Académique Documents
Professionnel Documents
Culture Documents
H
-------------
\ .
| |
)
`
=
y s ( )
F
1
EA
0
---------s
F
2
e
----- sinh
1
F
3
H
-----
\ .
| |
sinh
1
F w
0
H
-------------
\ .
| |
)
`
=
z s ( )
F
3
EA
0
---------s
F
1
2EA
0
------------
H
w
---- 1
F
3
H
-----
\ .
| |
1 2
+
)
`
1 1
F
2
w
0
H
----------------
\ .
| |
2
+
)
`
1 2
=
Fig. 1. Spatially Suspended Elastic Catenary Cable
Fig. 2. Elastic Catenary Cable Element Subjected to Multiple Con-
centrated Loads
Nonlinear Dynamic Analysis of Spatially Suspended Elastic Catenary Cable with Finite Element Method
Vol. 12, No. 2 / March 2008 123
(6)
where,
x(s
0
) = 0 y(s
0
) = 0 z(s
0
) = 0 @s(s
0
) = 0
(7)
x(s
N+1
) = l
x0
y(s
N+1
) = l
y
z(s
N+1
) = l
z
@(s
N+1
) = L
0
x(s
N+1
) = l
x
= l
x0
A
1
+ A
4
= f(H
x
, H
y
, V
z
, L
0
)
y(s
N+1
) = l
y
= l
y0
A
2
+ A
5
= g(H
x
, H
y
, V
z
, L
0
) (8)
z(s
N+1
) = l
z
= l
z0
A
3
+ A
6
= h(H
x
, H
y
, V
z
, L
0
)
where, l
x0
, l
y0
, l
z0
are initial coordinates of the j-end of the cable,
and l
x
, l
y
, l
z
are the displaced coordinates. A
n
is a displcement of
n
th
degree of freedom of the end nodes i and j.
The nodal forces of the i-node, H
x
, H
y
, and V
z
are determined
using the Newton-Raphson iterative procedure. In order to
implement the procedure, the flexibility matrix [F] as given in
Eq. (9). This is also essential in order to estimate displacement
interpolation functions and the tangential stiffness matrix. Each
component of the flexibility matrix is obtained from the
compatibility condition of Eq. (8) as given in Eq. (10).
(9)
where,
(10)
The tangential stiffness matrix of the cable element is required
in order to perform the nonlinear finite element analysis. This is
derived from derivatives of the nodal forces according to the
nodal displacements for each six degrees of freedom as given in
Eq. (11).
(11)
where, F
A
is the local nodal force vector due to self-weight,
internal concentrated loads, and local nodal displacements as
given in Eq. (12). Also, A represents local nodal displacement
vector.
(12)
The components of the first column for the tangential stiffness
matrix, [K
c
], can be derived from the differentiation of the
flexibility matrix, [F], to displacement A
1
as given in Eq. (13).
The same approach provides another five columns for the stiffness
matrix. It should be noted that the tangent stiffness matrix of a
cable element is calculated from the flexibility matrix of Eq. (9).
(13)
x s ( ) f( )
H
x
s
EA
0
-----------
1
EA
0
--------- f
xj
s
j
s ( ) { }
j 1 =
i
_
+ = =
1
w
0
----- H
xk
VR
k 1
( ) ln VR
k
( ) ln { } | |
k 1 =
i 1 +
_
y s ( ) g( )
H
x
s
EA
0
-----------
1
EA
0
--------- f
yj
s
j
s ( ) { }
j 1 =
i
_
+ = =
1
w
0
----- H
yk
VR
k 1
( ) ln VR
k
( ) ln { } | |
k 1 =
i 1 +
_
z s ( ) h( )
V
z
s
EA
0
---------
w
0
s
2
2EA
0
------------
1
EA
0
--------- + f
zj
s
j
s ( ) { }
1
w
0
----- R
k 1 +
R
k
( )
k 1 =
i 1 +
_
j 1 =
i
_
+ = =
H
xk
H
x
f
xj
( )
j 1 =
k 1
_
= H
yk
H
y
f
yj
( )
j 1 =
k 1
_
= V
zk
V
z
f
zj
( )
j 1 =
k 1
_
=
VS
k 1
V
zk
w
0
S
k 1
+ = VS
k
V
zk
w
0
S
k
+ = R
k
H
xk
2
H
yk
2
H
k
2
+ + =
VR
k 1
VS
k 1
R
k 1
+ = VR
k
VS
k
R
k
+ =
dl
x
dl
y
dl
z
df
dH
x
---------
df
dH
y
---------
df
dV
z
--------
dg
dH
x
---------
dg
dH
y
---------
dg
dV
z
--------
dh
dH
x
---------
dh
dH
y
---------
dh
dV
z
--------
dH
x
dH
y
dV
z
F | |
dH
x
dH
y
dV
z
= =
df
dH
x
---------
s
EA
0
---------
1
w
0
----- VR
k 1
( ) ln VR
k
( ) ln { } | |
k 1 =
i 1 +
_
=
H
xk
2
1 VR
k 1
R
k 1
( ) 1 VR
k
R
k
( ) { } ( ) + | |
df
dH
y
---------
1
w
0
----- H
xk
H 1 VR
k 1
R
k 1
( ) 1 VR
k
R
k
( ) { } | |
dg
dH
x
--------- =
k 1 =
i 1 +
_
=
df
dV
z
--------
1
w
0
----- H
xk
1 R
k 1
1 R
k
( ) { }
dg
dH
x
--------- =
k 1 =
i 1 +
_
=
dg
dH
y
---------
s
EA
0
---------
1
w
---- VR
k 1
( ) ln VR
k
( ) ln { } | |
k 1 =
i 1 +
_
=
H
yk
2
1 VR
k 1
R
k 1
( ) 1 VR
k
R
k
( ) { }} + |
dg
dV
z
--------
1
w
0
----- H
yk
1 R
k 1
1 R
k
( ) { }
dh
dH
y
--------- =
k 1 =
i 1 +
_
=
dh
dV
z
--------
s
EA
0
---------
1
w
0
----- VS
k 1
R
k 1
VS
k
R
k
( )
k 1 =
i 1 +
_
=
dF
A
dA
--------- K
c
| |
cH
x
cA
1
---------
cH
x
cA
2
--------- .
cH
x
cA
6
---------
cH
y
cA
1
---------
cH
y
cA
1
--------- .
cH
y
cA
1
---------
cV
z
cA
1
--------
cV
z
cA
2
-------- .
cV
z
cA
6
--------
cH
x
cA
1
---------
cH
x
cA
2
---------.
cH
x
cA
6
---------
cH
y
cA
1
---------
cH
y
cA
2
---------.
cH
y
cA
6
---------
cV
z
cA
1
--------
cV
z
cA
2
--------.
cV
z
cA
6
--------
=
F
A
F
1
F
2
F
3
F
4
F
5
F
6
, , , , , { }
T
=
H
x
H
y
V
z
H
x
f
xj
( )
j 1 =
N
_
H
y
f
yj
( )
j 1 =
N
_
V
z
f
zj
( )
j 1 =
N
_
, , , , ,
)
`
T
=
1
0
0
F | |
dH
x
dA
1
dH
y
dA
1
dV
z
dA
1
=
Sung-Pil Chang, Jung-Il Park and Kyoung-Chan Lee
124 KSCE Journal of Civil Engineering
2.3 Three Dimensional Cable Element Subjected to Dis-
tributed Load
A cable structure can easily be subjected to distributed load
from external environements such as wind action and water
pressure as well as its own self-weight. However, these effects
are not easily implementable by a mathematical model due to the
complex behavior of the catenary cable itself. This study suggests
a method for solving this problem using multiple Gaussian
points. If a cable is suspended arbitrarily in a three dimensional
space, the tension force, T, at a point p(x, y, z) of the cable can be
divided with three component forces F
x
, F
y
, and F
z
as shown in
Fig. 3. The equilibrium equation for the point is given in Eq. (14).
For components of the tension forces H
x
, H
y
and V
z
in a cable at a
point 0 s s
1
, geometric compatibility equation x(s), y(s) and
z(s) can also be estimated by substituting s for s
1
into Eq. .
for (14)
The numerical difficulties of the distributed load arise from the
continuous integration of acting forces F(s) in Eq. (14). If a
sufficient number of Gauss points are assigned to an element, the
integration can be discritized with equivalent forces acting at
Gaussian length with Gaussian weight as given in Eq. (15).
Therefore, the element formulation, subjected to multiple con-
centrated loads from Eqs. (6)-(13), can also be applicable to a
cable element subjected to a distributed load substituted with
equivalent multi-Gaussian point loads as shown in Fig. 4.
(15)
where, the subscript i represnets the x, y and z directions. Also,
f
ij
represents an equivalent Gaussian point load in the i-direction
at the j
th
Gaussian point
j
with Gaussian weight of w
j
and total
number of n Gauss points.
The above discretization is generally acceptable when the
distributed loads do not vary significantly along the element,
which is standard for usual environmental loading conditions.
2.4 Displacement Interpolation Function
In the conventional finite element method, the displacement
interpolation functions are derived from the displacement boundary
conditions of nodes. However, the interpolation function of a cable
element can not be obtained directly from the boundary conditions
because of its inherent flexbible behavior. Therefore, this study
suggests a new method for determining the interpolation function
from the flexibility matrix, [F], given in Eq. (9).
In the case of the first degree of freedom, which is in the x-
direction of the i-node given as A
1
in Fig. 2, variations of tension
forces (dH
x
, dH
y
, dV
z
) due to the infinitesimal virtual displa-
cement dA
1
to the first degree of freedom can be reduced to Eq.
(16) from Eq. (13). Fig. 5 depicts a conceptual diagram of this
process.
(16)
where [K] is an inverse matrix of the flexibility matrix, [F].
Subsequently, the new profile of the deformed cable in the new
virtual coordinate system (x
1
, y
1
, z
1
) is given as Eq. (17).
Therefore, the displacement interpolation function related to the
first degree of freedom is determined as Eq. (18) whose right
hand terms can be expressed as components of the flexibility and
tangential stiffness matrixes as given in Eq. (19) according to the
chain rule. Furthermore, the same procedure provides another
T
dx s ( )
dp
------------ H
x
F
x
s ( ) s d
0
s
1
)
=
T
dy s ( )
dp
------------ H
y
F
y
s ( ) s d
0
s
1
)
= s
1
s L
0
s s
T
dz s ( )
dp
------------ V
z
F
z
s ( ) s w
0
s d
0
s
1
)
=
F
i
s ( ) s d
0
s
1
)
F
i
j
s
1
( )
j 1 =
n
_
w
j
s
1
f
ij
j 1 =
n
_
= =
dH
x
dA
1
dH
y
dA
1
dV
z
dA
1
F | |
1
1
0
0
=
dH
x
dH
y
dV
z
K | |
dA
1
0
0
=
Fig. 3. Spatially Suspended Cable Subjected to Distributed Load
Fig. 4. Distributed Load Substituted to Equivalent Multiple Gauss-
ian Point Loads
Nonlinear Dynamic Analysis of Spatially Suspended Elastic Catenary Cable with Finite Element Method
Vol. 12, No. 2 / March 2008 125
displacement functions related to remaining five degrees of
freedom.
(17)
(18)
(19)
2.5 Mass and Damping Matrixes
The mass matrix of a cable element is derived from the derived
displacement interpolation functions and the discritization
concept. Each component of the consistent mass matrix, [M
c
], is
discritized to Gaussian masses as given in Eq. (20).
(20)
where m denotes mass per unit length and n is the total number
of Gaussin points. This equation follows the tensor notation: q
and r denote degrees of freedom from one to six, and i represents
x, y and z.
In addition, the equation constructs the structural damping
matrix, [C
c
], that substitutes mass per unit length of cable m in
Eq. as material damping per unit length of cable c.
2.6 Dynamic Motion of the Elastic Catenary Cable
In dynamic motion, the mass matrix is generally considered as
an invariant value during the time history and as an iterative
procedure during nonlinear analysis. However, in the cable element,
the mass and damping matrixes are composed of terms of
tangential stiffness and flexibility matrix, which are composed of
variant values from the time history and iterative procedure.
Therefore, inertia and damping internal force terms should be
estimated at every time interval, as well as the tangential stiffness
itself. These inertial and damping forces are also discretized in
this study as multiple inertial loads acting on Gaussian point
masses along the profile of the cable element as shown in Fig. 4.
Accordingly, all the nodal forces are estimated together with
inertia, damping and nodal forces due to self-weight and nodal
displacement. The discretized inertial force is derived as given in
Eq. (21) where s is equal to s
1
for unstrained cable length L
0
. The
Gaussian point damping loads can also be evaluated by
substituting the material damping per unit length of cable and
nodal velocity vector in place of mass per unit length of cable
and nodal acceleration vector as given in Eq. (22). In these
equations,
j
L
0
represents the j
th
positions of s along the cable
length on which the j
th
Gaussian mass m L
0
w
j
is acting.
(21)
(22)
where, m denotes the mass per unit length of cable.
c denotes the material damping per unit length of cable.
L
0
denotes the unstrained cable length
w
j
denotes the Gaussian weight at Gauss point
j
.
Additionally, subscript i denotes x, y and z, and f
i
ij
and f
c
ij
denote internal inertia and damping forces of j
th
Gaussian point
applied in the i-direction, respectively.
2.7 Coordinate Transformation Matrix
The coordinate transformation matrix, [T], is constructed as
described in Eq. (23). This matrix is used to transform the nodal
force, displacement, velocity, acceleration vectors and mass and
damping matrix from the global Cartesian coordinate system to
the local Cartesian coordinate system.
(23)
where,
U
x
, U
y
and U
z
denote unit direction cosines of a vector
directing from the i-node to the j-node of a cable as shown in Fig.
2 according to the global Cartesian coordinate axes. Also, x
i
, y
i
,
and z
i
are global nodal coordinates of the i-node and x
j
, y
j
, and z
j
are those of the j-node.
x
1
s
i
( ) f H
x
dH
x
+ H
x
dH
x
+ H
x
dH
x
s
i
, + , , ( ) =
y
1
s
i
( ) g H
x
dH
x
+ H
x
dH
x
+ H
x
dH
x
s
i
, + , , ( ) =
z
1
s
i
( ) h H
x
dH
x
+ H
x
dH
x
+ H
x
dH
x
s
i
, + , , ( ) =
|
x1
s
i
( ) 1 dx s
i
( ) dA
1
+ =
|
y1
s
i
( ) dy s
i
( ) dA
1
=
|
x1
s
i
( ) dz s
i
( ) dA
1
=
|
x1
s
i
( ) 1
df s
i
( ) dH
x
dH
x
dA
1
-------------------------
df s
i
( ) dH
y
dH
y
dA
1
-------------------------
df s
i
( ) dV
z
dV
z
dA
1
------------------------ + + + =
|
y1
s
i
( ) 1
dg s
i
( ) dH
x
dH
x
dA
1
--------------------------
dg s
i
( ) dH
y
dH
y
dA
1
--------------------------
dg s
i
( ) dV
z
dV
z
dA
1
------------------------- + + + =
|
z1
s
i
( ) 1
dh s
i
( ) dH
x
dH
x
dA
1
--------------------------
dh s
i
( ) dH
y
dH
y
dA
1
--------------------------
dh s
i
( ) dV
z
dV
z
dA
1
------------------------- + + + =
m
qr
m |
ir
s ( ) |
iq
s ( ) s d
0
L
)
m e
j
L |
ir
j
L ( ) |
iq
j
L ( ) | |
j 1 =
n
_
= =
f
ij
i
m w
j
L
0
|
ki
k 1 =
6
_
j
L
0
( ) = A
k
f
ij
c
c w
j
L
0
|
ki
k 1 =
6
_
j
L
0
( ) = A
k
T | |
U
x
U
z
1 + ( )
2
----------------- 0 0 U
x
U
z
1 ( )
2
---------------- 0 0
U
y
U
z
1 + ( )
2
----------------- 0 0 U
y
U
z
1 ( )
2
---------------- 0
U
z
U
z
1 + ( )
2
----------------- 0 0 U
z
U
z
1 ( )
2
----------------
U
x
U
z
1 + ( )
2
----------------- 0 0
Symmetry U
y
U
z
1 + ( )
2
----------------- 0
U
z
U
z
1 + ( )
2
-----------------
=
U
x
x
j
x
i
( )
x
j
x
i
--------------- = U
y
j
j
j
i
( )
y
j
y
i
------------- and U
z
y
j
y
i
( )
y
j
y
i
--------------- = = ,
Fig. 5. Virtual Displacement dA
1
to the First Degree of Freedom
Sung-Pil Chang, Jung-Il Park and Kyoung-Chan Lee
126 KSCE Journal of Civil Engineering
3. Numerical Examples
3.1 Free Vibration of a Suspended Cable
In order to evaluate the applicability and correctness of the
cable element presented herein, data from an experiment (Koh,
2007) is compared to results from analysis of the numerical
model of the experiment on the developed element. In addition,
the numerical model is analyzed using the plane hybird beam
element (B21H) of ABAQUS for the same example.
In the experiment, a cable is suspended with hinged ends. The
material and geometric properties of the cable are presented in
Table 1 and Fig. 6. Initially, a 100 Kg mass is hung statically at
the middle point of the cable. The mass is then suddenly
removed, and the displacement and acceleration at the middle
point of the cable are measured at every 0.01 second intervals
while the cable is in a free vibration.
The numerical model employs fourteen cable finite elements
and three Gaussian points to substitute a distributed load for
multiple concentrated loads in the analysis. When modeling with
ABAQUS, on the other hand, twenty B21H beam elements are
applied. In the beam analysis, the statically deformed coordinates
of nodes are determined from the elastic catenary cable theory in
order to follow the exact initial configuration of the cable under
static configuration. In the experiment report (Koh, 2007), the
average modal damping ratio is estimated to 0.00155, so that the
numerical models follow the same damping ratio.
Fig. 7 shows the displacement time history for the first four
seconds of the free vibration. These result provide evidence that
the present study gives better agreement to the experimental
results than the beam analysis of ABAQUS. In addition, the
Fourier transform of the time history response produces the
natural frequencies, the results of which are given in Table 2 up
to fifth mode. This shows that the proposed cable element gives
results that are close to the experiment results. The ABAQUS
beam element, on the other hand, gives results with about 2.5% ~
6.2% error.
3.2 Forced Vibration of a Suspension Cable by Spatial
Support Excitations
Another example is presented in order to verify the numerical
performance of the element in three-dimensional analysis. Fig. 7
depicts the numerical model, which is a hinge supported cable on
the xy-plane. The material properties of the cable are given in
Table 3 and the displacement of the center point of the cable is
inspected.
This example applies rotational displacements about the x-
direction on the yz-plane at the left support, which is point A as
shown in Fig. 8. The excited rotational displacements are given
in Eq. (24).
(24)
where, A
sy
and A
sz
denote excited support displacements in the
y and z directions, respectively.
A
sy
3 1 tt ( ) cos ( ) A
sz
3 tt ( ) sin = =
Fig. 6. Illustration of Models for Verification
Table 1. Material Properties of the Cable
Youngs modulus 200 kN/mm
2
Mass per unit length 20.3 kg/m
Effective section area 2348 mm
2
Unstrained cable length 44.304 m
Fig. 7. Displacement Response to Free Vibration of a Suspended
Catenary Cable
Table 2. Natural Frequencies of a Suspension Cable
Mode Experiment [Hz] Present [Hz] (error) ABAQUS Beam [Hz] (error)
1
st
1.51 1.51 (0.0%) 1.56 (3.3%)
2
nd
4.41 4.39 (0.5%) 4.30 (2.5%)
3
rd
7.43 7.37 (0.8%) 7.03 (5.4%)
4
th
10.61 10.50 (1.0%) 9.96 (6.1%)
5
th
13.96 13.72 (1.7%) 13.09 (6.2%)
Nonlinear Dynamic Analysis of Spatially Suspended Elastic Catenary Cable with Finite Element Method
Vol. 12, No. 2 / March 2008 127
For the purpose of comparison, analysis of the ABAQUS
beam element, B31 is performed with a flexural rigidity to axial
stiffness ratio of 1/1000. Fig. 9 shows the displacement time
history by means of 10 and 50 elements of the present cable
element and the ABAQUS beam element. Results from the
present study tend to converge as number of element increases;
however, the beam element produces a noisy response due to the
increase in the degree of freedom. It seems that the present cable
element is adequate for analyzing the three-dimensional behavior
of a spatially suspended catenary cable.
4. Conclusion
A consistent interpolation function for a catenary cable has
long been an obstable to the application of the finite element
method to dynamic analysis of the cable. Although the elastic
catenary cable finite element had previously been formulated
(Chang and Park, 1989, Lee, 1990, Ahn, 1991, Park, 1992), it
was limited to plane problems, and the displacement interpolation
function was only a rough estimate. This paper proposes an
improved interpolation function for the spatially suspended elastic
catenary cable. The newly formulated displacement interpolation
function is obtained from the flexibility matrix calculated by
means of the exact geometircal compatibility conditions. Con-
sequently, this paper proposes a spatially suspended elastic
catenary cable element that is applicable to both static and
dynamic analyses. The element is derived from equations where
the geometrical compatibility condition of a cable are subjected
to multiple concentrated loads based on the elastic catenary cable
theory of Irvine (1981) in a three-dimensional space. The
element is further expaned to analyze the distributed load case
with appropriate numerical discretization using multiple Gaussian
points. The discretization seems to be acceptable when the
distributed loads do not vary significantly along the element,
which is standard for usual environmental loading conditions.
The similar discretization concept derives internal forces
relating to dynamic motion, inertial and damping forces. These
internal forces include terms from the flexibility and stiffness
matrices so that they generate inherently geometrical nonlinear
behavior. This supports results from previous studies (Mesarovic
and Gasparini, 1992b, Volokh et. al., 2003), where the dynamic
behavior of a catenary cable is found to be definitely nonlinear
based on the numerical parametric study. Therefore, the total
internal force at every time interval and loading step should be
estimated together with the inertial, damping as well as elastic
forces of the cable element.
The proposed element is compared to an experiment on the
free vibration of a suspension cable in order to prove the
correctness and applicability, and to the ABAQUS beam element
in order to demonstrate its numerical efficiency. Displacement
responses in the time history show better agreement with the
experiment than with the ABAQUS beam element. In addition,
the natural frequencies are almost the same as those from the
experiment. Another example presents three-dimensional rotational
displacement excitations at the support of a suspension cable
structure. In this example, as the number of elements is
increased, the present study converges to consistent results.
However, the ABAQUS beam element shows noisy vibration as
the number of element is increased.
As a result, this paper presents a consistent formulation for
analyzing the static and dynamic behavior of spatially suspended
elastic catenary cables under multiple internal and distributed
loads. The proposed element will be a valuable engineering tool
in providing accurate analyses of cable structures under various
kinds of environmental load such as earthquake, wind and water
pressures.
Acknowledgements
This work has been part of a research project supported by the
Table 3. Material Properties of the Cable
Youngs modulus 340 kN/mm
2
Mass per unit length 5.0 kg/m
Effective section area 100 mm
2
Unstrained cable length 15.0 m
Fig. 8. Illustrative Model for Support Excitation Example
Fig. 9. Displacement at the Center Point of a Cable under Support
Excitation
Sung-Pil Chang, Jung-Il Park and Kyoung-Chan Lee
128 KSCE Journal of Civil Engineering
Korea Ministry of Construction & Transportation (MOCT)
through the Korea Bridge Design & Engineering Research
Center at Seoul National University. The authors wish to express
their gratitude for the financial support from these organizations.
References
Ann, S. S. (1991). Static and Dynamic Nonlinear Analysis of Spatial
Cable Networks using Elastic Catenary cable Element, Master
thesis, Seoul National University, Seoul, Korea.
Chang, S. P. and Park, J. I. (1989). A study on the Nonlinear Finite
Cable Element Journal of Korean Society of Coastal and Ocean
Engineers, Vol. 1, No. 1, pp. 93-101.
Irivine, H. M. (1981). Cable Structures, MIT press.
Kim, J. and Chang, S. P. (2001). Dynamic stiffness matrix of an
inclined cable Engineering Structures, Vol. 23, pp. 1614-1621.
Koh, H. M. (2007). Estimation of the Modal Damping Ratio in Stay
Cable by Free Vibration and Forced Vibration Test, Korea Bridge
Design and Engineering Research Center, Seoul, Korea.
Lee, D. I. (1990). The Static Analysis of the Cable Member by FEM,
Master thesis, Seoul National University, Seoul, Korea.
Leonard, J. W and Recker, W. W (1972). Nonlinear dynamics of cables
with low initial tension Jounral of Engineering Mechacnics,
ASCE., Vol. 98, pp.293-309.
Mesarovic, S. and Gasparini, D. A (1992b). Dynamic bahavior of non-
linear cable system. II Journal of Engineering Mechanics, ASCE,
Vol. 118, No. 5, pp. 904-920.
Park, Y. Y. (1992). Dynamic Analysis of Cable-Suspended Roof System
Subjected to a Wind Load, PhD thesis, Seoul National University,
Seoul, Korea.
Park, J. I. (1993). Dynamic Analysis of Tension Leg Platform Using 3D-
Curved Surface Boundary Element Methods, PhD thesis, Seoul
National University, Seoul, Korea.
Veletsos, A. S. and Darbre, G. R. (1983). Free Vibration of Parabolic
Cables Journal of Sturctural Engineering, ASCE, Vol. 109, No. 2,
pp. 503-519.
Volokh, K. Y., Vilnay, O., Averbuh, I. (2003). Dynamics of Cable
Structures Journal of Sturctural Engineering, ASCE, Vol. 129, No.
2, pp. 175-180.