You are on page 1of 11

ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 94-S11

Cyclic Behavior of Concrete Bridge Bents

by Robert Sexsmith, Donald Anderson, and Daryl English

During the planning of the seismic retrofit of Vancouver’s 40-


year-old Oak Street Bridge, it became evident that a test program
of a typical bridge pier under slow cyclic loading would assist in
the determination of appropriate retrofit techniques. A series of
0.45 scale models of an as-built pier were constructed. One was
tested in the as-built condition, while the others received various
retrofits. Slow cyclic lateral load was applied at the model deck
elevation. Hysteretic behavior of the bents and the output from
numerous strain gages and displacement transducers were
recorded. The test program provided significant design information
for the retrofit of the bridge and for the many similar bridge bents
currently in use and needing seismic retrofit.

Keywords: bridge piers; ductility; earthquake-resistant structures; hysteresis;


reinforced concrete.

The British Columbia Ministry of Transportation and


Fig. 1—Oak Street Bridge looking northeast.
Highways has conducted seismic assessment of a number of
bridges, and it has been noted that many of the reinforced
concrete bents that support the approach spans require retro-
fitting. Owing to the number of bents and the significant
retrofit cost per bent, the seriousness of the expected failure
modes, and the likelihood that a variety of retrofit designs by
various consultants would produce inconsistent levels of
safety, it was decided that a test program of two-column
bents would be beneficial.
Vancouver’s Oak Street Bridge, currently under retrofit,
was the basis for the test program, which included five speci-
mens constructed to the original design of Bent S28, considered
to be a typical bent of moderate height. The specimens were
at 45 percent scale to the prototype, comprising the upper
half of the bent (above the midheight inflection point). Of the
five specimens, four were retrofitted, using a variety of
schemes. An overall view of the bridge is shown in Fig. 1, Fig. 2—Delivery of test specimen at laboratory.
while Fig. 2 shows one of the test specimens being delivered
to the laboratory. patterns, extensive photographic records, and numerous
The tests consisted of constant application of a vertical reinforcing steel strain records.
load to simulate the structure dead load. In addition, a Several of the retrofits were demonstrated to be acceptable
cyclic lateral load, simulating earthquake load transverse to for the Oak Street Bridge, with the final retrofit choice
the bridge deck, was applied slowly and through many cycles.
The imposed deflection was increased in successive groups ACI Structural Journal, V. 94, No. 2, March-April 1997.
of cycles until extreme damage levels were reached. The Received Mar. 13, 1995, and reviewed under Institute publication policies. Copy-
right © 1997, American Concrete Institute. All rights reserved, including the making
behavior of the specimens was observed during the tests in of copies unless permission is obtained from the copyright proprietors. Pertinent
discussion will be published in the January-February 1998 ACI Structural Journal if
terms of load-deflection hysteresis, observed cracking received by Sept. 1, 1997.

ACI Structural Journal/March-April 1997 103


The older bridges in British Columbia, such as the Oak
Robert Sexsmith, FACI, is an associate professor in the Department of Civil Engi-
neering, University of British Columbia, Canada. He is a member of ACI Committees Street Bridge, were designed under AASHO (now AASHTO,
348, Structural Safety, and 341, Earthquake Resistant Concrete Bridges. He is a American Association of State Highway and Transportation
member of the Canadian National Committee on Earthquake Engineering and other Officials) provisions. Prior to 1958, only vague references
bodies responsible for development of codes and standards. He has been responsible
for the design for a number of bridge seismic retrofits, and is currently involved in were made to seismic design requirements, although
research for seismic retrofit and retrofit decisions for bridges and other structures. CALTRANS (California Department of Transportation)
had developed specific seismic provisions in 1943.
Donald Anderson is a professor in the Department of Civil Engineering, University of
British Columbia. He is a member of the Canadian National Committee for Earthquake AASHTO in 1958 introduced interim provisions
Engineering and the code committee for CSA S-304, Masonry Structures. His research following the 1943 CALTRANS provisions. These required
interests include nonlinear seismic response and design of structures.
a lateral seismic force as a percentage of the weight. The
Daryl English is an engineer-in-training with Sandwell, Inc., a major Vancouver force depended on footing type: 2 percent for spread footings
consulting firm, and a former graduate student in the Department of Civil Engi- on firm soil, 4 percent for spread footings on soft soil, and 6
neering, University of British Columbia.
percent for piled footings. These values were used by
AASHTO, and in British Columbia, until the mid-1970s.
The magnitude 6.6 1971 San Fernando Earthquake was
instrumental in highlighting the seismic vulnerability of
dependent on issues such as experience of the Ministry bridge structures designed according to the previous stan-
consultants, costs, and appearance. dards. This resulted in a number of changes to practice in
California and elsewhere, and to revision of the standards.
RESEARCH SIGNIFICANCE AASHTO 19751 and the similar CAN3-S6-M78,2 adopted
The behavior of old, lightly reinforced concrete members in 1980 in British Columbia, provided for a seismic force of
such as bridge bents has often been found to be inadequate up to about 8 percent of the weight, depending on the zone.
under modern seismic criteria. Prediction of such behavior is A response spectrum method was recommended for
difficult because it is often dominated by shear and by inad- complex cases, and requirements for restraining elements
equate anchorage and confinement. were included. These force levels were used with working
To design cost-effective upgrades of old concrete struc- stress design rules.
tures such as the bents of the Oak Street Bridge, a great deal In the mid-1980s, AASHTO 19833 was adopted for
of information is needed to establish and quantify improve- British Columbia.The peak ground velocity parameters from
ments to ductility and strength and construction strategies the National Building Code of Canada4 were substituted for
that will best achieve the improvement with minimum cost the A values of the guidelines.
and disruption. The design philosophy prior to the 1985 adoption of
The Oak Street bridge bents provided an ideal basis for the AASHTO 1983 was primarily to provide for the design
test program because the results were directly utilized for the seismic forces. These forces were then treated like all other
retrofit design of the bridge, and they also provided a useful load cases in the design of the structure. Since that time,
basis for retrofit of many similar structures. however, the principles of “capacity design” have been
accepted, along with much increased emphasis on details
PROGRAM OBJECTIVES that provide the required post-yield performance.
The test program was planned to meet a number of objec-
tives, including the following:
Structural description of existing bents
1. To experimentally determine the critical failure modes
Fig. 3 illustrates the details of a typical prototype bridge
of typical double column bents, as they exist in many major
bent. By far the most common bent arrangement comprises
bridges in the greater Vancouver area of British Columbia
48 in. (1219 mm) square columns with 16 #11 (35M) vertical
and elsewhere.
bars and #3 (10M) ties (single closed hoop and two single leg
2. To quantify postelastic performance of the as-built cross ties) spaced at 12 in. (304 mm) on centers. The beams
bents and the various retrofits, including both critical failure are 42 in. (1067 mm) wide by 60 in. (1524 mm) deep with #4
modes and ductility reserves. (15M) stirrups at 36 in. (914 mm) centers in the central
3. To demonstrate the behavior of the beam-column joint region and at 3-1/2 in. (89 mm) in the 24 in. (609 mm) zone
in the as-built and retrofitted bents. of maximum shear nearest the column faces. Beam flexural
4. To demonstrate that the criteria established for the steel consists of 13 #11 (35M) bars at the locations of
detailed design of the bent retrofits provides acceptable maximum positive and maximum negative moment, with
margins of safety against undesired failure modes, and to cutoffs located to suit the design moment demand very
identify areas of conservatism leading to unnecessary expen- closely. There are a total of 21 bents with these details,
ditures in retrofit designs. varying only in height, and an additional 24 taller piers with 24
bar column reinforcing arrangements and similar cap beam
PROTOTYPE BENT details. Bent S28 was chosen as a typical pier as it was near
Code seismic provisions in the past
Identification of potential problems in bridge seismic the average height and one in which the major deficiencies
performance depends upon an understanding of the code were in the cap beam.
design provisions that would have been applicable at the The prototype bridge bents have a number of important
time of design. seismic deficiencies:

104 ACI Materials Journal/March-April 1997


1. Stirrups comprise only two legs of closed hoops spaced
at 36 in. (914 mm) centers (75 percent of beam depth) in
potential plastic hinge zones. The volume and arrangement
of stirrup steel is inadequate for shear, concrete confinement,
longitudinal bar anchorage, and restraint of main bars from
buckling in hinge zones.
2. Horizontal beam-column joint reinforcing comprises
only 3 #3 (10M) ties at 12 in. (305 mm) centers. Vertical
column reinforcement terminates without hooks at the top of
the joints. The four bottom beam bars that are embedded in
the joint are curtailed halfway (24 in. [609 mm]) into
columns. While joint details would be deficient by modern
standards, in this case they were generally not the weakest
link in the system.
3. Beam flexural reinforcement curtails at the earliest
possible points. These locations were determined using
moment diagrams that treated all loads as elastic rather than
considering demands arising from members reaching their
flexural resistances (capacity design principles were not
known at the time of the original design).
4. Positive moment reinforcement (reinforcement at the
bottom of the beams) adjacent to joints was reduced to only
4 #11 (35M) bars due to reinforcement curtailment, while
negative moment reinforcing (top bars) at the same location
comprised up to 13 #11 (35M) bars. The unbalanced positive
and negative moment capacity of the beam may be expected
in some cases to increase the cumulative inelastic strains
experienced by the positive reinforcement.
5. The anchorage of the positive moment reinforcement at
24 in. (609 mm) into the columns is questionable, although
an architectural “fillet” detail at the beam soffit-column Fig. 3—Details of prototype bent (1 ft = 305 mm; 1 in. =
intersection alleviates this condition to some degree. 25.4 mm).
6. As was the case with the beam stirrups, tie steel is inad-
equate for shear in some columns, concrete confinement, and
restraint of main bars in plastic hinge zones.

Material properties
Samples of reinforcing steel and concrete cores were
obtained from the prototype bridge. Based on the tests of
samples, and a knowledge of the market in these products at
the time of construction, reinforcing steel was assumed to be
Grade 40 (40 ksi [275 MPa]), but with a likely variation up
to about 55 ksi (380 MPa). Concrete was assumed to have an
equivalent cylinder strength of about 6 ksi (41 MPa).
A test report by Anderson et al.5 contains further details of Fig. 4—Test setup.
the prototype bent and material properties.

EXPERIMENTAL ARRANGEMENTS The test setup is presented in Fig. 4. The vertical dead load
The test models were scaled to 0.45 of the prototype. This was applied with the use of spreader beams and Dywidag
was convenient in terms of scaling bar diameters (#11 scaled bars connected to hydraulic jacks anchored to the structural
to #5, 35M scaled to 15M), and resulted in overall dimensions test floor. These dead load forces were maintained constant
that provided adequate clearance under the overhead crane throughout the duration of each of the tests by controlling the
of the test laboratory and an overall weight that was about the hydraulic pressure at the jacks, and were applied at five
same as the allowable crane-handling capacity. The models bearing points consistent with the location of the bridge deck
represent the upper half of the prototype; the pinned bases of girders. Based on the anticipated structural behavior of the
the model are at midheight of the prototype columns, which diaphragms between the five girder spans, it was assumed
is the approximate location of the point of inflection for that the two interior girder bearings closest to the columns
inelastic behavior assuming yield at the top and bottom of would transfer most of the lateral load of the deck into the
the columns. bent. It was also anticipated that the cap beam would elongate

ACI Structural Journal/March-April 1997 105


lateral stability. The entire test frame was post-tensioned to
the laboratory’s strong floor.
The major external instrumentation used during testing
is shown in Fig. 5. The arrows in the figure indicate direction
of measured displacements or load. The vertical joint
displacements were measured with potentiometers, and
horizontal displacements were recorded with LVDTs
(displacement transducers) located at the center of the two
joints. The lateral force was measured with a load cell
located as shown. Relative horizontal displacements
between specimen and load frame were measured at the
actuator location. Displacement measurements were
taken at the north joint with a system of five LVDTs to
determine joint shear deformation. There were approxi-
mately 80 strain gages located in critical regions on the
reinforcement of each specimen. The signals from all
instrumentation were recorded every 2 sec and stored on
a personal computer for further processing.

LOADING PROGRAM
To simulate dead load on the specimen, the prototype dead
load from the superstructure and the self-weight were scaled,
resulting in a required dead load of 39 kips (173 kN) at each
Fig. 5—Instrumentation (1 ft = 305 mm). of the five bearing locations, as shown in Fig. 4. The nomi-
nally vertical Dywidag bars that transmitted load from the
Table 1—Load sequences jacks to the top of the test specimen took on a slight slope as
Sequence No. of cycles Amplitude Period, min. the specimen displaced laterally; hence there was a small
A 2 or 3 * 10
horizontal component that countered the lateral applied
10 or 20 kips
force. The error introduced by this second order effect was
B 2 or 3 20 or 30 kips* 10
about 2 kips (9 kN) per inch of lateral displacement. Hyster-
C 2 or 3 † 10
x etic curves presented herein have not been corrected for this
D 3 D‡ 10 minor effect.
E 3 1.5D 10 The lateral load program consisted of several sequences of
F 3 2.0D 10 load or displacement cycles in a “sawtooth” pattern at
G 3 3.0D 10 increasing amplitude. Within each sequence there were typi-
H 3 4.0D 10 cally three complete cycles at one amplitude. The sequences
I 3 6.0D 10 at low amplitude were performed to check the loading
J 3 9.0D 10 pattern and data acquisition systems and to establish the
K 3 12.0D 10 initial (uncracked) stiffness of the test specimen. Typically
*1 kip = 4.448 kN. the low sequences, up to about 75 percent of the estimated

x to be set at 75 percent of predicted first yield force.

yield, were on load control. The sequence at 75 percent of esti-
D is predicted displacement at first yield force based on 133 percent of measured
displacement at load x. mated yield load was carried out to predict the yield
displacement, which was taken as 4/3 of the displacement
at 75 percent of the estimated yield load. Subsequent
sequences were at multiples of this yield displacement, i.e.,
as it underwent inelastic deformations. It was for these at ductility levels of 1, 1.5, 2, 3, 4, 6, 9, and 12, or up to
reasons that the lateral load was applied to the bent through failure where that occurred earlier.
a determinate truss system, which permitted slight elonga- Table 1 indicates the sequences with the corresponding
tions of the cap beam and applied the lateral load equally to amplitudes and period. For a given sequence, the number of
the bent at the two interior dead load bearing locations. cycles and amplitude varied slightly from test to test. This
The lateral load was applied by a horizontal actuator located variation arose from some specimens being loaded to more
above the top of the cap beam to simulate the deck inertial than three different amplitude levels in the linear range of
loads. A load cell was mounted between the actuator and the response. These variations are evident in the specific test data.
top of the truss.
The specially constructed frame provided the horizontal TEST MODELS
reaction to the lateral load actuator. The system consisted of As-built models
two triangular trusses connected at the top with a spreader Fig. 2 shows Specimen OSB1, the as-built model, being
girder onto which the actuator was attached. Lateral bracing delivered to the structural testing laboratory at the University
perpendicular to the main axis of the test setup provided of British Columbia. Fig. 6 provides the dimensions of the

106 ACI Structural Journal/March-April 1997


as-built models. All of the specimens were initially
constructed to be identical to OSB1, and were subsequently
retrofitted. The retrofitted specimens were numbered OSB2
through OSB5, and are listed in Table 2. The successive
numbering indicates the order of testing.
OSB2—This prototype retrofit consisted solely of coring
through the longitudinal axis of the cap beam and then
post-tensioning with two tendons of 12K15 bundles of
strand (12 strands at 5/8 in. [15 mm] diameter each). The
scale model retrofit used two 1/2 in. (13 mm) and one 5/8 in.
(15 mm) diameter strands in each hole, located 12-1/4 in.
(311 mm) below the top of the cap beam. The net post-
tensioning in the cap beam produced an average compres-
sive stress of 417 psi (2.9 MPa). The tendons were pressure Fig. 6—Dimensions of OSB1, as-built model (1 ft = 305
grouted, insuring a fully bonded post-tensioned element. The mm; 1 in. = 25.4 mm).
columns were not altered in this retrofit.
The increased compressive stress of 417 psi (2.9 MPa) was
intended to improve the moment and shear capacity of the
cap beam and of the cap beam-column joint region and force
yielding into the column tops. If the cap beam did not yield,
the effects of the insufficient bar anchorage length, severe
bar cutoff locations, and poor confinement of the longitu-
dinal bars would be of lesser concern. This retrofit was not
capable of improving the poor column tie situation.
OSB3—This was termed the “underbeam” retrofit, and
appears in Fig. 7. A reinforced beam was cast to the soffit of
the existing beam and vertical post-tensioning rods, placed
in holes cored through the existing beam, anchored the two
together as a unit. The underbeam longitudinal bars were
drilled and grouted into the columns. Also added was a
longitudinal steel plate acting as a bearing plate for the Fig. 7—Details of OSB3 (1 ft = 305 mm; 1 in. = 25.4 mm).
vertical rod anchors on the top face of the existing cap beam.
These measures provided significant additional positive and Table 2—Specimen designation
negative moment capacity to the cap beam, enough to force Specimen Description
yielding into the columns. The shear capacity of the cap OSB1 Oak St. as-built
beam was greatly improved due to the additional vertical OSB2 Cap beam post-tensioned longitudinally
bars. To improve the column shear capacity, the columns OSB3 Underbeam, vertically post-tensioned to cap beam
were retrofitted with circular 3/16 in. (5 mm) steel jackets OSB4 Cap beam post-tensioned longitudinally and vertically
extending to within 1 in. (25 mm) of the underside of the new OSB5 Longitudinal post-tensioning with fiberglass jackets
beam. The jackets were grouted to the original rectangular
columns.
OSB4—Based upon the performance of the cap beam been increased with the use of the vertically post-tensioned
during the test of OSB2, the cap beam of OSB4 was also bars. Thus, the post-tensioning for the prototype was by one
retrofitted with bonded longitudinal post-tensioning and, in 19K15 tendon rather than two 12K15 tendons. This was
addition, with vertically post-tensioned rods to improve modeled in the scale bent by using 4-5/8 in. (15 mm) diameter
shear capacity. The vertical reinforcement was designed to strands, stressed to give an average compressive stress in the
prevent the wide shear cracks observed in OSB2 and also to cap beam of 342 psi (2.4 MPa), compared to 417 psi (2.9
add a vertical compressive force that would improve the MPa) in OSB2. Fig. 8 shows the details of Specimen OSB4.
bond strength of the cap beam top longitudinal reinforce- OSB5—The OSB5 retrofit featured fiberglass wraps in the
ment over the region of the cap beam between the columns, regions of high shear on the cap beam and columns, and a
where bond demands are high near the column faces and cap-beam longitudinal prestress of 342 psi (2.4 MPa) using
at bar cutoff regions and confinement is poor. Steel jackets external Dywidag bars. The fiberglass wrap was intended to
similar to those used in OSB3 were added to the columns to meet the shear demand in the cap beam and in the columns,
prevent column shear failures. The steel jackets were while the longitudinal prestress improved the joint behavior
expected to slightly increase the yield strength of the and the flexural capacity of the cap beam. Fig. 9 provides
columns due to the improved confinement and thus increase details of this retrofit. The fiberglass wrap has its principal
the seismic demands on the cap beam. direction perpendicular to the axis of the member; thus the
The longitudinal post-tensioning stress was reduced from flexural behavior is unaffected, while the shear behavior is
that in OSB2 because the shear strength of the cap beam had enhanced. The design of the fiberglass wrap was in general

ACI Structural Journal/March-April 1997 107


Fig. 8—

Fig. 10(a)—Hysteresis for OSB1 (1 in. = 25.4 mm; 1 kip =


4.448 kN.)

Test OSB1—As-built specimen


This as-built specimen showed very poor, brittle behavior,
as evidenced by the hysteretic curve shown in Fig. 10(a). The
maximum lateral load was only 60 kips (266 kN), and was
not sustained at subsequent load cycles. Fig. 11, showing the
Fig. 9—Details of OSB5 (1 ft = 305 mm; 1 in. = 25.4 mm). specimen on the first cycle at a displacement of 0.9 in. (23 mm),
shows a large diagonal crack forming in the cap beam under
negative moment when the shear is highest. This crack
increased in width with each cycle of load until ultimately
conformance to the recommendations of the developers of
the concrete compressive zone suddenly failed, allowing a
the system, Hexcel-Fyfe.6 The thickness in the cap beam was
portion of the beam to drop down and to be restrained only
calculated to provide a vertical capacity, at half the ultimate
by the beam’s bottom steel (Fig. 12). The top of the crack
capacity of the fiberglass, equal to the capacity of the vertical
coincided with the cutoff of several of the top bars. The wide
Dywidag bars of OSB4, which was in excess of that required
cracks were likely due to bond failure, as evidenced by hori-
for shear demand. This resulted in three wraps on the beam,
zontal splitting at the level of the top reinforcement, as seen
and a nominal thickness of 0.15 in. (4 mm). The column
in Fig. 12. Neither the joint region nor the columns showed
wraps were designed to resist the entire shear demand. This
any serious cracking as the demand on these elements was
could have been met with a single wrap, but two wraps were
limited by the cap beam shear failure.
considered the minimum that would ever be used, and this
resulted in a nominal thickness of 0.10 in. (2.5 mm). The The hysteretic plot, Fig. 10(a), shows rapid degradation of
fiberglass wraps were laid up by hand by an experienced crew. strength and stiffness with each cycle. In addition, the theo-
retical lateral load capacity of 62 kips (275 kN) base shear
TEST RESULTS based on flexural member strengths was not achieved
Overview because of the force limiting cap beam shear failure.
An overall view of the results for the Oak Street Bridge
can be made by comparing the hysteretic curves for the five Test OSB2—Cap beam internal post-tensioned
test specimens, given in Fig. 10(a) to 10(e). The joint retrofit
displacement in the figures is the average horizontal The results for the prestressed retrofit OSB2 showed a
displacement measured at the two joints, and the base shear marked improvement in both strength and ductility when
is the total base shear. Observation of the five figures shows compared to the as-built specimen, as evidenced by hyster-
the tremendous improvement in the load-deflection behavior etic curves of Fig. 10(b). The displacement at Ductility 1
of the retrofitted specimens when compared to the as-built corresponds to estimated structure yield displacement.
specimen results. In OSB2, diagonal cracks formed in the beam in nearly the
In this section each test is discussed separately, and then same location as in OSB1; however, they did not open as
comparisons are made among the results for the different wide, even at large load levels, and ultimately the beam did
retrofit schemes. not fail in shear. Fig. 13, showing OSB2, was taken at a

108 ACI Structural Journal/March-April 1997


Fig. 10(b)—Hysteresis for OSB2 (1 in. = 25.4 mm; 1 kip = Fig. 10(d)—Hysteresis for OSB4 (1 in. = 25.4 mm; 1 kip =
4.448 kN). 4.448 kN).

Fig. 10(c)—Hysteresis for OSB3 (1 in. = 25.4 mm; 1 kip = Fig. 10(e)—Hysteresis for OSB5 (1 in. = 25.4 mm; 1 kip =
4.448 kN). 4.448 kN).

greater displacement than for OSB1 in Fig. 12, and shows load reduced to about 20 kips (89 kN), caused by the over-
the much improved crack control in OSB2. Flexural shear turning moment from the lateral loading. This sudden shear
cracks formed in the columns of OSB2 in the region of the failure at a ductility of 6 produced a wide diagonal crack in
fillet, gradually becoming wider and longer as the ductility the column and resulted in the sudden drop in load. Unfortu-
level increased. Up until the last load cycle, when a sudden nately the ultimate strength of the cap beam retrofit could not
column shear failure occurred, there was very little strength be determined because of the column shear failure.
degradation, with the three hysteresis curves at each However, the size of the shear crack was of concern and indi-
sequence nearly falling on top of each other [Fig. 10(b)]. On cated the possibility of a brittle shear failure in the cap beam
the last loading cycle, which was initially a push, the “tension” had the columns been strengthened, and therefore further
column failed suddenly in shear on the subsequent pull retrofit measures that increased the shear capacity of the cap
cycle. Here “tension” column refers to the column with axial beam were considered.

ACI Structural Journal/March-April 1997 109


Fig. 11—OSB1 at Displacement Ductility 3.

Fig. 14—OSB3 near failure in columns immediately above


steel jacket.

The hysteretic force displacement curves show very good


performance to a ductility level of 9 with the strength
continuing to increase. At a ductility level of 6, the column
and the joint region had many cracks but there was no signif-
icant spalling of the cover concrete. At a ductility level of 9,
minor spalling took place at the bottom of the underbeam.
There was about a 5 percent strength loss in each subsequent
cycle of this sequence. The peak strength at a ductility level
of 12 dropped from that of Ductility 9, although there was
an increase from the last cycle of the previous sequence. The
Fig. 12—OSB1 at Displacement Ductility 4. strength loss in subsequent cycles at Ductility 12 was
approximately 10 percent in each cycle. At the final ductility
level of 12 spalling was extensive, as shown in Fig. 14. All
the vertical column bars along the exterior face of the
column buckled, and some fractured as the number of cycles
increased at this ductility level.
The addition of the underbeam and vertical reinforcement
to the cap beam made the beam sufficiently strong in both
flexure and shear to force flexural hinging into the columns
with little damage to the beam, although there was some
damage in the joint region. The cap beam developed nega-
tive moment and shear cracks near the joints at early stages
of loading, but cracks did not grow as the displacements
increased, indicating that the retrofit scheme adequately
protected the cap beam from damage.

Test OSB4—Vertically and horizontally


prestressed cap beam with column jackets
Fig. 13—OSB2 at Displacement Ductility 6. In many respects the test results for this specimen are very
similar to those for OSB3. The specimen eventually failed in
flexure in the column region above the column jackets by
Test OSB3—Underbeam and column jacket retrofit spalling of the cover concrete and buckling of the vertical
The hysteretic response of Specimen OSB3 is shown in column reinforcement. The hysteretic force displacement
Fig. 10(c). The specimen eventually failed in flexure in the curves are shown in Fig. 10(d) and show very good perfor-
column immediately above the steel jackets. At high mance up to a ductility level of 9. The first cycle to a ductility
ductility levels the concrete cover above the steel jackets level of 12 showed a strength loss when compared to the
spalled, allowing the column longitudinal bars to buckle and previous sequence, and subsequent cycles had approxi-
eventually fracture. mately a 15 percent loss in each cycle.

110 ACI Structural Journal/March-April 1997


Fig. 15—OSB4 at Displacement Ductility 6.
Fig. 16—OSB4 at Displacement Ductility 12.
At low ductility levels, flexure and shear cracks developed
in the cap beam but these did not widen or grow as the test
continued and the beam retrofit was successful in forcing the
damage into column flexure. Fig. 15 shows the column and
joint region at a ductility level of 6.
Fig. 16 shows the damage at a ductility of 12. At this level,
there was much more spalling of the cover concrete, more
bars had buckled, and some bars had fractured. As the
cycling progressed, more of the vertical bars fractured and
this lead to the subsequent reduction in strength with each
cycle as shown in Fig. 10(d). Fig. 16 also shows that some of
the diagonal cracks in the joint had joined up to form one
diagonal crack extending the full length of the joint from
lower right to upper left. However, this crack did not open to
any appreciable extent.
Fig. 17—OSB5 with fiberglass wrap, before testing.
Test OSB5—Prestressed cap beam, fiberglass
wrap on beam and column
The hysteretic load displacement response for this test is
shown in Fig. 10(e). The performance of this retrofit was
very good up to a displacement ductility level of 9, where the
test had to be terminated because of limitations in the
displacement capacity of the loading system. Fig. 17 shows
the specimen at the beginning of the test and shows the
extent of the fiberglass wrap. This specimen was more flex-
ible than the two previous specimens, which had full-length
steel column jackets; thus the yield displacement was larger,
and so even though the maximum displacement was greater
than in the previous tests the ductility demand was not as
high, although the plastic rotations were greater.
As the loading progressed, more cracks developed in the
joint area but not in the short central length of the beam that
was visible. Fig. 18 shows the cracks at Ductility 9 (note the
ductility shown on the caption board is in error). At peak Fig. 18—OSB5 at Displacement Ductility 9.
deflection all these cracks were relatively small, including
those in the joint region. A large crack did develop in the reinforcement. The size of the crack did not seem to affect
fillet portion of the column just above the top of the column closure on load reversal or give rise to spalling.
wrap, but this only reflects the large distance from the After this test, the fiberglass wrap was removed to
surface where the crack is seen and the column and beam examine the crack patterns. In addition, a test of the bare

ACI Structural Journal/March-April 1997 111


structure was carried out. The specimen was subjected to Because of the failure in shear and bond at each end of the
more cycles at Ductility 9. Although this specimen was now cap beam, the column-cap beam joint region and columns
similar to OSB2 in that it was longitudinally prestressed, were not subjected to a large demand; hence their behavior
the well-distributed cracking that had taken place rendered was not tested to capacity. Their behavior has to be assessed
it much more flexible than OSB2. The prestress was from the retrofit tests, for which the lateral load was higher.
slightly lower and was provided by unbonded bars. During
these cycles the beam did not develop many more cracks Retrofits
although there did appear to be a tendency for spalling of In contrast to the “as-built” OSB1 specimen, all the retrofit
the top cover. The major damage and load-reducing mech- specimens behaved very well, and any one of them would
anism came from spalling and subsequent buckling of the form the basis of an acceptable retrofit scheme for the Oak
reinforcement on the inside faces of the columns. The joint Street Bridge bents, although OSB2 was less adequate than
suffered no additional damage during these last cycles; the others. The prestress in OSB2 enhanced the shear
however, this was expected since the force levels were capacity by increasing the aggregate interlock effect. The
lower and the damage in the column was occurring well prestressing tendons also provided increased flexural
below the fillet and away from the joint. capacity, thus making up for the loss of negative moment
Recall that OSB2 suffered a column shear failure at a capacity through the debonding of the top steel. OSB2 devel-
displacement ductility of about 6. A possible explanation for oped a satisfactory lateral load capacity and a ductility of
the difference in the failure modes is that the strength and about 4, or a displacement of about 2 in. (50 mm). The load
stiffness of the fiberglass wrap specimen had been reduced remained stable through several cycles at each load level. At
by the time the fiberglass wrap was removed, and thus the the maximum load, however, the columns, which were not
applied column shear for comparable displacement was not retrofitted, were close to failure. At this load, the cap beam
as large as the shear in OSB2. Thus a shear failure did not had developed shear cracks in the same location as for OSB1
take place even though OSB5 suffered more flexural damage and a stable mechanism had formed with rotation at two loca-
in the columns. tions in the cap beam. At this stage the cracking became
significant, and in an actual structure would have been cause
for alarm. This indicates that when shear deficiency is
DISCUSSION AND CONCLUSIONS
important, it is worth considering a direct improvement in
As-Built Specimen OSB1 shear rather than the indirect improvement that is gained by
The performance of the as-built specimen OSB1 clearly post-tensioning, for example.
shows the need for seismic retrofitting of these types of two- If the column had been retrofitted, it is likely that the load
column bents. The failure of the test specimen was brittle. In would not have increased, but stable cyclic behavior would
addition, the lateral load capacity was about 60 percent of have been maintained. Increase in ductility demand might
that desired by the consultants responsible for the Oak Street have initiated brittle failure of the cap beam at the shear
Bridge retrofit, based on current BC Ministry requirements crack location. The energy dissipation would have been less
that use the AASHTO 19833 recommendations as the design than in subsequent retrofits because the behavior appeared to
basis. The bent demonstrated many of the effects of deficien- be dominated by elastic strain in the prestressing tendons in
cies typical of older concrete structures. Each of these defi- the region of the diagonal cracks.
ciencies has to be considered for retrofit.
The OSB3 test included a column retrofit and an under-
Analytical methods currently available are not very accurate beam post-tensioned to the cap beam. As part of the vertical
in predicting the shear capacity of concrete members such as post-tensioning, a steel plate member extended along the top
the cap beam when the amount of shear reinforcement is of the cap beam, providing additional moment capacity.
minimal as in this case. The OSB1 test helped to verify With both the cap beam and the columns improved, it is not
predictions based on the modified compression field theory.7 surprising that the failure occurred in the joint region.
The predicted lateral base shear at initiation of yielding in the Maximum load and displacement were highly satisfactory.
cap beam a distance equal to cap beam depth from the More significantly, cracks were small and widely distrib-
column was 51 kips (226 kN), controlled by yielding of uted. This would indicate that performance of the prototype
the upper longitudinal cap beam reinforcement due to in an earthquake would be better than OSB2. The behavior of
combined shear and flexure. The peak test value, after some the region immediately above the steel jackets is of some
yielding and cracking, was 60 kips (267 kN) at a lateral concern. The rotation at the joint is concentrated due to the
displacement of about 0.4 in. (10 mm). high flexural stiffness of the jacketed region. All the tests
The failure modes that were observed included the early with column jackets showed spalling and eventual bar buckling
and large shear cracks that developed in the expected locations, in the local region. This occurred, however, at very high
i.e., on a diagonal near each end of the cap beam, where ductility. OSB3 however, suffered from two real problems:
shear demand is high and shear reinforcement is inadequate. its appearance was less desirable due to the heavy under-
In addition, the top reinforcing steel or negative moment beam and the post-tensioned rods extending below the soffit
steel demonstrated bond failure. These bars were detailed and the likelihood of higher costs due to the extensive form
with numerous cutoffs and the lack of significant confining work needed for the underbeam.
steel from shear stirrups made it impossible for the top steel Although the capacity of the joints was of concern, it was
to develop. felt desirable to avoid retrofit of the joint region, as possible

112 ACI Structural Journal/March-April 1997


retrofit strategies for the joint would be unsightly and test, but was not considered a retrofit possibility for the Oak
expensive. Numerous joint regions in California bridges Street Bridge. The test has shown the potential of this type of
have been retrofit by complete removal of the joint concrete retrofit, and should provide encouragement to materials
and alteration of the reinforcing. This would require expen- researchers to investigate and overcome the drawbacks of
sive false work and access problems. this material.
Many of the advantages of OSB3 were characteristic of The test specimens were all constructed in the as-built
the OSB4 specimen. Longitudinal and vertical prestressing configuration first. Thus each of the retrofits was performed
of the cap beam in OSB4 provided vertical shear steel and on an existing test structure. This provided some insights
horizontal compressive stress for shear strength and into the retrofit process, which provided the designers of the
provided confinement for the horizontal flexural steel. The prototype retrofit with advance experience.
column jackets were again present. The cap beam retrofit
was able to force plastic hinging into the columns and so the ACKNOWLEDGMENTS
performance was very similar to that of OSB3. OSB4 This research program was sponsored by the British Columbia Ministry
suffered less joint damage than OSB3, and has the advantage of of Transportation and Highways. The cooperation with the Ministry's con-
sultants, Klohn Crippen International, Ltd., and their staff, especially Ole
eliminating the large underbeam.
Simonsen, Don Kennedy, and Dan Jennings, is gratefully acknowledged.
One of the more interesting results was the performance of Graduate students Markus Seethaler and Tony Cigic provided significant as-
OSB5, which was externally post-tensioned and then sistance to the project. Additional support was provided by the Canadian
wrapped with fiberglass cloth impregnated with epoxy. This Natural Sciences and Engineering Research Council.
is another example of direct shear improvement. The fiber-
glass was oriented perpendicular to the axis of the members CONVERSION FACTORS
on both cap beam and column, and hence was expected to 1 kip = 4. 448 kN
1 in. = 25.4 mm
offer some confinement but not contribute to flexural
1 ksi = 6.895 MPa
strength and stiffness. The specimen performed well up to
the displacement limits of the test system. In contrast to REFERENCES
OSB3 and OSB4, the fiberglass on OSB5 permitted well- 1. “Standard Specifications for Highway Bridges,” American Association of
distributed curvatures in the upper portions of the columns, State Highway and Transportation Officials, Washington, D.C., 1975.
contributing to the greater displacements. Since the yield 2. “Design of Highway Bridges (CAN/CSA-S6-1978),” Canadian
Standards Association, Rexdale, Ontario, 1978.
displacement was also increased, the ductility demand was 3. “Guide Specifications for Seismic Design of Highway Bridges,”
less than that for OSB3 and OSB4, but the plastic hinge rota- American Association of State Highway and Transportation Officials,
tions were approximately the same. The confinement Washington, D.C., 1983.
provided by the fiberglass contributed to maintaining the 4. “National Building Code of Canada 1985,” National Research Council
of Canada, Ottawa, 1985.
bond on the longitudinal bars, resulting in distributed cracks
5. Anderson, D. L.; Sexsmith, R. G.; English, D.; Kennedy, D.; and Jen-
and thus limiting crack widths. nings, D., “Oak Street and Queensborough Bridges Two-Column Bent
The fiberglass wrap on the prototype would require Tests,” University of British Columbia Earthquake Engineering Research
minimal drilling and related heavy construction work and Facility Technical Report 95-02, Vancouver, 1995, 52 pp.
minimal alteration to the appearance of the structure. Fiber- 6. Hexcel-Fyfe Co., “TYFO S Fiberwrap System,” Hexcel-Fyfe Co., Del
Mar, Calif., 1994.
glass does suffer from exposure to sunlight and vulnerability
7. Vecchio, F. J., and Collins, M. P., “The Modified Compression Field
to external damage. In this test program, the fiberglass wrap Theory for Reinforced Concrete Elements Subjected to Shear,” ACI
retrofit was considered as an extra, relatively inexpensive JOURNAL, Proceedings V. 83, No. 2, Mar.-Apr. 1986, pp. 219-231.

ACI Structural Journal/March-April 1997 113

You might also like