Transverse Panel-to-panel Connections For Full-depth Precast . - PCI

10 1/2 in.5 ft 6 /4 in.1 in.5 ft 6 1/4 in.35 ft 11 1/2 in.Transverseprestressingstrands7 ft 0 in.HorizontalshearconnectorpocketsLooped no. 4 barsno. 6 bar extendingHollow structural steel tube into hollow structuralwith drop-in no. 6 barsteel tube to splicewith drop-in barDrop-in barClosed loopbarLooped barconnectionconnectionFigure 4. Plan view of test specimen with nonprestressed connections. Note: No. 4 13M; no. 6 19M; 1 in. 25.4 mm.placed in 1 4 in. (25 100 mm) flat plastic duct (Fig.5). One of the connections with a stress of 340 psi(2.34 MPa) had an epoxy bonding agent applied to thefaces of the panels prior to placement of the grout.significantly contribute to the laboratory connection studybecause of the short time between stressing and testing.The companion paper14 describes the procedures used todetermine prestress losses and stress redistributions toarrive at the required initial stress. The method used wasinvestigated by Bowers15 to recommend initial prestress forboth simply supported and continuous steel and concretegirder bridges. An age-adjusted effective modulus methodwas used to calculate long-term losses and stress redistributions between panels and girders due to creep, shrinkage, and relaxation. The precompression remaining at theend of service is summed with the stresses caused by liveload and superimposed dead load. An initial prestress isdeemed adequate if the final worst-case tensile stress in thetop of the deck is below a limit set to control cracking andprevent leakage at the connection. Long-term losses did not5 ft 11 1/2 in.1 in.5 ft 11 in.One objective of the physical tests was to determine the netservice stress limit required to control cracking and preventleaking under service load moment. The first two posttensioned connections had four strands stressed to providea design compressive stress of 164 psi (1.13 MPa) in thepanels. This prestress was designed to keep the maximumtotal tensile stress in the connection at service load below6(whereis the specified compressive strength ofconcrete) or 380 psi (2.90 MPa) for the 4000 psi(37.6 MPa) compressive strength connection grout.16The second two connections had eight strands stressed toprovide a design compressive stress of 328 psi (2.26 MPa)in the panels. This prestress was designed to keep themaximum total tensile stress in the connection below215 psi (1.48 MPa), or approximately 3.4(based on4000 psi).1 in.5 ft 11 1/2 in.2 ft 0 in.1 in. x 4 in. flat ductfor posttensioning7 ft 0 in.3 ft 0 in.2 ft 0 in.Blockouts forhorizontal shearconnectorsBlockout tocouple tendonductTransverse prestress1 in. wide female-female keyedconnectionFigure 5. Plan view of test specimens with posttensioned connections. Note: 1 in. 25.4 mm; 1 ft 0.305 m.PCI Journal S p r i n g 201465

8 in.Panel 2Panel 11 1/2 in.1 ft 9 1/2 in.Panel 3W21 x 101 beams5 ft 6 in.Centerlineload3 ft 0 in.Centerlineconnection3 ft 0 in.CenterlinesupportCenterlineconnection5 ft 6 in.Centerlinetie downFigure 6. Longitudinal elevation view of test specimen. Note: 1 in. 25.4 mm; 1 ft 0.305 m.Panel-to-panel connectionconstruction and test setupSpecimen constructionA double cantilever test setup was created to simultaneously test two transverse connections in negative bending(Fig. 6). A hydraulic actuator applied load to one end ofthe specimen while the opposite end was held down withtie rods. A roller was placed at the midpoint beneath thebottom flanges of the beams to create a double cantilever.Figure 2 shows the connections in each specimen, and Fig.4 and 5 show the specimens in plan view. Each specimenwas constructed with two W21 101 steel rolled shapes,three full-depth deck panels, and two connections. Eachpanel had four shear stud pockets (Fig. 4 and 5).The panels were cast by a precast concrete manufacturer andtransported to the structural engineering laboratory wherethe specimens were constructed. The panels were placed onthe supporting girders and set to the proper elevation withleveling bolts cast into the panels. The panel-to-panel connection formwork and haunch formwork were then placed.For the nonprestressed connections, grout for all connections was placed in a single operation. For the posttensionedconnections, the panel-to-panel connections were placed inthe first operation and the haunches and shear connectorsgrouted in a second step after posttensioning.A prepackaged grout was used for all of the shear studpockets and transverse connections. A previous study byScholz et al. suggested that this grout would perform wellin the transverse connections due to its low shrinkage,rapid strength gain, good flow, and good bond to hardenedconcrete.17 The manufacturer’s guidelines for mixing andproportioning were followed. One posttensioned connection(170 psi [1.17 MPa], extended grout) contained grout witha 3 8 in. (9 mm) nominal maximum size pea gravel extensionper the manufacturer’s recommendation. Another posttensioned connection (340 psi [2.34 MPa], neat grout, epoxied)had an epoxy bonding agent applied to the faces of thepanels prior to placement of the grout. All of the connectionswere vibrated with a pencil vibrator to consolidate the grout.66S pri n g 2 0 1 4 PCI JournalPrecompression was carefully monitored during theconstruction of the deck panels with posttensioning. Theposttensioning tendon ducts were spliced across the connections, the formwork for the connections was constructed, and the grout was placed in the connection. When thegrout achieved the required strength of 4000 psi(27.6 MPa) the tendons were threaded and stressed. Loadcells were attached to each end (the dead and stressingends), and the loads were monitored during stressing andseating to ensure that the correct total precompressionwas applied. All posttensioned panels had two 4-strandflat ducts. In the specimen with the lesser prestress, twostrands were placed and stressed in each duct. The specimen with the greater prestress had four strands per duct.The load cells indicated some loss of force in the strandsbetween stressing and grouting of the tendons (less than1 ksi [7 MPa]). The stress based on load cell readings immediately prior to grouting the ducts was considered to bethe effective prestress in the strands.Specimen testingThe specimens were tested both cyclically and statically. Astatic test was performed to establish initial behavior withthe load increased incrementally to 75 kip (334 kN). Thenthe specimens were loaded cyclically from 5 to 75 kip(22 and 334 kN), with periodic static and ponding tests, asdescribed in the following paragraphs.The specimens were loaded such that the stress range oneach connection simulated the worst-case tensile stressesin the deck caused by an HS-20 design truck (the designtruck used at that time). The maximum stress range overan interior bent was computed by doing a live load analysis of the actual bridge with the design vehicle. A load of75.0 kip (334 kN) applied to the cantilever systemresulted in the same stress at the panel-to-panel connections in the test specimens as the connections wouldexperience in the actual structure. The connections weresubjected to 1,000,000 cycles of stress applied at 1 Hzwith the actuator. The number of load cycles was basedon the number of trucks expected during the design lifeof the real bridge.

Discs for demountablemechanical extensometerDiscs for demountablemechanical extensometerVibratingwire gaugesTransverseconnectionPanelsGirdersElectrical resistancestrain gaugesVibrating wiregaugesElectrical resistancestrain gaugesElevationCross sectionFigure 7. Instrumentation at connections.During load application, data were collected during a staticload test every 10,000 cycles for the first 50,000 cycles andthen every 100,000 cycles until the end of the cyclic load.Strain gauges on the steel girders and vibrating wire gauges(VWGs) inside the concrete panels measured the strains atseveral locations through the depth of the composite girderso strain profiles could be investigated. Locating discs foruse with a demountable mechanical extensometer(DEMEC) gauge were attached to the top surface of thepanels directly adjacent to the connections to measure longitudinal strain across the connections. The DEMEC gaugepoints were used to measure the strain above the two girderlines and at the midpoint between the girders. Figure 7shows the instrumentation at each transverse connection.Crack developments in the connections were carefullymonitored and recorded throughout the cyclic tests. Thecracks were measured with a crack gauge prior to firstloading and after all loads had been applied. The patternswere recorded during the initial static tests and periodicallyduring cyclic loading. The top, bottom, and sides of thetransverse connections were monitored for cracks.The first ponding test was performed before the first staticload. Approximately 1 4 in. (7 mm) of water was pondedon the top surface of the bridge deck for two hours todetermine whether water penetrated the full depth of thedeck. When water was observed to penetrate through theentire depth at cracks in the panel-to-panel connections, thelocation of the leak and the time were recorded. Pondingwas conducted at 10,000, 50,000, 100,000, 500,000, and1,000,000 cycles, after which a ponding test was run concurrently with an additional 100 load cycles.Strengths of the concrete materials were determinedthroughout the test program. This included the panelconcrete strength when posttensioning was applied and onthe day of the first test. Grout strength was measured twodays after construction and on the first and last days oftesting. The actual strengths were used when analyzing thesystem.ResultsConcrete and grout materialpropertiesTable 1 presents the age and compressive strengths of theconcrete in the deck panels and the grout in the connections of the three specimens at the time of initiation oftesting. The concrete strength and modulus of elasticitywere determined with 4 8 in. (100 200 mm) cylinders. Grout strengths were determined with 2 in. (50 mm)cubes, so no modulus tests were performed. The actualdeck concrete strengths of over 7900 psi (54 MPa) wereconsiderably higher than the design compressive strengthof 6000 psi (41 MPa). The design grout strength was4000 psi (28 MPa), and the actual strengths at the time oftesting were between 3970 (27 MPa) and 5360 psi(37 MPa). The duration of the cyclic testing was between21 and 29 days, and material properties were measured atthe end of testing as well. Concrete strengths changed lessthan 300 psi (2 MPa), and the grout strengths increased anaverage of 500 psi (3.5 MPa) over the course of testing.Deck panel prestressThe load cells on the strands were used to measure theforce applied to the strands during stressing, after seating,and up until the time of testing. The first set of posttensioned panels was designed to have an effective stress of164 psi (1.13 MPa) during testing. The actual precompression during testing was 170 psi (1.17 MPa). The second setof posttensioned panels was designed for a compressivestress of 328 psi (2.26 MPa), and the actual precompressionduring testing was 340 psi (2.34 MPa). The stress in eachset of panels was slightly higher than the design value.Initial cracksin transverse connectionsDEMEC gauge and VWG strain readings were used todetermine the load at which the load versus strain behaviorPCI Journal S p r i n g 201467

Table 1. Material properties for specimens at time of test initiationSpecimenAge at start of test, daysCompressive strength,psiModulusof elasticity, ksi398330514074640n.d.12586506450Connection neat grout283970n.d.Connection extended grout284700n.d.10279505700695360n.d.MaterialPanel concrete1: nonprestressedconnectionsConnection groutPanel concrete2: posttensionedconnections, 170 psiPanel concrete3: posttensionedconnections, 340 psiConnection groutNote: n.d. no data. 1 psi 6.895 kPa; 1 ksi 6.895 MPa.first deviated from linearity. This was considered the firstcracking load.behavior until after 5000 cycles, at 70 kip (311 kN) ofapplied load.The DEMEC gauge points were read at 5 kip (22.2 kN)intervals during the initial static loading and at 10 kip(44.4 kN) intervals during static loading performed atstages during cyclic testing. Plots of load versus strainwere created; Fig. 8 presents the plots for the firststatic load test. The strains presented are the average ofthe three DEMEC gauge readings across each connection. All connections initially exhibited linear load versus strain behavior, and four connections had a distinctload at which the slope of the line decreased dramatically. This point was selected as the first cracking load(Table 2). The more highly stressed posttensionedconnections did not exhibit nonlinear load versus strainVibrating wire gauge readings were collected every30 seconds during each static load test. At each load step,the load was held constant to permit at least two readingsof the VWGs. Figure 9 presents the plots of VWG strainversus load. Before cracking, the strains increased linearly with applied load. Because the VWGs are immediatelyadjacent to the crack but do not cross it, after crackingthe strain in the gauges did not continue to increase atthe same rate. This is because there can be no increasein tension across the open crack. Therefore, based on theload versus strain plots, the cracking load was taken asthe load at which the slope of the line distinctly increases.Table 2 presents these loads.80Applied load, kip70Looped bar60Drop-in bar50170 psi neat grout40170 psi extended grout3020340 psi neat grout10340 psi neat grout epoxied faces00500100015002000250030003500Strain, microstrainFigure 8. Strains from DEMEC gauge readings for each connection on first loading. Note: DEMEC demountable mechanical extensometer. 1 kip 4.448 kN; 1 psi 6.895 kPa.68S pri n g 2 0 1 4 PCI Journal

Table 2. First cracking loads and stressesDEMECappliedfirstcrackingstress, psiVWGappliedfirstcrackingstress, psi127327340140040551Posttensioned170 psi,extendedgrout5550Post‑tensioned340 psi,neat grout70Posttensioned340 ksi,neat grout,epoxied70VWG netcrackingstress, 4705000510510340170170705000510510340170170VWG firstcrackingload, 55Post‑tensioned170 psi,neat groutConnectionServiceload cyclesat firstcrackDEMEC netcrackingstress, psiDEMEC firstcrackingload, kipInitial precompression, psiNote: DEMEC demountable mechanical extensometer gauge; VWG vibrating wire gauge. 1 kip 4.448 kN; 1 psi 6.895 kPa.For four of the connections, the cracking loads determinedwith the two methods were the same or similar (Table 2).For the drop-in bar connection and the 170 psi (1.2 MPa)with neat grout connection, there was a significant discrepancy between the cracking load as determined with eachmethod. It is difficult to determine which method moreaccurately represents the load at the first crack, so both arepresented. For all six connections the two methods detectedthe first crack on the same service load cycle.The key piece of information from the data is the nettensile stress at the time of first cracking. Table 2 presentsthe calculated stress at the extreme top fiber of each connection based on the moment at the face of the connectionclosest to the interior support and the uncracked, transformed section properties of the composite section. For theposttensioned connections, the net tension is calculated asthe tension due to the applied load minus the precompression. The net tensile stress to cause first crack varied from122 psi (0.8 MPa) to 400 psi (2.8 MPa) (Table 2). Averagestress at first crack was around 240 psi (1.7 MPa).Table 3 compares the cracking stresses with the compressive strength of the grout in the connections. The stress inthe connection at first cracking load varied from 1.9to 5.9, with an average based on both strain measurements of 3.3.Figures 10 and 11 present the load versus strain databased on DEMEC gauges and VWGs during the static loadtest performed after 1,000,000 cycles. The posttensionedconnections with 340 psi (2.34 MPa) precompressionperformed the best based on both types of strain measurements, with the plot remaining linear to a significantlyhigher load than the other connections. Also the maximumstrain at service load (75 kip [330 kN]) as measured withthe DEMEC gauge, which crossed the connections, wasthe smallest for the highly precompressed connections.PCI Journal S p r i n g 201469

80Applied load, kip70Looped bar60Drop-in bar5040170 psi neat grout30170 psi extended grout20340 psi neat grout100340 psi neat grout - epoxiedfaces020406080100120Strain, microstrainFigure 9. Strains from vibrating wire gauge readings for each connection on first loading. Note: 1 kip 4.448 kN; 1 psi 6.895 kPa.Strain profilein transverse connectionsThe strain profiles were determined based on the measurements from strain gauges on the girders and VWGsinside the concrete panels. Profiles were developed foreach static load test. The theoretical strain distribution was computed using the service load moments andthe uncracked transformed section properties for theprestressed cross sections and with the cracked transformed section properties for the two nonprestressedconnections.Figures 12, 13, and 14 show the strain profiles at eachconnection at loads of 40 kip and 75 kip (180 kN and330 kN), which were recorded during the static load testsperformed after 1,000,000 cycles of load had been completed. The calculated strain profiles at the 75 kip serviceload are shown. The nonprestressed connections exhibiteda nonlinear strain distribution (Fig. 12). The cracking atthe joint prevented the concrete adjacent to the VWG fromdeveloping significant tensile stresses and strains, so thestrain remained near zero throughout loading. The neutralaxis location was somewhat higher than predicted with thecracked transformed section analysis.Table 3. Cracking stresses relative to grout compressive strengthDEMEC net crackingstress, psiVWG net crackingstress, psiGrout compressivestrength, psiDEMEC net crackingstress/ fc'VWG net crackingstress/ fc'Looped reinforcingbars27327346404.04.0Drop-in reinforcingbars40029246405.94.3Posttensioned 170 psi,neat grout12223139701.93.7Posttensioned 170 psi,extended grout23119447003.42.8Posttensioned 340 psi,neat grout17017053602.32.3Posttensioned 340 psi,neat grout, epoxied17017053602.32.3ConnectionNote: DEMEC demountable mechanical extensometer gauge; fc' specified compressive strength of concrete; VWG vibrating wire gauge.1 psi 6.895 kPa.70S pri n g 2 0 1 4 PCI Journal

8070Looped barApplied load, kip60Drop-in bar50170 psi neat grout40170 psi extended grout30340 psi neat grout20340 psi neat grout epoxied faces1000500100015002000250030003500Strain, microstrainFigure 10. Strains from demountable mechanical extensometer gauge readings for each connection after 1,000,000 cycles. Note: 1 kip 4.448 kN;1 psi 6.895 kPa.Figures 13 and 14 present the strain profiles for the posttensioned connections. All indicate that some crackinghad occurred, but the performance of the more highlyprestressed (340 psi) connections was better. The strain distribution was closer to linear, and the neutral axis locationwas higher in the cross section. For all connections, someloss of composite action and shear lag effects due to theshort distance between load and connection could contribute to the non-linear strain distributions.Long-term cracking patternsThe crack patterns on the connection surfaces are anindication of the expected durability of the panel-to-panelconnections. The grout developed high early strengthbut shrank within weeks of placement. Hairline cracksappeared at many of the grout-to-concrete interfaces.Choosing a low-shrinkage grout is essential to limitinginitial cracking in the transverse connections. Although the8070Looped barApplied load, kip60Drop-in bar50170 psi neat grout40170 psi extended grout3020340 psi neat grout10340 psi neat grout epoxied faces0020406080100120Strain, microstrainFigure 11. Strains from vibrating wire gauge readings for each connection after 1,000,000 cycles. Note: 1 kip 4.448 kN; 1 psi 6.895 kPa.PCI Journal S p r i n g 201471

Position from bottom flange, in.30252040 kip looped bar1575 kip looped bar40 kip drop-in bar1075 kip d

the transverse panel-to-panel connections are referred to simply as panel-to-panel connections. Panel-to-panel connections can be nonprestressed or post- . (25 100 mm) flat plastic duct (Fig. 5). One of the connections with a stress of 340 psi (2.34 MPa) had an epoxy bonding agent applied to the