You are on page 1of 34

Multibody System Dynamics 1: 189–222, 1997.

189
c 1997 Kluwer Academic Publishers. Printed in the Netherlands.

Flexible Multibody Dynamics: Review of Past and


Recent Developments

AHMED A. SHABANA
Department of Mechanical Engineering, University of Illinois at Chicago, Chicago, IL 60607-7022,
U.S.A.

(Received: 2 September 1996; accepted in revised form: 12 March 1997)

Abstract. In this paper, a review of past and recent developments in the dynamics of flexible multibody
systems is presented. The objective is to review some of the basic approaches used in the computer
aided kinematic and dynamic analysis of flexible mechanical systems, and to identify future direc-
tions in this research area. Among the formulations reviewed in this paper are the floating frame of
reference formulation, the finite element incremental methods, large rotation vector formulations, the
finite segment method, and the linear theory of elastodynamics. Linearization of the flexible multi-
body equations that results from the use of the incremental finite element formulations is discussed.
Because of space limitations, it is impossible to list all the contributions made in this important area.
The reader, however, can find more references by consulting the list of articles and books cited at
the end of the paper. Furthermore, the numerical procedures used for solving the differential and
algebraic equations of flexible multibody systems are not discussed in this paper since these proce-
dures are similar to the techniques used in rigid body dynamics. More details about these numerical
procedures as well as the roots and perspectives of multibody system dynamics are discussed in
a companion review by Schiehlen [79]. Future research areas in flexible multibody dynamics are
identified as establishing the relationship between different formulations, contact and impact dynam-
ics, control-structure interaction, use of modal identification and experimental methods in flexible
multibody simulations, application of flexible multibody techniques to computer graphics, numerical
issues, and large deformation problem. Establishing the relationship between different flexible multi-
body formulations is an important issue since there is a need to clearly define the assumptions and
approximations underlying each formulation. This will allow us to establish guidelines and criteria
that define the limitations of each approach used in flexible multibody dynamics. This task can now
be accomplished by using the absolute nodal coordinate formulation which was recently introduced
for the large deformation analysis of flexible multibody systems.
Key words: flexible multibody dynamics, finite element formulations, incremental methods, large
rotations, finite segment method, linear theory of elastodynamics, elastic body inertia, rigid body
inertia, impact in flexible body dynamics, control-structure interaction, computer graphics, large
deformation problem.

1. Introduction
Flexible multibody dynamics is the subject concerned with the computer modeling
and analysis of constrained deformable bodies that undergo large displacements,
including large rotations. The large displacement includes rigid body motion as well
as elastic deformations. As shown in Figure 1, a flexible multibody system may
consist of elastic and rigid components which are connected by joints and/or force
elements such as springs, dampers, and actuators. Because of the joint constraints,
190 A.A. SHABANA

Figure 1. Flexible multibody systems.

the displacements of the bodies in the system are not totally independent. In addition
to the problems encountered in the analysis of multi-rigid-body systems [79],
one must consider the problems associated with modeling the deformation when
flexible multibody systems are analyzed. Therefore, in addition to large rotation
and constraint problems, modeling the body deformation is of a major concern in
the analysis of flexible multibody systems. Furthermore, the deformation can have a
significant effect on the way the large rotation problem and the joint constraints are
formulated. Regardless of the amount of deformation (small or large), the dynamic
equations of the deformable bodies that undergo large rigid body displacements are
highly nonlinear and exhibit a strong nonlinear coupling between different modes
of displacements. Furthermore, the number of coordinates required to obtain a
reasonable mathematical model for flexible multibody systems can be very large as
compared to the number of coordinates used in the analysis of rigid body systems.
Because of the nonlinearity and large dimensionality, the use of modern digital
computers in the analysis of flexible multibody systems becomes a necessity.
Flexible multibody dynamics emerged as a new field in the early seventies
as the result of the need to simulate many industrial and technological systems.
Examples of these systems, where the deformation can have a significant effect
on the system dynamics, are ground and space vehicles, mechanisms, robotics,
space structures, and precision machines. Traditionally, many of these systems
were designed using bulky solids and were analyzed using the assumptions of rigid
body dynamics. The development of new light weight materials and the demand
for higher operating speeds required the development of more accurate method-
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 191

ologies that take into account the effect of deformation of the system components.
In the last two decades, extensive research efforts in flexible multibody dynamics
have led to many new concepts and approaches that were utilized in the analysis
and computer simulations of a wide range of flexible multibody applications. The
new concepts and approaches were the result of research activities in different areas
such as machine dynamics, aerospace, and mechanics (Finite Element). In machine
dynamics, flexible multibody research activities were motivated by several impor-
tant technological and industrial applications such as mechanisms and robotics.
The approach taken in machine dynamics was to utilize the developments in rigid
body dynamics and build on the rigid body methodologies in order to be able to
treat deformable bodies. This approach led to three methods known as the floating
frame of reference formulation, linear theory of elastodynamics, and finite segment
method. Examples of contributions made by researchers in machine dynamics and
robotics are the work of Song and Haug [95], Shabana and Wehage [80], Schiehlen
and Rauh [77], Bakr and Shabana [7, 8], Book [14], Chedmail et al. [18], Chu
and Pan [20], Huston [36], Sadler and Sandor [75], Melzer [58], and Ambrosio
and Pereira [4]. A literature survey on the effect of flexibility on mechanism and
machine systems can be found in the survey articles by Erdman and Sandor [22],
Lowen and Chassapis [49], and Lowen and Jandrasits [50].
Some of the concepts used in the floating frame of reference formulation were
originally presented in the aerospace literature. The flexible multibody research
in the aerospace field was motivated by the need to simulate and analyze flexible
space structures. In these applications, the effect of the body deformations cannot
be ignored. Space structures also experience large rigid body displacements. In the
aerospace literature, some of the fundamental problems associated with deformable
bodies, such as the selection of the deformable body coordinate systems, were
discussed and the basic modeling differences between rigid and deformable bodies
were identified. Examples of contributions made by researchers in aerospace are
the work of Frisch [26], Ho [30], Ho et al. [31], Hooker [32], Hughes [33], Likins
[47], Roberson [75], Meirovitch [54–56], Laskin et al. [45], Modi et al. [62], and
Pascal and Sylia [68].
In the early sixties, the finite element method emerged as a powerful tool for
the analysis of deformation. Several finite element procedures were proposed for
solving large rotation and deformation problems. In most of these procedures,
which are incremental in nature, the large displacements of the finite elements
are described using the nodal coordinates. Such a motion description, as will be
discussed in this review article, does not lead to an exact modeling of the rigid body
dynamics when certain types of elements are used. Examples of contributions from
the finite element community are the work of Belytschko and Hsieh [9], Belytschko
and Glaum [10], Belytschko and Schwer [11], Benson and Hallquist [12], Rankin
and Brogan [70], Cardona and Geradin [16], Geradin et al. [28], and Park et al.
[65]. Several finite element formulations which are not purely incremental were
also proposed [3, 63].
192 A.A. SHABANA

In this paper, some of the basic methods used in flexible body dynamics are
reviewed and future research directions are identified. Among these directions
are establishing the relationship between flexible multibody formulations, impact
and contact problems, control-structure interaction, use of experimental identifica-
tion methods in flexible multibody simulations, application of flexible multibody
methodologies to computer graphics, and large deformation problem. We also dis-
cuss the absolute nodal coordinate formulation which can be used in the large and
small deformation analysis of flexible multibody systems. This formulation can
also be used to establish the relationship between different methodologies used in
flexible multibody dynamics. This paper is organized as follows. In Section 2, the
kinematic approaches used to describe the motion of flexible bodies are reviewed.
In Section 3, different approaches that can be used for the dynamic analysis of
flexible multibody systems are presented. In Section 4, the analytical methods used
for formulating the multibody dynamic equations are discussed, while Section 5 is
devoted to some research topics which are of current interest. Section 6 presents
the main conclusions drawn from this review article.

2. Basic Methods

In this section, the most widely used methods for the kinematic description of
the motion of deformable bodies that undergo large displacements are briefly
discussed. Among these methods are the floating frame of reference, the convected
coordinate system, finite segment method, and the large rotation vector. We will also
discuss the subtle and fundamental problem of the selection of the deformable body
coordinate system when the floating frame of reference formulation is used. Large
deformation problems in flexible multibody dynamics can be efficiently solved
using the absolute nodal coordinate formulation. This formulation represents a
departure form the conventional finite element formulations used in engineering
analysis in the sense that no infinitesimal or finite rotations are used as nodal
coordinates. Using the absolute nodal coordinate formulation, beam and plate
elements, which are considered as non-isoparametric elements in the classical
finite element literature, can be treated as isoparametric elements.

2.1. FLOATING FRAME OF REFERENCE

The floating frame of reference formulation is currently the most widely used
method in the computer simulation of flexible multibody systems. It is implemented
in several commercial as well as research general purpose multibody computer
programs. In the floating frame of reference formulations, two sets of coordinates
are used to describe the configuration of the deformable bodies; one set describes the
location and orientation of a selected body coordinate system, while the second set
describes the deformation of the body with respect to its coordinate system. Using
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 193

Figure 2. Floating frame of reference.

this description, the global position vector of an arbitrary point on the deformable
body i can be written as [46, 80, 83]
ri = Ri + Ai (
uio + u
 if ); (1)

where all the vectors that appear in this equation are shown in Figure 2, and Ai is
the transformation matrix that defines the orientation of the body coordinate system
with respect to the global coordinate system. In the preceding equation, subscripts
o and f refer, respectively, to the undeformed position and deformation variables.
Several orientation parameter sets can be used to formulate the transformation
matrix Ai such as Euler angles and Euler parameters. The kinematic description
used in the preceding equation, leads to exact modeling of the rigid body dynamics
when the deformation is equal to zero. This an important issue in flexible multibody
dynamics since some of the methods that will be discussed in this review article do
not lead to exact modeling of the rigid body dynamics, and as a consequence, their
use in flexible multibody simulation must be carefully examined. It is important also
to point out that the preceding kinematic description does not imply any separation
between the rigid body motion and the elastic deformation of the flexible body,
since any coordinate system can be selected as the body reference [87].
In many investigations, a kinematic description based on the 4 4 transformation
matrix was used [96, 97]. The 4  4 transformation matrix is a function of four
parameters; the link length, link twist, link offset and joint angle. The first of these
two parameters are constant, while the second two are variables. In the case of
flexible links, however, the link length and link twist are no longer constant, and
194 A.A. SHABANA

therefore, the change in these two parameters as the link deforms must be taken
into consideration in order to be able to develop a consistent formulation.
In the kinematic description given in the preceding equation, the deformation
of the bodies can be described using component modes [46, 48] or using the finite
element method [80, 83]. When the finite element method is used, the component
mode technique [25] can be used to reduce the dimensionality of the problem
that results from the use of the finite element formulation. Song and Haug [95]
used the preceding equation to describe the kinematics of each finite element. A
coordinate system is attached to each finite element of the deformable body and an
independent set of coordinates are used to describe the orientation of each finite
element. The connectivity between elements are defined using nonlinear constraint
equations instead of the linear Boolean matrix approach used in the standard finite
element procedure. The use of this approach makes it difficult to eliminate high
frequency modes using modal truncation methods.

2.1.1. Reference Conditions


One of the fundamental problems that arises when using the floating frame of
reference formulation is the selection of the deformable body coordinate system.
The shape of deformation of the body is defined in its coordinate system, and as
a consequence, the selections of the deformation shapes and the body coordinate
system cannot be considered as two independent issues. In the case of rigid body
dynamics, the use of a centroidal body coordinate system leads to simplification
of the dynamic equations by decoupling the rigid body translation and rotational
motion. Many attempts have also been made in the past to simplify the dynamic
equations of the deformable bodies by a proper selection of the deformable body
coordinate system [1, 5, 17, 21, 29, 42, 61]. In many of these investigations,
deformable body coordinate systems that satisfy the mean axis conditions were
chosen. These conditions are obtained by minimizing the relative kinetic energy
with respect to an observer stationed on the body. Applying the deformable body
mean axis conditions leads to a weak coupling between the reference motion and
the elastic deformation. The selection of the deformable body coordinate system
and its relationship with the boundary conditions, when the floating frame of
reference formulation is used, is one of the most fundamental and interesting
problems in modeling deformable bodies. It can be demonstrated that two different
sets of vibration modes that correspond to different sets of natural frequencies and
define two different coordinate systems for the deformable body lead to the same
displacement solutions [1, 29, 87]. This is an important result which implies that
the natural frequencies of the linear problem do not have a significant effect on
the solution of the multibody dynamic equations. This result can also be used to
demonstrate that the kinematic description used in the floating frame of reference
formulation does not imply a separation between the rigid body motion and the
elastic deformation.
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 195

2.2. CONVECTED COORDINATE SYSTEM

The incremental finite element formulations that employ the convected coordinate
system have been widely used to solve large rotation and deformation problems.
There are two types of finite elements that are used in the static and dynamic
analysis of deformable bodies. These are the isoparametric and non-isoparametric
elements. An isoparametric element is defined as an element which has a shape
function that can be used to interpolate the location as well as the deformation of
the material points of the elements. As a consequence, the isoparametric elements
can describe an arbitrary rigid body displacement and can be used to obtain exact
modeling of the rigid body dynamics. Examples of these elements are the rectangu-
lar, triangular, solid, and tetrahedral elements. In general, these elements use only
nodal displacements as coordinates and do not employ rotations or slopes as nodal
coordinates. The other type of elements is the non-isoparametric type. These ele-
ments use infinitesimal rotations as nodal coordinates. Examples of these elements
are beam, plate, and shell elements. These elements cannot be used to model exact
rigid body motion and they do not produce zero strain under an arbitrary rigid body
displacement. For this reason, extensive research efforts were devoted to devel-
oping computational procedures for the large rotation and deformation problems
using the widely used non-isoparametric elements.
One of the most widely used computational procedure for non-isoparametric
elements is the incremental finite element approach. In the incremental methods,
it is assumed that the element shape function can describe small rotations. A
convected coordinate system that shares the large rotation of the element is chosen
for each element [9]. In the convected coordinate system approach, the kinematic
equations are first defined in the element coordinate system with the assumption that
the element coordinate system coincides with a selected fixed coordinate system.
By assuming that the rotation between two consecutive configurations that are
associated with two fixed coordinate systems is small, the element shape function
and nodal coordinates can be used to describe this small rotation. Hughes and
Winget [34] presented a numerical procedure in which the displacement increment
over a time step was defined and the gradient of this displacement was used to define
the strain and rotation tensors. Flanagan and Taylor [24] presented a numerical
algorithm which is based on the choice of the un-rotated configuration as the frame
of reference. The polar decomposition theorem was used to define the rigid body
rotation tensor. Rankin and Brogan [70] developed a corotational procedure that
enables finite element formulations to be used in large rotation problems. In this
procedure, the contribution of the rigid body motion of the element to the total
displacement field is removed before the element computations are performed.
The incremental finite element formulations, which are implemented in several
general purpose finite element computer programs, do not lead to exact modeling
of the rigid body dynamics when the structures rotate as rigid bodies. Furthermore,
non-isoparametric elements used in the incremental methods do not lead to zero
196 A.A. SHABANA

Figure 3. Finite element coordinate systems.

strain under an arbitrary rigid body displacement. In order to solve this problem,
a finite element formulation for the dynamic analysis of multibody systems that
consist of interconnected deformable bodies was presented [80, 83]. In this formu-
lation, which assumes that the element shape function can be used to describe only
an arbitrary large rigid body translation, four coordinate systems are employed to
define the finite element configuration, as shown in Figure 3. The first is a global
coordinate system which is fixed in time and serves as a unique standard for all the
deformable bodies in the system. Connectivity conditions that describe mechanical
joints between different deformable bodies in the system, such as revolute, pris-
matic, spherical and cylindrical joints, are defined in this global coordinate system.
The second coordinate system is a floating body coordinate system whose origin
does not have to be rigidly attached to a point on the deformable body. Connec-
tivity conditions between the finite elements of this body are defined in the body
coordinate system using a Boolean matrix approach. The third coordinate system is
the element coordinate system which has an origin rigidly attached to the element.
The fourth coordinate system is an intermediate element coordinate system which
has an origin that is rigidly attached to the origin of the body coordinate system,
and the axes of this intermediate coordinate system are selected such that they
are parallel to the axes of the element coordinate system in the undeformed initial
configuration. Using the fact that the element shape function can describe large
rigid body translation, exact modeling of the rigid body inertia can be obtained
in the intermediate element coordinate system whose orientation with respect to
the body coordinate system is defined by a constant transformation matrix. Using
the concept of the intermediate element coordinate system, which is similar to
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 197

the parallel axis theorem used in rigid body dynamics, an exact modeling of the
rigid body dynamics can be obtained [88]. Furthermore, one obtains zero strain
under an arbitrary rigid body motion when non-isoparametric elements that employ
infinitesimal rotations as nodal coordinates are used.

2.3. FINITE SEGMENT METHOD


In the finite segment approach, the deformable body is assumed to consist of a
set of rigid bodies which are connected by springs and/or dampers [35–37, 71,
77]. The configuration of each rigid segment is identified using three indepen-
dent Cartesian coordinates and the four dependent Euler Parameters. The system
elasticity represented by spring coefficients can be determined using the finite
element method. This technique has the advantage that rigid body methodologies
can be used in formulating the equations of motion of deformable bodies. Some
problems, however, remain to be solved when the finite segment method is used.
Among these problems are the selection of the number, size, and location of the
rigid segments as well as the representation of the inertia coupling between these
segments. Nonetheless, the finite segment method has been used in solving many
multibody applications including vehicle crash simulations [64]. The accuracy and
consistency of the finite segment approach can be significantly improved by using
a set of mode shapes that introduce inertia coupling between the rigid bodies that
form the deformable structure.

2.4. LARGE ROTATION VECTORS


As previously pointed out, the incremental finite element approach was used to
solve the finite rotation problem because classical beam and plate element shape
functions cannot be used to describe large rotations. This mainly due to the fact that
infinitesimal rotations are used as nodal coordinates in the case of non-isoparametric
elements. As a consequence, the finite element incremental approach leads to
a linearization of the equations of motion of the structures that undergo large
reference displacements. In order to avoid this linearization, large rotation vectors
are used to describe the element configuration. In this case, the element nodal
coordinates are defined with respect to the global coordinate system [28, 91, 92].
Finite rotations of the element cross section are treated as field variables that can
be interpolated in the same manner the displacement coordinates are interpolated.
As a result, the nodal coordinates of the finite element consist of displacements as
well as finite rotation coordinates.
The large rotation vector formulations are not widely used in flexible multibody
computer simulations. They have not also been implemented in many commercial
finite element codes that are used for the large rotation and deformation analysis of
structural systems. The limited success of the large rotation vector formulation in
flexible multibody dynamics is attributed to several factors. One of the important
198 A.A. SHABANA

Figure 4. Absolute nodal coordinate formulation.

factors is the redundancy of representing the large rotation of the cross section.
It is known from differential geometry that when the coordinates of the points
on a curve are represented as field variables, a Frenet frame defined by a tangent
and a normal vector can be obtained using the derivatives of the coordinates of
the points on the curve. In the case of beams, the Frenet frames can be used to
define the orientation of the cross sections. Therefore, the large rotation of the cross
sections of the beam can be defined using the displacement coordinates. Introducing
finite rotation coordinates in the large rotation vector formulations leads to two
independent representations of the large rotation of the cross sections, even when
shear deformations are considered since the interpolating polynomials that define
the displacements have coefficients which are independent from the coefficients
used in the polynomials that describe the large rotation coordinates. Because of
the redundancy, large rotation vector formulations can lead to singularity problems
when slender structures are considered and these formulations can also lead to
unrealistic shear forces since the shear angle is no longer defined as an infinitesimal
deviation from the normal to the cross section. This singularity problem also arise
if the effect of shear deformation is neglected.

2.5. ABSOLUTE NODAL COORDINATE FORMULATION


As previously pointed out, beam, plate and shell elements are not considered in the
classical finite element literature as isoparametric elements because these elements
cannot be used to obtain exact modeling of the rigid body dynamics. This is mainly
due to the fact that infinitesimal rotations are used as nodal coordinates. In the
absolute nodal coordinate formulation, no infinitesimal or finite rotations are used
as nodal coordinates, instead absolute displacements and global slopes are used
as the element coordinates [88–90]. The locations and the deformations of the
material points on the finite element are defined in the global coordinate system
using the element shape function and the nodal coordinates. As shown in Figure 4,
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 199

the global position vector of a material point on a beam element is defined by the
vector r. In the absolute nodal coordinate formulation, this vector can be written as
r = Se; (2)
where e is the vector of the element nodal coordinates that consists of absolute
displacements and slopes, and S is called here the element global shape function
in order to distinguish it from the local shape function used in the floating frame of
reference formulation. The global shape function Smust have a complete set of rigid
body modes that can describe an arbitrary rigid body translational and rotational
displacements. Using the absolute nodal coordinate formulation, beam and plate
elements can be used to obtain exact modeling of the rigid body dynamics, and as
a consequence, these elements can be considered as isoparametric elements. The
absolute nodal coordinate formulation has a potential in solving large deformation
problems in many engineering applications in general, and in flexible multibody
dynamics in particular.

3. Dynamic Formulations
Using different methods for the kinematic description, various dynamic approaches
can be used for formulating the dynamic equations of flexible multibody systems.
In this section, some of these dynamic approaches are reviewed and the approxi-
mations and assumptions underlying some of these methods are discussed. Exact
modeling of the rigid body dynamics will be one of the criteria which will be
given special consideration since deformable bodies in multibody systems undergo
large reference displacements. Furthermore, general purpose multibody computer
programs are designed for the simulation of systems that consist of interconnected
rigid and deformable bodies.

3.1. FLOATING FRAME OF REFERENCE


Using the dynamic description presented in Section 2.1, the principle of virtual
work in dynamics or Lagrange’s equation can be used to systematically develop the
dynamic equations of motion of the deformable bodies that undergo large reference
displacements. In the floating frame of reference formulation, the equations of
motion are expressed in terms of a coupled set of reference and elastic coordinates.
The reference coordinates define the location and the orientation of a selected
body coordinate system, while the elastic coordinates define the deformation of the
body with respect to its reference. The elastic coordinates can be introduced using
component mode methods, the finite element method or experimental identification
techniques. Using the floating frame of reference formulation, the equations of
motion of a deformable body in the multibody system can be written in the following
general form:
Mi yi + Ki yi = qie + qiv + qic ; (3)
200 A.A. SHABANA

where superscript i refers to the body number, M is the mass matrix, K is the
stiffness matrix, y is the vector of the system generalized coordinates, qe is vector
of externally applied forces, and qv is the vector of Coriolis and centrifugal forces,
qc is the vector of the constraint forces. The vector of the constraint forces can also
be expressed in terms of the vector of Lagrange multipliers  as

qic = ,QiT ;

where Q is the Jacobian matrix of the kinematic constraint equations that define
the joint constraints and specified motion trajectories. The equations of motion of
the multibody system can be written as

My + Ky = qe + qv + qc :

The vector of coordinates y can be partitioned as


h iT
y = yTr yTf ;

where subscripts r and f refer, respectively, to reference and elastic coordinates.


Using this coordinate partitioning, the equations of motion of the flexible multibody
system can be written as [83]
" #" # " # " # " # " #
Mrr Mrf 
yr (qe )r (qv )r (qc )r
Mfr Mff yf
+
0 0
0 Kff
=
(qe )f
+
(qv )f
+
(qc )f
:
The floating frame of reference formulation leads to a highly nonlinear mass matrix
as the result of the inertia coupling between the reference motion and the elastic
deformation. The stiffness matrix, on the other hand, is the same as the stiffness
matrix used in structural dynamics due to the fact that the elastic coordinates are
defined with respect to the body coordinate system.

3.1.1. Inertia Shape Integrals


It can be shown that the nonlinear inertia forces of the deformable body that
undergoes large reference displacements can be expressed in terms of a unique set
of inertia shape integrals that depend on the assumed displacement field. In the
spatial analysis, there are seven unique inertia shape integrals which are defined
for the deformable body i as
Z
 i = i Si dV i ;
S
Vi
Z
Sikl = iSikT Sil dV i ; k; l = 1; 2; 3;
Vi
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 201

where i and V i are the mass density and volume of the deformable body, Si is
the shape function that defines the deformation of body i and subscript k and l
refer to the k th and lth row of the shape function. In the case of the finite element
formulation, similar integrals can be obtained for each finite element. The body
shape integrals can be obtained by assembling the integrals of its elements. The
integrals of different types of finite elements have been developed and reported
in the literature. These integrals can also be evaluated using lumped or consistent
mass formulation [16, 58, 81, 85]. The use of lumped mass formulations or other
numerical methods for evaluating the inertia shape integrals allows general purpose
flexible multibody computer programs to be directly coupled with existing finite
element computer programs.

3.1.2. Implementation
The floating frame of reference formulation is the most widely used method in the
computer simulation of flexible multibody systems. This formulation leads to an
exact modeling of the rigid body dynamics, and also leads to zero strain under an
arbitrary rigid body displacement. Crucial to the successful and efficient computer
implementation of the floating frame of reference formulation is the identification
of the body inertia shape integrals. These inertia shape integrals can be evaluated
in a pre-processor computer program using lumped or consistent mass approach,
and can be expressed in a modal form. By so doing, the main computer program
can be designed to be independent of the type of the finite element used and also
independent of the method used to define the body inertia (consistent or lumped).

3.2. LINEAR THEORY OF ELASTODYNAMICS


The finite element floating frame of reference formulation based on the concept
of the intermediate element coordinate system was developed in the early eighties.
As the result of introducing this formulation, it became clear that finite element
computer codes must be used as preprocessors for flexible multibody computer
programs in order to evaluate the inertia shape integrals required to formulate
the nonlinear dynamic equations of the deformable bodies. Prior to introducing
the finite element floating frame of reference formulation, another sequence of
computations was adopted in order to predict the deformations and stresses of the
deformable components in multibody systems. The solution strategy which was
used in the past is to treat the multibody system first as a collection of rigid bodies.
General purpose multibody computer methodologies and programs can then be used
to solve for the inertia and reaction forces. These inertia and reaction forces obtained
from the rigid body analysis are then introduced to a linear elasticity problem in
order to solve for the deformation of the flexible components in the system. The
total motion of the deformable bodies is then obtained by superimposing the small
elastic deformation on the gross rigid body motion. This approach is referred to
202 A.A. SHABANA

as the linear theory of elastodynamics. In the linear theory of elastodynamics, the


rigid body motion and the elastic deformation are not solved simultaneously. It is
assumed that the elastic deformation does not have a significant effect on the rigid
body displacements, and therefore, the inertia terms in the reference equations are
assumed to be independent of the elastic deformation [22, 49, 50, 98, 101, 102]. In
order to understand the assumptions used in the linear theory of elastodynamics, we
rewrite the equations of motion of the deformable body presented in Equation (3)
as
Mirr yir + Mirf yif = (qie )r + (qiv )r ;

Mifr yir + Miff yif + Kiff yif = (qie )f + (qiv )f :


In the linear theory of elastodynamics, the effect of the deformation on the rigid
body displacement is neglected. In this case, the equations of the linear theory of
elastodynamics can be obtained as a special case of the preceding equations as
Mirr yir = (qie )r ;
Miff yif + Kiff yif = (qie )f + (qiv )f , Mifr yir :
In the first of these two equations, all terms, including the inertia tensor and
forces, are assumed to be independent of the deformations. The first equation
can be solved for the reference coordinates, velocities and accelerations as well
as the reaction forces using rigid multibody computer programs. The obtained
coordinates, velocities, and forces are substituted into the second equation in order
to determine the deformation of the bodies using standard finite element techniques.
The accuracy of the results obtained using the linear theory of elastodynamics
is questionable, particularly, when high speed, light weight mechanical systems
are considered. In many of these systems, the inertia coupling between the rigid
body motion and the elastic deformation is significant such that the effect of
this coupling cannot be neglected in the dynamic simulations. Furthermore, there
are some applications where the instability of the elastic modes can lead to the
instability of the rigid body modes.

3.3. INCREMENTAL FINITE ELEMENT APPROACH


The incremental finite element formulations are conceptually different from the
multibody formulations that employ the reference and elastic coordinates. Incre-
mental formulations have been successfully applied to many large deformation
problems in structural systems. In most of the incremental formulations, the large
displacements of the finite elements are described incrementally using the element
nodal coordinates. As previously mentioned, in the incremental finite element for-
mulations, a sequence of fixed coordinate systems are introduced, and at a given
instant of time, the finite element coordinate system is assumed to coincide with one
of these fixed coordinate systems. The equations of motion are first defined in the
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 203

Figure 5. Modeling of rigid body dynamics.

convected element coordinate system, leading to a simple expression for the mass
and stiffness matrices. Using the nodal coordinates, the transformation matrix that
defines the orientation of the convected element coordinate system with respect to
the global coordinate system can be evaluated and used to define the global inertia
and stiffness matrices. These global element matrices can then be used to define
the structure mass and stiffness matrices. The resulting structure equations can be
solved in order to determine the solution at the next time step.
The basic assumption used in the incremental finite element formulation is that
the rotation between two consecutive configurations is small such that the element
shape functions of beams and plates can be used to describe this rotation. Recall
that these shape functions can only describe infinitesimal rotations, and as such, the
use of the incremental finite element formulations involves a linearization of the
equations of motion. It can be shown that many of the currently used incremental
formulations do not lead to exact modeling of the rigid body dynamics when the
structure rotates as a rigid body [86, 88]. Therefore, the use of these formulations
must be carefully examined in the case of highly nonlinear problems as in the case
of multibody applications.
In order to avoid the linearization of the dynamic equations when the incre-
mental finite element formulation is used, the concept of the intermediate element
coordinate system was used in the floating frame of reference formulation. As pre-
viously pointed out, the use of the intermediate element coordinate system leads to
exact modeling of the rigid body dynamics. In order to demonstrate this fact, we
use Figure 5 which shows a uniform slender rod which has length l, cross sectional
area a, mass density , volume V , and mass m. The coordinate system of this beam
element is assumed to be attached to its left end which is defined by point O as
shown in the figure. Let
204 A.A. SHABANA

u = [x 0]T
be the position vector of an arbitrary point on the center line of the uniform slender
beam shown in Figure 5. Since the origin of the coordinate system of the slender
beam is attached at point O , the mass moment of inertia of the beam IO is defined
as
Z Z1
IO = u u dV = ax2 dx = ml3
2
T
V 0
and the moment of mass mO is
2 ml 3
Z Zl
mO = u dV = au dx = 4 2 5:
V 0 0
The mass, the mass moment of inertia and the moment of mass are the three
inertia quantities which form the mass matrix of the beam when the beam is treated
as a planar rigid body. The exact values of these inertia elements can be evaluated
using the finite element shape functions, provided that these shape functions can
describe rigid body translations [86, 88]. In order to demonstrate this, we consider
the beam element shape function defined as
" #
1, 0 0  0 0
S=
0 1 , 3 2 + 2 3 l( , 2 2 +  3 ) 0 3 2 , 2 3 l( 3 ,  2 )
; (4)
where  = x=l. The vector of nodal coordinates associated with the shape function
defined in the preceding equation is
e = [e1 e2 e3 e4 e5 e6 ]T ;
where e1 and e2 are the translational coordinates of the node at O , e4 and e5 are the
translational coordinates of the node at A, and e3 and e6 are the slopes at the two
nodes. Using the coordinate system shown in Figure 5 and assuming that the beam
is in the undeformed state, the vector of nodal coordinates defined with respect to
the coordinate system whose origin is attached to point O is
eO = [0 0 0 l 0 0 ]T :
Note that the use of this vector of nodal coordinates leads to
 
SeO =
x =u
0
and as a consequence, the exact mass moment of inertia can be defined using the
element nodal coordinates as [88]
2 3
Z Z
IO = uT u dV = eTO 4 ST S dV 5 eO = eTO Mff eO = ml3 ;
2

V V
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 205

which is the exact mass moment of inertia of the beam, and Mff is the symmetric
matrix
21 1 3
63 0 0 0 0 7
6 6 7
6
60 13 11 l 9
, 13 l 7
7
6 0
420 7
6 35 210 70 7
6 2 7
Z
6
60 11 l l 2
0
13 l , l 7
7
6 140 7
Mff = ST S dV = m 6 61
210 105 420 7:
7
6 1
0 7
V 6
66
0 0
3
0 7
7
6 7
6 13l 11l 7
60
6 70
9
420
0
13
35
, 210 7
7
6 7
4 13l l 2 11l l 2 5
0 , , 0 ,
420 140 210 105
Similarly, the exact moment of mass of the beam can be evaluated using the shape
function and the element nodal coordinate vector eO as
2 ml 3
Z
mO = u dV = Sm eO = 4 2 5 ;
V 0
where Sm is the matrix
Z
Sm = S dV =
m 6 0 0 6 0 0

:
12 0 6 l 0 6 ,l
V
It can be demonstrated that exact modeling of the inertia of the rigid beam
can be obtained in any intermediate element coordinate system which differs from
the element coordinate system by an arbitrary translation by using the element
shape function and the nodal coordinates [88]. Therefore, the use of the concept
of the intermediate element coordinate system which is similar to the parallel axis
theorem used in rigid body dynamics leads to exact modeling of the rigid body
motion.

3.3.1. Incremental Formulations and Linear Theory of Elastodynamics


Recall that in the incremental finite element formulations, a convected coordi-
nate system is introduced for each finite element. In some beam applications, the
convected system is selected such that one of its axes is defined as the vector
connecting the two nodal points of the beam element. The assumptions underlying
the use of this approach are basically different from the ones used in the linear
theory of elastodynamics discussed in the preceding section. In the linear theory of
elastodynamics, the rigid body motion, including the orientations of the coordinate
206 A.A. SHABANA

systems, are determined first using multi-rigid body formulations, while the elastic
deformations are determined in a postprocessing step using a structural dynamics
formulations. In the incremental finite element formulations, on the other hand,
the displacements of the elements are determined first using structural dynamics
formulations and the orientation of the element convected coordinate system is
defined using the displacements of the nodes. In simple terms, the linear theory of
elastodynamics uses exact model for the rigid body dynamics; it neglects, however,
the effect of the elastic deformation on the gross rigid body motion. In the incre-
mental formulations, the gross motion of the element is defined using the element
nodal coordinates; but these formulations do not lead to exact modeling of the rigid
body inertia since infinitesimal rotations are used as nodal coordinates. The lin-
earization that results from the use of the infinitesimal rotations can be avoided by
using the absolute nodal coordinate formulation which leads to exact modeling of
the rigid body dynamics and takes into account the effect of the coupling between
different modes of displacements.

3.4. LARGE ROTATION VECTORS AND OTHER FORMULATIONS


In order to avoid the linearization resulting from the use of the incremental finite
element formulations, several investigators [28, 91, 92] used a set of finite rotation
coordinates for the finite elements. These rotations coordinates define the orien-
tation of the finite element cross section with respect to the global coordinate
system. Geradin et al. [28] presented a method for the large displacement and
deformation analysis of structural components in flexible multibody systems. A
three dimensional rotation vector is used to define the element frame orientation.
The singularity as the result of using three parameters representation for the rota-
tion was avoided by restricting the angle of rotation about the axis of rotation to be
less than  .
Garcia de Jalon et al. [27] presented a methodology for the nonlinear dynamic
analysis of rigid and flexible bodies of constrained mechanical systems. In this
method, the configurations of the bodies and the constraints are described using
the Cartesian coordinates of points, Cartesian components of unit vectors, joint
coordinates (optionally), and modal coefficients (optionally). At least two points
and one non co-linear unit vector are required in order to completely define the
motion of the body, no angular coordinates are needed. The bodies, however,
must contain sufficient number of points and unit vectors so that their motion
is completely defined. The use of this description makes the formulation of the
joint constraints easy and straight forward. By using the Cartesian components of
points and unit vectors, the degree of nonlinearity of the constraint equations is
significantly reduced.
As an example of the large rotation vector formulation, we consider the work
of Simo and Vu-Quoc [92]. In this work, a finite element procedure was developed
for the large displacement and rotations of beams. Finite rotations are used as
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 207

Figure 6. Large rotation vector formulations.

field variables leading to a set of nodal coordinates that consist of displacement


coordinates as well as finite rotation coordinates. In order to explain the basic idea
used in this formulation, we consider the beam element shown in Figure 6. The
element has length l, cross sectional area a, volume V , mass density , and mass m.
In Figure 6, defines the orientation of the cross section without shear deformation,
defines the shear angle, and defines the final rotation of the cross section of
the beam element, that is = + . In the large rotation vector formulation the
configuration of the finite beam element is defined by three field variables. These
are the two components of the vector r that defines the location of an arbitrary
point on the beam and the angle that defines the orientation of the beam cross
section. As will be shown in the following subsection, the large rotation of the cross
section as defined by the angle can be determined using the components of the
vector r. As a consequence, this large rotation has two unrelated representations,
one in the vector r and the other in the angle . Such a motion description leads to
a redundant representation and can lead to fundamental problems in defining the
generalized forces associated with the beam generalized coordinates. Because of
this redundancy problem, large rotation vector formulations can lead to singularity
problems when slender structures are considered [89]. Introducing the shear effect
does not solve the problem of the coordinate redundancy even though the shear,
in principle, can be treated as an independent variable. In order to solve this
redundancy problem, the angle must be expressed in terms of the derivatives of
the components of the vector r. This solution, however, will lead to a nonlinear
mass matrix for the element and will make large rotation vector formulations much
less attractive to use. Moreover, large rotation vector formulations also require
the interpolation of finite rotations. Some of the problems associated with the
large rotation vector formulations can be circumvented by using the absolute nodal
coordinate formulation.
208 A.A. SHABANA

3.5. ABSOLUTE NODAL COORDINATE FORMULATION


In the absolute nodal coordinate formulation, the element nodal coordinates are
defined in the inertial frame. These nodal coordinates are used with a global shape
function which has a complete set of rigid body modes. Therefore, the global
position vector of an arbitrary point on the element can be described using the
global shape function and the absolute nodal coordinates as
r = Se; (5)
where S is the global shape function, and e is the vector of element nodal coor-
dinates. In the absolute nodal coordinate formulation, no infinitesimal or finite
rotations are used as nodal coordinates. The element coordinates are expressed in
terms of nodal displacements and slopes which can be determined in the unde-
formed reference configuration using simple rigid body kinematics. Using this
motion description, beams and plates can be treated as isoparametric elements
without the need to introduce orientation coordinates to describe the rigid body
rotation of the deformable element. As previously pointed out, introducing such
an orientation coordinate leads to a redundant set of variables since the tangent
and normal vectors at an arbitrary point on the deformed center line of the element
can be obtained using the derivatives of the position vector with respect to the
spatial coordinates. This fact can be demonstrated by considering the beam shown
in Figure 6. Using the Euler–Bernoulli beam assumptions, the orientation of the
coordinate system defined by the tangent vector t and the normal vector n can be
described in the inertial frame using the transformation matrix
 
cos , sin
Ac =
sin cos
;
where
(@r1 =@x) (@r2 =@x)
cos = q ; sin = q ;
(@r1 =@x)2 + (@r2 =@x)2 (@r1 =@x)2 + (@r2 =@x)2
where r1 and r2 are the components of the vector r that defines the global position
vector of the arbitrary point as described by Equation (5). Using the Frenet frame
whose orientation is defined in the inertial frame by the angle , the position
and orientation of the beam cross section in Euler–Bernoulli beam theory can be
uniquely defined using the vector r and the angle .

3.5.1. Rigid Body Motion


In the case of an arbitrary rigid body motion of the beam, the global position vector
of an arbitrary point on the beam element can be written as
" # " #
r R 1 + x cos 
r2 = R2 + x sin  ;
1
r=
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 209

where R1 and R2 are the global coordinates of the endpoint O and  in this case is
the angle that defines the beam orientation as shown in Figure 5. It follows that the
slopes in the case of a rigid body motion are defined as
@r1 = cos ; @r2 = sin :
@x @x
In this section, we consider cubic polynomials to define the elements of the
vector r. It is justified to use the same representation for the elements of this vector
since they are both defined in the inertial frame when the absolute nodal coordinate
formulation is used. In this case, the global shape function is given by
S= (6)
h i
1 , 32 + 23
0 l( , 22 + 3 )
0 32 , 23 0 l(3 , 2 ) 0
0 , 32 + 23
1 0 l( , 22 + 3 ) 0 32 , 23 0 l(3 , 2 )
and the vector of nodal coordinates is
e = [e1 e2 e3 e4 e5 e6 e7 e8 ]T ;
where e1 , e2 and e5 , e6 are, respectively, the absolute coordinates of the nodes at
O and A, and
e3 = @r1 (@x x = 0) ; e = @r2 (x = 0) ;
4
@x
e7 = @r1 (@x x = l) ; e = @r2 (x = l) :
8
@x
Using the simple rigid body kinematic equations previously obtained in this section,
one can show that in the case of an arbitrary rigid body motion defined by the
translations R1 and R2 of the endpoint O and the rotation defined by the angle  ,
the vector of the nodal coordinates e can be written as
e = [R1 R2 cos  sin  R1 + l cos  R2 + l sin  cos  sin  ]T :
Using this vector of nodal coordinates, and the shape function of Equation (6), it
can be verified that
" # " #
Se =
R 1 + x cos 
=
r1 = r
R2 + x sin  r2
which demonstrates that the element shape function of Equation (6) can describe
an arbitrary rigid body motion when global slopes instead of infinitesimal rotations
are used as nodal coordinates. Similar comments apply to the shape function of
Equation (4).

3.5.2. Inertia Coefficients


Consider the planar beam element shown in Figure 4. Assuming that the shape
function can describe an arbitrary rigid body displacement, the global position
210 A.A. SHABANA

vector of an arbitrary point on the center line of the element can be defined using
Equation (5). By differentiating this equation with respect to time, the kinetic
energy of the element can be written as

T = 12 e_ T Ma e_ ;
where Ma is the constant mass matrix of the element defined as
Z
Ma = ST S dV:
V
Note that this mass matrix is constant, and it is the same mass matrix that appears in
linear structural dynamics. While in the absolute nodal coordinate formulation the
mass matrix takes a simple form, it can be shown using the kinematic description
of Equation (5) that the strain energy is a highly nonlinear function of the element
coordinates even in the case of a linear elastic model [89, 90].
The absolute nodal coordinate formulation can be efficiently used in the large
deformation analysis of flexible multibody systems. It can also be effectively used
in the analysis of curved structures. The fact that this formulation leads to exact
modeling of the rigid body inertia can be used to obtain the mass moment of
inertia of complex structures. Table I shows the mass moment of inertia of simple
structures obtained using the absolute nodal coordinate formulation. The results
presented in this table are obtained using the simple equation
I = eT Ma e;
where I is the mass moment of inertia defined in the coordinate system in which the
vector e is defined. The results obtained using different number of finite elements
are compared with the exact solution. These results are obtained using a structure
which has total length of 2.4 m, cross sectional area of 0.0018 m2 , and total mass
of 11.9664 kg. The mass moments of inertia presented in the table are in kg.m2 .

4. Analytical Methods
As previously pointed out, the floating frame of reference approach is, at present,
the most widely used method for the dynamic simulation of complex flexible multi-
body systems, for it allows nonlinear constraints and forcing functions to be easily
and systematically introduced to the dynamic equations. Using the floating frame
of reference approach or any other modeling technique, there are several meth-
ods which can be used for formulating the acceleration equations of constrained
multibody systems. These methods, which can be obtained using different analyt-
ical approaches such as the methods of Lagrangian and Newtonian mechanics, are
briefly discussed in this section since they are in the most part similar in concept
to these methods used in rigid multibody systems,. The methods presented in this
section can be broadly classified as augmented or recursive methods.
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 211
Table I. Mass moments of inertia determined using the absolute nodal coordinate formulation.

4.1. AUGMENTED METHODS


In the augmented formulation, the kinematic constraint equations are adjoined to the
system dynamic differential equations using the technique of Lagrange multipliers
[100]. The resulting system of algebraic equations can be solved for the reference
and elastic accelerations as well as the vector of Lagrange multipliers. Lagrange
multipliers can be used to determine the generalized constraint forces, while the
accelerations can be integrated numerically in order to determine the reference and
elastic coordinates and velocities. In the augmented formulation, the equations of
motion of the system are written in terms of Lagrange multipliers as
" #" # " T# " #
Mrr Mrf 
yr Qr qr
Mfr Mff yf
+
QTf
=
qf
;
212 A.A. SHABANA

where in this case, the vectors qr and qf contain the external, Coriolis and centrifu-
gal forces as well as the elastic forces. Differentiation of the constraint equations
twice with respect to time yields
" #

yr
[Qr Qf ] = Qc :
yf

Combining the preceding two equations, one obtains


2 32 3 2 3
Mrr Mrf QTr yr qr
6 76 7 6 7
4 Mfr Mff QTf 5 4  y f 5 = 4 qf 5 :
Qr Qf 0  Qc

This system can be solved using sparse matrix techniques for the reference and
elastic accelerations as well as the vector of Lagrange multipliers [80, 95]. The
augmented formulation leads to a system of differential and algebraic equations
which must be solved simultaneously. The solution of the differential and algebraic
equations can also be avoided in flexible multibody dynamics by using the recursive
methods.

4.2. RECURSIVE AND PROJECTION METHODS


In the recursive methods [15, 19, 69], the vector of absolute and orientation coor-
dinates can be expressed in terms of the independent joint and elastic variables.
By formulating the joint kinematic equations, one can write the vector of system
velocities in terms of the system independent joint and elastic velocities as

y_ = Hy_ i ;

where H is a velocity influence coefficient matrix that depends on the joint types.
Differentiating the preceding equation with respect to time, one obtains

y = Hyi + ;


where is a vector that absorbs terms which are quadratic in the velocities. The
system equations of motion can be written in terms of the absolute and elastic
coordinates as

My = q + fq ;

where q contains all forces except the constraint forces which are represented by
the vector fc . Using the preceding equations, premultiplying the equation of motion
by HT , and using the fact that

HT fc = 0;
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 213

one obtains the following reduced system of differential equations which does not
include the constraint joint forces:

Mi yi = qi ;

where subscript i refers to vectors or matrices associated with the system inde-
pendent coordinates. The preceding equation is the general form of the recursive
equations which have a dense and highly nonlinear inertia matrix.
In order to efficiently solve the acceleration equations, several recursive order
n formulations were also developed for flexible multibody systems [43]. Such
formulations exhibits a strong coupling between the independent joint and elastic
coordinates. Furthermore, the solution for the accelerations becomes dependent on
the number of elastic coordinates.
The basic equations used for developing the independent recursive equations
can be used to obtain a larger system of loosely coupled system of equations which
can be written as
2 32 3 2 3
M I 0 
y q
4 I 0 H 5 4 ,fc 5 = 4 5 :
6 76 7 6 7
0 HT 0 yi 0

This amalgamated formulation can be used to systematically decouple the joint


and elastic acceleration equations [84]. It can also be used to obtain a formulation
in which the solution for the acceleration equations becomes independent of the
number of elastic degrees of freedom.

5. Current and Future Research Directions


There are several research topics which are of current interest. Among these topics
are the study of the effect of geometric centrifugal stiffening [38, 51, 52, 99], the
analysis of the solutions obtained for multibody dynamic equations [53], and the
use of symbolic manipulations [23, 78]. There are also several numerical issues
associated with the computer simulation of flexible multibody systems. These
issues are related to the use of detailed finite element models that contain high
frequencies as well as the need to efficiently solve the differential and algebraic
equations of the flexible multibody systems. Another interesting and emerging
application of flexible multibody techniques is in the area of computer graphics
where physics-based models are developed to describe the deformation of graphics
objects [59, 60]. Computational flexible multibody techniques are used to predict
the displacements and forces required to change the object shapes. Important
potential applications of these physics-based graphic techniques are in the field of
bio-engineering.
In this section we select, for a more detailed discussion, several topics which
are of current interest in the area of flexible multibody dynamics.
214 A.A. SHABANA

5.1. ESTABLISHING THE RELATIONSHIPS BETWEEN DIFFERENT FORMALISMS


For a subject to reach a certain degree of maturity, it is important to have a
clear understanding of the relationship between different methodologies. There are
many formulations which are used in the computer simulation of multibody appli-
cations. There are no, however, guidelines or criteria that define the assumptions
and approximations underlying each method. This often leads to a confusion as
to which method is most suited for a particular application. With introducing, the
absolute nodal coordinate formulation, it is now possible to establish the relation-
ship between different formulations, and clearly define the assumptions used in
each of these methods. This important objective can be accomplished because in
the absolute nodal coordinate formulation, beams and plates can be considered
as isoparametric elements. Using this fact, the equivalence between the floating
frame of reference formulation and the absolute nodal coordinate formulation can
be demonstrated and used to examine the effect of using the consistent and lumped
mass distribution on the inertia representation of deformable bodies that undergo
large reference displacements. In order to demonstrate the equivalence of the float-
ing frame of reference formulation and the absolute nodal coordinate formulation,
the relationship between the absolute and local slopes must be first defined and then
used to establish the relationship between the coordinates used in the two different
formulations [93].

5.2. IMPACT AND CONTACT PROBLEM


Two basic approaches were used in the impact dynamics of flexible mechanical
systems. In the first approach, a pieced interval analysis method is used. In this
method, the impact duration is assumed to be small such that the system configura-
tion does not significantly change. At the instant of impact, the generalized impulse
momentum equations can be formulated. By including the restitution condition,
one obtains a number of equations equal to the number of the unknowns (veloc-
ities, joint reaction forces, and the impulse). The resulting generalized impulse
momentum equations and the restitution conditions can be solved for the jump
discontinuity in the system velocities as well as the joint reaction forces [39, 40,
44]. The validity of using the concept of the coefficient of restitution in the impact
analysis of constrained flexible mechanical systems was examined analytically,
numerically, and experimentally. Rismantab-Sany and Shabana [74] used a simple
impact model to examine analytically and numerically the validity of using the
generalized impulse momentum equations and the coefficient of restitution in the
impact analysis of constrained deformable bodies. Using the generalized impulse
momentum equations, a series solution was obtained for the jump discontinuity in
the velocity vector and the joint reaction forces. Also, a series solution was obtained
for the generalized impulse as a function of the number of modes. It was demon-
strated that as the number of modes approaches infinity, the generalized impulse
approaches zero. Yigit et al. [43, 44] developed a numerical and experimental
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 215

model to examine the validity of using the concept of the coefficient of restitution
in the impact analysis of flexible rotating beams. The equations of motion were
developed using the modal expansion method and experimentally measured data
were used to demonstrate that the generalized impulse momentum equations and
the coefficient of restitution can be used with confidence in the impact analysis of
flexible mechanical systems.
In the second approach used in the impact analysis of flexible mechanical sys-
tems, a continuous force model is used to represent the force of interaction between
the impact surfaces. A stiffness and damping coefficients are used in this force mod-
el to account for the impact surface compliance and energy dissipation during the
impact process. In some investigations, the stiffness and damping coefficients are
determined using a momentum balance approach [39, 40]. Unlike the generalized
impulse momentum approach where algebraic equations are used, in the continu-
ous force model approach, the system equations of motion are integrated during
the impact duration, and as a consequence, the integration of these equations can
be very inefficient if high stiffness coefficients are used in the continuous impact
force models.
Several issues in the impact analysis of flexible multibody systems which need
further consideration are:
1. Effect of the finite rotation on the impact induced waves in constrained flexible
mechanical systems. Our understanding of the change of the wave motion
as the result of the finite rotations will significantly contribute to a better
understanding of the dynamics of flexible multibody systems.
2. Experimental, analytical, and numerical study of the definition of the restitution
condition in flexible body dynamics. The relationships between the velocities
before and after impact and their dependence on the structural flexibility and
the locations of the impact regions on the flexible bodies need to be examined.
3. Development new methodologies that characterize the force of impact between
deformable bodies.

5.3. CONTROL-STRUCTURE INTERACTION

There are many fundamental issues associated with the control of flexible multibody
systems which differs significantly from the control of rigid body systems even
when the simplest open loop control strategies are used [13, 75]. For example,
the feed forward control of rigid body systems can be obtained by solving an
inverse dynamics problem which requires only the solution of system of algebraic
equations. This is not the case when flexible multibody systems are considered,
since the inverse dynamics problem requires the solution of a set of differential
equations even in the cases where finite dimensional models are used. The elastic
coordinates are set of independent variables which introduce many problems that
make the control of flexible multibody systems a challenging task that has been the
216 A.A. SHABANA

subject of many investigations. Among the topics which need further investigations
in the area flexible multibody control are:
1. The coupling between the independent joint and elastic degrees of freedom
and its effect on the control parameters.
2. The effect of the selection of the reference conditions and the associated mode
shapes on the design of the control systems. It is known in flexible multibody
dynamics that the natural frequencies of the linear problem do not have any
effect on the solution of the nonlinear multibody dynamic equations. The
system frequencies that define the control bandwidth, stability and response
characteristics need to be identified.
3. The nature of the control forces associated with the independent elastic degrees
of freedom. Any successful control strategy must take into consideration the
effect of these forces.
4. Effect of different multibody dynamics linearization schemes on the control
and mechatronics system parameters. There is a need for the development
of linearization schemes for flexible multibody systems that are efficient and
accurate enough for the use in practical control system applications.
5. Use of general purpose multibody codes in the control of flexible mechanical
systems. Many of flexible multibody simulation codes have the capabilities for
control analysis. These capabilities, however, are not fully exploited.

5.4. MODAL IDENTIFICATION METHODS


A literature survey indicates that while there is a very large number of papers
published on the use of the finite element method in flexible multibody dynamics,
very few investigations were devoted to the use of experimental modal analysis
techniques in flexible multibody simulations in particular [82], and the frequency
domain analysis of flexible mechanical systems in general [66–68]. Many of the
finite element modeling problems in flexible multibody systems were addressed and
solved. Several of the fundamental modeling problems that arise from the coupling
between experimental modal identification techniques and general purpose flexible
multibody computer programs have not been addressed. Some of the issues which
need to be addressed are:
1. Development of accurate methods for evaluating the inertia shape integrals
that appear in the floating frame of reference formulation of the flexible bodies
using the experimentally identified modal parameters. These shape integrals
which define the inertia characteristics of the deformable bodies need to be
accurately evaluated in terms of the measured data.
2. Development of measurement procedures by which the component modes
can be experimentally identified from the experimental testing of the entire
system. This can be one of the most challenging and interesting problems in
the analysis of flexible multibody systems which may require introducing new
measurement concepts for the experimental identification techniques.
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 217

3. Use the experimental data to identify more accurately the boundary conditions
of constrained deformable bodies and compare the obtained modal character-
istics with those obtained using the finite element method.

6. Conclusions
The subject of flexible multibody dynamics is an interdisciplinary field that encom-
passes several other subjects such as rigid body dynamics, continuum mechan-
ics, finite element method, and numerical, matrix and computer methods. Active
research in this field started in the early seventies and significant accomplishments
have been made in the last two decades. This results in a better understanding
of the dynamics and constrained motion of deformable bodies that undergo large
displacements. As demonstrated in this review article, an accurate modeling of the
dynamics of deformable bodies requires the use of nonlinear equations. Regardless
of the approach used, these equations exhibit a strong coupling between different
modes of displacements as the result of the finite rotations of the bodies.
Several methods that are based on different concepts were reviewed in this arti-
cle. Roughly speaking, the methods used thus far in flexible multibody dynamics
can be categorized into three different approaches. These are the floating frame
of reference formulation, the incremental finite element formulation, and the large
rotation vector formulation. Other procedures such as the linear theory of elasto-
dynamics and the finite segment methods can be considered as derivatives of the
above mentioned three basic approaches. The floating frame of reference formula-
tion is currently the most widely used method in the computer simulation of flexible
multibody systems. The floating frame of reference formulation has been used for
the most part in the large reference displacement and small deformation analysis.
The incremental finite element formulation has been successfully used in the large
deformation analysis of structural systems. Using this approach in multibody sim-
ulation, however, must be carefully examined since the incremental methods do
not lead to exact modeling of the rigid body dynamics when non-isoparametric
elements are used. Large rotation vector formulations had thus far limited success
in the simulation of flexible multibody applications. A fourth approach which is
recently introduced is the absolute nodal coordinate formulation. This approach
is conceptually different from the previous methods in the sense that there is no
infinitesimal or finite rotations are used as nodal coordinates. Global displacement
and slope coordinates are used as the nodal coordinates of the finite element. Using
the absolute nodal coordinate formulation, beams and plates can be treated as
isoparametric elements. The absolute nodal coordinate formulation can be used in
the large and small deformation analysis of flexible multibody systems. It can also
be effectively used in the analysis of curved structures.
Much has been accomplished in the last two decades in the field of flexible
multibody dynamics. While significant progress has been made, much remains to
be achieved in this very important area. In the author’s opinion, the field of flexible
218 A.A. SHABANA

multibody dynamics can reach a high degree of maturity if the relationships between
different formulations are established. This will allow us to define the coordinate
transformations that relate various formulations, and as a result, the assumptions
and approximations underlying each method can be clearly defined. It is believed
that the absolute nodal coordinate formulation can be effectively used to achieve
this important goal. Several other topics which can be the subject of future research
were identified in this review article. Examples of these topics are the impact-
contact dynamics, the integration of experimental modal analysis techniques and
flexible multibody methodologies, the control-structure interaction, the application
of flexible multibody techniques to computer graphics, and large deformation
problems.

Acknowledgement
This research was supported, in part, by the U.S. Army Research Office, Research
Triangle Park, NC.

References
1. Agrawal, O.P., and Shabana, A.A., ‘Dynamic analysis of multibody systems using component
modes’, Computer and Structures 21(6), 1985, 1301–1312.
2. Agrawal, O.P. and Shabana, A.A., ‘Application of deformable body mean axis to flexible
multibody dynamics’, Computer Methods in Applied Mechanics and Engineering 56, 1986,
217–245.
3. Ambrosio, J.A.C. and Nikravesh, P., ‘Elasto-plastic deformation in multibody dynamics’, Non-
linear Dynamics 3, 1992, 85–104.
4. Ambrosio, J.A.C. and Pereira, M.S., ‘Flexibility in multibody dynamics with applications to
crashworthiness’, in Computer-Aided Analysis of Rigid and Flexible Mechanical Systems, M.S.
Pereira and J.A.C. Ambrosio, Kluwer Academic Publishers, Dordrecht, 1994, pp. 199–232.
5. Ashley, H., ‘Observations on the dynamic behavior of large flexible bodies in orbit’, AIAA
Journal 5(3), 1967, 460–469.
6. Avello, A., de Jalon, G. and Bayo, E., ‘Dynamics of flexible multibody systems using Cartesian
co-ordinates and large displacement theory’, International Journal for Numerical Methods in
Engineering 32(8), 1991, 1543–1564.
7. Bakr, E.M. and Shabana, A.A., ‘Geometrically nonlinear analysis of multibody systems’, Com-
puters and Structures 23(6), 1986, 739–751.
8. Bakr, E.M. and Shabana, A.A., ‘Timoshenko beams and flexible multibody system dynamics,
sound and vibration’, Sound and Vibration 116(1), 1987, 89–107.
9. Belytschko, T. and Hsieh, B.J., ‘Nonlinear transient finite element analysis with convected
coordinates’, International Journal for Numerical Methods in Engineering 7, 1973, 255–271.
10. Belytschko, T. and Glaum, L.W., ‘Application of higher order corotational stretch theories to
nonlinear finite element analysis’, Computers and Structures 1, 1979, 175–182.
11. Belytschko, T. and Schwer, L., ‘Large displacement transient analysis of space frames’, Inter-
national Journal for Numerical Methods in Engineering 11, 1977, 65–84.
12. Benson, D.J. and Hallquist, J.D., ‘A simple rigid body algorithm for structural dynamics pro-
grams’, International Journal for Numerical Methods in Engineering 22, 1986, 723–749.
13. Boland, P., Samin, J.C. and Willems, P.Y., ‘Stability analysis of interconnected deformable
bodies in a topological tree’, AIAA Journal 12(8), 1974, 864–867.
14. Book, W.J., ‘Analysis of massless elastic chains with servo controlled joints’, ASME Journal of
Dynamic Systems, Measurement, and Control 101, 1979, 187–192.
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 219

15. Book, W.J., Recursive Lagrangian dynamics of flexible manipulator arms’, The International
Journal of Robotic Research 3, 1984, 87–101.
16. Cardona, A. and Geradin M., ‘Modeling of superelements in mechanism analysis’, International
Journal for Numerical Methods in Engineering 32(8), 1991, 1565–1594.
17. Cavin, R.K. and Dusto, A.R., ‘Hamilton’s principle: Finite element methods and flexible body
dynamics’, AIAA Journal 15(2), 1977, 1684–1690.
18. Chedmail, P., Aoustin, Y. and Chevallereau, C., ‘Modeling and control of flexible robots’,
International Journal for Numerical Methods in Engineering 32(8), 1991, 1595–1620.
19. Changizi, K. and Shabana, A.A., ‘A recursive formulation for the dynamic analysis of open
loop deformable multibody systems’, ASME Journal of Applied Mechanics 55, 1988, 687–693.
20. Chu, S.C. and Pan, K.C., ‘Dynamic response of a high speed slider crank mechanism with an
elastic connecting rod’, ASME Journal of Engineering for Industry 97, 1975, 542–550.
21. De Veubeke, B.F., ‘The dynamics of flexible bodies’, International Journal for Engineering
Science 14, 1976, 895–913.
22. Erdman, A.G. and Sandor, G.N., ‘Kineto-elastodynamics – A review of the state of the art and
trends’, Mechanism and Machine Theory 7, 1972, 19–33.
23. Fisette, P., Samin, J.C. and Willems, P.Y., ‘Contribution to symbolic analysis of deformable
multibody systems’, International Journal for Numerical Methods in Engineering 32(8), 1991,
1621–1636.
24. Flanagan, D.P. and Taylor, L.M., ‘An accurate numerical algorithm for stress integration with
finite rotations’, Computer Methods in Applied Mechanics and Engineering 62, 1987, 305–320.
25. Friberg, O., ‘A method for selecting deformation modes in flexible multibody dynamics’,
International Journal for Numerical Methods in Engineering 32(8), 1991, 1637–1656.
26. Frisch, H.P., ‘A vector dyadic development of the equations of motion for N-coupled flexible
bodies and point masses’, NASA TN D-7767, 1974.
27. Garcia de Jalon, J., Cuadrado, J., Avello, A. and Jimenez, J.M., ‘Kinematic and dynamic
simulation of rigid and flexible systems with fully Cartesian coordinates’, in Computer-Aided
Analysis of Rigid and Flexible Mechanical Systems, M.S. Pereira and J.A.C. Ambrosio (eds),
Kluwer Academic Publishers, Dordrecht, 1994, pp. 285–323.
28. Geradin, M., Cardona, A., Doan, D.B. and Duysens, J., ‘Finite element modeling concepts in
multibody dynamics’, in Computer-Aided Analysis of Rigid and Flexible Mechanical Systems,
M.S. Pereira and J.A.C. Ambrosio, Kluwer Academic Publishers, Dordrecht, 1994, pp. 233-284.
29. Gofron, M., ‘Driving elastic forces in flexible multibody systems’, Ph.D. Thesis, University of
Illinois at Chicago, 1995.
30. Ho, J.Y.L., ‘Direct path method for flexible multibody spacecraft dynamics’, Journal of Space-
craft and Rockets 14, 1977, 102–110.
31. Ho, J.Y.L. and Herber, D.R., ‘Development of dynamics and control simulation of large flexible
space systems’, Journal of Guidance, Control, and Dynamics 8, 1985, 374–383.
32. Hooker, W.W., ‘Equations of motion of interconnected rigid and elastic bodies’, Celestial
Mechanics 11(3), 1975, 337–359.
33. Hughes, P.C., ‘Dynamics of chain of flexible bodies’, Journal of Astronautical Science 27(4),
1979, 359–380.
34. Hughes, T.J.R. and Winget, J., ‘Finite rotation effects in numerical integration of rate constitutive
equations arising in large deformation analysis’, International Journal for Numerical Methods
in Engineering 15(12), 1980, 1862–1867.
35. Huston, R.L., ‘Multi-body dynamics including the effect of flexibility and compliance’, Com-
puters and Structures 14, 1981, 443–451.
36. Huston, R.L., ‘Computer methods in flexible multibody dynamics’, International Journal for
Numerical Methods in Engineering 32(8), 1991, 1657–1668.
37. Huston, R.L. and Wang, Y., ‘Flexibility effects in multibody systems’, in Computer-Aided
Analysis of Rigid and Flexible Mechanical Systems, M.S. Pereira and J.A.C. Ambrosio (eds),
Kluwer Academic Publishers, Dordrecht, 1994, pp. 351–376.
38. Kane, T.R., Ryan, R.R. and Banerjee, A.K., ‘Dynamics of a cantilever beam attached to a
moving base’, AIAA Journal of Guidance, Control, and Dynamics 10(2), 1987, 139–151.
220 A.A. SHABANA

39. Khulief, Y.A. and Shabana, A.A., ‘Dynamic analysis of constrained system of rigid and flexible
bodies with intermittent motion’, ASME Journal of Mechanisms, Transmissions and Automation
in Design 108(1), 1986, 38–45.
40. Khulief, Y.A. and Shabana, A.A., ‘Dynamics of multibody systems with variable kinematic
structure’, ASME Journal of Mechanisms, Transmissions and Automation in Design 108(2),
1986, 167–175.
41. Khulief, Y.A. and Shabana, A.A., ‘A continuous force model for the impact analysis of flexible
multibody systems’, Mechanism and Machine Theory 22(3), 1987, 213–224.
42. Koppens, W.P., ‘The dynamics of systems of deformable bodies’, Ph.D. Thesis, Technical
University of Eindhoven, 1989.
43. Lai, H.J., Haug, E.J., Kim, S.S. and Bae, D.S., ‘A decoupled flexible-relative co-ordinate
recursive approach for flexible multibody dynamics’, International Journal for Numerical
Methods in Engineering 32(8), 1991, 1669–1690.
44. Lankrani, H.M. and Nikravesh, P.E., ‘Canonical impulse momentum equations for impact
analysis of multibody systems’, ASME Journal of Mechanical Design 114, 1992, 180–186.
45. Laskin, R.A., Likins, P.W. and Longman, R.W., ‘Dynamical equations of a free-free beam
subject to large overall motions’, Journal of Astronautical Sciences 31(4), 1983, 507–528.
46. Likins, P.W., ‘Modal method for analysis of free rotations of spacecraft’, AIAA Journal 5(7),
1967, 1304–1308.
47. Likins, P.W., ‘Dynamic analysis of a system of hinge-connected rigid bodies with nonrigid
appendages’, International Journal of Solids and Structures 9, 1973, 1473–1487.
48. Likins, P.W., ‘Hybrid-coordinate spacecraft dynamics using large deformation modal coordi-
nates’, Astronautical Acta 18(5), 1973, 331–348.
49. Lowen, G.G. and Chassapis, C., ‘The elastic behavior of linkages: An update’, Mechanism and
Machine Theory 21(1), 1986, 33–42.
50. Lowen, G.G. and Jandrasits, W.G., ‘Survey of investigations into the dynamic behavior of
mechanisms containing links with distributed mass and elasticity’, Mechanism and Machine
Theory 7, 1972, 13–17.
51. Mayo, J., ‘Geometrically nonlinear formulations of flexible multibody dynamics’, Ph.D. Dis-
sertation, Department of Mechanical Engineering, University of Seville, Spain, 1993.
52. Mayo, J., Dominguez, J. and Shabana, A., ‘Geometrically nonlinear formulations of beams in
flexible multibody dynamics’, ASME Journal of Vibration and Acoustics, 1995 (to appear).
53. Meijaard, J.P., ‘Direct determination of periodic solutions of the dynamical equations of flexible
mechanisms and manipulators’, International Journal for Numerical Methods in Engineering
32(8), 1991, 1691–1710.
54. Meirovitch, L., ‘A new method of solution of the eigenvalue problem for gyroscopic systems’,
AIAA Journal 12, 1974, 1337–1342.
55. Meirovitch, L., ‘A modal analysis for the response of linear gyroscopic systems’, ASME Journal
of Applied Mechanics 42, 1975, 446–450.
56. Meirovitch, L., ‘A stationary principle for the eigenvalue problem for rotating structures’, AIAA
Journal 14, 1976, 1387–1394.
57. Melzer, F., ‘Symbolic computations in flexible multibody systems’, in Proceedings of the NATO
Advanced Study Institute on the Computer Aided Analysis of Rigid and Flexible Mechanical
Systems, Vol. 2, Troia, Portugal, June 26–July 9, 1993, pp. 365–381.
58. Melzer, F., 1994, Symbolisch-Numerische Modellierung Elastischer Mehrkorpersysteme mit
Anwendung auf Rechnerische Lebensdauervorhersagen, Fortschr.-Ber., VDI Reihe 20, Nr. 139,
VDI-Verlag, Dusseldorf, Germany, 1994.
59. Metaxas, D.N., Physics-Based Deformable Models: Applications to Computer Vision, Graphics
and Medical Imaging, Kluwer Academic Publishers, Dordrecht, 1997.
60. Metaxas, D., and Koh, Eunyoung, ‘Flexible multibody dynamics and adaptive finite element
techniques for model synthesis and estimation’, Computer Methods in Applied Mechanics and
Engineering 136, 1996, 1–25.
61. Milne, R.D., ‘Some remarks on the dynamics of deformable bodies’, AIAA Journal 6(3), 1968,
556–558.
FLEXIBLE MULTIBODY DYNAMICS: REVIEW 221

62. Modi, V.J., Suleman, A., Ng, A.C. and Morita, Y., ‘An approach to dynamics and control of
orbiting flexible structures’, International Journal for Numerical Methods in Engineering 32(8),
1991, 1727–1748.
63. Nikravesh, P.E. and Ambrosio, J.A.C., ‘Systematic construction of equations of motion for
rigid-flexible multibody systems containing open and closed kinematic loops’, International
Journal for Numerical Methods in Engineering 32(8), 1991, 1749–1766.
64. Nikravesh, P., Chung, I. and Bendict, R.L., ‘Plastic hinge approach to vehicle crash simulation’,
Computers & Structures 16, 1983, 395–400.
65. Park, K.C., Downer, J.D., Chiou, J.C. and Farhat, C., ‘A modular multibody analysis capability
for high precision, active control and real time applications’, International Journal for Numerical
Methods in Engineering 32(8), 1991, 1767–1798.
66. Pascal, M., ‘Dynamics analysis of a system of hinge-connected flexible bodies’, Celestial
Mechanics 41, 1988, 253–274.
67. Pascal, M., ‘Dynamical analysis of a flexible manipulator arm’, Acta Astronautica 21(3), 1990,
161–169.
68. Pascal, M. and Sylia, M., ‘Dynamic model of a large space structure by a continuous approach’,
La Recherche Aérospatiale 2, 1993, 67–77.
69. Pereira, M.S. and Proenca, ‘Dynamic analysis of spatial flexible multibody systems using
joint co-ordinates’, International Journal for Numerical Methods in Engineering 32(8), 1991,
1799–1812.
70. Rankin, C.C. and Brogan, F.A., ‘An element independent corotational procedure for the treat-
ment of large rotations’, ASME Journal of Pressure Vessel Technology 108, 1986, 165–174.
71. Rauh, J., Ein Beitrag zur Modellierung Elastischer Balkensysteme, Fortschr.-Ber. VDI Reihe
18, Nr. 37, VDI-Verlag, Dusseldorf, Germany, 1987.
72. Rauh, J. and Schiehlen, W., ‘A unified approach for the modeling of flexible robot arms’,
in Proceedings of the 6th CISM-IFToMM Symposium on Theory and Practice of Robots and
Manipulators, Cracow, September 9–12, 1986.
73. Rauh, J. and Schiehlen, W., ‘Various approaches for modeling of flexible robot arms’, in
Proceedings of Euromech Colloquium 219, Refined Dynamical Theories of Beams, Plates, and
Shells, and Their Applications, Kassel, Germany, September 23–26, 1986.
74. Rismantab-Sany, J. and Shabana, A.A., ‘On the use of momentum balance in the impact analysis
of constrained elastic systems’, ASME Journal of Vibration and Acoustics 112(1), 1990, 119–
126.
75. Roberson, R.E., ‘A form of the translational dynamical equation for relative motion in systems
of many non-rigid bodies’, Acta Mechanica 14, 1972, 297–308.
76. Sadler, J.P. and Sandor, G.N., ‘A lumped parameter approach to vibration and stress analysis of
elastic linkages’, ASME Journal of Engineering for Industry 95, 1973, 549–557.
77. Schiehlen, W.O. and Rauh, J., ‘Modeling of flexible multibeam systems by rigid-elastic super-
elements’, Revista Brasiliera de Ciencias Mecanicas 8(2), 1986, 151–163.
78. Schiehlen, W., ‘Symbolic computations in multibody systems, in Computer-Aided Analysis
of Rigid and Flexible Mechanical Systems, M.S. Pereira and J.A.C. Ambrosio (eds), Kluwer
Academic Publishers, Dordrecht, 1994, pp. 101–136.
79. Schiehlen, W.O., ‘Multibody system dynamics – Roots and perspectives’, Multibody System
Dynamics 1, 1997, 149–188.
80. Shabana, A. and Wehage, R.A., ‘Coordinate reduction technique for transient analysis of spatial
substructures with large angular rotations’, Journal of Structural Mechanics 11(3), 1983, 401–
431.
81. Shabana, A.A., ‘Automated analysis of constrained inertia-variant flexible systems’, ASME
Journal of Vibration, Acoustic, Stress, and Reliability in Design 107(4), 1985, 431–440.
82. Shabana, A.A., ‘Dynamics of inertia variant flexible systems using experimentally identified
parameters’, ASME Journal of Mechanisms, Transmission, and Automation in Design 108,
1986, 358–366.
83. Shabana, A., Dynamics of Multibody Systems, John Wiley & Sons, New York, 1989.
84. Shabana, A., ‘Constrained motion of deformable bodies’, International Journal for Numerical
Methods in Engineering 32(8), 1991, 1813–1831.
222 A.A. SHABANA

85. Shabana, A.A., ‘Computer implementation of flexible multibody equations’, in Computer-Aided


Analysis of Rigid and Flexible Mechanical Systems, M.S. Pereira and J.A.C. Ambrosio, Kluwer
Academic Publishers, Dordrecht, 1994, pp. 325–349.
86. Shabana, A.A., ‘Incremental finite element formulation and exact rigid body inertia’, in Pro-
ceedings of the 1995 ASME Design Automation Conference, Boston, MA, September 1995,
pp. 617–623.
87. Shabana, A.A., ‘Resonance conditions and deformable body coordinate systems’, Journal of
Sound and Vibration 192(1), 1996, 389–398.
88. Shabana, A.A., ‘Finite element incremental approach and exact rigid body inertia’, ASME
Journal of Mechanical Design 118(2), 1996, 171–178.
89. Shabana, A.A., ‘An absolute nodal coordinate formulation for the large rotation and deformation
analysis of flexible bodies’, Technical Report No. MBS96-1-UIC, Department of Mechanical
Engineering, University of Illinois at Chicago, March 1996.
90. Shabana, A.A., Vibration of Discrete and Continuous Systems, 2nd edn, Springer-Verlag, New
York, 1997.
91. Simo, J.C., ‘A finite strain beam formulation. The three-dimensional dynamic problem, Part I’,
Computer Methods in Applied Mechanics and Engineering 49, 1985, 55–70.
92. Simo, J.C. and Vu-Quoc, L., ‘A three-dimensional finite strain rod model, Part II: Computational
aspects’, Computer Methods in Applied Mechanics and Engineering 58, 1986, 79–116.
93. Shabana, A. and Schwertassek, R., ‘Floating frame of reference formulation and definition of
the nodal coordinates’, International Journal of Non-Linear Mechanics, 1997 (in press).
94. Simo, J.C. and Vu-Quoc, L., ‘On the dynamics of flexible beams under large overall motions –
The plane case: Parts I and II’, ASME Journal of Applied Mechanics 53, 1996, 849–863.
95. Song, J.O. and Haug, E.J., ‘Dynamic analysis of planar flexible mechanisms’, Computer Meth-
ods in Applied Mechanics and Engineering 24, 1980, 359–381.
96. Sunada, W. and Dubowsky, S., ‘The application of the finite element methods to the dynamic
analysis of flexible spatial and co-planar linkage systems’, ASME Journal of Mechanical Design
103(3), 1981, 643–651.
97. Sunada, W. and Dubowsky, S., ‘On the dynamic analysis and behavior of industrial robotic
manipulator with elastic members’, ASME Journal of Mechanisms, Transmissions, and Automa-
tion in Design 105(1), 1983, 42–51.
98. Turcic, D.A. and Midha, A., ‘Dynamic analysis of elastic mechanism systems, Parts I & II’,
ASME Journal of Dynamic Systems, Measurement, and Control 106, 1984, 249–260.
99. Wallrapp, O. and Schwertassek, R., ‘Representation of geometric stiffening in multibody system
simulation’, International Journal for Numerical Methods in Engineering 32(8), 1991, 1833–
1850.
100. Wehage, R.A., ‘Generalized coordinate partitioning in dynamic analysis of mechanical systems’,
Ph.D. Thesis, The University of Iowa, Iowa City, Iowa, 1980.
101. Winfrey, R.C., ‘Elastic link mechanism dynamics’, ASME Journal of Engineering for Industry
93, 1971, 268–272.
102. Winfrey, R.C., ‘Dynamic analysis of elastic link mechanisms by reduction of coordinates’,
ASME Journal of Engineering for Industry 94, 1972, 577–582.
103. Yigit, A.S., Ulsoy, A.G. and Scott, R.A., ‘Dynamics of a radially rotating beam with impact,
Part I: Theoretical and computational model’, ASME Journal of Vibration and Acoustics 112,
1990, 65–70.
104. Yigit, A.S., Ulsoy, A.G. and Scott, R.A., ‘Dynamics of a radially rotating beam with impact,
Part II: Experimental and simulation results’, ASME Journal of Vibration and Acoustics 112,
1990, 71–77.
105. Zienkiewicz, O.C., The Finite Element Method, McGraw-Hill, New York, 1979.

You might also like