Integral Steel Box-Beam Pier Caps (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.

7CHAPTER 2 FINDINGS 2.1 STATE-OF-THE-ART SUMMARY Two different approaches to collecting data on the state of the art of integral connections were studied. First, the state of practice was studied through the questionnaire on past use of integral connections. Second, the state of research was stud- ied through the literature search. The summary of the results of these studies is presented below. The details of the state of practice and the literature search are presented in Appendixes A and B, respectively (which are included on the accompa- nying CD-ROM). 2.1.1 State of Practice The questionnaire on past use of integral connections was sent to all AASHTO voting and nonvoting members (i.e., DOTs in all states and Canadian provinces, and quasi- governmental authorities such as turnpike authorities). In addition, the questionnaire was slightly modified and sent to domestic researchers and bridge designers and inter- national bridge designers. The modifications were intended to allow the person responding to comment on the perfor- mance of bridges he or she did not design but is familiar with, provide contact information on the designer of bridges with integral pier caps, or forward the questionnaire to the bridge designer. The modified questionnaire was sent to 30 domes- tic bridge designers and 20 international bridge designers in 12 countries (i.e., Japan, France, the United Kingdom, Canada, Switzerland, Denmark, Poland, Portugal, Germany, Austria, Spain, and Egypt). In addition to the questionnaire, information was solicited from the American Iron and Steel Institute (AISI), the Amer- ican Institute of Steel Construction (AISC), and the National Steel Bridge Alliance (NSBA). Also, the manufacturers of prestressing systems were asked to provide information on the bridges where their systems were used to post-tension concrete integral pier caps. In total, 111 copies of the questionnaire were sent out. A total of 67 responses were received. The breakdown of the responses is given in Appendix A (provided on the accom- panying CD-ROM). Among the responses, 11 state DOTs indicated past use of integral pier caps. However, two of these responses indicated that the pier caps were integral with the superstructure but were supported on bearings (not integral with the substructures). Among the responses from domestic practicing engineers, six indicated designing bridges with integral pier caps in the past. Only one of the responses received from abroad indicated past use of integral pier caps. The analyses of the responses to the questionnaire are pro- vided in Appendix A. The main conclusions from the responses to the questionnaire were as follows: • The main reason for using integral pier caps in the past has been to increase underclearance and to avoid placing the pier caps at a sharp skew (94 percent of the cases). Enhancing seismic performance was cited in 33 percent of the cases. • The total number of bridges with integral pier caps reported in the responses is 59 (i.e., 47 bridges reported by state DOTs and 12 bridges reported by domestic prac- ticing engineers; however, some bridges may have been reported in both groups). • Plate girders were used in the superstructures of most bridges with integral pier caps. • Most integral pier caps (76 percent) are supported on single-column piers, 8 percent are supported on multi- column piers, and the type of piers was not defined for the remaining bridges. • Most integral pier caps (90 percent) are made of con- crete. The remaining pier caps are made of steel. • No accurate cost data were available. The respondents estimated that the weight of the steel was decreased by 5 to 10 percent and the cost of fabrication and the cost of erection both increased 5 to 10 percent. However, in one case it was estimated that the use of integral pier caps eliminated the need to raise the approach roadway elevation leading to an estimated savings of $250,000 for each 305 mm (12 in.) difference in elevation. • In most cases, the forces acting on the superstructure and the substructures were calculated taking into account the frame action between the superstructure and substructure. In a few cases the frame action was ignored in designing the superstructure or in designing both the superstructure and the substructure. • In general, past performance of integral pier cap bridges appears to be satisfactory. Deck cracking transverse to the girders is the most cited problem. However, it was concluded that this type of cracking does not seem to be

8seismic response in the longitudinal and transverse directions. Using simplified strut-and-tie models to represent the joint force conditions, these methods established efficient rein- forcement details suitable for integral box-girder concrete bridge joints. The investigation of alternative design methods included large-scale experimental tests and detailed analytical studies (4–10), which are summarized in Appendix B (pro- vided on the accompanying CD-ROM). Despite common application in non-seismic regions and in some seismic regions such as in Washington State, precast con- crete girders have seldom been used in California. The limited use of precast girders in California was attributed to the lack of design methods and past experimental research confirming sat- isfactory seismic behavior. Consequently, the Precast Concrete Manufacturers Association of California (PCMAC) and Cal- trans sponsored a research program at UCSD to investigate the seismic response of precast spliced-girder bridges with integral piers (11). Similar to the previous UCSD studies, cast-in-place columns and cap beams as well as cast-in-place integral con- nections between the cap beam and column were used in this research project. However, the soffit slab, which was assumed to be assisting the joint force transfer in the previous box girder studies, did not exist in the precast, spliced-girder bridge. Using both the bulb-tee and bathtub precast girders, the project demonstrated that the integral precast splice-girder bridge sys- tem without soffit slab can be constructed cost-effectively to produce satisfactory seismic response. A logical next step in the seismic bridge design advance- ment was the investigation of the integral bridge system with concrete columns and a steel superstructure, because the use of steel members in the superstructure can provide additional benefits over those realized with the precast concrete girder system (12). Although such a design is the subject of the NCHRP 12-54 Project, Caltrans initiated a similar investiga- tion at UCSD around the same time that NCHRP 12-54 was approved for research. Using the prototype structure from the precast spliced-girder research project, the UCSD study exam- ined an integral steel girder bridge system with a concrete cap beam (13,14) whereas the NCHRP Project 12-54 has investi- gated the use of steel girders supported on, and connected inte- grally to, steel box-beam pier caps. The pier caps are assumed to be supported on reinforced concrete columns. The con- nection between the columns and the pier cap is accomplished by extending the column longitudinal reinforcement through holes in the bottom flange of the box-beam pier cap. The com- partment of the box-beam pier cap directly above a column is then filled with concrete to anchor the column reinforcement. The UCSD study included several component tests repre- senting the integral connection region, which included a con- crete cap beam, two interior girders, and a support block mod- eling the concrete column. The loads were applied directly to the girders with the main objective of investigating the con- nection between the steel girder and the concrete cap beam. Two girder parameters were investigated in the component tests, namely girders with web stiffeners in the cap-beam caused by the integral connection as it also has been reported for conventional bridges. Many of the bridges reported in the responses are new bridges and, hence, long-term performance data were not available. • The difficulties faced during construction of integral pier cap bridges included the congestion of reinforce- ment at the columns-to-pier-cap joint and the need for a substantial temporary shoring system. See Appendix A for more detailed analyses of the question- naire responses. 2.1.2 Summary of Literature Review A comprehensive literature review was performed as part of the NCHRP 12-54 Project. Over the course of the project, other newer relevant articles and reports were reviewed and added to the documentation. At the time of the literature study, some critical aspects of the prototype structure to be used in the NCHRP 12-54 Proj- ect were not finalized. These included the type of cap beam (i.e., steel versus concrete), type of concrete bent (i.e., single column versus multiple column), and the critical loading direc- tion (i.e., transverse, longitudinal, or both). In fact, informa- tion obtained from the literature review was used in deciding that the research of the NCHRP Project 12-54 should focus on the longitudinal direction seismic response of a prototype bridge with a single-column concrete bent and a steel cap beam. Consequently, the literature review generally focused on past research and past use of integral pier bridges in both seismic and non-seismic regions. Most of the research materials relevant to seismic applica- tions of integral column bridges were found to be based on research sponsored by the California Department of Trans- portation (Caltrans) following the 1989 Loma Prieta Earth- quake. This event demonstrated that integral connections of concrete bridges in California were not sufficiently designed and detailed to permit bridges to experience ductile seismic response. As a result of damage to the cap beam-to-column integral connections, several bridges experienced significant damage, including collapse, in the Loma Prieta Earthquake (3,4). Throughout the 1990s, researchers at the San Diego and Berkeley campuses of the University of California conducted vigorous experimental seismic research on various aspects of concrete box-girder bridges with integral concrete columns. This particular bridge type was singled out because of its wide- spread application in California. The research effort, which was generally initiated to characterize the behavior of existing struc- tural members, investigated retrofit techniques, repair methods and improved design methods for columns, cap beams, foun- dations, and cap beam-to-column integral connections. Of particular interest to the investigation undertaken in the NCHRP 12-54 project were the alternative methods studied at the University of California at San Diego (UCSD) for designing cap beam-to-column concrete integral joints for

region and girders without web stiffeners. Two conditions for the cap beam were also considered—a conventional reinforced concrete cap beam and a post-tensioned cap beam. Four com- ponent tests were conducted under longitudinal seismic load- ing to investigate several possible combinations of the above- mentioned parameters. Based on the test results, an integral bridge system composed of stiffened steel girders and a post- tensioned cap beam was selected for system level investigation. The system test was conducted at 40-percent scale and showed that the proposed integral bridge system with stiff- ened girders and post-tensioned cap beams would provide ductile seismic response. In the test unit, the plastic hinge was fully developed in the column while the superstructure exhibited essentially elastic response. Concrete box-girder bridges and precast concrete and steel girder bridges designed generally with post-tensioned pier caps were the subjects of these articles, most of which dis- cuss the application of integral pier bridges in non-seismic regions. Review of these articles made it apparent that inte- gral pier concrete bridges have gained increasing popularity both in the United States and overseas since the mid-1970s. At present, several state DOTs use the integral pier concept for new bridges. Given the benefits of steel girder integral pier bridges in seismic regions, an investigation of this bridge concept with emphasis on seismic issues was needed. 2.2 FEASIBLE INTEGRAL PIER CONCEPTS To identify integral connection concepts for this study, the research team studied the connection concepts used in the past and developed several other systems. In total, 14 different pier cap systems were examined. The attributes of each pier cap system were determined, and the systems were grouped based on the number of columns per pier, type of girders, the need for shoring during construction, and the pier cap material. In addition, two main types, concrete and steel, of the integral connections were examined. A detailed description of all sys- tems examined and their attributes is provided in Appendix C (provided on the accompanying CD-ROM). Criteria were developed to assist in selecting a system for further detailed studies. These selection criteria were based on listing several desirable features and giving each of the 14 systems considered a score for each feature. Some of the features considered were related to construction and others were related to the long-term performance and economy of the system. Several practicing engineers participated in the selection process and the scores were averaged. The systems with highest score were as follows: • Steel I-girders on single-column piers and post-tensioned concrete pier cap, • Steel I-girders on multi-column piers with columns located under each girder, and • Steel I-girders on single-column piers and steel box-beam pier cap. 9 The first of the three systems was not selected in order not to duplicate the work started at UCSD (see Section 2.1.2 above). The second system was excluded because multi-column piers have not been used in the past and also because the results of studying the connection of a single-column pier to the super- structure can be applied to the design of the connection of the columns of a multi-column pier. The third system was selected for further study. The details of the selection process are presented in Appendix C (provided on the accompanying CD-ROM). 2.3 PROTOTYPE BRIDGE CONFIGURATION AND DESIGN Following is a summary of the description of the prototype bridge and its design. More details on the prototype bridge are provided in Appendix D (provided on the accompanying CD-ROM). 2.3.1 Configuration A continuous, two-span bridge with a single-column reinforced-concrete intermediate pier, steel box-beam pier cap, and steel girders as shown in Figure 2 was selected as the prototype bridge. The prototype contains an integral con- nection between the column and the pier cap and integral con- nections between the girders and pier cap and is simply sup- ported at the abutments. Each span of the prototype bridge was 30.5 m (100 ft.). The column height measured from the bottom of the pier cap to the top of the footing was taken as 12.2 m (40 ft.). The bridge was assumed to have four girders spaced at 3.050 m (10 ft.). A compartment centered above the column and bounded by the two box-beam webs and internal diaphragms aligned with the interior girder was assumed to be filled with con- crete. The column longitudinal bars were assumed to pass through holes in the bottom flange of the pier cap and to be anchored in the concrete inside the cap. Shear studs welded to the inside of the pier cap were assumed to transfer the col- umn forces from the concrete inside the cap to the cap itself. 2.3.2 Design The bridge superstructure and substructure components were designed in accordance with the 1998 AASHTO LRFD Bridge Design Specifications (1). In addition to traffic loads, the bridge was designed for seismic loads as a bridge in Seis- mic Zone 4, assuming a ground acceleration of 0.4 g. The seismic design of the prototype structure was based on two criteria: (1) the 1998 AASHTO-LRFD provisions as rep- resentative of typical design procedure and (2) the ATC-32 recommendations in consideration of current seismic design philosophy. In addition, it was decided that a minimum rein- forcement ratio of 2 percent would be considered in order to

ensure that a sufficient demand would be placed on the con- nection region to illustrate the connection performance under high loads. The minimum reinforcement ratio was determined to govern the design, and the corresponding required volumet- ric ratio of transverse reinforcement was calculated as 0.00727. 2.4 TEST SPECIMEN CONFIGURATION AND TESTING 2.4.1 Test Specimen Configuration Two one-third-scale test specimens were constructed and tested in the Iowa State University Structures Laboratory. The two specimens are referred to throughout this report as SPC1 and SPC2. The specimens were used to evaluate the lateral distribution of load between steel girders and the cor- 10 responding torsional demand on the cap beam and to verify the accuracy of the analytical models used to analyze the test specimens. Testing of these specimens also provided data to evaluate the effectiveness of the design details for the inte- gral connection between the reinforced concrete column and the steel box-beam pier cap. Test specimen SPC1 was tested in October 2001. Specimen SPC2 was designed based on results and observations of SPC1 and was tested in Septem- ber 2002. Both test specimens were one-third-scale models of the center portion of the two-span prototype bridge (Figure 3). The general test configuration selected is shown in Figure 4. This configuration was based on the scaled prototype dimen- sions and the laboratory fixture requirements. With a length of 6.1 m (20 ft.) and a width of 3.76 m (12 ft, 4 in.), both spec- imens modeled the center 18.3 m (60 ft.) of the prototype 200 mm (8 in.) concrete deck 30.5 m (100 ft) (100 ft) 30.5 m Steel cap beam 2290 mm x 1680 mm (7.5 ft x 5.5 ft) Steel girder Reinforced concrete column 1830 mm (6 ft) diameter 12 .2 m (40 ft ) (a) Longitudinal Elevation Steel plate girders @ 3050 mm (10 ft) on center Cap beam 1830 mm (6 ft) diameter Reinforced concrete column 200 mm (8 in.) concrete deck (b) Transverse elevation Figure 2. Prototype bridge.

bridge. The specimens were constructed and tested in an inverted position, and the girder ends were simply supported. The column in each specimen was first subjected to low- level loads to simulate service loads. Then, the column was subjected to cyclic loading to simulate the effects of earth- quake loads on the corresponding prototype structure, specif- 11 ically, the overall seismic performance of the structure and the performance of the connection details of the pier cap-to-col- umn connection and the pier cap-to-girder connection regions. 2.4.2 Design Details For the specimen column diameter of 2 ft., the required reinforcement ratio of 2 percent resulted in longitudinal steel of 20 reinforcing bars, 19 mm (3/4 in.) in diameter. The result- ing design for the plastic hinge region, to provide the required volumetric ratio, was a #10 (#3) spiral with a pitch of 63 mm (2.5 in.). The clear concrete cover provided outside the lon- gitudinal reinforcement was 1 in., as shown in Figure 5a for SPC1. The connection region in SPC2 differed slightly, as shown in Figure 5b, because of the use of mechanical anchor- age for the column longitudinal bars. A moment-curvature analysis was performed to predict the behavior of the plastic hinge region of the column. The same column design was used for both SPC1 and SPC2. In specimen SPC1, 610 × 101 (W24 × 68) rolled shapes were used for the four girders. The girders were decreased to 460 × 60 (W18 × 40) rolled shapes in SPC2. The 610 mm (24 in.) depth of SPC1 girders corresponds to a depth of 1,830 mm (72 in.) for the prototype bridge. This is larger than the typical girder depth for bridges with spans comparable to the prototype bridge. This depth was selected to provide adequate development length for the column lon- gitudinal bars inside the connection region of the test speci- men. For SPC2, the 457 mm (18 in.) girder depth corresponds to 1,370 mm (54 in.) in the prototype bridge. This depth is representative of actual bridges of span comparable to that of the prototype bridge. The depth of the connection region in SPC2 was not sufficient to fully develop the column longi- tudinal bars inside the connection region. The ends of the col- umn bars were threaded and mechanical anchorage, in the form of nuts threaded at the column bar ends, was added to provide full anchorage for these bars. Grillage model analyses were performed to predict the force-displacement responses of SPC1 and SPC2. A nonlinear Selected portion for testing 18.3 m (60 ft) 7. 5 m (24 .5 ft) Figure 3. Modeled portion of the prototype bridge. Vertical load (8' -2 5/8 ") 25 05 m m Horizontal reaction beam Lateral load Steel wide-flange Steel box beam girder 610 mm (24 in.) diameter reinforced concrete (RC) column (a) Elevation Horizontal reaction beam Roller supports Steel girder Steel diaphragm RC slab N 610 mm (24 in.) dia. RC column Steel box beam Roller supports 10 16 m m (3' -4" ) (3' -4" ) 10 16 m m (3' -4" ) 10 16 m m 37 60 m m (12 '-4 ") 5800 mm (19'-0") 6100 mm (20'-0") (b) Plan view Figure 4. General configuration of the test specimen.

spring was used to model the plastic hinge region of the col- umn. The spring’s properties reflected the moment-rotation behavior developed from the moment-curvature analysis con- ducted for the column section. 2.4.3 Instrumentation Load cells were used to measure the horizontal and verti- cal column loads applied at the free end of the specimen col- 12 umn. In SPC2, two load cells were added to measure interior girder support reactions. Displacement and rotation transducers were used on SPC1 and SPC2 to monitor selected displacements and rotations. These transducers measured horizontal deflection and rota- tion of the column free end, relative vertical deflections of the column used to determine column curvature, vertical deflec- tions of the specimen girders, relative horizontal girder dis- placements used to determine girder curvature, girder elon- Section X-X @ 100 mm (4 in.) c-to-c #10 (#3) bar Slab reinforcement Located on vertical cap plate 138 mm (5 7/16 in.) horizontal spacing 135 mm (5 5/16 in.) vertical spacing 20 - 16x68 mm (5/8 x 2 11/16 in.) shear studs #10 (#3) spiral @ 90 mm (3.5 in.) 20 - #19 (#6) longitudinal bars #10 (#3) spiral @ 65 mm (2.5 in.) X Gauged long bars 13x80 mm (1/2 x 3 1/8 in.) shear studs 760x560x13 mm 13x54 (1/2 x 2 1/8 in.) shear studs Deck, 67 mm (2 5/8 in.) thick 13x80 mm (1/2 x 3 1/8 in.) shear studs Cap beam Gauged short bars 16x106 mm (5/8 x 4 3/16 in.) shear studs W610x101 (W24 x 68) girder #10 (#3) spiral X Load (30 x 22 x 1/2 in.) both directions (a) SPC1 Section X-X @ 100 mm (4 in.) c-to-c #10 (#3) bar Slab reinforcement Located on vertical cap plate 100 mm (3 15/16 in.) horizontal spacing 135 mm (5 5/16 in.) vertical spacing 20 - 16x70 mm (5/8 x 2 11/16 in.) shear studs #10 (#3) spiral @ 90 mm (3.5 in.) X 13x80 mm (1/2 x 3 1/8 in.) shear studs 760x410x10 mm 13x54 mm (1/2 x 2 1/8 in.) shear studs 70 mm (2 5/8 in.) deck 13x80 mm (1/2 x 3 1/8 in.) shear studs Cap beam Bars without mechanical 16x106 mm (5/8 x 4 3/16 in.) shear studs W460x60 (W18x40) girder #10 (#3) spiral X Load anchorage 20 - #19 (#6) longitudinal bars #10 (#3) spiral @ 63 mm (2.5 in.) Longitudinal bar threaded end 16 mm (5/8 in.) nut 10 mm (3/8 in.) washer PL Cap beam PL (30 x 16 1/8 x 3/8 in.) both directions (b) SPC2 Figure 5. Column-to-cap beam connection detail.

gation, cap beam dilatation, and overall horizontal specimen translation. Strain gages were used to measure strains at selected loca- tions in both specimens. These locations included the gird- ers; column spiral reinforcement; column longitudinal rein- forcement; slab reinforcement; and cap beam webs, flanges, diaphragms, and shear studs. 2.4.4 Seismic Load Simulation To simulate seismic loading of the prototype, the load sequence for both specimens consisted of applying an appro- priate column axial load downward to the top of the (inverted) column to simulate the prototype gravity effect. While main- taining the gravity load at a constant level, seismic effects were simulated by using a cyclic, lateral load with full reversals. To determine appropriate column axial loading, the expected shear and moment values for the prototype bridge and the test specimens were compared. Based on the results of this investigation, most of the test of SPC1 was conducted with a column axial load of 270 kN (60 kips) to produce a better prototype/specimen moment comparison, and the axial load was increased to 580 kN (130 kips) in a later stage of cyclic testing to carefully evaluate the shear transfer. The load was returned to 270 kN (60 kips) for the remainder of the test. In SPC2, a similar pattern was used except an axial load of 220 kN (50 kips) was used instead of the 270 kN (60 kips) used in SPC1. Seismic effects were simulated using a cyclic, lateral load pattern with full reversals as shown in Figure 6. Details of the test fixture and the loading system are presented in Appendix E (provided on the accompanying CD-ROM). 2.5 TEST RESULTS A summary of the test results is presented below. A detailed description of the test observations and measure- 13 ments is presented in Appendix F for SPC1 and in Appen- dix G for SPC2. (These appendixes are provided on the accompanying CD-ROM.) 2.5.1 Seismic Test Results General observation of SPC1 revealed that the specimen behaved as expected under lateral loading (i.e., a plastic hinge was formed in the column adjacent to the cap beam). The super- structure behaved elastically throughout the entire test, also in accordance with the intent of the design. As expected, flexural cracking of the column occurred at loads below the predicted yield of the column longitudinal reinforcement. Defining the displacement ductility, µD, as the ratio between the maximum displacement during a load cycle during the test divided by the displacement required to cause the yield of the column longi- tudinal reinforcement, concrete spalling at the column-to- cap beam connection began to occur at displacement ductility µ∆ = 1.5. At ductility µ∆ = 4.0, several column longitudinal bars were visible, with a few showing indications of buckling. At ductility µ∆ = 6.0, the three extreme column longitudinal bars on each side of the column fractured, as shown in Figure 7. The buckling is believed to have been caused by loss of confinement because of interaction effects between the steel cap beam and concrete column. The steel flange of the cap beam interrupts the concrete of the column and represents a discontinuity of the concrete. In addition, the column spiral was terminated at the flange of the cap beam and was restarted on the other side of the flange. Each end of the spiral was anchored with an addi- tional two turns of the spiral as required by current design prac- tices for concrete members. However, it appears that, because of the discontinuity presented by the steel cap beam flange, this anchorage is not sufficient. Specimen SPC2 also displayed satisfactory seismic per- formance, exhibiting the formation of a plastic hinge in the 1.0 µ∆ −∆y' ∆y' -1.0 µ∆ Displacement Control Force Control # of Cycles = 1 4 steps 3 µ∆ = 1 3 µ∆ = 1.5 3 µ∆ = 2 3 µ∆ = 3 3 µ∆ = 4 3 µ∆ = 6 A xi al lo ad La te ra l l oa d se qu en ce 270 kN 580 kN Figure 6. Load sequence selected for simulation of seismic effects on test specimen SPC1 (1 kN = 0.225 kips). Figure 7. Fracture of column longitudinal bars at µ∆ = 6.0 × 1 (SPC1, column tension side while pull direction load is being applied).

column adjacent to the cap beam and showing elastic behav- ior of the superstructure. Early stages of the test revealed similar behavior to SPC1. Increased cracking was seen in the slab as could be expected for the more flexible superstructure in SPC2. The primary difference in results in SPC2 was the failure mechanism of the longitudinal bars, which appeared to lose anchorage in the connection region and fractured the mechanical connections. (Mechanical connections were not used in SPC1 because the increased cap beam height in SPC1 provided adequate anchorage length for the longitudinal rein- forcement.) The plastic-hinge region of SPC2 following test- ing is shown in Figure 8. Views of the columns from SPC1 and SPC2 following seismic testing are shown in Figures 9 and 10. Figures 11 and 12 show the experimental load-displacement hysteresis for specimens SPC1 and SPC2, respectively. Both specimens exhibit satisfactory seismic behavior through duc- tility µ∆ = 4.0 as indicated by the regular shape of the hys- teresis and the gradual degradation of stiffness. In the load- displacement response for SPC1 (Figure 11), the decreased load resistance resulting from the fracture of several of the column longitudinal bars at µ∆ = 6.0 is indicated by the lower stiffness of the last cycles in Figure 11a. SPC2 also exhibits decreased load resistance at ductility µ∆ = 6.0 because of the 14 fracture of the mechanical anchorage of several of the lon- gitudinal bars as shown in Figure 12a. The predicted load- displacement relationships developed from the grillage analy- ses are also shown in Figures 11b and 12b. The actual column behavior is quite consistent with the predicted behavior for the initial load steps. The pull direction response of SPC2 is seen to begin to differentiate from the predicted behavior at Figure 8. Partial view of column at the completion of seismic testing (SPC2, tension side of the column in the pull loading direction). Figure 9. Tension side of the column in the push loading direction after seismic testing (SPC1).

ductility µ∆ = 3.0 because of slippage of the extreme bar that was not mechanically anchored. To investigate the necessity of extending the column lon- gitudinal bars into the bridge deck, the strains measured on the longitudinal bars in the connection region were investi- gated. In Figure 13, the strain profiles from SPC1 of a longi- 15 tudinal bar extending through a hole in the cap beam flange into the deck are compared with the strain profiles of the opposite bar which was terminated in the connection region next to the flange near the deck (i.e., was not extended into the bridge deck). The comparison is shown for two different load steps, 0.5 Fy and µ∆ = 1.0. Most of the load is seen to be dissipated near the column-to-cap beam interface (embed- ment length = 0 in.) for the smaller load step for both bars, while both bars exhibit a more linear distribution at the higher load step. The good comparison indicates that extension of the bars into the bridge deck is not necessary if adequate anchorage length is provided within the connection region. The dilatation of the cap beam in the connection region was investigated to determine the adequacy of the steel box beam in providing confinement. This investigation revealed that, although the cap beam behaved elastically throughout the test, the dilatation began increasing more drastically at loads exceeding 1.0 Fy, indicating that the cap beam by itself would perhaps become ineffective in providing confinement at higher loads and that spiral reinforcement in the connec- tion region is required for confinement at higher loads. 2.5.2 Simulated Service Load Testing Four simulated service load tests were conducted as illus- trated in Figure 14: SLC1, SLC2, SLC3, and SLC4 with two different support conditions (i.e., all girders supported and only exterior girders supported) and two types of loading (i.e., vertical and lateral). The primary purpose of these tests was to investigate the distribution of moments between the inte- rior and exterior girders and to validate the analytical model by comparing the analytical results to test results. Because of the small magnitude of loads and strains for these four load conditions, three load conditions from the seis- mic loading with similar support and load configurations were also used in the analysis of the girder distribution factors. Girder strains were used to determine the experimental load distribution. Analytical models were used to predict the load distribution for loading in both the vertical and horizontal directions. Comparisons of the experimental and analytical load distributions from SPC1 and SPC2 are shown in Tables 1 and 2. The strain in the flanges of the interior and exterior gird- ers is shown in Figure 15. The comparisons for SLC1 and SLC3 are very good for both specimens. However, in both specimens the experimental distribution for SLC2 and SLC4 revealed a significant percentage of the load being carried by the interior girders, whereas the analytical model indicates a distribution of 100 percent to the exterior girders. At least two explanations are possible. First, the experimentally based load distributions may not be accurate because of experimental errors at the low strain levels for these two tests. Second, and much more likely, the transverse stiffness of the concrete slab and end diaphragm distributed forces to the interior girders even though they were not supported. The analytical model did not account for transverse stiffness between the girders. Figure 10. Tension side of the column in the push loading direction after seismic testing (SPC2).

16 -400 -300 -200 -100 0 100 200 300 400 -250 -150 -50 50 150 250 Displacement at column end (mm) La te ra l f or c e re si st an ce (k N) (a) Entire simulated seismic test 0 -400 -300 -200 -100 100 200 300 400 0-200 -150 -100 -50 50 100 150 200 Displacement at column end (mm) La te ra l f o rc e re si st an c e (kN ) Experimental Predicted µ∆ = 1.5 3.0 4.0 µ∆ = -4.0 -2.0 -1.0-3.0 (b) Up to µ∆ = 4.0 Figure 11. Column lateral force-displacement response of SPC1 (1 kN = 0.225 kips, 1 mm = 0.039 in.).

-400 -300 -200 -100 100 0 200 300 400 -250 -200 -150 -100 -50 0 50 100 150 200 250 Displacement at column end (mm) La te ra l f or ce re si st an ce (k N) (a) Entire seismic test 0 -400 -300 -200 -100 100 200 300 400 -200 -150 -100 -50 0 50 100 150 200 Displacement at column end (mm) La te ra l f or ce re si st an ce (k N) Experimental Predicted -4.0 -3.0 -2.0 -1.0 µ∆ = µ∆ = 1.5 3.0 4.0 (b) Up to µ∆ = 4.0 Figure 12. Experimental force-displacement response of SPC2 (1 kN = 0.225 kips, 1 in. = 25.4 in.). 0 100 200 300 400 500 600 700 0 500 1000 1500 2000 2500 Microstrain D is ta nc e fro m th e ca p be am in te rfa ce (m m) −1.0 µ 1.0 µ -1.0 Fy (short) -0.75 Fy (short) -0.5 Fy (short) -0.25 Fy (short) +0.25 Fy (long) +0.5 Fy (long) +0.75 Fy (long) 1.0 Fy (long) Figure 13. Tension strain profiles for the reinforcing bars (SPC1).

P = 90 kN (20 kips) H = 45 kN (10 kips) H = 45 kN (10 kips) P = 90 kN (20 kips) (a) SLC1 (b) SLC2 (d) SLC4(c) SLC3 Figure 14. Service-load test condition. Load Distributions Load Condition Girder Average Experimental Analytical Exterior 50% 45% SLC1 Interior 50% 55% Exterior 76% 100% SLC2 Interior 24% 0% Exterior 35% 28% SLC3 Interior 65% 72% Exterior 72% 100% SLC4 Interior 28% 0% Load distributions Load Condition Girder Average Experimental Analytical Exterior 47% 46% SLC1 Interior 53% 54% Exterior 75% 100% SLC2 Interior 25% 0% Exterior 32% 30% SLC3 Interior 68% 70% Exterior 60% 100% SLC4 Interior 40% 0% -90 -60 -30 0 30 60 M ic ro st ra in SLC1 SLC2 SLC3 SLC4 (a) Strains at gages S3 and S11 for SLC1 through SLC4 Exterior Girder Interior Girder S1, S3 S9, S11 (b) Cross-sectional view showing the relative location for strain gages S1, S3, S9, and S11 TABLE 1 Comparison of experimental and analytical load distributions (SPC1) TABLE 2 Comparison of experimental and analytical load distributions (SPC2) Figure 15. Top flange strains for the exterior and interior girders for SLC1 through SLC4.