Flange-to-Flange Moment Connections for Precast Concrete ...

86 PCI JOURNAL

Flange-to-Flange Moment Connections for Precast Concrete Deck Bulb-Tee Bridge Girders

M. Reza Kianoush, Ph.D., P. Eng.ProfessorRyersonUniversityToronto,ON,Canada

Khaled M. Sennah, Ph.D., P. Eng.

AssociateProfessorRyersonUniversity

Toronto,ON,Canada

Bijal N. Shah, M.A.Sc., P. Eng.SeniorStructuralEngineerBechtelCorp.Houston,Tex.

Clifford Lam, Ph.D., P. Eng.HeadofBridgeResearchOntarioMinistryofTransportationSt.Catharines,ON,Canada

Siyin TuGraduateStudentRyersonUniversityToronto,ON,Canada

November–December 2006 87

and inconvenience to the public. Higher initial costs for pre-cast concrete bridge systems, compared with other materi-als, are justified in situations where extended lane closure on high-traffic-volume roadways must be avoided. Cost-benefit analysis makes a precast concrete bridge system made of deck bulb-tee (DBT) girders an attractive choice for bridge replacement projects. In this system, the concrete deck slab is cast with the prestressed, precast concrete girders at the fabrication facility and then shipped to the bridge site, where a cast-in-place concrete closure strip is placed between the girders, making the deck continuous and allowing for the dis-tribution of live load.

Currently, the Canadian Highway Bridge Design Code (CHBDC) and the AASHTO LRFD Bridge Design Specifica-tions do not provide guidance for the design of precast con-crete DBT girder connections.1,2 At this time, the literature has no information on the design of these joints. In particular, there are no supporting test data on connection failure modes and ultimate load-carrying capacities to give the designer confidence when designing connections for DBT girders.

The objectives of this study are twofold: to develop mo-ment connections for flanges of precast concrete DBT bridge girders and to provide experimental data on the ultimate strengths of these full-scale bridge deck connection systems. These objectives include identifying the crack formation and propagation patterns, failure modes, and ultimate load- carrying capacities of the connections. Detailed experimen-tal results from testing of moment connections for flanges of precast concrete DBT bridge girders and data from six full-scale bridge models with moment-transferring connec-tions are presented in this paper. The main study parameters considered are joint width between DBT girders, reinforce-ment details of the connections, and the position of the ap-plied wheel load. The experimental wheel load is compared with the design wheel loads specified in the CHBDC and in the American Association of State Highway and Transporta-tion Officials (AASHTO) load and resistance factor design (LRFD) specifications.

LITERATURE REVIEW

Various types of precast, prestressed concrete tee girders have evolved over the past few decades for use in building short-span bridges.3,4 Common girder shapes include single-tee, double-tee, and multiple-tee sections, which are placed contiguously and are suitable for bridges in the 6-m- to 24-m-span (20 ft to 80 ft) range. Some single-tee girder sec-tions can span up to 36 m (120 ft). Precast concrete sections are produced in standard widths of 1.2 m, 1.8 m, and 2.4 m (4 ft, 6 ft, and 8 ft). During construction, precast concrete bridge girders are transported to the site and erected adjacent to each other. V-shaped joints between the edges of adjacent girder flanges are filled with nonshrink mortar grout, and the girders are transversely post-tensioned to provide for lateral resistance and continuity for load transfer.5–7

In 1959, Concrete Technology Corp. developed the bulb-tee series, which had a 1.2-m-wide (4 ft) flange in several standardized depths, to increase span capabilities of precast

The many economic, logistical, and environmental benefits of precast concrete elements and systems in bridge construction have been the subject of growing

interest among bridge jurisdictions in North America. Sever-al bridges are deteriorated and require repair, rehabilitation, or replacement. Considerable economic benefit (in reduced on-site construction time and labor) can be achieved through mass production of precast concrete materials, repetitive use of forms, and plant quality control.

Bridge rehabilitation/replacement is time consuming and costly and presents major concerns related to work-zone safety and traffic disruptions. Bridge remediation efforts that necessitate lane closures of heavily traveled roadways are costly because of the disruption of commercial and industrial activities. Because of these costs and concerns, precast con-crete bridge technology is seen as the preferred solution for bridge rehabilitation and replacement.

Precast concrete elements and systems save time and money because these systems minimize design efforts, are quickly assembled, reduce adverse environmental site im-pacts, and substantially reduce traffic delays, lane closures,

This paper reports findings on new flange-to-flange moment connections for precast concrete deck bulb-tee (DBT) bridge girders. In bridge systems made of precast concrete DBT girders, the concrete deck slab is cast with the prestressed, precast concrete girders at the production facility and then shipped to the bridge site. To make the deck continuous and allow for live load distribution, a concrete closure strip is placed between the precast concrete DBT girders. The objectives of this study are to develop a moment connection for DBT bridge girders and provide experimental data on the ultimate strength of the connection system. Study parameters include identifying crack formation and propagation pat-terns, failure modes, and ultimate load-carrying capacities. Four moment connections were designed and six full-sized bridge panels were fabricated and loaded to failure to test the developed connections. Effects of steel reinforcement, joint width, and loca-tion of the simulated wheel load on the connection designs are discussed. A joint design is considered successful if, prior to failure, the experimental wheel load satisfies North American bridge code require-ments. Data show that most of the connection designs satisfy the wheel load specifications. Results indicate that the location of the wheel load on the deck slab affects the ultimate load-carrying capac-ity of the developed connections. Connections typi-cally failed due to flexure.

88 PCI Journal

AASHTO girders with composite decks.12 Two specimens used a full-scale, 200-mm-thick (8 in.), cast-in-place, nor-malweight concrete deck, and the other two specimens used 100-mm-thick (4 in.), lightweight precast concrete panels with a 100-mm-thick (4 in.), cast-in-place, normalweight concrete deck. The load-deflection curves and strengths for each pair of specimens were almost identical.

Abendroth investigated the nominal flexural and shear strengths of a composite slab system that used precast, pre-stressed concrete panels as a bridge subdeck.13 Five full-scale models of composite slab specimens were constructed and tested. Experimental results were compared with analytical results using yield-line and punching-shear theories. Aben-droth concluded that full-composite behavior was maintained between the reinforced concrete topping and the precast con-crete panel and that punching shear governed the nominal strength of the slabs.

In deck rehabilitation of frequently traveled bridges, full-depth precast, prestressed concrete panels are placed trans-versely on the supporting girders and post-tensioned lon-gitudinally. Yamane et al. developed a full-depth precast, prestressed concrete bridge deck panel system.14 Issa et al. investigated and evaluated this idea further, and a complete system was subsequently developed.15,16 The full-depth panel system includes stemmed precast concrete panels; transverse grouted joints between panels; longitudinal post-tensioning among panels; and welded, threaded, headless studs that make the panels act compositely with the girders.

Prefabricated Girder Connections

The PCI Committee on Connection Details published typi-cal details and design methods for standard connections for precast, prestressed concrete double-tee girders, including connections between girders in longitudinal and transverse joints.17,18 In this publication, a typical flange-to-flange dou-ble-tee girder connection is made of an inclined steel plate anchored to the concrete flange using a specially shaped steel rod. In this connection, shear is transferred between the plate and the rod through welding. Similar details were also pre-sented in Applied Technology Council’s Report ATC-8 for a tee girder flange connection with some modifications.19

Pincheira et al. carried out a pilot series of tests on double-

concrete girders. Arthur Anderson improved this design in 1969, developing the innovative DBT series with greater standardized flange widths of 1.5 m and 3 m (5 ft and 10 ft), each with several standardized depths ranging from 700 mm to 1900 mm (28 in. to 75 in.) and spanning capabilities up to 57 m (187 ft).8–10 Placed contiguously, DBT girders pro-vide a ready-made deck, eliminating the need for a cast-in-place concrete deck. Anderson also developed the Washing-ton Series 14 bulb-tee girder, which was standardized with some modifications in 1988 as the AASHTO-PCI bulb-tee series.11

Precast Concrete Bridge Deck Systems

Precast concrete DBT girder systems offer significant sav-ings in construction time and labor because bridge compo-nents are fabricated off-site and are efficiently assembled on-site. Other advantages of precast concrete DBT girders are the elimination of deck placement from the project schedule’s critical path and increased product quality as a result of fac-tory-controlled conditions. Proper design and construction of the joints between DBT girders, however, must be addressed to ensure adequate system performance.

Burns and Centennial performed experiments on Type I

Fig. 1.Cross-sectionalviewoftheprecastconcretedeckbulb-teegirderbridgesystem.Note:C/C=center-to-centerspacing.

2%2%

ShoulderShoulder Lane width Shoulder Lane widthShoulder

Barrier wallBarrier wallAsphalt pavement

Cast-in-place concrete Prefabricated girder element

C/C of girders C/C of girders C/C of girders C/C of girders

Fig. 2.IsometricviewofMTC-1connectiondetailsandreinforcement.


and Mufti presented load distribution in shear-connected con-crete plank bridges.22,23 The main purpose of this testing was to determine the suitability of various methods for analysis of bridge girder connections.

ExPERIMENTAL PROGRAM

Proposed Moment Connections for DBT Girders

In the precast, prestressed concrete industry, DBT gird-ers are manufactured in any combination of four depths and three flange widths. Available girder depths are 800 mm, 1100 mm, 1500 mm, and 1800 mm (32 in., 44 in., 60 in., and 72 in.). Available flange widths are 1200 mm, 1500 mm, and 2000 mm (48 in., 60 in., and 80 in.). Figure 1 shows a cross section of the prefabricated DBT girder bridge system.

All DBT girders are constructed from normalweight con-crete with the exception of the DBT 1800 (72 in.) series. The units in the 1800 series are cast with semi-lightweight con-crete for the bulb and the web but use normalweight concrete for the flange to increase the resistance of the girder’s top surface to the effects of salt penetration and abrasion. The relatively flexible DBT girders require full-depth diaphragms between girders to provide transverse deck stiffness. Also, the specified strength of the connection between the DBT girder

tee girders with flange-to-flange connectors to examine the strength and deformation capacity of the connectors when subjected to multi-axial and cyclic loading.20 The connector consisted of a steel plate with two filet-welded reinforcing bars embedded in a 50-mm-thick (2 in.) concrete slab.

Arockiasamy et al. studied the fatigue strength of joints between precast, prestressed concrete double-tee bridge gird-ers.6 A 1:3.5 scale model of a two-span, transversely and longitudinally post-tensioned, continuous, double-tee girder bridge system was tested in static and fatigue loading. The structural integrity of the bridge system under fatigue loading was checked, and experimental deflection of the system was compared with the results from the finite element analysis. Results of the analysis showed that the bridge system main-tained its structural integrity after 8 million cycles.

Hariatmadar studied the seismic response of connections between precast concrete double tees.21 Connections consist-ed of angles welded with anchor bars, headed studs, or both. Design interaction curves and associated equations for each connection type were developed. A practice-oriented method was developed to determine the connection strength under shear and axial forces and a combination of these forces at a joint between precast concrete double tees.

Bakht et al. carried out two field tests on shear-connected concrete plank bridges with a welded shear key, and Bakht

Fig. 3.CrosssectionoftheMTC-1connectiondetails.Note:C/C=center-to-centerspacing;T/B=topandbottom;1mm=0.03937in.;15M≈No.5steelbar.

of girder

Girder spacing

40

440200 200

15M @ 300 C/C T/B

CL

15M @ 300 C/C T/B

4045

5227070

90 PCI Journal

200-mm-wide (8 in.), 70-mm-thick (2.8 in.) projection slab that acts as a stay-in-place form for the concrete closure strip. The MTC-1 joint has a 75-mm-deep (3 in.), 40-mm-wide (1.6 in.) trapezoidal-shaped shear key along the girder length. It is assumed that DBT girders will be aligned such that a 25 mm (1 in.) gap is provided that is filled with a 25-mm- diameter form backer rod.

The second MTC (MTC-2) also has a joint width between DBT girders of 425 mm (17 in.). Figure 4 shows an iso-metric view of the joint details and the mild steel reinforce-ment, and Fig. 5 shows a cross section of MTC-2 and the structural details. In MTC-2, one side of the girder flange has a flat vertical face with a shear key, while the other side is formed with a 400-mm-wide (16 in.) and 70-mm-deep (2.75 in.) extended slab. Extending 400 mm beyond the inner edge of the joint are 15M (No. 5) top reinforcing bars and dowels. Bottom reinforcement with 180-degree hooks projects from the girder flange for embedment in the cast-in-place con-crete joint. Both sides of the joint have a 75-mm-deep (3 in.), 40-mm-wide (1.6 in.), trapezoidal-shaped shear key along the girder length. It is assumed that DBT girders will be aligned such that a 25 mm (1 in.) gap is provided that is filled using a 25-mm-diameter form backer rod. The MTC-1 joint is sym-metrical and the MTC-2 joint is not.

MTC-3 has a joint width between DBT girders of 225 mm (9 in.), and the top reinforcement and dowels extend 150 mm (6 in.) beyond the inner edge of the joint and bend 90 degrees with a hook length of 250 mm (10 in.) to provide the speci-fied development length of the mild steel reinforcement. The bottom reinforcing steel, however, does not extend into the joint. DBT girder flange ends are formed with a 100-mm-wide (4 in.), 155-mm- deep (6.1 in.) extended slab. The joint has a 75-mm-deep (3 in.), 40-mm-wide (1.6 in.), trapezoidal-shaped shear key along the girder length. Figure 6 shows an isometric view of the joint details and reinforcement, while Fig. 7 shows a cross section of the DBT girder MTC-3 and the structural details.

The fourth MTC (MTC-4) has a joint width of 325 mm (13 in.). Figure 8 shows an isometric view of the joint de-tails and reinforcement, and Fig. 9 shows the cross section of the girder connection with structural details. In MTC-4, the top reinforcement and dowels extend 270 mm (11 in.) beyond the inner edge of the joint and end with horizontal 180-degree hooks. Also with the 180-degree hooks, the bot-tom reinforcement projects from the girder flange for embed-ment in the cast-in-place concrete joint. DBT girder flange ends are formed with a 150-mm-wide (6 in.), 155-mm-thick (6.1 in.) extended slab. The joint has a 75-mm-deep (3 in.), 40-mm-wide (1.6 in.), trapezoidal-shaped shear key along the girder’s length.

Loading and Details

Taking into account the location of the simulated wheel load during testing, six bridge models, incorporating the four proposed connection details were tested to collapse. Figure 10 summarizes the description of these bridge connections and simulated wheel load locations. Note that two of the MTCs were loaded both eccentrically and concentrically,

flanges should provide continuity for live load distribution.Based on information obtained from the literature review

and from the CHDBC and AASHTO LRFD requirements for the design of deck slabs under wheel loading, four moment- transferring connections (MTCs) were developed. All four MTCs maintain the structural integrity of the bridge cross sec-tion and provide local deck-slab resistance based on the wheel load specifications. To provide the desirable monolithic con-nection between precast concrete elements, each MTC includes placement of a cast-in-place concrete closure strip between the precast concrete DBT girders.

The first MTC (MTC-1) has a 425-mm-wide (17 in.) joint between DBT girders and was chosen based on the develop-ment length of 15M (No. 5) straight, mild steel reinforce-ment. Figure 2 shows an isometric view of the joint details and reinforcement. Figure 3 shows a cross section of MTC-1 and the structural details. In MTC-1, the top reinforcement and dowels extend 400 mm (16 in.) beyond the inner edge of the joint. Bottom reinforcement with 180-degree hooks projects from the girder flange for embedment in the cast-in-place concrete joint. The flange end is formed with a

Fig. 4.IsometricviewofMTC-2jointdetailsandreinforcement.

Left side of joint

Right side of joint


of 300 mm (12 in.). This reinforcement represents 0.3% of the cross-sectional area, as CHBDC requires.

All girders and joints were cast with high-early strength con-crete with a minimum specified compressive strength of 50 MPa (7250 psi) at seven days. Joints were filled with grout with the same specified concrete compressive strength as that of the girders. After the grout reached its specified strength, the gird-

while the other two connections were only loaded concentri-cally. Table 1 summarizes the description of these models with respect to load conditions. The last letter, C or E, shown in bridge model symbols in the second column of Table 1 refers to concentric or eccentric loading, respectively, with respect to the joint’s centroid.

The experimental program investigated the behavior and ultimate load-carrying capacities of the MTCs under mono-tonic loading. Experimental data for deflection, concrete and steel strain, cracking load, cracking patterns, and ultimate load-carrying capacity were recorded and summarized. The DBT girders were constructed with different connection pat-terns at each girder end to maximize model use. Girders were placed side by side, and the concrete closure strip was placed to form an integral slab-girder system. For all models, the width of the girder’s deck slab (between girders) was 1800 mm (72 in.), the total span length was 2000 mm (80 in.), and the flange thickness was 225 mm (9 in.).

The DBT girders were simply supported, over neoprene bearing pads, with a clear span of 1500 mm (60 in.). Each gird-er was reinforced with three 15M (No. 5) mild steel reinforcing bars at the top and bottom, and 10M (No. 3) stirrups at a center-to-center spacing of 200 mm (8 in.). The precast concrete deck slab was reinforced in both directions, at the top and bottom, with 15M (No. 5) reinforcing bars at a center-to-center spacing

Fig. 5.CrosssectionoftheMTC-2girderconnectionwithstructuraldetails.Note:C/C=center-to-centerspacing;T/B=topandbottom;1mm=0.03937in.;15M≈No.5steelbar.

40

4045

5227070

44025

400

15M @ 300 C/C T/B

40050

15M @ 300 C/C T/B

Girder spacingof girderCL


92 PCI Journal

a summary of the details of MTCs, as well as the location of the simulated wheel load on the joint cross section.

Description of Bridge Models

Figure 11 shows the first bridge deck connection model M1, which includes MTC-1, as well as details of the shear key, closure strip, and form backer rod. Model M1 was loaded concentrically at the joint while a similar model, M2, was loaded eccentrically (Fig. 10). M2 was intended to examine whether the offset of the wheel load from the midspan location could affect the ultimate load-carrying capacity of the slab. Figure 12 shows the third bridge deck connection model, M3, which includes MTC-2, and the re-spective connection details. The fourth bridge deck connec-tion model, M4, was similar to bridge model M3 but was loaded eccentrically at the joint.

Figure 13 shows model M5, the fifth bridge deck con-nection model, which includes MTC-3. This connection is similar to MTC-1 in terms of the general connection cross-sectional profile. The main differences between the connec-tions, however, were that the 70-mm-thick (2.8 in.) extended

er system was loaded at the joint using simulated wheel load, distributed over the smallest wheel load contact area given in CHBDC, 250 mm × 250 mm (10 in. × 10 in.). Figure 10 shows


15M @ 300 C/C T/B

15M @ 300 C/C T/B

440150

40

100

Girder spacing

4045

5227070

of girderCL




440

40

120 150

15M @ 300 C/C T/B 15M @ 300 C/C T/B

150440

40

120 150

15M @ 300 C/C T/B 15M @ 300 C/C T/B

150

Girder spacing

4045

7070

of girderCL

Table 1. DescriptionofTestedModels,CrackingLoads,andUltimateLoadValues

Model Symbols Description of Symbols Cracking Load, kN

Ultimate Load, kN

Remarks

M1 MTC-1-C Moment-transferring connection No. 1 with concentric loading

333 541 Loading 1

–* 400 Loading 2

M2 MTC-1-E Moment-transferring connection No. 1 with eccentric loading

201 435


200 384

M4 MTC-2-E Moment-transferring connection No. 2 with eccentric loading

178 310


76 183


116 378

*Concrete already cracked.

Note: Letters C and E after connection type represent concentric and eccentric loading, respectively. 1 kN = 224.8 lbf.

94 PCI Journal

Fig. 10.Detailsofmoment-transferringconnections(MTCs)modelsM1throughM6.Note:LettersCandEafterconnectiontyperepresentconcentricandeccentricloading,respectively.

CONNECTION TYPE LOADING TYPE MO DEL M1 – MTC -1-C

MO DEL M2 – MTC -1-E

MO DEL M3 – MTC -2-C

MO DEL M4 – MTC -2-E




closure strip was reduced from 425 mm to 225 mm (17 in. to 9 in.). In this connection, the reinforcement in the 70-mm-thick (2.8 in.) extension slab did not extend into the cast-in-place concrete joint. Figure 13 shows details of the joint shear key for model M5.

The sixth bridge deck connection model M6, shown in Fig. 14, includes MTC-4. In this model, MTC-4 has the

slab at the joint was only 100 mm (4 in.) wide on each side of the girder flanges instead of the 200 mm (8 in.) width in models M1 and M2. Top reinforcement and dowels ex-tend about 175 mm (7 in.) from the inner face of the joint in MTC-3 and bend horizontally with a free length of 250 mm (10 in.) to provide the required development length. By bend-ing the reinforcement 90 degrees in plan view, the size of the

Fig. 11.Cross-sectionaldetailsofmodelsM1andM2afterjointgroutinganddetailsofshearkey(DetailA),jointopening,andformbackerrod(DetailB)formodelsM1andM2.Note:C/C=center-to-centerspacing;T/B=topandbottom.Alldimensionsareinmillimeters;1mm=0.03937in.;15M≈No.5steelbar.

1825

225

1800

200 25

1800

0010

200

Cross section of models M1 and M2 after grouting

40

2540

7555

200

DETAIL A

40

10

10

DETAIL B

30

NOTE:

Details of shear key, joint opening, and form backer rod for models M1 and M2

This detail was usedfor all specimens

25 mm diameterform backer rod

Sealant

Grout

Longitudional reinforcement2–15M T/B

See detail BSee detail A

of girderCLof girderCL

96 PCI Journal

Materials

Concrete with a seven-day minimum specified compres-sive strength of 50 MPa (7300 psi) was used in every bridge model. The concrete consisted of CSA Type 10 (equivalent

same reinforcement arrangement, joint shape, and size as MTC-1 except for a reduced width of 150 mm (6 in.) for the 70-mm-thick (2.8 in.) extension slab. Figure 14 shows details of the shear key for model M6. Figure 15 shows the joints prior to grouting of models M1 through M6.

Fig. 12.CrosssectionofmodelM3afterjointgroutinganddetailsofshearkeyandjointopeningformodelsM3andM4.Note:C/C=center-to-centerspacing;T/B=topandbottom.Alldimensionsareinmillimeters:1mm=0.03937in.;15M≈No.5steelbar.

1825

2518001800

200 200

Cross section of model M3 after grouting

Details of shear key and joint opening for models M3 and M4

225

0010

Grout


See detail ASee detail B

of girderCLof girderCL


eter and 200 mm (8 in.) high, were cast concurrently with the bridge models. The cylinders were kept beside the models to ensure the same curing conditions. The concrete cylinders were tested on the same day the bridge model was tested to

to ASTM Type I) normal portland cement; natural gravel with nominal maximum aggregate size of 9.5 mm (0.4 in.); natural sand; and a polynaphthalene, sulfonic-acid-based su-perplasticizer. Standard cylinders, 100 mm (4 in.) in diam-

Fig. 13.CrosssectionofmodelM5aftergroutingoperationanddetailsofshearkeyandjointopeningformodelM5.Note:C/C=center-to-centerspacing;T/B=topandbottom.Alldimensionsareinmillimeters:1mm=0.03937in.;15M≈No.5steelbar.

1825

200 25 200

18001800


Details of shear key and joint opening for model M5


Grout

See detail A

0010

225

of girderCL of girderCL

98 PCI Journal

Instrumentation

Electrical strain gauges measured the longitudinal strain in the mild steel reinforcement at critical locations in the bridge deck slab. The length of the steel strain gauge was 10 mm

obtain the concrete’s actual compressive strength. An aver-age of three cylinders were cast for each bridge model test. Commercially available mild steel reinforcement with a 400 MPa (60 ksi) yield stress was used in the concrete deck slabs and DBT girders for every bridge model.

Fig. 14.CrosssectionofmodelM6aftergroutingoperationanddetailsofshearkeyformodelM6.Note:C/C=center-to-centerspacing;T/B=topandbottom.Alldimensionsareinmillimeters:1mm=0.03937in.;15M≈No.5steelbar.

20025

1800

200

1825

1800


Details of shear key and form backer rod for model M6

225

0010

Grout

Longitudional reinforcement2–15m T/B

See detail A

of girderCL of girderCL


Load was applied at the midspan of each model using a hy-draulic jack. A load cell was connected to the data acquisition system to measure the applied jacking load. All models except M2 and M4 were tested under monotonic, increasing concen-tric loading at the centroid of the joint. Models M2 and M4 were loaded eccentrically, in which the edges of the wheel load lined up with the edge of the joint. Figure 16 shows the test setup for model M1, and Fig. 17 is a photo of the DBT girder models in the testing facility.

Concrete and steel strain gauges and LVDTs were connected to a data acquisition system to record their readings. Jacking load was applied in increments and was kept constant during deflection and strain data recording. Crack propagation and changes in model geometry were observed during testing. Cracks in the bottom and top surfaces of the deck slab and on the sides of the models were traced and marked with the cor-responding load values in kips (Fig. 18). Subsequent to model failure, the applied load was released slowly (failure is assumed to occur when the model cannot carry any additional load).

(0.4 in.), and it had a resistance of 120 ± 2% Ohms and a gauge factor of 2.11 ± 1%. Electrical strain gauges of the same resis-tance and gauge factor measured the strain on the top surface of the deck slab. All of the strain gauges were placed at the mid-span of each bridge deck. Linear variable differential transduc-ers (LVDTs) with a 50 mm (2 in.) electrical stroke measured deck slab deflections for each model.

Experimental Setup and Testing Procedure

The DBT girders in each bridge model were supported on elastomeric bearing pads at their ends to allow rotation of the girders under load. At each girder end, two 12-mm-thick (0.5 in.) plates were inserted between the bottom surface of the DBT girder and the top surface of the bearing pad to transfer the load over the entire pad surface. LVDTs were placed under the deck slab at five locations to measure bridge deflection under increasing load. Models were tested under increasing monotonic concentric or eccentric vertical load until collapse.

Fig. 15. DesignofmodelsM1andM2jointreinforcement;close-up,cross-sectionalviewofjointreinforcingsteelformodelsM3andM4;reinforcingsteelinmodelM5;andreinforcingsteelinplanviewofmodelM6.

Models M1 and M2

Model M5

Models M3 and M4

Model M6

100 PCI Journal

ExPERIMENTAL RESULTS

Table 1 lists the experimental cracking load and ultimate load capacities of the six bridge models. Cracking load is typically taken as that which causes the first crack to form on the sides of the joint or at the top or bottom of the bridge

deck. In all models, the first crack occurred along the cold joint. Ultimate load is taken as the load that causes the failure of the bridge model.

In all models, the maximum tensile strains in the mild steel reinforcement were observed at the joint (Fig. 19). Substantial-ly less strain was recorded in strain gauges close to the girder

Fig. 16.Testsetup,elevation,andplanviewsformodelM1.Note:Alldimensionsareinmillimeters;1mm=0.03937in.

1800

4025

25

PLAN

100

1825

250

ELEVATION

L1

2000

1000

L5 L2

1800

100

250

1500

250

250

L4L3

P

2100

502

C loadL

L

LL

C joint

W section

Hydraulic jack

C of girder

LVDT

Floor level

C joint

Elastomericbearing pad

C of girder

C o

f su

ppor

tC

of

supp

ort

Steel plate

C of girder

C of girder

Spacer

Load cell

Steel plate

Elastomeric bearing pad Steel plate

250 × 250 × 25 mm

2 nos. 300 × 600 × 12 mm250 × 250 × 15 mm

250 × 250 × 15 mm

2–300 × 600 × 12 mm

L

L

LL

L


(Fig. 20). Maximum concrete compressive strains were noted at the top of the deck in the center line of the joint (Fig. 21). In all of the models, cracks developed diagonally from the bot-tom of the cast-in-place concrete joint and extended upward to the top of the deck on the side of the joint. Cracks developed diagonally outward from the load location toward the web at the bottom of the deck.

In every model, failure was initiated in the connections due to flexure behavior consistent with monolithic construction. Testing was terminated when the slab could not absorb any ad-ditional load and when significant deformations were observed at the connections. No sign of punching-shear failure or other failure modes was observed. Additional details of the experi-mental results are reported by Sennah et al.24

Table 2 shows the experimental ultimate wheel load (ex-cluding load and dynamic load allowance [DLA] factors) and design wheel loads specified by CHBDC and AASHTO LRFD. Wheel load specifications of 87.5 kN and 72.5 kN (19.7 kip and 16.3 kip) were used from the CHBDC and AAS-HTO LRFD, respectively. The CHBDC live load factor of 1.7, the AASHTO LRFD live load factor of 1.75, and the DLA fac-tors of 0.4 and 0.33 (for CHBDC and AASHTO, respectively) were applied. The ratio of the experimental wheel load to the specified wheel load is considered satisfactory when this ratio is greater than 1.0, as was the case for all bridge models except for model M5. In model M5, the ratio (0.88) was not satisfac-tory for the specified CHBDC design wheel load, but it was satisfactory (1.08) based on the AASHTO LRFD wheel load.

Effect of Loading Condition among MTCs

Models M1 and M2—Model M1 was first tested without bearing pads at the DBT girder supports, leading to spalling of the girder webs without any observed sign of joint dis-tress. Subsequently, the model was tested with bearing pads used for girder support. For model M1, the load and deflec-tion values were taken from the latter test because they more- closely matched the boundary conditions of other models. The experimental ultimate load capacity of the concentrically load-ed bridge model M1 was 8% less than that of the eccentrically loaded model M2 (Table 1). Model M1, however, deflected 4% more than model M2 did at the center of the joint (Fig. 22) be-cause model M1 was previously loaded (without bearing pads) and had already cracked. In model M1, compressive strains in the concrete deck were much less than those in model M2 (Fig. 21). The DBT girders in models M1 and M2 behaved similarly with even stress distribution on either side of the joint.

Models M3 and M4—The experimental ultimate load ca-pacity of the concentrically loaded model M3 was 24% more than that of the eccentrically loaded model M4 (Table 1). Model M3 deflected 7% more than model M4 at the center of the joint at failure (Fig. 22). In both models, the strains in the reinforcement on the vertical side of the joint (left side) were less than those on the other side of the joint. Model M3 demon-strated more ductile behavior than model M4 (Fig. 22). Stress distributions were uneven in both models, however, due to the nonsymmetric shape of the cast-in-place concrete strip.

Based on observations, it is concluded that the load-carrying

capacities of MTCs depend on the location of the wheel load on the joint, and the stress distribution depends on the joint’s cross-sectional profile. From the failure mode of the connec-tion and crack pattern, it may be concluded that the weakest section in the connection is at the cold joint between the grout and the vertical face of the precast concrete DBT girder flange end.

Effect of Connection Details among MTCs

Models M1 and M5—Major differences between models M1 and M5 are joint width and reinforcement details. Both models were loaded concentrically at the joint. The experimen-tal cracking load of model M1 was observed to be three times that of model M5, and the ultimate load capacity of model M1 was twice that of model M5 (Table 1). Model M1 deflected 1.3 times more than did model M5 at ultimate load (Fig. 22). In model M5, due to bent reinforcement at the joint, stresses concentrated at the reinforcement located at the center of the joint, causing the reinforcement to yield locally before load was transferred away from the joint. Model M1 behaved in a more ductile manner, however, than model M5 (Fig. 22). In model M5, reinforcement details caused the stress distribution at the joint and in the girder to be uneven.

Models M1 and M3—The major difference between mod-els M1 and M3 is the cross-sectional profile of the precast con-crete DBT girder flanges at the joint. Both models were loaded concentrically at the joint. The experimental cracking load for model M1 was observed to be 12% more than that of model M3, and the ultimate load capacity of model M1 was 4% more than that of model M3 (Table 1). At the center of the connec-tion, model M1 deflected 19% less than did model M3 at the ultimate load (Fig. 22). In model M3, reinforcement strains in the girder at the vertical side of the joint were less than those on the other side of the joint (with the concrete extension slab projection). In model M1, strain distributions in the reinforce-ment were even on both sides of the joint. Both models M1 and M3 behaved monolithically, but stress distribution in the girders at the joint were uneven in model M3 due to the joint’s lack of symmetry.

Fig. 17.Deckbulb-teegirdermodelsinthesteeltestingfacilityattheStructuralLaboratoryatRyersonUniversityinToronto,ON,Canada.

102 PCI Journal

Models M1 and M6—The major differences between mod-els M1 and M6 are joint width and reinforcement details. Both models were loaded concentrically at the joint. The experimen-tal cracking load for model M1 was observed to be 1.9 times that of model M6, and the ultimate load capacity of model M1

was 6% more than that of model M6 (Table 1). Model M1 deflected 10% more than model M6 at ultimate load (Fig. 22). At the same load increments, model M6 exhibited more com-pressive strain in the top surface at the center of the joint. In model M6, due to bent reinforcement at the joint, stresses con-

Fig. 18.ViewsofcrackpatternsonfrontandbacksidesofjointsaftertestingformodelsM1,M2,M3,M4,M5,andM6.

Model M1 Model M2

Model M3 Model M4

Model M5 Model M6


Fig. 19.Load-to-straincurvesforreinforcingsteelstraingaugeatjointformodelsM1throughM6.Note:1kN=224.8lbf;1m=3.2808ft.

0

100

200

300

400

500

0 1000 2000 3000 4000 5000

M1

M2

M3

M4

M5

M6

Strain (µm/m)

Ap

plie

d lo

ad (

kN)

Fig. 20.Load-to-straincurvesforreinforcingsteelstraingaugeatgirderwebsformodelsM1throughM6.Note:1kN=224.8lbf;1m=3.2808ft.

0

100

200

300

400

500

-50 0 50 100 150 200

M1

M2

M3

M4

M5

M6

Strain (µm/m)

Ap

plie

d lo

ad (

kN)

104 PCI Journal

centrated at the reinforcement in the center of the joint and less load was transferred away from the joint compared with model M1. Models M1 and M6 both behaved in a ductile manner (Fig. 22).

Models M5 and M6—The major differences between models M5 and M6 were joint width and reinforcement de-tails. Both models were loaded concentrically at the joint. The observed experimental cracking load of model M5 was 35% less than that of model M6, and the ultimate load capac-

ity of bridge model M5 was 52% less than that of model M6 (Table 1). Model M5 deflected 20% less than model M6 at ultimate load (Fig. 22). Model M5 exhibited higher compres-sive strain at the top surface at the center of the joint compared with model M6.

Based on the observations described, it may be concluded that the load capacities of MTCs and their stress distribu-tions depend on the joint width and reinforcement details. Furthermore, the geometry of the side faces of the DBT girder flanges at the joint affects the load capacities of the

Fig. 21.Load-to-straincurvesforstraingaugesattopofdeckandcenterofjointformodelsM1throughM6.Note:1kN=224.8lbf;1m=3.2808ft.

0

100

200

300

400

500

-4000 -3000 -2000 -1000 0

M1

M2

M3

M4

M5

M6

Strain (µm/m)

Ap

plie

d lo

ad (

kN)

Table 2 RatiooftheExperimentallySpecifiedWheelLoadtoCHBDCandAASHTOLRFDSpecifiedValues

Model Description Experimental Ultimate Wheel Load Pu, kN

CHBDC (2000) AASHTO LRFD (1998)

Experimentally Specified Wheel Load* = Pu /(1.7× 1.4), kN

Ratio = (Column 4 Value)/87.5

Experimentally Specified Wheel Load* = Pu/(1.75 × 1.33), kN

Ratio = (Column 6 Value)/72.5

Column 1 Column 2 Column 3 Column 4 Column 5 Column 6 Column 7

M1 MTC-1-C 400 168 1.92 172 2.37

M2 MTC-1-E 435 183 2.09 187 2.58

M3 MTC-2-C 384 161 1.84 165 2.27

M4 MTC-2-E 310 130 1.49 133 1.84

M5 MTC-3-C 183 77 0.88 79 1.08

M6 MTC-4-C 378 159 1.82 162 2.24*Service design load = ultimate load/(load factor × DLA).

Note: AASHTO = American Association of State Highway and Transportation Officials; CHBDC = Canadian Highway Bridge Design Code; DLA = dynamic load allowance; LRFD = load and resistance factor design; 1 kN = 224.8 lbf.


MTCs. In model M5, due to a small joint width and few bent joint reinforcing bars, stresses concentrated at the re-inforcement located at the center of the joint, and there was also less load transfer away from the joint compared with model M6. Model M6 behaved in a more ductile manner than model M5 (Fig. 22).

DESIGN CONSIDERATIONS

The tested bridge models represent one panel of a typical precast concrete DBT girder bridge system prototype, with a girder spacing of 1.8 m (6 ft) and a deck slab thickness of 225 mm (9 in.). Results of bridge model testing may be applied with confidence to bridge prototype designs using multiple pre-cast concrete DBT girders with cross sections used in this study (Fig. 1), but with the following considerations:• In reference to the precast concrete DBT girder bridge

system prototype shown in Fig. 1, the presence of more than two precast concrete DBT girders would increase girder rotational restraint near the wheel loads, thereby increasing the load capacity of the locally loaded deck slab. In other words, the improved lateral restraint pro-vided by adjacent girders in a prototype bridge would allow arching action to develop in the deck slab, which would significantly increase the deck’s load capacity.

• The presence of more than two DBT girders (in refer-ence to Fig. 1) would significantly enhance load- distribution characteristics such that the additional gird-ers contribute resistance to the localized wheel loads over the deck slab connection.

• Typically, bridge prototypes have transverse dia-phragms at abutments, piers, and intermediate locations between supports. Transverse diaphragms enhance girder rotational restraint against local deck slab defor-mation, magnify arching-action effects, and encourage punching-shear deck slab failure rather than flexural failure. This restraint mechanism is evident in model M1, where the maximum jacking load reached 541 kN (122 kip) for load case 1, while the jacking load in load case 2 reached only 400 kN (90 kip). The release of lateral restraint for load case 2 was caused by the use of bearing pads at the DBT girder end supports, which prevented local bearing failure of the concrete.

• The experimentally simulated wheel load was applied over a surface area of 250 mm × 250 mm (10 in. × 10 in.) as specified in the CHBDC. The presence of a 75-mm-thick (3 in.) wearing surface, however, would increase the dispersion area of the wheel load, thereby increasing the load capacity of the deck slab at the joint location.

• Use of a girder model length of 2 m (7 ft) created a deck-slab aspect ratio of slightly more than two. In the prototype bridge, the slab aspect ratio is large enough to force the wheel load to be transferred over a greater length to the nearest girder webs. An increased aspect ratio in a prototype bridge also improves the longitu-dinal restraint provided to the deck slab, thus allowing a more effective arching action to be developed in the slab and a consequent increase in its load capacity.

• In the bridge models, the joints are filled with grout of the same compressive strength as the deck slab

Fig. 22.Load-to-deflectioncurvesatthecenterofthejointfromlinearvariabledifferentialtransducers(LVDTNo.1)formodelsM1throughM6.Note:1kN=224.8lbf;1mm=0.03937in.

0

100

200

300

400

500

0 5 10 15 20

M1

M2

M3

M4

M5

M6

Deflection (mm)

Ap

plie

d lo

ad (

kN)

106 PCI Journal

concrete. In practice, a nonshrink, rapid-set grout is used to reduce joint curing time. This research assumes the strength of the grout is equal to or more than that of deck slab concrete.

• It may be possible to reduce the overall shear-key joint width by designing for joint confinement in order to re-duce the development length of reinforcing bars. Details of this type of connection are reported by Tadros and Baishya and by Badie et al.25,26

CONCLUSIONS

This paper reports the test results of four flange-to-flange MTCs for DBT bridge girders to determine the load-deflec-tion and load-strain behavior, cracking and ultimate loads, and failure modes. Results address two aspects of precast concrete connection design: specimen failure mode and factors affecting connection load capacity. The following conclusions can be drawn from Tables 1 and 2 and Fig. 18 through Fig. 22:

• The failure mode of all bridge models is due to flexure at the connections; none of the models failed due to punching shear of the bridge deck. Flexural failure is primarily due to the geometric details of the test specimens used in this study at the connections; arching action was not appreciable because of a lack of proper confinement in the deck slab. Consequent-ly, test results for the connection load capacity are regarded as conservative, lower-bound values.

• The ultimate capacity of an MTC depends on the po-sition of the simulated wheel load with respect to the location of the joint. Concentric loading conditions have higher ultimate load capacities than do eccentri-cally loaded conditions.

• The ultimate capacity of the MTC depends on the joint’s geometric cross-sectional profile and rein-forcement details. The ultimate load capacity of model M1 was 4% more than that of model M3 due to different joint geometry for the same loading condition. The ultimate load capacity of model M5, however, was 50% less than that of model M1 due to a smaller joint width and different reinforcement details.

• With the exception of model M5, all models satis-fied the CHBDC wheel load requirement of 87.5 kN (19.7 kip). Only model M5 had 12% less load capac-ity than that required by the CHBDC.

• All models satisfied the AASHTO LRFD wheel load requirement of 72.5 kN (16.3 kip).

ACKNOWLEDGMENTS

The authors gratefully acknowledge the financial sup-port of the Ministry of Transportation of Ontario for this research. Opinions expressed in this paper are those of the authors and do not necessarily reflect the views and policies of the Ministry.

REFERENCES

1. Canadian Standard Association (CSA). 2000. Canadian High-way Bridge Design Code: A National Standard of Canada. CAN/CSA-S6-00. Toronto, ON, Canada: CSA.

2. American Association of State Highway and Transportation Officials (AASHTO). 1998. LRFD Bridge Design Specifica-tions. 2nd ed. Washington, DC: AASHTO.

3. Curtis, R. B. 1967. Single T-Beam Bridges. PCI Journal, Vol. 12, No. 2 (March–April): pp. 76–81.

4. Kwei, G. C. S. 1967. Discussion of Single T-Beam Bridges. PCI Journal, Vol. 12, No. 5 (September–October): pp. 82–83.

5. Shahaway, M. E. 1990. Feasibility Study of Transversely Pre-stressed Double-T Bridges. PCI Journal, Vol. 35, No. 5 (Sep-tember–October): pp. 56–69.

6. Arockiasamy, M., A. P. Badve, B. V. Rao, and D. V. Reddy. 1991. Fatigue Strength of Joints in a Precast Prestressed Con-crete Double Tee Bridge. PCI Journal, Vol. 36, No. 1 (Janu-ary–February): pp. 84–96

7. Shahaway, M. E., and M. Issa. 1992. Load Testing of Trans-versely Prestressed Double-T Bridges. PCI Journal, Vol. 37, No. 2 (March–April): pp. 86–99.

8. Anderson, A. R. 1957. How Beam Design Affects Prestressed Concrete Bridge Costs. Engineering News-Record (October): pp. 326–328.

9. Anderson, A. R. 1972. Systems Concepts for Precast Pre-stressed Concrete Bridge Construction. In Special Report 132, pp. 9–21. Washington, DC: Highway Research Board.

10. Anderson, A. R. 1973. Stretched-Out AASHTO-PCI Beams, Type III and IV, for Longer-Span Highway Bridges. PCI Journal, Vol. 18, No. 5 (September–October): pp. 32–49.

11. Geren, K. Y., and M. K. Tadros. 1994. The NU Precast Pre-stressed Concrete Bridge I-Girder Series. PCI Journal, Vol. 39, No. 3 (May–June): pp. 26–39.

12. Burns, N. H., and Z. Centennial. 2001. Composite Action be-tween Girder and Bridge Deck with Precast Panels. PCI Jour-nal, Vol. 46, No. 1 (January–February): p. 85.

13. Abendroth, R. E. 1995. Nominal Strength of Composite Prestressed Concrete Bridge Deck Panels. ASCE Journal of Structural Engineering, Vol. 121, No. 2: pp. 307–318.

14. Yamane, T., M. K. Tadros, S. S. Badie, and M. C. Baishya. 1998. Full Depth Precast, Prestressed Concrete Bridge Deck System. PCI Journal, Vol. 43, No. 3 (May–June): pp. 50–66.

15. Issa, M. A., A. A. Yousif, and M. A. Issa. 2000. Experimental Behavior of Full-Depth Precast Concrete Panels for Bridge Rehabilitation. ACI Structural Journal, Vol. 97, No. 3 (May–June): pp. 397–407.

16. Issa, M. A., A. A. Yousif, M. A. Issa, I. I. Kaspar, and S. Y. Khayyat. 1998. Analysis of Full Depth Precast Concrete Bridge Deck Panels. PCI Journal, Vol. 43, No. 1 (January–February): pp. 74–85.

17. PCI Committee on Connection Details. 1995. Addendum to

ResultsreportedinthispaperarepartofanongoingresearchprojectatCanada’sRyersonUniversityinToronto.Thenextphaseofresearchwillberelatedtoconnectionfatigueloading,animportantaspectofdesignforservice-loadconditions.


Design and Typical Details of Connections for Precast and Prestressed Concrete. PCI Journal, Vol. 40, No. 5 (Septem-ber–October): pp. 46–57.

18. PCI Committee on Connection Details. 1998. Standard Pre-cast Connections. PCI Journal, Vol. 43, No. 4 (July–August): pp. 42–58.

19. Applied Technology Council (ATC). 1981. Proceedings of a Workshop on Design of Prefabricated Concrete Buildings for Earthquake Loads. Report ATC-8. Berkeley, CA: ATC.

20. Pincheira, J. A., M. G. Oliva, and F. I. Kusumo-Rahardjo. 1998. Tests on Double-Tee Flange Connectors Subjected to Monotonic and Cyclic Loading. PCI Journal, Vol. 43, No. 3 (May–June): pp. 82–96.

21. Hariatmadar, H. 1997. Seismic Response of Connections in Precast Concrete Double Tees. Ph.D. diss. Department of Civil Engineering, McGill University, Montreal, QC, Canada.

22. Bakht, B., A. A. Mufti, and D. D. Sargent. 2001. Assessment of Stresses in Welds of Shear-Connected Concrete Plank

Bridges. From the 29th Annual Conference of the Canadian Society for Civil Engineering. Victoria, BC, Canada.

23. Bakht, B., and A. A. Mufti. 2001. Load Distribution in Shear-Connected Concrete Plank Bridges. From the 29th Annual Conference of the Canadian Society for Civil Engineering. Victoria, BC, Canada.

24. Sennah, K., M. R. Kianoush, B. Shah, S. Tu, and M. Al-Hashimy. 2004. Innovative Precast/Prestressed Concrete Bridge Systems and Connection Technology: Experimental Study. Ministry of Transportation of Ontario Highway Infra-structure Innovation Funding Program, ON, Canada.

25. Tadros, M. K., and M. Baishya.1998. Rapid Replacement of Bridge Decks. Report 407. Transportation Research Board, National Research Council.

26. Badie, S. S., M. C. Baishya, and. M. K. Tadros. 1999. Innova-tive Bridge Panel System a Success. Concrete International, V. 21, No. 6.

Design, Build, and Maintain Concrete Parking Structures

ACI/PCI One-day Seminar

For dates and locations,or more information on this

or other seminars, please visit www.concreteseminars.com or www.pci.org

Flange-to-Flange Moment Connections for Precast Concrete ...

Documents