Original Article

Proc IMechE Part C:

J Mechanical Engineering Science
Buckling analysis of beam structure with 0(0) 1–8
! IMechE 2020
absolute nodal coordinate formulation Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/0954406220947117
journals.sagepub.com/home/pic

Jia Wang1 and Tengfei Wang2

Abstract
Absolute nodal coordinate formulation (ANCF) was applied to the buckling analysis. A delicate analysis scheme based on
dichotomy method was proposed to solve the buckling problem with beam elements whose tangent stiffness matrix is a
highly nonlinear function of nodal coordinates. Three existing planar beam elements are employed to show the appli-
cation. The accuracy and capability of the ANCF beam for buckling analysis was validated with benchmark cases.
Additionally, the influence of the shear effect on the buckling load is thoroughly investigated by comparing the solutions
associated with different shear stiffness and slenderness.

Keywords
Structure buckling, absolute nodal coordinate formulation, dichotomy method
Date received: 21 March 2020; accepted: 13 July 2020

Introduction
buckling analysis. The intention of present paper is to
The absolute nodal coordinate formulation (ANCF) propose such method for buckling analysis with
developed by Shabana1 in which the global position ANCF beam elements. The performance of the
and slope vectors are used to describe the element ANCF beams in the buckling application will be thor-
kinematics has gained a wide range of applications. oughly investigated.
Compared to classical nonlinear beam elements, The buckling of a column under the axial compres-
ANCF element generally leads to a constant mass sive force has been thoroughly investigated over cen-
matrix and zero Coriolis and centrifugal forces turies. The pioneering contribution of Euler has to be
which is advantageous in the dynamics modeling of mentioned. In his monograph,4 he determined the
the multi-body system. The simple inertia description critical load and studied the deformed configurations
makes ANCF such attractive that a series of beam of a column with several loading and boundary con-
elements based on ANCF emerge. The convergence ditions. Besides the bending deformation, the elastic
and accuracy of ANCF beam elements has been val- beam normally also undergoes shearing and compres-
idated by both static and dynamic analyses while its sion. By introducing the influence of shear force to the
performance in the buckling problem is seldom dis- lateral deflection, Timoshenko5 obtained the critical
cussed. Nachbagauer and Gerstmayr2 applied their load of the shear-deformable cantilever beam. In
element to the buckling analysis, the lateral critical order to obtain exact solutions, the accurate geomet-
buckling load of the right angle beam structure is ric description of beam deformation is necessary since
determined using a nonlinear static iterative method. the shape of the structure may change substantially
To induce buckling, a perturbation force has to be when buckling occurs. Hummer predicated the criti-
applied according to the buckling mode. Thus, the cal loads of the column with four classical boundary
predicable buckling mode is necessary to correctly
apply the perturbation. Luo et al.3 discussed the per- 1
School of Mechanical Engineering, Shenyang University of Technology,
turbation technique, the strategy for nonlinear buck- Shenyang, China
2
ling analysis was proposed. The perturbation would State Key Laboratory for Strength and Vibration of Mechanical
inevitably destroy the initial configuration of the Structures, School of Aerospace Engineering, Xi’an Jiaotong University,
Xi’an, China
structure. Consequently, the buckling character
Corresponding author:
becomes different. Therefore, it is necessary to devel- Jia Wang, School of Mechanical Engineering, Shenyang University of
op a general method that does not need specify the Technology, No. 111, Shenliao West Road, Shenyang, China.
buckling modes and perturbations in advance for the Email: [email protected]
2 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

conditions.6 The equilibrium equation is derived two other element formulations, i.e., the enhanced con-
based on Reissneerum geometrically exact relation tinuum mechanics and the higher order beam element,
for a planar beam, in which the stress resultants which have been proved to alleviate the locking prob-
depend linearly on the generalized strain measures. lem would be cited here to exclude the influence of the
The fundamentals of the structural buckling has locking on the buckling analysis. In this section, the
been studied in detail.7 Generally, three methods strain energy rather than the elastic force formulation
have been successfully used in determining the critical is presented since it is more straightforward to be used
load for the elastically instable structure. The equilib- in the following buckling analysis based on the energy
rium approach4,6 is the most classical one, that can criterion.
always give exact solution by virtue of accurate
description of beam deformation. In the case of com- Continuum mechanics-based formulation (CMF)
plex structures, difficulties will be encountered when
deducing the equilibrium equation due to the lack of For the planar shear-deformable beam, the orientation
properly specified deformed configuration. The of the cross-section of the beam does not remain per-
second method is usually called the imperfection pendicular to the tangent vector of the neutral axis.
method, the defection response with respect to exter- According to Omar and Shabana,10 the gradient
@r
nal load can be traced by the nonlinear iterative vector @y is used to define the orientation of the
method2 or more elegantly with the arc-length cross-section and a planar shear deformable ANCF
method which can capture the snap through phenom- beam was proposed firstly. The displacement field for
enon. While it is worth to note that the imperfect this element can be defined as
approach should not be associated with the stability " # " #
of the perfect system, but simply characterize the r1 a0 þ a1 x þ a2 y þ a3 xy þ a4 x2 þ a5 x3
r¼ ¼
response of the imperfect system. The third approach r2 b0 þ b1 x þ b2 y þ b3 xy þ b4 x2 þ b5 x3
based on the energy criterion of stability can be stated
(1)
as: if the total potential has a relative minimum at an
equilibrium position, then the configuration is stable.7 The global position vector of an arbitrary point on
The energy approach has been widely used in finite the beam can be written as
element method to solve linear buckling problems in
the commercial software such as ANSYS. The stiff-  
r1
ness matrix of the classical finite element is a linear r¼ ¼ Sq (2)
r2
function of the axial force. The buckling problem is
solved as an eigenvalue problem. In contrast to the where S is the shape function matrix, q is the vector of
classical finite element, the stiffness of ANCF ele- the element nodal coordinates defined as
ments is a function of nodal coordinates with high  
nonlinearity. As a result, special strategies have to @ri @ri @rj @rj
q ¼ ri rj (3)
be conducted for the buckling analysis. @x @y @x @y
An elegantly designed scheme for buckling analysis
with ANCF element is proposed in the present paper. Where, the subscript i and j indicate the two tip nodes
The paper is focused on the flexural buckling since the of the beam element. The matrix of the displacement
analysis is based on the planar ANCF beam element, gradients F is defined as
which means the three dimensional torsional buckling
issues are not discussed. It is worth to note that the 2 3
@r1 @r1 2 31
scheme can also be used with other nonlinear formu- @r 6 @x @y 7
@r01 @r01
6 74 @x @y
5
lations such as the geometrically exact formulation.8,9 F¼ ¼6 7 (4)
@r0 4 @r2 @r2 5 @r02 @r02
@x @y
@x @y
Review of ANCF beam formulations
Three beam elements based on the absolute nodal Where r0 is the corresponding position vector in the
coordinate formulation(ANCF) are shortly reviewed reference configuration. The Green-Lagrange strain
in this section. The original planar shear-deformable tensor E is defined as
ANCF beam element was proposed by Omar and
Shabana,10 whose elastic force is defined with nonlin- 1
E ¼ ðFT F  IÞ (5)
ear Green-Lagrange strain tensor and the second 2
Piola-Kirchhoff stress tensor based on the continuum
The strain tensor and stress tensor can be written
mechanics. The shear locking problem is the main
drawback of Omar and Shabana’s element. The tech- in the vector form as
nique to alleviate locking phenomenon has been dis-    
 ¼ Exx Eyy 2Exy and r ¼ Sxx Syy Sxy
cussed in detail by Patel and Shabana11 while it is
beyond the scope of the current paper. Alternatively, (6)
Wang and Wang 3

In the case of plane stress state, the elasticity inconsistent interpolation order in the axial and trans-
matrix is degenerated to verse direction.13,14 Shen et al.15 proposed beam ele-
2 3 ments with quadratic interpolation for cross-section
1 v 0 deformation which can be considered as the higher-
E 6v 1 0 7
D¼ 4 1  v5 (7) order beam element (HOBE). In the planar case, the
1  v2 0 0 displacement field for the higher-order beam element
2 with quadratic interpolation in the transverse direc-
where E is Young’s modulus and v is Poisson’s ratio. tion can be defined as
The stress is obtained as " # " #
r1 a0 þ a1 x þ a2 y þ a3 xy þ a4 x2 þ a5 y2 þ a6 xy2 þ a7 x3
r¼ ¼
r ¼ D (8) r2 b0 þ b1 x þ b2 y þ b3 xy þ b4 x2 þ b5 y2 þ b6 xy2 þ b7 x3
(13)
The strain energy for a beam element with rectan-
gular cross section can be written as The nodal coordinate vector of higher-order beam
Z H Z L
is given as
1 2
UCMF ¼ W T Ddxdy (9) " #
2 H2 0 @ri @ri @ 2 ri @rj @rj @ 2 rj
q¼ ri rj (14)
@x @y @y2 @x @y @y2
Where, W and H are the width and height of the cross
section.
Based on the continuum mechanics formulation,
Enhanced continuum mechanics formulation equations (4) to (8) can be used to define the strain
and stress vectors. Accordingly, the strain energy for
(ECMF)
higher-order beam is written as
The beam element constructed based on the energy
Z
form given in equation (9) suffers from the Poisson 1
UHOBE ¼ T Ddv (15)
locking since the Poisson ratio couples the axial strain 2 v
and the transverse normal strains. The Poisson lock-
ing needs to be alleviated to obtain a reasonable
result. Gersmayr et al.12 suggested to divide the
Energy criterion for structural buckling
strain energy into two parts as
To determine the critical buckling load with FEM,
Z H Z L Z L the energy criterion can be employed. Accordingly,
1 2 1
UECMF ¼ W T D0 dxdy þ A T Dv dx when the structure becomes unstable, the second
2 H2 0 2 0
order variation turns to zero. The corresponding gov-
(10) erning equation is

The first part in equation (10) is related to the d2 Uelastic ¼ 0 (16)

modified elasticity matrix D0 which does not take
into account the Poisson effect where Uelastic is the strain energy of the deformed
2 3 structure. Equation (16) is equivalent to
E 0 0
D0 ¼ 4 0 E 0 5 (11) jKt j ¼ 0 (17)
0 0 Gks
In equation (17), jKt j means the determinate of
Where G ¼ 2ð1þvE
Þ is the shear modulus. ks is the cor- tangent stiffness matrix. Once the determinate of tan-
rection factor to shear stiffness. The second part con-
gent stiffness matrix Kt turns to zero, the structure
siders the Poisson effect only at the beam axis in
becomes unstable. For the traditional FEM element
which the elasticity matrix Dv is defined as
such as BEAM188 in ANSYS, the tangent stiffness
2 3 can be written as
v 1 0
Ev 41
Dv ¼ v 05 (12)
1  v2 Kt ¼ K0 þ Kg (18)
0 0 0

where K0 is a constant matrix and Kg is linearly asso-

ciated with the elementte axial force. In this case,
Higher-order beam element formulation (HOBE) equation (17) can be solved as an eigenvalue problem.
The enhanced continuum mechanics formulation still While for the ANCF beam elements, Kt is obtained
does not have optimal convergence due to the from the second derivative with respect to the nodal
4 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

coordinate as Where Km is the mth eigenvalue of Kt . Specifically, the

eigenvalue K instead of the determinate of the stiffness
@2U matrix is adopted in the analysis procedure to determine
Kt ¼ (19)
@q2 the mth critical buckling load. An iterative search with
increasing compressive load begins at Pstart with incre-
which is a nonlinear function of the nodal coordi- ment step Pstep. At the beginning of the iterative, two
nates. To solve equation (17), the dichotomy load condition Pa ¼ Pstart and Pb ¼ Pstart þ pstep are
method can be employed. The scheme of dichotomy considered. The mth eigenvalue which relates to the mth
method is illustrated in Figure 1. It should be noticed critical load are extracted at the two load condition
that numerical difficulties are expected in the direct denote as Km a and Kb . If the eigenvalue is not converged
m

calculation of the determinate of the stiffness matrix to the acceptable tolerance Tol, Pa and Pb are updated
due to its large amplitude. If the Young’s modulus is according to the sign of Km a  Kb . Otherwise, if the tol-
m
th
large enough, the value of the determinate can even erance is admitted, m critical load is saved as Pm cr , the
th
lead to overflow, which means the variable is too large procedure goes to next search for ðm þ 1Þ critical load
to represent as the conventional floating-point values. until m > NSolve or Pb > Pend. The proposed scheme is
While the determinate of the stiffness can be calculat- implemented using MATLAB.
ed by its eigenvalues as
Buckling of a simply supported beam
Y
n
jKt j ¼ K m
(20) The shear deformable beams are used to calculate the
m¼1 critical compressive load of a simply supported beam

Figure 1. Flow chart for buckling solver with absolute coordinate formulation.
Wang and Wang 5

structure. The influence of shear effect on the critical calculated. The length of the beam is L ¼ 10 m. The
load is discussed. The simply supported beam under cross section of the beam is assumed to be square
axial compression Fx is depicted in Figure 2. The with a side length of 0.5 m. The Young’s modulus E ¼
beam is assumed to be with rectangular cross section, 2:05  1011 Pa, the Poisson ratio v ¼ 0.3. The shear
of which the area is A and inertia moment is I. The modulus is calculated as G ¼ E=ð2 þ 2vÞ. The shear
material is assumed to be isotropic and homogeneous correction factor ks ¼ 10ð1 þ vÞ=ð12 þ 11vÞ. Thus the
with the Young’s modulus E, the shear modulus G slenderness of the beam is k ¼ 20:0532p, the ratio
and Poisson ratio v. Additionally, ks is the shear cor- between the shear and axial stiffness of the beam is
rection factor. To discuss the influence of shear effect g ¼ 0:3268. Additionally, the convergence properties
on the critical load, the following dimensionless of ANCF beam elements in the buckling analysis are
parameters are used to describe the beam model. also verified with this example. The first three critical
8 loads are calculated with increasing numbers of ANCF
rffiffiffiffi
>
> A elements. The relative error which is the difference
>
<k ¼ L
I between solutions given by different number of ele-
(21) ments is defined as
>
> ks GA ks
>
: g ¼ EA ¼ 2ð1 þ vÞ
Error ¼ jjSolðnÞ  Solðn  1Þjj (23)
where k is the ratio between the length L and the radius
of inertia of the beam to indicate the slenderness of the Where SolðnÞ is solution with element number 2n and
beam structure, g is the ratio between the shear and n ¼ 2 . . . 8. The relative errors of the first three critical
axial stiffness. According to Euler, in the pure bending loads with respect to the numbers of elements are
case the first Euler buckling load is calculated as shown in Figure 3. The figure reveals that the two
elements, i.e., the CMF and the HOBE, based on
p2 EI
Fe ¼ (22) the continuum mechanics exhibit comparable conver-
L2 gence, while the element based on the enhanced con-
However, once the shear deformation is taken into tinuum mechanics formulation converges faster.
account, the result will be different. The converged solution of ANCF beams is com-
pared with the ANSYS solution given by a mesh of
Accuracy and convergence 210 BEAM188 elements to verify the accuracy of the
ANCF beams. The comparison is listed in Table 1. In
Firstly, to verify the validity of the proposed scheme,
Table 1, Fcr1 ; Fcr2 and Fcr3 are the first three critical
the critical load of simply supported beam structure is
loads of the beam obtained by the proposed method

Table 1. Comparison of convergence solution with ANSYS.

Element type Fcr1 =Fe Fcr2 =Fe Fcr3 =Fe

CMF 1.088164 4.228439 9.080279

ECMF 0.991029 3.860028 8.319556
HOBE 0.990927 3.858505 8.312729
ANSYS BEAM188 0.992571 3.883927 8.433018
Figure 2. Simply supported beam under axial compression.

(a) (b) (c)

102 102 102
HOBE Element HOBE Element HOBE Element
ECMF Element ECMF Element ECMF Element
CMF Element 0 CMF Element CMF Element
0 10 0
10 10
Relative Error
Relative Error
Relative Error

-2 -2
10 10
10-2

-4 -4
10 10
-4
10

101 102 101 102 101 102

Number of Elements Number of Elements Number of Elements

Figure 3. Relative error of the first three critical load. (a) First order. (b) Second order. (c) Third order.
6 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

with ANCF elements, Fe is the first critical load based locking problem. The solutions given by ECMF and
on Euler’s theory which can be calculated as equation HOBE are smaller than Hummer’s because the
(22). For the convenience of comparison, the dimen- ANCF beams relax Timoshenko’s assumption,
sionless parameters of Fcr =Fe is used. For the first which means the cross-section of the beam is deform-
critical load, the accuracy of the HOBE and ECMF able in these two formulations. As the shear stiffness
element reach the level of 102 while the CMF ele- increases, the critical loads approach to Euler’s
ment only yields 101 . For the second and third crit- solution.
ical load, the absolute error becomes large while The dependence of the first three critical loads on
ECMF element performs best. the slenderness k is depicted in Figure 5. The critical
loads approach to those of the classical elastic theory
The influence of shear effect on the critical load as the slenderness increases which indicates that the
influence of the shear deformation is significant for
When the shear deformation is considered, the critical
short and thick beams with small slenderness. The
load can be very different to Euler’s theory. To give
difference between different elements is quite large
thorough discussion on the influence of the shear in the case of smaller slenderness, while it is worth
effect, the shear-deformable ANCF beams are used to note that even Euler’s theory is non-admissible
to calculate the critical load of the simply supported for short and thick beams since few materials can
beam structure by changing the values of g and k. suffer such intensive compression. Again, due to the
Herein, the beam is meshed with 20 ANCF beam shear locking problem, the standard CMF element
elements. Based on the assumption that the cross- gives much higher solutions than the others and
section remains plane but not necessarily perpendic- even fails in the first order critical load case.
ular to the axis, Hummer6 investigated the buckling
problems of shear-deformable beam with different
boundary conditions. Here, Hummer’s results6 are Buckling of right-angle frame
considered as reference solutions. The right-angle frame loaded by constant force P and
The first three critical loads plotted against the its buckling modes are illustrated in Figure 6. The
shear stiffness ratio is shown in Figure 4 for slender- frame is made up by two components of same mate-
ness of pk ¼ 20. For structure that is very flexible in rial with Elastic modulus E ¼ 2:05e11 Pa, Poisson
shear deformation, the critical loads significantly ratio v ¼ 0.3, and shear modulus G ¼ E=ð2 þ 2vÞ.
decrease compared to the classical elastic theory. The length of each component is L ¼ 120 m, and the
For the first critical load, the solutions given by cross-section is set to be square with a side length
ECMF and HOBE element can follow Hummer’s 0.1 m. As shown in Figure 6, one tip of the frame is
solution well while the difference on higher-order crit- fully fixed and the other tip is fixed by pin joint. The
ical loads becomes obvious. The critical loads calcu- ANCF beams are adopted to determine the critical
lated with standard CMF formulation are much loads and each component is meshed with 32 ele-
higher than the others and Hummer’s solution. ments. The method proposed by Shabana and
Even for the first critical load, the CMF element Mikkola16 is used to model the slope discontinuity
fails in following the reference. It is proved again with ANCF formulation. The exact solution by
that the shear locking problem prevents the standard Timoshenko5 and that from ANSYS are adopted as
CMF beams from giving the accurate solution in the the reference. The solutions of the first three critical
buckling analysis. The ECMF and HOBE formula- loads are calculated and dimensionless critical load
tions perform better due to the alleviation of the coefficient m ¼ Pcr L2 =EI is shown in Table 2.

(a) (b) (c)

1.1 4.5 10

1.05 9

4
1 8
Fcr/Fe

Fcr/Fe
Fcr/Fe

0.95 7
3.5
Hummer Hummer Hummer
0.9 ECMF 20 Element ECMF 20 Element 6 ECMF 20 Element
High Order 20 Elements High Order 20 Elements High Order 20 Elements
CMF 20 Elements CMF 20 Elements CMF 20 Elements
0.85 3 5
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2

Figure 4. Critical load against g for constant pk ¼ 20. (a) First order. (b) Second order. (c) Third order.
Wang and Wang 7

(a) (b) (c)

1.1 4.5 10

9
1.05 4
8

Fcr/Fe
Fcr/Fe
Fcr/Fe

1 3.5 7

6
Hummer Hummer Hummer
0.95 3
ECMF 20 Element ECMF 20 Element ECMF 20 Element
5
High Order 20 Elements High Order 20 Elements High Order 20 Elements
CMF 20 Elements CMF 20 Elements CMF 20 Elements
0.9 2.5 4
10 15 20 25 30 10 15 20 25 30 10 15 20 25 30
λ/π λ/π λ/π

Figure 5. Critical load against k

p for constant g ¼ 12. (a) First order. (b) Second order. (c) Third order.

(a) (b) (c) (d)

20 20 20

0 0 0

-20 -20 -20

-40 -40 -40

-60 -60 -60

-80 -80 -80

-100 -100 -100

-120 -120 -120

0 50 100 150 0 50 100 0 50 100 150

Figure 6. Geometry of right-angle frame and its buckling modes. (a) Geometry setup. (b) First buckling mode. (c) Second buckling
mode. (d) Third buckling mode.

Table 2. Dimensionless critical load coefficient m ¼ Pcr L2 =EI. ANCF beams are capable to determine reasonable
critical load. The influence of the shear deformation
Element type First order Second order Third order
on the buckling load is thoroughly investigated by
CMF Lower 11.0771 35.0194 73.5516 comparing the solution with different shear stiffness
ECMF 14.8611 46.3552 96.1228 and slenderness. Generally, the shear deformation
CMF HOBE 14.7958 46.3729 96.8758 makes the critical load smaller than the Euler buckling
ANSYS BEAM188 14.7331 46.6873 99.2111
load. The critical load decreases obviously with small
Analytical 14.72 – –
shear stiffness and slenderness, while as the shear stiff-
ness and slenderness increase, the influence of the shear
The table shows that the results predicted by the effect becomes neglectable and the critical loads
ECMF element and HOBE element agree with the approach to Euler’s solution. According to the perfor-
analytical solution very well while the element based mance of the selected ANCF beam elements, the
on the standard continuum mechanics fails in giving higher order beam element (HOBE) and the beam ele-
similar solution to the reference. ment based on the enhanced continuum mechanics for-
mulation (ECMF) are recommended for the structure
buckling analysis, while the CMF element based on the
Conclusions standard continuum should be used with care. Finally,
The beam elements based on the absolute nodal coor- it is worth to note that the procedure proposed is also
dinate formulation (ANCF) were applied to predict applicable to other nonlinear beam element formula-
the critical load of structures. Since the tangent stiff- tions such as the geometrically exact formulation.
ness matrix of the ANCF element is highly nonlinearly
related to the nodal coordinates, a delicate analysis Declaration of Conflicting Interests
scheme was proposed with the dichotomy method to The author(s) declared no potential conflicts of interest with
conduct the buckling analysis. Two benchmark prob- respect to the research, authorship, and/or publication of
lems were analyzed and the result shows that the this article.
8 Proc IMechE Part C: J Mechanical Engineering Science 0(0)

Funding 8. Simo JC and Vu-Quoc L. On the dynamics of flexible

beams under large overall motions – the plane case:
The author(s) received no financial support for the research,
Part I and II. J Appl Mech 1986; 53: 849–863.
authorship, and/or publication of this article. 9. Sonneville V, Cardona A and Brüls O. Geometrically
exact beam finite element formulated on the special
ORCID iD Euclidean group SE(3). Comput Meth Appl Mech Eng
Tengfei Wang https://orcid.org/0000-0002-4742-5298 2014; 268: 451–474.
10. Omar MA and Shabana AA. A two-dimensional shear
deformable beam for large rotation and deformation
References
problems. J Sound Vib 2001; 243: 565–576.
1. Shabana AA. Definition of the slopes and the finite 11. Patel M and Shabana AA. Locking alleviation in the
element absolute nodal coordinate formulation. large displacement analysis of beam elements: the strain
Multibody Syst Dyn 1997; 1: 339–348. split method. Acta Mech 2018; 229: 2923–2946.
2. Nachbagauer K and Gerstmayr J. Structural and con- 12. Gerstmayr J, Matikainen MK and Mikkola AM. A
tinuum mechanics approaches for a 3d shear deform- geometrically exact beam element based on the absolute
able ANCF beam finite element: application to nodal coordinate formulation. Multibody Syst Dyn
buckling and nonlinear dynamic examples. J Comput 2008; 20: 359–384.
Nonlinear Dyn 2014; 9: 011013. 13. Sopanen JT and Mikkola AM. Description of elastic
3. Luo K, Liu C, Tian Q, et al. An efficient model reduction forces in absolute nodal coordinate formulation.
method for buckling analyses of thin shells based on Nonlinear Dyn 2003; 34: 53–74.
IGA. Comput Meth Appl Mech Eng 2016; 309: 243–268. 14. Mikkola AM, Vallejo DG and Escalona JL. A simple
4. Euler L. Methodus inveniendi lineas curvas maximi procedure to avoid lockings in a shear deformable beam
minimive proprietate gaudentes, sive solutio problematis element based on the absolute nodal coordinate formu-
isoperimetrici lattissimo sensu accepti. Lausanne: Marc- lation. Nonlinear Dyn 2007; 50: 249–264.
Michel Bousquet et Soc, 1744. 15. Shen Z, Li P, Liu C, et al. A finite element beam model
5. Timonshenko S and Gere J. Theory of elastic stability. including cross-section distortion in the absolute nodal
New York: McGraw-Hill international Book Company, coordinate formulation. Nonlinear Dyn 2014; 77:
1961. 1019–1033.
6. Humer A. Exact solutions for the buckling and post- 16. Shabana AA and Mikkola AM. Use of the finite ele-
buckling of shear-deformable beams. Acta Mech 2013; ment absolute nodal coordinate formulation in model-
224: 1493–1525. ing slope discontinuity. J Mech Des 2003; 125: 342–350.
7. Simitses G and Dewey HH. Fundamentals of structural
stability. Oxford: Butterworth-Heinemann, 2006.