Connection of Simple-Span Precast Concrete Girders for Continuity (2004)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

B-1 APPENDIX B DETAILS OF THE EXPERIMENTAL PROGRAM GENERAL The experimental program consisted of two parts: con- nection capacity tests and full-size tests. The design concrete strength for girders was 4,000 psi at release and 5,500 psi at 28 days. Slabs and diaphragms had design 28-day strengths of 4,000 psi Actual concrete strengths are provided in Table B-1. All mild reinforcing bars were Grade 60, but the actual yield strength was 83 ksi. The prestressing strand was Grade 270 low relaxation. CONNECTION CAPACITY: STUB SPECIMEN TESTS Connection Details The stub specimens consisted of two 16-ft-long AASHTO Type II girders connected by a diaphragm (see Figure B-1). A total of six specimens were tested (see Chapter 2 for details). The original plan was to fabricate six stub girders and to cre- ate the six specimens by using both ends of the stubs. Two stubs would be used to create a specimen for a test, and then after the test, the stubs would be cut from the diaphragm. The stub girders could then be turned around, and the other ends used to create a second specimen. This was done for the first four specimens; however, the last two specimens had the girders embedded in the diaphragms, which made cutting them from the diaphragm impractical. Initially, a total of seven stub girders (six for use plus one spare) were fabricated. To make the final specimen, an additional stub beam was fabricated and used with the spare stub girder to make Specimen 6. The stub specimens were designed to simulate an interior girder in a multispan I-girder bridge. This bridge had 50-ft spans and used AASHTO Type II girders. The live-load inflec- tion points were approximately 15 ft from the diaphragm, so the stub specimens approximated the area between the live- load inflection points. All six specimens used a 10-in. gap between the ends of the girders (see Figure B-1). A 7.5-in.- thick by 8-ft-wide slab was cast on top of the girders. The diaphragms were 8 ft wide. The thickness of the dia- phragms varied from 10 in. to 26 in., depending on whether the girders were embedded in the diaphragm. Except for Specimen 5, the diaphragms had No. 5 stirrups placed 12 in. on center with the first stirrup 3 in. from the face of the bot- tom flange (see Figure B-1). Specimen 5 had two additional stirrups placed 3 in. on center, just outside of the bottom flange of the girder. Two types of positive moment connections were used: bent strand or bent bar. All connections were designed to have a nominal capacity of 1.2 Mcr. The cracking moment, Mcr, was calculated using the composite cross section but assuming the entire section was made from slab/diaphragm concrete. The cracking moment was determined to be 2,950 k-in., so 1.2 Mcr is 3,540 k-in. The bent-strand connection was made by extending the strand extend from the face of the girder and then bending it at a 90° angle. To determine the length of strand, the equations developed by Salmons and others (1–4) were used. These equations provide both an ultimate strength and a working stress. The connection was designed to have an ultimate strength capacity of 1.2 Mcr and a working stress capacity of Mcr. There are two variables: the area of the strand and length of the strand. For the stub specimens, the entire bottom layer of six strands was used. This provided an area of 0.918 in2. The original work by Salmons and others allowed for the use of horizontal bars attached to the web and perpendicular to the girder (see Chapter 2). If these horizontal bars are not used, the required working stress in the bent strand is given by fps (req’d) where M = positive moment, Aps = area of the bent strand crossing the connection joint, and jdps = distance between the compressive and tensile resultants. The required embedment length is then found from fp = (Le − 8.25 in.)/0.228 < 150 ksi, and Le = embedment length (in.). Ultimate strength is calculated by fpu = (Le − 8.25 in.)/0.163, a = (Aps fpu − As fy)/0.85fc′b, Mu = (φAps fpu [dps − a/2] + As fy [d − a/2]), b = width of compression face (rectangular section assumed), fc′ = compressive strength of concrete, and φ = understrength factor. = M A jdps ps

It was assumed that the working stress condition controlled and that the connection should be able to develop Mcr at working stress. For six strands, the cracked section has jd = 40 in. with Mcr = 243 k-ft, Le = 26 in. If an Le of 26 in. is then substituted into the ultimate strength equations, the capacity is found to be 315 k-ft, which exceeds 1.2 Mcr = 292 k-ft. For all calculations, the composite cross section is used, but it is assumed to be made entirely of diaphragm concrete. This is consistent with the nor- mal design assumption that failure occurs in the diaphragm concrete. The section is treated as a regular reinforced concrete section and uses 0.9 as the understrength factor. For the bent-bar connection, the composite cross section was designed as a regular reinforced concrete section with an ultimate capacity of 1.2 Mcr. Using the nominal concrete strength of 4 ksi and 60 ksi for yield of the steel, the section required five No. 5 bars. Experimental Set-Up Figure B-2 shows the experimental set up. For ease of understanding, the ends of the specimen were designated B-2 “east” and “west.” In Figure B-2, east is foreground. The spec- imen is held down at the center by a tie down device and is loaded by a yoke system at each end (see Figures B-3 and B-4). Load was applied through two −40 kip hydraulic cylin- ders. Load cells were attached to the rod of each cylinder. The system was controlled by servohydraulic valves and an electronic controller. The girders were supported by load cells at the center sup- port placed under the ends of the girders (see Figure B-5). According to data collected in a survey (see Chapter 2), this is the usual method of support. Direct current linear variable differential transformers (DCLVDTs) were attached to the girders and targeted to the diaphragms to measure crack open- ings (see Figure B-5). Crack openings were measured on both sides of the girders and on both faces of the diaphragm—a total of four joints. The DCLVDTs were placed at the mid- height of the top and bottom flanges and at the midheight of the web. Vibrating wire gages were placed in the diaphragm area, as shown in Figure B-6. Vibrating wire strain gages were placed 1.5 in. from the bottom of the diaphragm, at a distance of 3 in. and 12 in. from the outside face of the bottom flange of the girder. Additional external vibrating wire gages were attached to one side of each girder (see Figure B-5). The original idea was to lift both ends of the specimen, but this proved to be unstable. The same load effect couldFigure B-1. Stub specimen. TABLE B-1 Concrete compressive strength* (a) (b) Figure B-2. Stub experimental setup. Stub Specimens Concrete Compressive Strength psi (MPa) Girders 10,500 (73.5) Diaphragm/Slab 1 7,200 (50.4) Diaphragm/Slab 2 8,200 (57.3) Diaphragm/Slab 3 7,500 (52.5) Diaphragm/Slab 4 5,500 (38.4) Diaphragm/Slab 5 5,700 (39.9) Diaphragm/Slab 6 6,400 (44.8) Full-Size Girders 11,500 (80.5) Partial Diaphragm 1 @ 28 Days 5,300 (37.1) Partial Diaphragm 1 @ Test 5,500 (38.5) Slab 1 @ 28 Days 6,550 (45.9) Slab 1 @ Test 6,750 (47.3) Diaphragm/Slab 2 7,100 (49.7) *Strength is at time of test unless noted.

be obtained by loading the specimen as a cantilever (see Fig- ure B-2). One end was loaded while the other remained sta- tionary. A load cell placed under the stationary end verified that the loading was correct. The specimen was loaded to simulate a condition where the time dependent and the temperature effects created a positive moment of Mcr, and then traffic loads cycled the moment at the connection above or below Mcr. As a result, a cyclic load- ing pattern was developed in which the specimen was loaded between Mcr − MLL− and Mcr + MLL+ where MLL− and MLL+ are the positive and negative live-load moments at the connec- tion. From analysis, the maximum negative live-load moment was found to be MLL− = 365 k-ft, and the maximum positive live-load moment was determined to be MLL+ = 90 k-ft. All live-load moment values were at the face of the diaphragm. Mcr was calculated as 245 k-ft. Prior to the cyclic loading, the specimen was loaded to develop the negative live load moment and then the positive live-load moment at the joint. This loading was to establish the initial stiffness of connection and was repeated three times. To ensure consistency, the test was done twice. First the east end was loaded and the west end was kept station- ary, then the west end was loaded while the east end was kept stationary. B-3 Next, the specimen was loaded to develop Mcr at the joint to simulate the assumed time-dependent and temperature effects. This loading was repeated three times for each end by first applying three cycles to one end and then three cycles to the other end. After the initial loads had been applied, the load was cycled to develop moments between Mcr − MLL− and Mcr + MLL+. This represented an assumed worst-case loading for the connection. The servohydraulic controller applied the load and counted the cycles. The cyclic loading was to last for 1,000,000 cycles or to failure. The intent was to load the east side and keep the west side stationary for 100,000 cycles and then switch sides. How- ever, this was only done for Specimens 5 and 6 because they were the only ones to survive past 100,000 cycles. Load was applied at a rate of 2 hertz as long as the specimen would take the load at that rate. As the connection deteriorated, the cycle length had to be slowed down to ensure the correct loads were applied; however, this only occurred near failure, and the slowest loading rate was 1 hertz. The cyclic tests were stopped at various intervals, and the specimen was loaded under static loads to Mcr − MLL− and Mcr + MLL+. Even though only one side was being loaded under cyclic load, the static tests were performed twice, once on each side. Static tests were to be performed at 5,000; Figure B-3. Schematic of loading yoke.

10,000; 15,000; 20,000; 25,000; 30,000; 50,000; 75,000; and 100,000 cycles (if any test had continued beyond 200,000 cycles, static tests would have been conducted every 100,000 cycles). Failure of the specimen was accompanied by a catastrophic event such as the concrete on the bottom of the diaphragm B-4 spalling off or the specimen pulling out of the diaphragm. At this point, the specimen would not hold the required load under static testing. The stub girders from Specimens 1 and 3 were reused for Specimens 2 and 4. On both Specimens 1 and 3, failure was accompanied by the concrete spalling off the bottom of the diaphragm, exposing the positive moment bar or strand. This bar or strand could then be cut, if needed (in some cases the bar or strand was already broken). The deck was cut on both sides of the diaphragm, and the girders were removed from the diaphragm. The stub girders were turned end-for-end and reused in Specimens 2 and 4. FULL-SIZE SPECIMENS The full-size specimens were constructed of two 50-ft Type III AASHTO I girders joined with a 10-in. diaphragm. An 8-ft-wide × 7.5-in. thick composite concrete deck slab was cast on top (see Figure B-7). Each end of the girder spec- imens were cast with both bent-bar and extended strands so that either end could be used for either type of connection. When the girders were cast, shrinkage specimens were made. One was left with the girder to measure field shrink- age, and the other was used for the standard shrinkage test (AASHTO T160). At 180 days after the girders were fabri- cated, the field specimen showed a shrinkage of 710 micro- Figure B-4. Loading yoke. LOAD CELLS GIRDER VIBRATING WIRE STRAIN GAGE DCDT DIAPHRAGM Figure B-5. External instrumentation.

strain and the standard specimen shrank 570 microstrain. Cylinders were cast for creep tests. Two cylinders were used for the standard creep test. At 180 days after the girders were cast, the average creep coefficient (creep strain/elastic strain) was 2.4. Monitoring Specimen 1 Full-size Specimen 1 used a bent-bar connection with the girders not embedded in the diaphragm. The positive moment connection was designed as a reinforced concrete section using the composite cross section and the strength of the dia- phragm concrete. As with the stub specimens, the connection was designed for a capacity of 1.2 Mcr. Using the design strengths of the concrete and steel, it was determined that eight No. 5 bars, in two rows of four, were needed to provide a capacity of 1.2 Mcr = 500 k-ft (see Figure B-8). The bars had standard hooks. The diaphragm was 10-in. wide. As noted in Chapter 2, a partial diaphragm was cast for Spec- imen 1 when the girders were 28 days old. The remainder of the diaphragm and the slab was cast 28 days later. As with the gird- ers, both field and standard shrinkage specimens were made for the slab. At 120 days after the slab was cast, the shrinkage was approximately 560 microstrain for both specimens. After the slab was cast, the specimen was monitored for approximately 120 days. Girder reactions were monitored by six vibrating wire strain gage–type load cells. These load B-5 cells are very stable and do not drift. Three load cells were placed under each girder, two at the end near the diaphragm and one at the far end. Two vibrating wire strain gages were embedded in each girder at midspan. A vibrating wire strain gage was embed- ded in the slab at midspan directly above the girders (see Fig- ure B-9). A vibrating wire strain gage was also embedded 6 in. from the end of each girder, 4 in. from the bottom. Vibrat- ing wire strain gages and bonded foil strain gages were placed in the diaphragm, as shown in Figure B-10. Another vibrat- ing wire strain gage was placed 1.5 in. from the bottom of the diaphragm at a distance of 3 in. from the outside edge of the girder bottom flange. DCLVDTs were installed to measure crack openings. There were four joints, one on each side of the girder on each side of the diaphragm. Three DCLVDTs were placed on each joint at 3 in., 23 in., and 42 in. from the bottom of the girder. The bottom DCLVDTs were placed 1 day after the partial diaphragm was poured by drilling holes in the formwork to allow the DCLVDTs to be targeted on the diaphragm. These DCLVDTs were removed for a few hours when the forms were removed and then reinstalled. The DCLVDT can never be reinstalled to exactly the same reading as before removal, so several data points were taken before and after removal. The difference between these two readings was used to cor- rect future readings. Deflections were monitored by wire potentiometers placed at midspan of the girders. After the Figure B-6. Placement of instruments in the diaphragm of the stub specimen.

B-6 Figure B-7. Full-size composite section. Figure B-8. Full-size bent bar specimen.

forms were removed, two additional wire potentiometers were attached to the diaphragm. The instruments were monitored by an electronic data acquisition system, except for the foil gages. The system did not have the ability to monitor the foil gages and the foil gages tend to drift, so any data obtained would have been questionable. The gages were read every hour. The tests were done at an outdoor facility located at Prestress Services in B-7 Melbourne, Kentucky. The facility is approximately 16 miles from the University of Cincinnati campus. Because of the distance, data was collected every 3 to 4 days. Although the computers were furnished with battery backups and surge protectors, there were still occasions in which the research team would arrive to collect data and find that one or more of the computers had reset and that the data were lost. This is why some gaps appear in the data in Chapter 2. Figure B-9. Location of vibrating wire gages at midspan. Figure B-10. Location of vibrating wire gages in the full-size specimen diaphragm.

After monitoring the first specimen for 120 days after the slab was cast, the specimen was tested for continuity. These tests are described in a following section. Full-Size Specimen 2 Full-size Specimen 2 was a bent-strand connection with the girder ends not embedded in the diaphragm. The strand length was kept at 26 in. to match the stub girder specimens. To achieve a capacity of 1.2 Mcr = 500 k-ft, 10 strands were needed (see Figure B-11). The number of strands was deter- mined from the same equations used for determining strand capacity for the stub girder specimens. The second full-size specimen was constructed by cut- ting the diaphragm out of the first full-size specimen and turning the girders end-for-end. When the deck slab was cast for the first full-size specimen, approximately 12 ft of slab had not been cast at the end of each girder. When the girders were turned end-for-end, there was a gap in the slab over the diaphragm area. When the diaphragm for Speci- men 2 was cast, a deck slab was cast in this gap to provide continuity. Reinforcing bars had extended from the end of the existing slab to tie the new slab to the old slab. Full-size Specimen 2 was not monitored, but was tested for continu- ity when the diaphragm and deck slab were 28 days old. Instru- mentation for full-size Specimen 2 was the same as for full- size Specimen 1. B-8 Testing the Full-Size-Specimens for Continuity One purpose of the full-size specimens was to determine whether cracking at the diaphragm affected continuity. The plan was to apply the cracking moment to connection and then to apply external loads. In an effort to duplicate the kind of deformation that would occur due to creep and shrinkage of the girders and would crack the diaphragm, a post-tensioning system was used. A post-tensioning bar was placed through the bottom flange of the girder, and a post-tensioning force was applied (see Fig- ures B-12 and B-13). This would cause the girders to camber up—simulating the formation of additional positive moment. Note that the post-tensioning bars only went through the girders and did not go through the diaphragm (i.e., the bar was dead-ended at the end of the girder where the girder was attached to the diaphragm). To determine the size and placement of the post-tensioning bar, each span was modeled as a propped cantilever (see Figure B-14) with a constant internal moment applied. Note that for each 1 k-ft of internal moment, the moment at the connection increases by 1.5 k-ft. To generate the cracking moment of 417 k-ft, an internal moment of 278 k-ft was required. This could be achieved with a 1.375-in.-diameter Dywidag bar placed as shown in Figure B-12 and tensioned to the maximum allowable force of 160 k. Four 40-kip-capacity hydraulic cylinders were used to sim- ulate the live load. Two cylinders were placed in each span at 22.5 ft from the face of the diaphragm (see Figures B-15 Figure B-11. Full-size bent strand specimen.

B-9 and B-16) because analysis showed that this was the most efficient position for the load. The loading sequence was as follows: 1. Span 1 was loaded, 2. Span 2 was loaded, 3. Span 1 was unloaded, and 4. Span 2 was unloaded. The upward loads created positive moment, and downward loads created negative moment. The general steps of testing for both full-size specimens were as follows: 1. The load was applied such that when both spans were loaded, the negative live-load moment (365 k-ft) was developed at the connection. 2. The load was applied such that when both spans were loaded, the positive live-load moment (90 k-ft) was developed at the connection. Figure B-12. Positioning of post-tensioning duct in the full-size specimen. Figure B-13. End of Specimen 1 showing extended bar, extended strand, and post-tensioning bar. Figure B-14. Model for developing moment from post-tensioning moment.

B-10 3. One-quarter of the post tensioning force (40 kips) was applied to the girders; the girders were post-tensioned one at a time. 4. Steps 1 through 3 were repeated until all the post- tensioning force had been applied. 5. When all the required post-tensioning force has been applied, the specimen was loaded to the negative and positive live-load moments. 6. If the girders still appeared to be continuous, the loading system was used to apply additional positive moment until the specimen failed or the system lifted off the supports. This sequence was followed for both full-size specimens. For full-size Specimen 2, additional positive moment was gener- ated by jacking up and shimming the ends of the specimen to increase the end reaction (see Chapter 2). REFERENCES FOR APPENDIX B 1. Salmons, J.R., and McCrate, T.E. Bond of Untensioned Pre- stress Strand. Interim Report 73-5A, Missouri Cooperative High- way Research Program, Missouri State Highway Department, August 1973. 2. Salmons, J.R., and May, G.W. Strand Reinforcing for End Con- nections of Pretensioned I-Beam Bridges. Interim Report 73-5B, Missouri Cooperative Highway Research Program, Missouri State Highway Department, May 1974. 3. Salmons, J.R. End Connections of Pretensioned I-Beam Bridges. Final Report 73-5C, Missouri Cooperative Highway Research Program, Missouri State Highway Department, Nov 1974. 4. Salmons, J.R. Behavior of Untensioned-Bonded Prestressing Strand. Final Report 77-1, Missouri Cooperative Highway Research Program, Missouri State Highway Department, June 1980. Figure B-15. Full-size specimen with loading cylinders. Figure B-16. Overall photograph of the full-size specimen.