3-D NONLINEAR SEISMIC BEHAVIOR O F CABLE-

STAYED BRIDGES
By Ahmed M. Abdel-Ghaffar,1 Member, ASCE and Aly S. Nazmy, 2
Associate Member, ASCE

ABSTRACT: The dynamic behavior of three-dimensional (3-D) long-span cable-

stayed bridges under seismic loadings is studied. The cases of synchronous and
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

nonsynchronous support motions due to seismic excitations of these flexible struc-

tures are considered; furthermore, effects of the nondispersive traveling seismic
wave on the bridge response are studied. Different sources of nonlinearity for such
bridges are included in the analysis. Nonlinearities can be due to: (1) Changes of
geometry of the whole bridge due to its large deformations, including changes in
the geometry of the cables due to tension changes (known as the sag effect); and
(2) axial force and bending moment interaction in the bridge tower as well as the
girder elements. A tangent stiffness iterative procedure is used in the analysis to
capture the nonlinear seismic response. Numerical examples are presented in which
a comparison is made between a linear earthquake-response analysis (based on the
utilization of the tangent stiffness matrix of the bridge at the dead-load deformed
state) and a nonlinear earthquake-response analysis using the step-by-step inte-
gration procedure. In these examples, two models having center (or effective) spans
of 1,100 ft (335.5 m) and of 2,200 ft (671 m) are studied; this range covers both
present and future designs. The study sheds some light on the salient features of
the seismic analysis and design of these long contemporary bridges.

INTRODUCTION

Cable-supported bridges, e.g., classic suspension bridges and contem-

porary cable-stayed bridges, are very important lifeline structures. Suspen-
sion bridges efficiently cover the range of 1,500-7,000 ft (457.5-2,100 m)
of center or effective span, while cable-stayed bridges cover the range of
500 ft (152.5 m) to 2,000 ft (610 m); see Figs. 1 and 2 (Gimising 1986; Nazmy
and Abdel-Ghaffar 1987). Cable-stayed bridges, in particular, are becoming
very popular in the United States, Japan, Europe, and the third-world
countries. Thus, it is essential that the pursuit of economical performance,
of design rationality, and of the confirmation of safety and durability proceed
parallel to one other. Fig. 2 shows some of the most famous steel and
concrete design cable-stayed bridges in the world, built in the last three
decades, along with their center-span lengths.
As of 1989, the total number of steel and concrete cable-stayed bridges
having central (or effective) span lengths of 800 ft (245 m) or longer is
estimated to be about 90 bridges worldwide.
In the United States, eight such bridges have been constructed, seven
bridges are under construction, and about eight bridges are in the design
and consideration processes. Due to the fact that these long-span flexible
structures lend themselves to a unique class of vibration problems due to
environmental factors, e.g., wind and earthquakes, it is essential to develop
realistic analysis procedures to predict their responses due to these loads.
'Prof., Civ. Engrg, Dept., Univ. of Southern California, Los Angeles, C A 90089-
0242.
2
Asst. Prof., Civ. Engrg. Dept., Ain Shames Univ., Cairo, Egypt.
Note. Discussion open until April 1, 1992. To extend the closing date one month,
a written request must be filed with the A S C E Manager of Journals. The manuscript
for this paper was submitted for review and possible publication on November 4,
1988. This paper is part of the Journal of Structural Engineering, Vol. 117, No. 11,
November, 1991. © A S C E , ISSN 0733-9445/91/0011-3456/$1.00 + $.15 per page.
Paper N o . 26387.

3456

J. Struct. Eng. 1991.117:3456-3476.

 Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

RANGE OF CENTER SPAN FOR CABLE-SUPPORTED BRIDGES

CABLE-STAYED BRIDGES SUSPENSION BRIDGES

imnii

en
~4

III
R.C.C.S

2000
so
3000 4000 5000 6000 7000 8000 9000 10000
CENTER SPAN IN FT.

J. Struct. Eng. 1991.117:3456-3476.

200 400 800 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 3200
CENTER SPAN IN M.

FIG. 1. Practical Range of Center or Effective Span for Cable-Supported Bridges; Each Vertical Line Represents Existing Bridge
 900 - 3000 -

800 -
2500 —
700 - STEEL CABLE-STAYED BRIDGES
<n -
2000 —
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

rr 600
LU -
^ UJ
In
2 LU
U.
500
1500
z z
7. z
< 400 D. <
u.
« W -
rr 300 - a .30
in LU
K i-
7
LU z
UJ
200
_
o o 500
100

000 I i I. . . . I ... . I
000
1950 1960 1970 1980 1990 2000

YEAR OF COMPLETION AND/OR CONSIDERATION

FIG. 2. Some of World Long-Span Cable-Stayed Bridges; Both Steel and Concrete
Design Bridges

In this study, a seismic response analysis program has been developed spe-
cifically for long-span cable-stayed bridges. The analysis takes into consid-
eration: (1) The static and dynamic nonlinear behavior of these bridges,
including both beam-column and cable members; and (2) multiple-support
as well as uniform seismic loadings of such bridges. Cost-effective compu-
tational procedures are demonstrated for a nonlinear dynamic analysis to
reduce the size of the problem, and consequently the computer time. Finally,
comments and recommendations are made concerning the seismic analysis
and design of cable-stayed bridges.

MODELS

Fig. 3 shows the models considered in the study; the side and center spans
of one of the models are half the length of the spans of the other one (Fig.
3). Model 1 has a center span of 1,100 ft (335.5 m) and side spans of 480
ft (146.4 m), while for model 2, these span lengths are 2,200 ft (671 m) for
the center span and 960 ft (292.8 m) for the side spans). Model 1 represents
the current trend, while model 2 represents the future trend in cable-stayed
bridge design. Two horizontal and vertical elastic links were provided at
the deck-tower connections (Fig. 3). For more structural details, see Nazmy
and Abdel-Ghaffar (1987).

DYNAMIC ANALYSIS

The equation of motion of the three-dimensional vibration of the bridge

(with N degrees of freedom) when subjected to uniform or nonuniform
3458

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE

3-D FINITE ELEMENT MODEL

Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

MODEL 2 - SPAN 2200 FT

^ 36 37 38 39 40 fi 42 43 44 45 46 47 46 72 73 74 75 76 7 7 ^ 79 80 81 82 83 p

• fti 0 9 I' i i—L-J-j—^ A 6 & $ 4 i O 4 O O 9 O—r«j—ft Q A • A 1*

'^ 12 13 14 15 16 ^ f 18 19 20 21 22 23 24 49 50 51 52 53 54 ^ 56 57 58 59 60 £7*

(a)

TOP STRUT

A\ DECK
CROSS - SECTION

TWO HORIZONTAL &

I VERTICAL ELASTIC LINKS |-«—

MODEL 1 MODEL 2

(b)

FIG. 3. One of Two Models (Model 2) Considered in Present Investigation: Deck

Nodal Points; (b) Tower Configuration

multiple-support seismic excitations at the two anchor piers and the two
tower bases (see Fig. 3), can be expressed in matrix form as (Abdel-Ghaffar
and Nazmy 1986a); Abdel-Ghaffar et al. 1983; Baron et al. 1976- Cloueh
and Penzien 1975)

K„
Mgs Mg: + + K„ . (1)
LK^
where the subscript g = the degrees of freedom corresponding to the points
of application and directions of ground motion; and the subscript s = all
other structural degrees of freedom of the bridge model. Thus, the matrices
M sg , C sg , and Ksg = rectangular mass, damping, and stiffness matrices,
respectively, which represent the coupling between the structure nodes not
3459

J. Struct. Eng. 1991.117:3456-3476.

 connected to the ground and the support displacements due to ground
motion.
The total nodal displacements may be decomposed into quasi- (or pseudo-)
static displacements and relative (or vibrational) displacements as follows:
(Abdel-Ghaffar et al. 1983; Baron et al. 1976; Clough and Penzien 1975)

(2
feH":M°o"} »
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

For the linear analysis, this equation can be written as

te}-1 fc}/iw+.?.{*}«•« (3
»
where gpsi = the rth quasi-static function that results from unit displacement
in the rth degree of freedom at a supporting point; f,{t), i = 1 , 2 , . . . , G
= the input displacement ground motions to the supporting points of the
bridge in the three orthogonal directions; gpgl = a G x 1 vector of which
the rth element is equal to unity, with all its other elements being zero; {<}>„}
= the «th vibration fixed-base mode shape; q„(t) = the nth generalized
coordinate; and P = the number of mode shapes used in the modal analysis
(in this study a total of 30 modes were considered). The eigenvalue problem
was solved based on the utilization of the tangent stiffness matrix of the
bridge in the dead-load deformed state (Abdel-Ghaffar and Nazmy 1986a,
1986b; Fleming and Egeseli 1980; Nazmy and Abdel-Ghaffar 1987), which
is obtained from the geometry of the bridge under gravity load conditions
as shown in Fig. 4. The first six computed three-dimensional (3-D) mode
shapes for the two models are shown in Figs. 5(a) and 5(b); it is evident
that a three-dimensional motion is associated with almost every mode of
vibration. Furthermore, these modes are closely spaced in terms of
frequencies.
As indicated previously, in the linear (based on the tangent stiffness and
modal superposition) or the nonlinear (based on a step-by-step time-inte-
gration procedure) dynamic analysis, the total nodal displacements may be
decomposed into: (1) Noninertial, quasi-static displacements; and (2) in-
ertial or relative (i.e., vibrational) displacements. The quasi-static displace-
ment is caused by the nonuniform or nonsynchronous motion of the sup-
porting points at any time. Fig. 6 shows some of the quasi-static functions
that result from unit displacements at the supporting points of model 1.
Such shapes of flexibility functions are multiplied by the earthquake ground
displacement at the corresponding supporting points of the bridge to provide
the quasi-static displacements.
A tangent stiffness, iterative procedure is used to capture the nonlinear
seismic response. The nonlinear equations of motion are solved using a step-
by-step integration technique. The structure is discretized in space into finite
elements, mainly beam-column elements and cable elements, and the Wil-
son-6 method (Nazmy and Abdel-Ghaffar 1987), with 8 = 1.4, is used for
the time discretization, to ensure numerical stability of the algorithm for all
time increments. Direct integration in the real displacement coordinate
space and, alternatively, integration in the modal coordinate space, using
the normal mode shapes as an orthogonal basis for the purpose of coordinate
transformation, are used to integrate the incremental equations of motion.
The latter approach takes less computation time than the former when the
3460

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE - MODEL 2 - SPAN
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

STATIC DEFORMATION DUE TO D . L.

(NONLINEAR ANALYSIS)

FIG. 4. Geometry of Bridge Model under Gravity Load Conditions

modes of vibration used for the coordinate transformation do not need to

be recomputed at the beginning of each time step, which is the case for
systems with mild nonlinearity. Finally, the modified Newton-Raphson tech-
nique was used to solve the equation of equilibrium at the end of each time
step. For more details on the nonlinear analysis and algorithm see Nazmy
and Abdel-Ghaffar (1987).

EARTHQUAKE-INPUT MOTIONS

Existing strong motion records can be used to define representative and

appropriately correlated multiple-support seismic inputs. Some of the ground-
motion records taken from the Imp^ial Valley, California, (El Centra)
earthquake (with local Richter magnitude of ML = 6.6) of October 15,
1979, Brady et al. (1980) are employed in this study to define the multiple-
input as well as uniform support motions (Abdel-Ghaffar and Nazmy 1987;
1986a, 1986b; Abdel-Ghaffar et al. 1983; Nazmy and Abdel-Ghaffar 1987).
The displacements of this recorded motion have a very strong component
at periods close to 3-5 sec; in addition, the ground accelerations are rich
in high-frequency components. Three cases of three-component earthquake
input motions were considered (see Fig. 7): (1) Multiple-support or non-
3461

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE - MODEL 1 - SPAN 1100 FT
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

MODE 1

MODE 2 f =0.3109 Hz
T =3.2161 sec

MODE 3 f =0.4105 Hz
T = 2.4362 sec

MODE 4
f =0.5154 Hz
T = 1.9401 sec

MODE 5 f =0.6498 Hz
T = 1.5390 sec

MODE 6
f = 0.6626 Hz
T = 1.5093 sec

f = 0.6994 Hz
T = 1.4298 sec

FIG. 5. (a). First Six Commuted 3-D Mode Shapes for Two Models: Model 1;

synchronous seismic inputs; (2) uniform seismic inputs; and (3) traveling
seismic waves with different speeds of propagation (covering the range 400,
800, 3,200, and 4,600 ft/sec) but with an invariant wave pattern (i.e., non-
dispersive or time-lag propagation).
3462

J. Struct. Eng. 1991.117:3456-3476.

 3-D ' CABLE - STAYED BRIDGE - MODEL 2 - SPAN 2200 FT
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

MODE 1

MODE 2

f =0.1924 Hz
T =5.1973 sec

MODE 3

f = 0.2520 Hz
T = 3.9689 sec

MODE 4

f = 0.3009 Hz
T = 3.3233 sec

MODE 5

f =0.3327 Hz
T =3.0060 sec

MODE 6

f = 0.4009 Hz
T = 2.4941 sec

f =0.4142 Hz
T = 2.4146 sec

FIG. 5. (b). First Six Commuted 3-D Mode Shapes for Two Modes: Model 2

RESPONSE CHARACTERISTICS

Linear and Nonlinear Dynamic Responses (Fig, 8)

The first 15 sec (which is the duration of the phase of strong shaking of
the recorded earthquake motion of array 6 of the 1979 Imperial Valley
3463

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE - MODEL 1 - SPAN 1100 FT

QUASI-STATIC FUNCTIONS
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

UNIT LONG DISPL

AT RIGHT ABUTMENT

UNIT VERTICAL DISPL

AT RIGHT ABUTMENT

UNIT LATERAL DISPL

AT RIGHT ABUTMENT

UNIT LONG DISPL

AT RIGHT TOWER

UNIT VERTICAL DISPL

AT RIGHT TOWER

UNIT LATERAL DISPL

AT RIGHT TOWER

FIG. 6. Pseudo- or Quasi-Static Functions Resulting unit Displacements at Sup-

porting Points

3464

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE

SUBJECTED TO THREE TYPES

OF
AY GROUND MOTION
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

1. MULTIPLE EARTHOUAKE
INPUTS
ARRAYS 4 , 5 . 6 , 7

WAVE SPEED 400 FT/SEC 2 . UNIFORM EARTHQUAKE

WAVE SPEED 800 FT/SEC INPUT
WAVE SPEED 1600 FT/SEC
ARRAY # 6
WAVE SPEED 3200 FT/SEC
WAVE SPEED 6400 FT/SEC
WAVE PROPAGATION CASE
ARRAY I 6

FIG. 7. Three Cases of Three-Component Earthquake Input Motions

earthquake) are used for the uniform or synchronous seismic inputs at the
two end abutments and the two tower bases of the bridge models. In all
the time-history response analyses, the damping ratio was assumed constant
and equal to 2% (Nazmy and Abdel-Ghaffar 1987) for all modes. Fig. 9
shows the locations of the response displacements and member forces con-
sidered in this study, while Figs. 10(A), 10(b) and 11 show a comparison
between the linear and nonlinear dynamic analysis results for these response
quantities. By examining Figs. 10(a) and 10(6) it is evident that there is not
much difference between the results of the linear- and nonlinear dynamic
analyses for the 1,100-ft center-span modal. Since the difference between
linear- and nonlinear dynamic analyses for model I is small, it is important
to examine the validity of using linear static analysis and to investigate the
necessity of performing a nonlinear analysis under dead loads to start the
dynamic analysis.
For this investigation, three types of analysis were performed. The first
is a linear static analysis followed by linear earthquake analysis; it is called
linear-linear (or L-L). The second analysis is a nonlinear static analysis
followed by a linear earthquake analysis (based on the utilization of the
tangent stiffness matrix of the bridge in the dead-load deformed state); it
is called nonlinear-linear (or NL-L). The third analysis is a nonlinear static
analysis followed by a nonlinear earthquake analysis; it is called nonlinear-
nonlinear (or NL-NL). Fig. 8 shows a qualitative sketch of the three types
of analysis.
Fig. 10(a) shows a comparison among the results obtained by the three
analysis methods. It is evident, by examining this figure, that although the
difference between the NL-L and the NL-NL analysis results is very small,
these results drifted from the L-L analysis results by a considerable amount
for most of the computed response quantities. Thus, although for the present
range of center spans (up to 1,400 ft) linear dynamic analysis is adequate
3465

J. Struct. Eng. 1991.117:3456-3476.

 Generalized l CABLE -STAYED BRIDGES
Force

Non-cable Structures
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

/ Cable Structures

Generalized
Displacement

Generalized .
Force

Dynamic Load

Dead Load

Generalized
Displacement

FIG. 8. Nonlinearities of Cable-Stayed Bridges

(Fleming and Egeseli 1980; Morris 1974), nonlinear static analysis under
dead loads is still essential to start the linear dynamic analysis.
For model 2, the results of the nonlinear response quantities are compared
with those obtained by linear dynamic modal analysis. Figs. 10(b), and 11
show this comparison. It is evident from these figures that the nonlinear
dynamic behavior of this long-span bridge model is more pronounced than
in the case of model 1. Furthermore, there is a frequency shift observed in
the response time-history; this is due to the fact that the overall stiffness of
the bridge increases by the increase in the dynamic displacements as well
as the forces. This result is consistent with the fact that the nonhnearity of
this type of structure is of the geometric-hardening type, which is mainly
due to large deformations and an increase of the center-span length. Thus,
3466

J. Struct. Eng. 1991.117:3456-3476.

Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

LOCATIONS OF CALCULATED RESPONSE QUANTITIES

FIG. 9. Locations of Calculated Response Quantities

a geometrically nonlinear dynamic analysis is necessary for computing the

response of long-span cable-stayed bridges subjected to strong ground shak-
ing. Again, the future trend in designing these structures to have a longer
center of effective span ( s 2,000 ft) makes such nonlinear analyses inevi-
table.
Since strong nonlinearity is observed in the response of model 2 (with a
long span) to uniform earthquake inputs, it is now essential to examine the
nonlinear behavior of the same model under nonsynchronous multiple-sup-
port seismic inputs. Records from arrays 4, 5, 6, and 7 of the 1979 Imperial
Valley earthquake are used for the input motion (with three orthogonal
components) at the left abutment, left tower base, right tower base, and
the right abutment, respectively (see Fig. 7).
Fig. 11 shows both the comparison between linear and nonlinear dynamic
analysis results, and a comparison of (or the difference between) the vi-
brational and total response displacements and member forces. It is evident
from this figure that there is strong nonlinear behavior in this long-span
model. Furthermore, nonuniform multiple-support seismic excitations tend
to considerably change (increase or discrease) the response, and thus it can
have a significant effect and should be considered in the earthquake-re-
sponse analysis of such long and complex three-dimensional structures. This
effect is especially important when the structural redundancy is high. For
statically determinate structures, phase differences in support motions would
not influence the nature of the dynamic response of the bridge to an earth-
3467

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE - MODEL 1 - SPAN 1100 FT
UNIFORM EARTHQUAKE INPUT

Y-DISPL OF JOINT 23
H 1 1-
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

TIME IN SECONDS

Y -SHEAR IN MEM 61 AT JOINT 23

3.60-
TOTAL
3.00-
o
—2.40-
X

1.80-

1.20- \ I '\
W
0.60-
/^^ P \ :'\ j\ I ;' \ ! A
Q. I \ { ! \ i '. / A ^
in
Z -0,60- '! \ ; I \'l': '\
J
ORCE

-1.20- I: j • / • •

-1.80-
LL LINEAR
-2.10-
NONLINEAR
1 1 —h- H 1 1 1 —
0.60 0.80 1.00

TIME IN SECONDS t X 10'

(a)

FIG. 10. (a). Some of Calculated Time-History, Earthquake-Response Quantities:

Model 1 with uniform earthquake motion

3468

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE - MODEL 2 - SPAN 2200 FT
UNIFORM EARTHQUAKE INPUT
,Y-DISPL OF JOINT 23
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

TIME IN SECONDS i x io>

MULTIPLE EARTHQUAKE INPUTS

Y-DISPL OF JOINT 23

TIME IN SECONDS i x io' >

(b)

FIG. 10. (b). Some of Time-History, Earthquake-Response Quantities: Model 2

with Uniform and Nonuniform Earthquake Motions

3469

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE - MODEL 2 - SPAN 2200
MULTIPLE EARTHQUAKE INPUTS
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

FIG, 11. Some of Calculated Time History of Generalized Force of Model 2 Due
to Nonuniform Multiple Earthquake Inputs

quake; for statically indeterminate or redundent structures, a noninertial or

kinematic motion will be induced (due to the nonuniform support motions)
and should be added to the vibration motion of the bridge.

Spatial Distribution of Maximum Displacement and Member Forces

For practical design purposes, it is important to provide the design en-
gineer with the absolute maximum values of response displacements and
member forces in the structure during the earthquake excitation. Figs. 12,
13(a), 13(b) show the distribution of some of these response quantities along
the bridge towers and deck for models 1 and 2 when subjected to nonuniform
or multiple-support seismic excitations. It should be emphasized that all the
earthquake-induced sectional forces and deformations illustrated in these
figures are absolute (maximum) spectral values, are completely reversible
in sign, and did not occur at the same time. The general observation from
all these figures is that the nonuniform or multiple-support seismic inputs
induce much higher (or lower) member forces and joint displacements than
those induced by the uniform input-motion. That is evident from the dif-
ference between the vibrational response and the total response (caused by
the quasi-static support-motion effects). It is also evident from these figures,
that earthquake loading conditions can induce deformations and forces of
considerable magnitude.

Effect of Seismic Wave Propagation

The three orthogonal components of array 6 of the El Centra 1979 earth-
quake are assumed to be traveling along the centerline of the bridge with
different propagation speeds (covering the range of 400, 800, 1,600, and
6,400 ft/sec) without changing the wave pattern (i.e., nondispersive waves),
see Fig. 7. Propagation of the earthquake wave train is simulated by taking
the three orthogonal components as the input at the left anchorage of the
3470

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE - MODEL 1 - SPAN 1100 FT
MULTIPLE EARTHQUAKE INPUTS
HB50L. HHX (-5HEHH IN THE TOHEFI
^fl8S0L. MAX Z-HOMENT IN THE TOWER
VI BFiRT I nNHL
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

B.H. IKIPHFT) S . F . IMPS)

3-D CABLE-STAYED BRIDGE - MODEL 2 - SPAN 2200 FT

RB50L. MRX T-HOMENT IN THE TOWER FlBSOL. MffX. X-DISPL 1IJ THE TOWER
BflRTIONftL --- VI6RHT I 0*iHL
— TOTfiL

B.H. (KIPKFTI
DISPL IN INCHES
I
FIU. 12. Tower Spatial Distribution of Maximum Response Quantities Due to Non-
uniform Earthquake Inputs

bridge and adding the appropriate time delays to the other inputs based on
the traveling distance and the propagation speed. The choice of this range
of propagation speeds can be rationalized as follows.
For model 1, with a center-span length of 1,100 ft, if the seismic wave-
length is assumed to be in the same order as the center-span length, and if
the wave has the same period as the fundamental mode (about 3.2 sec),
then the propagation speed of this wave is Cs = 1,100/3.2 = 420 ft/sec.
Therefore, the range of 400-6,400 ft/sec was chosen to represent a wide
band of possible seismic waves that can cause out-of-phase or differential
displacements in the bridge elements.
Fig. 14 shows the effect of seismic-wave propagation, in the time and
frequency domains, on the longitudinal shear force at the tower base for
model 2. It is obvious that at low wave speeds, out-of-phase motion occurs
at the support of the bridge, which induces high values of quasi-static mem-
ber forces. At high wave speeds, the motion approaches the uniform input
case, and the quasi-static member forces are almost zero (see Fig. 14 for
the case of wave speed = 6,400 ft/sec).
Finally, a summary of the bridge seismic behavior is shown in Fig. 15,
which depicts the Fourier transform of the response time histories of the
axial force in cable 7 (Fig. 9). Both bridge models and the two earthquake-
input cases were considered. The following response characteristics are ev-
ident from Fig. 15:

1. There is a multi-modal contribution from several modes of bridge vibration

to the total response.
3471

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE - MODEL 1 - SPAN 1100 FT
MULTIPLE EARTHQUAKE INPUTS
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

ELEVATION

RBSOL. MRX. TOR MOMENT IN THE DECK

VIBRATIONAL
TOTAL

RBSOL. MRX. Z-SHERR IN THE DECK

VIBRATIONAL
TOTAL

RBSOL. MAX. T-.MOMENT IN THE DECK

VIBRATIONAL
TOTAL
Q-

FIG. 13. (a). Deck Spatial Distribution of Maximum Response Quantities Due to
Nonuniform Earthquake Inputs: Model 1

2. The kinematic or noninertial effect of the quasi-static displacements (in-

duced by nonuniform ground motion) is more pronounced in the low-frequency
or long-period range of the response.
3. For linear or nonlinear earthquake response, the nonsynchronous multiple-
support motions have a significant effect on the bridge response, as indicated
by the difference between the total and vibrational response values.
CONCLUSIONS
Based on this investigation of the earthquake-response characteristics of
cable-stayed bridge models under uniform and nonuniform multiple-support
seismic excitations, the following conclusions and remarks can be made:
3472

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAYED BRIDGE - MODEL 2 - SPAN 2200 FT
MULTIPLE EARTHQUAKE INPUTS.
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

RBSOL. MRX. RXIRL FORCE IN THE DECK

VIBRATIONAL
TOTRL
O

RBSOL. MRX. T-SHERR IN THE DECK

VIBRATIONAL
TOTAL

RBSOL. MPX. T-DISPL IN THE DECK

VIBRATIONAL
TOTAL

FIG. 13. (b). Deck Spatial Distribution of Maximum Response Quantities Due to
Nonuniform Earthquake Inputs: Model 2

1. For these typically three-dimensional structures, there is strong coupling

in the three orthogonal directions within each mode of vibration. Therefore, a
two-dimensional dynamic analysis is not adequate for this type of structure.
2. Generally, there is a multimodal contribution from several modes of vi-
bration to the total response of the structure for both displacements and member
forces.
3. Nonsynchronous multiple-support seismic excitations can have a significant
effect and should be considered in the earthquake-response analysis of such
long and complex three-dimensional structures. Spatially varying input motions

3473

J. Struct. Eng. 1991.117:3456-3476.

 3-D CABLE-STAVED BRIDGE - MODEL 2 - SPAN 2200 FT (TRAVELLING WAVE EFFECT)

Z-SHEAR FORCE AT LEFT TOWER BASE

WAVE S P E E D J 0 0 T O S E C _ ^ _ _ _WAVE_SPEED MOFT/SEC

Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

WAVE SPEED 1600 FT/SEC WAVE SPEED 6400 FT/SEC

FREO (CPS) FREQ (CPS)

FIG. 14. Time-History and Frequency-Domain Response of Tower-Base Shear of

Model 2 at Different Speeds of the Traveling Inputs

resulted in a much more severe response; stresses induced by the kinematic

effect of ground motions augmented those induced by the vibrational effect.
4. Depending on the dynamic properties of the local soils at the supporting
points, as well as the soils at the surrounding bridge site, the traveling seismic
wave effect should be considered in the seismic analysis of these bridges.
5. Geometrical, as well as general nonlinear dynamic analysis, is necessary
for computing the response of long-span cable-stayed bridges subjected to strong
ground shaking. The future trend in constructing these structures with a longer
center span makes the need for such nonlinear analysis inevitable; this is essential
not only for evaluating the stress and deformation induced by environmental
loads, e.g., wind and earthquakes, but also for ensuring safety during construc-
tion.
6. Although for the present range of center spans [up to 1,400 ft (450 m)],
linear dynamic analysis is adequate [this was concluded by Fleming and Egeseli
(1980)], nonlinear static analysis under dead load is still essential to start the
linear dynamic analysis from the dead-load deformed state.
3474

J. Struct. Eng. 1991.117:3456-3476.

 3-D ,CABLE - STAYED BRIDGE - MODEL 1 - SPAN 1100 FT
MULTIPLE EARTHQUAKE INPUTS

LINEAR ' " s

Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

UNIFORM EARTHQUAKE INPUT

:
NONLINEAR
o -

LU A F.T. OF THE •
l\ AXIAL FORCE •
* - I IN CABLE 7 -

cc
UJ
cc
D
o
1

3-D CABLE-STAYED
I
1-
V L.
\
VIBRATIONAL .
TOTAL

BRIDGE - MODEL 2 - SPAN 2200 FT

MULTIPLE EARTHQUAKE INPUTS
:;
o - NONLINEAR
LU '
W A,
* ° 9
a. „
F-T. OF THE
* -'
AXIAL FORCE
i
cc
IN CABLE 7

cc
VIBRATIONAL
o TOTAL
Li.

FREQ (CPS)
FIG. 15. Fourier Spectra of Earthquake-Induced Cable Axial Force of Two Models,
Showing Multi-Modal Contribution, Noninertial, or Kinematic Effect of Nonuniform
Ground Motions and Nonlinearity Effect

3475

J. Struct. Eng. 1991.117:3456-3476.

 ACKNOWLEDGMENTS

This research is supported by the National Science Foundation through

Grants No. ECE-8501067 and N o . CES-8717252 with Dr. S.C. Liu as the
Program Director. This support is greatly appreciated. The paper is based
on the Proceedings of the 4th Structures Congress, entitled "Effects of
Three-Dimensionality and Nonlinearity on the Dynamic and Seismic Be-
havior of Cable-Stayed Bridges."
Downloaded from ascelibrary.org by Drexel University on 06/06/13. Copyright ASCE. For personal use only; all rights reserved.

APPENDIX I. CONVERSION TO SI UNITS

To convert To Multiply by

ft m 0.305

APPENDIX II. REFERENCES

Abdel-Ghaffar, A. M., and Nazmy, A. S. (1986a). "Earthquake resistant analysis

of cable-stayed bridges in eastern and central United States." EERI Proc, 3rd
U.S. Nat. Conf. on Earthquake Engrg.
Abdel-Ghaffar, A. M., and Nazmy, A. S. (1986b). "Seismic behavior of long span
cable-stayed bridges." Proc. 3rd ASCE Engrg. Mechanics Specialty Conf. on Dy-
namic Response of Structures.
Abdel-Ghaffar, M , and Nazmy, A. S. (1987). "Effects of three-dimensionality and
nonlinearity on the dynamic and seismic behavior of cable-stayed bridges." Proc.
4th Structures Cong. 1987; Volume on bridges and transmission line structures,
ASCE, New York, N.Y.
Abdel-Ghaffar, A. M., Scanlan, R. H., and Rubin, L. I. (1983). "Earthquake re-
sponse of long-span suspension bridges." Report No. 83-SM-13, Dept. of Civil
Engineering, Princeton University, Princeton, N.J.
Baron, F., Arikan, M., and Hamati, E. (1976). "The effects of seismic disturbances
on the Golden Gate Bridge." Report No. EERL 76-31, University of California,
Berkeley, Calif.
Baron, F., and Venkatesan, M. S. (1971). "Nonlinear analysis of cable and truss
structure." Struct. Div., ASCE, 97(2), 679-710.
Brady, A. G., Perez, V., andMork, P. N. (1980). "The Imperial Valley earthquake,
October 15, 1979: Digitization and processing of acceleration records." Open-File
Report 80-703, U.S. Geological Survey, Seismic Engineering Branch, Menlo Park,
Calif.
Clough, R. W., and Penzien, J. (1975). Dynamics of structures. McGraw-Hill, New
York, N.Y.
Fleming, J. F., and Egeseli, E. A. (1980). "Dynamic behavior of a cable-stayed
bridge." Int. J. Earthquake Engrg. Struct. Dyn., 8(1), 1-16.
Gimising, N. J. (1986). "Recent and future developments of cable-stayed bridges."
Report No. 214, Series R, Dept. of Struct. Engrg., Technical Univ. of Denmark,
Copenhagen, Denmark.
Morris, N. F. (1974). "Dynamic analysis of cable-stiffened structures." /. Struct.
Div., ASCE, 100(5), 971-981.
Nazmy, A. S., and Abdel-Ghaffar, A. M. (1987). "Seismic response analysis of
cable-stayed bridges subjected to uniform and multiple-support excitations." Re-
port No. 87-SM-l, Dept. of Civil Engineering, Princeton University, Princeton,
N.J.

3476

J. Struct. Eng. 1991.117:3456-3476.