Cast-in-Place Concrete Connections for Precast Deck Systems (2011)

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.

22 Chapter 3 PCSSS: Background 3.0 Introduction to Background Interest in the development of precast composite slab span system (PCSSS) bridges in Minnesota led to research on the topic at the University of Minnesota prior to the initiation of the NCHRP 10-71 project. The results were valuable in selecting appropriate design parameters for the laboratory portion of the NCHRP 10-71 project, and provided important information considered in the development of the comprehensive design guide for PCSSS bridges as part of the NCHRP project. The first section of this chapter includes a summary of a phone survey that was completed at the outset of the current study. The primary focus of the survey was to identify the principle behaviors and concerns that engineers, contractors, fabricators, and researchers had with the precast slab span bridge system. The results of the phone survey proved very useful in the development of specific characteristics that the PCSSS should be expected to satisfy. In addition, this chapter includes observations of an existing three-span PCSSS bridge (i.e., Bridge No. 13004) constructed in Center City, Minnesota, by the Minnesota Department of Transportation (Mn/DOT). This bridge was one of the initial implementations of the PCSSS in Minnesota, and was instrumented during construction to measure the load distribution and the potential for the development of reflective cracking. Near midspan of the center span of the bridge, three adjacent longitudinal joints were instrumented to investigate reflective cracking using a series of strain gages placed transversely above the joint. The instruments were monitored for several years following construction. The data collected from this bridge provided a useful data set regarding the behavior of an in-service PCSSS bridge, which was analyzed and utilized during the NCHRP 10-71 study to correlate the laboratory results with measured field data. In addition to the Center City Bridge, Mn/DOT sponsored the construction and testing of a two-span laboratory specimen located at the University of Minnesota, referred to as the Concept 1 laboratory bridge elsewhere in the report. The Mn/DOT study included the investigation of the development of restraint moments in the Concept 1 laboratory bridge specimen. In addition, Eriksson (2008) completed a numerical and parametric study of restraint moments as part of the Mn/DOT investigation. Because of the previous research conducted regarding restraint moments on the PCSSS, no further investigation on the topic was conducted during the NCHRP 10-71 study, however the experimental data and other available resources from the previous studies were analyzed to provide guidance regarding restraint moments that is included in the design guide developed during the NCHRP 10-71 study. 3.1. Introduction to Survey Results This section provides a summary of the comments received from the individual respondents who participated in the phone survey that was used to gather information including the respondents’ experience with similar systems, their input on important performance criteria, and their feedback on proposed connection concepts. A more extensive summary of the comments received is included in Appendix C. Nearly 60 people were interviewed during the phone surveys which were conducted in partial fulfillment of project Tasks 1 and 2. The individuals interviewed were selected through consultation with the project team, David Beal (Senior Program Officer, NCHRP), and interviewees who recommended

23 others who should be contacted. The respondents represent bridge engineers (including many individuals who serve as State Bridge Engineers), consulting engineers, fabricators, material suppliers, industry representatives, and technical committee contacts. The distribution is presented in Table 3.1.1. Table 3.1.1: Distribution of phone survey respondents Bridge Engineer Material Specialist Fabricator Researcher Industry Rep Contractor Other 37 9 6 13 3 0 3 A few survey respondents were identified under multiple categories in Table 3.1.1, so the total number in the table (71) exceeds the number of survey respondents (59). Also note, that although there are no individuals listed under the “contractor” category, a number of bridge engineers and fabricators were able to provide firsthand experience on constructability issues encountered in the field. Attempts were made to include contractors in the survey, but because of the high level of activity with the contractors in the field towards the end of the construction season, and the difficulty in reaching the individuals in the field, this was not possible. The contractor on site during construction of the Mn/DOT implementation of the Poutre Dalle (PD) concept was queried by members of the project team regarding constructability issues, which influenced some of the decisions made in the concept development implemented in the University of Minnesota laboratory specimens described in Chapters 5 and 6. A complete list of the phone survey participants is included in Appendix C. The phone survey was primarily developed to provide insight into what the respondents considered to be the most important aspects and behaviors that should be considered for a rapid construction bridge system. In addition, respondents were provided with an overview and figure of the Poutre Dalle system as well as the modifications considered in the development o f the Mn/DOT PCSSS, and were asked to share their comments, concerns, and any changes they might consider if they were to implement a similar composite slab span system. Some respondents (44) provided insight into the relative importance and prioritization of certain characteristics in regards to the PCSSS. The participating respondents were asked to provide a rank ordering, from one to eight (one being the most important), of the following categories: durability, strength, fatigue, seismic, constructability, rapid construction, serviceability/performance, and economy. Some participants did not rank all eight performance categories, and only ranked the one or two criteria considered to be most important. The category that received the highest number of votes was durability, with 91 percent of the 44 participants providing a ranking. A total of 75 percent of the respondents that provided a ranking in the durability category selected it as being the most important, with a ranking of one. The second and third most ranked categories were constructability and rapid construction, with a total of 68 and 57 percent of the respondents providing a ranking. Economy, serviceability/performance, and strength were the fourth, fifth, and sixth most ranked categories, with 45, 43, and 34 percent of the respondents providing a ranking, respectively. The two performance categories least likely to be ranked at all were fatigue and seismic, both with a total of 20 percent of the respondents providing a ranking. A complete breakdown of the rank ordering is provided in Appendix C. The vast majority of the respondents considered durability to be the primary issue that must be considered with the PCSSS. Many respondents provided additional comments suggesting that reflective cracking between the precast panels would likely be a primary issue with the system. A primary concern among the respondents was in regards to the development of cracking and subsequent ingress of chlorides and the potential for corrosion of the reinforcement. One respondent suggested that the joints between the precast elements would be the weak link of the system, and should therefore be well

24 designed and thought out. The overwhelming consensus that durability was a primary concern regarding the PCSSS was a major driver in the fact that the control of reflective cracking was an essential objective of the project. 3.1.1. Mn/DOT PCSSS and Poutre Dalle System The survey respondents provided insightful comments and suggestions regarding the original PD system as well as the first implementation of the Mn/DOT PCSSS. Most, if not all, respondents liked the overall concept of the PCSSS, and furthermore thought it could be implemented by their DOT’s. Some however, considered the system to be economical only in shorter spans, and did not envision the system as a highway bridge. Several respondents commented specifically on the applicability of the system in locations with vertical clearance requirements. The use of 90 degree hooked transverse reinforcement over the precast joint was favored strongly to the 180 degree version in the French Poutre Dalle system, especially as this allowed for the cage reinforcement to be pre-tied and dropped into position as one piece. A few respondents commented on the use of embedded transverse hooked bars, as it would require perforations of the formwork, which may not be favored by fabricators. One solution to perforated formwork would be the use of threaded bars, where the hooked reinforcement was threaded on after casting, which could therefore be done on-site. One respondent suggested that, if the embedded bars were selected for this system, that a standard bar spacing be selected. The bar spacing should either be constant, or an even multiple, thereby allowing fabricators to perforate their formwork in a given pattern which would not be expected to change. The use of voids in the precast elements was brought up by a majority of the respondents, both for and against. Proponents of incorporating voids into the system suggested that the large dead weight of the system would be a hindrance, and that fabricating precast beams with voided elements had been successfully completed in the past (Grafton; Hyzak; Khalegi; Tadros). Several respondents however were opposed to the use of voids primarily due to previous performance issues, including durability issues reported by Minnesota and South Dakota. The use of CIP concrete was discouraged by a number of respondents, primarily because they felt it would be difficult to achieve a significant benefit in terms of rapid construction, and can be costly and difficult to procure in rural areas. The use of CIP concrete in only the trough between each panel was suggested, where the top of the precast web could then be used as the driving surface, however limiting the quantity of CIP concrete was not expected to provide a significant savings. Furthermore, the use of a CIP deck overlay was favored by many respondents because it provided a uniform driving surface, regardless of differential camber in the precast elements, as well as covered the longitudinal joints and ensured the entire driving surface was a uniform color, which was considered to be important to several respondents. 3.2. Center City PCSSS Bridge The first implementations of the Mn/DOT PCSSS began in 2005 with the design and construction of two precast composite slab-span structures. The first was in Beltrami County, Minnesota with three 45 ft. spans and a precast section depth of 16 in. The second implementation was Mn/DOT Bridge No. 13004 in Center City, Minnesota about 40 miles northeast of Minneapolis. The Center City Bridge was also a three-span bridge with 22, 27, 22 ft. spans and had a 12 in. precast section depth. The depth of the CIP concrete over the webs of the precast elements was 6 in. in both implementations. The CIP above the

25 webs was reinforced with typical deck reinforcement in both the longitudinal and transverse directions. The plan view and construction stages of the Center City Bridge are shown in Figure 3.2.1. Figure 3.2.1: Plan view and construction stages of Mn/DOT Bridge No. 13004 in Center City, Minnesota (Bell et al. 2006) The Center City Bridge was instrumented to investigate two primary behaviors, reflective cracking and live load distribution over a continuous pier. Reflective cracking was anticipated to originate from the discontinuity created by the joint between precast panels and/or at the corners of the precast webs near the top of the section, as illustrated in Figure 3.2.2. To facilitate the monitoring of reflective cracking and the overall performance of the structure, the bridge was instrumented with a total of 45 transversely oriented concrete embedment vibrating wire (VW) strain gages and 21 transversely oriented spot-weldable VW strain gages in the trough area between precast panels. The instrumentation to investigate reflective cracking was located solely in the center span of the bridge constructed during Stage 1, as shown in Figure 3.2.1. The instrumentation was distributed over three adjacent longitudinal joint regions, shown in Figure 3.2.3, to provide insight into the performance of each joint as well as the interaction between precast panels.

26 Figure 3.2.2: Anticipated locations of reflective cracking in Mn/DOT PCSSS (Bell et al. 2006) Figure 3.2.3: Location of instrumented joints in the Center City Bridge (Bell et al. 2006) The concrete embedment gages provided a useful means of exploring the condition of the CIP concrete in the longitudinal trough region and were the primary instrumentation investigated during the NCHRP 10-71 study. All concrete embedment gages in the field bridge were nominally located at midspan of the center span in each of the three instrumented joints. The gages were oriented transversely in the precast trough in two vertical rows. The lower row, nominally placed at the same vertical location as the Reflective cracking Web corner of precast section Longitudinal precast joint Flange corners of precast sections

27 transverse hooks, contained five gages, while the upper row, placed just above the precast web consisted of ten gages. The instrumentation layout, which was identical at each of the instrumented joints, is shown in Figure 3.2.4. As shown in the figure, the instrumentation was overlapped to better determine the crack location to within a 2 in. region (i.e., gage length was 6 in., gage overlap was 2 in.). If an increase in strain was observed in two adjacent gages, it could be deduced that the crack developed in the region where the gages overlapped. The lower row of gages was embedded only in the CIP concrete and therefore cracking or separation along the vertical precast web interface could not be observed by the instrumentation, unless such a crack extended into the CIP topping. Figure 3.2.4: Location of transverse concrete embedment gages in each of the three instrumented joints at midspan of the center span of the Center City Bridge (Bell et al. 2006) In addition, seven transversely oriented spot-weldable strain gages were included in each joint to provide insight into the stress demands on the reinforcement bridging the joint between adjacent precast members. The seven gages were installed on two immediately adjacent transverse hooked bars near the group of concrete embedment gages, as shown in Figure 3.2.5.

28 Figure 3.2.5: Lower level of concrete embedment and spot-weldable VW gages utilized in observation of reflective cracking in the Center City Bridge Furthermore, the bridge was instrumented with longitudinally oriented spot-weldable gages to investigate the behavior of the structure in terms of longitudinal live load distribution over a continuous pier. Because of symmetry, only one pier was considered, as shown in Figure 3.2.1. The longitudinal instrumentation was selected to provide for the calculation of the longitudinal curvature at various cross sections on either side of the continuous pier. The longitudinal instrumentation layout is shown in Figure 3.2.6, where the vertical groups of gages are represented by either a diamond or circle. The gage groups represented by a diamond include three gages vertically distributed through the depth, with one gage on the lower No. 5 longitudinal cage bar, the second gage on the top No. 5 longitudinal cage bar, and the third was attached to the deck reinforcement. At cross sections near the pier, where the longitudinal cage reinforcement was not continuous, only two gages were present through the depth of the section, which is represented by the circle symbols. At these locations, the bottom gage was located on the longitudinal No. 8 bar that was provided near the precast flanges for positive restraint moments, and the top gage was located on the deck reinforcement. In addition, the deck reinforcement above two of the webs on the center span side of the pier were instrumented with single longitudinally oriented spot- weldable strain gages, represented by the triangle symbols in Figure 3.2.6.

29 Figure 3.2.6: Plan view of longitudinal instrumentation locations for investigation of live load distribution over the continuous pier (Bell et al. 2006) The Center City Bridge was constructed with a 5-1/4 in. thick precast flange, which controlled the depth of the transverse hooks and subsequently the vertical location of the instrumentation. The lower level of instruments were installed at a nominal depth, measured from the bottom of the section, of 8.5 in., while the upper row of instrumentation for monitoring reflective cracking was located at a nominal depth of 13.5 in. The nominal and measured instrumentation locations were tabulated by Smith et al. (2008). The nominal and as measured gage locations in the Center City Bridge are included in Appendix D. The instrumentation naming scheme was described in detail by Smith et al. (2008), and is summarized here. The first two letters of the name refer to the type of gage and general location, “CJ” refers to a concrete embedment gage near the precast joint. The third digit refers to the joint number; therefore “CJ1” refers to instrumentation in joint 1. The fourth digit refers to the longitudinal location of the gage, “5” corresponds to midspan of the center span, which is where all concrete embedment gages were located to monitor the potential for developing reflective cracks (in conjunction with spot-weldable gages on transverse reinforcement). The fifth digit refers to the vertical depth of the gage, with “1” corresponding to the lower row of gages, and “3” corresponding to the top level of gages. Finally, the sixth digit corresponds to the individual gage number in a given row of instruments, and is incremented between one and five for the lower row of gages, and between one and ten for the top row of gages. Therefore, “CJ1-51-3” refers to a concrete embedment gage near the joint region in Joint 1 at midspan of the center span of the bridge, in the bottom layer of instruments and is the third gage in the row of five gages, which was centered directly over the joint. The data from the instruments was recorded every two hours starting at 10:00am on October 1, 2005. The transverse strains measured at the three instrumented joints at midspan of the center span were monitored over the initial 24 month observation period during the study commissioned by Mn/DOT, as well as afterwards to augment data for the NCHRP 10-71 study. To provide insight into the mechanical strains observed in the bridge, the data collected from the concrete embedment gages was further

30 analyzed. Due to the different values of the coefficient of thermal expansion between the concrete itself and the steel wire inside of the concrete embedment VW gages, taken to be 5.67 µε/°F and 6.78 µε/°F, respectively, a correction based on the difference between the values of the coefficients of thermal expansion was required to determine the mechanical strain, which was calculated as shown in Eqn. (3.2.1). εmechanical = εmeasured where ΔT is the change in temperature, in degrees Fahrenheit, measured from the base temperature recorded when the initial readings of the strain gages were taken. No correction was necessary for the spot-weldable gages, as they measured the mechanical strain directly. + (6.78 µε/°F – 5.67 µε/°F) * ΔT (3.2.1) Large changes in strain observed in Joints 1 and 3 of the structure during the first spring after construction indicated reflective cracking initiated in Joints 1 and 3 on April 25, 2006. Figure 3.2.7 shows the data obtained for Joint 1 which was representative of the behavior of Joint 3 as well. The strain measured in the blue and red gages, which were located directly over and immediately next to the joint, show a large increase and divergence from the other gages starting on April 25, 2006. The blue and red gages correspond with a “3” and “2” as the last number in their name shown in the legend in Figure 3.2.7, respectively; the red gage data series is completely hidden from view by the blue gage data series. The behavior observed in Joints 1 and 3 suggested that cracking was initiated in these joints on April 25, 2006, which was attributed to the effects of solar radiation. Because the bridge was constructed in September of 2005, the bridge was not subjected to significant effects of solar radiation until the spring of 2006. Because the superstructure had been in service and instrumented for a total of 206 days before an increase in the transverse strain was observed, it is unlikely that traffic loading initiated the crack. In fact, strains measured in the transverse joints during subsequent truck load tests conducted in April 2007, as discussed in Section 3.2.1.1, were an order of magnitude smaller than those generated due to the solar radiation effects. Consequently, the reflective cracking was attributed to the effects of large daily variations in the thermal gradient encountered during the spring season due to solar radiation which produced large transverse tensile stresses between the precast panels. Also shown in the figure are the approximate changes in strain over the course of a day during a given summer, with values of 150µε and 220µε measured during the summers of 2008 and 2009, respectively. A similar pattern of large measured strains and divergence of Gages 3 and 4 was observed in Joint 3 of the bridge. The approximate relative changes in transverse strain measured in Joint 3 during the summers of 2008 and 2009 were 230µε and 300µε, respectively. No sudden increase or divergence of the transverse strain was measured in any of the gages located in Joint 2 for the period between October 1, 2005 and July 14, 2009. The presence of cracking in both Joints 1 and 3 was expected to relieve the tensile stresses in the structure which may explain why cracking was not observed in Joint 2; that is, reflective cracking was not originally observed to develop in the adjacent joint.

31 19-Sep-2005 11-Mar-2006 03-Sep-2006 26-Feb-2007 21-Aug-2007 13-Feb-2008 06-Aug-2008 29-Jan-2009 24-Jul-2009 -200 0 200 400 600 S tr ai n ( 10 -6 in /in ) Date CJ1-51-1 CJ1-51-2 CJ1-51-3 CJ1-51-4 CJ1-51-5 19-Sep-2005 11-Mar-2006 03-Sep-2006 26-Feb-2007 21-Aug-2007 13-Feb-2008 06-Aug-2008 29-Jan-2009 24-Jul-2009 -30 -20 -10 0 10 20 30 T em p er at u re ( o C ) Date Figure 3.2.7: Measured transverse mechanical strain and temperature in Joint 1 of Center City Bridge [note results of red gage (black dashed line) are obscured by those of the blue gage (blue line) in the figure] The transverse mechanical strains measured in the Center City Bridge, primarily the relative changes in the strain during a given summer outlined above, provided a quantitative measurement of the performance of an existing bridge in terms of reflective cracking. The measured transverse strains were utilized during the laboratory tests of the NCHRP 10-71 study as a means of determining adequate loading parameters in the laboratory and qualifying the results. The application of the data extracted from the Center City Bridge to the laboratory research specimens is discussed further in Chapters 4 and 5. Smith et al. (2008) also observed large longitudinal strain readings on the reinforcement near mid-depth of the section at the east pier, which suggested that cracking due to positive restraint moment was present at that location. The daily strain changes of approximately 45 µε prior to cracking increased to more than 700 µε on April 23, 2006, which was two days before large increases in the daily strain fluctuations were observed in the transverse directions near the longitudinal precast joints at midspan of the center span. The crack at the east pier was monitored for a little over a year, however the monitoring did not continue past June 21, 2007, when the gage measuring the large daily strain changes began to malfunction. Two longitudinally oriented strain gages located near the top of the section failed to record any large increases in strain over this period, which suggested that the crack had not propagated to the top of the section. 3.2.1. Live Load Distribution Tests at the Center City Bridge A live load truck test at the Center City Bridge was completed by Smith et al. (2008). The objectives of the live load truck test were to evaluate the response of the structure to known static vehicle loads using the embedded instrumentation, which was subsequently analyzed to determine the applicability Initiation of cracking δ≈150µε δ≈220µε

32 of the design assumptions made with respect to transverse live load distribution and continuity over the shared piers. A total of seven single truck and five paired truck configurations were selected for the truck tests, however due to time constraints associated with the single night test, the various configurations were prioritized, with the authors designating the primary test configurations with numbers, and the secondary configurations with letters. The locations of the centroid of the rear truck axles are shown in Figures 3.2.8a and 3.2.8b. The width of the truck axles was measured to be 6 ft. As shown in Figure 3.2.8a, configurations 1, 2, 3, 4, A, and B were located at midspan of the center span of the bridge, and consisted only of single truck configurations, where the rear tandem of the single truck was centered at midspan of the center span for various lateral positions. The primary purpose of these test locations was to provide several load points across the width of the bridge, which would be useful in calculating longitudinal curvature profiles for many locations along the width of the bridge. Furthermore, data from the transverse instrumentation collected during these test configurations was utilized to measure the transverse tensile strains in the trough region between the precast panels, where the potential for reflective cracking was a concern. The six truck configurations were selected such that the centroid of the wheels were located either over the precast web (i.e., configurations 2, 4, and A) or directly over the precast joint (i.e., configurations 1, 3, and B), which was expected to represent practical bounds for the minimum and maximum transverse tensile stress demands near the precast joint, respectively. An additional truck position was located at midspan of the outer span (configuration 6) to provide further insight into longitudinal continuity over the pier.

33 (a) Single truck positions (b) Paired truck positions Figure 3.2.8: Single and paired truck positions during live load truck tests at the Center City Bridge (Smith et al. 2008) Paired truck configurations are shown in Figure 3.2.8b which included configurations 8, 9, 10 and 11, in which cases, the two trucks were parked side by side as closely as possible. These configurations were centered over Joint 1, due to the abundance of longitudinal instrumentation near that joint, and were primarily used to investigate longitudinal continuity over the pier. A fifth configuration, designated E,

34 had the rear tandem of the two trucks spaced laterally at 12 ft. on center at midspan of the center span. This orientation located the centroid of one truck directly over Joint 1 and the centroid of the other paired truck directly over Joint 3, thereby loading all three joints by wheel loads transversely located 3 ft. to either side of the three instrumented longitudinal joints. This configuration enabled evaluation of the transverse strains due to global plate bending without the effects of concentrated loads directly over the joints. The researchers completed the live load tests overnight on April 18, 2007. A total of two tandem sand trucks were utilized, with measured gross vehicle weights of 52.0 and 51.7 kips, which were measured by the contractor away from the test site. Drive-on scales were utilized at the test site to determine the rear axle weights of the individual trucks, however fluctuations in the readings were observed during various weigh-ins due to the fact that the scales elevated the wheels being measured. Therefore, an average rear axle weight was calculated, which was estimated to be 18.6 kips and was applicable to both trucks. Complete tabulated strain measurements from the live load truck tests are given by Smith et al. (2008). 3.2.1.1. Observation of Reflective Cracking during Live Load Truck Tests The transverse instrumentation located in the three joint regions at midspan of the center span was monitored while wheel loads were individually placed above each joint to investigate the effects of the static loading on the joint regions. When wheel loads were placed above each joint, an increase in the transverse mechanical strains of 19, 7, and 32 µε were measured in Joints 1, 2, and 3, respectively. The results obtained for Joints 1 and 2 were calculated as an average over two tests, while three tests were completed over Joint 3. In all cases, the maximum transverse strain was recorded by the middle of the five concrete embedment gages that were located in the lower layer of instrumentation. The increase in the transverse mechanical strains observed during the various test configurations are shown in Table 3.2.1. Instrumentation designation and measured gage locations are given in Appendix D. Table 3.2.1: Increases in transverse mechanical strains immediately over longitudinal joint during static live load truck tests on the Center City Bridge (Smith et al. 2008) Truck Configuration Strain Value(s) (με) Joint 1 3 19, 19 Joint 2 1 6, 6 Joint 2 3 8, 8 Joint 3 1 30, 31 Joint 3 B 34 When wheel loads were located directly above the longitudinal joints, the change in transverse strain was positive, indicating that there was an induced tensile stress in each joint, and therefore that loading tended to open the joint. Due to the fact that the three instrumented joints were designed identically, it was expected that each of the three joints should see similar increases in transverse strains under similar loading. Therefore, the results of the live load truck tests suggested that pre-existing cracking was likely to be present in Joints 1 and 3, because of the larger increases in transverse strain observed in those locations. This conclusion was also supported by the long term strain monitoring described in Section 3.2, where it was suggested that reflective cracking was observed in Joints 1 and 3. Furthermore

35 the relative values of transverse strain observed during the live load truck tests were significantly less than the daily strain fluctuations observed in the same Joints during the long term observation, where, as discussed in Section 3.2, daily transverse strain fluctuations of more than 220 µε were observed. This suggests that transverse strains induced due to thermal gradients can be an order of magnitude larger than those observed due to traffic loading, and therefore should be considered carefully in the design of precast composite slab span systems. Smith et al. also investigated the strains measured in the transverse hooked bars using seven spot- weldable strain gages placed on two adjacent hooked bars (i.e., three on one hooked bar and four on an immediately adjacent hooked bar) at midspan of the center span in each of the three Joints. The researchers observed that the strains measured in the transverse hooked bars were slightly smaller than those measured in the concrete embedment gages, despite the fact that the spot-weldable gages were located 1-¼ in. lower in the section. The authors suggested that the reduction in the measured strain in the transverse hooked bars may be due to slip between the transverse hooked bar and CIP concrete due to the size of the bar (No. 6) and epoxy coating. Another potential reason for the reduction in the measured strain may be due to localized slip due to reduced bond immediately near the strain gage as a result of the strain gage cover. The measured increases in transverse mechanical strains in the adjacent hooked bars observed in Joint 1 under a wheel load during the live load tests are shown in Figure 3.2.9. It can be seen in the figure that each of the adjacent transverse hooked bars shared the load approximately equally, and the strain profile for each hooked bar was similar, regardless of the hook orientation, which suggested that the hooked end was equally effective as the embedded end of each bar.

36 Hook End Center Embedded End Embedded End Hook End Off-Center Off-Center 0 2 4 6 8 10 12 14 16 -8 -6 -4 -2 0 2 4 6 8 Transverse Distance from Joint (in) S tra in (μ ε) East Hook West Hook Figure 3.2.9: Change in mechanical tensile strain in transverse hooked bars at Joint 1 immediately under wheel load during live load truck tests on the Center City Bridge (Smith et al., 2008) 3.2.1.2. Transverse Load Distribution During Live Load Truck Tests Smith et al. (2008) utilized the longitudinal curvature data collected during the live load truck tests to investigate the ability for the Center City Bridge to transfer live loads transversely to adjacent precast panels. In addition, the researchers were interested in the applicability of the AASHTO 2004 Specification Article 4.6.2.3 slab-type bridge load distribution factors, which were utilized in the design of the Center City Bridge (note that the AASHTO Article numbering and content has not changed between the 2004 and 2010 specifications). The authors subsequently utilized the effective width calculations for slab-type bridges to analyze the performance of the Center City Bridge during the truck tests. Configurations 1, 2, 3, 4, A, and B of the truck test were utilized to investigate live load distribution. In each case, longitudinal curvatures calculated at midspan of the center span were measured using the spot-weldable gages located on the longitudinal reinforcement in Joints 1 and 2. Using this procedure, the longitudinal curvature at each joint was calculated as the distance from the joint to the center of the truck load was varied from 0 in. when the center of the truck was immediately over the joint, to as much as 180 in., which was the distance from the center of the truck to Joint 1 in Configuration B. The longitudinal curvatures were estimated as the slope of the best fit line through three spot-weldable gages nominally located at 9, 12.5, and 15.5 in. from the bottom of the precast section, though measured locations were utilized in the calculation of the curvatures. The longitudinal curvatures measured as a function of the distance from the center of the truck load to Joints 1 and 2 are shown in Figure 3.2.10. Also included in the figure are the results of a finite element analysis assuming an isotropic flat plate with a smeared stiffness representing the composite section stiffness of the Center

37 City Bridge. The authors included the parapets in their model, however found that the results were affected only locally by the parapets, and therefore were able to superimpose the data for each of the six configurations onto the single plot. Also included in the figure is a simple-span model, which was calculated assuming that the center span was modeled alone on roller bearings located at the center of each pier, thus providing a center to center of bearing simple span of 306 in. Finally, Figure 3.2.10 also illustrates the design equations provided by AASHTO LRFD (2004) Article 4.6.2.3, which provided the effective width factors for slab-type bridges. -5.0 -4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 -180 -144 -108 -72 -36 0 36 72 108 144 180 Transverse Distance from Center of Truck Load (in) C ur va tu re ( με /i n) Joint 1 Data Joint 2 Data Continuous Model Simple Span Model Design Assumption Figure 3.2.10: Longitudinal curvatures at midspan due to a single truck located at midspan of the center span of the Center City Bridge (Smith et al. 2008) It is evident from Figure 3.2.10 that the measured longitudinal curvatures in the Center City Bridge followed the continuous isotropic model predictions well. Also observed from the figure are the relative similarities in the magnitudes of longitudinal curvatures at common distances from the center of the truck loads. This suggested that loads applied at various distances from Joints 1 and 2 tended to affect each joint similarly, despite suspected reflective cracking in Joint 1 (see Section 3.2). For this reason, the authors suggested that the PCSSS could be adequately modeled using an isotropic plate, which was expected to be simpler because it removed the need to model the discontinuity created by the longitudinal precast joint. Furthermore, the simple span model was found to be more conservative than the continuous model, and would also adequately serve as a method for numerical analysis of the PCSSS. These results also provided significant insight into the applicability of the current design equations provided for slab-type bridges in the AASHTO (2004) LRFD Specification. As shown in the figure, the curvatures determined via the slab-type design equations were approximately three times larger than the curvatures measured in the Center City Bridge when the wheel load was centered directly over the longitudinal joint. For this reason, the slab-type effective lane widths presented in the AASHTO LRFD Specifications (2004) were deemed to be adequate and conservative for the design of precast composite

38 slab span systems. It can be assumed that longer spans would behave similarly because they would have a deeper overall section, and the height of the gap at the flange tips would be unchanged, so longer span bridges should behave more like monolithic systems than shorter ones. 3.2.1.3. Continuity over the Continuous Piers during Live Load Truck Tests Another behavior of interest considered by Smith et al. (2008) was the level of continuity achieved between the piers. For this case, midspan of the center span was loaded directly over Joint 1 and the longitudinal strains in the deck reinforcement of the adjacent span were recorded and compared to the values calculated using the isotropic continuous finite element model. When the center span was loaded, changes in longitudinal strains in the deck steel of the adjacent span were 1.2 and 1.5 µε for single and double truck configurations, respectively. These values were slightly smaller than those predicted by the continuous FEM model, where values of 1.9 and 3.7 µε were predicted for single and double truck configurations, respectively. A likely reason for the discrepancy between the measured and predicted values was attributed to moment transferred into the pier cap, whereas idealized rollers were assumed in the finite element model. In addition, the midspan curvatures in the center span were observed to be smaller than those predicted in the model, as discussed in Section 3.2.1.2. The overall performance of the Center City Bridge during the truck tests suggested that the assumption of full continuity appeared to be conservative for the precast composite slab span system. 3.3. Restraint Moment Multi-span precast composite bridge structures made continuous with CIP concrete develop time- dependent and thermal restraint moments at the continuous piers. The sign and magnitude of restraint moments are affected by shrinkage, creep, age of the precast members at the time of continuity, and thermal gradients. Positive and negative restraint moments are illustrated in Figure 3.3.1. Negative restraint moments are caused by differential shrinkage of the CIP concrete, where the rate of shrinkage of the CIP concrete is larger than the rate of shrinkage and creep of the precast member. When the precast member is at a relatively old age, defined as greater than 90 days in AASHTO (2010) Article 5.14.1.4.4, the shrinkage of the newly placed CIP concrete will tend to “shorten” the top fiber of the bridge structure and subsequently induce longitudinal tensile stresses in the top of the bridge at the piers. The reinforcement included in the deck of the structure over the piers in continuous systems provides the tension ties necessary to counteract negative restraint moments.

39 (a) Negative restraint moment induces tension near the top surface at the pier (b) Positive restraint moment induces tension near the bottom surface at the pier Figure 3.3.1: Positive and negative restraint moments in continuous bridge superstructures (Molnau 2007) Positive restraint moments at the piers in continuous systems may be due to both time-dependent and thermal effects. When the precast member is at a relatively young age at the time of continuity, the rate of shrinkage of the precast member and the CIP may be similar, however, the precast member would also undergo creep. The creep of the precast section would tend to “shorten” the bottom fiber of the bridge structure and subsequently induce longitudinal tensile stresses in the bottom of the bridge at the pier. In addition, thermal gradients in the section cause the top surface of the structure to expand, again inducing a positive restraint moment in the structure. For this reason, both time-dependent and thermal gradient effects must be considered in the design of positive restraint moments. Because positive restraint moments induce longitudinal tensile stresses near the bottom of the section, reinforcement must be provided to carry the tensile force at the piers. Because of the sectional geometry of the PCSSS, all reinforcement provided for positive restraint moments must be located within the longitudinal trough regions between precast panels. Consequently, this reinforcement must be placed in groups centered between panels, generally six feet apart, thereby prohibiting the distribution of the reinforcement along the face of the bottom surface. Smith et al. (2008) monitored the Center City Bridge for evidence of cracking due to restraint moment. Gages located at the pier indicated that cracking initiated due to the effects of positive restraint moment. The crack was believed to develop as the bridge underwent its first large thermal gradient effects due to solar radiation in the spring. As a consequence, it was suspected that the behavior was driven by thermal gradients in the bridge superstructure where the solar radiation heated the top of the bridge. This caused the individual spans of the bridge to camber which generated positive restraint moments. Eriksson (2008, pp. 56) stated, “Because [the] time-dependent effects on [the day the crack

40 was observed] should not have varied significantly from the previous day, researchers speculated thermal effects may have played a role in the crack development. Based on this conjecture, both time- dependent and thermal gradient effects on restraint moment were investigated analytically.” Eriksson (2008) completed a parametric study to investigate the effects of differential shrinkage, creep, and thermal gradient effects on the development of restraint moments. In an effort to predict the restraint moment in a section based on the time-dependent properties of the system, Eriksson completed a numerical parametric study using Pbeam, which is a fiber-based finite element code developed by Suttikan at the University of Texas in 1978 (Suttikan, 1978). The program allows for inputs including material strength, age, creep, shrinkage, steel relaxation, dead loads and support conditions. The program provides output in the form of stresses, strains, reactions and deformations at user specified time intervals (Suttikan 1978). Furthermore, Eriksson utilized a modified version of Pbeam created by Le (1998), called TPbeam, which enables incorporation of thermal gradient effects in the analysis. After finding that, when using functions based on measured quantities for the input values (i.e., creep, shrinkage, concrete strength gain with age, etc.), Pbeam predicted restraint moments that corresponded reasonably well with the measured results from the Concept 1 laboratory specimen, Eriksson utilized both Pbeam and TPbeam to conduct a parametric study to determine reasonable bounds for expected restraint moments in PCSSS bridges. In general, the purpose of the parametric study was to predict the maximum positive and negative restraint moments that would be expected in PCSSS bridges. Precast strengths of 6 ksi and 12.9 ksi were used with assumed continuity dates of 7, 28, 60, and 90 days to develop an expected envelope of the positive and negative restraint moments. Because of the difficulty in providing reinforcement for positive restraint moment due to the geometry of the PCSSS, the necessity to design for such moments was of interest. Eriksson found that, “positive restraint moment cracking due to time-dependent effects is not expected for [PCSSS] for spans between 20 and 50 ft” as seen in Figure 3.3.2 where the ratio of positive restraint moment to cracking moment is always less than one when only time-dependent effects were included. Also, positive restraint moment generally induced at the pier was due to creep of the precast member, therefore increasing the age of the precast member at continuity will reduce the positive restraint moment due to time-dependent effects in the section. The thermal gradient due to solar radiation has the same effect as placing CIP on a young precast section. The results from TPbeam substantiate NCHRP 519 findings that state restraint moments due to thermal effects are significant (Miller et al. 2004). Eriksson (2008) found that the positive restraint moments caused by thermal effects induced restraint moments that were two to seven times larger than the positive restraint moments caused by time-dependent effects as shown in Figure 3.3.3. As a consequence, positive restraint moment cracking due to thermal gradient effects was expected to occur in nearly all of the designs studied, and therefore should be considered by the designer.

41 Figure 3.3.2: Ratio of 20-year positive restraint moment (due to time-dependent effects only) to cracking moment comparison 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 15 30 45 60 75 90 Precast Age at Continuity (days) 20 Y ea r R es tr ai nt M om en t/M cr 50 ft - 6 ksi 50 ft - 12.9 ksi 20 ft - 6 ksi 20 ft - 12.9 ksi CRACK

42 Figure 3.3.3: Ratio of 20-year positive restraint moment (due to time-dependent and thermal effects) to cracking moment comparison AASHTO (2010) Article 3.12.3 provides general guidelines for design of thermal gradients based on regional zones, but indicates temperature gradient should be evaluated on a project specific basis. Judgment is reserved for experienced designers indicating thermal gradient can be neglected if previous structures have not experienced distress. These basic guidelines provide little guidance regarding when thermal gradients are important. Based on the fiber-based finite element model results from Pbeam and TPbeam and the suspected positive moment crack in the Center City Bridge, thermal effects have a significant effect on the development of positive restraint moments and should be considered in the restraint moment design. Restraint moment design for time-dependent properties is complicated by the need to investigate the interaction of the variation in time-dependent effects over time. The PCA and P-method both provide options for how to design for restraint moments due to time-dependent effects. However, design for restraint moments caused by thermal effects does not include the time variation and should be summed with restraint moments due to time-dependent effects. Barker and Puckett (2007) provide a hand calculation for determining the restraint moment due to thermal effects. Assuming the beam is a simple span between supports, the design thermal gradient is applied to the section and used to calculate the resulting curvature in the beam. The curvature from a temperature gradient can be expressed as ( )∫ ⋅⋅⋅= dAyyTI αφ (3.3.1) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 30 60 90 Precast Age at Continuity (days) 20 Y ea r R es tr ai nt M om en t/M cr 50 ft - 6 ksi 50 ft - 12.9 ksi 20 ft - 6 ksi 20 ft - 12.9 ksi CRACK 50 ft - 6 ksi THERMAL 50 ft - 12.9 ksi THERMAL 20 ft - 6 ksi THERMAL 20Ft - 12.9 ksi THERMAL

43 where α is the coefficient of thermal expansion, T(y) is the temperature gradient (°F) through the depth y of the member (in.), and I is the moment of inertia of the entire cross section (in.4) (Barker and Puckett, 2007). This equation is also found in AASHTO (2010) Article C4.6.6. The end rotation (θ ) can be found by integrating the curvature from midspan to the end of the span. Then, the restraint moment, or the moment restraining the rotation, can be found using the three- moment equation. The equation assumes pinned-end supports and is expressed as L IEM ⋅⋅⋅= θ3 (3.3.2) where θ is the rotation, E is the elastic modulus (for the composite system), and L is the span length. The moment, M, is the restraint moment at the pier of a continuous system to resist the rotations induced by the thermal gradient. If the span lengths on each side of the pier are not equal, then the different spans will induce different moments at the pier (i.e., the rotation would be different, leading to different moments).The restraint moments induced by thermal gradients in each span can be calculated using Eqn. (3.3.2) and the design thermal gradient should be for the largest restraint moment. The effects of thermal gradients and time-dependent effects can be calculated independently and then combined with the appropriate load factors. Using the above methodology for prediction of the thermally-induced restraint moments in the 20 and 50 ft. span beams in the parametric study conducted by Eriksson (2008) provided conservative results as compared to the TPbeam results, as shown in Figure 3.3.4. The hand calculations over-predicted the calculated positive restraint moment by a range of 20 to 40 percent. The calculations agreed with TPbeam results that the highest ratio of positive restraint moment induced by a thermal gradient to the cracking moment (i.e., 2.9) was for the 20 ft. span with 12.9 ksi concrete. The shorter span with greater concrete strength had the greatest stiffness and the least flexibility of the sections studied. The 50 ft. span with 6 ksi concrete, the most flexible section studied, had the lowest ratio of positive restraint moment induced by thermal gradient to cracking moment (i.e., 0.95).

44 Figure 3.3.4: Comparison of calculated and TPbeam results for ratio of 20-year restraint moment (due to thermal effects only) to cracking moment The magnitude of the positive restraint moments, with the inclusion of the effects of thermal gradient, may negate any benefit from continuity at service. In this case, if positive moment reinforcement steel is placed at the pier to tie the structure together, the possibility of developing a positive moment at the pier cannot be neglected and the restraint moment should be taken as the lesser of the calculated restraint moments and the moment that can be developed using the positive moment reinforcement placed at the pier. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 10 20 30 40 50 60 T he rm al R es tr ai nt M om en t / M cr Span Length (ft.) 20 ft., 12.9 ksi Hand Calculation 20 ft., 12.9 ksi TPbeam 20 ft., 6 ksi Hand Calculation 20 ft., 6 ksi TPbeam 50 ft., 6 ksi Hand Calculation 50 ft., 6 ksi TPbeam 50 ft., 12.9 ksi Hand Calculation 50 ft., 12.9 ksi TPbeam CRACK