Professional Documents
Culture Documents
Free Vibration Analysis of Beams With Non-Ideal Clamped Boundary Conditions
Free Vibration Analysis of Beams With Non-Ideal Clamped Boundary Conditions
discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/257775206
CITATIONS
1 AUTHOR:
Jinhee Lee
Hongik University
24 PUBLICATIONS 214 CITATIONS
SEE PROFILE
DOI 10.1007/s12206-012-1245-2
Abstract
A non-ideal boundary condition is modeled as a linear combination of the ideal simply supported and the ideal clamped boundary conditions with the weighting factors k and 1-k, respectively. The proposed non-ideal boundary model is applied to the free vibration analyses of Euler-Bernoulli beam and Timoshenko beam. The free vibration analysis of the Euler-Bernoulli beam is carried out analytically,
and the pseudospectral method is employed to accommodate the non-ideal boundary conditions in the analysis of the free vibration of
Timoshenko beam. For the free vibration with the non-ideal boundary condition at one end and the free boundary condition at the other
end, the natural frequencies of the beam decrease as k increases. The free vibration where both the ends of a beam are restrained by the
non-ideal boundary conditions is also considered. It is found that when the non-ideal boundary conditions are close to the ideal clamped
boundary conditions the natural frequencies are reduced noticeably as k increases. When the non-ideal boundary conditions are close to
the ideal simply supported boundary conditions, however, the natural frequencies hardly change as k varies, which indicate that the proposed boundary condition model is more suitable to the non-ideal boundary condition close to the ideal clamped boundary condition.
Keywords: Euler-Bernoulli beam; Free vibration; Natural frequency reduction; Non-ideal boundary conditions; Timoshenko beam
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
1. Introduction
In the problems of mechanical systems the boundary conditions of structures play a very important role and are usually
represented in the idealized forms such as clamped, simply
supported, and free boundary conditions. It is always assumed
that those ideal boundary conditions are satisfied exactly in the
process of the problem solution. In practice, however small,
deviations from the ideal boundary conditions may exist and
the ideal boundary condition assumptions sometimes lead to
unsatisfactory solutions. The types of boundary conditions
with small deviations from the ideal boundary conditions are
referred as the non-ideal boundary conditions.
Often the non-ideal boundary conditions were represented
by using perturbation methods. Cherki et al. investigated the
uncertain boundary conditions of imperfectly clamped joints
and represented the uncertainties as fuzzy parameters with
assumed membership functions [1]. Pakdemirli and Boyaci
considered the vibrations of Euler-Bernoulli beams with simply supported and sliding-clamped boundary conditions [2, 3].
The non-ideal boundary conditions were modeled using perturbation theory, and it was shown how the non-ideal boundary conditions affected the natural frequencies and mode
*
Corresponding author. Tel.: +82 10 7321 2589, Fax.: +82 44 862 2664
E-mail address: jinhlee@hongik.ac.kr
298
parameters [8]. Pai et al. employed the two spring model and
considered the vibration of Euler-Bernoulli beams with damages and imperfect boundary conditions [9]. A signal decomposition method was suggested, where an experimental operational deflection shape was processed using a sliding-window
least squares curve fitting technique to separate central and
boundary solutions. The extracted boundary solutions and the
central solutions were used to identify the damage locations
and the boundary conditions, respectively. Ritto et al. developed a finite element model based on the Timoshenko beam
theory, in which the uncertain boundary conditions were represented by pinned while the rotation was restrained by a torsional spring [10]. A parametric stochastic approach was used
to model the stiffness of the torsional spring. Sari and Butcher
applied the shifted Chebyshev polynomials to compute the
natural frequencies of a tapered Euler beam, where the damaged boundary conditions were modeled as the two spring
model [11]. Wang and Yang modeled the boundary condition
of a tapered cantilever beam by two springs [12]. The relationship between the static flexibility and boundary conditions
was deduced based on the static equilibrium equation, and a
set of linear equations for identifying the boundary conditions
was formed by static flexibility measurements.
The clamped boundary condition is considered to be
achieved by fixing a beam between a top and a bottom plate
using bolts. In practice, however, the clamped boundary condition may be degenerated from its ideal state when the
clamping force is reduced due to the aging and/or the damages
on its support. The simply supported boundary condition is
hard to come by because it is very difficult to install a moment-free sharp edge support at the extremity of a beam. In
this study, the non-ideal boundary condition is represented by
a linear combination of the ideal clamped boundary condition
and ideal simply supported boundary condition, and its effect
on the natural frequencies of uniform beams is discussed.
(0 x L) .
(1)
Y ( x) = C1 cos x C2 sin x
+ C3 cosh x + C4 sinh x
Y ( x) = C1 2 sin x C2 2 cos x
+ C3 2 sinh x + C4 2 cosh x .
(2a)
(2b)
(2c)
(3a)
(3b)
ideal clamped:
Y ( xb ) = 0, Y ( xb ) = 0.
ideal free:
(3c)
In this study the non-ideal boundary condition is represented by a linear combination of the ideal clamped and the
ideal simply supported boundary conditions. It is worthwhile
to note that Eq. (3a) and Eq. (3b) have Y ( xb ) = 0 in common,
which is also employed in the non-ideal boundary condition as
it is. The other condition is obtained by the addition of the
other conditions Y ( xb ) = 0 and Y ( xb ) = 0 with the
weighting factors k and 1 k , respectively, so that the proposed non-ideal boundary condition is expressed as
Y ( xb ) = 0, kLY ( xb ) (1 k ) Y ( xb ) = 0, ( 0 k 1) .
(4)
In Eq. (4) L is multiplied to Y ( xb ) to match the dimension of Y ( xb ) . Eq. (4) indicates that for k = , where is
a small positive value, the originally clamped boundary condition is deteriorated to allow a reduced clamping force. It also
describes the case that the originally simply supported boundary condition is stiffened for k = 1 . Eq. (4) becomes the
ideal clamped boundary condition for k = 0 , and the ideal
simply supported boundary condition for k = 1 . It is found
that the sign convention in Eq. (4) is different for the left end
and the right end of the beam, and the non-ideal boundary
condition is rewritten as
Y = 0, k L LY (1 k L ) Y = 0 at left end ( xb = 0 )
(5)
Y = 0, k R LY + (1 k R ) Y = 0 at right end ( xb = L ) .
(1 kL ) Y (0) kL LY (0) = 0
Y ( L) = 0, Y ( L) = 0 .
(6a, b)
(6c, d)
kL L
1 kL
k L L C 2 0
1 kL
. (7)
=
sin L cos L sinh L cosh L C3 0
299
kL
0.01
0.02
0.03
0.04
0.05
kL
kR
0
0
0.01
0.01
0.02
0.02
0.03
0.03
0.04
0.04
0.05
0.05
1L
2L
3L
4L
5L
1.8751
4.6941
7.8548
10.996
14.137
1.8566
4.6493
7.7820
10.897
14.014
1.8386
4.6085
7.7195
10.817
13.919
1.8210
4.5712
7.6654
10.751
13.846
1.8039
4.5369
7.6183
10.697
13.787
1.7872
4.5055
7.5770
10.651
13.740
1L
2L
3L
4L
5L
4.7300
7.8523
10.996
14.137
17.279
4.6405
7.7090
10.799
13.892
16.988
4.5599
7.5867
10.642
13.707
16.779
4.4870
7.4818
10.514
13.563
16.624
4.4205
7.3908
10.408
13.448
16.505
4.3597
7.3113
10.319
13.354
16.411
1 kL
1 kL
kL L
kL L
det
=0.
sin L cos L sinh L cosh L
(8)
(1 k L ) Y ( 0 ) kL LY ( 0 ) = 0
(1 kR ) Y ( L ) + kR LY ( L ) = 0 .
(9a, b)
(9c, d)
Eqs. (2a-c) are substituted into Eqs. (9a-d) and are rearranged in a matrix form
0
1
0
1
C1 0
1 kL
kL L
1 kL
k L L
C 0
sin L
cos L
sinh L
cosh L 2 = .
C
0
k
L
k
L
k
L
1
cos
sin
1
cosh
1
( R)
( R)
( kR ) sinh L C3 0
(
R)
4
+ k R L sinh L + k R L cosh L
k R L sin L k R L cos L
(10)
For the non-trivial solution of Eq. (10), it is required that the
determinant of the matrix vanishes
0
1
0
1
1 kL
1 kL
kL L
k L L
=0.
sin
cos
sinh
cosh
L
L
L
L
det
k
cos
L
k
sin
L
1
k
cosh
L
1
k
sinh
L
(
)
(
)
(
)
(
)
R
R
R
R
(11)
0.95
0.95
0.96
0.96
0.97
0.97
0.98
0.98
0.99
0.99
1
1
1L
2L
3L
4L
5L
3.1581
6.2915
9.4303
12.571
15.711
3.1547
6.2898
9.4292
12.570
15.711
3.1514
6.2881
9.4281
12.569
15.710
3.1481
6.2864
9.4269
12.568
15.709
3.1448
6.2848
9.4258
12.567
15.709
3.1416
6.2832
9.4248
12.566
15.708
300
E* E
1/2
1/3
1/4
1/5
1L
2L
3L
4L
5L
1.8751
4.6941
7.8548
10.996
14.137
1.8659
4.6718
7.8186
10.947
14.076
1.8566
4.6501
7.7844
10.902
14.022
1.8481
4.6308
7.7549
10.864
13.977
1.8395
4.6119
7.7267
10.826
13.937
2x L
1, 1 .
L
z=
(13)
+ Gh W * = 2 I ,
L2
L
2
L
G * +
(14a)
4 **
W = 2 W
L2
(14b)
ideal free:
(15a)
(15b)
2 *
W =0
L
(15c)
at the extremities zb = 1 . The rotation and the lateral deflection W are approximated by the series expansion
( z)
N +2
a T ( z ),
n n 1
W (z)
n =1
N +2
b T (z) .
n n 1
(16a, b)
n =1
(16a, b) are substituted into Eq. (14a, b), and are collocated at
the Chebyshev grid points
zi = cos
( 2i 1)
2N
i = 1, , N
(17)
a
n =1
4 EI **
2 Gh
bnTn*1 ( zi )
2 Tn 1 ( zi ) GhTn 1 ( zi ) +
L
L
N +2
(18a)
= 2 I an 1Tn 1 ( zi )
n =1
4 G
2 G
anTn*1 ( zi ) + 2 bnTn**1 ( zi )
L
L
n =1
N +2
(18b)
N +2
= 2 bnTn 1 ( zi )
n =1
( i = 1, , N ) .
Eqs. (18a, b) can be rearranged in the matrix form
EI + Gh (W ) = 2 I ,
(12a)
Gh ( + W ) = 2 hW
(12b)
(0 < x < L) .
{ }
{ })
K {d } + K d = M {d } + K d
(19)
301
{d } = {a1 a2 aN b1 b2 bN }
{d } = {aN +1 aN + 2 bN +1 bN + 2}T .
T
(20a)
(20b)
*
W = 0, 2k R + (1 k R ) = 0
at left end ( zb = 1)
at right end ( zb = 1)
(21)
Using the expressions of Eq. (16a, b), Eq. (21) can be realized by
N +2
N +2
*
R n 1
n =1
n 1
n =1
(22a)
N +2
N +2
b T ( 1) = 0, a {2k T ( 1) (1 k )T ( 1)} = 0.
n n 1
*
L n 1
n =1
n 1
n =1
(22b)
The ideal free boundary condition of Eq. (15c) is also expressed as
N +2
N +2
2b
a T ( z ) = 0, L
*
n n 1
n =1
n =1
Tn*1 ( zb ) anTn 1 ( zb ) = 0
(23)
at zb = 1 .
Lets consider a situation that a Timoshenko beam is restrained by the non-ideal condition at the left end and free at
the right end. The combination of Eq. (22a) and Eq. (23)
makes up the four additional equations required to match the
number of unknowns:
N +2
bnTn 1 ( 1) = 0
n =1
N + 2
*
an 2k LTn 1 ( 1) (1 k L ) Tn 1 ( 1) = 0
n =1
N +2
anTn*1 (1) = 0
n =1
N +2
2bn *
n =1 L
(24a)
{ }
Since {d } is expressed as
{d } = V
U {d } .
(26)
(24b)
(24c)
(K K V
(24d)
(25)
U {d } = 2 M M V U . (27)
1
Eq. (27) can be solved for the estimates of the natural frequencies and corresponding expansion coefficients. This procedure can also be applied the case that both ends of the beam
are restrained by the non-ideal boundary conditions. In such a
case, Eq. (24a-d) would be replaced with
302
4. Conclusions
bnTn 1 ( 1) = 0
n =1
N +2
*
an 2k LTn 1 ( 1) (1 k L ) Tn 1 ( 1) = 0
n =1
N +2
bnTn 1 (1) = 0
n =1
N +2
(28a)
(28b)
(28c)
(28d)
Acknowledgment
This work was supported by 2011 Hongik University Research Fund.
References
[1] A. Cherki, G. Plessis, B. Lallemend, T. Tison and P. Level,
Fuzzy behavior of mechanical systems with uncertain
boundary conditions, Computer Methods in Applied Mechanics and Engineering, 189 (2000) 863-873.
[2] M. Pakdemirli and H. Boyaci, Vibrations of a stretched
beam with non-ideal boundary conditions, Mathematical &
Computational Applications, 6 (3) (2001) 217-220.
[3] M. Pakdemirli and H. Boyaci, Effects of non-ideal boundary
conditions on the vibrations of continuous systems, Journal
of Sound and Vibration, 249 (4) (2002) 815-823.
[4] M. Pakdemirli and H. Boyaci, Non-linear vibrations of a
simple-simple beam with a non-ideal support in between,
Journal of Sound and Vibration, 268 (2003) 331-341.
[5] M. Aydogdu and M. C. Ece, Buckling and vibration of nonideal simply supported rectangular isotropic plates, Me-
303