Worsening traffic congestion and public impatience with delays have made accelerated bridge construction (ABC) a necessity.
Particularly in urban environments, closing traffic lanes for extended periods to construct a bridge is no longer acceptable. Therefore, owners, engineers and planners need to find ways of constructing bridges that shorten the time needed on-site.
Some approaches to ABC have been based on administrative methods and include innovative financial incentives, while others depend on new construction methods, such as prefabrication, to reduce the site time. In extreme cases, bridge superstructures are now constructed near the site and are transported to their final location and erected there in a single piece using self-propelled modular transporters (SPMTs). This technique is, of course, possible only if a convenient construction site is available close to the bridge location, and such conditions are rare in a dense urban environment.
The work described here leads to a method of constructing bridge bents that makes use of precast concrete components and is intended to be applied to common bridge types, such as freeway overpasses. However, prefabrication implies connections that must be made on-site. For ease of fabrication and transportation, straight elements (e.g., beams and columns rather than cruciform pieces) are preferable, but the site connections must then be made at the beam-column and column-foundation intersections. This construction constraint leads to a potential structural problem when the bridge is located in a seismic zone; the largest bending moments and shears, and the site of any potential inelastic behavior, lie exactly where the connection is to be made. The connection must then be sufficiently robust to resist the forces and deformations imposed by earthquake loading, but also it must be easy to assemble and have generous allowances for site adjustment. This combination of demands presents significant design and construction challenges.
The bridge bent concept described here was developed jointly by a multidisciplinary team consisting of personnel from the University of Washington, the Washington State Department of Transportation (WSDOT) and BergerABAM Engineers, in consultation with general contractors and precast concrete fabricators. The competing structural- and construction-related demands were therefore resolved in an integrated fashion.
The system is illustrated in an exploded view (Figure 1). Its goal is to allow a bent to be erected rapidly, so that girders can start to be set in the shortest possible time. It is shown here with a drop bent cap on which the girders are to be set. That system is widely used in Washington state, but the proposed system could be adapted for other cap-beam systems.
First the footing is excavated and, if needed, a temporary concrete pad is cast in its center. Next, the footing reinforcement is placed, the precast column is erected on the pad in the excavation and the concrete is cast in place around the column. No reinforcement crosses the interface between the cast-in-place footing and the precast column, but the vertical steel projects from the top of the column. The precast cap beam has ducts that match the projecting column bars, and it is placed on the column and the ducts are then grouted. At this stage the girders can be set, and construction proceeds as with a conventional bridge built with precast, prestressed concrete girders and a CIP deck.
The new features of the system are the column connections to the footing and cap beam.
The footing connection is referred to as a “socket” connection. The major questions associated with its performance focus on its ability to resist vertical gravity load and cyclic lateral load. At the cap beam, the questions are associated with the strength of the connection under cyclic lateral loading. Because the footing and cap beam are both typically much stiffer than the column, the column itself behaves, under lateral load, as if it were fixed against rotation at top and bottom, with an inflection point at mid-height. Lateral load testing could therefore be conducted using cantilever specimens in which the height of the test column was half that of the prototype.
Creating better footing
A number of design challenges arise for the column-to-footing connection. For ease of fabrication and transportation, the column should be straight and have no bent bars projecting into the footing. Thus, to achieve anchorage, headed bars were used. Their use causes a flow of forces in the footing that differs from that in a traditional cast-in-place system. In the latter, structural criteria call for the bars to be bent inward. This can be seen in Figure 2 by comparing the footing to a beam-column joint in a building frame, turned through 90º, so the beam in the frame corresponds to the footing in the bridge bent.
In the frame joint, the beam steel must be bent inward in order to achieve good force transfer at the nodes in the strut-and-tie mechanism. The same is true for the bridge footing, but the bars in a circular column cannot be conveniently bent inward for lack of space, and the column cage would not be stable while the footing was being cast. Thus, the compromise in CIP systems is to bend the bars outward for constructability and to compensate in other ways for the structural shortcomings of the internal load path. The compensation consists of providing top steel and ties in the footing. With straight, headed bars, both of these are theoretically unnecessary and the connection performance should be expected to be better than with bent-out bars. Thus the use of straight headed bars appears to provide both structural- and construction-related benefits.
Two other design issues arise. The first is whether the column might push through the footing by sliding along the precast/cast-in-place interface. The column surface was roughened in the region, with the goal of providing shear friction resistance to any potential sliding. Reinforcement is then needed to clamp the footing against the column, and sets of four bars, arranged in a square but at 45º to the main footing steel, were included in the design. The need for them was not clear, because an argument can be made that the main footing steel can provide both flexural and shear-friction resistance. Thus the need for separate shear-friction steel was included as a variable in the testing program.
Second, if the CIP footing is cast around the PC column, no footing steel can pass under the column, but rather must be moved to a location just beside the column, where it can be bundled with other bars already there. The question of whether steel in that location would engage properly with the column steel to transfer moments from the column to the footing also was studied during the test program.
To investigate the structural properties, tests were conducted at the University of Washington on three column-footing subassemblies at approximately 42% scale. The cantilever columns were loaded vertically to simulate gravity and simultaneously subjected to a series of ever-increasing cyclic lateral displacements to study resistance to idealized seismic loading. The footings were cast with a lip around the bottom so that all the support occurred at the perimeter, and the column was free to move downward if the shear-friction resistance to gravity load proved inadequate.
In the first two spread-footing specimens (SF-1 and SF-2), the footing thickness (22 in.) was approximately the same as the column diameter (20 in.). This thickness was chosen because it was the smallest for which the footing concrete could resist the one-way shear forces without tie reinforcement. The reinforcement in the two specimens also was different, with SF-1 being as close as possible to code-compliant and containing the full complement of shear friction, steel top steel and footing ties, as shown in Figure 3. Furthermore, NS and EW slots were formed in the bottom of the column through which some of the bottom footing steel could pass in order to improve the moment transfer between elements.
Specimen SF-2 represented a less conservative design that contained only half the quantity of footing ties and one quarter of the shear-friction steel. The column also contained no slots, so all of the footing reinforcement that would otherwise have fallen under the column was moved to its sides.
The third specimen, SF-3, was built with a 10-in.-deep footing, with the goal of forcing failure into the footing connection region.
In SF-1 and SF-2, the lateral loading eventually caused failure in the column, as shown in Figure 4. In both cases, the footing remained uncracked, and strain-gauge readings showed peak bar stresses in the diagonal shear-friction steel and the footing ties of approximately 5% of yield. These low strains demonstrated that those two types of reinforcement were not necessary. The load-deflection plots were essentially identical to that of a reference cast-in-place specimen tested in an earlier study. The implication is that the improved force flow permitted by the bar heads is effective in reducing joint shear demands, and the footing, even with the lighter reinforcing used in SF-2, is not the weak link. Rather, the column properties control the system strength. This is exactly the desired outcome.
Specimen SF-3 contained much more flexural steel to compensate for the smaller footing depth. During the first part of the loading program, it behaved like the other two specimens, with damage concentrated in the column. However, at 10% drift, it finally failed by punching through the footing as a result of combined vertical load and bending of the connection. That failure provided information on the strength of the connection, which allows the formulation of design recommendations.
Specimens SF-1 and SF-2 also were tested under pure vertical load. This was done after the column had failed in bending during the lateral load tests, and the spiral steel had fractured. The goal was to determine the shear-friction resistance to vertical load at the precast/cast-in-place interface. In both cases the applied load reached 3.5 times the factored DL + LL that would occur in the prototype bridge, scaled to lab dimensions. At that load, the column crushed extensively and lost strength, but there was no sign of cracking in the footing or slip of the column through the footing. This result represents a very large margin of safety, and suggests that the shear-friction mechanism is easily adequate to transfer the load from the column to the surrounding footing if the footing thickness is equal to the column diameter.
Setting the cap
The column-to-cap beam connection consists of bars projecting from the column and grouted into ducts in the cap beam, as shown in Figure 1. This concept has been used before, but direct translation from a cast-in-place to a precast design leads to the use of a large number of bars and ducts, with the attendant risks of alignment problems on-site because the ducts have to be quite small. The approach adopted here was to facilitate site assembly by using a small number of large bars in large ducts. The system was tested (and proven) to work with six No. 18 bars, each in an 8-in.-diam. duct in a 4-ft-diam. column. Input from contractors confirmed that those dimensions would allow easy assembly.
The structural questions raised by such a detail concern the anchorage of the large bars in the precast cap beam, which is normally approximately 3 ft 6 in. deep and nominally too shallow to permit full anchorage of a No. 18 bar.
The testing was conducted in two stages. In the first, monotonic pullout tests were conducted of bars grouted into ducts. These were at full scale and used bar sizes from No. 8 to No. 18. The measured values showed that anchorage of bars in ducts occurs in a length that is much shorter than that specified for development of the same bar in concrete and that any commercially available bar can easily be anchored in a grouted duct within the depth of a typical cap beam.
Lateral load tests were then conducted, at 42% scale, of a column-to-cap beam connection that used the large grouted bar concept. The longitudinal bars were No. 8, which, when scaled up to prototype size, are slightly larger than No. 18. Three variants on the detail were tested, with different amounts of deliberate debonding of the bar to relieve high local strains. In all cases, the connection behaved essentially identically to the cast-in-place reference specimen tested earlier, with failure occurring in the column and only superficial flaking of the concrete of the cap beam.
A Grand entrance
The system was implemented on-site in the summer of 2011 in the replacement of the I-5 Grand Mound to Maytown Interchange, U.S. Rte. 12 over I-5, Br. No. 12/118 Replacement, near Olympia Wash. Figure 5 shows a column being placed in the footing, and the completed bridge is shown on p. 36. In it, two additional features were introduced: The columns were cast segmentally and the cap beam was made in two parts that were joined by a cast-in-place closure pour. Segmental columns were selected, not because they were needed, but because the demonstration project offered the chance to try the technology for future use in bridges with heavier columns. The central pier was a four-column bent, so the cap beam was quite long and heavy, and fabricating it as two pieces reduced the maximum pick weight.
Construction proved relatively straightforward, thanks largely to careful planning by WSDOT and close coordination with the contractor. The most notable problems occurred with grouting the column segments. That step is not an integral part of the system, and the problems were easily resolved.
The use in a precast concrete bridge bent of a socket connection at the footing and a large-bar grouted duct connection at the cap beam provides a design concept that allows rapid assembly on-site and provides excellent resistance to cyclic lateral loads such as earthquakes. Laboratory testing demonstrated the good structural properties, and site implementation confirmed that it was easy to construct. R&B