Professional Documents
Culture Documents
Mechanics Research Communications: E.J. Sapountzakis, J.A. Dourakopoulos
Mechanics Research Communications: E.J. Sapountzakis, J.A. Dourakopoulos
a r t i c l e
i n f o
Article history:
Received 30 November 2007
Received in revised form 6 June 2008
Available online 18 June 2008
Keywords:
Flexural-torsional buckling
Timoshenko beam
Nonuniform torsion
Warping
Flexural
Bar
Composite beam
Twist
Boundary element method
Shear deformation
a b s t r a c t
In this paper, a boundary element method is developed for the general exural-torsional
linear buckling analysis of Timoshenko beams of arbitrarily shaped composite cross-section. The composite beam consists of materials in contact, each of which can surround a
nite number of inclusions. The materials have different elasticity and shear moduli with
same Poissons ratio and are rmly bonded together. The beam is subjected to a compressive centrally applied load together with arbitrarily axial, transverse and/or torsional distributed loading, while its edges are restrained by the most general linear boundary
conditions. The resulting boundary value problem, described by three coupled ordinary differential equations, is solved employing a boundary integral equation approach. Besides
the effectiveness and accuracy of the developed method, a signicant advantage is that
the method can treat composite beams of both thin and thick walled cross-sections taking
into account the warping along the thickness of the walls, while the displacements as well
as the stress resultants are computed at any cross-section of the beam using the respective
integral representations as mathematical formulae. All basic equations are formulated with
respect to the principal shear axes coordinate system, which does not coincide with the
principal bending one in a nonsymmetric cross-section. To account for shear deformations,
the concept of shear deformation coefcients is used. Six coupled boundary value problems
are formulated with respect to the transverse displacements, to the angle of twist, to the
primary warping function and to two stress functions and solved using the analog equation
method, a BEM based method. Several beams are analysed to illustrate the method and
demonstrate its efciency. The signicant inuence of the boundary conditions and the
shear deformation effect on the buckling load are investigated through examples with
great practical interest.
2008 Elsevier Ltd. All rights reserved.
1. Introduction
Elastic stability of beams is one of the most important criteria in the design of structures subjected to compressive loads.
This beam buckling analysis becomes much more complicated in the case the cross-sections centroid does not coincide with
its shear center (asymmetric beams), leading to the formulation of the exural-torsional buckling problem. Also, composite
structural elements consisting of a relatively weak matrix material reinforced by stronger inclusions or of materials in contact are of increasing technological importance. Steel beams or columns totally encased in concrete, ber-reinforced materials or concrete plates stiffened by steel beams are most common examples. Moreover, unless the beam is very thin the
* Corresponding author. Tel.: +30 2107721718; fax: +30 2107721720.
E-mail addresses: cvsapoun@central.ntua.gr (E.J. Sapountzakis), jdour@mail.ntua.gr (J.A. Dourakopoulos).
0093-6413/$ - see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mechrescom.2008.06.007
498
error incurred from the ignorance of the effect of shear deformation is substantial, and an accurate analysis requires its inclusion in it.
The rst published work on exural-torsional buckling appeared in 1899 by Michell (1899) and Prandtl (1899) for thin,
rectangular, solid beams. Since then, the exural-torsional buckling problem of thin-walled homogeneous beams (based on
the assumptions of the thin tube theory) (Vlasov, 1961; Timoshenko and Gere, 1961; Rao and Carnegie, 1970; Mei, 1970;
Hodges and Peters, 1975; Reissner, 1979; Milisavljevic, 1995; Hodges, 2001) or of symmetrical cross-section beams ignoring
warping (Orloske et al., 2006) has been studied by many researchers noting that Reissner (1979) was the rst who included
transverse shear in his analysis. Moreover, the exural-torsional buckling problem in the case of composite beams of thinwalled or laminated cross-sections has also been examined ignoring (Lee and Kim, 2001; Sapks and Kollr, 2002) or taking
into account (Kollr, 2001; Machado and Cortnez, 2005; Yu et al., 2002; Cortnez and Piovan, 2006) shear deformation effect
employing again the assumptions of the thin tube theory and using either the rened models or shell stress resultants. To
the authors knowledge publications on the solution to the general exural-torsional buckling analysis of Timoshenko beams
of arbitrarily shaped composite cross-section do not exist.
In this investigation, an integral equation technique is developed for the solution of the aforementioned problem. The composite beam consists of materials in contact, each of which can surround a nite number of inclusions. The materials have
different elasticity and shear moduli with same Poissons ratio and are rmly bonded together. The beam is subjected to a
compressive centrally applied load together with arbitrarily axial, transverse and torsional distributed loading, while its edges
are restrained by the most general linear boundary conditions. The resulting boundary value problem, described by three coupled ordinary differential equations, is solved employing the concept of the analog equation (Katsikadelis, 2002). According to
this method, the three coupled fourth order hyperbolic partial differential equations are replaced by three uncoupled ones
subjected to ctitious load distributions under the same boundary conditions. All basic equations are formulated with respect
to the principal shear axes coordinate system, which does not coincide with the principal bending one in a nonsymmetric
cross-section. To account for shear deformations, the concept of shear deformation coefcients is used. Six boundary value
problems are formulated with respect to the transverse displacements, to the angle of twist, to the primary warping function
and to two stress functions and solved using the analog equation method (Katsikadelis, 2002), a BEM based method. The
essential features and novel aspects of the present formulation compared with previous ones are summarized as follows:
(i) The proposed method can be applied to beams having an arbitrary composite constant cross-section and not to a necessarily thin-walled one.
(ii) All basic equations are formulated with respect to the principal shear axes coordinate system, which does not necessarily coincide with the principal bending one.
(iii) Shear deformation effect is taken into account on the exural-torsional buckling analysis of beams of nonsymmetric
constant composite cross-section avoiding the restrictions of the thin-walled theory.
(iv) Torsional warping arising from nonuniform torsion is taken into account.
(v) The beam is supported by the most general linear boundary conditions including elastic support or restraint.
(vi) The proposed method overcomes the shortcoming of the induced error in the case of the utilization of a thin tube theory solution.
(vii) The shear deformation coefcients are evaluated using an energy approach, instead of Timoshenko and Goodiers
(1984) and Cowpers (1966) denitions, for which several authors (Schramm et al., 1994; Schramm et al., 1997) have
pointed out that one obtains unsatisfactory results or denitions given by other researchers (Stephen, 1980; Hutchinson, 2001), for which these factors take negative values.
(viii) With the exception of the structural models presented by Machado and Cortnez (2005), Cortnez and Piovan (2006)
which make use of shell stress resultants, previous formulations concerning composite beams of thin-walled crosssections or laminated cross-sections are analyzing these beams using the rened models. However, these models
do not satisfy the continuity conditions of transverse shear stress at layer interfaces and assume that the transverse
shear stress along the thickness coordinate remains constant, leading to the fact that kinematic or static assumptions
cannot be always valid.
(ix) The proposed method employs a pure BEM approach (requiring only boundary discretization) resulting in line or parabolic elements instead of area elements of the FEM solutions (requiring the whole cross-section to be discretized into
triangular or quadrilateral area elements), while a small number of line elements are required to achieve high
accuracy.
Several beams are analysed to illustrate the method and demonstrate its efciency. The signicant inuence of the
boundary conditions and the shear deformation effect on the buckling load are investigated through examples with great
practical interest.
499
Fig. 1. Prismatic element of an arbitrarily shaped composite cross section occupying region X (a) subjected in bending and torsional loading (b).
modulus Gj , occupying the regions Xj j 1; 2; . . . ; K of the y; z plane (Fig. 1). The materials of these regions are assumed
homogeneous, isotropic and linearly elastic. Let also the boundaries of the nonintersecting regions Xj be denoted by
Cj j 1; 2; . . . ; K. These boundary curves are piecewise smooth, i.e. they may have a nite number of corners. In Fig. 1a
CYZ is the coordinate system through the cross-sections centroid C, while yC , zC are its coordinates with respect to Syz principal shear system of axes through the cross-sections shear center S. The beam is subjected to a compressive load P, to the
combined action of the arbitrarily distributed axial loading pX pX X, transverse loading pY pY X, pZ pZ X acting in the
Y and Z directions, respectively, and to the arbitrarily distributed twisting moment mx mx x (Fig. 1b).
Under the aforementioned loading the displacement eld of the beam with respect to the Syz system of axes is given as
x; y; z ux hY xZ hZ xY
u
dhx x P
/S y; z
dx
x; y; z vx zhx x
v
y; z wx yhx x
wx;
1a
1b
1c
and therefore the displacement components of the cross-sections centroid can be written as
dhx x P
/S y; z
dx
vC vx zC hx x
2b
wC wx yC hx x
2c
uC ux
2a
where ux, vx and wx are the beam axial and transverse displacements of the shear center S with respect to x, y and z
axes, respectively, hY , hZ are the angles of rotation of the cross-section due to bending, dhx =dx denotes the rate of change
of the angle of twist hx regarded as the torsional curvature and /PS is the primary warping function with respect to the shear
center S of the cross-section of the beam (Sapountzakis and Mokos, 2003).
500
Moreover, according to the linear theory of beams (small deections), the angles of rotation of the deection line with
respect to the shear center x and to the centroid b in the xz and xy planes of the beam subjected to the aforementioned
loading and taking into account shear deformation effect satisfy the following relations
sin xy xy
cos by 1;
sin by
dw
hY cxz ;
dx
sin xz xz
dv
hZ cxy
dx
cos bz 1
dwC
;
dx
sin bz
3a; b
3c; d
dvC
dx
3e; f
while employing the stressstrain relations of the three-dimensional elasticity after ignoring the shear strain due to warping,
the arising shear stress resultants Q z , Q y are given as
dw
hY
dx
Xj
j1
K Z
X
dv
hZ
sxy dXj G1 Ay
Qy
dx
Xj
j1
Qz
K Z
X
sxz dXj G1 Az
4a
4b
where the rst material is considered as reference material, cxz , cxy are the additional angles of rotation of the cross-section
due to shear deformation (Fig. 2a) and G1 Az , G1 Ay are the cross-sections shear rigidities of the Timoshenkos beam theory, with
Az jz AG
K
1 G
1 X
G j Aj
A
az
az j1 G1
5a
Ay jy AG
K
1 G
1 X
G j Aj
A
ay
ay j1 G1
5b
the shear areas with respect to z, y axes, respectively, jz , jy are the shear correction factors and az , ay the shear deformation
coefcients. It is worth here noting that the reduction of Eq. (4) using the shear modulus G1 of the rst material, could be
achieved using any other material, considering it as reference material.
Referring to Fig. 2, the stress resultants Rx , Ry , Rz acting in the x, y, z directions, respectively, are related to the axial N and
the shear Q y , Q z forces as
6a
Ry N sin bz Q y cos bz
6b
Rz N sin by Q z cos by
6c
which by virtue of the small deection theory and Eqs. (2) and (3) become
dwC
dvC
Qy
dx
dx
dvC
dv
dhx
zC
Qy N
Qy
Ry N
dx
dx
dx
dwC
dw
dhx
Qz
yC
Qz N
Rz N
dx
dx
dx
Rx N Q z
7a
7b
7c
The second and third terms in the right hand side of Eq. (7a), express the inuence of the shear forces Q y , Q z on the horizontal
stress resultant Rx . However, since initial shear forces are not taken into account, these terms can be neglected since Q y , Q z
are much smaller than N (Rothert and Gensichen, 1987; Ramm and Hofmann, 1995) and thus Eq. (7a) can be written as
Rx N
Employing Eq. (1a) to the straindisplacement equations of the three-dimensional elasticity and ignoring axial deformations,
the normal strain component ex can be written as
2
ex
dhY
dhZ
d hx
Z
Y 2 /PS
dx
dx
dx
MY
K Z
X
j1
MZ
Ei ex Z dXj E1 IYY
Xj
K Z
X
j1
Xj
dhY
dhZ
E1 IYZ
dx
dx
Ei ex Y dXj E1 IZZ
dhZ
dhY
E1 IYZ
dx
dx
10a
10b
501
Fig. 2. Displacements (a) and equilibrium of an element in the xz (b) and xy (c) planes.
where
IYY
Z
K
X
Ej
Z 2 dXj ;
E1 Xj
j1
IZZ
Z
K
X
Ej
Y 2 dXj ;
E1 Xj
j1
IYZ
Z
K
X
Ej
YZdXj
E 1 Xj
j1
11a; b; c
are the moments and the product of inertia of the cross-section with respect to its centroid C. Substituting Eqs. (4a) and (4b)
in Eqs. (10a) and (10b) the bending moments M Y , M Z can be written as
MY E1 IYY
MZ E1 IZZ
!
!
2
2
1 dQ z d w
d v
1 dQ y
2 E1 IYZ
G1 Az dx
dx
dx2 G1 Ay dx
!
!
2
2
d v
1 dQ y
1 dQ z d w
E1 IYZ
2
dx2 G1 Ay dx
G1 Az dx
dx
12a
12b
The governing equations of the problem at hand will be derived by considering the equilibrium of the deformed element.
Thus, referring to Fig. 2 we obtain
502
dRx
pX 0;
dx
dRy
pY 0;
dx
dMY
Q z 0;
dx
dRz
pZ 0
dx
13a; b; c13ac
dM Z
Qy 0
dx
13d; e
Substituting Eqs. (8), (7b), and (7c) into Eqs. (13ac) we obtain
dN
pX
dx
14a
dQ y dN ov
ohx
zC
N
dx ox
dx
ox
dQ z dN ow
ohx
N
yC
dx ox
dx
ox
o2 v
o2 hx
zC 2
2
ox
ox
o2 w
o2 hx
yC 2
ox2
ox
pY 0
14b
!
pZ 0
14c
Substituting Eqs. (14b) and (14c) into Eqs. (12a) and (12b) we obtain the expressions of the bending moments M Y , M Z as
!!
2
2
dw
dhx
d w
d hx
pZ pX
yC
N
yC
dx
dx2
dx
dx2
!!
2
2
2
d v E1 IYZ
dv
dhx
d v
d hx
E1 IYZ 2
zC
pY pX
zC
N
dx
dx
dx2
G1 Ay
dx
dx2
!!
2
2
2
d v E1 IZZ
dv
dhx
d v
d hx
zC
pY pX
z
MZ E1 IZZ 2
N
C
dx
dx
dx2
G 1 Ay
dx
dx2
!!
2
2
2
d w E1 IYZ
dw
dhx
d w
d hx
N
E1 IYZ 2
yC
pZ pX
yC
dx
dx
dx2
G1 Az
dx
dx2
2
d w E1 IYY
MY E1 IYY 2
G1 Az
dx
15a
15b
d v
d w
E1 IYZ 3
dx3
dx
!
!!
2
2
3
3
E1 IZZ dpY dpX dv
dhx
d v
d hx
d v
d hx
zC
2pX
zC
zC
N
dx2
dx3
G1 Ay dx
dx dx
dx
dx2
dx3
!
!!
2
2
3
3
E1 IYZ dpZ dpX dw
dhx
d w
d hx
d w
d hx
N
yC
yC
yC
2pX
dx2
dx3
G1 Az dx
dx dx
dx
dx2
dx3
Q y E1 IZZ
16a
d w
d v
E1 IYZ 3
dx3
dx
!
!!
2
2
3
3
E1 IYY dpZ dpX dw
dhx
d w
d hx
d w
d hx
N
yC
y
2pX
C
C
dx2
dx3
G1 Az dx
dx dx
dx
dx2
dx3
!
!!
2
2
3
3
E1 IYZ dpY dpX dv
dhx
d v
d hx
d v
d hx
2pX
zC
zC
zC
N
dx2
dx3
G1 Ay dx
dx dx
dx
dx2
dx3
Q z E1 IYY
16b
and eliminating these forces from Eqs. (14b) and (14c) we obtain the rst two coupled partial differential equations of the
problem of the beam under consideration subjected to the combined action of axial, bending and torsional loading as
" 2
!
!
4
4
2
2
2
3
3
d v
d w E1 IZZ d pY d pX dv
dhx
dpX d v
d hx
d v
d hx
3p
z
E
I
z
z
3
1
YZ
C
C
C
X
dx4
dx4 G1 Ay dx2
dx3
dx2 dx
dx
dx dx2
dx2
dx3
!#
"
!
!
4
4
2
2
2
2
3
3
d v
d hx
E1 IYZ d pZ d pX dw
dhx
dpX d w
d hx
d w
d hx
3pX
zC
yC
yC
yC
N
3
dx4
dx3
dx4
G1 Az dx2
dx2 dx
dx
dx dx2
dx2
dx3
!#
!
4
4
2
2
d w
d hx
dv
dhx
d v
d hx
N
p
0
N
z
p
z
C
C
C
Y
X
dx4
dx
dx2
dx4
dx
dx2
E1 IZZ
17a
"
!
!
4
4
2
2
2
2
3
3
d w
d v E1 IYY d pZ d pX dw
dhx
dpX d w
d hx
d w
d hx
3p
E
I
y
3
1 YZ
C
C
X
C
dx4
dx4 G1 Az dx2
dx3
dx2 dx
dx
dx dx2
dx2
dx3
!#
"
!
!
4
4
2
2
2
2
3
3
d w
d hx
E1 IYZ d pY d pX dv
dhx
dpX d v
d hx
d v
d hx
yC
zC
zC
zC
N
3
3pX
dx4
dx3
dx4
G1 Ay dx2
dx2 dx
dx
dx dx2
dx2
dx3
!#
!
4
4
2
2
d v
d hx
dw
dhx
d w
d hx
p
0
y
z
y
N
N
C
Z
X
C
C
dx4
dx
dx2
dx4
dx
dx2
503
E1 IYY
17b
Finally, the angles of rotation of the cross-section due to bending hY , hZ are given from Eqs. (4a) and (4b) as
3
dw E1 IYY d w E1 IYZ d v
dx G1 Az dx3 G1 Az dx3
"
!
2
2
E1 IYY dpZ dpX dw
dhx
d w
d hx
2pX
yC
2 2
y
N
C
dx2
dx dx
dx
dx2
G1 Az dx
"
!
2
2
E1 IYZ dpY dpX dv
dhx
d v
d hx
N
z
zC
2
2pX
C
dx2
dx dx
dx
dx2
G1 Ay Az dx
hY
!#
3
3
d w
d hx
y
C
dx3
dx3
!#
3
3
d v
d hx
z
C
dx3
dx3
dv E1 IZZ d v E1 IYZ d w
dx G1 Ay dx3 G1 Ay dx3
"
!
!#
2
2
3
3
E1 IZZ dpY dpX dv
dhx
d v
d hx
d v
d hx
N
2 2
zC
zC
zC
2pX
dx2
dx3
dx dx
dx
dx2
dx3
G1 Ay dx
"
!
!#
2
2
3
3
E1 IYZ dpZ dpX dw
dhx
d w
d hx
d w
d hx
yC
y
2pX
N
2
C
C
dx2
dx3
dx dx
dx
dx2
dx3
G1 Ay Az dx
hZ
18a
18b
Equilibrium of torsional moments along x axis of the beam element, after taking into account the additional shear stresses
due to the presence of the axial force N (Timoshenko and Gere, 1961), which employing Eq. (2) are written as
dw
dw
dhx
rj
y
dx
dx
dx
dhx
j dv
j dv
r
z
r
dx
dx
dx
sjxz rj
19a
sjxy
19b
Mt;add
dw
dv
IS dhx
NzC
N E
dx
dx
A dx
20
where
IS
K
X
21
j1
E
Pk
j1 Ej Aj
22
E1
leads to the third (coupled with the previous two) partial differential equation of the problem of the beam under consideration as
dM t dMt;add
mx pZ yc pY zc
dx
dx
23
which employing Eq. (20) and having in mind that the torsional moment M t is given as (Sapountzakis and Mokos, 2003)
3
Mt E1 C S
d hx
dhx
G1 It
dx3
dx
24
can be written as
4
d hx
d hx
d w
d v I S d hx
E1 C S 4 G1 It
N y c 2 zc 2 E
dx
dx
dx
dx2
A dx2
dw
dv
IS dhx
pX E
zc
mx pZ yC pY zC pX yc
dx
dx
A dx
25
504
where IS is the polar moment of inertia with respect to the shear center S, E1 C S and G1 It are the cross-sections warping and
torsional rigidities, respectively, with C S , It being its warping and torsion constants, respectively, given as (Sapountzakis and
Mokos, 2003)
Z
K
X
Ej
uP 2 dXj
E1 Xj S
j1
Z
K
X
Gj
ouP
ouP
y2 z2 y S z S dXj
It
G1 Xj
oz
oy
j1
CS
26a
26b
It is worth here noting that the primary warping function uPS y; z can be established by solving independently the Neumann
problem (Sapountzakis and Mokos, 2003)
r2 uPS 0 in X [Kj1 Xj
Gi
ouPS
on
i
Gj
ouPS
on
j
27
2
oq
12 Gi Gj osS
on Cj j 1; 2; . . . ; K
28
where r2 o2 =ox2 o2 =oy2 is the Laplace operator; Gi 0 at the free part of the boundary
of Xj region or Gi is the shear
p
modulus of Xi region at the common part of the boundaries of Xi and Xj regions; qS x2 y2 is the distance of a point
on the boundary Cj from the shear center S; o=onj denotes the directional derivative normal to the boundary Cj and
o=osj denotes differentiation with respect to its arc length s.
As it is already mentioned, Eqs. (17a), (17b), and (25) constitute the governing equations of the beam subjected to the
combined action of axial, bending and torsional loading taking into account shear deformation effect. For the special case
of an axially compressive load N P pX 0. The aforementioned equations are also subjected to the pertinent boundary
conditions of the problem, which are given as
a1 vx a2 Ry x a3 ; a 1 hZ x a 2 MZ x a 3
29a; b
1 hY x b
2 MY x b
3
b
30a; b
b1 wx b2 Rz x b3 ;
c1 hx x c2 Mt x c3 ; c1
dhx x
2 Mw x c
3
c
dx
31a; b
at the beam ends x 0; l, where Ry , Rz and M Z , M Y are the reactions and bending moments with respect to y and z axes,
respectively, obtained from Eqs. (7b), (7c), and (15a), (15b), (16a), (16b) as
!
3
3
3
3
d v E1 IZZ
d v
d hx
d w
Ry E1 IZZ 3
N
zC
E1 IYZ 3
dx
dx3
dx
G1 Ay
dx3
!
3
3
E1 IYZ
d w
d hx
dv
dhx
N
N
y
z
C
C
dx3
dx
G1 Az
dx3
dx
!
3
3
3
3
d w E1 IYY
d w
d hx
d v
E1 IYZ 3
Rz E1 IYY
N
yC
3
3
3
dx
dx
dx
G1 Az
dx
!
3
3
E1 IYZ
d v
d hx
dw
dhx
N
yC
N
zC
dx3
dx
G 1 Ay
dx3
dx
!
!
2
2
2
2
2
2
d w E1 IYY
d w
d hx
d v E1 IYZ
d v
d hx
E1 IYZ 2
MY E1 IYY 2
N
yC
N
zC
dx
dx2
dx
dx2
G1 Az
dx2
G1 Ay
dx2
!
!
2
2
2
2
2
2
d v E1 IZZ
d v
d hx
d w E1 IYZ
d w
d hx
MZ E1 IZZ 2
E1 IYZ 2
N
zC
N
yC
dx
dx2
dx
dx2
G 1 Ay
dx2
G1 Az
dx2
32a
32b
33a
33b
the angles of rotation due to bending hY , hZ are evaluated from Eq. (18) as
3
dw E1 IYY d w EIYY
d w
d hx
hY
N
yC
dx G1 Az dx3 G21 A2z
dx3
dx3
!
3
3
E1 IYZ
d v
d hx
2
N
zC
3
dx
dx3
G 1 Ay Az
E1 IYZ d v
G1 Az dx3
34
hZ
dv E1 IZZ d v E1 IZZ
d v
d hx
N
zC
dx G1 Ay dx3 G21 A2y
dx3
dx3
!
3
3
E1 IYZ
d w
d hx
2
N
y
C
dx3
dx3
G1 Ay Az
505
E1 IYZ d w
G1 Ay dx3
35
while in Eq. (31b) M w is the warping moment given as (Sapountzakis and Mokos, 2003)
2
Mw E1 C S
d hx
dx2
36
k ; c ; c
k ; bk ; b
k k 1; 2; 3 are functions specied at the beam ends x 0; l. Eqs. (29)(31) describe the most genFinally, ak ; a
k
eral linear boundary conditions associated with the problem at hand and can include elastic support or restraint. It is apparent that all types of the conventional boundary conditions (clamped, simply supported, free or guided edge) can be derived
from these equations by specifying appropriately these functions (e.g. for a clamped edge it is a1 b1 c1 1,
a 1 b1 c1 1, a2 a3 b2 b3 c2 c3 a 2 a 3 b2 b3 c2 c3 0).
The solution of the boundary value problem given from Eqs. (17), (25) subjected to the boundary conditions (29)(31)
which represents the exural-torsional buckling of beams, presumes the evaluation of the shear deformation coefcients
ay , az , corresponding to the principal shear axes coordinate system Syz. These coefcients are established equating the
approximate formula of the shear strain energy per unit length (Schramm et al., 1997)
U appr:
ay Q 2y
G
2A G1
az Q 2z
37
2AG G1
U exact
Z
K
X
sxz 2j sxy 2j
E1
dXj
Ej Xj
2G1
j1
38
ay
39a
39b
where sxz j ; sxy j are the transverse (direct) shear stress components, r iY o=oY iZ o=oZ is a symbolic vector with
iY ; iZ the unit vectors along Y and Z axes, respectively, D is given from
40
m is the Poisson ratio of the cross-section materials, e and d are vectors dened as
"
e
Y 2 Z2
IYY
IYZ YZ
2
"
d
m IZZ YZ IYZ
Y 2 Z2
2
!#
iY
!#
"
Y 2 Z2
IYY YZ IYZ
2
"
iY m IZZ
!#
Y 2 Z2
IYZ YZ
2
iZ
41a
!#
iZ
41b
and HY; Z, UY; Z are stress functions, which are evaluated from the solution of the following Neumann type boundary
value problems (Mokos and Sapountzakis, 2005):
r2 Hj 2IYZ Z IYY Y in Xj j 1; 2; . . . ; K
Ej
oH
on j
Ei
oH
on i
Ej Ei n e on Cj j 1; 2; . . . ; K
r2 Uj 2IYZ Y IZZ Z in Xj j 1; 2; . . . ; K
Ej
oU
on j
Ei
oU
on i
Ej Ei n d on Cj j 1; 2; . . . ; K
42a
42b
43a
43b
where Ei is the modulus of elasticity of the Xi region at the common part of the boundaries of Xj and Xi regions, or Ei 0 at
the free part of the boundary of Xj region, while o=onj nY o=oYj nZ o=oZj denotes the directional derivative normal to
the boundary Cj . The vector n normal to the boundary Cj is positive if it points to the exterior of the Xj region, while the
normal derivatives across the interior boundaries vary discontinuously. It is also worth here noting that the boundary con-
506
ditions (28), (42b), (43b) have been derived from the physical consideration that the traction vectors in the direction of the
normal vector n on the interfaces separating the j and i different materials are equal in magnitude and opposite in direction,
while it vanishes on the free surface of the beam.
3. Integral representations numerical solution
3.1. For the transverse v, w displacements and the angle of twist hx
According to the precedent analysis, the exural-torsional buckling problem of composite beams reduces in establishing
the displacement components vx, wx and hx x having continuous derivatives up to the fourth order with respect to x,
satisfying the coupled governing equations (17) and (25) inside the beam and the boundary conditions (29)(31) at the beam
ends x 0; l.
Eqs. (17) and (25) are solved using the analog equation method (Katsikadelis, 2002) as it is developed for hyperbolic differential equations (Sapountzakis, 2005; Sapountzakis and Tsiatas, 2007). This method is applied for the problem at hand as
follows. Let vx, wx and hx x be the sought solution of the aforementioned boundary value problem. Setting as
u1 x vx, u2 x wx, u3 x hx x and differentiating these functions four times with respect to x yields
4
d ui
qi x i 1; 2; 3
dx4
44
Eq. (44) indicate that the solution of Eqs. (17) and (25) can be established by solving Eq. (44) under the same boundary conditions (29)(31), provided that the ctitious load distributions qi x i 1; 2; 3 are rst established. These distributions can
be determined using BEM as follows:
The solution of Eq. (44) is given in integral form as
ui x
"
d ui du d ui d u dui d u
qi u dx u
ui
dx3
dx dx2
dx2 dx
dx3
#l
45
0
u
r 3
r 2
1 3
l 2 3
12
l
l
46
with r x n, x, n points of the beam, which is a particular singular solution of the equation
4
d u
dx n
dx4
47
ui x
"
qi K4 rdn K4 r
d ui
d ui
dui
K3 r 2 K2 r
K1 rui
dx3
dx
dx
#l
48
0
1
r
K1 r sgn
2
l
r
1
K2 r l 1
2
l
1 2 r r
r
K3 r l 2 sgn
4 l l
l
r 3
r 2
1 3
l 2 3
K4 r
12
l
l
49a
49b
49c
49d
Notice that in Eq. (48) for the line integral it is r x n, x, n points inside the beam, whereas for the rest terms it is r x f, x
inside the beam, f at the beam ends 0, l.
Differentiating Eq. (48) with respect to x, results in the integral representations of the derivatives of ui as
dui x
dx
"
d ui
d ui
dui
qi K3 rdn K3 r 3 K2 r 2 K1 r
dx
dx
dx
0
"
#
l
Z
2
3
2
l
d ui x
d ui
d ui
qi K2 rdn K2 r 3 K1 r 2
2
dx
dx
dx
0
0
#l
50a
0
50b
d ui x
dx3
Z
0
"
d ui
qi K1 rdn K1 r 3
dx
507
#l
50c
0
d ui x
qi x
dx4
50d
The integral representations (48) and (50), when applied for the beam ends 0; l, together with the boundary conditions (29)(31) are employed to express the unknown boundary quantities ui f, ui;x f, ui;xx f and ui;xxx f f 0; l in
terms of qi . This is accomplished numerically as follows. The interval 0; l is divided into L equal elements (Fig. 3), on
which qi x is assumed to vary according to certain law (constant, linear, parabolic, etc.). The constant element assumption
is employed here as the numerical implementation becomes very simple and the obtained results are very good. Employing the aforementioned procedure for the coupled boundary conditions (29) and (30) the following set of linear equations
is obtained:
D11
6 0
6
6
6 E31
6
6 0
6
6
6 0
6
6 0
6
6
4 0
0
D14
D22
D23
D24
D27
E32
E33
E34
E42
E43
E44
D54
D55
D63
D64
D66
D67
0
0
0
0
0
0
E31
0
E32
E42
E33
E43
38 ^ 9 8 9 2 3
2 3
0
D18 > u
0
a3 >
1 >
>
>
>
>
>
>
>
>
>
7
7
6
607
>
>
>
>
^
>
>
>
D28 7>
a
0
u
3
1;x
6 7
6 7
> >
>
>
>
>
7>
6 7
>
>
> 6 7
> >
>
^ 1;xx >
6 F3 7
607
>
>
>
>
0 7>
0>
>
>u
>
>
>
>
>
7
7>
6
6 7
>
>
>
>
<u
= < 0 = 6F 7
607
^
0 7
1;xxx
7
6 47
6 7
6 7q1 6 7q2
7
^
>
>
>
>
7
7
6
607
b
D58 7>
0
u
2
>
>
>
3
> >
>
> 6 7
> >
6 7
>
>
>
>
>
>
7
7
6
607
>
>
>
>
^
07
D68 7>
6 7
> b3 >
> u2;x >
> 6
> >
>
>
>
7
7>
6
6 7
>
>
>
>
>
>
>
>
405
4 F3 5
^
E34 5>
0
u
>
>
>
2;xx
>
>
: >
:
;
; >
^ 2;xxx
0
E44
F4
0
u
51
E11
6 0
6
6
4 E31
0
E12
E22
0
E23
E32
E33
E42
E43
9 8 9 2 3
38
^ 3 > > c3 >
E14 > u
0
>
>
> >
>
< c >
< ^ >
= 607
= >
0 7
3
3;x
7 u
6 7
6 7q3
7
^ 3;xx >
>
>u
4 F3 5
E34 5>
0>
>
>
>
>
>
>
>
:
; : ;
^ 3;xxx
0
E44
F4
u
52
where D11 , D14 , D18 , D22 , D23 , D24 , D27 , D28 , D54 , D55 , D58 , D63 , D64 , D66 , D67 , D68 , E22 , E23 , E1j , j 1; 2; 4 are 2 2 known square
j ; c ; c
3 , c3 , c3 are 2 1 known
j ; bj ; b
j j 1; 2 of Eqs. (29)(31); a3 ,
matrices including the values of the functions aj ; a
a3 , b3 , b
j
column matrices including the boundary values of the functions a3 ; a3 ; b3 ; b3 ; c3 ; c3 of Eqs. (29)(31); Ejk ,
j 3; 4; k 1; 2; 3; 4 are square 2 2 known coefcient matrices resulting from the values of the kernels Kj r
j 1; 2; 3; 4 at the beam ends and Fj j 3; 4 are 2 L rectangular known matrices originating from the integration of
the kernels on the axis of the beam. Moreover
^ i fui 0; ui lgT
u
T
dui 0
dui l
^ i;x
;
u
dx
dx
(
)T
2
2
d ui 0
d ui l
^ i;xx
u
;
dx2
dx2
(
)T
3
3
d ui 0
d ui l
^ i;xxx
;
u
dx3
dx3
53a
53b
53c
53d
are vectors including the two unknown boundary values of the respective boundary quantities and qi fqi1 qi2 . . . qiL gT
i 1; 2; 3 is the vector including the L unknown nodal values of the ctitious load.
Nodal points
1
2
l
Fig. 3. Discretization of the beam interval and distribution of the nodal points.
508
Discretization of Eqs. (48) and (50) and application to the L collocation points yields
^ i H2 u
^ i;x H3 u
^ i;xx H4 u
^ i;xxx
ui C4 qi H1 u
^
^
^
ui;x C3 qi H1 ui;x H2 ui;xx H3 ui;xxx
54a
54b
^ i;xx H2 u
^ i;xxx
ui;xx C2 qi H1 u
54c
^ i;xxxx
ui;xxxx C1 qi H1 u
54d
ui;xxxx qi
54e
where Cj j 1; 2; 3; 4 are L L known matrices; Hj j 1; 2; 3; 4 are L 2 also known matrices and ui , ui;x , ui;xx , ui;xxxx , ui;xxxx
are vectors including the values of ui x and their derivatives at the L nodal points.
The above equations, after eliminating the boundary quantities employing Eqs. (51) and (52), can be written as
ui Ti qi Tij qj ti ;
i; j 1; 2;
i 6 j
55a
u3 T3 q3 t3
55b
i; j 1; 2;
i 6 j
55c
55d
i; j 1; 2;
i 6 j
55e
55f
i; j 1; 2;
i 6 j
55g
55h
i 1; 2; 3
55i
where Ti , Tix , Tixx , Tixxx , Tij , Tijx , Tijxx , Tijxxx are known L L matrices and ti , tix , tixx , tixxx are known L 1 matrices. It is worth here
3 c c
3 b3 b
3 0 it is ti tix tixx tixxx 0.
noting that for homogeneous boundary conditions a3 a
3
In the conventional BEM, the load vectors qi are known and Eq. (55) are used to evaluate ui x and their derivatives at the
L nodal points. This, however, can not be done here since qi are unknown. For this purpose, 3L additional equations are derived, which permit the establishment of qi . These equations result by applying Eqs. (17) and (25) to the L collocation points,
leading to the formulation of the following set of 3L simultaneous equations
8 9
>
< q1 >
=
A NB C q2 f
>
: >
;
q3
56
E1 IZZ
6
A 4 E1 IYZ
2
E1 IYZ
E1 IYY
E1 CS G1 It T3xx
6
1
B6
4 G1 Ay E1 IYZ T21xx
yC T21xx zC T1xx
3
C11 C12 C13
6
7
C 4 C21 C22 C23 5
C31 C32 C33
2
3
7
5
57a
zC
G1 Ay
zC
G1 Ay
7
E1 IYZ Gy1CAz E1 IYY yC T3xx 7
5
yC T2xx zC T12xx
IS
AE
57b
T3xx
57c
E1 IZZ
C11 pX T1x
p T1x 3pX;x T1xx 3pX T1xxx
G1 Ay X;xx
E1 IYZ
pX;xx T21x 3pX;x T21xx 3pX T21xxx
G1 Az
E1 IZZ
p T12x 3pX;x T12xx 3pX T12xxx
C12 pX T12x
G1 Ay X;xx
E1 IYZ
pX;xx T2x 3pX;x T2xx 3pX T2xxx
G1 Az
58a
58b
E1 IZZ
C13 zC pX T3x
zC pX;xx T3x 3zC pX;x T3xx 3zC pX T3xxx
G1 Ay
E1 IYZ
yC pX;xx T3x 3yC pX;x T3xx 3yC pX T3xxx
G1 Az
E1 IYY
C21 pX T21x
p T21x 3pX;x T21xx 3pX T21xxx
G1 Az X;xx
E1 IYZ
pX;xx T1x 3pX;x T1xx 3pX T1xxx
G 1 Ay
E1 IYY
C22 pX T2x
p T2x 3pX;x T2xx 3pX T2xxx
G1 Az X;xx
E1 IYZ
pX;xx T12x 3pX;x T12xx 3pX T12xxx
G 1 Ay
E1 IYY
C23 yC pX T3x
y p T3x 3yC pX;x T3xx 3yC pX T3xxx
G1 Az C X;xx
E1 IYZ
zC pX;xx T3x 3zC pX;x T3xx 3zC pX T3xxx
G 1 Ay
509
58c
58d
58e
58f
58g
58h
58i
8 9 8
9
8
9
t1xx zC t3xx
>
>
>
< f1 >
= >
< 0 >
=
<
=
t2xx yC t3xx
f f2
0
N
>
>
>
>
>
: ; :
;
: y t z t IS t >
;
G1 It t3xx
f3
C 1xx
C 2xx
AE 3xx
59
with
E1 IZZ
p
pX;xx t1x zC t3x
G1 Ay Y;xx
E1 IYZ
p
3pX;x t1xx zC t3xx 3pX t1xxx zC t3xxx
G1 Az Z;xx
pX;xx t2x yC t3x 3pX;x t2xx yC t3xx 3pX t2xxx yC t3xxx
E1 IYY
p pX;xx t2x yC t3x
f 2 pZ pX t2x yC t3x
G1 Az Z;xx
E1 IYZ
p
3pX;x t2xx yC t3xx 3pX t2xxx yC t3xxx
G1 Ay Y;xx
f 1 pY pX t1x zC t3x
60a
60b
60c
In the above set of equations the matrices E1 IYY , E1 IZZ , E1 IYZ , E1 CS , G1 It are L L diagonal matrices including the values of
the corresponding quantities, respectively, at the L nodal points. Moreover, pX , pX;x , pX;xx are diagonal matrices and pY , pY;xx ,
pZ , pZ;xx and mx are vectors containing the values of the external loading and their derivatives at these points.
Solving the linear system of Eq. (56) for the ctitious load distributions q1 , q2 , q3 the displacements and their derivatives
in the interior of the beam are computed using Eq. (55).
3.1.1. Buckling equation
3 c c
3 b3 b
3 0 (homogeneous boundary conditions) and pX pX;x pX;xx
In this case it is a3 a
3
pY pY;xx pZ pZ;xx mx 0, N P. Thus, Eq. (56) becomes
8 9
>
< q1 >
=
A PB q2 0
>
: >
;
q3
61
The condition that Eq. (61) has a nontrivial solution yields the buckling equation
detA PB 0
62
510
2
!
!3
K Z
1 X
o/PS
o/PS 5
4
ds on Cj j 1; 2; . . . ; K
CS
B Ej
Ei
E1 j1 Cj j
on
on
j
It
63a
K Z
h
i
1 X
Gj Gi yz2 zuPS j cos b zy2 yuPS j sin b ds on Cj j 1; 2; . . . ; K
G1 j1 Cj
63b
and using constant boundary elements for the approximation of these line integrals. In Eqs. (63a) and (63b) Cj
j 1; 2; . . . ; K is an interface between regions Xj and Xi , while Ej Gj 0 in the case Cj is a free boundary. Moreover, in
these equations the normal n to the boundary Cj points to the exterior of the region Xj and Cj is traveled only once, while
By; zj is a ctitious function dened as the solution of the following Neumann problem
r2 Bj uPS j
Ej
oB
on j
Ei
oB
on i
in Xj j 1; 2; . . . ; K
64
0 on Cj j 1; 2; . . . ; K
65
b2
z,Z
E1
C
h1
E2
E2
h2
b1
Fig. 4. Cross-section of the composite beam of Example 1.
Table 1
Buckling load P (kN) of the composite beam of Example 1
E2 =E1
Hingedhinged
Without shear
deformation
Fixedhinged
With shear
deformation
Present
study
Without shear
deformation
FEM
Fixedxed
With shear
deformation
Present
study
Without shear
deformation
FEM
With shear
deformation
Present
study
FEM
b2 h2 0:02 m
0.5
479
1
783
2
1487
476
780
1483
502
815
1490
530
834
1547
527
831
1544
541
841
1512
573
882
1612
570
879
1609
556
849
1520
b2 h2 0:08 m
0.5
7046
1
13,558
2
27,076
6845
13,089
25,828
6949
13,288
26,299
13,343
25,306
50,698
12,586
23,646
46,425
13,010
24,550
48,521
22,288
39,929
75,155
20,571
36,892
69,625
21,164
38,117
72,321
b2 h2
0.5
1
2
23,524
38,082
54,383
23,591
38,301
54,978
49,235
80,601
116,685
46,666
74,580
105,070
47,666
76,766
108,940
96,012
157,223
228,201
87,510
137,653
190,992
92,800
147,284
206,875
0:20 m
24,098
39,442
57,031
511
The evaluation of the primary warping function uPS and the ctitious function By; z is accomplished using BEM as this is
presented in Sapountzakis and Mokos (2001) and in Sapountzakis and Mokos (2003), respectively.
3.3. For the stress functions HY; Z and UY; Z
The evaluation of the stress functions HY; Z and UY; Z is accomplished using BEM as this is presented in Mokos and
Sapountzakis (2005). Moreover, since the exural-torsional buckling problem of composite beams is solved by the BEM, the
domain integrals for the evaluation of the area, the bending moments of inertia and the shear deformation coefcients (Eqs.
(39a) and (39b)) have to be converted to boundary line integrals, in order to maintain the pure boundary character of the
method. This is achieved using integration by parts, the Gauss theorem and the Green identity (Mokos and Sapountzakis,
2005).
Fig. 5. 3-D views of the buckling mode shapes of the FEM solution of the beams of Example 1 with E2/E1 = 0.5 (numbers in parentheses correspond to the
FEM solution).
512
4. Numerical examples
On the basis of the analytical and numerical procedures presented in the previous sections, a computer program has been
written and representative examples have been studied to demonstrate the efciency and the range of applications of the
developed method. In all the examples treated, each cross-section has been analysed employing N 300 constant boundary
elements along the boundary of the cross section, which are enough to ensure convergence at the calculation of the sectional
constants, while the beam interval is divided into L 60 constant equal elements.
Example 1. A monosymmetric beam of length l 3:0 m, with a composite cross section consisting of three rectangular parts
b1 h1 0:40 m in contact (reference material 1: E1 3:0 107 kN=m2 , m1 0:20, materials 2, 3: E2 E3 , m2 m3 0:20,
as this is shown in Fig. 4, has been studied. Three different types starting from a thin-walled and ending with a thick-walled
cross-section are considered, that is (i) b2 h2 0:02 m, (ii) b2 h2 0:08 m and (iii) b2 h2 0:20 m. In Table 1 the
computed values of the buckling load P for the cases of hingedhinged, xedhinged and xedxed boundary conditions
and for various values of the ratio E2 =E1 are presented taking into account or ignoring shear deformation effect as compared
with those obtained from a FEM solution (MSC/NASTRAN, 1999) employing 2600, 21,600 and 6000 solid brick elements for
the three cases, respectively (the buckling mode shapes of the latter are presented in Fig. 5). From the obtained results the
inuence of the inclusion of the aforementioned effect is remarkable leading to the conclusion that it has to be taken into
account. Moreover, it can be concluded that the inuence of the boundary conditions on the buckling load is signicant,
while the buckling load is increasing monotonically with the ratio E2 =E1 .
Example 2. To demonstrate the range of applications of the proposed method a slab-and-beam structure of length
l 40:0 m, with a composite cross section consisting of a rectangular concrete C20/25 plate (reference material 1:
E1 2:9 107 kN=m2 , m1 0:20 stiffened by two concrete C35/45 I-section beams (material 2: E2 3:35 107 kN=m2 ,
m2 0:20, as this is shown in Fig. 6, has been studied. In Table 2 the geometrical and inertia properties of the composite
cross-section are presented together with its warping, torsion constants and its shear deformation coefcients referred to
Z,z
330cm
25cm
30cm
15cm
30cm
S
Y
7cm
C20/25
C35/45
35cm
177cm
190cm
20cm
35cm
50cm
50cm 90cm
320cm
90cm
Table 2
Geometric, inertia, torsion and warping constants and shear deformation coefcients of the composite cross-section of example 2
IYY 4:049 m4
IZZ 17:182 m4
IYZ 0:00 m4
C S 16:572 m6
It 0:205 m4
IS 31:403 m4
ay 3:349
az 3:618
ayz 0:0
yC 0:0 m
zC 1:55 m
AG 4:246 m2
Table 3
Buckling load P (kN) of the composite beam of Example 2
Hingedhinged
Fixedhinged
Fixedxed
Without shear
deformation
With shear
deformation
Without shear
deformation
With shear
deformation
Without shear
deformation
With shear
deformation
674,313
661,137
1,082,168
1,045,833
1,834,828
1,744,819
513
its principal shear system of axes (the directions of which coincide with the principal bending ones due to the monosymmetric property of the cross-section). In Table 3 the computed values of the buckling load P for the cases of hingedhinged,
xedhinged and xedxed boundary conditions are presented taking into account or ignoring shear deformation effect.
From the obtained results the inuence of both the aforementioned effect and the boundary conditions is once more
remarkable.
Example 3. To demonstrate the range of applications of the proposed method, three different cases of a nonsymmetric beam
of length l 1:0 m, with a rectangular composite cross-section consisting of a rectangular part (reference material 1:
E1 2:9 107 kN=m2 , m1 0:30 stiffened by an L-section of unequal legs (material 2: E2 2:1 108 kN=m2 , m2 0:30 having three different thickness t values, namely t t1 1 cm (composite cross-section properties AG 3:188 102 m2 ,
It 6:800 105 m4 , C S 1:057 108 m6 , IS 1:147 104 m4 , t t 2 3 cm (composite cross-section properties
AG 5:394 102 m2 , It 8:611 105 m4 , C S 2:420 108 m6 , IS 2:024 104 m4 , t t 3 5 cm (composite crosssection properties AG 8:181 102 m2 , It 1:137 104 m4 , C S 4:009 108 m6 , IS 2:518 104 m4 , as this is shown
in Fig. 7, has been studied. Since the proposed method requires the coordinate system CYZ through the cross-sections centroid C to have Y, Z axes parallel to the principal shear axes, in the rst column of Tables 46 the geometric, the inertia constants and the shear deformation coefcients of the three aforementioned cases of the examined cross-section are given with
t
Principal shear axes (at S)
~
Z
h=15.5cm
~
Y
y
t
b=10.5cm
Fig. 7. Cross-section of the nonsymmetric composite beam of Example 3.
Table 4
Geometric, inertia constants and shear deformation coefcients of the composite cross-section of Example 3 for t t1 1 cm
eZ
e
Coordinate system C Y
514
Table 5
Geometric, inertia constants and shear deformation coefcients of the composite cross-section of Example 3 for t t2 3 cm
eZ
e
Coordinate system C Y
Table 6
Geometric, inertia constants and shear deformation coefcients of the composite cross-section of Example 3 for t t3 5 cm
eZ
e
Coordinate system C Y
Table 7
Buckling load P (kN) of the composite beam of Example 3
Hingedhinged
Fixedhinged
Without shear
deformation
With shear
deformation
Present
study
Fixedxed
Without shear
deformation
With shear
deformation
FEM
Without shear
deformation
Present
study
FEM
With shear
deformation
Present
study
FEM
t t 1 1 cm
9791
9364
9430
20,029
18,154
18,824
39,158
33,126
36,128
t t 2 3 cm
13,392
12,917
12,916
27,373
25,268
25,931
53,422
46,580
50,000
t t 3 5 cm
16,423
15,994
15,948
33,572
31,654
32,371
65,539
59,202
63,341
Table 8
Buckling load P(kN) of the composite beam of Example 3, for various beam lengths
Beam length
(m)
Hingedhinged
Fixedxed
Without shear
deformation
With shear
deformation
Discrepancy
(%)
t t 1 1 cm
1.00
0.90
0.80
9791
12,087
15,297
9364
11,444
14,281
4.56
5.62
7.11
t t 2 3 cm
1.00
0.90
0.80
13,392
16,530
20,915
12,917
15,811
19,778
t t 3 5 cm
1.00
0.90
0.80
16,423
20,271
25,649
15,994
19,621
24,618
Without shear
deformation
With shear
deformation
Discrepancy
(%)
39,158
48,334
61,155
33,126
39,464
47,616
18.21
22.48
28.43
3.68
4.55
5.75
53,422
65,877
83,224
46,580
55,776
67,736
14.69
18.11
22.87
2.68
3.31
4.19
65,539
80,835
102,157
59,202
71,408
87,554
10.70
13.20
16.68
515
Fig. 8. 3D views of the buckling mode shapes of the FEM solution of the beams of Example 3 (numbers in parentheses correspond to the FEM solution).
e Z,
e followed by the evaluation of the angle of rotation hS (Mokos and Sapountzarespect to an original coordinate system C Y
kis, 2005) giving the nal coordinate system CYZ and the new geometric, inertia constants and shear deformation coefcients
given in the second column of the aforementioned tables. In Tables 7 and 8 the computed values of the buckling load P for
various boundary conditions are presented taking into account or ignoring shear deformation effect. More specically, to
demonstrate the accuracy of the proposed method, in Table 7 the obtained results are compared with those obtained from
a FEM solution (MSC/NASTRAN, 1999) employing 16,500, 17,600 and 16,000 solid brick elements for the three cases, respectively (the corresponding buckling mode shapes of the latter method are presented in Fig. 8), while in Table 8 the discrepancy of the obtained results taking into account or ignoring shear deformation effect for various beam lengths demonstrates
the inuence of this effect.
5. Concluding remarks
In this paper a boundary element method is developed for the general exural-torsional linear buckling analysis of Timoshenko beams of arbitrarily shaped composite cross-section without initial effects. The main conclusions that can be drawn
from this investigation are:
516
(a) The numerical technique presented in this investigation is well suited for computer aided analysis for composite
beams of arbitrary cross-section, subjected to any linear boundary conditions and to an arbitrarily distributed or concentrated loading.
(b) The proposed method can treat composite beams of both thin and thick walled cross-sections taking into account the
warping along the thickness of the walls, while the displacements as well as the stress resultants are computed at any
cross-section of the beam using the respective integral representations as mathematical formulae.
(c) All basic equations are formulated with respect to the principal shear axes coordinate system, which does not necessarily coincide with the principal bending one.
(d) The discrepancy of the obtained results arising from the ignorance of shear deformation especially in thick-walled
cross-sections is remarkable and necessitates its inclusion in these cases.
(e) The developed procedure retains the advantages of a BEM solution over a pure domain discretization method since it
requires only boundary discretization.
Acknowledgements
This work has been funded by the Project PENED 2003. The project is conanced 75% of public expenditure through EC
European Social Fund and 25% of public expenditure through Ministry of Development General Secretariat of Research and
Technology and through private sector, under measure 8.3 of OPERATIONAL PROGRAM COMPETITIVENESS in the 3rd Community Support Program.
References
Cortnez, V.H., Piovan, M.T., 2006. Stability of composite thin-walled beams with shear deformability. Computers and Structures 84, 978990.
Cowper, G.R., 1966. The shear coefcient in Timoshenkos beam theory. Journal of Applied Mechanics, ASME 33 (2), 335340.
Hodges, D.H., 2001. Lateral-torsional utter of a deep cantilever loaded by lateral follower force at the tip. Journal of Sound and Vibration 247 (1), 175183.
Hodges, D.H., Peters, D.A., 1975. On the lateral buckling of uniform slender cantilever beams. International Journal of Solids and Structures 11, 12691280.
Hutchinson, J.R., 2001. Shear coefcients for Timoshenko beam theory. ASME Journal of Applied Mechanics 68, 8792.
Katsikadelis, J.T., 2002. The analog equation method, a boundary-only integral equation method for nonlinear static and dynamic problems in general
bodies. Theoretical and Applied Mechanics 27, 1338.
Kollr, L.P., 2001. Flexural-torsional buckling of open section composite columns with shear deformation. International Journal of Solids and Structures 38,
75257541.
Lee, J., Kim, SE., 2001. Flexural-torsional buckling of thin-walled I-section composites. Computers and Structures 79, 987995.
Machado, S.P., Cortnez, V.H., 2005. Lateral buckling of thin-walled composite bisymmetric beams with prebuckling and shear deformation. Engineering
Structures 27, 11851196.
Mei, C., 1970. Coupled vibrations of thin-walled beams of open-section using the nite element method. International Journal of Mechanical Science 12,
883891.
Michell, A.G.M., 1899. Elastic stability of long beams under transverse forces. Philosophical Magazine 48 (5th Series), 298309.
Milisavljevic, B.M., 1995. On lateral buckling of a slender cantilever beam. International Journal of Solids and Structures 32 (16), 23772391.
Mokos, V.G., Sapountzakis, E.J., 2005. A BEM solution to transverse shear loading of composite beams. International Journal of Solids and Structures 42,
32613287.
MSC/NASTRAN for Windows, 1999. Finite element modeling and postprocessing system. Help System Index, Version 4.0, USA.
Orloske, K., Leamy, M.J., Parker, R.G., 2006. Flexural-torsional buckling of misaligned axially moving beams: I. Three-dimensional modeling, equilibria, and
bifurcations. International Journal of Solids and Structures 43, 42974322.
Prandtl, L., 1899. Kipperscheinungen, Dissertation der Universitat Munchen.
Ramm, E., Hofmann, T.J., 1995. Stabtragwerke, Der Ingenieurbau. In: Mehlhorn, G. (Ed.), Band Baustatik/Baudynamik. Ernst&Sohn, Berlin.
Rao, J.S., Carnegie, W., 1970. Solution of the equations of motion of coupled-bending torsion vibrations of turbine blades by the method of RitzGalerkin.
International Journal of Mechanical Science 12, 875882.
Reissner, E., 1979. On lateral buckling of end-loaded cantilever beams. ZAMP 30, 3140.
Rothert, H., Gensichen, V., 1987. Nichtlineare Stabstatik. Springer-Verlag, Berlin.
Sapks, A., Kollr, L.P., 2002. Lateral-torsional buckling of composite beams. International Journal of Solids and Structures 39, 29392963.
Sapountzakis, E.J., 2005. Torsional vibrations of composite bars of variable cross-section by BEM. Computer Methods in Applied Mechanics and Engineering
194, 21272145.
Sapountzakis, E.J., Mokos, V.G., 2001. Nonuniform torsion of composite bars by boundary element method. Journal of Engineering Mechanics, ASCE 127 (9),
945953.
Sapountzakis, E.J., Mokos, V.G., 2003. Warping shear stresses in nonuniform torsion of composite bars by BEM. Computer Methods in Applied Mechanics and
Engineering 192, 43374353.
Sapountzakis, E.J., Tsiatas, G.C., 2007. Flexural-torsional buckling and vibration analysis of composite beams. Computers, Materials and Continua 6 (2), 103
115.
Schramm, U., Kitis, L., Kang, W., Pilkey, W.D., 1994. On the shear deformation coefcient in beam theory. Finite Elements in Analysis and Design 16, 141
162.
Schramm, U., Rubenchik, V., Pilkey, W.D., 1997. Beam stiffness matrix based on the elasticity equations. International Journal for Numerical Methods in
Engineering 40, 211232.
Stephen, N.G., 1980. Timoshenkos shear coefcient from a beam subjected to gravity loading. ASME Journal of Applied Mechanics 47, 121127.
Timoshenko, S.P., Goodier, J.N., 1984. Theory of Elasticity, third ed. McGraw-Hill, New York.
Timoshenko, S.P., Gere, J.M., 1961. Theory of Elastic Stability. McGraw-Hill, Tokyo.
Vlasov, V.Z., 1961. Thin-walled elastic beams. Israel Program for Scientic Translations, Jerusalem.
Yu, W., Hodges, D.H., Volovoi, V., Cesnik, C.E.S., 2002. On Timoshenko-like modeling of initially curved and twisted composite beams. International Journal
of Solids and Structures 39, 51015121.