INTERNATIONALJOURNAL FOR NUMERICAL METHODS IN ENGINEERING,VOL.

32, 1657-1668 (1991)

COMPUTER METHODS I N FLEXIBLE MULTIBODY

DYNAMICS

RONALD L. HUSTON
Department of Mechanical, Industrial and Nuclear Engineering, University of Cincinnati, Cincinnati, Ohio 45221-0072,
U.S.A.

INTRODUCTION
One of the most difficult problems in modern mechanics is the study of the non-linear dynamics of
large flexible systems-particularly, flexible systems undergoing rapid gross motion (for example,
rapid rotation). In spite of numerous studies and extensive research efforts, there are still many
unresolved issues. Perhaps the most difficult of these is determining the effect of the gross ‘rigid-
body’ motion upon the movement permitted by the flexibility.
Analysts are in disagreement about which are the best approaches to this problem. There are
disagreements about the consistency of various approximation methods. There are disagreements
about which dynamical procedures are the most effective and most efficient. There are even
disagreements about the interpretation and application of the results of the analyses.
References 1-56 outline a sampling of the numerous studies dedicated to understanding the
phenomena. In spite of many differences, these references provide a basis for the analysis
discussed in the following sections.
It is believed by this investigator that non-linear dynamics in large flexible systems can give rise
to chaotic effects where a small change in initial movement and configuration can produce
significant changes in the motion at subsequent times.
In the sequel we outline a method of analysis and computation which can account for these
non-linear and chaotic effects. The method is based upon Kane’s equation^^*-^^ and procedures
developed by this writer and others during the past decade.21-23*44-47 It is a lumped parameter
method with the flexibility and damping modelled at the joints. It uses relative co-ordinates, Euler
parameters and generalized speeds.45*52
The balance of the paper is divided into seven sections with the following sections providing a
summary of definitions and procedures useful in the analysis. The next two sections discuss the
system modelling, the kinematics, the kinetics and the equations of motion. The incorporation of
flexibility effects with application to long, slender members is discussed in the subsequent two
sections. The final sections present some examples and concluding remarks.

PRELIMINARY NOTATIONS AND RESULTS

Flexible multibody systems

A multibody system is simply a collection of bodies with a given connection configuration. The
bodies of the system may be either rigid or flexible; they may be physically connected (for

oO29-598 1/91/161657-12$06.oO Received 13 December 1990

0 1991 by John Wiley & Sons, Ltd. Revised 6 June 1991
 1658 R. L. HUSTON

Figure I. A flexible multibody system

Figure 2. A multibody model of a flexible structural member

example, with spherical or pin joints) or separated (for example, with spring connections); and
they may form closed loops or they may be open.
Of all these options, the one most difficult to analyse is the flexibility. Indeed, more assump-
tions and more approximations are needed to study the dynamics of flexible systems than all the
others together.
Flexibility effects in multibody systems are important if the system is large and massive; if it
contains long slender numbers; or if there is rapid movement of the system. Alternatively,
multibody systems may be used to model flexible members of rapidly moving physical systems.
Figures 1 and 2 depict a flexible multibody system and a multibody system model of a structural
system member.

Transformation matrices
A major difficulty in dynamic analyses of multibody systems is accounting for the geometry
and movement of the individual bodies of the system. The orientation of the bodies may be
described by transformation matrices defined as follows. Consider a typical pair of adjoining
bodies of the system as depicted in Figure 3. The orientation of Bk relative to Bj may be defined in
terms of the relative inclinations of the unit vector sets nij and nkm.Let SJK be a matrix whose
elements are defined as
SJK,, = nji * nkm (1)
Then SJK is seen to be an orthogonal transformation matrix and the unit vectors nji and nki may
 COMPUTER METHODS IN FLEXIBLE MULTIBODY DYNAMICS 1659

Figure 3. Two typical adjoining bodies

be related to each other through the equation49

nji = SJKimnkm (2)
Regarding notation, the J and K in SJK and the first subscripts on the unit vectors refer to the
bodies Bj and B,. Repeated indices, such as the m in equation (l),represent a sum over the range
of the index.
Finally, the orientation of a typical body Bk relative to an inertial reference frame R may be
defined in terms of a transformation matrix SOK where 0 refers to R. Then SOK may be
expressed as4’
SOK = SO1 S12. . . SJK (3)
where the matrix multiplications are determined by the numbering of the bodies through the
system branch containing B,.

Euler parameters
From a computational prospective, analysts have found that it is convenient to describe the
relative orientation of adjoining bodies in terms of Euler parameters, as opposed to orientation
angles.45*48*49 An advantage of Euler parameters is that they are linearly related to the relative
angular velocity components. (Orientation angles have a non-linear (trigonometric) relationship
with the angular velocity components. This non-linearity produces singularities, and hence
computational difficulties, in some configurations of the bodies.) The linear relationship of Euler
parameters to the relative angular velocity components is a computational advantage since these
components are often identified with the ‘generalized speeds’ in Kane’s m e t h ~ d . ~ Al -dis- ~~
advantage of Euler parameters is that four are required for the orientation of each body, whereas
only three orientation angles are needed. Finally, if the movement of the bodies is restricted to a
plane, only one orientation angle is needed for each body and the singularity can be avoided.
Thus, for planar motion Euler parameters are not needed.
Euler parameters may be introduced in a number of ways. For a multibody system it is
convenient to consider two typical adjoining bodies Bj and B,. Let Bk have a given general
orientation relative to Bj. Then B, can be brought into this orientation, from any other
orientation, by a single rotation about an appropriate a ~ i s . ~If*hk* is~ a~ unit vector along this
axis and if 0, is the rotation angle, the Euler parameters are defined as
&ki = Aki sin(8k/2) i = 1, 2, 3, &k4 = cos(8k/2) (4)
where the &i are the nji components of hk.
1660 R. L. HUSTON

The &ki are not independent: from equation (4)they are seen to be related by the expression
&il + + +
&i2E i 3 &f4 = 1 (5)
The transformation matrix SJK may be expressed in terms of the &ki as48349

Finally, the time derivatives of the Euler parameters, &ki,may be expressed in terms of the nji
components, &k,, of the angular velocity of Bk relative to Bj as45*49

Diferentiation algorithms
A feature of Kane’s method is that it employs vectors and vector projections. Hence, for
computational purposes it is useful to have algorithms for vector differentiation. Such algorithms
are readily developed from the fundamental concept of differentiating a vector fixed in a rigid
body. Specifically, let r be a vector fixed in a typical body Bk. Then the derivative of r in a
reference frame R in which Bk moves
dr/dt = mk x r (8)
where mk is the angular velocity of Bk in R.
Equation (8) forms the basis for algorithm development. It shows that differentiation can be
accomplished by a multiplication-a useful computational procedure.
Equation (8) can also be used to develop differentiation algorithms for scalars which are the
result of vector operations. For example, for the elements of the transformation matrix SOK we
have47
d(SOKmn)/dt= d(nom*non)/dt
= nom*d(nkn)/dt

= nom*mkx nkn= nom*cokrnor

x SOKPnnod (9)
= emnp O k r soKpn
where the ( P = 1, 2, 3) are the no, components of o k and the emrpare the elements of the
permutation matrix.
In matrix form equation (9) can be written as
d(SOK)/dt = WOK SOK (10)
where WOK is the 3 x 3 array with elements
= enwpwkr (1 1)
 COMPUTER METHODS IN FLEXIBLE MULTIBODY DYNAMICS 1661

SYSTEM MODELLING AND KINEMATICS

Flexible system modelling, generalized co-ordinates, general speeds

To model a flexible system with a lumped parameter system it is necessary to incorporate
springs and dampers at the joints. This means that there needs to be additional degrees of freedom
(in general, both translation and rotation) introduced at the joints to accommodate the move-
ment of the springs and dampers.
Consider again two typical adjoining bodies as in Figure 3. Let Qk and 0, be connection points
of the bodies. That is, Qk and 0, coincide where the bodies are not separated. Let 5, measure the
separation of the bodies as the displacement of 0, relative to Qk.Let t,k be expressed in terms of
the nji as skinji.
If there are N bodies of the system, 6N co-ordinates xl (1 = 1,. . . , 6 N ) are needed to describe
the configuration and motion of the system. Let 6 N parameters y, (1 = 1,. . . ,6N)be introduced
as linear combinations of the derivatives of the x,.Specifically, let the first 3N of the y, be defined
as
Y 3 ( k - l ) + i = &,i i 1, 2, 3; k = 1,. . . ,N (12)

Let the remaining 3N y, be defined as

yJ(N+k-l)+i= i k i

where &ki and &&are nki components of the relative angular velocity and relative displacement
vectors.
The y, are called 'generalized speeds'. They are not always integrable in closed form-
particularly those of equation (12). That is, in general, there do not exist 'orientation functions'
whose derivatives are the &ki. Hence, in computational analyses it is convenient to use equations
(7), with the Euler parameters, to define the relative orientation of the bodies.

Angular velocities and accelerations

The angular velocity u), of a typical body B, of the system, in a reference frame R, may be
expressed using the addition f ~ r m u l a ~asl . ~ ~
mk=c;)1 + " ' +d, (14)
where the terms on the right are relative angular velocities of the bodies of the system from a
reference body B , through the branches to Bk. By using equation (14) may be expressed in the
form4',
"k = wkZmYZnOm (15)
where the u,,, ( k = 1,. . . ,N ; 1 = 1,. . . , 6N; m = 1, 2, 3) form a block array of coefficients
representing scalar components of the 'partial angular velocity vectors' used by Kane and co-
worker~.~~-~~
By differentiating, the angular accelerations of the bodies in R may be expressed as

Explicit expressions and efficient algorithms for computing the mklm and hklm
arrays are
recorded in References 45,50 and 51.
1662 R. L. HUSTON

Mass centre velocities and accelerations

Let G, be the mass centre of typical body B,. A position vector P k locating G, relative to a fixed
point in an inertial reference frame R can be constructed in terms of vectors fixed in the bodies of
the branches containing Bk and, in terms of the displacement vectors. Then differentiation of p,
produces the velocity and acceleration of G,. These derivatives may be evaluated using equations
(8) and (10).
The results of these operations show that the velocity and accelerations can be written in the
forms45.
vk = Uklm Y l (17)

where t?ktm, like the O k l m , the ( k = 1,. . . , N ; 1 = 1,. . . ,6N; m = 1, 2, 3) form a block array of
coefficients representing scalar components of the partial velocity vectors used by Kane and co-
w o r k e r ~ . ~ ’As
- ~with
~ the w ,,, and
,,&t efftcient algorithms have been written and recorded in
References 45, 50 and 51.

KINETICS AND EQUATIONS O F MOTION

Kane’s equations are ideally suited for obtaining the governing equation of motion of multibody
systems-particularly for lumped parameter models of flexible systems. Kane’s equations are
based upon a balance of generalized forces. The generalized forces in turn are obtained as
projections of forces and moments onto the partial velocity and partial angular velocity vectors.
Let the applied forces on typical body B, be equivalent to a force F, passing through mass
centre G, together with a couple with torque M,. Then the generalized applied (or ‘active’) force
on B,, associated with the generalized speed y,, is defined as
Fk = vklm Fkm + O k l mMkm (19)
where Fkmand M,, are the no, components of Fk and Mk and, as before, there is a sum over the
repeated indices.
Similarly, let the inertia forces on B, be equivalent to a force F: passing through mass centre Gk
together with a couple with torque M:. Then, analogous to equation (19), the generalized inertia
force on B,, associated with the generalized speed y , , is defined as
1* = ukIm :rn + O k f mM:m (20)
From the principles of classical mechanics FZ and MZ may be written as41,43
FZ =-mkmak and MZ = - I k ’ b l k - U ) k X ( I k * U ) k ) (no Sum) (21)
where m, is the mass of B, and 1, is its central inertia dyadic.
Kane’s equations assert that the sum of the generalized forces is zero. That is,
Fl + F? = 0 1 = 1,. . . , 6N (22)
By substituting from equations (15), (16), (18), (19), (20) and (21) into (22) the equations of
motion take the form
a@3, = A (I, P = 1,. . ., 6N) (23)
 COMPUTER METHODS IN FLEXIBLE MULTIBODY DYNAMICS 1663

where alpand fl are

(24)
and

+ %m Iksn O k f mO k q r O k p n Y p Y q ) (25)
Equations (23), (7), (12) and (13) form a set of 13N first order differential equations for the 6 N y p ,
the 3 N t k iand the ~ N E , Since,. the coefficients of equations (23) depend upon the four block arrays
O k l m , Ci)kgm, Vklm and Ijklm, and since efficient algorithms have been written for the computation of
these arrays, algorithms can thus be written for the numerical development and solution of the
governing equations. One set of such algorithms is found in the program DYNOCOMBS.”

FLEXIBILITY EFFECTS
When a flexible multibody system model is subjected to forces, either directly (‘actively’) through
externally applied forces or indirectly (‘passively’) through inertia forces, the springs and dampers
at the joints will develop forces and moments between adjoining bodies. That is, the forces on the
system cause relative movement of the bodies which in turn generates spring and damper forces.
The spring and damper forces enter the governing equations through the generalized active forces
Ff of equation (25).
When the generalized active forces are defined with the generalized speeds being relative
angular velocity components and relative displacement derivatives, the spring and damper force
and moment components occur singly in the governing equations. That is, the equations are
uncoupled in the spring and damper force and moment components.
To see this, consider again two typical adjoining bodies as in Figure 3. If tkmeasures the
displacement of 0, relative to Q k ,then the velocity of O k in an inertial frame R may be expressed
as 41.43

VOk = VQk + wj x + gj skinji (26)

where VQk is the velocity of Qk in R. If t jis small or if Qk is that point of Bj (or Bj extended) which
coincides with O k , then equation (26) reduces to
vOk = vQk + {kinji (27)
Similarly, the angular velocities of the bodies are related by the expression
a k = mj + &k = mj + hkinji (28)
Let the force system exerted through the springs and dampers at the connection joints be
represented as follows. Let the forces exerted on Bj by B, be equivalent to a single force Qij passing
through Qk together with a couple with torque mkij.Then by the law of acti~n-reaction,’~ for
small t k (or an extended Bj), the force system on B k by B j is equivalent to a single force -fkij
together with a couple with torque ,mkij.
Consider now the contribution Fl to a typical generalized active force Ff from the forces
transmitted across the joint between Bj and Bk:From equation (19) Ft may be expressed as22*23
= (avQk/ayl)’(fk[j) + (awj/ayl)*(mk/j)
+ (dvOk/ayf)’( - fk/j) + ( d o k / d y f ) * ( - mk/j) (29)
 1664 R. L. HUSTON

Consider the following cases:

Case I: yj is not equal to either &,i or ojki. Then from equations (27) and (28) the partial
velocities and partial angular velocities of equation (29) are respectively equal. That is,
av,,/ay, = av,,/ay, and aaj/ay, = aa,/ay, (30)
Hence, from equation (29), 2,is zero.
Case 2: y, is equal to one of the Then the equations (27) and (28) the partial velocities and
&is

partial angular velocities are

and

Hence, from equation (29), 2,is

where& is the nji component of bij.

Case 3: yI is equal to one of the &kj. Then from equations (27) and (28) the partial velocities and
partial angular velocities are

and
ao,/ay, = a a k / d o j k i = nji (35)
Hence, from equation (29), 2, is
*
F 1 --- m . . 31 (34)
where F~~is the nji component of m,,,.
The F, of equations (33) and (36) are to be inserted in the generalized forces, F,, of equation (25).
It is seen that there is a one-to-one correspondence between the joint force and moment
components and individual Fi. Those joint force and moment components are thus the contribu-
tions to the governing equations by the springs and dampers which simulate the flexibility in the
lumped parameter model.

APPLICATION: LONG, SLENDER BODIES

To illustrate the simplicity of the method consider a model of a beam as in Figure 4. That is, let
the beam be divided into rigid segements with the flexibility represented by springs at the segment
ends. These springs can simultaneously represent the extensional, flexural and torsional charac-
teristics of the beam.
Consider two typical adjoining segments Bj and Bk as in Figure 5. Let G, and G, be the mass
centres o f B jand_B, and let R, and R, be reference frames fixed in Bj and B, with origins at G, and
G,. Let O j and 0, be the connecting joints of the springs between Bj and Bk and the springs of
their adjacent bodies, as shown.
With this modelling and nomenclature there are 12 displacements and rotations associated
with each beam segment-six at each end. If the beam is undeformed the reference frames are
mutually aligned. As the beam deforms, the reference frames translate and rotate relative to each
other. For Bj the displacement of 0, relative to G, may be expressed in terms of three components,
and the rotation of kj relative to Bj (or R j ) may be described i,n terms of three parameters.
Similarly, the displacement of Gk relative to G, and the rotation of R, relative to B, (or R j ) may be
 COMPUTER METHODS IN FLEXIBLE MULTIBODY DYNAMICS 1665

Beam

Figure 4. A model of an elastic beam Figure 5. Typical beam segments and reference frames

Table I. Displacement and rotation notation

Definition Notation
Displacement of Gjrelative to Gj
Rotation of Rjrelative to Rj
Displacementof 6 k relative to cj
Rotation of Rk relative to Rj
Displacement of Rk(or Bk)relative to Rj (or Bj) * * A

ukx, uky? Ukn

Rotation of Rk(or Bk)relative to Rj(or Bj) @kx, @ky, @h

described in terms of six parameters. Table I presents a typical listing and notation of these
parameter^.^' Regarding notation, the first subscript of ujlxrefers to the segment, the second to
the segment end and the third to the direction. The over hat of lik, signifies a relative displacement.
The displacement and rotation components of B k relative to Bjmay be expressed in terms of the
end displacements and rotations by the expressions
ukx = uj2x - uklx? uky = uj2y - ukly, ukz = uj2z - U k l z (37)
and
6k.x = 0j2x - 0klx9 6ky = 0j2y - okly, 6kz = 0jzz - 0k1z
(38)
Observe that, although there are 12 local displacement and rotation components associated with
each segment (6 at each end), there are only 6 relative displacement and rotation components.
Using the principles of matrix structural analysis, the displacement and rotation components
may be expressed in terms of the spring force and moment component^.^^ For many beams there
is a linear relation between the loading and deformation. This means that in many cases the force
and moment components of the generalized forces of equations (33) and (36) may be directly
related to the displacement and rotation components of equations (37) and (38).
For beams with non-linear material properties and/or with damping the modelling of the
spring and damper parameters is more difficult. The procedure for determining the parameters in
this case depends upon the properties of the beam and thus requires a judgement and evaluation
by the analyst.

ILLUSTRATIONS/CONVERGENCE
To illustrate the efficacy of the method consider the axial (extensional) oscillation of an
unsupported beam. Let the beam have length: 1.0 m; cross-section area: 0.196 cm2; elastic
1666 R. L. HUSTON

Table TI. Comparison of natural frequencies of continuum and finite segment models for beam
axial vibrations49

Natural Continuum Finite segment models Difference (YO)

frequencies model (Hz) 10 segments 30 segments 10 segments 30 segments

1 108.8 108.4 108.8 0.4 0.0

2 2176 214.1 2 17.6 16 01
3 326.5 3 14.5 325.2 3.7 0.2
4 435.3 407.2 432.1 6.4 0.7
5 544.1 489.9 537.9 9.9 1.1
6 652.9 560.5 642.3 142 1.6
7 76 1fi 6 17.3 744.9 18.9 2.2

Table 111. Effect of beam rotation upon natural frequency

Mode dwo 1 2 3 4 5

0 3.50 21.69 6002 116.0 180.5

5 3.97 23.59 65.31 121.1 194.9
10 4 81 28.81 78.71 135.3 212.8
50 7.23 96.54 252.3 368.2 526.3
100 7.61 188.0 490.1 706.1 996.4
CM 3.51 22.03 61.71 1209 199.9

modulus: 1.2 x lo9 dyn/cm2; and density 1 g/cm3. Then by modelling the beam with 10 and 30
segments, the natural frequencies can be calculated and compared with values computed from a
continuum model. The results are shown in Table 11.
Next, consider a rotating beam with clamped support at the hub. Let the beam modelled by 10
segments. Then the foregoing method may be used to determine the effect of the rotation upon the
natural frequencies. Numerical results obtained by Y. Wang5’ are listed in Table I11 where o is
the angular speed and oois
~ r =) [EI/pAL4]”2
~ (39)
where E, I , p, A and L are the elastic modulus, the second moment of area, the mass density, the
cross-section area and the beam length. Finally, in Table 111, CM represents a continuum
models6 of a non-rotating beam and the data are normalized by coo. (The results were obtained by
numerically integrating the differential equations using a fourth-order Runge-Kutta routine).

DISCUSSION
These examples illustrate the ability of the method to model flexibility effects in beams and long
flexible numbers. It is seen that the accuracy of the method increases as the number of segments
increases. However, the method is not restricted to such applications. Indeed, the applicability of
the method extends far beyond simple beam applications. For example, the method may readily
be used to model non-linear effects in beams arising either from geometric or material non-
linearities. Also, the method may be used to model systems that are not elastic at all, such as
cables and chains which are completely compliant.
 COMPUTER METHODS IN FLEXIBLE MULTIBODY DYNAMICS 1667

The principal feature of the method, however, is not flexibility modelling of individual numbers
of a multibody system, but instead the ability to simultaneously model both flexibility effects and
the inertia loading associated with general motion.
This broad range of application and the ease and efficiency of the modelling stems from the use
of Kane’s equations using partial velocities, partial angular velocities and generalized speeds.
Kane’s equations provide for the explicit development of the multibody dynamics equations as in
equations (23),(24) and (25). The use of generalized speeds decouples the joint force and moment
components so that there is only one such component per equation.
The governing differential equations may be integrated using standard numerical procedures.
To date, fourth-order Runge-Kutta methods have proved to be the most successful.
Finally, in the opinion of this writer, the full potential of this method is yet be developed.
Creative modelling of non-linear systems and the use of new computer hardware will greatly
expand the range of applicability of the method. As these applications are developed new insights
will be obtained about the dynamic behaviour of high speed flexible machines, of space vehicles
and of biosystems.

ACKNOWLEDGEMENT

Research for this paper was partially supported by the National Science Foundation under Grant
MSS8912521 to the University of Cincinnati.

REFERENCES
1. 0. P. Agrawal and A. A. Shabana, ‘Dynamic analysis of multibody systems using component modes’, Comp. Struct.,
21, 1303-1312 (1985).
2. H. Ashley, ‘On passive damping mechanisms in large space structures’, J . Spacecrafi Rockets, 21,448-455 (1984).
3. P. M. Bainum, R. Krishna and V. K. Kumar, ‘The dynamics of large flexible earth painting structures with a hybrid
control system’, J . Astronaut. Sci., xxx, 251-267 (1982).
4. A. K. Banerjee and J. M. Dickens, ‘Dynamics of an arbitrary flexible body in large relations and translation’, J. Guid.
Control Dyn., 13, 221-227 (1990).
5. H. Baruh and S. S. K. Tadikonda, ‘Issuesin the dynamics and control of flexible robot manipulators’, J . Guid. Control
Dyn., 12, 659-671 (1989).
6. C. E. Benedict and D. Tesar, ‘Dynamic response analysis of quasi-rigid mechanical systems using kinematic influence
coefficients’,J . Mechanisms, 6, 383-403 (1971).
7. P. Boland, J. C. Samin and P. Y. Willems, ‘Stability analysis of interconnected deformable bodies in a tobological tree’,
A I A A J., 12, 1025-1030 (1974).
8. H. Bremer, ‘Dynamics of flexible hybrid structures,’ J . Guidance and Control, 2, 86-87 (1979).
9. R. H. Cannon, Jr. and E. Schmitz, ‘Initial experiments on the end-point control of a flexible one-link robot,’ Int. J .
Robotics Res., 3, 62-75 (1984).
10. N. G. Chalhoub and A. G. Ulsoy, ‘Control of a flexible robot arm: Experimental and theoretical results’, A S M E Paper
87- WAIDSC-3 (1987).
11, S. Dubowsky and T. N. Gardner, ‘Design and analysis of multilink flexible mechanisms with multiple clearance
connections’, J . Eng. lndustry ASME, 99,88-96 (1977).
12. S. Dubowsky and M. F. Moening, ‘An experimental and analytical study of impact forces in elastic’, Mech. Mach.
Theory, 13,451-465 (1978).
13. A. G. Erdman, G. N. Sandor and R. G. Oakberg, ‘A general method for kineto-elastodynamic analysis and synthesis
of mechanisms’, J . Eng. Industry ASME, 94, 1193-1205 (1972).
14. W. B. Gevarter, ‘Basic relations for control of flexible vehicles’, A I A A J., 8, 666-678 (1970).
15. J. Y. L. Ho, ‘Direct path method for flexible multibody spacecraft dynam ics’, J . Spacecrafi Rockets, 14, 102-110
(1977).
16. J. Y. L. Ho and R. Gluck, ‘Inductive methods for generating the dynamic equations of motion for multibody flexible
systems Part 2’, Synthesis of’ Vibrating Systems, ASME, New York, 1971.
17. J. Y. L. Ho and D. R. Herber, ‘Development of dynamics and control simulation of large flexible space systems’,
J . Guid. Control Dyn., 8, 374-383 (1985).
18. Y. Huang and C. S. 6. Lee, ‘Generalization of Newton-Euler formulation of dynamic equations to nonrigid
manipulators’, J . Dyn. Syst. Meas. Control, 110, 308-315 (1988).
1668 R. L.HUSTON

19. P. C. Hughes,‘Dynamics of a chain of flexible bodies’, J . Astronaut. Sci., XXVII, 359-380 (1979).
20. P. C. Hughes, ‘Dynamics of a flexible space vehicle with active attitude control’, Celestial Mech. J., 9, 21-39 (1974).
21. R. L. Huston, ‘Flexibility effects in multibody systems’, Mech. Res. Commun., 7, 261-268 (1980).
22. R. L. Huston, ‘Multibody dynamics including the effects of flexibility and compliance’, Comp. Struct., 14, 443-451
(1981).
23. R. L. Huston, ‘Multibody dynamics: Analysis of flexibility effects’, Proc. Ninth U.S. National Congress of Applied
Mechanics, Cornell University, ASME, New York, 1982.
24. I. Imam, G. N. Sandor and S. N. Kramer, ‘Deflection and stress analysis in high speed planar mechanisms with elastic
links’, J . Eng. Industry ASME, 95, 541-548 (1973).
25. W. Jerkovsky, ‘Exact equations of motion for a deformable body’, Aerospace Corporation, SA, SO-TR-77-133,1977.
26. J. J. Kalker and G. J. Olsder, ‘Robots, with flexible links: Dynamics, control and stability’, N T l S Report PB86-I 74885,
1985.
27. T. R. Kane, P. R. Ryan and A. K. Banerjess, ‘Dynamics of a cantilever beam attached to a moving base’, J . Guid.
Control Dyn., 10, 139-151 (1987).
28. D. A. Levinson, ‘Large motions of unrestrained structures’, Proc. Ninth U S . National Congress of Applied Mechanics,
Cornell University, ASME, New York, 1982, pp. 259-264.
29. P. W. Likins, ‘Dynamic analysis of a system of hinge-connected rigid bodies with nonrigid appendages’, Int. J . Solids
Struct., 9, 1473-1487 (1973).
30. P. W. Likins, ‘Quasi-coordinate equations for flexible spacecraft’, A I A A J., 13, 524-526 (1975).
31, P. W. Likins and H. K. Bouvier, ‘Attitude control of nonrigid spacecraft’, Astronaut. Aeronaut., 9, No. 5,64-71(1971).
32 K. W. Lips and V. J. Modi, ‘Transient attitude dynamics df satellites with deploying flexible appendages’, Acta
Astronaut., 5, 797-815 (1978).
33 K. W. Lipps and V. J. Modi, ‘General dynamics of a large class of flexible satellite systems’, Acta Astronaut., 7,
1349-1360 (1980).
34. V. J. Modi, ‘Attitude dynamics of satellites with flexible appendages-A
_. brief review’, J . Spacecrajt Rockets, 11,
No. 11, 743-751 (1974):
35. M. Pascal. ‘Dvnamical analvsis of a flexible maniuulator arm’. Acta Astronaut.. 21. 161-169 (1990).
36. S. Rajaram a i d J. L. Junk&, ‘Identification of vibrating flexible structures’, J.’Guid. Control,’ 8,463-470 (1985).
37. J. P. Sadler and G. N. Sander, ‘A lumped parameter approach to vibration and stress analysis of elastic linkages’,
J . Eng. Industry ASME, 95, 549-557 (1973).
38. A. A. Shabana, ‘Substructure synthesis methods for dynamic analysis of multi-body systems’, Comp. Struct., 20,
737-744 (1985).
39. R. C. Winfrey, ‘Elastic link mechanisms dynamics’, J . Eng. Industry ASME, 93, 268-272 (1971).
40. T. R. Kane, ‘Dynamics of nonholonomic systems’, J . Appl. Mech. ASME, 28, 574-578 (1961).
41. T. R. Kane, Dynamics, Holt, Rinehart and Winston, New York, 1968.
42. T. R. Kane, P. W. Likins and D. A. Levinson, Spacecraft Dynamics, McGraw-Hill, New York, 1983.
43. T. R. Kana and D. A. Levinson, Dynamics: Theory and Application, McGraw-Hill, New York, 1985.
44. R. L. Huston and C. E. Passerello, ‘On multi-rigid body-systems dynamics’, Comp. Struct., 10, 439-446 (1979).
45. R. L. Huston, C. E. Passerello and M. W. Harlow, ‘Dynamics of multirigid body systems’, J . Appl. Mech. ASME, 45,
889-894 (1978).
16. R. L. Huston and C. E. Passerello, ‘Multibody structural dynamics including translation between the bodies’, Comp.
Struct., 18, 999-1003 (1984).
47. R. L. Huston, ‘Multibody dynamics formulations via Kane’s equations’, in J. L. Junkins (ed.), Mechanics and Control
ofLarge Space Structures, Vol. 129, of Progress in Astronautics and Aeronautics, AIAA, 1990, pp. 71-86.
48. E. T. Whittaker, Analytical Dynamics, Cambridge, 1957.
49. R. L. Huston, Multibody Dynamics, Butterworth-Heinmann, Stoneham, MA, 1990.
50. R. L. Huston, ‘Computing angular velocity in multibody systems’, Eng. Computations, 3, 223-230 (1986).
51. J. W. Kamman, R. L. Huston and T. P. King, ‘UCIN-DYNOCOMBS-Software for the dynamic analysis of
constrained multibody systems’, in W. Schielen (ed.), Multihody Systems Handbook, Springer-Verlag, Berlin, 1990, pp.
103-121.
52. T. R. Kane and C. F. Wang, ‘On the derivation of equations of motion’, J . Soc. Industrial Appl. Math., 13,487-492
(1965).
53. T. R. Kane, Analytical Elements of Mechanics, I , Academic Press, New York, 1959.
54. R. L. Huston and C. E. Paserello, Finite Element Methods-An Introduction, Marcel Dekker, New York, 1985.
55. Y. Wang and R. L. Huston, ‘Dynamic analysis of elastic beam-like mechanism systems’, in A. Midha (ed.), Trends and
Developments in Mechanisms, Machines and Robotics-1988, DE-15-2, 1988, pp. 457-460.
56. W. Weaver, Jr., S. P. Timoshenko and D. H. Young, Vibration Problems in Engineering, 5th edn, Wiley, New York,
1990, pp. 427-428.