Академический Документы
Профессиональный Документы
Культура Документы
102
TEST BRIDGE
Synopsis
The general plan, elevation, and cross
section of the experimental bridge are An experimental prestressed con-
shown in Fig. 1. The bridge consisted of crete segmental bridge was con-
two identical simply supported girders structed and tested at the Pennsyl-
with segments and joints numbered as vania Transportation Institute of The
shown. Each independent girder con- Pennsylvania State University.
sisted of seventeen segments which The bridge was designed by the
were tied together with longitudinal bar Pennsylvania Department of Trans-
or strand post-tensioning tendons plus portation as two independent single-
diagonal bar post-tensioning tendon. span curved girders with a length of
The ducts containing the tendons were 121 ft (36.9 m). Each girder was com-
grouted after post-tensioning. Steel posed of seventeen segments. The
shear dowels were used to achieve bridge was initially field tested at ser-
alignment during construction and to vice load levels and subsequently
transfer torsional moment after the gird- tested for overloads when one girder
ers were built. Epoxy was used as the was tested to failure.
main jointing material between the The incremental loading to failure is
segments. discussed and these results are com-
End diaphragms were introduced in pared with those obtained from a fi-
the end segments, which were ample in nite element analysis (SAP IV), which
size to take the substantial reaction models the cracking patterns and
forces from the neoprene bearing pads material nonlinearities. In addition, the
and torsional anchorages and to provide results of classical simplified analyses
room for the post-tensioning end an- are compared with selected experi-
chorage plates. In addition, an opening mental and finite element results.
was made to allow easy access by re- Conclusions are given that relate to
searchers to the inside of the box sec- the application of the research.
tion. An open longitudinal joint be-
tween the girders was selected to allow
an independent comparison of the two
girders. The overload tests were per-
formed on Girder B, At the bottom of the webs the frame was
The segments for the experimental assumed to he simply supported. Each
bridge were cast individually at a fabri- girder was analyzed for torsion as a hori-
cation plant by the short line method in zontally curved beam with eccentric
one steel form with provisions for ad- loads. The cross section of the segments
justments. They were then hauled about was approximated as a box section with
100 miles (161 km) to the test track the flanges neglected.
facility where they were erected on The design strength of the concrete at
steel scaffolding-type falsework. 28 days was 5750 psi (39.6 Nlmm 2 ). In-
The curved box girder was designed termediate grade ASTM A615 reinforc-
for longitudinal moment using straight ing bars with a specified minimum yield
beam theory for the dead load, AASHTO stress of 40,000 psi (275 NImr 2 ) were
HS20-44 live loading, and prestress. used for all mild steel reinforcement.
The design was made using allowable Post-tensioning steel bars with a
stresses and checked for ultimate specified ultimate stress of 160,000 psi
strength. For transverse moment, the (1100 NImm 2 ), or steel strand with a
segments were designed elastically as a specified ultimate stress of 270,000 psi
box frame with side cantilever flanges. (1860 N/mm z), were used for all post-
Girdefr A
Segment
IIA \2A\4A 5A 6A 7A 8A 9A 110A 112-A I3A 14A 15A 16A Number
lilA
Gird r B ~^
^IB 2B 3B 4B 5 6B 7B BB 9B IOB IIB 1213 13B 146 I5B IGB 17B
r 7'
Radius = 553,62
LSegment
121'D" C/C Bearings
PLAN
Segment
— Joint Number
ii
Diq gonal Posi-Tensioning Tendons
1 ELEVATION
iB'9-3/4"-- J i8°-3/4"
HH
R q di q]
Open Joint
Slope=0.1040 t/ft J Curb and
Curb and —=•— Parapet
Parapet (Cost-in-ploce)
(Precast)
jv ^Segmeni
I gO^
VLOnailudinal P q st-
Torsionol Shear Dowels Tensioning Tendons
TYPICAL CROSS SECTION
Note : t'=0.3O4Bm
1"- 2t.4mrn
Fig. 1. General plan, elevation and typical section of an experimental segmental bridge.
MI
Table 1. Material properties.
Properties of concrete
Age (year) 4.5
Compressive strength (psi) 7,350
Tensile strength (psi) 646
Modulus of elasticity (psi) 5,232,000
Poisson's ratio 0.189
Properties of prestressing bars and strands
Bars Strands
Diameter (in.) 1.25 0.50
Yield strength (ksi) 156.5 256.0
Ultimate strength (ksi) 169.6 283.7
Modulus of elasticity (ksi) 30,555. 28,000.
Percent of elongation 5.5 6.4
for the test bridge can he found in the model ,'- was chosen for the analysis of
final report on Research Project 72-9, the bridge.
Penn DOT Publication No. 118.'
Octahedral Stresses and Strains
FINITE ELEMENT ANALYSIS It was assumed in this work that com-
pressive stresses and strains were pos-
Modeling of Materials itive and that v,, ar , and Q, represent
Concrete in the bridge has complex the maximum, intermediate, and
stress distributions resulting from a minimum principal stresses, respec-
number of sources. Success in analyzing tively. The orthogonal coordinate sys-
tem o, a Z , and a-3 , which defines the
such a structure requires knowledge of
stress space, was transformed into a
the deformational behavior and strength
cylindrical coordinate system in whichz
properties of concrete under multiaxial
coincides with the space diagonal (Q, =
states of stress. Different models based
Uq = o) of the original system, and r and
upon different theories have been pro-
posed for the stress-strain law of con- 9 are the radius and rotational variables,
respectively, on the plane perpendicu-
crete under short-term loads. These are
lar to the z axis (octahedral plane), as
based on plasticity, nonlinear elasticity,
shown in Fig. 2.
model of microstructures, measure of
The two coordinate systems are re-
damage, endochronic theory of plastic-
lated by the following equations:
ity, and mathematical functions. A com-
plete explanation of all these models can z=(a1 +Q2 +v 3 )/f = YWifo (1)
be found in Ref. 2.
After examining these various models, r=(11 3)x
it was concluded that, until more ex- •l 0-1 — O'2)2 + (Q8 — ( •,) 2. +(o- —Q^)s
tensive and appropriate investigations =3 To (2)
become available, a type of model that
allows a direct inclusion of the experi-
mental data should be preferred. Q, + Qz — 2 63
cos O = (3)
Therefore, Kostovos and Newman's r ,^ 6
C3
l3Z
106
U.S (Ultimate Stress)
IQa -^----
0
W - --- - - OUFT
8o
FS.(FaS}ress}
z I
60 o}
aF
w cn o J^
a - OSFP
CJ, a 40
N j I =1 >I
20 ^,i ml m^ @I
(n o^ of al
COMPRESSIVE STRAIN
Fig. 3. Uniaxial stress- strain curve for concrete.
tiplies and propagates, but in a slow sta- considered four stages which will be
ble manner. If Ioading is stopped and explained in the following sections.
the stress level is maintained at a certain Elastic Concrete (Stage I) — Under
value, crack propagation ceases. The in- combined states of stress, the stress-
creasing internal damage, revealed by strain relationship is generally non-
deviation of the linear elastic behavior, linear. However, when stress is below
causes irrecoverable deformation upon 45 percent of the ultimate stress, the
unloading. The start of such deformation material characteristics are unaffected
behavior has been termed "onset of sta- by the fracture processes explained pre-
ble fracture propagation" (OSFP). viously, the deformation is recoverable,
A third stage (Stage III in Fig. 3) and the stress-strain relationship is al-
applies up to the ultimate strength. most linear. Therefore, it is assumed
Interface microcracks are linked to each here that concrete is isotropic, homoge-
other by mortar cracks, and void forma- neous, linearly elastic, and that its
tion (dilation) begins to have its effect stress-strain relations are described
on deformation. The start of this stage completely by two elastic constants,
has been termed "onset of unstable Poisson's ratio (v) and Young's modulus
fracture propagation" (OUFP). The (F).
level is easily defined since it coincides Inelastic Concrete (Stage II) — This
with the level at which the overall vol- stage represents states of stress between
ume of the material becomes a 45 percent of the ultimate stress, which
minimum. has been termed the "onset of stable
A fourth stage defines the region be- fracture propagation" (OSFP), and 85
yond the ultimate strength. In this re- percent of the ultimate stress, which has
gion (Stage IV in Fig. 3), the energy re- been termed the "onset of unstable
leased by the propagation of a crack is fracture propagation" (OUFP). In this
greater than the energy needed for zone of behavior, Kostovos and Newman
propagation. Thus, the cracks become consider deformations to be composed
unstable and self-propagating until of the following components:
complete disruption and failure occurs. 1. A component dictated by the ma-
Similarly, the multiaxial behavior terials characteristics and unaf-
model of Kostovos and Newman has fected by the fracture process.'
C-
U)
U)
U
W tt
J
U)
w
Et 2ET
TENSILE STRAIN
Fig. 4. Assumed stress-strain curve for concrete in tension.
108
tively. And, eo and yo are the hydrostatic hardening of the material was defined
and deviatoric components at ultimate, according to the experimental stress-
respectively, strain curve.
110
C)
L
O Joint 1 2 3 4 5 6 8
n
oSegment O 2O Q ® lJ © lf1 f
3 (19) — (20)
m 39)
199)
Section I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 !71819 20 22
(1), (2), (3),....• Points
E lement Numbering
of Loadings
1
West East
30" 30" 8 16" 24" 24" 24" I6" 8 30" 30"
7 ••
6.25" 7 67
10.25"
6.75' _
7.37"
5.38
6.00"
3.50•,
3.50° Note: I ° = 25.4 mm.
certain truss members were given dif- procedures 1 ° and the equivalent load
ferent values along the beam, depend- method 1° 1 ' was used to compute the
ing on the areas of steel lumped in each forces applied by the tendons on the
division. The effective values of the structure.
modulus of elasticity for these truss The effective prestress force after
members were used where appropriate. losses are accounted for is called Pe.
Symmetry was used in the analysis of These effective prestressing forces were
the bridge with one-half of the total represented as nodal forces on the finite
structure represented by finite ele- element model. The effects of the
ments. change in the vertical alignment of the
prestressing tendons and the horizontal
curvature in the beam axis were consid-
Prestressing Analysis ered in this representation.
The material properties of concrete The initial stresses in the bars and
and steel depend on the stress or strain tendons, caused by the effective pre-
state of the material. The state of stress stressing forces Pe, were calculated by
in a prestressed concrete element is dividing the force Pe + A P, in any bar,
produced by the prestressing forces and by the cross-sectional area of that bar.
toads applied to the structure. There- Here, is the force caused by elastic
fore, at any state of loading, stresses shortening, and it was added to correct
caused by prestressing forces should be for the elastic shortening effect which
calculated and added to those caused by takes place when the effective pre-
the applied loads in order to define the stressing forces are applied as nodal
material properties of the element. forces on the finite element nodel.
The force that a tendon will exert
upon the structure is a function of sev-
eral variables. The most important of Cases of Loading
these are the jacking forces, P, applied The cases of loading needed in the
at the anchors, losses of prestressing analysis procedure are defined as fol-
force, and position and geometry of the lows:
tendon. In this study, the effects of all — Experimentally, the deflections
these variables were taken into account. and surface strains due to live load (LL)
The time-dependent losses were calcu- only were measured. Thus, to compare
lated in detail according to standard these experimental results, theoretical
112
LL LL
— The beam
DL
P ------------
e -----------
e
Cracked region
The beam
DL
P: ----------------------------------- P.
114
Fig. 8. The loading frame of the bridge.
distance from the girder, and two level overload testing began. Most of these
rods, which were permanently mounted cracks were caused by temperature and
at the midspan of the girder. shrinkage. They were traced with black
felt-tipped pens to differentiate them
Strains at Midspan from those caused by live load.
Strains at the middle of Segment 9B The load was applied in increments of
were measured at each load increment 100 kips (445 kN), one increment each
using metal foil electrical resistance day. After each increment of loading, the
strain gages. This segment, which is at bridge was completely unloaded and
reloaded incrementally in the next day
midspan, was chosen because of the
large bending moment at that location. of testing. The testing of the bridge was
The strain gages were placed on the completed in 9 days. For a typical day of
upper surface, the lower surface, and testing, the following steps were carried
both sides of the bridge girder as shown out:
in Fig. 9. Only longitudinal gages were 1. The pressure on the hydraulic rams
used since bending was of primary was released and the steel loading
interest. All strains were used with a beams were lifted up until there was no
Model P-350 Budd Strain Indicator contact between the loading beams and
using a half-bridge circuit with temper- bearing points on the concrete pedestals.
ature compensation gages. 2. Initial readings were taken at zero
load for strains, deflections, horizontal
offsets, and elevations of the deflection
Testing Procedure points on top of the girder.
A crack survey was made for the out- 3. The four rams were activated by an
side and inside of the girder before electric pump, and the pressure read-
64.d" 64.0°
I
TS.D^ ^15.0^
o
Longitudinal
Strain Gage io
N-
Yi
O
116
was added to the measured deflection. first yielding. The remaining part of the
This was necessary so that all experi- curve was defined by two points, The
mental deflections were measured from first point was the first yielding of the
the same origin. Finite element deflec- bars and the second point was the ulti-
tions were obtained by subtracting the mate condition assuming that concrete
deflections due to load case 2 from load fails by crushing when the compression
case 1 (see Fig. 7). The standard theo- strain reaches a value of 0.003. In this
retical analysis procedure for prestressed part of the curve, the theoretical deflec-
concrete structures was used to calcu- tions are much larger than the experi-
late curvature at different load levels. mental ones.
Theoretical deflections were calculated. Table 2 shows the observed, finite
by loading the conjugate beam with the element and the standard calculated
curvature diagram. values for loads and deflections at first
The experimental and theoretical cracking, first yielding of bars, and at
midspan deflections are shown in Fig. bridge failure, The first yielding load for
10. The figure shows good agreement all solutions was taken to be equal to the
between finite element and observed theoretical yielding load, and the corre-
deflections with the finite element sponding deflections were compared
model showing less stiffness than the accordingly. The percentage differences
real structure. That is, under a given between the observed and theoretical
load, the simulated structure deforms values are also shown in Table 2.
more than the actual structure.
Fig. 10 also shows the results of the
Longitudinal Stresses/Strains at
standard analysis for prestressed con-
crete beams which shows good agree- Midspan
ment with experimental values up to Finite element stresses were obtained
1000-
Ullimate Load
900-
y • ^C7•
800'
First Yielding '" ^'
of Bars i
700-
In
600.
Y
a- 500
-0-0 Experimental
C
400 -.------Finite
-Finite Element
/ Frst Crack
300 —r-- Standard Theory
200
by subtracting the stresses of load case 2 up to the first cracking load of 376 kips
from the stresses of load case 1 (see (1670 kN). At a load of 476 kips (2120
Fig.7), Then the longitudinal strains kN), the strain gages located near the
were calculated using the generalized bottom became inoperative due to
Hooke's law for a three-dimensional cracks developing. 12.13
state of stress. Here again, the perma- A sample for the experimental and fi-
nent set strains should be added to the nite element strains obtained is shown
measured strains in order to obtain the by Fig. 11 at a load on the bridge equal to
absolute surface strains which should he 476 kips (2120 kN), which is a load
compared with the finite element below first yielding. The figure shows a
strains. However, permanent set strains reasonable agreement between ob-
were not measured, and this was one of served and finite element strains. The
the main reasons for the deviation in trend observed for deflections is sus-
strain results between the finite ele- tained here; that is, the experimental re-
ment analysis and test results. sults are smaller than the finite element
A comparison of measured and finite results.
element strains did show fair agreement Based on the standard theoretical
up to first yielding where permanent set analysis, compression strains at the top
strains were small but did show a devia- surface of the girder were calculated at
tion of results above first yielding where loads of 439, 600, and 700 kips (1950,
permanent set strains were relatively 2670, and 3110 kN). These strains were
large. Strains increased almost linearly compared with the average compression
118
Observed Strains (/1. in/in)
Finite Element Strains (p.m/in)
-3 7W 7 —^--__ - 377 -346 -312 -299
279) (-322) (-343) (-270) ^--- (-253:
-150 -129
(-177) (-124)
^ Observed
Element
129 1! 166
(99) (114)
Fig. 11, Longitudinal surface strains at middle of Segment 9 for load of 476 kips (2120 kN).
strains obtained experimentally and by were heard at different times and the
the finite element method. The results pressure gage readings dropped down
are shown in Table 3. This table indi- slightly. It sounded like a strand or bar
cates that the finite element method tendon had broken each time. At this
gives better agreement with the ob- load, the cracks at Joints 8 and 9 opened
served results than does the standard widely and extended upward toward the
theoretical analysis. The same indica- top slab.
tion was obtained by the load-deflection In the last day of testing, at load P =
response (see Fig. 10). The percentage 945 kips (4203 kN), two events occurred:
differences between the observed and first, a noise was heard and the deflec-
theoretical values are also shown in tion increased suddenly by 0.25 in. (6.35
Table 3. mm); second, two loud sounds, simlarto
those which occurred at P = 876 kips
Cracking and Failure of the Bridge (3896 kN), were heard, Again, de-
In the first and second days of testing, flection increased suddenly by another
up to load P = 276 kips (1228 kN), there 0.25 in. (6.35 min). Pressure reading
was no visible cracking on the bottom started to fall off, but reached a constant
surface of.the bridge. In the third clay of value. As the load was slightly increased
testing, at load P – 376 kips (1672 kN), to the failure load of P = 953 kips (4250
the first visible cracking was observed at kN), the crack at one of the middle joints
the bottom surface between the points (Joint 8) opened widely and the con-
of loading. As the load was increased, crete in the compression zone crushed
cracks increased in number, and those and spalled on the surface. The mode of
between the points of loading widened failure of the bridge is shown in Fig. 12.
and extended toward the compression Upon inspection of Joint 8, it was found
zone. In the transverse direction, it was that all the strands were broken and the
noticed that cracks began at the inner solid bars were holding the bridge in
side of the girder and progressed gradu- place.
ally toward the outer side. The finite element load at first crack-
Nothing unusual was noticed until the ing was estimated at P = 420 kips (1868
eighth day of testing, at load P = 876 kN) which is 11.70 percent larger than
kips (3876 kN), when two Io>icl sounds the observed value, The cracking load
120
® ^o
it 8 w Jt 9
Girder B, at load P = 776 kips (3453 kN), 2. The experimental deflections and
and at failure. Fig. 13 shows the outside strains (stresses) at midspan from lon-
crack pattern for the bottom of Girder B gitudinal bending were always less than
at failure for the three segments nearest the corresponding finite element values.
midspan. At failure most of the cracking This indicates that the actual structure is
occurred on these three segments. Dot- stiffer than that predicted by theory.
ted lines on this figure represent the 3. Cracking, first yielding, and ulti-
cracks that occurred after load P = 776 mate loads were found to be in good
kips (3450 kN) up to failure. The thicker agreement with their respective finite
line indicates the joint that opened wide element values.
and caused the failure. 4. The general analysis approach
It should be noted that all cracks be- based upon the finite element method
tween the points of loading were flexural gives a great amount of information on
cracks, and as one goes farther from the deflections, strains, stresses, and forces
points of loading, the cracks bend more in the prestressing steel, which can be
in a diagonal direction, These are used in judging the behavior of bridges
known as flexure-shear cracks. It can be if the structure is properly modeled.
noted also that at the maximum load 5. Theoretical deflection and strains
level, there was a tendency in segmental (stresses) obtained by the standard
structures to concentrate strains or cur- analysis procedure for prestressed con-
vatures and yielding of tendons at one or crete structures were found to agree
more joint locations. This is because the closely with observed values up to first
mild reinforcing bars are not continuous yielding, but after first yielding agree-
across the joint. For Girder B, at load P ment was not good.
= 945 kips (4200 kN), Joints 8 and 9 6. The standard prestressed analysis
were wide open, and strain concen- can be used to calculate loads, deflec-
trations were approximately equal at tions, and strains (stresses), tip to the
these joints. At the failure load of P = first yielding of the steel, but after first
955 kips (4250 kN), however, Joint 9 yielding, calculations are not reliable.
started to close, all strain concentrations The ultimate load still can be conserva-
appeared at Joint 8, and a flexural type tively predicted by these standard
failure occurred at this joint methods with a good degree of accu-
rac y.
7. A bending type failure occurred
CONCLUSIONS between the points of loading. There
was no sign of a shear type distress near
The material reported in this paper the ends. The bridge had an adequate
does not cover all of the topics that were but conservative factor of safety against
covered in the research study. More failure.
comprehensive findings have been re- 8. The standard prestress analysis was
ported by McClure, West, and Abdel- used in the design of the experimental
Halim. 12,1 The following conclusions segmental bridge for longitudinal
can be made as a result of this study: bending, shear, and torsion. For future
1. The results of the three-dimen- designs of this type, the standard pre-
sional finite element analysis, deflec- stress analysis can he safely used. How-
tions and stresses (strains), which used ever, an analysis based on the finite
the Kostovos and Newman material element method would also give safe re-
model for concrete, compared rea- sults which are less conservative. Some
sonably well with experimental values economy in design might be achieved
in the elastic and the post-cracking by using the finite element method of
ranges. analysis in the design.
122
ACKNOWLEDGMENT 6. Kostovos, M. D., and Newman, J. B.,
"Behavior of Concrete Under Multiaxial
This study covers a portion of a major Stress," ACI Journal, V. 74, No. 9, Sep-
6-year investigation on an experimental tember 1977, pp. 443-446.
segmental bridge which was conducted 7. Liu, T. C. Y., Nilson, A. H., and Slate,
at the Pennsylvania Transportation In- F. 0., "Biaxial Stress-Strain Relations for
stitute located at The Pennsylvania Concrete," journal of the Structural Di-
vision, American Society of Civil En-
State University. The study was spon- gineers, V. 98, No. ST5, Proceedings
sored and funded by the Pennsylvania Paper 8905, May 1972, pp. 1-25-1934.
Department of Transportation and the 8. Hand, F. R., Pecknold, D. A., and
Federal Highway Administration. The Schnobrich, W. C., "A Layered Finite
contents of this paper reflect the views Element Nonlinear Analysis of Rein-
of the authors who are responsible for forced Concrete Plates and Shells,"
the facts and the accuracy of the data. Structural Research Series No. 389, Civil
The contents do not necessarily reflect Engineering Studies, University of Il-
the official views or policies of the spon- linois, Urbana--Champaign, August
sors. 1972.
9. Bathe, K. J., Wilson, E. L., and Paterson,
F. E., "SAPIV—A Structural Analysis
REFERENCES
Program for Static and Dynamic Analysis
1. Koretzky, H. P., and Tschernelf, A. T., of Linear Structural Systems," EERC
Final Report for Research Project No. Report No. 73-1I, College of Engineer-
72-9 on the Design of an Experimental ing, University of California, Berkeley,
Post-tensioned Segmental Concrete Box June 1973, Revised April 1974.
Girder Bridge, Pennsylvania Depart- 10. Nilson, A. H., Design of Prestressed
ment of Transportation, Publication No. Concrete, John Wiley & Sons, New York,
118, September 1974. 1978.
2. Chen, W. F., Plasticity in Reinforced 11. Lin, T. Y., and Burns N. H., Design of
Concrete, McGraw-Hill Book Company, Prestressed Concrete Structures, Third
New York, N.Y., 1982. Edition, John Wiley & Sons, New York,
3. Kostovos, M. D., and Newman, J. B., N.Y., 1981.
"Generalized Stress-Strain Relations for 12. McClure, R. M., West, H. Ff., and
Concrete," Jou rn al of the Enginee ri ng Abdel-Halim, M., Overload Testing of
Mechanics Division, American Society of an Experimental Segmental Bridge,
Civil Engineers, V. 104, No. EM4, Au- Interim Report, Project 75-3, The
gust 1978, pp. 845-856. Pennsylvania Transportation Institute,
4. Kostovos, M. D., and Newman, J. B., "A University Park, Pennsylvania, July
Mathematical Description of the Defor- 1982.
mational Behavior of Concrete Under 13. Abdel-Halim, M. A. H., Nonlinear Anal-
Complex Loading," Magazine of Con- ysis of a Segmental Concrete Bridge by
crete Research (London), V. 31, No. 107, Finite Element Method, PhD Thesis,
June 1979, pp. 77-90. The Pennsylvania State University,
5. Kostovos, M. D., "A Mathematical De- March 1982.
scription of the Strength Properties of Con- 14, American Association of State Highway
crete Under Generalized Stress," Maga- and Transportation Officials, Standard
zine of Concrete Research (London), V. Specifications for Highwa y, Bridges,
31, No. 108, September 1979, pp. 151-158. Washington, D.C., 1983.