapunov function

2
V ¼ 2A3 T
G2 þ G2
A23 r 20 : ð7:21Þ

Here, T and G2 are defined by equalities (7.14) and (7.13). It is easy to check with
the help of (7.14), (7.13) that function V vanishes in the case of steady motion, i.e.,
for p ¼ q ¼ r
r0 ¼ 0. Assuming r ¼ r0 + x, we write V in the form

2
V ¼ A1 ðA3
A1 Þp2 þ A2 ðA3
A2 Þq2 þ A21 p2 þ A22 q2 þ A23 x2 þ 2A23 r 0 x ;

then, it follows that, for A3 > A1, A3 > A2, function V for sufficiently small x is a
positive definite function of p, q, x. On the other hand, from (7.15) and G ¼ 0, it
follows that V_  0 due to the equations of motion (7.12). Hence, by the Lyapunov
theorem, we make the conclusion about stability of the solution p ¼ q ¼ 0, r ¼ r0 for
A3 > A1, A3 > A2. This result obtained for small Reynolds numbers is quite consis-
tent with the known research result concerning stability of stationary rotations of a
body with a viscous fluid for arbitrary Reynolds numbers [9].
The results of studying the motion equations of a body with a viscous fluid (7.12)
or (7.18), i.e., the trajectories of vector G on the unit sphere, are shown in Fig. 7.2,
where we assume A1 > A2 > A3. The rotation about the axis of the middle moment
of inertia A2 is unstable (singular point of the saddle type), the rotation about the
axis of the smallest moment of inertia A3 is also unstable (the singular point is an
unstable focus), whereas the rotation about the axis of the largest moment of inertia
A1 is stable (the singular point is a stable focus). This is the only stable steady
motion of the system. The separatrices emanating from the saddle enter one of
the foci.
7.3 Free Three-Dimensional Motion of a Body with a Viscous Fluid 149

Let us turn to the quantitative study of the nonlinear transition process leading to
a stable motion. Of greatest interest is to obtain the law of decrease in time of
kinetic energy or another related quantity.
Consider first the motion under the condition 2TA1  G2 > 2TA2 corresponding
to the trajectories in Fig. 2.1 which envelope axis A1. We introduce the function


ðA2
A3 Þ 2TA1
G2
k2 ¼
2 , 0  k  1, ð7:22Þ
ðA1
A2 Þ G
2TA3

which is univalently connected with energy T. The value k ¼ 0 corresponds to a

rotation about axis A1 and k ¼ 1 to a motion along the separatrix (see Fig. 2.1). With
the help of (7.22), (7.15), we can express the derivative dk2/dt in terms of p, q, r and
the constant angular momentum G. After that, we express functions p, r in terms of
G, T, q from equations (7.13), (7.14) and then express T in terms of k2 and G from
formula (7.22). Finally, we get

dk2

¼ v
1 f G; k2 ; q , ð7:23Þ
dt

where function f is a fourth-degree polynomial in q.

The function k2, as well as T, is a slowly varying variable. Therefore, in the first
approximation, one can substitute into (7.23) the function q(t) from the unperturbed
Euler–Poinsot motion [8] (see also (2.30)):
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2TA1
G2 ht
t0 i
q¼ sn 4K ðkÞ; k : ð7:24Þ
A 2 ð A1
A 2 Þ τ

Here, τ is the period of motion depending on T, G, whereas t0 is an arbitrary

constant. We substitute (7.24) into (7.23), eliminate T again using (7.22), and
average the right-hand side of equation (7.23) over the period τ of the Euler–Poinsot
motion. In the averaging, as usual, the slow variable k2 is considered to be constant.
Omitting the cumbersome intermediate calculations which use the formulas for the
integrals of elliptic functions [5], we obtain, similar to equation (7.9), the averaged
equation in the form

dk2
  EðkÞ
¼ ð1
χ Þ 1
k2
ð1
χ Þ þ ð1 þ χ Þk2 ,
dξ K ðkÞ

 
3A2 A21 þ A23
A2 ðA1 þ A3 Þ
χ¼ , ð7:25Þ
ðA1
A3 Þ½A2 ðA1 þ A3
A2 Þ þ 2A1 A3 
t
t∗ 3vA21 A22 A23
ξ¼ , N¼ :
N ρP0 G ðA1
A3 Þ½A2 ðA1 þ A3
A2 Þ þ 2A1 A3 
2
150 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

The arbitrary constant t∗ is chosen so that the time moment t ¼ t∗ corresponds to

the separatrix crossings, i.e., k ¼ 1, 2TA2 ¼ G2 for ξ ¼ 0. Up to now it has been
assumed that 2TA2 < G2. Due to monotonic decrease of T, this condition holds for
t > t∗, i.e., for ξ > 0. Therefore, we must find for ξ  0 the solution of equation
(7.25) with the initial condition k2(0) ¼ 1. This solution k2(χ, ξ) depending on the
dimensionless parameter χ has been found numerically for different χ. It is easy to
check that the following equalities hold for the quantity χ from (7.25)
χ1
χ2
χ¼ , χ 1 ¼ ðA1
A2 Þ½A3 ðA1 þ A2
A3 Þ þ 2A1 A2  > 0,
χ1 þ χ2
χ 2 ¼ ðA2
A3 Þ½A1 ðA2 þ A3
A1 Þ þ 2A2 A3  > 0:

This implies that |χ|  1 for any physically acceptable A1, A2, A3. For all
admissible χ, function k2 decreases monotonically from 1 to 0 under the increase
of ξ from 0 to 1, while for ξ ¼ 0 we have dk2/dξ ¼ 0. Figure 7.3 shows the graphs of
functions k2(χ, ξ) obtained by calculations for χ ¼ 1,
0.5, 0, 0.5, and 1, where to
larger χ there correspond more rapidly decreasing functions.
Similarly to (7.10), we can find the asymptotics of the solution of equation (7.25)
for small k2, i.e., for large ξ,

k2 ðχ; ξÞ ¼ C2 exp½
ð3 þ χ Þξ=2, C2 > 0, ξ ! þ1: ð7:26Þ

The obtained functions k2(χ, ξ) and formulas (7.22), (7.25) describe the
motion for t  t∗, i.e., the decrease of T from G2/2A2 to G2/2A1.
For t < t∗, the inequalities 2TA2 > G2  2TA3 hold which correspond to the
trajectories (see Fig. 2.1) enveloping axis A3. This case is considered in a similar
fashion, and to obtain the solution, one just needs to interchange the letters A1 and
A3 in all the formulas (7.22), (7.24), (7.25).
Since in such an interchange χ and N just change signs, we can keep unchanged
formulas (7.25) for χ и N, but replace χ by
χ in equation (7.25) for k2, and,
besides, change the relationship between ξ and t by putting in (7.22)

ξ ¼ ðt∗
tÞ=N:

Therefore, to describe the motion for t  t∗, we need to find the solution of
equation (7.25) for ξ  0, t  t∗ with the initial condition k(0) ¼ 1. Therefore, we
can use the already found solutions by putting k2 ¼ k2(
χ, ξ). Here, the value k ¼ 0
corresponds to a rotation about the axis with the moment of inertia A3, while, as
previously, k ¼ 1 corresponds to the motion along the separatrix.
The change of energy T in any particular case can be obtained from the obtained
relations by a simple recalculation. First, we need to calculate constants χ, N from
(7.25) for the given A1, A2, A3, ρ, v, P0 and take from the presented graphs or
calculate again functions k2(χ, ξ) and k2(
χ, ξ) satisfying equation (7.25) and the
condition k(0) ¼ 1. On the basis of function k2(χ, ξ) and with the help of (7.22) and
7.3 Free Three-Dimensional Motion of a Body with a Viscous Fluid 151

Fig. 7.3 Graphs of

the squared modulus of
elliptic functions for a body

the formula t
t∗ ¼ Nξ, we can find T(t) for t  t∗, i.e., in the case T  G2/2A2.
Using function k2(
χ, ξ) and the formulas similar to (7.22) and (7.25)


ðA2
A1 Þ 2TA3
G2
k ¼
2

 , t∗
t ¼ Nξ, ð7:27Þ
ðA3
A2 Þ G2
2TA1

we determine T(t) for t  t∗, i.e., in the case T  G2/2A2. The constant t∗ can be
determined as in Sect. 7.2, provided the energy T at a certain fixed time is given.
Thus, function T(t) will be completely determined. Thus, we obtain a description of
the entire nonlinear transition process, which is the Euler–Poinsot motion with
slowly varying energy. In particular, using (7.26), we find with the help of (7.22)
and (7.27) the asymptotic behavior of T(t):
 
G2 G2 ðA1
A2 ÞðA1
A3 Þ ð3 þ χ Þðt
t∗ Þ
T ¼ þ C 2 exp
, t ! þ1,
2A1 2A21 ðA2
A3 Þ 2N
 
G2 G2 ðA2
A3 ÞðA1
A3 Þ ð3
χ Þðt
t∗ Þ
T ¼
C 2 exp , t !
1:
2A3 2A23 ðA1
A2 Þ 2N

The solution for the angular velocities in this approximation is determined by the
equations of the Euler–Poinsot motion of the type (7.24) (see [8]), in which we need
to substitute the obtained slowly varying functions T(t) or k2(t). The characteristic
time of the transition process is of the order of N from (7.25). The motion of the
body finally tends to a single stable stationary motion: a rotation about the axis of
the largest moment of inertia A1.
As an example, Fig. 7.4 presents the graphs of variation of the dimensionless
energy T1 ¼ 2TA1G
2 as a function of dimensionless time ξ1 ¼ (t
t∗)N
1. For all
the curves in Fig. 7.4, we take A1 ¼ 8, A3 ¼ 4, and A2 ¼ 5, 6, 7 for curves 1, 2, and
3, respectively. The value T1 ¼ 2 corresponds to a rotation about axis A3 (unstable
motion) and T1 ¼ 1 to a rotation about axis A1 (stable motion). Note that for ξ1 ¼ 0
(the separatrix crossing), the curves in Fig. 3.1 have horizontal tangent lines
(inflection points).
Note that the considered above motion of a free rigid body with a cavity filled
with a fluid of high viscosity can serve as a qualitative model of the motion of
planets containing the masses of molten magma.
152 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

Fig. 7.4 Graphs of the

dimensionless energy

The obtained above general equations (5.28) allow studying other problems of
the dynamics of a body with cavities containing fluid.
Let us qualitatively discuss here the issue of the motion relative to the center of
mass of a satellite filled with viscous fluid, under the action of gravitational torques
on the Keplerian orbit. An analysis of this problem is given in paper [10].
As follows from the results of Chap. 6, in the case of a satellite, which is a rigid
body, its motion about the center of mass is the superposition of the Euler–Poinsot
motion about the angular momentum vector G and the movement of the vector
G itself in the space. This vector performs the precession motion about the normal
to the orbit plane, on which finer oscillatory effects are superimposed. The value of
the kinetic energy T of the motion of the satellite relative to the center of mass and
the value of its angular momentum G with respect to the center of mass remain
constant with high accuracy.
On the other hand, if the satellite is acted upon only by the viscous torques from
the fluid (the Reynolds number is assumed to be small), then the magnitude and
direction of vector G do not change, whereas the energy T decreases monotonically.
Under the combined action of the gravitational and viscous torques, the resulting
motion will be a superposition of these two motions. To calculate it approximately,
we can do the following: First, on the basis of the solutions obtained in this section,
we find the law of decrease of kinetic energy T(t) and then substitute it into the
averaged equations derived in Sect. 6.2, for example, into equations (6.35), (6.48).
These equations describe the movement of vector G in space. After a large enough
time, the satellite will rotate about the axis of the largest principal central moment
of inertia; vector G will be directed precisely along this axis. Then, vector G itself
will move in space almost in the same way as in the case of a satellite which is a
rigid body.
In [10], the rapid rotational motion under the action of gravitational torque is
investigated for a satellite with a spherical cavity which is entirely filled with a
viscous fluid with a low Reynolds number. A system of equations of motion is
obtained containing the slow and fast variables. The averaging procedure over the
Euler–Poinsot motion and modified averaging method described in Sects. 4.6 and
4.4 are applied. It is established that the angular momentum vector G remains a
7.4 On the Motion of a Rigid Body Containing a Damper 153

constant vector inclined at a constant angle δ to the vertical of the orbit plane. In this
case, the end of vector G moves along a sphere of constant radius G0, first
counterclockwise due to the existing initial kinetic energy and then clockwise.
Moreover, the kinetic energy decreases to the value equal to one which corresponds
to stable rotation of the satellite about the axis of largest inertia A1. The analysis of
the rotational motion of the satellite in the vicinity of the largest moment of inertia
is conducted.

7.4 On the Motion of a Rigid Body Containing a Damper

M. A. Lavrentyev proposed the following model for the simulation of the motion of
a rigid body with a cavity filled with a viscous fluid (see [11]). He considered a rigid
body with a spherical cavity in which there is another rigid body of a spherical
shape. Between the sphere and the cavity walls, there is a narrow gap in which
viscous forces act (lubricating layer). This model with a finite number of degrees of
freedom has certain mechanical properties of a body with a cavity containing a
viscous fluid. Therefore, its study is of interest. Besides, such systems are of interest
also from the viewpoint of applications as they allow, in principle, performing the
damping of the relative motion of a body due to internal forces.
In Sects. 7.4–7.5, we study some properties of the described system, which will
be called “a rigid body with a damper.”
Let a rigid body G with mass m1 have a spherical cavity D of radius a. Inside the
cavity there is a solid sphere of mass m0 and radius close to a, the mass distribution
in which is spherically symmetric (e.g., uniform). The thickness h of the gap
between the sphere and the cavity walls is considered constant and small (h a)
so that the displacement of the sphere’s center relative to the center O of the cavity
D can be ignored (Fig. 7.5). Let us construct the equations of motion of the system.
The equation of motion of the center of inertia has the form (m1 + m0)wc ¼ R,
where wc is the acceleration of the center of inertia of the system and R is the
principal vector of all external forces acting on the system.
Let O1 be any point rigidly connected with the rigid body (e.g., the center of
inertia of the system or a fixed point if there is one). We introduce two Cartesian
coordinate systems: O1y1y2y3 whose axes move in an arbitrarily given manner (e.g.,
progressively) and O1x1x2x3 rigidly connected with the rigid body (see Fig. 7.5).
Let us write the equation of moments in system O1y1y2y3
ð
dK
¼ M; K ¼ r1
vdm: ð7:28Þ
dt
GþD

Here, K is the angular momentum of the body with a damper relative to point O1
when it moves relative to coordinate system O1y1y2y3; M is the principal moment
relative to point O1 of all the external forces acting on the body with a damper in
154 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

Fig. 7.5 Rigid body with

damper

this system; r1 is the radius vector measured from the point O1; v is the velocity of
the system of coordinates O1y1y2y3; and dm is the mass element. The moment
M includes, in particular, the inertia torque due to the motion of coordinate system
O1y1y2y3. The velocity v of a point of the system is equal to v ¼ ω
r1 + u1, where
ω is the angular velocity of the body relative to the system O1y1y2y3 and u is the
velocity of this point relative to the coordinate system O1x1x2x3. For the points of
the rigid body, it is, obviously, u ¼ 0, and therefore formula (7.28) for K takes the
form
ð ð
K¼ r1
ðω
r1 Þdm þ L ¼ J  ω þ L; L ¼ r1
u dm: ð7:29Þ
GþD D

Here, J is the inertia tensor of the entire system relative to point O1, the
components of which are constant in the system O1x1x2x3. The gyrostatic moment
L is the angular momentum of the damper in the coordinate system O1x1x2x3. It is
easy to see that it does not depend on the choice of the pole, and in this case it equals

L ¼ IΩ; Ω ¼ ω1
ω, ð7:30Þ

where I is the moment of inertia of the damper relative to its diameter and ω1 and Ω
are the angular velocities of the damper in the coordinate systems O1y1y2y3 and
O1x1x2x3, respectively.
Assume that the external forces acting on the damper do not generate torque
relative to its center O. The torque M1 of the damper interaction with the body
(relative to point O) is assumed equal to (
kΩ), where k is a constant coefficient of
proportionality. Then, the equation of motion of the damper relative to its center
takes the form
dω1
I ¼ M1 ¼
kΩ, k > 0: ð7:31Þ
dt
7.4 On the Motion of a Rigid Body Containing a Damper 155

We note that, from the quantities included in equation (7.31), one can construct a
dimensionless criterion R1 ¼ I/(kT) which is similar to the Reynolds number for a
body with a cavity containing fluid. Here, the constant T is a characteristic time of
the process.
Equations (7.28)–(7.31) describe the motion of a body with a damper in the
coordinate system O1y1y2y3. Generally speaking, they must be complemented by
the conventional kinematic relations, as well as the equations of motion for the
center of inertia and possibly other closing equations. Let us rewrite equations
(7.28)–(7.31), indicating by the dot the derivatives in the system of coordinates
O1x1x2x3 associated with the rigid body:

K_ þ ω
K ¼ M, K ¼ J  ω þ I ðω1
ωÞ,
ð7:32Þ
I ω_ 1 þ Iω
ω1 ¼
kðω1
ωÞ:

Next, we calculate the quantity k assuming that the interaction of the damper
with the body takes place via a thin spherical layer of an incompressible viscous
fluid with density ρ1 and kinematic viscosity v1. In the coordinate system O1x1x2x3,
the body is at rest, and the damper rotates with angular velocity Ω. The point of its
diameter which forms angle θ with vector Ω (see Fig. 7.5) has velocity Ωa sin θ in
the system O1x1x2x3. In the fluid layer of thickness h, the velocity gradient
Ωa sin θ/h occurs which causes tangential stress on the damper surface equal to
ρ1v1Ωa sin θ/h. Let us calculate the magnitude of the torque generated by these
stresses relative to point O, and then, comparing it with formula (7.31), we
determine k:
ðπ
ρ1 v1 Ωa sin θ 8π ρ1 v1 a4
M1 ¼ a sin θ  2πa sin θ  a dθ ¼ Ω,
h 3 h
0 ð7:33Þ
8πρ1 v1 a4
k¼ :
h

Let the angular velocities ω and ω1 be of the magnitude order of T
1, where T is
a characteristic time of the process; and let their derivatives ω_ , ω_ 1 in the coordinate
system O1x1x2x3 be of the order of T
2 and ω, € ω€ 1 of the order of T
3. In addition, we
assume that number R1 is small: R1 1. Without loss of generality, we can take
T as a unit of time, the radius a of the damper as a unit of length, and the quantity I/a2
as a unit of mass (the damper’s mass normalized by its radius). Then, the moment of
inertia I, angular velocities ω, ω1, and their first and second derivatives are of the
order of one, while R1 ¼ I/k with k  1 (the lubricant’s viscosity is high). Rewrite
equation (7.31), and then differentiate both its sides with respect to time in the
system O1x1x2x3:

I ω_ 1 þ Iω
ω1 ¼
kΩ, € 1 þ I ω_
ω1 þ Iω
ω_ 1 ¼
kΩ_ :
Iω ð7:34Þ

According to the assumptions made, the left-hand sides of equations (7.34) are
   
the quantities of the order of one, and thus Ω  Ω_   k
1 1. We now
156 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

substitute the equality ω1 ¼ ω + Ω into the first equation (7.34) and express Ω from
it up to small quantities of higher order:


Ω ¼
Ik
1 ω_ þ O k
2 , k  1:

Substituting this formula into equalities (7.29) and (7.30), we obtain the follow-
ing representation for angular momentum K:


K ¼ J  ω þ L, L ¼
I 2 k
1 ω_ þ O k
2 , k  1: ð7:35Þ

Let us compare this expression with expression (5.24) for the angular momen-
tum G of a body with a cavity filled with a viscous fluid at a low Reynolds number
R:

ρ a2
G ¼ J  ω
P  ω_ , R¼ 1: ð7:36Þ
v vT

Here, P is the tensor introduced in Sect. 5.3, a is a characteristic linear dimension

of the cavity, v is the kinematic viscosity of the fluid, and T is a characteristic time
of the process which is equal in the order of magnitude to ω
1. Recall that,
according to (5.30), for a spherical cavity of radius a, tensor P is spherical, and
its diagonal elements P0 are equal to

8πa7
P0 ¼ : ð7:37Þ
525

Comparing relations (7.35) and (7.36), we see that, under the imposed assump-
tions (the dimensionless numbers R ¼ I/kT and R ¼ a2/vT are small), the body with a
spherical damper is dynamically equivalent to the body with a spherical cavity
filled with an incompressible viscous fluid. These systems are described by the
same equations. For the complete mechanical equivalence of systems (with the
same rigid body, the same radius a of the cavity, the same external forces and
external torques), we need to require the equality of:
1. The mass of the damper and of the fluid in the cavity (for the equations of motion
of the center of mass to be equivalent)
2. The moments of inertia of the damper and the fluid relative to the diameter
3. The coefficients at dω/dt in equations (7.35) and (7.36), i.e., I2k
1 ¼ ρv
1P0
The mass and the moment of inertia of the lubricating layer are neglected.
Hence, we obtain the equalities which are necessary and sufficient (under the
assumptions made) for the equivalence of the systems

4 8 vI 2 56
m0 ¼ πρa3 , I¼ πρa5 , k¼ ¼ πρva3 : ð7:38Þ
3 15 ρP0 3

In equalities (7.38) we take into account the value of P0 from (7.37).

7.4 On the Motion of a Rigid Body Containing a Damper 157

Using formula (7.33), the last equality (7.38) can be written in the form
ρ1 v1 7ρv
¼ :
h a

This formula like (7.33) is valid for h a. If the conditions of equivalence

(7.38) are fulfilled, then it is easy to check that there is a connection between the
Reynolds numbers:
 
I a2
R ¼ 35R1 R1 ¼ ; R ¼ :
kT vT

If we abandon the condition of identity of rigid bodies in the simulation, then a

body with the spherical damper with R1 1 can simulate a body with a viscous
fluid with R 1 for those shapes of cavities, for which tensor P is of the form
P ¼ PE0, where P is a scalar and E0 is the unit tensor (e.g., for the cavities in the
shape of a sphere, cube, etc.).
Consider also the dampers with not spherical but axially symmetric shape: a
rigid body of revolution placed in the cavity of the same shape (in the cavity,
between the body and walls, there is still a lubricating layer). Suppose that there are
s cavities in a rigid body, which contain such axially symmetric dampers. Denote by
dj the unit vector directed along the axis of the j-th damper; by Ij its moment of
inertia about the rotation axis; by ωj and Ωj the angular velocities of the j-th damper
in the coordinate systems O1y1y2y3 and O1x1x2x3, respectively; by Lj the gyrostatic
moment of the j-th damper; and by Mj ¼
kjΩj the moment of the viscous forces
acting on the j-th damper from the body, with kj being constant proportionality
coefficients, j ¼ 1 , 2 , . . . , s. Then, the expressions for the angular momentum
K and the gyrostatic torque take a form similar to (7.29), (7.30):
X
s
K ¼ J  ω þ L, L¼ Lj , Lj ¼ I j Ωj , Ωj ¼ ωj
ω ¼ dj Ωj ,
j¼1

j ¼ 1, . . . , s: ð7:39Þ

It is taken into account here that vector Ωj is collinear with the axis dj of the
damper. The equation of motion of the j-th damper about its axis of rotation dj can
be written in a form similar to (7.31):

dωj
I j dj ¼
Mj dj ¼
kj Ωj , j ¼ 1, 2, . . . , s: ð7:40Þ
dt

Differentiate both sides of equation (7.40) with respect to time in the coordinate
system O1x1x2x3 associated with the body. Given that in this system dj is a
constant vector, we obtain by analogy with (7.34)
158 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid


I j dj ω_ j þ ω
ωj ¼
kj Ωj ,


I j dj ω€ j þ ω_
ωj þ ω
ω_ j ¼
kj Ω_ j , ð7:41Þ
j ¼ 1, . . . , s:

As above, let us assume that ω, ωj are of the order of T
1, ω_ , ω_ j are of the order
of T
2, and ω,€ ω€ j are of the order of T
3, where T is a characteristic time of the
process, j ¼ 1 , . . . , s. The Reynolds numbers Rj ¼ Ij/(kjT) for all dampers are
assumed small for j ¼ 1 , . . . , s, that is, Rj 1.
By choosing the units of measurement, we can assume, as above, that ω, ωj and
their derivatives are the quantities of the order of one and kj  1 for j ¼ 1 , . . . , s.
Then, it follows from (7.41) that Ωj and Ω_ j are small quantities of the order of k
1j .
Substituting ωj ¼ ω + djΩj in the first equation of (7.41) and taking into account the
presented estimates, we get, up to small quantities of higher order, that


Ωj ¼
I j =kj ω_ dj , j ¼ 1, . . . , s:

We substitute the obtained equation into formula (7.39), also taking into account
the relation ω_ ¼ dω=dt:

X
s X
s I2  
j dω
L¼ Lj ¼
dj dj  :
j¼1 j¼1
kj dt

Let e1, e2, e3 be the unit vectors of axes O1x1, O1x2, O1x3, while αjk ¼ dj  ek are
the direction cosines of the unit vectors dj in these axes, j ¼ 1 , . . . , s; k ¼ 1 , 2 , 3.
Then, the previous formula for L can be written in the form

dω X
s I2
j
L ¼
Q ; Qik ¼ αji  αjk , i, k ¼ 1, 2, 3: ð7:42Þ
dt j¼1
kj

Here, Q is a tensor, and Qik are its components in the coordinate system
O1x1x2x3. These components, as well as the quantities αjk, are constant, where,
obviously, Qjk ¼ Qki. It is easy to show that the quadratic form defined by matrix Qik
is nonnegative definite. Indeed,
! !2
X 3 Xs X 3 Xs X3
2
1 2
1
Qik ξi ξk ¼ Ij kj αji ξi αjk ξk ¼ Ij kj αji ξi  0:
i, k¼1 j¼1 i, k¼1 j¼1 i¼1

Comparing equations (7.36) and (7.42), we find that a rigid body with a cavity of
arbitrary shape, filled with a viscous fluid for R 1, can be simulated with the help
of a rigid body containing several axially symmetric dampers for Rj 1. For the
equivalence of these systems, it is sufficient that tensors P and Q in equations
(7.36), (7.42) are related by the equality P ¼ λQ, where λ is a coefficient of
7.5 Stability of Motion of a Rigid Body with a Damper 159

proportionality. In this case, to simulate a body with a cavity of arbitrary shape, it is

enough to have three dampers (flywheels with viscous damping), whose axes are
mutually perpendicular and parallel to the principal axes of tensor P. It is easy to see
from formulas (7.36), (7.42) that the parameters of the systems should be connected
in this case by the relations similar to (7.38):
Pji kj
¼ λ ¼ const, j ¼ 1, 2, 3:
I 2j

Thus, the equation of motion of a rigid body containing spherical or axially

symmetrical dampers are equivalent in the case of low Reynolds numbers to the
equations of motion of a body with a cavity filled with a fluid of high viscosity. In
particular, the results obtained in Sects. 7.1–7.3 are fully applicable, under the
assumptions made, also to a body with dampers. Thus, from the results of Sect. 7.3,
it follows that the only stable stationary rotation of a free rigid body with a spherical
damper and R1 1 is a rotation about the axis of the largest principal central
moments of inertia of the entire system.

7.5 Stability of Motion of a Rigid Body with a Damper

The study of the motion of a rigid body with a damper, besides similarities with a
body containing a viscous fluid, is of certain independent interest. Consider the
motion of a free rigid body with a spherical damper, no longer requiring the
condition R1 1. Putting M ¼ 0 in the equations (7.32) and subtracting the third
equation from the first, we get (the dot still means the derivative of vector in the
coordinate system O1x1x2x3 associated with the body)
J0 ω_ þ ω
ðJ0  ωÞ ¼ kðω1
ωÞ,
I ω_ 1 þ I ðω
ω1 Þ ¼
kðω1
ωÞ, ð7:43Þ
J0 ¼ J
IE0 :

Here, E0 is the unit tensor, and J0 is the tensor of inertia of the system relative to
point O1 under the condition that the entire mass of the damper is concentrated at its
center. Equations (7.43) form a closed system. They describe the motion of a free
rigid body with a damper relative to the center of inertia of the system. In addition,
O1 is the center of inertia of the body with a damper, and the coordinate system
O1y1y2y3 moves translationally.
We associate the coordinate system O1x1x2x3 with the principal axes of the
tensor of inertia J relative to point O1. Obviously, these axes will be the principal
ones also for tensor J0. Denote by p, q, r the projections of vector ω on axes O1x1,
O1x2, O1x3, respectively; by p1, q1, r1 the projection of vector ω1 on the same axes;
by A1, A2, A3 the principal moments of inertia of the entire system relative to these
axes; and by A10, A20, A30 the principal values of tensor J0 in these axes which,
160 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

respectively, equal A1
I, A2
I, A3
I. Equations (7.43) in the scalar notation

take the form
A10 p_ þ ðA30
A20 Þqr ¼ kðp1
pÞ, I ðp_ 1 þ qr 1
rq1 Þ ¼ kðp
p1 Þ,
A20 q_ þ ðA10
A30 Þrp ¼ kðq1
qÞ, I ðq_ 1 þ rp1
pr 1 Þ ¼ kðq
q1 Þ,
A30 r_ þ ðA20
A10 Þpq ¼ kðr 1
r Þ, I ðr_ 1 þ pq1
qp1 Þ ¼ kðr
r 1 Þ:
ð7:44Þ

Let us determine the possible steady motions of the body. If ω is constant ( ω_

¼ 0), then it follows from the first equation of (7.43) that ω1 is also constant and
ω1 ¼ 0. Then, multiplying scalarly both sides of the second equation (7.43) by
ω1
ω, we obtain ω1 ¼ ω. It is seen from equations (7.44) that such motion is
possible only in the case when the rotation takes place about one of the principal
axes of inertia of the system. Thus, the only possible steady motions of the system,
as for a body with a viscous fluid, are uniform rotations of the system as a rigid body
about one of the principal axes of inertia.
Let us investigate the stability of these motions. Let the unperturbed motion, the
rotation of the system about axis O1x1 with a constant angular velocity ω0 6¼ 0, be
described by the equations
p ¼ p 1 ¼ ω0 , q ¼ q1 ¼ r ¼ r 1 ¼ 0: ð7:45Þ

Let us put in the perturbed motion p ¼ ω0 + x, p1 ¼ ω0 + y and linearize equation

(7.44) near the solution (7.45):
A10 x_ ¼ kðy
xÞ, I y_ ¼ kðx
yÞ,
A20 q_ þ ðA10
A30 Þω0 r ¼ kðq1
qÞ, I q_ 1 þ Iω0 ðr
r 1 Þ ¼ kðq
q1 Þ,
A30 r_ þ ðA20
A10 Þω0 q ¼ kðr 1
r Þ, I r_ 1 þ Iω0 ðq1
qÞ ¼ kðr
r 1 Þ:
ð7:46Þ

The top two equations (7.46) are independent of other four equations, and
therefore the characteristic equation for system (7.46) splits into two equations.
After expanding the determinants, the characteristic equations are reduced to the
form

A10 Iλ2 þ kðA10 þ I Þλ, a0 λ4 þ a1 λ3 þ a2 λ2 þ a3 λ þ a4 ¼ 0,

a0 ¼ A20 A30 I 2 , a1 ¼ Ikð2A20 A30 þ A20 I þ A30 I Þ,
a2 ¼ I 2 ω20 ½ðA10
A20 ÞðA10
A30 Þ þ A20 A30  þ k2 ðA20 þ I ÞðA30 þ I Þ, ð7:47Þ
a3 ¼
Iω20 k½2ðA10
A20 ÞðA10
A30 Þ þ I ðA10
A20 Þ þ I ðA10
A30 Þ,
a4 ¼ I 2 ω20 þ k2 ðA10
A20 ÞðA10
A30 Þω20 :

Zero root of the first equation (7.47) corresponds to the fact that the initial
perturbation of the angular momentum of the whole system will remain constant.
In view of the zero root, from the consideration of linearized system (7.31), one can
obtain only necessary but not sufficient conditions for stability of motion (7.30). For
7.5 Stability of Motion of a Rigid Body with a Damper 161

the stability, it is necessary that the real parts of all roots λ of equations (7.47) are
nonpositive: Reλ  0. For the roots of the first equation (7.47), this condition is
satisfied for k  0. In order that it takes place also for the second equation, it is
necessary that the Lienard–Chipart conditions are satisfied [12], in which we can
allow the equality signs, as it is required by the inequalities Reλ  0, and not
Reλ < 0, as usual. Since a0 > 0, the Lienard–Chipart conditions for the second
equation (7.47) assume the form

a1  0, a2  0, a4  0, a1 a2 a3  a0 a23 þ a21 a4 : ð7:48Þ

Using formulas (7.47), the last inequality (7.48) can be reduced, after cumber-
some but elementary algebraic manipulations, to the form

2I 2 ω20 ðA10
A20
A30 Þ2 ½A20 ðA10
A30 Þ þ A30 ðA10
A20 Þþ

þk2 ð2A20 A30 þ A20 I þ A30 I Þ ðA10
A20 ÞðA20 þ I Þþ

þðA10
A30 ÞðA30 þ I Þ  0: ð7:49Þ

From the condition a4  0, it follows that it is necessary that either A10  A20 and
A10 A30 or A10  A20 and A10  A30. However, in the first case it is easy to see
that condition (7.49) is violated, while in the second case for k  0 all the conditions
(7.48), (7.49) are satisfied. Inequalities k  0, A10  A20, A10  A30 are the necessary
conditions of stability of motion (7.45).
Turning to sufficient conditions, we note that system (7.44) has the first integral

K 2 ¼ ðJ0  ω þ Iω1 Þ2 ¼
¼ ðA10 p þ Ip1 Þ2 þ ðA20 q þ Iq1 Þ2 þ ðA30 r þ Ir 1 Þ2 ¼ const, ð7:50Þ

which expresses the preservation of the angular momentum of the entire system.
The kinetic energy H, defined by the equality


2H ¼ A10 p2 þ A20 q2 þ A30 r 2 þ I p21 þ q21 þ r 21 , ð7:51Þ

as can be easily verified, does not increase in the motion due to equations (7.44),
i.e., H_  0, if k  0. Let us compose the Lyapunov function, following the idea of
N. G. Chetaev’s method:
h i2
V ¼ 2ðA10 þ I ÞH
K 2 þ K 2
ðA10 þ I Þ2 ω20 : ð7:52Þ

It is easily seen that function V vanishes in the unperturbed motion (7.45). As

above, we perform in the perturbed motion the substitution p ¼ ω0 + x, p1 ¼ ω0 + y
in equations (7.50), (7.51) and write V from (7.52) as a function of variables x, y, q,
q1, r, r1. In this case, the linear terms cancel each other, and after collecting similar
terms, we obtain
162 7 Motion of a Rigid Body with a Cavity Filled with a Viscous Fluid

h i
V ¼ A10 I ðx
yÞ2 þ 4ω20 ðA10 þ I Þ2 ðA10 x þ IyÞ2 þ

þ A20 ðA10 þ I
A20 Þq2
2A20 Iqq1 þ A10 Iq21 þ

þ A30 ðA10 þ I
A30 Þr 2
2A30 Irr 1 þ A10 Ir 21 þ . . . ð7:53Þ

The dots denote the terms of the third and higher orders of smallness. The first
square bracket in (7.53) is a positive definite (for ω0 6¼ 0) quadratic form of x, y. For
the positive definiteness of two other quadratic forms in (7.53), it is sufficient to
require
A10 ðA10 þ I
A20 Þ > A20 I, A10 ðA10 þ I
A30 Þ > A30 I:

Expanding the brackets and canceling out the factor (A10 + I), we transform these
inequalities to the form A10 > A20, A10 > A30. Under these conditions, function V is
positive definite. Since K_ ¼ 0, and H_  0 for k  0, the derivative of function
(7.52) is nonpositive due to equations of motion, V_  0 for k  0. Therefore, by the
Lyapunov theorem, the motion is stable under the indicated conditions (ω0 6¼ 0,
k  0, A10 > A20, A10 > A30). Note that inequalities A10 > A20, A10 > A30 are equiv-
alent to inequalities A1 > A2, A1 > A3 for the principal moments of inertia of the
entire system.
Thus, for the stability of the stationary rotation of a free rigid body with a damper
about axis O1x1 (motion (7.45)) for ω0 6¼ 0, k  0, it is necessary that the conditions
A1  A2, A1  A3 are fulfilled and sufficient that the strict inequalities A1 > A2,
A1 > A3 hold. In other words, a stable stationary rotation of a free rigid body with
a damper can only take place about the axis of the largest principal moment of
inertia.
The obtained steady motions and the conditions of their stability (necessary and
sufficient) are fully consistent with the corresponding conditions for a body with a
cavity filled with a fluid of high viscosity (see Sect. 7.3). These sufficient conditions
(A1 > A2, A1 > A3) take place also in the general case of a free rigid body with a
cavity filled with a viscous fluid (see [9]).

References

1. Chernousko, F.L.: Motion of a rigid body with cavities filled with viscous fluid at
small Reynolds number. USSR Comput. Math. Math. Phys. 5(6), 99–127 (1965)
2. Chernousko, F.L.: Motion of a Rigid Body with Viscous-Fluid-Filled Cavities. Computing Center
AN SSSR, Moscow (1968) in Russian
3. Chernousko, F.L.: The Movement of a Rigid Body with Cavities Containing a Viscous Fluid.
NASA, Washington, DC (1972)
4. Zhuravsky, A.M.: Handbook of Elliptical Functions. Academy of Sciences Press, Moscow
(1941) in Russian
References 163

5. Gradshtein, I.S., Ryzhik, I.M.: Tables of Integrals, Sums, Series and Products. Academic Press,
San Diego, CA (2000)
6. Bogolubov, N.N., Mitropolsky, Y.A.: Asymptotic Methods in the Theory of Nonlinear Oscil-
lations. Gordon and Breach Science Publisher, New York, NY (1961)
7. Volosov, V.M., Morgunov, B.I.: The Averaging Method in the Theory of Non-linear Oscil-
latory Systems. Moscow State Univ., Moscow (1971) in Russian
8. Landau, L.D., Lifshitz, E.M.: Course of Theoretical Physics, Mechanics, vol. 1. Pergamon Press,
Oxford (1976)
9. Moiseev, N.N., Rumyantsev, V.V.: Dynamics and Stability of Bodies Containing Fluid.
Springer, New York, NY (1968)
10. Akulenko, L.D., Leshchenko, D.D., Rachinskaya, A.L.: Evolution of rotations of a satellite with
cavity filled with viscous liquid. Mekh. Tverd. Tela. 37, 126–139 (2007) in Russian
11. Ishlinsky, A.Y.: The work of Mikhail Alekseyevich Lavrent’yev at the Academy of Sciences of
the Ukrainian SSR. Zh. Prikl. Mekh. Tehn. Fiz. 3, 16–19 (1960) in Russian
12. Gantmakher, F.R.: The Theory of Matrices. Chelsea Publishing, New York, NY (1959)
Chapter 8
Evolution of Rotations of a Rigid Body
in a Medium

The scheme of asymptotic solution suggested in Sects. 4.6 and 6.2 is applicable not
only to the problems of motion of a satellite relative to its center of mass but also to
other problems of fast motion of a rigid body. In Sect. 8.1, we consider fast motion
of a rigid body about a fixed point [1].
In Sect. 8.2, we study fast rotational motion of a heavy rigid body about a
fixed point in the presence of external resistance [2, 3].
In Sect. 8.3, following [4, 5], we consider the motion of a satellite relative to its
center of mass under the action of the gravitational torque and the resistant torque.

8.1 Fast Motion of a Heavy Rigid Body About a Fixed Point

Consider the fast motion of a nonsymmetric heavy rigid body about a fixed point.
We call fast motions such motions for which the moment of the applied forces
relative to the fixed point is small compared with the current value of the
kinetic energy of rotation.
We will use the same notations and coordinate systems as in Sect. 4.5 and in
Figs. 4.1 and 4.2. The origin O of all coordinate systems is taken at the fixed point of
the body. The axis Ox3 of the fixed coordinate system Ox1x2x3 is directed upward.
We assume that the body is acted upon by the force of gravity. The position of
the center of mass is determined by its coordinates (l1, l2, l3) in the body reference
system Oz1z2z3, as well as by radius vector r0 drawn from the fixed point O.
The gravitational torque equals

L ¼ mgx3  r0 , ð8:1Þ

where m is the mass of the body, g is the gravitational acceleration, and x3 is the
unit vector directed along axis Ox3.

© Springer International Publishing AG 2017 165

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_8
166 8 Evolution of Rotations of a Rigid Body in a Medium

We project equalities (8.1) on the axes of coordinate system Oy1y2y3 (Fig. 4.1),
using the direction cosines αij from (1.31). We get

X
3 X
3  
L1 ¼ mg cos δ lj α2j , L2 ¼ mg lj α3j sin δ þ α1j cos δ ,
j¼1 j¼1
ð8:2Þ
X
3
L3 ¼ mg sin δ lj α2j :
j¼1

Substituting equalities (8.2) into (4.42) and (4.46), we obtain a closed system of
equations of motion. In this case, small parameters ε, μ (see (4.47)) equal

mgr 0 ω21
ε ¼ 0, μ  ,
A 1 ω2 ω2

where ω is the magnitude of the angular velocity of rotation about the fixed point
and ω1 is the angular frequency of small oscillations of the rigid body relative to a
stable equilibrium position (as a physical pendulum). We assume the parameter μ to
be small and construct a solution in the first approximation with respect to μ.
The role of slow variables is played by G, δ, λ, T. To obtain the equations of the
first approximation for these variables, we average the right-hand sides of equa-
tions (4.42), (6.11) according to the scheme (4.55).
First, we average over ψ the projections of torque Li from (8.2), included in the
right-hand sides of equations (4.42), as well as the right-hand side of equation
(6.11), which we denote by f. The values of the direction cosines αij are substituted
into (8.2) from (1.31). The result is expressed in terms of projections Gi of the
angular momentum vector (4.45). So we arrive at
Mψ ðL1 Þ ¼ Mψ ðL3 Þ ¼ 0,
Mψ ðL2 Þ ¼ mgG1
 sin δðl1 G1 þ l2 G2 þ l3 G3 Þ,
mg cos δ 1 1
Mψ ð f Þ ¼  l1 G2 G3 þ ð8:3Þ
G A
 2 A 3   
1 1 1 1
þ  l2 G 1 G 3 þ  l3 G1 G2 :
A3 A1 A1 A2

Next we average relations (8.3) over the trajectories of the angular momentum
vector (Fig. 2.1), bearing in mind that these trajectories are symmetric with respect
to the coordinate planes. Therefore, the second averaging of function f yields zero.
Let us assume for definiteness, as before, that A1 > A2 > A3, and let us consider
the trajectories of the angular momentum vector enveloping axis Oz1 (G2 > 2TA2 on
them, and G1 does not change sign along the trajectory).
Substituting the averaged expression (8.3) into (4.42) and (6.11), we obtain a
system of the first approximation:
8.1 Fast Motion of a Heavy Rigid Body About a Fixed Point 167

G_ ¼ 0, δ_ ¼ 0, T_ ¼ 0,
mg
λ_ ¼ 2 l1 M1 ðG1 Þ: ð8:4Þ
G

Now we average the projection G1 over the Euler–Poinsot motion. This projec-
tion for the trajectories with G2 > 2TA2 is equal in absolute value to [6]:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
G1  ¼ A1 G2  2TA3
t  t0 
dn 4K ðkÞ; k , ð8:5Þ
A1 ð A1  A3 Þ τ

where τ is the period of motion, t0 is an arbitrary constant, and k is the modulus of

elliptic functions. The average value of the elliptic function in (8.5) equals [7]
π
Mu ðdn uÞ ¼ : ð8:6Þ
2K ðkÞ

Taking into account relations (8.5), (8.6), we can write the last equation (8.4) in
the form
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
πmgl A G2  2TA3
λ_ ¼ 1 1
sgnG1 : ð8:7Þ
2G2 K ðkÞ A1 ðA1  A3 Þ

Thus, the angular momentum vector slowly rotates about the vertical. The quan-
tities G, δ, and T are constant in this approximation, and the constant angular velo-
city of rotation of the angular momentum vector is determined according to (8.7)
(where k2 is given by equality (4.32)) for the case G2 > 2TA2. When G2 < 2TA2, we
must interchange A1 and A3 in formulas (4.32), (8.7) and replace l1 by l3 in (8.7).
The asymptotic solution is valid with the accuracy of μ (for G, δ, λ, T ) on the
time interval of the order of

1 ω
Δt  ¼ 2:
ωμ ω1

Let us consider some special cases of formula (8.7). For the rotation of the body
about axis Oz1, we have

1
G ¼ A1 ω, T ¼ A 1 ω2 , k ¼ 0,
2

and then it follows from (8.7) that

mgl1
λ_ ¼ 
A1 ω
168 8 Evolution of Rotations of a Rigid Body in a Medium

for the rotation in the positive and negative directions, respectively. The formula
has an analogous form also for the rotation about other principal axes of inertia.
In the case of dynamic symmetry (A1 ¼ A2), we arrive at the well-known formula
for a fast top in the gravitational field [6]:

mgl3 cos θ
λ_ ¼ ,
G

where θ is the angle of nutation (between the angular momentum vector and the
dynamic symmetry axis), which in this case is constant in the first approximation.

8.2 Rotation of a Heavy Rigid Body About a Fixed Point

in a Viscous Medium

8.2.1 Statement of the Problem and the Averaging Procedure

Following [2, 3], we consider the fast motion about a fixed point of an unsymmetric
heavy rigid body in a resistive medium.
We will use the same notation and coordinate systems as in Sect. 4.5 and in
Figs. 4.1 and 4.2. The origin O of all coordinate systems is taken at the fixed point of
the rigid body; the axis Ox3 of the fixed coordinate system Ox1x2x3 is directed
upward.
The relations between the direction cosines and Euler angles are represented by
formulas (1.31).
The equations of motion of the body relative to the fixed point can be written in
the form [8] (see also (4.42), (4.46), (6.1), (6.2)):
dG dδ L1 dλ L2
¼ L3 , ¼ , ¼ ,
dt dt G 
dt G sinδ
dθ 1 1 L2 cos ψ  L1 sin ψ
¼ G sin θ sin φ cos φ  þ ,
dt  A 1 A2  G
dφ 1 sin 2 φ cos 2 φ L1 cos ψ þ L2 sin ψ ð8:8Þ
¼ G cos θ   þ ,
dt  2 3 A A 1 A 2 G sin θ
dψ sin φ cos φ 2
L1 cos ψ þ L2 sin ψ L2
¼G þ  θ  δ:
dt A1 A2 G G

Here, Li are the moments of the applied forces relative to axes Oyi, i ¼ 1 , 2 , 3;
G is the magnitude of the angular momentum; A1, A2, A3 are the principal moments
of inertia of the body relative to the axes of the coordinate system Oz1z2z3 associ-
ated with the principal axes of inertia of the body.
The projections of vector G on the axes of the body reference system Oz1z2z3 are
represented by formulas (2.34).
8.2 Rotation of a Heavy Rigid Body About a Fixed Point in a Viscous Medium 169

As above, together with the introduced variables that form a complete set, it will
be convenient to use an important characteristic, the kinetic energy T (see (2.7)), the
time derivative of which has the form (6.11)
  2 
2T sin φ cos 2 φ 1
T_ ¼ L3 þ G sin θ cos θ þ  ðL2 cos ψ  L1 sin ψ Þþ
G A1 A2 A3
  
1 1
þ sin φ cos φ  ðL1 cos ψ þ L2 sin ψ Þ : ð8:9Þ
A1 A2

We assume that the body is acted upon by the force of gravity and the resis-
tant force of the medium. The position of the center of mass is determined by its
coordinates (l1, l2, l3) in the body reference system Oz1z2z3, as well as by radius
vector r0 drawn from fixed point O. The gravitational torque Lg is written according
to (8.1).
The dependence of the dissipative moment of the resistant forces on the
angular velocity vector ω of rotation of the rigid body is assumed to be linear.
Following (5.20), we can write expressions for the components of the moment of
viscous friction forces in the body reference system Oz1z2z3:
0 10 1
I 11 I 12 I 13 p
L ¼ @ I 21 I 22 I 23 A@ q A:
r
ð8:10Þ
I 31 I 32 I 33 r

Here, Iij are constant coefficients of the resistant torque hindering the rotation of
the body. In expression (8.10), the dissipation matrix is considered positive definite
and independent of the shape of the body and the properties of the medium.
The total applied torque equals L ¼ Lg + Lr.
Let us project relations (8.1) (8.10) on the axes of coordinate system Oy1y2y3,
using the direction cosines αij from (1.31):
X3 X3  
I i1 I i2 I i3
L1 ¼ mg cos δ lj α2j  G α31 α1i þ α32 α1i þ α33 α1i ,
j¼1 i¼1
A1 A2 A3
X
3  
L2 ¼ mg lj α3j sin δ þ α1j cos δ 
j¼1
3   ð8:11Þ
X I i1 I i2 I i3
G α31 α2i þ α32 α2i þ α33 α2i ,
i¼1
A1 A2 A3
X 3 
X I i1
3 
I i2 I i3
L3 ¼ mg sin δ lj α2j  G α31 α3i þ α32 α3i þ α33 α3i :
j¼1 i¼1
A1 A2 A3

T 0  ε  1 to be small, where
As we study the fast motion, we assume the ratio mgl
l is the distance from the center of mass to the fixed point. The resistance of the
medium is assumed to be weak of the same order of smallness kI k /G0  ε  1,
170 8 Evolution of Rotations of a Rigid Body in a Medium

where kIk is the norm of the matrix of the resistance coefficients. Here, we can take
as G0 the initial value of G.
We investigate the solution of system (8.8), (8.9) for small ε on a large time
interval t  ε1. To solve this problem, we apply the method of averaging according
to the scheme proposed in (4.55). The accuracy of the averaged solutions for the
slow variables is of the order of ε on the interval of time during which the body
makes ε1 revolutions.
Let us first consider the unperturbed motion (ε ¼ 0), in which the external
torques are zero. In this case, the rotation of the rigid body is the Euler–Poinsot
motion. The quantities G, δ, λ, T become constants, while θ, φ, ψ are some functions
of time t. The slow variables in the perturbed motion will be G, T, δ, λ and the
fast ones, the Euler angles θ, φ, ψ.
The averaging is carried out over the Euler–Poinsot motion for
nonresonant cases. Since the frequencies of the Euler–Poinsot motion ω1 ¼ τðG;T 2π
Þ
and ω2 ¼ τ0 ð2π
G;T Þ depend on G, T, then the condition of their incommensurability
may be violated (resonance phenomena). The study of resonances requires further
consideration. However, since the system is not “stuck” on a resonance (it is shown
below that variables G и T, on which the frequencies ω1, ω2 depend, decrease
monotonically), after passing over the resonance, the body motion is again
described by the equations for the nonresonant case. In this case, the accuracy in
determining the slow variables on the interval Δt  ε1 is of the magnitude of O
pffiffiffi
ð εj ln εjÞ [9, 10], which tends to zero as ε ! 0.
We assume for definiteness that A1 > A2 > A3 and consider the motion under the
condition 2TA1  G2 > 2TA2, which corresponds to the trajectories of the angular
momentum vector enveloping axis Oz1. We introduce the quantity k2 according to
(2.29), which is a constant for the unperturbed motion: the squared modulus of the
elliptic functions describing the Euler–Poinsot motion. In the perturbed motion, k2
will be a slow variable.
To construct a system of averaged equations of the first approximation, we
substitute the solution of the unperturbed Euler–Poinsot motion (2.30) into the
right-hand sides of equations (8.8), (8.9) and perform averaging with respect to
variable ψ, and then with respect to time t, taking into account the dependencies of
θ, φ on t. Such a scheme of averaging was used above in Chap. 6 in studying the
motion of a body with a triaxial ellipsoid of inertia about its center of mass. We
keep the previous notation for slow averaged variables. As a result, using formulas
for the integrals of elliptic functions [11], we obtain
8.2 Rotation of a Heavy Rigid Body About a Fixed Point in a Viscous Medium 171

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2 ffi
πmgl A G  2TA
λ_ ¼ , δ_ ¼ 0,
1 1 3
2G2 K ðkÞ A 1  A3
G  
G_ ¼  I 22 ðA1  A3 ÞW ðkÞ þ I 33 ðA1  A2 Þ k2  W ðkÞ þ
Rð k Þ 
þ I 11 ðA2  A3 Þ½1  W ðkÞ ,
2T  
T_ ¼  I 22 ðA1  A3 ÞW ðkÞ þ I 33 ðA1  A2 Þ k2  W ðkÞ þ
Rð k Þ  ð8:12Þ
ðA1  A2 ÞðA1  A3 ÞðA2  A3 Þ I 33 2 
þ k  W ðk Þ þ
Sðk Þ A3 
I 22   I 11 ðA2  A3 ÞRðkÞ
þ 1  k W ðkÞ þ
2
½ 1  W ðk Þ ,
A2 A1 Sð k Þ
RðkÞ ¼ A1 ðA2  A3 Þ þ A3 ðA1  A2 Þk2 ,
Eð k Þ
W ðkÞ ¼ 1  , SðkÞ ¼ A2  A3 þ ðA1  A2 Þk2 :
K ðk Þ

Here, K(k) and E(k) are the complete elliptic integrals of the first and second
kind, respectively.
Differentiating the expression for k2 from (2.29) and using equations (8.12) for
_
G and T_ , we obtain the differential equation:

dk2    E ðk Þ
¼ ð1  χ Þ 1  k2  ð1  χ Þ þ ð1 þ χ Þk2 ,
dξ K ðk Þ ð8:13Þ
χ ¼ ð2I 22 A1 A3  I 11 A2 A3  I 33 A1 A2 Þ=½ðI 33 A1  I 11 A3 ÞA2 ,
ξ ¼ ðt  t∗ Þ=N, N ¼ A1 A3 =ðI 33 A1  I 11 A3 Þ:

Here, t∗ is a constant. To the value k2 ¼ 1, there corresponds the equality

2TA2 ¼ G2, which is related to the separatrix for the Euler–Poinsot motion. If for
some solution of equation (8.13) the equality k2 ¼ 1 is attained, then we choose t∗
so that k2 ¼ 1 for ξ ¼ 0, t ¼ t∗. Note that, depending on the relations between the
quantities I11, I33, A1, A3, parameter N may assume positive or negative values;
it has the dimension of time.
It follows from equations (8.12) that the presence of the medium resistance leads
to the evolution of both the kinetic energy T of the body and the magnitude of the
angular momentum G. It can be directly observed that, in the first approximation,
the change of T and G is influenced only by the medium resistance force; moreover,
the equations include only the diagonal coefficients Iii of the matrix of the friction
torque introduced in (8.10). The terms containing the off-diagonal components Iij,
i 6¼ j drop out upon averaging. The angular velocity λ_ of rotation of the angular
momentum vector about the vertical depends on the action of both the gravity force
and the medium resistance force. Note that the action of these forces does not
lead to a change in the angular variable δ, and the deviation from the vertical
remains constant in this approximation.
172 8 Evolution of Rotations of a Rigid Body in a Medium

Equation (8.13) describes the averaged motion of the end of the angular momen-
tum vector G on the sphere of the radius G. The third equation (8.12) describes the
change with time of the sphere radius.
The expression in the braces in the right-hand side of the equation for G is
positive (for A1 > A2 > A3) because of the inequalities (1  k2)K E K (see [11]).
Each coefficient at Iii is a nonnegative function of k2; moreover, they cannot all be
zero. Therefore, dG/dt < 0 for G > 0, that is, variable G is strictly decreasing for all
k2 2 [0, 1]. Similarly, it can be shown that the kinetic energy is also strictly
decreasing.
Equations (8.12), (8.13) for G, T, k2 admit integration in quadratures. We write
these equations in the form
     2 
 
G_ ¼ Gf G k2 , T_ ¼ Tf T k2 , k ¼ f k k2 , ð8:14Þ

where fG, fT, and fk are functions defined by equations (8.12), (8.13). Hence, we find
2 2 3 2 2 3
kð kð
  6 7   6 7
G k2 ¼ G0 exp4 FG ðsÞds5, T k2 ¼ T 0 exp4 FT ðsÞds5,
k20 k20
ð8:15Þ
  kð
2

 2  f G, T k 2 ds
FG , T k ¼   , ¼ t  t0 :
f k k2 f k ðsÞ
k20

Evaluating function fG from (8.14), we find that the following differential

inequality holds

G_
f G f Gþ , k2 2 ½0; 1, ð8:16Þ
G

where f G , f Gþ are positive constants. Hence, integrating (8.15), we obtain an

estimate for G:

G0 expðf G tÞ G G0 expðf Gþ tÞ: ð8:17Þ

Similar to (8.17), inequalities hold for T; they are obtained by replacing G by T.

8.2.2 Analysis of the Equation for k2

The main stage in the study of the body motion is the analysis of equation (8.13).
Interestingly, this equation coincides with the equation (7.25), obtained for the case
of free spatial motion of the body with a cavity filled with a fluid of high viscosity
[12–14]. Note that equation (8.13) does not include the acceleration of gravity. The
8.2 Rotation of a Heavy Rigid Body About a Fixed Point in a Viscous Medium 173

evolution of k2 is influenced only by the resistance of the medium; and, due to the
fact that this equation is integrated independently, a partial separation takes place of
the influence of resistance and gravity. In this case, complete separation does not
take place, because the slowly decaying variables G, T are included in the right-
hand side of the expression for λ_ . Note that in Sect. 8.1, we have studied the effect
of the gravitational perturbation torque on the rigid body motion (no resistance);
in that case, G and T remain constant.
It is easy to verify that quantity χ from (8.13) can be represented in the form
A3 χ 1  A1 χ 2
χ¼ , χ 1 ¼ I 22 A1  I 11 A2 , χ 2 ¼ I 33 A2  I 22 A3 :
A3 χ 1 þ A1 χ 2

Since quantities χ 1, χ 2 can take arbitrary values, then, depending on the para-
meters of the problem, quantity χ changes from 1 to +1.
Note that in [12–14], where equation (8.13) was obtained for the first time, the
inequalities χ 1 > 0, χ 2 > 0 were fulfilled, and, therefore, |χ| 1 (see Sect. 7.3). In
paper [15], an equation of the type (8.13) for a rigid body with a cavity of arbitrary
shape filled with a highly viscous fluid was considered, where parameter χ varied in
the range |χ| 3 (see Sect. 7.3). In these works, the numerical integration of
equation (8.13) with the initial condition k2(0) close to 1 was performed. It was
shown that function k2 decreases monotonically from 1 to 0 with the growth of ξ;
the greater the χ, the faster it decreases.
The equation of the form (8.13) was also obtained in [16] in the study of the
influence of the eddy currents on the rotation and orientation of a satellite with a
triaxial ellipsoid of inertia and in paper [17] in the consideration of the motion about
the center of mass of a fast rotating rigid body under the action of forces appearing
in the movement of a conductor in a uniform magnetic field.
Next we study the family of solutions of equation (8.13) corresponding to
various χ 2 (1, +1). Note that for χ <  3 new qualitative effects appear,
while for χ > 3 the character of the solution is the same as for |χ| 3. Indeed, as
can be seen from the graphs of functions k2(χ, ξ) shown in Fig. 8.1 for χ ¼  3, 0, 1,
3, 5, 8, to the larger χ there correspond more rapidly decreasing functions of the
argument ξ.
For χ <  3, equation (8.13) for k2 admits stationary points k2 ¼ k2∗ , i.e.,
regardless of G and T, the value of k2 remains constant due to equation (8.13) for
an appropriate choice of the initial conditions. Note that for χ >  3 there are no
such stationary points (besides k ¼ 0, k ¼ 1).
Let us determine the quasi-stationary solutions k2 ¼ k2∗ . To this end, we equate
the right-hand side of (8.13) to zero. Then we solve the resulting equation with
respect to χ:
174 8 Evolution of Rotations of a Rigid Body in a Medium

Fig. 8.1 Graphs of the

squared modulus of
elliptic functions for a body
in viscous medium

Fig. 8.2 Numerical

determination of quasi-
stationary motions

 
k2  1 þ 1 þ k2 EðkÞ=K ðkÞ
χ¼   : ð8:18Þ
1  k2 ½EðkÞ=K ðkÞ  1

The graph of the dependence of χ on k2, which has been determined numerically,
is depicted by curve 1 in Fig. 8.2. It follows that for any χ <  3 there exists a
unique value k2∗ 2 ð0; 1Þ corresponding to the quasi-stationary motion
k2 ¼ k2∗ ¼ const. Curve 2 in Fig. 8.2 will be discussed below, in Sect. 8.2.4.
A numerical analysis of the solutions of (8.13) was carried out for χ <  3. For
the given values k2∗ 2 ð0; 1Þ corresponding to the quasi-stationary motion, the
appropriate values of χ were determined by formula (8.18). Figure 8.3 presents
the typical graphs of functions k2(χ, ξ), obtained by numerical integration of
equation (8.13). Here the continuous curves correspond to the value k2∗ ¼ 0:8,
whereas the curves with markers, to the value k2∗ ¼ 0:2. To each of these values,
there correspond three branches. As the initial conditions for the upper branches, we
8.2 Rotation of a Heavy Rigid Body About a Fixed Point in a Viscous Medium 175

Fig. 8.3 Graphs of the

squared modulus of
elliptic functions obtained
by numerical integration

choose k2(0) ¼ 1  δ, where δ  1. Two lower branches on each graph were drawn
for the initial conditions k2 ð0Þ ¼ 0:5k2∗ . In this case, the increasing branch corre-
sponds to the integration for ξ > 0, while the decreasing branch is a mirror image
with respect to the line ξ ¼ 0 of the dependence k2(χ, ξ) obtained for ξ < 0.
The depicted curves allow constructing a solution of equation (8.13) for these
parameters for any initial condition. Indeed, equation (8.13) for k2 being auto-
nomous, the solution k2(χ, ξ) for any initial conditions is determined by the shift of
the reference point along the axis ξ. Therefore, for any initial value k2 ¼ k20 ,
selecting the appropriate branch of the graphs, it is possible to describe the
subsequent change of k2 in this branch. If k20 > k2∗ , then we take the upper branch;
if 0:5k2∗ k20 < k2∗ , the middle one.
On the other hand, if k20 < 0:5k2∗ , we take the lower branch, the movement over
which with the increasing of ξ occurs in the negative direction till k2 ¼ 0:5k2∗ , and
then we move to the middle branch. For k20 ¼ k2∗ , we have a stationary solution.

8.2.3 Qualitative Investigation of the Specific Cases

of the Rigid Body Motion

Let us consider some particular cases of motion. When I33A1 ¼ I11A3, we have
|N| ! 1, |χ| ! 1 in relations (8.13). After evaluation of indeterminate forms,
instead of equation (8.13), we get
176 8 Evolution of Rotations of a Rigid Body in a Medium

 
dk2 2 Eð k Þ
¼ ðI 11 A3  I 22 A1 Þ 1  : ð8:19Þ
dt A1 A2 K ðk Þ

Therefore, when I11A3 > I22A1, the variable k2 increases and approaches one; for
I11A3 < I22A1, the quantity k2 decreases and tends to zero, i.e., the motion tends to a
rotation about the axis Oz1 corresponding to the maximum moment of inertia A1
(see Fig. 2.1).
It follows from (8.13) that the solution k2 ¼ 0 satisfies the equation. Such quasi-
stationary motion corresponds to a slow rotation about the axis with the largest
moment of inertia.
From equations (8.12) obtained for the motions under the condition 2TA1G2
> 2TA2, we get the following expressions for the variables G and T when k2 ¼ 0:
   
I 11 I 11
G ¼ G0 exp  t , T ¼ T 0 exp 2 t :
A1 A1

Setting formally k2 ¼ 1, which corresponds to the motion along the separatrix of

the Euler–Poinsot case, we have
   
I 22 I 22
G ¼ G0 exp  t , T ¼ T 0 exp 2 t :
A2 A2

Thus, in the particular cases of the rigid body rotation about axis Oz1 and the
motion along the separatrix, the presence of the medium resistance force leads to
the fact that the magnitude of the angular momentum and kinetic energy decrease
exponentially.
A similar reasoning can be carried out in the region 2TA2>G2  2TA3. More-
over, the value k2 ¼ 0 is connected with the rotation about axis Oz3 corresponding to
the minimum moment of inertia A3 (see Fig. 2.1).
For small k2 corresponding to the motions of a rigid body close to rotations about
axis Oz1, the right-hand side of equation (8.13) can be simplified by using the
decomposition of complete elliptic integrals into the series with respect to k2
[11]. In this case, equation (8.13) can be integrated, and its asymptotic solution is
written as
   
ð3 þ χ Þξ A3 χ 1 þ A2 ðI 33 A1  I 11 A3 Þ
k ¼ C1 exp 
2
¼ C2 exp  t , ð8:20Þ
2 A1 A2 A3

where C1 > 0, C2 > 0 are constant.

In the case of small k2, the analytical expressions for the magnitude of the
angular momentum and kinetic energy (8.15) can be obtained in an explicit form.
For example, with an error O(k4), formula (8.15) for G can be written in this case as
follows:
8.2 Rotation of a Heavy Rigid Body About a Fixed Point in a Viscous Medium 177

    
I 11 A3 χ 1 þ A2 ðI 33 A1  I 11 A3 Þ
G ¼ G0 exp  t þ b1 exp  t 1 ,
A1 A1 A2 A3
ð8:21Þ
C2 ðI 33 A1  I 11 A3 Þ
b1 ¼  :
2ðA1  A3 ÞðI 22 A1 A3 þ I 33 A1 A2  2I 11 A2 A3 Þ

The dependence T(t) can be expressed similarly.

For the values of k2 close to one and corresponding to the motions of a rigid body
near the separatrix, the right-hand side of equation (8.13) can be written applying
the asymptotic expansions of E(k) and K(k) for k2  1 [11]. As a result of expansion
and subsequent integration of equation (8.13), we obtain
  
1  k2 4 1
ξ¼ logpffiffiffiffiffiffiffiffiffiffiffiffiffi þ : ð8:22Þ
2 1k 2 2

Note that the derivative of function k2 equals zero for ξ ¼ 0. Besides, equation
(8.13) allows an exact particular solution k2 ¼ 1; therefore for k2 ¼ 1 the uniqueness
of solutions is lost. This is due to the fact that for k2 ¼ 1 the periodic Euler–Poinsot
motions degenerate into aperiodic motion along the separatrix, and the conditions
of applicability of the averaging method are violated. However, as follows from
[9, 18, 19], the averaging method is suitable for describing the motion for all initial
conditions, except for a set of small measure; only the accuracy of the method gets
worse.

8.2.4 Investigation of the Stability of Quasi-stationary

Motions

We analyze the stability of quasi-stationary motions found in Sects. 8.2.2 and 8.2.3
within the framework of averaged equation (8.13). To do this, we determine the
sign of function ∂Φ/∂k2 for the value of χ corresponding to quasi-stationary
motions (here Φ(k2, χ) is the right-hand side of equation (8.13)). In Fig. 8.2, curve
2 shows the graph of function ∂Φ/∂k2 obtained by numerical calculation. As it
follows from the graph, ∂Φ/∂k2 < 0, i.e., all quasi-stationary motions in Sect. 8.2.2
are asymptotically stable with respect to variable k2(in the sense of [20] for ξ  0).
This can also be seen from the graph in Fig. 8.3.
In the true time t  t∗, there is stability for N > 0 or I33A1 > I11A3 (see (8.13)). In
the opposite case, when N < 0, I33A1< I11A3, these quasi-stationary motions are
unstable.
Asymptotic stability is understood in the sense that for small deviations of the
trace of vector G on the unit sphere from the motion along the Euler–Poinsot
trajectory corresponding to quasi-stationary motion, the trace of vector G is trying
to return to this trajectory as times goes.
178 8 Evolution of Rotations of a Rigid Body in a Medium

According to (8.20), when ξ > 0, the quasi-stationary motion k2 ¼ 0 is asymp-

totically stable for χ >  3 and unstable for χ <  3. In the true time for t  t∗, this
motion can be asymptotically stable or unstable, depending on the value of χ and
the sign of parameter N.
Based on the conducted analysis, we obtain the following qualitative picture of
the motion. Consider first the case N > 0. For t  t∗, i.e., in the region 2TA1  G2
> 2TA2, the motion is described by function (8.20), formulas (2.29) for k2, and also
equations (8.12) and (8.13). When t t∗, the inequalities 2TA2 > G2  2TA3 are
fulfilled corresponding to the trajectories of the angular momentum vector
enveloping axis Oz3 (Fig. 2.1). In this case, we must interchange A1 and A3, I11
and I33 in formulas (2.29), (8.12), (8.13), as well as replace l1 by l3 in (8.12). Then
equation (8.13) keeps its form, but we need to replace χ by χ and N by N there.
The motion is similarly determined for N < 0. It is assumed that at the time
moment t ¼ t∗ the motion (one of the branches in Fig. 8.3) passes over the
separatrix; however, as already noted, there is a possibility of “getting stuck” during
indeterminate time for an initial data set of small measure [9, 18, 19].
Figure 8.4 shows the character of variation of the quantity k2 in its dependence
on χ and N in the true time t. The points are indicated corresponding to the quasi-
stationary motions, while the arrows show the direction of motion. The letters z1, z2,
z3 are the axes of the body, to which corresponds a given value of k2; moreover, on
the left from z2, there is the region where 2TA1  G2 > 2TA2 and on the right, the
region in which 2TA2 > G2  2TA3.
The following interpretation can be given to the obtained results. We introduce
the notation
μi ¼ I ii =Ai , i ¼ 1, 2, 3, βi ¼ μi =μ2 , i ¼ 1, 2, 3: ð8:23Þ

The rotation of the body about one of the principal axes, for example, Oz1,
under the action of dissipative moment is described by the relations

dω
A1 ¼ I 11 ω, ω ¼ const
expðμ1 tÞ:
dt

Therefore, the quantities μi in (8.23) are the braking coefficients for rotations
about the axes of inertia Ozi. The dimensionless quantities βi are equal to the corre-
sponding coefficients divided by μ2; in particular, β2 ¼ 1. We rewrite relations
(8.13) for χ and N in terms of βi as
2  β1  β3 1
χ¼ , N¼ : ð8:24Þ
β3  β 1 μ 2 ðβ 3  β 1 Þ

In the plane β1β3, we draw the line β3 ¼ β1, on which N changes its sign, and the
lines 1 + β1 ¼ 2β3 and 1 + β3 ¼ 2β1 which by (8.24) correspond to equalities
χ ¼  3. These lines divide the quadrant β1 > 0, β3 > 0 into six regions shown in
Fig. 8.5. The numbers of the regions refer to the number of qualitative pictures of
8.2 Rotation of a Heavy Rigid Body About a Fixed Point in a Viscous Medium 179

Fig. 8.4 Variation of k2 for

different cases

Fig. 8.5 Types of quasi-

stationary motions

motions shown in Fig. 8.4. It can be seen that the number of quasi-stationary modes
of motion and their stability depend on the relative value of the attenuation coeffi-
cients μi of rotations about the principal axes of inertia.
Thus, in the considered approximation, the perturbed motion of the body is
composed of the fast Euler–Poinsot motion about the vector G and the slow evol-
ution of the parameters of this motion.
The magnitudes of the angular momentum and kinetic energy are strictly
decreasing, and their change depends only on the medium resistance. The motion
of the vector G itself in the space is described by the first two equations of system
(8.12) and occurs with a constant deviation from the vertical δ ¼ const. In contrast
to the case of the action of the gravity force only, considered in Sect. 8.1, the
velocity of rotation of vector G about the vertical is variable. The evolution of
parameters of the Euler–Poinsot motion in the coordinate system associated with
the body is described by equation (8.13) and qualitatively represented in Figs. 8.4
and 8.5.
180 8 Evolution of Rotations of a Rigid Body in a Medium

8.2.5 The Case of Dynamic Symmetry

For an axially symmetric body (A1 ¼ A2), the system of averaged equations (8.12)
for the slow variables takes the form [21]

mgl3
λ_ ¼ cos θ, δ_ ¼ 0,
G
 2 
G_ sin θ I 33
¼ ðI 11 þ I 22 Þ þ cos 2 θ ,
G 2A1 A3

 
_θ ¼ sin θ cos θ I 33  I 11 þ I 22 : ð8:25Þ
A3 2A1

Equations (8.25) indicate that the medium resistance leads to the evolution of
variables G and θ, which remained constant in Sect. 8.1 in the absence of the
medium resistance.
Integrating the last equation from (8.25), we obtain
 
I 33 I 11 þ I 22
tan θ ¼ tan θ0 exp  t : ð8:26Þ
A3 2A1

þI 22
It is seen from (8.26) that for IA333 > I112A 1
angle θ increases and tends to π2. The
final motion here is the rotation about an axis perpendicular to the axis of
dynamic symmetry.
When I11 ¼ I22 ¼ I33 ¼ I and A1 > A3, a dynamically elongated body overturns.
This result was obtained in [22] for a dynamically symmetric satellite under the
þI 22
action of the forces of aerodynamic dissipation. In the case I112A 1
> IA333 , angle θ
decreases and tends to zero. In this case, for I11 ¼ I22 ¼ I33 ¼ I, the stability condi-
tion A3 > A1 obtained in [23] is fulfilled.
Thus, the dynamically compressed body is stabilized about its axis of symmetry,
which also coincides with the conclusion [22]. The final motion is a rotation about
the axis of dynamic symmetry. It can be concluded that the motion tends to a
rotation about the axis with the largest moment of inertia.
The third equation of (8.25), by substituting expression (8.26) for θ, is integrated
explicitly:
    
I 33 I 11 þ I 22
G ¼
2
G20 cos 2 θ0 exp 2 t þ tan θ0 exp 
2
t : ð8:27Þ
A3 A1

Let us study the rotation velocity λ_ of the angular momentum vector about the
vertical. From equations (8.25) and relations (8.26), (8.27), it follows that
8.3 Fast Rotation of a Satellite About Its Center of Mass Under the Action of. . . 181

1=2
mgl3 ð1 þ tan 2 θ0 Þ
λ_ ¼ n
 h
 io : ð8:28Þ
G0 exp  IA333 t þ tan 2 θ0 exp IA333  I11AþI1 22 t

Moreover, according to (8.28), in the case þI22

> I112A1
the angular velocity λ_ of
I 33
A3
rotation of the angular momentum vector G about the vertical decreases and λ_ ! 0
mgl ð1þtan θ0 Þ
2 1=2

as t ! 1. On the other hand, if I11 þI22

2A1 ¼ IA333 , then λ_ ! 3 G0 tan 2 θ0 as t ! 1.
I 11 þI 22
Note that for 2A1 > I 33
the denominator of expression (8.28) decreases and
A3
tends to zero as t ! 1. Consequently, λ_ increases in this case and λ_ ! 1 as
t ! 1. Thus, in all cases we have carried out a complete analysis of the
body motion in a resistive medium with linear resistance.
The fast motion of a heavy rigid body with a spherical cavity filled with a fluid of
high viscosity about a fixed point in a resistive medium is considered in [24].

8.3 Fast Rotation of a Satellite About Its Center of Mass

Under the Action of Gravitational Torque in a Resistive
Medium

Let us consider the motion of a satellite about its center of mass under the combined
action of the moments of the gravitational forces and the forces of resistance [4, 5].
We introduce three Cartesian coordinate systems as indicated in Sect. 4.6. The
position of the angular momentum vector G relative to the center of mass of the
satellite in the coordinate system Oxi, i ¼ 1 , 2 , 3, associated with the orbit, is
determined by angles λ and δ as shown in Fig. 4.1.
We consider a dynamically nonsymmetric satellite, whose moments of inertia,
for definiteness, satisfy the inequality A1>A2 > A3 under the assumption that the
angular velocity ω of the satellite motion relative to the center of mass is substan-
tially greater than the angular velocity ω0 of the orbital motion, i.e., ε ¼ ω0/ω 
A1ω0/G  1. In this case, the kinetic energy of the body rotation is large compared
with the perturbation torques.
The orbit of the satellite is assumed to be circular, so the atmosphere density
can be considered constant during the motion.
We write the equations of motion of the body relative to the center of mass in the
form (4.42), (4.46).
The projections Li of the moment of external forces are composed of the gravi-
g
tational torque Li and the moment Lir of the external resistance forces. The
projections of these moments on the axes Oyi, i ¼ 1 , 2 , 3, associated with the
angular momentum vector (see Fig. 4.1), can be written in the form (5.18) and
182 8 Evolution of Rotations of a Rigid Body in a Medium

(5.20). Below the projection on axis Oy1 is given; the projections on the other axes
have a similar form:

X
3  
L1 ¼ L1g þ L1r ¼ 3ω20 σ 2 σ j s3j  σ 3 σ j s2j 
j¼1
X3  
I i1 α1i α31 I i2 α1i α32 I i3 α1i α33
G þ þ ,
A1 A2 A3 ð8:29Þ
i¼1
X
3
smj ¼ Ap αjp αmp , σ 1 ¼ cos ðv  λÞ cos δ,
p¼1
v ¼ ω0 t, σ 2 ¼ sin ðv  λÞ, σ 3 ¼ cos ðv  λÞ sin δ:

Here, it is assumed that the moment Lr of resistance forces can be represented in

the form Lr ¼  Iω similar to that considered in Sect. 5.2.
In relations (8.29), we also use formulas (5.17) and (5.19), where v ¼ ω0t is the
true anomaly measured from an arbitrary point of the circular orbit chosen as the
initial one.
The derivative of the kinetic energy has the form (8.9). The goal is to investigate
the motion for small ε on a large time interval t  ε2. To solve this problem, we use
the averaging method according to (4.55).
As in the problem of the rigid body motion about a fixed point in a viscous
medium (Sect. 7.2), in the considered problem, in the case of unperturbed motion
(the Euler–Poinsot motion), the quantities G, δ, λ, T turn to constants, whereas φ, ψ,
θ are some functions of time t. The slow variables in the perturbed motion will be G,
δ, λ, T and the fast ones, the Euler angles φ, ψ, θ.
Consider the motion under the condition 2TA1  G2 > 2TA2, corresponding to
the trajectories of the angular momentum vector enveloping the axis Oz1 of the
maximum moment of inertia. We introduce the quantity k2 according to (2.29),
which is a constant in the unperturbed motion: the squared modulus of the
elliptic functions describing this motion.
For the construction of the averaged system of the first approximation, we
carry out, following (4.55), the averaging with respect to variable ψ and then
with respect to time t taking into account the dependencies φ and ψ on t.
In addition, we keep the previous notation for the slow variables δ, λ, G, T.
As a result, we get
8.3 Fast Rotation of a Satellite About Its Center of Mass Under the Action of. . . 183

dδ 3ω2 dλ 3ω20
¼  0 σ2σ3 N∗ , ¼ σ1 σ3N∗,
dt 2G dt 2G sin δ
dG G 
¼ I 22 ðA1  A3 ÞW ðkÞþ
dt Rð k Þ  
þ I 33 ðA1  A2 Þ k2  W ðkÞ þ I 11 ðA2  A3 Þ½1  W ðkÞ ,
dT 2T   ð8:30Þ
¼ I 22 ðA1  A3 ÞW ðkÞ þ I 33 ðA1  A2 Þ k2  W ðkÞ þ
dt Rð k Þ 
ðA1  A2 ÞðA1  A3 ÞðA2  A3 Þ I 33 2 
þ k  W ðk Þ þ
Sðk
Þ A3 
I 22   I 11 ðA2  A3 ÞRðkÞ
þ 1  k W ðkÞ þ
2
½1  W ðkÞ :
A2 A1 Sð k Þ

Here,
  
∗ 2A1 T K ð k Þ  Eð k Þ
N ¼ A2 þ A3  2A1 þ 3 1 A3 þ ð A2  A3 Þ ,
G2 K ðkÞk2

whereas R(k), W(k), S(k) are introduced according to (8.12).

The third and fourth equations of system (8.30) coincide with the equations of
change of the angular momentum and kinetic energy (8.12). This is due to the
fact that the evolution of these quantities takes place only under the action of the
medium resistance.
The differential equation for the change of the quantity k2 is again of the form
(8.13). From this it follows that the averaged motion of the end of the angular
momentum vector G on the sphere of radius G is of the same nature as the one
studied in Sect. 8.2.
Consider the system consisting of the first two equations of system (8.30).
In view of equalities (5.19) and v ¼ ω0t, we obtain

3ω20 ∗
δ_ ¼ N sin δ sin ðv  λÞ cos ðv  λÞ,
G
3ω 2 ð8:31Þ
λ_ ¼ 0 N ∗ cos δcos 2 ðv  λÞ,
2G
v_ ¼ ω0 :

We have obtained a system of a special type. To solve it, we apply averaging

with respect to variable v, considering δ, λ as slow variables. The system takes the
form
2
3ω
δ_ ¼ 0, λ_ ¼ 0 N ∗ cos δ: ð8:32Þ
4G

Thus, vector G retains a constant angle δ with the normal to the orbital plane.
184 8 Evolution of Rotations of a Rigid Body in a Medium

The system of equations consisting of equations (8.32), the last two of equations
(8.30) and the equation (8.13), which we rewrite as
 
dk2 I 33 A1  I 11 A3   
2 Eðk Þ
¼ ð1  χ Þ 1  k  ð1  χ Þ þ ð1 þ χ Þk
2
, ð8:33Þ
dt A 1 A3 K ðk Þ

has been integrated numerically. The integration was carried out under the follow-
ing initial conditions: k2(0) ¼ 0.99; δ(0) ¼ 0.785 rad; λ(0) ¼ 0.785 rad and the
values of the principal central moments of inertia of the body A1 ¼ 3.2; A2 ¼ 2.6;
A3 ¼ 1.67. The numerical calculation was performed for a circular orbit (e ¼ 0).We
considered two possible options for the resistance coefficients: I11 ¼ 2.32,
I22 ¼ 1.31, I33 ¼ 1.43 and I11 ¼ 0.92, I22 ¼ 5.23, I33 ¼ 1.67. In the first case, the
value of χ in equation (8.33) is negative and equal to 4.48, and in the second it is
positive and equal to 3.85.
Some calculation results in the normalized dimensionless variables are presented
in Figs. 8.6 and 8.7, which show the dependencies of G and T on time for the two
indicated cases. Curves 1 correspond to the value χ ¼  4.48 and curves 2 to the
value χ ¼ 3.85. It is seen that for χ > 0 functions G and T decrease faster than for
χ < 0.
In the case of a dynamically symmetric satellite (A1 ¼ A2), the equations of
motion are greatly simplified. The equations for the magnitude of the angular
momentum G and the nutation angle θ, averaged over the Euler–Poinsot motion,
are reduced to the following:
 2 
_ sin θ I 33
G ¼ G ðI 11 þ I 22 Þ þ cos θ ,
2

 2A1  A3
ð8:34Þ
_θ ¼ I 33  I 11 þ I 22 sin θ cos θ,
A3 2A1

which coincide with equations (8.25) for the rotation of a body about a fixed point in
a resistant medium. Thus, the evolution of quantities G and θ takes place only under
the influence of resistance forces.
Recall that this system (8.34) can be integrated analytically; its exact solution is
given by relations (8.26) and (8.27).
The averaged equations (8.32) in the case A1 ¼ A2 under consideration assume
the form
 
3ω2 cos δ 3
δ_ ¼ 0, λ_ ¼ 0 ðA1  A3 Þ 1  sin 2 θ : ð8:35Þ
2G 2

Here are some results of calculations for the values of the satellite’s moments of
inertia A1 ¼ 4.18, A3 ¼ 1.67 and two variants of the resistance coefficients:
8.3 Fast Rotation of a Satellite About Its Center of Mass Under the Action of. . . 185

Fig. 8.6 Graphs of the

satellite angular momentum

Fig. 8.7 Graphs of the

satellite kinetic energy


1 I 11 ¼ 2:32, I 22 ¼ 1:31, I 33 ¼ 1:42;
ð8:36Þ
2 I 11 ¼ 2, I 22 ¼ 1, I 33 ¼ 0:5:

In the first case (8.36), the quantity in the brackets in equation (8.34) for θ is
positive, while in the second case, negative.
Figures 8.8 and 8.9 show the dependencies of G(τ) and θ(τ) on the normalized
time τ, where curves 1 and 2 correspond to the variants 1 and 2 from (8.36).
It can be seen that angle θ increases in case 1 and decreases in case 2.
Substituting the solution (8.26), (8.27) into equation (8.35) for λ, we obtain the
differential equation
186 8 Evolution of Rotations of a Rigid Body in a Medium

Fig. 8.8 Dependence of the

angular momentum on the
normalized time

Fig. 8.9 Dependence of the

nutation angle on the
normalized time

Fig. 8.10 Dependence of

the orientation angle of the
angular momentum on time
(case 1)
8.3 Fast Rotation of a Satellite About Its Center of Mass Under the Action of. . . 187

Fig. 8.11 Dependence of

the orientation angle of the
angular momentum on time
(case 2)

Fig. 8.12 Dependence of

the orientation angle of the
angular momentum on time
(case 1, small time)


dλ 3ω20 cos δðA1  A3 Þ 1  12tan 2 θ0 expð2btÞ expðμtÞ
¼ , ð8:37Þ
dt 2G0 cos θ0 ½1 þ tan 2 θ0 expð2btÞ3=2

where we introduce the notation

I 33 I 11 þ I 22 I 33
b¼  , μ¼ :
A3 2A1 A3

The results of solving equation (8.37) for the two options (8.36) are shown in
Figs. 8.10 and 8.11.
Note that the direction of change of angle λ, that is, the derivative λ_ , according to the
second equation of (8.35), depends on the sign of the expression [1  (3/2)sin2θ]. If
angle θ is in the interval [0, θ∗], where θ∗ ¼ 0.955, then this expression is positive,
188 8 Evolution of Rotations of a Rigid Body in a Medium

and for θ > θ∗ it is negative. Therefore, for the option 2 from (8.36) in which θ
decreases as shown in Fig. 8.9, we have θ < θ∗, and angle λ increases as shown in
Fig. 8.11. More complicated is the behavior of angle λ in the variant 1 from (8.36).
Here, angle θ increases (curve 1 in Fig. 8.9) and passes through the value θ ¼ θ∗.
Accordingly, angle λ first increases on a small interval of time and then decreases (see
Fig. 8.12, where the dependence λ(τ) is presented in a larger scale).
Thus, the conducted study allows following the motion of a satellite relative to
the center of mass under the combined action of the gravitational torque and the
moment of linear resistance.

References

1. Klimov, D.M., Kosmodem’yanskaya, G.V., Chernousko, F.L.: Motion of a gyroscope with

contactless suspension. Izv. Akad. Nauk SSSR Mekh. Tverd. Tela. 2, 3–8 (1972) in Russian
2. Leshchenko, D.D.: Motion of a ponderous rigid body with a fixed point in a mildly resisting
medium. Soviet Appl. Mech. 11(3), 299–303 (1975)
3. Akulenko, L.D., Leshchenko, D.D., Chernousko, F.L.: Fast motion of a heavy rigid body about
a fixed point in a resistive medium. Mech. Solids. 17(3), 1–8 (1982)
4. Akulenko, L.D., Leshchenko, D.D., Rachinskaya, A.L.: Evolution of the satellite fast rotation
due to the gravitational torque in a dragging medium. Mech. Solids. 43(2), 173–184 (2008)
5. Akulenko, L.D., Leshchenko, D.D., Rachinskaya, A.L.: Evolution of the dynamically sym-
metric satellite fast rotation due to the gravitational torque in a dragging medium.
Mekh. Tverd. Tela. 36, 58–63 (2006) in Russian
6. Landau, L.D., Lifshitz, E.M.: Course of Theoretical Physics, Mechanics, vol. 1. Pergamon Press,
Oxford (1976)
7. Jahnke, E., Emde, F., Losch, F.: Tables of Higher Functions. McGraw-Hill, New York, NY
(1960)
8. Chernousko, F.L.: On the motion of a satellite about its center of mass under the action of
gravitational moments. J. Appl. Math. Mech. 27(3), 708–722 (1963)
9. Arnold, V.I., Kozlov, V.V., Neishtadt, A.I.: Mathematical Aspects of Classical and Celestial
Mechanics. Springer, Berlin (2007)
10. Neishtadt, A.I.: On passing through resonances in the two-frequency problem. Dokl. Akad.
Nauk SSSR. 221(2), 301–304 (1975) in Russian
11. Gradshtein, I.S., Ryzhik, I.M.: Tables of Integrals, Sums, Series and Products. Academic Press,
San Diego, CA (2000)
12. Chernousko, F.L.: Motion of a rigid body with cavities filled with viscous fluid at
small Reynolds number. USSR Comput. Math. Math. Phys. 5(6), 99–127 (1965)
13. Chernousko, F.L.: Motion of a Rigid Body with Viscous-Fluid-Filled Cavities. Computing Center
AN SSSR, Moscow (1968) in Russian
14. Chernousko, F.L.: The Movement of a Rigid Body with Cavities Containing a Viscous Fluid.
NASA, Washington, DC (1972)
15. Smirnova, E.P.: Stabilization of free rotation of an asymmetric top with cavities completely
filled with fluid. J. Appl. Math. Mech. 38(6), 931–935 (1974)
16. Martynenko, Y.G.: Motion of a Rigid Body in Electric and Magnetic Fields. Nauka, Moscow
(1988) in Russian
17. Martynenko, Y.G., Pankrat’eva, G.V.: Fast rotations of a conductive ellipsoid in the uniform
magnetic field. Proceedings of the Moscow Power Engineering Institute. Problems of mecha-
nics of controlled systems, machines and mechanisms. 77, 3–10 (1985) in Russian
References 189

18. Arnold, V.I.: Geometrical Methods in the Theory of Ordinary Differential Equations. Springer,
London (2012)
19. Neishtadt, A.I.: Passing over separatrix in the resonant problem with a slow varying parameter.
Prikl. Mat. Mekh. 39(4), 621–632 (1975) in Russian
20. Rumyantsev, V.V., Oziraner, A.S.: Stability and Stabilization of Motion with Respect to
Some of the Variables. Nauka, Moscow (1987) in Russian
21. Akulenko, L.D., Leshchenko, D.D., Rachinskaya, A.L., Zinkevich, Y.S.: Perturbed and Con-
trolled Rotations of a Rigid Body. Mechnikov Odessa Nation. Univ., Odessa (2013) in Russian
22. Beletsky, V.V.: Motion of an Artificial Satellite about its Center of Mass. Israel Program for
Scientific Translation, Jerusalem (1966)
23. Boichuk, O.P.: Stability of motion of an axially symmetric rigid body (gyroscope) on the
spherical support. Dop. Akad. Nauk Ukr. SSR. 1, 31–34 (1963) in Ukranian
24. Akulenko, L.D., Leshchenko, D.D.: Rapid rotation of a heavy gyrostat about a fixed point in a
resisting medium. Soviet Appl. Mech. 18(7), 660–665 (1982)
Chapter 9
Motion of a Rigid Body with Internal Degrees
of Freedom

In Sect. 9.1, following [1], we study the problems of motion of a free rigid body
carrying a movable mass connected with the body by an elastic coupling in the
presence of viscous friction. The cases of complete dynamic symmetry of the body
and the case of the axially dynamically symmetric body are considered.
In Sect. 9.2, following paper [2], we carry out an analysis of the motion of a
dynamically symmetric rigid body carrying a movable point mass, connected with
the body by an elastic coupling with quadratic friction.

9.1 Dynamics of a Rigid Body with a Movable

Internal Mass

9.1.1 The Case of Complete Body Symmetry

Consider first the case of complete dynamic symmetry of the body. Assume that a
rigid body D∗, in which a material point P is placed at a point O1, which is fixedly
connected to the body, has complete dynamic symmetry with respect to its center of
inertia C, i.e.,

J∗
C ¼ JE, ð9:1Þ

where E is the unit tensor, J is a scalar.

Thus, we study the perturbation of a free motion of a rigid body with a spherical
ellipsoid of inertia caused by one of its points P not being fixed absolutely. Consider
three cases of attachment of point P.
1. Let the point P be connected to O1 by means of elastic spring, and let it have
three degrees of freedom relative to the body. The equations of motion of point
P are written in the form (5.42) (see also formulas (5.43), (5.44)). Inequalities (5.45)

© Springer International Publishing AG 2017 191

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_9
192 9 Motion of a Rigid Body with Internal Degrees of Freedom

are considered fulfilled. Under the condition (9.1), from relations (5.46) to (5.51),
we successively find performing the steps outlined in paragraph 2 of Sect. 5.4,

2  2


ω_ ¼ On Ωh , r ¼ Ω iω  ðω  ρÞ þ o O Ω4 ,


g ¼ mΩ2 ω 2ω2 ρ2  ðω  ρÞ2  ρ2 ω2 ðω  ρÞ þ O Ω4 , ð9:2Þ
2 2

4 
Φ ¼ mΩ ω ðω  ρÞω  ρ þ O Ω :

The terms containing λ will be involved here only in higher approximations.

Equation (5.50) can be written with the accepted accuracy as

ω_ ¼ mρ2 Ω2 J 1 ω2 ðω  eÞω  e: ð9:3Þ

Here, e is the unit vector corresponding to vector ρ, i.e., ρ ¼ ρe. It follows from
equation (9.3) that


ω2 ¼ 2ω  ω_ ¼ 0, ðω  eÞ ¼ ω_  e ¼ 0, ð9:4Þ

i.e., vector ω preserves its magnitude and the projection on the direction of e.The
position of vector ω in a coordinate system rigidly connected with the rigid body
will be characterized by the angle θ with vector e and the angle φ between the
projection of ω on the plane perpendicular to e and a certain fixed direction in this
plane. Angle φ is measured in the positive direction—counterclockwise when
viewed from the end of vector e (see Fig. 9.1).
In these variables, we have from equation (9.3)

ω ¼ const, θ ¼ const, φ_ ¼ mρ2 Ω2 J 1 ω3 cos θ ¼ const, ð9:5Þ

which corresponds to the uniform rotation of vector ω about the unit vector e (see
Fig. 9.1). The rest points for equation (9.3) on the sphere ω ¼ const are the points
θ ¼ 0, θ ¼ π, θ ¼ π/2, i.e., the poles and the equator of the sphere. These points
correspond to the stationary rotations of the system about a fixed axis.

Fig. 9.1 Angular velocity

vector in a coordinate
system connected with the
rigid body
9.1 Dynamics of a Rigid Body with a Movable Internal Mass 193

2. Suppose that, as before, point P is attached to point O1 by means of elastic

coupling but has only one degree of freedom relative to the body and can move
along a straight line passing through point O1 in the direction of the unit vector h.
Then the radius vector of point P relative to the body has the form r ¼ ξh, where ξ is
the displacement of the point. Quite similar to formulas (9.2), we obtain (with the
same error estimates)

r ¼ ξh, ξ ¼ Ω2 ½ω  ðω  ρÞ  h ¼ Ω2 ðω  ρÞ  ðω  hÞ,

g ¼ mΩ2 ½2ωðρ  hÞ  hðω  ρÞ  ρðω  hÞξ, ð9:6Þ
Φ ¼ mΩ2 ½ω  hðω  ρÞ þ ω  ρðω  hÞξ:

Substituting relations (9.1) and (9.6) into (5.50), we obtain an equation for the
vector ω similar to (9.3):

ω_ ¼ mρ2 Ω2 J 1 ½ðω  eÞ  ðω  hÞ½ðω  eÞω  h þ ðω  hÞω  e: ð9:7Þ

Here, we again denote ρ ¼ ρe. It follows from equation (9.7) that ω  ω_ ¼ 0;

therefore, equation (9.7) has the first integral ω2 ¼ const. Using equation (9.7), it is
easy to verify the identity

ðω_  eÞ  ðω  hÞ þ ðω  eÞ  ðω_  hÞ ¼ 0,

from which it follows that equation (9.7) has another first integral

ðω  eÞ  ðω  hÞ ¼ const: ð9:8Þ

Consider the motion of the trace of vector ω on a sphere of constant unit radius,
where the center of the sphere is placed at point C, while the sphere itself is rigidly
connected to the rigid body.
Vectors e and h are fixed relative to this sphere. Without loss of generality, angle
α between them can be assumed in the range 0  α  π/2; if α > π/2, we can just
reverse the direction of the unit vector h. We choose a Cartesian coordinate system
associated with the rigid body so that its basis vectors e1, e2, e3, drawn from point C,
is equal to (see Fig. 9.2)
   
e1 ¼ h  e=h  e, e2 ¼ e3  e1 , e3 ¼ ðe þ hÞ=e þ h: ð9:9Þ

We assume so far that α > 0. Vectors e, h in this coordinate system are

represented as

e ¼ e2 sin ðα=2Þ þ e3 cos ðα=2Þ,

ð9:10Þ
h ¼ e2 sin ðα=2Þ þ e3 cos ðα=2Þ:

Denoting by ω1, ω2, ω3, the projections of vector ω on the indicated axes, we can
write the first integral (9.8) in terms of the introduced projections:
194 9 Motion of a Rigid Body with Internal Degrees of Freedom

Fig. 9.2 Motion of the

trace of the angular velocity
vector on a unit sphere

ðω  eÞ  ðω  hÞ ¼ ω21 cos 2 α þ ω22 cos 2 ðα=2Þ  ω23 sin 2 ðα=2Þ ¼ const: ð9:11Þ

The trajectories of vector ω on the introduced unit sphere are the lines of inter-
section of this sphere with a family of hyperboloids described by equation (9.11) for
different values of the constant. According to (9.11), the following equation corre-
sponds to the zero value of the integral

ðω  eÞ  ðω  hÞ ¼ ω21 cos 2 α þ ω22 cos 2 ðα=2Þ  ω23 sin 2 ðα=2Þ ¼ 0, ð9:12Þ

which describes an elliptical cone, the axis of which is directed along the axis e3.
This cone intersects the unit sphere ω2 ¼ 1 along two closed curves, encircling axis
e3, and passing through the points ω ¼  e and ω ¼  h. One of these curves is
shown in Fig. 9.2 by the dashed line. It follows from (9.12) to (9.7) that on these
curves we have ω_ ¼ 0, i.e., these curves consist of the rest points of equation (9.7).
In the polar coordinates θ, φ, introduced by the formulas
ω1 ¼ ω sin θ cos φ, ω2 ¼ ω sin θ sin φ, ω3 ¼ ω cos θ, ð9:13Þ

the equation of curves (9.13) on the unit sphere assumes the form
 
tan 2 θ cotan2 ðα=2Þ  cos 2 φ ¼ 1: ð9:14Þ

The other rest points are obtained by equating to zero the second square bracket
in (9.7), i.e.,
ω  ½hðω  eÞ þ eðω  hÞ ¼ 0: ð9:15Þ

Equation (9.15) means that either vector ω is collinear to the expression in

square brackets or this expression itself equals zero. The first possibility leads to the
equality ω ¼ C1h + C2e, where C1, C2 are constant.
9.1 Dynamics of a Rigid Body with a Movable Internal Mass 195

Substituting this representation into relation (9.15), we find


ω  ½hðω  eÞ þ eðω  hÞ ¼ ðe  hÞ½ω  ðC2 e  C1 hÞ ¼ e  h C22  C21 ¼ 0: ð9:16Þ

If α > 0, then e  h 6¼ 0, and from relation (9.16) we obtain C2 ¼  C1 and

ω ¼ C1  (h  e). Comparing the last equality with formulas (9.10), we see that the
rest points on the unit sphere will be the points e2 and e3.
The second possibility of (9.15) turning to zero yields the condition h(ω  e) ¼
 e(ω  h). Since vectors e and h are not collinear for α > 0, this condition can be
satisfied only if ω  e ¼ ω  h ¼ 0, i.e., when ω is collinear to vector e3 from (9.9).
So, we have identified all the rest points of equation (9.7). On the unit sphere,
these points lie on curves (9.14) and also at the ends of vectors e1, e2, e3.
Among the last six points, four are singular points of the center type (points e2 and
e3), and two points are saddle points (points e1). This can be established with the
help of the first integral (9.11). The pattern of the trajectories of equation (9.7) on
the unit sphere is shown in Fig. 9.2. All curves, except the separatrices passing
through the saddles e1, are closed curves enveloping either axis e2 or axis e3. On
the trajectories in Fig. 9.2, the arrows indicate the direction of motion. Note that on
curves (9.14), consisting of rest points, the direction of motion changes.
The separatrices, passing through the saddles e1 and separating the curves of
different families, are two great circles whose planes are perpendicular to vectors
e and h. To prove this, let us take a vector ω in the plane ω  e ¼ 0. Under the
condition ω  e ¼ 0, from equation (9.7), we obtain ω_  e ¼ 0, i.e., in the motion,
vector ω does not leave this plane. Similarly, we can prove that ω_  h ¼ 0, if
ω  h ¼ 0.Therefore, these great circles on the unit sphere, passing through the
singular points e1, are, indeed, integral curves of equation (9.7).
Let us consider also the particular cases α ¼ π/2 and α ¼ 0. When α ¼ π/2,
vectors e and h are orthogonal, and, therefore, the separatrix lying on the plane
ω  e ¼ 0 contains vector h, whereas the separatrix lying on the plane ω  h ¼ 0,
vector e. In this case, the separatrices coincide with curves (9.14) and consist of
rest points.
For α ¼ 0 we have e ¼ h, and equation (9.7) is simplified:

ω_ ¼ 2mρ2 Ω2 J 1 ðω  eÞ2 ðω  eÞðω  eÞ: ð9:17Þ

For equation (9.17), the properties (9.4) are satisfied. In polar coordinates, ana-
logous to (9.5), equation (9.17) yields the relations

ω ¼ const, θ ¼ const, φ_ ¼ 2mρ2 Ω2 J 1 ω3 sin 2 θ cos θ ð9:18Þ

and describes a uniform rotation of vector ω about axis e (Fig. 9.1).

3. As another example, we consider the case when point P has two degrees of
freedom relative to the attachment point O1 and connected to O1 by a solid rod of
small length l. As in case 1, the position of point P in the first approximation is
196 9 Motion of a Rigid Body with Internal Degrees of Freedom

determined by the direction of centrifugal force, and, similarly to (9.2), we obtain

for vector r:
 
r ¼ ω  ðω  ρÞlω  ðω  ρÞ1 : ð9:19Þ

Thus, this case differs from case 1 by the scalar coefficient |ω  (ω  ρ)|1 in
expression (9.19). In the derivation of equation (9.3), in the case of complete
dynamic symmetry, this coefficient is constant, since it depends on the slowly
changing variable ω (see estimates (9.2)). Therefore, we obtain similarly to (9.3)
 
ω_ ¼ mρlJ 1 ω2 ðω  eÞðω  eÞω  ðω  eÞ1 : ð9:20Þ

Here, properties (9.4) hold again. In polar coordinates, equation (9.20) yields the
expressions

ω ¼ const, θ ¼ const, φ_ ¼ mρlJ 1 ωθ ¼ const: ð9:21Þ

All relations (9.5), (9.18), and (9.21) describe a uniform rotation of the vector ω
about vector e (Fig. 9.1), but they differ from one another by the dependence of the
angular velocity φ_ of rotation on angle θ. Note that these cases do not include the
effect of energy dissipation due to viscous internal friction, because this effect is
manifested under the complete symmetry of the body in the terms of higher order
than in the general case.

9.1.2 Motion of Dynamically Symmetric Rigid Body

with a Movable Mass

Let us investigate the motion of a rigid body D∗ having axial dynamic symmetry,
and a movable point P attached by an elastic coupling to a point O1 on the axis of
symmetry. We will use the equations and assumptions of paragraph 2 of Sect. 5.4.
The origin of the Cartesian coordinate system, connected with the rigid body, is
placed at the center of inertia C of body D∗, whereas the basis vectors e1, e2, e3 of
this system are directed so that the basis vector e3 coincides with the dynamic
symmetry axis of body D∗. Then ρ ¼ ρe3, where, without loss of generality, we
assume ρ > 0 (otherwise, we can reverse the direction of the basis vector e3). In this
coordinate system, the inertia tensor JC∗ of rigid body D∗, for which point P is
superposed with O1, has the form
9.1 Dynamics of a Rigid Body with a Movable Internal Mass 197

 
 A1 0 0 
 
J C∗ ¼
 0 A1 0 , ð9:22Þ
 0 0 A3 

where A1 is the equatorial, and A3 is the axial moments of inertia. The equations of
the zero approximation (5.51) are written in the form

p_ ¼ ð1  A3 =A1 Þqr, q_ ¼ ð1  A3 =A1 Þpr, r_ ¼ 0: ð9:23Þ

Here p ¼ ω  e1, q ¼ ω  e2, r ¼ ω  e3. Using relations (9.23) for the components
of vector ω_ and ρ ¼ ρe3, we obtain from (5.47)

a ¼ ρr ðpe1 þ qe2 ÞA3 =A1  ρðp2 þ q2 Þe3 ,

ð9:24Þ
a_ ¼ ρr 2 A3 ðA1  A3 ÞA2
1 ðqe1  pe2 Þ:

Substitute relations (9.24) into equality (5.46) and then expression (5.46) into
formula (5.48) for g. After transformations using equalities (9.23), we define


 
g ¼ mρ2 Ω2 2p2 þ 2q2 þ A23 A2 1 2
1 r ðpe1 þ qe2 Þ þ A3 A1 p þ q re3 
2 2

mρ2 λΩ4 A23 ðA1  A3 ÞA3

1 r ðqe1  pe2 Þ:
3
ð9:25Þ

Finally, we substitute formula (9.25) into the right-hand side of equation (5.49),
and then again we apply equalities (9.23) for the calculation of derivatives p_ , q_ , r_ .
We obtain the desired equation (5.50) for vector ω. In the projections on axes e1, e2,
e3, this equation has the form

A1 p_ þ ðA3  A1 Þqr ¼ Lqr þ Br 4 p,

A1 q_ þ ðA1  A3 Þpr ¼ Lpr þ Br 4 q, ð9:26Þ
A3 r_ ¼ A1 A1
3 Br ðp þ q Þ,
3 2 2

where we introduce the notation

2 2 
L ¼ mρ2 Ω2 A3 A3
1 A1 p þ A1 q þ A3 r ,
2 2 2 2
ð9:27Þ
B ¼ mρ2 λΩ4 A33 ðA1  A3 ÞA4
1 :

Multiplying three equations (9.26) by A1p, A1q, and A3r, respectively, and
adding them, we see that the system has the first integral

G2 ¼ A21 p2 þ A21 q2 þ A23 r 2 ¼ const, ð9:28Þ

which expresses constancy in this approximation of the magnitude of vector G of

the angular momentum of rigid body D∗ relative to point C. We introduce angles θ,
φ, defining the orientation of vector G relative to the rigid body, according to the
equalities
198 9 Motion of a Rigid Body with Internal Degrees of Freedom

A1 p ¼ G sin θ cos φ, A1 q ¼ G sin θ sin φ, A3 r ¼ G cos θ: ð9:29Þ

Let us move in equations (9.26) to variables (9.29), taking into account the
constancy of G. Solving the resulting equations with respect to the derivatives φ_ , θ_
and substituting formulas (9.27), we obtain the equations

φ_ ¼ β cos θ, θ_ ¼ γcos 3 θ sin θ: ð9:30Þ

Here, we introduce the notation

 1 1
β ¼ A3  A1  mρ2 Ω2 A3 A3
1 G A3 A1 G ¼ const,
2
ð9:31Þ
γ ¼ mρ2 λΩ4 ðA1  A3 ÞA5 1 4
1 A3 G ¼ const:

The quantities β, γ have the dimension of angular velocity and remain constant
during the motion. Under the imposed conditions (5.45), the quantity β is of the
order of O(1), whereas γ ¼ O(λΩ4)  1. Note that in the case of complete sym-
metry, i.e., when A1 ¼ A3, equalities (9.30) and (9.31) turn into relations (9.5).
For λ ¼ 0, relations (9.30) and (9.31) assume the form
γ ¼ 0, θ ¼ const, φ_ ¼ β cos θ ¼ const: ð9:32Þ

Vector G in the coordinate system associated with the body rotates uniformly
about the symmetry axis e3.The only difference from the case of absolutely rigid
body is that the quantity β in (9.31) differs by a small quantity O(Ω2) from the
value ðA3  A1 ÞA1 1
1 A3 G that holds for the rigid body; this leads to a small change
in the angular velocity of rotation of vector G about vector e3.
Now, we move to the general case λ > 0. Integrating the second equation (9.30),
we find
 
ð1=2Þtan 2 θ þ log tan θ ¼ γt þ const: ð9:33Þ

Satisfying the initial condition θ(t0) ¼ θ0, we obtain from (9.33)



tan 2 θexp tan 2 θ ¼ tan 2 θ0 exp tan 2 θ0 exp½γ ðt  t0 Þ: ð9:34Þ

Relation (9.34) implicitly defines the dependence of angle θ on t. The left-hand

side of this relation is a monotonically increasing function of | tan θ|, whereas the
right-hand side, a monotonic function of t. Therefore, (9.34) defines a univalent
monotonic function θ(t). According to (9.31), the sign of the quantity γ is deter-
mined by the value of the difference A1  A3. Let us suppose, for definiteness, that
θ0 < π/2. Then, in view of (9.34), for A1 > A3 angle θ increases monotonically with
the increasing of t, and θ ! π/2 as t ! 1. On the other hand, if A1 < A3, then θ
decreases monotonically and approaches zero as t ! 1. In the case θ0 > π/2,
exactly the same picture will be for angle θ1 ¼ π  θ. Thus, in all cases, the axis
9.1 Dynamics of a Rigid Body with a Movable Internal Mass 199

of the angular momentum G of the rigid body in the coordinate system connected
with the body approaches the axis of the largest moment of inertia. The time
dependence of angle φ is determined by quadrature from the first equation (9.30).
The trajectories of vector G on the sphere connected with the rigid body will be
spirals coiling on the poles and equator of the sphere. The inclination angle of these
spirals with respect to circles θ ¼ const is small and is of the order of λΩ4.
Recall that the angular momentum of the entire system GC, which remains
constant in the fixed coordinate system, differs, according to equation (5.41),
from vector G by vector g, which is small under conditions (5.45) and is of the
order of O(Ω2). Therefore, the obtained results indicate that the motion of the
system in the presence of internal dissipation tends to a rotation about the axis of the
largest moment of inertia as t ! 1. This qualitative conclusion is well known (see,
e.g., [3–8]) and follows from the energy considerations. In fact, the kinetic energy
of the rigid body equals
1
 1  
T¼ A1 p2 þ A1 q2 þ A3 r 2 ¼ A1 1
1 G 1 þ ðA1  A3 ÞA3 cos θ :
2 2
ð9:35Þ
2 2

Here, we have used substitution (9.29). It follows from (9.35) that, for A1 > A3,
the minimum value T corresponds to the angle θ ¼ π/2, whereas for A1 < A3, to the
angle θ ¼ 0; this is in accord with the above result. In the case under consideration,
the presence of a movable point inside the body insignificantly changes its angular
momentum but can lead to significant energy dissipation. The above analysis
provides a quantitative picture of change of the body motion under the presence
of internal elastic and dissipative forces. The characteristic time τ of this change is
equal to

A21 Ω4
τ γ 1   :
A1  A3 mρ2 ω4 λ

Papers [9, 10] consider the motion about the center of inertia of a rigid body
which is close to a dynamically spherical one with a viscoelastic element.
The problem of motion about the center of inertia of a dynamically nonsym-
metric rigid body carrying a movable mass connected to the body by an
elastic coupling with viscous friction is studied in [11].
200 9 Motion of a Rigid Body with Internal Degrees of Freedom

9.2 Motion of a Rigid Body with a Movable Mass

Connected to the Body by an Elastic Coupling
with Quadratic Friction

Let us study the motion of a dynamically symmetric rigid body carrying a movable
point mass m, connected to the body at a certain point O1 on the axis of symmetry. It
is assumed that, in the relative motion, the point m is acted upon by the restoring
elastic force and the resistance force proportional to the squared velocity (quadratic
friction).
The origin of the Cartesian coordinate system connected to the body is placed at
the center of inertia C of the system, consisting of the body and the point mass
located at point O1. The basis vectors e1, e2, e3 of the system are directed so that the
basis vector e3 is oriented along the dynamic symmetry axis of the system. Then the
radius vector r of point O1 equals ρ ¼ ρe3, where, for definiteness, we assume ρ > 0.
In this coordinate system, the inertia tensor JC∗ is a diagonal one:
JC∗ ¼ diagðA1 ; A1 ; A3 Þ. The quantities A1 and A3 are the equatorial and axial
moments of inertia, respectively. Since p ¼ ω  e1, q ¼ ω  e2, r ¼ ω  e3, the equation
of the zero approximation of the type (5.112) can be written in scalar form:
p_ ¼ dqr, q_ ¼ dpr, r_ ¼ 0 ðd ¼ 1  A3 =A1 Þ: ð9:36Þ

To determine the right-hand side of the equation of motion of the type (5.113),
we calculate the quantities a and a_ in the expression (5.109) for r. With the help of
relations (9.36), we find

a ¼ ρr ðpe1 þ qe2 ÞA3 A1

1  ρω⊥ e3 ,
2
ð9:37Þ
¼ ρr 2 A3 A1
1=2
a_ 1 d ðqe1  pe2 Þ, ω⊥ ¼ ðp2 þ q2 Þ :

As a result, for the conventional angular momentum vector g of the movable

mass from (5.111), we obtain an explicit expression in terms of variables p, q, r with
the error of O(Ω4):


g ¼ mρ2 Ω2 2ω2⊥ þ A3 A2
1 r ðpe1 þ qe2 Þþ
2

 3 3 3  
 
þA3 A1
1 ω⊥ re3  mρ ΛΩ A3 A1 d d ω⊥ r ðqe1  pe2 Þ:
3 5
ð9:38Þ

Now, the desired expression for derivative g_ is calculated analogously using

formulas (9.36). The calculation procedure is fairly simple. It should be noted that,
by virtue of (9.36), the derivative ω_ ⊥ ¼ 0.
Now, projecting equation (5.113) on the basis vectors e1, e2, e3, we obtain the
desired equations of motion:
9.2 Motion of a Rigid Body with a Movable Mass Connected to the Body by an. . . 201

p_  dqr ¼ Sqr þ Qpr 6 , pð0Þ ¼ p0 ,

q_ þ dpr ¼ Spr þ Qqr 6 , qð0Þ ¼ q0 , ð9:39Þ
r_ ¼ QA21 A2
3 ω⊥ r ,
2 5
r ð 0Þ ¼ r 0 :

For brevity, we introduce the notation

S ¼ mρ2 Ω2 A3 A4 2

1 G , G2 ¼A21 ω2⊥ þ A23 r 2 ,
3 4 5   ð9:40Þ
Q ¼ mρ ΛΩ A3 A1 d d ω⊥ :
3

Adding equations (9.39), respectively, multiplied by A21 p, A21 q, and A23 r, we find
the first integral of motion—the modulus G ¼ |G| of the angular momentum:

G ¼ G0 ¼ const, G20 ¼ A21 ω2⊥0 þ A23 r 20 : ð9:41Þ

System (9.39) allows further integration. To find the quantity ω, we determine

the projections of vector G on the principal central axes of inertia by the formulas

A1 p ¼ G sin θ cos φ, A1 q ¼ G sin θ sin φ, A3 r ¼ G cos θ: ð9:42Þ

Since G ¼ const according to (9.41), then, differentiating relations (9.42) and

taking into account equations (9.39) and expressions (9.40) for spherical angles θ,
φ, we obtain the differential equations
 
φ_ ¼ β cos θ, θ_ ¼ γ sin θ sin θcos 5 θ;
ð9:43Þ
φðt0 Þ ¼ φ0 , θðt0 Þ ¼ θ0 :

Coefficients β and γ in (9.43) are constant and equal:

 1
β ¼  d þ mρ2 Ω2 A3 A4 2
1  G A3 G,
2   7 ð9:44Þ
γ ¼ mρ3 ΛΩ3 A6
1 A3 d d G :

In the particular cases of spherical symmetry (d ¼ 0) or ρ ¼ 0, it follows from

(9.44) that the constant γ ¼ 0, and equations (9.43) can be integrated explicitly as
θ ¼ θ0 , φ ¼ βðt  t0 Þ cos θ0 þ φ0 ; θ0 , φ0 ¼ const:

The components p, q, r of angular velocity are also calculated explicitly by

means of (9.42)
p ¼ ω⊥ cos φ, q ¼ ω⊥ sin φ, r ¼ r0 ðω⊥ ¼ ω⊥O Þ:

Let us now consider the general case γ 6¼ 0. Integration of the second equation of
(9.43) leads to the relation
202 9 Motion of a Rigid Body with Internal Degrees of Freedom



2sec 4 θ cosec θ þ 5 sec 2 θ  3 cosec θþ
 
þ3log tan ðπ=4 þ θ=2Þ ¼ 8γt þ const: ð9:45Þ

For definiteness, we assume that the value of θ0 belongs to the interval (0, π/2).
By (9.44), the sign of γ is determined by the sign of parameter d, i.e., the difference
A1  A3. It follows from (9.43) and (9.45) that, for A1 > A3(elongated body), angle θ
monotonically increases and tends to π/2 as t ! 1, whereas r ! 0. In the case
A1 < A3 (flattened body), the quantity θ decreases monotonically: θ ! 0 as t ! 1,
while φ_ ! β ¼ const. Thus, the direction of the angular momentum vector G in the
body reference system tends to a stationary state: to the directions of the axes
corresponding to the largest moments of inertia. If the dependence θ(t) is
constructed according to (9.45), then function φ(t) can be found by quadrature
from the first equation of (9.43). Dividing one equation of (9.43) by the other and
then carrying out a quadrature, we obtain the dependence φ(θ), which, together with
(9.45), implicitly gives the solution of system (9.39) using formulas (9.42).
Note that the constant τ ¼ |γ|1 has the dimension of time and characterizes the
rate of change of the rigid body motion, i.e., the rate of convergence of the angle of
nutation θ: θ ! 0 or θ ! π/2. In the case of a viscoelastic connection between the
body and the mass, a similar time constant τ determines the time interval during
which the angle of nutation in the linear approximation decreases or increases by
e times.
In the considered problem with quadratic friction, for small θ the nutation angle
tends to zero much more slowly since θ_
τ1 θ2 .The initial value problem (9.43)
in the main quadratic approximation with respect to θ(φ is arbitrary) leads to the
following expression for θ:
 
θ ¼ θðtÞ ¼ θ0 ½1  γθ0 ðt  t0 Þ1 , θ  1: ð9:46Þ

The analysis of change of angle θ in its dependence on t and parameters θ0, γ,

ψ 0 is elementary and standard. The polar angle φ ¼ φ(t) is obtained by quadrature
according to the first equation of (9.43) and is calculated in elementary functions in
the quadratic approximation with respect to θ. Note that, according to (9.46), the
relation holds
θðt  τÞ  θðtÞ ¼ γτθðt  τÞθðtÞ, γτ ¼ 1ðd≶0Þ:

In the approximation linear with respect to θ, we have

θðtÞ ¼ θ0 ¼ const:

Paper [12] considers the motion of a dynamically symmetric body which has a
spherical cavity filled with a high viscosity fluid and which carries a movable mass
attached by means of an elastic coupling with viscous or quadratic friction to a point
on the axis of symmetry.
References 203

References

1. Chernousko, F.L.: On the motion of rigid body with moving internal masses. Izv. Akad. Nauk
SSSR Mekh. Tverd. Tela. 4, 33–44 (1973) in Russian
2. Akulenko, L.D., Leshchenko, D.D.: Some problems of the motion of a solid with a
moving mass. Mech. Solids. 13(5), 24–28 (1978)
3. Chernousko, F.L.: Motion of a rigid body with cavities filled with viscous fluid at
small Reynolds number. USSR Comput. Math. Math. Phys. 5(6), 99–127 (1965)
4. Chernousko, F.L.: Motion of a Rigid Body with Viscous-Fluid-Filled Cavities. Computing Center
AN SSSR, Moscow (1968) in Russian
5. Chernousko, F.L.: The Movement of a Rigid Body with Cavities Containing a Viscous Fluid.
NASA, Washington, DC (1972)
6. Haseltine, W.R.: Passive damping of wobbling satellites: general stability theory and example.
J. Aerosp. Sci. 29(5), 543–549, 557 (1962)
7. Colombo, G.: The motion of satellite 1958 Epsilon around its center of mass. Smithsonian
Contrib. Astrophys. 6, 149–163 (1963)
8. Thomson, W.T.: Introduction to Space Dynamics. Dower, New York, NY (1986)
9. Kushpil, T.A., Leshchenko, D.D., Timoshenko, I.A.: Some problems of evolution of rotations
of a solid under the action of perturbation torques. Mekh. Tverd. Tela. 30, 119–125 (2000) in
Russian
10. Akulenko, L., Leshchenko, D., Kushpil, T., Timoshenko, I.: Problems of evolution of rotations
of a rigid body under the action of perturbing moment. Multibody Syst. Dyn. 6(1), 3–16 (2001)
11. Leshchenko, D.D.: Motion of a rigid body with movable point mass. Mech. Solids. 11(3),
33–36 (1976)
12. Leshchenko, D.D., Sallam, S.N.: Some problems on the motion of a rigid body with
internal degrees of freedom. Intern. Appl. Mech. 28(8), 524–528 (1992)
Chapter 10
Influence of the Torque Due to the Solar
Pressure upon the Motion of a Sun Satellite
Relative to Its Center of Mass

In Sect. 10.1, we describe the coordinate systems used in the sequel. We present a
phenomenological formula for the torque L due to the light pressure acting on a Sun
satellite. The equations of the perturbed motion of the satellite in the presence of the
force function are written. We note some results obtained in [1, 2] in the study of
motion of a dynamically nonsymmetric or symmetric satellite relative to its center
of mass under the action of the light pressure torque.
In Sect. 10.2, which is based on the results of [3], we consider the motion due to
the light pressure torque and relative to the center of mass of a satellite which is a
body of revolution close to a dynamically spherical one. Moreover, the coefficient
of the light pressure torque is approximated by a trigonometric polynomial of
arbitrary degree. As an example, we consider taking into account the zero and
first harmonics in the approximation of the coefficient of the light pressure torque.
In Sect. 10.3, following [4], we study the evolution of rotations of a satellite
close to a dynamically spherical one, under the action of the light pressure torque, in
the approximation of which the third and even harmonics are taken into account.
The numerical and qualitative analysis of the phase plane is carried out.

10.1 Equations of Spacecraft Rotation Under the Action

of the Torque Due to the Solar Pressure

Consider the motion of a spacecraft along an orbit around the Sun. We introduce
three Cartesian coordinate systems Ox1x2x3, Oy1y2y3, and Oz1z2z3 in accordance
with Sect. 4.6 (see Figs. 4.1 and 4.2). We place the origin of these systems at the
satellite’s center of inertia.
The relative positions of the principal central axes of inertia Ozi of the body and
the axes Oyi,i ¼ 1 , 2 , 3 are determined by the Euler angles φ, ψ, θ. The direction

© Springer International Publishing AG 2017 205

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_10
206 10 Influence of the Torque Due to the Solar Pressure upon the Motion of a Sun. . .

cosines of the system Oy1y2y3 are expressed in terms of the Euler angles φ, ψ, θ by
formulas (1.31).
We will assume that the spacecraft moves around the Sun along an elliptical
orbit and also suppose that we can neglect all torques except the light pressure one.
Suppose that the surface of the spacecraft is a surface of revolution, while the unit
0
vector k of the symmetry axis is directed along axis Oz3. For the light pressure
torque L, acting on the satellite in the form of a body of revolution, we adopt the
formula (5.115):

ac ðεs ÞR20
L¼ e r  k0 , ð10:1Þ
R2

where

R20
ac ð ε s Þ ¼ pc Sðεs Þz00 ðεs Þ: ð10:2Þ
R2

Here, R is the current distance from the Sun’s center to the center of inertia of the
satellite; R0 is a fixed value of R; er is the unit vector in the direction of the radius
0
vector R; k is the unit vector in the direction of the symmetry axis of the satellite; εs
0
is the angle between these directions, so that |er  k | ¼ sin εs; ac(εs) is the coeffi-
cient of the light pressure torque determined by the properties of the surface; S is the
area of “shadow” on the plane perpendicular to the flow; and z00 is the distance from
the center of mass to the center of pressure. We assume that ac ¼ ac(cosεs) and
approximate ac by polynomials in powers of cosεs. The light pressure torque has a
force function that depends only on the orientation of the body’s symmetry axis in
space. We represent function ac(cosεs) in the form

ac ¼ a0 þ a1 cos εs þ . . . þ aN cos N εs : ð10:3Þ

The equations of perturbed motion of the satellite in the presence of the force
function U have the form (4.60) in the variables G, δ, λ, θ, φ, ψ:
1 ∂U 1 ∂U 1 ∂U ∂U
λ_ ¼ , δ_ ¼  þ cotan δ , G_ ¼ ,
G sin δ ∂δ G sin δ ∂λ G ∂ψ ∂ψ

 1 ∂U 1 ∂U
θ_ ¼ G sin θ sin φ cos φ A1
1  A2
1
 þ cotan θ ,
G sin θ ∂φ G ∂ψ

 1 ∂U
φ_ ¼ G cos θ A1 1 1
3  A1 sin φ  A2 cos φ þ
2 2
,
 G sin θ ∂θ  ð10:4Þ

1 2  1 ∂U ∂U
ψ_ ¼ G A1 sin φ þ A12 cos φ 
2
cotan δ þ cotan θ :
G ∂δ ∂θ

To the system of equations (10.4), we must add an equation describing the time
change of the true anomaly:
10.2 Evolution of Rotation of a Spacecraft with Moments of Inertia Close to. . . 207

3=2
dv ω0 ð1 þ e cos vÞ2 2π η1=2 ð1  e2 Þ
¼ , ω0 ¼ ¼ , ð10:5Þ
dt ð1  e2 Þ3=2 Q0 p3=2

where ω0 is the average angular velocity of motion of the center of mass along the
elliptical orbit, Q0 is the period of revolution of the satellite, e and p are the
eccentricity and the focal parameter of the orbit, respectively, and η is the product
of the gravitational constant and the mass of the Sun.
In [2], using the averaging method, the rotational motion of a dynamically
nonsymmetric satellite with an axially symmetric surface relative to the center of
mass under the action of the light pressure torque is studied. The satellite moves
along an elliptical orbit around the Sun.
It is noted that the study of motion of a nonsymmetric satellite in the case
ac ¼ a1 cos εs can be carried out in the way it is done in [1] for the case of a
dynamically symmetric satellite in the gravitational field on a circular orbit. Earlier,
it was found in [1] that the equations of motion in the coordinates δ, λ for a
dynamically symmetric satellite under the action of the light pressure torque
coincide with the equations of motion under the influence of gravitational torques
if we put e ¼ 0 (circular orbit).

10.2 Evolution of Rotation of a Spacecraft with Moments

of Inertia Close to Each Other

10.2.1 Basic Assumptions and Statement of the Problem

Consider the motion of a spacecraft relative to its center of mass under the action of
the light pressure torque. We introduce three right-handed Cartesian coordinate
systems Ox1x2x3, Oy1y2y3, Oz1z2z3 (see Sects. 10.1 and 4.6, Figs. 4.1 and 4.2); the
origin of coordinates is placed at the center of inertia of the satellite. Angles δ, λ and
the Euler angles φ, ψ, θ, determining the orientation of the body reference system
Oz1z2z3 relative to the system Oy1y2y3, were introduced in Sect. 4.6.
We neglect the moments of all forces except the forces of light pressure. As in
Sect. 10.1, we assume that the surface of the spacecraft is a surface of revolution,
whereas the unit vector k0 of the symmetry axis is directed along the axis Oz3. In
this case, formula (10.1) holds for the light pressure torque L acting on the satellite.
Next, we assume that ac ¼ ac(cosεs) and represent the function ac(cosεs) in the form
(10.3). The equations of the perturbed motion of the satellite in the presence of the
force function in the variables G, δ, λ, θ, φ, ψ are of the form (10.4). The force
function U depends on time t via the true anomaly v(t) and the direction cosines α3,
β3, γ 3 of axis Oz3 relative to the coordinate system Ox1x2x3:
208 10 Influence of the Torque Due to the Solar Pressure upon the Motion of a Sun. . .

U ¼ U ðvðtÞ; α3 ; β3 ; γ 3 Þ: ð10:6Þ

Here, according to [1],

α3 ¼ sin ψ sin θ cos δ sin λ  cos ψ sin θ cos λ þ cos θ sin δ sin λ,
β3 ¼  sin ψ sin θ sin δ þ cos θ cos δ, ð10:7Þ
γ3 ¼ sin ψ sin θ cos δ cos λ þ cos ψ sin θ sin λ þ cos θ sin δ cos λ:

The equation for true anomaly (10.5) should be added to the system of equations
(10.4).
The torque (10.1) corresponds to the force function
ð
2 2
U ð cos εs Þ ¼ R0 R ac ð cos εs Þdð cos εs Þ:

Consider first the case

ac ð cos εs Þ ¼ an cos n εs : ð10:8Þ

The force function in this case has the form

an R20
U n ð cos εs Þ ¼  cos nþ1 εs , ð10:9Þ
ðn þ 1ÞR2
cos εs ¼ γ 3 cos v þ α3 sin v:

The direction cosines α3, β3, γ 3 are expressed in terms of δ, λ, θ, ψ according to

formulas (10.7).
Suppose that the principal central moments of inertia of the satellite are close to
one another and can be represented as

A1 ¼ J 0 þ εA01 , A2 ¼ J 0 þ εA02 , A3 ¼ J 0 þ εA03 , ð10:10Þ

where 0 < ε  1 is a small parameter. Assume also that a0  ε, a1  ε, ..., aN  ε,

i.e., the light pressure torque, has the same magnitude order of ε as the gyroscopic
moment. From (10.9) it follows that Un  ε. Let us study the solution of systems
(10.4) and (10.5) for a small ε on a large time interval t  ε1. The accuracy of the
averaged solution for slow variables is of the order of O(ε) on the interval of time
during which the body performs ε1 revolutions. Independent averaging over ψ, v is
carried out as in the nonresonant cases [5] (see Sect. 4.7).
10.2 Evolution of Rotation of a Spacecraft with Moments of Inertia Close to. . . 209

10.2.2 Transformation of the Expression for the Force

Function: Averaging Procedure and Construction
of the First Approximation System

Consider the unperturbed motion (ε ¼ 0), when the equations (10.4) and (10.5)
describe the motion of a spherically symmetric body and the light pressure torque
(10.1) is zero. In this case, we find from system (10.4) that λ, δ, G, θ, and φ are
constants, while ψ is determined by formula (4.57).
As shown in [5] and in Sect. 4.7, averaging of functions depending on v is
reduced to averaging over v according to (4.59).
The expression for cosεs from (10.9), taking into account relations (10.7), can be
represented as

cos εs ¼ d þ g cos υ,
d ¼ cos θ sin δ cos ðλ  vÞ, υ ¼ ψ  ζ, ð10:11Þ
 1=2
g ¼ sin 2 θ½sin 2 ðλ  vÞsin 2 δ þ cos 2 δ :

Then factor cosn + 1εs in the expression of the force function (10.9) can be
represented in the following form using Newton’s binomial formula:
X
nþ1

cos nþ1 εs ¼ ðd þ g cos υÞnþ1 ¼ k
Cnþ1 cos k υ gk dnþ1k ,
k¼0
  1=2
cos ζ ¼ sin θ sin ðλ  vÞ sin 2 θ sin 2 ðλ  vÞsin 2 δ þ cos 2 δ ,
 2  2 1=2
sin ζ ¼ sin θ cos δ cos ðλ  vÞ sin θ sin ðλ  vÞsin 2 δ þ cos 2 δ : ð10:12Þ

With the help of known expressions for the direction cosines α3, β3, γ 3 of axis
Oz3 relative to the coordinate system Ox1x2x3 (10.7), we obtain the average with
respect to ψ value of the force function. To do this, we define
2ðπ
1 X
nþ1
ðd þ g cos υÞnþ1 dυ ¼ k
Cnþ1 gk dnþ1k I k ,
2π k¼0
0
2ðπ
ð10:13Þ
1 ð2m  1Þ!!
Ik ¼ cos k υ dυ, I 2mþ1 ¼ 0, I 2m ¼ :
2π ð2mÞ!!
0

As a result of averaging Un from (10.9) over ψ, we obtain, in view of (10.13),

210 10 Influence of the Torque Due to the Solar Pressure upon the Motion of a Sun. . .

Eðnþ1
2 Þ
X
2m nþ12m  ð2m  1Þ!!
U∗ ¼ C2m
nþ1 g d ,
n
m¼0
ð2mÞ!! ð10:14Þ
an R20
U n ¼ σ n U ∗
n, σn ¼ ,
ðn þ 1ÞR2

where E(z) means the integer part of number z.

To simplify notation, we will denote the averaged expressions by the same
symbols as before averaging.
Let us perform averaging of Un with respect to v according to (4.59). Note that,
in view of the equation of motion of the center of mass along the elliptical orbit, we
have R ¼ p(1 + e cos v)1; therefore, on the basis of (4.59) and (10.14), the expres-
sion (1 + e cos v)2 in the formula for Un is canceled out.
Denote u ¼ λ  v, then d ¼ h cos u, where h ¼ cos θ sin δ; represent the expres-
sion g2m in (10.14) by means of Newton’s binomial formula in the form

m
X
m

g2m ¼ ðb þ qsin 2 uÞ ¼ Cmk sin 2k u qk bmk ,
k¼0
b ¼ sin 2 θcos 2 δ, q ¼ sin 2 θsin 2 δ:

For the second averaging of expression (10.14) with respect to variable

u ¼ λ  v, we should consider the integral of the form

2ðπ
1
m
b þ qsin 2 u ðh cos uÞn2mþ1 du ¼
2π
0

X ð
2π
m
1
¼ h nþ12m
Cmk k mk
qb sin 2k uð cos uÞnþ12m du:
k¼0
2π
0

The obtained integral is calculated explicitly in book [6]; in particular, for n ¼ 2l

it equals to zero due to oddity of the degree of the second factor in the integral.
Let n ¼ 2l + 1 be an odd number, then we have
 
nþ1
k ¼ 0, 1, . . . , m; m ¼ 0, 1, . . . , E ¼ l þ 1:
2

After averaging (10.14) with respect to u ¼ λ  v, we get

10.2 Evolution of Rotation of a Spacecraft with Moments of Inertia Close to. . . 211

U 2lþ1 ¼ σ l U ∗2lþ1 ,
lþ1 X
X m
U∗
2lþ1 ¼ Almk ð cos θÞ2ðlþ1mÞ sin 2m θ
m¼0 k¼0
 ð sin δÞ2ðlþ1mþkÞ ð cos δÞ2ðmkÞ , ð10:15Þ
3=2
a2lþ1 R20 ð1  e2 Þ
σl ¼ ,
2ðl þ 1Þp2
k ð2m  1Þ!!ð2k  1Þ!!½2ðl þ 1  mÞ  1!!
Almk ¼ C2m
2ðlþ1Þ Cm :
ð2mÞ!!½2ðk þ l þ 1  mÞ!!

The force function U for the coefficient of the light pressure torque of the form
(10.3) can be written as follows:

X
Q  
N1
U ðθ; δÞ ¼ U 2lþ1 ðθ; δÞ, Q¼E , ð10:16Þ
l¼0
2

where N is taken from the approximation (10.3).

Thus, in the first approximation, the coefficient of the light pressure torque (10.3)
is equivalent to the following expression:

X
Q
ac  e
ac ¼ a2lþ1 ð cos εs Þ2lþ1 , ð10:17Þ
l¼0

because the even harmonics of the coefficient of the light pressure torque drop out
in the averaging.
Calculating the partial derivatives ∂U/∂δ, ∂U/∂θ of the function (10.15) in view
of (10.16) and taking into account the identities ∂U/∂λ ¼ ∂U/∂ψ ¼ ∂U/∂φ  0, we
find that the averaged system of the first approximation has the form
1 ∂U ∗
λ_ ¼ 2σ l ,
G sin δ ∂δ

δ_ ¼ 0, G_ ¼ 0, θ_ ¼ G sin θ sin φ cos φ A1 1  A2 ,
1

 1 ∂U ∗
φ_ ¼ G cos θ A1  A 1
sin 2
φ  A 1
cos 2
φ  2σ l ,
3 1 2
G sin θ ∂θ
∂U ∗ XQ X lþ1 X
m
¼ Almk ð cos θÞ2ðlþ1mÞ sin 2m θð sin δÞ2ðlmþkÞþ1  ð10:18Þ
∂δ l¼0 m¼0 k¼0
 ð cos δÞ2ðmkÞ1 ½ðl þ 1Þcos 2 δ þ k  m,
∂U ∗ X Q Xlþ1 Xm
¼ Almk ð sin δÞ2ðlþ1mþkÞ ð cos δÞ2ðmkÞ 
∂θ l¼0 m¼0 k¼0
2ðlmÞþ1
 ð cos θÞ ð sin θÞ2m1 ½m  ðl þ 1Þsin 2 θ:

The main analytical result is as follows: coefficients σ l and Almk are defined in
(10.15), whereas the coefficients a2l of expansion (10.3) containing even powers
212 10 Influence of the Torque Due to the Solar Pressure upon the Motion of a Sun. . .

disappear during averaging. The angular momentum vector remains constant in

modulus and constantly inclined to the orbital plane.

10.2.3 Investigation of Equations for the Angles of Nutation

and Proper Rotation

The equations for determining the nutation angle θ and the proper rotation angle φ
(10.18) describe the motion of the angular momentum vector G relative to the body
and are reduced to the form (with respect to slow time ξ)
θ0 ¼ sin θ sin φ cos φ,
∗
1 ∂U ðθ; δ0 Þ
φ0 ¼ cos θðη1  sin 2 φÞ  2σ l β1 G2 ð sin θ Þ , ð10:19Þ
0
∂θ
γ
ξ ¼ G0 βt, η1 ¼  , β ¼ A1 1
1  A2 , γ ¼ A1 1
2  A3 :
β

Parameter σ l is defined in (10.15), whereas the function ∂U∗(θ, δ0)/∂θ is

constructed in accordance with formulas (10.18). The quantities G0, δ0 in (10.19)
are the values of G, δ at the initial moment of time. In view of relations (10.10) and
the assumption a2l + 1  ε, l ¼ 0 , . . . , Q, we find that β, γ, σ l  ε. For the system
(10.19), the following first integral exists:

c ¼ sin 2 θðη1  sin "2

φÞ  4σ l β1 G20 f ðθ; δ0 Þ,
XQ Xlþ1 Xm
m X
lm
ð1Þi ð sin θÞ2ðmþiÞ
f ðθ; δ0 Þ ¼ Almk i
Clm 
2 i¼0 mþi ð10:20Þ
l¼0 m¼0 k¼0 #
1 X
lm
ð1Þi ð sin θÞ2ðmþiþ1Þ
 ð l þ 1Þ ð sin δ0 Þ2ðlþ1mþkÞ ð cos δ0 Þ2ðmkÞ ,
2 i¼0
m þ i þ 1

which can be directly verified.

10.2.4 Taking into Account the Zero and First Harmonics

in the Approximation of the Solar Pressure Torque

Let us represent the function ac(cosεs) in the form

ac ¼ a0 þ a1 cos εs : ð10:21Þ

In this case, the equations for obtaining θ and φ, as follows from (10.19), assume
the form (the terms remain that are determined by a1)
10.2 Evolution of Rotation of a Spacecraft with Moments of Inertia Close to. . . 213

θ0 ¼ sin θ sin φ cos φ, φ0 ¼ cos θðη∗   sin φÞ,

2

∗ αγ 1

2 3=2 2 2 2 3 2 ð10:22Þ
η ¼ , α¼ 1e a1 R0 p G0 1  sin δ0 :
β 2 2

The quantities γ, β are defined by formulas (10.19).

Here, G0 is the value of G at the initial moment of time. For system (10.22), the
following first integral takes place:



c1 ¼ sin 2 θ η∗  sin 2 φ ¼ sin 2 θ0 η∗  sin 2 φ0 ¼ const: ð10:23Þ

This result can also be obtained from the expression for the first integral (10.20)
for the functions ac of the form (10.3) for n ¼ 1 (l ¼ 0).
For definiteness, we set A1 > A2 > A3, then, β < 0, γ < 0, η∗ < 0. We introduce
the variable x ¼ cos θ. After a series of transformations in view of (10.23), equation
(10.22) for the determination of θ allows separation of variables and leads to the
relation

ðx


 1=2
Cðt  t0 Þ ¼  x21  h b21  x21 dx1 ,

0
x
 ∗ ∗ ð10:24Þ
C ¼ G0 A1 1
2  A1 ½η ðη  1Þ
1=2
,
c1 c 1
h ¼ 1  ∗ , b1 ¼ 1  ∗
2
:
η η 1

Thus, the problem is reduced to quadrature: on the right in (10.24), there is an

elliptic integral. The inversion of integral (10.24) for obtaining the solution of
equations (10.22) is carried out differently depending on the parameters in the
integrand of (10.24).
From (10.24), it is easy to see that the following inequalities hold
c1
h¼1 1, h b21 1:
η∗

Let us carry out the inversion of integral (10.24) for the case h 0. In this case,
|x| b1; therefore, we make the change of variables x ¼ b1 cos ω, and integral
(10.24) is reduced to the form

ðω

1=2
τ ¼ Θðt  t∗ Þ ¼ 1  k2 sin 2 ω dω,
0 ð10:25Þ

1=2 b2
Θ ¼ C b21  h , k2 ¼ 2 1 1:
b1  h
214 10 Influence of the Torque Due to the Solar Pressure upon the Motion of a Sun. . .

Here, t∗ is a fixed moment of time. Thus, an elliptic integral of the first kind is
obtained. The inversion of this integral yields the expression

ω ¼ am τ, cos ω ¼ cn τ, cos θ ¼ b1 cn τ: ð10:26Þ

With respect to argument τ, functions cn τ, sn τ are periodic with the period

Tτ ¼ 4K(k2). The period of oscillations of angle θ with respect to time is equal to
4K ðk2 Þ
T θ ¼ Θ . Thus, the time dependence of θ is computed in a known manner by the
Jacobi elliptic functions.
To calculate the unknown φ(t), it suffices to know how the functions sinφ sin θ
and cosφ sin θ change with time. To determine these functions, we turn to the first
integral (10.23). Using this integral, we obtain (for definiteness, in (10.24) we take
the plus sign for sinθ [7])

 1=2
sin φ sin θ ¼ η∗ h  b21 dn τ,
ð10:27Þ
cos φ sin θ ¼ ð1  η∗ Þ1=2 b1 sn τ:

Thus, all the direction cosines of the angular momentum vector are periodic with
the period Tθ.
Now, we integrate equation (10.22) in the case h
0. Denote h ¼ b22 and
represent integral (10.24) in the form

bð1


 1=2
C ðt  t 0 Þ ¼ b21  x21 x21  b22 dx1 : ð10:28Þ
x

We make the change of variables x21 ¼ b21 cos 2 ω þ b22 sin 2 ω. After a series of
calculations, integral (10.28) takes the form (10.25), where we perform the changes
Θ ! b1C, t∗ ! t0, whereas the modulus k is described by the relation


k2 ¼ b21  b22 b21 < 1. The inversion of this integral can be written as follows:
ω ¼ am τ, τ ¼ b1C(t  t0). Then

x ¼ cos θ ¼ b1 dn τ: ð10:29Þ

Using the first integral (10.23), we get (up to the sign)


 1=2
sin φ sin θ ¼ η∗ b22  b21 cn τ,

2  1=2 ð10:30Þ
∗
cos φ cos θ ¼ ð1  η Þ b1  b22 sn τ:

Thus, in the first approximation of the averaging method, an analogy of the

problem being solved with the Euler–Poinsot case takes place. In the slow time τ,
the problem of motion of a rigid body close to a dynamically spherical one under
the action of light pressure forces is equivalent to that of the motion of a fictitious
10.2 Evolution of Rotation of a Spacecraft with Moments of Inertia Close to. . . 215

Fig. 10.1 Admissible

values of parameters

rigid body with arbitrary moments of inertia. This is due to the adopted approxi-
mation of function (10.21) and is the main qualitative result of the research.
Let us carry out a qualitative analysis of the phase plane of the angle variables
(θ, φ). We study system (10.22) for θ and φ with the first integral (10.23). In this
system, variables θ and φ vary in the ranges 0 θ π, 0 φ < 2π, whereas the
parameter η∗ may assume arbitrary values  1 < η∗ < + 1 (depending on the
relations between the moments of inertia). The domain D of admissible values of
parameters (η∗, c1) is shown in Fig. 10.1.
We identify three subdomains D1 , 2 , 3: subdomain D1 is defined by the inequal-
ities η∗
c1
0 and η∗
1, subdomain D2 is defined by the following relations: η∗

c1
η∗  1 and 0 η∗ 1, and, finally, for D3 we have 0
c1
η∗  1 and η∗
0. The admissible domain D of the system parameters (c1, η∗), that is,
D ¼ D1 [ D2 [ D3 is shown in Fig. 10.1.
The boundaries of subdomains D1 , 2 , 3 are singular subsets of system (10.22). In
the domains D1 and D3, the motion is the oscillations with respect to θ and
oscillations or rotation with respect to φ. The separatrix for domain D1 is given
by the relation

1
sin 2 θ ¼ ðη∗  1Þ η∗  sin 2 φ ,

whereas for domain D3 we get

1
sin 2 θ ¼ η∗ η∗  sin 2 φ 1:

In domain D2, oscillations with respect to θ and φ take place.

216 10 Influence of the Torque Due to the Solar Pressure upon the Motion of a Sun. . .

Fig. 10.2 Phase portrait of

system (10.22), case η* ¼ 0

Fig. 10.3 Phase portrait of

system (10.22), case
η* ¼ 1.7

Some typical cases of the phase portraits of system (10.22) in the plane φ, θ for
various values of η∗ are shown in Figs. 10.2–10.5. Figure 10.2 corresponds to the
case η∗ ¼ 0. The dependence of θ on φ for η∗ ¼ 1 can be obtained from the
dependence shown in Fig. 10.2 by shifting by π/2 along the φ axis. Figure 10.3
shows the phase portrait obtained with the help of the first integral (10.23) of system
(10.22) for η∗ ¼  1.7. Here, for the motions inside the separatrix, there take place
oscillations with respect to φ and θ, while outside it, rotations. For η∗ > 1, the phase
portraits are similar but shifted by π/2 along the φ axis. For η∗ !  1, the
dependencies θ(φ) become straight lines parallel to the φ axis. Figure 10.4 corre-
sponds to the value η∗ ¼ 0.95, while Fig. 10.5, to the value η∗ ¼ 0.5.
Let us investigate the evolution of the angular momentum vector. System
(10.18) for the function ac of the form (10.21) can be written as follows:
10.2 Evolution of Rotation of a Spacecraft with Moments of Inertia Close to. . . 217

Fig. 10.4 Phase portrait of

system (10.22), case
η* ¼ 0.95

Fig. 10.5 Phase portrait of

system (10.22), case
η* ¼ 0.5

 


_λ ¼ 1 1  e2 3=2 a1 R2 G1 p2 1  3sin 2 θ cos δ,
0
2 2

_δ ¼ 0, G_ ¼ 0, θ_ ¼ G sin θ sin φ cos φ A1  A1 ,

1 1  2 ð10:31Þ
φ_ ¼ cos θ G A3  A1 1 sin φ 
2 1
 A2 cos φ 

2

1
 3=2 3
 1  e2 a1 R20 G1 p2 1  sin 2 δ :
2 2

Equation (10.31) for λ in the case h 0, after substitution (10.26) and a series of
transformations, assumes the form
218 10 Influence of the Torque Due to the Solar Pressure upon the Motion of a Sun. . .

dλ

¼ d 1  3b21 cn2 τ , τ ¼ Θðt  t∗ Þ,
dτ ð10:32Þ
1
3=2 1 1=2

d ¼ 1  e2 a1 R20 G2 2
0 p A1 A2 cos δ0 ðA1  A2 Þ ðc1 Þ ,
4

where δ0 is the value of δ at the initial moment of time. After integration of equation
(10.32), we get
h   i
λ ¼ λ0 þ d  τ 1 þ 3b21 k0 =k2  3Eðg; kÞb1 =k2 :
2
ð10:33Þ

Here, E(g, k) is the incomplete elliptic integral of the second kind; k2 is the
0
squared modulus of elliptic functions; k 2 ¼ 1  k2 is the squared complementary
modulus; λ0 is the value of λ at the initial moment of time; and g ¼ am τ is the
elliptic amplitude.
For small values of k, one can use the series for function E(g, k) [8]. Substituting
them into (10.33), we get
 h  i
λ ¼ λ0 þ d  ð2Ky=π Þ 1 þ 3b21 k2 k0  E=K 
2



 1 

3b21 1  k2 =4 2 sin 2y þ k2 sin 4y þ 1 þ k2 sin y þ k2 sin 3y 
4


 

 1  k2 cos y þ k2 cos 3y 1  k2 =2 þ O k4 : ð10:34Þ

Here
h  i

1 þ 3b21 k2 k0  E=K ¼ O k2 , y ¼ πτð2K Þ1 ,
2

K, E are complete elliptic integrals of the first and second kind, respectively.
Formula (10.34) is valid for any y and small k and consists of a linear and an
oscillation term with respect to t.
For the value of modulus k close to one and small values of g, one can use the
0
corresponding decomposition for function E(g, k) from [8, 9]. Keeping the terms k 2,
we get, after substituting them into (10.33), the following expression:
    
λ ¼ λ0 þ d  τ 1 þ 3b21 k0 k2  3b21 k2 1  k0 =4 
2 2


logðð1=2Þam τ þ π=4Þ þ ð1=4Þk0 sn τ cn2 τ :
2
ð10:35Þ

For h
0, equation (10.31) for λ with substitution (10.29), after a series of
transformations, takes the form
10.2 Evolution of Rotation of a Spacecraft with Moments of Inertia Close to. . . 219

dλ

¼ dþ 1  3b21 dn2 τ , τ ¼ b1 Cðt  t0 Þ,
dτ
1
3=2 1 ð10:36Þ
dþ ¼ 1  e2 a1 R20 G2 2
0 p A1 A 2 ð A1  A 2 Þ 
4
 ½η∗ ðη∗  1  c1 Þ1=2 cos δ0 :

After integrating equation (10.36), we find


λ ¼ λ0 þ d þ τ  3b21 Eðg; kÞ : ð10:37Þ

For small k, using the expansion of function E(g, k) in powers of the quantity k2
[9], we obtain analogously to (10.34)



λ ¼ λ0 þ dþ ð2Ky=π Þ 1  3b21 E=K  ð51=8Þb21 k2 sin 2y þ O k4 : ð10:38Þ

Formula (10.38) is valid for any y and small k and contains a linear and an
oscillating term.
With k close to one and small g, using the expansion of E(g, k) [9], we obtain
analogously to (10.35)

λ ¼ λ0
( " !   #)
k0
2
þ 1 π 1 02 2
þd τ 3b21 1 log tan am τ þ þ k sn τ cn τ :
4 2 4 4
ð10:39Þ

As follows from equation (10.31) for variable λ, with the value of θ close to zero
or π, the rate of change λ_ is alternating in sign; it is negative for cosδ > 0 and
positive for cosδ < 0.
In general, variable λ may have the character of a rotational or oscillatory
motion. If θ is highly variable, then the expression 1  (3/2)sin2θ can be alternating
in sign. As a result, the value of λ can be almost constant for

Mt 1  ð3=2Þsin 2 θ ¼ 0:

An analysis shows that there exist the values of parameters η∗, θ0, φ0, for which
λ  const.
In the problem under consideration, the pattern of the motion evolution turns out
to be more complicated in comparison with the case of a symmetric satellite [1]
(A1 ¼ A2 6¼ A3), because the system contains a larger (by one) number of slow
variables.
Consider particular cases of the body motion. The value θ ¼ 0 is a stationary
point of the first equation (10.22). The equation φ for θ ¼ 0 assumes a form
allowing separation of variables. After integrating this equation, we get
220 10 Influence of the Torque Due to the Solar Pressure upon the Motion of a Sun. . .



tan φ ¼ l tan rξ þ arctan l1 tg φ0 ,
ð10:40Þ
l ¼ ½η∗ =ðη∗  1Þ1=2 , r ¼ ½η∗ ðη∗  1Þ1=2 , ξ ¼ G0 βt:

The upper and lower signs in front of r in (10.40) correspond to the cases η∗ > 1
and η∗ < 0. On the other hand, if 0 < η∗ < 1, then we have

tan φ ¼ Ξ½zexpðJξÞ  w½zexpðJξÞ þ w1 ,

h i1=2
z ¼ 1 þ ð1  η∗ Þðη∗ Þ1 tan φ0 , J ¼ 2½η∗ ð1  η∗ Þ1=2 , ð10:41Þ
h i1=2
w ¼ 1  ð1  η∗ Þðη∗ Þ1 tan φ0 , Ξ ¼ ½η∗ =ð1  η∗ Þ1=2 :

For small values of angle θ, system (10.31) is written as follows:

1
3=2
λ_ ¼  1  e2 a1 R20 G1 p2 cos δ0 ,
2
δ ¼ δ0
, G ¼ G0 , θ_ ¼ G0 β sin θ sin  φ cos φ, ð10:42Þ
φ_ ¼ G0 A1 3  A 1
1 sin 2
φ  A 1
2 cos 2
φ 
1

2 3=2 2 1 2

 1e a1 R0 G p 1  ð3=2Þsin 2 δ0 ,
2

where β is defined in (10.19).

Here, we take into account the terms of the order of θ. In the case of small θ, the
equation for the determination of φ coincides with the corresponding equation for
the case θ ¼ 0, and its solution can be represented in the form of relations (10.40),
(10.41). After integrating the equation for determining angle θ (10.42) and using
formula (10.40), we obtain the relation

1  

θ2 ¼ θ20 l2 l2 cos 2 φ0 þ sin 2 φ0 cos 2 rξ þ arctan l1 tan φ0 þ

 
þl2 sin 2 rξ þ arctan l1 tan φ0 1 : ð10:43Þ

The upper and lower signs «» in (10.43) correspond to η∗ > 1 and η∗ < 0,
respectively. If 0 < η∗ < 1, then in view of (10.41), we have for θ the expression
1=2
θ ¼ θ
0 ½Gexpð2JξÞ þ Hexp
ðJξÞ þ V  exp½ð21=2

ÞJξ, ð10:44Þ
G ¼ z 1 þ Ξ , H ¼ 2zw 1  Ξ , V ¼ w 1 þ Ξ2 :
2 2

As a result of the integration of equation (10.42) for the determination of angle λ,

we get
1
3=2
λ ¼  1  e2 a1 R20 G1 2
0 p t cos δ0 þ λ0 : ð10:45Þ
2

Note that, for a strictly dynamically symmetric satellite [1] (A1 ¼ A2 + O(ε2)),
equations (10.31) can be integrated in the following form:
10.3 Taking into Account the Third and Even Harmonics in the Approximation of. . . 221

θ ¼ θ0 , G ¼ G0 , δ ¼ δ0 , φ ¼ G0 ðα  γ Þt cos θ0 þ φ0 ,
1
3=2  ð10:46Þ
λ ¼  1  e2 a1 R20 G1
0 p
2
1  ð3=2Þsin 2 θ0 t cos δ0 þ λ0 :
2

Thus, we have studied the evolution of rotations of a satellite close to a

dynamically spherical one under the action of the light pressure torque, in the
approximation of which we take into account the zero and first harmonics; some
qualitative effects have been revealed.

10.3 Taking into Account the Third and Even Harmonics

in the Approximation of the Solar Pressure Torque

Let us represent the function ac(cosεs) from (10.3) in the form

X
Q
ac ¼ a2k cos 2k εs þ a3 cos 3 εs : ð10:47Þ
k¼0

In this case, in the slow time ξ, the equation for obtaining θ and φ, as follows
from (10.19), takes the form (the terms are left that are determined by a3)

θ0 ¼ sin θ sin φ cos φ,

0 γ
φ ¼ cos θðη2  sin 2 φ  α1 sin 2 θÞ, η2 ¼   α1 s,
β
3a3 R20


2 3=2
 ð10:48Þ
α1 ¼ 1e 8  40sin δ0 þ 35sin 4 δ0 ,
2
64G0 p β
2 2

4sin 2 δ0 ð4  5sin 2 δ0 Þ
s¼ :
8  40sin 2 δ0 þ 35sin 4 δ0

The quantities ξ, β, γ can be found from relations (10.19). In view of relation

(10.10) and the assumptions a3  ε, we conclude that β, α, γ are small quantities of
the order of ε.
For system (10.48), there holds the first integral which can be obtained directly
or from the expression of the first integral (10.20) for the general case of depen-
dence of ac, approximated by a trigonometric polynomial of arbitrary degree; for
n ¼ 3(l ¼ 1) we have

c2 ¼ sin 2 θ η2  sin 2 φ  ð1=2Þα1 sin 2 θ ¼

¼ sin 2 θ0 η2  sin 2 φ0  ð1=2Þα1 sin 2 θ0 ¼ const: ð10:49Þ

Let us conduct a qualitative analysis of the phase plane (θ, φ). We investigate the
system (10.48) for θ и φ with the first integral (10.49). In this system, variables θ
and φ vary within limits 0 θ π and 0 φ 2π, respectively, while parameter η2
222 10 Influence of the Torque Due to the Solar Pressure upon the Motion of a Sun. . .

Fig. 10.6 Domain of

admissible values of
parameters

can take arbitrary values  1 < η2 < + 1 (depending on the relations between the
moments of inertia). The domain D∗ of admissible values of parameters η2, α1 is
shown in Fig. 10.6.
Let us define stationary points of equations (10.48), equating their right-hand
sides to zero. We have:
1. cosθ ¼ 0, θ ¼  π/2, φ ¼ 0, π, π/2; these points exist in the entire plane (η2, α1).
pffiffiffiffiffi pffiffiffiffiffi
2. sinθ ¼ 0, θ ¼ 0, π, η2  sin2φ ¼ 0, 0 η2 1, φ ¼ arcsin η2 , φ ¼ arcsin η2
þπ; in this case, stationary points exist in the strip 0 η2 1.
pffiffiffiffiffiffiffiffiffiffiffiffi
3. φ ¼ 0, π, η2  α1sin2θ ¼ 0, 0 < η /α 1, sin θ ¼  η2 =α1 ,
pffiffiffiffiffiffiffiffiffiffiffi
ffi pffiffiffiffiffiffiffiffiffiffiffiffi 2 1
θ ¼ arcsin η2 =α1 , θ ¼ arcsin η2 =α1 þ π; the obtained points exist in the
interior of the angles, hatched horizontally in Fig. 10.6, i.e., under the fulfillment
of inequalities

η2 α1 ðη2 ; α1 > 0Þ, η2

α1 ðη2 ; α1 < 0Þ:
qffiffiffiffiffiffiffiffi
4. φ ¼  π/2, η2  1  α1sin2θ ¼ 0, 0 η2α1 1, θ ¼ arcsin η2 1
α1 , θ ¼ arcsin
qffiffiffiffiffiffiffiffi 1

η2 1
α1 þ π; the obtained stationary points are defined in the interior of the angles,
hatched vertically, i.e., under the fulfillment of the inequalities

η2  1 α1 ðη2  1; α1 > 0Þ, η2  1

α1 ðη2  1; α1 < 0Þ:
10.3 Taking into Account the Third and Even Harmonics in the Approximation of. . . 223

We can consider various specific cases of selection of parameters η2, α1 (see

Fig. 10.6): (a1 , 2) unhatched parts of the plane outside the corners, (b) the unhatched
parallelogram in the center, (c) the half-strip below on the left, (d) the half-strip at
the top on the right, (e) the half-strip at the top, ( f ) the half-strip below, (g1) the
corner at the top on the right, and (g2) the corner below on the left.
Below we present the phase portraits of the averaged system, constructed
numerically for the cases listed above. Note that, due to (10.49), all the phase
trajectories are symmetric with respect to the lines θ ¼ π/2 and φ ¼ π/2. Therefore,
it is sufficient to depict a quarter of the phase portrait of the system for 0 θ π/2,
0 φ π/2(in the first quadrant of the (θ, φ) plane).
The family of phase trajectories of the averaged system in the (θ, φ) plane for
η2 ¼  5, α1 ¼  2 (the case (a2)) is shown in Fig. 10.7. These graphs correspond to
the oscillations with respect to angle θ, whereas with respect to φ, either oscillations
(inside the separatrix) or rotations (outside the separatrix) occur. Also in the case
(a1) (η2 ¼ 5, α1 ¼ 2), the pattern of dependence of θ on φ is similar. In the case (a2),
in the phase plane (θ, φ), there is stationary point (π/2, π/2) of the “center” type and
point (π/2, 0) of the “saddle” type, whereas in the case (a1), point (π/2, 0) of the
“center” type and (π/2, π/2) of the “saddle” type.
For η2 ¼ 0.8, α1 ¼ 0.5 (case (b)), the phase trajectories are represented in
Fig. 10.8. In the case, the stationary points are (π/2, 0), (π/2, π/2), (0, 1.107).

Fig. 10.7 Phase

trajectories of the averaged
system, case (a2)

Fig. 10.8 Phase

trajectories of the averaged
system, case (b)
224 10 Influence of the Torque Due to the Solar Pressure upon the Motion of a Sun. . .

Fig. 10.9 Phase

trajectories of the averaged
system, case (e)

Fig. 10.10 Phase

trajectories of the averaged
system, case (g1)

For η2 ¼ 0.8, α1 ¼ 3 (case (e)), the phase curves describe oscillations with respect to
θ and oscillations or rotations with respect to φ separated by separatrices (see
Fig. 10.9). In this case, the stationary points are as follows: (π/2, π/2), (π/2, 0),
(0, 1.107), (0.543, 0).
Such phase trajectories are obtained also in the case ( f ) (η2 ¼ 0.8; α1 ¼  1.5).
Here, the stationary points are (π/2, 0), (π/2, π/2), (0, 1.107), (0.374, π/2).
For η2 ¼ 3, α1 ¼ 5 (case (g1)), the phase trajectories are shown in Fig. 10.10 and
describe oscillations with respect to θ, oscillations with respect to φ inside and
above the separatrix passing through the point θ ¼ π/2, φ ¼ 0.323, and rotations
with respect to φ below the separatrix. The stationary points are (π/2, 0), (π/2, π/2),
(0.685, π/2), (0.886, 0). In the case (g2)(η2 ¼  2, α1 ¼  5), the pattern of the
phase curves is similar. The separatrix passes through the point θ ¼ π/2, φ ¼ 0.955;
the stationary points are (π/2, 0), (π/2, π/2), (0.886, π/2), (0.685, 0).
The value θ ¼ 0 is a stationary point of the first equation (10.48). The equation
for φ for θ ¼ 0 assumes a form allowing separation of variables. After integrating
this equation, we obtain expressions (10.40) and (10.41), where η2 should be put in
place of η∗ (see (10.22)). For small θ, system (10.48) is written as follows:

θ0 ¼ θ sin φ cos φ, φ0 ¼ η2  sin 2 φ: ð10:50Þ

References 225

Here, we take into account the terms of the order of θ. The equation for the
determination of φ in the case of small θ coincides with the corresponding equation
for θ ¼ 0, and its solution can be written in the forms (10.40) and (10.41). After
integrating equation (10.50) for θ, taking into account the solution (10.40), we
obtain expressions (10.43) and (10.44) (replacing η∗ by η2).
Thus, we have studied the evolution of rotations of a satellite close to a
dynamically spherical one under the action of the light pressure torque, in the
approximation of which we take into account the third and even harmonics. It is
established that taking into account the harmonics higher than the first one greatly
complicates the analysis of the motion. The main effects of the influence of the light
pressure torque on the motion of the satellite relative to the center of mass are
revealed while taking into account the first and even harmonics.

References

1. Beletsky, V.V.: Motion of an Artificial Satellite about its Center of Mass. Israel Program for
Scientific Translation, Jerusalem (1966)
2. Leshchenko, D.D., Shamaev, A.S.: Motion of a satellite relative to the center of mass under the
action of moments of light-pressure forces. Mech. Solids. 20(1), 11–18 (1985)
3. Akulenko, L.D., Leshchenko, D.D.: Evolution of rotation of a nearly dynamically spherical
triaxial satellite under the action of light pressure torques. Mech. Solids. 31(2), 1–10 (1996)
4. Leshchenko, D.D.: Evolution of rotation of a triaxial body under the action of the torque due to
light pressure. Mech. Solids. 32(6), 12–20 (1997)
5. Chernousko, F.L.: On the motion of a satellite about its center of mass under the action of
gravitational moments. J. Appl. Math. Mech. 27(3), 708–722 (1963)
6. Timofeev, A.F.: Integration of Functions. Gostekhizdat, Moscow (1948) in Russian
7. Beletsky, V.V.: Spacecraft Attitude Motion in Gravity Field. Moscow State Univ., Moscow
(1975) in Russian
8. Zhuravsky, A.M.: Handbook of Elliptical Functions. Academic Science Press, Moscow (1941)
in Russian
9. Gradshtein, I.S., Ryzhik, I.M.: Tables of Integrals, Sums, Series and Products. Academic Press,
San Diego, CA (2000)
Chapter 11
Perturbed Motions of a Rigid Body Close
to Lagrange’s Case

In Sect. 11.1, we describe an averaging procedure for slow variables of a perturbed

motion of a rigid body, where the motion is close to Lagrange’s case in the first
approximation [1]. It turns out that a number of applied problems admit averaging
over the nutation angle θ.
Thus, in Sect. 11.2, following [1], we consider a perturbed motion close to
Lagrange’s case, taking into account the torques acting on a rigid body from the
external medium. In Sect. 11.3, we study perturbed rotational motions of a rigid
body close to the regular precession in Lagrange’s case under various degrees of
smallness of the projections of the perturbation torque vector [2].

11.1 General Properties of the Averaging Procedure over

the Lagrange Motion

In Sect. 4.8.1, we discussed perturbed motions of a rigid body close to Lagrange’s

case. The conditions for the possibility of averaging the equations of motion with
respect to the angle of nutation (4.64), (4.66), or (4.67) (together with (4.68)) were
formulated. The system of equations of the perturbed motion of a rigid body close
to Lagrange’s case is represented in the form (4.69). We use all equations, assump-
tions, and notation of Sect. 4.8.
The idea is to conduct a study of the perturbed motion in slow variables ui,
i ¼ 1 , 2 , 3, connected with G1, H, r by relations (4.62), which are more convenient
to work with. Moreover, there is no need to solve the cubic equation (3.18) with
respect to ui. We assume that variables ui, i ¼ 1 , 2 , 3, are different. Slow variables
G1, H, r can be expressed in terms of ui from (4.62) in the following way:

© Springer International Publishing AG 2017 227

F.L. Chernousko et al., Evolution of Motions of a Rigid Body About its Center of
Mass, DOI 10.1007/978-3-319-53928-7_11
228 11 Perturbed Motions of a Rigid Body Close to Lagrange’s Case

G1 ¼ ðA1 mglÞ1=2 ðu1 þ u2 þ u3 þ u1 u2 u3 þ δ1 RÞ1=2 

  δ2 sgnð1 þ
u1 u2 þu1 u3 þ u2 u3 Þ,

1 A1 A1
H ¼ mgl ðu1 þ u2 þ u3 Þ 1 þ þ ð δ 1 R  u1 u2 u3 Þ 1  ,
2 A3 A3 ð11:1Þ
r ¼ δ2 A1
3 ðA1 mglÞ
1=2
ðu1 þ u2 þ u3 þ u1 u2 u3  δ1 RÞ1=2 ,
    1=2
R ¼ 1  u21 1  u22 u23  1 ,
 2 
δ1 ¼ sgn G1  A23 r 2 , δ2 ¼ sgn r:

At the initial moment of time, quantities δ1, δ2 are determined by the initial
conditions for G1, r. If, in the process of moving, one or both of quantities
G21  A23 r 2 , r pass through zero, the sign change for δ1, δ2 is possible; to determine
these signs, we can use the original system (4.69).
We write relations (4.62) in a concise form:

Si ðu1 ; u2 ; u3 Þ ¼ Φi ðG1 ; H; r Þ, i ¼ 1, 2, 3, ð11:2Þ

where Si, Φi are known functions of their arguments (see (4.62)). Differentiating
(11.2) with respect to time and substituting expressions (4.69) in place of G_ 1 , H_ , r_ ,
we arrive at the relations:

X3
∂Si
u_ j ¼ εZ i ðu1 ; u2 ; u3 ; θÞ, i ¼ 1, 2, 3: ð11:3Þ
j¼1
∂uj

Here,

∂Φi ∂Φi ∂Φi

Zi ¼ F1 þ F2 þ F3 , i ¼ 1, 2, 3: ð11:4Þ
∂G1 ∂H ∂r

Next, we need to solve linear equations (11.3) with respect to derivatives u_ i .The
determinant D of the linear system (11.3) equals


∂Si
D ¼ det ¼ ðu1  u2 Þðu1  u3 Þðu2  u3 Þ,
∂uj

and, by the assumption, it is different from zero. The partial derivatives, appearing
in (11.4), can be represented, with the help of equalities (11.1), as functions of
variables ui only.
Thus, after solving system (11.3) with respect to derivatives, the desired system
of equations for slow variables ui assumes a form similar to (4.69):
11.1 General Properties of the Averaging Procedure over the Lagrange Motion 229

u_ i ¼ εV i ðu1 ; u2 ; u3 ; θÞ, i ¼ 1, 2, 3,
V i ¼ V i1 F∗ ∗ ∗
1 þ V i2 F2 þ V i3 F3 , ð11:5Þ
V ij ¼ V ij ðu1 ; u2 ; u3 Þ, j ¼ 1, 2, 3,

where we have

G1  A3 ru1
V 11 ¼ ,
A1 mglðu1  u2 Þðu1  u3 Þ
u21  1
V 12 ¼ , ð11:6Þ
mglðu1  u2 Þðu1  u3 Þ
 
A3 A3 G1
V 13 ¼  1 ru1  u1 þ r :
2
mglðu1  u2 Þðu1  u3 Þ A1 A1

Here, instead of G1, r, we need to substitute the appropriate formulas (11.1). The
functions V2j, V3j, j ¼ 1 , 2 , 3 can be obtained from the corresponding expressions
(11.6) for the same value of j by a cyclic permutation of the indices for variables ui.
Functions F∗ i are obtained by substituting the expressions for integrals (11.1) into Fi
from (4.69). The initial values ui0 for variables ui are calculated from the initial data
G10, H0, r0 by means of relations (4.62).
An averaging procedure for equations (11.5) for slow variables ui of the first
approximation consists in the following. We substitute into the right-hand sides of
system (11.5), the fast variable θ from expression (3.29) for the unperturbed
motion:

θ ¼ arccos u1 þ ðu2  u1 Þsn2 ðαt þ βÞ : ð11:7Þ

After substituting (11.7), the right-hand sides of system (11.5) will be periodic
functions of t with the period 2K(k)/α, where k, α are defined by relations (3.24) and
(3.26). Averaging the right-hand sides of the resulting system with respect to t, we
arrive at an averaged system of the first approximation in the slow time τ ¼ εt (the
prime denotes differentiation with respect to τ, while we keep the previous notation
for the averaged variables):
u0i ¼ U i ðu1 ; u2 ; u3 Þ, ui ð0Þ ¼ ui0 , i ¼ 1, 2, 3: ð11:8Þ

Here,

ð
2K=α
α
U i ð u1 ; u 2 ; u3 Þ ¼ V i ðu1 ; u2 ; u3 ; θðtÞÞdt, ð11:9Þ
2K ðkÞ
0

while expression (11.7) is inserted into (11.9) instead of θ ¼ θ(t).

Thus, according to the proposed method, the study of a perturbed Lagrange
motion is performed as follows. Let perturbation torques εLi satisfy (4.64) or, in
230 11 Perturbed Motions of a Rigid Body Close to Lagrange’s Case

particular, (4.66), (4.67) (together with (4.68)). We successively calculate functions

L∗ ∗
i , Fi , Vi, i ¼ 1 , 2 , 3, with the help of relations (4.64), (4.69), (11.5), and (11.6).
Then, according to (11.9), we average functions Vi using expressions (3.24), (3.26),
(3.29), and (11.7) and form the averaged system (11.8). System (11.8) is much
simpler than the initial system (4.61), since it has a lower order (three instead of
six), is autonomous, and does not contain fast oscillations. After researching and
solving system (11.8) for ui, the initial slow variables G1, H, r are reconstructed by
formulas (11.1). Fast variables φ, θ, ψ can be found with the help of (3.15), (3.16),
and (11.7). In this case, according to general theorems of the averaging method,
slow variables ui or G1, H, r are determined with an accuracy of the order of ε, while
the fast variables, with an accuracy of the order of unity on the interval of time of
the order of ε1.

11.2 Perturbed Motion of a Body Under Linear Dissipative

Torques

As an example of the developed procedure, we investigate a perturbed Lagrange

motion, while taking into account the torques acting on a rigid body from the
external medium. We assume that the perturbation torques εLi, i ¼ 1 , 2 , 3 are of the
form
L1 ¼ ap, L2 ¼ aq, L3 ¼ br, a, b > 0, ð11:10Þ

where a and b are certain constant proportionality coefficients, depending on the

properties of the medium and the shape of the body. The torques (11.10) satisfy
sufficient conditions (4.66) and (4.68), which determine the possibility of averaging
with respect to the nutation angle θ only. System (4.63) can be written as follows:

G_ 1 ¼ ε½aðp sin φ þ q cos φÞ sin θ þ br cos θ,

ð11:11Þ
H_ ¼ ε aðp2 þ q2 Þ þ br 2 , r_ ¼ εbC1 r:

Integrating the third equation (11.11), we obtain (r0 is an arbitrary initial value of
the axial velocity of rotation)
 
r ¼ r 0 exp εbA1
3 t : ð11:12Þ

According to the procedure from Sect. 11.1, we proceed to new slow variables ui
and perform averaging according to (11.9). We get the averaged system (11.8) with
respect to the slow time τ ¼ εt in the following form:
11.2 Perturbed Motion of a Body Under Linear Dissipative Torques 231


1 A3 A3  
u01¼ b r 2 u2  a r 2 u21  1 
mglðu1  u2 Þðu1  u3 Þ A1 1 A1


a   a   A3 A3 2
 2 mgl u21  1 υ þ 2 H u21  1  ba r u1 υ
A1 A1 A1 A1


 
A3 a b a 2 1 A3
 þ G1 ru1 þ 2 G1 þ ba G1 rυ ,
A1 A1 A3 A1 A1 A1
EðkÞ
υ ¼ u3  ð u3  u1 Þ : ð11:13Þ
K ðkÞ

Here, instead of G1, H, r, k, we substitute the expressions (3.24) and (11.1) for
them. The equations for u2, u3 are obtained from (11.13) by a cyclic permutation of
the indices of ui. However, under this permutation, the expression for υ, where K(k),
E(k) are complete elliptic integrals of the first and second kind, should be left
unchanged in all three equations.
The averaged system (11.13) was integrated numerically for τ  0 and various
initial conditions and parameters of the problem. Let us present the results of
calculation for three cases that correspond to the following initial data:

u10 u20 u30 θ0

0:913 0:996 1:087 5 ð11:14Þ
0 0:5 2 60 ð11:15Þ
0:992 0:985 2:992 170 ð11:16Þ

These data correspond to a top which, at the initial moment of time, was given
the angular velocity of rotation about the axis of dynamic symmetry equal to r 0
pffiffiffi
¼ 3 and was deflected by angle θ0 from the vertical. Besides, we assume that
A1 ¼ 1.5, A3 ¼ 1, a ¼ 0.125, b ¼ 0.1, mgl ¼ 0.5. Using the values ui obtained by
numerical integration, we determine G1, H, r by formulas (11.1).
Figures 11.1, 11.2 and 11.3 present the graphs of functions ui (i ¼ 1 , 2 , 3), G1, H,
r for the indicated three cases. The total energy H decreases monotonically,
asymptotically approaching the value H ¼  mgl ¼  0.5. The projection G1 of
the angular momentum vector on the vertical in the cases (11.14) and (11.15)
monotonically decreases; in the case (11.16), it monotonically increases, and in
all cases, it tends to zero. The quantities u1 and u2 decrease monotonically and tend
to 1, whereas u3 asymptotically approaches +1. In this case, as follows from
(3.29), we have cosθ !  1 (θ ! π).
Thus, under the influence of external dissipation, the rigid body tends to be the
only stable (lower) state of equilibrium. The correctness of calculation was con-
trolled by the values r, obtained by numerical data and formulas (11.1), practically
coincided with the exact solution (11.12).
Note that paper [3] studies a perturbed motion of the Lagrange top under the
influence of a dissipative torque depending on slow time.
232 11 Perturbed Motions of a Rigid Body Close to Lagrange’s Case

Fig. 11.1 Graphs of the

first integrals and roots of
the cubic polynomial in
Lagrange case for initial
conditions (11.14)

Fig. 11.2 Graphs of the

first integrals and roots of
the cubic polynomial in
Lagrange case for initial
conditions (11.15)

Fig. 11.3 Graphs of the

first integrals and roots of
the cubic polynomial in
Lagrange case for initial
conditions (11.16)
11.3 Evolution of Rotation of a Rigid Body Under Various Assumptions for the. . . 233

11.3 Evolution of Rotation of a Rigid Body Under Various

Assumptions for the Perturbation Torque

11.3.1 General Approach

In Sect. 4.8.2, we examine the perturbed motions of a rigid body close to regular
precession in Lagrange’s case. It is assumed that the angular velocity of the body is
large enough, its direction is close to the axis of dynamical symmetry of the body,
and two projections of the perturbation torque vector on the principal axes of inertia
of the body are small compared to the restoring torque, whereas the third one is of
the same order with it. A small parameter is introduced in a special way. We arrive
at system (4.81) which is more convenient for further study. We use all notation,
equations, and assumptions of Sect. 4.8.2.
Since variables L∗i , i ¼ 1 , 2 , 3, introduced by formulas (4.73), are periodic with
respect to φ with the period 2π, then, according to substitution (4.78)–(4.80),
functions Lio from (4.82) are periodic functions of α and γ with periods 2π. Then
the system (4.83) contains two rotating phases α and γ, while the corresponding
frequencies A3 A1 1
1 r and ðA3  A1 ÞA1 r are variable. When averaging the system
(4.81) or (4.83), we must distinguish between two cases: the nonresonance one
when the frequencies A3 A1 1
1 r and ðA3  A1 ÞA1 r are incommensurable and the
resonance one when these frequencies are commensurable. A very significant
feature of the system (4.83) is that the ratio of frequencies is constant:

ðA3  A1 ÞA1
1 r
1
¼ 1  A1 A1
3
A3 A1 r

and the resonance case takes place when

A3 i i
¼ ,  2, ð11:17Þ
A1 j j

where i, j are natural relatively prime numbers, whereas in the nonresonance case,
A3/A1 is an irrational number. As a consequence of (11.17), averaging of nonlinear
system (4.83), in which X does not depend on t, is equivalent to averaging of a
quasi-linear system with constant frequencies. This is achieved by introducing an
independent variable γ.
In the nonresonance case (A3/A1 6¼ i/j), the averaged system of the first approx-
imation is obtained by independently averaging the right-hand sides of system
(4.81) with respect to both fast variables α and γ. As a result, we obtain the
following equations for the slow variables:
234 11 Perturbed Motions of a Rigid Body Close to Lagrange’s Case

a_ ¼ εA1 1 1 2 2 s
1 μ1  εKA3 r b cos θ þ εKA3 r μ3 sin θ,
b_ ¼ εA1 μ2 þ εKA3 r a cos θ þ εKA2
1 1 1 2 c
3 r μ3 sin θ, ð11:18Þ
r_ ¼ εA3 μ3 , ψ_ ¼ εKA3 r , θ_ ¼ 0,
1 1 1

where we introduce notation

2ðπ 2ðπ
1  
μ1 ða; b; r; ψ; θÞ ¼ 2 L01 cos γ þ L02 sin γ dα dγ,
4π
0 0
2ðπ 2ðπ
1  
μ2 ða; b; r; ψ; θÞ ¼ L01 sin γ  L02 cos γ dα dγ,
4π 2
0 0
ð 2π
2π ð
1
μ3 ða; b; r; ψ; θÞ ¼ L03 dαdγ, ð11:19Þ
4π 2
0 0
2ðπ 2ðπ
1
μ3s ða; b; r; ψ; θÞ ¼ L03 sin α dα dγ,
4π 2
0 0
2ðπ 2ðπ
1
μ3c ða; b; r; ψ; θÞ ¼ L03 cos α dα dγ:
4π 2
0 0

Solving averaged system (11.18) for perturbation torques L0i of a specific type,
we determine the motion in the nonresonance case with the error of the order of ε on
the time interval of the order of ε1. Note that the last equation of system (11.18)
can be integrated, and it yields θ ¼ θ0.
This system is equivalent to a two-frequency system with constant frequencies,
since both frequencies are proportional to the axial component r of the angular
velocity vector. Therefore, we can substantiate the applicability of the averaging
method as it is done for a quasi-linear system. The main assertion is as follows. Let
a function X be sufficiently smooth in α and γ, and let it satisfy the Lipschitz
condition with respect to x with a constant independent of α, γ. Then on the plane of
admissible values of parameters A3 and A1, there exists a set Λ of measure zero such
that if (A3, A1) 2
= Λ, then, for the solutions of systems (4.83) and (11.18), we have the
estimate
  
xðt; εÞ  ξðεtÞ  Dε, t 2 0; Θε1 ,

where ξ(εt) is the solution of system (11.18) averaged with respect to phases α, γ,
ξ ¼ (a, b, r, ψ, θ), and D, and Θ are constants. The proof is by using Gronwall’s
lemma and a standard procedure of changing variables in the averaging method [4],
11.3 Evolution of Rotation of a Rigid Body Under Various Assumptions for the. . . 235

as well as an arithmetic lemma used to estimate the “small denominators” [5],

appearing in the construction of the indicated change of variables.
In the resonance case (11.17), system (4.83) is a single-frequency one. Indeed,
let us introduce instead of α a new slow variable: a linear combination of phases
with integer coefficients:
i i
λ ¼ α  iði  jÞ1 γ, 6¼ 1,  2, i, j > 0: ð11:20Þ
j j

System (4.83) takes the form of a standard system with a rotating phase:

x_ ¼ εX x; iði  jÞ1 γ þ λ; γ ,
 ð11:21Þ
λ_ ¼ εY x; iði  jÞ1 γ þ λ , γ_ ¼ ðA3  A1 ÞA1
1 r,

and its right-hand sides are periodic with respect to γ with the period 2|i  j|π. The
system of the first approximation is constructed by averaging the right-hand sides of
system (11.21) over the specified period of change of the argument γ. As a result,
we obtain a system of equations for slow variables:

a_ ¼ εA1 ∗ 1 1 2 2 ∗s
1 μ1  εKA3 r b cos θ þ εKA3 r μ3 sin θ,

b_ ¼ εA1 ∗ 1 1 2 2 ∗c
1 μ2 þ εKA3 r a cos θ  εKA3 r μ3 sin θ,

r_ ¼ εA1 ∗
3 μ3 , ψ_ ¼ εKA1 1
3 r , θ_ ¼ 0, λ_ ¼ εKA1 3 r
1
cos θ,
 
2π ij
ð
1  0 
μ∗
1 ða; b; r; ψ; θ; λÞ ¼   L1 cos γ þ L02 sin γ dγ,

2π i  j 
0 
2π ij
ð
1  0 
μ∗
2 ða; b; r; ψ; θ; λÞ ¼   L1 sin γ  L02 cos γ dγ,
2π i  j

0 
2π ij
ð
1 ð11:22Þ
μ∗
3 ða; b; r; ψ; θ; λÞ ¼   L03 dγ,
2π i  j
0 
2π ij
ð
1
μ∗s
3 ða; b; r; ψ; θ; λÞ ¼   L03 sin α dγ,
2π i  j
0 
2π ij
ð
1
μ∗c
3 ða; b; r; ψ; θ; λÞ ¼   L03 cos αdγ:
2π i  j
0
236 11 Perturbed Motions of a Rigid Body Close to Lagrange’s Case

It is assumed that variable α in the integrands is replaced by λ according to

(11.20). Note that the next to last equation (11.22) has a solution θ ¼ θ0.
Solving the averaged system (11.22) for the perturbation torques of a certain
kind, we determine the motion of the body in the resonance case with an error of the
order of ε on a time interval of the order of ε1. The justification can be performed
in the standard way [4, 6, 7]. Next, using the described technique, we will consider
some specific examples of the perturbed motion of a rigid body.

11.3.2 Influence of External Dissipative Torques

As the first example for the developed method, we examine a perturbed Lagrange
motion, taking into account the torques acting on the rigid body from the external
medium. We assume that the perturbation torques Li, i ¼ 1 , 2 , 3 are linearly
dissipative:

L1 ¼ εI 1 p, L2 ¼ εI 1 q, L3 ¼ εI 3 r, I 1 , I 3 > 0: ð11:23Þ

Here, I1, I3 are some constant proportionality coefficients depending on the

properties of the medium and the shape of the body.
Let us write the perturbation torques, taking into account relations (4.73) for
p and q:

L1 ¼ ε2 I 1 P, L2 ¼ ε2 I 1 Q, L3 ¼ εI 3 r: ð11:24Þ

For nonresonance case, we proceed to new slow variables a, b, r, ψ, and θ and

obtain the averaged system (11.18) of the form

a_ ¼ εI 1 A1 1 1
1 a  εKA3 r b cos θ,
b_ ¼ εI 1 A1 b þ εKA1
1 1
3 r a cos θ, ð11:25Þ
r_ ¼ εI 3 A3 r, ψ_ ¼ εKA3 r , θ_ ¼ 0:
1 1 1

Integrating the third equation (11.25), we obtain (r0 is an arbitrary initial value of
the axial velocity of rotation)
 
r ¼ r 0 exp εI 3 A1
3 t , r 0 6¼ 0: ð11:26Þ

In view of (11.26), equation (11.25) for ψ_ is integrated and yields (ψ 0 is a

constant equal to the initial value of the angle of precession at t ¼ 0)
  
ψ ¼ ψ 0 þ KI 1 1 1
3 r 0 exp εI 3 A3 t  1 : ð11:27Þ
11.3 Evolution of Rotation of a Rigid Body Under Various Assumptions for the. . . 237

Moreover, as it is clear from (11.25), the nutation angle keeps its constant value
θ ¼ θ0. Substituting expression (11.26) instead of r into the first two equations
(11.25), we obtain a completely integrable linear system of the form
 
a_ ¼ εI 1 A1 1 1 1
1 a  εKA3 r 0 expεI 3 A3 tb cos θ,
b_ ¼ εI 1 A1 1 1 1
1 b þ εKA3 r 0 exp εI 3 A3 t a cos θ,

the solution of which can be written as follows:

 
a ¼ expεI 1 A1 1 1
1 tP0 cos η þ Q0 sin η  KA3 r 0 sin θ 0 sin ðη þ φ0 Þ ,
b ¼ exp εI 1 A1
1 t  P0 sin
 η  Q0 cos η þ KA1 1
3 r 0 sin θ 0 sin ðη þ φ0 Þ ,
1
η ¼ KI 1 1
3 r 0 cos θ 0 exp εI 3 A3 t  1 :
ð11:28Þ

As a result of substitution of expressions for a, b from (11.28) and r from (11.26)

into relations (4.78) and (4.73) for P, Q, p, and q, we find
 
p ¼ exp εI 1 A1 1 t p0 cos ðγ  ηÞ  q 0 sin ðγ  ηÞþ
þ kA1 1
3 r 0 sin  θ0 sin ðγ  η  φ0 Þ þ
1 1 1

þ kA
 3 0 1 εI 3 A3 t sin θ0 sin φ,
r exp
q ¼ exp εI 1 A1 t p0 sin ðγ  ηÞ þ q0 cos ðγ  ηÞ

 kA1 1
θ0 cos ðγ  η  φ0 Þ þ ð11:29Þ
3 r 0 sin  
1 1 1
þ kA3 r 0 exp εI 3 A3 t sin θ0 cos φ,
A3 ðA3  A1 Þr 0   
γ ¼ 1  exp εI 3 A1
3 t ,
I 3 A1 ε
p0 ¼ εP0 , q0 ¼ εQ0 :

Thus, the solution for the system of first approximation for slow variables in the
case of dissipative torque (11.23) is constructed. Let us note some qualitative
features of the motion in this case. The modulus of the axial velocity of rotation
r monotonically and exponentially decreases according to (11.26). The increment
ψ  ψ 0 of the precession angle slowly exponentially increases according to (11.27).
It follows from (11.28) that slow variables a, b monotonically and exponentially
approach zero.
According to (11.29), the summands of the projections p, q due to the initial
values p0, q0 decay exponentially. At the same time, the projections p, q include
exponentially increasing terms proportional to the restoring torque k which leads to
exponential increase of the quantity ( p2 + q2)1/2.
If resonance relation (11.17) holds, then the averaging should be carried out
according to the scheme (11.22). In this case, all the integrals μ∗ i from (11.22)
coincide with the corresponding integrals μi from (11.19). Therefore, in fact, there
is no resonance in this case, and the obtained solution is suitable for describing the
motion for any value of the ratio A3/A1 6¼ 1.
238 11 Perturbed Motions of a Rigid Body Close to Lagrange’s Case

Note that we can similarly investigate a more general than (11.23) case of a
linear dependence of dissipative torques on the angular velocity of rotation, namely,
L ¼  εIω. Here, I is a tensor defined by the matrix
 
 I 1 εI 12 εI 13 
 
 εI 21 I 2 εI 23 ,
 
 εI 31 εI 32 I 3 

in which the off-diagonal elements are small in comparison with the diagonal ones.

11.3.3 Action of a Small Constant Torque, Applied Along

the Symmetry Axis

Consider a motion of a rigid body in Lagrange’s case under the action of a small
torque which is constant in the body-connected axes and applied along the axis of
symmetry. In this case, the perturbation torques Li, i ¼ 1 , 2 , 3 have the form

L1 ¼ L2 ¼ 0, L3 ¼ εL∗
3 ¼ const: ð11:30Þ

Turning to new slow variables a, b, r, ψ, θ, we obtain in the nonresonance case

the averaged system of the type (11.18):

a_ ¼ εKA1 1
3 r b cos θ, b_ ¼ εKA1 1
3 r a cos θ, ð11:31Þ
r_ ¼ εA3 L3 , ψ_ ¼ εKA3 r , θ_ ¼ 0:
1 ∗ 1 1

Integrating the third equation (11.31), we get

r ¼ r 0 þ εA1 ∗
3 L3 t: ð11:32Þ

Substitute (11.32) into (11.31) and integrate the equation for ψ:

 1  
ψ ¼ ψ 0 þ K L∗
3 log1 þ εA1 ∗ 1 
3 L3 r 0 t : ð11:33Þ

Here, ψ 0 and r0 are arbitrary initial values of the precession angle and the axial
velocity of rotation.
As follows from (11.31), the angle of nutation θ does not change during the
motion of the body: θ ¼ θ0.
The solution of the system of the first two equations (11.31), after substituting
expression (11.32) instead of r, is written as follows:
11.3 Evolution of Rotation of a Rigid Body Under Various Assumptions for the. . . 239

a ¼ P0 cos β þ Q0 sin β  KA1 1

3 r 0 sin θ 0 sin ðβ þ φ0 Þ,
1 1
b ¼ P0 sin β  Q0 cos β þ KA3 r 0 sin θ0 cos ðβ þ φ0 Þ, ð11:34Þ
 1  
β ¼ K L∗ 3 cos θ0 log1 þ εA1 1 ∗ 
3 r 0 L3 t :

Substituting the obtained expression for a, b from (11.34) and r from (11.31) into
formulas (4.78) and (4.73), we find

p ¼ p0 cos ðγ  βÞ  q0 sin ðγ  βÞþ

þ kA1 1
3 r 0 sin θ 0 sin ðγ  β  φ0 Þþ
1
 
∗ 1
þ kA3 r 0 þ εA1 3 L3 t sin θ0 sin φ,
q ¼ p0 sin ðγ  βÞ þ q0 cos ðγ  βÞ
ð11:35Þ
 kA1 1
3 r 0 sin θ 0 cos ðγ  β  φ0 Þþ
 
∗ 1
þ kA13 r0 þ
εA13 L3 t sin θ0 
cos φ,
1 1 1 ∗ 2
γ ¼ ðA3  A1 ÞA1 εA L t þ r 0 t , p0 ¼ εP0 , q0 ¼ εQ0 :
2 3 3

According to (11.32), the quantity |r(τ)|, τ ¼ εt, increases if parameters r0, L∗

3
have the same sign and decreases if the signs are different. The precession angle ψ
(11.33) contains a variable component, the modulus of which in both cases
it is bounded for finite τ  1; in the
increases monotonically: in the first case,
second, it tends to infinity as τ !  A3 r 0
L∗
, while r ! 0.
3

Variable β in (11.34) and (11.35) changes similarly to ψ if θ0 6¼ 12 π; it has the

  1
meaning of the oscillation phase. The oscillation frequency dβ dt r 0 . Slow vari-
ables a, b are bounded 2π-periodic functions of β. According to (11.35), the
components p, q of the angular velocity vector contain bounded oscillating terms
resulted from nonzero initial data p0, q0, and also a similar term due to the restoring
torque (4.71). The oscillation frequency is determined by the derivative of variable
(γ  β) which has the meaning of phase.

11.3.4 The Case of a Body Close to the Dynamically

Symmetric One

Let us briefly consider the case of a heavy rigid body, whose ellipsoid of inertia with
respect to point O is close to an ellipsoid of revolution, so that its principal moments
of inertia are of the form

A1 ¼ A0 ð1 þ εσ 1 Þ, A2 ¼ A0 ð1 þ εσ 2 Þ, A3 6¼ A0 :
240 11 Perturbed Motions of a Rigid Body Close to Lagrange’s Case

Here, σ 1 and σ 2 are dimensionless constants of the order of unity; A0 is a

characteristic magnitude of the moments of inertia. In addition, the body’s center
of gravity may be displaced relative to point O∗ lying on the principal axis of
inertia, with respect to which the moment is A3, by a quantity of the order of ε. In
this case, the problem of motion of a heavy rigid body can be reduced to the one
considered above by introducing auxiliary perturbation torques satisfying condition
(4.73). It turns out that, in this case, L3  ε2, so L∗ ∘
3 ¼ L3 ¼ 0. Following (11.19) and
(11.22), we get

μ3 ¼ μ3s ¼ μ3c ¼ 0, μ∗ ∗s ∗c
3 ¼ μ3 ¼ μ3 ¼ 0:

Thus, the last three equations (11.18) take the form

r_ ¼ 0, ψ_ ¼ εKA1 1
3 r , θ_ ¼ 0:

In the considered approximation, the kinematic Euler’s equations are not

perturbed, and the motion of the body is a regular precession.
Note that, as follows from the equations of the first approximation (11.18) and
(11.22), if there are several perturbation torques of the form (4.73), the results of
their actions are added, and the corresponding to these perturbation integrals
(11.19) and (11.22) is represented as the sum of integrals for individual
perturbations.
In [8, 9], perturbed rotational motions of a rigid body close to regular precession
in Lagrange’s case under the action of a restoring torque depending on slow time,
the nutation angle and also a perturbation torque slowly changing over time are
considered. In paper [8], the body is rapidly spun, while the restoring and pertur-
bation torques are assumed to be small, with a certain hierarchy of smallness of the
components.
In [9, 10], the body is also assumed to be rapidly spun, while the projections of
the perturbation torque on the principal axes of inertia of the body are small
compared with the restoring torque.

References

1. Akulenko, L.D., Leshchenko, D.D., Chernousko, F.L.: Perturbed motions of a rigid body, close
to the Lagrange case. J. Appl. Math. Mech. 43(5), 829–837 (1979)
2. Akulenko, L.D., Leshchenko, D.D., Chernousko, F.L.: Perturbed motions of a rigid body that
are close to regular precession. Mech. Solids. 21(5), 1–8 (1986)
3. Kozachenko, T.A., Leshchenko, D.D., Rachinskaya, A.L.: Perturbed rotation of Lagrange top
under the action of nonstationary dissipative torques. Vestn. Odes. Nats. Univ. Mat. Mekh. 16
(16), 152–157 (2011) in Russian
4. Volosov, V.M., Morgunov, B.I.: The Averaging Method in the Theory of Non-linear Oscilla-
tory Systems. Moscow State Univ., Moscow (1971) in Russian
References 241

5. Arnold, V.I.: Geometrical Methods in the Theory of Ordinary Differential Equations. Springer,
London (2012)
6. Bogolubov, N.N., Mitropolsky, Y.A.: Asymptotic Methods in the Theory of Nonlinear Oscil-
lations. Gordon and Breach Science, New York, NY (1961)
7. Mitropolsky, Y.A.: The Method of Averaging in Nonlinear Mechanics. Naukova Dumka, Kiev
(1971) in Russian
8. Akulenko, L.D., Kozachenko, T.A., Leshchenko, D.D.: Evolution of rotations of a rigid body
under the action of restoring and control moments. J. Comput. Syst. Sci. Int. 41(5), 868–874
(2002)
9. Akulenko, L.D., Kozachenko, T.A., Leshchenko, D.D.: Perturbed rotational motions of a rigid
body under the action of nonstationary restoring moment. Mekh. Tverd. Tela. 32, 77–84
(2002) in Russian
10. Leshchenko, D.D., Shamaev, A.S.: Perturbed rotational motions of a rigid body that are close
to regular precession in the Lagrange case. Mech. Solids. 22(6), 6–15 (1987)