Definition of ANCF Finite Elements coordinate redundancy problem as discussed in the literature [1].

In
order to circumvent the problems associated with the infinitesimal
rotation elements and the large rotation vector elements, the corota-
Ahmed A. Shabana tional approach which is based on an incremental-rotation algo-
rithm has been used as the solution procedure when these elements
Department of Mechanical and Industrial Engineering,
are used in the large displacement analysis. Nonetheless, the
University of Illinois at Chicago, incremental-rotation corotational formulation is not ideally suited
842 West Taylor Street, for multibody system (MBS) implementation, and there is no need
Chicago, IL 60607 for using this approach with finite elements that can describe
e-mail: [email protected] correctly rigid body motion and have a unique rotation field.
Over the past two decades, ANCF finite elements have been
widely used to develop new complex physics and engineering
models that capture details that cannot be captured using other FE
Since the absolute nodal coordinate formulation (ANCF) was and MBS formulations. These ANCF elements will serve as the
introduced, a large number of fully parametrized and gradient basis for developing new mechanics-based CAD/analysis systems
deficient finite elements were developed. Some of the finite ele- that will allow developing the geometry and analysis mesh,
ments (FE) proposed do not fall into the ANCF category, and for including mechanical joints, in one step. Developing such
this reason, this technical brief describes the general require- mechanics-based CAD/analysis systems represents a significant
ments for ANCF finite elements. As discussed in this paper, some change since existing CAD systems are envisioned for dealing
of the conventional isoparametric finite elements can describe ar- with the geometry of one component at a time. Future mechanics-
bitrary rigid body displacements and can be used with a nonincre- based CAD/analysis systems will allow for developing the geome-
mental solution procedure. Nonetheless, these isoparametric try and analysis mesh of a complete MBS model, thereby allowing
elements, particularly the ones that employ position coordinates for component shape manipulation, imposing continuity condi-
only, are of the C0 type and do not ensure the continuity of the tions and mechanical joint constraints, and adjusting dimensions
position vector gradients. It is also shown that the position vector and distances between components. Nonetheless, some of the ele-
gradient continuity conditions can be described using homogene- ments used are mistakenly called ANCF finite elements when
ous algebraic equations, and such conditions are different from they are used with a nonincremental solution procedure [2,3]. The
those conditions that govern the displacement vector gradients. objective of this technical brief (communication) is to discuss
The use of the displacement vector gradients as nodal coordinates what constitutes an ANCF finite element and shed light on the
does not allow for an isoparametric representation that accounts characteristics that distinguish these elements from other finite
for both the initial geometry and displacements using one kine- elements. The main goal of this technical brief is to demonstrate
matic description, can make the element assembly more difficult, the importance of concepts rather than the assumed polynomials
and can complicate imposing linear algebraic constraint equa- when dealing with ANCF finite elements. As discussed in this
tions at a preprocessing stage. Understanding the ANCF geomet- paper, ANCF is a solution framework for solving dynamics prob-
ric description will allow for the development of new mechanics- lems, and therefore, the formulation of the inertia forces using a
based computer-aided design (CAD) /analysis systems as briefly constant mass matrix is one of the main requirements. ANCF fi-
discussed in this paper. [DOI: 10.1115/1.4030369] nite elements are not C0 elements, C0 elements do not ensure the
continuity of the rotation and gradient fields. Furthermore, the use
Keywords: absolute nodal coordinate formulation, finite element, of absolute position and position gradient coordinates as nodal
position vector gradients, displacement vector gradients, coordinates is necessary in order to be able to develop the new
mechanics-based CAD/analysis systems mechanics-based CAD/analysis systems that will allow for a gen-
eral nonincremental solution procedure that accounts for the ini-
1 Introduction tial geometry of the element. The elements cannot be judged by
There are several conventional finite elements, such as the the interpolating polynomials only because the type of nodal coor-
brick, rectangular, triangular, and tetrahedral elements, that cor- dinates employed significantly impacts the generality of the
rectly describe rigid body motion and can be used in a nonincre- method, as will be discussed in this paper.
mental solution framework. However, these conventional finite
elements cannot be classified as ANCF finite elements. These con- 2 ANCF Finite Elements
ventional finite elements are C0 elements that ensure position con-
tinuity only, do not employ position vector gradients as nodal In order to be able to develop accurate models for engineering
coordinates, and do not ensure the continuity of the rotation and and physics systems, it is necessary to understand some basic
gradient fields at the nodal points. Such elements, while widely geometry and analysis concepts that motivated introducing the
used in the FE literature, are not good candidates for modeling ANCF solution framework. There has been a misunderstanding on
structures that are subjected to bending loading. They produce dis- what constitutes ANCF finite elements. As described in the litera-
continuous stress and curvature fields and their use requires a very ture [4], there are several conditions that must be met in order to
fine mesh in order to compensate for the problems associated with have an ANCF solution framework. These conditions are summar-
the low order of interpolation used to define the displacement ized below:
fields of these elements.
Other conventional finite elements, such as beams, plates, and (1) Dynamic Problem: the problem must be a dynamics prob-
shells, that employ infinitesimal rotations as nodal coordinates are lem that requires addressing the formulation of the inertia
based on a linearized kinematic description that is not suited for forces which are expressed in terms of a constant inertia
the analysis of spinning structures. Furthermore, large rotation matrix, and therefore, the Coriolis and centrifugal forces
vector finite element formulations, while capable of correctly are identically zero. This requirement is particularly
describing rigid body displacements, suffer from a serious important in MBS implementations. Static solution proce-
dures do not address the formulation of the inertia forces.
Because the ANCF mass matrix is constant, an identity
1
Corresponding author. generalized mass matrix can be obtained using a Cholesky
Contributed by the Design Engineering Division of ASME for publication in the
JOURNAL OF COMPUTATIONAL AND NONLINEAR DYNAMICS. Manuscript received March
transformation [4]. The use of the ANCF Cholesky coordi-
18, 2015; final manuscript received April 9, 2015; published online June 25, 2015. nates has a significant effect on the structure of the solution
Assoc. Editor: Ahmet S. Yigit. algorithm.

Journal of Computational and Nonlinear Dynamics SEPTEMBER 2015, Vol. 10 / 054506-1

C 2015 by ASME
Copyright V

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/ on 04/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 (2) Position Vector Gradients: the vector of nodal coordinates gradients associated with the midsurface parameters x and
must consist of absolute position and position gradient y. If z ¼ 0, one obtains a simple bilinear rectangular finite
coordinates and the gradients must be interpreted correctly element defined by position coordinates only.
during the entire solution procedure. The use of position (2) As mentioned in the abstracts of Betsch and Stein’s paper,
vector gradients is conceptually different from the use of their element is C0 which implies that no continuity condi-
the displacement vector gradients. As discussed in this tions are imposed on the gradients at the midsurface of the
paper, the position vector gradients and the displacement element. The stress and strain fields at the midsurface are
vector gradients are not tangents to the same coordinate also discontinuous. The rotations along two axes are also
lines and they are not governed by the same connectivity discontinuous. ANCF finite elements ensure the continuity
conditions when elements are assembled or linear algebraic of the gradients at the nodal points, and therefore, the rota-
constraint equations are imposed at a preprocessing stage. tion field is continuous at those points.
For instance, having a vector that is tangent to a certain (3) If z ¼ 0 (midsurface) and for a given value of y, rm defines
coordinate line, such as a beam centerline, gives this gradi- a truss-like element. Some authors have used this element
ent vector a clear geometric and physical meaning that to model tires and called it an ANCF element. For a tire
allows for using it to impose more general connectivity and model, at z ¼ 0 and for a given value of y, the circle is
continuity conditions. Furthermore, B-spline and NURBS defined using straight line segments (truss-like elements).
control points can be related to the position vector gradients This leads to a poor geometric description which leads to
using a system of homogeneous algebraic equations. This discontinuous slopes along the circumferential curve of the
cannot be achieved, for example, by using the displacement tire tread midsurface.
gradient vectors, as will be discussed later in this paper. (4) As a direct consequence of using this non-ANCF element,
(3) Rigid Body Displacement: the finite elements must the midsurface displacement field cannot be used to define
correctly describe an arbitrary rigid body motion including curvature and the midsurface cannot bend. As in the case
arbitrary finite rotations using the assumed displacement of the conventional brick element, the use of this element
field rðx; tÞ ¼ SðxÞeðtÞ, where r is the global position vec- to describe bending accurately may require a very fine
tor on the element, S is the element shape function, e is the mesh particularly in the case of large bending deforma-
vector of nodal coordinates, x ¼ ½ x y z T is the vector tion. Such elements cannot also be used to describe accu-
of the element spatial coordinates, and t is the time. rately initial stress-free curved configurations as in the
(4) Nonincremental Solution Procedure: the equations of case of tires.
motion are solved nonincrementally, and therefore, there is (5) The solution procedure used in the Betsch and Stein’s paper
no need for the use of the corotational approach which was has nothing to do with the ANCF nonincremental solution
introduced to remedy the problems associated with finite procedure. The corotational formulation is used by employ-
elements that employ infinitesimal rotations. Corotational ing an updating scheme as clearly stated in their paper.
formulations are not ideally suited for MBS implementa- Using the corotational approach, the mass matrix is not, in
tions, and there is no need to use such corotational solution general, constant and one has to formulate the gyroscopic
procedures with elements that correctly describe rigid body and Coriolis forces.
motion. (6) The paper by Betsch and Stein also uses an update proce-
dure for the extensible director based on simple vector
It is important to point out that, in the FE literature, there has addition [7]. This updating scheme cannot be applied to
been frequent reference to gradients and slopes. Nonetheless, all position gradient vectors which must be updated using a
or some of the above conditions are not often met as the result of different multiplicative scheme that employs tensor opera-
improper treatment of the gradients, the use of incremental proce- tions. Therefore, while derivative with respect to z is used,
dures, and/or the focus on static procedures without addressing the directors are not treated as position gradients since addi-
the formulation of the element inertia when the element experien- tive updating is used. One must carefully examine Eqs. (23)
ces arbitrary finite rotations [5,6]. and (24) in Ref. [7] to see that the directors are not inter-
preted or dealt with as position gradient vectors throughout
the solution procedure. Equations (23) and (24) in Ref. [7]
3 Gradient Continuity describe the interpolation of the magnitude of a vector mul-
tiplied by a unit vector. Equations (60)–(64) of the same pa-
The concept of gradients must be understood when ANCF finite per [7] can also be examined to understand the fundamental
elements are assembled. A tensor transformation consistent with differences between the way the position gradients and
the definition of gradient vectors as tangents to coordinate lines directors are handled. This is again a fundamental issue
must be used when the finite elements are assembled. Some when the finite elements are assembled [8].
authors have mistakenly referred to elements used in the literature (7) Some of the issues related to the use of two interpolations
as ANCF finite elements. To explain the fundamental differences in which the first interpolation depends on the second,
between ANCF finite elements and elements that were wrongly while the second interpolation does not depend on the first
labeled as ANCF finite elements, the plate element employed by are discussed in Sec. 11 of Chap. 7 of the second edition of
Betsch and Stein [5] is used as an example in this section. In their the book [9] and subsequent editions of the same book.
excellent paper, Betsch and Stein did not call their element an There are fundamental issues related to the formulations of
ANCF element, it was subsequently called ANCF element by the forces and kinematic equations that need to be consid-
other authors. Betsch and Stein’s element is defined by the dis- ered when using such assumed displacement fields.
placement field rðx; y; zÞ ¼ rm ðx; yÞ þ zrz ðx; yÞ, where rm ðx; yÞ
defines the position of an arbitrary point on the midsurface of the
finite element, and rz ðx; yÞ ¼ @r=@z was called extensible director. Slopes, strains, stresses, and gradients have been frequently
Both rm ðx; yÞ and rz ðx; yÞ are interpolated using bilinear polyno- used in the finite element literature long before ANCF finite ele-
mials in x and y. This element which was used to model vehicle ments were introduced. However, slopes, strains, stresses, and
components by other authors is not an ANCF finite element, as gradients have been used in different frameworks under different
explained below. assumptions. For example, in the FE literature, it is quite often to
refer to the infinitesimal rotations as slopes, but as it is known
(1) At the midsurface of the Betsch and Stein’s element, the rotations are not slopes, and vectors are not necessarily gradients
displacement field is not expressed in terms of any that are tangents to coordinate lines.

054506-2 / Vol. 10, SEPTEMBER 2015 Transactions of the ASME

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/ on 04/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 4 Position and Displacement Gradient Vectors position vector gradients are equated using tensor transformation
The ANCF description is based on using absolute position and with the understanding that two gradient vectors associated with
gradient coordinates that define the location of material points on the nodes of two different elements can be equal if these nonzero
the finite element in a fixed global coordinate system. The vector vectors are tangents to the same coordinate lines. In the case of a
rðx; tÞ ¼ SðxÞeðtÞ defines the absolute position vector of an arbi- beam element, for example, the vector of position vector gradients
trary point on the ANCF finite element with respect to the global is tangent to the beam centerline, while the vector of displacement
fixed coordinate system. There is a fundamental difference gradients is not tangent to the beam centerline. More discussion
between this ANCF description and other descriptions that on this important topic will be provided before concluding this
employ displacement vectors instead of the position vectors. section.

4.1 Background. Milner [6] introduced a method for the large 4.4 Initially Curved Configurations. The ANCF description
displacement analysis of beams. The method employs positions rðx; tÞ ¼ SðxÞðeo ðtÞ þ ed ðtÞÞ, where eo is the vector of nodal coor-
and displacement gradients as nodal coordinates. Milner’s method dinates in the reference configuration, and ed is the nodal displace-
was applied to static analysis with no mention of the formulation ment vector, allows for dealing with general geometry and for
of the inertia forces of the beams. The same interpolation was accurately calculating the inertia of curved structures. For exam-
used for the displacement coordinates. While the formulation was ple, the general ANCF implementation allows for modeling ini-
focused on the static analysis, it is important to mention that the tially stress-free curved structures. If J ¼ jrx j is the determinant
use of the displacement gradients as nodal coordinates is different of the matrix of position vector gradient J ¼ rx at the current con-
from the use of the position vector gradients. The position gradi- figuration, and if Jo ¼ @ ðSeo Þ=@x is the matrix of position vector
ent vectors used with ANCF finite elements are tangents to coordi- gradients at the initial stress-free configuration, then the initial
nate lines, while the displacement gradient vectors are not curvature can be handled in an ANCF implementation in a straight
tangents to the same coordinate lines. If rðx; tÞ is the position of forward manner using the product JJ1 o . Furthermore, important
an arbitrary point on the element, this vector can be written using concepts for imposing constraints at a preprocessing stage will be
the continuum mechanics description as rðx; tÞ ¼ x þ uðx; tÞ, much more difficult if the displacement gradient vectors are used
where u is the displacement vector. It follows that the relationship instead of the position vector gradients [10,11]. For instance, the
between the position gradient vectors and the displacement gradi- determinant Jo ¼ jJo j is required in order to determine the inertia
ent vectors is defined by the equation rx ¼ I þ ux , where of curved structures and also to account for the stress-free initial
rx ¼ @r=@x and ux ¼ @u=@x. This equation shows that the posi- curvature.
tion gradient vectors which form the columns of the tensor
rx ¼ ½ rx ry rz  are not parallel to the displacement gradient 4.5 Interpolation of Displacements. In the case of using
vectors which form the columns of the tensor ux ¼ ½ ux uy uz . position or displacement coordinates, the interpolation polyno-
Furthermore, the displacement gradient vectors can be identically mials must correctly describe rigid body motion. In the case of
zeros, while the position gradient vectors cannot be the zero vec- rigid body motion of an Euler–Bernoulli-beam element, the dis-
tors. In the case of the Euler–Bernoulli beam considered by Mil- placement vector u is defined as uðx; tÞ ¼ RðtÞ þ ðA  IÞx, where
ner [6], if s is the arc length parameter, rs must always satisfy the R ¼ ½ R1 R2 T is the displacement of the first node of the ele-
condition jrs j ¼ 1, while jus j can assume any value including the ment, A is the transformation matrix defined by the first matrix in
zero value. Eq. (1), and x ¼ ½ x 0 T . As described by Milner [6], who used
the arc length instead of the element spatial coordinate in his static
4.2 Rigid Body Rotation. In the case of a simple planar rigid analysis, u can be interpolated as u ¼ Sed , where S
body rotation defined by the angle h, the position gradient vectors ¼ ½ s1 I s2 I s3 I s4 I , I is the 2  2 identity matrix, ed is the
and the displacement gradient vectors are defined, respectively, vector of nodal displacements, and sk ; k ¼ 1; 2; 3; 4, are the shape
by functions defined as
 
cosh sinh
rx ¼ ½ rx ry  ¼ ;   )
sinh cosh s1 ¼ 1  3n2 þ 2n3 ; s2 ¼ l n  2n2 þ n3 ;
    (2)
cosh  1 sinh
ux ¼ ½ ux uy  ¼ (1) s3 ¼ 3n2  2n3 ; s4 ¼ l n2 þ n3
sinh cosh  1
In the definition of these shape functions, n ¼ x=l, where l is the
It is clear from these equations that in the case of simple rigid length of the element. In the case of a general planar rigid body
body rotation, rTx ux ¼ rTy uy ¼ 1  cosh, rTx uy ¼ sinh, and displacement, the vector of nodal displacements ed is defined as
rTy ux ¼ sinh, that is, the two sets of gradient vectors do not assume
the same orientation. In this simple case of rigid body rotation, the ed ¼ ½R1 R2 ðcosh  1Þ sinh R1 þ lðcosh  1ÞR2
difference in the orientation between the vectors of position gra-
dients and the vectors of displacement gradients can be signifi- þ lsinh ðcosh  1Þ sinhT (3)
cant. These equations also show that in the case of simple rigid
body rotation, the displacement gradient vectors can be equal to Using this definition of the nodal displacements, one can show
zero, while the position vector gradients cannot be equal to zero. that
That is, the matrix of position vector gradients is always nonsin-
gular, while the matrix of displacement vector gradients can be " #
singular. Furthermore, in the case of rigid body motion the posi- R1 þ xðcosh  1Þ
tion gradient vectors are unit vectors, while the displacement gra- u ¼ Sed ¼ (4)
dient vectors are not unit vectors. R2 þ xsinh

which is the correct rigid body displacement. Note that if the

4.3 Coordinate Lines. The fact that the displacement gradi- absolute position coordinates are used, a simpler expression for
ent vectors are not tangents to the coordinate lines used to define the nodal coordinates is obtained as
these displacement gradients sheds light on the need for using the
position vector gradients in formulating the ANCF finite element ed ¼ ½ R1 R2 cosh sinh R1 þ lcosh R2 þ lsinh cosh sinh T
kinematic and dynamic equations. In the assembly process, the (5)

Journal of Computational and Nonlinear Dynamics SEPTEMBER 2015, Vol. 10 / 054506-3

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/ on 04/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Which when substituted in r ¼ Se gives the correct absolute posi-
tion of an arbitrary point on the element as
 
R1 þ xcosh
r ¼ Se ¼ (6)
R2 þ xsinh

The interpolation of the displacements, however, makes the ele-

ment assembly more cumbersome and complicates imposing
ANCF algebraic constraints equations at the preprocessing stage
[12]. For instance, in the assembly process, the gradients of
the elements should be defined using the same coordinate lines
in order to apply the connectivity and continuity conditions
correctly. If x is the vector that defines the structure coordinate
Fig. 1 Mechanics-based CAD/analysis models (• chassis/
lines, and xj is the vector that defines the spatial coordinates of
shackle revolute joint, $ shackle/leaf spring revolute joint, 䉱
the finite element j, one has @r=@x ¼ ð@r=@xj ÞAj , where leaf spring/axle revolute joint, and 䊏 chassis/leaf spring revo-
Aj ¼ @xj =@x is the matrix of coordinate line transformation. For lute joint)
the same  point on another element k, one has @r=@x
¼ @r=@xk Ak , where Ak ¼ @xk =@x. It follows that the connec-
tivity conditions using position vector gradients are governed by distances and dimensions can be adjusted. The resulting solid
the homogenous algebraic equations ð@r=@xj ÞAj ¼ @r=@xk Ak , model of the entire vehicle is the analysis mesh used in the MBS
while the same conditions using the displacement vector gradients computer code. This mesh, which includes tires, leaf springs,
are governed by  the nonhomogeneous relationships axles, and chassis, has a constant mass matrix, which can be made
 an identity generalized inertia matrix associated with the ANCF
ð@u=@xj ÞAj  @u=@xk Ak ¼ Ak  Aj . In terms of the structure
parameters, the displacement gradients along the element coordi- Cholesky coordinates. The rigidity of some of the ANCF refer-
  1 ence nodes can be imposed in the MBS algorithm using nonlinear
nate lines can be written as ð@u=@xj Þ ¼ ð@u=@xÞ  Aj þ I Aj ,
which is not a simple equation to recover the displacement gra- algebraic equations that are satisfied at the position, velocity, and
dients defined in the element coordinate system as compared to acceleration levels.
the position vector gradient equation. The multiplicative decom-
position of the position vector gradients allows for defining the
same tangents at a point shared by two elements. This is not, how- 6 Summary
ever, the case when the displacement gradients are used, as it is CAD systems were envisioned as tools for the development of
clear from the equations presented in this section. the geometry and solid model of one component, and therefore,
Despite the fact that Milner [6] considered only the static case existing CAD systems are not suited for handling models that con-
with no mention to dynamics or a dynamic solution procedure, sist of many components connected by mechanical joints. Further-
Milner’s work is an example of contributions in which the use of more, existing CAD systems employ B-spline and NURBS
certain interpolation, even if it were used in dynamics, does not representations which were developed without considering the
necessarily define an ANCF finite elements because this interpola- concept of the degrees-of-freedom which is fundamental at the
tion is used for static problems and because of the type of coordi- analysis stage. ANCF finite elements will serve as the basis for
nates used. The use of the displacement coordinates will not allow developing new mechanics-based CAD/analysis systems. To suc-
for developing the ANCF meshes that allows for systematically cessfully develop such CAD/analysis systems, the definition of
integrating geometry and analysis algorithms [11,12]. These algo- what constitutes ANCF finite elements must be understood. In this
rithms will serve as the basis for developing new mechanics- technical brief, the requirements for defining ANCF finite ele-
based CAD/ analysis systems that will allow from the outset ments are summarized. As discussed in this paper, ANCF is a
developing the geometry of multicomponent models. solution framework for solving dynamics problems and therefore,
the formulation of the inertia forces using a constant mass matrix
5 Mechanics-Based CAD/Analysis Systems is one of the main issues. ANCF finite elements are not C0 ele-
ments, C0 elements do not ensure the continuity of the rotation
While CAD systems were envisioned as tools for creating the and gradient fields. The use of absolute position and gradient
geometry and solid models of one component at a time, under- coordinates is necessary in order to be able to develop a general
standing the generality of the ANCF geometric description, which nonincremental solution procedure that accounts for the initial
employs position vector gradients, will allow developing new geometry of the element. The elements cannot be judged by the
mechanics-based CAD/analysis systems designed to handle multi- interpolating polynomials only because the type of nodal coordi-
component models, define their geometry, introduce their connec- nates employed significantly impacts the generality of the method.
tivity including mechanical joints, and produce a solid model that This was demonstrated in this paper by discussing the
represents the analysis mesh. Such new mechanics-based analysis
systems cannot be developed using the B-spline and NURBS geo-
metries that have rigid recurrence structure. As demonstrated in
the literature, the concept of the knot multiplicity used in B-spline
and NURBS representations does not correctly account for the
concept of the degrees-of-freedom which is fundamental at the
analysis stage [4]. ANCF finite element geometry allows for
imposing algebraic equations that define mechanical joints as well
as other connectivity and continuity conditions. Figure 1 shows an
ANCF mesh that represents the entire vehicle shown in Fig. 2
[13]. This mesh is assembled at a preprocessor stage using the
concept of the ANCF reference nodes, which allows for assem-
bling rigid and flexible bodies in a straight forward manner [10].
The mesh includes continuum-based ANCF tires and leaf springs,
and rigid chassis and axles. The geometry of all these compo-
nents, treated as one CAD model, can be manipulated and Fig. 2 Vehicle model

054506-4 / Vol. 10, SEPTEMBER 2015 Transactions of the ASME

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/ on 04/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 fundamental differences between the position and displacement [7] Betsch, P., and Stein, E., 1996, “A Nonlinear Extensible 4-Node Shell Element
vector gradients. Based on Continuum Theory and Assumed Strain Interpolations,” Nonlinear
Sci., 6(2), pp. 169–199.
[8] Shabana, A. A., and Mikkola, A. M., 2003, “Use of the Finite Element Absolute
References Nodal Coordinate Formulation in Modeling Slope Discontinuity,” ASME J.
[1] Ding, J., Wallin, M., Wei, C., Recuero, A. M., and Shabana, A. A., 2014, “Use Mech. Des., 125(2), pp. 342–350.
of Independent Rotation Field in the Large Displacement Analysis of Beams,” [9] Shabana, A. A., 1998, Dynamics of Multibody Systems, 2nd ed., Cambridge
Nonlinear Dyn., 76(3), pp. 1829–1843. University, Cambridge.
[2] Dombrowski, S. V., 2002, “Analysis of Large Flexible Body Deformation in [10] Shabana, A. A., 2015, “ANCF Tire Assembly Model for Multibody System
Multibody Systems Using Absolute Coordinates,” Multibody Syst. Dyn., 8(4), Applications,” ASME J. Comput. Nonlinear Dyn., 10(2), p. 024504.
pp. 409–432. [11] Patel, M., Orzechowski, G., Tian, Q., and Shabana, A. A., “A New MBS
[3] K€ubler, L., Eberhard, P., and Geisler, J., 2003, “Flexible Multibody Systems Approach for Tire Modeling Using ANCF Finite Elements,” Proc. Inst. Mech.
With Large Deformations Using Absolute Nodal Coordinates for Isoparametric Eng., Part K (published online).
Solid Brick Elements,” ASME Paper No. DETC2003/VIB-48303. [12] Hamed, A. M., Jayakumar, P., Letherwood, M. D., Gorsich, D. J., Recuero, A.
[4] Shabana, A. A., 2012, Computational Continuum Mechanics, 2nd ed., Cam- M., and Shabana, A. A., 2015, “Ideal Compliant Joints and Integration of Com-
bridge University, Cambridge. puter Aided Design and Analysis,” ASME J. Comput. Nonlinear Dyn., 10(2),
[5] Betsch, P., and Stein, E., 1995, “An Assumed Strain Approach Avoiding Artifi- p. 021015.
cial Thickness Straining for a Non-Linear 4-Node Shell Element,” Commun. [13] Yu, Z., Liu, Y., Tinsley, B., and Shabana, A. A., 2015, “Integration of
Numer. Methods Eng., 11(11), pp. 899–909. Geometry and Analysis for Vehicle System Applications: Continuum-
[6] Milner, H. R., 1981, “Accurate Finite Element Analysis of Large Displacements Based Leaf Spring and Tire Assembly,” IMechE J. Multibody Dyn. (in
in Skeletal Frames,” Comput. Struct., 14(3–4), pp. 205–210. review).

Journal of Computational and Nonlinear Dynamics SEPTEMBER 2015, Vol. 10 / 054506-5

Downloaded From: http://computationalnonlinear.asmedigitalcollection.asme.org/ on 04/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use