Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title no. 97-S86

Torsional Behavior of Prestressed Concrete Box-Girder


Bridges with Corrugated Steel Webs
by Y. L. Mo, Chyuan-Hwan Jeng, and Y. S. Chang
The type of prestressed concrete box-girder bridges with corrugated steel webs is a major improvement on traditional prestressed
concrete box-girder bridges. First of all, steel webs, instead of concrete webs, are efficient in reducing self-weight. Moreover, using
corrugated webs allows one to avoid using stiffeners usually
needed in flat plate webs, and provides a convenient solution to the
problem of prestress losses. Since the first prestressed concrete
box-girder bridge with corrugated steel webs was built in France
in 1986, the performance of this type of bridge has been receiving
deserved attention and has proved its superiority over the others.
In the past, however, most of the relevant studies were primarily
concerned with the response of the corrugated steel webs. The flexural behavior of the bridge as a whole was not experimentally
studied until 1998. This paper describes the experimental results of
four scaled prestressed concrete box-girder bridges with corrugated steel webs subjected to torsion, and presents an analytical
model to predict the torsional behavior of such bridges. It is found
that the torsional performance looks quite promising and that the
analytical model fits the experimental results satisfactorily.
Keywords: bridges; concrete; prestressed; torsional stress.

INTRODUCTION
The high-speed rail project to improve Taiwans transportation system is a part of the effort of the Taiwanese Government to further the islands economic development. From
the planned route of 345 km, approximately 207 km consist
of viaducts and bridges. Since the end of the 1970s, prestressing has been going through a spectacular revival of innovations for civil engineering structures, especially bridges.
This revival is due to the evolution of construction methods,
the growth of the field of precasting, the research on reduced
cross sections, and the use of modern external prestressing.1
Thus, erection methods have become easier, safer, and
quicker. In fact, when tendons are arranged outside the concrete, the independence between structures and prestressing
opens up new horizons, which engineers are now mastering.
From this point of view, the idea of external prestressing was
the catalyst for a boom in geometry and material innovation
that could be associated with prestressing.
The need for improving the efficiency of classical concrete
box girders and reducing the weight and cost of cross sections led to prestressing of composite box girders, which is
achieved by combining concrete, prestressing tendons, and
steel plates. Contrary to prestressed steel plate webs, which
have to be dimensioned for stability rather than for structural
strength, the use of corrugated webs, capable of withstanding
shear forces without absorbing unwanted axial stresses due
to prestressing, is a very attractive concept.2 As a matter of
fact, the only stresses appearing in corrugated webs are almost pure shear stresses. The local stability of each steel strip
between two folds can be studied using existing design rules.
Then, the form and wavelength of the corrugation are deACI Structural Journal/November-December 2000

signed to ensure the general stability of the webs with an adequate safety factor. Corrugation leads to a minimum web
thickness, which is therefore chosen on the basis of the allowable shear stresses in steel. The corresponding web flexural rigidity has the advantage of eliminating any
longitudinal or transverse stiffeners, which become unnecessary. Because these innovative composite structures judiciously combined steel and concrete, the stability, strength,
and efficiency of the materials used are increased.
As a consequence, since the first prestressed concrete boxgirder bridge with corrugated steel webs was built in France
in 1986,3 the performance of this type of bridge has been receiving attention worldwide4-6 and proving its superiority
over the others. Although none have been constructed in Taiwan so far, bridges of this type are undoubtedly excellent candidates in the future, and are appropriate for the high-speed
rail project.
In the past, however, most of the relevant studies were primarily concerned with the response of the corrugated steel
webs,7-10 and consequently, little was known about the structural behavior of the bridge. The flexural behavior of the structure as a whole was not experimentally studied until 1998.11
In this paper, the torsional behavior of prestressed concrete
box-girder bridges with corrugated steel webs is investigated
by tests on four reduced-scale specimens subjected to reversed
cyclic torsion, and an analytical model is presented. It is found
that the torsional performance looks quite promising, and
that the analytical model predicts the experimental results
satisfactorily.
RESEARCH SIGNIFICANCE
Based on equilibrium and compatibility conditions, as
well as a stress-strain relationship for softened concrete, the
softened truss model theory for torsional behavior of prestressed concrete members presented previously has been
extended to include prestressed concrete box girders with
corrugated steel webs by using both the 1991 Belarbi and
Hsu model and the 1993 Vecchio and Collins Model A as the
constitutive law of softened concrete. To evaluate the accuracy of the proposed theory, a series of four specimens were
tested. It is found that the analytical results predicted by the
1991 Belarbi and Hsu model and the 1993 Vecchio and Collins Model A agree well with the test outcome up to the maximum torque.

ACI Structural Journal, V. 97, No. 6, November-December 2000.


MS No. 00-032 received February 4, 2000, and reviewed under Institute publication
policies. Copyright2000, American Concrete Institute. All rights reserved, including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion will be published in the September-October 2001 ACI Structural
Journal if received by May 1, 2001

849

ACI member Y. L. Mo is Professor of Civil and Environmental Engineering at the


University of Houston, Houston, Tex. He received his PhD in structural engineering
from the University of Hannover, Hannover, Germany. He is a member of Joint ACIASCE Committee 445, Shear and Torsion.
Chyuan-Hwan Jeng is a PhD student in the Department of Civil and Environmental
Engineering at the University of Houston. He received his MS degree in Engineering
Mechanics from the University of Wisconsin-Madison, Madison, Wis.
Y. S. Chang was formerly a MS student in the Department of Civil Engineering,
National Cheng Kung University, Tainan, Taiwan.

Fig. 2Concept of decompression of concrete.

Fig. 1Illustration of equilibrium of RC/PC box-section


member subjected to torsion.

quantities are all expressed in terms of the maximum strain


at the surface of concrete struts ds
2

ANALYTICAL MODEL
Truss model theory for prestressed members
Based on equilibrium and compatibility conditions, as
well as a stress-strain relationship for softened concrete, a
truss model theory for reinforced members was presented by
Hsu and Mo,12 and was later extended to include uniformly
prestressed concrete by using the concept of the decompression of concrete for prestressing steel.13 The equations of
this theory will be summarized. The symbols used in the
equations are defined in the notations list.
Equilibrium equationsThe space truss model shown in
Fig. 1 provides four basic equilibrium equations. From these
equations the torque T, the angle of inclination of concrete
struts , and the effective wall thickness td can be obtained
in terms of the average stress in the concrete struts d
T = 2A o d t d sin cos

(1)

A l f l + A ps f ps
2
cos = ----------------------------------po d td

(2)

A l f l + A ps f ps A t f t
t d = ----------------------------------- + ----------po d
s d

(3)

Compatibility equationsThe compatibility condition of


the space truss model provides five basic equations. From
these basic compatibility equations, the strains in the transverse and longitudinal steel t and l, respectively, the angle
of twist , and the shear strain can be derived.12 These
850

Ao d

t = -------------------- 1--- ds
p o T tan 2

(4)

Ao d
1
l = -------------------- --- ds
p o T cot 2

(5)

2A
ds
= --------o- -------------------------------p o 2t d sin cos

(6)

ds
= -------------------------------2t d sin cos

(7)

Constitutive laws
Prestressing steelTo obtain the stress in the prestressing steel fps , the strain in the prestressing steel should
first be found. Using the concept of the decompression of
concrete,13 the strain in the prestressing steel should be the
strain in the nonprestressed steel plus the strain at decompression
ps = dec + l

(8)

Figure 2 is a superposition of the stress-strain curves of


prestressed and nonprestressed steel to explain the strain at
decompression. When the prestress is applied, there is a
compressive strain in the longitudinal reinforcing bars li and
a tensile strain in the longitudinal prestressing steel pi. This
ACI Structural Journal/November-December 2000

condition is represented by Point A in Fig. 2. When a torsional


moment that corresponds to AD is applied to a prestressed
member, the longitudinal reinforcing bars will become zero
(represented by B), and the strain in the longitudinal prestressing steel will become the decompression strain dec
(represented by C). Therefore, dec can be calculated as
dec = pi + li

(9)

f pi
pi = ----Ep

(10)

where

Fig. 3Belarbi and Hsus 1991 model.


A ps f pi
li = ----------------------------------------------------------------------------A l ( E s E c ) + ( A c A h A ps )E c

(11)

For prestressing steel, there is an initial elastic response up


to a proportional elastic limit. If the stress is increased beyond this proportional limit, a nonlinear relation between
stress and strain will follow until failure. To describe this
nonlinear relation, Ramberg-Osgoods equation can be employed as follows
1
---

E p ps R R
f ps = E p ps 1 + ------------ f

(12)

pu

Mild steelThe stress-strain curve of mild steel is idealized as two straight lines. This steel is assumed to be elastic
up to the yield point, followed by a yield plateau.
ConcreteThe truss model has been applied to treat shear
and torsion of reinforced concrete since the turn of the 20th
century. The prediction based on the truss model, however,
consistently overestimated the shear and torsional strengths
of tested specimens. This nagging mystery has plagued researchers for over half a century. The source of this difficulty
was first understood by Robinson and Demorieux in 1972,14
who realized that a reinforced concrete membrane element
subjected to shear stresses is actually subjected to biaxial
compression-tension stresses. Viewing the shear action as a
two-dimensional problem, they discovered that the compressive strength in one direction was reduced by cracking due to
tension in the perpendicular direction. Applying the softening effect of concrete struts to the thin webs of eight test
beams with I-section, they were able to explain the equilibrium of stresses in the webs according to the truss model.
They concluded that a reduction of 15% for the effective
compressive strength should be taken into account in biaxial
compression-tension stresses. Apparently, the mistake in applying the truss model theory before 1972 was the use of the
compressive stress-strain relationship of concrete obtained
from the uniaxial tests of standard cylinders without considering the two-dimensional softening effect.
The tests of Robinson and Demorieux could not delineate
the variables that govern the softening parameter because of
the technical difficulties in the biaxial testing of large panels.
The quantification of the softening phenomenon, therefore,
had to wait for a decade until a unique shear rig test facility
was built in 1981 by Vecchio and Collins.15 Based on their
ACI Structural Journal/November-December 2000

tests of 17 panels of 89 cm square by 7 cm thick, they proposed a softening parameter that was a function of the ratio
of the tensile principal strains to the compression principal
strain r /d.
The discovery and the quantification of this softening phenomenon have provided the major breakthrough in understanding the shear problem in reinforced concrete. During
the past 20 years, a number of diverse analytical models have
been proposed according to the test results. The effect of
these softening models on low-rise framed shear walls was
studied by Mo and Rothert.16 According to the studies reported by Mo and Rothert,16 two out of all 14 softening models
give the best predictions,17 namely, Belarbi and Hsus 1991
model18,19 and Vecchio and Collinss Model A.20 Therefore,
these two models are briefly introduced as follows and will
be used in the analytical model.
Belarbi and Hsus 1991 model18,19As shown in Fig. 3,
the Hognestad parabola is used as the base curve for the softening model. In the ascending branch
d d 2
d = f c 2 ------- -------
o
o

(13a)

in the descending branch


2
d
------- 1
o

-
d = f c 1 ---------------- --2- 1

(13b)

where
0.9
= -------------------------1 + 600 r

(14)

The average stress of the concrete block d in the diagonal


compression struts is defined as k1fc , where k1 is the ratio
of the average stress to the peak stress. This coefficient can
be obtained by integrating Eq. (13)
when ds p,
851

K f = 0.1825 f c ( MPa ) 1.0

(21)

If d < o, d = dbase

(22a)

with f p = f c and p = o
(22b)

If o < d, d = dbase

(22c)

with fp = fc and p = o.
The ratio of the average stress to the peak stress k1 can be
obtained by integrating Eq. (22).
When < o,

Fig. 4Vecchio and Collinss 1993 Model A.


ds
1 ds
- 1 --- -----k 1 = -----3 p
p

If o < d < o, d = f c

(15a)

B 2
1 A 3
k 1 = -------------- --- ds + --- ds

p ds 3
2

when ds > p,

(23a)

when o < < o,


2
p

- 1 1--- ------ +
k 1 = 1 -----------------
2
3 ds

(2 )

(15b)

Vecchio and Collinss 1993 Model A20As shown in Fig.


4, the Thorenfeldt et al. curve is used as the base curve for
the softening model

B 2
1
A 3
k 1 = ----------------- --- ds + --- ds + f c ( o o ) +

f c ds 3
2

(23c)

2
m 2
( f c + m o ) ( ds o ) ---- ( ds o )

where
(16)
2--- f
pt2 o
3 o c
A = ---------------------------------------------

f p ( MPa )
n = 0.80 + -------------------17

(17)

If d p, k = 1.0

(18a)

f p ( MPa )
If d > p, k = 0.67 + -------------------62

(18b)

1
= -------------------------1.0 + K c K f

(19)

0.80

K c = 0.35 ----r- 0.28


1.0

(20)

where

852

(23b)

when > o,

2
ds
ds

------------------- ------ 1 1--- -----2


3
p
(2 ) p

n ----d-

p
dbase = f p ------------------------------------- d nk

( n 1 ) + ----

B 2
1
A 3
k 1 = ----------------- --- ds + --- ds + f c ( ds o )

f c ds 3
2

2 4--- 2 2 f
o c
o pt2 9
B = ---------------------------------------------------------

(24)

(25)

2 2 2
= --- o
9

(26)

2
--- n
3
pt2 = f p ------------------------------------2 nk
( n 1 ) + ---
3

(27)

f c u
m = -------------------u o

(28)

ACI Structural Journal/November-December 2000

Table 1Material properties of specimens


Concrete
compressive
Prestress
Specimen strength, in each
N/mm2 bar, kN
no.

Fig. 5Illustration of modification of Ao (shaded area).

n ----u-

o
u = f c ------------------------------------- u nk

( n 1 ) + ----
o

(29)

Fundamental principle for prestressed concrete


box girders with corrugated steel webs
In the prestressed concrete box girders with corrugated steel
webs, the top and bottom slabs are made of prestressed concrete and the webs are corrugated steel only. Hence, the area
enclosed by the centerline of shear flow on the prestressed
concrete portions (Fig. 5) Ao and the perimeter of the centerline of shear flow on the prestressed concrete portions po are
calculated according to the contributions of the top and bottom flanges, written as follows
1
A o = --- ( b t b ) ( h t d )
2

(30)

po = 2 ( b tb )

(31)

In other words, the torsional moment contributed by the


top and bottom flanges Tf can be determined by substituting
Eq. (30) and (31) into Eq. (1) to (6).
The torsional moment contributed by the webs is calculated
by the shear force on the webs times the distance between the
couple of shear forces
Tw = w Aw ( b tb )

(32)

The total torsional moment is


T = Tf + Tw

(33)

In addition, it is assumed that the cross section of a box


girder under torsion rotates along the girder axis as a rigid
body. In other words, the angle of twisting is uniformly distributed on the cross section. Hence, the shear strain in the
corrugated steel webs is the same as that in the top and bottom flanges, as shown in Eq. (6).
From the concept of mechanics of materials, the relationship between the yields stress in shear, and the yield stress in
tension21 is
ACI Structural Journal/November-December 2000

Yielding stress, N/mm2

N5

46.4

61

Corrugated
steel
webs
265

Longitu- Transdinal
verse Prestressreinforc- reinforcing
ing bars ing bars strands
420
600
1100

N8

46.4

99

265

420

600

1100

H5

63.3

66

265

420

600

1100

H8

64.4

96

265

420

600

1100

fy
y = -----3

(34)

If the elasto-plastic relationship is employed for the shear


stress-shear strain relationship of the corrugated steel webs,
the following equations can be obtained

w = G web for web < ----y


G

(35)

web > ----y


G

(36)

w = y for

It should be noted that G is the shear modulus of corrugated


steel webs that can be found from Reference 22.
Solution
The relationships among the set of ten quantities (T, , td,
t, l, , , , k1, and ds) can be obtained by assuming the
value of one quantity ds, and then solving simultaneously
the nine equations (Eq. (1) to (7), (13) or (22), and (15) for
Belarbi and Hsus 1991 model, or Eq. (23) for Vecchio and
Collinss 1993 Model A). The algorithm of the analytical
model is shown in Fig. 6.
EXPERIMENTAL PROGRAM
Specimens
The Shinkai bridge built in 1994 in Japan22 was selected
as the reference of the prototype. A scaling factor of 6 was
chosen due to the capacity limitation of the actuator of 500 kN.
Four model specimens were designed for cyclically reversed
torsional tests. Experimental parameters include compressive
strength of concrete and magnitude of prestressing force.
The dimensions and configurations of the specimens are
indicated in Fig. 7. Four specimens were designated as N5,
N8, H5, and H8, respectively. Character N stands for normal-strength concrete (designed fc = 45 MPa), and H for
higher-strength concrete (designed fc = 65 MPa); Character
5 corresponds to a designed effective prestress fpe = 0.5 fpy,
while 8 corresponds to fpe = 0.8 fpy. All of the specimens
were of the same size, which is 2 m long with a 0.8 x 0.45 m
rectangular cross section. There were end diaphragms of 0.4 m
thickness at both ends of each specimen. Both longitudinal and
transverse steel was designed according to ACI 318-95.23
Based on the considerations of local and global buckling, the
corrugated steel webs were designed according to the formulas suggested by Galambos et al.24 and Masayasu et al.22 The
material properties of the specimens are shown in Table 1.
853

(a)

(b)

(c)

(d)

Fig. 6Algorithm of analytical model.


Test setup and procedures
The test setup is shown in Fig. 8. The specimens were
clamped at both ends by I-shaped steels. The clamping Ishaped steels were fastened by nuts and threaded rods. One
end of the specimens was fixed. The specimen was held on a
steel seat supported by the strong floor. The other end was a
854

Fig. 7Dimensions and configurations of specimens (unit:


mm): (a) configuration of cross section; (b) corrugated steel
(side view); (c) dimensions of corrugated steel (top view);
and (d) schematic diagram of shear stud installation.
rotating one, and was subjected to a torsion via a lever arm
of 1 m. Each specimen was instrumented with load cells, linear variable displacement transducers (LVDTs), and strain
gages to monitor the applied displacement and corresponding loads as well as the resulting strains and relative deforACI Structural Journal/November-December 2000

Fig. 9Displacement history.

(a)

Fig. 10Hysteretic loop of Specimen N5.


(b)

Fig. 11Hysteretic loop of Specimen N8.


mations. During the tests, the displacement-controlled
procedures were used, as is shown in Fig. 9. The tests were
continued until the failure occurred, which was defined as
the point when the 80% of the maximum force in the descending branch of torque-twist relationship was reached.

(c)
Fig. 8Test setup: (a) side view; (b) Section A-A; and (c)
Section B-B.
ACI Structural Journal/November-December 2000

EXPERIMENTAL RESULTS
General observation
The hysteretic loops of all four specimens are shown in
Fig. 10 to 13. Each specimen was subjected to 14 loading cycles. Typically, the first observed crack on the top or bottom
855

Table 2Experimental results


Yielding state

Ultimate state

Specimen
no.

Twist,
degree/m

Torque,
kN-m

Twist,
degree/m

Torque,
kN-m

Ductility
factor

Energy
dissipation,
kN-m

Failure
mode

N5

0.555

123.34

2.024

216.63

3.64

37.7

Concrete
crushing

N8

0.643

156.64

2.290

234.95

3.56

47.4

Concrete
crushing

H5

0.541

127.89

2.509

226.93

4.64

41.5

Concrete
crushing

H8

0.623

148.49

2.227

242.74

3.57

42.7

Concrete
crushing

Note: Ultimate state defined as state in which maximum torque is reached.

Fig. 14Crack pattern and failure mode.

Fig. 12Hysteretic loop of Specimen H5.

Fig. 15Comparison of energy dissipation between Specimens H5, N5, and N8.
crushing of the top and bottom flanges. The remaining specimens had the same failure mode as Specimen H8, as shown
in Table 2.
Fig. 13Hysteretic loop of Specimen H8.
flange slab occurred in the third cycle. In the fifth cycle, the
corrugated steel webs started yielding. In the eighth or ninth
cycle, the inclined cracks were found to increase rapidly, and
the transverse steel also yielded followed by the reduction of
wave height of the corrugated steel webs. In the ninth or
eleventh loading cycle, the maximum torque was reached
due to concrete crushing. In the eleventh cycle, concrete cover
spalled. The crack pattern and failure mode of Specimen H8 is
shown in Fig. 14. It can be seen from Fig. 14 that the inclined
cracks are due to torsion and the failure results from the concrete
856

Ductility factor and dissipated energy


The ductility factor is defined as the twist corresponding to
the maximum torque divided by the twist at the occurrence
of longitudinal or transverse steel yielding. The dissipated
energy is determined by integrating the areas enclosed by all
hysteretic loops. Table 2 gives the experimental results of the
yielding state, ultimate state, ductility factor, and dissipated
energy for all four specimens. It should be noted that, in Table
2, the values for the yielding state, ultimate state, and ductility
factor are based on the experimental results in the positive direction, and the values for the dissipated energy are based on
the experimental results in both the positive and negative directions. It can be seen from Table 2 that the range of ductilACI Structural Journal/November-December 2000

Table 3Comparison between experimental and analytical results


test /calc

Ttest /Tcalc

Specimen
no.
N5

1991 Belarbi and Hsu


model
Yielding
Ultimate
state
state

1993 Vecchio and Collins


Model A
Yielding
Ultimate
state
state

1991 Belarbi and Hsu


model
Yielding
Ultimate
state
state

test /calc

1993 Vecchio and Collins


Model A
1991 Belarbi 1993 Vecchio
Yielding
Ultimate
and Collins
and Hsu
state
state
Model A
model
1.013
0.784
0.798
0.772

0.923

0.993

0.923

0.980

1.014

0.810

N8

1.076

1.054

1.078

1.045

1.153

0.935

1.151

0.907

0.810

0.787

H5

0.925

0.987

0.925

0.973

1.005

0.803

1.003

0.834

0.899

0.832

H8
Average

0.984
0.977

1.024
1.015

0.979
0.976

1.016
1.004

1.140
1.078

0.840
0.847

1.136
1.076

1.791
0.829

0.735
0.811

0.694
0.771

Standard
deviation

0.0718

0.031

0.0726

0.034

0.0794

0.061

0.0786

0.057

0.068

0.058

Note: Ttest /Tcalc represents ratio of test torque to calculated torque; test /calc represents ratio of calculated twist to test twist; and test /calc represents ratio of calculated ductility to
test ductility.

Fig. 16Twist-torque relationships of all four specimens.

Fig. 18Experimental and analytical torque-twist curves of


Specimen N8.

Fig. 17Experimental and analytical torque-twist curves of


Specimen N5.

Fig. 19Experimental and analytical torque-twist curves of


Specimen H5.

ity factor for all four specimens is from 3.56 to 4.64. Figure 15
indicates the loading cycle-dissipated energy diagrams. It can
be seen from Fig. 15 that Specimen N8 with greater prestressing
has greater dissipated energy than Specimen N5, and that Specimen H5 with greater concrete compressive strength has greater
dissipated energy than Specimen N5 before the occurrence of
maximum torque; after the maximum torque, the dissipated energy of Specimen H5 is less than that of Specimen N5.

Torque-twist relationships
The torque-twist relationship (primary curve) of each of
all four specimens, which was obtained from the envelopes
of the hysteretic curves, is shown in Fig. 16. It can be seen
from Fig. 16 and Table 2 that the maximum torque increases
with increasing prestress (for example, N8 versus N5) and
concrete compressive strength (for example, H8 versus N8),
and that the ductility increases with decreasing prestress (for

ACI Structural Journal/November-December 2000

857

Fig. 21Comparison of ductility factors.

Fig. 20Experimental and analytical torque-twist curves of


Specimen H8.
example, H5 versus H8) and with increasing concrete compressive strength (for example, H5 versus N5). The differences, however, were not large among these cases.

COMPARISON OF EXPERIMENTAL DATA WITH


ANALYTICAL RESULTS
Figure 17 to 20 indicate the comparison of experimental
data with analytical results. There are three torque-twist
curves in each figure. The solid thin curve represents the experimental data. The solid thick curve represents the analytical predictions using the 1991 Belarbi and Hsu model. The
dashed thick curve represents the analytical predictions using the 1993 Vecchio and Collins Model A. It can be seen
from Fig. 17 to 20 that both the 1991 Belarbi and Hsu model
and the 1993 Vecchio and Collins Model A can well predict
the experimental curves up to the maximum torque. The analytical predictions, however, do not show a descending
branch as the experimental curve does. The reason for this
may be explained below. As shown in Fig. 3 and 4, both the
1991 Belarbi and Hsu model and the 1993 Vecchio and Collins Model A do not show a drastic decrease after the peak
stress in the stress-strain curves.
Table 3 shows the comparison of yielding and maximum
torques, yielding twist corresponding to the maximum
torque, and ductility factor. The following results can be
found from Table 3.
1. The average Ttest / Tcalc values for the maximum torque
from the 1991 Belarbi and Hsu model and the 1993 Vecchio
and Collins Model A are 1.015 and 1.004, respectively, and
the corresponding standard deviations are 0.031 and 0.034,
respectively;
2. The average test /calc values corresponding to the
maximum torque from the 1991 Belarbi and Hsu model and
the 1993 Vecchio and Collins Model A are 0.847 and 0.829,
respectively, and the corresponding standard deviations are
0.061 and 0.057, respectively; and
3. The average test /calc values from both the 1991 Belarbi
and Hsu model and the 1993 Vecchio and Collins Model A are
0.811 and 0.771, respectively, and the corresponding standard
deviations are 0.068 and 0.058, respectively. The experimental
ductility factors of all four specimens are compared with the
analytical predictions from both models, as shown in Fig. 21. It
can be seen from Fig. 21 that the ductility factor of each spec858

imen predicted by the 1991 Belarbi and Hsu model is closer


to the experimental value. Although the analytical torquetwist relationships from both the 1991 Belarbi and Hsu model
and the 1993 Vechhio and Collins Model A agree well with
the test outcome up to the maximum torque, both models do
not provide a descending branch.
CONCLUSIONS
Based on the studies reported in this paper, the following
conclusions can be made.
1. The softened truss model theory for torsional behavior
of prestressed concrete members presented previously was
extended to include prestressed concrete box girders with
corrugated steel webs by using both the 1991 Belarbi and
Hsu model and the 1993 Vecchio and Collins Model A as the
constitutive law of softened concrete;
2. The analytical results predicted by the 1991 Belarbi and
Hsu model and the 1993 Vecchio and Collins Model A agree
well with the test outcome up to the maximum torque. The
predicted torque-twist curves from these two models, however, do not show a descending branch as the experimental
curve does. More research is needed in this area;
3. The test specimens of prestressed concrete box-girder
bridges with corrugated steel webs have acceptable seismic
torsional performance because their ductility factors are in
the range from 3.56 to 4.64; and
4. Specimens with greater concrete strength have both
greater strength and ductility. Specimens with greater prestressing have higher strength, but less ductility.
ACKNOWLEDGMENTS
This research was funded by the National Science Council, Taiwan,
through Grant NSC88-2211-E-006-018. Thanks are due to S. J. Wang and
I. C. Nien for their assistance in constructing and testing the specimens.

NOTATION
A
Ac
Ah
Al
Ao
Aps
At
Aw
B
b
Ec
Ep
Es
fc
fl
fp
fpe

=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=

parameter, as shown in Eq. (24)


area enclosed by outer perimeter of hollow cross section
area enclosed by inner perimeter of hollow cross section
cross section area of longitudinal reinforcing steel
area enclosed by center line of shear flow, shaded area in Fig. 5
cross section area of prestressing steel
cross section area of transverse reinforcing steel
cross section area of corrugated steel web
parameter, as shown in Eq. (25)
height of corrugated steel web
Youngs modulus of concrete
Youngs modulus of prestressing steel
Youngs modulus of reinforcing steel
concrete compressive strength
stress of longitudinal reinforcing steel
peak stress of softened concrete
effective prestress

ACI Structural Journal/November-December 2000

fpi
fps
fpu
fpy
ft
fy
G
h
Kc

=
=
=
=
=
=
=
=
=

Kf

k1
m
n

=
=
=

p0
q
R

=
=
=

s
=
T
=
Tf
=
Tw
=
t
=
tb
=
td
=

=
d
=
dec =
ds
=
l
=
li
=
o
=
p
=
pi
=
=
ps
r
=
t
=
u
=

=
web =

=
d
=
dbase =
pt2
u
w
y

=
=
=
=

initial prestress of prestressed steel


stress of prestressing steel
ultimate tensile strength of prestressing steel
yield prestress
stress of transverse reinforcing steel
tensile yield stress of corrugated steel webs
shear modulus of corrugated steel webs
thickness of concrete flange slab
parameter used in Vecchio and Collins 1993 Model A, as shown
in Eq. (20)
parameter used in Vecchio and Collins 1993 Model A, as shown
in Eq. (21)
parameter used in Vecchio and Collins 1993 Model A, as shown
in Eq. (18)
ratio of average stress to peak stress in concrete struts
parameter, as shown in Eq. (28)
parameter used in Vecchio and Collins 1993 Model A, as shown
in Eq. (17)
perimeter of area enclosed by center line of shear flow
shear flow
Ramberg-Osgoods coefficient to describe stress-strain curve of
prestressing steel
spacing of transverse hoop bars
torque
torque contributed by concrete flange slab
torque contributed by corrugated steel web
thickness of corrugated steel web
thickness of corrugated steel web
effective thickness of concrete flange slab
angle of inclination of concrete struts
softened coefficient of concrete
parameter, as shown in Eq. (26)
strain in diagonal concrete struts
decompression strain of prestressed steel
compressive strain of concrete at outer surface
tensile strain of longitudinal reinforcing steel
initial compressive strain of reinforcing steel
peak strain of nonsoftened concrete
peak strain of softened concrete
initial tensile strain of prestressing steel
tensile strain of prestressing steel
average principal tensile strain in r-direction
tensile strain of transverse reinforcing steel
ultimate compressive strain of concrete
shear strain
shear strain of corrugated steel web
angle of twist
average stress in diagonal concrete struts
concrete stress obtained by using Thorenfeldt curve as base
curve, as shown in Eq. (16)
parameter, as shown in Eq. (27)
parameter, as shown in Eq. (29)
shear stress in corrugated steel web
shear yielding stress of corrugated steel web

REFERENCES
1. Grimm, R., and Zink, M., New Bridge for High-Speed Trains,
Darmstadt Concrete, Institut fuer Massivbau, Technische Universitaet
Darmstadt, Germany, V. 7, 1992, pp. 141-153.
2. Cheyrezy, M., and Combault, J., Composite Bridges with Corrugated
Steel WebsAchievements and Prospects, IABSE Symposium on Mixed
Structures Including New Materials, Brussels, 1990, pp. 479-484.
3. Combault, J., The Maupre Viaduct Near Charolles, France Proceeding of 1988 National Steel Construction Conference, June 1988.

ACI Structural Journal/November-December 2000

4. Knig, G.; Duda, H.; and Zink, M., Neue Entwicklungen im Pannbetonbrckenbau, New Development in Prestressed Concrete Bridges,
Beton-und Stahlbetonbau, V. 89, No. 4, 1994, pp. 85-89. (in German)
5. Combault, J.; Lebon, J. D.; and Pei, G., Box-Girders Using Corrugated Steel Webs and Balanced Cantilever Construction, FIP Symposium,
Kyoto, Japan, Oct. 17-20, 1993, pp. 417-424.
6. Elgaaly, M., and Seshadri, A., Girders with Corrugated Webs under
Partial Compressive Edge Loading, Journal of Structural Engineering,
ASCE, V. 123, No. 6, June 1997, pp. 783-791.
7. Elgaaly, M.; Seshadri, A.; and Hamilton, R. W., Bending Strength of
Steel Beams with Corrugated Webs, Journal of Structural Engineering,
ASCE, V. 123, No. 6, June 1997, pp. 772-782.
8. Elgaaly, M.; Hamilton, R. W.; and Seshadri, A., Shear Strength of
Beams with Corrugated Webs, Journal of Structural Engineering, ASCE,
V. 122, No. 4, Apr. 1996, pp. 390-398.
9. Johnson, R. P., and Cafolla, J., Local Flange Buckling in Plate Girders
with Corrugated Webs, Institute of Civil Engineers, Structures and Buildings, V. 123, May 1997, pp. 148-156.
10. Johnson, R. P., and Cafolla, J., Corrugated Webs in Plate Girders for
Bridges, Institute of Civil Engineers, Structures and Buildings, V. 123,
May 1997, pp. 157-164.
11. Mo, Y. L., and Krawinkler, H., Behavior of Prestressed Concrete
Box Bridges with Corrugated Steel Webs, Research Report, Department
of Civil and Environmental Engineering, Stanford University, Sept. 1998.
12. Hsu, T. T. C., and Mo, Y. L., Softening of Concrete in Torsional
MembersTheory and Tests, ACI JOURNAL, Proceedings V. 82, No. 3,
May-June 1985, pp. 290-303.
13. Hsu, T. T. C., and Mo, Y. L., Softening of Concrete in Torsional
MembersPrestressed Concrete, ACI JOURNAL, Proceedings V. 82, No. 5,
Sept.-Oct. 1985, pp. 603-615.
14. Robinson, J. R., and Demorieux, J. M., Essais de Traction-Compression sur Modeles dame de Poutre en Bton Anne, Part I, June 1968;
Part II, May 1972, Institut de Recherches Appliquees du Bton Arm
(IRBA).
15. Vecchio, F., and Collins, M. P., Stress-Strain Characteristics of
Reinforced Concrete in Pure Shear, Final Report, IABSE Colloquium on
Advanced Mechanics of Reinforced Concrete, International Association
for Bridge and Structural Engineering, Zurich, 1981, pp. 211-225.
16. Mo, Y. L., and Rothert, H., Effect of Softening Models on Behavior
of Reinforced Concrete Framed Shearwalls, ACI Structural Journal, V. 94,
No. 6, Nov.-Dec. 1997, pp. 730-745.
17. Hsu, T. T. C., Discussion on Effect of Softening Models on Behavior of Reinforced Concrete Framed Shearwalls, by Mo and Rothert, ACI
Structural Journal, V. 95, No. 6, Nov.-Dec. 1998, pp. 768-770.
18. Belarbi, A., and Hsu, T. T. C., Constitutive Laws of Reinforced
Concrete in Biaxial Tension-Compression, Research Report UHCEE 9122, University of Houston, 1991.
19. Hsu, T. T. C., Unified Theory of Reinforced Concrete, CRC Press,
Boca Raton, Fla., 1993, 313 pp.
20. Vecchio, F. J., and Collins, M. P., Compression Response of
Cracked Reinforced Concrete, Journal of Structural Engineering, ASCE,
V. 119, No. 12, Dec. 1993, pp. 3590-3610.
21. Salmon, C. G., and Johnson, J. E., Steel Structures, Harper Collins
Publishers Inc., New York, 1996, 54 pp.
22. Masayasu, K.; Yoichi, S.; Kinishi, K.; and Masaaki, H., Design and
Construction of the Shinkai Bridge-Prestressed Concrete Bridge Using
Corrugated Steel Webs, Bridge and Foundation, Sept. 1994, pp. 13-20. (in
Japanese)
23. ACI Committee 318, Building Code Requirements for Structural
Concrete (ACI 318-95) and Commentary (318R-95), American Concrete
Institute, Farmington Hills, Mich., 1995, 369 pp.
24. Galambos, T. V., Guide to Stability Design Criteria for Metal Structure, John Wiley & Sons, Inc., New York, 1988.

859

You might also like