Documente Academic
Documente Profesional
Documente Cultură
Abstract: A new concept has been developed for connecting spread footings and precast columns in bridges. The socket connection is con-
structed by precasting the column, erecting it, and casting the reinforced concrete footing around it. This system saves construction time on site
Downloaded from ascelibrary.org by New York University on 05/15/15. Copyright ASCE. For personal use only; all rights reserved.
because, in little more than the time needed to construct the footing, both the column and footing can be constructed. Site erection is facilitated
by the fact that the field tolerances are essentially unlimited. The longitudinal column bars are straight and are terminated with mechanical
anchors. This arrangement improves constructability, because no bars cross the interface between the column and footing, and it provides better
transfer of forces in the connection region than is possible with conventional bent-out longitudinal bars. The surface of the column is roughened
to improve adhesion to the surrounding cast-in-place concrete. Axial-load tests demonstrated that the connection can resist column axial loads
far above those expected in practice. Cyclic, lateral-load tests demonstrated that the seismic performance of the connection is at least as good as,
if not better than, that of a comparable cast-in-place system. The recent deployment of the new system in a highway overpass provided both field
experience and initial estimates of the potential time savings. DOI: 10.1061/(ASCE)BE.1943-5592.0000413. © 2013 American Society of
Civil Engineers.
CE Database subject headings: Bridges; Columns; Connections; Precast concrete; Footings; Seismic effects; Cast in place.
Author keywords: Bridges; Accelerated bridge construction; Connection; Column; Spread footing; Precast concrete; Cast-in-place; Seismic.
Introduction In contrast, precast concrete has seldom been used for bridge
substructures, and then only in states with low seismicity. For ex-
Extensive construction will be needed to replace numerous func- ample, Matsumoto et al. (2001, 2008) tested four precast column-to-
tionally or structurally obsolete bridges in the United States (FHWA bent-cap connections suitable for use by the Texas DOT. In regions of
2008). If such construction is performed following conventional moderate or high seismicity, connections between precast concrete
practices, existing traffic congestion will increase further, because components must satisfy a number of challenging design criteria. In
cast-in-place concrete construction requires time-consuming shor- addition to being easy to construct rapidly, a column-to-footing
ing, form construction, reinforcement placement, and concrete connection must be able to transfer to the footing both the gravity
casting and curing. This congestion leads to extensive traffic delays, loads from the bridge weight and traffic, and lateral forces. To fa-
wasted fuel, and corresponding increases in CO2 emissions. New cilitate column fabrication and transportation, connections are typi-
structural systems and construction methods are needed that disrupt cally located at beam-to-column and column-to-footing interfaces.
society less. But during an earthquake, these are the locations that experience the
One solution for reducing construction-related delays is to pre- largest moments. Therefore, such connections must accommodate
cast structural elements off site and then assemble them rapidly on large column inelastic deformations without loss of stiffness and
site. This approach, which has been used for bridge girders for over strength, while simultaneously resisting vertical loads caused by
50 years, also improves construction quality (through plant pre- a combination of gravity and seismic overturning effects. Because
casting) and worker safety (through reduced time on site). The foundations are difficult to inspect and repair, these load transfers
precast, prestressed concrete girder bridge has now become the most must be accomplished without significant damage to the footing.
common type of bridge in most states. Recent research on accelerated bridge construction has focused
on the development of precast concrete bridge substructures for
seismic regions, as discussed by Marsh et al. (2011) and Weinert
1
Research Assistant, Dept. of Civil and Environmental Engineering, (2011). For example, Pang et al. (2008, 2010) tested three variations
Univ. of Washington, Seattle, WA 98195-2700 (corresponding author). of a large-bar, large-duct beam-to-column connection, in which a few
E-mail: olafurh@uw.edu large bars [e.g., D44 or D57 (#14 or #18)] are grouted into large
2
Research Assistant, Dept. of Civil and Environmental Engineering, ducts [e.g., 152 mm (6 in.)] (Steuck et al. 2007, 2009). This connection
Univ. of Washington, Seattle, WA 98195-2700. is easy to assemble, because it provides generous construction tol-
3
Professor, Dept. of Civil and Environmental Engineering, Univ. of erances. Matsumoto (2009) describes three tests of column-to-beam
Washington, Seattle, WA 98195-2700. connections, including one connection with grouted ducts and two
4
Professor, Dept. of Civil and Environmental Engineering, Univ. of with pocket connections. The pocket connection concept is to precast
Washington, Seattle, WA 98195-2700.
the cap-beam with a void created by a corrugated pipe, which is then
Note. This manuscript was submitted on January 24, 2012; approved
on July 17, 2012; published online on August 15, 2013. Discussion period filled with site-cast concrete to connect the cap-beam and column.
open until February 1, 2014; separate discussions must be submitted for Tobolski (2010) describes tests of a hybrid system that is intended to
individual papers. This paper is part of the Journal of Bridge Engineer- minimize residual displacements. The dual-shell system combines
ing, Vol. 18, No. 9, September 1, 2013. ©ASCE, ISSN 1084-0702/2013/ two steel shells, cast-in-place concrete, and posttensioning. Mild
9-910–919/$25.00. reinforcing bars project from the top of the column and are grouted
plumbed, leveled, and braced. After the footing reinforcement is presented in Table 1. Specimen SF-1 represented a precast equiv-
placed around the column, the foundation is cast. alent of a typical Washington DOT (WSDOT) cast-in-place column-
The structural details of the proposed connection differ from to-footing connection, with a socket embedment depth of 1.1 times
those of a conventional, cast-in-place system in two important ways: the column diameter. Specimen SF-2 had the same overall geometry,
(1) no bars cross the column-footing interface, so the resistance to but the reinforcement was reduced and simplified to facilitate con-
vertical loads relies on shear friction across the roughened interface, struction. These first two tests revealed that the embedment depth
and (2) the longitudinal column bars are not bent out at the bottom, as ratio of 1.1 was sufficient to protect the footing by promoting failure
is common in cast-in-place bridges. Instead, these bars are straight in the column, as is required in bridge design. However, because the
and terminated with headed anchors. This choice simplifies trans- connection region was not damaged, the tests only provided a lower
portation and handling, and reduces the hazards posed by protruding bound on the capacity of the load transfer mechanism. To better
bars. The configuration also provides a more direct transfer of in- understand that mechanism, Specimen SF-3 was designed with
ternal forces than is possible with bent-out bars, as illustrated by the a thinner footing to force the connection to fail within the footing.
strut-and-tie model shown in Fig. 2. With the headed bar detail, the
diagonal strut force within the column is transferred to the column Design of Test Specimens
bars via a CCC node, which connects three compression struts, so the
transfer depends on direct bearing. By contrast, with the conven- The key dimensions and reinforcement of the three test column-
tional bent-out bar detail, the force must be transferred by bond to the footing subassemblies (SF-1, SF-2, and SF-3) are shown in Fig. 3.
curved part of the bar. With a few exceptions discussed later, the three specimens were
designed to confirm to the AASHTO Load Resistant Factor Design
Specifications (AASHTO 2009b), the AASHTO Guide Specifications
for LRFD Seismic Design (AASHTO 2009a), the WSDOT Bridge Design Manual (WSDOT 2008) for the ends of prestressed girders.
Design Manual (WSDOT 2008), and the California DOT (CalTrans) This roughness detail satisfied the AASHTO LRFD (AASHTO
Seismic Design Criteria (CalTrans 2006). 2009b) requirement for surface-transferring shear friction in terms of
All three subassemblies had identical 508-mm (20-in.)-diameter minimum amplitude. In this test program the roughened region of the
columns, with a height from top of the footing to the loading point column had an octagonal cross-section (inscribed inside a 508-mm
of 1.524 m (60 in.), resulting in a flexure-critical column with a circle) to facilitate forming the saw-tooth with wood strips, as shown
cantilever span/depth of 3.0. To facilitate comparison with previous in Fig. 4. The circular cross-section used in the main body of the
research tests [i.e., Pang et al. (2008, 2010)], this ratio was in- column could be used in the foundation as well, provided that
tentionally not matched to the expected point of inflection of the a simple way of forming the ridges could be developed.
prototype column, which would have resulted in a span/depth of 2.5 The footing depth of 1.372 m (4.5 ft) in the prototype bridge was
at specimen scale. Each column was embedded in a cast-in-place controlled by the design goal of resistingpffiffiffithe
ffi one-waypffiffiffiffi
shear in the
footing [2.286-m (90-in.) long and 1.727-m (68-in.) wide]. This concrete alone, using a stress of 0:166 f 9c ðMPaÞ ½2 f 9c ðpsiÞ. At
geometry represented, at 42% scale, a 1.219-m (48-in.)-diameter scale, this depth corresponded to 0.572 m (22.5 in.) in specimens
column embedded in a 5:486-m ð18-ftÞ 3 5:486-m ð18-ftÞ cast-in- SF-1 and SF-2. Top and bottom flexural reinforcement was con-
place footing. In the tests, the footing width had to be reduced trolled by the minimum reinforcement requirement. Specimen SF-1
slightly below the scaled value [to 1.727 m, (68 in.)] to fit in the test was designed to replicate as closely as possible in precast concrete
machine. the details commonly used in cast-in-place systems. In it, slots were
The column longitudinal reinforcement consisted of eight D19 formed in the bottom of the column in each direction to allow some
bars, corresponding to a longitudinal reinforcement ratio, r, of of the bottom footing reinforcement to pass through and engage
1.12%. The transverse reinforcement consisted of a smooth wire directly with the tension steel in the column (Fig. 5). In addition,
spiral of 6.2-mm (0.24-in.) diameter at a pitch of 32 mm (1.25 in.), three sets of four diagonal bars (12-D13) were stacked, in the bottom,
resulting in a transverse reinforcement ratio, rs , of 0.88%. The part in the horizontal plane around the column to provide the normal
of the column that was embedded in the cast-in-place footing was force needed to induce shear friction resistance to the column
intentionally roughened using the saw-tooth pattern, scaled down to pushing through the footing. One set of four diagonal bars was
42% of the dimensions of the prototype [18-mm (0.71-in.) amplitude placed in the top mat for trimming purposes. Footing ties were
at 36-mm (1.41-in.) spacing], required by the WSDOT Bridge included to just satisfy the CalTrans criterion (CalTrans 2006). The
friction resistance. Last, the footing ties required by the CalTrans included in SF-3, in the top.
(2006) criteria were reduced by half, because strut-and-tie modeling
suggested that these ties are necessary only if bent-out column bars
are used (Xiao et al. 1996). The ties were placed in the locations Testing Protocol and Test Setup
where they would most effectively engage with 45 compressive
struts. All three subassemblies were first loaded with an axial load. SF-2
was subjected to an axial load of 1,068 kN (240 kips), which at
42% scale, represented the factored axial dead and live loads on the
connection (1:25DL 1 1:75LL). Specimens SF-1 and SF-3 were
subjected to higher axial loads (4,083 kN [918 kips] for SF-1, and
1,521 kN [342 kips] for SF-3), which corresponded, respectively, to
3.8 times and 1.4 times the factored gravity axial loads.
The axial load was then reduced to the unfactored dead load value
of 707 kN (159 kips), and the specimen was subjected to displacement-
controlled cyclic lateral loading. The displacement history was a
modification of the loading history for precast structural walls rec-
ommended in the National Earthquake Hazards Reduction Program
(BSSC 2004). The two specimens that suffered no footing failure
during the lateral-loading sequence were then loaded axially to failure.
The three specimens were tested using a loading rig in which the
vertical load was applied by a 10.7 MN (2,400 kip) Baldwin uni-
versal testing machine, acting through a sliding spherical bearing.
The horizontal load was applied by a servocontrolled ram connected
to a steel reaction frame. The specimens were anchored down to the
base of the test machine, as shown in Fig. 6, but a gap of 1.5 mm
(1/16 in.) was left between the load cell and the nut above it. This
precaution was taken in case the columns proved to be unexpectedly
strong, and the lateral strength of the system was controlled by
overturning rather than column strength. In all specimens, a void was
maintained beneath the column so as not to preclude punching
Fig. 4. Embedded octagonal cross-section
failure, and a displacement sensor monitored any vertical movement
of the column.
Fig. 5. Base detail comparison of specimens (a) SF-1 and (b) SF-2
Fig. 7. Comparison of specimens’ drift ratios for the major damage states
Moment-Drift Response
the lateral-load tests. In contrast, Specimen SF-3 suffered significant
damage in the footing. Cracks in the sides of it were first observed at The measured moment-drift ratio responses of the three sub-
a drift ratio of 0.3%. Radial cracks on its surface began to propagate assemblies are plotted in Fig. 9. The responses of SF-1 and SF-2 are
from the column early in the test, and spalling of the footing started nearly indistinguishable. This similarity is consistent with the ob-
at 7% drift. During the last large cycle (maximum drift ratio 5 10.7%) servation that the damage in these tests occurred only in the columns,
the column punched through the footing as the column was returning which were nominally the same. The responses of these two
from the peak displacement. specimens were stable to about 6% drift, at which point the lateral-
After the cyclic lateral-load tests were completed, SF-1 and SF-2 load resistance dropped rapidly. The resistance of SF-3 began to
were loaded axially to failure. In both specimens, failure occurred by decay earlier (starting at about 2.5% drift) but the decay was more
crushing of the concrete in the already damaged column plastic hinge gradual. All specimens maintained a strength of at least 80% of the
region, with no sign of damage to the connection region. Specimens peak resistance up to 7% drift.
SF-1 and SF-2 sustained maximum axial loads of 3,745 kN (842 The maximum measured base moments in both specimens SF-1
kips) and 3,634 kN (819 kips), respectively. These values corre- and SF-2 were approximately 350 kN×m (3,100 kip-in.). In con-
sponded to about 3.5 times the factored gravity load (1:25DL 1 trast, Specimen SF-3 was approximately 10% stronger (383 kN×m).
1:75LL). This final axial load test was not performed on Specimen Material tests showed that all the column longitudinal bars were
a
OT 5 overturning.
virtually identical, but the concrete in Specimen SF-3 was about 50%
stronger than that in specimens SF-1 and SF-2. A moment-curvature
analysis of the specimens indicated that the stronger concrete should
be expected to cause only a 3–4% increase in flexural strength. The
source of the remaining 5–6% increase in the column strength is not
known.
calculated using measured material properties, with resistance the demand exceeded the capacity for punching shear but not for
factors (f) taken as 1.0. Three sets of loads were used. In the first four shear friction. Last, joint shear in the footing was evaluated using the
rows of the table, the measured quantity, such as Mu , is based on procedure given in the AASHTO Guide Specifications (AASHTO
the loads measured during the lateral load testing. In rows 5 and 6, 2009a). It was not critical in any of the specimens.
the punching and push-through shear demands, Vu , are based on the
maximum vertical load ever applied [4,083, 3,645, and 1,521 kN
(918, 820, and 342 kips), respectively]. In rows 7 and 8, the joint Constructability
shear is based on moment-curvature analyses using the measured
material properties. That procedure was necessary because all of the The socket connection was developed to be used with a precast
bar forces were needed but strains were measured in only two. column and cast-in-place footing. This connection, together with the
The flexural demands in the footings were computed at the col- large-bar, large-duct column-to-beam connection (Pang et al. 2008),
umn face. The cracking moment capacity was calculated using the were incorporated in the construction of the bridge carrying US 12
whole width, but the nominal moment capacity was calculated using over Interstate 5 in Washington State. That demonstration project
the effective width, beff , as specified by the AASHTO Guide provided an opportunity for verifying the constructability of the
Specifications (AASHTO 2009a). The demands were approximately precast bent system.
10% below the calculated cracking capacities for the two thick The bridge had two spans, tall abutments at the ends, and a four-
footings (SF-1 and SF-2), but they greatly exceeded the nominal column bent at the center. Although the columns were small enough
cracking moment for the thin footing (SF-3), as expected. These to be handled in one piece, they were constructed in segments to
findings are consistent with the observed damage patterns. The evaluate that technology for use on future projects with larger col-
moments experienced in all of the footings were below their nominal umns. Fig. 11(a) shows a column base segment plumbed and leveled
flexural strengths. In specimens SF-1 and SF-2 the flexural steel was in a foundation, with only the bottom flexural reinforcement in place.
controlled by minimum requirements, so the ratio was expected to be The steel framing was used as a guide. Fig. 11(b) shows the
less than 1.0. Specimen SF-3 was controlled by strength require- placement of the precast cross beam.
ments, but the ratio is still less than 1.0 because of the 1.2 over- The construction of the socket connections was successful. The
strength factor required by AASHTO (2009b), the f-factor, and the contractor placed the bottom footing steel before installing the
rounding up of bar sizes. column, but that was out of choice and not necessity. No major
As required by the AASHTO Guide Specifications (AASHTO problems were encountered, and the minor ones that did occur were
2009a), the one-way shear demand was evaluated at the column face, associated with grouting the column segments, which are not an
and the shear concrete capacity was calculated using the effective essential part of the system. A precast cap beam was placed on the
width and included the stirrup contribution. The footing shear precast columns, and the large bars and ducts made alignment
demands were only 43–75% of the nominal capacities. straightforward. Even though this was the first time that the socket
concept described here had been used, placing and plumbing the
column base segments took less than an hour per column. Given the
Table 3. Nominal Demand/Capacity Ratios connection’s great strength in resisting vertical loads, it is expected
that the cap beam could be placed very soon after casting the foot-
Design criterion SF-1 SF-2 SF-3
ings. It is likely that the precast column system described here will
Footing cracking, Mu =Mcra
0.91 0.89 3.97 prove most advantageous in multicolumn bents, especially if site
Footing flexural strength, Mu =Mn a 0.57 0.57 0.75 access is limited. In another bridge project, with 14 columns, the
Footing one-way shear strength, Vu =Vn a 0.43 0.56 0.75 contractor requested a change from cast-in-place to precast socket
Punching shear and moment transfer, 0.20 0.25 1.27 columns to accelerate the construction process.
Vu =Vn b
Pure punching shear strength, Vu =Vn b 1.19 1.35 1.82
Shear-friction push-through strength, 1.44 1.90 0.74 Summary and Conclusions
Vu =Vn b
c =pc
limit c
Joint shear, (compressive) pcalc 0.23 0.21 0.32 A new column-to-footing socket connection, suitable for precast
Joint shear, (tensile) pt =pt
calc limit c
0.24 0.22 0.44 bridge bents in seismic regions, has been developed. The connection
a concept is to precast the column and to cast the spread footing
Based on loads measured during lateral load testing.
b around it, with no bars crossing the interface between the two. The
Based on maximum load ever applied.
c connection’s performance in the cyclic, lateral loading tests was
Based on moment curvature analysis using the measured material
properties. excellent and is believed to be at least as good as, if not better than,
Fig. 11. Deployment of precast system over I-5 in Washington State: (a) socket connection; (b) placement of cross-beam
that of a conventional cast-in-place system. The basis for this as- When the longitudinal column steel consists of straight, headed
sertion is that, in the new system, the critical node in the strut and tie bars rather than the conventional bent-out bars, the prescriptive
model is a CCC node, at which the forces are transferred by direct vertical footing ties specified by the California DOT Seismic Design
bearing, whereas the force transfer occurs by bond in the traditional Criteria (CalTrans 2006) perform no useful function and can be
bent-out bar system and implicitly depends on the tension strength of omitted. This conclusion applies only to the prescriptive ties and not
the concrete. These model results are supported by test evidence that to ties that are needed to resist the computed shear demands.
shows that diagonal joint shear cracking and failure occurs with The bottom diagonal bars, included in the thick footing test
bent-out bars [e.g., Saunders et al. (1995)], and that footing ties are specimens (SF-1 and SF-2) and intended to provide shear friction
needed to resist it (Xiao et al. 1996), whereas in the new system no resistance, experienced low stresses during the factored axial-load
diagonal cracks occurred and essentially no force was recorded in the test and during the cyclic lateral-load tests, during which the foot-
ties. Under vertical load, the capacity of the new system capacity was ing remained essentially uncracked and elastic. They therefore
much greater than the expected demands. proved unnecessary and can also be eliminated.
The constructability advantages of the system are that it is quick
and simple to build, and the column is easy to transport, because no
bars project from the bottom or sides of the column. This configu- Acknowledgments
ration has several construction advantages over other systems that
use precast columns: This project was supported by the Federal Highway Administration
• It avoids any potential problems of fit-up of bars in sleeves or (FHWA) (Grant No. DTFH61-09-G-00005), and by the Valle Foun-
ducts at the base of the column; dation of the University of Washington. This work would not have
• It allows generous placement tolerances; been possible without valuable assistance of Dr. Bijan Khaleghi,
• The column detailing can be almost identical to that of a cast in Mr. Jugesh Kapur, and Mr. Eric Schultz of the WSDOT, Dr. Lee
place column; and Marsh of Berger/ABAM Engineers, Mr. Steve Seguirant of Con-
• No grouting is needed. crete Technology Corporation, and the staff of Tri-State Construc-
The socket connection was incorporated successfully into the tion. The support of all of them is gratefully acknowledged.
construction of a bridge in Washington State.
In the lateral-loading tests, the response of all three specimens
was essentially identical to that of comparable cast-in-place sub- Notation
assemblies. The precast columns can therefore be designed fol-
lowing the same specifications used to design cast-in-place columns. The following symbols are used in this paper:
The longitudinal bar strains directly behind the anchor heads Ag 5 gross area of column cross section;
showed that the stress there reached 20% of yield in the two speci- fc9 5 specified compressive strength of concrete;
mens with the thicker footings (SF-1 and SF-2), and exceeded yield Paxial 5 applied axial load;
in Specimen SF-3, with its thinner footing. The heads therefore r 5 longitudinal reinforcement ratio; and
provided at least partial anchorage and should be included in future rs 5 volumetric ratio of transverse reinforcement.
implementations of the socket connection.
The shear friction strength of the connection under pure axial
load, using the shear key roughness detail of typical WSDOT precast References
girder ends (WSDOT 2008), was sufficient to prevent any slip in any
of the three test specimens. In specimens SF-1 and SF-2, in which AASHTO. (2009a). AASHTO guide specification for LRFD seismic bridge
the footing was relatively undamaged after the lateral load test pro- design (2009 interim revisions), Washington, DC.
AASHTO. (2009b). LRFD bridge design specifications, 4th Ed., Wash-
gram, the connection interface withstood a shear stress of 4.21 MPa
ington, DC.
(611 psi) with no sign of damage even after the specimens had American Concrete Institute (ACI). (2008). “Building code requirements for
been subjected to large lateral deformations. Specimen SF-3 reached structural concrete and commentary.” ACI 318-08, Farmington Hills,
its predicted capacity during the cyclic test, but failed in combined MI.
punching shear and moment transfer at the end of the test. This Berry, M. P., and Eberhard, M. O. (2003). “Performance models for
mode of failure is not considered in AASHTO (2009a, b). flexural damage in reinforced concrete columns.” Rep. 2003/18, Pacific