Content uploaded by Islam Mantawy
Author content
All content in this area was uploaded by Islam Mantawy on Mar 29, 2016
Content may be subject to copyright.
Earthquake Shake Table Testing of a Self-centering ABC Bridge
Islam Mantawy, Research Assistant, University of Nevada Reno, (775)537-9019, imantawy@unr.edu
Travis Thonstad, Research Assistant, University of Washington, (503)347-1723, thonstat@uw.edu
David Sanders, Professor, University of Nevada Reno, (775)784-4288, sanders@unr.edu
John Stanton, Professor, University of Washington, (206)543-6057, stanton@u.washington.edu
Marc Eberhard, Professor, University of Washington, (206)543-4815, eberhard@u.washington.edu
ABSTRACT
This paper describes the verification by shake table testing of a bridge bent system that was designed to
be rapidly constructible and to provide superior seismic performance through re-centering and reduction of
damage. The system used precast concrete elements, and the re-centering was achieved by means of
unbonded pre-tensioning in the columns. Column damage was suppressed by a steel rocking detail that
confined the ends of the columns. The development of the system concept and its initial quasi-static testing
are described in a separate paper. A two-span, three-bent bridge was tested seismically on the shake tables
at the NEES Facility at the University of Nevada, Reno. The bridge was quarter scale, had two-column
bents with 12 in. (0.305 m) diameter columns, and 30 ft. (9.14 m) span lengths. The overall bridge geometry
was similar to that of one previously tested at the University of Nevada, Reno that used conventional non-
prestressed, cast-in-place concrete columns. This paper describes the test bridge, the loading procedures
and the preliminary experimental results. Comparisons are made with the previously tested conventional
bridge. During a 1.66 g PGA earthquake excitation, numerous longitudinal bars fractured at drift ratio
between 10.2 and 13.2%. Minimal architectural spalling occurred due to column geometric change above
the steel rocking details. The residual drift was about 2.5% from the maximum drift at 1.66 g PGA
earthquake excitation.
INTRODUCTION
Within the United States, design of reinforced concrete bridges in seismic regions has changed little since
the mid-1970s, when ductile details were first introduced. Many bridge bents in seismic regions are
constructed of cast-in-place reinforced concrete. Cast-in-place bridges with proper confinement have
performed well in the past, but to meet modern design expectations for bridges, new structural systems and
construction methods are needed to improve: 1) speed of construction, 2) seismic resilience and 3)
durability.
The new system that was originally developed at the University of Washington has the following key
features: 1) columns and beams are cast off-site and then assembled rapidly once they arrive on site, 2)
construction is further accelerated by using a “wet socket” connection between the column and the spread
footing [Haraldsson et al 2013(1)] and a “hybrid-bar-socket” connection between the column and the precast
beams [Davis et al 2014(2)], 3) post-earthquake residual displacements are reduced by pre-tensioning the
precast bridge columns with unbonded tendons, which are designed to return the system to its original
position when the ground motion stops [Eberhard et al 2014(3)], and 4) damage to the system is minimized
by incorporating a confined rocking detail, or “shoe”, at the ends [Eberhard et al 2014(3)].
BRIDGE SPECIMEN
The shake-table specimen was designed to investigate the global response of the pre-tensioned, rocking
bent system. The bridge geometry, illustrated in Figure 1 and 2, was chosen to match that a specimen
previously tested at University of Nevada, Reno [Johnson et al 2008(4), Johnson et al 2006(5)] that used
conventional non-prestressed, cast-in-place concrete columns. The bridge specimen was quarter scale
with octagon columns ended by steel shoes. Bent dimensions and column reinforcement details of the
shake-table specimen are shown in Figures 3 and 4. The bridge length was 69.25 ft. (21.11m); The clear
height of the specimen was 127 in. (3.21 m); the total superimposed weight on the bridge was 170.2 kips
(757.4 kN ).
Figure 1. Shake table specimen
Figure 2. Overall dimension of the shake-table specimen.
Figure 3. Bent dimensions.
Figure 4. Typical top and bottom reinforcement details of the columns
SPECIMEN DESIGN
Column Design
The column clear heights for each bent, from the top of the footing to the bottom of the bent caps, varied.
This matched the column heights of the previous bridge experiment tested that used conventional non-
prestressed columns. Clear heights of 6 ft. (1.83 m), 8 ft. (2.44 m), and 5ft (1.52 m) were used for bents 1,
2 and 3 respectively. The column bases were embedded 22 in. (0.56 m) inside the cast-in-place combined
footing using a wet socket connection [Haraldsson et al 2013(1)]. The tops of the column were integrally
grouted into the superstructure using a hybrid-bar-socket connection [Davis et al 2014(2)]. The column
longitudinal reinforcement consisted of 6~#3 (10 mm) bars and 4~3/8 in. (10 mm) diameter epoxy coated
prestressing strands. The longitudinal reinforcement was debonded at each rocking interface over sufficient
length to prevent bar failure at designed deformation. The strands were bonded in the footing and bent cap
and unbonded through the column clear height. The concrete at the column-to-footing and the column-to-
cap-beam connections were confined by a steel rocking detail, which consisted of a circular steel tube
welded to an annular end plate. The end plate was intended to concentrate column rotations at the two
interfaces, creating rigid body rotation of the columns in between. Supplementary reinforcing was welded
to the end plate, extending into the clear height of the column to distribute compressive forces and arrest
cracks that could form at the boundary of the steel confining tube.
Superstructure Design
The superstructure of the bridge consisted of six precast slabs post-tensioned together to provide a stiff
deck. The slabs has been designed for the conventionally designed bridge and were reused in this bridge.
Each span, consisting of three slabs, was assembled on the lab floor and post-tensioned transversely using
ten 1.25” (32 mm) diameter rods. Each rod was prestressed to 100 kips (445 kN) of force, to rigidly connect
the precast slabs preventing slippage and providing flexural capacity in the transverse direction. Each set
of beams was placed between the bents and aligned with longitudinal post-tension ducts embedded into
the three precast caps. The deck was longitudinally post-tensioned to a total force of 720 kips (3204 kN) to
provide rigid connections between the bent caps and slabs.
SPECIMEN CONSTRUCTION
The key stages of bridge construction are shown in Figure 5. Six columns and three bent caps were cast
at the University of Washington and shipped to the University of Nevada, Reno. The columns were aligned
in the footing formwork and the spread footings were cast in place in an outdoor staging area. The footing
and columns were then moved as a single piece onto the shake tables. Spacer blocks were used between
the bottom of the footings and the shake tables to maintain a level superstructure due to the variations in
column heights. Two different types of grout were used to connect the cap beams to the columns. A fiber
reinforced grout pad was used at the column-to-cap-beam interface to allow for the adjustment of the cap
elevation and level. A standard, non-fiber reinforced, grout was used to connect the column’s reduced
section and longitudinal reinforcing to the bent cap. The placement of the non-fiber grout was postponed
until after the longitudinal post-tensioning was completed to reduce secondary moments in the columns
due to slab shortening.
After the bent and spacer blocks were aligned, grouted and vertically bolted to the shake tables the
preassembled spans were supported on formwork between each bent cap. Hydrostone was placed
between the bent caps and the slabs, and the spans were lowered onto the bent cap ledges. The post-
tensioning was conducted in stages to allow the placement of the superimposed mass. Eight concrete
blocks with a total weight 160 kips (712 kN) and 10.2 kips (45.4 kN) of steel plates were placed on the
superstructure to provide a representative structural mass, scaled from the superstructure of the prototype
bridge.
INSTRUMENTATION
The bridge was instrumented with 395 channels to record accelerations, displacements, bar/strand strains,
and change in the strand forces using load cells. A summary of the instrumentation plan is shown in Table
1. Transverse, longitudinal, and vertical accelerations of the superstructure at each bent and mid spans
were measured using accelerometers. Superstructure displacement and column curvatures were
measured using displacement transducers. The strain level in the longitudinal reinforcement, transverse
reinforcement, longitudinal strands and steel shoe within critical column sections were measured with strain
gauges. Potential slippage of strands at top of the columns was measured using load cells.
Figure 5. Photographs of the construction phases of the shake table specimen: (A) Precast columns at
University of Washington; (B) Fiber grout between column and bent cap; (C) Non-fiber grout between
column reduced section and bent cap; (D) Superimposed masses.
Table 1. Instrumentation Summary
Recorded response Count
Potentiometers
Slab displacements (T, L, V) 25
Column curvatures 72
Table displacements (T, L) 6
Accelerometers
Slab accelerations (T, L, V) 15
Table accelerations (T, L) 6
Table velocities (T, L) 6
Strain gauges
Longitudinal reinforcement strain 165
Transverse reinforcement strain 24
Strand strain 41
Steel shoe (Rosette) 6
Load cell
Strand load cell 23
Actuator load cell 6
EXPERIMENT SCHEDULE
Both low-amplitude and high-amplitude earthquake excitations were used to investigate the bridge
response. A summary of the experiment schedule is shown in Table 2. The excitations were based on the
90 deg. and 360 deg. components of the Century City Country Club North (CCN90/CCN360) record from
the 1994 Northridge California Earthquake, the 360 deg. component of the Sylmar- Olive View Med. Center
(SYL360) record from the 1994 Northridge California Earthquake and the 0 deg. component of the Takatori
(TAK000) record from the 1995 Kobe, Japan Earthquake. Low-amplitude motions consisted of coherent,
incoherent and biaxial motions, whereas high-amplitude motions consisted of only coherent motions in the
transverse direction of the bridge due to the absence of abutments. White-noise and square wave
excitations were distributed throughout testing to track the bridge properties including the bridge periods
and damping. Sinusoidal waves were added to evaluate the dynamic response of the bridge subjected to
harmonic motions. Because of the one-quarter geometric scale, the time coordinate of the input was
multiplied by a factor of 0.5.
Since the main objective of this study was to compare the response of the precast, pre-tensioned bridge
bent system with the conventional cast-in-place bridge previously tested at the University of Nevada, Reno,
a majority of the motions used were the same as in the previous experiment. Preliminary OpenSees models
were used to evaluate the effects of adding Sylmar, Takatori motions and sinusoidal motions to the test
schedule. The intent was to add these motions without altering the system performance during later motions
that were comparable to the previous experiment. The final motion schedule eliminated some low-amplitude
motions from the previous experiment and added Sylmar and Takatori motions at high-amplitude motions.
To investigate the bridge behavior with different excitations including near fault motions, Sylmar and
Takatori motions were added after motion 14. The acceleration histories were scaled to have similar
structural demands to the Century City motion.
Table 2. Experiment Schedule with motion description.
Test Test Type Description Test Test Type Description
1A Low Level
Coherent
Motion
CCN90 (0.08g PGA) S4
S5
Sinusoidal Motion 0.15g 0.30sec
0.10g 0.30sec
1B
CCN90 (0.15g PGA)
4
5
6
Low Level
incoherent
Motion
CCN90
(0.07g-0.18g-0.18g)
CCN90
(0.18g -0.07g-0.18g)
14B1
14B2
14C
15
16
17
18
19
20A
20B
21A
21B
21C
High Level coherent
Motion
SYL360 (0.20g PGA)
SYL360 (0.40g PGA)
TAK000 (0.20g PGA)
CCN90 (0.5g PGA)
CCN90 (0.75g PGA)
CCN90 (1.00g PGA)
CCN90 (1.33 g PGA)
CCN90 (1.66 g PGA)
CCN90 (0.75g PGA)
SYL360 (0.843g PGA)
TAK000 (0.40g PGA)
TAK000 (0.611g PGA)
TAK000 (0.80g PGA)
CCN90
(0.18g -0.18g-0.07g)
9A Biaxial
Motion
CCN90/CCN360
(0.08g PGA)
9B CCN90/CCN360
(0.15g PGA)
12
High Level
coherent
Motion
CCN90 (0.08g PGA)
13 CCN90 (0.15g PGA)
14A CCN90 (0.25g PGA)
S1
Sinusoidal
Motion
0.05g 0.25sec
S2 0.10g 0.25sec
S3 0.15g 0.25sec
9C Biaxial
Motion CCN90/CCN360
(0.25g PGA)
OBSERVED DAMAGE
No damage was observed in the columns or the superstructure during the low-amplitude motions. Similarly,
no damage (Concrete cracking, slippage of the longitudinal post-tensioning and cracking the non- fiber
grout) was observed in the superstructure during the high-amplitude motions.
The first yield of the longitudinal reinforcement occurred during Motion 13. The first rebar fracture occurred
at bent 1 during Motion 17, at a maximum column drift ratio of 5.7%. Spalling of the column concrete first
occurred above the steel shoe at the change in column geometry during Motion 16. Bulging of the steel
confining tube occurred in Bent 1 and Bent 3 during Motion 18. Multiple rebar fractures and grout pad loss
occurred during Motion 18 after exceeding drift ratios of 9% and 6% for bents 1 and 3 respectively, and
during Motion 19, after exceeding drift ratios of 11% and 13% for bents 1 and 3 respectively. Figure 6
shows the damage progression of the column concrete and steel shoe for bent 1 at the end of Motion 19.
Figure 6. Damage progression for bent 1 at bottom of each column: (A) spalling at bottom of south
column of bent 1 at motion 16; (B) flaking at bottom of north column of bent 1 at motion 17; (C) shoe
bulge at bottom of south column of bent 1 at motion 19.
MEASURED RESULTS
The bridge induced low displacement/drift levels during the low-amplitude motions; while during the high-
amplitude the displacement/drift level was high (maximum drift was 13.2% for Bent 3). The maximum
residual drift was 0.2% for Bent 3 during Motion 19. The bridge was subjected to three design-level
earthquakes after the highest motion. Motion 19 was equivalent to 2.2 times the design earthquake. The
bridge showed superior resistance for these motions with maximum residual drift equals to 0.1%. Due to
the bar fracture after Motion 19, the bridge period was shifted which allowed the bridge to resist the motions
from 20A to 21C.
COMPARISON WITH PREVIOUS BRIDGE EXPERIMENT
One of main purposes of the experiment was to compare the performance of the new system with one
previously tested at the University of Nevada, Reno that used conventional non-prestressed columns. The
results show that the new system produced the same displacement/drift ratios up to Motion 16. Starting
from Motion 17, the new system produced higher displacements than the conventional bridge. The new
system had a maximum drift of 13.20% for Bent 3, while the previous bridge had 8.00%. The residual
displacements for the new system were much lower than the conventional system. The new system had
0.2% for Bent 3 for Motion 19, while the pervious bridge had 0.5%. Figures 8 and 9 show the comparison
between the new design and the conventional design, for both maximum drift ratio and residual drift ratio
of Bent 3. The researchers are currently evaluating other possible reasons for differences in the response;
reasons could include difference in the shake table response, changes in bridge period, and/or the
differences in the cyclic, force-deformation characteristics of the bridges.
The new system incurred less damage than the conventional bridge. Figure 7 shows the comparison
between the damage at end of Motion 19 for both specimens. The new system experienced minimal
spalling and rebar fracture, whereas the conventional bridge sustained total failure of Bent 3 including
excessive spalling, spiral fractures and bar buckling. After Motion 19, equivalent 2.2 times the design
earthquake, the specimen continued to resist lateral forces and showed excellent re-centering. The
maximum residual drift ratio for these motions was less than 0.3%. This is in contrast to the previous
experiment where, after Motion 19, the superimposed mass was removed from bent 3 due to concerns of
collapse, allowing the experiment to continue.
A) New Bridge System B) Conventional Bridge [Johnson et al., 2006(5)]
Figure 7. Damage comparison between the new bridge system and conventional bridge
Figure 8. Maximum drift ratio comparison between the new system and conventional bridge for bent 3
Figure 9. Residual drift ratio comparison between the new system and conventional bridge for bent 3
*Note: Since motions were added to the conventional loading protocol for the new system, straight lines
are used for the results of conventional bridge
CONCLUSIONS
A new bridge system has been developed for use in any seismic region. It accelerates bridge construction,
it re-centers after extreme earthquakes, and it minimizes seismic damage.
1. Damage is minimized by rocking, confinement details.
2. Re-centering is achieved by pre-tensioned strands
3. Compared to the conventional bridge, the new bridge induced less observed damage with no
exposure of column reinforcement occurring during any test.
4. Compared to a conventional bridge, the peak transverse displacements for the new bridge system
were higher on average when subjected to a high-amplitude ground motion, while the new bridge
system demonstrated lower residual displacements for all experiments.
ACKNOWLEDGMENTS
This research is supported by the National Science Foundation George Brown Network for Earthquake
Engineering Simulation Research Program Award No. 1207903.
The shake table test was conducted at University of Nevada, Reno. The writers are indebted to the
dedicated support of Patrick Laplace, Chad Lyttle, Todd Lyttle and Paul Lucas of the Earthquake
Engineering Laboratory at the NEES Site at the University of Nevada, Reno. The dedicated help from
undergraduate students Taylor Nielsen, Osvaldo Arias, Guillermo Munoz, Mimi Mungedi, Lisa Bryant and
REU student Eric Ramirez.
REFERENCES
(1) Haraldsson, O.S., Janes, T.M., Eberhard, M.O., and Stanton, J.F. (2013). “Seismic Resistance of
Socket Connection between Footing and Precast Column.” Journal of Bridge Engineering.
10.1061/(ASCE)BE.1943-5592.0000413 (September 2013)
(2) Davis P.M., Janes T.M., Eberhard M.O., Stanton J.F., Haraldsson, O.S. (2014). “Unbonded pre-
tensioned columns for accelerated bridge construction in seismic regions”, ASCE, J. Bridge Eng.,
submitted for publication.
(3) Eberhard, M., Stanton, J., Sanders, D., Schaefer, J., Kennedy, B., Thonstad, T., Harldsson, O. and
Mantawy, I., (2014) “Shaking Table Tests on a New Bridge System Designed to Re-Center.” Proc. of
10th U.S. National Conference on Earthquake Engineering: Frontiers of Earthquake Engineering,
Anchorage, AK. July 21-25.
(4) Johnson, N., Ranf R., Saiidi, M., Sanders, D., Eberhard, M, (2008)“Seismic Testing of a Two-Span
Reinforced Concrete Bridge,” Journal of Bridge Engineering, ASCE March/April 2008, pp 173-182
(5) Johnson, N., Saiidi, M., and Sanders D., (2006)“Large-Scale Experimental and Analytical Studies of a
Two-Span Reinforced Concrete Bridge System,” Center for Civil Engineering Earthquake Research,
Department of Civil Engineering, University of Nevada, Reno, Nevada, Report No. CCEER-06-02,
March 2006
