U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
REPORT |
This report is an archived publication and may contain dated technical, contact, and link information |
|
Publication Number: FHWA-HRT-17-093 Date: February 2018 |
Publication Number: FHWA-HRT-17-093 Date: February 2018 |
Four connection design details were investigated with full-scale testing in this study using AASHTO type BII-36 box beams. These beams have a cross section of 36 inches (914 mm) wide, 33 inches (838 mm) deep, and 50 ft (15.2 m) long. Each test consisted of two box beams connected using one of the four connections being investigated. This chapter introduces the connection designs, construction materials, and construction procedures. It then presents the setup of the thermal loading and cyclic structural loading tests, including the loading protocols and instrumentation used.
Four connection designs were evaluated in this study. The first two used conventional high-strength non-shrink grout. One had a partial-depth connection, and the other had a full-depth connection, as shown in figure 1 and figure 2, respectively. It should be noted that, in accordance with common practice, the designs with conventional grout utilized transverse PT. Photographs of the partial-depth conventional grout shear key prior to casting and the means of applying the transverse PT force are presented in figure 3 and figure 4, respectively.
The other two connections investigated were new design details that take advantage of the enhanced mechanical and durability properties of UHPC. The UHPC connections used the same two basic details: one had a partial-depth connection, and the other had a full-depth connection, as shown in figure 5 and figure 6, respectively.
The connection designs included two different surface finishes. The conventionally grouted connection used a sandblasted (SB) surface finish (see figure 7), while the UHPC connection used an exposed aggregate (EA) surface finish (see figure 8). The EA surface finish was created by applying a gelatinous set retarder to the formwork. This delayed the hydration reaction in the concrete, and the unhydrated paste was washed off with water after the formwork was removed. The EA surface preparation has been suggested for field applications of UHPC.(15–17)
The design of the UHPC connection included protruding reinforcing steel from each of the precast box beams. These rebars formed a non-contact lap splice when filled with UHPC. With UHPC, a reduced embedment length for deformed reinforcements allowed for simplified design and construction of connection details. An embedment length of 5.5 inches for a No. 4 bar (140 mm for an M13 bar) in the UHPC connections was used in this study and has been demonstrated to be sufficient to develop the yield strength of the steel bar.(18,19) Transverse PT was only applied to the UHPC beams during the casting of the connection to ensure stability of the system during construction. The partial-depth UHPC connection can be seen during and after alignment, as shown in figure 9 and figure 10, respectively.
The concepts for the partial- and full-depth connections are presented in figure 11 and figure 12, respectively. The box beam designs and connection detail dimensions are presented in figure 113 through figure 116 in the appendix. Each box beam has two connection details, one on each side. Figure 13 and figure 14 show the same box beams aligned for the conventional grout and the UHPC connections, respectively. With this design, each beam could be tested with both a conventional grout and a UHPC connection. The conventionally grouted connection was tested first, and the two beams were then separated and arranged to test the UHPC connection.
Figure 11. Illustrations. Partial-depth connection details for the conventional and UHPC connections.
Figure 12. Illustrations. Full-depth connection details for the conventional and UHPC connections.
Source: FHWA.
Figure 13. Photo. Partial-depth beams aligned for connection with the conventional grout shear key connection.
Source: FHWA.
Figure 14. Photo. Partial-depth beams aligned for connection with the UHPC shear key connection.
Two field-cast cementitious materials were used in the connection details. One was a conventional non-shrink grout, and the other was UHPC.
The conventional grout material used in this study was a portland cement-based, prepackaged, non-shrink grout. A water-to-solids ratio of 0.17 was used, and the grout reached an average compressive strength of between 7,800 and 8,120 psi (54 and 56 MPa) at the time of testing. Further assessment of the properties of this grout can be found elsewhere.(20,21) A summary of the properties of the grout is available in table 1. More details about the non-shrink grout and the UHPC can be found in research by Graybeal et al. (See references 16–18 and 20–24.)
1 ksi = 6.89 MPa.
1 inch = 25.4 mm.
1 lb/ft3 = 16 kg/m3.
aValues recorded in this study.
bValues reported by De la Varga and Graybeal.(20)
Advances in the science of concrete materials have led to the development of a new class of advanced cementitious materials, namely UHPC. These concretes tend to contain high amounts of cementitious materials and a very low water-to-cementitious materials ratio as well as exhibit high compressive and tensile strengths. Discrete steel fiber reinforcement is included in UHPC and allows the concrete to sustain a tensile load after the cracking of the cementitious matrix.
UHPC has been defined as follows:
UHPC is a cementitious composite material composed of an optimized gradation of granular constituents, a water-to-cementitious materials ratio less than 0.25, and a high percentage of discontinuous internal fiber reinforcement. The mechanical properties of UHPC include compressive strength greater than 21.7 ksi (150 MPa) and sustained post-cracking tensile strength greater than 0.72 ksi (5 MPa). UHPC has a discontinuous pore structure that reduces liquid ingress, significantly enhancing durability compared to conventional concrete. (p. 1)(22)
Typical field-cast UHPC material properties are presented in table 2, which represent average values for a number of test parameters relevant to the use of UHPC as obtained from independent testing of a commercially available product.(23) This research, which was published in 2006 by the Federal Highway Administration (FHWA), investigated a number of material properties of a UHPC.(23) It also analyzed both mechanical- and durability-based behaviors of UHPC to assess its potential for use in future highway and bridge construction projects.
1 lb/ft3 = 16 kg/m3.
1 ksi = 6.89 MPa.
1 inch = 25.4 mm.
1 inch/inch/°F = 1.8 mm/mm/°C.
1 lb/yd3 = 0.593 kg/m3.
1 oz = 28.35 g.
The UHPC used for this research was a proprietary product produced by a major materials supplier. The UHPC formulation contained a premix powder, water, a modified phosphonate plasticizer, a modified polycarboxylate high-range water-reducing admixture, a non-chloride accelerator, and non-deformed, cylindrical, high-tensile strength steel fibers. The steel fibers had a diameter of 0.008 inch (0.2 mm), a length of 0.5 inch (12.7 mm), tensile strength greater than 290 ksi (2,000 MPa), and a thin brass coating that provided lubrication during the drawing process and provided corrosion resistance. The proportions used in this mix are shown in table 3. The UHPC used for this study had an average compressive strength of 26 ksi (179 MPa) at the time of testing.
1 lb/yd3 = 0.593 kg/m3.
The primary objective of this research was to evaluate the connection performance under thermal and cyclically structural loading. To ensure that the connection material was not biased toward poor performance from early age degradation, such as shrinkage or debonding prior to structural loading, the following standard procedures were adopted:
After the connection was cured, the top surface of the connection was ground to achieve a smooth surface. This served to assist in examining the shear key and in identifying interface cracking. An example of the ground connection surface is presented in figure 18.
Source: FHWA.
Figure 18. Photo. Comparison of the connection surface and interface between the box beams before and after grinding.
Thermal loading was simulated by pumping steam through copper tubes cast in the top flange of the box beams to create a temperature gradient through the depth of the beams. The copper tube arrangement is illustrated in figure 19 and can be seen in the box beam before the concrete was cast in figure 20.
Source: FHWA.
1 inch = 25.4 mm.
PS = prestressing.
O.C. = on the center.
Figure 19. Illustration. Copper tube arrangement within the top flange of the beams.
The beams were kept inside the laboratory with a room temperature of 70 °F (21 °C). The rate of thermal heating was approximately 18 °F/h (10 °C/h). The steam was cut off when the temperature gradient between the flanges was approximately 50 °F (28 °C). A total of 10 cycles were performed on each connection configuration. Temperatures were recorded using thermocouples embedded within the beams.
Two box beams were connected, and four-point bending loads were applied. Three boundary configurations were used: one configuration with both beams simply supported, a second condition with limited end transverse rotation of both beams, and a third with the same limited end rotation plus an additional restraint on one beam’s mid-span deflections. The latter two conditions provided more stiffness to the system.
Transverse PT was used with the conventionally grouted connections. The transverse PT force varied from 0 to 100 kip (0 to 445 kN) at each transverse PT location. These applied PT levels ranged from 8 to 0 kip/ft (0 to 117 kN/m) (i.e., high to low). The transverse PT force was monitored using the load cells (see figure 21). The PT was applied through the internal diaphragms of the girders. No transverse PT was applied for the UHPC connections.
Transverse PT is widely used in precast prestressed box beam bridge systems. Annamalai and Brown conducted experimental studies to investigate the effect of transverse PT on the behavior of small assemblies.(38) They concluded that transverse PT exhibited a high degree of monolithic behavior and significantly improved the shear strength of grouted shear key connections. However, according to Russell, more than 80 percent of designers surveyed as part of a study did not complete any design calculations to determine the level of transverse PT force needed.(1) When box girder bridges were examined in the field by Huckelbridge et al., they concluded that transverse PT was ineffective at resisting Δδ after the connection partially fractured, despite observing a satisfactory load distribution among beams.(4)
The loading setup on the two connected box beams is illustrated in figure 22. The box beams were simply supported at each end, providing a span length of 48 ft (14.6 m). Each individual beam was loaded with a spreader beam attached to an actuator. The loading points were 3 ft (0.9 m) on either side of the mid-span. The load was intentionally placed 6 inches (15 cm) off the centerline of the box beam to create a more severe tension force at the connection interface. A photograph showing the laboratory setup is presented in figure 23.
Source: FHWA.
1 inch = 25.4 mm.
B. End view.
Figure 22. Illustrations. Loading setup for the simply supported configuration.
The simply supported two-beam system is more flexible than a multi-beam bridge. In a multi-beam bridge, adjacent members are forced to deflect simultaneously when load is applied to a single beam as vertical shear force is transferred through the connections. When the deflection of an adjacent member is restrained by other beams in the bridge, a higher shear force at the connection can be expected compared to the case with two members.
With this consideration, two strategies were used to provide additional stiffness in the two-beam system tested in this study. The first required clamping the beam ends to restrain end transverse rotation. The second employed end clamping in tandem with providing in-span support under one beam at the in-span diaphragms. These are referred to as “partially stiffened” and “fully stiffened” boundary conditions, respectively, while the simply supported condition is referred to as “unstiffened.” The clamp-down force at each end was applied with two double C channels with a total clamp-down force of 100 kip (445 kN). For the support at the in-span diaphragms, the inside edge of the beam sat on a 6- by 24- by 2-inch (15- by 61- by 5-cm) neoprene pad, while the outside edge of the beam was tied down with a 35-kip (156-MPa) force to reduce the torsional rotation of the cross section when the beam was loaded. The setup is illustrated in figure 24 and pictured in figure 25 and figure 26.
Source: FHWA.
1 inch = 24.5 mm.
B. Section A-A.
Source: FHWA.
1 inch = 24.5 mm.
C. Section B-B.
Figure 24. Illustrations. Boundary conditions intended to increase structural stiffness.
The basic cyclic loading protocol included load application to the two beams through a 2-Hz sinusoidal wave ranging from the minimum to the maximum load. A 180-degree phase angle between the actuators on each beam was included to generate load transfer through the connection. Load data from load cells on the actuators in one of the tests is shown in figure 27. A minimum load of 5 kip (34.5 kN) was employed to avoid actuator lift off and potential actuator movement relative to the beams.
Source: FHWA.
1 kip = 4.448 kN.
Figure 27. Graph. Example of force read by the actuator load cells in the simply supported and partially stiffened beam setups.
When the partially stiffened configuration was used, the beams were loaded following the same loading protocol as the unstiffened configuration. For the fully stiffened configuration, the cyclic load was applied to the unrestrained beam, while the load on the restrained beam was held constant at 5 kip (22 kN). Some load data from load cells on the actuators in one of the tests are shown in figure 28. Minor fluctuations in beam B can be seen, which were caused by the loading on beam A.
Source: FHWA.
1 kip = 4.448 kN.
Figure 28. Graph. Example of force read by the actuator load cells for beams with the fully stiffened boundary condition.
An analysis of a representative adjacent box beam bridge indicated that an 18-kip (80-kN) loading amplitude was approximately the force effect on a single beam induced by a fatigue truck in the AASHTO LRFD Bridge Design Specifications.(6) Based on this information, loading ranges of 18, 36, 54, 72, and 90 kip (80, 160, 240, 320, and 400 kN) were applied in the study. The loading summaries for the cyclic loading scenarios are presented in table 4 through table 7. They include the boundary conditions, loading ranges, PT levels, and number of cycles performed for each of the four test specimens.
1 kip = 4.448 kN.
1 kip/ft = 14.59 kN/m.
*Number of cycles listed is for each individual loading range, not the total for all loading ranges listed.
1 kip = 4.448 kN.
1 kip/ft = 14.59 kN/m.
*Number of cycles listed is for each individual loading range, not the total for all loading ranges listed.
1 kip = 4.448 kN.
1 kip/ft = 14.59 kN/m.
*Number of cycles listed is for each individual loading range, not the total for all loading ranges listed.
1 kip = 4.448 kN.
1 kip/ft = 14.59 kN/m.
*Number of cycles listed is for each individual loading range, not the total for all loading ranges listed.
The beams were loaded with a computer-controlled servo-hydraulic loading system, and the structural response of each specimen was captured through the use of electronic instrumentation. The test utilized thermocouples, load cells, linear variable differential transformers (LVDTs), and strain gauges to record critical data on the structural performance. A high-speed data acquisition system was used for the capture of data from the instruments during both the thermal and cyclic loading.
Each beam was equipped with seven embedded thermocouples. There was one placed in the bottom flange at the mid-span, while the other six were distributed along the top flange. Of these six, three were placed around the mid-span (i.e., one at the mid-span and two 6 ft (1.83 m) longitudinally on each side of the mid-span). Two thermocouples were positioned to monitor incoming and outgoing steam temperatures at one beam end. The final thermocouple was at the opposite beam end.
The load was monitored by six load cells placed under the end supports of each beam, as shown in figure 29. The west end of the beams utilized a single load cell under each beam, while the east end had two load cells under each beam. Load cells inside each actuator captured the force imparted to the structure through the actuators. A load cell was also used to monitor the PT force in each PT bar.
The vertical displacements of the beams were recorded by four LVDTs placed under the beam at the mid-span 2 inches (51 mm) away from each edge of each beam. Figure 30 shows the installation of these vertical transformers. Six transverse LVDTs measured separation of the connection between the beams. Three transverse LVDTs spanned the connection at the top of the beam, and three spanned the connection at the bottom of the beam. They were placed at the mid-span and 6 ft (1.83 m) longitudinally from the mid-span in each direction. The transverse bottom LVDT at the mid-span is visible in figure 30. Figure 31 shows a typical transverse LVDT spanning the top of the connection.
Source: FHWA.
Figure 30. Photo. Five LVDTs (four vertical and one transverse) at the mid-span used to measure vertical deflection of the beams and the transverse connection opening.
Source: FHWA.
Figure 31. Photo. Transverse LVDT and strain gauge on the top surface spanning the connection.
Electrical resistance strain gauges were used to capture longitudinal and transverse strain in the concrete. Each beam had six embedded strain gauges placed on sister rebar. The sister rebar was then coupled with the reinforcing steel of the beams. Figure 32 shows one of the longitudinal strain gauges at the mid-span in the bottom flange of one of the beams coupled with the top layer of prestressing strands. Transverse embedded strain gauges were placed at four locations: in the top and bottom flanges at the mid-span and in the top flange 6 ft (1.83 m) longitudinally on each side of the mid-span. Longitudinal strain gauges were placed in both the top and bottom flanges and at the mid-span. All embedded strain gauges were placed along the centerline of the beams. Some of the embedded strain gauges were not used in some of the tests. Table 8 shows what gauges were included in which tests.
Source: FHWA.
Figure 32. Photo. Longitudinal strain gauge and thermocouple located at the mid-span in the bottom flange of one of the beams.
1 ft = 0.305 m.
*The southern mid-span gauge did not capture any data.
Surface strain gauges were installed after the beams were connected. Seven transverse surface strain gauges were installed along the length of the connection to measure PT confinement strain and the development of tensile strain during the tests. Gauges were located at the mid-span 6 ft (1.83 m) from the mid-span in each direction and at PT locations. Additional transverse gauges were located on the top of the girder at the mid-span and 6 ft on each side of the mid-span. Three gauges were placed on each beam at the centerline and 6 inches (152 mm) from each edge. The location 6 ft (1.83 m) east of the mid-span had all six gauges. At the mid-span, the southern-most gauge was missing. The location 6 ft (1.83 m) west only had the two gauges nearest the connection. Four longitudinal surface strain gauges were placed on the bottom flange at the mid-span 6 inches (152 mm) from each edge. It should be noted that the surface strain gauges were not used on every specimen. Table 8 shows what gauges were included in which tests. The arrangement of all of the instrumentation is shown in figure 33.