Professional Documents
Culture Documents
International Journal for Computational Methods in Engineering Science and Mechanics, 7:425431, 2006
c Taylor & Francis Group, LLC
Copyright
ISSN: 15502287 print / 15502295 online
DOI: 10.1080/15502280600826381
In this paper, the exact two-node Timoshenko beam finite element is formulated using a new model for representing beam rotation in a shear deformable beam. An exact relationship between
bending rotation and shear rotation was achieved using an analytical bending and shear rotation interdependent shape functions
obtained from a consideration of the asymmetrical beam flexural
mode, which is shown to embody bending and shearing kinematics. These functions enable the total beam cross sectional rotation
to be expressed in terms of bending and shear rotation, and subsequently lead to the use of the usual cubic interpolation and linear interpolation to model the bending rotation and shear rotation
based beam curvatures respectively. The formulation ensures the
circumvention of the shear-locking phenomenon, permitting complete interaction between bending and shear deformation fields
and thus allows for a straightforward derivation of the exact Timoshenko beam stiffness matrix and consistent nodal load vector as
obtained in classical structural analysis.
Keywords
1. INTRODUCTION
The Euler-Bernoulli beam theory (EBT) frequently used for
the analysis of isotropic beams, which have extensive use in engineering structures, describes beam kinematics completely in
terms of flexural deformation. In neglecting the contribution of
shearing deformation the EBT requires that plane sections remain plane and perpendicular to the neutral axis after deformation. Consequently, this theory is best suited for thin or slender
beams as shear strains have a considerable influence on the deformation of thick beams. A more accurate representation of beam
flexure which allows for the inclusion of shear strains present
in isotropic beams and more suited for thick beam analysis is
the Timoshenko beam theory [1]. This theory, a first order shear
deformation theory (FSDT), relaxes the normality assumption
of plane sections evident in the EBT. By allowing for the inclusion of a constant through thickness shear strain, it violates the
no-shear boundary condition at the top and bottom horizontal
beam surfaces, requiring a problem dependent shear correction
factor [2].
Displacement-based beam finite elements founded on the
Timoshenko beam theory have been beset by numerical difficulties in the thin beam limit, such as the shear locking phenomenon. This is characterized as the inability of the beam
to yield zero shear strains as the beam becomes progressively
slender, thus leaving the deformation considerably underestimated. The locking phenomenon is observed when using equal,
particularly linear interpolation polynomial for transverse displacement and sectional rotation. Traditionally, shear locking
has been alleviated using reduced or selective integration. This
basically involves using an integration rule one order less for
the evaluation of the shear stiffness terms while full integration
is used to evaluate the bending stiffness terms [3]. The genesis of the shear locking phenomenon has been ascribed to a
number of causes. Carpenter et al. [4] credited the coupling
between bending and shear rotation in the Timoshenko beam
equations as causing shear locking and suggested that these be
decoupled through the use of appropriate strain fields. Using
the concept of field consistency, Prathap [5, 6] attributed shear
locking to the choice of interpolation functions used for displacement fields, which impose spurious constraints that occur
because the strains developed from the displacement approximation are inconsistent. Discussions of different interpolation
425
426
I. B. EDEM
(1)
427
b1,1 =
(2)
b1,2
(3)
mL
.
6E I
Q
.
G A
(5)
(8)
s1,2
(9)
In the above two equations, the notation of the superscripts defined earlier still applies. The total shear rotation at node 1 is the
addition of (8) and (9)
s1 =
2m
.
LG A
(10)
b1
.
b1 + s1
(11)
s1
.
+ b1
(13)
.
1+
(14)
b + s = 1.
(15)
s =
Notice that
m1
.
LG A
m2
=
.
LG A
s1
(4)
s =
(12)
=
12E I
L 2 G A
(16)
1
1+
2 1 (1
+ )()2
(17)
dw
b + s (x)s .
dx
(18)
428
I. B. EDEM
(x) =
M(x)
d x.
EI
(19)
Using the expression for (x) from (18) and substituting in (19)
gives
L
dw
M(x)
b + s (x)s =
.
dx
EI
0
M(x)
d dw
b + s (x)s =
.
dx dx
EI
(20)
(21)
FIG. 2.
On expanding the bracket on the left, two distinct momentcurvature equations can be written
E I1
d 2w
= M1 (x) and
dx2
E I2
ds
= M2 (x).
dx
(22)
d 2w
dx
and s =
ds
.
dx
(23)
x
1.
L
x2 ), a non-
(24)
d() d
1
= .
d d x
L
(25)
d()
.
dx
(26)
429
(30)
w
1 = w(0), w
2 = (0) = 1 ,
w
3 = w(l), w
4 = (L) = 2 .
dw
dx
s (x ) = c1 + c2 x .
= D T c.
s1 = 1 = s (0),
s2 = 2 = s (L).
s = [1
2 ]T .
s = Sc.
c = S1 s .
(33)
where
w = Ca.
(34)
a = C 1 w.
(35)
1
S=
1
0
L
Solving for a
C=
1
1
L
0
L2
2L
C 1
1
0
=
3/L 2
2/L 3
L3
(36)
3L 3
0
1
0
0
2/L
3/L 2
1/L 2
2/L 3
0
0
1/L
N1s = (1 ),
(37)
1/L 2
w(x) = N w.
(48)
b = B w,
(38)
(39)
N2w = L (1 )2 ,
N3w
N4w
= L (1 )
2
N2s = .
(50)
s = N s s .
s = Bs s .
and
(51)
(52)
s = (N ) s = Bs s .
(53)
(49)
and
(47)
= (3 2 )
s (x) = N s s
1
0
1/L 1/L
(45)
(46)
where
1
(44)
(43)
(32)
w = [w
1 1 w
2 2 ]T
(41)
(42)
(31)
H is the column vector [1 x x 2 x 3 ],T a stands for the column vector of ai for 1 i 4 and x = x x1 . The nodal displacements
are labeled as
where =
(40)
Bw =
B
N1w
N2w
[(N1s ) (N2s ) ].
N3w
N4w
and
(54)
430
I. B. EDEM
and
B2s =
1
.
L
Rearranging (57),
1
T
( w)
[(Bw )T E I1 Bw w]Ld
+ ( s )T
0
1
1
s T
s
[(B ) E I2 B s ]Ld ( w)
q(x)(N w )T Ld = 0.
K e = K eb + K es .
( w)
0
1
q(x)(N w )T Ld = 0.
0
1
T
( s )
[(Bs )T E I2 Bs s ]Ld = 0.
e
d = [w
1 1 w
2 2 ]T .
12
6L
12
6L
4L 2
6L
2L 2
E I b
6L
K eb =
. (68)
L 3 12 6L
12
6L
(59)
(60)
=
[(Bs )T E I2 Bs ]Ld .
and
K es =
E I s
L
q(x)(N w )T Ld = 0. (61)
0
0
1
[(Bs )T E I2 Bs s ]Ld = 0.
(62)
(67)
2L 2
1
.
1
6L
6L
s
2
4L + L 2
b
6L
s
2
2L L 2
b
12
6L
1
1
4L 2
(69)
(66)
e
d is the vector of nodal displacement for the assembled beam
element given as
(58)
(65)
e
K e d f e = 0.
K eb
(64)
(56)
+
B s E I2 Bs s Ld
0
0
1
q(x)(N w )T w Ld = 0.
(57)
q(x)(N w )T Ld .
and
B1s =
f =
e
K es
(63)
E I b
K =
L3
e
12
6L
12
6L
6L
12
6L
6L
s
2L 2 L 2
b
.
6L
s 2
2
4L + L
b
(70)
12
6L
6L
2
4L + L 2
12
6L
12
2L 2 L 2
6L
6L
6L
12
6L
2L 2 L 2
6L
4L 2 + L 2
(71)
qL L
e
f =
.
(72)
6
12
12
6L
12
6L
6L
6L
2L 2 L 2
4L 2 + L 2
EI
L (1 + ) 12
6L
12
6L
2L 2 L 2
w
1
6
qL
2
L
6
w
2
12
L
2
6L
6L
4L 2 + L 2
(73)
431