Figure 26: Effect of internal and external post-tensioning on performance of test units. Figure 27: Phase II test unit. Figure 20: Cross Section of test units. Figure 24: Load versus vertical displacements of units with internally bonded tendons. Figure 22: Mid-span joint after failure of Test Unit 1001NTCIP Figure 21: Test specimen and loading fram. Figure 23: Mid-span joint after failure of Test Unit 1001EXT Figure 25: Load versus vertical displacements of units with different ratios of internal to external post-tensioning. Figure 26: Effect of internal and external post-tensioning on performance of test units.

The first phase of the seismic research project conducted at the University of California San Diego (UCSD) on precast segmental bridges has been completed. This research program was initiated by the American Segmental Bridge Institute (ASBI) and funded by Caltrans. The objective of Phase I of the project was to study the seismic performance of segment-to-segment joints in precast segmental bridge superstructures with different ratios of internal to external post-tensioning. Phase I focused on joints subjected to high bending moments combined with low shears. Four large-scale units with different ratios of internal to external post-tensioning were tested at UCSD; the test matrix is given in Table 1 (see page 5). Each test unit consisted of six precast segments which were epoxy bonded. Test Units 1 and 2 used 100 percent internal post-tensioning (bonded tendons). Test Unit 3 was similar to Unit 1 but it was post-tensioned with external tendons only. Unit 4 was also similar to Unit 1, except that half of the post-tensioning was achieved by external tendons and the remaining half by internally bonded tendons (see Table 1). Test Units 1 to 4 are designated as 100INT, 100INTCIP, 100EXT, and 50INT50EXT, respectively. The terms 100INT, 100EXT, and 50INT50EXT, in test unit designation indicate the ratio of internal to external post-tensioning. In the second test unit, CIP in unit designation stands for the cast-in-place deck closure joint. In Units 100INT, 100EXT, and 50INT50EXT, no mild reinforcement were present across the segment-to-segment joints, whereas Test Unit 100INTCIP had reinforced cast-in-place deck closures at the location of segment-to-segment joints. The remaining portion of the joints in Unit 100INTCIP, along the web and bottom flange, were epoxy bonded. The reinforced cast-in-place deck detail across each joint in the 100INTCIP specimen was similar to the detail proposed for the new East Bay Skyway Structure of the San Francisco Oakland Bay Bridge. Only one half of a single cell box girder in the shape of an equivalent I-section was modeled at 2/3-scale in the experiments to simplify the test setup (Fig. 20). Figure 21 shows the load frame and a typical test unit simply supported at both ends. The test units were loaded by four vertical hydraulic actuators as shown in Fig. 2 in two testing stages. The first testing stage represented service load conditioning, whereas the second testing stage represented vertical seismic loading of the superstructure. In the seismic testing stage each unit was subjected to fully reversed cyclic displacements with increasing amplitude up to failure. The forces in all the four actuators were maintained equal at all times.

Under downward loading of all test units, the first crack occurred at the mid-span joint in the concrete cover adjacent to the epoxy bonded joint. This was followed by cracking at the two joints adjacent to the mid-span joint of Test Units 100INT, 100INTCIP and 50INT50EXT which had internally bonded tendons. However, only the mid-span joint opened during downward loading of Unit 100EXT with external tendons. In the upward loading direction, cracking occurred only at the mid-span joint in Units 100INT, 100EXT, and 50INT50EXT, whereas several closely spaced cracks developed inside the segments of Unit 100INTCIP because of the continuity of the deck at joint locations. Unit 100INT failed when the prestressing strands fractured at the mid-span joint. Failure of Unit 100INTCIP initiated by buckling of the cast-in-place deck reinforcing bars which was followed, after repeated displacement cycles, by compression failure of the deck at mid-span. Figure 22 shows the mid-span joint of Unit 100INTCIP after failure. Failure of Test Unit 100EXT (with external tendons) initiated by crushing of the deck at the mid-span joint. In contrast to the explosive failure of Units 100INT and 100INTCIP, failure of Unit 100EXT occurred gradually. The load carrying capacity dropped gradually with increased displacement in the post-peak range. The test was terminated when a relatively high displacement was reached. Figure 23 shows the mid-span joint of Unit 100EXT after termination of the experiment. Finally, Unit 50INT50EXT failed at a relatively low displacement when the internal tendon ruptured (as for Unit 100INT).

Figure 24 shows the history of total applied load versus vertical displacement measured at 6 inches from mid-span for Test Units 100INT and 100INTCIP; the variable of these two test units was presence of the mild reinforcement in the deck across the segment-to-segment joints. Figure 25 is similar to Figure 24 but it shows the load versus displacement for Units 100INT, 100EXT and 50INT50EXT with the test variable being the ratio of internal to external post-tensioning. Sign convention in Figs. 5 and 6 is positive for downward loading and displacement. The reference load level indicated by the horizontal solid lines in Figs. 24, 25, and 26 represents the total load on each test unit before application of the fully reversed cyclic vertical displacements. This load was selected to represent the total dead load on the prototype structure. The performance of Units 100INT and 100INTCIP was similar in the downward loading direction as shown in Fig. 24. However, the performance of the two test units was substantially different under upward loading. Unit 100INT did not have any continuous deck reinforcement crossing the joints; thus once the mid-span joint opened there was no other mechanism to carry the load and there was a large drop in the applied upward load. The yield strength of the deck reinforcing bars in Unit 100INTCIP could be developed resulting in a maximum total upward load of 327 kip, rather than 93 kip for Unit 100INT. Figure 26 also shows the enhancement in energy dissipation with the cast-in-place deck joints in Unit 100INTCIP.

As indicated by Fig. 25 and 26(a), the failure load of specimen 100EXT was less than the failure loads of specimens 100INT or 100INTCIP which had an equal number of bonded internal tendons, or the failure load of specimen 50INT50EXT where the total number of tendons was also equal. This difference in behavior is routinely considered in the AASHTO design process for static loads which requires a larger number of unbonded tendons to provide capacity equivalent to members with bonded tendons. The capacity of all four specimens in this series was substantially greater than the gravity load capacity required to meet AASHTO load factor or load and resistance factor bridge design specifications. From the perspective of seismic performance, the ductility and maximum displacement reached before failure are substantially improved by use of 100 percent external post-tensioning, as indicated by Fig. 26(a). Structures designed for seismic loading with 100 percent external tendons would have tendons near the top of the section to resist load reversals at mid-span. Unit 50INT50EXT with combined internal and external post-tensioning had a load carrying capacity that falls nearly halfway in between that of Units 100INT and 100EXT. The force in the internal tendon of Unit 50INT50EXT was higher than the force in the external ones as evidenced from the recorded tendon strains. At the same displacement levels of the test units, the measured strains in the internal tendon in Unit 50INT50EXT were higher than the strains in the internal tendons of Units 100INT and 100INTCIP with 100 percent internal post-tensioning. Thus, the internal tendon in Unit 50INT50EXT failed at a relatively small displacement compared to the other units. Thus in terms of seismic performance of superstructure segment-to-segment joints, combination of internal and external post-tensioning is not recommended. The maximum applied load, Vu, and the maximum displacement before failure, Δu, are given in Table 1 (see page 5) for all test units. Failure displacements, Δu, given in Table 1 (see page 5) are indicated by the solid circles in Fig. 26(a). Figure 26(b) shows the load versus displacement curve during the downward loading portion of the 3 in. displacement cycle of units 100INT, 100EXT and 50INT50EXT. The displacement measured in the unloading portion of any of the curves shown in Fig. 26(b) at the reference load level represents the permanent residual displacement, Δr, after earthquake occurrence. Values of Δr measured after 3 in. maximum displacement of all test units are also given in Table 1 (see page 5). Comparison of the curves shown in Fig. 26(b) indicates that residual displacements can be minimized by use of 100 percent external post-tensioning. This is because the strains in external tendons are significantly less than the corresponding strains in internal bonded tendons. Inelastic strains in internally bonded tendons result in loss of the prestressing force, large permanent displacements and joint openings.

Phase II of the project is currently in progress at UCSD. This second phase focuses on the seismic performance of segment-to-segment joints subjected to high bending moments combined with high shears. Four units will be tested in Phase II; the test units are similar to those of Phase I (Table 1). Figure 27 shows an elevation view of one test unit. Only one joint will be studied in these experiments. The test unit is supported on a concrete footing as shown in Fig. 27. The test units were designed again at a 2/3-scale of the same prototype structure used for Phase I tests with the only difference in the layout of prestressing tendons. The prototype structure of Phase II is post-tensioned by a horizontal tendon close to the bottom soffit of the superstructure in addition to a harped-shape tendon. Both the horizontal and the harped-shape tendons will be modeled in the experiments (Fig. 27). A steel beam will be attached to the end of the concrete test unit. Two vertical hydraulic actuators will be used to apply vertical loads on the steel beam. Each test unit will be subjected to fully reversed cyclic vertical displacements up to failure.

Table 1. Test Matrix and Test Results
Test Unit No.	Unit Designation	Test Description	Vu′ kip (kN)	Δu′ in. (mm)	Δu′ in.* (mm)
1	100INT	100% Internal Post-Tensioning	490 (2180)	4.8 (122)	1.17 (30)
2	100INTCIP	100% Internal Post-Tensioning and Cast-in-Place deck joint	480 (2135)	5.9 (150)	0.53 (13)
3	100EXT	100% External Post-Tensioning	417 (1855)	6.6 (168)	0.14 (4)
4	50INT50EXT	50% Internal + 50% External Post-Tensioning	451 (2006)	3.8 (97)	0.82 (21)

* Measured during the 3 in. displacement cycle.