Pylons of Cable-Stayed Bridges A Comparison

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Pylons of cable-stayed bridges: a comparison

FAKHRY
ABOUL-ELLA
College of Architecture and Planning, Department of Building Science and Technology, King Faisal University,
Dammam P.O. Box 2397, Saudi Arabia
Received May 6 , 1987
Revised manuscript accepted January 25, 1988

A new analytical model for the static analysis of cable-stayed bridges is presented taking into consideration the effect of
soil-structure interaction. Bridges having pylons and piers of different degrees of restraint, such as fixed or hinged base
pylons, fixed base piers, and pile-supported piers, are analyzed and compared to investigate the effect of these restraints on the
response of the entire cable-stayed bridge system. The validity of the assumption of fixed support for pylons or piers is also
examined.
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by San Diego (UCSD) on 01/11/15

Key words: bridges, cables, foundation, towers, pylons, piers, piles, caissons, soil mechanics.

Un nouveau modkle analytique pour le calcul statique des ponts ii cdbles inclines est present6 en tenant compte de l'interac-
tion structure-sol. Des ponts dont les pylbnes et les piliers posskdent des niveaux d'encastrement varies, par exemple des
pylbnes a base fixe ou articulke, des piliers i base fixe et des piliers sur piles, sont Ctudits et compar6s afin d'expliquer l'inci-
dence des encastrements sur le comportement global d'un pont ii cdbles inclines. La validit6 de la thtorie de l'appui fixe pour
les pylbnes ou les piliers est egalement examinee.
Mots clCs : cibles, fondation, tours, pylbnes, piliers, piles, caissons, m6canique de sols.
[Traduit par la revue]
Can. I. Civ. Eng. 15,516-523 (1988)

Introduction tions of pylons and piers has been carried out from which
Cable-stayed bridges are flexible structures whose responses general conclusions and recommendations have been drawn.
are affected by the degree of tower restraint. The pylon can be Analytical model
For personal use only.

of various types and may have hinged support or it can be


The structural system (Fig. l a ) consists mainly of caissons
rigidly connected at the base to a pier. A pylon with its base
(or piles), piers, stiffening girder, pylon, and cables. In the
rigidly connected to the pier may be more practical from an
well-known nonlinear stiffness method presented by Saafan
erection point of view and may be less costly than a heavy
(1963) and others, the member end forces are defined in terms
pinned bearing as stated in the Commentary on the tentative
of the stiffness matrices and end displacements of each mem-
recommendations for cable-stayed bridge structures (Task Com-
ber, then the structural stiffness matrix, which relates forces
mittee on Cable-suspended Structures 1977b). In the analysis
and displacements at all joints, is assembled. Nonlinearity due
of cable-stayed bridges, many authors (for example, Gimsing
to change of geometry of the bridge system, change of axial
1983; Khalil et al. 1983; Kavanagh 1977; Podolny and Scalzi
stiffness of cables, and change of bending stiffnesses of girder
1976) assume that the pylon has fixed support at its base and
and pylon caused by axial forces can be taken into account.
neglect the deformations of the pier. Actually the pier and the
The stiffness method can analyze any system with any cable
pylon are forming a tower which is a free-standing element and
configuration, but it usually needs huge computer memory.
is often supported by piles or caissons. The tower is more
The new analytical model presented in this paper is linear
flexible than the pylon alone and therefore it would have a
and assumes that the continuous girder is supported on the
considerable effect on both the internal forces and the resDonse
piers and passes through the pylon legs (i.e., there is no moment
of the entire bridge. transfer between the pylons and the girder) and the cables are
The effects of soil-structure interaction (SSI) on free-standing
fixed to the pylons. The model is based on the theorem of
towers have been studied by Novak (1974), on buildings by
virtual work in which a base system (statically determined) is
Aboul-ella (1984) and many others. The effects that SSI and
assumed and then be corrected. The base system of the girder
tower flexibility may have on cable-stayed bridges still need to
is shown in Fig. 1b in which each simple beam k is subjected
be explored, which is one of the objectives of the present
to the external distributed and concentrated loads plus concen-
paper.-~hepaper also presents a new iterative procedure based
trated loads Cifrom the supporting cables. According to the
on the potential energy of the bridge girder and utilizes the
energy concept, the deflection yo at any point distance x from
method of the three moment equations. This makes it possible
the left end of beam k can be calculated using Fourier coeffi-
to analyze and to perform parametric studies on largebridges
cients as follows:
using- small computers or PCs because the method does not 30
require large computer memory. The new procedure also in-
cludes tower flexibility and SSI and compares very well with [l] yo = 1 ak. sin j-
j=1 / Lk
~ x

the well-known nonlinear stiffness method.


Working example using the new procedure has been given where ak, is the Fourier coefficient for beam k of mode j given
and a comparison between bridges having different end condi- by Hegab (1986) as

(9)* I "I

NOTE:Written discussion of this paper is welcomed and will be


j r x o - cos !YE - 1 wisin
received by the Editor until December 31, 1988 (address inside front [2] akj= [COS LI: i=l Lk
cover). EL(j~lLk)~(Lk/2)
Printed in Canada: Imprim6 au Canada
ABOUL-ELLA

Support !_1
I
2 1 1 lm-1
I II
Lk
I l l
m+l I
cl

X.
(b) -- ' C1=Fi s i n Qi

Cable force
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by San Diego (UCSD) on 01/11/15
For personal use only.

FIG. 1. Typical multi-span cable-stayed bridges: (a) system; (6) deflections of the base system due to live loads and cable forces; ( c )deflec-
tions due to unit end moments.

in which Wi = Ci or -Pi; Ci = vertical component of the i, and Li = length of cable i. Deflection Ai can be calculated
cable force Fi, i.e., Ci = FisinOi (see Fig. lb); Pi = external from the flexibility matrix of the tower which include the foun-
concentrated live load; Elk = bending stiffness of beam k; p = dation flexibilities. This matrix can be obtained by applying a
intensity of the distributed live load extending from xo to x,; unit horizontal force at each joint at a time and calculating the
Lk = span length of beam k; Oi = slope angle of cable i; and horizontal deflections at all joints. The deflection St, of joint t
n l = number of cables and concentrated live loads. on the tower due to unit horizontal load acting at joint s (Fig.
The slope angles, B 1 and B2, at the two ends of each beam 2b) is of the form:
(Fig. 1b) can be determined and a set of three moment equa-
tions are established. These equations, one at each intermedi-
ate support, are solved for each mode j to obtain the connect- in which xf and Of = horizontal displacement and rotation of
ing moments Mmj, where m is the support number. The final the pier's base due to unit load at s and u,, = deflection of fixed
deflection at distance x from the left support of any beam k is base tower. xf and Of can be calculated as follows:
k$$ - hs kf,fI
[61 Xf =
(kflflk$$ - ku+ k$II)
where y l l and yz2 are the deflections at distance x due to unit - ku+ + hs kuu
r71 of = (kuuk$+ - ku$k$u)
moments at the left and the right supports respectively of beam
k (Fig. lc). The final value of the cable force Fi can be esti- in which kuu = caisson horizontal stiffness at head, k++ =
mated from the final deflection yi and from the tower deflection caisson rocking stiffness, and ku9 = caisson cross stiffness.
Ai (Fig. 2a) as follows: Static stiffnesses of piles and caissons can be obtained from
EOAi Poulos and Davis (1980) and Poulos (1979) or from the real
[4] Fi = 7 (yi sin Oi + Ai cos Oi) parts of the dynamic stiffnesses, at very low frequency, given
by Novak and El Sharnoby (1983) and Novak and Aboul-ella
in which Eo = Young's modulus of cables, Ai = area of cable (1978).
518 CAN. J. CIV. ENG. VOL. 15, 1988

Positive O
-
(0)
A
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by San Diego (UCSD) on 01/11/15

(b)
Stiff cower = Convergence

f l c s l b l e tower =

Tou'er=P~,lon dlvergcnrc.

rigidly
c o n n e c t ~ dt o c A
a pier
For personal use only.

FIG. 2. Deflections: (a) cable and ( b ) tower.


FIG.3. Definitions and presentation of the new procedure: (a)
Iterative procedure tower and bridge stiffnesses; ( b ) convergence and divergence accord-
The previous analysis assumes that cable forces are known. ing to Hegab 1986; ( c ) idea of the new procedure.
Actually they are not and they depend on the final deflections
of the girder ([3]). Therefore an iterative procedure must be obtains directly the required deflection A, at which Qt and Q,
used to have a solution. Hegab (1986) suggests an iterative are equal. At a certain value of A, the calculation of deflections
procedure in which the value of the tower sway A is improved yi and forces 6 needs further iteration, during which A is kept
during iteration cycles. This procedure does not converge un- constant.
less the tower possesses sufficient stiffness as will be described The iterative procedure for obtaining Q, with fixed value of
later in this section. In this paper a new technique is presented A is simple and efficient. Nine or ten Fourier terms are suffi-
which converges for any tower having any degree of flexibil- cient to have a difference in yi less than 0.5% at any point and
ity. The new technique defines two stiffnesses at each cable- seven or eight cycles are sufficient to have a difference in Fi
tower intersection, the first is the stiffness Kt of the tower alone less than 1% in any cable. The iterative procedure is summa-
and the second is the stiffness Kc of the bridge without the rized in the following steps:
tower. The two stiffnesses and their corresponding forces Q, 1. Assign a certain fixed value of A (A will not be changed
and Q, are shown for a radiating system in Fig. 3a. It was during iteration cycles) and set 6 = 0.
found (Fig. 3b) that Q, varies linearly with A . Actually Q, is 2. Determine Fourier coefficients ak, according to 6.
equal to the resultant of the horizontal components of cable 3. Determine girder deflections yi.
forces, assuming that the top ends of cables have given a 4. Add more Fourier terms (i.e., add more modes) until
certain deflection A without the tower being there. Now the convergence takes place.
idea of Hegab (1986) and the new technique can be discussed. 5. Find F, (positive for tension) from yi and A.
Figure 3b shows the variation of Q, and Q, with A for stiff and 6. Check convergence; if yes, calculate
flexible towers. It is clear from Fig. 3b that the iterative proce- 11 1

dure suggested by Hegab (1986) converges only for stiff towers Qc = I =1 -Fi cos O i
i.e., when K, is greater than -Kc.
The new technique calculates the stiffnesses Kt and Kc and if no, go back to step 2.
ABOUL-ELLA
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by San Diego (UCSD) on 01/11/15

FIG.4. Examples: ( a ) double tower radiating system (SSI is included); (b) comparison with the nonlinear stiffness method.
For personal use only.

This procedure is applicable to most systems, but for bridges The above stiffnesses and forces are calculated using the bridge
that have stiff cable systems, it would be more efficient if the properties given in the next section and are found to be
forces F, (at the beginning of an iteration cycle) can be taken
equal to the average of fi at the beginning (step 2) and F, at Kt,I= Kt,, = 531kN/m
the end (step 6) of the previous cycle. and
If A is taken equal to unity, then the stiffness Kc is equal to
the difference (Q, - Q,), where Q l is Q, at A = 0. For a Kt,, = Kt21= 0
single-tower radiating system, Kt and Kc will have single val- KtIIis the force that makes the top deflection of the tower alone
ues, but in cases of double-tower radiating systems or harp equal unity.
systems, Kt and Kc are stiffness matrices. From the loaded bridge (without the tower):
The required sway A, of the tower is the sway at which Q, is
equal to Q,, i.e., at the intersection of lines Qt and Q, (see (1) for A1 = A2 = 0 ,
Fig. 3c) and is given by Q l l = 301kN and Q I 2 = -301kN
(2) for A1 = 1 and A2 = 0 ,
in which Q1 is Q, at A = 0.
KcI1= -3190 kN/m and = -302 kN/m
The final procedure can be outlined in the following steps:
1. perfor& the analysis assuming A = 0 to obtainthe hori- (3) for A, = 1 and A, = 0,
zontal force (or forces) Q l ;
2. Do the same as in step 1 but with A = 1 to obtain the KC22= -3 190 kN/m and KC],= -302 kN/m
stiffness Kc = Qc - Q l (for a harp system, step 2 is repeated n
times to obtain an n x n stiffness matrix, where n is the
number of cable-tower intersections).
3. Solve [8] to obtain A,.
Therefore [9] becomes

[lo1
[ A , ] - [3721
A2
302
3021-'
3721
I!:-[
4. Perform the analysis again with a fixed value of A equal
to A, to find the bridge response and internal forces. which gives A1 = 0.088 m and A2 = -0.088 m.
In the above analysis a 2 x 2 matrix is solved, while in the
Worked example nonlinear stiffness method, at least a 57 x 15 matrix has to be
To explain the procedure, the double-tower radiating system solved many times. The present approach gives exactly the
shown in Fig. 4a is analyzed in details. The bridge has two same results as the linear stiffness method and compares very
cable-tower intersections, one at the top of each tower, at well with the nonlinear stiffness analysis as shown in Fig. 4b.
which [8] is given as This indicates that nonlinearity due to axial load, change of
geometry, and change in cable axial stiffness has a small effect
on the bridge's response and also shows the efficiency of the
present approach.
CAN. J . CIV. ENG. VOL. 15. 1988

(a)
System 1 2 3 4

L-J
P v l o n of Pvlon o f Tower o f Calsson-supported
pinned base f'ixed b a s e rixed base Lowei- (SSI)
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by San Diego (UCSD) on 01/11/15

Moments
For personal use only.

FIG. 5. (a) Systems of pylons and towers used for comparisons; (b) effect of pier flexibility and SSI on the tower of the system shown in
Fig. 4.

TABLE1. Caisson stiffnesses*

Soil G (MPa) KliLl( W I m ) K,,, (kN) K++( W . m)


-

(i) Firm 400 0.105 x lo7 -0.372 x lo7 0.394 x 10'


(ii) Moderate strength 80 0.363 X lo6 -0.226 x lo7 0.235 x 10'

Comparisons The following properties are assumed:


Four different systems of pylons and piers are considered in 1. Axial stiffness of all cables is EA/L = 15 000 kN/m.
the following comparisons. These systems are (Fig. 5a) 2. Girders and pylons are assumed to be steel elements
system 1: pylon of pinned base; having bending stiffness E I = 4 x 109kN.m2. The girder is
system 2: pylon of fixed base; continuous and is supported on many supports; one is hinged
system 3 : tower of fixed base; and the others are roller (movable) bearings. This leads to a
system 4: caisson-supported tower (i.e., SSI is included). self-anchored system applied in almost every cable-stayed bridge
The comparison between fixed base pylon (system 2) and built up till present (Gimsing 1983). The girder will therefore
caisson-supported tower (system 4) is actually the comparison be entirely in compression which can be resisted, in small and
between the usual analysis assumed by many authors and the moderate spans, with a small increment of the plate thickness
refined analysis that includes SSI. near the pylons (Gilsanz and Biggs 1983) avoiding the cost of
Three cable-stayed bridges similar to the real bridges pre- the anchorages.
sented in Change and Edward (1981) and The Sub-committee 3. The pier is assumed to be a concrete element with EI = 4
on Cable-Stayed Bridges (1977) have been chosen for these x 109kN-m2.
comparisons. These bridges are (1) double-tower radiating sys- 4. The pier is supported by a caisson whose stiffnesses at
tem having pylon and pier of height equal to 50 m; (2) single- the head are given in Table 1, in which G is the average shear
tower radiating system having pylon and pier of height equal to modulus of the soil surrounding the caisson.
75 m; and (3) single-tower harp system having pylon and pier 5. Live load is assumed to be a uniformly distributed load
of height equal to 75 m. of 40kN/m. This load is estimated according to the recom-
ABOUL-ELLA
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by San Diego (UCSD) on 01/11/15

O X - d ?' $ 1'2 ;G O
; ;/I
For personal use only.

FIG.6 . Deflections and bending moments of pylons and piers of single-tower radiating system.

FIG.7. Deflections and bending moments of pylons and piers of single-tower harp system.
522 CAN. J . CIV. ENG , VOL. 15, 1988

TABLE
2. Single-tower cable-stayed bridges pier flexibility and soil structure interaction have to be consid-
ered in the analysis of such important structures.
M , x 104
System y , (cm) (kN.m) T , (kN) 7.2 (kN) ABOUL-ELLA, F. 1984. Dynamic behavior of pile-supported struc-
tures with reference to Jubail soil. The Arabian Journal of Science
and Engineering, 9(3): 209-220.
(a) Radiating system h = 75 m F. K., and EDWARD,
CHANGE, C. 1981. Long-span bridges, state
55.1 24.8 1410.0 of the art. ASCE Journal of the Structural Division, 107(ST7):
48.2 21.8 1729.0 1145-1 159.
53.9 24.3 1469.0 GILSANZ, E., and BIGGS,M. 1983. Cable-stayed bridges: degrees of
54.4 24.5 1445.0 anchoring. ASCE Journal of Structural Engineering, 109(1):200-220.
GIMSING, N. J. 1983. Cable-supported bridges: concept and design.
(b) Harp system h = 75 m John Wiley and Sons Ltd., New York, NY.
HEGAB,H. I. A. 1986. Energy analysis of cable-stayed bridges.
65.0 29.3 1427.0 ASCE Journal of Structural Engineering, 112(5): 1182-1 195.
57.7 26.2 1939.0
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by San Diego (UCSD) on 01/11/15

KAVANAGH, T. C. 1972. Cable-supported bridges. Structural steel


66.2 29.8 1395.0 designers handbook. McGraw-Hill Book Co., Inc., New York,
67.0 30.1 1347.0 NY.
KHALIL, M. S., DILLGER, W. H., and GHALI, A. 1983. Time-depen-
*Definition of terms are shown in Figs. 6 and 7
+Negative indicates compression. dent analysis of PC cable-stayed bridges. ASCE Journal of Struc-
tural Engineering, 109(8): 1980- 1996.
NOVAK, M. 1974. Effect of soil on structural response to wind and
mended design loads for bridges (The Committee on Loads earthquake. International Journal of Earthquake Engineering and
and Forces on Bridges 1981), assuming four lanes. The deflec- Structural Dynamics, 3(1): 79-96.
tions under this load are found to be less than the specified NOVAK, M., and ABOUL-ELLA, F. 1978. Impedance functions of piles
limits given by the tentative recommendations for cable-stayed in layered media. ASCE Journal of the Engineering Mechanics
Division, 104(EM3): 643-661.
bridge structures (Task Committee on Cable-suspended Struc-
NOVAK,M., and EL SHARNOBY, B. 1983. Stiffness constants of
tures 1 9 7 7 ~ ) . single piles. ASCE Journal of Geotechnical Engineering, 109(7):
961-974.
Typical results PODOLNY, J. R., and SCALZI, J. B. 1976. Construction and design of
For personal use only.

The results obtained are presented in Figs. 5b, 6, 7, and cable-stayed bridges. John Wiley and Sons Inc., New York, NY.
Table 2. Comparing these results, a number of observations POULOS,H. G. 1979. Group factors for pile-deflection estimation.
can be made: ASCE Journal of the Geotechnical Engineering Division, lOS(GT12):
1. Figure 5b shows the effect of soil on the response and 1489-1509.
internal forces of the tower shown in Fig. 4. Two kinds of POULOS, H. G., and DAVIS,E. H. 1980. Pile foundation analysis and
design. John Wiley and Sons, New York, Chichester, Brisbane,
sandy soils are used; soil (i) is firm and soil (ii) has moderate
Toronto.
strength. SSI decreases considerably the bending moments of SAAFAN, S. A. 1963. Nonlinear behavior of structural plane frames.
the tower and increases its top deflection. These changes, of ASCE Journal of the Structural Division, 89(ST4): 557-579.
course, depend on the soil strength. Also, SSI has a negligible TASKCOMMITTEE ON CABLE-SUSPENDED STRUCTURES. 1977a. Ten-
effect on the girder response as can be seen from Table 2. tative recommendations for cable-stayed bridge structures. ASCE
2. The top deflection of the pylon having pinned bearing Journal of the Structural Division, 103(ST5): 929-939.
(system 1) does not differ much from that of caisson-supported 1977b. Commentary on the tentative recommendations for
tower (system 4). Although the bending moment of the pinned cable-stayed bridge structures. ASCE Journal of the Structural Divi-
pylon is nil or very small, system 4 is recommended from sion, 103(ST5): 941-959.
rigidity and erection points of view. THECOMMITTEE O N LOADS AND FORCES ON BRIDGES OF THE COM-
MITTEE ON BRIDGES OF THE STRUCTURAL DIVISION. 1981. Recom-
3. The assumption of fixed-base pylons underestimates both
mended design loads for bridges. ASCE Journal of the Structural
the deflections of the pylon (Figs. 6 and 7) and the deflections Division, 107(ST7): 1161- 1213.
of the girder (Table 2). Under this assumption some of the THESUB-COMMITTEE ON CABLE-STAYED BRIDGES OF THE COMMIT-
tension forces in cables are underestimated (Table 2) and there- TEE ON LONG-SPAN STEELBRIDGES OF THE COMMITTEE ON MET-
fore it can lead to unsafe design. On the other hand, it greatly ALS.1977. Bibliography and data on cable-stayed bridges. ASCE
overestimates the bending moments of the pylons (Figs. 6 and Journal of the Structural Division, 103(ST10): 1971-2004.
7).
4. The harp system produces larger deflection and larger
bending moments of the girder than the radiating system does. List of symbols
cross-sectional area of member
Summary and conclusion Fourier coefficient
A new iterative method for the analysis of cable-stayed bridges Young's modulus
is presented, which makes it possible to analyze pylons having bending stiffness
any degree of flexibility with SSI included. The method is tension force in cable i
simple and efficiently programmed. The numerical results are shear modulus
identical to those obtained from the linear stiffness method and height of pier or pylon
compare very well with those of the nonlinear method. It was cross-sectional moment of inertia
found that the analysis that assumes pylons with fixed bases stiffnesses of caisson at head
predicts unrealistic response and internal forces. Therefore, stiffness of bridge without tower
ABOUL-ELLA

[Kt] stiffness of tower alone xf horizontal displacement of caisson


Lk span of bridge beam k y girder deflection
M end moment, bending moment Ai deflection of tower at the point of attachment
P concentrated live load of cable i
p intensity of uniformally distributed live load Of angle of rotation of caisson head
[Qc] resultant of horizontal components of the
cable forces
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by San Diego (UCSD) on 01/11/15
For personal use only.

You might also like