Вы находитесь на странице: 1из 10

Finite Elements in Analysis and Design 47 (2011) 237246

Contents lists available at ScienceDirect

Finite Elements in Analysis and Design


journal homepage: www.elsevier.com/locate/finel

Nonlinear static and dynamic analysis of cable structures


Huu-Tai Thai, Seung-Eock Kim n
Department of Civil and Environmental Engineering, Sejong University, 98 Gunja Dong Gwangjin Gu, Seoul 143-747, Republic of Korea

a r t i c l e in f o

abstract

Article history:
Received 14 June 2010
Received in revised form
11 October 2010
Accepted 24 October 2010
Available online 24 November 2010

This paper presents a catenary cable element for the nonlinear analysis of cable structures subjected to
static and dynamic loadings. The element stiffness matrix and element nodal forces, which account for
self-weight and pretension effects, are derived based on exact analytical expressions of elastic catenary.
Cables encountered in cable networks as well as cable-supported bridges can be modeled using the
proposed element. An incremental-iterative solution based on the Newmark direct integration method
and the NewtonRaphson method is adopted for solving the nonlinear equation of motion. The accuracy
and reliability of the present element are verified by comparing the predictions with those generated by
commercial finite element package SAP2000, and the results given by other authors using different
analytical or numerical approaches.
& 2010 Elsevier B.V. All rights reserved.

Keywords:
Catenary element
Cable structures
Nonlinear analysis
Dynamic analysis

1. Introduction
In recent decades, cable element has been widely used in
tension structures such as cable-supported bridges and roofs of
structures covering large unobstructed areas due to their aesthetic
appearances as well as the structural advantages of cables. Since
the highly nonlinear behavior exhibits in this element, the effects of
flexibility and large deflection in the cable should be considered in
establishing the equilibrium equations. In general, the cable
member can be modeled using two different approaches: (1) the
finite element approach based on the polynomial interpolation
functions and (2) the analytical approach based on analytical
expressions of elastic catenary.
In the first method, the interpolation functions are adopted to
represent the nonlinear effects of the cable. This method has been
employed to formulate two-node element, multi-node element,
and curved element with rotational degrees of freedom. The twonode element is the most common element used in the modeling of
cables, and was adopted by several researches [13]. This element
is only suitable for modeling the cables with high pretension. To
account for the sag effect, the elastic modulus is modified by the
equivalent modulus proposed by Ernst [4]. Several researchers
have adopted the equivalent modulus for modeling the cables,
which have been proved to be sufficiently accurate for the cases of
cable under relatively high stress and small length [5]. For cables
with large sag, a series of straight elements is used to model the
curved geometry of cables. The multi-node element was developed

Corresponding author. Tel.: + 82 2 3408 3291; fax: + 82 2 3408 3332.


E-mail addresses: taispkt@yahoo.com (H.-T. Thai), sekim@sejong.ac.kr
(S.-E. Kim).
0168-874X/$ - see front matter & 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.finel.2010.10.005

instead of using many two-node elements. The multi-node element


is based on the higher order polynomials for the interpolation
functions [68]. Since the formulations of this element are complex
expressions, the tangent stiffness matrix, and nodal force vector are
obtained using the isoparametric formulation. These elements give
accurate results for cables with small sag. When the large sag cable
is modeled by several elements, the continuity of slopes is violated.
The continuity of the slopes can be enforced by adding rotational
degrees of freedom to the nodes. Such an element was developed
by Gambhir and Batchelor [9]. In general, the polynomial based
elements are only appropriate to model the cable with small sag.
For cable element with large sag, it is necessary to use a large
number of elements to model the curved geometry of cable.
Therefore it causes computational costs.
In the second approach, exact analytical expressions of elastic
catenary are used to describe the realistic behavior of cables. This
method was originally proposed by OBrien and Francis [10] and later
developed by Jayaraman and Knudson [11], Wang et al. [12], Andreu
et al. [13], Yang and Tsay [14], and Such et al. [15]. In this method, the
curved cables are modeled by a single two-node catenary element
without internal joints. This element can be used to model the small
sag cables in cable-stayed bridges as well as large sag cables in
suspension bridges. Compared to the finite element method, the
analytical approach has some advantages such as requiring fewer
number of degree of freedom and exactly considering the nonlinear
effects of the cable. The catenary cable element presented in this
study is derived based on the second approach.
The purpose of this paper is to develop a spatial two-node catenary
cable element for the nonlinear analysis of cable structures subjected
to static and dynamic loadings. The tangent stiffness matrix and
internal force vector of the element are derived explicitly based on the
exact analytical expressions of elastic catenary. Self-weight of the

238

H.-T. Thai, S.-E. Kim / Finite Elements in Analysis and Design 47 (2011) 237246

cables can be directly considered without any approximations. The


effect of pre-tension of cable is also included in the element
formulation. It should be noted that most of the finite element
package still lack suitable cable element; therefore the proposed
element is also implemented in a computer program for practical use
in design. An incremental-iterative solution based on the Newmark
direct integration method and the NewtonRaphson method is
adopted for solving the nonlinear equation of motion. Several
numerical examples are presented and discussed to illustrate the
accuracy and efficiency of the proposed element in predicting the
static and dynamic responses of cable structures.

2. Catenary cable element


To accurately simulate the realistic behavior of cable structures,
the cable element presented in this paper is derived based on the
exact analytical expressions of the elastic catenary element. It is
assumed that the cable is perfectly flexible with the self-weight
distributed along its length, and the cross-sectional area of the
cable is kept constant. Fig. 1 shows the cable suspended between
two points I and J which have the Cartesian coordinates (0, 0, 0) and
(lx, ly, lz), respectively. The Lagrangian coordinates of the undeformed and deformed configurations are s and p. The equations for
the equilibrium condition of the cable can be expressed as follows:
 
dx
F1
1a
T
dp

dy
F2
T
dp


1b


dz
F3 ws
dp

1c

where F1, F2 and F3 are the projected components of cable tension in


the x-, y- and z-axis, respectively; w the self-weight of the cable;
and T the cable tension at the Lagrangian coordinate s given by
q
Ts F12 F22 F3 ws2
2

The cable tension T is related to the strain e by Hooks law as






dpds
dp
1
3
EA
T EAe EA
ds
ds

where E and A are the elastic modulus and cross-sectional area of


the cable. The relationships between the Lagrangian coordinate s
and the Cartesian coordinate are as follows:
Z
Z
dx
dx dp
xs
ds
ds
4a
ds
dp ds
ys

dy
ds
ds

dy dp
ds
dp ds

4b

zs

dz
ds
ds

dz dp
ds
dp ds

4c

F6
y

F5

and the boundary conditions at the two ends of cable are


x0 y0 z0 0

5a

xL0 lx , yL0 ly , zL0 lz

5b

Substituting Eqs. (1)(3) into Eq. (4) and applying the boundary
conditions in Eq. (5), the projected lengths of the cable can be
derived as follows:
 q

F1 L0 F1

ln
F12 F22 wL0 F3 2 wL0 F3
lx
EA
w
q

6a
ln
F12 F22 F32 F3
 q

F2 L0 F2

ln
F12 F22 wL0 F3 2 wL0 F3
EA
w
q

2
ln
F1 F22 F32 F3

ly

lz

F3 L0 wL20
1

EA
2EA w

q q
F12 F22 wL0 F3 2 F12 F22 F32

F2

lx f F1 , F2 , F3

7a

ly gF1 , F2 , F3

7b

lz hF1 , F2 , F3

7c

The tangent stiffness matrix and corresponding internal force


vector of the element can be derived using an iteration procedure in
solving Eq. (6). By differentiating both sides of Eq. (6), the following
equations are obtained as:
dlx

@f
@f
@f
dF1
dF2
dF3
@F1
@F2
@F3

8a

dly

@g
@g
@g
dF1
dF2
dF3
@F1
@F2
@F3

8b

dlz

@h
@h
@h
dF1
dF2
dF3
@F1
@F2
@F3

8c

or in matrix form
9 2
8
f11 f12
>
=
< dlx >
dly 6
4 f21 f22
>
;
: dl >
f31 f32
z

9
9
8
38
f13 >
>
=
=
< dF1 >
< dF1 >
7
f23 5 dF2 F dF2
>
>
>
;
: dF >
f33 : dF3 ;
3

where F is the flexibility matrix given as follows:






Tj F6
F2
L0
1
1
1
f11
1
log

EA w
Ti F3
w Ti Ti F3 Tj Tj F6

f12 f21

10a





F1 F2
1
1
F1 1 1

, f13 f31

w Ti Ti F3 Tj Tj F6
w Tj Ti
10b

lz

lx
w

F1

6c

where L0 is unstressed length of the cable. The above expressions


for lx, ly, and lz can be rewritten in terms of the end forces (F1, F2, F3)
as

F4
F3

6b





Tj F6
F2
L0
1
1
1
2
f22
log

EA w
Ti F3
w Ti Ti F3 Tj Tj F6

ly

I
Fig. 1. Three-dimensional catenary cable element.

x
f23 f32



F2 1 1

,
w Tj Ti

f33



L0 1 F6 F3

EA w Tj
Ti

10c

10d

H.-T. Thai, S.-E. Kim / Finite Elements in Analysis and Design 47 (2011) 237246

where Ti and Tj are the cable tension at nodes I and J, defined as


q
Ti F12 F22 F32
11a
Tj

q
F42 F52 F62

11b

and nodal forces (F4, F5, F6) at node J are obtained from equilibrium
equations as
F4 F1

12a

F5 F2

12b

F6 F3 wL0

12c

The stiffness matrix is obtained by taking the inverse of the


flexibility matrix F as
2
31
f11 f12 f13
6
7
13
K F 1 4 f21 f22 f23 5
f31 f32 f33
The tangent stiffness matrix and corresponding internal force
vector of cable element can be expressed in terms of the six degrees
of freedom as


K
K
KT
14
K
K
T

15

zs lL3 12xl sin yx1x=3

17

Fint F1


F2

F3

F4

F5

F6

Once the tangent stiffness matrix and internal force vector are
determined, the cable length S and the cable sag zs (Fig. 2) can be
obtained as follows [11,14]:
s
2
sinh l
S l2z l2x l2y
16
2

where

w q
l2x l2y =F12 F22
l
2
L

18a

q
l2x l2y l2z

18b

q
x x= l2x l2y

18c

An elasticplastic hinge model is adopted herein for representing the inelastic behavior of cable element. According this model,
the spread of plasticity of cable element is assumed to be lumped at
two ends of element, while the whole element remains elastic. If
the axial force of cable is greater than the yield force Py Asy, the
elastic modulus of the cable element will become zero and the axial
force of cable element will be equal to the yield force Py.

239

3. Procedure for computing the stiffness matrix


The tangent stiffness matrix and internal force vector of cable
element are evaluated using an iteration procedure. This procedure
requires the initial values of end forces (F1, F2, F3). Based on
the well-known catenary expressions, the initial values of end
forces are obtained as follows [11]:
F1

wlx
2l0

19a

F2

wly
2l0

19b

F3



w
coshl0
lz
L0
2
sinhl0

in which
8
>
106 if l2x l2y 0
>
>
>
>
2
2
2
2
>
>
< 0:2 if L0 r lx ly lz

l0 v
!
>u
u L2 l2
>
>
>
> t3 20 z2 1
if L20 4l2x l2y l2z
>
>
lx ly
:

19c

20

The iteration procedure for obtaining tangent stiffness matrix


and internal force vector of cable element is briefly presented as
follows:
Step 1 Input w, E, A, L0, nodes I(xi, yi, zi) and J(xj, yj, zj)
Step 2 Calculate lx0 xjxi, ly0 yjyi and lz0 zjzi
Step 3 Initialize end forces (F1, F2, F3) using Eq. (19)
Step 4 Update (lx, ly, lz) using Eq. (6)

Step 5 Calculate misclosure vector dL lx0 lx ly0 ly
T
lz0 lz g
Step 6 If dL is smaller than the preset tolerances, calculate KT
using Eq. (14) and Fint using Eq. (15). Otherwise continue to
next step
Step 7 Calculate the correction vector of end forces dFF 1dL
Step 8 Update the end forces Fi + 1 Fi + dF and go to Step 4
When the initial cable tension T0 is given instead of the
unstressed cable length L0, a similar iteration procedure can be
adopted to determine the unstressed cable length. The iteration
procedure for obtaining the unstressed cable length is summarized
as follows:
Step 1 Input w, E, A, nodes I(xi, yi, zi) and J(xj, yj, zj)
Step 2 Calculate lx0 xjxi, ly0 yjyi and lz0 zjzi
Step 3 Initialize unstressed length L0 and end forces (F1, F2, F3) as
follows:
q
21a
L0 l2x0 l2y0 l2z0
F1

lx0
T0
L0

21b

F2

ly0
T0
L0

21c

F3

lz0
T0
L0

21d

x tan

zs
J
Fig. 2. The sag zs of the inclined cable.

Step 4 Update (lx, ly, lz) using Eq. (6)



Step 5 Calculate misclosure vector dL lx0 lx ly0 ly
lz0 lz gT
Step 6 If dL is smaller than the given tolerance, the current L0 is
unstressed cable length. Otherwise continue to next step

240

H.-T. Thai, S.-E. Kim / Finite Elements in Analysis and Design 47 (2011) 237246

Step 7 Calculate [C] by differentiating Eq. (6) with respect to F2,


F3 and L0 as
2
3
@f
@f
@f
6 @F2 @F3 @L0 7
6
7
6 @g
@g @g 7
7
6
C 6
22
7
6 @F2 @F3 @L0 7
7
6
4 @h @h @h 5
@F2 @F3 @L0

T
C1 dL
Step 8 Calculate dF2 dF3 dL0
Step 9 Update F2 F2 + dF2, F3 F3 + dF3, L0 L0 + dL0 and go to
Step 4

4. Solution algorithm
For the nonlinear static analysis, the residual forces in each load
increment can be dissipated using the NewtonRaphson method.
For the nonlinear timehistory analysis, an incremental-iterative
solution based on the Newmark direct integration method and the
NewtonRaphson method is employed to solve the nonlinear
equation of motion. The incremental equation of motion of a
structure can be written as
CfDDg
_ KfDDg fDFg
MfDDg

23

DD,
_ and [DD] are the vectors of incremental accelwhere DD,
eration, velocity, and displacement, respectively; [M], [C], and [K]
are the mass, damping, and tangent stiffness matrices, respectively; {DF} is the external load increment vector. The viscous
damping matrix [C] can be defined as
C aM M aK K

24

where aM and aK are mass- and stiffness-proportional damping


factors, respectively. With the adoption of the average acceleration
method of the Newmark family (g 1/2, b 1/4), the incremental
acceleration and velocity at the first iteration of each time step can
be written as
n
o
4  4 n_ o n o
25
DD 2 DD
D n 2 D n
Dt
Dt
n

DD_

2  n_ o
DD 2 D n
Dt

26

force {DR} as
^ DDDg fDRg
Kf

33

where the residual force {DR} is determined based on the total


external force {F}, inertial force, damping force, and updated
internal force {Fint} as
n 1 gCfD
_ n 1 gfFint g
fDRg fFn 1 gMfD

Once the convergence criterion is satisfied, the structural


response is updated for the next time step as
fDDk 1 g fDDk g fDDDg

35

fDn 1 g fDn g fDDk 1 g

36

n
o
n o 2 n
o
_ n1 D
_n
DDk 1
D
Dt

37

n
o
n o 4 n o
n
o
_ n 4 DDk 1
n
n1 D
D
D
2
Dt
Dt

38

5. Numerical verifications
A computer program is developed based on the abovementioned algorithm. The flow chart of the proposed program
for the application of the Newmark method and the Newton
Raphson method is illustrated in Fig. 3. Two earthquake records of
the El Centro and the Loma Prieta as shown in Fig. 4 are used as
ground excitation in the dynamic analysis. Their peak ground
accelerations and time steps are listed in Table 1. In the dynamic
timehistory analysis, the mass- and stiffness-proportional damping factors are chosen based on the first two modes of the structure
so that the equivalent viscous damping ratio is equal to 5%. Several
numerical examples are presented and discussed to verify the
accuracy and efficiency of the proposed program in predicting the
nonlinear response of cable structures subjected to static and
dynamic loadings. For the verification purpose, the predictions
obtained from proposed program are compared with available
results reported in the literature, and those generated by SAP2000.
It should be noted that the cable element provided by SAP2000
ignores the self-weight and inelastic effects, whereas the proposed
element can consider these effects.

Substituting Eqs. (25) and (26) into Eq. (23), the incremental
displacement can be calculated from
27

At the first iteration of each time step, the total displacement,


velocity, and acceleration at the time t + Dt is updated based on the
incremental displacement vector {DD} as

n
n

_ n1
D

n1
D

n o 2 
_n
D
DD
Dt

n o 4 n o

_ n 4 DD
n
D
D
Dt
Dt2

30
31

Form the tangent stiffness matrix

Next iteration

^ are the effective stiffness matrix and incremen^ and fDFg


where K
tally effective force vector, respectively, given as
h i
4
2
M
C K
K^
28
Dt
Dt2
n o
n o  4
n o
_ n 2M D
n
M 2C D
DF^ DF
29
Dt

fDn 1 g fDn g fDDg

Previous time step

Current time step

^ DDg fDFg
^
Kf

Solve for the increment displacement

Update element force

Calculate the residual force R


No

Check convergence of R?
Yes

Update structural response D,D,D

32

For the second and subsequent iterations of each time step,


the structural system is solved under the effect of the residual

34

Next time step


Fig. 3. Flow chart of the proposed program.

241

H.-T. Thai, S.-E. Kim / Finite Elements in Analysis and Design 47 (2011) 237246

P = 35.586 kN

1.2

Acceleration (g)

0.8

30.48 m

0.4

121.92 m
152.4 m

0
0

10

15

20

25

30

304.8 m

-0.4

Fig. 5. Isolated cable under concentrated load.

-0.8
-1.2

Time (s)

Table 2
Initial properties of isolated cable under concentrated load.

1.2

Acceleration (g)

0.8
0.4
0
0

10

15

20

25

30

Item

Data

Cross-sectional area
Elastic modulus
Cable self-weight
Sag under self-weight at load point
Unstressed cable length of sections 12
Unstressed cable length of sections 23

5.484 cm2
13100.0 kN/cm2
46.12 N/m
29.276 m
125.88 m
186.85 m

40

-0.4
Table 3
Comparison of displacements of isolated cable under concentrated load.

-0.8
-1.2

Researcher

Time (s)

Element type

Fig. 4. Earthquake records. (a) El Centro and (b) Loma Prieta.

Table 1
Peak ground acceleration and its corresponding time step of earthquake records.
Earthquake

PGA (g)

Time
step (s)

El-Centro (1940) (Array, #9, USGS Station 117)


Loma Prieta (1989) (Capitola, 000, CDMG Station 47125)

0.319
0.529

0.020
0.005

Michalos and Birnstiel [16]


OBrien and Francis [10]
Jayaraman and Knudson [11]
Jayaraman and Knudson [11]
Tibert [17]
Andreu et al. [13]
Yang and Tsay [14]
SAP2000
Present work

Elastic
Elastic
Elastic
Elastic
Elastic
Elastic
Elastic
Elastic
Elastic

straight
catenary
straight
catenary
catenary
catenary
catenary
catenary
catenary

Displacements (m)
Vertical

Horizontal

5.472
5.627
5.471
5.626
5.626
5.626
5.625
5.626
5.626

0.845
0.860
0.845
0.859
0.859
0.860
0.859
0.860
0.859

40

5.1. Static analysis


Example 1. The first example is an isolated cable spanning of
304.8 m between two supports at the same elevation, where the
sag at the mid-span is 30.48 m. This problem was first considered
by Michalos and Birnstiel [16], and later analyzed by OBrien and
Francis [10], Jayaraman and Knudson [11], Tibert [17], Andreu et al.
[13], and Yang and Tsay [14]. The initial configuration and the data
for this structure are presented in Fig. 5 and Table 2. Table 3 shows
the comparison of displacements generated by the proposed
program, SAP2000 and other researchers. It can be seen that the
results obtained by the proposed element are almost identical with
those predicted by other authors. It should be noted that the set of
catenary type elements shows a more deformable behavior than
the straight type element. Moreover, the straight type element
can achieve the same level of accuracy of the catenary type
elements when using much more elements per member. The
loaddisplacement curves obtained by the proposed program
and SAP2000 are compared in Fig. 6. A good agreement is obtained.
Example 2. A hyperbolic paraboloid cable network as shown in
Fig. 7 has been numerically analyzed by several researches as well

Applied load, P (kN)

35

uz

ux
Present
SAP2000

30
25
20
15
10
5
0
0

-1

-2

-3

-4

-5

-6

Displacement at node 2 (m)


Fig. 6. Loaddisplacement curves of isolated cable under concentrated load.

as experimentally tested by Lewis et al. [18]. The structure consists


of a grid in a hyperbolic paraboloid form with 31 cable segments
subjected to concentrated load of 15.7 N at some joints. The elastic
modulus and cross-sectional area of cables are 128.3 kN/mm2 and
0.785 mm2. The cables are pretensioned to carry 200 N prior to the

242

H.-T. Thai, S.-E. Kim / Finite Elements in Analysis and Design 47 (2011) 237246

8
3

12

23

18

17

11

22

16

10

21

15

9
4x0
.4 m

0.45 m

13

26

20

14

m
0.8

25

19

24

m
0.4
3x

0.9
m

Fig. 7. Hyperbolic paraboloid net.

Table 4
Comparison of vertical displacements (mm) of hyperbolic paraboloid net.
Node Experiment
(Ref. [18])

Dynamic relaxation
(Ref. [18])

Minimum energy
(Ref. [19])

Present

5
6
7
10
11
12
15
16
17
20
21
22

19.30( 1.03)a
25.30(0.00)
23.00(0.88)
25.90(1.97)
33.80(0.60)
29.40(2.08)
26.40(4.76)
31.70(3.59)
21.90(4.29)
21.90(4.29)
20.50(3.54)
14.80(4.23)

19.30( 1.03)
25.50(0.79)
23.10(1.32)
25.80(1.57)
34.00(1.19)
29.40(2.08)
25.70(1.98)
31.20(1.96)
21.10(0.48)
21.10(0.48)
19.90(0.51)
14.30(0.70)

19.56(0.31)
25.70(1.58)
23.37(2.50)
25.91(2.01)
34.16(1.67)
29.60(2.78)
25.86(2.62)
31.43(2.71)
21.56(2.67)
21.57(2.71)
20.14(1.72)
14.55(2.46)

19.50
25.30
22.80
25.40
33.60
28.80
25.20
30.60
21.00
21.00
19.80
14.20

a
Numbers in parentheses are the percentage error with respect to experiment
values.

x
6x4.0

0m
= 24.

12
4

z
2
6 1
13
=

4x

4.0

20

16

.0

14

9
15

26

18
17

Example 5. An inclined cable subjected to earthquake loading is


shown in Fig. 10 with its associated data. The horizontal displacement responses of the cable obtained by the proposed program and
SAP2000 for the nonlinear elastic analysis are compared in Fig. 11
for two different earthquakes. It can be seen that a strong
agreement of displacement responses of the cable generated using
the proposed program and SAP2000 is obtained. A comparison of
the peak displacement responses is shown Table 7. It can be
observed that the proposed program and SAP2000 give nearly

25

16

24
23

22

21

5.2. Dynamic timehistory analysis

19

11

5
10

31
30

28

27

application of external load. Lewis et al. [18] and Sufian and


Tempelman [19] analyzed this problem using the dynamic relaxation method and minimum energy method, respectively. Table 4
shows a comparison of the vertical displacements obtained by the
present work with those predicted by the other authors using
different numerical methods. It can be seen that the proposed
element can accurately predict the behavior of cable with the
maximum difference of 2.78%.
Example 3. A spatial cable network of plan dimension
24 m  16 m consists of 38 pretensioned cable segments spaced
at 4 m  4 m grid as shown in Fig. 8. The structure has mirror
symmetry about both centerlines, and the z-coordinates for a
quarter of the structure are given in Table 5. The initial geometry of
the structure is achieved by means of the pretension force of 90 kN
in the x-direction and 30 kN in the y-direction. The structure
is subjected to a vertically concentrated load of 6.8 kN at all
internal nodes. The cross-sectional area of cables is 350 mm2 in
the x-direction and 120 mm2 in the y-direction, and the elastic
modulus of all cables is 160 kN/mm2. The displacements obtained
in present study are compared with those predicted by Lewis [20]
as shown in Table 5. It is observed that the results predicted by the
present study are very close to those given by Lewis.
Example 4. The saddle net shown in Fig. 9 consists of 142
pretensioned cable segments spaced at 5 m  5 m grid. The
structure has mirror symmetry about both centerlines, and the
z-coordinates for a quarter of the structure are given in Table 6. All
cable segments have the same pretension force of 60 kN. The
cross-sectional area and elastic modulus of all cables are 306 mm2
and 147 kN/mm2. The saddle net is subjected to the external loads
of 1 kN in the x- and z-directions at all the free nodes on one-half of
the net. They are nodes 11-15, 22-26, 33-37, 44-48, 55-59,
66-70 and 77-81. Table 6 shows the comparison of the nodal
displacements obtained by the proposed program and Kwan [21]. It
can be seen that the results obtained in present work are in good
agreement with those predicted by Kwan.

29

Fig. 8. Spatial net.

Table 5
Comparison of displacements (mm) of spatial net.
Node

1
2
3
6
7
8
9
13
14
15
16
b

z-coord.

1000.0
2000.0
3000.0
0
819.5
1409.6
1676.9
0
687.0
1147.8
1317.6

Lewis [20]

Present

dx

dy

dz

dx

dy

dz

5.14
2.26
0

0.42
0.47
2.27

30.41
17.70
3.62

5.03(2.14)b
2.23(1.33)
0

0.41(2.38)
0.46(2.13)
2.31( 1.76)

29.86(1.81)
17.29(2.32)
3.61(0.28)

4.98
2.55
0

0
0
0

43.49
44.47
41.65

4.92(1.20)
2.55(0.00)
0

0
0
0

42.85(1.47)
44.26(0.47)
42.08( 1.03)

Numbers in parentheses are the percentage error with respect to Lewis [20] values.

243

H.-T. Thai, S.-E. Kim / Finite Elements in Analysis and Design 47 (2011) 237246

identical results with the maximum difference of 0.24%. The effect


of cable self-weight on the dynamic response of the proposed
element is also investigated. Fig. 12 shows the displacement
responses of the cable subjected to El Centro earthquake for three
different cases of w0 (i.e. ignored self-weight, the result is
identical with that generated by SAP2000), w0.01 N/mm, and
w0.05 N/mm. It can be observed that the uniformly distributed
load has significant influence on the dynamic response of cable. The
elastic and inelastic responses of cable subjected to El Centro
earthquake are shown in Fig. 13. Since the proposed element
considers the self-weight and inelastic effects of cable, it is proved
to be more efficient than the cable element provided by SAP2000.
Example 6. A plane cable net subjected to both static and
dynamic loadings is shown in Fig. 14. The static behavior of this

5
4

15
14

13

2
12

26
25

24

E = 200 GPa

37
36

3m

48

59 70
81
69 80
91
57
22
90
68 79
45
10
33
56
.0m
89
21
67 78
44
50
=
32
55
.0
88
66
8x5
43
x5
77
.0 =
10
54
40.
87
0m 65 76

11

47

35

23

34

46

58

y = 220 MPa

M
A = 100 mm2
3m

structure was presented by Jayaraman and Knudson [11]. In this


research, both static and dynamic analyses are performed.
The lumped mass at each node is assumed to be 8.76 Ns2/mm
(50 lbs2/in). Additional data of the problem are given in Table 8. The
static load is applied first to the structure by a static analysis, and
then the earthquake loading is applied in the vertical direction of
the structure by a nonlinear timehistory analysis. The vertical
displacements of cable net under static load are compared in
Table 9. It can be seen that a strong agreement of displacements of
study cable net predicted by proposed program, SAP2000, and
Jayaraman and Knudson [11] is obtained. The displacement
responses of cable net under dynamic load are shown in Fig. 15.
It can be seen that all the results generated by proposed program
and SAP2000 are nearly the same, which prove the accuracy of the

Ground motion

M = 1.0 Ns2/mm
Lu = 4995 mm

4m

4m
Fig. 10. Inclined cable subjected to earthquake.

Fig. 9. Saddle net.

Table 6
Comparison of displacements (mm) of saddle net.
Node

1
2
3
4
5
11
12
13
14
15
22
23
24
25
26
33
34
35
36
37
44
45
46
47
48
52
72
81
85
c

z-coord.

1368
2432
3192
3648
3800
1032
1835
2408
2752
2867
792
1408
1848
2118
2200
648
1152
1512
1728
1800
600
1067
1400
1600
1667
600
1848
2867
1032

Kwan [21]

Present

dx

dy

dz

dx

dy

dz

15.55
11.50
7.38
5.34
4.11
14.43
11.27
7.25
5.67
4.77
11.71
9.55
6.30
4.92
4.65
10.63
8.80
5.83
4.64
4.55
0.92
3.85
4.11
5.40

4.46
5.55
4.20
3.11
2.80
3.53
4.47
2.97
2.12
0.60
1.71
2.11
1.15
0.23
0.52
0
0
0
0
0
0
0.78
2.80
1.87

81.70
61.22
33.31
17.88
11.16
97.14
72.90
31.98
10.54
11.34
92.44
66.94
20.21
14.05
35.79
88.73
62.83
13.99
22.52
45.89
5.86
30.12
11.16
32.17

15.55(0.00)c
11.5(0.00)
7.38(0.00)
5.34(0.00)
4.10(0.24)
14.42(0.07)
11.26(0.09)
7.25(0.00)
5.67(0.00)
4.77(0.00)
11.70(0.09)
9.54(0.10)
6.30(0.00)
4.91(0.20)
4.65(0.00)
10.62(0.09)
8.79(0.11)
5.83(0.00)
4.63(0.22)
4.54(0.22)
0.92(0.00)
3.85(0.00)
4.10(0.24)
5.40(0.00)

4.46(0.00)
5.54(0.18)
4.19(0.24)
3.11(0.00)
2.80(0.00)
3.53(0.00)
4.46(0.22)
2.97(0.00)
2.11(0.47)
0.60(0.00)
1.71(0.00)
2.11(0.00)
1.15(0.00)
0.23(0.00)
0.52(0.00)
0
0
0
0
0
0
0.78(0.00)
2.80(0.00)
1.87(0.00)

81.66(0.05)
61.18(0.07)
33.28(0.09)
17.87(0.06)
11.15(0.09)
97.10(0.04)
72.84(0.08)
31.94(0.13)
10.52(0.19)
11.34(0.00)
92.40(0.04)
66.89(0.07)
20.17(0.20)
14.06( 0.07)
35.77(0.06)
88.68(0.06)
62.77(0.10)
13.95(0.29)
22.52(0.00)
45.87(0.04)
5.86(0.00)
30.10(0.07)
11.15(0.09)
32.15(0.06)

Numbers in parentheses are the percentage error with respect to Kwan [21] values.

244

H.-T. Thai, S.-E. Kim / Finite Elements in Analysis and Design 47 (2011) 237246

60
Present
SAP2000

60
40
20
0
-20

10

15

20

25

30

-40
-60
Time (s)

-80

Horizonal displacement (mm)

Horizonal displacement (mm)

80

Present
SAP2000

40
20
0
0

10

15

20

25

30

35

40

-20
-40
-60

Time (s)

Fig. 11. Displacement responses of inclined cable. (a) El Centro earthquake and (b) Loma Prieta earthquake.

Table 7
Comparison of displacement response of the inclined cable in elastic analysis.
Earthquake type

Max/min

Displacement(mm)

SAP2000

Present

Error (%)

El Centro

Max

Horizontal
Vertical
Horizontal
Vertical
Horizontal
Vertical
Horizontal
Vertical

66.932
77.038
57.832
89.102
52.913
70.742
53.014
71.030

66.884
77.076
57.730
89.093
52.999
70.717
52.888
70.912

0.07
0.05
0.18
0.01
0.16
0.03
0.24
0.17

Min
Loma Prieta

Max
Min

Present, w = 0
Present, w = 0.01N/mm
Present, w = 0.05N/mm

80

Horizonal displacement (mm)

60
40
20
0
0

10

15

20

25

30

-20
-40
-60
-80
Time (s)
Fig. 12. Effect of distributed loading on response of inclined cable under El Centro.

proposed program in predicting the seismic response of cable


structure. Using the same personal computer configuration (AMD
Phenom II, 3.2 GHz) the analysis time of the proposed program and
SAP2000 of the plane cable net subjected to Loma Prieta earthquake, which is the problem having the longest analysis time, are
5 s and 708 s, respectively. This result demonstrates the high
computational efficiency of the proposed program.

6. Conclusion
An accurate and effective catenary cable element is presented for
the nonlinear analysis of cable structures subjected to static and
dynamic loadings. The explicit form of tangent stiffness matrix and
corresponding internal force vector are presented. With the present
element, each cable member in cable nets can be modeled by using

245

H.-T. Thai, S.-E. Kim / Finite Elements in Analysis and Design 47 (2011) 237246

80

Horizonal displacement (mm)

60

Elastic
Inelastic

40
20
0
5

10

15

20

25

30

-20
-40
-60
Time (s)

-80

Fig. 13. Effect of material nonlinearity on response of inclined cable under El Centro.

6
2

900

10

Present (inelastic)
Present (elastic)
SAP2000

30.48
m

f
P
3

5
f

f
7

30.48
m

m
.48
30

12

f
9
11
8

Vertical displacement (mm)

m
.48
30

m
.48
0
3

30.48
m

f = 9.144 m
Fig. 14. Plane cable net.

700

500

300

100
0

10

Table 8
Initial properties of plane cable.
Item

Data

20

25

30

700

146.45 mm
82,737 MPa
420 MPa
1.459 N/m
24.283 kN
23.687 kN
35.586 kN

Table 9
Comparison of displacements of plane cable under static loading.

Vertical displacement (mm)

Cross-sectional area
Elastic modulus
Yield stress
Cable self-weight
Prestressed force of horizontal members
Prestressed force of inclined members
Vertical load at all internal nodes

Researcher

15
Time (s)

600
500
400
Present (inelastic)
Present (elastic)
SAP2000

300

Displacements of node 4 (mm)

200
Jayaraman and Knudson [11]
SAP2000
Present

x-direction

y-direction

z-direction

39.62
40.28
40.13

40.20
40.28
40.13

446.32
448.88
446.50

only one element. The computer program developed for this research
is verified for accuracy and computational efficiency through several
numerical examples. The good results obtained in a short analysis
time prove that the proposed element can be effectively used in
predicting the nonlinear behavior of cable structures instead of using
the time-consuming commercial structural software.

10

15

20

25

30

35

40

Time (s)
Fig. 15. Displacement responses at node 4 of the plane cable: (a) El Centro and (b)
Loma Prieta earthquake.

Acknowledgement
This research has been supported by the Brain Korea 21 Project
of the Korea Research Foundation.

246

H.-T. Thai, S.-E. Kim / Finite Elements in Analysis and Design 47 (2011) 237246

References
[1] J. Argyris, D. Scharpf, Large deflection analysis of prestressed networks, Journal
of the Structural Division 98 (3) (1972) 633654.
[2] M. Gambhir, B. Batchelor, Finite element study of the free vibration of a 3-D
cable networks, International Journal of Solids and Structures 15 (2) (1979)
127136.
[3] H. Ozdemir, A finite element approach for cable problems, International Journal
of Solids and Structures 15 (5) (1979) 427437.
[4] H. Ernst, Der E-modul von seilen unter beruecksichtigung des durchhanges,
Der Bauingenieur 40 (2) (1965) 5255.
[5] J. Liew, N. Punniyakotty, N. Shanmugam, Limit-state analysis and design of
cable-tensioned structures, International Journal of Space Structures 16 (2)
(2001) 95110.
[6] J. Coyette, P. Guisset, Cable network analysis by a nonlinear programming
technique, Engineering Structures 10 (1) (1988) 4146.
[7] H. Ali, A. Abdel-Ghaffar, Modeling the nonlinear seismic behavior of cablestayed bridges with passive control bearings, Computers and Structures 54 (3)
(1995) 461492.
[8] Z.H. Chen, Y.J. Wu, Y. Yin, C. Shan, Formulation and application of multi-node
sliding cable element for the analysis of Suspen-Dome structures, Finite
Elements in Analysis and Design 46 (9) (2010) 743750.
[9] M. Gambhir, B. Batchelor, A finite element for 3-D prestressed cablenets,
International Journal for Numerical Methods in Engineering 11 (11) (1977)
16991718.
[10] W. OBrien, A. Francis, Cable movements under two-dimensional loads, Journal
of the Structural Division, ASCE 90 (3) (1964) 89123.

[11] H. Jayaraman, W. Knudson, A curved element for the analysis of cable


structures, Computers and Structures 14 (3-4) (1981) 325333.
[12] C. Wang, R. Wang, S. Dong, R. Qian, A new catenary cable element, International
Journal of Space Structures 18 (4) (2003) 269275.
[13] A. Andreu, L. Gil, P. Roca, A new deformable catenary element for the analysis of
cable net structures, Computers and Structures 84 (29-30) (2006) 18821890.
[14] Y.B. Yang, J.Y. Tsay, Geometric nonlinear analysis of cable structures with
a two-node cable element by generalized displacement control method,
International Journal of Structural Stability and Dynamics 7 (4) (2007)
571588.
[15] M. Such, J.R. Jimenez-Octavio, A. Carnicero, O. Lopez-Garcia, An approach based
on the catenary equation to deal with static analysis of three dimensional cable
structures, Engineering Structures 31 (9) (2009) 21622170.
[16] J. Michalos, C. Birnstiel, Movements of a cable due to changes in loading,
Journal of Structural Division ASCE 127 (1962) 267303.
[17] G. Tibert, Numerical analyses of cable roof structures: Royal Institute of
Technology, Dept. of Structural Engineering, 1998.
[18] W. Lewis, M. Jones, K. Rushton, Dynamic relaxation analysis of the non-linear
static response of pretensioned cable roofs, Computers and Structures 18 (6)
(1984) 989997.
[19] F. Sufian, A. Templeman, On the non-linear analysis of pre-tensioned cable net
structures, Structural Engineering 4 (2) (1992) 147158.
[20] W. Lewis, The efficiency of numerical methods for the analysis of prestressed
nets and pin-jointed frame structures, Computers and Structures 33 (3) (1989)
791800.
[21] A. Kwan, A new approach to geometric nonlinearity of cable structures,
Computers and Structures 67 (4) (1998) 243252.

Вам также может понравиться