Impact of construction loads on steel diaphragm bridge design

GEORGIA DOT RESEARCH PROJECT 18-02 FINAL REPORT
IMPACT OF CONSTRUCTION LOADS ON STEEL DIAPHRAGM BRIDGE DESIGN
OFFICE OF PERFORMANCE-BASED MANAGEMENT AND RESEARCH
600 WEST PEACHTREE NW ATLANTA, GA 30308

1. Report No.:

2. Government Accession No.:

FHWA-GA-21-1802

N/A

4. Title and Subtitle:

Impact on Construction Loads on Steel Diaphragm Bridge

Design

7. Author(s): Lauren K. Stewart (PI), PhD, PE;

3. Recipient's Catalog No.: N/A
5. Report Date: November 2022
6. Performing Organization Code: N/A
8. Performing Organization Report No.: 18-02

Lawrence Kahn (co-PI), PhD, PE;

Yang Wang (co-PI), PhD;

Nadine Fahed

9. Performing Organization Name and Address:

10. Work Unit No.:

Georgia Tech Research Corporation

N/A

926 Dalney Street NW

11. Contract or Grant No.:

Atlanta, GA 30332-0420

PI#0016159

Phone: (404) 385-1919

Email: lauren.stewart@ce.gatech.edu

12. Sponsoring Agency Name and Address:

13. Type of Report and Period Covered:

Georgia Department of Transportation

Final; Start Sept 2018 End Sept 2021

Office of Performance-based Management and Research 600 West Peachtree St. NW

14. Sponsoring Agency Code: N/A

Atlanta, GA 30308

15. Supplementary Notes:

Prepared in cooperation with the US Department of Transportation, Federal Highway Administration.

16. Abstract:

Bridges are critical structures, serving an important function that is vital to the safe and economical conveyance of

people and goods throughout Georgia. They are designed with specifications to carry loads including their self-

weight and a design vehicle load, among others, when they are in service. Satisfying all design specifications is

crucial to the structure's strength, stiffness, stability, and durability throughout its lifetime. In addition to the in-

service dead and live load conditions, bridges are also designed to accommodate various loading conditions during

the construction process. In some cases, these construction load and associated stability requirements are the

governing load conditions for some of the bridges' components. Georgia Department of Transportation (GDOT) has

recently allowed the substitution of steel diaphragms for concrete diaphragms in its bridges. This substitution is

gaining popularity among contractors for its ease of construction and subsequent reduction of cost. Currently, there is

no standardized design for GDOT steel diaphragms, and contractors are allowed to produce their own designs based

on loading scenarios currently specified in the 2018 GDOT Bridge and Structures Design Manual. These scenarios

include full long-term wind loadings and are thought to be overly conservative because the actual loads to which the

bridges are subjected during the construction process are poorly understood. This project seeks to provide the data

and recommendations for a more efficient, yet safe, steel diaphragm design. Specifically, this project will (1) observe

and measure GDOT construction practices through visual observations by experts and by electronic sensors, (2)

quantify the effects of the construction practices in terms of loadings via observations and computational models, (3)

assess the overall impact of construction load variations on bridge designs, and (4) make recommendations to GDOT

for loading specifications and for a standardized steel diaphragm design.

17. Keywords:

18. Distribution Statement:

Diaphragm, Construction, Monitoring

No Restriction

19. Security Classification 20. Security Classification (of

(of this report):

this page):

Unclassified

Unclassified

Form DOT 1700.7 (8-69)

21. No. of Pages: 175

22. Price: Free

GDOT Research Project No. 18-02
Final Report
IMPACT OF CONSTRUCTION LOADS ON STEEL DIAPHRAGM BRIDGE DESIGN
By
Lauren Stewart, PhD, PE Associate Professor School of Civil and Environmental Engineering
Lawrence Kahn, PhD, PE Emeritus Professor School of Civil and Environmental Engineering
Yang Wang, PhD Associate Professor School of Civil and Environmental Engineering
Nadine Fahed Graduate Research Assistant
Georgia Institute of Technology
Contract with Georgia Department of Transportation
In cooperation with US Department of Transportation Federal Highway Administration
November 2022
The contents of this report reflect the views of the authors, who are responsible for the facts and accuracy of the data presented herein. The contents do not necessarily reflect the official views of the Georgia Department of Transportation or the Federal Highway Administration. This report does not constitute a standard, specification, or regulation.
ii

EXECUTIVE SUMMARY Georgia Department of Transportation (GDOT) recently allowed the substitution of steel diaphragms for concrete diaphragms in its bridges. This substitution is gaining popularity among contractors for its ease of construction and subsequent reduction in cost. Currently, there is no standardized design for GDOT steel diaphragms, and contractors are allowed to produce their own designs based on loading scenarios currently specified in the 2018 GDOT Bridge and Structures Design Manual. These scenarios were thought to be overly conservative because the actual loads to which the bridges are subjected during the construction process are poorly understood.
Through in-situ bridge monitoring and finite element modeling, this project quantified the loads on multiple k-frame diaphragms on a single bridge during the construction process, specifically during concrete deck pouring. The monitoring and modeling determined that the wind loads, specified by American Association of State Highway and Transportation Officials (AASHTO), produced strains that were greater than the construction loads for the members that were monitored (diagonal members and chords). Additional testing is needed to determine the behavior of the gusset plate and to verify connections under the wind loading, which was not monitored as part of the research effort.
iii

* SI is the symbol for the International System of Units. Appropriate rounding should comply with Section 4 of ASTM E380. (Revised March 2003)
iv

TABLE OF CONTENTS
CHAPTER 1. INTRODUCTION .................................................................................... 1 RESEARCH MOTIVATION....................................................................................... 1 BACKGROUND............................................................................................................ 2 OBJECTIVES................................................................................................................ 5 REPORT ORGANIZATION ....................................................................................... 6
CHAPTER 2. EXAMPLE CURRENT DESIGN PROCESS ....................................... 7 CHAPTER 3. BRIDGE VISITS AND OBSERVATIONS.......................................... 11 CHAPTER 4. LABORATORY TESTING .................................................................. 16
MARLET SYSTEM .................................................................................................... 16 LABORATORY SPECIMEN .................................................................................... 18 TEST SETUP............................................................................................................... 21 INSTRUMENTATION............................................................................................... 23 TESTING PROCEDURE........................................................................................... 26 RESULTS AND VALIDATION ................................................................................ 27
Load Case 1 (LC1): Loading in Steps .................................................................... 27 Load Case 2 (LC2): Loading Continuously .......................................................... 29 CHAPTER 5. BRIDGE MONITORING...................................................................... 32 BRIDGE MONITORING 1........................................................................................ 32 Instrumentation ....................................................................................................... 34 Installation Procedure ............................................................................................. 35 Data Collection......................................................................................................... 37 BRIDGE MONITORING 2........................................................................................ 37 Data Analysis............................................................................................................ 46 CHAPTER 6. FINITE ELEMENT MODELS............................................................. 52 SAP2000 MODEL ....................................................................................................... 52 CSIBRIDGE MODEL ................................................................................................ 54 Wind Load Analysis ................................................................................................ 62 Staged Construction Analysis................................................................................. 63
v

CSIBridge Data Processing..................................................................................... 64 ABAQUS MODEL 1: LABORATORY SIMULATION ......................................... 69 ABAQUS MODEL 2: FIELD SIMULATION ......................................................... 72 COMPARISONS ......................................................................................................... 78 CHAPTER 7. CONCLUSIONS..................................................................................... 80 APPENDIX A: DIAPHRAGM CALCULATIONS ..................................................... 82 APPENDIX B: TIME LAPSE FIGURES................................................................... 117 APPENDIX C: MATLAB CODE................................................................................ 127 APPENDIX D: RAW DATA........................................................................................ 157 APPENDIX E: CONCRETE CONSTRUTION LOAD CALCULATIONS........... 161 ACKNOWLEDGMENTS ............................................................................................ 162 REFERENCES.............................................................................................................. 163
vi

LIST OF FIGURES
Figure 1. Engineering drawing. Partial section at intermediate diaphragms ...................... 2 Figure 2. Photo. Intermediate k-frame diaphragms in situ ................................................. 2 Figure 3. Engineering Drawing. Example of existing GDOT solid plate with MC section diaphragms .......................................................................................................................... 4 Figure 4. Engineering Drawing. Example of existing GDOT k-frame diaphragm ............ 4 Figure 5. Excerpt. Table 3.8.1.2.1.1. from AASHTO......................................................... 5 Figure 6. Equations. Bridge wind load calculations. .......................................................... 7 Figure 7. Equations. Concentrated wind force calculations. .............................................. 8 Figure 8. Engineering Drawing. Example wind loading of k-frame Diaphragm ............... 8 Figure 9. Calculation. Example wind load internal forces (from RISA) ............................ 9 Figure 10. Calculation. Example wind load member design ............................................ 10 Figure 11. Photo. Installation of k-frame diaphragm........................................................ 12 Figure 12. Photo. Still from time-lapse camera during concrete deck pour ..................... 13 Figure 13. Map. Bridge location ....................................................................................... 14 Figure 14. Map. Satellite view of bridge span to be instrumented ................................... 14 Figure 15. Photo. Martlet wireless sensing unit (WSU) ................................................... 16 Figure 16. Photo. 24-bit ADC sensor board ..................................................................... 17 Figure 17. Photo. Cabled DAQ setup using NI ................................................................ 18 Figure 18. Engineering Drawing. Fabricated k-frame replica (1 of 3) ............................. 19 Figure 19. Engineering Drawing. fabricated k-frame replica (2 of 3) .............................. 19 Figure 20. Engineering Drawing. Fabricated k-frame replica (3 of 3) ............................. 20 Figure 21. Photo. Frabricated k-frame replica .................................................................. 20 Figure 22. Schematic. Laboratory setup ........................................................................... 21 Figure 23. Photo. Laboratory setup of the steel diaphragm .............................................. 22 Figure 24. Photo. Laboratory setup with crane................................................................. 22 Figure 25. Schematic. Strain gauge instrumentation layout ............................................. 24 Figure 26. Photo. Cleaning of surface and instrumentation of strain gauges onto the diaphragm ......................................................................................................................... 24 Figure 27. Schematic. Displacement sensor layout .......................................................... 25 Figure 28. Photo. Setup of LVDT displacement sensors to measure in-plane deflections25 Figure 29. Photo. Setup of string potentiometer to measure out-of-plane deflections ..... 26 Figure 30. Graph. Variation of the load as a function of time when loaded in steps (on the left) and when loaded continuously (on the right) ............................................................ 26 Figure 31. Graph. Variation of the load as a function of time for LC1 ............................ 27 Figure 32. Graph. Variation of the displacement measure by the string potentiometer as a function of time for LC1 ................................................................................................... 28 Figure 33. Graph. Variation of the displacement measure by LVDT2 as a function of time for LC1.............................................................................................................................. 28
vii

Figure 34. Graph. Plot of the strain as a function of time using Martlet and NI DAQ (left) at locations and (right) for LC1.................................................................... 29 Figure 35. Graph. Plot of the strain as a function of time using Martlet and NI DAQ (left) at locations and (right) for LC1 ................................................................... 29 Figure 36. Graph. Variation of the load as a function of time for Load Case 1 ............... 30 Figure 37. Graph. Plot of the strain as a function of time using Martlet and NI DAQ (left) at locations and (right) for LC2.................................................................... 31 Figure 38. Graph. Plot of the strain as a function of time using Martlet and NI DAQ (left) at locations and (right) for LC2 ................................................................... 31 Figure 39. Engineering drawing. Elevation view, bridge Sheet 3 of 64, Bridge 11B plans
32 Figure 40. Engineering Drawing. Plan view Span 27RT, Bridge 11B ............................. 33 Figure 41. Engineering Drawing. Half-Section Span 27RT, Bridge 11B ........................ 33 Figure 42. Engineering drawing. Partial section at intermediate diaphragm.................... 33 Figure 43. Photo. Panoramic picture of Span27RT .......................................................... 34 Figure 44. Schematic. Instrumentation layout and sensor numeric identifiers................. 35 Figure 45. Photo. Collection of pictures depicting strain gauge installation process ....... 36 Figure 46. Photo. Setup for data acquisition..................................................................... 37 Figure 47. Photo. Steel reinforcement in deck prior to pour ............................................ 38 Figure 48. Photo. End of span prior to concrete deck pour .............................................. 38 Figure 49. Photo. Panoramic photo of span prior to deck pour ........................................ 38 Figure 50. Drawings. Elevation view, Bridge 11B ........................................................... 39 Figure 51. Drawings. Plan view, Bridge 11B ................................................................... 39 Figure 52. Drawings. Deck plan of Span 14, Bridge 11B ................................................ 39 Figure 53. Drawings. Four Half-Section of Span 14, Bridge 11B.................................... 40 Figure 54. Engineering Drawings. K-frame steel diaphragm ........................................... 40 Figure 55. Schematic. Instrumentation layout and sensor numeric identifiers................. 41 Figure 56. Photo. Cleaning and sanding of the surface (left), bonding of strain gauges (right) ................................................................................................................................ 42 Figure 57. Photo. Strain gauges after bonding to surface ................................................. 43 Figure 58. Photo. Weatherproofing of strain gauges ........................................................ 43 Figure 59. Photo. Final product of strain gauges installation and connection to wireless sensing units for Bay 1 (left) and Bay 3 (right) ................................................................ 44 Figure 60. Photo. Deck pour on August 26, 2020 ............................................................ 45 Figure 61. Photo. Base station setup during data collection ............................................. 46 Figure 62. Photo. Time-lapse camera footage on both sides of the bridge....................... 46 Figure 63. Graph. Raw data (black) versus filtered data (Red) ........................................ 47 Figure 64. Data. Strain gauge data from top of Bay 1 ...................................................... 48 Figure 65. Data. Strain gauge data from bottom of Bay 1 ................................................ 49 Figure 66. Data. Strain gauge data from top of Bay 3 ...................................................... 50
viii

Figure 67. Data. Strain gauge data from bottom of Bay 3 ................................................ 51 Figure 68. Schematic. Effective length used in SAP2000 model ..................................... 53 Figure 69. Schematics. SAP2000 K-frame diaphragm reaction forces ............................ 53 Figure 70. Schematics. SAP2000 analysis axial forces .................................................... 53 Figure 71. Model. CSIBridge model of Span 14 of Bridge11B of the I-16 I-75 Interchange project with prestressed girders..................................................................... 54 Figure 72. Model. Concrete end diaphragms and supports in CSIBridge model ............. 54 Figure 73. Model. K-frame intermediate steel diaphragms in CSIBridge model ............. 55 Figure 74. Engineering Drawing. Bulb-Tee section at midpoint and end ........................ 55 Figure 75. Drawing. Prestressed concrete girder tendons................................................. 56 Figure 76. Drawing. Fixed-expansion end supports of Span 14....................................... 56 Figure 77. Interface. Elevation, plan, and section drawings of the final bridge tendon layout display .................................................................................................................... 57 Figure 78. Model. Model tendon objects (in green) ......................................................... 57 Figure 79. Model. Changes to end releases of steel diaphragms in CSIBridge................ 59 Figure 80. Drawing. Abutment drawings for Span 14 of Bridge 11B .............................. 60 Figure 81. Model. Bearing and substructure elevation in CSIBridge............................... 60 Figure 82. Drawing. Partial section at intermediate diaphragm of Span 14 of Bridge 11B
61 Figure 83. Model. Frame joint offset location for diagonal members of the diaphragm.. 61 Figure 84. Model. Stress S11 in elements from step 1 of the multistep linear wind load analysis.............................................................................................................................. 63 Figure 85. Model. Stress S11 in elements from step 2 of the multistep linear wind load analysis.............................................................................................................................. 63 Figure 86. Sketch. Strain gauge location on angle............................................................ 65 Figure 87. Data. Strain gauge data from top of Bay 1 ...................................................... 65 Figure 88. Data. Strain gauge data from bottom of Bay 1 ................................................ 66 Figure 89. Data. Strain gauge data from top of Bay 3 ...................................................... 67 Figure 90. Data. Strain gauge data from bottom of Bay 3 ................................................ 68 Figure 91. Model. Front view of k-frame diaphragm Abaqus model ............................... 69 Figure 92. Model. Back view of k-frame diaphragm Abaqus model ............................... 69 Figure 93. Model. 3D Isometric view of k-frame diaphragm Abaqus model................... 70 Figure 94. Model. Mesh detail of k-frame diaphragm...................................................... 70 Figure 95. Model. Interactions between members of the k-frame .................................... 71 Figure 96. Model. Bolt reference points and no-slip constraints ...................................... 71 Figure 97. Model. Bolt node to surface contact interaction.............................................. 72 Figure 98. Model. Concrete girders in Abaqus in undeformed (left) and deformed (right) states 73 Figure 99. Model. Tie constraint between the steel diaphragm and concrete girders ...... 74 Figure 100. Model. Concrete pour simulation by quarters. .............................................. 75
ix

Figure 101. Data. Diagonal member strain results from Abaqus simulation for concrete only 75 Figure 102. Data. Bottom chord strain results from Abaqus simulation for concrete only
76 Figure 103. Photo. Identification of the concrete pavement machine as a major construction load from time-lapse camera ........................................................................ 77 Figure 104. Model. Simulation of halfway point of concrete pouring process: concentrated load at midspan and distributed load ........................................................... 77
x

LIST OF TABLES Table 1. Example Bridge Properties ................................................................................... 7 Table 2. Summary of Bridge Site Visits ........................................................................... 11 Table 3. Summary of Martlet Units and Sensors .............................................................. 35 Table 4. Martlet Units and Sensors used for External Diaphragm ................................... 41 Table 5. of Martlet Units and Sensors used for Internal diaphragm ................................. 42 Table 6. Summary of micro-strain values for varying load cases in Abaqus ................... 78 Table 7. Comparison of maximum (absolute value) micro-strain .................................... 79
xi

CHAPTER 1. INTRODUCTION
RESEARCH MOTIVATION Bridges are critical structures, serving an important function that is vital to the safe and economical conveyance of people and goods throughout Georgia. They are designed with specifications to carry loads including their self-weight and a design vehicle load, among others, when they are in-service. Satisfying all design specifications is crucial to the strength, stiffness, stability, and durability of the structure throughout its lifetime. In addition to the in-service dead and live load conditions, bridges are also designed to accommodate various loading conditions during the construction process. In some cases, these construction load and associated stability requirements are the governing load conditions for some of the bridges' components.
Georgia Department of Transportation (GDOT) has recently allowed the substitution of steel diaphragms for concrete diaphragms in its bridges, as shown in Figure 1 and Figure 2.(1) This substitution is gaining popularity among contractors for its ease of construction and subsequent reduction of cost. Currently, there is no standardized design for GDOT steel diaphragms, and contractors are allowed to produce their own designs based on loading scenarios currently specified in the 2018 GDOT Bridge and Structures Design Manual. These scenarios include full long-term wind loadings and are thought to be overly conservative because the actual loads to which the bridges are subjected during the construction process are poorly understood. This project seeks to provide the data and recommendations for a more efficient, yet safe, steel diaphragm design. Specifically, this project will (1) observe and measure GDOT construction practices through visual
1

observations by experts and by electronic sensors, (2) quantify the effects of the construction practices in terms of loadings via observations and computational models, (3) assess the overall impact of construction load variations on bridge designs, and (4) make recommendations to GDOT for loading specifications and for a standardized steel diaphragm design.
Figure 1. Engineering drawing. Partial section at intermediate diaphragms
Figure 2. Photo. Intermediate k-frame diaphragms in situ BACKGROUND Standard specifications from multiple states' departments of transportation (DOTs) show a wide variety of construction load approaches for dead, live, wind, and impact
2

loadings.(2-16) The specifications from each state adopt their own combination of practices, the majority of which are derived from the AASHTO Guide Design Specifications for Bridge Temporary Works and ASCE 37: Design Loads during Construction.(17-18) The state DOTs' approaches vary in terms of the magnitude of load and the construction phases in which the loads apply.
In addition to the loading specifications, the state agencies also differ in terms of their acceptance and design requirements for diaphragms. GDOT has historically solely recommended the use of concrete diaphragms in its bridges; however, the current 2018 GDOT Bridge and Structures Design Manual has the following provision regarding the substitution of steel diaphragms for certain conditions in Section 3.9.1.1:
"Steel Diaphragms at the contractor's option, steel diaphragms may be used in lieu of the concrete diaphragms shown in the plans. At a minimum, steel diaphragms are to be designed for applied wind load. Stability of the beams and structure during all phases of construction are the sole responsibility of the contractor. Submit shop drawings and calculations for the steel diaphragms to the engineer for review and acceptance."
Since the introduction of this provision, a relatively small number of contractors have chosen the steel diaphragm option and have provided GDOT with new designs and supporting calculations for acceptance checks. Two examples of these designs are shown in Figure 3 and Figure 4. From the figures, it is apparent that two designs, while meeting the current standard, are drastically different in both geometry and sizing. GDOT expects that the number of instances of steel diaphragm substitution will continue to increase due to its ease of construction and reduced cost for the contractor. Because of this, GDOT is
3

interested in understanding the construction and other loads on these systems such that the design can be standardized.
Figure 3. Engineering Drawing. Example of existing GDOT solid plate with MC section diaphragms
Figure 4. Engineering Drawing. Example of existing GDOT k-frame diaphragm Currently, the load considered for the design of the steel diaphragms is wind load. The calculations are done in accordance with the windward load as illustrated in AASHTO Table 3.8.1.2.1-1, shown in Figure 5. As indicated in the table, the minimum load used
4

for design should be 0.3 klf in the plane of the windward chord. Chapter 2 provides an example of the current design practice.
Figure 5. Excerpt. Table 3.8.1.2.1-1. from AASHTO(17) OBJECTIVES The objectives of this research are as follows:
1) To understand the loads on steel diaphragm bridges during the construction process through visual observation of bridge construction.
2) To measure the effect of construction loads on steel diaphragm bridges during construction via sensors.
3) To quantify the construction loads on the structure using the observations and measured data combined with computational and analytical models.
4) To draft recommended practices (e.g., construction loads) for GDOT steel diaphragm design.
5

REPORT ORGANIZATION Chapter 2 of this report gives an example of a typical diaphragm design based on AASHTO 3.8.1.2.1.1. This example will be referenced in additional examples of the report. Chapter 3 describes the site visits that were conducted throughout the project and discusses the selection of the bridge for monitoring. Chapter 4 details a laboratory effort that was used to validate the sensors for field monitoring. Chapter 5 contains details on the bridge monitoring effort, including logistics, setup, and results. Chapter 6 explains the three finite-element modeling efforts that were conducted based on the monitoring. Chapter 7 provides the conclusions and recommendations. The five appendices contain example design calculations (Appendix A), time-lapse photos (Appendix B), MATLAB codes (Appendix C), unfiltered data (Appendix D), and concrete construction load calculations (Appendix E).
6

CHAPTER 2. EXAMPLE CURRENT DESIGN PROCESS

To illustrate the typical process used for design per AASHTO, this chapter provides an

example calculation. This example was submitted to GDOT by a contractor. In this

example, A Bulb Tee 63 (BT-63) girder-type bridge with the properties and dimensions

was considered (Table 1).

Table 1. Example Bridge Properties

Bridge Element

Dimensions / Property

Longest Girder Length, L 124 ft

Girder Material

Concrete

Girder Type

BT-63

Girder Cross Section Area, A 4.95 ft2

Girder Height, H

4.5 ft

The wind pressure per linear foot, wplf, is calculated according to AASHTO Table 3.8.1.2.1.1 as the 50 psf multiplied by the girder height and should not be taken to be less than 300 plf. For this bridge, the calculations are shown the grouping of equations given in Figure 6.

Figure 6. Equations. Bridge wind load calculations. To obtain the concentrated wind force, wTWL, the value of load per linear foot, wplf, is multiplied by the longest girder length as shown by the grouping of equations (Figure 7).
7

Figure 7. Equations. Concentrated wind force calculations. The diaphragm is loaded with 50 percent of the girder length for the wind load. Therefore, the total wind load is divided by two, and thus half of the wind load, , applied to the diaphragm is 18.6 kips.
The application of load is dependent on the type of diaphragm chosen. In the case of a solid plate with MC sections, the one-half wind load is applied at one end of the diaphragm for the analysis. For the k-frame diaphragms with L sections, the half wind load is divided by two, and that value is applied to the top diagonal and bottom member horizontal leg as a lateral load on the wind face of the diaphragm, chosen to be the right side in Figure 6. Because k-frame diaphragms are most of interest to GDOT, the analysis will be continued for that example.
Figure 8. Engineering Drawing. Example wind loading of k-frame Diaphragm Analysis of the structure was completed in RISA-3D and the internal forces and moments in all the members were determined (Figure 9).(19) The Allowable Strength Design direct
8

analysis method is used for design of the steel members and bolts, whereby no load factors are used. The members of the diaphragm are designed with this method (Figure 8). Appendix A contains additional design calculations provided to GDOT contractor. These calculations provide additional design checks on the bolted connections.
Figure 9. Calculation. Example wind load internal forces (from RISA) 9

Figure 10. Calculation. Example wind load member design 10

CHAPTER 3. BRIDGE VISITS AND OBSERVATIONS

Because construction loads are not typically included in the AASHTO design, it was

important for the team to monitor a bridge throughout its construction. Multiple active

bridge sites with different steel diaphragm configurations, particularly k-frame and single

chord diaphragms, were visited to identify bridges that would potentially be used in the

monitoring stage and to plan the monitoring. A summary of the bridge visits with purpose

is given in Table 2.

Date 02/01/2019 03/20/2019 03/21/2019 05/29/2019 10/24/2019 11/01/2019 08/13/2020 08/14/2020 08/21/2020 08/26/2020

Table 2. Summary of Bridge Site Visits

Location

Purpose

Macon, I16-I75 Bridge

Conduct preliminary site visit to

https://goo.gl/maps/YjS87Azzw9ZsYjgP8 observe the different installed

diaphragms on the bridge

Macon, I16-I75 Bridge

Observe status of the bridge post

https://goo.gl/maps/YjS87Azzw9ZsYjgP8 construction updates and identify

potential spans to instrument

Macon, I16-I75 Bridge

Install time-lapse camera and

https://goo.gl/maps/YjS87Azzw9ZsYjgP8 capture time-lapse footage during

deck pour

Longstreet Bridge

Observe diaphragm installation

Gainseville, GA 30506

and placement

https://goo.gl/maps/z6s3z5eBUVDBmYQD6

Macon, I16-I75 Bridge, Bridge 11B, Span Conduct dry run installation of

25RT

strain gauges onto a single chord

diaphragm

Macon, I16-I75 Bridge, Bridge 11B, Span Test wireless data communication

25RT

during deck pouring process and

capture time-lapse footage

Macon, I16-I75 Bridge, Bridge 11B, Span 14 Instrumentation of exterior bay

diaphragm (bay 1) with strain

gauges

Macon, I16-I75 Bridge, Bridge 11B, Span 14 Instrumentation of interior bay

diaphragm (bay interior 3) with

strain gauges

Macon, I16-I75 Bridge, Bridge 11B, Span 14 Troubleshoot installed sensors on

site and test wireless

communication post installation

Macon, I16-I75 Bridge, Bridge 11B, Span 14 Conduct wireless data collection

during concrete deck pouring

process

11

During the site visit on May 29, 2019, the installation of the steel diaphragm was observed (Figure 7). The diaphragms are preassembled according to the steel drawing and then lifted and placed between the girders via a crane. The side angles are then bolted into the girders. This method is both practical and efficient in terms of resources including cost, time, and labor and is often the reason that contractors choose to replace the concrete diaphragms with the steel option.
Figure 11. Photo. Installation of k-frame diaphragm During site visit on March 21, 2019, span 25RT of Bridge 11B in the I-16/I-75 Interchange project in Macon was monitored with two time-lapse cameras during the concrete deck pouring. Figure 8 shows a single image of this time lapse. The time lapse was examined to identify the sources of construction loads acting on the diaphragms. After reviewing the captured footage, the main sources of loading identified were the concrete pour and the concrete pavement machine. Before the slab is placed and the concrete sufficiently hardens, the steel cross frame diaphragms help limit rotations and
12

twisting distortions in the concrete girders, as well provide lateral stiffness to the bridge. Appendix B contains a series of images from the time lapse.
Figure 12. Photo. Still from time-lapse camera during concrete deck pour After visiting multiple bridge sites using steel diaphragms in prestressed reinforced concrete girder bridges in Georgia, a bridge in Macon was selected for instrumentation and monitoring. The bridge was selected mainly for the installation and deck pouring time frame as well as accessibility to the diaphragms using a bucket truck. More specifically, the bridge studied was located at the interchange of Interstate I75 in Macon, Georgia. Figure 9 shows a map of the location, and Figure 10 shows a satellite image of the bridge. The interchange is currently under construction in the satellite image.
13

Figure 13. Map. Bridge location
Figure 14. Map. Satellite view of bridge span to be instrumented The selected bridge consists of prestressed concrete girders (PSC) with steel diaphragms and concrete deck slab. More specifically, span 25RT was chosen for instrumentation and span 14 were chosen for instrumentation and monitoring. Span 25RT was used to ensure our instrumentation process was adequate and test our data acquisition system using our
14

wireless sensing system. Span 14 was chosen for instrumentation and monitoring during the concrete deck pouring process. Chapter 5 explains each field test in detail.
15

CHAPTER 4. LABORATORY TESTING Before the monitoring occurred, an experimental effort was conducted to validate the wireless sensing system, Martlet, which was used for the bridge test. This chapter explains the Martlet system, provides details on the test setup, and summarizes the results of the validation testing. MARTLET SYSTEM The wireless sensing system, named Martlet, which was developed by Georgia Tech researchers and collaborators.(20-22) The data acquisition was done wirelessly, via the wireless sensing system Martlet (Figure 11) and via National Instrument cabled data acquisition (NI DAQ). An advantage of the wireless sensing system is that it can be conveniently installed and used, particularly in a cluttered construction site. The wireless sensing unit (WSU) consists of the battery board and the mother board as well as a 24bit ADC board (Figure 12) used for data collection. Martlet uses a Texas Instruments Piccolo microcontroller (TMX320F28069) as the core processor, which can be programmed based on different needs and tasks.
Figure 15. Photo. Martlet wireless sensing unit (WSU) 16

Figure 16. Photo. 24-bit ADC sensor board In terms of wireless data acquisition, all units were prepared and programmed accordingly. Note that in addition to strain gauges, a thermistor was also used in each Martlet unit to obtain temperature readings. For the cabled data acquisition, all the electrical wiring was connected appropriately. A file was prepared in LabView to process the strain, load, and deflection signals. The different cabled sensors used along with the corresponding NI cards are delineated in Figure 13.
17

Figure 17. Photo. Cabled DAQ setup using NI LABORATORY SPECIMEN At Georgia Tech, a full-scale replica of one of the k-frame steel diaphragms approved by GDOT was fabricated in the machine shop at Georgia Tech. Drawings of the k-frame are shown in Figure 16, Figure 17, and Figure 18. The k-frame consists of a L5x5x0.5 bottom chord member connected to two L3.5x3.5x0.5 diagonal members via a 0.25-inchthick gusset plate. Furthermore, the opposite ends of the diagonal members are connected to L8x6x0.5 angles. From the drawings, the steel members were cut and assembled by bolting together the appropriate members. The full-scale replica is shown in Figure 19.
18

Figure 18. Engineering Drawing. Fabricated k-frame replica (1 of 3)
Figure 19. Engineering Drawing. fabricated k-frame replica (2 of 3) 19

Figure 20. Engineering Drawing. Fabricated k-frame replica (3 of 3)
Figure 21. Photo. Frabricated k-frame replica 20

TEST SETUP The purpose of the experiment was to validate the wireless sensing network against cabled sensors and confirm that the wireless sensors can perform continuous monitoring. To apply loads onto the test specimen and measure the induced strain, an experimental setup was designed and built in the Structural Engineering and Materials Laboratory at the Georgia Institute of Technology. A sketch of the setup is shown in Figure 19. A load cell was mounted onto a column, which was held in position by two transverse beams screwed tightly into the strong floor, also shown in Figure 19. The specimen was bolted to the plate, which, in turn, was bolted onto the rigid frame, as shown in Figure 20.
Figure 22. Schematic. Laboratory setup
21

Frame Plate

Steel Diaphragm Specimen to be tested

Stand/Support

Figure 23. Photo. Laboratory setup of the steel diaphragm The specimen was either supported by a stand or by the crane at its middle, as shown in Figure 21. The crane was used for support before loading. Once enough load was placed onto the system to ensure stability, the crane was released, and the testing was continued.

Figure 24. Photo. Laboratory setup with crane 22

INSTRUMENTATION The diaphragm specimen was equipped with a network of both cabled and wireless strain gauges to measure strains at various locations, as well as linear variable displacement transducers (LVDTs), a string potentiometer to measure displacement, and thermistors to measure temperature.
An array of different strain gauges was used that included 12 quarter bridge strain gauges and 12 full bridge strain gauges. Figure 25 illustrates the locations of the instrumented strain gauges along with the type and mode of data acquisition. To install the gauges, the surface of the diaphragm was thoroughly cleaned until bare steel surface was reached, and the strain gauges were mounted onto the diaphragm at the selected locations (Figure 26).
To obtain displacement measurements, three displacement sensors were used, and their locations are shown in Figure 27. One string potentiometer (SP1A) was used to measure the out-of-plane deflections, and two linear variable differential transformers (LVDT1 and LVDT 2) were used to measure in-plane deflection in both directions shown. The LVDTs were fixed to the support columns (Figure 28). The string potentiometer was connected to the angle and placed on the ground vertically below the member (Figure 29).
23

Figure 25. Schematic. Strain gauge instrumentation layout
Figure 26. Photo. Cleaning of surface and instrumentation of strain gauges onto the diaphragm 24

Figure 27. Schematic. Displacement sensor layout
Figure 28. Photo. Setup of LVDT displacement sensors to measure in-plane deflections 25

Figure 29. Photo. Setup of string potentiometer to measure out-of-plane deflections

TESTING PROCEDURE

The tests consisted of either loading continuously up to a certain limit, which was

determined via a finite element model or by loading in steps of 2 kips until the maximum

load was reached to mimic concrete deck pouring stages. Once the maximum load was

reached, the load was kept for a specified duration. Unloading was also done either in

steps or continuously until fully unloaded. Data were continuously collected throughout

the entire process. Figure 30 provides a summary of the loading procedure.

Load (kips) Load (kips)

12 10
8 6 4 2 0 -2
0

Load Cell

500

1000

1500

2000

2500

time(sec)

12 10
8 6 4 2 0 -2
0

Load Cell

200

400

600

800

1000

1200

1400

time(sec)

Figure 30. Graph. Variation of the load as a function of time when loaded in steps (on the left) and when loaded continuously (on the right)

26

RESULTS AND VALIDATION A MATLAB code was written to analyze the result and compare the readings obtained wirelessly and via cabled data acquisition. The code is given in Appendix C. Two load cases were considered as described in the following two subsections. Load Case 1 (LC1): Loading in Steps The first load case involved loading in steps of 2 kips until the maximum limit of 10 kips is reached. The load is then held for approximately 10 minutes. Note that for initial support, the overhead crane held the diaphragm horizontally. Once the load cell indicated 2 kips and the hydraulic jack was in contact with the diaphragm, therefore supporting the diaphragm laterally, the crane support was released. Similarly, the unloading was done in steps of 2 kips until fully unloaded. Figure 31 shows the loading and unloading.
Figure 31. Graph. Variation of the load as a function of time for LC1 The data collection was continuous throughout the loading and unloading stage. The displacement results collected by the string potentiometer are shown in Figure 28, and the maximum displacement can be seen to be around 0.70 inches in the vertical direction, in the direction normal to the plane containing the diaphragm.
27

displacement(inches)

0.8 0.6 0.4 0.2
0 -0.2
0

StringPot

500

1000 1500 2000 2500

time(sec)

Figure 32. Graph. Variation of the displacement measure by the string potentiometer as a function of time for LC1

Furthermore, the displacements in the two lateral directions were measured during the loading and unloading. Displacement from LVDT2 is shown in Figure 33, where the maximum displacement can be seen to around 0.3 inches.
LVDT2 0.3

displacement(inches)

0.2

0.1

0

0

500

1000 1500 2000 2500

time(sec)

Figure 33. Graph. Variation of the displacement measure by LVDT2 as a function of time for LC1

The results from the adjacent strain gauges were plotted for comparison purposes to ensure that the strain collected through wired NI DAQ and those obtained wirelessly through Martlet agree within a reasonable margin of errors. The overlaying plots for some strain gauge locations are shown in Figure 34 and Figure 35. and 5 are the

28

strains measured at the location shown in Figure 34 and Figure 35, respectively, by Martlet using full bridge strain gauges, and 0 and 5 are the strains measured by NI in a similar location using a full bridge and quarter bridge strain gauge, respectively.
0

strain ( )

-50
-100 0

martlet M
0
cabled FBSG
0

500

1000

1500

2000

2500

time (sec)

Figure 34. Graph. Plot of the strain as a function of time using Martlet and NI DAQ (left) at locations and (right) for LC1

0

strain ( )

-20

-40

martlet M

-60

5
cabled QBSG

1

0

500

1000

1500

2000

2500

time (sec)

Figure 35. Graph. Plot of the strain as a function of time using Martlet and NI DAQ (left) at locations and (right) for LC1

The plots that the measured strain by the two different methods of data acquisition agree within 2.8% in the loading phase and 7.8% overall.
Load Case 2 (LC2): Loading Continuously To test the data acquisition system further, a second loading scenario was also tested. For that purpose, a 2-kip load was applied, at which point the crane supporting the diaphragm horizontally was relieved. After that, the load was increasing continuously until the 10
29

kips limit load was reached. The diaphragm was held at that loading for a few minutes. Similarly, for unloading, the hydraulic jack was released until the load cell read 2 kips. At this stage, the crane was activated again to avoid any unnecessary stresses to the diaphragm, then fully unloaded and no longer supported by the hydraulic jack and load cell setup. This loading scenario is illustrated in Figure 36.
Figure 36. Graph. Variation of the load as a function of time for Load Case 1 Overlaying plots of the load strain obtained from Martlet and NI are provided in Figure 37 and Figure 38 for the same locations presented for LC1. A similar observation can be made by comparing the strains for LC2 using the different data acquisition systems. The plots show that the measured strain by the two different methods of data acquisition agree within 5.1% margin of error for loading and constant regime and 3.6% margin of error overall. These results were deemed satisfactory and, overall, validated the wireless sensing system that will be used for field testing.
30

0

strain ( )

-50
-100 0

martlet M
5
cabled QBSG
1

200

400

600

800

1000

1200

1400

time (sec)

Figure 37. Graph. Plot of the strain as a function of time using Martlet and NI DAQ (left) at locations and (right) for LC2

0

strain ( )

-50
-100 0

martlet M
0
cabled FBSG
0

200

400

600

800

1000

1200

1400

time (sec)

Figure 38. Graph. Plot of the strain as a function of time using Martlet and NI DAQ (left) at locations and (right) for LC2

31

CHAPTER 5. BRIDGE MONITORING The bridge monitoring consisted of two monitorings. The first monitoring effort was conducted as a "dry run" to check the installation processes at the construction site. The second monitoring effort was conducted during the pouring of the concrete deck. The following sections discuss these two efforts. BRIDGE MONITORING 1 The first monitoring effort that was the "dry run" of the installation was conducted on October 24 and November 11, 2019. It was located at Bridge 11B in Macon I16-I75 Interchange Project and consisted of Span 27RT, as shown in Figure 39 and Figure 40. More specifically, the intermediate diaphragm of the exterior bay of the span, shown in Figure 41, was instrumented. The diaphragm consists of a single chord, specifically a MC18x42.7 structural steel section connected to two plates shown in Figure 42. A photo of Span 27RT is shown in Figure 43.
Figure 39. Engineering drawing. Elevation view, bridge Sheet 3 of 64, Bridge 11B plans
32

Figure 40. Engineering Drawing. Plan view Span 27RT, Bridge 11B Figure 41. Engineering Drawing. Half-Section Span 27RT, Bridge 11B
Figure 42. Engineering drawing. Partial section at intermediate diaphragm 33

Figure 43. Photo. Panoramic picture of Span27RT Instrumentation Figure 44 provides the instrumentation layout for the strain gauges and temperature sensors instrumented on the diaphragm. Each strain gauge is labeled "SG," followed by an arbitrary numeric identifier. Also, temperature sensors are labeled "TS," followed by an arbitrary numeric identifier. Four strain gauges were installed at the locations shown, in addition to two temperature gauges. Strain gauges were installed 6 inches away from the centerline on each side of the centerline on the inner side of the top and bottom flange. The temperature gauges were fixed to the bottom flange of the steel surface. Note that TS1 was fixed adjacent to SG2, while TS2 was fixed adjacent to SG4. Two Martlet units were used, unit 106 (U106) and unit 128 (U128). Each unit was connected to two strain gauges and one temperature sensor. A summary of the sensors is given in Table 3. Summary of Martlet Units and Sensors.
34

Figure 44. Schematic. Instrumentation layout and sensor numeric identifiers

Unit 119 Unit 103 Unit 118 Unit 101

Table 3. Summary of Martlet Units and Sensors

CH_0

CH_1

SG1

SG2

SG3

SG4

SG5

SG6

SG7

SG8

CH_3 TS1 TS2 TS3 TS4

Installation Procedure To install the sensors, namely the strain gauges and the thermistors, a few keys steps had to be taken to ensure proper installation. First, the planned location for the strain gauges to be instrumented were marked. Next, using an electric belt sander, the corresponding surfaces were sanded down as needed. The surfaces were then thoroughly cleaned using alcohol and a cloth. The locations of the gauges were re-marked as needed. The appropriate epoxy was applied onto one end of the gauges and attached to the surface. To ensure proper bonding, a cutout piece of plexiglass was placed onto the gauges for protection, and clamps were used to apply pressure. The clamps and plexiglass were then removed once proper bonding had been ensured. Next, the temperature sensors were secured onto the steel members at the appropriate locations. Antennas were then
35

connected to the wireless sensing units and placed onto the steel members such that it maintained line of sight with the base station during data collection. Furthermore, all the sensors were connected to the appropriate sensor boards. The wireless sensing units were then placed in waterproof boxes, which were then screwed shut and securely attached onto the steel diaphragms. Finally, the necessary waterproofing was applied onto the strain gauges to ensure longevity during different weather conditions. Figure 45 illustrates some steps of the installation procedure. Note that, following data collection, all units were disassembled and taken back to the laboratory for inspection.
Figure 45. Photo. Collection of pictures depicting strain gauge installation process 36

Data Collection For the data collection, the laptop, base station, and antenna were set up below the bridge. Data was continuously sent wirelessly and plotted in MATLAB in addition to being saved to the secure digital (SD) card to ensure continuous successful receipt of the data. The data acquisition setup worked as intended and was deemed adequate for future data collection.
Figure 46. Photo. Setup for data acquisition BRIDGE MONITORING 2 The second monitoring effort monitored Span 14 of Bridge 11B of the I-16/I-75 Interchange project, shown in Figure 47 through Figure 51. The diaphragm was instrumented and monitored during the concrete deck pours to determine the strain induced in the steel diaphragms due to the weight of the concrete and other equipment during the pours. This span consists of nine 125-foot PSC girders and eight intermediate k-frame steel diaphragms located midspan, shown in Figure 52.
37

Figure 47. Photo. Steel reinforcement in deck prior to pour
Figure 48. Photo. End of span prior to concrete deck pour
Figure 49. Photo. Panoramic photo of span prior to deck pour 38

Figure 50. Drawings. Elevation view, Bridge 11B Figure 51. Drawings. Plan view, Bridge 11B
Figure 52. Drawings. Deck plan of Span 14, Bridge 11B 39

Two bays with steel k-frame diaphragms of the mentioned span were chosen for instrumentation, shown in Figure 53. More specifically, an intermediate diaphragm of the exterior bay of the span and an intermediate diaphragm of the interior bay of the span, were instrumented. The diaphragm consists of an L5x5x0.5 and two diagonal L3.5x3.5x0.5 structural steel sections connected to two plates shown in Figure 54.
Figure 53. Drawings. Four Half-Section of Span 14, Bridge 11B
Figure 54. Engineering Drawings. K-frame steel diaphragm 40

Figure 55 provides the instrumentation layout for the strain gauges and temperature sensors to be instrumented on the diaphragm. Each strain gauge is labeled "SG" followed by an arbitrary numeric identifier. Also, temperature sensors are labeled "TS" followed by an arbitrary numeric identifier. Eight strain gauges per diaphragm were installed at the locations shown, in addition to four temperature gauges per diaphragm for total of 16 strain gauges and 8 temperature sensors. Strain gauges were installed 20 inches from the exterior end of all members, at the locations on the cross sections shown in the left of Figure 55. The temperature gauges were fixed to the steel surface. Four Martlet units were used per diaphragm for a total of eight Martlet units. Each unit was connected to two strain gauges and one temperature sensor. The units were programmed with the latest version of the software code prior to installation. Details of the connections are shown in Table 4 and Table 5.

Figure 55. Schematic. Instrumentation layout and sensor numeric identifiers

Table 4. Martlet Units and Sensors used for External Diaphragm

CH_0

CH_1

CH_3

Unit 128

SG1

SG2

TS1

Unit 116

SG3

SG4

TS2

Unit 102

SG5

SG6

TS3

Unit 100

SG7

SG8

TS4

41

Table 5. of Martlet Units and Sensors used for Internal diaphragm

CH_0

CH_1

CH_3

Unit 119

SG1

SG2

TS1

Unit 125

SG3

SG4

TS2

Unit 118

SG5

SG6

TS3

Unit 101

SG7

SG8

TS4

The installation process was like that described for Span 27RT and is illustrated in Figure 56 through Figure 59. First, the surfaces were sanded and cleaned and then the gauges were bonded to the surface. Next, the gauges were weatherproofed. Finally, the wireless units were connected.

Figure 56. Photo. Cleaning and sanding of the surface (left), bonding of strain gauges (right)
42

Figure 57. Photo. Strain gauges after bonding to surface
Figure 58. Photo. Weatherproofing of strain gauges 43

Figure 59. Photo. Final product of strain gauges installation and connection to wireless sensing units for Bay 1 (left) and Bay 3 (right)
Following the installation, the data acquisition systems were tested by collecting ambient data to ensure proper communication and transfer of data between the wireless sensing units and the base station. The batteries were collected after installation and taken to the laboratory to ensure they were fully charged for the day of the pour. The deck was poured on August 26, 2020, at approximately 4:00 a.m. The trucks just prior to the pour time are shown in Figure 60. The fully charged batteries were reconnected to the units, which were turned on and ready to collect data at the beginning of the concrete pour.
44

Figure 60. Photo. Deck pour on August 26, 2020 The base station (Figure 61), laptop, and antenna were setup under the bridge, maintaining line of site with the wireless sensing units. Data was continuously collected during the concrete pour and was uninterruptedly sent wirelessly to the base station over the entire duration of the data acquisition. The concrete pour ended around 8:00 a.m. ET. Data collection stopped a few hours after the concrete pouring was completed. In addition to data collection, two time-lapse cameras were set up on both sides of the bridge to capture the events on top of the of the bridge, shown in Figure 62. The footage was reexamined in conjunction with the collected strain data.
45

Figure 61. Photo. Base station setup during data collection
Figure 62. Photo. Time-lapse camera footage on both sides of the bridge Data Analysis The raw data, which was collected in volts, was appropriately converted to the physical quantity of strain and temperature. Moreover, a smoothing filter was applied to better visualize the trend of the collected data. More specifically, the "rloess" function, a more robust version of the "loess" filter assigning lower weights to outliers in the regression, was used for that purpose. This filter performs local regression using weighted linear least squared and a second-degree polynomial model to provide a filtered version of the raw data and reduce the noise, as shown in Figure 63.
46

Figure 63. Graph. Raw data (black) versus filtered data (Red) The filtered data from the concrete deck pouring process is summarized in Figure 64 through Figure 67. Unfiltered data can be found in Appendix D. Note that three strain gauges were damaged and did not collect any meaningful data. The strain entries for these gauges are represented by an `X' in the table.
47

Bay 1 (between external and internal girders)

Left Diagonal Member Strain Gauge 5

)

Strain (

WSU 102 CH1 50

40

30

20

10

0 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Right Diagonal Member Strain Gauge 7

Left Diagonal Member Strain Gauge 6

WSU 102 CH2 20

0

-20

-40

-60

-80

-100

-120 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Right Diagonal Member Strain Gauge 8

)

Strain (

WSU 100 CH2 60

50

40

30

20

10

0

-10 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Figure 64. Data. Strain gauge data from top of Bay 1

)

Strain (

48

Bay 1 (between external and internal girders)

Bottom Member (Left) Strain Gauge 1

)

Strain (

WSU 128 CH1 20

0

-20

-40

-60

-80 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Bottom Member (Right) Strain Gauge 3

)

Strain (

WSU 116 CH1 80

60

40

20

0

-20 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Bottom Member (Left) Strain Gauge 2

Bottom Member (Right) Strain Gauge 4

)

Strain (

WSU 116 CH2 0

-5

-10

-15

-20

-25

-30 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Figure 65. Data. Strain gauge data from bottom of Bay 1

49

Bay 3 (between two internal diaphragms)

Left Diagonal Member Strain Gauge 5

)

Strain (

WSU 118 CH1 40

30

20

10

0

-10 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Right Diagonal Member Strain Gauge 7

)

Strain (

WSU 101 CH1 30

20

10

0

-10

-20

-30 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Left Diagonal Member Strain Gauge 6

WSU 118 CH2 20

0

-20

-40

-60

-80

-100 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Right Diagonal Member Strain Gauge 8

Figure 66. Data. Strain gauge data from top of Bay 3

)

Strain (

50

Bay 3 (between two internal diaphragms)

Bottom Member (Left) Strain Gauge 1

)

Strain (

WSU 119 CH1 50

40

30

20

10

0

-10

-20 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Bottom Member (Right) Strain Gauge 3

)

Strain (

WSU 103 CH1 40

30

20

10

0

-10

-20

-30 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Bottom Member (Left) Strain Gauge 2

)

Strain (

WSU 119 CH2 20

0

-20

-40

-60

-80 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Bottom Member (Right) Strain Gauge 4

)

Strain (

WSU 103 CH2 5

0

-5

-10

-15

-20

-25 08:00

08:30

09:00

09:30

10:00

10:30

11:00

Time UTC

11:30

12:00

12:30

13:00

Aug 26, 2020

13:30

Figure 67. Data. Strain gauge data from bottom of Bay 3

51

CHAPTER 6. FINITE ELEMENT MODELS
To better understand how the loads are distributed within the diaphragm elements, three models were created with three different software: a SAP2000 Model, a CSIBridge model, and an Abaqus model. This chapter summarizes the results from the models and provides comparisons with the monitoring effort.
SAP2000 MODEL A simplified model of the k-frame diaphragm of Span 14 of Bridge 11B was constructed using the commercial finite element program SAP2000.(23) The purpose behind the model was to provide a preliminary analysis of the internal forces and moments in the diaphragm when subjected to concentrated wind forces. The model was also used to reproduce the calculations provided by GDOT using the finite-element model software RISA.
The SAP model consisted of the structural steel angle members, where the effective lengths of the members were taken from the innermost slotted bolt holes. The effective lengths are represented by the red lines in Figure 68. The support conditions were modeled as pin supports at one end of the diaphragm members and roller supports at the opposite end. Wind load calculations for this specific span were performed, and the load was applied to the model as a concentrated force at one end of the top diagonal member and the bottom chord member. A linear analysis was run with the applied concentrated forces. As a result, the reactions at the supports were calculated and are shown in Figure
52

69. Additionally, the resulting stresses and internal forces in each member were obtained (Figure 70). The values obtained matched those provided by GDOT.
Figure 68. Schematic. Effective length used in SAP2000 model
Figure 69. Schematics. SAP2000 K-frame diaphragm reaction forces
Figure 70. Schematics. SAP2000 analysis axial forces 53

CSIBRIDGE MODEL The commercial finite element software CSIBridge was used to construct an initial model of Span 14 of Bridge11B of the I-16 I-75 Interchange that was to be instrumented.(23) This model consisted of the prestressed reinforced concrete girders, the concrete end diaphragms, and the intermediate steel diaphragms that are the focus of this research. These elements are shown in Figure 71, Figure 72, and Figure 73.
Figure 71. Model. CSIBridge model of Span 14 of Bridge11B of the I-16 I-75 Interchange project with prestressed girders
Figure 72. Model. Concrete end diaphragms and supports in CSIBridge model 54

Figure 73. Model. K-frame intermediate steel diaphragms in CSIBridge model The appropriate member sections and constitutive properties obtained from the provided drawing were used to model the intermediate k-frame diaphragms as well as the reinforced concrete Bulb-Tee girders and the prestressed tendons, shown in Figure 74 and Figure 75. The fixed-expansion support was used to model the end of the spans, as indicated in the drawings in Figure 76.
Figure 74. Engineering Drawing. Bulb-Tee section at midpoint and end 55

Figure 75. Drawing. Prestressed concrete girder tendons
Figure 76. Drawing. Fixed-expansion end supports of Span 14 The tendons were modeled as object elements, using the Bridge Tendon Wizard in CSIBridge. The material property was defined appropriately as A416 Grade 27 steel. The tendon area, load, and layout were adequately calculated and modeled based on the information provided in the shop drawings provided by GDOT. The elevation, plan, and section drawings of the final bridge tendon layout display are shown in Figure 77. Figure 78 shows the modeled tendon objects in green for each of the Bulb-Tee reinforced concrete beams.
56

Figure 77. Interface. Elevation, plan, and section drawings of the final bridge tendon layout display
Figure 78. Model. Model tendon objects (in green) 57

An investigation of different model variables was performed to determine the impact of these variables on the model. Variables included end supports (abutments and bearing properties), diaphragm frame joint offset, tendons modeling, and boundary conditions for the diaphragm members. The initial model was reviewed and updated based on the study findings. The major changes to the model are delineated in what follows.
The boundary conditions of the diaphragms were examined. CSIBridge allows the user to model steel intermediate diaphragms at the desired location and with the desired sectional properties in reinforced concrete girder bridges. By default, CSIBridge releases moments at both ends and torsion at one end of the members of the diaphragm when modeling a kframe steel diaphragm in a concrete girder bridge. Investigation of the internal forces of the members of the diaphragms under different loading conditions, namely dead load and staged construction, showed that this method did not provide adequate and meaningful results. Therefore, the releases were deemed inadequate and had to be updated for better and more representative results. Consequently, rotational springs were added at the diagonal and bottom chord frame diaphragms ends with some stiffness, shown in Figure 79, and the analysis was rerun to show improved results.
58

Figure 79. Model. Changes to end releases of steel diaphragms in CSIBridge The end supports of the span were modeled as abutment links, with specified values for substructure location elevation as well as bearing assignment elevation at layout line. According to CSIBridge documentation, Substructure Location, Elevation (Global Z) is the bearing seat elevation, or the elevation at the top of the bent cap or abutment cap; and Bearing Assignment, Elevation at Layout Line (Global Z) is the elevation at the bearing action point. Preliminary values were obtained from the abutment section drawings provided by GDOT (Figure 80). The bearing and substructure elevation in CSIBridge, with values, is given in Figure 81.
Analysis of the model was performed for different values of substructure location elevation as well as bearing assignment elevation and lead to negligible impact on the analysis results, and hence this variable was deemed to have little to no significance on the analysis.
59

Figure 80. Drawing. Abutment drawings for Span 14 of Bridge 11B
Figure 81. Model. Bearing and substructure elevation in CSIBridge Another important modification to the steel diaphragms modeled in CSIBridge was to add frame joint offsets to the diagonal members to better model the diaphragms. Examining the drawings provided by GDOT suggests that the diaphragms are connected to a member at a certain distance from the top of the beam (Figure 82). Consequently, frame joint offsets of cardinal points were added to the diagonal members of the
60

diaphragm (Figure 83), to better model the in-situ design of the diaphragms. The updated model was then used to run a multitude of tests for wind load analysis as well as nonlinear staged construction analysis. The next section describes the methodology and the subsequent results.
Figure 82. Drawing. Partial section at intermediate diaphragm of Span 14 of Bridge 11B
Figure 83. Model. Frame joint offset location for diagonal members of the diaphragm 61

Wind Load Analysis A wind load pattern was defined in CSIBridge using an AASHTO2018 auto lateral load patten. An automatic AASHTO 2018 lateral wind load pattern was created with the aim of running the analysis and comparing the stresses in the members of the diaphragms with those obtained in the previous section. For the defined automatic load pattern, the wind load is calculated on the substructure, superstructure, and on live load if present. The defined parameters are specified according to the code chosen, and the wind forces are described as function of the exposed areas and the height. A multistep linear static load case analysis is automatically defined for the AAHSTO 2018 automatic wind load pattern. For the analysis, the wind load is applied at a multitude of different angles to the transverse direction of the bridge. Consequently, the different angles are analyzed as a multistep linear load case. According to the CSIBridge documentation, multistepped load patterns represent several separate and independent loading patterns applied in sequence. Multistepped load patterns can be applied in a multistep static load case, which performs a series of independent linear analysis of the defined load patterns. The analysis, therefore, resulted in six steps, with each step representing an independent linear analysis. The subsequent internal forces and strains in the diaphragms members were examined. Examples of the output is given in Figure 84 and Figure 85 for stress in the 1-1 direction. Additional results are provided in the Comparisons section at the end of this chapter.
62

Figure 84. Model. Stress S11 in elements from step 1 of the multistep linear wind load analysis
Figure 85. Model. Stress S11 in elements from step 2 of the multistep linear wind load analysis
Staged Construction Analysis A nonlinear staged construction analysis was performed to model the concrete deck pouring process. This type of analysis allows the user to define multiple stages, with a sequence of operations for each stage. Each operation can include different object types, which can be added and removed as needed to mimic the construction process. Eight stages were defined in the concrete pouring load case. Guides for the top slab are defined in the first stage of the concrete pouring load case. Defining guides is important to correctly position the object when added to a deformed structure. This step will allow the slab object, when added, to follow the deformed shape of the girders. The second stage used the "add structure" operation to add the girders as well as the concrete end diaphragms and the intermediate steel diaphragms, and "load object" operation to load the mentioned added objects with their self-weight. Next, five concrete pours were defined in the slab wet concrete load assignment to model five different parts of the
63

pouring process and obtain a time history of the loads under the concrete pouring process. Wet concrete allows the user to model the concrete before it hardens and reaches its full stiffness. Consequently, this load applies only the weight of the concert onto the girders without applying any of their stiffness, as would be the case when the concrete is poured in situ. During the concrete deck pouring process, the slab should have no composite action until the concrete hardens and cures. The pour concrete and remove operations offer a convenient way to model the concrete deck pouring process before and after the concrete cures and hardens. These operations are used in the subsequent steps. Stages 3 to 7 use the "pour concrete operation" along with the wet concrete load defined to model the continuous on-site pouring of wet concrete in five different stages. The pour concrete operation adds in the weight of the concrete pour using the equivalent point and bracket load based on the tributary width for each girder. Finally, the "remove pour" operation is used for all five wet concrete loads applied. The slab is now treated as a structural object and not just a load, as would be the case when the wet concrete cures and hardens. Therefore, stages 3 to 7 model the data collection during the concrete deck pouring process and are used in the comparisons at the end of the chapter.
CSIBridge Data Processing At each step in the analysis, the internal forces, including the axial forces and the moments, were exported at different stations of the discretized members of the steel diaphragms. A MATLAB script was written to import the mentioned internal forces and calculate the corresponding axial strain at the location of the instrumented sensors (Figure 86). The variation of the strain as a function of the concrete pour stages is shown in Figure 87 through Figure 90.
64

Figure 86. Sketch. Strain gauge location on angle Bay 1 (between external and internal girders)

Strain (micro)

Strain (micro)

0

-0.2

-0.4

-0.6

-0.8

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Strain (micro)

5

4

3

2

1

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

0.5

0.4

0.3

0.2

0.1

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Strain (micro)

8

6

4

2

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Figure 87. Data. Strain gauge data from top of Bay 1

65

Strain (micro)

Bay 1 (between external and internal girders)

0

-5

-10

-15

-20

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Strain (micro)

0

-5

-10

-15

-20

-25

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

20

15

10

5

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Strain (micro)

15

10

5

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Figure 88. Data. Strain gauge data from bottom of Bay 1

Strain (micro)

66

Strain (micro)

Bay 3 (between two internal diaphragms)

0.5

0.4

0.3

0.2

0.1

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Strain (micro)

8

6

4

2

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

5

4

3

2

1

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Strain (micro)

0

-0.2

-0.4

-0.6

-0.8

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Figure 89. Data. Strain gauge data from top of Bay 3

Strain (micro)

67

Bay 3 (between two internal diaphragms)

Strain (micro)

15

10

5

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

20

15

10

5

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Strain (micro)

15

10

5

0

1

1.5

2

2.5

3

3.5

4

4.5

5

Concrete Pour Stage

Figure 90. Data. Strain gauge data from bottom of Bay 3

Strain (micro)

68

ABAQUS MODEL 1: LABORATORY SIMULATION To gain a better understanding of the stress distribution in the steel diaphragms, a detailed finite element model of the same scale as the in-situ diaphragm tested was constructed in the commercial finite element software Abaqus.(24) The model consists of 284,000 total nodes and 215,000 linear hexahedral elements, each of type C3D8R. The model is shown in Figure 91 through Figure 93. A fine mesh was chosen such that there are three elements per thickness as shown in Figure 93 to avoid problems with aspect ratio and element bending.
Figure 91. Model. Front view of k-frame diaphragm Abaqus model
Figure 92. Model. Back view of k-frame diaphragm Abaqus model 69

Figure 93. Model. 3D Isometric view of k-frame diaphragm Abaqus model
Figure 94. Model. Mesh detail of k-frame diaphragm To model the interaction between the members of the diaphragm, different constraints were used. Weld constraints were used as interaction between the members and the middle gusset plate for a total of three weld constraints. Additionally, surface-to-surface constraints were used to model the interaction between any two other members in contact. The different interactions are shown in Figure 95.
70

Figure 95. Model. Interactions between members of the k-frame The bolts connecting the different members were modeled as rigid body cylinders. To better model the bolt behavior, two steps were defined in the job wizard in Abaqus, a contact step followed by a loading step. A reference point was created for each bolt and was linked to the corresponding bolt. Moreover, a no-slip boundary condition was used for the pins, whereby the reference points linked to the bolts were restrained in the contact step (Figure 96). Additionally, a node to surface contact interaction was modeled between the outer surface of the pins and the inner surface of the bolt holes, shown in Figure 97.
Figure 96. Model. Bolt reference points and no-slip constraints 71

Figure 97. Model. Bolt node to surface contact interaction The model was used to determine the maximum load to which the diaphragm can be subjected during the same-scale laboratory testing process described earlier. Consequently, the boundary conditions of the diaphragm were modified to match the laboratory test setup. For that purpose, one end of the diaphragm was modeled as fixed, where the rotation and translation were restricted in all directions. The other side of the diaphragm was restrained in the x-direction. A distributed load was applied on the latter end of the diaphragm, and the corresponding strain was examined. It was determined that an appropriate load to use for the test without causing any permanent damage was 10 kips, which was what was executed in the laboratory.
ABAQUS MODEL 2: FIELD SIMULATION To gain a better understanding into the impact of the construction loads on the k-frame steel diaphragms, additional changes were made to the model described in the previous section to make it more representative of the diaphragm in the field. The concrete girders were modeled in Abaqus to explore the load distribution during the concrete pouring. Since the focus was understanding the strain distribution in the steel k-frame diaphragm,
72

simplified versions of the prestressed reinforced concrete girders were modeled and are shown in Figure 98.
Figure 98. Model. Concrete girders in Abaqus in undeformed (left) and deformed (right) states
The cross section and length of the girders were accurately modeled based on the girder drawings for Span 14 of Bridge 11B. As mentioned, only a simplified version of the girder was modeled, and therefore the prestressing was not taken into consideration. Furthermore, the girder was given an elastic constitutive model with a Young's modulus equal to that of reinforced concrete. To model the boundary conditions, a fixed-expansion model was adopted, whereby one end of the girders had translation constraints in all directions, and the other had translation constraints in the y- and z- directions. The side angles of the diaphragm were connected to the inside of the girders using a tie constraint with the master being the outer side of the angle and the slave being the inner side of the girders, as shown in Figure 99. A tie constraint ties together two separate objects such that there is no relative motion between them and allows two regions with dissimilar meshes to be tied together.
73

Figure 99. Model. Tie constraint between the steel diaphragm and concrete girders As mentioned, the model was used to study the strain in the diaphragms caused by the construction loads. This was attained by applying construction loads on the beams to view the resulting effects on the k-frame. For that purpose, the load to be applied was calculated by estimating the weight of concrete needed for the deck pouring.
The deck and beam cross section drawings were used to calculate the weight of the concrete during the concrete deck pouring. Using the dimensions of the deck, the volume of concrete needed was calculated. Moreover, the weight of the concrete was estimated by multiplying the volume with the weight density of concrete. Furthermore, using the tributary width, the loads carried by each girder was then found. Finally, using the surface area of the top of the beams, the distributed load was calculated and used for the simulations. Detailed calculations are given in Appendix E. The pressure caused by the concrete was found to be 1.24 lb/in2. For that purpose, the beams were partitioned into four parts, with the pressure applied on each quarter representing four stages of concrete deck pour, shown in Figure 100. Strain results of these simulations are calculated and shown in Figure 101 and Figure 102.
74

Figure 100. Model. Concrete pour simulation by quarters.

Strain (micro)

0

-2

-4

-6

-8

1

1.5

2

2.5

3

3.5

4

Concrete Pour Stage

Strain (micro)

-1

-2

-3

-4

-5

1

1.5

2

2.5

3

3.5

4

Concrete Pour Stage

Strain (micro)

0

-2

-4

-6

1

1.5

2

2.5

3

3.5

4

Concrete Pour Stage

Strain (micro)

0

-2

-4

-6

1

1.5

2

2.5

3

3.5

4

Concrete Pour Stage

Figure 101. Data. Diagonal member strain results from Abaqus simulation for concrete only

75

Strain (micro)

-1

-2

-3

-4

-5

-6

1

1.5

2

2.5

3

3.5

4

Concrete Pour Stage

Strain (micro)

-2

-3

-4

-5

-6

1

1.5

2

2.5

3

3.5

4

Concrete Pour Stage

Strain (micro)

0.2

0.1

0

-0.1

1

1.5

2

2.5

3

3.5

4

Concrete Pour Stage

Strain (micro)

0.2

0

-0.2

-0.4

-0.6

1

1.5

2

2.5

3

3.5

4

Concrete Pour Stage

Figure 102. Data. Bottom chord strain results from Abaqus simulation for concrete only

During the examination of the time-lapse footage of the concrete deck pouring, a major construction load identified was the concrete pavement machine, shown in Figure 103. Certain models of these concrete pavement machine weigh up to 20,000 lbs. Consequently, to account for this significant weight, a concentrated force of varying magnitude was added to the mid span of the beams in the Abaqus model, in addition to the distributed load explained previously, as shown in Figure 104.

76

Figure 103. Photo. Identification of the concrete pavement machine as a major construction load from time-lapse camera
Figure 104. Model. Simulation of halfway point of concrete pouring process: concentrated load at midspan and distributed load
The distributed load was placed onto half of the beam length to simulate half of the concrete pour. The aim was to aim to simulate the steel k-frame diaphragm strain during the halfway point of the concrete deck pouring process. This would consequently mean that half of the concrete deck had been poured and that the concrete paving machine is at
77

the midspan, or directly on top of the diaphragms. Results for this simulation are detailed
in the next section. The analysis is run for different magnitude values of the concentrated
load modeling the weight of the concrete pavement machine. For all the different load
cases, the nodal displacement outputs were exported at two adjacent nodes close to the
location of the instrumented strain gauges. Coordinates at the location of the
instrumented strain gauge were also exported and used for the calculations. These values
were then imported into MATLAB. A MATLAB script was written to read the initial and
final nodal displacement values at certain locations and calculated the change in
displacement after loading and consequently the induced strain. The strain (in )
obtained at different locations for all members for different values of the concentrated
force applied are summarized in Table 6.
Table 6. Summary of micro-strain values for varying load cases in Abaqus
Load Case: Distributed Distributed + Distributed + Distributed + Distributed + 5,000lb Force 7,000lb Force 10,000lb Force 7,000lb Force + (on each side) (on each side) (on each side) 8,000lb Force (one on each side)
Strain Gauge 1 -5.034943936 -5.602265686 -5.830869055 -6.179215768 -6.408809117 Strain Gauge 2 0.016107379 -0.020134246 -0.015995304 -0.006298190 -0.000849146 Strain Gauge 3 -5.554543038 -6.165171939 -6.406660206 -6.777852824 -7.022282063 Strain Gauge 4 -0.448738082 -0.547207164 -0.580249465 -0.619059108 -0.647317706 Strain Gauge 5 -5.072218373 -6.208841369 -4.309247012 4.3071912585 -5.211049619 Strain Gauge 6 -3.619596354 -4.732260319 -6.724004692 -5.612233975 -7.632418176 Strain Gauge 7 -4.876654816 -5.339982398 -3.182135119 -5.540292969 -6.208876527 Strain Gauge 8 -4.291231245 -7.606567976 -6.615146514 -5.185465123 -6.974399311
COMPARISONS
Table 7 provides a comparison of the strain values at each of the strain gauge locations.
The comparisons include values for both construction loads and wind loads with all the
methods developed. The maximum value of strain for each element type (e.g., diagonal)
78

is in bold. From the table, it is evident that there is a wide variation in the strains between

the analysis methods. It is important to note that while the field data was recorded during

the concrete deck pouring process specifically, the strains obtained could encompass

wind loads in addition to the concrete deck pouring and equipment. Generally, the

magnitudes of strains recorded during field monitoring are between those provided from

the CSIBridge staged construction analysis and CSIBridge wind load analysis.

Additionally, it is also evident that the maximum strains are all caused by the wind

condition as opposed to the construction loading, in terms of the locations monitored.

Table 7. Comparison of maximum (absolute value) micro-strain

Field Monitoring

Construction Loads

CSIBridge

Abaqus

Wind Loads CSIBridge SAP2000

Strain Gauge 1 Strain Gauge 2 Strain Gauge 3 Strain Gauge 4 Strain Gauge 5 Strain Gauge 6 Strain Gauge 7 Strain Gauge 8

76.003 68.610 34.264 26.720 49.879 102.367 22.8455 51.884

36.083 16.231 20.624 25.763 37.399 39.410 33.083 31.946

6.179 0.020 7.022 0.647 6.209 7.632 6.209 7.607

83.763 144.074 63.625 64.454 93.887 96.809 94.745 94.840

107.008 107.238 68.058 68.058 58.112 56.723 136.764 134.430

79

CHAPTER 7. CONCLUSIONS
This research project quantified the effects of construction loads on steel k-frame diaphragms using a combined field monitoring and modeling effort. The main conclusions and recommendations from the research project are as follows:
1. The use of commercial software to model construction loads produced widelyvarying strain values for the locations documented in this research effort. It is recommended that engineers use this software with care when specifying boundary and loading conditions, in particular.
2. By comparing the strain values from construction loads determined from various methods with those caused by wind load, it was determined that the wind load is the governing load case. Current AASHTO guidance to design diaphragms using the wind loading condition was verified by this research, at least in terms of the diagonals and chords considered. Therefore, the wind loading case is sufficient to design these elements.
3. Because the bridge was not monitored during the wind load condition, no data is available for some of the diaphragm components. Additional test(s) should be conducted to verify the behavior of the gusset plate, in particular, when subjected to the wind load condition.
80

APPENDICES 81

APPENDIX A: DIAPHRAGM CALCULATIONS 82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

CAMERA 1:

APPENDIX B: TIME-LAPSE FIGURES

117

118

119

120

121

CAMERA 2: 122

123

124

125

126

APPENDIX C: MATLAB CODE

function [data_out] = Plot_DAQ_Results_Martlet_3(folder_num)

% Plot Narada and Martlet DAQ Data % **Caution: this code assumes that Unit # of Martlet is larger than 100** %close all;

Gain = 4; Vex = 3.3; GF = 2.05; v = 0.3 ;

if(nargin==1) if(strcmp(folder_num,'last')) % Plot the last result dirc = dir('.\DAQResults\'); [A,I] = max([dirc(:).datenum]); if ~isempty(I) run_num = dirc(I).name; end else run_num = folder_num; end
else % get folder number from user run_num = input('Enter Folder Name: ');
end

% if run_num >= 100000

%

error('badInput:unusable:tooLarge','%s','Input is too

large!');

% end

% load parameters from .txt file path_base = sprintf('.\\DAQResults\\%s\\',run_num); path = [path_base 'TestName.txt']; [DAQset] = load_DAQ_settings_Martlet(path);

% Find actual points collected points1 = DAQset.fs * DAQset.T; points2 = DAQset.points_per_poll; if points2 > points1
points = points1; elseif mod(points1,points2) ~= 0

127

points = (floor(points1/points2)+1)*points2; else
points = points1; end num_poll_cycles = ceil(points1/points2);
% Preallocate the memory data = zeros(DAQset.num_units, max(max(DAQset.channel_num_list(:,:))), num_poll_cycles*DAQset.points_per_poll);
% Load the data: time = 1/DAQset.fs*[1:points]'; for k = 1:DAQset.num_units
chan = DAQset.channel_num_list(k,:); for j = 1:chan
if (DAQset.chans(k,j,1) == 65 && DAQset.chans(k,j,2) == 49)
filename = [path_base sprintf('U%02d_ADC_A1', DAQset.unit_list(k,1))];
elseif(DAQset.chans(k,j,1) == 65 && DAQset.chans(k,j,2) == 50)
filename = [path_base sprintf('U%02d_ADC_A2', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 65 && DAQset.chans(k,j,2) == 52));
filename = [path_base sprintf('U%02d_ADC_A4', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 65 && DAQset.chans(k,j,2) == 53));
filename = [path_base sprintf('U%02d_ADC_A5', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 65 && DAQset.chans(k,j,2) == 54));
filename = [path_base sprintf('U%02d_ADC_A6', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 66 && DAQset.chans(k,j,2) == 48));
filename = [path_base sprintf('U%02d_ADC_B0', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 66 && DAQset.chans(k,j,2) == 49));
filename = [path_base sprintf('U%02d_ADC_B1', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 66 && DAQset.chans(k,j,2) == 50));
128

filename = [path_base sprintf('U%02d_ADC_B2', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 66 && DAQset.chans(k,j,2) == 52));
filename = [path_base sprintf('U%02d_ADC_B4', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 66 && DAQset.chans(k,j,2) == 53));
filename = [path_base sprintf('U%02d_ADC_B5', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 66 && DAQset.chans(k,j,2) == 54));
filename = [path_base sprintf('U%02d_ADC_B6', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 72 && DAQset.chans(k,j,2) == 49));
filename = [path_base sprintf('U%02d_EXTADC_CH1', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 72 && DAQset.chans(k,j,2) == 50));
filename = [path_base sprintf('U%02d_EXTADC_CH2', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 72 && DAQset.chans(k,j,2) == 51));
filename = [path_base sprintf('U%02d_EXTADC_CH3', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 72 && DAQset.chans(k,j,2) == 52));
filename = [path_base sprintf('U%02d_EXTADC_CH4', DAQset.unit_list(k,1))];
end

tempdata = []; ppp = DAQset.points_per_poll; for i = 1:num_poll_cycles
filename_i = [filename '_' num2str(i,'%05d') '.dat'];
if DAQset.chans(k,j,1) == 72 tempdata((i-1)*ppp*2+1:i*ppp*2,1) =
load(filename_i); else tempdata((i-1)*ppp+1:i*ppp,1) =
load(filename_i); end
end

%

for i=1:length(tempdata)

129

if DAQset.chans(k,j,1) == 72 if DAQset.chans(k,j,2) == 52 for data_i = 1:length(tempdata)/2 data_volt(data_i) =
bitshift(tempdata((data_i-1)*2+1),16) + tempdata(data_i*2); data_volt(data_i) =
typecast(uint32(data_volt(data_i)),'int32'); data_volt(data_i) =
data_volt(data_i)*2.442/2^31; end
else for data_i = 1:length(tempdata)/2 data_volt(data_i) =
bitshift(tempdata((data_i-1)*2+1),16) + tempdata(data_i*2); data_volt(data_i) =
typecast(uint32(data_volt(data_i)),'int32'); data_volt(data_i) =
data_volt(data_i)*2.442/2^31/1.084/1.759/Gain; end
end

if DAQset.chans(k,j,1) == 72 for data_i = 2:length(data_volt) if abs(data_volt(data_i) -
data_volt(data_i-1)) > 0.3 data_volt(data_i) =
data_volt(data_i-1); end
end end

if DAQset.chans(k,j,2) == 52 data_volt(1) = data_volt(2); data_v = -data_volt; data_tmp = 5.8145*data_v.^3 +
3.5922*data_v.^2 + 30.245*data_v + 16.111; else

for (i=1:length(data_volt)) data_str(1,i) =
data_volt(1,i)*2/Vex/GF/(1+v - data_volt(1,i)/Vex*(1v))*10^6; %mod by N
end

% to delete the spikes if needed.

%

for i=1:5

130

%

[~,b] = max(data_str);

%

if b>1

data_str =(data_str - mean(data_str));

data_str = data_str - data_str(1,1);

%

end

else

data_volt(data_i) = tempdat;

for data_i=1:length(tempdata)a(data_i,1) * 3.3

/ 4095;

end

data_str = data_volt*15100;

data_str =(data_str - mean(data_str));

data_str =(data_str - mean(data_str))-

(data_str(:,1) - mean(data_str));

end

if DAQset.chans(k,j,1) == 72

if DAQset.chans(k,j,2) == 52 figHand = figure; set (figHand, 'Position',[200 200 600
200]); plot(time, data_v) xlabel('Tims(s)'); ylabel(['Voltage (V)']); display(['Mean value V: '
num2str(mean(data_v))]); failn = find(data_v == 0); display(['Failure percentage: '
num2str(length(failn)/length(data_v)*100) '%']); else figHand = figure; set (figHand, 'Position',[200 200 600
200]); plot(time, data_volt) xlabel('Tims(s)'); ylabel(['Voltage (V)']); display(['Mean value V: '
num2str(mean(data_volt))]); failn = find(data_volt == 0); display(['Failure percentage: '
num2str(length(failn)/length(data_volt)*100) '%']); end

if DAQset.chans(k,j,2) == 52 figHand = figure;

131

set (figHand, 'Position',[200 200 600

200]);

plot(time, data_tmp)

title('thermistor 2')

xlabel('time(s)');

ylabel(['temperature (\circ C)']);

display(['Mean value: '

num2str(mean(data_tmp))]);

display(['Noise level: '

num2str(std(data_tmp))]);

else

figHand = figure;

set (figHand, 'Position',[200 200 600

200]);

plot(time, data_str)

xlabel('Tims(s)');

ylabel(['Strain (\mu\epsilon)']);

%

ylim([-100 5])

legend ('martlet')

%

ylabel(['Acc (g)']);

display(['Mean value: '

num2str(mean(data_str))]);

display(['Noise level: '

num2str(std(data_str))]);

end

else figHand = figure; set (figHand, 'Position',[200 200 600 200]); plot(time, data_volt) xlabel('Tims(s)'); ylabel(['Voltage (V)']); display(['Mean value V: '
num2str(mean(data_volt))]); failn = find(data_volt == 0); display(['Failure percentage: '
num2str(length(failn)/length(data_volt)*100) '%']);

figHand = figure; set (figHand, 'Position',[200 200 600 200]);

plot(time, data_str)

ylim([-110 5])

xlabel('Tims(s)');

ylabel(['Strain (\mu\epsilon)']);

%

ylabel(['Acc (g)']);

132

display(['Mean value: ' num2str(mean(data_str))]);
display(['Noise level: ' num2str(std(data_str))]);
end
end
end end
function [DAQset] = load_DAQ_settings_Martlet(path) % Use this function to automaticlly load the DAQ settings for a Narada DAQ % run using the automatically generated .txt file.
fid = fopen([path]); tline = fgets(fid); tline = fgets(fid); tline = fgets(fid); tline = fgets(fid); tline = fgets(fid); DAQset.PCTime = sscanf(tline, '\t%s'); %get name tline = fgets(fid); tline = fgets(fid); [DAQset.timestamp] = sscanf(tline, '\t%s'); %get timestamp tline = fgets(fid); tline = fgets(fid); tline = fgets(fid); tline = fgets(fid); DAQset.fs = sscanf(tline, '\t\t%d Hz'); % get sample rate tline = fgets(fid); tline = fgets(fid); DAQset.T = sscanf(tline, '\t\t%d seconds'); % get number of seconds tline = fgets(fid); tline = fgets(fid); DAQset.points_per_poll = sscanf(tline, '\t\t%d samples'); % get points per polling cycle tline = fgets(fid); tline = fgets(fid); tline = fgets(fid);
133

DAQset.num_units = sscanf(tline, '\t- %d %*s'); % get points per polling cycle %tline = fgets(fid) DAQset.unit_list = zeros(DAQset.num_units,1);
tline = fgets(fid); for k = 1:DAQset.num_units
temp1 = sscanf(tline,'\t\t- %*s %d'); % flush junk DAQset.unit_list(k,1) = temp1; num_chan = sscanf(tline,'\t\t- %*s %*d %*s %*s %*s %d'); % flush junk %Assemble Channel Lists:
for kk = 1 : num_chan
tline = fgets(fid); if tline == -1
break ; elseif tline(3)=='-'
break ; end
if tline(6) == 'E' temp = tline(14:15); eval(['chans' int2str(k) '(' int2str(kk) ',:)'
'=' 'temp(1:2)' ';']); % it is only for Narada. else temp = tline(10:11); eval(['chans' int2str(k) '(' int2str(kk) ',:)'
'=' 'temp(1:2)' ';']); % it is only for Narada. end
end tline = fgets(fid); end fclose(fid);
DAQset.channel_num_list = zeros(DAQset.num_units,1); for k = 1:DAQset.num_units
temp = eval(['size(chans' int2str(k) ');']); DAQset.channel_num_list(k,1) = temp(1); end DAQset.chans = zeros(DAQset.num_units,max(DAQset.channel_num_list),2); for k = 1:DAQset.num_units for j = 1:DAQset.channel_num_list(k,1)
eval(['DAQset.chans(k,j,:) = chans' int2str(k) '(j,:);']);
134

end end

function data = lvm_import(filename,verbose)

%LVM_IMPORT Imports data from a LabView LVM file

% DATA = LVM_IMPORT(FILENAME,VERBOSE) returns the data from

a LVM (.lvm)

% ASCII text file created by LabView.

%

% FILENAME The name of the .lvm file, with or without

".lvm" extension

%

% VERBOSE

How many messages to display. Default is 1

(few messages),

%

0 = silent, 2 = display file information and

all messages

%

% DATA

The data found in the LVM file. DATA is a

structure with

%

fields corresponding to the Segments in the

file (see below)

%

and LVM file header information.

%

%

% This function imports data from a text-formatted LabView

Measurement File

% (LVM, extension ".lvm") into MATLAB. A LVM file can have

multiple

% Segments, so that multiple measurements can be combined

in a single

% file. The output variable DATA is a structure with

fields named

% 'Segment1', 'Segment2', etc. Each Segment field is a

structure with

% details about the data in the Segment and the actual

data in the field

% named 'data'. The column labels and units are stored as

cell arrays that

% correspond to the columns in the array of data.

% The size of the data array depends on the type of x-axis

data that is

% stored in the LVM file and the number of channels

(num_channels).

% There are three cases:

% 1) No x-data is included in the file ('No')

% The data array will have num_channels columns (one

column per channel

135

% of data).

% 2) One column of x-data is included in the file ('One')

% The first column of the data array will be the x-

values, and the data

% array will have num_channels+1 columns.

% 3) Each channel has its own x-data ('Multi')

% Each channel has two columns, one for x-values, and one

for data. The

% data array will have num_channels*2 columns, with the

x-values and

% corresponding data in alternating columns. For example,

in a Segment

% with 4 channels, columns 1,3,5,7 will be the x-values

for the data in

% columns 2,4,6,8.

%

% Note: because MATLAB only works with a "." decimal

separator, importing

% large LVM files that use a "," (or other character) will

be noticeably

% slower. Use a "." decimal separator to avoid this issue.

%

% The LVM file specification is available at:

% http://zone.ni.com/devzone/cda/tut/p/id/4139

%

%

% Example:

%

% Use the following command to read in the data from a

file containing two

% Segments:

%

% >> d=lvm_import('testfile.lvm');

%

% Importing testfile.lvm:

%

% Import complete. 2 Segments found.

%

% >> d

% d =

%

X_Columns: 'One'

%

user: 'hopcroft'

%

Description: 'Pressure, Flowrate, Heat, Power, Analog

Voltage, Pump on, Temp'

%

date: '2008/03/26'

%

time: '12:18:02.156616'

%

clock: [2008 3 26 12 18 2.156616]

136

%

Segment1: [1x1 struct]

%

Segment2: [1x1 struct]

%

% >> d.Segment1

% ans =

%

Notes: 'Some notes regarding this data set'

%

num_channels: 8

%

y_units: {8x1 cell}

%

x_units: {8x1 cell}

%

X0: [8x1 double]

%

Delta_X: [8x1 double]

% column_labels: {9x1 cell}

%

data: [211x9 double]

%

Comment: 'This data rulz'

%

% >> d.Segment1.column_labels{2}

% ans =

% Thermocouple1

%

% >> plot(d.Segment1.data(:,1),d.Segment1.data(:,2));

% >> xlabel(d.Segment1.column_labels{1});

% >> ylabel(d.Segment1.column_labels{2});

%

%

%

% M.A. Hopcroft

%

< mhopeng at gmail.com >

%

% MH Sep2017

% v3.12 fix bug for importing data-only files

%

(thanks to Enrique Alvarez for bug reporting)

% MH Mar2017

% v3.1 use cellfun to vectorize processing of comma-

delimited data

%

(thanks to Victor for suggestion)

% v3.0 use correct test for 'tab'

% MH Aug2016

% v3.0 (BETA) fixes for files that use comma as delimiter

%

improved robustness for files with missing columns

% MH Sep2013

% v2.2 fixes for case of comma separator in multi-segment

files

%

use cell2mat for performance improvement

%

(thanks to <die-kenny@t-online.de> for bug report

and testing)

% MH May2012

137

% v2.1 handle "no separator" bug

%

(thanks to <adnan.cheema@gmail.com> for bug report

and testing)

%

code & comments cleanup

%

remove extraneous column labels (X_Value for "No X"

files; Comment)

%

clean up verbose output

%

change some field names to NI names

("Delta_X","X_Columns","Date")

% MH Mar2012

% v2.0 fix "string bug" related to comma-separated

decimals

%

handle multiple Special Headers correctly

%

fix help comments

%

increment version number to match LabView LVM

writer

% MH Sep2011

% v1.3 handles LVM Writer version 2.0 (files with decimal

separator)

%

Note: if you want to work with older files with a

non-"." decimal

%

separator character, change the value of

"data.Decimal_Separator"

% MH Sep2010

% v1.2 bugfixes for "Special" header in LVM files.

%

(Thanks to <bobbyjoe23928@gmail.com> for

suggestions)

% MH Apr2010

% v1.1 use case-insensitive comparisons to maintain

compatibility with

%

NI LVM Writer version 1.00

%

% MH MAY2009

% v1.02 Add filename input

% MH SEP2008

% v1.01 Fix comments, add Cells

% v1.00 Handle all three possibilities for X-columns

(No,One,Multi)

%

Handle LVM files with no header

% MH AUG2008

% v0.92 extracts Comment for each Segment

% MH APR2008

% v0.9 initial version

%

%#ok<*ASGLU>

138

% message level if nargin < 2, verbose = 1; end % use 1 for release and 2 for BETA if verbose >= 1, fprintf(1,'\nlvm_import v3.1\n'); end
% ask for filename if not provided already if nargin < 1
filename=input(' Enter the name of the .lvm file: ','s');
fprintf(1,'\n'); end
%% Open the data file % open and verify the file fid=fopen(filename); if fid ~= -1, % then file exists
fclose(fid); else
filename=strcat(filename,'.lvm'); fid=fopen(filename); if fid ~= -1, % then file exists
fclose(fid); else
error(['File not found in current directory! (' pwd ')']);
end end
% open the validated file fid=fopen(filename);
if verbose >= 1, fprintf(1,' Importing "%s"\n\n',filename); end
% is it really a LVM file? linein=fgetl(fid); if verbose >= 2, fprintf(1,'%s\n',linein); end % Some LabView routines create an LVM file with no header; just a text file % with columns of numbers. We can try to import this kind of data. if isempty(strfind(linein,'LabVIEW'))
try data.Segment1.data = dlmread(filename); if verbose >= 1, fprintf(1,'This file appears to be
an LVM file with no header.\n'); end
139

if verbose >= 1, fprintf(1,'Data was copied, but no other information is available.\n'); end
return catch fileEx
error('This does not appear to be a text-format LVM file (no recognizeable header or data).');
end end

%% Process file header % The file header contains several fields with useful information

% default values data.Decimal_Separator = '.'; text_delimiter={',',' ','\t'}; data.X_Columns='One';

% File header contains date, time, etc. % Also the file delimiter and decimal separator (LVM v2.0) if verbose >= 2, fprintf(1,' File Header Contents:\n\n'); end while 1

% get a line from the file linein=fgetl(fid); % handle spurious carriage returns if isempty(linein), linein=fgetl(fid); end if verbose >= 3, fprintf(1,'%s\n',linein); end % what is the tag for this line? t_in = textscan(linein,'%s','Delimiter',text_delimiter); if isempty(t_in{1}{1})
tag='notag'; else
tag = t_in{1}{1}; end % exit when we reach the end of the header if strfind(tag,'***End_of_Header***')
if verbose >= 2, fprintf(1,'\n'); end break end

% get the value corresponding to the tag

%

if ~strcmp(tag,'notag')

140

%

v_in = textscan(linein,'%*s

%s','delimiter','\t','whitespace','','MultipleDelimsAsOne',

1);

if size(t_in{1},1)>1 % only process a tag if it has

a value

%

val = v_in{1}{1};

val = t_in{1}{2};

switch tag case 'Date' data.Date = val; case 'Time' data.Time = val; case 'Operator' data.user = val; case 'Description' data.Description = val; case 'Project' data.Project = val; case 'Separator' % v3 separator sanity check if strcmpi(val,'Tab') text_delimiter='\t'; if strfind(linein,',') fprintf(1,'ERROR: File header
reports "Tab" but uses ",". Check the file and correct if necessary.\n');
return end elseif strcmpi(val,'Comma') || strcmpi(val,',') text_delimiter=','; if strfind(linein,sprintf('\t'))
fprintf(1,'ERROR: File header reports "Comma" but uses "tab". Check the file and correct if necessary.\n');
return end end

case 'X_Columns' data.X_Columns = val;
case 'Decimal_Separator' data.Decimal_Separator = val;
end if verbose >= 2, fprintf(1,'%s: %s\n',tag,val); end

141

end

%

end

end

% create matlab-formatted date vector if isfield(data,'time') && isfield(data,'date')
dt = textscan(data.Date,'%d','Delimiter','/'); tm = textscan(data.Time,'%d','Delimiter',':'); if length(tm{1})==3
data.clock=[dt{1}(1) dt{1}(2) dt{1}(3) tm{1}(1) tm{1}(2) tm{1}(3)];
elseif length(tm{1})==2 data.clock=[dt{1}(1) dt{1}(2) dt{1}(3) tm{1}(1)
tm{1}(2) 0]; else data.clock=[dt{1}(1) dt{1}(2) dt{1}(3) 0 0 0]; end
end

if verbose >= 3, fprintf(1,' Text delimiter is "%s":\n\n',text_delimiter); end

%% Process segments % process data segments in a loop until finished segnum = 1; val=[]; tag=[]; %#ok<NASGU> while 1
%segnum = segnum +1; fieldnm = ['Segment' num2str(segnum)];
%% - Segment header if verbose >= 1, fprintf(1,' Segment %d:\n\n',segnum); end % loop to read segment header while 1
% get a line from the file linein=fgetl(fid); % handle spurious carriage returns/blank lines/end of file while isempty(linein), linein=fgetl(fid); end if feof(fid), break; end if verbose >= 3, fprintf(1,'%s\n',linein); end
% Ignore "special segments"

142

% "special segments" can hold other types of data. The type tag is
% the first line after the Start tag. As of version 2.0,
% LabView defines three types: % Binary_Data % Packet_Notes % Wfm_Sclr_Meas % In theory, users can define their own types as well. LVM_IMPORT % ignores any "special segments" it finds. % If special segments are handled in future versions, recommend % moving the handler outside the segment read loop. if strfind(linein,'***Start_Special***')
special_seg = 1; while special_seg
while 1 % process lines until we find the end of the special segment
% get a line from the file linein=fgetl(fid); % handle spurious carriage returns if isempty(linein), linein=fgetl(fid); end % test for end of file if linein==-1, break; end if verbose >= 2, fprintf(1,'%s\n',linein); end if strfind(linein,'***End_Special***')
if verbose >= 2, fprintf(1,'\n'); end
break end end
% get the next line and proceed with file % (there may be additional Special Segments) linein=fgetl(fid); % handle spurious carriage returns/blank lines/end of file while isempty(linein), linein=fgetl(fid); end if feof(fid), break; end
143

if isempty(strfind(linein,'***Start_Special***'))
special_seg = 0; if verbose >= 1, fprintf(1,' [Special Segment ignored]\n\n'); end end end end % end special segment handler

% what is the tag for this line? t_in = textscan(linein,'%s','Delimiter',text_delimiter); if isempty(t_in{1}{1})
tag='notag'; else
tag = t_in{1}{1}; %disp(t_in{1}) end if verbose >= 3, fprintf(1,'%s\n',linein); end % exit when we reach the end of the header if strfind(tag,'***End_of_Header***') if verbose >= 3, fprintf(1,'\n'); end break end

% get the value corresponding to the tag

% v3 assignments use dynamic field names

if size(t_in{1},1)>1 % only process a tag if it has

a value

switch tag

case 'Notes'

%

%d_in = textscan(linein,'%*s

%s','delimiter','\t','whitespace','');

%

d_in = linein;

data.(fieldnm).Notes = t_in{1}{2:end};

case 'Test_Name'

%

%d_in = textscan(linein,'%*s

%s','delimiter','\t','whitespace','');

%

d_in = linein;

data.(fieldnm).Test_Name =

t_in{1}{2:end}; %d_in{1}{1};

case 'Channels'

%

numchan =

textscan(linein,sprintf('%%*s%s%%d',text_delimiter),1)

%

data.(fieldnm).num_channels =

numchan{1};

144

data.(fieldnm).num_channels =

str2num(t_in{1}{2});

case 'Samples'

%

numsamp =

textscan(linein,'%s','delimiter',text_delimiter);

%

numsamp1 = numsamp{1};

numsamp1 = t_in{1}(2:end);

%

numsamp1(1)=[]; % remove tag

"Samples"

num_samples=[];

for k=1:length(numsamp1)

num_samples = [num_samples

sscanf(numsamp1{k},'%f')]; %#ok<AGROW>

end

%numsamp2=str2num(cell2mat(numsamp1));

%#ok<ST2NM>

data.(fieldnm).num_samples =

num_samples;

case 'Y_Unit_Label'

%

Y_units =

textscan(linein,'%s','delimiter',text_delimiter);

%

data.(fieldnm).y_units=Y_units{1}';

data.(fieldnm).y_units=t_in{1}';

data.(fieldnm).y_units(1)=[]; % remove

tag

case 'Y_Dimension'

%

Y_Dim =

textscan(linein,'%s','delimiter',text_delimiter);

%

data.(fieldnm).y_type=Y_Dim{1}';

data.(fieldnm).y_type=t_in{1}';

data.(fieldnm).y_type(1)=[]; % remove

tag

case 'X_Unit_Label'

%

X_units =

textscan(linein,'%s','delimiter',text_delimiter);

%

data.(fieldnm).x_units=X_units{1}';

data.(fieldnm).x_units=t_in{1}';

data.(fieldnm).x_units(1)=[];

case 'X_Dimension'

%

X_Dim =

textscan(linein,'%s','delimiter',text_delimiter);

%

data.(fieldnm).x_type=X_Dim{1}';

data.(fieldnm).x_type=t_in{1}';

data.(fieldnm).x_type(1)=[]; % remove

tag

case 'X0'

%[Xnought, val]=strtok(linein);

145

val=t_in{1}(2:end); if ~strcmp(data.Decimal_Separator,'.')
val = strrep(val,data.Decimal_Separator,'.');
end X0=[]; for k=1:length(val)
X0 = [X0 sscanf(val{k},'%e')]; %#ok<AGROW>
end data.(fieldnm).X0 = X0; %data.(fieldnm).X0 = textscan(val,'%e'); case 'Delta_X' %, %[Delta_X, val]=strtok(linein); val=t_in{1}(2:end); if ~strcmp(data.Decimal_Separator,'.')
val = strrep(val,data.Decimal_Separator,'.');
end Delta_X=[]; for k=1:length(val)
Delta_X = [Delta_X sscanf(val{k},'%e')]; %#ok<AGROW>
end data.(fieldnm).Delta_X = Delta_X; end end
end % end reading segment header loop % Done reading segment header
% after each segment header is the row of column labels linein=fgetl(fid); Y_labels = textscan(linein,'%s','delimiter',text_delimiter); data.(fieldnm).column_labels=Y_labels{1}'; % The X-column always exists, even if it is empty. Remove if not used. if strcmpi(data.X_Columns,'No')
data.(fieldnm).column_labels(1)=[]; end % remove empty entries and "Comment" label if any(strcmpi(data.(fieldnm).column_labels,'Comment'))
data.(fieldnm).column_labels=data.(fieldnm).column_labels(1 :find(strcmpi(data.(fieldnm).column_labels,'Comment'))-1);
146

end % display column labels if verbose >= 1
fprintf(1,' %d Data Columns:\n | ',length(data.(fieldnm).column_labels));
for i=1:length(data.(fieldnm).column_labels) fprintf(1,'%s |
',data.(fieldnm).column_labels{i}); end fprintf(1,'\n\n');
end
%% - Segment Data % Create a format string for textscan depending on the number/type of % channels. If there are additional segments, texscan will quit when % it comes to a text line which does not fit the format, and the loop % will repeat. if verbose >= 1, fprintf(1,' Importing data from Segment %d...',segnum); end
% How many data columns do we have? (including X data) switch data.X_Columns
case 'No' % an empty X column exists in the file numdatacols = data.(fieldnm).num_channels+1; xColPlural='no X-Columns';
case 'One' numdatacols = data.(fieldnm).num_channels+1; xColPlural='one X-Column';
case 'Multi' numdatacols = data.(fieldnm).num_channels*2; xColPlural='multiple X-Columns';
end
% handle case of not using periods (aka "dot" or ".") for decimal point separators
% (LVM version 2.0+) if ~strcmp(data.Decimal_Separator,'.')
if verbose >= 2, fprintf(1,'\n (using decimal separator "%s")\n',data.Decimal_Separator); end
147

% create a format string for reading data as

numbers

fs = '%s'; for i=2:numdatacols, fs = [fs ' %s'];

end

%#ok<AGROW>

% add one more column for the comment field

fs = [fs ' %s'];

%#ok<AGROW>

% v3.1 - use cellfun to process data

% Read columns as strings

rawdata =

textscan(fid,fs,'delimiter',text_delimiter);

% Convert ',' decimal separator to '.' decimal

separator

rawdata = cellfun(@(x)

strrep(x,data.Decimal_Separator,'.'), rawdata,

'UniformOutput', false);

% save first row comment as The Comment for this

segment

data.(fieldnm).Comment =

rawdata{size(rawdata,2)}{1};

% Transform strings back to numbers

rawdata = cellfun(@(x) str2double(x), rawdata,

'UniformOutput', false);

% else is the typical case, with a '.' decimal

separator

else

% create a format string for reading data as

numbers

fs = '%f'; for i=2:numdatacols, fs = [fs ' %f'];

end

%#ok<AGROW>

% add one more column for the comment field

fs = [fs ' %s'];

%#ok<AGROW>

% read the data from file

rawdata =

textscan(fid,fs,'delimiter',text_delimiter);

% save first row comment as The Comment for this

segment

data.(fieldnm).Comment =

rawdata{size(rawdata,2)}{1};

end

% v2.2 use cell2mat here instead of a loop for better performance
% consolidate data into a simple array, ignore comments data.(fieldnm).data=cell2mat(rawdata(:,1:numdatacols));

148

% If we have a "No X data" file, remove the first column (it is empty/NaN)
if strcmpi(data.X_Columns,'No') data.(fieldnm).data=data.(fieldnm).data(:,2:end);
end
if verbose >= 1, fprintf(1,' complete (%g data points (rows)).\n\n',length(data.(fieldnm).data)); end
% test for end of file if feof(fid)
if verbose >= 2, fprintf(1,' [End of File]\n\n'); end
break; else
segnum = segnum+1; end
end % end process segment
if verbose >= 1 if segnum > 1, segplural='Segments'; else segplural='Segment'; end fprintf(1,'\n Import complete. File has %s and %d Data
%s.\n\n',xColPlural,segnum,segplural); end
% close the file fclose(fid); return
%% input Gain = 2; Vex = 3.3; GF = 2.05; v = 0.3 ; count = 0; n_channels = 8 ; run_num ='LAPTOP-288A8P0F_20190730_111223_PCTime';% 'LAPTOP-288A8P0F_20190730_123107_PCTime'; %'LAPTOP288A8P0F_20190730_111223_PCTime'; % martlet file name ni_filename = '20190730_test_9'; % '20190730_test_11' ;% %ni file name
149

%run_num = 'LAPTOP-288A8P0F_20190808_120341_PCTime';
% run_num = 'LAPTOP-288A8P0F_20190730_123107_PCTime'; % martlet file name % ni_filename = '20190730_test_11'; %ni file name
dirc = dir('.\DAQResults\'); path_base = sprintf('.\\DAQResults\\%s\\',run_num); path = [path_base 'TestName.txt'];
[DAQset] = load_DAQ_settings_Martlet(path);
% Find actual points collected points1 = DAQset.fs * DAQset.T; points2 = DAQset.points_per_poll; if points2 > points1
points = points1; elseif mod(points1,points2) ~= 0
points = (floor(points1/points2)+1)*points2; else
points = points1; end num_poll_cycles = ceil(points1/points2);
% Preallocate the memory data = zeros(DAQset.num_units, max(max(DAQset.channel_num_list(:,:))), num_poll_cycles*DAQset.points_per_poll); % Load the data: time = 1/DAQset.fs*[1:points]'; data_str = zeros(n_channels , length (time)); for k = 1:DAQset.num_units
chan = DAQset.channel_num_list(k,:); for j = 1:chan
count = count + 1; if((DAQset.chans(k,j,1) == 72 && DAQset.chans(k,j,2) == 49))
filename = [path_base sprintf('U%02d_EXTADC_CH1', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 72 && DAQset.chans(k,j,2) == 50))
filename = [path_base sprintf('U%02d_EXTADC_CH2', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 72 && DAQset.chans(k,j,2) == 51))
150

filename = [path_base sprintf('U%02d_EXTADC_CH3', DAQset.unit_list(k,1))];
elseif((DAQset.chans(k,j,1) == 72 && DAQset.chans(k,j,2) == 52))
filename = [path_base sprintf('U%02d_EXTADC_CH4', DAQset.unit_list(k,1))];
end
tempdata = []; ppp = DAQset.points_per_poll; for i = 1:num_poll_cycles
filename_i = [filename '_' num2str(i,'%05d') '.dat'];
if DAQset.chans(k,j,1) == 72 tempdata((i-1)*ppp*2+1:i*ppp*2,1) =
load(filename_i); else tempdata((i-1)*ppp+1:i*ppp,1) =
load(filename_i); end
end
if DAQset.chans(k,j,1) == 72 if DAQset.chans(k,j,2) == 52 for data_i = 1:length(tempdata)/2
data_volt(data_i) = bitshift(tempdata((data_i-1)*2+1),16) + tempdata(data_i*2);
data_volt(data_i) = typecast(uint32(data_volt(data_i)),'int32');
data_volt(data_i) = data_volt(data_i)*2.442/2^31;
end else
for data_i = 1:length(tempdata)/2 data_volt(data_i) =
bitshift(tempdata((data_i-1)*2+1),16) + tempdata(data_i*2); data_volt(data_i) =
typecast(uint32(data_volt(data_i)),'int32'); data_volt(data_i) =
data_volt(data_i)*2.442/2^31/1.084/1.759/Gain; end
end
if DAQset.chans(k,j,1) == 72 for data_i = 2:length(data_volt)
151

if abs(data_volt(data_i) data_volt(data_i-1)) > 0.3
data_volt(data_i) = data_volt(data_i-1);
end end end
if DAQset.chans(k,j,2) == 52 data_volt(1) = data_volt(2); data_v = -data_volt; data_tmp = 5.8145*data_v.^3 +
3.5922*data_v.^2 + 30.245*data_v + 16.111; else for i=1:length(data_volt) data_str(count,i) =
data_volt(1,i)*2/Vex/GF/(1+v - data_volt(1,i)/Vex*(1v))*10^6; %mod by N
% data_str(count,i) = data_volt(1, i)*2/3.3/2.04/0.697*1000000;
end
data_str = data_str - data_str(:,1);
end end end end
data_str = data_str([1 2 4 5 7 8 10 11 ],:);
for i=1:n_channels for e=1:4 [~,b] = max(data_str(i,:)); if b>1 data_str(i,b) = data_str(i,b-1); end [~,b] = min(data_str(i,:)); if b>1 data_str(i,b) = data_str(i,b-1);
152

end end end
%% NI n_qbsq = 12; n_fbsg = 4; cd DataAcquisitionNI %DAQ = lvm_import([num2str(testdate),'_test_', num2str(testn) ,'.lvm'], 0); DAQ = lvm_import(ni_filename, 0); cd ..
t = DAQ.Segment1.data(:,1); QBSG(:,1:n_qbsq) = DAQ.Segment1.data(:,2:2+n_qbsq-1)*10^6; FBSG(:,1:n_fbsg) = -DAQ.Segment1.data(:,14:14+n_fbsg1)*10^6; lvdt1 = DAQ.Segment1.data(:,19); lvdt2 = DAQ.Segment1.data(:,18); sp1a = DAQ.Segment1.data(:,20); LC = -DAQ.Segment1.data(:,21);
n_cycles = time(end)/60; t_loss = 3; % 2 sconds lost newtime = [time', time(end)+1:time(end)+t_loss*n_cycles]'; for i=1:n_cycles
if t_loss == 2 if i<=1 newdata_str(:,1:60*i+(i-1)*t_loss) =
data_str(:, 1:60*i); newdata_str(:,60*i+(i-1)*t_loss + 1) =
data_str(:, 60*i); newdata_str(:,60*i+(i-1)*t_loss + 2) =
data_str(:, 60*i);
else
newdata_str(: , 60*(i-1)+(i-1-1)*t_loss+ t_loss +1 : 60*i+(i-1)*t_loss) = data_str(:, 60*(i-1)+1:60*i);
newdata_str(:,60*i+(i-1)*t_loss + 1) = data_str(:, 60*i);
newdata_str(:,60*i+(i-1)*t_loss + 2) = data_str(:, 60*i);
end
153

else if t_loss == 3 if i<=1 newdata_str(:,1:60*i+(i-1)*t_loss) =
data_str(:, 1:60*i); newdata_str(:,60*i+(i-1)*t_loss + 1) =
data_str(:, 60*i); newdata_str(:,60*i+(i-1)*t_loss + 2) =
data_str(:, 60*i); newdata_str(:,60*i+(i-1)*t_loss + 3) =
data_str(:, 60*i);
else
newdata_str(: , 60*(i-1)+(i-1-1)*t_loss+ t_loss +1 : 60*i+(i-1)*t_loss) = data_str(:, 60*(i1)+1:60*i);
newdata_str(:,60*i+(i-1)*t_loss + 1) = data_str(:, 60*i);
newdata_str(:,60*i+(i-1)*t_loss + 2) = data_str(:, 60*i);
newdata_str(:,60*i+(i-1)*t_loss + 3) = data_str(:, 60*i);
end end end end
%% Plots figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]); plot(time, data_str(0+1,:), t, FBSG(:,0+1)) legend('martlet', 'cabled') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)') figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]) plot(time, data_str(1+1,:), t, FBSG(:,1+1)) legend('martlet', 'NI') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)') figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]) plot(time, data_str(2+1,:), t, FBSG(:,2+1)) legend('martlet', 'NI') xlabel ('time (sec)')
154

ylabel('strain (\mu\epsilon)') figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]) plot(time, data_str(3+1,:), t, FBSG(:,3+1)) legend('martlet', 'NI') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)')
figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]); plot(time, data_str(4+1,:), t, QBSG(:,0+1)) legend('martlet', 'NI') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)') figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]) plot(time, data_str(5+1,:), t, QBSG(:,1+1)) legend('martlet', 'NI') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)') figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]) plot(time, data_str(6+1,:), t, QBSG(:,2+1)) legend('martlet', 'NI') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)') figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]) plot(time, data_str(7+1,:), t, QBSG(:,3+1)) legend('martlet', 'NI') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)')
%% Plots (with new data for time compensation) figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]); plot(newtime, newdata_str(0+1,:), t, FBSG(:,0+1), 'LineWidth',1.5) legend('martlet M_0', 'cabled FBSG_0','FontSize', 10) xlabel ('time (sec)','FontSize', 18) ylabel('strain (\mu\epsilon)','FontSize', 18) ax = gca; ax.FontSize = 14;
155

figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]) plot(newtime, newdata_str(1+1,:), t, FBSG(:,1+1)) legend('martlet', 'NI') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)') figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]) plot(newtime, newdata_str(2+1,:), t, FBSG(:,2+1)) legend('martlet', 'NI') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)') figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]) plot(newtime, newdata_str(3+1,:), t, FBSG(:,3+1)) legend('martlet', 'NI') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)')
figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]); plot(newtime, newdata_str(4+1,:), t, QBSG(:,0+1)) legend('martlet', 'NI') xlabel ('time (sec)') ylabel('strain (\mu\epsilon)')
figHand(i) = figure; set (figHand(i), 'Position',[200 200 600 200]) plot(newtime, newdata_str(5+1,:), t, QBSG(:,1+1), 'LineWidth',1.5) legend('martlet M_5', 'cabled QBSG_1','FontSize', 10) xlabel ('time (sec)','FontSize', 18) ylabel('strain (\mu\epsilon)','FontSize', 18) ax = gca; ax.FontSize = 14;
156

APPENDIX D: RAW DATA Bay 1 (between external and internal girders) (LD | RD) Left Diagonal Member Strain Gauge 5 Right Diagonal Member Strain Gauge 7 Left Diagonal Member Strain Gauge 6 Right Diagonal Member Strain Gauge 8
157

Bay 1 (between external and internal girders) (LBM | RBM) Bottom Member (Left) Strain Gauge 1 Bottom Member (Right) Strain Gauge 3 Bottom Member (Left) Strain Gauge 2 Bottom Member (Right) Strain Gauge 4
158

Bay 3 (between two internal diaphragms) (LD | RD) Left Diagonal Member Strain Gauge 5 Right Diagonal Member Strain Gauge 7 Left Diagonal Member Strain Gauge 6 Right Diagonal Member Strain Gauge 8
159

Bay 3 (between two internal diaphragms) (LBM | RBM) Bottom Member (Left) Strain Gauge 1 Bottom Member (Right) Strain Gauge 3 Bottom Member (Left) Strain Gauge 2 Bottom Member (Right) Strain Gauge 4
160

APPENDIX E: CONCRETE CONSTRUTION LOAD CALCULATIONS
1 = 81.5; % = 1500; % = 7.375; % = 1 / 12^3; %^3 = 150; % /^3 , ; = ; % = 3 12 + 6; % = ; % ^2 = / ; % /^2
161

ACKNOWLEDGMENTS The following individuals at GDOT provided many valuable suggestions throughout this study: Mr. Steve Gaston, Assistant State Bridge Engineer; Mr. Jason Waters, Concrete Branch Chief; Mr. Chris Watson, Bridge Engineer; Ms. Sarah Lamothe, Research Program Manager; and Ms. Supriya Kamatkar, Assistant Office Head, Office of Performance-based Management and Research. The opinions and conclusions expressed herein are those of the authors and do not represent the opinions, conclusions, policies, standards, or specifications of GDOT or of other cooperating organizations. The authors would like to thank the Mr. Peter Lander for the assitance in collecting the data in this research, and Mr. Jeremy Mitchell for his assistance in the laboratory experiment. The authors express their profound gratitude to all of these individuals for their assistance and support during the completion of this research project.
162

REFERENCES
1. Georgia Department of Transportation (GDOT). Bridge and Structures Design Manual. Atlanta Georgia, 2018.
2. California Department of Transportation (CALTRANS). Standard Specifications. Sacramento, California, 2006.
3. Colorado Department of Transportation (CDOT). Standard Specifications for Road and Bridge Construction. Denver, Colorado, 2005.
4. Florida Department of Transportation (FDOT). Standard Specifications for Road and Bridge Construction. Tallahassee, Florida, 2010.
5. Idaho Transportation Department (ITD). Standard Specifications for Highway Construction. Boise, Idaho, 2004.12.
6. Illinois Department of Transportation (IDOT). Standard Specifications for Road and Bridge Construction. Springfield, Illinois, 2007.
7. Kentucky Transportation Cabinet (KYTC). Standard Specifications for Road and Bridge Construction. Frankfort, Kentucky, 2008.
8. Maryland Department of Transportation (MDOT), State Highway Administration. Standard Specifications for Construction and Materials. Baltimore, Maryland, 2008.
9. Michigan Department of Transportation (MDOT). Standard Specifications for Construction. Lansing, Michigan, 2005.
10. Minnesota Department of Transportation (Mn/DOT). Standard Specifications for Construction. St. Paul, Minnesota, 2005.
11. New York State Department of Transportation (NYSDOT). Standard Specifications. Albany, New York, 2008.18. Ohio Department of Transportation (ODOT). Construction and Material Specifications. Columbus, Ohio, 2010.
12. Pennsylvania Department of Transportation (PennDOT). Construction Specifications. Harrisburg, Pennsylvania, 2007.
13. Texas Department of Transportation (TxDOT). Standard Specifications for Construction and Maintenance of Highways, Streets, and Bridges. Austin, Texas, 2004.
14. Washington Department of Transportation (WSDOT). Standard Specifications for Road, Bridge, and Municipal Construction. Olympia, Washington, 2010.
163

15. Wisconsin Department of Transportation (WisDOT). Standard Specifications for Highway and Structure Construction. Madison, Wisconsin, 2011.
16. McPherson, D., McCullough, B., Bowman, M., Structural Impact of Construction Loads. Joint Transportation Research Program, 2012.
17. American Association of State Highway and Transportation Officials (AASHTO). Guide Design Specifications for Bridge Temporary Works. Washington, D. C., 2017.
18. American Society of Civil Engineers (ASCE). ASCE 37: Design Loads of Structures During Construction. Reston, V. A., 2014.
19. RISA Tech Incorporated. RISA-3D Rapid Interactive Structural Analysis Version 19 General Reference, 2020.
20. Dong, X., Liu, X., Wright, T., Wang, Y., and DesRoches, R. (2016). "Validation of wireless sensing technology densely instrumented on a full-scale concrete frame structure." Proceedings of International Conference on Smart Infrastructure and Construction (ICSIC), Cambridge. United Kingdom, June 27-29, 2016.
21. Liu, X., Dong, X., and Wang, Y. (2016). "Field testing of Martlet wireless sensing system on an in-service pre-stressed concrete highway bridge." Proceedings of SPIE, Health Monitoring of Structural and Biological Systems, 9805, Las Vegas, NV, USA, March 20-24, 2016.
22. Dong, X., Zhu, D., Wang, Y., Lynch, J. P., and Swartz, R. A. (2014). "Design and validation of acceleration measurement using the Martlet wireless sensing system." Proceedings of the ASME 2014 Conference on Smart Materials, Adaptive Structures and Intelligent Systems (SMASIS), Newport, RI, USA, September 8-10, 2014.
23. Computers and Structures, Incorporated. CSI Analysis Reference Manual For SAP2000, ETABS, SAFE and CSIBridge. 2016.
24. Simulia. Abaqus Analysis User's Guide, 2020.
164