Experimental validation of

flexible multibody dynamics beam formulations∗

Olivier A. Bauchau∗, Shilei Han∗, Aki Mikkola†,
Marko K. Matikainen†, and Peter Gruber‡
∗
University of Michigan-Shanghai Jiao Tong University Joint Institute
800 Dong Chuan Road, Shanghai, 200240, China
†
Department of Mechanical Engineering
Lappeenranta University of Technology, 53851 Lappeenranta, Finland
‡
Austrian Center of Competence in Mechatronics GmbH
4040 Linz, Austria
July 12, 2015

Abstract
In this paper, the accuracies of the geometrically exact beam and absolute nodal coordi-
nate formulations are studied by comparing their predictions against an experimental data set
referred to as the “Princeton beam experiment.” The experiment deals with a cantilevered
beam experiencing coupled flap, lag, and twist deformations. In the absolute nodal coordinate
formulation, two different beam elements are used. The first is based on a shear deformable
approach in which the element kinematics is described using two nodes. The second is based on
a recently proposed approach featuring three nodes. The numerical results for the geometrically
exact beam formulation and the recently proposed three-node absolute nodal coordinate formula-
tion agree well with the experimental data. The two-node beam element predictions are similar
to those of linear beam theory. This study suggests that a careful and thorough evaluation of
beam elements must be carried out to assess their ability to deal with the three-dimensional
deformations typically found in flexible multibody systems.

1 Introduction
The simulation of helicopter rotor blade dynamics is a difficult problem because nonlinearities affect
the response of the system. In 1958, Houbolt and Brooks [1] derived the differential equations of
motion for the combined flapwise bending, chordwise bending, and torsion of twisted nonuniform
rotor blades. Their approach was based on a modal representation of the blade’s response and the
equations of motion were developed in a frame of reference that rotated with the rotor. Due to
poor correlation with flight test measurements, the accuracy of this formulation was questioned.
In 1974, Hodges and Dowell [2] developed a more accurate set of equations for the same problem:
nonlinear terms were handled carefully, although terms of order higher than second were neglected.
To assess the accuracy of these equations, Dowell and Traybar [3, 4] carried out an experimental
study of the nonlinear stiffness of a rotor blade undergoing flap, lag, and twist deformations, often
∗
Multibody System Dynamics, 34(4): pp 373-389, 2015

1
referred to as the “Princeton beam experiment.” Although the correlation of the Hodges-Dowell
predictions with this data set was not very good, the analysis was deemed sufficiently accurate for
hingeless helicopter rotor blade dynamics and aeroelasticity, because they involve moderately large
deformations only.
The Princeton beam experiment presents the static deflections of a simple cantilevered beam
under various loading conditions. Yet, this static experiment was used for the validation of beam
models developed for rotor dynamics. In this paper, a similar approach is followed: the same static
experimental data will be used for the validation of flexible multibody dynamics beam formulations.
The dynamic response of bearingless rotor blades is affected more significantly by nonlinear
behavior than that of articulated blades. In the late 1970s, it was recognized that bearingless rotor
problems demanded a more accurate theory, and many researchers started focusing on geometrically
exact beam theories. The kinematics of the geometrically exact beam formulation was first presented
by Simo et al. [5, 6], but similar developments were proposed by Borri and Merlini [7], Danielson
and Hodges [8, 9], or Bauchau and Hong [10, 11, 12]. The Princeton beam experiment was used as
a benchmark to assess the accuracy of these formulations.
The previous paragraphs have summarized in a succinct manner some the milestones in the
development of accurate equations for the modeling of helicopter rotor blades. Similar steps can
be found in the development of simulation tools for flexible multibody dynamics, which originally
focused on simple tree-like topologies composed of rigid bodies. As the need to model flexibility
arose, the floating frame of reference formulation [13, 14, 15] was developed; this solution strategy
is akin to the approach of Houbolt and Brooks for helicopter rotors.
Because floating frame of reference formulations were found to yield inaccurate results when
elastic deformations become large, the multibody dynamics community began to turn its attention
to geometrically exact beam formulations (GEBF). other approaches have been proposed for the
analysis of beams undergoing large motion, the GEBF is probably the best established formulation.
The authors cited above have proposed different numerical implementations of the theory, but
all solve the geometrically exact beam equations. In recent years, the absolute nodal coordinate
formulation (ANCF) has been developed and has received considerable attention for the modeling
of flexible multibody systems. Clearly, the developments in the multibody dynamics community
parallel those of the rotorcraft community. The GEBF and ANCF involve fewer assumptions
than the floating frame of reference approach, but little effort has been devoted to the systematic
comparison of these two approaches.
Romero [16] has presented a comparison of both qualitative and quantitative aspects of the two
approaches. He concluded that the ANCF is more straightforward, while GEBF involves thorny
issues, such as the treatment of finite rotations. Unfortunately, the ANCF suffers from a number
of locking mechanisms that must be eliminated to obtain accurate results. As pointed out by
Gerstmayr et al. [17], this can be accomplished in a number of ways, but the proposed techniques
complicate the description of elastic forces, leading to more arduous implementations and moving
away from the simplicity of the initial implementation. In some of the examples treated by Romero,
the ANCF and GEBF did not converge to the same solution. In all cases, the computational
efficiency of the GEBF was found to be far superior to that of the ANCF.
Bauchau et al. [18] further compared the GEBF and ANCF to identify the causes of their differing
computational efficiencies for the two-dimensional beam case. First, they performed a kinematic
solution, in which the exact nodal displacements were prescribed and the interpolated displacement
and strain fields were compared for the two methods. The accuracies of the interpolated strain fields
were found to differ markedly: the predictions of the GEBF were far more accurate than those of
the ANCF. They attributed this phenomenon to the fact that the curvature field is obtained as a
second derivative of the displacement in the ANCF, but as a first derivative only for the GEBF.
Next, they carried out a static solution to determine the solution of the problem. For the GEBF,
the predictions of the static solution are far more accurate than those obtained from the kinematic

2
solution; in contrast, the same order of accuracy was obtained for the two solution procedures when
using ANCF. In all cases, they reported that the predictions of the GEBF are more accurate than
those of the ANCF.
The comparative studies of Romero and Bauchau et al. were of a purely numerical nature, and
hence, validation against experimental data is needed to come to a definite conclusion. Because
the Princeton beam experiment focused on applications to helicopter rotor blades, this data set
has received little attention outside the rotorcraft community. The goal of this paper is to validate
several beam theories used in flexible multibody dynamics simulations against the Princeton beam
experiment.

2 Formulation of beam elements

In Euler-Bernoulli beam and Kirchhoff plate formulations, transverse shear strains are assumed to
vanish and hence, rotation of the cross-sectional plane and of the normal material line, respectively,
are obtained from derivatives of the displacement field, and curvatures are expressed in terms second
derivatives of the same field [19]. In contrast, shear deformable beams and plates, often called
Timoshenko beams [20, 21] and Reissner-Mindlin plates [22, 23], respectively, are Cosserat solids:
the kinematics of these structural components are described in terms of two independent fields, a
displacement field and a rotation field. Reissner investigated beam theory for large strains [24] and
large displacements of spatially curved members [25, 26].
In two- and three-dimensional elasticity, the rotation field is not independent of the displacement
field. Indeed, the polar decomposition theorem can be used to decompose the deformation gradient
tensor into a stretch tensor and an orthogonal rotation tensor [27]. Because this decomposition
is unique, the deformation gradient tensor, a function of the displacement field only, defines the
rotation field unambiguously. This contrasts with Cosserat solids for which the displacement and
rotation fields are independent.
A beam is defined as a structure having one of its dimensions much larger than the other two,
as depicted in fig. 1. The axis, or reference line, of the beam is defined along that longer dimension
and its cross-section is normal to this axis. The cross-section’s geometric and physical properties
are assumed to vary smoothly along the beam’s span.

Deformed
configuration

B
P

Reference
line

Reference
P B configuration

Figure 1: Curved beam in the reference and deformed configurations.

Figure 1 depicts an initially curved and twisted beam of length L, with a cross-section of arbitrary
shape and area A. The volume of the beam is generated by sliding the cross-section along the
reference line of the beam, which is defined by an arbitrary curve in space. Curvilinear coordinate
α1 defines the intrinsic parameterization of this curve, i.e., it measures length along the beam’s

3
reference line. Point B is located at the intersection of the reference line with the plane of the
cross-section.

2.1 Kinematics of the problem

In the reference configuration, an orthonormal basis, B0 (α1 ) = (b̄01 , b̄02 , b̄03 ), is defined at point B.
Vector b̄01 is the unit tangent vector to the reference curve at that point, and unit vectors b̄02 and
b̄03 define the plane to the cross-section. An inertial reference frame, FI = [O, I = (ı̄1 , ı̄2 , ı̄3 )], is
defined, and the components of the rotation tensor that brings basis I to B0 , resolved in basis I,
are denoted R0 (α1 ).
The position vector of point B along the beam’s reference line is denoted u0 (α1 ). The position
vector of material point P of the beam then becomes x(α1 , α2 , α3 ) = u0 (α1 ) + α2 b̄02 + α3 b̄03 , where
α2 and α3 are the material coordinates along unit vectors b̄02 and b̄03 , respectively. Coordinates α1 ,
α2 , and α3 form a natural choice of coordinates to represent the configuration of the beam.
In the deformed configuration, all the material points located on a cross-section of the beam
move to new positions. This motion is decomposed into two parts, a rigid-body motion and a
warping displacement field. The rigid-body motion consists of a translation of the cross-section,
characterized by displacement vector u(α1 ) of reference point B, and of a rotation of the cross-
section, which brings basis B0 to B(α1 ) = (b̄1 , b̄2 , b̄3 ), see fig. 1. The components of the position
vector of point B in the deformed configuration is denoted x(α1 ) and the components of the rotation
tensor that brings basis B0 to B are denoted R(α1 ); all tensor components are resolved in basis I.
For shear deformable beams, the deformation is characterized by six sectional strains: the axial
strain, the two transverse shear strains, the twist rate, and the two bending curvatures. Of course,
for Euler-Bernoulli beams, the two transverse shear strains are assumed to vanish.

2.2 Geometrically exact beam formulation

In this section, the geometrically exact beam theory will be summarized. The kinematic description
of the problem developed in section 2.1 accounts for arbitrarily large displacements and rotation,
hence the term “geometrically exact,” although the strain components are assumed to remain small.

2.2.1 Definition of strain measures

The sectional strain measures for beams are defined as
   ′
x0 + u′ − (R R0 ) ı̄1

Γ
E= = , (1)
Λ axial(R′ RT )

The sectional strain vector is defined as ΓT = Γ11 , Γ12 , Γ13 , where Γ11 is the sectional axial


strain, and Γ12 and

 Γ13 the two components of transverse shearing strains. The curvature vector is
T
defined as Λ = Λ1 , Λ2 , Λ3 , where Λ1 is the sectional twist rate, and Λ2 and Λ3 the two bending
curvatures. Notation (·)′ indicates a derivative with respect to α1 . The strain components resolved
in the convected material basis, B, are denoted Γ∗ = (R R0 )T Γ. The curvature components resolved
in the same material basis are denoted Λ∗ = (R R0 )T Λ. Notation (·)∗ indicates the components of
vectors and tensors resolved in the material basis.
By definition, a rigid-body motion is a motion that generates no strains. This implies that the
following rigid-body motion, u(α1 ) = uR + (RR − I)x0 (α1 ), R(α1 ) = RR , consisting of a translation,
uR , and a rotation about the origin characterized by a rotation tensor, RR , should not strain the
beam. It can be verified readily with the help of eqs. (1) that such rigid-body motion results in
Γ = 0 and Λ = 0, as expected.

4
2.2.2 Equations of motion
The derivation of the equations of motion for geometrically exact beams can be found in textbooks
such as those of Hodges [28] or Bauchau [29] and will not be repeated here. The governing equations
are

ḣ − N ′ = f , (2a)
ġ + ũ˙ h − M −
′
(x̃′0 ′
+ ũ )N = m. (2b)

The externally applied forces and moments per unit span of the beam are denoted f and m,
respectively. The beam’s sectional forces and moments are denoted N and M , respectively. Finally,
the components of the sectional linear and angular momenta are denoted h and g, respectively.
Notation (·)· indicates a derivative with respect to time. The governing equations of motion (2)
express the dynamic equilibrium conditions of the beam at each instant. To form a complete set,
constitutive laws for both sectional forces and momenta are required.
The sectional forces and moments are related to the sectional strain and curvature components
through the following constitutive laws
 ∗  ∗
N ∗ Γ
=D , (3)
M∗ Λ∗

where N ∗ and M ∗ are the beam’s sectional forces and moments, respectively, and Γ∗ and Λ∗ the
sectional strains and curvatures, respectively, resolved in the material basis. In eq. (3), D ∗ is the
beam’s 6×6 sectional stiffness matrix. This matrix is a byproduct of a two-dimensional finite element
analysis over the beam’s cross-section, as discussed in refs. [30, 28, 31]. For homogeneous sections of
simple geometry, exact or approximate analytical expressions are available for the stiffness matrix.
The strain energy stored in the beam is then

1 L ∗T ∗ ∗
Z
A= E D E dα1 . (4)
2 0
The linear and angular momenta are related to the sectional linear and angular velocities through
the following constitutive laws
 ∗ " #
mI mη̃ ∗T (R R0 )T u̇

h
= , (5)
g∗ mη̃ ∗ ̺∗ (R R0 )T ω

where h∗ and g ∗ the components of the sectional linear and angular momenta resolved in the
material system, respectively, and u̇ and ω the inertial linear and angular velocities of the cross-
section, respectively. The components of linear and angular momenta resolved in the material basis
are related to their counterparts in the inertial system as h = (R R0 )h∗ and g = (R R0 )g ∗ . In
eq. (5), the following sectional mass constants were defined

1
Z Z Z
∗ ∗ ∗
m= ρ dA, η = ρ s dA, ̺ = ρs̃∗ s̃∗T dA, (6)
A m A A

where m is the mass of the beam per unit span, η∗ the components of the position vector of the
sectional center of mass with respect to point B, see fig. 1, and ̺∗ the components of the sectional
tensor of inertia per unit span, all resolved in the material basis.

5
2.3 Absolute nodal coordinate formulation
According to Gerstmayr et al. [17], beam elements based on the ANCF fall into two groups. The
first group is based on Euler-Bernoulli kinematics, i.e., transverse shear deformations are assumed
to vanish. These elements, often referred to as “gradient deficient elements,” are approximated
as line elements and use position coordinates and longitudinal slopes variables only. The second
group is based on Timoshenko kinematics, i.e., transverse shear deformations do not vanish. In
addition to the variables found in their gradient deficient counterparts, these elements, referred
to as “fully parameterized elements,” also feature variables that describe transverse slope vectors,
thereby allowing the description of transverse shear and cross-sectional deformations.
In this study, two- and three-node shear deformable beam elements [32, 33] are used to assess
the ANCF. The two-node elements introduced by Dufva et al. [32] have longitudinal and transverse
slope vectors, while the recently introduced three-node elements [33] have transverse slope vectors
only. Both elements are based on a structural mechanics approach, but use different strain energy
definitions. Furthermore, both elements use a different interpolation order.
The fully parameterized, two-node beam elements first introduced by Shabana and Yakoub [34]
used cubic interpolation of displacements along the beam’s span and linear interpolation over its
cross-section. This element converged slowly due to transverse shear locking: indeed, the transverse
and longitudinal slope vectors were approximated using different polynomial orders. This mismatch
of interpolation orders led to shear locking, which can be eliminated using a mixed interpolation
approach, as presented Dufva et al. [32].
Nachbagauer et al. [33] proposed three-node beam elements featuring quadratic interpolation of
displacements along the beam’s span and linear interpolation over its cross-section. This approach,
combined with a proper description of elastic forces, led to higher accuracy and convergence rates
than those observed for the two-node elements.

2.3.1 Kinematics of the ANCF beam elements

Like typical three-dimensional finite element formulations, the ANCF uses shape functions and
inertial coordinates. The position vector of an arbitrary material point in inertial frame I is
interpolated as
r = S m (α1 , α2 , α3 )e, (7)
where matrix S m stores the shape functions expressed in terms of the element local coordi-
nates α1 , α2 , and α3 , and array e stores the nodal coordinates. Array e is partitioned as
e = {e(1)T , e(2)T , . . . , e(n)T }, where n is the number of nodes of the element. For fully parame-
terized ANCF beam elements the nodal coordinates are
h iT
(i)T (i)T (i)T
e(i) = r(i)T r ,1 r ,2 r,3 , (8)

(i) (i) (i)

where r (i) is the inertial position of node i, r,1 the longitudinal slope vector, and r ,2 and r,3 the
transverse slope vectors used to define cross-sectional orientations and deformations. Notation (·),j
indicates a derivative with respect to coordinate αj , i.e., r,j = ∂r/∂αj for j = 1, 2, 3.

2.3.2 Two-node ANCF beam element

In this section, the two-node beam element introduced by Dufva et al. [32] is summarized succinctly.
The kinematic description of the element is similar to that proposed by Yakoub and Shabana [35],
while the strain energy is different. The rotation of the cross-section is defined using linear inter-
polation over the beam’s span. In the case of the two-node element, transverse shear deformations
vary linearly over the beam length to alleviate shear locking [32]. For this reason, the kinematics

6
uses local coordinate systems that describe distinct cross-section rotations due to bending and shear
deformations. The position vector of a material point in the beam element is expressed as

r p = r0 + R3 A2 y + R2 A3 z, (9)

where r0 = r(α1 , 0, 0) is the position vector of the beam’s reference line and vectors y = {0, α2 , 0}T
and z = {0, 0, α3}T define the plane of the cross-section. The position vector of a material point is
defined using third-order interpolation. The shape functions for the element are defined in terms of
the non-dimensional axial variable ξ = 2α1 /L and transverse variables η = 2α2 /H and ζ = 2α3 /W ,
where L, H, and W are the element length, height, and width, respectively. For the present element,
the shape functions are

S1 = (ξ + 1)(ξ − 2)2 /4, S2 = Lξ(ξ − 2)2 /8,

S3 = Hη(ξ + 2)/4, S4 = W ζ(ξ + 2)/4,
2 (10)
S5 = ξ (ξ + 3)/4, S6 = Lξ 2 (ξ − 2)/8,
S7 = Hξη/4, S8 = W ξζ/4.

Rotation matrices A2 and A3 describe the rotations of the cross-section due to bending defor-
mations [32] about the local α2 - and α3 -axes, respectively. The first rotation tensor is written as
A2 = [t̄2 , n̄2 , b̄2 ], where the tangent to the reference line is t̄2 = r ,1 /kr,1 k, b̄2 = (r ,1 × r ,2 )/(kr,1 × r ,2 k)
and n̄2 = (b̄2 × r ,1 )/(kb̄2 × r ,1 k). Similarly, the second rotation tensor is A3 = [t̄2 , n̄3 , b̄3 ], where
n̄3 = (r,3 × r ,1 )/(kr,3 × r ,1 k) and b̄3 = (r,1 × n̄3 )/(kr,1 × n̄3 k).
Rotation tensor R2 gives the additional rotation due to transverse shear deformation along the
local α2 -axis. To obtain this rotation tensor, the kinematics of the ANCF are used to approximate
the shear angle in terms of the slope vectors as γ2 ≈ sin γ2 = (r T,1 r ,3 )/(kr,1 kkr ,3 k). The nodal
(1) (2)
shear strains are denoted γ2 and γ2 and to alleviate shear locking, the shear angle within the
(1) (2)
element is interpolated using linear shape functions, γ2 = N1 γ2 + N2 γ2 , where N1 and N2 are the
conventional linear shape functions. Finally, for small shear angles, the following approximations
are made: R2 ≈ I + γ2 b̃∗2 . A similar development yields rotation tensor R3 that gives the additional
rotation due to transverse shear deformation along the local α3 -axis. Nodal values are evaluated as
γ3 ≈ sin γ3 = −(r T,1 r ,2 )/(kr,1 kkr,2 k) and linear interpolation of nodal values leads to Rz ≈ I + γz b̃∗3 .
The components of the Green-Lagrange strain tensor are obtained readily as

ǫxx = (rTp,1 rp,1 − 1)/2, ǫyy = (r T,2 r,2 − 1)/2, ǫzz = (rT,3 r ,3 − 1)/2,
(11)
ǫxy = (rTp,1 rp,2 )/2, ǫxz = (r Tp,1 r p,3 )/2, ǫyz = (r Tp,2 rp,3 )/2.

For a beam made of a homogeneous isotropic material of Young’s modulus E and shear modulus
G, the strain energy stored in one beam element of volume V is
1
Z
E(ǫ2xx + ǫ2yy + ǫ2zz ) + 4G(k2 ǫ2xz + k3 ǫ2xy + ǫ2yz ) dV,
 
A= (12)
2 V
where k2 and k3 are the shear correction factors.

2.3.3 Three-node ANCF beam element

The second ANCF element to be used in this study is the three-node ANCF beam element intro-
duced by Nachbagauer et al. [33], in which the position vector and sectional rotations are inter-
polated quadratically and linearly, respectively, over the element. Due to a different definition of
the strain energy and an improved integration scheme, the accuracy and convergence order of this
three-node element are better than those of the two-node element described in the previous section.

7
 This three-node element has a long history. Initially, the choice of degrees of freedom was
proposed by Kerkkänen et al. [36] and Garcı́a-Vallejo et al. [37], in which longitudinal slope vectors
were omitted. The strain components are identical to those presented by Gerstmayr et al. [38]
and Nachbagauer et al. [39]. A two-node version of the element was proposed by Matikainen et
al. [40]. Additional elements without longitudinal slope vectors are described by Matikainen et
al. [41]. In addition to these changes in kinematics, the strain energy expression proposed by Simo
was introduced in two- and three-dimensional ANCF by Gerstmayr et al. [38] and Nachbagauer et
al. [39], respectively.
In Nachbagauer et al. [33], the kinematics of the beam element is defined using three nodes,
each of which has nine degrees of freedom. The vector of nodal coordinates for the beam element
is expressed as
n o
eT = r (1)T r (1)T
,2 r
(1)T
,3 r (2)T
r
(2)T
,2 r
(2)T
,3 r (3)T
r
(3)T
,2 r
(3)T
,3 , (13)

(i) (i)
where r (i) are the inertial position vectors of the nodes, and r,2 and r,3 the transverse slope vectors
that define the orientation of the beam’s cross-section. The shape functions for the element are
written as
S1 = ξ(ξ − 1)/2, S2 = HηS1 , S3 = W ζS1,
S4 = ξ(ξ + 1)/2, S5 = HηS4 , S6 = W ζS4, (14)
S7 = 1 − ξ 2 , S8 = HηS7 , S9 = W ζS7,
and matrix S m defined by eq. (7) becomes S m = [S1 I, . . . , S9 I].
 
The kinematics of the element rely on a single rotation tensor, (R R0 ) = b̄1 , b̄2 , b̄3 , where
unit vectors b̄1 , b̄2 , and b̄3 are shown in fig. 1. In the present formulation, the axial unit vector
b̄1 = (r,2 × r,3 )/kr,2 × r,3 k and the transverse unit vectors are selected as b̄2 = r ,2 /kr,2 k, and b̄3 =
(r ,3 × b̄1 )/kr ,3 × b̄1 k. The sectional strain components are now given by eq. (1), where x′0 + u′ = r ,1 .
The strain energy stored in the beam is given by eq. (4), where the section stiffness matrix is
selected as D ∗ = diag(S, K22 , K33 , H11 , H22 , H33 ), where S = EA is axial stiffness, H22 = EI2 and
H33 = EI3 the bending stiffnesses about unit vectors b̄2 and b̄3 , respectively, and H11 = GJ is
torsional stiffness. The shear stiffnesses are K22 = GAk2 and K33 = GAk3 along unit vectors b̄2
and b̄3 , respectively, where k2 and k3 are the shear correction factors.
In the present formulation, the transverse slope vectors, r ,2 and r ,3 , do not remain unit vectors
nor do they remain mutually orthogonal. Hence, they describe cross-sectional deformations and
neglecting Poisson’s effect, the associated strain energy is found as

1 L
Z
2 2 2

Acsd = EA(Eyy + Ezz ) + 2GAEyz dα1 , (15)
2 0

where Eyy = (rT,2 r ,2 −1)/2 and Ezz = (r T,3 r ,3 −1)/2 are the in-plane strains and Eyz = r T,2 r ,3 /2 the in-
plane shear strain, all computed at the beam’s reference line. The beam’s total strain energy is the
sum of two components, A and Acsd , defined by eqs (4) and (15), respectively. To alleviate locking,
selective integration is used in the element: a two-point Gauss and a three-point Lobatto integration
schemes were used to integrate the strain energies defined by eqs. (4) and (15), respectively.

2.4 Summary
This section has summarized in a very succinct manner three beam element formulations that are
widely used in flexible multibody dynamics simulations. The geometrically exact beam formulation
summarized in section 2.2 is a classical formulation that has been used for decades; initially proposed
by Simo et al. [5, 6], then by Borri and Merlini [7], Danielson and Hodges [8, 9], and Bauchau and

8
Hong [10, 11, 12]. These formulations are characterized by the sectional strains and strain energy
defined by eqs. (1) and (4), respectively. The various authors mentioned above have used different
shape functions and various integration schemes.
Furthermore, different parameterizations of rotation and many rational techniques for their in-
terpolation have been proposed [42, 43, 44]. It is also possible to use a redundant representation of
rotation, such as that proposed by Betsch et al. [45, 46, 47], who used nine degrees of freedom to
represent rotation. Six constraint equations are then imposed, because three independent parame-
ters only are required to represent three-dimensional rotations. Yet all these formulations use the
sectional strains and strain energy definitions that characterize geometrically exact beams.
The ANCF element presented in section 2.3.2 is rooted in the original formulation of Shabana
and Yakoub [34, 35] and presents a radical departure from the GEBF. Indeed, rather than using
the sectional strains and strain energy defined by eqs. (1) and (4), respectively, it is based on three-
dimensional expressions for the Green-Lagrange strain components and strain energy defined by
eqs. (11) and (12), respectively. Here again, various authors have used different shape functions
and various integration schemes.
Finally, the formulation presented in section 2.3.3 can be viewed as a GEBF with a redundant
representation of rotation. Indeed, it uses the sectional strains and strain energy defined by eqs. (1)
and (4), respectively, that characterize the GEBF. The two transverse slope vectors use six degrees
of freedom to represent three rotation components. Instead of imposing holonomic constraints,
this formulation enforces these three constraints via the penalty method. Indeed, eq. (15) can be
interpreted as a penalty term that enforces the unit norms and mutual othogonality of the transverse
slope vectors using penalty coefficients EA and GA, respectively.

3 Experimental validation
The various formulations presented in the previous section will be validated by comparing their
predictions against the measurements of the Princeton beam experiment.

3.1 The Princeton beam experiment

The Princeton beam experiment [3, 4] is a study of the large displacement and rotation behavior of
a simple cantilevered beam under a gravity tip load. A straight aluminum (T 7075) beam of length
L = 0.508 m with a rectangular cross-section of nominal thickness t = 3.175 mm and nominal height
h = 12.7 mm was cantilevered at its root and subjected to a static concentrated load P at its tip.
_ _
i3 θ b3
Root
bearing 0o
Cross-section
at beam root
_
i2
-90o +90o
Reference axis
_ of the beam
b2
Applied load
180o (at beam tip)

Figure 2: Configuration of the Princeton beam experiment.

Figure 2 shows an end-view of the test set-up. An inertial frame of reference is selected as
F = [O, I = (ı̄1 , ı̄2 , ı̄3 )] and material frame F B = [O, B = (b̄1 , b̄2 , b̄3 )] is attached at the beam’s
I

9
root section, which is cantilevered into a bearing that allows rotation of the beam about its reference
axis by an angle θ, called the “loading angle.” The gravity load applied at the beam tip is acting in
the opposite direction of unit vector ı̄3 . Variation of the loading angle from 0 to 90 degrees yields
a wide range of nonlinear problems where torsion and bending in two directions are coupled.
Experimental results [3] consist of measurements of the beam’s tip deflection along the material
unit vectors b̄2 and b̄3 , denoted u2 and u3 , respectively, and called the “flapwise” and “chord-
wise displacements,”
 E E E  respectively. Furthermore, the beam’s tip twist was also measured. Let
E
R = b̄1 , b̄2 , b̄3 denote the rotation tensor of the beam’s tip cross-section. In the absence
of tip load, RE (P = 0) = b̄E
 E E
 ET

1 , b̄2 , b̄3 , where b̄3 = 0, sin θ, cos θ , and it then follows that
E E
θ = arctan(R23 (P = 0)/R33 (P = 0)). Under a tip load P , the orientation of the tip section is
E E
defined as arctan(R23 (P )/R33 (P )) and the beam’s tip twist is defined as
E
R23
φ = arctan( E
) − θ. (16)
R33

The procedure used to measure the twist angle experimentally is detailed in the report by Dowell
and Traybar [3].
Data was acquired at loading angles of θ = 0, ±15, ±30, ±45, ±60, ±75, ±90, and 180 degrees.
For a perfect system, symmetry implies that the absolute values of the tip displacements and twist
should be identical for loading angles ±θ. In the experimental setting, these measurements differed,
providing an estimate of their accuracy. Three loading conditions were used, P1 = 4.448 N (1 lb),
P2 = 8.896 N (2 lbs), and P3 = 13.345 N (3 lbs). Figures 3, 4, and 5 show the experimental
measurements for the absolute value of the tip flapwise displacement, chordwise displacement, and
twist, respectively. Symbols indicate the average of the measurements and error bars are also
provided.
Note that the Dowell and Traybar report [3] provides no measurements for loading condition
P2 at loading angles θ = 75 and 90 degrees and for loading condition P3 at loading angles θ = 60,
75, and 90 degrees. A cursory look at fig. 3 reveals that those loading cases would result in large
flapwise deflections, which could generate permanent plastic deformations in the beam. It is likely
that the authors of the study did not want to damage the test article and hence, did not acquire
data at those loading conditions.

3.2 Correlation using linear theory

The linear solution of the problem is found using the shear deformable beam theory described in
structural analysis textbooks such as that of Bauchau and Craig [19]. The tip transverse displace-
ment components are

P L3
 
T PL
u2 = + sin θ, (17a)
3H33 K22
P L3
 
T PL
u3 = + cos θ. (17b)
3H22 K33

Of course, for the linear theory, the tip twist vanishes.

For θ = 0 or 180 and θ = ±90, the beam undergoes planar deformation and elementary formulæ
of Timoshenko beam theory (17) provide the tip deflection in the linear regime. Using the Young’s
modulus of T 7075 aluminum as E = 71.7 GPa and Poisson’s ratio ν = 0.31, hand calculations yield
uT2 = 5.004 and uT3 = 80.034 mm for the chordwise and flapwise tip displacements, respectively,
at loading level P1 . This compares favorably with experimental measurements of uT2 = 5.3594 and
uT3 = 77.635 mm, respectively, resulting in -6.6 and +3.1% error, respectively. In this effort, the

10
dimensions of the cross-section were adjusted slightly to achieve good correlation between measure-
ments and predictions of linear theory in these two cases. The following data was used for this
correlation effort. Beam dimensions: L = 0.508 m, t = 3.2024 mm and height h = 12.377 mm;
material properties: E = 71.7 GPa, Poisson’s ratio ν = 0.31, and shear modulus G = E/2(1 + ν)
= 27.37 GPa.
Because the distributed mass of the beam is far smaller than the applied tip weight, it was
neglected in the simulations. Consequently, rather than rotating the beam, it was kept at a fixed,
vertical orientation at the root, and the direction of the applied load was varied from 0 to 90 degrees.
2.5

DISPLACEMENT, u2 [10-1m] 2.0

1.5

1.0

0.5

0.0
0 15 30 45 60 75 90
LOADING ANGLE [DEG]

Figure 3: Flapwise displacement at the beam tip versus loading angle for three loading conditions.
Experimental measurements: with error bars. Predictions: Linear solution (✷), GEBF (△),
two-node ANCF (▽), and three-node ANCF (♦). Loading condition: P1 , dashed-dotted line, P2 ,
dotted line, P3 solid line.

Figures 3 and 4 show the flapwise and chordwise displacements, respectively, predicted by the
linear theory expressed by eq. (17a) and (17b), respectively. The flapwise displacements are pre-
dicted accurately for the smaller applied load level, P1 , but significant discrepancies are observed
for the P2 and P3 loading levels.
In contrast, the chordwise displacement predictions shown in fig. 4 are poor, even for the lowest
loading level. At a loading angle of 45 degrees, the linear solution underpredicts the measured
displacement by about 27% for loading level P3 . Note that the linear predictions decrease mono-
tonically as the loading angle increases, while the experiments show an initial rise before a steeper
decline. This behavior, which is particularly noticeable for the highest loading level, stems from the
nonlinear coupling between torsion and chordwise bending, a phenomenon that is not captured by
linear theory.
Of course, linear theory predicts a vanishing tip twist angle because for the present configuration,
twisting is attributable to nonlinear effects entirely. For this experimental setup, torsion is the
phenomenon exhibiting the most pronounced nonlinear behavior. Tip twist, per se, is due to
nonlinear effects only, and the indirect effect of twisting on chordwise bending is very important,
as shown in fig. 4.

3.3 Correlation using GEBF

The GEBF uses sectional stiffness parameters as input data. These parameters were computed
from the physical data given in the previous section and the following values were used: axial
stiffness S = 2.84 MN, chordwise bending stiffness H22 = 36.28 N·m2 , flapwise bending stiffness
H33 = 2.43 N·m2 , torsion stiffness H11 = 3.10 N·m2 , flapwise shear stiffness K22 = 0.64 MN, and

11
chordwise shear stiffness K33 = 0.90 MN. The stiffness parameters of the beam’s rectangular cross-
section were evaluated using the three-dimensional elasticity solution developed by Bauchau and
Han [48]. If the strain energy stored per unit span of the beam is evaluated with the help of
these stiffness coefficients, it is identical to that stored in the three-dimensional structure of infinite
length. Different shear coefficients result for the beam’s flapwise and chordwise directions.
The beam was modeled with 12 four-node elements. Each node features six degrees of freedom,
three displacement and three rotation components. Cubic interpolation functions were used for both
displacement and rotation fields. Interpolation of the rotation field is based on the interpolation of
the relative rotation parameter vector [44]. To eliminate shear locking effects, a three-Gauss point
reduced integration scheme was used.
DISPLACEMENT, u3 [10-2m] 2.0

1.6

1.2

0.8

0.4

0.0
0 15 30 45 60 75 90
LOADING ANGLE [DEG]

Figure 4: Chordwise displacement at the beam tip versus loading angle for three loading conditions.
Experimental measurements: with error bars. Predictions: Linear solution (✷), GEBF (△), two-
node ANCF (▽), and three-node ANCF (♦). Loading condition: P1 , dashed-dotted line, P2 , dotted
line, P3 solid line.

The predictions of the GEBF for the three loading cases are shown in figs 3, 4, and 5 that depict
the tip flapwise displacement, chordwise displacement, and twist, respectively. For all loading cases
and loading angles, excellent correlation with experiment is observed.

3.4 Correlation using two-node ANCF beam element

Next, the predictions of the two-node ANCF beam element were correlated with experimental
results. The beam was modeled with eight two-node elements. Twelve degrees of freedom are
defined at each node and include three displacement and nine slope components that describe
cross-sectional rotation and deformation. As explained in section 2.3.2, each two-node element
employs cubic and linear shape functions in the axial and transverse directions, respectively. In
this study, the beam has a rectangular cross-section and volume integrals were performed using full
Gaussian integration: four points in the axial direction and 2 × 2 points over the cross-section.
The predictions the two-node ANCF beam element for the three loading cases are shown in figs 3,
4, and 5 that depict tip flapwise displacement, chordwise displacement, and twist, respectively. For
the tip flapwise displacement, the predictions are in good agreement with the measured data. In
contrast, for the tip chordwise displacement and twist, poor correlation is observed. The predictions
are close to those of linear theory, because the two-node element does not predict torsion correctly,
as shown in fig 5.

12
3.5 Correlation using three-node ANCF beam element
Finally, the predictions of the three-node ANCF beam element were correlated with experimental
results. The beam was modeled with eight three-node elements. Nine degrees of freedom are defined
at each node and include three displacement and six slope components that describe cross-sectional
rotation. Each element uses quadratic and linear shape functions in the axial and transverse direc-
tions, respectively. For the three-node beam, line integrals were evaluated using selective Gaussian
integration: a two-point Gauss and a three-point Lobatto integration schemes were used to integrate
the strain energies defined by eqs. (4) and (15), respectively.
The predictions for the three-node ANCF beam element for the three loading cases are shown in
figs 3, 4, and 5 that depict tip flapwise displacement, chordwise displacement, and twist, respectively.
For all loading cases and loading angles, excellent correlation with experiment data is observed. Note
that the predictions of the three-node ANCF element are nearly identical to those of the GEBF.
This is to be expected since both formulations are nearly identical.
7.0
TWIST ANGLE [10-2 RAD]

6.0

4.0

2.0

1.0

0.0
0 60 90
LOADING ANGLE [DEG]

Figure 5: Twist at the beam tip versus loading angle for three loading conditions. Experimental
measurements: with error bars. Predictions: GEBF (△), two-node ANCF (▽), and three-node
ANCF (♦). Loading condition: P1 , dashed-dotted line, P2 , dotted line, P3 solid line. Note that the
linear solution predicts a vanishing twist angle.

3.6 Convergence study for the three beam elements

To further compare the three elements used in this paper, a convergence study is presented next.
Four different cases will be compared: the GEBF 3-node element (i.e., with quadratic shape func-
tions), the GEBF 4-node element (i.e., with cubic shape functions), the two-node ANCF element,
and three-node ANCF element. For each element, simulations were run with increasingly finer
meshes for loading case P3 . In each case, the reference solution was selected as that obtained with
the finest mesh and relative errors with respect to the reference solutions were then obtained.
Figure 6 shows the result of the convergence study for tip flapwise displacement at loading angle
θ = 0 deg and fig. 7 the corresponding results at loading angle θ = 30 deg. Finally, fig. 8 shows the
relative error in tip chordwise displacement at loading angle θ = 30 deg. Note that all results are
presented on log-log plots.
As expected from the studies of Romero [16] and Bauchau et al. [18], at equal numbers of
degrees of freedom, the predictions of the GEBF elements are more accurate than those of the
ANCF elements.

13
Figure 6: Relative error in the tip flapwise displacement at loading angle θ = 0 deg, loading
conditions P3 : GEBF 3-node element (✷), GEBF 4-node element (⋄), two-node ANCF (▽), and
three-node ANCF (△).

Figure 7: Relative error in tip flapwise displacement at loading angle θ = 30 deg, loading conditions
P3 : GEBF 3-node element (✷), GEBF 4-node element (⋄), two-node ANCF (▽), and three-node
ANCF (△).

4 Conclusions
In this paper, the accuracies of the geometrically exact beam and absolute nodal coordinate formu-
lations were assessed by comparing their respective predictions against experimental data. In the
experiment, a cantilever beam was subjected to coupled flap, lag, and twist deformations.
The predictions of the two-node beam element based on the ANCF did not correlate well with
the experimental data. In fact, for the tip chordwise displacement and twist, predictions were close
to those of linear theory. These poor predictions stemmed from the inability of this formulation
to capture the torsional behavior of the beam accurately. The inadequate modeling of torsional
behavior resulted, in turn, in poor predictions of the coupled chordwise displacements that were
found to be up to 27% in error compared to experimental measurements. This two-node beam
element is rooted in the original ANCF proposed by Shabana and Yakoub. Although accurate
predictions can be obtained for planar beam problems, this study suggests that this formulation
should not be used for three-dimensional beams undergoing coupled bending and torsion.
The numerical predictions of the GEBF and of the recently proposed three-node element based

14
Figure 8: Relative error in tip chordwise displacement at loading angle θ = 30 deg, loading conditions
P3 : GEBF 3-node element (✷), GEBF 4-node element (⋄), two-node ANCF (▽), and three-node
ANCF (△).

on the ANCF agree well with the experimental data at all loading angles and for the three loading
cases. Although the numerical implementations of the elements differ, both elements share the
sectional strain and strain energy definitions that characterize the GEBF. The GEBF presented
here uses a minimum set of variables, three nodal displacements and rotations. In contrast, the
three-node ANCF element uses a redundant set of coordinates: six degrees of freedom are used to
represent rotation of the cross-section. Constraints are enforced via the penalty method.
This study demonstrates the crucial need for thorough validation of the beam elements used
for the simulation of flexible multibody systems before they are used to solve practical problems.
Comparison with available experimental data seems indispensable. Predictions of the original ab-
solute nodal coordinate formulation have been widely presented in the literature, primarily for
planar problems, yet this study shows that this formulation cannot handle three-dimensional beam
problems accurately.

5 Acknowledgements
The authors would like to thank the Academy of Finland (Application No. 259543) for supporting
Marko K. Matikainen.

References
[1] J.C. Houbolt and G.W. Brooks. Differential equations of motion for combined flapwise bending,
chordwise bending, and torsion of twisted nonuniform rotor blades. Technical Report 1348,
NACA Report, 1958.

[2] D.H. Hodges and E.H. Dowell. Nonlinear equations of motion for the elastic bending and
torsion of twisted nonuniform rotor blades. Technical report, NASA TN D-7818, 1974.

[3] E.H. Dowell and J.J. Traybar. An experimental study of the nonlinear stiffness of a rotor blade
undergoing flap, lag, and twist deformations. Aerospace and Mechanical Science Report 1257,
Princeton University, 1975.

15
 [4] E.H. Dowell, J.J. Traybar, and D.H. Hodges. An experimental-theoretical correlation study
of non-linear bending and torsion deformations of a cantilever beam. Journal of Sound and
Vibration, 50(4):533–544, February 1977.

[5] J.C. Simo. A finite strain beam formulation. The three-dimensional dynamic problem. Part I.
Computer Methods in Applied Mechanics and Engineering, 49(1):55–70, 1985.

[6] J.C. Simo and L. Vu-Quoc. A three-dimensional finite strain rod model. Part II: Computational
aspects. Computer Methods in Applied Mechanics and Engineering, 58(1):79–116, 1986.

[7] M. Borri and T. Merlini. A large displacement formulation for anisotropic beam analysis.
Meccanica, 21:30–37, 1986.

[8] D.A. Danielson and D.H. Hodges. Nonlinear beam kinematics by decomposition of the rotation
tensor. Journal of Applied Mechanics, 54(2):258–262, 1987.

[9] D.A. Danielson and D.H. Hodges. A beam theory for large global rotation, moderate local
rotation, and small strain. Journal of Applied Mechanics, 55(1):179–184, 1988.

[10] O.A. Bauchau and C.H. Hong. Finite element approach to rotor blade modeling. Journal of
the American Helicopter Society, 32(1):60–67, 1987.

[11] O.A. Bauchau and C.H. Hong. Large displacement analysis of naturally curved and twisted
composite beams. AIAA Journal, 25(11):1469–1475, 1987.

[12] O.A. Bauchau and C.H. Hong. Nonlinear composite beam theory. Journal of Applied Mechan-
ics, 55:156–163, March 1988.

[13] A.A. Shabana and R.A. Wehage. A coordinate reduction technique for dynamic analysis of
spatial substructures with large angular rotations. Journal of Structural Mechanics, 11(3):401–
431, March 1983.

[14] O.P. Agrawal and A.A. Shabana. Application of deformable-body mean axis to flexible multi-
body system dynamics. Computer Methods in Applied Mechanics and Engineering, 56(2):217–
245, 1986.

[15] A.A. Shabana. Flexible multibody dynamics: Review of past and recent developments. Multi-
body System Dynamics, 1(2):189–222, June 1997.

[16] I. Romero. A comparison of finite elements for nonlinear beams: the absolute nodal coordinate
and geometrically exact formulations. Multibody System Dynamics, 20:51–68, 2008.

[17] J. Gerstmayr, H. Sugiyama, and A. Mikkola. An overview on the developments of the absolute
nodal coordinate formulation. In Proceedings of the Second Joint International Conference on
Multibody System Dynamics, Stuttgart, Germany, May 2012.

[18] O.A. Bauchau, S.L. Han, A. Mikkola, and M.K. Matikainen. Comparison of the absolute
nodal coordinate and geometrically exact formulations for beams. Multibody System Dynamics,
32(1):67–85, June 2014.

[19] O.A. Bauchau and J.I. Craig. Structural Analysis with Application to Aerospace Structures.
Springer, Dordrecht, Heidelberg, London, New-York, 2009.

[20] S.P. Timoshenko. On the correction factor for shear of the differential equation for transverse
vibrations of bars of uniform cross-section. Philosophical Magazine, 41:744–746, 1921.

16
[21] S.P. Timoshenko. On the transverse vibrations of bars of uniform cross-section. Philosophical
Magazine, 43:125–131, 1921.

[22] E. Reissner. The effect of transverse shear deformation on the bending of elastic plates.
Zeitschrift für angewandte Mathematik und Physik, 12:A.69–A.77, 1945.

[23] R.D. Mindlin. Influence of rotatory inertia and shear on flexural motions of isotropic elastic
plates. Journal of Applied Mechanics, 18:31–38, 1951.

[24] E. Reissner. On one-dimensional finite-strain beam theory: the plane problem. Zeitschrift für
angewandte Mathematik und Physik, 23:795–804, 1972.

[25] E. Reissner. On one-dimensional large-displacement finite-strain beam theory. Studies in

Applied Mathematics, 52:87–95, 1973.

[26] E. Reissner. On finite deformations of space-curved beams. Zeitschrift für angewandte Math-
ematik und Physik, 32:734–744, 1981.

[27] L.E. Malvern. Introduction to the Mechanics of a Continuous Medium. Prentice Hall, Inc.,
Englewood Cliffs, New Jersey, 1969.

[28] D.H. Hodges. Nonlinear Composite Beam Theory. AIAA, Reston, Virginia, 2006.

[29] O.A. Bauchau. Flexible Multibody Dynamics. Springer, Dordrecht, Heidelberg, London, New-
York, 2011.

[30] V. Giavotto, M. Borri, P. Mantegazza, G. Ghiringhelli, V. Carmaschi, G.C. Maffioli, and

F. Mussi. Anisotropic beam theory and applications. Computers & Structures, 16(1-4):403–
413, 1983.

[31] O.A. Bauchau and S.L. Han. Three-dimensional beam theory for flexible multibody dynamics.
Journal of Computational and Nonlinear Dynamics, 9(4):041011 (12 pages), 2014.

[32] K.E. Dufva, J.T. Sopanen, and A.M. Mikkola. A three-dimensional beam element based on a
cross-sectional coordinate system approach. Nonlinear Dynamics, 43(4):311–327, 2005.

[33] K. Nachbagauer, P. Gruber, and J. Gerstmayr. Structural and continuum mechanics ap-
proaches for a 3d shear deformable ancf beam finite element: Application to static and
linearized dynamic examples. ASME Journal of Computational and Nonlinear Dynamics,
8(2):021004, 2013.

[34] A.A. Shabana and R.Y. Yakoub. Three dimensional absolute nodal coordinate formulation for
beam elements: Theory. ASME Journal of Mechanical Design, 123:606–613, 2001.

[35] R.Y. Yakoub and A.A. Shabana. Three dimensional absolute nodal coordinate formulation
for beam elements: Implementation and applications. ASME Journal of Mechanical Design,
123:614–621, 2001.

[36] K. S. Kerkkänen, J. T. Sopanen, and A. M. Mikkola. A linear beam finite element based on the
absolute nodal coordinate formulation. Journal of Mechanical Design, 127(4):621–630, 2005.

[37] D. Garcı́a-Vallejo, A. Mikkola, and J.L. Escalona. A new locking-free shear deformable finite
element based on absolute nodal coordinates. Nonlinear Dynamics, 50(1-2):249–264, 2007.

17
[38] J. Gerstmayr, M.K. Matikainen, and A.M. Mikkola. A geometrically exact beam element
based on the absolute nodal coordinate formulation. Multibody System Dynamics, 20(4):359–
384, 2008.

[39] K. Nachbagauer, S.A. Pechstein, H. Irschik, and J. Gerstmayr. A new locking free formulation
for planar, shear deformable, linear and quadratic beam finite elements based on the absolute
nodal coordinate formulation. Journal of Multibody System Dynamics, 26(3):245–263, 2011.

[40] M.K. Matikainen, R. von Hertzen, A.M. Mikkola, and J. Gerstmayr. Elimination of high
frequencies in the absolute nodal coordinate formulation. Proceedings of the Institution of
Mechanical Engineers, Part K: Journal of Multi-body Dynamics, 224(1):103–116, 2010.

[41] M.K. Matikainen, O. Dmitrochenko, and A.M. Mikkola. Beam elements with trapezoidal cross-
section deformation modes based on the absolute nodal coordinate formulation. In International
Conference of Numerical Analysis and Applied Mathematics, Rhodes, Greece, 19-25 September
2010.

[42] M.A. Crisfield and G. Jelenić. Objectivity of strain measures in the geometrically exact three-
dimensional beam theory and its finite-element implementation. Proceedings of the Royal So-
ciety, London: Mathematical, Physical and Engineering Sciences, 455(1983):1125–1147, 1999.

[43] O.A. Bauchau, A. Epple, and S.D. Heo. Interpolation of finite rotations in flexible multibody
dynamics simulations. Proceedings of the Institution of Mechanical Engineers, Part K: Journal
of Multi-body Dynamics, 222(K4):353–366, 2008.

[44] O.A. Bauchau and S.L. Han. Interpolation of rotation and motion. Multibody System Dynamics,
31(3):339–370, 2014.

[45] P. Betsch. The discrete null space method for the energy consistent integration of constrained
mechanical systems. Part I: Holonomic constraints. Computer Methods in Applied Mechanics
and Engineering, 194(50-52):5159–5190, 2005.

[46] P. Betsch and S. Leyendecker. The discrete null space method for the energy consistent integra-
tion of constrained mechanical systems. Part II: Multibody dynamics. International Journal
for Numerical Methods in Engineering, 67:499–552, 2006.

[47] S. Leyendecker, P. Betsch, and P. Steinmann. The discrete null space method for the en-
ergy consistent integration of constrained mechanical systems. Part III: Flexible multibody
dynamics. Multibody System Dynamics, 19(1-2):45–72, 2008.

[48] S.L. Han and O.A. Bauchau. Nonlinear three-dimensional beam theory for flexible multibody
dynamics. Multibody System Dynamics, 34(3):211–242, July 2015.

18