Performance Concrete (UHPC) Longitudinal Joints. A dissertation presented to. the faculty of. In partial fulfillment

Transcription

1 Behavior of Adjacent Prestressed Concrete Box Beam Bridges Containing Ultra High Performance Concrete (UHPC) Longitudinal Joints A dissertation presented to the faculty of the Russ College of Engineering and Technology of Ohio University In partial fulfillment of the requirements for the degree Doctor of Philosophy Ali A. Semendary May Ali A. Semendary. All Rights Reserved.

2 2 This dissertation titled Behavior of Adjacent Prestressed Concrete Box Beam Bridges Containing Ultra High Performance Concrete (UHPC) Longitudinal Joints by ALI A. SEMENDARY has been approved for the Department of Civil Engineering and the Russ College of Engineering and Technology by Eric P. Steinberg Professor of Civil Engineering Dennis Irwin Dean, Russ College of Engineering and Technology

3 3 ABSTRACT SEMENDARY, ALI A., Ph.D., May 2018, Civil Engineering Behavior of Adjacent Prestressed Concrete Box Beam Bridges Containing Ultra High Performance Concrete (UHPC) Longitudinal Joints Director of Dissertation: Eric P. Steinberg Adjacent precast prestressed concrete box beam bridges have been used for short and medium spans for decades. This type of bridge is preferred due to its high torsional rigidity, vertical clearances, and overall aesthetics. However, one of the major issues with this type of bridges is cracking that can occur between adjacent beams, at the longitudinal connection, which could lead to reflective cracks in composite deck or overlay surfaces. These cracks can cause water to leak through the joints, which can accelerate the corrosion of steel reinforcement from salt water exposure, causing the load transfer, between adjacent beams, to be reduced or diminished significantly. Despite several studies, the root cause of the cracking has yet to be addressed, therefore, a grout material with superior mechanical properties, interface bond strength, and long term durability is needed for this type of bridge in order to reduce or eliminate longitudinal cracks. In this doctorate study, ultra-high performance concrete (UHPC), which is becoming a common grout material in bridge connections, was used in the longitudinal joints of prestressed concrete box beam bridge. The new longitudinal joint design consisted of UHPC as the grout, with threaded dowel bars spaced evenly along the shear key joint length. This new shear key connection led to the elimination of transverse post-

4 4 tensioning, composite deck, and transverse tie rods. The performance of this type of connection was investigated in laboratory testing conducted by the Federal Highway Administration s (FHWA) Turner-Fairbank Highway Research Center for a pair of box beams under cyclic and temperature loads. However, the field performance can vary greatly from grout produced under ideal lab conditions. Further investigations of this type of connection are needed because most of the studies indicated that the longitudinal cracks were initiated by temperature effects, and propagate due to applied load. To test this, a bridge utilizing FHWA Turner-Fairbank Highway UHPC connections was constructed, instrumented, and tested in the field to investigate the behavior of the bridge due to truck and environmental loads. Results of this research showed that the fabrication of the new UHPC connections was not difficult and that proper placement of the beams ensured no conflict between dowel bars. No longitudinal cracks in shear key or at the interface were observed at early age based on field observation and data analysis. The bridge behaved monolithically, which indicates the ability of the reinforced UHPC shear keys to transfer and resist the applied load. Distribution factors determined by AASHTO LRFD Design Specifications were conservative, although the beams were assumed as being sufficiently connected to behave as a unit. The dynamic amplification factor matches that recommended by AASHTO LRFD Design Specifications. Monitoring of the bridge during various environmental changes did not show any detrimental effects. The cold temperature was found to have more effect on the behavior of UHPC shear key connections than warmer temperatures. Results from finite element model compared well with the field

5 5 measurements, which indicates the ability of the model to capture the behavior of the bridge. The parametric studies that were conducted on bridges with different widths, skews, depths and lengths showed the ability of the new shear key design to transfer and resist the applied load.

6 6 DEDICATION To my whole family, I dedicate this work

7 7 ACKNOWLEDGMENTS First of all, I would like to thank my supervisor, Professor Eric Steinberg for his suggestions and recommendations that played a vital role in shaping me as a researcher I am today. I would also like to thank my committee members, Dr. Shad Sargand, Dr. Kenneth Walsh, Dr. Issam Khoury, Dr. Michael Pollino, and Dr. Annie Shen for providing me with invaluable advice, comments, and recommendations. I would like to thank Dr. Teruhisa Masada for his suggestions and recommendations. I would especially like to thank Mr. Abdul Ridah Saleh for his assistance throughout my research. Many thanks to the staff of the Civil Engineering Department, Ohio University, especially to Mr. Joshua Jordan for his assistance with the field portion of this research. Special thanks to my family: father, mother, brothers and sisters for all of the sacrifices they have made and the patience that they have while I am away from them. In the end, I would like express appreciation to my wife, Rusul, who was always there to support and encourage during all the stages of my research. Finally, I would like to thank my sponsor, the Higher Committee for Education Development (HCED) in Iraq, for their financial support during my PhD journey.

8 8 TABLE OF CONTENTS Page Abstract... 3 Dedication... 6 Acknowledgments... 7 List of Tables List of Figures Chapter 1: Introduction Background Ultra High Performance Concrete Objectives Research Significance Dissertation Outline Chapter 2: Literature Review Adjacent Box Beam Bridges Shear Key Geometry and Grout Materials Transverse Post-Tensioning and Diaphragms Thermal Behavior Ultra High Performance Concrete (UHPC)... 55

9 9 2.3 Dowel Bars Summary Chapter 3: Methodology Bridge Description Beam Fabrication and Instrumentation Truck Test Static Truck Test Dynamic Truck Test Environmental Test Chapter 4: Behavior of an Adjacent Prestressed Concrete Box Beam Bridge Under Static and Dynamic Truck Loads Static Truck Load Test Results Deflection Longitudinal Strain Monitored by Using Exterior Strain Gauges Transverse and Longitudinal Strains Monitored by Using Interior Strain Gauges Live Load Moment Distribution Factors Dynamic Truck Load Test Results Deflection and Strain of the Bridge

10 4.2.2 Dynamic Load Allowance Chapter 5: Behavior of an Adjacent Precast Prestressed Concrete Box Beam Bridge Under Temperature Load Behavior after UHPC Placed in Joints from July 17-25, Beam Longitudinal Strain Results Longitudinal Shear Key Strain Results Beam Transverse Strain Results Shear Keys Transverse Strains Dowel Bar Strains Discussion August 8-16, Beam Longitudinal Strain Results Longitudinal Shear Key Strain Results Beam Transverse Strain Results Shear Key Transverse Strains Dowel Bar Strains January 7 - January 10, Beam Longitudinal Strain Results Longitudinal Shear Key Strain Results

11 Beam Transverse Strain Results Shear Key Transverse Strains Dowel Bar Strains Deflections Joint Movement Longitudinal Beam Movement Discussion Chapter 6: Finite Element Analysis and Parametric Studies Model Description Geometry Material Properties, Mesh Size, and Element Type Interaction, Boundary Conditions and Loading FEM Calibration Boundary Conditions Beam-Shear Key Interaction FEM Validation Mid-Span Deflection Mid Span Exterior Bottom Strain Transverse and Longitudinal Interior Strain

12 Strain in Shear Keys 1-3 in the Longitudinal and Transverse Directions Strain in the Dowel Bars for the Portion Embedded in the Top Flange of Beams 1-3 and the Portion Embedded in Shear Keys Discussion Longitudinal Strains in the Beams Behavior of the Shear Keys Behavior of the Dowel Bras Behavior of the Bridge under AASHTO LRFD Standard Design Truck Load Parametric Studies Bridge Width Effect on the Behavior of the UHPC Dowel Shear Key Connections Behavior of the UHPC Dowel Shear Key Connections with Different Skew Angles Bridge Depth and Length Effect on the Behavior of the UHPC Dowel Shear Key Connections Chapter 7: Summary, Conclusions and Recommendations Summary Conclusions Recommendations

13 Study Limitations and Future Work References Appendix A: Live Load Distribution Factors Determination for Case F Appendix B: Live Load Distribution Factors Determination for Case G Appendix C: Dynamic Truck Test Results

14 14 LIST OF TABLES Page Table 2.1 UHPC mix design (adopted from Yuan and Graybeal, 2014a) Table 2.2 Typical filed-cast material properties of UHPC (adopted fromyuan and Graybeal 2014a) Table 3.1 Locations of strain gauges embedded in beams Table 3.2 Locations of thermocouples embedded in beam Table 3.3 Locations of strain gauges embedded in shear keys Table 3.4 Location of strain gauges and LVDTs Table 3.5 Axle dimensions and axle and tire loads (adopted from Semendary et al.2017a) Table 3.6 Date and time of data collection Table 4.1 Interior top and bottom flange longitudinal strains in beam 1 at mid and quarter span Table 4.2 Top flange transverse strains in beams Table 4.3 Top and bottom flange longitudinal strains in beams 1-3 at mid span Table 4.4 Top and bottom flange longitudinal strains in beams 1-3 at quarter span Table 4.5 Longitudinal strains in shear keys 1 and 3 at mid and quarter span Table 4.6 Transverse strains shear keys 1-3 at mid and quarter span

15 15 Table 4.7 Axial strains in dowel bars embedded in shear keys 1-3 at mid and quarter span Table 4.8 Axial strains in dowel bars embedded in beams 1-3 at mid and quarter span 112 Table 4.9 Live load moment distribution factors Table 4.10 Common deck superstructures covered in articles and (adopted from AASHTO LRFD 2016) Table 4.11 AASHTO LRFD moment-distribution factors of live loads per lane Table 4.12 Dynamic amplification factor (DAF) from strain data (adopted from Semenderay et al. 2017c) Table 4.13 Dynamic amplification factor (DAF) from deflection data (adopted from Semenderay et al. 2017c) Table 5.1 Longitudinal strain in top and bottom flange of beams 1-3 at mid and quarter span due to temperature and static truck loads Table 5.2 Longitudinal strain in shear keys 1 and 3 at mid and quarter span due to temperature and static truck loads Table 5.3 Transverse strain in the top flange of beams 1-3 at mid span due to temperature Table 5.4 Transverse strain in shear keys 1-3 at mid and quarter span due to temperature and static truck loads

16 16 Table 5.5 Strain in dowel bar embedded in right side of the cross section of beams 1-3 at mid and quarter span due to temperature and static truck loads Table 5.6 Strain in the dowel bar embedded shear keys 1-3 at mid and quarter span due to temperature and static truck loads Table 6.1 Materials properties (adopted from Semendary et al. 2017a) Table 6.2 Maximum principal strain in the shear keys and dowel bars for different types of interactions Table 6.3 Top flange transverse strains in beams Table 6.4 Top and bottom flange longitudinal strains in beams 1-3 at mid span Table 6.5 Top and bottom flange longitudinal strains in beams 1-3 at quarter span Table 6.6 Longitudinal strain in the shear keys 1 and 3 at mid and quarter span (adopted from Steinberg et al.2016) Table 6.7 Transverse strain in shear keys 1-3 at mid and quarter span (adopted from Steinberg et al.2016) Table 6.8 Axial strain in the dowel bars embedded in beams 1-3 at mid and quarter span Table 6.9 Axial strain in the dowel bars embedded in shear keys 1-3 at mid and quarter span (adopted from Steinberg et al.2016) Table 6.10 Results for 33% dynamic load allowance with different load configurations

17 17 Table 6.11 Results for 75% dynamic load allowance with different load configurations Table 6.12 Results for 33% dynamic load allowance with different bridge width under load configuration A Table 6.13 Results for 75% dynamic load allowance with different bridge width under load configuration A Table 6.14 Results for 33% dynamic load allowance with skews under load configuration A Table 6.15 Results for 75% dynamic load allowance with skews under load configuration A Table 6.16 Results for 33% dynamic load allowance with skews under load configuration A Table 6.17 Results for 75% dynamic load allowance with skews under load configuration A

18 18 LIST OF FIGURES Page Figure 2.1 Typical section of an adjacent member bridge: (a) composite superstructure, (b) non- composite superstructure, (c) shear key details (adopted from PCI 2012, Steinberg and Semendary 2016) Figure 2.2 Box beam used in about two-thirds of the States (Russell 2011, with permission from the author) Figure 2.3 Longitudinal cracks and leaking in box beam bridges: (a) an asphalt riding surface and (b) in a composite concrete deck (Russell 2011; with permission from the author) Figure 2.4 Frequency of longitudinal deck cracking: (a) by year built, (b) by age at time of inspection (adopted from Lall et al. 1998) Figure 2.5 Comparison of precast concrete sections: (a) original precast section; (b) modified precast section at diaphragm (adopted from El-R y et al. 1996) Figure 2.6 Presteressing force for mid span diaphragms (adopted from El-R y et al. 1996) Figure 2.7 Effective of bridge width on the required post-tensioning for 0.0 deg. skew and span-to-depth ratio of 30 at mid span diaphragm (adopted from Hanna et al. 2009) 48 Figure 2.8 Comparison between precast prestressed concrete bridge design manual charts with updated chart (adopted from Hanna et al. 2009)... 49

19 19 Figure 2.9 Girder movement and opening of shear key due to temperature loading (adopted from Miller et al. 1999) Figure 2.10 Compression of compressive and tensile behavior between UHPC and HSC (adopted from Li et al. 2015) Figure 2.11 Shear key details between girders in Hodder Avenue Underpass bridge (adopted from Li et al. 2015) Figure 2.12 Pair of adjacent box beam connected by UHPC and dowel bars: (a) partial depth, and (b) full depth (Yuan and Graybeal 2014b; with permission from the author) 62 Figure 2.13 Shear transfer by friction: a) external compressive force across the joint, b) relative joint displacement, c) internal compressive force generated by transverse bars across the joint (adopted from fib 2007) Figure 2.14 Shear key geometry (adopted from fib 2007) Figure 3.1 Bridge description: (a) erection diagram, (b) bridge section (adopted from Semendery et al. 2017a,b) Figure 3.2 Shear key detail (all dimensions in inches, adopted from Steinberg et al. 2015) Figure 3.3 Dowel bar details: (a) dowel bars splice, and (b) straight dowel bar splicers. 72 Figure 3.4 Connection details: (a) cross section, and (b) top view Figure 3.5 Shear key form and beam dowel parts in place... 73

20 20 Figure 3.6 Instrumentation in the top and bottom flanges in beams 1-3: (a) quarter span, (b) mid span Figure 3.7 Vibrating wire strain gauge (adopted from Semendary et al. 2017b, Steinberg et al. 2015) Figure 3.8 Instrumentation in the top flange at mid and quarter span (adopted from Semendary et al. 2017b) Figure 3.9 Instrumented dowel bar and dowel bar covered with a protective shield (adopted from Steinberg et al. 2015,2016) Figure 3.10 Thermocouples through the depth of Beam 3 (adopted from Steinberg et al. 2015) Figure 3.11 Shear key preparation Figure 3.12 Instrumenting and installed a shear key dowel bar and shear key with dowels (adopted from Semendary et al. 2017b, Steinberg et al. 2016) Figure 3.13 Transverse and longitudinal shear key gauge (Semendary et al. 2017b with permission from ASCE) Figure 3.14 UHPC placement into longitudinal shear keys and shear keys after plywood forms removed Figure 3.15 Waterproofing installation Figure 3.16 Instrumentation frame and bridge instrumentation Figure 3.17 Bridge instrumentation... 86

21 21 Figure 3.18 Truck loading (adopted from Semendary et al.2017a) Figure 3.19 Data acquisitions system Figure 4.1 Mid span deflection for load configuration 1 (1 truck on left) Figure 4.2 Mid span deflection for load configuration 2 (1 truck on right) Figure 4.3 Mid span deflection for load configuration 3 (1 truck on left and 1 truck on right) Figure 4.4 Mid span deflection for load configuration 4 (2 trucks on left back to back). 97 Figure 4.5 Mid span deflection for all load configurations Figure 4.6 Mid span bottom strain for load configuration 1 (1 truck on left) Figure 4.7 Mid span bottom strain for load configuration 2 (1 truck on right) Figure 4.8 Mid span bottom strain for load configuration 3 (2 trucks at midspan) Figure 4.9 Mid span bottom strain for load configuration 4 (2 trucks on left back to back) Figure 4.10 Mid span bottom strain for all load configurations Figure 4.11 Live load moment distribution factors versus width of the bridge (adopted from Semendary et al.2017a) Figure 4.12 Live load moment distribution factors versus width of the bridge (1 design lane loaded) (adopted from Semendary et al.2017a) Figure 4.13 Live load moment distribution factors versus width of the bridge (2 design lanes loaded) (adopted from Semendary et al.2017a)

22 22 Figure 4.14 Mid span bottom strain versus time for truck speed at 5 mph (adopted from Semenderay et al. 2017c) Figure 4.15 Mid span bottom deflection versus time for truck speed at 5 mph (adopted from Semenderay et al. 2017c) Figure 4.16 Peak strain in bottom of each beam for truck speed at 5 mph Figure 4.17 Peak deflection in bottom of each beam for truck speed at 5 mph Figure 4.18 Peak mid span strain in the bottom of each beam for different truck speeds (adopted from Semenderay et al c) Figure 4.19 Peak mid span deflection in the bottom of each beam for different truck speeds (adopted from Semenderay et al. 2017c) Figure 4.20 Mid span strain in the bottom of each beam due to static load and dynamic load for different truck speeds (adopted from Semenderay et al. 2017c) Figure 4.21 Mid span deflection in the bottom of each beam due to static load and dynamic load for different truck speeds (adopted from Semenderay et al. 2017c) Figure 4.22 Dynamic amplification factor (DAF) from strain data of beam 7 versus truck speeds Figure 4.23 Dynamic amplification factor (DAF) from deflection data of beam 5 versus truck speeds Figure 4.24 Dynamic amplification factor (DAF) for strain data of beams 1-7 versus bridge width

23 23 Figure 4.25 Dynamic amplification factor (DAF) for deflection data of beams 1-7 versus bridge width Figure 5.1 Beam 1 longitudinal strains/temperatures versus time at mid-span Figure 5.2 Beam 3 longitudinal strains/temperatures versus time at mid span Figure 5.3 Shear key 3 longitudinal strains/temperatures versus time at mid span (adopted from Semendery et al.2017b) Figure 5.4 Shear key 3 longitudinal strains/temperatures versus time at quarter span (adopted from Semendery et al.2017b) Figure 5.5 Beam 3 transverse top strains/temperatures versus time at mid span (adopted from Semendery et al.2017b) Figure 5.6 Shear key 1 transverse strains/temperatures versus time at mid span (first 24 hours) Figure 5.7 Shear key 1 transverse strains/temperatures versus time at quarter span (first 24 hours) Figure 5.8 Shear key 1 transverse strains/temperatures versus time at mid span (after 24 hours) (adopted from Semendery et al.2017b) Figure 5.9 Shear keys 1-3 transverse strains versus time at mid span Figure 5.10 Dowel bar in beam 1 strains/temperatures versus time at mid span (adopted from Semendery et al.2017b)

24 24 Figure 5.11 Dowel bar in shear key 1 strains/temperatures versus time at mid span (adopted from Semendery et al.2017b) Figure 5.12 Beam 1 and shear key 1 longitudinal strains/temperatures versus time: (a) mid span (adopted from Semendery et al.2017b), (b) quarter span Figure 5.13 Dowel bar in Beam 1 and shear key 1 strains/temperatures versus time at mid span (adopted from Semendery et al.2017b) Figure 5.14 Beam 1 longitudinal strains/temperatures versus time at mid-span Figure 5.15 Beam 1 bottom surface longitudinal strains/temperatures versus time at midspan Figure 5.16 Exterior bottom longitudinal strains versus time at mid span Figure 5.17 Shear Key 1 longitudinal strains/temperatures versus time at mid span Figure 5.18 Beam 3 transverse top strains/temperatures versus time at mid span Figure 5.19 Shear Key 2 transverse strains/temperatures versus time at quarter span Figure 5.20 Dowel bar in Beam 1 strains/temperatures versus time at mid span Figure 5.21 Strains/temperatures versus time for dowel bar embedded in shear key 3 at mid-span Figure 5.22 Strains/temperatures in embedded dowel bar and shear key 1 at mid span 157 Figure 5.23 Strains/temperatures in embedded dowel bar and shear key 1 at quarter span Figure 5.24 Beam 1 longitudinal strains/temperatures versus time at mid-span

25 25 Figure 5.25 Beam 1 bottom surface longitudinal strains/temperatures versus time at midspan Figure 5.26 Exterior bottom longitudinal strains versus time at mid span Figure 5.27 Shear key 1 longitudinal strains/temperatures versus time at mid span Figure 5.28 Beam 3 transverse top strains/temperatures versus time at mid span Figure 5.29 Shear key 1 transverse strains/temperatures versus time at quarter span Figure 5.30 Dowel bar in beam 3 strains/temperatures versus time at mid span Figure 5.31 Strains/temperatures versus time for dowel bar embedded in shear key 3 at mid-span Figure 5.32 Deflections and bottom surface temperatures versus time for beam 1 at mid span Figure 5.33 Joint movement versus time across joint Figure 5.34 Beam 7 longitudinal movement versus time at the rear abutment Figure 5.35 Beam 7 longitudinal movement versus time at the forward abutment Figure 6.1 Extrusion of components: (a) beams; (b) diaphragms; (c) shear keys; (d) rebar; (e) strands and (f) dowel bars Figure 6.2 View of the assembled finite element model Figure 6.3 Finite element model mesh Figure 6.4 Overclosure pressure relationship: (a) hard contact; (b) linear contact (adopted from Abaqus user s manual 2006)

26 26 Figure 6.5 Slip regions with limit critical shear stresses for the friction model (adopted from Abaqus user s manual 2006) Figure 6.6 Boundary conditions Figure 6.7 Mid span deflection and strain of the bridge for load configuration 2 (1 truck on right) based on different boundary conditions Figure 6.8 Deflection and strain at mid span of the bridge for load configuration 2 (1 truck on right) (adopted from Semendary et al. 2017a) Figure 6.9 Behavior of the dowel bars under different types of constraints: (a) surface to surface; (b) tie Figure 6.10 Deflection at mid span compared with FEM of the bridge for load configuration Figure 6.11 Deflection at mid span compared with FEM of the bridge for load configuration Figure 6.12 Deflection at mid span compared with FEM of the bridge for load configuration Figure 6.13 Deflection at mid span compared with FEM of the bridge for load configuration Figure 6.14 Mid span strain compared with FEM of the bridge for load configuration Figure 6.15 Mid span strain compared with FEM of the bridge for load configuration

27 27 Figure 6.16 Mid span strain compared with FEM of the bridge for load configuration Figure 6.17 Mid span strain compared with FEM of the bridge for load configuration Figure 6.18 Transverse strain from FEM for load configuration Figure 6.19 FEM top transverse strain across bridge width at mid-span for load configuration Figure 6.20 Transverse strain and maximum principal tensile strain in the shear key for load configuration 4 (adopted from Steinberg et al.2016) Figure 6.21 Maximum principal tensile stresses in the dowel bar for load configuration 4 (adopted from Steinberg et al.2016) Figure 6.22 Truck load combinations and position in transverse direction: (a) one lane loaded; (b, c) two lane loaded (adopted from Semendary et al. 2017a) Figure 6.23 Truck load position in longitudinal direction Figure 6.24 Behavior of the bridge at mid span under load configuration A1: (a) strain (b) deflection Figure 6.25 Behavior of the bridge at mid span under load configuration A2: (a) strain (b) deflection Figure 6.26 Behavior of the bridge at mid span under load configuration A3: (a) strain (b) deflection

28 28 Figure 6.27 Bridges with different widths Figure 6.28 Top view of the bridge with different skew angles Figure 6.29 Bridge cross section with different depths Figure 6.30 Bridge with different span lengths Figure 7.1 Shear key and dowel bars details for future application in other designs: (a) shear key details, (b) dowel bar details

29 29 CHAPTER 1: INTRODUCTION 1.1. Background Adjacent precast prestressed concrete box beam bridges have often been preferred for medium to short spans for ease and speed of constructions as well as cost and effectiveness. Due to these advantages, adjacent box beam bridges are used by 2/3 of the State Departments of Transportation (DOTs) (Russell 2011). Adjacent box beam bridges have been used in the United States since 1950 (PCI 2012). Typical span lengths of the adjacent box beam bridges ranges between 20 to 100 ft. Adjacent box beam bridges are constructed by placing the box beams side by side and grouting the shear keys between the adjacent beams. The partial or full depth shear key assists in load transfer and prevents water leakage between the beams. Limited information is available about the magnitude of the forces in the shear keys and the ability of shear keys to resist these forces (Russell 2009). Therefore, the size, location of the shear keys, and type of grout materials vary among the state DOTs. Transverse post-tensioning may be used to enhance the integrity of the bridge in the transverse direction to reduce the differential deflection. The transverse connections consisted of either threaded grouted or un-grouted tie rods or post tensioning bars /strands and placed at discrete locations along the bridge s span. However, limited design specifications are available about the number, spacing, and the force in the transverse tie rods/strands. The magnitude of the force in the tie rods/strands and stage that the force should be applied vary among the state DOTs. Longitudinal cracks in the shear keys and overlays have been identified as a recurring issue for adjacent box beam bridges during their service life (El-R y et al.

30 , Hlavacs et al. 1996; Miller et al. 1999; Dong 2002; Russell 2009; Ulku et al. 2010; Russell 2011, Grace et al. 2012; Attanayake and Aktan 2014; Yuan and Graybeal 2016). The poor performance of the shear keys is considered as the main cause of this problem (Fu et al. 2011). The longitudinal cracks have several adverse effects on the bridge, including the allowance of water and deicing chemicals to penetrate into the joints between the box beams, leading to accelerated corrosion of the steel reinforcement. These cracks can also reduce or cause a loss of load transfer between beams (Russell 2011). This could cause the loaded beam to carry the applied load entirely without any contribution from the other beams. According to the AASHTO LRFD (2016) the shear keys tend to crack due to wheel loads, warping, and environmental effects, leading to leaking of the keys and decreased the shear transfer. The relative movement between adjacent panels tends to cause cracking in the overlay if present. Different studies have been done to determine the causes of the longitudinal cracks in the shear keys between adjacent box beams. Some of these studies have shown that the cracks are initiated by thermal stresses and then propagate due to applied loads (Hlavacs et al. 1996; Miller et al. 1999; Grace et al. 2012; Attanayake and Aktan 2014). Different solutions have been suggested to reduce the cracking in the shear keys by using different shear key configurations and grout materials (Gulyas et al. 1995; Issa et al. 2003; Lall et al. 1998) and/or increasing the number of diaphragms and using different magnitudes of transverse post-tensioning with or without a concrete composite deck (El- R y et al. 1996; Hanna et al. 2009; Fu et al. 2011; Grace et al. 2012; Hansen et al. 2012). However, the cracks in the longitudinal joint persist after these studies. One way

31 31 to address the longitudinal cracking in adjacent box-beam bridges is to use an improved grouting material with enhanced tensile and bond strengths. 1.2 Ultra High Performance Concrete A grout material with high mechanical properties and bond strength along with superior durability should be used in the joints to eliminate the longitudinal cracks. Ultra-High Performance Concrete (UHPC) represents a new class of concrete, which has high strength and durability characteristics compared with normal or even high strength concrete. UHPC has been defined as a cementitious composite material composed of an optimized gradation of granular constituents, a water-to-cementitious ratio less than 0.25, and a high percentage of discontinuous internal fiber reinforcement. The mechanical properties of UHPC include compressive strength greater than 22 ksi and sustained postcracking tensile strength greater than 1 ksi (Yuan and Graybeal 2014a). UHPC has a discontinuous pore structure that reduces liquid ingress, significantly enhancing durability compared to conventional concrete (Yuan and Graybeal 2014a). The essential component in the mix of the UHPC is the steel fibers. The steel fibers used in the mix have a tensile strength specified to be greater than 290 ksi. These fibers have a diameter of in. and length of 0.5 in. To combat against corrosion, the steel fibers have a brass coating, which also provides lubrication during the drawing process (Yuan and Graybeal 2014a). The superior mechanical properties and durability of UHPC makes it an excellent candidate to be used as a grout material for bridge connections. Field-cast UHPC connections have been used between prefabricated bridge elements in thirty projects in

32 32 the United States (Graybeal, 2014). UHPC was also used to fill the connections in the Mackenzie River twin bridges. This work is considered the largest field-cast UHPC connection project in North America. The field cast UHPC was used to fill transverse joints between deck panels, shear pockets, and haunches between the underside of the deck panels and steel girders (Perry et al. 2014). Perry et al. (2007) documented the use of UHPC as a grout material in the skewed 24.4 m (80 ft) single span bridge in Rainy Lake, Ontario, Canada. UHPC also has been used in the longitudinal and transverse joints of the 3-span side-by-side Eagle River Bridge in Canada (Perry et al. 2010). For this bridge, the UHPC was successfully placed and the required strength was achieved after the curing period. However, no instrumentation was available on the bridge s performance. Perry and Royce (2010) studied the use of UHPC as grout material in New York State. The state program consisted of the development and testing of prototype panels connected by reinforced joints filled with UHPC material. Three-rebar configurations (straight, hairpins, and headed studs) with three types of rebar (epoxy coated, galvanized, and black) were employed. A significant bond was found between the rebar and UHPC which led to a reduction of the development length for the rebar and a reduced joint width. The result of tests on a pair of High Performance Concrete (HPC) precast deck panels showed no leakage or degradation after applying a million cycles of live loading. After the laboratory testing, the UHPC was used to fill the joints of a single span deck bulb-tee bridge in the Village of Lyons, located in Wayne County, NY. UHPC has also been suggested for use as grout material in the shear keys with dowel bars of adjacent box beam bridges because of its superior mechanical properties.

33 33 Yuan and Graybeal (2014b) studied two full-scale adjacent box beams to evaluate four connection details. The first configuration included a partial or full depth shear key grouted with non-shrink grout in combination with transverse post-tensioning. The second configuration included a partial or full depth shear key grouted with UHPC and lap spliced reinforcement bars without post-tensioning. Cyclic loading was applied and the results showed that the post-tensioned non-shrink grout and the UHPC shear keys did not crack and had the same behavior. However, the cyclic loading was able to propagate a pre-existing crack in the conventional grout connection independent of the level of post-tensioning. The UHPC connection exhibited superior performance even when direct tension was applied on the top to intentionally create cracks. Cracks were observed in the box beams, but no cracks developed in the shear keys or at the beam-shear key interface. 1.3 Objectives The main objectives of this doctoral dissertation are: 1. Monitor the field performance of the UHPC shear key configurations and dowel bars along with the behavior of the bridge. In order to investigate the field performance, a bridge utilizing new shear key connections was instrumented and monitored under static, moving, and temperature loads. 2. Collect and analyze the data. Then, compared the results with the allowable limits and the standard codes. 3. Create a three dimensional finite element model using commercial software. Calibrate and validate the model with the field measurement in order investigate the ability of the model to capture the behavior of the bridge.

34 34 4. Conduct a parametric studies using the validated model and investigate the ability of new UHPC dowel shear keys to transfer and resist the standard design load for adjacent box beam bridges with different widths, skews, depths and lengths. 1.4 Research Significance Longitudinal cracks in shear keys continue to be the primary cause of deterioration in adjacent box beam bridges. Eliminating or at least reducing the cracks can lead to higher durability, less maintenance costs, and a longer service life. A comprehensive study incorporating a combination of traffic and temperature loads can help engineers understand the behavior of box beam bridges and identify the main causes for the cracks. 1.5 Dissertation Outline The dissertation is organized into seven chapters as follows: Chapter 1 documents an introduction to the research along with objectives, research significance and outline to the dissertation. Chapter 2 presents a literature review of past and current practices used in the construction of adjacent box beam bridges. This chapter is divided into several subsections which include an introduction to the adjacent box beam bridges, shear key geometry and grout material, the transverse post tensioning practice, thermal behavior of adjacent box beam bridges, ultra high performance concrete (UHPC) and dowel bar behavior, and the last subsection is a summary of the literature review. Chapter 3 describes the beam fabrication, instrumentation, and processes of construction of the bridge. This includes the fabrication and the instrumentation of the

35 35 beams, shear keys, and dowel bars. Static and moving loads tests, as well as the environmental monitoring, are described in this chapter. Chapter 4 presents the static and moving loads test results. The short-term behavior of the bridge is described in this chapter. Figures and tables are used to document the field performance. Chapter 5 presents the long-term behavior of the bridge using the data which was collected at the early stages as well as in the summer and winter seasons. Chapter 6 presents a description of finite element (FE) model, calibration, validation and discussion. This chapter also presents the parametric studies that were conducted based on validated FEM model. Finally, Chapter 7 provides a summary and conclusions of the main funding of this dissertation. This chapter also includes recommendations for future work.

36 36 CHAPTER 2: LITERATURE REVIEW 2.1 Adjacent Box Beam Bridges Adjacent precast prestressed concrete box beam bridges have often been used as medium and short span bridges. According to the Precast/Prestressed Concrete Institute (PCI) (2012), adjacent box beam bridges have been used in the United States since 1950 and have typically spanned between 20 and 100 feet. Adjacent prestressed concrete box beam bridges are constructed by placing beams side-by-side and grouting a longitudinal shear key between the beams after they are set in position (Figure 2.1a and b). The longitudinal shear keys covered either a partial depth of the beam (partial depth shear key) of an approximately the full depth of the beam (full depth shear key) as shown in Figure 2.1 c. Transverse post-tensioning can be used to enhance the integrity of the bridge in the transverse direction. The bridges can be either composite by using cast in place concrete deck on the top or non-composite by using the top flange as a riding surface. The non-composite bridge is often covered with an asphalt overlay surface as shown in Figure 2.1b. This type of bridges is preferred due to its relative ease and speed of construction, high torsional rigidity, vertical clearances, and overall aesthetics. Box beam bridges are considered to be especially economical due to deck elimination. The hollow section of box beam reduces the overall weight of the beams and allows a protected route for utilities such as water and gas pipes (Hanna, 2008)

37 37 (a) (b) Partial depth shear key Full depth shear key Figure 2.1 Typical section of an adjacent member bridge: (a) composite superstructure, (b) non- composite superstructure, (c) shear key details (adopted from PCI 2012, Steinberg and Semendary 2016) According to Russell (2011), two-thirds of the United States use adjacent box beam bridges as shown in Figure 2.2.

38 38 Figure 2.2 Box beam used in about two-thirds of the States (Russell 2011, with permission from the author) However, one of the major issues with this type of bridges is cracking occurring between adjacent beams in the longitudinal connection. The cracks allow water to penetrate and eventually cause the corrosion of reinforcement as shown in Figure 2.3. From this, load transfer between beams may be lost (Russell 2011). (a) (b) Figure 2.3 Longitudinal cracks and leaking in box beam bridges: (a) an asphalt riding surface and (b) in a composite concrete deck (Russell 2011; with permission from the author)

39 39 Research has been done to study the main causes of the cracks in the shear keys and to understand the effect of the cracks on the overall behavior of the bridge. Two bridges were tested in Ohio to investigate the behavior of the adjacent precast prestressed concrete box beam bridges (Huckelbridge et al. 1995). The first bridge was a simply supported, non-composite box beam bridge. The second bridge was a four span, noncomposite box beam bridge with four-traffic lanes. A relative displacement of 0.02 in. was observed in the joint adjacent to the center girder in the first tested bridge, which indicated leaking in the joint. However, the joint still transferred shear load to the adjacent beam. A relative displacement with the same magnitude also noticed in the second bridge, especially for the joint adjacent to the wheel load position. However, the strain readings indicated the bridge still had a reasonable capacity despite the damage. The bridge was then repaired, new shear keys were cast, and the bridge was tested after repairs were made. A large relative displacement was also observed in the same joint, after repairs. The authors concluded that the intact shear key should have relative displacement less than in., otherwise the shear key will fracture along at least part of the length of the key. The observed relative displacement in the joint were between in. and 0.02 in., which indicated partial fracture. The shear transfer across the joints was either reduced or totally lost depending on the fracture. However, the load was still distributed among the beams even though the bond line between the shear key and the beam was broken. It was concluded that the load still transferred through other mechanical, but the tie rods had little to no effect on the shear key performance.

40 40 In another research, destructive testing was conducted to study the behavior of a 43 year, standard adjacent box beam bridge with different degrees of damage (Huffman 2012). The bridge consisted of three spans. Each span contained nine adjacent box beams connected together with partial depth shear keys, and 1 in. transverse tie rods at two positions along each span. Each tie rod connected two beams in the transverse direction because of the skew angle of the bridge. Two spans (center and west) of the bridge were instrumented to monitor the top and bottom strain and deflection of the bridge. The damage was created artificially by cutting a 2 in. deep notch in the bottom of the three interior beams in the west span. Load was applied to the bridge using three steel frames with a hydraulic cylinder at the center of each frame. The results showed that the load was distributed in transverse direction which emphasize the ability of shear keys and tie rods to distribute the load despite damage in the beams. In addition, the beam adjacent to the damaged beams of the bridge carried more load than the damaged beams. After the collapse of the bridge, no corrosion was observed in the tie rod after its 43 years in service and the shear keys only exhibited de-bonding failure. There was no instrumentation within the shear key or on the tie rods in order to monitor the actual strain during the test. Several bridges in Michigan were inspected to identify the changes in the design of side-by-side box beam bridges (Attanayake and Aktan 2014). The design changes included full depth non shrinkage grout shear keys, high transverse post-tensioning, and a cast in place deck with seven-day moist curing. However, the box beam bridges inspected still experienced longitudinal reflective cracks, irrespective of age. One such bridge under

41 41 construction showed that cracks were developing at the grout-beam interface within 3 days after grouting, and before the application of post-tensioning. Even after post tensioning, the shear key still experienced cracking. When the bridge was inspected three weeks later before the deck placement, cracks were observed along every beam/shear key interface. Reflective cracks were first observed within 15 days after casting the deck and before the deck was subjected to barrier and live loads. The observed cracks were over the abutments and ran vertically through the full depth of the deck over the shear keys. Furthermore, the study also documented shear key conditions during box beam bridge demolition. Cracking was observed as de-bond failures between the shear key and box beam. The authors concluded that current box beam designs be redesigned to consider the effect of live load and intrinsic loads. Instrumentation was not used by the authors to identify the main causes for the cracks. Recently, Steinberg and Semendary (2016) conducted a truck load test on 50- year-old, simple span adjacent box beam bridge with total length of 54 ft. The bridge consisted of eight box beam, each 36 in. wide by 27 in. deep. The bridge was instrumented to measure the strain and deflection in beams as well as the strain in the tie rods. The strains were recorded while the bridge was loaded with truck load. The strain in tie rods was also monitored during the demolition of the shear keys. The bridge behaved satisfactory despite of damage of the exterior beams. Furthermore, the tie rods exhibited a relatively low strain during the truck load test due to the functionality of the shear key in transferring the load. The force calculated from the strain that measured in tie rods during shear key demolition consisted with the current standard magnitude used in state of Ohio.

42 Shear Key Geometry and Grout Materials When cracking occurs in adjacent box beam bridges, it is typically developed in the longitudinal direction of the bridge. The shear key geometry and grout materials have been found to be important factors that affect this cracking behavior. Studies have been conducted to investigate the effect these attributes have on eliminating cracks in shear keys. In one study, non-shrink grout and magnesium ammonium phosphate (Mg-NH4- PO4) mortar were tested to evaluate their strength (Gulyas et al. 1995). Three types of tests were used which included vertical shear, direct tension, and longitudinal shear. The vertical shear test was used to simulate the wheel loads on only one member and not distributed amongst members. The direct tension test was used to simulate the transverse shortening of precast member due to creep and shrinkage, as well as the drying shrinkage of the keyway grout. The longitudinal shear test was used to simulate the shortening in the precast concrete member due to creep from the prestressing force and shrinkage in concrete while the grout material does not exhibit the same degree of shortening. Sixteen specimens were cast and tested. The results showed that for all cases Mg-NH4-PO4 mortar displayed higher strength than non-shrinkage grout. The drying shrinkage of the Mg-NH4- PO4 was also reported lower than that of non-shrinkage coarse aggregate grout with a 0.3 water cement ratio. Lower bond strength was recorded because the carbonation was used at the interface surface of Mg-NH4-PO4. The performance of different grouting materials, used in joints between concrete panels, were evaluated by using experimental and finite element methods (Issa et al. 2003). In the study, set grout, set 45 for normal temperature, set 45 for hot temperature,

43 43 and polymer concrete grouting materials were evaluated. Vertical shear and flexure tests were used to simulate the shear and bending due to the wheel loads crossing the transverse joints. Direct tension teste was used to simulate the drying shrinkage of the grout and the transverse shortening of the precast components. In order to evaluate the grouting materials, 36 full-scale specimens were fabricated and tested. The results showed that the polymer concrete exhibited the highest performance in all three tests. The ultimate load from the finite element model was kips with less crushing and cracks. Minor cracks were observed in the lower neck of joints, but propagated to the concrete side. The polymer concrete recorded lower shrinkage and permeability compared to the other types of the grouts. The authors recommend using female-to-female joints grouted with polymer concrete. In another study, the shear key geometry was investigated to study the improvement of the longitudinal connection in New York adjacent box beam bridges (Lall et al. 1998). Bridges built prior to 1990, consisted of a 12 in. partial depth shear key with transverse post-tensioning of 30 kips per location. The frequency of transverse posttensioning varied depending on bridge length. For spans up to 50 ft, no transverse posttensioning was used. Transverse post-tensioning occurred at the center of the bridge for spans between 50 and 75 ft. For spans more than 75 ft, transverse post-tensioning was used at the quarter points. Bridges from this era usually utilized a 6 in. reinforced concrete composite deck. However, longitudinal cracking was observed shortly after construction and propagate along the full length of the keyway with time. By 1992, a new bridge design was adopted which used full depth shear keys. The shear key depth

44 44 was increased to connect approximately the full depth of the beam. Furthermore, the number of transverse post-tensioning was increased to three for spans up to 50 ft and five for longer spans. A survey was conducted to investigate the effect design improvements had on the behavior of the shear key. Fifty-four percent of inspected bridges, built between 1985 and 1990, had cracks in their shear keys. However, for bridges built between 1992 and 1996, the cracks were observed in 23 percent of inspected bridges as shown in Figure 2.4. (a) (b) Figure 2.4 Frequency of longitudinal deck cracking: (a) by year built, (b) by age at time of inspection (adopted from Lall et al. 1998) Transverse Post-Tensioning and Diaphragms Studies have been conducted to investigate the effects of using transverse posttensioning on the behavior of an adjacent box beam bridges. When used, post-tensioning can help to reduce the differential deflection between beams and reduce the tensile stresses in the shear key by the application of compressive stresses. The Load and Resistance Factor Design (LRFD) Specification (AASHTO, 2016) section d states that the required transverse stresses after all losses should not be less than 250 psi.

45 45 However, the area that the transverse prestressing should be applied is not clear. Therefore, many studies have been done to determine the required transverse stresses that making the bridge behave monolithically in the transverse direction. The methodology for transverse post-tensioning of adjacent box beam bridges in the transverse direction was studied by comparing the design criteria of box beam bridges built in Japan with those built in the United States (El-R y et al. 1996). The box beam bridges in Japan were essentially the same as those in the U.S. except the shear keys were wider, deeper, and filled with concrete and there was a higher magnitude of post-tensioning. However, some states in the U.S. used smaller shear keys with little or no transverse post-tensioning, while others used higher transverse post-tensioning without theoretical justification. The diaphragms represented an important member to transfer the load between adjacent beams. Box beams without diaphragms exhibited large differential deflections between beams which led to cracking in the shear keys. The study assumed that the diaphragms helped to transfer truck loading in transverse direction. Five diaphragms were used for span lengths up to 100 ft. The diaphragms were distributed as two at the ends and three at the quarter points, in order to keep differential deflection less than 0.02 in. A few modifications were created in the diaphragm locations by providing a 1 in. pocket on each side of the beam and grouting the whole depth of the shear key at diaphragm location for distance 8 in. in longitudinal direction as shown in Figure 2.5. The grid method was used to obtain the member forces. The beams were modeled as a series of beam elements connected with crossing of beam element representing the

46 diaphragms. The shear key was considered to transfer bending and tension in addition to the shear. 46 (a) (b) Figure 2.5 Comparison of precast concrete sections: (a) original precast section; (b) modified precast section at diaphragm (adopted from El-R y et al. 1996) The transverse post-tensioning was constant per unit length of the bridge span, but varied with bridge width. The post-tensioning was the same for mid and quarter diaphragms. However, a minimum transverse post-tensioning stress of 250 psi was used for ends diaphragms in order to assure that the ends of the beams were adequately stiff. A design chart was proposed to calculate the required post tensioning as shown in Figure 2.6 for both mid and quarter spans. The authors recommended applying the posttensioning after grouting the shear key and that a full depth shear key should be used at each diaphragm. The transverse post-tensioning should be divided between the top and bottom for each diaphragm.

47 47 Figure 2.6 Presteressing force for mid span diaphragms (adopted from El-R y et al. 1996) The design chart of PCI s Bridge Design Manual (2003) was developed by El- R y et al. (1996) for standard AASHTO box girders. The design chart considered bridges with mild skew angles (less than 15 degrees), average span lengths and it was based on the HS-25 truck loading A new design chart was developed by Hanna et al. (2009), which was accounted for different spans, widths, and skews. Furthermore, the new chart was based on AASHTO LRFD specifications for truck and lane live load (HL- 93), as well as a dynamic allowance of 33% for truck load. The grid method was used to calculate member forces for the mid span diaphragm. Self-weight of solid concrete barriers was considered at 0.48 kip/ft. To develop the chart, the required post-tensioning for four standard box beams was investigated for different bridge widths with zero-degree skew and span to depth ratios of 30. It was noticed that the required transverse posttensioning increased as the bridge width increased. In addition, shallow box beams required higher post-tensioning than deep beams for the same bridge width. This was related to the reduction in the transverse stiffness in shallow beams compared with deep beams as shown in Figure 2.7.

48 48 Figure 2.7 Effective of bridge width on the required post-tensioning for 0.0 deg. skew and span-to-depth ratio of 30 at mid span diaphragm (adopted from Hanna et al. 2009) The updated chart showed an increase in the required post-tensioning of up to 40%, when the previous AASHTO specification for live load and dynamic allowance was considered as shown in Figure 2.8. The results also showed that there were variable effects from span to depth ratio on the required post-tensioning force. Higher transverse post-tensioning was required when the positive moment controlled the design. In addition, the skew angle had a minimum effect on the required post-tensioning especially for deeper beams with longer spans. However, the shallow beams required higher posttensioning when skew angle increased

49 49 Figure 2.8 Comparison between precast prestressed concrete bridge design manual charts with updated chart (adopted from Hanna et al. 2009) A simple equation was proposed to calculate the transverse post-tensioning for intermediate diaphragm per unit length of bridge span by considering all the variables. The difference in the required post-tensioning force was 7.7% between the proposed equation and grid analysis. P = ( 0.9W D Where: D = Box depth W = Bridge width 1.0) K LK S ( 0.2W D + 8.0) K LK S Eq.1-1 K L = Correction factor for span to depth ratio = ( L D 30) K S = Correction factor for skew angle more 0 deg. = θ L = Bridge span θ = Skew angle

50 50 The behavior of adjacent solid (without hollow section) precast box beam bridges and the design of the transverse post-tensioning connection were studied based on a shear friction concept by Fu et al. (2011). The effect of transverse post-tensioning in the tie rod was evaluated between two bridges. The first bridge was used to calculate the required transverse post-tensioning by using a finite element model. The deflection results showed that in the absence of the post-tensioning, each beam behaved independently. However, when the post-tensioning was increased, the bridge behaved monolithically and the load was distributed between the beams. When the post-tensioning exceeded 76 kips, there was no difference in the distribution of the load between the beams. The results showed that the 30 kip post-tensioning used by Maryland Department of Transportation was insufficient to transfer the load. The required post tensioning was determined to be 70 kips for span lengths less than 40 ft and 110 kips for bridge spans equal to or greater than 40 ft. The coefficient of friction between precast beam-shear key interfaces was an important factor to identify the required post-tensioning. The parametric study showed the required post-tensioning decreased when the coefficient of friction increased. The second bridge analyzed in the study consisted of eleven box beams using partial depth shear keys and a combination of transverse tie rods and a composite deck. Two beams from the bridge were instrumented at the bottom with two strain gauges for each beam at the tie rod position. The bridge was loaded by one truck located in three different positions. The results showed that vary transverse post-tensioning from 30 kip to 80 kips had no effect on the load transfer between beams before cracking of the shear key. The longitudinal strain did not change when the magnitude of post-tensioning was changed.

51 51 However, after cracks in the shear key and deck had occurred, the post tensioning was considered an important factor to transfer the load between beams. The post-tensioning force assisted to control on the cracks and also reinforced the bridge in transverse direction, which caused the bridge to behave as a single unit. The authors recommended applying the post-tensioning in two stages: one -sixth of design level should be applied before grouting the shear key in order to tie the beams together and the remaining part after the shear key was filled with grout. The effects of eliminating mid and end diaphragms on the behavior of adjacent box beams were investigated by Hansen et al. (2012). Four box beams were connected together by steel rods in transverse direction staggered every 8 ft. in both top and bottom of the box beam. The shear key was partial depth except at steel rods location which was a full depth of the beam. Ducts were used instead of diaphragms for high strength post tension steel rods. Strains and deflection were recorded at the both sides of the location of the joins that exposed to the maximum tensile stress while the system was loaded with static and cyclic loads. Two load configurations were used to create high tension in the joint. The first load configuration was used to create a high tension in the top flange. The second load configuration was applied with the load at mid span to create high tension in the bottom of flange. However, no cracks were observed in the joint. The box beams without diaphragms will be lower cost, and quicker to fabricate. The behavior of adjacent box beam bridges due to a combination of thermal and traffic load was investigated by Grace et al. (2012). Field observations, experimental testing, and numerical modeling were used in the research. Field observations were done

52 52 on two bridges. The first bridge was instrumented with thermocouples, and temperature gradients were monitored over a one-year period. The second bridge was only inspected to identify the position of the longitudinal cracks. Experimental testing consisted of four box beams connected by using full depth shear keys and transverse diaphragms. Finite element modeling was used to study the behavior of the bridge due to traffic and thermal load. These models were verified with experimental results. The results showed that the cracks were unlikely to have occurred due to the traffic load when transverse post - tensioning was used. However, when the bridge was subjected to a positive temperature gradient, cracks were observed. Under positive temperature gradients, the beams at the top tried to separate from one another, which generated a high tensile stress at the bottom of the deck causing longitudinal cracks. In addition, cracks from thermal loads propagated during traffic loading. Bridges with different widths, spans, and depths were studied using FE models to identify the number of diaphragms and post-tensioning required to prevent cracking in box beam bridges. The results showed that the transverse pressure distribution was not uniform and concentrated only in the diaphragm locations when post-tensioning was applied. The required number of diaphragms to eliminate the longitudinal cracking in the deck increased as bridge span enlarged. An increase in bridge width led to an increase in the required transverse post tensioning forces. The required post tension slightly decreased when the deck strength increased and was independent of the single box beam width.

53 Thermal Behavior Although, stresses caused by temperature changes are expected to have more effects on the behavior of adjacent box beam bridges, limited research has focused on the effect of the thermal stresses on the performance of the box beam bridges. The performance of a partial depth shear key due to thermal and cyclic loading was investigated by Hlavacs et al. (1996). In the research, a full scale model, consisted of four box beams connected together using non-shrink grout shear keys, was tested. Cracks were identified by using the pulse velocity and dye penetration methods. Box beams and shear keys were instrumented with embedded strain gauges. In the first test, the shear key was grouted in late autumn. Because of the cold weather, the grout was treated to prevent freezing. The bridge was covered and heated up from underneath for five days. Minor shrinkage cracks were observed. When the shear key was inspected after a few days, cracks were observed at the shear key interface with the beams. The cracks completely penetrated the middle shear key at mid-span, and partially penetrated another shear key near the abutment. The embedded strain gauges in the shear key recorded a high jump in strain. When the bridge was subjected to cyclic loading, the thermal cracks grew with no additional cracks due to load. The second test was completed in May. The shear keys were cracked after three days and cracking was observed in the middle and exterior keys near the abutment. No cracks were observed at the mid-span for all the shear keys. A high change in strain was recorded at the top flange of the beams due to the heating from the sun. The recorded thermal strain and deflection were higher than those from the

54 54 loading. The authors conclude that the cracks happened in young shear keys due to thermal load and propagated after applying the load in both cold and warm weather. The shear key configurations and grout materials in a full scale bridge were investigated by Miller et al. (1999). A top shear key was tested with non-shrink and with epoxy grout, while a mid-depth shear key was tested only with non-shrink grout. The non-shrink grouted top shear key was cast in November and just minor shrinkage cracks were observed after inspection. Due to some cold, snowy weather, load testing was done in January. After inspection, the shear key was found to be cracked before any load was applied. The strain gauges embedded in beams, as well as the shear key, recorded a high jump in the strain when the temperature reached 0 F. The bridge was tested with cyclic load and the cracks from temperature were propagated without any new cracks coming from the loading. The top shear key was also tested in May, and the key showed cracking after one week. In addition, the temperature was found to be the main cause of the cracking. The authors showed the temperature led to differential heating at the top of the bridge. Throughout the course of a day, the temperature caused the bridge to deflect up and down. The daily movement led some joints to open and others to close because the beams were not set perfectly on the abutment as shown in Figure 2.9. This movement created high strain causing the cracking in the shear key. Epoxy was also used as grout material in the shear keys. No cracks were observed under temperature effect or loads. Epoxy is undesirable due to the high difference in thermal expansion between epoxy and concrete. The mid depth shear key was constructed with non-shrinkage grout. Cracks were reduced because the throat was not grouted which helped to reduce temperature

55 55 stresses. The author s suggestions were to use post-tensioning to stop joint movement, neutral axes shear key with non-grouted throat, or full depth shear key. Rotation of the girders due to heat from the sun Figure 2.9 Girder movement and opening of shear key due to temperature loading (adopted from Miller et al. 1999) 2.2 Ultra High Performance Concrete (UHPC) Adjacent precast prestressed concrete box beam bridges are still exhibiting cracks in the shear key, and therefore, a grout material with high mechanical properties, durability, and bond strength may be able to eliminate or reduce cracking. Ultra high performance concrete (UHPC) represents a new class of concrete, which has high strength and long term durability characteristics. UHPC has been defined as cementitious composite material composed of an optimized gradation of granular constituents, a waterto-cementitious ratio less than 0.25, and a high percentage of discontinuous internal fiber reinforcement. The mechanical properties of UHPC include compressive strength greater

56 56 than 21.7 ksi and sustained post-cracking tensile strength greater than 0.72 ksi. UHPC has a discontinuous pore structure that reduces liquid ingress, significantly enhancing durability compared to conventional concrete (Yuan and Graybeal, 2014a). The most popular UHPC in North America that typically used for both research and applications is a commercial product that is known as Ductal. Typical mix design of UHPC is shown in Table 2.1. The premix powder usually consists Portland cement, fine sand, slice fume, and ground quartz. Table 2.1 UHPC mix design (adopted from Yuan and Graybeal, 2014a) Materials Amount (Ib/yd 3 ) Premix powder 3700 Water 219 Premia 150 * 30 Optima 100 ** 20 Turbocast 650 A + 39 Steel Fiber (2 %) * A modified phosphonate plasticizer ** A modified polycarboxylate + A non-chloride accelerator ++ Steel fiber content 2% by volume An essential component in the UHPC mixture is steel fibers. Steel fibers used in the mix are often non-deformed, cylindrical, high tensile strength steel, and have a diameter of in. and length of 0.5 in. The fiber tensile strength is typically specified to be greater than 290 ksi. The steel fibers have the ability to resist corrosion because it has a brass coating which provides lubrication during the drawing process (Yuan and Graybeal, 2014a). The UHPC has superior compressive and tensile behavior compared with convention or even high strength concrete (HPC) as shown in Figure 2.10 a and b (Li et al. 2015).

57 57 (a) (b) Figure 2.10 Compression of compressive and tensile behavior between UHPC and HSC (adopted from Li et al. 2015) The superior mechanical properties and durability of UHPC has led it to be used as a grout material for bridge connections. The bridge connection using UHPC becomes small in size, simple and more practical to implement as well as strong and durable, which improves the overall quality of the structure (Li et al. 2015). UHPC with dowels have been used in the longitudinal and transverse connections in precast prestressed concrete bridges (Graybeal 2014). The typical material properties of the field-cast UHPC is shown in Table 2.2. These material properties represent the average values for a number of tests on field-cast UHPC. Graybeal (2014) documents the design and construction of UHPC connections.

58 58 Table 2.2 Typical filed-cast material properties of UHPC (adopted fromyuan and Graybeal 2014a) Materials Characteristic Average Results Density Ib/ft Compressive Strength (ASTM C39;28-day strength) 18.3 ksi Modulus of Elasticity (ASTM C469;28-day modulus) 6200 ksi Splitting Cylinder Cracking Strength (ASTM C496) 1.3 ksi Prism Flexure Cracking Strength (ASTM C1018; 12 in. span) 1.3 ksi Mortar Briquette Cracking Strength (AASHTO T132) 0.9 Direct Tension Cracking Strength (Axial tensile load) ksi Prism Flexural Tensile Toughness (ASTM C1018; 12 in. span) I 30 =48 Long-Term Creep Coefficient (ASTM C512; 11.2 ksi load) 0.78 Long- Term Shrinkage (ASTM C157; initial reading after set) 555 microstrain Total Shrinkage (Embedded Vibrating wire gauge) 790 microstrain Coefficient of Thermal Expansion (AASHTO TP60-00) 8.2 x 10-6 in./in. F Chloride Ion Penetrability (ASTM C1202;28-day test) 360 coulombs Chloride Ion Permeability (AASHTO T259;0.5 in. depth) 3 < 0.10 Ib/yd Scaling Resistance (ASTM C672) No Scaling Abrasion Resistance (ASTM C944 2x weight; ground surface) oz. lost Freeze-Thaw Resistance (ASTM C666A;600 cycles) 112% Alkali-Silica Reaction (ASTM C1260; tested for 28 days) Innocuous Based on Federal Highway Administration (FHWA), the first bridge built with UHPC in North America was a pedestrian bridge in Sherbrook, Canada in However, the Mars Hill Bridge in Wapello County, Iowa was considered the first bridge in the U.S. which was constructed in In this bridge, all beams were made of UHPC. Perry et al. (2007) employed UHPC as grout material in the skewed 80 ft, single-span bridge at Rainy Lake, Ontario, Canada. The original bridge, which was constructed in 1962, consisted of steel plate girders and a cast in place reinforced concrete deck. The bridge was replaced with precast deck components, connected longitudinally and transversally by using UHPC which were supported on the original steel plate girders. The precast deck panel was 19 ft by 12 ft and reinforced with Glass Fiber Reinforced Polymer (GFRP) in the top mat and normal reinforcement in the bottom mat. The

59 59 longitudinal connection consisted of standard headed studs welded on the top flange of the girder and embedded in the pockets of the panels which were filled with UHPC. In addition to the headed studs, two layers of reinforcement were embedded 7.5 in. in the pocket joint. One row of reinforcement, beneath the top layer, was added in the longitudinal direction. The transverse panel-to-panel connection was constructed with a simple steel configuration without looped bars. The joint width was reduced to 8 in. compared to the conventional design of 24 in. for both longitudinal and transverse joints. Bar embedment lengths of 6 in. and 4 in. were used for the top and bottom reinforcement, respectively. UHPC was implemented in the longitudinal and transverse joints of the three span an adjacent box beam Eagle River Bridge (Perry et al. 2010). A non-corrosive glass fiber reinforced polymer (GFRP) rebar was used in both longitudinal and transverse joints. Additional GFRP rebar was added to the transverse joint. The UHPC grout material was used successfully, and provided the required strength after the curing period. No instrumentation or connection details for both the longitudinal and transverse joints were mentioned. UHPC was also used in longitudinal, transverse and approach slab connections in two spans of the Hodder Avenue Underpass Bridge that was constructed in northwestern Ontario, Canada. The bridge consisted of two spans with total length of 33.5 ft each. Sixteen adjacent box girder bridges were connected using UHPC longitudinal shear key joints reinforced with GFRP (Li et al 2015). The shear key details used in the bridge is

60 shown in Figure The bridge did not utilize a reinforced concrete deck but covered with asphalt overlays only. 60 Figure 2.11 Shear key details between girders in Hodder Avenue Underpass bridge (adopted from Li et al. 2015) UHPC has been also used as grout material in the state of New York (Perry and Royce 2010). A prototype HPC panels were connected by a joint filled with UHPC material. Three-rebar configurations, (straight, hairpins, and headed stud) with three types of rebar (epoxy coated, galvanized and black) were tested. A significant bond was found between rebar and UHPC which reduced development length for rebar and therefore reduced the joint width. The joint between a pair of HPC precast deck panels showed no leaking or degradation after applying a million cycles of wheel loadings. After the laboratory test of deck panel connections, the UHPC was used to fill the joints of a single span deck bulb-tee bridge in the Village of Lyons, located in Wayne County, NY. The UHPC installation was completed successfully. The connection performance between two precast concrete deck specimens, connected by using Ultra High Fiber Reinforced Concrete (UHPFRC) with two different splice details (straight bars and U-bars), was analyzed. Both joint width and splice length

61 61 were studied while the specimens were tested under both flexural and fatigue tests (Lee et al. 2014). In the flexural test, two precast concrete deck specimens, with joint widths of 10, 8, and 6 in., were used for both straight and U-bars. The splice length was changed from 6.3 in. to 5.5 in. and 4.3 in. The results showed that the flexural response was not affected by the type of splice. The behavior of the monolithic precast concrete deck was similar to the connected precast concrete deck specimens. The splice length of 4.3 in. was adequate for both straight and U- bars. In fatigue behavior, the joint width between the precast concrete deck specimens was set to 7.1 in. with a splice length of 6.3 in. No appreciable damage was exhibited after 2 million load cycles. Early age cracks were observed, but they did not progress even after increasing the number of load cycles. The cracks at the interface were less than in. upon the completion of the test. The authors recommended using simple splice details with short joint widths when UHPFRC is used as the grouting material. Recently, a full-scale model of two adjacent box beams was used to evaluate four connection details by Yuan and Graybeal (2014b, 2016). The first configuration included partial and full depth shear keys grouted with non-shrinkage grout in combination with transverse post tensioning. The second configuration included partial and full depth shear keys grouted using UHPC and lap spliced reinforcement bars as shown in Figure During all the testing, the load was cyclically applied. The results showed that the shear key without cracks had the same behavior when using non-shrinkage grout or UHPC. The cyclic load was unable to cause any cracks regardless of post-tensioning. However, the load was able to propagate the pre-existing cracks in conventional grout connection,

62 62 independent of the level of post-tensioning. The UHPC connection exhibited a superior performance even when the direct tension was applied to the top of the beam, the cracks developed in the precast concrete beam and de-bonding at the interface or cracks at the shear key itself did not occur. (a) (b) Figure 2.12 Pair of adjacent box beam connected by UHPC and dowel bars: (a) partial depth, and (b) full depth (Yuan and Graybeal 2014b; with permission from the author) The behavior of a pair of box beams which was tested by FHWA was investigated using finite element modeling under live and temperature gradient loads by Steinberg et al. (2014). Two shear key (partial and full depth) configurations including dowels spaced at 4 in were modeled to connect a pair of box beams. The partial depth shear key

63 63 configuration model was calibrated by using available experimental data. The full depth shear key was modeled by using the same parameters used with partial shear key model because experimental data was not available. Maximum principal tensile stresses in both shear keys and dowel bars were investigated under different temperature gradients and dowel bar spacing. The results showed that maximum principal tensile stresses increased in the partial depth shear key after applying the positive temperature gradients. However, there was a decrease in maximum principal stresses for the full depth shear key. In addition, shear keys exhibited larger principal tensile stresses as dowel bar spacing increased. The principal stress in the dowel bars increased for both partial and full depth shear keys when temperature gradients were included. Both shear key configurations experienced differential deflection lower than The authors suggested partial depth shear key with dowel bar spacing of 12 in. 2.3 Dowel Bars In the application of UHPC as a field cast grout material for the bridge connections between prefabricated bridge components, a cold joint is typically developed between two materials if they were cast at different times. The interface between the two materials, transfers load using adhesion until specific value, which depends on the surface preparation, moisture content at the interface, and dimension stability of the grout material. If the stresses from the applied load exceeded the adhesion value, interface cracking develops. If there was no reinforcement placed across the interface, an interfacial failure occurs. However, the reinforcement across the interface will transfer the load after cracking until either yield or fracture in the reinforcement occurs.

64 64 Therefore, UHPC connections are typically utilize reinforcement (dowel bars) across their interfaces. The main function of the dowel bars is to transfer the load if cracking occurs either in the shear key or at the beam-shear key interface (Sargand et al. 2017). The stress that the dowel carries depends of the type of stress that the joint must resist during its service life. For instance, the main faction of the dowel bars in a joint subjected to moment is to resist the flexural cracking. Furthermore, dowels carry more shear stress if the main function of the joint was to resist shear stresses. The main function of the dowel in the case of adjacent box beam bridge connections would be to resist both flexural and shear stresses. In the International Federation for Structural Concrete (fib 2007), shear force across concrete elements is transferred by adhesion or friction that occurs at the interface of the joints, dowel action, and shear keys effects (fib 2007). The adhesion mechanism is used for un-cracked joints with low shear application. The adhesion bond depends on the surface preparation and can be lost due to poor preparation of the joint surfaces. The shear friction mechanism occurs when the joint has sufficient roughness and a compressive force acting on the joint s interface can enhance this behavior. The compressive force is either internal or external. Figure 2.13 (a) shows the external compressive force (N c ) across the joint. As can be seen from the figure, the friction mechanism is enhanced by applying the compressive force. Figure 2.13 b and c show the internal compressive force generated by the transverse bars across the joint. When the shear force (F v ) is applied, one part slips relative to the other with differential movement (s). If the surface is rough, this slip tries creating a joint movement (w) in the transverse

65 65 direction. The transverse movement creates tensile stresses in the reinforcement across the joint. An internal compressive force, which is equal to the tensile force, will then be applied to the joint. (a) (b) (c) Figure 2.13 Shear transfer by friction: a) external compressive force across the joint, b) relative joint displacement, c) internal compressive force generated by transverse bars across the joint (adopted from fib 2007) The dowel action mechanism depends on the way that the reinforcement is utilized across the joint. When end anchorage and/or sufficient bond is provided for the reinforcement, axial restraint will be developed in the joint. Both axial and flexural stress develop in the dowel when the connection is subjected to the shear force. As already discussed, the axial restraint will improve joint capacity through shear friction. However, the dowel action will be insignificant because the dowel will yield due to small joint separation and high axial restraint before the development of the dowel action. If the dowel is without end anchorage or lacks bond, the dowel will resist shear displacement by bending deformation. Axial restraint is typically neglected in this case. If the bond is ineffective to yield the bar with small shear slip, large shear slip will develop and will force the bar to act as a dowel before yielding. In this situation, the shear resistance will

66 66 be a function of both shear friction and the dowel action, (fib 2007). A shear key works as a mechanical interlock to prevent sliding along the shear key joint. The shear key geometry (tooth length, tooth depth, and tooth inclination) can also enhance the shear transfer across the joint by preventing any significant uncontrolled separation across the joints as shown in Figure Figure 2.14 Shear key geometry (adopted from fib 2007) According to ACI , shear transfer should be considered when there is an existing or potential crack plane, an interface between dissimilar materials, or at the interface of two concretes cast at different times (ACI 2014). The shear transfer concept assumes that the cracks will form and reinforcement should be provided across the crack to resist the relative displacement. When the shear force is applied, one side will slide relative to the other side. The sliding is accompanied by separation of the joint due to the roughness at the interface. The separation will lead to stress in the reinforcement across the joint, and a clamping compressive force equal to the tensile force in reinforcement (Avf fy) will be applied to the interface. The applied shear will be resisted by friction at the interface, by resistance to the shearing along the protrusions on the crack faces, and the dowel action of reinforcement.

67 67 The same concept of shear friction is used in AASHTO LRFD Bridge Design Specifications section C The shear displacement along the interface might be resisted by two components, which are cohesion-aggregate interlock and shear friction. These components differ from the ACI because it considers the cohesion effect and the aggregate interlock. In addition, the interfacial shear resistance is directly proportional to the net normal clamping force (A vf f y + P c ) through the friction coefficient (µ). The clamping force is generated by the reinforcement cross the interface (Avf fy) and/or by an external compressive force normal to the shear plane (P c ), (AASHTO LRFD 2016). However, the interface shear strength equations were developed for the monolithic concrete or for precast concrete and cast in place concrete (cold joint). 2.4 Summary The literature review explored the research that has been done to investigate the behavior of box beam bridges. The longitudinal cracking in the longitudinal shear key joints between adjacent beams is considered a major issue. Many solutions including shear key configurations, shear keys grouting materials, and transverse post-tensioning have been suggested to eliminate the cracks in the longitudinal joints. However, cracks are still observed in some adjacent box beam bridges due to either live loads or a combination of live and thermal loads. Therefore, in this research UHPC was used as a grout material with dowel bars in the longitudinal shear key connections of an adjacent box beam bridge. Testing at Turner Fairbank Highway Research Center (TFHRC) involved connecting two box beams together and applying a concentrated load. The results show that the new design was sufficient to make the bridge behave as a unit with

68 68 no cracks recorded in the shear keys even after numerous cyclic loading. Finite element modeling of the laboratory testing along with a parametric analysis was also performed and shown that larger spacing of dowel bars could be used (Ubbing 2014; Steinberg et al. 2014). However, the design was already far along and field behavior can sometimes greatly differ from laboratory conditions. Therefore, a conservative decision was made to utilize the same dowel spacing in the bridge as utilized in the Turner Fairbank Highway Research Center (TFHRC) testing. Similar shear key designs have been used in bridges that are in service in Ontario, Canada. Unfortunately, no data exists to quantify the performance of this type of shear key design. The behavior of the first adjacent box beam bridge in the U.S. to utilize this type of connection will be investigated in this research.

69 69 CHAPTER 3: METHODOLOGY In this chapter, the bridge description, beam fabrication, instrumentation, and bridge construction utilized in this research will be discussed. The static and dynamic truck load tests will also be described. The last section will discuss the environmental test. 3.1 Bridge Description The adjacent prestressed concrete box beam bridge was constructed on Sollars Road in Fayette County, Ohio near the town of Washington Court House. The bridge consisted of seven beams adjacent to one another. The bridge length was 61 ft. long with zero skew angle and had a width of 28 ft. Each beam was 48 in. wide and 21 in. deep. For reference, the beams were numbered 1 to 7 from left to right while facing the forward abutment as shown in Figure 3.1. A total of 28 half-inch diameter seven-wire low relaxation strands with an ultimate strength of 270 ksi were used in each box beam. The mild shear reinforcement had a yield strength of 60 ksi. A 33 in. long diaphragm was used at each end of each box beam. The beams were set on abutments which were supported on three drilled shafts. The beams were placed on two bearing pads at the end of each beam and one dowel bar at each beam end to connect the beam to the abutment. For each beam, one dowel bar was grouted to create fixed condition on one end while the dowel was filled with material that allowed for expansion on another end. Ohio typically requires transverse tie rods; however, no tie rods were used in this bridge. In addition, transverse post-tensioning or a composite deck, was not implemented. The bridge utilized a new shear key design that was developed and tested at the Turner Fairbank Highway

70 70 Research Center (TFHRC) as shown in Figure 3.2. The new design consisted of using UHPC as the grout material with equally spaced dowel bars in each joint. The shear key was larger than a typical shear key and the dowel bars were threaded into the beams and were staggered at a 4 in. spacing. Each beam had shear keys on both faces except for the exterior beams, which only had one interior face shear key. The dowel bar system had two parts. The first part was embedded in the beam 18 in. and contained a female threaded end. The part that was embedded in the shear key had a length of 4.75 in. and had a male threaded end allowing it to be screwed into the part embedded in the beam as shown in Figure 3.3. The connection details and dowel bars spacing are shown in Figure 3.4.

71 71 (a) (b) Figure 3.1 Bridge description: (a) erection diagram, (b) bridge section (adopted from Semendery et al. 2017a,b) Figure 3.2 Shear key detail (all dimensions in inches, adopted from Steinberg et al. 2015)

72 72 Figure 3.3 Dowel bar details: (a) dowel bars splice, and (b) straight dowel bar splicers (a) (b) (all dimensions in inches) Figure 3.4 Connection details: (a) cross section, and (b) top view

73 Beam Fabrication and Instrumentation The box beams were fabricated in Kalamazoo, Michigan during May of 2014 in a precast prestressed concrete manufacturing facility. The typical box beam form was used, except the shear key shape was modified using wood forms. For exterior Beams 1 and 7, the shear key modification was used only on the inside face of the beams. For the interior beams, the shear key form was used on both sides. The wood form for the new shear key can be seen in Figure 3.5. The form was coated with a retarder and the female threaded ends of the dowel bar parts internal to the beam were placed on red plastic tabs to assure concrete did not enter the threads. Figure 3.5 also shows the internal dowel bar parts installed in the beam form. Figure 3.5 Shear key form and beam dowel parts in place The first three box beams were instrumented with vibrating wire strain gauges embedded in the beams and on the dowel bars. Fifteen vibrating wire strain gauges were used in the three beams. Five strain gauges were used in each beam to monitor the strain in longitudinal and transverse directions. Two vibrating wire strain gauges, one in the top

74 74 flange and one in the bottom flange, were placed longitudinally at the quarter span as shown in Figure 3.6 a. Three vibrating wire strain gauges, one longitudinal and one transverse in the top flange and one longitudinal in the bottom flange, were used at midspan as shown in Figure 3.6 b. (a) (b) Figure 3.6 Instrumentation in the top and bottom flanges in beams 1-3: (a) quarter span, (b) mid span The bottom gauges were positioned between strands and the top gauges were mounted between the shear reinforcement. Figure 3.7 shows a longitudinal vibrating wire strain gauge positioned in the form between the strands at the bottom flange. The strain gauges positions for Beams 1-3 are shown in Table 3.1. The wiring for the instrumentation was routed to extra plastic drain forms and sealed with tape. This allowed access to the wiring after the beams were delivered to the bridge site. Figure 3.8 shows the gauges on the top flange at quarter span and mid span, respectively. The wiring for this instrumentation was run in between the Styrofoam void inserts and into a drain form.

75 75 Figure 3.7 Vibrating wire strain gauge (adopted from Semendary et al. 2017b, Steinberg et al. 2015) Figure 3.8 Instrumentation in the top flange at mid and quarter span (adopted from Semendary et al. 2017b)

76 76 Table 3.1 Locations of strain gauges embedded in beams 1-3 Location Quarter span Mid Span + Longitudinal *Transverse Beam No Direction General Depth Distance from L-to- R (in.) Distance from Bottom (in.) Distance from Rear Abutment (in.) L + Top L Bottom L Top L Bottom L Top L Bottom L Top T * Top L Bottom L Top T Top L Bottom L Top T Top L Bottom Dowel bars in each beam were instrumented using vibrating wire strain gauges, as shown in in Figure 3.9. The gauge was installed at a distance of 2 in. from the threaded end of the dowel bar. A total of six dowel bars were instrumented using the vibrating wire strain gauges. Beams 1, 2, and 3 contained the instrumented dowel bars on the right side of the cross section. Two instrumented dowel bars were installed for each beam, one at mid span and one at the quarter span. The installed instrumented dowel bar can be seen in Figure 3.9 and included a protective shield to avoid damage from the concrete.

77 77 Figure 3.9 Instrumented dowel bar and dowel bar covered with a protective shield (adopted from Steinberg et al. 2015,2016) Beam 3 was also instrumented with four thermocouples through the depth on the left side of the cross section, as shown in Figure 3.10 to measure temperature gradients. Table 3.2 shows position details for the thermal couples in Beam 3. Figure 3.10 Thermocouples through the depth of Beam 3 (adopted from Steinberg et al. 2015)

78 78 Table 3.2 Locations of thermocouples embedded in beam 3 Position from Thermocouples Thermocouple the bottom Position Name (in.) Mid Span Distance from L-to R (in.) Distance from Rear Abutment (in.) TH TH TH TH The first beam was cast on May 7, The following day, the pre-stressed strands were cut and the box beam was lifted from the form and loaded onto a truck. While the beam was on the truck, power washing with water was used to obtain an exposed aggregate surface in the shear key as shown in Figure The beam was then moved to the yard for storage. The same procedures were used for the remaining beams. Figure 3.11 also show the shear key upon removal from the box beam form and after power washing. The result was a rough shear key surface with exposed aggregate that enhanced bond between the beams and UHPC. Figure 3.11 Shear key preparation

79 79 On May 28, 2014, the contractor began to remove the previous bridge and the new bridge s foundations and abutments were constructed. On Saturday July 12, 2014, the box beams were transported to the site. Six vibrating wire strain gauges were installed on six male threaded dowel bars. Each dowel bar was instrumented with one gauge at distance of 1.5 in. from the threaded end as shown in Figure While the beams were on the truck, the dowel bars were screwed into the female threaded part of the dowel bar embedded in the beams. Instrumented dowel bars were installed to the left side of the cross section of Beams 2, 3, and 4. Two instrumented dowel bars were used in each beam, one at the quarter span and one at mid span at a distance of 208 in. and 368 in. from the rear abutment, respectively, as shown in Figure The beams were then moved from the truck and set on the abutments using a crane. Each end of each beam had bearing pads between the abutment and beam. In addition, one positioning vertical dowel bar at each end of each beam was placed through the beam into the abutment. The forward abutment vertical dowel bars allowed for expansion by using a joint sealer around the dowels. The rear abutment vertical dowel bars were grouted into place to create longitudinal translation fixed conditions. Figure 3.12 provides a view of the shear key with dowels after beam placement.

80 80 Figure 3.12 Instrumenting and installed a shear key dowel bar and shear key with dowels (adopted from Semendary et al. 2017b, Steinberg et al. 2016) On July 16, 2014, the three shear keys between Beams 1-4 were instrumented with vibrating wire strain gauges. Six short length vibrating wire strain gauges were set in a transverse direction of all three shear keys as shown in Figure Each shear key was instrumented with one transverse gauge at the quarter span and one transverse strain gauge at the mid span. Four vibrating wire strain gauges were used in the longitudinal direction as shown in Figure Shear Key 1 and 3 were instrumented with one longitudinal gauge at quarter span and one longitudinal gauge at mid span. The strain gauges distribution is shown in Table 3.3.

81 81 Figure 3.13 Transverse and longitudinal shear key gauge (Semendary et al. 2017b with permission from ASCE) Table 3.3 Locations of strain gauges embedded in shear keys 1-3 Position Shear Key Number L- to R Direction Gauge s location from Top (in.) Quarter Span Mid Span Distance from Rear Abutment (in.) Longitudinal Transverse Transverse Longitudinal Transverse Longitudinal Transverse Transverse Longitudinal Transverse After installation, excess expandable filler material between the beams was removed, and the joints were covered with plywood except for the larger openings at the quarter points along the shear key. On July 17, the shear key joints were cast using UHPC. Two mixers were used to properly mix the UHPC. The JS1000 UHPC was mixed and moved to the joints in wheelbarrows and placed into chimneys made of plastic

82 82 buckets located at the plywood openings as shown in Figure The UHPC flowed into the joints and the filling of the joints was assured by a hydraulic head of the UHPC in the chimneys. Instrumentation was connected to data acquisition systems in order to monitor the bridge as the UHPC cured. On July 22, the plywood forms were removed from the joints. No cracks were observed from inspection in the shear keys as shown in Figure Figure 3.14 UHPC placement into longitudinal shear keys and shear keys after plywood forms removed On July 24, a waterproofing membrane was installed on the top of the bridge as work continued on the approaches to the bridge as shown in Figure An asphalt wearing surface was paved on the bridge on August 5.

83 83 Figure 3.15 Waterproofing installation The following day, an instrumentation frame was set up underneath the bridge at mid-span as shown in Figure On August 7, a total of 16 strain gauges, seven LVDT s, and three thermocouples were installed to monitor the bridge. Seven electrical resistance strain gauges, WFLM LT, were glued to the bottom of the beams in the longitudinal direction at the mid span to measure the strain from truck loading. Seven KM- 100B gauges, were installed in gauge brackets that had been epoxied to the bottom of the beams. These KM-100B s were placed in the longitudinal direction at mid span to measure the strain from temperature loading. Two KM-100B s were also epoxied to the frame in the vertical direction on two columns at a distance of in. of the left column and in. of right column from the end of beams of the rear abutment to measure the strain in the frame from temperature as shown in Figure Three thermocouples were used: one thermocouple on the left column of the frame, one thermocouple on the right column of the frame and one on the bottom of Beam 4. The seven LVDT s were mounted to the frame using brackets. The frame was used as a reference surface. Figure 3.16 also shows the LVDT mounted to the frame along with a KM-100B mounted in a bracket

84 84 temporarily held in place with duct tape as the epoxy set, and the electrical resistance strain gauge (white). The gauges and LVDT s positions are shown in Table 3.4. Figure 3.16 Instrumentation frame and bridge instrumentation

85 85 Table 3.4 Location of strain gauges and LVDTs Beam Number Instrumentation Type Direction Distance from L-to- R (in.) Distance from Rear Abutment (ft) WFLM LT Longitudinal KM- 100B Longitudinal LVDT Perpendicular to the bottom WFLM LT Longitudinal KM- 100B Longitudinal LVDT Perpendicular to the bottom WFLM LT Longitudinal KM- 100B Longitudinal LVDT Perpendicular to the bottom WFLM LT Longitudinal KM- 100B Longitudinal LVDT Perpendicular to the bottom WFLM LT Longitudinal KM- 100B Longitudinal LVDT Perpendicular to the bottom WFLM LT Longitudinal KM- 100B Longitudinal LVDT Perpendicular to the bottom WFLM LT Longitudinal KM- 100B Longitudinal LVDT Perpendicular to the bottom On December 15, 2014 five LVDT s and three thermocouples were added to the instrumentation. Joints 4, 5, and 6 were instrumented with three LVDT s at mid span at a distance of 366 in. from the beam ends of the rear abutment. Two brackets were used to install the LVDTs transversely across the joint as shown in Figure Two LVDT s were installed at the two edges of Beam 7 to monitor the beam movement due to

86 86 temperature changes. One thermocouple was installed on Beam 1 and another on Beam 7 to monitor the temperature on both sides of the bridge. One thermocouple was installed on the bottom of Beam 3 at mid span to monitor the temperature at the bottom of the beam in addition to the four thermocouples which had been already embedded in this beam. Figure 3.17 Bridge instrumentation On July 9, 2015 seven LVDT s and two KM-100B were disconnected because the frame was removed. Two thermocouples were added to the instrumentation. The temperature through the depth of Beam 3 was measured using the five thermocouple distributed through the depth. Four thermocouples were interior and one was exterior. In addition, the temperature at the bottom of Beams 1-7 was measured. 3.3 Truck Test Static Truck Test One August 8, 2014 two trucks were used to test the bridge. The gross weights of the trucks were 56.1 kip and 53.4 kip. The axle dimensions of each truck were recorded

87 87 and each axle load was determined using truck scales. The trucks dimensions and weights are shown in Table 3.5. Four static load configurations were used in the tests, and the trucks were positioned to obtain the maximum moment at mid-span. These load configurations were: o A single 56.1 kip truck load placed in the left lane o A single 53.4 kip truck load placed in the right lane o Two trucks placed side-by-side creating a kip total load o Two trucks placed back to back in the left lane for a kip total load Table 3.5 Axle dimensions and axle and tire loads (adopted from Semendary et al.2017a) Truck Number 1 2 Axle Axle Width, (in.) Distance Between Axles (in.) Left Tire (Kip) Right Tire (Kip) Axle Load (kip) Truck Total Load (Kip) The truck load positions are shown in Figure Data was collected with two data acquisition systems as shown in Figure 3.19 while the truck(s) were static on the bridge. The exterior instrumentation, which included seven WFLM LT gauges and seven LVDT s, were connected to the data acquisition 1. All embedded instruments, KM-100B s, and thermocouples were connected to the data acquisition 2. The data acquisition 2 was programed to take readings every fifteen minutes. The data acquisition 1 took readings while the truck(s) were placed on the bridge. Because the time required to

88 take the readings for the data acquisition 2 was set for fifteen minutes, the truck(s) were left on the bridge until data acquisition 2 recorded the data. 88 (a) (b)

89 89 (c) (d) Figure 3.18 Truck loading (adopted from Semendary et al.2017a)

90 90 Figure 3.19 Data acquisitions system Dynamic Truck Test After finishing the static truck load testing, the bridge was loaded with a moving load by driving one truck at five speeds of 5, 10, 15, 25, 30 mph as the data was collected. After finishing the dynamic truck load test, the seven LVDT s which were used for truck test were connected to data acquisition 2 for long term monitoring. 3.4 Environmental Test The first environmental behavior of the bridge was monitored from July 17-25, 2014 when the joints were filled with UHPC. The collected data was used to understand the behavior of the bridge during the curing period of the UHPC grouted materials in the shear keys. Data was collected by using a Campbell Scientific CR3000 data acquisition system. Data was collected with thirty-two of the thirty-seven vibrating wire strain gauges, which were embedded in the beams, shear keys, and on dowel bars due to the data acquisition s capacity. The data collection started from 13:05 on July 17, 2014 until

91 91 10:30 of July 25. Data was recorded every five minutes. Strain and temperature were recorded by the data acquisition for each gauge. The second environmental behavior of the bridge was monitored from August 8-16, The data collection started from 12:30 on August 8, 2014 after finishing the truck test until 17:45 of August 16. Data was recorded every fifteen minutes. Strain and temperature were recorded by the data acquisition. The same vibrating wire strain gauges were connected as in July. In addition, nine KM-100B strain gauges, seven LVDT s, and seven thermocouples were connected. All instrumentation was connected to the CR3000 data acquisition system. The third environmental behavior was monitored in December 2014 and January 2015 by using the same vibrating wire strain gauges connected as in July, nine KM-100B strain gauges, which were the same ones as connected in August. However, twelve LVDT s and ten thermocouples were connected for this time periods. All instruments were connected to CR 3000 data acquisition system. The data was collected in different short time periods because the cold weather caused the battery to be replaced frequently. The date and time of the data collection is shown in Table 3.6. Table 3.6 Date and time of data collection Start of collecting data End of collecting data December 15, 2014 at 14:45 December 17, 2014 at 19:45 December 19, 2014 at 10:45 December 25, 2014 at 0:00 December 30, 2014 at 10:00 January 1, 2015 at 19:00 January 7, 2015 at 13:45 January 10, 2015 at 7:00 January 15, 2015 at 10:45 January 19, 2015 at 10:45

92 92 The forth environmental behavior was monitored by using the same vibrating wire strain gauges connected as in July. In addition, seven KM-100B strain gauges were connected. The KM-100Bs on both columns of the frame were not connected because the frame was removed. Five LVDT s and twelve thermocouples were connected. Because the frame was removed, only the LVDT s used to monitored the joint movement (Joints, 4, 5, and 6) and the LVDT s used to measures the longitudinal movement at Beam 7 were connected. All instruments were connected to the CR 3000 data acquisition system. The data collection started from 13:30 July 9, 2015 until 4:15 July 14, 2015.

93 93 CHAPTER 4: BEHAVIOR OF AN ADJACENT PRESTRESSED CONCRETE BOX BEAM BRIDGE UNDER STATIC AND DYNAMIC TRUCK LOADS 4.1 Static Truck Load Test Results Data was collected by using two data acquisition systems during the truck load test. One data acquisition was used to read interior instruments and the other for exterior instruments. The static truck test included data collected from strain gauges (WFLM LT) which were glued to the bottom surface of each beam and from LVDT s. Data from embedded instruments was also collected while the truck was in a static position on the bridge. Initial readings were taken by both data acquisition systems before the truck was positioned on the bridge. The truck for each loading configuration was left on the bridge until the CR3000 data acquisition read the embedded gauges because the CR3000 collected data every fifteen minutes Deflection The deflection was measured by using LVDT s when the bridge was loaded with the first load configuration, which was one truck parked on the left side. The total truck load was 56.1 kips. Readings were taken for both the initial non-loaded state and when the truck was on the bridge. Deflection was calculated by subtracting the initial reading from the final reading and the results are shown in Figure 4.1. The results showed that the loaded beams generally exhibited the largest deflection. The reading of LVDT 1 was removed because it was unreliable. LVDT installation caused the tip of LVDT 1 to stick during the test and did not register a reasonable measurement. Beam 2 showed the largest deflection of in. The LVDT on Beam 4 did not register any measurement.

94 94 Beams 5-7 all deflected under the load configuration, which emphasizes the ability of the new shear key configuration to transfer the live load between beams, thereby causing the bridge to deflect as a system. Figure 4.1 Mid span deflection for load configuration 1 (1 truck on left) The deflection was recorded when the bridge was loaded with a second load configuration which was one truck on the right side. The total truck load was 53.4 kips. The results showed that loaded Beam 7 had the highest deflection of 0.32 in. However, non-loaded beams still exhibited deflection due to the load transfer mechanism, as shown in Figure 4.2.

95 95 Figure 4.2 Mid span deflection for load configuration 2 (1 truck on right) Then, the bridge was loaded with a third load configuration, which was two trucks at mid span. The total weight of the trucks was kips. Deflections were recorded by the data acquisition systems. The maximum deflection of in. was recorded for this load configuration on Beam 7 as shown in Figure 4.3. All beams exhibited deflection due to load transfer mechanism. The right side of the bridge exhibited more deflection than the left side. However, the truck on the right side was 53.4 kips and the truck on the left side was 56.1 kips. The difference in the stiffness between the beams could be one reason. Lower stiffness leads to an increase in deflection. Therefore, higher deflection of the right side indicated that Beams 5-7 had lower stiffness compared with the left side. The difference in the stiffness between beams could also be related to the difference in compressive strength. This could result from differences in the mix design, temperature during and after the casting, and also the curing periods. The difference in the prestressing force at the release could also cause a difference in stiffness between beams. The locations of the LVDT s were also not exactly at mid span. LVDTs (1-7) were

96 96 located at ft, ft, ft, ft, ft, 31.58, and ft, respectively. The difference between the locations of LVDTs 1 and 2 and center of the bridge were 0.42 ft, 0.75 ft toward the rear abutment, and the difference between the locations of LVDTs 3-7 and center of the bridge were 0.25 ft, 0.54 ft, ft, 1.08 ft, and 1.42 ft toward the forward abutment. This indicates that LVDTs 3-7 were subjected to higher load as the locations were toward the rear tire during the test of load configurations 3. For example, the distance between LVDT 2 and the center of the first tire in back axle was approximately 2 ft while the distance for LVDT 7 was 0.26 ft which mean the LVDT was directly under the load. This could be another reason for the highest deflection on the right side compared with the left side. Figure 4.3 Mid span deflection for load configuration 3 (1 truck on left and 1 truck on right) The deflection, when the bridge was loaded with two trucks back to back on the left side, was measured as shown in Figure 4.4. The total load of the trucks on left side was kips. The maximum deflection was in. occurred on Beam 2. However,

97 97 Beam 7 still had deflection due to transferring of the load between beams by the new shear key configuration, which caused the bridge to deflect as a system. Figure 4.4 Mid span deflection for load configuration 4 (2 trucks on left back to back) A comparison between the deflections for the different load configurations is shown in Figure 4.5. The figure shows that load configuration 3 had the largest deflection. The load configurations 1 and 2 were combined and compared with load configuration 3. The results were consistent in the trend even though there was a difference in magnitude. The slight difference in magnitude was related to the difference in the stiffness between beams. The varies in stiffness could be related to the difference in mix design, curing conditions, and prestressing forces between beams. It may be also related to the position of the trucks between the configurations that were compared. The location of the truck on the right side was the same in load configurations 2 and 3. However, there was slight difference in the location of the truck on the left side. The

98 98 difference in LVDTs locations could also have an effect of the comparison. If the all LVDTs were located at the center of the bridge, a better comparison may be achieved. Figure 4.5 Mid span deflection for all load configurations Longitudinal Strain Monitored by Using Exterior Strain Gauges The exterior bottom strains were measured during the truck load test. The strain was calculated by taking the reading for both the initial non-loaded state and when the bridge was loaded with trucks. The final strain was calculated by subtracting the initial non-loaded state strain from the final strain. The positive strain constitutes tension while the negative strain constitutes compression. The bridge was loaded with one truck on the left side (configuration 1) with a truck weight of 56.1 kips. The loaded beams exhibited maximum strain on the bottom at mid-span compared to the non-loaded beams as shown in Figure 4.6. The maximium strain recorded by Beam 1 was 61 µε. The results also show that Beam 7 had higher strain than Beams 4, 5, and 6, though it was located farthest from the applied load. This difference might be due to the beams having slightly lower

99 99 stiffness. The lower stiffness led to increase the strain and therefore the higher strain in Beam 7 indicates that this beam had a lower stiffness. The strain gauges were not exactly at mid span, 30.5 ft from the end of the beam, but were placed at ft, ft, ft, ft, ft, 30, and ft, respectively for strain gauges (1-7). The difference between the locations of strain gauges 1-7 and center of the bridge were 2.92 ft, 2.67 ft, 1.92 ft, 1.75 ft, 1.75 ft, 1.5 ft and 1.17 ft toward the rear abutment, respectively, and were located before the center line, toward the rear abutment. This indicates that the gauges with less distance were closer to the first tire of the back axle. This could explain why Beam 7 had the highest strain; it was the closest gauge to the center of the bridge. Figure 4.6 Mid span bottom strain for load configuration 1 (1 truck on left) The bottom strain at mid span when the bridge was loaded with one truck on the right side (configuration 2) with a total truck weight of 53.4 kips is shown in Figure 4.7. The loaded beams exhibited larger strains than the non-loaded beams. The maximium strain, which was recorded in Beam 7, was 68 µε. Beam 7 exhibited slightly larger strain

100 100 than Beam 1 in the previous load configuration. However, the weight of the truck in load configuration 2 was less than the truck weight in load configuration 1. This difference may be related to the beams having slightly different stiffness. Also, the lateral postion of the truck was slightly different. Figure 4.7 Mid span bottom strain for load configuration 2 (1 truck on right) The bottom strain at mid span when the bridge was loaded with two trucks positioned at mid span (configuration 3) with a total weight of kips is shown in Figure 4.8. Beams 1-3 had approximately the same strain, while Beam 4 exhibited the lowest value because it was not directly loaded and the bridge was symmetric about this beam. The maximium strain was 109 µε which occurred on Beam 7.

101 101 Figure 4.8 Mid span bottom strain for load configuration 3 (2 trucks at midspan) Finally, the bridge was loaded with two trucks back to back on the left side (configuration 4) with a total weight of kips. Figure 4.9. provides the bottom strains at mid span for load configuration 4. Beam 1 had larger strain than the other beams. The maximium recorded strain was 95 µε. The results show that Beam 7 again had higher strain than Beams 4, 5, and 6. This difference might be due to the beams having slightly different stiffness

102 102 Figure 4.9 Mid span bottom strain for load configuration 4 (2 trucks on left back to back) The strains from all load configurations were compared as shown in Figure Configuration 3 showed the maximium recorded strain. Load configurations 1 and 2 were combined and compared with load configuration 3. The results were consistent in both trend and magnitude. However, the right side compered better than the left side because the location of the truck on the right side was the same in load configurations 2 and 3. However, there was slight difference in the location of the truck on the left side.

103 103 Figure 4.10 Mid span bottom strain for all load configurations The maximum strain was 109 µε on Beam 7 when the bridge was loaded with two trucks at mid span (configuration 3) with a total weight of kips. This strain was lower than the cracking strain which was calculated by dividing the cracking stress on modulus of elasticity. The compressive strength of concrete box beam was determined from cylinders with a dimentions of 6 in. by 12 in. during the construction of the beams. The cylinders were tested for compressive strength according to ASTM C 39/C 39M 05 (ASTM 2015) in structural lab during the time of truck test. The test was conducted approximately three months after casting. The beams reached a compressive strength of 11 ksi at the time of the truck test. The modulus of elasticity was calculated to be 5988 ksi by using the following equation from ACI (ACI 2014). E c = f c Eq. 4-1 Where: f c = Compressive strength of the concrete at the time of the truck test.

104 104 The tensile strength of the prestressed box beams was assumed to be ksi by using the following equation from ACI (ACI 2014). f ct =7.5 f c Eq. 4-2 The cracking strain was determined from static to be f ct E = 131 µε. The recorded strain was lower than the cracking strain. However, the recorded strains were due to external load only and did not include the effects of self weight and prestressing forces. The strain due to prestressed force and the beam self weight after losses should be added to field measurment readings. Data from Beam 1 is shown in Table 4.1 Strains from the gauges embedded in the top and bottom flange of Beam 1, at mid and quarter span, were recorded before and after cutting the prestressing strands. The redaings perior casting of the beam were considered as initial readings and another redings were taken after cutting. By substracting the intial readings from readings the after cutting, the strain were calculated as shown in first colum in Tbale 4.1. The gauges at bottom flange were at the location of the strands. The age of the beam at the time of truck testing was approximately three months. Therefore, the initial strain readings changed from prestress losses due to steel relaxation, concrete creep, and concrete shrinkage. The measured strains prior to truck loading are shown in the second column of Table 4.1. The results show a compressive strain of 237 µε at the level of the strands at mid-span. The strains were also calculated using the basic equations from prestressed concrete under the effects of self weight and prestressing force and the results are shown in the third column of Table 4.1. The losses were calculated using the approximate method in AASHTO LRFD (AASHTO LRFD 2016). The calculated strains were lower than the measured

105 105 strains and could be the result of differences in losses, initial prestressing force, and loadings from self-weight and the wearing surface. The calculated strain at the level of the strand at mid-span was 187 µε (compression). The tensile strain in the bottom of the beam (77 µε) at the level of the strands, from the truck loading, was not high enough to overcome the pre-compressive strains and therefore cracking of the beams was not expected or observed. This includes the extreme bottom of the beams because the highest tensile strain was oberved at the extreme bottom face of the beam. The precompression at the extreme bottom face was anticipated to be higher. Table 4.1 Interior top and bottom flange longitudinal strains in beam 1 at mid and quarter span Position Prestress + selfweight Strain before truck test Strain due to truck load for different load configurations M a C b Load 1 Load 2 Load 3 Load 4 Bottom longitudinal at quarter span Top longitudinal at quarter span Bottom longitudinal at mid span Top longitudinal at mid span a Measured b Calculated Transverse and Longitudinal Strains Monitored by Using Interior Strain Gauges The strain data obtained using strain gauges embedded in Beam 1-3 was analyzed both at mid and quarter span. The strains recorded by using the transverse strain gauges, which were embedded in the top flange of Beams 1-3 at mid span, are shown in Table 4.2 for different load configurations. The readings were taken every fifteen minutes because

106 106 the data acquisition was set at this frequency for long-term environmental loading of the embedded gauges. The readings were taken prior to and while the bridge was loaded with trucks for each load configuration. The total time for testing all configurations was relatively short and therefore the readings were not corrected for temperature. The strain for both the transverse and longitudinal strain gauges embedded in Beams 1-3 were determined as follows: R = f 2 Eq. 4-3 Where: R = Microstrain f = Frequency reading in Hz The actual strain was calculated by using: ε = B (R 1 R O ) Eq. 4-4 Where: B = Nominal batch factor = 0.97 R 1 = Gauge reading with bridge loaded R 0 = Initial reading prior to loading Overall the transverse strains were relatively low. The results showed that when the bridge was loaded with one truck on the left, the maximum strain was 22 µε in the exterior beam (Beam 1). However, when the bridge was loaded with one truck in the right lane, the strain was approximately equal for Beams 1-3. This was due to none of the beams being directly loaded. When the bridge was loaded with two trucks, the exterior beam (Beam 1) had the maximum strain of 28 µε. The exterior beam (Beam 1)

107 107 also had the largest strain of 17 µε when the bridge was loaded with two trucks back to back. The largest transverse strain in Sollars Road Bridge from the field was 28 µε, for load configuration 3. This magnitude of strain was within the range of strains (20-40 µε) found by Yuan and Graybeal (2016). The top transverse tensile strain was measured close to the connection for full depth shear key filled with UHPC between two adjacent box beams tested under a concentrated load in the laboratory (Yuan and Graybeal 2016). However, the maximum transverse strain from the field measurement of the bridge was lower than the maximum value measured by Yuan and Graybeal (2016). The measured strain of the bridge was at the center of the top flange and the gauge was located at distance 3.5 in. down from the top. The top exterior strain near to the connection was anticipated to be higher. Furthermore, the magnitude and location of the applied load between two studies were slightly different. Table 4.2 Top flange transverse strains in beams 1-3 Time Load Configuration Beam 1 (µε) Beam 2 (µε) 8/8/ :30 One Truck on Left lane (1) /8/ :45 One Truck on Right lane (2) /8/ :00 Two Trucks on Mid Span (3) /8/ :15 Two Trucks back to back on Left lane (4) Beam 3 (µε) The strains from the strain gauges embedded in the top and bottom flanges in a longitudinal direction were calculated by using the equations (4-3 and 4-4). The results are shown in Tables 4.3 and 4.4. Note that N/A means that the gauges were not connected due to data acquisition sensor capacity. The results show that the top flange exhibited compressive strain and the bottom flange had tensile strain. The maximum compressive

108 108 and tensile strains under truck load only were -79 µε and 77 µε, respectively. The recorded strains were from load configuration 4, when the bridge was loaded with two trucks back to back. Table 4.3 Top and bottom flange longitudinal strains in beams 1-3 at mid span Time Load Configuration Gauge Position Beam 1 (µε) Beam 2 (µε) Beam 3 (µε) 8/8/ :30 One Truck on Left Top lane (1) Bottom /8/ :45 One Truck on Right Top lane (2) Bottom /8/ :00 Two Trucks on Mid Top Span (3) Bottom /8/ :15 Two Trucks back to Top back on Left lane (4) Bottom Table 4.4 Top and bottom flange longitudinal strains in beams 1-3 at quarter span Time Load Configuration Gauge Position Beam 1 (µε) Beam 2 (µε) Beam 3 (µε) 8/8/ :30 One Truck on Left Top N/A lane (1) Bottom 28 N/A N/A 8/8/ :45 One Truck on Right Top N/A lane (2) Bottom 18 N/A N/A 8/8/ :00 Two Trucks on Mid Top N/A Span (3) Bottom 47 N/A N/A 8/8/ :15 Two Trucks back to Top N/A back on Left lane (4) Bottom 49 N/A N/A N/A: disconnected strain gauges due to data acquisition capacity The strain in the shear keys in the longitudinal direction was calculated by using equations (4-3 & 4-4) because the same type of gauges was used in shear keys in longitudinal direction as those embedded in the beams. In a transverse direction, the strain in the shear keys was measured by using smaller strain gauges in order to be fit in the shear key in the transverse direction. The strain due to the loading was calculated by

109 109 using the following equation (4-5). A nominal batch factor was not used in the equations because it was set in the data acquisition program. ε = (R 1 R O ) Eq. 4-5 Where: R 1 = Gauge reading with loading R 0 = Initial reading prior to loading The strains in the shear keys for both mid and quarter span in longitudinal and transverse directions are shown in Table 4.5 and 4.6. Table 4.5 Longitudinal strains in shear keys 1 and 3 at mid and quarter span Time Load Configuration Shear key 1 (µε) Shear key 3 (µε) Mid Span Quarter Span Mid Span Quarter Span 8/8/ :30 One Truck on Left (1) /8/ :45 One Truck on Right (2) /8/ :00 Two Trucks on Mid Span (3) /8/ :15 Two Trucks on Left (4) Table 4.6 Transverse strains shear keys 1-3 at mid and quarter span Time Load Shear key 1 (µε) Shear key 2 (µε) Shear key 3 (µε) Configuration Mid Span Quarter Span Mid Span Quarter Span Mid Span Quarter Span 8/8/ :30 One Truck on Left (1) 8/8/ :45 One Truck on Right (2) 8/8/ :00 Two Trucks on Mid Span (3) 8/8/ :15 Two Trucks on Left (4)

110 110 The shear key in the longitudinal direction was subjected to bending. Therefore, compressive strain was recorded due to the high bond between the beams and shear keys which allow them to behave together. However, in the transverse direction, the shear keys were subjected to tensile strain. Cylinders with dimensions of 3 in. by 6 in. were prepared during the casting of UHPC shear keys and tested for compressive strength according to ASTM C 39/C 39M 05 (ASTM 2015) in structural lab with exception of the load rate was modified based on the recommendation by Graybeal (2006). The UHPC reached a compressive strength of 22 ksi at 28 days. In order to calculate the allowable tensile strain, the modulus of elasticity was calculated to be 7268 ksi by using Equation 4-6, (Russell and Graybeal 2013). E c =49000 f c Eq.4-6 The allowable tensile strength in the shear key was calculated to be 0.99 ksi by using Equation 4-7, (Russell and Graybeal 2013). f ct =6.7 f c Eq. 4-7 The cracking strain was calculated by dividing the allowable tensile strength over the modulus of elasticity and was found to be 137 µε. The recorded strains in the shear key were less than the cracking strain and hence were not high enough to cause cracks. The results show the ability of the shear key with UHPC grout to transfer the truck load between adjacent box beams. The recorded strain refers to fact that if the cracks happen in the shear keys, they are not related to the live load but to other factors such as temperature gradients or shrinkage. In the transverse direction, the higher strain was recorded when the bridge was loaded with two trucks back to back.

111 Strain in the Dowel Bars The axial strain in the dowel bars was measured with strain gauges. The strain due to the loading was calculated by using equation (4-5). The axial strains in the instrumented dowel bars at mid and quarter spans for the portion embedded in the shear keys and the portion embedded in the box beams are shown in Table 4.7 and 4.8, respectively. The portion of the dowel bar embedded in the beams exhibited higher strain than the portion embedded in the shear keys. The results also showed that the strains in the dowel bars embedded in the shear keys were consistent with the transverse measured strains in the keys. Furthermore, the strains in the dowel for the part embedded in the beams was consistent with the transverse measured strains in the beams. The recorded strains in in the dowel bars in both portions were relatively low. The dowel bars may exhibit flexural/shear strains in addition to axial strain but the strains due to flexural/shear were not captured because the dowel was instrumented using one strain gauge only. The low recorded strains may also imply that no cracks existed in the shear key. Theoretically, the dowel bar will exhibit high stresses and strains when large differential displacement developed between the beams. This differential displacement would make the joints open/close in transverse direction, which would result in high stresses and strains developing in the dowel bars. The dowel bars will try to prevent the joint movement by applying compressive stresses equal and opposite to the tensile stresses in a transverse direction according to the shear friction concept.

112 Table 4.7 Axial strains in dowel bars embedded in shear keys 1-3 at mid and quarter span Time Load Configuration Dowel in Shear Key1 (µε) Dowel in Shear Key 2 (µε) Dowel in Shear Key 3 (µε) Mid Span Quarter Span Mid Span Quarter Span Mid Span Quarter Span 8/8/ :30 One Truck on Left (1) 8/8/ :45 One Truck on Right (2) 8/8/ :00 Two Trucks on Mid Span (3) 8/8/ :15 Two Trucks on Left (4) Table 4.8 Axial strains in dowel bars embedded in beams 1-3 at mid and quarter span Time Load Configuration Dowel in Beam 1(µε) Dowel in Beam 2(µε) Dowel in Beam 3(µε) Mid Span Quarter Span Mid Span Mid Span 8/8/ :30 One Truck on Left (1) /8/ :45 One Truck on Right (2) /8/ :00 Two Trucks on Mid Span (3) /8/ :15 Two Trucks on Left (4) Live Load Moment Distribution Factors The live load moment distribution factor (LLMDF) refers to the portion of the total live load moment that distributes between beams. The live load moment distribution factor was calculated by dividing the strain, which was determined using the bottom exterior gauges at mid span from each beam, by the total strain recorded for the bridge as shown in Equation 4-8. This was done for each load configuration. This method is acceptable if all the beams are assumed to have the same stiffness. LLMDF i = ε i n i=1 ε i Eq. 4-8 Where: LLMDF i = Live load moment distributed factor of ith beam

113 113 ε i = Strain of ith beam n i=1 = Summation of all beam strains ε i n = Number of beams The results are shown in Table 4.9. The loaded beam and adjacent beam have higher live load moment distribution factors in all loading conditions. However, the highest live load moment distribution factor occurred when the bridge was loaded with one truck on the right side. When the bridge was loaded with one or two trucks on the left side, the live load moment distribution factors were approximately the same as shown in Figure The results show that Beam 7 exhibited the highest live load moment distribution factor for all loading configurations. The live load moment distribution factors were calculated using strain data. The lower stiffness of Beam 7 led to higher strain which caused the higher live load moment distribution factor. Table 4.9 Live load moment distribution factors Load Configuration Beam 1 Beam 2 Beam 3 Beam 4 Beam 5 Beam 6 Beam 7 One Truck on Left (1) One Truck on Right (2) Two Trucks on Mid Span (3) Two Trucks back to back on Left (4)

114 114 Figure 4.11 Live load moment distribution factors versus width of the bridge (adopted from Semendary et al.2017a) The live load moment distribution factor was compared with American Association of Highway and Transportation Officials (AASHTO) Load and Resistance Factor Design (LRFD) values (2012, 2016). The AASHTO LRFD equations for the calculation of the live load moment distribution factor does not have a case when the shear key has dowel bars in the joints. Two cases (f and g) were found in AASHTO LRFD (2012, 2016) and the classification was based on the utilizing deck or not as well as on the exiting of the post-tensioning as shown in Table Table 4.10 Common deck superstructures covered in articles and (adopted from AASHTO LRFD 2016) Supporting Components Type of Deck Typical Cross-Section Precast Solid, Voided or Cellular Concrete Boxes with Shear Keys Cast-in-place concrete overlay Precast Solid, Voided, or Cellular Concrete Box with Shear Keys and with or without Transverse Post- Tensioning Integral concrete

115 115 The beams act as monolithic unit if sufficiently interconnected. A fully interconnected joint is identified as a flexural shear joint. This type of interconnectivity is enhanced by either transverse post-tensioning of the intensity 0.25 ksi or by a reinforced structural overlay, or both. The use of transverse mild steel rods secured by nuts or similar unstressed dowels should not be considered sufficient to achieve full transverse flexural continuity unless demonstrated by testing or experience. If the intensity of 0.25 ksi for post-tensioning is achieved, it is thought to be more effective than a structural overlay (AASHTO LRFD 2016). According to the AASHTO LRFD sections b-1 and d-1, the live load moment distribution factor of case (g), by assuming that the beams are sufficiently connected to act as a unit. In this case, the live load moment distribution factor is the same as cross section (f) (see Appendix A). For the interior beam, the live load moment distribution factors were and for the one design lane and two design lanes, respectively. For the exterior beam, the live load moment distribution factors were and for the one design lane and two design lanes, respectively. However, if the beams are connected only enough to prevent relative vertical displacement at the interface (Case g), the live-load moment distribution factor can be calculated using different equations (see Appendix B) based on AASHTO LRFD (2012). However, theses equations are no longer used in AASHTO LRFD (2016) for adjacent box beam bridges but the equations were used for comparison with AASHTO LRFD (2012). The load moment distribution factor for interior beam was regardless of number of loaded lanes. For the exterior beam, live load moment distribution factors

116 116 were and for the one design lane and two design lanes, respectively. The results for load distribution factors for two cases is shown in Table Table 4.11 AASHTO LRFD moment-distribution factors of live loads per lane Case Type of beam LLDF (one lane) LLDF (two lanes) Case (f) Interior Exterior Case (g) Interior Exterior The live load moment distribution factors calculated using the assumption that assumed the beams are connected only enough to prevent relative vertical displacement at the interface are larger than the load distribution factors calculated using the assumption that assumed the beams are sufficiently connected to act as a unit. This was expected because Case (f) is for a system that distributes the loading better than Case (g). The values calculated using the two cases were compared with field results and were found to be conservative. The new shear key configuration with dowel bars provided a sufficient connectivity to make the bridge act as a unit. The comparison between AASHTO LRFD and field results are shown in Figures 4.12 and 4.13.

117 117 Figure 4.12 Live load moment distribution factors versus width of the bridge (1 design lane loaded) (adopted from Semendary et al.2017a) Figure 4.13 Live load moment distribution factors versus width of the bridge (2 design lanes loaded) (adopted from Semendary et al.2017a)

118 Dynamic Truck Load Test Results Deflection and Strain of the Bridge A moving truck test was used to monitor the response of the bridge and the shear key design subjected to a moving load. After finishing the static load test, one truck was driven on the left lane of the bridge at speeds of 5, 10, 15, 25, and 30 mph. Strain and deflection data were collected from strain gauges glued to the bottom surface of each beam and from the LVDT s, respectively. Figures 4.14 and 4.15 show the strain and deflection data versus time in each beam for the truck driven at a speed of 5 mph. The results show that Beams 1-3 had higher values of strain because the truck was on the left side and directly loaded these beams. However, Beam 7 recorded a higher strain than Beams 4-6 as previously observed under static loading. It was also noticed that Beams 1-3 peaked after Beams 4-7. For deflection, loaded Beams 1-3 have a higher deflection than the other beams because the truck passed directly over these beams. However, beam 7 recorded a lowest deflection compared with other beams. All beams peaked at the same time.

119 119 Figure 4.14 Mid span bottom strain versus time for truck speed at 5 mph (adopted from Semenderay et al. 2017c) Figure 4.15 Mid span bottom deflection versus time for truck speed at 5 mph (adopted from Semenderay et al. 2017c) The peak values of the strain and deflection for each beam is shown in Figures 4.16 and 4.17, respectively. Beams 1-3 have highest strains and Beams 1 and 3 have highest deflection.

120 120 Figure 4.16 Peak strain in bottom of each beam for truck speed at 5 mph Figure 4.17 Peak deflection in bottom of each beam for truck speed at 5 mph Same behavior with slightly difference in magnitude was observed when the truck was driven over the bridge in the left lane at a speed of 10, 15, 25 and 30 mph and therefore the results are not shown (see Appendix C). Peak strains and peak deflections were compared for different speeds as shown in Figure 4.18 and 4.19, respectively. The

121 121 results show that the highest strain and deflection was recorded for a truck speed of 5 mph in Beam 1. Figure 4.18 Peak mid span strain in the bottom of each beam for different truck speeds (adopted from Semenderay et al c) Figure 4.19 Peak mid span deflection in the bottom of each beam for different truck speeds (adopted from Semenderay et al. 2017c)

122 122 Static and dynamic peak strain truck load response was compared as shown in Figure The static load was taken when the bridge was loaded with one truck on the left. The results show that the loaded Beam 1 had the same strain for the static loading and dynamic loading at a speed of 5 mph when the truck was in the same lane. However, Beams 3-7 have dynamic peak strains slightly greater than the static strains which indicates an increase in the response due to the moving load. Figure 4.20 Mid span strain in the bottom of each beam due to static load and dynamic load for different truck speeds (adopted from Semenderay et al. 2017c) Static and dynamic truck load deflection responses were compared as shown in Figure The static load was taken when the bridge was loaded with one truck on the left. The results show that static deflection was lower than dynamic deflection when the truck was in the same lane. This indicates an increase in the deflection due to the moving load.

123 123 Figure 4.21 Mid span deflection in the bottom of each beam due to static load and dynamic load for different truck speeds (adopted from Semenderay et al. 2017c) Dynamic Load Allowance The dynamic load allowance is an increment to be applied to the static wheel load to account for wheel load impact from moving vehicles (AASHTO LRFD 2016). The moving vehicle on the bridge generates an increase in deflection and strains compared with static vehicle on the bridge. The dynamic load allowance, which is also known as dynamic amplification factor (DAF) is used to account for the irregularity of the deck surface, the deflection and stress in the bridge in both static and dynamic, and the load increasing due to vehicle /bridge interaction. The current AASHTO LRFD (2016) design value is 1.33, which means an increase in the response of the structure by 33%. The dynamic amplification factor was calculated for both strain and deflection. The dynamic amplification (DA) was calculated using the following equations, (Phares et al, 2009). For strain: DA = ε dyn ε stat ε stat Eq. 4-17

124 124 ε dyn = ε stat (1 + DA) For deflection: DA = dyn stat stat Eq dyn = stat (1 + DA) where: ε dyn = Peak dynamic strain when one truck was driven on the left lane of the bridge at speeds of 5, 10, 15, 25, and 30 mph ε stat = Static strain when the bridge was loaded with one truck on the left lane dyn = Peak dynamic deflection when one truck was driven on the left lane of the bridge at speeds of 5, 10, 15, 25, and 30 mph stat = Static deflection when the bridge was loaded with one truck on the left lane The dynamic amplification factor (DAF) = 1 + DA The peak strain and deflection for each beam from dynamic test when one truck was driven in the left lane of the bridge at speeds of 5, 10, 15, 25, and 30 mph and the static strain and deflection when the bridge was loaded with one truck on left lane one truck were used in the calculation of the dynamic amplification factor (DAF). The results are shown in Table 4.12 and The N/A means that the deflection was unavailable from field test for Beams 1and 4 for the same reasons previously discussed and therefore, the dynamic amplification factor for deflection was ignored for Beam 1 and 4.

125 125 Table 4.12 Dynamic amplification factor (DAF) from strain data (adopted from Semenderay et al. 2017c) Truck DAF Speeds, Beam 1 Beam 2 Beam 3 Beam 4 Beam 5 Beam 6 Beam 7 mph Table 4.13 Dynamic amplification factor (DAF) from deflection data (adopted from Semenderay et al. 2017c) Truck DAF Speeds, mph Beam 1 Beam 2 Beam 3 Beam 4 Beam 5 Beam 6 Beam 7 5 N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A The results show that Beam 7 has the maximum dynamic amplification factor of 1.17 based on strain. The value was recorded when the truck was driven with the speed of 15 mph. The results also show that Beam 5 has the maximum dynamic amplification factor of 1.33 for deflection. The value was recorded when the truck was driven at a speed of 15 mph. The results show that the dynamic amplification factor for strain was lower than the factor from AASHTO LRFD (2016) which is The results show that the dynamic amplification factor for deflection was the same the factor from AASHTO LRFD (2016). The results for DAF for Beam 7 for strain and Beam 5 for deflection are plotted in Figures 4.22 and 4.23 for different truck speeds, respectively. For both strain and deflection, the maximum dynamic amplification factors were observed when the truck speed was 15 mph.

126 126 The dynamic amplification factor from deflection was higher than the dynamic amplification factor from strain. The variability in the dynamic amplification factors between beams based on deflection data was higher than that from the strain data. This could be related to the fact that the measurement of the deflection was not as accurate as strain measurement. The strain was measured using strain gauges attached to the concrete. However, the deflection was measured using LVDTs installed on the farm, which may have also exhibited some movement during the moving load. Furthermore, the load configuration changes from a static concentrated load to approximately a distributed load due to the inertia of the effective bridge mass in case of the moving loading. This effect has been noted for bridges with span lengths more than 82 ft (Ludescher and Brühwiler, 2009). However, the mass of the adjacent box beam bridge is large relative to the truck load and the effect is also applicable for this bridge type at shorter spans. A simple calculation based on both deflection and strain for simply supported bridge under uniform and concentrated loads showed that the increase in the dynamic amplification factor based on deflection was higher than when it is based on the strain.

127 127 Figure 4.22 Dynamic amplification factor (DAF) from strain data of beam 7 versus truck speeds Figure 4.23 Dynamic amplification factor (DAF) from deflection data of beam 5 versus truck speeds The dynamic amplification factors for all beams at different truck speeds are shown in Figure 4.24 and 4.25 for both strain and deflection data, respectively. At truck speed of 30 mph, the lowest dynamic amplification factors were observed for all beams. The highest dynamic amplification factor for Beam 1, 5, 6 and 7 were at a truck speed of 15 mph. However, the highest dynamic amplification factor for Beam 2 occurred at a

128 128 truck speed of 5 mph and for Beam 3 at 25 mph. Beam 4 shows approximately the same dynamic amplification factors for truck speeds 5, 15 and 25 mph. For deflection data, the highest dynamic amplification factors occurred at a truck speed of 15 mph. Figure 4.24 Dynamic amplification factor (DAF) for strain data of beams 1-7 versus bridge width Figure 4.25 Dynamic amplification factor (DAF) for deflection data of beams 1-7 versus bridge width

129 129 CHAPTER 5: BEHAVIOR OF AN ADJACENT PRECAST PRESTRESSED CONCRETE BOX BEAM BRIDGE UNDER TEMPERATURE LOAD Data was collected from July 17-25, 2014 immediately after casting the UHPC joints. Data was also collected from August 8-16, 2014 after finishing truck load testing. Data was also collected in December 2014 and January In addition, data was collected from July 9-14, According to the strain gauges manual, the strain gauge s reading should be corrected for temperature when the gauge is attached to or embedded in the concrete. The difference in the coefficient of thermal expansions between concrete and gauge s steel should be considered. If the concrete expanded by exactly the same amount as the wire, then the wire tension would have remained constant and no correction would have been necessary. However, the coefficient of expansion of steel C 1 is 12.2 µ/ C whereas the coefficient of expansion of concrete C 2 is approximately 10.4 µ/ C so that a correction for temperature is required. The strains from longitudinal gauges which were embedded either in concrete or UHPC as well as from transverse gauges, which were embedded in the concrete, were calculated using Equation 5-1. ε = f 2 Eq. 5-1 Where: ε = Strain in microstrain f=frequency in Hz The actual strain was calculated by Equation 5-2 ε = B (R 1 R O ) + (T 1 T 0 ) (C 1 C 2 ) Eq. 5-2

130 130 Where: B=Nominal batch factor = 0.97 R 0 = Initial reading R 1 =Gauge reading T 0 = Initial temperature in C T 1 = Recorded temperature in C C 1 = The coefficient of expansion of steel (12.2 µ/ C), (Gauge design manual) C 2 = The coefficient of expansion of concrete (10.4 µ/ C), (Gauge design manual) C 2 = The coefficient of expansion of UHPC (15 µ/ C), (Russell and Graybeal 2013) The strains from transverse gauges, which were embedded in UHPC only, were calculated with Equation 5-3. ε = (R 1 R O ) + (T 1 T 0 ) (C 1 C 2 ) Eq. 5-3 Where: R 0 = Initial reading R 1 =Gauge reading, T 0 = Initial temperature in C T 1 = Recorded temperature in C C 1 = The coefficient of expansion of steel (12.2 µ/ C), (Gauge design manual) C 2 = The coefficient of expansion of UHPC (15µ/ C), (Russell and Graybeal 2013) The strains from gauges, which were attached on the dowel bars, were calculated without temperature correction because the gauge and dowel bar were assumed to have the same coefficient of thermal expansion.

131 131 ε = (R 1 R O ) Eq. 5-4 The temperature was recorded using thermistor for each strain gauge. The temperature for each strain gauge was used in the temperature correction. 5.1 Behavior after UHPC Placed in Joints from July 17-25, 2014 This data monitoring occurred immediately after the UHPC longitudinal joints were cast and prior to any asphalt overlay being placed on the bridge Beam Longitudinal Strain Results The strains near the top and bottom measured by the embedded strain gauges in the longitudinal direction for Beams 1 and 3 are shown in Figures 5.1 and 5.2, respectively. These strains were at the mid-span of the bridge and though the strains were measured at locations within the beam they are referred to as top and bottom strains. Measurements were also taken at the quarter span of the bridge but showed smaller magnitude values and little difference in behavior. Therefore, quarter span results were not included. Data was also collected from instrumentation installed in Beam 2. Beam 2 results are shown since they were typical of the data collected from Beam 1, 2 and 3. Figure 5.1 shows the results of the longitudinal strains and temperatures in Beam 1 at mid-span during the period data was collected. The top of Beam 1 exhibited hightensile strain on the day after casting, which was about 45 µε and represented the highest tensile strain during the first week readings of the top flange at mid span. The maximum compressive strain recorded during this week after joint casting for Beam 1 was -75 µε in the top flange when the temperature reached approximately 115 o F. The temperatures in the top flange ranged from approximately 65 o F to 115 o F. The results show that both

132 132 tension and compression strains exist from the temperature gradients and inversely follow the temperatures recorded in the top flange. This means that as the top of the beam increases in temperature it attempts to expand. The expansion is restrained from occurring due to abutments positioned at ends and this generates compressive strain. The opposite occurs as the beam cools. Figure 5.1 also provides the bottom longitudinal strain and temperature for Beam 1 at mid-span for the period data was collected. The recorded strains in the bottom flange were very small. This was due to bottom temperature range was only about 15 o F. Temperatures were also much lower in the bottom than the top during high peaks but similar at the low temperature peaks. It was also noticed that the strains in the bottom behaved, in general, in the opposite direction compared to the top. The bottom showed peak strains when the temperature peaked. However, there was a delay between the maximum bottom strain and maximum bottom temperature. The change in behavior for the bottom of the beam may be due to the bottom being influenced by the higher strains and temperature changes in the top. The higher strains in the top may have induced bending in the beam which offset the typical behavior noted in the top. For example, a high compressive strain in the top due a high temperature could cause tensile strain in the bottom that offset the compressive strain in the bottom caused by a lower temperature.

133 133 Figure 5.1 Beam 1 longitudinal strains/temperatures versus time at mid-span Figure 5.2 shows the longitudinal strains and temperatures over time for Beam 3 at mid-span. The top flange of Beam 3 exhibited the highest measured compressive strains of all beams monitored. However, the top tensile strain decreased from Beam 1 to Beam 3. The top of Beam 3 behaved in a similar manner as Beam 1 in that both had strain values that were mirror images of the temperature data. Beam 3 also appeared to have a slightly larger time delay from temperature peaks and strain peaks compared to Beam 1. The time delay between temperature peaks and strain peaks was between 15 minutes to 3 and a half hours and it was more pronounced when the temperature decreased. This indicates that the decreasing in temperature requires more time to induce strain compared with the increasing in temperature. This may have been caused by the interior nature of Beam 3 compared to Beam 1 being exterior. Figure 5.2 also shows the bottom longitudinal strains and temperatures over time for Beam 3 at mid-span. The bottom flange exhibited similar behavior to the bottom flange of Beam 1 in that the longitudinal strains and temperature changes were small compared to the top flange.

134 134 Also similar to Beam 1 was that bottom flange longitudinal strain of Beam 3 mimicked the bottom temperature. However instead of a delay in strain response, the peak strains in the bottom appear to occur prior to the peak temperatures. The maximum tensile strain in the bottom flange was 14 µε and it occurred before temperature peaks by 3 hours. This early response may also be the result of the influence of the top temperature which appears to peak prior to the bottom temperature. The top strain peaks were earlier than the bottom strain peaks by about approximately 7 hours. Figure 5.2 Beam 3 longitudinal strains/temperatures versus time at mid span Longitudinal Shear Key Strain Results Figures 5.3 and 5.4 provide the strains and temperatures in the longitudinal direction for Shear Key 3 (between Beams 3 and 4) at mid-span and quarter span, respectively. Shear Key 1 between Beams 1 and 2 was also measured but showed similar behavior and slightly less strains and is not included. The shear key behavior in the

135 135 longitudinal direction was the same as the top longitudinal beam behavior where increases in temperature produced compressive strains and decreases in temperature produced tensile strains. There was a time delay of approximately 6 hours between the temperatures and the strains as it might take some time for the temperatures to actually generate the strains. In addition, higher shear key tensile strains occurred within the first three days. Some of these tensile strains may have been caused by shrinkage in addition to thermal effects. There were higher tensile strains in the shear key compared to the compressive strains at mid span. However, the strains in shear key at the quarter span were approximately the same magnitude for tensile and compressive behavior. The largest tensile strain was approximately 80 µε and the largest compressive strain was approximately 50 µε. Figure 5.3 Shear key 3 longitudinal strains/temperatures versus time at mid span (adopted from Semendery et al.2017b)

136 136 Figure 5.4 Shear key 3 longitudinal strains/temperatures versus time at quarter span (adopted from Semendery et al.2017b) Beam Transverse Strain Results Figure 5.5 shows the top transverse strain and the temperature in Beam 3 at mid span. Beams 1 and 2 showed similar results and are not shown for brevity. Similar to the longitudinal direction, compressive strains were generated when the temperature increased and tensile strains were generated for temperature dropped. There was also a delay of approximately 7 hours in the strain response from the change in temperatures for the peak tensile strain.

137 137 Figure 5.5 Beam 3 transverse top strains/temperatures versus time at mid span (adopted from Semendery et al.2017b) Shear Keys Transverse Strains Data collection started approximately 4 hours after the placement of the UHPC. The transverse strains and temperatures in the shear key were difficult to fully understand, revealed the complexity of the joint behaviour due to the UHPC curing and gaining strength and causing higher temperatures while ambient temperatures changed and effected beam. Therefore, data was plotted for the first day and then again after the first day assuming the UHPC gained strength. The type of UHPC used in this research was JS1000. This type of UHPC did not gain strength until after 24 hours (Graybeal et al 2013). The results for the first day showed two different behaviors. One behavior is shown in Figure 5.6 for Shear Key 1 between Beams 1 and 2 at mid span. Shear Key 1 has negligible transverse strain as the temperature rises. As the UHPC cures and gains strength, the increasing temperature induces compressive strain into the UHPC. The

138 138 strain continues to increase in compression even as the temperature falls. The strain eventually levels off even though the temperature begins to change. This may be due to the same time lag seen in the transverse direction for the beams. As the temperature begins increasing again, the strain remained relatively constant. This same behavior also occurred in Shear Key 2 at the quarter span and Shear Key 3 at the quarter span and midspan. The other behavior within the first 24 hours can be observed in Figure 5.7 for Shear Key 1 at the quarter span. The initial behavior is the same with increasing compressive strain as the temperature rose. However, after a period of decreasing temperature, the compressive strain begins to reduce and continues to do so until tensile strain occurs. This same behavior also occurred in Shear Key 2 at mid span. These differences in behavior for different locations may be due to differences in how the joints expanded and contracted during UHPC placement and curing relative to the joint opening or closing. The transverse strains and temperatures in Shear Key 1 after re-zeroing the gauge upon 24 hours of curing are shown in Figure 5.8. This same behavior was noted in all other instrumented shear keys after the first 24 hours. The behavior is the same as the transverse behavior of the beams in that the strains are a mirror image of the temperatures. When the temperatures increase, the strains become more compressive and when the temperatures drop, the compressive strains are reduced. A difference in the transverse behavior of the shear keys is the continued increase in the compressive strains over time. This is likely due to the increasing compressive strength of the UHPC.

139 139 Figure 5.6 Shear key 1 transverse strains/temperatures versus time at mid span (first 24 hours) Figure 5.7 Shear key 1 transverse strains/temperatures versus time at quarter span (first 24 hours)

140 140 Figure 5.8 Shear key 1 transverse strains/temperatures versus time at mid span (after 24 hours) (adopted from Semendery et al.2017b) The transverse strains in Shear Keys1-3 at mid span over the entire time period of monitoring is shown in Figure 5.9. As can be seen in the Figure, strains in Shear Keys 1 and 3 become highly compressive and then are effected by the changing temperature but continue to show slightly higher compressive strains as the UHPC gains more strength. Shear Key 2 shows an initial compressive strain but then becomes tensile before becoming compressive once again after approximately 3 days. This behavior is likely caused by the difference in the joints relative to the UHPC curing and strength gain. Similar behavior was noted at the quarter span but Shear Key 1 exhibited the tensile behavior while Shear Keys 2 and 3 showed compressive strains.

141 141 Figure 5.9 Shear keys 1-3 transverse strains versus time at mid span Dowel Bar Strains The dowel bars embedded in Beams 1-3 had the same behavior as the transverse beam behavior. The strain increased when the temperature decreased and the strain decreased when the temperature increased. A slight delay of approximately 4 to 8 hours occurred between peaks of the temperature and the strains. Figure 5.10 shows the strain in the dowel bar embedded in Beam 1 near the mid-span. The data for other instrumented dowel bars embedded in the beams had the same behavior and are not shown. The largest observed tensile strain was 20 µε and occurred in the dowel bar embedded in Beam 1 at mid-span. The largest measured compressive strain was 60 µε and occurred in the dowel bar embedded in Beam 2 at mid-span.

142 142 Figure 5.10 Dowel bar in beam 1 strains/temperatures versus time at mid span (adopted from Semendery et al.2017b) The strains and temperatures of the dowel bar embedded in Shear Key 1 between Beams 1 and 2 at mid-span are shown in Figure Similar behavior was demonstrated in the other instrumented dowel bars embedded in the shear keys at quarter span. However, the dowel bars embedded in Shear Keys 2 and 3 at mid-span did not show any tensile strain. In addition, the behavior was similar to the transverse behavior measured in the UHPC of the shear key. The results showed initially that the strains became tensile as the temperature increased. However, this was during initial strength gain of the UHPC. Once the UHPC gained sufficient strength, the dowel bar began to show results as a mirror image to the temperature (increasing temperature resulted in decreasing strains). There was also a time lag of approximately 10 hours at the peak maximum tensile strain between temperature and strain peaks. Also, the strains became larger compressive as the UHPC gained more strength toward the end of monitoring.

143 143 Figure 5.11 Dowel bar in shear key 1 strains/temperatures versus time at mid span (adopted from Semendery et al.2017b) Discussion The longitudinal strain in the top flange of Beam 1 and in the shear key 1 at mid and quarter spans along with temperatures were compared in Figure The beam s age at the time of data was approximately three months, and it was anticipated that the measured strains are due to environmental effects rather than the curing process and therefore the creep and shrinkage effects in the beams would be minor. On the first day after casting, the beam strain increased/decreased in the opposite direction relative to the temperature. However, the shear key strain increased/decreased with the temperature change. The difference in behavior between the beam and key on the first day after casting may be attributed to the UHPC not having gained sufficient strength. After the first day, the increases/decreases in the shear key and beam strains were consistent, with both changing in the opposite direction with respect to the daily temperature fluctuations. During this time, there were substantial differences in longitudinal strain between the UHPC key and the beam for the first five days after casting, with those differences

144 144 diminishing toward the end of the data collection period. The early age differences in the longitudinal strains will produce shear stresses at the beam-key interface, which can lead to de-bonding if the magnitude of the stresses exceeds the interface shear strength. (a) (b) Figure 5.12 Beam 1 and shear key 1 longitudinal strains/temperatures versus time: (a) mid span (adopted from Semendery et al.2017b), (b) quarter span The cracking strain of the UHPC after 24 hours was calculated and compared to the maximum tensile strain measured in the longitudinal and transverse directions during

145 145 the first week of data collection. The cracking strain was calculated from the tensile strength and modulus of elasticity determined at 24 hours after casting. The tensile strength of the UHPC was determined based on the equation for untreated specimens, and was reported by Russell and Graybeal (2013) to be: f t = 6.7 f ct Eq. 5-5 In Eq. (Eq. 5-5), the compressive strength (f ct ) of the UHPC at 24 hours was determined using the equation from Russell and Graybeal (2013): f ct = f c [1 exp ( ( t )0.6 )]. Eq. 5-6 The modulus of elasticity of the UHPC at 24 hours may be calculated using Eq. (4-6). The compressive strength (f c ) from cylinder test at 28 days was found to be 22 ksi. Substitution of f c into Eqs. (5-6) and (5-7) gives the compressive strength and modulus of elasticity at 24 hours as 2.7 ksi and 2546 ksi, respectively. Substituting the 24 hour compressive strength into Eq. (5-5) gives the 24 tensile strength as 0.35 ksi. The cracking strain at 24 hours was calculated by dividing the tensile strength by the modulus of elasticity, and was found to be 137 µε. From the data collected during the first 24 hours after casting, the largest longitudinal tensile strain in the UHPC key was determined to be 42 µε, and the largest transverse tensile strain was determined to be 40 µε. Both of these values were well within the 24 hour tensile cracking strain of 137 µε. Furthermore, the maximum longitudinal and transverse tensile strains for the duration of the monitoring period were 80 µε and 40 µε, respectively. These values were also well within the 24 hour cracking strain value of 137 µε, which was anticipated to increase with the increase in UHPC strength.

146 146 The strains and temperatures for the dowel bar embedded in Beam 1 and the dowel bar embedded in Shear Key 1 at mid-span were compared as shown in Figure The results showed a difference in behavior during the first day as the UHPC cured and shrank. Afterward the behavior becomes consistent but the dowel bar in the shear key began to achieve higher compressive strains even though temperatures were almost identical. In general, the strains in the part of the dowel bar embedded in the beam were comparatively small and fluctuated between tensile and compressive strain. On the other hand, the strains in the part of the dowel embedded in the shear key were larger in magnitude than the strains in the beam portion. The strain was tensile for the first four days after casting, and then remained compressive for the remaining four days of the data collection period. Figure 5.13 Dowel bar in Beam 1 and shear key 1 strains/temperatures versus time at mid span (adopted from Semendery et al.2017b)

147 August 8-16, 2014 Data was collected again from the strain gauges embedded in the beams and shear keys and on dowel bars during the period of August 8-16, 2014 after the bridge had been open to traffic. In addition to the gauges used to measure the strain in July, seven exterior strain gauges were used to measure the strain on the bottom flange of each beam during this monitoring period. The data acquisition also was zeroed prior to truck testing and environmental monitoring occurred after truck testing was completed. Therefore, initial values may not be zero at the start of environmental monitoring. The strain from the exterior strain gauges (KM- 100B) was calculated by using the following equation from the gauge manual. ε 3 = c ε ε i + (c β γ) t Eq. 5-8 Where: ε 3 = Strain 10-6 c ε = Calibration coefficient ( / ) ε i = Change of measured value from the initial value (at K=2.0) ( 10-6 ) c β = Compensation coefficent ( / C) γ = Thermal coefficient of linear expansion of specimen ( / C) t = Temperature change ( C) The values of c ε and c β were taken from the gauge manual. The value of γ was taken at 10.4 x 10-6 which represented the coefficient of the thermal expansion for concrete. In addition, temperature gradients were measured by using four thermocouples embedded in Beam 3 and distributed through the depth. Two strain gauges KM- 100B were glued to

148 148 the columns of the instrumentation support frame, one on the left and one on the right, to monitor frame behavior. Two thermocouples were used with these gauges to measure the temperature. In addition, a thermocouple on the bottom of the KM- 100B gauge of Beam 4 was used to measure the temperature. This measured temperature was used in Eq. 5-8 for all KM- 100B gauges connected on the bottom of the bridge Beam Longitudinal Strain Results The strains and temperatures for the top and bottom of Beam 1 at mid span are shown in Figure 5.14 for the time period monitored. The results for Beams 2 and 3 are similar and are not shown. The behavior of the Beams during this period was similar as it was prior to placement of the asphalt overlay. The top of the beam had larger strains and temperatures compared to the bottom. The top exhibited compression strains when the temperatures increased and tensile strains when the temperatures dropped. The bottom of the beam also exhibited compressive strains when the temperature increased and tensile strain when the temperature dropped. This means that as the top of the beam increases in temperature it attempts to expand. The expansion is restrained from occurring and this generates compressive strain. The opposite occurs as the beam cools. The bottom of the beam peaks occurs before the temperature peaks. Furthermore, the bottom peaks occur before the top peaks. The bottom strains may also have been influenced more by the top behavior due to the larger magnitude of strains on the top. All strain values on both the top and bottom flanges were lower than the strain recorded in July. The decrease in strain might have occurred because the temperatures recorded in August were lower than in July. The jump in strain values was noticeable for all beams at the same time. These

149 149 jumps in the data may have been related to applied traffic load as data was being collected. Figure 5.14 Beam 1 longitudinal strains/temperatures versus time at mid-span The strain was also measured at the very bottom of beams by using exterior strain gauges glued to the bottom surface of each beam. The results for Beam 1 are shown in Figure The figure shows that the bottom strains are small and the strain increased when the temperature increased as was noted for the internal gauges near the bottom. However, the bottom surface gauges did not show any time lag between temperature peaks and strain peaks as noted for the internal strains measured by the internal gauges.

150 150 Figure 5.15 Beam 1 bottom surface longitudinal strains/temperatures versus time at midspan The strain measured on the exterior surface of the bottom flanges of Beams 1, 2, 3, 5, 6, and 7 in the longitudinal direction at mid span are shown in Figure The results show that the beams generally had the same behavior. The exception is Beam 3 that initially had the opposite behavior of the other beams but then eventually showed the same behavior as the remaining beams. Also the higher tensile strains were typically in the outer beams and the higher compressive strains were typically experienced in the beams toward the interior of the bridge. This is likely due to the heating and cooling condition differences.

151 151 Figure 5.16 Exterior bottom longitudinal strains versus time at mid span Longitudinal Shear Key Strain Results Figure 5.17 provides the measured longitudinal shear key strains and temperatures in Shear Key 1 (between Beams 1 and 2). Shear Key 1 at the quarter span and Shear Key 3 between Beams 3 and 4 were also measured but showed similar behavior. The shear key behavior in the longitudinal direction was similar to the beam behavior. When the temperature increased, the shear key was under compression and when the temperature decreased, the shear key was under tension. This behavior demonstrates possible bond behavior between the beams and shear keys because the bridge behaves as a unit, which generated the same type of strain in both materials. In addition, the recorded strain during this time period was lower than the strains observed in July.

152 152 Figure 5.17 Shear Key 1 longitudinal strains/temperatures versus time at mid span Beam Transverse Strain Results Figure 5.18 shows the top transverse strain in Beam 3 at mid span. Beams 1 and 2 showed similar results and are not shown. Compressive strains were generated when the temperature decreased and tensile strains were generated for temperature increased. However, there was also a delay in the strain response from the change in temperatures. The temperature peaks occurred before strain peaks by approximately two and a half hours.

153 153 Figure 5.18 Beam 3 transverse top strains/temperatures versus time at mid span Shear Key Transverse Strains Figure 5.19 shows the transverse strains and temperatures in Shear Key 2 (between Beams 2 and 3) at the quarter-span. The data for the other shear keys at the quarter and mid-spans showed similar behavior, but with slightly lower magnitude of strains. Therefore, this additional data is not provided. Similar to the longitudinal beam behavior and transverse beam behavior, the strains decreased when the temperature increased and increased when the temperature decreased

154 154 Figure 5.19 Shear Key 2 transverse strains/temperatures versus time at quarter span Dowel Bar Strains The strains and temperatures versus time of the dowel bar embedded in Beam 1 at mid-span are shown in Figure The results show that the strain increased when the temperature decreased which is similar to the beam behavior. The instrumented dowel bars embedded in Beams 2 and 3 had the same behavior as the transverse beam behavior and are not shown. A slight delay occurred between peaks of the temperature and the strains. The largest observed tensile strain was approximately 32 µε and occurred in the dowel bar embedded in Beam 2 at mid-span. The largest measured compressive strain was about 24 µε and occurred in the dowel bar embedded in Beam 1 at the quarter span. The strains and temperatures in Dowel Bar 1 at mid span compared very close to the transverse strains and temperatures in Beam 1 at mid span.

155 155 Figure 5.20 Dowel bar in Beam 1 strains/temperatures versus time at mid span The strains and temperatures of the dowel bar embedded in Shear Key 3 between Beams 3 and 4 at mid-span are shown in Figure Similar behavior was demonstrated in the other instrumented dowel bars embedded in Shear Keys 1 and 2 and at mid-span and quarter span. The results showed low strains and a fairly large time lag between the strain and temperature peaks. The strain increased when the temperature was increasing but then the strains began to decrease with further increase in temperature.

156 156 Figure 5.21 Strains/temperatures versus time for dowel bar embedded in shear key 3 at mid-span The strains and temperatures of the dowel bar embedded in Shear Key 1 and the transverse strain in Shear Key 1 at mid and quarter span were compared in Figures 5.22 and 5.23, respectively. The results showed that though the temperatures were the same, the dowel bar had higher peak strains than the shear key at mid-span. However, the opposite occurred at the quarter span. The peak strain of the shear key came before the peak of strain of the dowel bar at both locations. The strain in the dowel bars increased when the temperature increased, which was opposite to the behavior of the UHPC shear key. However, there was a delay in the strain response from the change in temperatures for the dowel bars. This behavior of the dowel bars indicates when the temperature increased, the UHPC shear key contracted and the dowel tried to reduce this contraction, which generated tensile strain. The opposite behavior occurred when the temperature decreased.

157 157 Figure 5.22 Strains/temperatures in embedded dowel bar and shear key 1 at mid span Figure 5.23 Strains/temperatures in embedded dowel bar and shear key 1 at quarter span 5.3 January 7 - January 10, 2015 Data was collected in December of 2014 and January of 2015 to monitor the behavior of the bridge when the temperatures were low during the winter months. The data was collected during the periods of December 15-17, 2014; December 19-25, 2014; December 30, January 1, 2015; January 7-10, 2015; and January 15-19, The

158 158 data was collected for different periods because the cold weather required frequent replacement of the battery of the data acquisition system. This also caused the data to appear less smooth as the timeframe of the data was shortened. The data from January 7-10, 2015 was chosen for presentation because it represents the coldest period during the winter monitoring Beam Longitudinal Strain Results The longitudinal strains and temperatures in the top and bottom flange of Beam1 at mid are shown in Figure Similar results were obtained at the quarter span and at the mid span for Beams 2 and 3 and therefore are not shown. As expected, the temperatures are much lower during this period. However, the peak high and low temperatures occur in the top flange of the beam and there is a delay in the peak temperatures for the top compared to the bottom. As previously observed, the strains are a mirror image of the temperatures in which a decrease in temperature caused an increase in strain and an increase in temperatures caused a decrease in strain. There was approximately a two and a half hours of the delay between the peak temperatures and strains. However, there was no delay for the bottom flange strain.

159 159 Figure 5.24 Beam 1 longitudinal strains/temperatures versus time at mid-span The strain was also measured at the bottom surface of the beams using exterior strain gauges mounted to the bottom of each beam. The results for Beam 1 are shown in Figure The figure shows that the bottom strains are small and the strain increased when the temperature decreased as was noted for the internal gauges near the bottom. The strain also was larger than the internal strain near bottom flange measured by the internal gauge. The bottom surface gauges did not show any time lag between temperature peaks and strain peaks which was consistent with the results of the internal strains.

160 160 Figure 5.25 Beam 1 bottom surface longitudinal strains/temperatures versus time at midspan The longitudinal strains on the bottom surface of Beams 1-7 at mid span are shown in Figure The beams had the same behavior and the exterior beams (Beams 1 and 7) had the higher tensile strains. This could be related to the locations of the exterior beams, which made them subjected to the temperature more compared with the other beams.

161 161 Figure 5.26 Exterior bottom longitudinal strains versus time at mid span Longitudinal Shear Key Strain Results Figure 5.27 provides the measured longitudinal shear key strains and temperatures in Shear Key 1 between Beams 1 and 2 at mid span. Shear Key 1 at the quarter span and Shear Key 3 between Beams 3 and 4 were also measured but showed similar behavior. The shear key behavior in the longitudinal direction was similar to the beam behavior. When the temperature increased, the shear key was under compression and when the temperature decreased, the shear key was under tension.

162 162 Figure 5.27 Shear key 1 longitudinal strains/temperatures versus time at mid span Beam Transverse Strain Results Figure 5.28 shows the top transverse strain in Beam 3 at mid span. Beams 1 and 2 showed similar results and are not shown. Similar to the longitudinal direction, compressive strains were generated when the temperature increased and tensile strains were generated for decreasing temperatures. There was also a delay in the strain response from the change in temperatures. The temperatures were much lower during this period but the strains showed slightly higher magnitudes compared to the summer data sets.

163 163 Figure 5.28 Beam 3 transverse top strains/temperatures versus time at mid span Shear Key Transverse Strains Figure 5.29 shows the transverse strains and temperatures in Shear Key 1 (between Beams 1 and 2) at the quarter-span. The data for the other shear keys at the quarter and mid-spans showed similar behavior, but with slightly lower magnitude of strains. Therefore, this addition data is not provided. Similar to the longitudinal beam behavior and transverse beam behavior, the strains decreased when the temperature increased and increased when the temperature decreased.

164 164 Figure 5.29 Shear key 1 transverse strains/temperatures versus time at quarter span Dowel Bar Strains The strains and temperatures versus time of the dowel bar embedded in Beam 3 at mid-span are shown in Figure The instrumented dowel bars embedded in Beams 1 and 2 had the same behavior with slightly lower strain magnitudes and are not shown. The results showed the same behavior as the transverse beam behavior where the strain increased when the temperature decreased, and the strain decreased when the temperature increased. A slight delay occurred between peaks of the temperature and the strains.

165 165 Figure 5.30 Dowel bar in beam 3 strains/temperatures versus time at mid span The strains and temperatures of the dowel bar embedded in Shear Key 3 between Beams 3 and 4 at mid-span are shown in Figure Similar behavior was demonstrated in the other instrumented dowel bars embedded in Shear Keys 1 and 2 and at mid-span and quarter span. The results showed very low strains and a fairly large time lag between the strain and temperature peaks. The strain decreased when the temperature was decreasing but then the strains began to increase with further decrease in temperature.

166 166 Figure 5.31 Strains/temperatures versus time for dowel bar embedded in shear key 3 at mid-span Deflections Deflection at the bottom of each beam was measured by using the LVDT s. The LVDT s were mounted to a frame that was supported on steel plates that were placed in the creek. The data was difficult to interpret as the frame could have moved over time of data retrieval as well as possible movement of the LVDT s within the brackets holding them to the frame or the LVDT s sticking from a combination of moisture and low temperatures. The mid span deflections and temperatures measured on the bottom surface of Beam 1 are shown in Figure Positive deflection is downward and negative is upward in the figure. In general, the results show that when the temperature decreased, the beam moved downward and the beam moved upward when the temperature increased. In addition, there was a time lag between the temperature and the bridge deflection. This general trend was also observed in the data for the other beams. However, the data for Beam 6 showed little change and may have been due to the LVDT reaching its limit or not properly moving. Overall deflections were small.

167 167 Figure 5.32 Deflections and bottom surface temperatures versus time for beam 1 at mid span Joint Movement The joint opening at mid span in the transverse direction at the bottom of Joints 4 (between Beams 4 and 5), 5 (between Beams 5 and 6), and 6 (between Beams 6 and 7) was measured using LVDT s. Unfortunately, the LVDT readings on Joint 6 were unreliable and therefore omitted. However, the data from Joints 4 and 5 were very similar and consistent. In addition, the LVDT s were mounted directly onto the beams so data was easier to interpret. The thermocouple readings on the bottom of Beam 4 at mid span as well as the Joint 4 movements are shown in Figure Although the temperature was measured at the bottom of Beam 4, it was very close to the location of the LVDT that was located across Joint 4. A positive movement is closing of the joint and negative is opening of the joint. The results show very small movements but the trend is a decrease in temperature led to a closing of the joint and the joint opened when the temperature increased. There was also a delay of approximately three hours in the

168 168 joint movement response in comparison to the temperature readings. The joint movement may be related to the higher coefficient of thermal expansion of UHPC compared to the concrete of the beams. Figure 5.33 Joint movement versus time across joint Longitudinal Beam Movement The longitudinal movements at the ends of Beam 7 relative to the abutments are shown in Figures 5.34 and The ends of each beam contained a 2 inch diameter holes with a 1 inch diameter dowel. The dowels were the height of the beams and were embedded 12 inches into the abutments. The dowels at the rear abutment were grouted to create resistance to longitudinal movement. The dowel holes at the forward abutment were the filled with joint sealer to allow for movement. Interestingly, the results show that when the temperature decreased, both ends had approximately the same small movements. A positive movement in the figures means the beam expanded and a negative movement means the beam contracted. When the temperature increased, the

169 169 beam expanded and the beam contracted when temperatures decreased. The beam end at the forward abutment had slightly more movement due to the joint sealer used in used in the grouting of the dowel hole. Figure 5.34 Beam 7 longitudinal movement versus time at the rear abutment Figure 5.35 Beam 7 longitudinal movement versus time at the forward abutment

170 Discussion Comparison between Thermal and Static Truck Loads Behavior The behavior of the bridge under temperature load was compared with the behavior of the bridge under static truck load in order to investigate which type of load had more effects on the behavior of UHPC shear key connections. The cracks in beams or in shear keys occurs when the measured tensile strain is higher than the allowable tensile cracking strain. Therefore, the maximum tensile strain under temperature or load will be of interest. The maximum longitudinal tensile strain at the exterior bottom of Beam 7 at mid span was about 40 µε from the winter period. However, the maximum exterior bottom tensile strain was 109 µε on Beam 7 when the bridge was loaded with two trucks on mid span with a total weight of kips. The results showed that the tensile strain from loading was twice as large as the value from temperature. The deflection due to temperature was also compared with that of static truck load. The maximum positive deflection (downward) was 0.11 in. and the maximum negative deflection (upward) was 0.2 in. from temperature. However, the maximum deflection of the bridge due to truck test was found to be on Beam 7 was 0.48 in. when the bridge was loaded with two trucks at mid span with total weight of kips. The deflection from the temperature was lower than the deflection from static truck load. The longitudinal interior strain in the top and bottom flanges of Beams 1-3 from the temperature from the specific period were compared with the longitudinal interior strains from static truck load at mid and quarter span as shown in Table 5.1. The results from the temperature show that the beam exhibited tensile strain when the temperature decreased in both the top and

171 171 bottom flanges. However, the results from the static truck load when the bridge was loaded with two trucks back to back with total weight of kips show that the top flange exhibited compressive strain while the bottom flange exhibited tensile strain at both mid and quarter spans. The strain in the bottom flange due to static truck load was higher than the values due to temperature. The strains due to truck load excluded temperature effects. Table 5.1 Longitudinal strain in top and bottom flange of beams 1-3 at mid and quarter span due to temperature and static truck loads Type of load Location Strain gauge Beam 1 (µε) Beam 2 (µε) Beam 3 (µε) location Temperature Mid Span Top load at Bottom January,8-12:00 PM Quarter Span Top N/A Two trucks back to back on Left lane Bottom 16 N/A N/A Mid Span Top Bottom Quarter Span Top N/A Bottom 49 N/A N/A Note: Negative strain constitutes compression N/A: the gauges were disconnected due to data acquisition capacity The longitudinal strains in the shear keys 1 and 3, due to temperature, were also compared with the longitudinal strains in the shear keys 1 and 3 due to static truck load as shown in Table 2. The maximum compressive strain was observed when the bridge was loaded with two trucks back to back. The results showed that the shear keys exhibited compressive strain due to static truck load. However, tensile strain was observed in the shear key due to temperature load. By comparing the results from Table 5.1 to the results from Table 5.2, it can be shown that the longitudinal tensile strains in the shear keys, due

172 to temperature load, were higher than the longitudinal tensile strain in the top flange of the beams although the temperature was the same. 172 Table 5.2 Longitudinal strain in shear keys 1 and 3 at mid and quarter span due to temperature and static truck loads Type of load Location Shear Key 1 (µε) Shear Key 3 (µε) Temperature load at January,8-12:00 PM Two trucks back to back on Left lane Mid Span Quarter Span Mid Span Quarter Span The transverse strain in the top flange the Beams 1-3 at mid span due to temperature was compared with the strain due to static truck load, as shown in Table 5.3. The maximum transverse tensile strain was observed when the bridge was loaded with two trucks at mid span with total weight of kips. The results show that the bridge exhibited tensile strain at the top of each beam in transverse direction. The transverse tensile strain, from temperature, was slightly higher than the strain from static truck load. Table 5.3 Transverse strain in the top flange of beams 1-3 at mid span due to temperature and static truck loads Type of load Beam 1 (µε) Beam 2 (µε) Beam 3 (µε) Temperature load at January,8-12:00 PM Two trucks on Mid Span The transverse strain in Shear Keys 1-3 at mid span and quarter span due to temperature was compared with the strain due to static truck load as shown in Table 5.4. The maximum transverse tensile strain was observed when the bridge was loaded with two trucks back to back with total weight of kips. The results showed that the shear

173 173 keys exhibited tensile strain from both the temperature and static truck load in transverse direction. The transverse tensile strains from temperature were much higher than the strains from static truck load. Table 5.4 Transverse strain in shear keys 1-3 at mid and quarter span due to temperature and static truck loads Type of load Location Shear Key 1 (µε) Shear Key 2 (µε) Shear Key 3 (µε) Temperature load at January,8-12:00 PM Two trucks back to back on Left lane Mid Span Quarter Span Mid Span Quarter Span The axial strain in the dowel bar for the part embedded in the beam and the part embedded in the shear keys at mid span and quarter span due to temperature were compared with the axial strains due to static truck load as shown in Table 5.5 and 5.6. The maximum axial strain from the truck load was observed when the bridge was loaded with two trucks back to back with total weight of kips. The results show that the dowel exhibited tensile strain from both the temperature and static truck load. The axial tensile strains from temperature were higher than the strains from static truck load for the part embedded in the beams and slightly higher for the part embedded in the shear keys. Table 5.5 Strain in dowel bar embedded in right side of the cross section of beams 1-3 at mid and quarter span due to temperature and static truck loads Type of load Location Dowel Bar 1 (µε) Dowel Bar 2 (µε) Dowel Bar 3 (µε) Temperature load at Mid Span January,8-12:00 PM Quarter Span 26 N/A N/A Two trucks back to Mid Span back on Left lane Quarter Span 18 N/A N/A N/A: the gauges were disconnected due to data acquisition capacity

174 Table 5.6 Strain in the dowel bar embedded shear keys 1-3 at mid and quarter span due to temperature and static truck loads Type of load Location Dowel Bar 1 (µε) Dowel Bar 2 (µε) Dowel Bar 3 (µε) Temperature load at Mid Span January,8-12:00 PM Two trucks back to back on Left lane Quarter Span Mid Span Quarter Span Compare the Measured Strains with the Allowable Code Limits The main objective of the shear keys is to transfer the load between beams in the transverse direction. In order to obtain adequate load transfer, the shear key tensile strength, the interface bond strength, and the beam transverse tensile strength, should be sufficient to resist the applied load which develops due to truck and/or thermal loads. The maximum transverse tensile strain at the center of the top flange under temperature loading was 31 µε in Beam 1, which is equivalent to 0.19 ksi by considering the modules of elasticity of the beams (Eq.4-1) at time of truck test. The maximum transverse tensile strain in the UHPC was 48 µε in Shear Key 1 at quarter span, which is equivalent to 0.35 ksi by considering the modules of elasticity (Eq.4-6) of the shear keys at 28 days. The maximum tensile strain in the dowel bar was 40 µε, which is equivalent to 1.16 ksi by considering the modulus of elasticity of steel (29,000 ksi). The results show that the applied stresses due to temperature and/or truck load were unable to cause any cracks in precast concrete beam or in UHPC shear keys by comparing with allowable cracking limits (Eq.4-2 and Eq. 4-7). The results indicate that the new shear key configuration had adequate capacity to resist both static load that was applied during the test and thermal load that was observed during the environmental monitoring.

175 175 CHAPTER 6: FINITE ELEMENT ANALYSIS AND PARAMETRIC STUDIES A 3D linear elastic finite element model (FEM) of the bridge was developed using Abaqus software to investigate the behavior of the bridge under field truck loading as well as under AASHTO LRFD (2016) design truck loading and to be used later in parametric studies. The FEM was also implemented to verify if the field measured values were within reasonable ranges. The details of the FEM, calibration, validation, discussion, and parametric studies are described in this chapter. In section 6.1, a description of the FEM is documented. This section was divided into several subsections. Subsection provides a description for each component used in the model. Subsection documents the material properties for all of the components, mesh details and element type. Subsection provides details for the different types of interactions used in the model, boundary conditions, and loadings. Section 6.2 explains the calibration of the FEM with the field measurement and it is also divided into several subsections to investigate the effect of boundary conditions and beam-shear key interactions on the results. In section 6.3, the validation of the FEM is discussed in detail. Section 6.4 covers a discussion based on both field and FEM results and is also divided into several subsections. Section 6.5 covers the behavior of the bridge under AASHTO LRFD (2016) design truck loading. Sections 6.6 describes the parametric studies that were completed to investigate the performance of the UHPC shear key connection for bridges with different widths, skews, depths, and lengths.

176 Model Description Geometry The FEM was constructed from different components, which included box beams, diaphragms, shear keys, longitudinal reinforcement, strands and dowel bars. Each component as shown in Figures 6.1 was created in Abaqus using the dimensions that were described in the bridge plans. After drawing each component using Abaqus, the components were assembled to create the whole model as shown in Figure 6.2. (a) (b) (c) (d)

177 177 (e) (f) Figure 6.1 Extrusion of components: (a) beams; (b) diaphragms; (c) shear keys; (d) rebar; (e) strands and (f) dowel bars Figure 6.2 View of the assembled finite element model Material Properties, Mesh Size, and Element Type All materials used in FEM were defined as linear elastic materials because the bridge was anticipated to behave in the linear stage as discussed in Chapter 4 as no cracks were anticipated or observed in beams, shear keys or at beam-shear key interfaces. The compressive strengths of concrete box beam and UHPC were determined from cylinders.

178 178 The beams reached a compressive strength of 11 ksi. The UHPC reached a compressive strength of 22 ksi. The modulus of elasticity and Poisson s ratio were defined for each material. The modulus of elasticity of concrete beam was calculated using Equation 4-1. For the UHPC, the modulus of elasticity was calculated by using the Equation 4-6. The modules of elasticities and Poisson ratio for steel components (longitudinal reinforcement and dowel bars) as well as prestressing strands were taken based on the recommended values in AASHTO LRFD (2016). The materials properties used in the FEM are shown in Table 6.1. Table 6.1 Materials properties (adopted from Semendary et al. 2017a) Part Modulus of Poisson s ratio Elasticity (ksi) Beams and Diaphragms 5, Steel Components 29, Strand 28, UHPC Shear Key 7, Each part was meshed individually: box beams, diaphragms, shear keys, longitudinal reinforcement and strands were seeded to 5 inches in a longitudinal direction. The cross section was meshed individually to obtain a more uniform shape of the elements. The dowel bar was seeded to 2 in. in axial direction of the dowel and the cross section of the dowel was divided in several elements. The mesh details are shown in Figure 6.3.

179 179 Figure 6.3 Finite element model mesh Three dimensional linear brick elements with reduced integration (C3D8R) were used to define every part. The C3D8R element is considered by Abaqus as a three dimensional element with eight nodes and each node has three degrees of freedom. The complete bridge model exceeded 290,000 elements with more than 500,000 nodes Interaction, Boundary Conditions and Loading Different interactions were used in the model. The interaction between the end diaphragms and box beams were modeled as a tie constraint using the surface to surface option in Abaqus as both components were casted monolithically. In a tie constraint, the two surfaces (slave and master) were tied together during the simulation and the translational and rotational motion as well as all other active degrees of freedom would be equal for the two surfaces. However, the matched mesh size is considered an important factor in the tie constraint because any slave nodes that are not satisfying the tolerance criteria will remain unconstraint for the duration of the simulation. This could cause a slave node to freely penetrate in to the master surface (Abaqus user s manual 2006).

180 180 The longitudinal reinforcement/dowel bars /strands were embedded in concrete/uhpc using embedded constraints. This assures the translational degrees of freedom of the nodes of the embedded elements are constrained to the interpolated values of the corresponding degrees of freedom of the host element nodes. The most important constraint in the model was the interface between the UHPC and the box beams as debonding and load transfer are totally dependent on the accuracy of this constraints. Furthermore, the frequently observed failure in box beam bridges was at the beam-shear key interface as discussed in Chapter 2. Different types of constraints were assumed to investigate the most accurate representation of the load transfer across the beam- shear key interface. One approach to model the beam- shear key interfaces was to use tie constraint. The shear keys were modeled as master surfaces due to higher stiffness compared with the beams, which were modeled as slave surfaces. The tie constraint tied the two surfaces together and translational and rotational motion as well as all other active degrees of freedom would be equal for both surfaces. This type of constraint ignored the effect of the cold joint (interface deformation) but assured monolithic behavior. The constraint between box beams and shear keys was also modeled as a surfaceto-surface option in Abaqus. The normal behavior was defined by using either a hard contact relationship or a linear pressure-overclosure relationship in order to investigate the effect of those different options on the results. According to the Abaqus user s manual (2006), penetration of the slave surface into the master surface could be minimized by using hard contact relationship at locations where surfaces were in contact.

181 181 However, no tensile stresses can be transmitted across the interface (in case of opening). The contact pressure is transmitted between the two surfaces, and the overclosure pressure activates when the contact state is closed. This indicates that the slave nodes are in contact with the master surface. This causes the contact nodes to experience only compressive contact forces during the analysis. However, the contact pressure reduces to zero when separation occurs and the two surfaces will not be in contact again until the clearance between them reduces to zero. The hard contact relationship is shown in Figure 6.4a. In linear overclosure relationship, the relationship between contact pressure and clearance at the interface is linear. The contact pressure transmits between the two surfaces when the overclosure, measured in normal direction between the two surfaces, is in contact. This type of contact is able to solve the contact conditions better than the hard contact. The linear pressure overclosure relationship is shown in Figure 6.4b. However, the linear pressure-overclosure relationship could behave as hard contact if a large value of contact stiffness was assigned. Therefore, three difference values of contact stiffness were used (i.e. 1 ksi/in, 10 ksi/in and 100 ksi/in) to investigate the effect of the contact stiffness on minimizing the penetration between two surfaces.

182 182 (a) (b) Figure 6.4 Overclosure pressure relationship: (a) hard contact; (b) linear contact (adopted from Abaqus user s manual 2006) The tangential behavior was defined using a friction coefficient based on a Coulomb friction model. Typically, both shear and normal forces can be transmitted between any two contacted surfaces. The relationship between the two forces is known as friction. The maximum shear stresses that can be carried by two contacted surfaces before sliding is known as critical shear stresses (τ crit ). The critical shear stresses are proportional to the normal contact pressures (p) in such form (τ crit = µp), where µ represents friction coefficient. The friction coefficient depends on contact pressure, slip rate, and temperature at the contact points. Two orthogonal components of shear stresses (τ 1, τ 2 ) and two slip components (δ 1, δ 2 ) developed in three dimensional analysis simulation along the interface. The slip at the interface occurs if the equivalent shear stresses (τ eq = τ τ 2 ) reaches the critical shear stresses (τ crit ). The critical shear stresses (maximum shear stresses) could be assigned using available experimental data. In this case, slipping occurs when the equivalent shear stress exceeded critical shear stresses regardless of the magnitude of equivalent contact pressure. After this state, the

183 two surfaces will start sliding relative to one another. The friction model with limit shear stress is shown in Figure Figure 6.5 Slip regions with limit critical shear stresses for the friction model (adopted from Abaqus user s manual 2006) The tangential behavior was defined using the penalty option in Abaqus. However, the elastic slip may not be zero using the penalty option because it allows some relative displacement (elastic slip) at the interface while the two surfaces should be sticking. The maximum elastic slip will usually be 0.5% of contact element s average length. In order to prevent any elastic slip during the sticking stage and to achieve sticking constrains, Lagrange multipliers should be used. However, this could increase the cost of the analysis and cause convergence issues. Three shear friction coefficients were investigated in the model; 0.1, 0.8 and 1. Although no effect was anticipated as the applied load was not high enough to cause slip. A friction coefficient of 0.1 was chosen because it correlated very well with the experimental results in the study by Chen and Graybeal (2012). The coefficient of friction of 0.8 was used based on PCI (2010) recommendation. The coefficient of friction

184 184 of 1 was chosen based on the recommendation by Russell and Graybeal (2013) and it is also recommended by AASHTO LRFD (2016) for rough surfaces. The boundary conditions were defined in the model as follows. A set of nodes, located at the bottom of the vertical dowel bar used to connect the beam to the abutment, were restrained at both ends of each beam. Three sets of boundary conditions were investigated at the beam ends in the model (pin-pin, pin-roller and one end was modeled as a pin and the other end by using a roller with springs in a longitudinal direction). The spring was set with a linear stiffness of 15 kip/in. This value was chosen because it led to good correlation with the field results. The vertical, transverse, and longitudinal movement were restrained for pin boundary condition. In the roller boundary condition, the roller support was free to move in the longitudinal direction. However, the roller with spring boundary condition, the longitudinal movement was resisted by the spring. The boundary condition locations are shown in Figure 6.6. Figure 6.6 Boundary conditions The magnitude and position of the loading applied to the model was the same as the trucks used in the field test for each load configuration. The tire area was measured in the field, and a uniform load was applied to elements used to represent the tire area in the model. The load was applied as a uniform load on the tire area to avoid stress

185 185 concentration. The distributed load value was found by dividing the load applied at the tire by the tire area. The effect of the prestressing and the self-weight was not included in the model because the field measured strain was only due to applied truck load. This allowed for direct comparison between the field measurements and FEM results. 6.2 FEM Calibration Boundary Conditions Three different boundary conditions were used in the calibration to compare the results with field measurements. Truck load configuration 2 (one truck on right) was used in the calibration by using a surface to surface interaction with contact stiffness of 10 ksi/in and a friction coefficient of 1. The difference between field and finite element results was calculated by using Equation 6-1. % Difference = { Field Results Finite Element Results Field Results } 100% Eq. 6-1 The maximum difference in deflection was 61%, 9% and 4% while for strain it was 71%, 22% and 14% for the pin-pin, pin-roller and pin-roller with spring, respectively. The pin-roller with a longitudinal 15 kip/in spring boundary condition compared well with the field measurement compared to the other conditions. The results indicated that the behavior of bridge was neither the pin-pin nor pin-roller condition but fell in between these two boundary conditions. The mid span deflection and strain results are shown in Figures 6.7a and 6.7b.

186 186 (a) (b) Figure 6.7 Mid span deflection and strain of the bridge for load configuration 2 (1 truck on right) based on different boundary conditions Beam-Shear Key Interaction The effect of the type of constraints between the precast beams and shear keys was studied in order to investigate the best model that represented the most accurate load transfer at the interface. The results showed that shear friction had insignificant effect on the deflection and strain for the whole bridge when a contact stiffness of 10 ksi/in. was used to define the normal behavior. The maximum differences between field and FEM were 4.3%, 4.1%, and 4.2% for deflection and 14.2%, 14.1%, and 14.1% for strain. Therefore, a coefficient of friction of 1 was chosen to model the beam-shear key interface

187 187 because the constructed bridge utilized an exposed aggregate shear key surface preparation which was close to AASHTO LRFD rough surface conditions. Furthermore, this value was recommended by Russell and Graybeal (2013). By using a coefficient of friction of 1, the maximum differences between field and FEM were 17%, 4% and 3% for deflection and 23%, 14% and 17% for strain measurements when contact stiffness values of 1 ksi/in, 10 ksi/in and 100 ksi/in were used. Therefore, the contact stiffness of 10 ksi/in was chosen in the analysis because it correlated better. The hard contact option was compared to the results from using a contact stiffness of 10 kip/in. The maximum differences between field and FEM for hard contact were 4% and 18% for deflection and strain measurements, respectively. Therefore, the contact stiffness of 10 ksi/in correlated better especially for strain with the field measurements. Tie constraints were also used and the results were compared. The tie constrains led to a decrease in the deflection and strain on the loaded portion and increase the deflection and strain for the unloaded portion for the bridge. The maximum difference between field and FEM were 10% and 23% for deflection and strain, respectively. Therefore, a coefficient of friction of 1 and a contact stiffness of 10 ksi/in were used in the validation as well as in the parametric studies because it led to better correlation compared with all other options based on an overall behavior of the bridge as shown in Figure 6.8 a and b.

188 188 (a) (b) Figure 6.8 Deflection and strain at mid span of the bridge for load configuration 2 (1 truck on right) (adopted from Semendary et al. 2017a) The effect of the type of constraints between the precast beams and shear keys on the strains in shear keys and dowel bars was also investigated. The maximum principal tensile strain in shear keys and dowel bars for different types of constraints determined from the FEM are shown in Table 6.2. The increase in the friction coefficient had insignificant effects on the strain in both the shear keys and dowel bars because no slip was anticipated under the load level that was considered. However, under higher load or temperature/shrinkage effects, the slip may be anticipated and the friction coefficient

189 189 effects will play an important role in load transfer. The slight difference in maximum principal tensile strain in shear key and dowel bars between hard contact compared with contact stiffness models was related to the penetration that occurred at the interface. When the contact stiffness of 100 ksi/in used, the difference between hard and contact stiffness models reduced due to the less penetration due to the high contact stiffness. The tie constraints led to an increase in the maximum principal tensile strain in the shear keys about twice the value from surface-to-surface. In tie constrains, the load transferred through the shear key and penetration or elastic slip were not anticipated because the effect of cold joint at the interface was ignored. This also caused a high decrease in the maximum principal tensile strain in the dowel bars. The dowel bars in this case behaved as regular reinforcement embedded in concrete/uhpc and not as shear reinforcement (SR) as shown in Figure 6.9 a and b. It can be seen from the figure that the increase in maximum tensile strain occurred at the beam-shear key interface. Table 6.2 Maximum principal strain in the shear keys and dowel bars for different types of interactions Type of constrains Coefficient of Maximum principal tensile strain (µε) friction Shear keys Dowel bars S-to-S-Contact stiffness S-to-S-Hard contact Tie

190 190 (a) (b) Figure 6.9 Behavior of the dowel bars under different types of constraints: (a) surface to surface; (b) tie 6.3 FEM Validation Mid-Span Deflection After calibration, the FEM was validated by comparing the FEM results with the field measurements from other loading locations. The deflection from the field test, which was recorded using (LVDT s), was compared with FEM results when the bridge was loaded with first load configuration (one truck on left side). The total truck load was 56.1 kips. The largest field deflection of 0.27 in. was in Beam 2. The largest deflection from FEM was 0.34 in Beam 1 because it was directly loaded and it was an exterior beam

191 191 as shown in Figure An agreement was observed between field and FEM results which indicates the validation of the FEM. Figure 6.10 Deflection at mid span compared with FEM of the bridge for load configuration 1 When the bridge was loaded with one truck on the right side, the results from the FEM was compared with the field test as show in Figure This load configuration was already used in calibration and therefore this comparison was not considered a validation. The total truck load was 53.4 kips. The results showed that loaded Beam 7 had the highest deflection of 0.32 in. However, beams not directly loaded still exhibited

192 deflection due to the load transfer mechanism. The maximum deflection from FEM was 0.33 in. on Beam 7. The FEM results correlated well with the field measurements. 192 Figure 6.11 Deflection at mid span compared with FEM of the bridge for load configuration 2 For the third load configuration (two trucks side by side at mid span), the total truck weight was kips. The maximum recorded deflection for this load configuration occurred on Beam 7 and was 0.48 in. as shown in Figure The field results were also compared with the finite element results. The results show that there was a difference between field and finite element especially for first three beams. The reason for that may be related to the beams on the left had higher stiffness. The higher

193 193 stiffness led to less deflection. All the beams were assumed to have the same stiffness in the FEM while the actual stiffness may be different between beams. This could be avoided in the future by testing the compressive strength of each beam in order to assign the actual stiffness for each beam in the model. However, the difference in stiffness may not be related just to the compressive strength as other factors could have an effect such as prestressing forces, and curing conditions. The maximum FEM deflection was in. on Beam 1, which was reasonable because the truck weight was higher on the left side. Figure 6.12 Deflection at mid span compared with FEM of the bridge for load configuration 3

194 194 The measured deflections when the bridge was loaded with two trucks back to back on the left side are shown in Figure The total truck load on left side was kips. The maximum deflection on Beam 2 was 0.38 in. The results from the FEM were compared with the field measurement. There was a difference between field and finite element especially for first three beams for the same reasons discussed earlier. However, the loaded side deflected more than unloaded side as expected. The maximum FEM deflection was 0.58 in. on Beam 1, which was reasonable because both trucks were on the left side. Figure 6.13 Deflection at mid span compared with FEM of the bridge for load configuration 4

195 Mid Span Exterior Bottom Strain Strain measured using exterior gauges on the bottom of each beam during the truck test was compared with the strain from the finite element models for different load configurations. The loaded beams exhibit the higher strains than unloaded beams. Figure 6.14 shows the strain when the bridge was loaded with one truck on the left side with a total weight of 56.1 kip. The maximium recorded strain from the field test was 61 µε in Beam 1. However, the maximium strain from the model was 66 µε in Beam 1. A fairly good agreement was found between the field and finite element results with the largest difference in the strains being 12 µε in Beam 7. Figure 6.14 Mid span strain compared with FEM of the bridge for load configuration 1

196 196 The strain when the bridge was loaded with one truck on the right side with a total weight of 53.4 kip is shown in Figure This load configuration was already used in calibration and therefore this comparison was not considered a validation. Good agreement was found between the field and finite element results with the largest difference in the strains to be 6 µε in Beam 5. The field results show that Beam 7 had the highest strain from both the field and finite element model results because it was an exterior beam and the load was directly on this beam. Beam 7 still has higher field strain than finite element strain which might be due to the difference in stiffness between the model and actual bridge. The maximium recorded strain from the field test was 68 µε on Beam 7. However, the maximium strain on Beam 7 from the model was 62 µε.

197 197 Figure 6.15 Mid span strain compared with FEM of the bridge for load configuration 2 Figure 6.16 shows the strain when the bridge was loaded with two trucks at mid span with a total weight of kip. Beams 1-3 had approximatly the same field strain. However, Beam 4 exhibited the lowest value because it was not directly loaded and the bridge was symmetric about this beam. Beam 7 exhibited the maximium strain from the field test because it was an exterior beam and a portion of the load was directly on this beam. Beam 7 still had higher field strain when compared to finite element strain which might be due to the difference in stiffness between model and actual beam stiffness. The results from the finite element show that the maximium recorded strain was on Beam 1. The maximium strain from the field test was 109 µε on Beam 7. However, the

198 198 maximium strain from the finite element model was 100 µε on Beam 1. The maxiumium strain from the finite element was on Beams 1 and 7 because the load was applied on Beams 1 and 7 directly and these beams were exterior beams. The difference in the strain from the model between Beams 1 and 7 might be related to the weight and position of the trucks. The largest difference between the field and finite element starins was 21 µε in Beam 4. Figure 6.16 Mid span strain compared with FEM of the bridge for load configuration 3

199 199 Figure 6.17 shows the strain when the bridge was loaded with two trucks back to back in the left lane with a total weight of kip. The field results show that Beam 7 had higher strain than Beams 4, 5, and 6. This difference might be due to the difference in stiffness between the beams and/or Beam 7 was an exterior beam and therefore carried more load than the interior beam. The finie element results were different. Beam 7 exhibited lower strain values than the other beams. This difference might be related to the asssumption that all beam have the same stiffness in the modeling. The maximium recorded strain rom the field test was 95 µε on Beam 1. However, the maximium strain on Beam 1 from the model was 97 µε. Agreement was found between the field and finite element model results with largest difference in the strains being 16 µε.

200 200 Figure 6.17 Mid span strain compared with FEM of the bridge for load configuration Transverse and Longitudinal Interior Strain The strains recorded by using the transverse strain gauges, which were embedded in the top flange of Beams 1-3 at mid span, for different load configurations are compared with the results from FEM for validation as shown in Table 6.3. The results show the ability of the finite element model to capture the transverse behavior of the bridge in comparison to the data from the interior strain gauges. The values from finite element were close to values from the field with a maximum difference in the strain of 8 µε. The results from both field test and finite element show that the bridge exhibited a tensile strain in the top flange in the transverse direction even when two trucks back to

201 back was placed on the one side. This observation will be discussed, in details, later in this chapter. 201 Table 6.3 Top flange transverse strains in beams 1-3 Load configuration Result Beam 1 (µε) Beam 2 (µε) Beam 3 (µε) type One truck on left lane (1) Field FEM One truck on right lane (2) Field FEM Two trucks on mid span (3) Field FEM Two trucks back to back on left lane (4) Field FEM The strains from the strain gauges embedded in the top and bottom flanges in a longitudinal direction at both mid and quarter span were compared with the values from FEM. The results show the ability of the finite element model to capture the behavior of the bridge as measured by the interior strain gauges. The values from finite element were close to values from the field, which emphasize the validation of the model as shown in Tables 6.4 and This results further indicate the ability of the FEM to capture the behavior of the bridge.

202 202 Table 6.4 Top and bottom flange longitudinal strains in beams 1-3 at mid span Load configuration Result type Gauge position Beam 1 (µε) Beam 2 (µε) Beam 3 (µε) One truck on left lane (1) Field Top FEM Top Field Bottom FEM Bottom One truck on right lane (2) Field Top FEM Top Field Bottom FEM Bottom Two trucks on mid Span (3) Field Top FEM Top Field Bottom FEM Bottom Two trucks back to back on left lane (4) Note: Negative strain constitutes compression Field Top FEM Top Field Bottom FEM Bottom Table 6.5 Top and bottom flange longitudinal strains in beams 1-3 at quarter span Load configuration Results types Gauge position Beam 1 (µε) Beam 2 (µε) Beam 3 (µε) One truck on left lane (1) Field Top N/A FEM Top Field Bottom 28 N/A N/A FEM Bottom One truck on right lane (2) Field Top N/A FEM Top Field Bottom 18 N/A N/A FEM Bottom Two trucks on mid Span (3) Field Top N/A FEM Top Field Bottom 47 N/A N/A FEM Bottom Two trucks back to back on left lane (4) Note: Negative strain constitutes compression N/A: gauges were disconnected due to data acquisition capacity Field Top N/A FEM Top Field Bottom 49 N/A N/A FEM Bottom

203 Strain in Shear Keys 1-3 in the Longitudinal and Transverse Directions The measured strain in the shear keys in longitudinal and transverse directions from the field test are compared with the finite element results for validation at both mid and quarter spans. The longitudinal strains in Shear Keys 1 and 3 are compared as shown in Table 6.6. Consistent agreement is observed between the field and finite element results. The maximum strain from the field results was observed for Shear Key 3 at mid span when the bridge was loaded with two trucks at mid span. However, the maximum strain from the finite element model was observed for Shear Key 1 at mid span when the bridge was loaded with two trucks back to back in the left lane. Table 6.6 Longitudinal strain in the shear keys 1 and 3 at mid and quarter span (adopted from Steinberg et al.2016) Load Configuration Results Types Shear key 1 (µε) Shear key 3 (µε) One Truck on Left (1) One Truck on Right (2) Two Trucks on Mid Span (3) Two Trucks on Left (4) Mid Span Quarter Span Mid Span Quarter Span Field FEM Field FEM Field FEM Field FEM The shear key was subjected to bending in the longitudinal direction. Therefore, compressive strains were recorded as the beams and shear keys behave monolithically. The maximum longitudinal strain values from the finite element model for different load configurations were higher at locations other than that of the gauges. The results show that the maximum longitudinal compressive strain was 83 µε for load configuration 1 on

204 204 the top of Shear Key 2 at mid span, 77 µε for load configuration 2 on the top of Shear Key 5 at mid span, 122 µε for load configuration 3 on the top of Shear Key 2 at mid span, and 122 µε for load configuration 4 on the top of Shear Key 2 at mid span. In the transverse direction, the strains in the shear keys from the field test are compared with the strain from the FEM for different load configurations as shown in Table 6.7 at both mid and quarter span. Good agreement was found between the field and finite element results with largest difference in the strains being 8 µε. The maximum transverse strain from both field and finite element model results was observed when the bridge was loaded with two trucks back to back in the left lane. Furthermore, the maximum field strain was observed in Shear Key 3 with a value of 12 µε at quarter span. However, the maximum transverse strain from finite element model was observed in Shear Key 1 with a value of 16 µε at mid span. Table 6.7 Transverse strain in shear keys 1-3 at mid and quarter span (adopted from Steinberg et al.2016) Load Results Shear Key 1 (µε) Shear Key 2 (µε) Shear Key 3 (µε) Configuration Types Mid Span Quarter Span Mid Span Quarter Span Mid Span Quarter Span One Truck on Field Left (1) FEM One Truck on Field Right (2) FEM Two Trucks on Field Mid Span (3) Two Trucks on Left (4) FEM Field FEM The maximum transverse tensile strain values from the FEM for different load configurations were higher at locations other than that of the gauges. The results show

205 205 that the maximum transverse tensile strain was 17 µε for load configuration 1 on the top of Shear Key 2 at mid span, 15 µε for load configuration 2 on the top of Shear Key 5 at mid span, 23 µε for load configuration 3 on the top of Shear Key 2 at mid span, and 23 µε for load configuration 4 on the top of Shear Key 2 at mid span Strain in the Dowel Bars for the Portion Embedded in the Top Flange of Beams 1-3 and the Portion Embedded in Shear Keys 1-3 The strains in the dowel bars embedded in the beams and in the shear keys from the field test and the FEM are compared for all load configurations as shown in Tables 6.8 and 6.9. The results were consistent between the field and finite element data. However, the slight difference in the results may be related to the elastic slip that occurred at the interface due to the Abaqus assumption. There was not much strain recorded in the dowel bars. The maximum recorded strain for the part embedded in Beam 1 at mid span from the field test was 23 µε when the bridge was loaded with two trucks back to back. The maximum recorded strain for the part embedded in Beam 1 at mid span from the FEM was 21 µε under same load configuration. The maximum recorded strain for the part embedded in Shear Key 3 at mid span from the field test was 15 µε when the bridge was loaded with two trucks back to back. The maximum recorded strain for the part embedded in Shear Key 1 at mid span from the FEM was 15 µε for the same load configuration. The dowel bar embedded in the beams exhibited higher strain than the part embedded in the shear keys.

206 Table 6.8 Axial strain in the dowel bars embedded in beams 1-3 at mid and quarter span Load Configuration Results Types Dowel in Beam 1 (µε) Dowel in Beam 2 (µε) Dowel in Beam 3 (µε) Mid Span Quarter Span Mid Span Quarter Span Mid Span Quarter Span One Truck on Left (1) One Truck on Right (2) Two Trucks on Mid Span (3) Field N/A 10 N/A FEM Field N/A 4 N/A FEM Field N/A 11 N/A FEM Two Trucks on Left (4) Field N/A 13 N/A FEM Table 6.9 Axial strain in the dowel bars embedded in shear keys 1-3 at mid and quarter span (adopted from Steinberg et al.2016) Load Configuration Results Types Dowel in Shear Key 1 (µε) Dowel in Shear Key 2 (µε) Dowel in Shear Key 3 (µε) Mid Span Quarter Span Mid Span Quarter Span Mid Span Quarter Span One Truck on Left (1) One Truck on Right (2) Two Trucks on Mid Span (3) Two Trucks on Left (4) Field FEM Field FEM Field FEM Field FEM The low recorded strain values are likely due to the fact that no cracks occurred in the shear key to engage the dowel bars to carry stress and/or the strain gauges were not placed at critical locations. The maximum axial tensile strain values from the FEM for different load configurations were higher at locations other than that of the gauges. The results show that the maximum axial tensile strain was 249 µε at mid span for load configuration 1 in the dowel bar embedded in Shear Key 3 which was located between loaded and unloaded side. The results show that the maximum axial tensile strain was

207 µε at mid span for load configuration 2 in the dowel bars embedded in Shear Key 4, which was located between loaded and unloaded side. The lower strain was anticipated as the truck weight in load configuration 2 was lower than that of load configuration1. The results show that the maximum axial tensile strain was 133 µε for load configuration 3 at mid span for the dowel bar embedded in Beam 7. Finally, the results show that the maximum axial tensile strain was 299 µε at mid span for load configuration 4 in the dowel bars embedded in Shear Key 3 which was located between loaded and unloaded side. In the future work with UHPC connections, the dowel bars should be instrumented with gauges at both side of the interface as well as at further distance from the interface to investigate an actual strain accurately. 6.4 Discussion Longitudinal Strains in the Beams The maximium longitudinal bottom tensile strain from the field test was 109 µε on Beam 7 and 105 µε on Beam 1 from the FEM both under load configuration 3. The cracking strain was determined to be 131 µε as expalined in Chapter 4. The measured and FEM strains were under the truck loading and did not include the strain due to the prestressing force, beam self-weight, and the weight of the asphalt wearing surface. The longitudinal bottom tensile strain for both the field and FEM was not high enough to overcome the pre-compressive strains and therefore cracking of the beams was not expected or observed. The bridge exhibited transverse tensile strain in the top based on both field measurement and FEM results. The largest measured transverse tensile strain in the

208 208 bridge from the field was 28 µε for load configuration 3. This magnitude of strain was within the range of transverse tensile strains (20-40 µε) found by Yuan and Graybeal (2016). The transverse tensile strains were measured at the top of the full depth shear key filled with UHPC connected two adjacent box beams and tested under concentrated load in the laboratory. However, the maximum transverse tensile strain from the field for the bridge in this research was at the center of the top flange and at 3.5 in. down from the top. The top exterior strain close to the connection was anticipated to be higher as the beams may rotate about the shear keys. The transverse strains from FEM under load configuration 3 (two truck at mid span) are shown in Figure The maximum transverse tensile strain from the FEM was 30 µε at the edge of Beam 1. This strain is equivalent to the stress 0.18 ksi when utilizing the modules of elasticity of the beam. Figure 6.18 also shows that the top transverse tensile strains are higher near regions close to the shear keys. The majority of the bottom of the bridge was under compression in the transverse direction. The same behavior was observed from the finite element by Huckelbridge and El-Esnawi (1997) who studied the behavior of a partial depth shear key with non-shrink grout under live load. The authors concluded that the high transverse tensile stresses could be the main reason for development of longitudinal cracks. This effect would be critical if the bond at beam-shear key interface was not strong enough to resist these stresses.

209 209 Figure 6.18 Transverse strain from FEM for load configuration 3 The top transverse strain across the bridge width at mid span for load configuration 3 from the FEM analysis is shown in Figure Beams 1 and 7 exhibited slight compressive strains at the center of the exterior surface of the top flange. This compressive strain was related to either load concentration and/or local frame action of the top flange and the two webs. Frame behavior was further implied by the bottom surface of the top flange in the model by showing tension in the transverse direction. Figure 6.19 FEM top transverse strain across bridge width at mid-span for load configuration 3

210 210 The high transverse tensile strain near the beam-shear key interface emphasizes the need for the tensile strength of UHPC, the interface bond strength at the beam-shear key interface, and the tensile strength of the precast beam to be high enough to resist this strain. The strain range close to the connection from the FEM was found to be in the range of 15 to 30 µε. The largest strain of 30 µε was equivalent to a stress of 0.18 ksi by considering the modulus of elasticity of the beams. This stress was investigated under load only and did not consider the effect of thermal changes, shrinkage, and prestressing which might produce additional transverse tensile stresses. The tensile strength of the concrete used in the prestressed box beams was determined to be ksi by ACI (ACI, 2014). The results show that the applied stress of 0.18 ksi from the truck loading was less than the tensile strength of the beams. However, additional stress may exist prior to loading from thermal changes, shrinkage, and prestressing. No cracks were anticipated at the beam and shear key interface as soon as the tensile strength across the interface was as strong as the tensile strength of the beam (Yuan and Graybeal, 2016). UHPC was placed in the shear keys on July 17 and the truck test was on August 8. This resulted in the UHPC having an age of 23 days. De la Varga et al. (2016) determined the bond strength at the interface between exposed aggregate concrete and UHPC under pull off bond testing at an age of 14 days to be ksi. The interfacial bond strength in the Sollars Road Bridge was anticipated to be greater due to the additional age and therefore the transverse tensile stresses due to load would not cause cracks at the interface. De la Varga et al. (2016) also investigated the interface bond strength between non-shrink grout and precast concrete under pull off test at age of

211 days and determined to be ksi. However, this bond strength was measured with exposed aggregate surface and anticipated to be much lower with sandblasting surface which is normally used with non-shrink grout material. Another study by Issa et al. (2003) determined the strength of the key way joint under direct tension test. The average tensile stress was found to be ksi and the failure mode was fracture through the joint and concrete. De la Varga et al. (2016) investigated the effect of surface treatment and type of the grout material under flexural beam test. The grout material exhibited low bond strength regardless of surface preparation. Furthermore, interface bond failure was observed during handling, prior to load application, when sandblasting interface surfaces. The interface bond strength of non-shrink grout using exposed aggregate surface were ksi and ksi at 7 and 28 days, respectively. However, the UHPC exhibited superior performance because the failure was observed in the precast concrete and no cracking in UHPC or at the interface. The grout material with non-shrink grout might be unable to resist the stress of 0.18 ksi and might fail at the interface. However, UHPC has superior bond strength compared with the non-shrink grout material. The test by Yuan and Graybeal (2014b) showed that the cyclic loading was able to propagate a pre-existing crack in the conventional grout connection independent of the level of post-tensioning. The allowable tensile strength (f ct ) of the UHPC was calculated to be 0.99 ksi. The results also show that the applied stress of 0.18 ksi from the truck loading was less than the tensile strength of the UHPC. However, the tensile strength of non-shrink grout material was measured after 14 days and was found to be ksi (De la Varga et al. 2016).

212 Behavior of the Shear Keys The behavior of the UHPC shear keys was investigated under load configuration 4 because the maximum transverse tensile strain was noted under this load condition. The transverse tensile strain and maximum principal tensile strain from FEM are shown in Figure Figure 6.20 Transverse strain and maximum principal tensile strain in the shear key for load configuration 4 (adopted from Steinberg et al.2016) The results show that the maximum transverse tensile strain of 23 µε was on the top of Shear Key 2 at mid span. The left rear tire of the truck, which was on Shear Key 2, might have caused the high tensile strain in the transverse direction at this location. The maximum principal tensile strain of 28 µε was on the end of Shear Key 3. The results also show that the maximum principal tensile stress was 0.19 ksi at the end of Shear Key 3. Since the maximum principal tensile stress from the FEM was lower than the estimated cracking stress of 0.99 ksi, the shear key was not expected to exhibit any cracking. The

213 213 estimated allowable tensile strength in the shear key was about five times greater than the stress in the shear key although the bridge was loaded with 109 kips on one side Behavior of the Dowel Bras Load configuration 4 produced the largest tensile strains in the dowel bars. The maximum axial tensile strain based of the FEM was 299 µε. This corresponds to a maximum tensile stress of 11 ksi. The maximum principal tensile strain was 370 µε for the dowel bar embedded in Shear Key 3, which was located between loaded and unloaded beams. This dowel bar was located in the non-loaded side. The strain corresponds to a maximum principal tensile stress of 12 ksi, which is lower than the yield strength of 60 ksi. The maximum principal tensile stress is shown in Figure The maximum principal tensile stress was located at the interface between the UHPC and precast element. The dowel bars had reserve capacity to resist the applied load with the specified dowel bar spacing. However, the high observed stresses in the dowel bars before interface cracking need more verification. The magnitude of the stresses was high at the interface. Therefore, an experiment test is still needed to identify the magnitude of stress in the dowel bars before cracking. Figure 6.21 Maximum principal tensile stresses in the dowel bar for load configuration 4 (adopted from Steinberg et al.2016)

214 Behavior of the Bridge under AASHTO LRFD Standard Design Truck Load The behavior of the bridge was investigated when the model was loaded with AASHTO LRFD (2016) standard truck. This load is typically used in design and therefore the behavior of the bridge under this load condition is of interest and warranted. The design truck load consists of live load and a dynamic allowance of either 33% or 75% of the truck load. According to the AASHTO LRFD article , the dynamic load allowance should be taken as 75% for deck joints and 33% for all other components. The design truck includes lane load of 0.64 kip/ft uniformly distributed in the longitudinal direction and uniformly distributed over 10 ft width in the transverse direction as specified by AASHTO LRFD (2016) article The design truck consisted of three concentrated loads with total weight of 72 kips distributed on three axels. The dynamic load allowance of 33% was used to investigate the maximum bottom strain in the beams and compared to the allowable tensile strength. This case was also used to investigate the maximum deflection of the bridge at mid span. The dynamic load allowance of 75% was used to investigate the stresses in the UHPC shear keys as well as in the dowel bars. Three load configurations were examined to produce a maximum force in the transverse direction. In load configuration A1 (where A refers to AASHTO), the lane load was placed at the edge of the design lane. The design truck was placed at a distance of 2 ft from the edge of the design lane in transverse direction to produce a maximum force effect according to the AASHTO LRFD (2016) article as shown in Figure 6.22a. This load configuration A1 was anticipated to produce maximum shear load on the

215 215 longitudinal joint as observed in field testing. In load configuration A2, two exterior lanes were loaded in order to produce maximum transverse negative moment in the bridge as shown in Figure 6.22b. In load configuration A3, the two loaded lanes were moved to the mid span in order to produce a maximum positive moment in the transverse direction as shown in Figure 6.22c. In longitudinal direction, the design truck was placed to produce a maximum moment at mid span. For one lane loaded, the truck load was multiplied by the multiple presence factor of 1.2 according to the AASHTO LRFD (2016) article For two lanes loaded, the truck load was multiplied by a multiple presence factor of 1. The load position in the longitudinal direction is shown in Figure 6.23.

216 216 (a) (b) (c) Figure 6.22 Truck load combinations and position in transverse direction: (a) one lane loaded; (b, c) two lane loaded (adopted from Semendary et al. 2017a) Figure 6.23 Truck load position in longitudinal direction The behavior of the bridge with the 33% dynamic load allowance was investigated under load configuration A1. The mid span exterior bottom strain and

217 217 deflection under load configuration A1 are shown in Figures 6.24a and b. The maximum exterior longitudinal bottom strain of 139 µε was observed on Beam 1. A maximum deflection of 0.84 in. was also observed on Beam 1 because this beam was directly loaded and it was an exterior beam. (a) (b) Figure 6.24 Behavior of the bridge at mid span under load configuration A1: (a) strain (b) deflection The exterior bottom strain and deflection at mid span for load configuration A2, which produced a maximum negative moment in transverse direction, are shown in Figures 6.25a and b, respectively. The maximum exterior longitudinal bottom strain,

218 218 which was observed on Beam 7, was 191 µε and the maximum deflection, which was observed on Beam 1, was 1.11 in. Beams 1 and 7 were directly loaded and exterior beams. (a) (b) Figure 6.25 Behavior of the bridge at mid span under load configuration A2: (a) strain (b) deflection The exterior bottom strain and deflection at mid span for load configuration A3, which produced a maximum positive moment in transverse direction, are shown in Figures 6.26 a and b, respectively. The maximum exterior longitudinal bottom strain of 189 µε was observed on Beam 4. The the maximum deflection of 1.11 in. was also

219 observed on Beam 4 because this beam was directly loaded and the bridge was symmetric about it. 219 (a) (b) Figure 6.26 Behavior of the bridge at mid span under load configuration A3: (a) strain (b) deflection The maximum tensile longitudinal bottom strain was observed under load configuration A2. The maximum longitudinal tensile strain was also less than the allowable cracking strain and therefore no cracks were anticipated because the bottom would still be under compression from prestressing effects, self weight and wearing surface as explained in Chapter 4. A summary for the results for both the 33% and 75%

220 220 dynamic load allowance factors are shown in Tables 6.10 and For the 33% dynamic load allowance, which applies to the beams, load configuration A2 led to the highest strain. For the 75% dynamic load allowance, load configuration A1 led to the highest stresses in both shear keys and dowel bars. Table 6.10 Results for 33% dynamic load allowance with different load configurations Load configuration Exterior bottom mid span strain (µε) Mid span deflection (in.) (Beam 1) (Beam 1) (Beam 7) (Beam 1) (Beam 4) (Beam 4) Table 6.11 Results for 75% dynamic load allowance with different load configurations Load Configuration Maximum principal tensile stresss in Maximum principal tensile stresses in dowel bars Transverse strain in the top of the bridge (µε) shear keys (ksi) (ksi) The maximum principal tensile stress in the shear key was ksi. The transverse tensile strain in the top of the bridge was 76 µε which was equivalent to 0.46 ksi by considering the modulus of elasticity of the beams to be 5988 ksi. The maximum principal tensile stress in the dowel bar was ksi. The allowable tensile strength of the UHPC was 0.99 ksi. The allowable tensile strength for beams was found to be ksi. No carcks were anticipated in UHPC shear key or in the beams since the principal stresses were less than the allowable stresses for the AASHTO LRFD loading. The maximum principal tensile stress in the dowel bar of ksi was lower than the yield strength of 60 ksi. The maximum principal tensile stresses in shear keys and dowel bars

221 221 under AASHTO LRFD loading were about twice what observed from the field test when the bridge was loaded with kips of the one side. This indicates that AASHTO LRFD loading is more critical than the truck loading used in the field. 6.6 Parametric Studies Bridge Width Effect on the Behavior of the UHPC Dowel Shear Key Connections The width of the bridge was increased for the same beam cross section (B21x48) and the same length (61 ft). The Sollars Road bridge consisted of seven adjacent box beams which led to total width of 28 ft. The width was increased to 56 ft and the bridge consisted of fourteen adjacent box beams in order to achieve this width. For 76 ft bridge width, the bridge consisted of nineteen adjacent box beams as shown in Figure The same cross section depth was used because the study by Hanna (2008) indicated that the required transverse post-tensioning was higher for the shallow depth for the same bridge width. The author reported that this related to the less transverse stiffness of shallow beam. Therefore, only the shallow section was investigated under different bridge width. Figure 6.27 Bridges with different widths The bridge was loaded with load configuration A1 which was consisted of standard truck and lane load of 0.64 k/ft. This load configuration was chosen because it

222 222 produced maximum stresses in shear keys and dowel bars. The dynamic load allowance of 33% and 75% were used. The 33% dynamic load allowance was used to investigate the longitudinal strain and the deflection in the bottom of the bridge at mid span. The 75% dynamic load allowance was used to investigate the stresses in the shear keys and the dowel bars. The strain and deflection were decreased as the bridge width increased as shown in Table Table 6.12 Results for 33% dynamic load allowance with different bridge width under load configuration A1 Bridge width (ft) Mid exterior mid span bottom strain (µε) Mid span bottom deflection (in.) (Beam 1) (Beam 1) (Beam 1) (Beam 1) (Beam 1) (Beam 1) The 75% dynamic load allowance was used to investigate the stresses in the shear keys and the dowel bars. The results show that the stresses in the shear key were slightly increased as the bridge width increased. However, the stresses in the dowel bar decreased as bridge width increased as shown in Table 6.13 due to the increase in the transverse stiffness. The increase in UHPC stresses may be related to the increase in the stiffness of the non-loaded side. The results consisted with the PCI design chart developed by (Hanna, 2008) which shows that the bridge width has an effect on the required transverse post-tensioning because the required transverse post-tensioning increased as the bridge width increased. The increasing in the required post-tensioning indicates that high tensile stresses developed in shear keys. However, the constructed bridge in this research did not utilize a transverse post-tensioning and the comparison was only used to identify the

223 223 effect of this factor on stresses in shear keys. The increasing in shear key stresses in the current study replicated that more transverse post-tensioning forces is need based on regular design. Table 6.13 Results for 75% dynamic load allowance with different bridge width under load configuration A1 Bridge width (ft) Maximum principal tensile stresss in Maximum principal tensile stresses in dowel bars Transverse strain in the top of the bridge (µε) shear keys (ksi) (ksi) The maximum transverse tensile stress in the shear key was 0.38 ksi. The transverse tensile strain in the top of the bridge was 0.46 ksi. Therefore, no cracks in the shear keys, in beams or at the interface were anticipated based on the results. The maximum principal tensile stress in the dowel bar of was lower than the yield strength Behavior of the UHPC Dowel Shear Key Connections with Different Skew Angles The Sollars Road Bridge had zero skew angle. To study the effect of the skew angle on the behavior of the bridge, the same bridge cross section, width and the length were used with both zero and 30-degree skew angles. The dowel bars were placed perpendicular to the bridge centerline and did not follow the skew angle as shown in Figure In addition, the end diaphragms were skewed with the bridge on both ends. However, in the standard design of Ohio Department of the Transportations (ODOT), the diaphragms are typically skewed in one side only. The bridge was loaded with load configuration A1. Dynamic load allowances of 33% and 75% were used.

224 224 (a) 0 Skew angle (b) 30 Skew angle Figure 6.28 Top view of the bridge with different skew angles For the 33% dynamic load allowance, the results show that the mid span longitudinal strain and deflection were decreased as skew angle increased as shown in Table This was related to the increase in the transverse strain. For the 75% dynamic load allowance, the results show that the maximum principle tensile stresses in the shear keys were higher for the skewed bridge as shown in Table The results also showed that the stresses in the dowel bar were increased for the skewed bridge. The results also consisted with the PCI design chart developed by (Hanna, 2008). Table 6.14 Results for 33% dynamic load allowance with skews under load configuration A1 Skew angle, deg. Mid exterior mid span bottom strain (µε) Mid span bottom deflection (in.) (Beam 1) (Beam 1) (Beam 1) -0.59(Beam 1)

225 Table 6.15 Results for 75% dynamic load allowance with skews under load configuration A1 Skew angle, degree Maximum principal tensile stresss in shear keys (ksi) Maximum principal tensile stresses in dowel bars (ksi) Transverse strain in the top of the bridge (µε) The maximum transverse tensile stress in the shear key was 0.54 ksi. The transverse tensile strain in the top of the bridge was 91 µε which was equivalent to 0.54 ksi. The maximum principle tensile stress in the dowel bar was 24 ksi. No cracks were anticipated in UHPC shear key or in the beams. The interface bond strength at the interface between concrete and UHPC and beam should be greater than 0.54 in order to assure no interfacial cracking. The maximum principal tensile stress in the dowel bar of 24 was lower than the yield strength of 60 ksi Bridge Depth and Length Effect on the Behavior of the UHPC Dowel Shear Key Connections The behavior of the bridge with different depths and span lengths were investigated. Three span to depth ratios were considered (L/D=35, L/D=30, and L/D=26.25). The study by Hanna (2008) indicates the higher span to depth ration requires larger transverse post tensioning force. The bridge cross section was B21-48 and span length was 61 ft. For this cross section and according to the Ohio Department of Transportation (ODOT), the depth of the shear key should be 7 in. Two other cross sections were investigated. Theses cross sections were chosen based on Ohio Department of Transportation (ODOT) design specifications. According to the standard design, cross sections (B33-48) and (B42-48) should have 12 in. shear key depth. In order to

226 226 investigate if the new UHPC shear key connections, which was 7 in. depth, is still efficient for this type of the cross section, two bridge were analyzed. The first bridge had a 28 ft width, 82.5 ft length and utilized the B33-48 cross section. The bridge was design using LEAP Bridge Enterprise software to investigate the number of the strands. The initial and final concrete compressive strengths were assumed to be 6.8 ksi and 8 ksi, respectively. The 0.5 in special diameter low relaxation strands were used in the design. The results showed that 24 strands were enough for this cross section depth, and beam length. The second bridge had a 28 ft width, 105 ft length, and utilized the B42-48 cross section. This cross section was chosen because it was the deepest section used in both ODOT and AASHTO/PCI design specifications for adjacent box beam bridges. The bridge was design using LEAP Bridge Enterprise software to investigate the number of the strands. The initial concrete compressive strength was assumed to be 6.8 ksi and final compressive strength was assumed to be 8 ksi. The 0.5 in diameter special low relation strands were used. The results showed that 28 strands were enough for this cross section depth and beam length. After investigating the design requirements, each bridge was analyzed using the ABAQUS software by loading the model with the load configuration 1A. In addition, the load was placed to maximize the moment at mid span. The considered bridges are shown in Figures 6.29 and 6.30.

227 227 Figure 6.29 Bridge cross section with different depths Figure 6.30 Bridge with different span lengths The results show that the both the mid span bottom longitudinal strain and the mid span deflection were reduced as the beam s depth and length were increased as shown in Table Table 6.16 Results for 33% dynamic load allowance with skews under load configuration A1 Section type Bridge length, (ft) Mid exterior mid span bottom strain (µε) Mid span bottom deflection (in.) B (Beam 1) (Beam 1) B (Beam 1) (Beam 1) B (Beam 1) (Beam 1)

228 228 For 75 % dynamic load allowance, the results showed that the stresses in the shear key and dowel bars were decreased as span to depth ratio decreased as shown in Table However, the transverse tensile strain was increased as the bridge depth and length increased. The results consisted with the PCI design chart developed by (Hanna, 2008). Table 6.17 Results for 75% dynamic load allowance with skews under load configuration A1 Section Bridge type length, (ft) Maximum principal tensile stresss in shear keys (ksi) Maximum principal tensile stresses in dowel bars (ksi) Transverse strain in the top of the bridge (µε) B B B The maximum transverse tensile stress in the shear key was 0.32 ksi. The transverse tensile strain in the top of the bridge was 85 µε which was equivalent to 0.51 ksi. The maximum principle tensile stress in the dowel bars was ksi. The allowable tensile strength of UHPC was 0.99 ksi. The allowable tensile strength for beams was found to be ksi. No carcks were anticipated in UHPC shear key or in the beams. The interfacial bond strength between concrete and UHPC and beam should be greater than 0.51 in order to assure that no cracking would occur at the interface. The maximum principal tensile stress in the dowel bar of was lower than the yield strength of 60 ksi. The results indicate that shallow shear key of 7 in. depth could be used with deeper cross section. The only concern about using UHPC shear key connection will be the bond strength at the interface. As long as good bond was achieved, the box beam bridges with UHPC connections will have no issue.

229 229 CHAPTER 7: SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 7.1 Summary This dissertation documented the field performance of an adjacent precast prestressed concrete box beam bridge utilizing reinforced UHPC shear key connections. Static and moving truck load tests were conducted to examine the performance of the bridge as well as the ability of the shear keys and dowel bars to transfer the truck loads. The behavior of the bridge was also explored directly after casting the UHPC as well as at different time periods in both hot and cold weather conditions. A three dimensional finite element model (FEM) was developed, calibrated and validated with the field measurements. After validation, the model was used to examine the behavior of the bridge along with the behavior of UHPC shear key connections under standard AASHTO LRFD truck loading. The performance of the UHPC shear key connection was also investigated for bridges with different widths, skews, depths, and lengths. Chapter 3 described beam fabrication, bridge construction, instrumentation and testing. The fabrication of the beams with the new shear key shape was not difficult. Wood was used to modify the existing formwork for the key and the application of a concrete retarder was easily brushed onto the wooden forms. Power washing the shear keys upon beam removal from the forms created an exposed aggregate finish that enhanced bond. Proper placement of the beams during bridge construction ensured no conflicts between dowel bars of adjacent beams. Gaps between beams were sealed with a Sprayable expanding insulation to ensure no leakage of UHPC through the shear key. Mixing of UHPC required mixers on site. The joints had to be covered with plywood for

230 230 curing purposes, and the UHPC was placed into chimneys and allowed to flow to completely fill the joints. The new shear key design with UHPC did not contain transverse ties and associated diaphragms and/or a reinforced concrete deck. No cracking was observed in the UHPC after plywood removal and prior to waterproofing membrane placement. In Chapter 4, data which was collected from the instrumentation installed in beams, shear keys and dowel bars was analyzed. The bridge was loaded with four static load configurations, and the behavior of the bridge along with the behavior of UHPC shear key connections was examined. The chapter also presented the results from the moving load when one truck was driven on one side at different speeds. The measured data from both static and moving loads were compared with allowable limits and standard codes. Chapter 5 examined the beahior of the bridge at early age after casting UHPC as well as at both hot and cold weather conditions. The data from environmental monitoring can only account for the behavior during the period being examined. The data collected from the instrumentation was analyzed, discussed and compared with allowable limits. In chapter 6, a three dimensional finite element model (FEM) of the bridge was developed to investigate the behavior of the bridge under field truck loading as well as AASHTO LRFD (2016) design truck loading. The FEM was also implemented to verify calibrate with field measured values and to perform a brief parametric study.

231 Conclusions Based on the results of this research, the following conclusions were drawn: The beams that were not loaded even experienced deflection and strain due to the load transfer. This indicated that the bridge behaved monolithically which emphasized the ability of the new shear key design with equal dowel bar spacing to transfer applied load. The tensile strain measured in the bottom of the beams was not high enough to overcome the pre-compressive strain from prestressing force. The transverse tensile strains in beams and in shear keys were also less than the cracking strains. Therefore, cracks in beams or in shear keys were not anticipated from the truck loading. Furthermore, the magnitudes of the strains recorded in the dowel bars were low because the shear key did not crack. The maximum live load moment distribution factors which were calculated based on the strain were found to be for exterior beams and for interior beams. The live load moment distribution factors were also calculated based AASHTO LRFD by assuming that the beams were sufficiently connected to act as a unit. The results indicated that the live load moment distributions determined by AASHTO LRFD design procedures are conservative even when assuming the beams to be sufficiently connected to act as a unit. The maximum dynamic amplification factors were 1.17 based on strain and 1.33 based on deflection data. This was anticipated based on literature and calculations. The dynamic amplification factors were lower than the factor from

232 AASHTO LRFD which is The value of 1.33 may be used conservatively for this type of bridge with UHPC shear key connections. 232 The longitudinal strains in the shear key showed the same behavior as the top longitudinal behavior of the beam. However, this behavior was not observed until 24 hours after casting. The bond at the beam-key interface begins to develop after 24 hours from casting based on the data analyzed from the first week after casting. From the data collected during the first 24 hours after casting, the largest longitudinal tensile strain in the UHPC shear key was determined to be 42 µε, and the largest transverse tensile strain was determined to be 40 µε. Furthermore, the maximum longitudinal and transverse tensile strains for the duration of the monitoring period were 80 µε and 40 µε, respectively. All values were well within the 24 hour tensile cracking strain of 137 µε. The strains in the part of the dowel embedded in the shear key at early age were generally larger in magnitude than the strains in the beam portion. The strain was tensile for the first four days after casting. The reason for the high tensile strain at early age is not completely understood and more investigation is still needed. The behavior observed after the bridge was opened to traffic was the same prior to being open to traffic. Strains were lower after the bridge was opened to traffic due to temperature ranges being lower.

233 233 The transverse results showed higher tensile strains in beams, shear keys and dowel bars in cold weather conditions which indicated that cold temperatures had more effect on the behavior. The maximum transverse tensile strain under cold temperatures was 31 µε, which is equivalent to 0.19 ksi. The maximum transverse tensile strain in the UHPC was 48 µε, which is equivalent to 0.35 ksi. Therefore, the interface bond strength at beam-shear key interface should be greater than 0.35 ksi to prevent the interface failure under cold temperature load. The cold temperature had the major effect on the behavior of UHPC shear key connections by generating a high tensile strain compared with the value observed under loading. The FEM was calibrated with the field data and the results showed the ability of to capture the behavior of the bridge. The maximum difference in deflection was 4% while for strain it was 14% for pin-roller with spring. The results indicated that the behavior of bridge was neither the pin-pin nor pin-roller condition but fell in between these two boundary conditions. The surface to surface constraint using coefficient of friction of 1 and a contact stiffness of 10 ksi/in led to better calibration with the field measurements. The estimated allowable tensile strength in the shear key was about five times greater than the stress in the shear key from the FEM and therefore no cracks were anticipated.

234 234 The strain in dowel bars based on the FEM for different load configurations were higher at locations other than that of the gauges. This indicated that the gauges were not at critical location. The maximum principal tensile strain was 370 µε. This strain corresponds to a maximum principal tensile stress of 12 ksi, which is lower than the yield strength of 60 ksi. Under AASHTO LRFD loading, the maximum principal tensile stress in the shear key was ksi. The transverse tensile stress in the top of the bridge was 0.46 ksi, and the maximum principal tensile stress in the dowel bar was ksi. These stresses were higher than that observed from the field load testing and implied that AASHTO loading is critical and conservative. Based on the parametric studies, the stresses in the shear key slightly increased as the bridge width increased. However, the stresses in the dowel bar decreased as bridge width increased due to the increase in the transverse stiffness. The stresses in the shear key and dowel bars also increased as skew angle increased and decreased as span to depth ratio decreased. However, the transverse tensile strain was increased as the bridge span to depth ratio decreased. The results were consistent with the PCI design chart developed by Hanna (2008). 7.3 Recommendations The new shear key design with dowel bars and UHPC investigated in this research performed well under static, moving as well as temperature loads. The new shear key design with dowel bars also performed well based on FEM. The new UHPC shear key connections with 4 in. dowel bar spacing also performed well when considering

235 235 different bridges widths, skews, depths and lengths. The FEM results indicated that the shallow shear key of 7 in. depth with dowel bars staggered at 4 in. spacing could be used with deeper cross sections then used in the Sollars Bridge. The shear key and dowel bars details for future application in other designs is shown below: (all dimensions in inches) (a) (b) Figure 7.1 Shear key and dowel bars details for future application in other designs: (a) shear key details, (b) dowel bar details The only concern about using UHPC shear key connection will be the bond strength at the interface. As long as good bond was achieved, the box beam bridges with UHPC connections will have no issue. 7.4 Study Limitations and Future Work The beams seem to have different stiffness. This could be verified in the future by testing the compressive strength of each beam in order to fully understand the higher strain that was observed in some beams (even those that were not directly loaded) compared with other beams.

236 236 The compressive, tensile and interface bond strengths of UHPC at different ages needs to be measured in order to accurately determine if the measured strains and corresponding stresses from truck load or temperature is high enough to cause failure. The dynamic amplification factors were not the same between beams based on deflection and strain data. Therefore, the dynamic load allowance factor needs more investigation. Strain gauges near the support should be installed in the shear keys in the future as the maximum principal tensile strain was observed at the end based on the validated FEM. The dowel bars need to be instrumented on both sides of the interface as well as further from the interface to determine the strain in the dowel from both truck and temperature loads. Furthermore, instrumented the dowels located near the support should be considered. The transverse strain in beam as well as shear key should be measured at both top and bottom through the depth. The strain could be used to determine the rotations in beams as well as in the shear keys. The stress in dowel bars under truck load, temperature effects as well as standard truck loads was not high enough to reach the yield. This indicates the possibility of increasing the dowel bar s spacing. However, an experimental study is still needed to determine the magnitude of strain in dowel bars before cracking.

237 237 Based on superior performance of the key in the conducted study, a smaller shear key similar to that used in previous designs filled with UHPC without dowel bars might perform satisfactorily. However, this would require further analytical and experimental study.

238 238 REFERENCES (AASHTO) American Association of State Highway and Transportation Officials. (2012). LRFD Bridge Design Specifications (6th ed). Washington, DC. (AASHTO) American Association of State Highway and Transportation Officials. (2016). LRFD Bridge Design Specifications (8th ed). Washington, DC. ABAQUS [Computer software]. Providence, RI, Dassault Systèmes Simulia Corp. ABAQUS. (2006). User s manual - Version Pawtucket, RI: Abaqus Inc. ACI (American Concrete Institute). (2014). Building code requirements for structural concrete. ACI , Farmington Hills, MI. ASTM C39. (2015). Standard Test Method for Compressive Strength of Cylindrical Concrete Specimens, ASTM International, West Conshohocken, PA. Attanayake, U., and Aktan, H. (2014). First-generation ABC system, evolving design, and half a century of performance: Michigan side-by-side box-beam bridges. Journal of Performance of Constructed Facilities.dio: / (ASCE) CF Chen, L., and Graybeal, B. (2012). Modeling Structural Performance of Second- Chicago, IL. De la Varga, I., Haber, Z.B., and Graybeal, B.A., Performance of Grouted Connection for Prefabricated Bridge Elements Part II: Material-Level Investigation on Shrinkage and Bond. Proceedings of the 2016 PCI National Bridge Conference, Nashville, TN, USA, 2016, 11 pp.

239 239 Dong, X. (2002). Traffic forces and temperature effects on shear key connections for adjacent box girder bridge. Doctoral dissertation. University of Cincinnati. El-R y, A., Tadros, M. k., Yamane, T., and Krause, G. (1996). Transverse design of adjacent precast prestressed concrete box girder bridges. PCI Journal, 41(4), Engineering, Fu, C. C., Pan, Z., and Ahmed, M. S. (2011). Transverse posttensioning design of adjacent precast solid multi beam bridges. Journal of performance of constructed facilities, 25(3), Generation Ultrahigh-Performance Concrete Pi-Girders. Journal of Bridge Grace, N. F., Jensen, E. A., and Bebawy, M. R. (2012). Transverse post -tensioning arrangement for side-by-side box-beam bridges. PCI Journal, 57 (2), Graybeal, B. A. (2006). Material property characterization of ultra-high performance concrete (No. FHWA-HRT ). Graybeal, B. A. (2013). Material Characterization of Field-Cast Connection Grouts. Tech Brief FHWAHRT McLean, VA: FHWA (Federal Highway Administration). Graybeal, B., Design and Construction of field-cast UHPC connections. U.S. Department of Transportation, Federal Highway Administration, FHWA-HRT , October 2014, 36 pp. Gulyas. R., J., Wirthlin, G., J., and Champa, J., T. (1995). Evaluation of keyway grout test methods for precast prestressed concrete bridge. PCI Journal, 40(1),

240 240 Hanna, K., E. (2008). Behavior of adjacent precast prestressed concrete box girder bridges. Ph.D. dissertation, Univ. of Nebraska Lincoln, Lincoln, NE. Hanna, K., E., Morcous, G., and Tadros, M., K. (2009). Transverse post-tensioning design and detailing of precast prestressed concrete adjacent box-girder bridges. PCI Journal, 54(4), Hansen, J., Hanna, K., and Tadros, M. k. (2012). Simplified transverse post-tensioning construction and maintenance of adjacent box girders. PCI Journal, 57(2), Hlavacs, G. M., Long, T., Miller, R. A., and Baseheart, T. M. (1996). Nondestructive determination of response of shear keys to environmental and structural cyclic loading. Transportation research record, 1574, Huckelbridge, A. A., El-Esnawi, H., and Moses, F. (1995). Shear key performance in multi beam box girder bridges. Journal of Performance of Constructed Facilities, 9(4), Huckelbridge, A., and El-Esnawi, H., (1997). Evaluation of improved shear key design for multi beam box girder bridges. U.S. Department of Transportation, Federal Highway Administration, FHWA/OH-97/009, November Huffman, J. M. (2012). Destructive Testing of a Full-Scale 43 Year Old Adjacent International Federation for Structural Concrete (fib). (2007). Structural connections for Issa, M., A., Ribeiro do Valle, C., L., Abdalla, H., A., Islam, S., and Issa, M., A. (2003). Performance of transverse joint grout materials in full-depth precast concrete bridge deck systems. PCI Journal, 48(4),

241 241 Lall, J., Alampalli, S., and DiCocco, E. F. (1998). Performance of full-depth shear keys in adjacent prestressed box beam bridges. PCI Journal, 43(2), Lee, J., K., and Lee, S., H., (2014) Evaluation of Flange-To-Flange Connection of Precast Concrete Deck Using UHPFRC. PCI/NBC. Washington, D.C. Li, Y. E., Guo, L., Rajlic, B., and Murray, P. (2014). Hodder Avenue Underpass: An Innovative Bridge Solution with Ultra-High Performance Fibre-Reinforced Concrete. Key Engineering Materials. Ludescher, H., and Brühwiler, E. (2009). Dynamic amplification of traffic loads on road bridges. Structural Engineering International, 19(2), Miller, R. A., Hlavacs, G. M., Long, T., and Greuel, A. (1999). Full-scale testing of shear keys for adjacent box girder bridges. PCI Journal, 44(6), Ohio University. PCI (Precast/Prestressed Concrete Institute). (2012). The State of Art of Precast/Prestresed Adjacent Box Beam Bridges. (1 st ed.). Chicago, IL. Perry, V., Dykstra, D., Murray, P., and Rajilic, B. (2010). Innovative field cast UHPC joints for precast bridge systems 3-span live load continuous. Annual Conference of the Transportation Association of Canada, Halifax, Nova Scota. Perry, V., H., and Royce, M. (2010). Innovative field-cast UHPC joints for precast bridge decks (side-by-side deck bulb-tees), Village of Lyons, New York: design, prototyping, testing and construction. 3rd fib International Congress.

242 242 Perry, V., Scalzop, and Weiss, G. (2007). Innovative field cast UHPC joints for precast deck panel bridge superstructures - CN overhead bridge at Rainy Lake, Ontario. Concrete Bridge Conference. Phares, B., Wipf, T., Bigelow, J., Bowers, R., and Klaiber, F. Precast Concrete Elements for Accelerated Bridge Construction: Laboratory Testing, Field Testing, Evaluation of a Precast Concrete Bridge, Madison County Bridge. IHRB Project TR-561, January precast concrete buildings (1st ed.). Lausanne, Switzerland. Precast/Prestressed Concrete Institute (PCI). (2010). PCI design handbook (7th ed.). Prestressed Concrete Box Beam Bridge: Middle and West Spans. Master Thesis, Russell, H. (2011). Adjacent precast concrete box-beam bridges: state of the practice. PCI Journal.56 (1) Russell, H., and Graybeal, B. (2013). Ultra-high performance concrete: a state-of the-art report for the bridge community. U.S. Department of Transportation, Federal Highway Administration, FHWA-HRT , pp.171. Russell, H., G. (2009). Adjacent precast concrete box-beam bridges: connection details. National Cooperative Highway Research Program (NCHRP). Sargand, S., Walsh, K., Hussein, H., Al-Rikabi, F., and Steinberg, E. (2017). Modeling the Shear Connection in Adjacent Box-Beam Bridges with Ultra-High Performance Concrete Joints - Part II: Load Transfer Mechanism. Reston VA: American Society of Civil Engineers

243 243 Semendary, A., Steinberg, E., Walsh, K., and Barnard, E (2017). Evaluation of Moment Live Load Distribution Factors for Adjacent Precast Prestressed Concrete Box Beam Bridges utilized Reinforced UHPC Shear Key Connections. Journal of Bridge Engineering, ASCE, Semendary, A., Walsh, K and Steinberg, E. (2017). Early Age Behavior of an Adjacent Prestressed Concrete Box-beam Bridge Containing UHPC Shear Keys with Transverse Dowels. Journal of Bridge Engineering, Semendary, A., Steinberg, E., and Walsh, K. Dynamic Response of Adjacent Prestressed Concrete Box Beam Bridge utilized Reinforced UHPC Shear Keys. 39th IABSE Symposium Vancouver Canada, Steinberg, E. P., and Semendary, A. A. (2016). Evaluation of Transverse Tie Rods in a 50-Year-Old Adjacent Prestressed Concrete Box Beam Bridge. Journal of Bridge Engineering, 22(3), Steinberg, E., Semendary, A., and Walsh, K. Implementing Ultra High Performance Concrete (UHPC) With Dowel Bars in Longitudinal Joints (Shear Key) in an Adjacent Box Beam Bridge. First International Interactive Symposium on UHPC, July Steinberg, E., Semendary, A., Walsh, K. (2015). Adjacent Precast Box Beam Bridges: Using UHPC Longitudinal Joints. 8. Buffalo, NY: The Construction Specifier; 68:

244 244 Steinberg., Ubbing, J., Giraldo-Londoño, O., and Semendary, A. (2014). Parametric Analysis of Adjacent Prestressed Concrete Box-Beams with UHPC-dowel shear keys. Proceedings of the 2014 PCI National Bridge Conference, DC, USA, 2016, 11 pp. Ubbing, J. L. (2014). Analytical Investigation of Adjacent Box Beam Ultra-High Performance Concrete Connections. Master Thesis, Ohio University. Ulku, E., Attanayake, U., and Aktan, H., M. (2010). Rationally designed staged posttension abates reflective cracking on side by side box-beam bridge deck. Transpiration Research Board (TRB) 89 th annual meeting, Washington, D.C. Yuan, J., and Graybeal, B., Bond behavior of reinforcing steel in ultra-high performance concrete. U.S. Department of Transportation, Federal Highway Administration, FHWA-HRT , October 2014,78 pp. Yuan, J., and Graybeal, B.,(2014). Adjacent box beam connection. PCI/NBC. Washington, D.C. Yuan, J., and Graybeal, B.,(2016). Full-scale testing of shear key details for precast concrete box-beam bridges. Journal of Bridge Engineering, Dio: / (ASCE) BE ).

245 245 APPENDIX A: LIVE LOAD DISTRIBUTION FACTORS DETERMINATION FOR CASE F A. Interior beams (AASHTO LRFD, b-1) One design lane loaded g = k[ b 33.3L ]0.5 [ I J ]0.25 Eq.A-1 Two design lanes loaded g = k[ b 305 ]0.6 [ b 12.0L ]0.2 [ I J ]0.06 Eq.A-2 Where: (values for bridge shown in parenthesis) N = Number of beams, 5 N 20, (N=7) k = 2.5(N) , (k =1.69) b = Beam width, 35 b 60, (b = 48 in. ) L = Span length, 20 L 120, (L = 61 ft) I = Beam moment of inertia, (I = 32,065.3 in 4 ) J = Tortional stiffness for closed thin-walled shapes and can be defined in Eq. A-3 (82,301.2 in 4 ) J = 4A 0 2 s/t Eq.A-3 Where: A 0 = Area enclosed by centerlines of elements (in. 2 ) s = Length of side element (in.) t = Width of plate-element (in.) B. Exterior beams (AASHTO LRFD, d-1)

246 246 One design lane loaded g = e g interior Eq.A-4 e = d e 1.0 Eq.A-5 30 de 2.0 Where: de = the distance from the centerline of the exterior web of exterior beam to the interior face of the traffic railing. It should be taken positive if the exterior web is inboard and negative if it is outboard of the curb or traffic barrier (2.75 in.). Two design lanes loaded g = e g interier Eq.A-6 e = d e 25 1 Eq.A-7

247 247 APPENDIX B: LIVE LOAD DISTRIBUTION FACTORS DETERMINATION FOR CASE G A. Interior beams (AASHTO LRFD b-1) g = S D Eq.B-1 Where: g = Distribution factor S = Spacing of beams or webs (ft) D = Width of distribution per lane (ft) D = 11.5 N L N L (1 0.2C) 2, when C 5 D = 11.5 N L, when C > 5 N L = Number of design lane 6 C = Stiffness parameter C = K ( W L ) K W = Edge-to-edge width of the bridge (ft) L = Span of beam (ft) K = Constant for different types of construction (1 + μ) I K = J μ = Poisson s ratio by substituting values for the bridge into the equations: ( ) K = = 0.684

248 248 C = ( 28 ) = D = ( ) 2 = ft g = = B. Exterior beams (AASHTO LRFD d-1) There are no design specifications for this case in AASHTO LRFD (2012). The same equations were used. The same e factor occurs as calculated for the Case (f) exterior beam.

249 249 APPENDIX C: DYNAMIC TRUCK TEST RESULTS Mid span bottom strain versus time for truck speed at 10 mph Mid span bottom deflection versus time for truck speed at 10 mph

250 250 Mid span bottom strain versus time for truck speed at 15 mph Mid span bottom deflection versus time for truck speed at 15 mph

251 251 Mid span bottom strain versus time for truck speed at 25 mph. Mid span bottom deflection versus time for truck speed at 25 mph

252 252 Mid span bottom strain versus time for truck speed at 30 mph Mid span bottom deflection versus time for truck speed at 30 mph

253 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Thesis and Dissertation Services!