Generalized Timoshenko Theory of

the Variational Asymptotic Beam

Sectional Analysis
Wenbin Yu and Dewey H. Hodges
Abstract
The generalized Timoshenko theory for composite beams embedded in the computer
program VABS has the same structure as Timoshenkos original theory for isotropic beams
without the restrictive assumptions of the original theory. An overview of this theory is
presented to show its general and rigorous framework. Certain theoretical details missing from previous developments are supplied, such as the proof of a kinematical identity
and the expression of the recovery theory in terms of sectional stress resultants. It has
been demonstrated that the VABS generalized Timoshenko theory reproduces the elasticity solution for the flexure problem of isotropic prisms. Numerical results are presented
in support of the long-term validation effort, focusing especially on calculation of sectional stiffnesses (including shear correction factors) and shear center location, making use
of the VABS model for composite beam analysis (including buckling and vibration), and
recovering three-dimensional field variables over the section. The accuracy of the VABS
generalized Timoshenko theory is demonstrated, and some of its practical advantages over
three-dimensional finite element analysis are exhibited.

Presented at the 44th Structures, Structural Dynamics and Materials Conference, Norfolk, Virginia,
April 7 10, 2003

Assistant Professor, Department of Mechanical and Aerospace Engineering, Utah State University, Logan, Utah 84322-4130. Formerly, Post Doctoral Fellow, School of Aerospace Engineering, Georgia Institute
of Technology, Atlanta, Georgia. Email: [email protected].

Professor, School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, Georgia, 303320150. Email: [email protected]. Fellow, AHS.

1 of 43

Nomenclature
h

beam cross-sectional characteristic dimension

wavelength of beam axial deformation

characteristic radius of initial curvatures and twists of the beam

A, B, C, D

matrices carrying both sectional geometry and material information in

the asymptotic energy model

x1

beam axial coordinate

local Cartesian coordinates for the beam section. Greek indices assumes
values 2 and 3 while Latin indices assume 1, 2, and 3. Repeated indices
are summed over their range except where explicitly indicated

( )0

the partial derivative with respect to x1

defined as e
= eijk k , with eijk as the permutation symbol

b 11 1 2 3 cT , column matrix of strain measures as defined in classical beam theory with 11 as the extensional strain, 1 as the torsional
strain, as the bending curvatures around x

b11 1 2 3 cT , column matrix of classical strain measures as defined

in Timoshenko beam theory with 11 as the extensional strain, 1 as
the torsional strain, as the bending curvatures around x

b212 213 cT , column matrix of transverse strain measures (212 and

213 ) defined in Timoshenko beam theory

the position vector of the deformed beam axis

b0 213 212 cT , Rodrigues parameters corresponding to the direction

cosine matrix between classical beam system and Timoshenko beam
system

k1 , k2 , k3

initial twist of the beam and initial curvatures of the beam reference
line

shear correction factor in the x direction

shear stiffness in the x direction

2 of 43

shear modulus of an isotropic material

cross-sectional area

the mass per unit span

i2 , i3

the mass moments of inertia per unit span about coordinates x2 and x3

Introduction
Because of the special geometric feature of beams, in which one dimension is much larger
than the other two, beam modeling has been regarded as an elasticity problem for centuries,
starting with Galileos inquiry (Ref. 1). In the more than three hundred years that followed,
investigators tried to simplify the analysis by taking advantage of the geometric features
to model beams as one-dimensional (1-D) problems. However, to obtain an accurate beam
representation for the physically three-dimensional (3-D) structure, one has to find a way
to reproduce the 3-D energy stored in the structure as accurately as possible. That is, one
must find a way to take into account the eliminated two-dimensional (2-D) cross-sectional
coordinates. The famous Euler-Bernoulli hypothesis was introduced to meet this requirement. Therein it is assumed that a normal cross-sectional plane remains plane and normal
to the reference line when the beam deforms. The theory resulting from this assumption can
account for extension and bending in two directions. To account for torsion, the cross section
cannot in general be constrained to remain plane. However, its shape and size in its own
plane are assumed to be preserved during torsion, and the cross section can warp out of its
plane freely. The simplest beam theories for isotropic beams account for extension, torsion
and bending in two directions based on these assumptions and, in spite of the incorporation
of Saint-Venant torsion, are sometimes called Euler-Bernoulli beam theories.
Refined theories are required for higher accuracy when the wavelength of the beam deformation is not very large compared to the cross-sectional dimensions. For isotropic beams the
next logical step beyond Euler-Bernoulli beam theory is Timoshenko beam theory, in which
there are six fundamental global deformations (bending and transverse shear in two directions, extension, and twist). The assumption of Euler-Bernoulli theory is relaxed. Rather,
3 of 43

a cross section of the undeformed beam normal to the reference line will be, in general,
oblique to the reference line of the deformed beam because of transverse shearing. To ensure
that the reduced 1-D strain energy is equivalent to the original 3-D model over a broader
range of slenderness ratios, shear correction factors are often introduced to modify the shear
stiffness for isotropic beams. These factors typically reduce the transverse shear stiffness of
a cross section relative to what it would have been were there no out-of-plane cross-sectional
warping due to transverse shear.
However, for beams made with generally anisotropic materials, the imposition of such
ad hoc kinematic assumptions can introduce significant errors. To accurately capture the
behavior of composite beams, one must abandon these restrictive assumptions and include
all possible deformation. In the limit of small h/l, the resulting theory governs fully coupled
extension, twist, and bending. Thus, instead of four fundamental stiffnesses there could
be as many as 10 (a fully populated 44 symmetric matrix). Such a generalized EulerBernoulli theory is typically termed classical beam theory and can predict well the static
and low-frequency dynamic behavior of slender, composite beams (Ref. 2). The stiffness
model of classical theory can be refined to take initial twist and curvature into account
without changing the types of deformation or the number of stiffness constants (Ref. 3).
The 3-D transverse shear effects are only taken into account for the calculation of extension,
twist, and bending stiffness constants.
Although for many cases of static and low-frequency dynamic deformation of beams it
is shown in Ref. 4 that an asymptotically correct classical model is at least as accurate as
refined theories based on ad hoc assumptions, classical theory has its limitations. It is not
only possible but desirable to generalize Timoshenko theory for composite beams, so that
transverse shearing beam variables and short-wavelength phenomena can be included in the
model. A significant literature review is beyond the scope of this paper. However, recent
research has resulted in significant advances in this field (Refs. 47).
One of the recent developments pointed out in the more recent review papers is the

4 of 43

computer program VABS (Variational Asymptotic Beam Sectional Analysis), a finite element based cross-sectional analysis for composite beam-like structures developed by Hodges
and his co-workers over the past decade (Refs. 815). VABS takes the Variational Asymptotic Method (VAM) as the mathematical foundation to decouple a general 3-D nonlinear
anisotropic elasticity problem into a linear, 2-D, cross-sectional analysis and a nonlinear,
1-D, beam analysis (Ref. 16). The cross-sectional analysis calculates the 3-D warping functions asymptotically and finds the constitutive model for the 1-D nonlinear beam analysis.
After one obtains the global deformation from the 1-D beam analysis, the original 3-D fields
(displacements, stresses, and strains) can be recovered using the obtained 3-D warping functions.
VABS can perform a classical analysis for beams with initial twist and curvature with
arbitrary reference cross sections. VABS is also capable of capturing the trapeze and Vlasov
effects, which are useful for specific beam applications. VABS is able to calculate the 1D sectional stiffness matrix with transverse shear refinement for any initially twisted and
curved, inhomogeneous, anisotropic beam with arbitrary geometry and material properties.
Finally, VABS can recover asymptotically correct 3-D displacement, stress and strain fields
within a cross section. To the best of the authors knowledge, there does not exist any
other published treatment of nonlinear composite beam modeling with such generality and
versatility. It should be emphasized, however, that the recovery operations within cross
sections that are near beam boundaries, concentrated loads, or sudden changes in the crosssectional geometry along the span are not accurate. Indeed, in these areas one has truly 3-D
behavior which cannot be modeled using a beam model.
To emphasize that ad hoc assumptions such as those used in the original Timoshenko
theory are not invoked in development of the VABS composite beam theory accounting for
transverse shear effects it is referred to as a generalized Timoshenko theory. For composite
beams, instead of six fundamental stiffnesses, there could be as many as 21 terms in a fully
populated 66 symmetric matrix. The purpose of this paper is to explain, validate, and

5 of 43

assess this theory embedded in VABS.

First, an overview of the VABS generalized Timoshenko theory is presented along with
a few theoretical details missing from previous treatments of this topic, such as Refs. 14, 15.
Then, the relation between VABS and elasticity theory is briefly reported as an analytical
validation of the theory, building on Ref. 17. Finally, several examples are presented as
numerical validations in which VABS results are compared with those from the commercial
finite element analysis (FEA) package ANSYS and other approaches in the literature where
available and appropriate.

The Generalized Timoshenko Theory of VABS

Construction of strain energy density

The first step of developing the generalized Timoshenko beam theory of VABS is to
find a strain energy asymptotically correct up to the second order of h/l and h/R. A
complete second-order strain energy is shown to be sufficient for the purpose of constructing
a generalized Timoshenko model because it is generally accepted that the transverse shear
strain measures are one order less than classical beam strain measures (extension, torsion
and bending in two directions). The 3-D strain field in terms of beam strain measures and
arbitrary warping functions can be obtained by formulating the beam kinematics exactly in
an intrinsic fashion. The 3-D warping functions can be solved by VAM asymptotically and
a strain energy asymptotically approximating the 3-D energy up to the second order can be
achieved (Ref. 14). The resulting asymptotically correct strain energy is
T

2U = T A + 2T B0 + 0 C0 + 2T D00

(1)

Although the strain energy expressed in Eq. (1) is asymptotically correct, it is difficult to use
in practical analyses because it contains derivatives of the classical strain measures, which
requires boundary conditions that are more complicated than necessary. Timoshenko beam
theory, commonly used in engineering practice, is free from such drawbacks. Therefore,
6 of 43

the second step is to fit the obtained asymptotically correct strain energy, Eq. (1), into a
generalized Timoshenko model of the form

2U = T X + 2T F s + sT Gs

(2)

Generally, it is impossible to achieve this transformation while keeping the resulting model
asymptotically correct. Although it is noted that Berdichevsky & Staroselskii (Ref. 18),
used changes of variable to achieve a model of the form of Eq. (2), the 1-D transverse shear
strain measures are not equivalent to those commonly used. Hence, direct use of the model
in Ref. 18 could produce misleading results if one does not take this subtlety into account
(Ref. 17). However, one should be able to recover 3-D fields based on that work that are
equivalent to those obtained in the VABS generalized Timoshenko theory.
To ensure that the generalized Timoshenko model represents the original asymptotically
correct model as accurately as possible, all the known information between these two models
should be used. First, the strain measures ( and s ) defined in the generalized Timoshenko
model can be expressed in terms of the strain measures () defined in the asymptotically
correct model. In fact, the strain measures of the asymptotically correct and generalized
Timoshenko models are associated with two different triads, Ti and Bi, respectively. As
shown in Fig. 1, they are related according to

B1

BT
B2 = C

B3

where
C BT

T1

T2

T3

1 212 213

=
1
0
212

213
0
1

(3)

(4)

7 of 43

Based on the definition of 1-D force-strain measures (Ref. 19), the identity

R0 = (1 + 11 )B1 + 21 B = (1 + 11 )T1

(5)

is obtained. Dot multiplying this equality with B1 , making use of Eq. (3), and assuming the
strain components are small, then
11 = 11

(6)

According to Ref. 19, the moment-strain measures can be related by

e 0
I 12
+ C BT KT
KB =
1 T
1 + 4

(7)

with I denoting the 33 identity matrix and

KB = + k
KT = + k

(8)

where and are the column matrices representing the moment-strain measures associated with bases Bi and Ti , respectively. The column matrix k contains the initial twist and
curvatures measured in basis bi . By virtue of the restriction to small strain, the generalized
Timoshenko constitutive model being sought is linear. Thus, Eq. (7) can be rewritten as

= 0 + k C BT k

(9)

which can be written explicitly as

1
0
21 k
1

0
=
+
+
212 k1
2
2
213

3 3 2 0 213 k1

12

(10)

8 of 43

It should be noted that despite the linearity of the constitutive model, which is a byproduct
of the restriction to small strain, the 1-D beam analysis, including both kinematical equations
and equations of motion, is geometrically exact (Ref. 19). Combining Eqs. (6) and (10), a
kinematical identity is obtained between these two sets of strain measures, namely

=  + Q s0 + P s

with

0 0

0 0

Q=

0 1

1 0

(11)

0
0

k2
k3

P =

k1 0

0 k1

(12)

Making use of Eq. (11), the asymptotically correct strain energy (Eq. (1)) can be expressed in terms of the generalized Timoshenko beam strain measures as
2U =T A + 2T AQs0 + 2T AP s+
T

0T

(13)

00

2 B +  C + 2 D

To fit Eq. (13) into the generalized Timoshenko form given by Eq. (2), the derivatives of strain
measures must be expressed in terms of the strain measures themselves. Fortunately, the
1-D equilibrium equations based on the generalized Timoshenko model provide a convenient
way to relate the derivatives of strain measures with the strain measures themselves. Taking
advantage of such relations, the unknowns in Eq. (2) can be calculated and a generalized
Timoshenko constitutive model can be obtained (Ref. 14).
Recovery relations

There are at least two applications of the generalized Timoshenko model. First, the
main application of the constitutive model obtained from VABS is to use it as input for
1-D beam analysis. Although VABS casts the strain energy into a form that has types of
9 of 43

deformation similar to those of the original Timoshenko beam theory, it does not make the
limiting kinematical assumptions of that theory. In fact, VABS considers all possible 3-D
deformation but still creates a seamless connection to traditional beam theories so that the
1-D beam analysis remains essentially the same. The additional 3-D information, which is
eliminated in the construction of the 1-D beam analysis, is included by introducing arbitrary
warping functions and is retained to a degree sufficient to give accurate stress, strain, and
displacement recovery. Note that any general 1-D beam solver can directly make use of the
VABS generalized Timoshenko constitutive model to carry out buckling, dynamic, static
and aeroelastic analyses as long as the 1-D beam solver uses 1-D strain measures that are
equivalent to the standard ones that are used in the VABS development (Ref. 19). A second
application is to use this model to calculate the shear center location for a cross section with
arbitrary geometry and material (Refs. 14, 15).
There are several criteria for evaluation of composite beam theories and modeling approaches. Although it is necessary for a methodology to provide accurate results for the
various types of beam global behavior (i.e., static deflections, natural frequencies, mode
shapes, nonlinear transient behavior, buckling loads, etc.), this is not sufficient. Indeed, it is
misleading to focus only on the 1-D behavior, per se, because an insufficiently detailed study
of published results may lead one to believe that differences among the various published
composite beam theories are insignificant. Actually, the adequacy of a composite beam modeling approach should be measured based on how well it predicts 3-D behavior of the original
3-D structure. Therefore, a full set of recovery relations should be provided to complete the
modeling. Recovery relations are expressions for the 3-D displacements, strains and stresses
in terms of 1-D beam quantities and the local cross-sectional coordinates, x .
Such relations are provided in the VABS generalized Timoshenko model. The recovery
relations of the 3-D strains and stresses in terms of the 1-D generalized strains and their
derivatives are reported briefly in Ref. 15. However, it is difficult to obtain the strain
derivatives if the 1-D analysis is solved by a 1-D finite element method using lower-order

10 of 43

shape functions. To be consistent with the procedure used to construct the generalized
Timoshenko theory, the final version of recovery theory in VABS for 3-D displacements,
strains and stresses is expressed in terms of sectional stress resultants and the applied and
inertial loads using the 1-D equations of motion.
For modeling an initially curved and twisted beam, the warping that is asymptotically
correct up to the order of h/R and h/l can be expressed as

w(xi) = (V0 + V1R ) + V1S 0

(14)

where w(xi ) are the 3-D warping functions, V0 , V1R , and V1S are the asymptotically correct
warping functions for classical modeling, the correction due to initial curvatures/twist, and
the refined warping of the order of h/l, respectively.
The recovered 3-D displacement field of the generalized Timoshenko model can be expressed as
Ui (x1 , x2 , x3 ) = ui(x1 ) + x [Ci (x1 ) i ] + wi (x1 , x2 , x3 )

(15)

where Ui are the 3-D displacements, ui are the 1-D beam displacements, Ci are components
of the direction cosine matrix representing the rotation of beam triads during deformation,
and i is the Kronecker symbol. Strictly speaking, although the warping field has been
calculated only to first order, the strain energy density is asymptotically correct through
the second order in the small parameters. With the first-order warping, the 3-D fields can
only be recovered through the first order. Recovering the 3-D fields up to the second order
requires calculation of the second-order warping field, which means additional complexity
and computation. Here the 3-D results will be recovered based on the first-order warping and
all the other information available. Numerical examples show that such recovery relations
yield accurate results without introducing additional computational cost.

11 of 43

The 3-D strain field can be recovered as follows:

= [(h + R )(V0 + V1R ) +  ]

+ [(h + R )V1S + l (V0 + V1R )] 0
+ l V1S 00

(16)

where the 3-D field is

= b 11 212 213 22 223 33 cT

(17)

All the operators in Eq. (16) can be found in Ref. 14. It is more useful to write the recovery
relations in terms of stress resultants because those quantities and their derivatives can be
obtained through the 1-D equilibrium equations. Denoting S as the 66 stiffness matrix for
the generalized Timoshenko beam model, the generalized strain measures can be obtained
in terms of sectional stress resultants as

t = S 1 F

(18)

where t = b 11 212 213 1 2 3 cT are the generalized 1-D strain measures obtained
from a generalized Timoshenko model, and F = b F1 F2 F3 M1 M2 M3 cT the crosssectional stress and moment resultants. To find the derivatives of stress resultants, the 1-D
nonlinear equilibrium equations can be arranged as

f
e
0
KB
F0 =
F f RF f
fB
ee1 + e
K

(19)

12 of 43

with

= b11 212 213 cT

(20)

e1 = b1 0 0cT

(21)

f = bf1 f2 f3 m1 m2 m3 cT

(22)

where f are the known distributed 1-D generalized applied and inertial loads. Note that the
generalized strains in Eq. (19) can be obtained from Eq. (18). It is clear that Eq. (19) can
be differentiated on both sides to get higher derivatives as

F 00 = (R2 F 0 )F + Rf f 0
F 000 = (R3 + RR0 + 2R0 R R00 )F + (R2 + 2R0 )f + Rf 0 f 00

(23)

Having F 0 , F 00 and F 000 , one can obtain 0t , 00t and 000
t from Eq. (18). Substituting these values
into Eq. (16) and using Eq. (11), the 3-D strain field can be expressed in terms of the known
quantities F , f , f 0 and f 00 from the 1-D beam analysis. Finally, the 3-D stress field can be
obtained using the 3-D constitutive law. The results obtained from use of these recovery
relations are identically the same as from those written in terms of 1-D strains and their
derivatives; see Eq. (16) of Ref. 15.

Validation of the VABS Generalized Timoshenko Theory

The aforementioned generalized Timoshenko theory for initially curved and/or twisted
composite beams has been implemented in the computer program VABS. VABS is a versatile
computer program for design and analysis of composite beam-like structures. Here, more
validation is provided to demonstrate further the accuracy and versatility of this theory.
Validation of VABS can be undertaken in two different ways. One is analytical and
the other numerical. Several examples are given in this section to support the continuing
validation of the generalized Timoshenko theory of VABS. The examples include isotropic
13 of 43

and anisotropic cases. The comparison is made against 3-D elasticity theory, 3-D FEA and
other published results whenever possible and appropriate.
Analytical validation

Analytical validation of VABS is accomplished by establishing the connection between

VABS and 3-D elasticity theory. The first such example appeared in Ref. 15 in which the
variational asymptotic cross-sectional analysis of an elliptic isotropic prism with transverse
shear refinement is solved by using Ritz method. Later, a one-to-one correspondence between
the 3-D elasticity theory and the VABS generalized Timoshenko theory was found (Ref.
17). There, the general formulation from VABS is specialized to treat isotropic, prismatic
beams. The governing differential equations and associated boundary conditions of the VABS
classical model and generalized Timoshenko model are exactly the same as those of the SaintVenant problem and flexure problem in 3-D elasticity theory, respectively. This discovery
demonstrates that VABS yields exact results for shear correction factors and shear center
location for isotropic prismatic beams. Therefore, any existing model that does not produce
the same results as VABS does for these kinds of beams should be considered inferior. The
fact that VABS reproduces the results of elasticity theory clearly confirms the fact that
VABS avoids the difficulties of dealing with 3-D elasticity while at the same time obtains
results that agree very well with exact solutions.
Although it may not be possible to analytically validate the general theory of VABS for
anisotropic beams in the same way, it is a natural deduction based on the derivation of Ref.
17 that the results for generally anisotropic beams should be the same as those calculated
by methods based on 3-D anisotropic elasticity theory, such as 3-D FEA. Indeed, VABS may
be considered as a means for going beyond those limits when considering the cross-sectional
analysis of beam-like structures just as 3-D FEA allows one to go beyond the limitations of
3-D elasticity. When coupled with a suitable 1-D beam solver, 3-D stresses on the interior
of a beam are obtained for a computational cost that is two to three orders of magnitude
less than that of 3-D FEA.
14 of 43

Numerical validation examples

Another means of validation is to compare results from VABS with those published and
with 3-D FEA. Most of the numerical examples presented in previous publications on VABS
are validations of this kind; see Refs. 9,10,1215. In Ref. 15 some VABS results are assessed
against the 3-D FEA package ABAQUS.
In this section, several numerical examples will be investigated to demonstrate the accuracy and advantages of the VABS generalized Timoshenko theory. The problems related
with generalized Timoshenko modeling of composite beams can be classified into the following three groups:
1. Obtaining the matrix of cross-sectional stiffness constants for general beams, particularly the transverse shear stiffnesses or shear correction factors for isotropic sections.
2. Locating the shear center, especially for structures that are anisotropic and not thinwalled.
3. Recovering 3-D results for the purpose of detailed analysis.
Stiffness model and shear correction factors. One of the main outcomes of a generalized
Timoshenko model is a 66 stiffness model including the transverse shear stiffnesses for a
cross section. Particularly, for isotropic sections, shear correction factors are often used to
obtain the equivalent shear stiffness to be used for a Timoshenko beam analysis. These
factors are defined as
c =

S
G

(24)

Taking for example a rectangular section with a length of 2a in the x2 direction and 2b
in the x3 direction, the factor 5/6 is normally used in engineering practice. However, this
number is only correct for the x2 direction when b << a and/or the Poissons ratio = 0.

15 of 43

The exact factor can be obtained by the VABS generalized Timoshenko theory as

c1
2

6
= +
5

1+

2 "

18 X tanh(m)
1

54 5 5 m=1
m5

(25)

with = a/b. As mentioned in Ref. 17, Eq. (25) is exactly the same as the flexure problem of
elasticity theory. The same result was also obtained independently in Refs. 20,21. There are
other values proposed in the literature (Refs. 18, 22, 23), but these are only approximations
to the exact value and not as accurate as what VABS calculates.
Eq. (25) is obtained by mathematically solving the governing differential equations of
the VABS generalized Timoshenko theory. Letting 2a = 1 in., 2b = 2 in., G = 109 psi and
= 0.3, then the exact value of c2 is 0.784442. To study that the computer program VABS
is a faithful implementation of the theory behind it, seven different meshes: 12, 24, 36,
48, 510, 612 and 1020, where the first number is the number of elements along x2
and the second is the number of elements along x3 , are created for this purpose. The section
is meshed by ANSYS shell93 elements and this mesh is imported to VABS using specialpurpose macros created for interfacing VABS and ANSYS. The results for the relative error
of c2 with respect to the exact result are plotted in Fig. 2 versus the number of elements
along x2 on a log-log scale. It is apparent that the numerical results from VABS converge
monotonically to the exact result as one refines the mesh. Even if the mesh is only comprised
of 8 elements, the difference between the VABS numerical result and the exact solution is
less than 1%.
To demonstrate that VABS can calculate the shear correction factors correctly for arbitrary sections, an irregular section as sketched and meshed in Fig. 3 is studied. VABS results
are compared with those of Ref. 24 and ANSYS beam capability (beam 188/189) in Table 1.
VABS results agree with those from Ref. 24 to within 0.2%, while ANSYS results are off by
as much as 1.3%. Note that the results from ANSYS are independent of Poissons ratio and
so can be only considered as approximations to the exact solution when = 0. Although

16 of 43

it is said in the ANSYS manual that the cross-sectional capability is valid for composite
sections, the word composite just means that one can build up several isotropic sections
into an arbitrary section. It is not applicable to sections made with anisotropic materials. It
should be noted that the work of Ref. 24 is devoted to calculate the shear correction factors
only for arbitrary isotropic sections, which is much less versatile than VABS because this
capability is just a small subset of the VABS functionalities.
Usually, when the six global deformations are uncoupled, it is sufficient to obtain the
shear correction factors instead of calculating the 66 stiffness matrix. However, more
often than not, the global deformations are coupled, and the full stiffness matrix is needed.
Especially for composite beams, the stiffness matrix is generally coupled (Refs. 14,15). Here
a more complex composite beam with a cross section described in Fig. 4 is modeled by
VABS. This cross section is comprised of two straight strips and two half circles, each of
which is made with two laminated layers. The material properties are listed in Table 2,
and the dimensions are given in Fig. 4. The cross-sectional properties are listed in Table 3,
from which extension-twist and shear-bending couplings are evident. The results labelled as
SVBT are produced by a computer program based on Ref. 25, a generalized application of
the Saint-Venant approach. Therein, the behavior of all quantities versus the axial coordinate
is represented by polynomials of zeroth and first degree, while the cross-sectional variations
are handled by means of finite element approximation. Although SVBT is not based on
asymptotic methods, the results from it and from VABS agree very well for most cases
compared to date. The generalized Saint-Venant methodology, however, is limited to linear
problems and lacks rigorous connections to treatments of end effects, such as are modeled
by Vlasov theory.
Based on the above, in combination with previously published demonstrations, it can be
concluded that the generalized Timoshenko stiffness model of VABS can be used confidently
to carry out structural analysis for a wide variety of beam cross-sectional geometries and
materials.

17 of 43

Locating the shear center. Another important outcome of generalized Timoshenko modeling is the shear center location. For thin-walled sections this is straightforward, but for
arbitrary sections there is not much published information. Although it is not a common
practice to carry out generalized Timoshenko modeling to find the shear center, it is indeed
very simple to accurately locate the shear center after an accurate 66 stiffness model such
as that from VABS is obtained.
The first example is an isotropic open channel-like section with unequal edges and equal
thickness (see Fig. 5) studied in Ref. 26. VABS results are compared against those of Ref.
26 and ANSYS. As shown in Table 4, the three sets of results agree within 1.7% for c2 and
to three significant figures for c3 .
Another example studied in Ref. 26 is a closed section with three cells (see Fig. 6 for
geometry and mesh). Again the results from VABS (Ref. 26) and ANSYS are compared in
Table 5 and agreement to within 0.5% is found.
While most research concerning the shear center is focused on isotropic, thin-walled
sections, VABS can also locate the shear center for generally anisotropic cross sections,
whether or not they are thin-walled. On this subject not much work can be found in the
literature.
Beam analysis with the VABS generalized Timoshenko model. Composite beam modeling
does not stop at the stage of obtaining the cross-sectional properties such as beam stiffness
matrix (including shear correction factors), and shear center location, etc. The ultimate
goal of beam modeling is to use the obtained sectional properties to carry out the 1-D beam
analysis to predict the global behavior, such as deflections, buckling load, natural frequencies
and so on. Here, we assume the cross section in Fig. 4 is cut from a cantilever beam with

18 of 43

length L = 20 in. The sectional mass properties obtained by VABS are

= 7.42513 102 lb/in.

i2 = 9.88336 103 lb.in.
i3 = 6.11827 102 lb.in.

(26)

The VABS stiffness model (Table 3) and sectional inertial properties in Eq. (26) are used
along with a 1-D nonlinear beam solver; here DYMORE (Ref. 27) is used to carry out a 1-D
dynamic analysis to obtain the natural frequencies and a 1-D buckling analysis to obtain
the critical Euler buckling load (compression). The natural frequencies for the first three
bending modes are listed in Table 6, where bi denotes the ith bending mode in the direction
of x . Model 1 is the generalized Timoshenko model given in Table 3, and Model 2 is the
corresponding classical model. It should be noted that the 44 matrix of classical stiffness
constants can be found by inverting the 66 cross-sectional stiffness matrix from the generalized Timoshenko model, removing the two columns associated with transverse shear, and
then inverting the resulting 44 matrix. Table 6 shows that there is a significant difference
between the natural frequencies obtained from the generalized Timoshenko and classical
models. It is clear that the classical model lacks predictive capability for the higher bending
modes and that to obtain accurate second and higher flapping and lagging frequencies for a
composite rotor blade, one should use the generalized Timoshenko model. Notice that there
is no difference between the two models for torsional frequencies.
A static stability problem has also been studied for this composite beam. The tip deflection is excited by a very small tip shear force. As apparent from the plots, when the
compression load approaches the critical load, there is an abrupt change of the tip deflection. There is a small difference (3.5%) between the results obtained from classical and
generalized Timoshenko models (see Fig. 7).
Recovering 3-D results. Another capability of VABS is that it can recover the asymptot-

19 of 43

ically correct distributions of the 3-D fields (displacements, strains, and stresses) over the
cross section after obtaining the 1-D global behavior, which is very important for analyzing
the failure of some critical areas. For demonstration purposes, consider a composite beam of
length 5 inches with a rectangular cross section, dimensions of which are b=0.25 inches, and
h=1 inch (Fig. 8). The material properties are given in Table 7, and the layup is a repeating
quasi-isotropic pattern with 80 layers.
A unit shear force was applied in the x3 direction at one end, and the other end was
constrained to have zero displacement at every node. This means that the VABS model
requires an input of F3 =1 lb, and M2 =2.5 lb-in for the purpose of recovering the stress
components on the cross section at the mid-span of the beam (x1 = 2.5 in); specifically, the
stress along the line x2 = 0 for various x3 locations. For the ANSYS run, the total number
of elements was 25,600, and about one hour of computer time was required. For the VABS
run, the number of elements was 640, and the computer time was less than two seconds. As
can be seen for stress components 13 and 12 in Figs. 9 and 10, respectively, the results from
VABS are almost on top of those of ANSYS.
The 3-D stress distributions clearly identify which areas within a cross section are experiencing extreme stresses. This capability of VABS enables the designer to make certain
adjustments to avoid possible damage in the structure before it is actually built and tested.
This means that the cost of a composite beam structure may be reduced. It should be
emphasized, however, that the VABS analysis can only provide accurate stresses away from
beam boundaries, concentrated loads, and sudden changes in the cross-sectional geometry
along the span. In these areas one has truly 3-D behavior, and the structure does not behave
as a beam.

Conclusions
An overview of the VABS generalized Timoshenko theory has been presented so that the
generality and rigor of the framework can be emphasized and exhibited. Certain theoretical
derivations that were missing from earlier publications have been presented. These include
20 of 43

a proof of the kinematical identity and a presentation of the recovery theory in terms of
sectional stress resultants. A complete formulation of the theory is now available. Examples
have been presented to demonstrate that the generalized Timoshenko theory in VABS can
reproduce the results of elasticity theory, accurately find the shear correction factors, and
locate the shear center for beams made from anisotropic materials. It is also shown that
the generalized Timoshenko model obtained from VABS can be used to carry out 1-D beam
analyses such as buckling, vibration, etc. Significant differences are found between the results
from the classical beam model and the generalized Timoshenko model. VABS can also be
used to recover the 3-D distribution over the cross section with much less modeling and
computational time relative to 3-D FEA.
The VABS generalized Timoshenko theory provides an accurate prediction of the behavior
of the original 3-D beam structure because it asymptotically approximates the 3-D elasticity
theory for the interior behavior of beams. It can predict the required results with much
less labor time and computational time. Designers should be able to use this tool at both
preliminary and detailed design stages to carry out needed tradeoffs more effectively, so that
better and more cost-effective composite beam-like structures can be produced.

Acknowledgements
This work was supported in part by the National Rotorcraft Technology Center and the
Rotorcraft Industry Technology Association through the Georgia Tech Center of Excellence
in Rotorcraft Technology. The authors gratefully acknowledge the help of Eduardo D. Fuchs
running ANSYS.

References
1

Love, A. E. H., A Treatise on the Mathematical Theory of Elasticity, Dover Publica-

tions, New York, New York, 4th ed., pp. 13.

2

Hodges, D. H., Atlgan, A. R., Cesnik, C. E. S., and Fulton, M. V., On a Simpli-

21 of 43

fied Strain Energy Function for Geometrically Nonlinear Behaviour of Anisotropic Beams,
Composites Engineering, Vol. 2, (57), 1992, pp. 513526.
3

Cesnik, C. E. S., Hodges, D. H., and Sutyrin, V. G., Cross-Sectional Analysis of

Composite Beams Including Large Initial Twist and Curvature Effects, AIAA Journal ,
Vol. 34, (9), Sept. 1996, pp. 19131920.
4

Volovoi, V. V., Hodges, D. H., Cesnik, C. E. S., and Popescu, B., Assessment of Beam

Modeling Methods for Rotor Blade Applications, Mathematical and Computer Modelling,
Vol. 33, (1011), 2001, pp. 10991112.
5

Hodges, D. H., A Review of Composite Rotor Blade Modeling, AIAA Journal ,

Vol. 28, (3), March 1990, pp. 561565.

6

Cesnik, C. E. S., Cross-Sectional Analysis of Initially Twisted and Curved Composite

Beams, Ph.D. thesis, Aerospace Engineering, Georgia Institute of Technology, May 1994.
7

Jung, S. N., Nagaraj, V. T., and Chopra, I., Assessment of Composite Rotor Blade

Modeling Techniques, Journal of the American Helicopter Society, Vol. 44, (3), July 1999,
pp. 188205.
8

Cesnik, C. E. S. and Hodges, D. H., Stiffness Constants for Initially Twisted and

Curved Composite Beams, Applied Mechanics Reviews, Vol. 46, (11, Part 2), 1993,
pp. S211S220.
9

Cesnik, C. E. S. and Hodges, D. H., Variational-Asymptotical Analysis of Initially

Twisted and Curved Composite Beams, International Journal for Engineering Analysis
and Design, Vol. 1, (2), Apr. 1994, pp. 177187.
10

Cesnik, C. E. S. and Hodges, D. H., Stiffness Constants for Composite Beams Including

Large Initial Twist and Curvature Effects, Applied Mechanics Reviews, Vol. 48, (11, Part
2), 1995, pp. S61S67.
11

Cesnik, C. E. S. and Hodges, D. H., VABS: A New Concept for Composite Rotor

22 of 43

Blade Cross-Sectional Modeling, Journal of the American Helicopter Society, Vol. 42, (1),
Jan. 1997, pp. 2738.
12

Popescu, B. and Hodges, D. H., On Asymptotically Correct Timoshenko-Like

Anisotropic Beam Theory, International Journal of Solids and Structures, Vol. 37, (3),
1999, pp. 535558.
13

Popescu, B., Hodges, D. H., and Cesnik, C. E. S., Obliqueness Effects in Asymptotic

Cross-Sectional Analysis of Composite Beams, Computers and Structures, Vol. 76, (4),
2000, pp. 533543.
14

Yu, W., Hodges, D. H., Volovoi, V. V., and Cesnik, C. E. S., On Timoshenko-Like

Modeling of Initially Curved and Twisted Composite Beams, International Journal of Solids
and Structures, Vol. 39, (19), 2002, pp. 51015121.
15

Yu, W., Volovoi, V. V., Hodges, D. H., and Hong, X., Validation of the Variational

Asymptotic Beam Sectional Analysis, AIAA Journal , Vol. 40, (10), Oct. 2002, pp. 2105
2113.
16

Berdichevsky, V. L., Variational-Asymptotic Method of Constructing a Theory of

Shells, Journal of Applied Mathematics and Mechanics, Vol. 43, (4), 1979, pp. 664687.
17

Yu, W. and Hodges, D. H., Elasticity Solutions versus Asymptotic Sectional Analysis

of Homogeneous, Isotropic, Prismatic Beams, Journal of Applied Mechanics, Vol. 71, (1),
2004, pp. 1523.
18

Berdichevsky, V. L. and Staroselskii, L. A., On the Theory of Curvilinear Timoshenko-

Type Rods, Journal of Applied Mathematics and Mechanics, Vol. 47, (6), 1983, pp. 809817.
19

Hodges, D. H., A Mixed Variational Formulation Based on Exact Intrinsic Equations

for Dynamics of Moving Beams, International Journal of Solids and Structures, Vol. 26,
(11), 1990, pp. 12531273.
20

Renton, J. D., Generalized Beam Theory Applied to Shear Stiffness, International

Journal of Solids and Structures, Vol. 27, (15), 1991, pp. 19551967.

23 of 43

21

Pai, P. F. and Schulz, M. J., Shear Correction Factors and an Energy-Consistent Beam

Theory, International Journal of Solids and Structures, Vol. 36, (14), 1999, pp. 15231540.
22

Stephen, N. G., Considerations on Second Order Beam Theories, International Jour-

nal of Solids and Structures, Vol. 17, (3), 1981, pp. 325333.
23

Cowper, G. R., The Shear Coefficient in Timoshenkos Beam Theory, Journal of

Applied Mechanics, Vol. 33, (2), June 1966, pp. 335340.

24

Gruttmann, F. and Wagner, W., Shear Correction Factors in Timoshenkos Beam

Theory for Arbitrary Shaped Cross-sections, Computational Mechanics, Vol. 27, (3), 2001,
pp. 199207.
25

Giavotto, V., Borri, M., Mantegazza, P., Ghiringhelli, G., Carmaschi, V., Maffioli,

G. C., and Mussi, F., Anisotropic Beam Theory and Applications, Computers and Structures, Vol. 16, (14), 1983, pp. 403413.
26

Paz, M., Strehl, C. P., and Schrader, P., Computer Determination of the Shear Center

of Open and Closed Sections, Computers and Structures, Vol. 6, (2), 1976, pp. 117125.
27

Bauchau, O., Computational Schemes for Flexible, Nonlinear Multi-Body Systems,

Multibody System Dynamics, Vol. 2, (2), 1998, pp. 169225.

24 of 43

List of Tables
1
2
3
4
5
6
7

Shear correction factors for an arbitrary section . . . . . . . . . . . .

Properties of the anisotropic pipe section . . . . . . . . . . . . . . . .
The sectional stiffness matrix of the anisotropic pipe . . . . . . . . .
Shear center for the channel section . . . . . . . . . . . . . . . . . . .
Shear center for the 3-cell closed section . . . . . . . . . . . . . . . .
Natural frequencies (rad./sec.) from different models . . . . . . . . .
Material properties of anisotropic beam with rectangular cross section

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

26
27
28
29
30
31
32

25 of 43

Table 1:

Shear correction factors for an arbitrary section

c2 /a

c3 /a

Resources
VABS
Ref. 24
ANSYS
VABS
Ref. 24
ANSYS

0
0.7404
0.7395
0.7402
0.6780
0.6767
0.6778

0.25
0.7367
0.7355
0.7402
0.6764
0.6753
0.6778

0.5
0.7306
0.7294
0.7402
0.6736
0.6727
0.6778

26 of 43

Table 2:

Properties of the anisotropic pipe section

right wall:
left wall:
upper wall:
lower wall:
Material properties:
Et = 1.42 106 psi
lt = tn = 0.42

[45 / 45 ]
[45 / 45 ]
[90 /0 ]
[90 /0 ]
El = 20.59 106 psi
Glt = Gtn = 8.7 105 psi
= 0.057 lb/in.3

27 of 43

Table 3:

The sectional stiffness matrix of the anisotropic pipe

Stiffness
VABS
S11 (lb)
1.03890 107
S22 (lb)
7.84299 105
S33 (lb)
3.29002 105
S14 (lb-in.)
9.82878 104
S25 (lb-in.) 8.18782 103
S36 (lb-in.) 5.18541 104
S44 (lb-in.2 )
6.86973 105
2
S55 (lb-in. )
1.88236 106
S66 (lb-in.2 )
5.38972 106

SVBT
1.03892 107
7.85310 105
3.29279 105
9.84575 104
8.21805 103
5.20981 104
6.87275 105
1.88238 106
5.38987 106

Rel. error (%)

0.0019%
0.13%
0.084%
0.17%
0.37%
0.45%
0.044%
0.0011%
0.0028%

28 of 43

Table 4:

Shear center for the channel section

Resources
VABS
Ref. 26
ANSYS

c2 (in.) c3 (in.)
-0.176
0.186
-0.179
0.186
-0.177
0.186

29 of 43

Table 5:

Shear center for the 3-cell closed section

Resources
VABS
Ref. 26
ANSYS

c2 (in.) c3 (in.)
4.356
1.000
4.337
1.000
4.356
1.001

30 of 43

Table 6:

Natural frequencies (rad./sec.) from different models

Modes
b13
b12
b23
b22
b33
b32

Model 1
8.417 102
1.410 103
4.488 103
7.244 103
1.054 104
1.665 104

Model 2
8.693 102
1.465 103
5.423 103
8.932 103
1.507 104
2.399 104

Difference
3.3%
3.9%
20.8%
23.2%
43.0%
44.1%

31 of 43

Table 7:

Material properties of anisotropic beam with rectangular cross section

Layup:
Material properties:
Et = 1.42 106 psi
lt = tn = 0.42

[(45/ + 45/0/90)10]s
El = 20.59 106 psi
Glt = Gtn = 8.7 105 psi
= 0.057 lb/in.3

32 of 43

List of Figures
1
2
3
4
5
6
7
8
9
10

Coordinate systems used for transverse shear formulation . . . . . . . . . . .

Relative error for shear correction factor c2 showing convergence with mesh
refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Geometry and mesh of an arbitrary cross section . . . . . . . . . . . . . . . .
Sketch of the section of an anisotropic pipe . . . . . . . . . . . . . . . . . . .
Sketch of an open section . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cross section having three closed cells . . . . . . . . . . . . . . . . . . . . . .
Transverse Deflection at the tip of of the Anisotropic Pipe . . . . . . . . . .
Schematic of rectangular beam cross section . . . . . . . . . . . . . . . . . .
Stress component 13 at mid-span and x2 = 0 . . . . . . . . . . . . . . . . .
Stress component 12 at mid-span and x2 = 0 . . . . . . . . . . . . . . . . .

34
35
36
37
38
39
40
41
42
43

33 of 43

T3

B3
2

13

T1

B1

Figure 1:

Coordinate systems used for transverse shear formulation

34 of 43

-2

Normalized Factor c2

-4

-6

-8

-10

0.5

1
1.5
Elements Along x2

Figure 2:
Relative error for shear correction factor c2 showing convergence with
mesh refinement

35 of 43

x3
a

2a

2a

2a

x2

Figure 3:

Geometry and mesh of an arbitrary cross section

36 of 43

x3

0.3
45o

45o

-45o

-45o

90

x2

0.4

90o
0o

Figure 4:

Sketch of the section of an anisotropic pipe

37 of 43

x3

0.5

0.5

0.06

1.0

0.06

0.06
x2

Figure 5:

Sketch of an open section

38 of 43

x3

0.1
2
0.05

x2

Figure 6:

Cross section having three closed cells

39 of 43

Transverse Deflection @inD

0
11000

11100

11200 11300 11400 11500

Axial Compression Forces @lbD

11600

11700

Figure 7:
Distribution of the 3-D stress 13 through the thickness. Solid line: exact
solution; dots: VAPAS; dashed line: FOSDT; long-dash/short-dash line: CLT.

40 of 43

Figure 8:

Schematic of rectangular beam cross section

41 of 43

Transverse shear stress t13 (psi)

VABS

ANSYS

0
-0.5

-0.4

-0.3

-0.2

-0.1

0.1

0.2

0.3

0.4

0.5

Location x 3

Figure 9:

Stress component 13 at mid-span and x2 = 0

42 of 43

30

Transverse shear stress t12 (psi)

20

VABS

ANSYS

10

-10

-20

-30
-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

Location x 3

Figure 10:

Stress component 12 at mid-span and x2 = 0

43 of 43