You are on page 1of 8

Application of the Absolute

Nodal Coordinate Formulation to


Large Rotation and Large
Deformation Problems
There are three basic finite element formulations which are used in multibody dynam-
A. A. Shabana ics. These are the floating frame of reference approach, the incremental method and
the large rotation vector approach. In the floating frame of reference and incremental
formulations, the slopes are assumed small in order to define infinitesimal rotations
H. A. Hussien that can be treated and transformed as vectors. This description, however, limits the
use of some important elements such as beams and plates in a wide range of large
Department of Mectianical Engineering, displacement applications. As demonstrated in some recent publications, if infinitesi-
University of lliinois at Ctiicago, mal rotations are used as nodal coordinates, the use of the finite element incremental
842 West Taylor Street formulation in the large reference displacement analysis does not lead to exact
Ctiicago, IL 60607-7022
modeling of the rigid body inertia when the structures rotate as rigid bodies. In this
paper, a simple non-incremental finite element procedure that employs the mathemati-
cal definition of the slope and uses it to define the element coordinates instead of the
J. L. Escalona infinitesimal and finite rotations is developed for large rotation and deformation
Department of IVIectianical Engineering, problems. By using this description and by defining the element coordinates in the
University of Seville,
global system, not only the need for performing coordinate transformation is avoided,
Av. Reina Mercedes s/n
41012 Seville, Spain but also a simple expression for the inertia forces is obtained. The resulting mass
matrix is constant and it is the same matrix that appears in linear structural dynamics.
It is demonstrated in this paper that this coordinate description leads to exact model-
ing of the rigid body inertia when the structures rotate as rigid bodies. Nonetheless,
the stiffness matrix becomes nonlinear function even in the case of small displace-
ments. The method presented in this paper differs from previous large rotation vector
formulations in the sense that the inertia forces, the kinetic energy, and the strain
energy are not expressed in terms of any orientation coordinates, and therefore, the
method does not require interpolation of finite rotations. While the use of the formula-
tion is demonstrated using a simple planar beam element, the generalization of the
method to other element types and to the three dimensional case is straightforward.
Using the finite element procedure presented in this paper, beams and plates can be
treated as isoparametric elements.

1 Introduction In flexible body dynamics, several formulations have been


developed to study the nonlinear motion of structural compo-
The finite elements used in the static and dynamic analysis nents that undergo large displacements (Belytschko and Hsieh,
of structures can be categorized into two groups. The first group 1973; Flanagan and Taylor, 1987; Hughes and Winget, 1980;
has elements which have only displacements as nodal coordi- and Simo and Vu-Quoc, 1986; Cardona and Geradin, 1991;
nates. Examples of these elements are the isoparametric planar Geradin et al., 1994). Among these approaches are the-floating
triangular and rectangular elements and the isoparametric spatial frame of reference method, the incremental finite element
solid and tetrahedral elements. These isoparametric elements method, and the large rotation vector method (Shabana, 1997).
can be efficiently used in the large deformation analysis since In the floating frame of reference approach, a coordinate system
they lead to a constant mass matrix when the element coordi- is assigned to each deformable body, and the large displacement
nates are defined in the global system. For this reason, the use of this coordinate system is described using a mixed set of
of these elements do not represent a challenging problem, and Cartesian and orientation coordinates. The deformation of the
as a consequence, these elements are not the subject of extensive body is defined with respect to its coordinate system using a
research activities. The second group, on the other hand, has modal representation or the finite element method. The resulting
elements in which slopes or rotations are used as nodal coordi- dynamic equations are highly nonlinear and exhibit a strong
nates. Examples of these elements are beams, plates and shells. nonlinear inertia coupling between the reference motion and the
These elements, in the classical finite element literature, are not elastic deformation of the body. The solution of this system
considered as isoparametric elements, and their use in the large requires the LU factorization of a nonlinear mass matrix, and
deformation analysis has been the subject of extensive research the inertia forces include Coriolis and centrifugal forces which
activities for many years. are quadratic in the velocities. This approach, however, remains
the most widely used approach in flexible multibody dynamics
(Likins, 1974; Shabana, 1998).
Contributed by the Design Automation Committee for publication in tlie JOUR-
NAL OF MECHANICAL DESIGN. Manuscript received April 1997. Associate Techni-
In the incremental finite element formulations, it is assumed
cal Editor: H. Lankavani. that the element shape functions and the nodal coordinates can
describe infinitesimal rotations. Infinitesimal rotations are used
as nodal coordinates, and therefore, the set of nodal coordinates

188 / Vol. 120, JUNE 1998 Copyright 1998 by ASME Transactions of the ASME

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


can be treated and transformed as vectors. The equations of we consider the uniform slender beam shown in Fig. I. The
motion are first defined in an element coordinate system, and beam has length I, cross sectional area a, mass density p, vol-
then transformed using orthogonal matrices in order to define ume V, and mass m. The coordinate system of this beam element
the equations in the inertial frame of reference. As demonstrated is assumed to be initially attached to its left end which is defined
recently (Shabana, 1996a), the incremental finite element ap- by point O as shown in the figure. The shape function of this
proach, which is often used to solve large rotation problems, element is assumed to be
does not lead to exact modeling of the inertia of simple struc-
tures when they rotate as rigid bodies because the physical nodal 1 - 3 ^ ' -)- 2^3 0
coordinates can not be used to describe large rotations in the 0 1 - 3 ^ ' -H 2 C '
case of beams and plates. Slopes are assumed to be small and
are used to describe infinitesimal rotations that can be treated
/(e - le + e) o
and transformed as vectors. This leads to a linearization of the
0 /(e - 2e + e) (1)
3C - 2C 0
equations of motion of rotating structures, and as such the use
0 2e - 2^
of the incremental formulation does not lead to the correct rigid
body motion. ne - e) 0
The third approach which has been proposed recently is the
0 ne - e)
large rotation vector. In this approach, large rotation coordinates where (, = xll. In Eq. (1), the two components of the displace-
are used as the nodal coordinates which are defined in the inertial ment are interpolated using the same polynomial since the coor-
frame. The use of this approach leads to a simple expression for dinates are defined in the global system. The vector of nodal
the kinetic energy, and the nonlinearities are transformed to the coordinates associated with the shape function of Eq. (1) is
stiffness matrix (Simo and Vu-Quoc, 1986). For example, in the
work of Simo and Vu-Quoc (1986), the kinetic and strain energies e = [e, 62 fij 64 65 e? esJ (2)
are expressed in terms of a large rotation coordinate. In order to
solve the resulting equations, an interpolation function is used to where ei and 2 are the translational coordinates at the node at
describe the finite rotation coordinates. In the three dimensional O, es and ef, are the translational coordinates at the node at A,
analysis, interpolation of finite rotations must be carefully handled fij and 64 are the spatial derivatives of the displacements of the
since finite rotations can not be treated as vectors. More signifi- node at O defined in the XY coordinate system, and ej and eg
cantiy, in many of the finite rotation formulations, the absolute are the spatial derivatives of the displacements of the node at
slopes and rotations of the element are treated as independent A defined in the XF coordinate system. An arbitrary rigid body
variables, leading to excessive shear forces. In such formulations, displacement of the beam is defined by the translation R =
the large rotation of the cross section of the element has two [R^ R^y of the reference point O, and a rigid body rotation
independent representations when the finite element approxima- 9. As the result of this arbitrary rigid body displacement, the
tion is used. One representation is defined by the finite rotation vector of nodal coordinates e can be defined in the global coordi-
coordinate and the other is defined by the derivatives of the dis- nate system (Shabana, 1996a) as
placement coordinates. This redundant representation remains a
problem even in the cases where the independent shear angle is e = [R, Ry cos I , + /cosS Ry + lsinff cose sin S]'
introduced (Shabana, 1998).
The non-incremental absolute nodal coordinate formulation (3)
used in this paper for the large rotation and deformation analy-
sis is conceptually different from the three approaches pre- Using Eqs. (1) and (3), it follows that the global position vector
viously mentioned, since infinitesimal or finite rotations are r of an arbitrary point on the beam can be written as
not used as nodal coordinates. In this formulation, the element
nodal coordinates are described in terms of absolute location Rj, + X cos
Se = (4)
coordinates and global slopes. The nodal coordinates are de- R, + X sin (
fined in the global coordinate system, and as such, a simple
expression for the inertia forces is obtained. Using this new
procedure (Shabana, 1996b), beams and plates can be treated which demonstrates that the element shape function and the
as isoparametric elements even in the cases in which conven- nodal coordinates can describe an arbitrary rigid body displace-
tional element shape functions are used. The resulting mass ment provided that the coordinates are defined in the global
matrix obtained in this paper is constant and it is the same coordinate system and the slopes are defined in terms of trigono-
matrix that arises in linear structural dynamics. The use of metric functions.
the absolute nodal coordinate formulation in large rotation
problems is demonstrated in this paper. The extension of the Rigid Body Inertia. Since exact representation of the posi-
method to three dimensional elements is straight forward as tion of an arbitrary point on the element, as a result of an
recently demonstrated (Shabana and Christensen, 1997). arbitrary rigid body displacement, can be obtained using the
element shape function and the nodal coordinates, it can be
demonstrated that exact modeling of the rigid body inertia can
2 Background be obtained using the shape function and the element nodal
Classical finite element shape functions of beams can be used coordinates. If we use point O as the reference point, the time
to describe an arbitrary rigid body displacements provided that derivatives of the nodal coordinates defined in Eq. (3) are

e = [R, Ry ' sm (R.-WsmO) {Ry +19 cos 6) -9 sin 9 9 cos 9V (5)

the following conditions are satisfied: (1) The coordinates of The kinetic energy of the element can be written as
the element are defined in an inertial coordinate system, and
(2) No infinitesimal rotations are used as nodal coordinates,
and instead global slopes that are obtained as the derivatives of
p.W.^J^,
pr'idV = ^e'l \ pS^'SdV (6)

the displacements are used. In order to demonstrate this fact. which can be written as

Journal of Mechanical Design JUNE 1998, Vol. 120 / 189

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Current configuration

Initial configuration
Fig. 1 Large displacement of beams

T = ie^M/^e (7) Using this coordinate transformation, one can show that the
mass matrix of the rigid beam is

where
1 0 sin :
2
Jv
M = B'^MffB = m cos 0 (11)
13 11/ 9 -13/
0 0 0 0
35 210 70 420 I . I
13 11/ 9 -13/ sin 6/ - cos 3
0 0 0 0 2 2
35 210 70 420
which is the exact form of the mass matrix in the case of rigid
11/ 13/ body displacement. Therefore, an exact modeling of the rigid
0 0 0 0
210 105 420 140 body motion can be obtained using the element shape function
11/ J^ 13/ ^ and the nodal coordinates.
0 0 0 0
210 105 420 140
9 13/ 13 -11/ 3 Absolute Nodal Coordinate Formulation
0 0 0 0
70 420 35 210 In the case of deformable bodies, we define the nodal coordi-
9 13/ 13 -11/ nates in a fixed global frame of reference. In this case, the large
0 0 0 0
70 420 35 210 displacement, including the rotation, is described by the element
shape function as
13/ 0
-/^
0
-11/
0 J^ 0
420 140 210 105
r = Se (12)
-13/ -11/
0 0 0 0
420 140 210 105
It is assumed that the shape function S has a complete set of rigid
(8) body modes and the vector of nodal coordinates e is defined in
the global system.
is the symmetric constant mass matrix of the element. The mass
matrix of the rigid beam can be obtained from the preceding Inertia Forces. The kinetic energy of the body is defined,
equations by writing in this case, as

e = Bq (9)
pt^tdV = '^e^'M.
i//= (13)
Jv
where q = [i?^ Ry 9]^, and
where r is the global position vector of an arbitrary point on
1 0 0 0 1 0 0 0 the element, as defined in Eq. (12) and M//is the constant mass
0 1 0 0 0 1 0 0 matrix defined in the preceding section. Since the mass matrix
0 0 -sin^ cos d -/sinS /cos 6 -sin 6 cos in this case is constant, the inertia forces are linear in the acceler-
ation and do not contain any quadratic velocity terms.
The mass moments of inertia of the straight and curved struc-
(10) tural systems can be evaluated using the matrix M// of Eq. (13)

190 / Vol. 120, JUNE 1998 Transactions of the ASME

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Table 1 Mass moments of inertia of different shapes Since rigid body rotations produce zero strains, and since
the components of the unit vectors i and j are time depen-
^~~_ NO, OP
EXACT dent, a simple expression for the strain energy U can be
BODY
^~~-..^ELEMENTS
^~~~~~,.,^__^
5 10 20 30 INERTIA
written as

duiV
V Ea + EI dx (18)
22.9755 22.9756 22.9765 22.9765 22.9765 2 Jo dx)

M
where E is the modulus of elasticity, / is the second moment
of area and a is the cross-sectional area. Equation (18), after
16.3685 16.3686 16.3586 16.3586 16.3586 using the expressions for the longitudinal and the transverse
displacements, can be written as

Lli 4.0900 4.0900 4.0900 4.0900 4.0900


U = WK, //*=
(19)

where in this case the matrix K// is a nonlinear function of time.

0.
While in Eq. (18), the small strain assumptions are used, in
the case of large deformation of the element, another expression
for the strain energy must be used.
1.6201 1.7380 1.7455 1.7458 1.7460
Equation of Motion. In the non-incremental absolute nodal
coordinate formulation, the equation of motion of the element
takes the following simple form

as e'^Mz/C. Table 1 shows the results obtained for the mass Mfft + K//e = Q (20)
moments of inertia using this equation for different structural
shapes using different numbers of elements. Crucial in the use where Q is the vector of generalized nodal forces. Note that
of this equation is the new interpretation of the vector of nodal in this formulation, there is no quadratic velocity inertia
coordinates using the global slopes. The structures shown in forces because the mass matrix is constant. Also, since the
Table 1 are assumed to have a total length of 2.4 m and cross mass matrix is constant, the accelerations can be efficiently
sectional area of 0.0018 m^. The structures are assumed to be determined and integrated forward in time to determine the
made of aluminum which has mass density p = 2770 Kg/m' coordinates and velocities (Belytschko and Hughes, 1983).
(Shabana, 1997). In order to numerically solve the preceding differential
equations, the initial nodal coordinates and slopes must be
Elastic Force. Since the mass matrix is constant, one determined. The slopes in the undeformed configuration can
expects nonlinearities in the formulation of the stiffness coef- be determined using simple rigid body kinematics. For in-
ficients. If we select point O as the reference point, the com- stance, if the beam element axis in the undeformed state
ponents of the relative displacement of point A with respect makes an angle a with the global X axis, the position of an
to point O can be defined in the inertial coordinate system arbitrary point on the element is defined using simple rigid
as body kinematics as

(S, - S,o)e X cos a


(14) (21)
(S2 820)6 X sin a

where Sj and S2 are the rows of the element shape function The value of the slope at this initial undeformed configuration
matrix, and Sio and S20 are the rows of the shape function can be obtained as
matrix defined at the reference point O. Since in the technical
beam theory, the displacements are defined in the element coor- dry . dvx
dinate system, one needs to define the longitudinal and trans- - = sin a, = cos a (22)
verse displacements of the beam. Let i be a unit vector along dx dx
a selected beam axis defined by the vector
In this case, the initial conditions for the nodal coordinates e^,
64, e^ and eg are
i = [4 (15)
Ir^i - to]
63 = 67 = cos a, 64 = eg = sm a (23)
A unit vector j perpendicular to i can be obtained as
A similar procedure can be used to determine the initial con-
j = [;, 7;]^ = k x i (16) ditions for the slopes of other element types.

where k is unit vector along the Z axis. Then, the longitudi-


nal and transverse deformations of the beam can be defined 4 Stiffness Shape Integrals
as In the non-incremental absolute nodal coordinate formula-
tion presented in this paper, the mass matrix is constant, and
U 1 X it is the same matrix that appears in linear structural dynam-
= = = (17) ics. The stiffness matrix, on the other hand, becomes nonlin-
ear function even in the cases of linear elastic problems. It

Journal of Mechanical Design JUNE 1998, Vol. 120 / 191

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


is clear from the analysis presented in this investigation that Effect of the Gravity Forces. In some applications, the
the calculations of the stiffness matrix in the case of the gravity force is introduced as a concentrated force acting at the
linear elastic analysis of beams that undergo large rotations center of mass of the finite element. It can be demonstrated that
requires the evaluation of the following integrals: such a description of the gravity force does not lead to the

d'S2Yfd'S2

A] = Ea

Using the shape function of Eq. (1), the explicit form of the correct motion of the element. Using the shape function of Eq.
matrices and the vectors given in Eq. (24) can be determined (1) and the virtual work, it can be shown that the generalized
(Shabana, 1996b). nodal forces associated with the element nodal coordinates re-
Using the expression for the strain energy and a beam element sulting from the use of the concentrated gravity force can be
shape function, the vector of the element generalized elastic written as
forces can be obtained as
F . = mg i 0 - 0 i 0 i (27)
j = KffC = Aei? + A22e/J + (A12 + A j O e y , + B e i ^ 2 2 8

Where g is the gravity constant.


+ 82267J + (B,2 + B2i)ejjy - Ai4 - Aiiy Using the same shape function, it can be shown that the
generalized forces associated with the nodal coordinates of the
+ ( e^Aueix + -e^(Ai2 + A2i)eiy- A[e beam element as the result of using a distributed gravity force
are

0
+ ( e''A22e;y + -e''(Ai2 + A2i)e!V-Ai'e)(--^ F,
-I p[S[ SJ] dV

1 / 1 /
+ ( e ^ B e y , + ie^(B,2 + B 2 , ) e ; , ) ( ^ mg\ 0 - 0 0 - - 0 (28)
2 12 2 12

djy
In order to examine the effect of the concentrated and the
+ I e''B22ej, + - e'"(B,2 + B2i)e7' (25) distributed gravity forces on the dynamics of the beam element
de using the absolute nodal coordinate formulation, the free falUng
of a beam of length 2.4 m, cross sectional area of 0.0018 m^,
The matrix equation of motion of the element can then be mass density of 2770 kg/m^ and a modulus of elasticity 0.700
written as X 10" N/m^ is considered. Figure 2 shows the results obtained
using the concentrated and distributed gravity force models.
M//e = QIff (26) The results presented in this figure show that the distributed
gravity model gives an accurate prediction of the beam dynam-
where the generaUzed force vector Q// contains the external and ics since the beam does not experience any deformation and
the elastic forces. The structure mass matrix and the structure the displacements of all the points are equal to -jgt^, where t
force vector can be obtained by assembling the element mass is the time. The concentrated gravity force model, on the other
matrices and the element generalized force vectors using a stan- hand, leads to incorrect results in the case of free falling since
dard finite element procedure. this model results in the deformed configurations of the beam
shown in Fig. 2.
5 Numerical Results Free Falling of a Flexible Pendulum. The second example
The use of the absolute nodal coordinate formulation in the considered in this section is the free falling of the flexible pendu-
large rotation and large deformation problems is demonstrated lum shown in Fig. 3. The pendulum is assumed to fall under
in this section. Several examples will be considered and the the effect of gravity. The beam has a length of 1.2 m, a cross
definition of the forces in this finite element procedure will be sectional area of 0.0018 m^, a second moment of area 1.215 X
discussed. 10"^ m'', a mass density of 5540 Kg/m^ and a modulus of

192 / Vol. 120, JUNE 1998 Transactions of the ASME

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


aoi
0 -t-aoos
? -0.01 -t"ao4s
I -0.02 -t-aoss
t -0.03 -t-0.08

I
> -0.04

0.0S
-0.1 as 0.8 1,3 1.8 2.3 X8 3.3 3.3
-0.0 X displacement (m) X displicement (m)

(a) Distributed (b) Concentrated


Fig. 2 Effect of using distributed and concentrated gravity force model

elasticity of 0.700 X 10*^ N / m ^ The beam is divided into 30 Large Deformation. The use of the non-incremental abso-
two dimensional beam elements whose shape function is defined lute nodal coordinate formulation in the large rotation analysis
byEq. (1). of flexible bodies was demonstrated by the pendulum example
Note that the beam is made of a relatively flexible and heavy previously discussed in this section. In the remainder of this
material, therefore, one may expect large deformations due to section, we consider another simple example to demonstrate the
the free falling. Figure 4 shows the deformed shape of the use of this finite element procedure in the analysis of large
pendulum at different time steps, while Fig. 5 shows the trajec- deformations. To this end, we consider the cantilever beam
tory of the pendulum tip and its comparison with the trajectory shown in Fig. 6. The beam is assumed to be subjected to a
of the tip of a rigid pendulum, which follows an exact circle. vertical concentrated force defined in Fig. 7. The cantilever
The results presented in Fig. 4 show that initially the part of beam has a length of 2.4 m, a cross sectional area of 0.0018
the beam away from the pin joint appears to fall undeformed m^, a second moment of area 1.215 X 10"* m", a mass density
as a rigid body since this part is not immediately influenced by of 2770 Kg/m^ a modulus of elasticity of 1.000 X 10" N/m^
the constraint forces of the pin joint. and the value F of the concentrated force is 0.01 N. The beam

P(x)= pgA

-1.5 -0.5 0.5 1.5


X
Fig. 3 Initial configuration of the flexible pendulum
Fig. 5 Trajectory of the tip of the pendulum: ( - - - ) rigid b e a m ; ( )
flexible b e a m

Y
J

F(t)

-1.4
i1 X
wmmm
>r

/=2.4m

Fig. 4 Deformed configuration of the pendulum at different points in


time Fig. 6 Cantilever b e a m subject to concentrated force at the tip

Journal of Mechanical Design JUNE 1998, Vol. 120 / 193

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


FI2 (l-cos(t/))if M i s 0
if ?>ls Ti -0.5
c n
l-H
-1 -

DC -1 5
^ -9 H
T3

Is Time ^ -2.5
+j
O
Fig. 7 Concentrated force
-3
-3 5

is divided into 30 elements. Figure 8 shows the deformed beam 0.2 0.4 0.6 0.8
at different points in time. The large deformation results pre-
sented in this figure clearly demonstrate the effect of the elastic Time (s)
coupling between the longitudinal and transverse displacements
of the beam. This effect becomes more significant as the defor- Fig. 9 Angular orientation of pendulum: (- -) absolute nodal coordi-
mation becomes large. nate formulation; (---) floating frame of reference

Comparison with Existing Formulations. In order to


check the accuracy of the results obtained using the absolute avoid the linearization problem resulting from the use of the
nodal coordinate formulation, another simple pendulum exam- finite element incremental approach, infinitesimal rotations are
ple is used. The results of the absolute nodal coordinate formula- not used as nodal coordinates. The finite element coordinates
tion are compared with the results obtained using the floating including the slopes are defined in the inertial frame. It is dem-
frame of reference approach which is the most widely used onstrated that such a definition of the coordinates, which can
method in flexible multibody dynamics. The floating frame of be used to describe an arbitrary rigid body displacement, leads
reference formulation is used for the large reference rotation- to the exact modeling of the rigid body dynamics when the
small deformation problems. For this reason, we choose a pen- structures rotate as rigid bodies. Using the absolute nodal coor-
dulum system with a relatively high stiffness as compared to dinate formulation, beams and plates can be treated as isopara-
the pendulum previously used in this section. The length of the metric elements. The formulation presented in this paper leads
beam is 0.4 m, the mass density is 5540 Kg/m', the cross to a simple expression for the inertia forces which are linear in
sectional area is 0.0018 m^ and the modulus of elasticity is the accelerations, do not include any quadratic velocity terms,
1.000 X 10^ N / m ^ The beam is divided into 10 elements. In and are expressed in terms of a constant mass matrix. The
the floating frame of reference formulation, 10 elastic modes resulting mass matrix is the same matrix which appears in linear
are used in the computer simulation of the free falling of the structural dynamics problems. Furthermore, the inertia forces
pendulum. While in the absolute nodal coordinate formulation, and the kinetic and strain energies are not expressed in terms
42 nodal coordinates are used. Figure 9 shows a comparison of finite rotation coordinates, and as a consequence, the interpo-
between the results obtained using the absolute nodal coordinate lation of large rotation coordinates is avoided by using the
formulation and the results obtained using the floating frame of procedure outlined in this paper. Equally important, the proce-
reference formulation for the angular orientation of the beam. dure presented in this investigation does not lead to a singular
It is clear from these results that the two methods are in good mass matrix when the transverse dimensions of the slender
agreement. Figure 10 shows the transverse deformation of the beam are neglected.
tip point measured with respect to a coordinate system attached The simple non-incremental finite element absolute nodal
to the end of the beam at the pin joint. coordinate formulation used in this investigation, can be used
to efficiently solve flexible multibody applications. Many of the
6 Summary and Conclusions important and conceptual problems encountered when using the
floating frame of reference formulations are not encountered
In this investigation, a non-incremental computational proce- when the proposed method is used. One of these important
dure is developed for the nonlinear transient dynamic analysis
of flexible bodies that undergo large displacements. In order to

S 0.006
0.5

0 o

ft
-0.5 \ \1 \ ^ ^
-0.002
-1 M
(U
\ 10s > -0.004
-1.5
15^ '
V -0.006
-2 1 r 0.2 0.4 0.6 0.8
0 0.5 1 1.5 2 2.5
Time (s)
X
Fig. 8 Deformation of a flexible cantilever beam subject to a concen- Fig. 10 Transverse displacement of pendulum tip: ( ) absolute nodal
trated force coordinate formulation; (---) floating frame of reference

194 / Vol. 120, JUNE 1998 Transactions of the ASME

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


problems is the selection of the deformable body coordinate 3 Cardona, A., and Geradin, M., 1991, "Modeling of Superelements in
Mechanism Analysis," International Journal for Numerical Methods in Engi-
system and its relationship with the component modes and neering, Vol. 32, No. 8, pp. 1565-1593.
boundary conditions (Shabana, 1998). However, there are in 4 Flanagan, D. P., and Taylor, L. M., 1987, "An Accurate Numerical Algo-
existence today, several numerical algorithms and commercial rithm for Stress Integration with finite Rotations,'' Computer Methods in Applied
computer codes which are based on the floating frame of refer- Mechanics and Engineering, Vol. 62, pp. 305-320.
5 Geradin, M., Cardona, A., Doan, D. B., and Duysens, J., 1994, "Finite
ence formulation. Using the procedure developed in this paper, Element Modeling Concepts in Multibody Dynamics," Computer Aided Analysis
modifications can be made in the algorithms based on the float- of Rigid and Flexible Mechanical Systems, M. F. O. S. Pereira, and J. Ambrosio,
ing frame of reference formulation to allow such algorithms eds., Kluwer Academic Publishers, pp. 233-284.
to be used in the large deformation analysis of constrained 6 Hallquist, J. O., 1979, "NIKE 2D: An Implicit Finite Deformation, Finite
Element Code for Analyzing the Static and Dynamic Response of Two-Dimen-
mechanical systems (Shabana, 1996b). sional Solids," Report UCRL-52678, Lawrence Livermore Laboratory, University
Several simple examples were presented in this paper to dem- of CaUfornia.
onstrate the use of the absolute nodal coordinate formulation in 7 Hughes, T. J. R., and Winget, J., 1980,' 'Finite Rotation Effects in Numeri-
cal Integration of Rate Constitutive Equations Arising in Large Deformation
the large rotation and deformation analysis of flexible bodies. Analysis," International Journal for Numerical Methods in Engineering, Vol. 15,
These examples include free falling of a beam under the effect No. 12, pp. 1862-1867.
of concentrated and distributed gravity forces, free falling of a 8 Likins, P. W., 1974,' 'Analytical Dynamics and Nonrigid Spacecraft Simu-
flexible pendulum and large deformation of a flexible cantilever lation," Jet Propulsion Laboratory, Technical Report 32-1593, California Institute
of Technology.
beam. 9 Shabana, A. A., 1998, Dynamics of Multibody Systems, Second Edition,
Cambridge University Press.
10 Shabana, A. A., 1996a, "Finite Element Incremental Approach and Exact
Rigid Body Inertia," ASME JOURNAL OF MECHANICAL DESIGN, Vol. 118, pp.
Acknowledgment 171-178.
11 Shabana, A. A., 1996b, "An Absolute Nodal Coordinate Formulation for
This research was supported by the U.S. Army Research the Large Rotation and Deformation Analysis of Flexible Bodies," Technical
Office, Research Triangle Park, N.C. Report #MBS 96-1-UIC, Department of Mechanical Engineering, University of
Illinois at Chicago, March 1996.
12 Shabana, A. A., 1997, "Flexible Multibody Dynamics: Review of Past
and Recent Development," Journal of Multibody System Dynamics, Vol. 1, No.
References 2, pp. 189-222.
13 Shabana, A. A., and Chritensen, A. P., 1997, "Three Dimensional Abso-
1 Belytschko, T., and Hsieh, B. J., 1973, "Nonlinear Transient Finite Ele- lute Nodal Coordinate Formulation," International Journal for Numerical Meth-
ment Analysis with Convected Coordinates," InternationalJournal for Numerical ods in Engineering, Vol. 40, No. 15, pp. 2775-2790.
Methods in Engineering, Vol. 7, pp. 255-271. 14 Simo, J. C , and Vu-Quoc, L., 1986, "On the Dynamics of Flexible Beams
2 Belytschko, T., and Hughes, T. J. R., 1983, Computational Methods for Under Large overall Motions-The Plane Case: Parts I and II," ASME Journal
Transient Analysis, Elsvier Science Publishers. of Applied Mechanics. Vol. 53, pp. 849-863.

Journal of Mechanical Design JUNE 1998, Vol. 120 / 195

Downloaded From: http://mechanicaldesign.asmedigitalcollection.asme.org/ on 05/20/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like