259
Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live Loading of Bridge Structures By Carl L. Schneeman, B.S.C.E. A Thesis submitted to the Faculty of the Graduate School, Marquette University, in Partial Fulfillment of the Requirement for the Degree of Master of Science Milwaukee, Wisconsin 2006

Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

Development and Evaluation of a Removable and Portable Strain

Sensor for Short-term Live Loading of Bridge Structures

By

Carl L. Schneeman, B.S.C.E.

A Thesis submitted to the Faculty of the Graduate School, Marquette University, in

Partial Fulfillment of the Requirement for the Degree of

Master of Science

Milwaukee, Wisconsin

2006

Page 2: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

i

Preface

Historically, bridges have been constructed in many different ways. As economic

conditions change, so do the materials and methods used in construction. To evaluate the

use of new materials for these structures, research and development is conduced to

determine their adequacy in modern construction. However, long-term results are

required for analysis of bridges, as they are constructed to remain in service for extended

periods of time. Additionally, a fundamental understanding and proper selection of the

tools used in bridge monitoring is required.

This thesis presents a detailed discussion of the state of bridge monitoring,

development and operation of a removable strain sensor to be used in collecting data on

bridge structures, and guidelines for a future load test of a bridge in northern Wisconsin.

Observations and conclusions of the study are made as well, with recommendations for

future work on this topic.

Page 3: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

ii

Acknowledgements

I would like to thank the many people who have helped in numerous ways

throughout my tenure at Marquette. Without their support and help none of this work

would have been possible.

The members of my advisory committee have been wonderful. Dr. Chris Foley

has been a tireless and selfless mentor to me. Never before have I worked with an

individual so passionate and dedicated. His constant drive to teach others has benefited

me countless ways. Drs. Stephen Heinrich and Baolin Wan both were extremely helpful

in my classes and during my thesis work.

The many people working for the laboratories of Marquette need to be

recognized. Dave Newman has been an instrumental person for me. His frequent

questions, new challenges, and often used humor have helped shape my education in so

many positive ways. Our students are fortunate to have such a wealth of experience to

learn from. Tom Silman’s help and guidance requires thanks also. Without his tireless

flexibility my late hours in the machine shop would not have been possible.

Additionally, John Boudnik’s willingness to help at any time is much appreciated.

To those in the working world who have shown me what being a professional

means. Many thanks are directed to John Pluta at MSI General, who was the first to truly

show me how to be an engineer. The generosity of Victory Steel of Milwaukee and

Construction Supply & Errection of Germantown, WI are to be commended – their

willingness to aid the university is instrumental to our successes.

Thanks are also due to so many people who have pushed me to go beyond what is

required. People like Andy Basta, Nick Hornyak, Kristine Martin, Panchito Ojeda, Brian

Page 4: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

iii

Porter, and Danny Stadig, have been a mainstay for me. Their support, suggestions and

assistance in all aspects of life have left are truly valued.

Many thanks are due to my girlfriend, Meg Taylor. She has been a pillar of

stability for me, constantly supporting me in my endeavors.

Finally, my family has also been wonderful. My parents, Chris and Cathy, and

siblings, Pat, Dan, Matt and Lucy, have always provided me with endless support. How

they have sustained my many engineering lectures and overly-scientific explanations of

daily occurrences is amazing.

Page 5: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

iv

Table of Contents

1. Introduction and Summary of Work

1.0 Introduction……………………………………………………………………. 1

1.1 The De Neveu Creek IBRC Bridge…………………………………………… 3

1.2 Objective of Thesis…………………………………………………………… 6

1.3 Scope of Thesis………………………………………………………………... 7

2. Literature Review and Synthesis

2.0 Introduction …………………………………………………………………… 8

2.1 Instrumentation Guidelines…………………………………………………… 8

2.2 The De Neveu Creek IBRC Bridge …………………………………………... 13

2.2.1. FRP Reinforcement in the De Neveu Bridge…………………………... 15

2.2.2. Development and Testing of the FRP-grillage Reinforced Deck ……... 17

2.3 Load Testing of the De Neveu Creek Bridge ………………………………… 20

2.3.1. Strain Gage Instrumentation of the Bridge ……………………………. 21

2.3.2. In-Situ Load Test of B-20-148 FRP Bridge …………………………… 23

2.3.3. In-Situ Load Test of B-20-149 Conventional Bridge …………………. 26

2.3.4. Results of B-148 and B-149 Load Tests ………………………………. 27

2.4 Ohio Bridge HAM-126-0881 ………………………………………………… 31

2.4.1. Data Acquisition System Employed …………………………………... 32

2.4.2. Load Testing and Results ……………………………………………… 34

2.5 South Carolina Route S655 …………………………………………………... 37

2.5.1. GFRP Panels …………………………………………………………... 38

2.5.2. Instrumentation and Load Testing …………………………………….. 40

2.6 Fairground Road Bridge ……………………………………………………… 44

2.6.1. Study of Composite Action and Strain Measurements ……………….. 45

2.6.2. Load Testing and Results………………………………………………. 46

2.7 The Bridge Street Bridge……………………………………………………… 50

2.7.1. Materials Used ………………………………………………………… 51

2.7.2. Instrumentation ………………………………………………………... 52

2.7.3. Load Test and Results …………………………………………………. 53

2.8 Synthesis of Literature ………………………………………………………... 55

Page 6: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

v

3. Data Acquisition and Strain Measurement

3.0 Introduction …………………………………………………………………… 59

3.1 Signal Processing ……………………………………………………………... 59

3.1.1. Analog to Digital Conversion ………………………………………… 60

3.1.2. Sampling Rates………………………………………………………… 63

3.1.3. Signal Amplification…………………………………………………… 64

3.1.4. Signal Filtering………………………………………………………… 69

3.2 Measurement with Electrical Resistance Strain Gages ………………………. 75

3.3 Strain Gage Measurement Errors …………………………………………….. 83

3.4 String Potentiometers and Linear Position Sensors …………………………... 90

3.5 DASYLab Data Acquisition Software ……………………………………….. 93

3.5.1. Installation of ADC Modules …………………………………………. 96

3.5.2. Installation of Digital Filtering ………………………………………... 99

3.5.3. The Black Box Module ………………………………………………... 102

3.5.4. Offset Adjustment of Signals………………………………………….. 104

3.5.5. Establishment of Calibration Modules …………………………………108

3.5.6. Continuous Unit Conversion ………………………………………….. 113

3.5.7. Duplicating the Black Box for use with Transducers ………………… 115

3.5.8. Configuring the Black Box for Use with Different Channels…………. 117

4. Development and Testing of a Portable Strain Sensor

4.0 Introduction …………………………………………………………………… 123

4.1 Quarter Bridge Circuit Selection ……………………………………………... 124

4.2 Material Experimentation and Selection ……………………………………... 127

4.3 Description of Portable Strain Sensor ………………………………………… 130

4.4 Anchorage of the Sensor ……………………………………………………… 133

4.5 Laboratory Validation…………………………………………………………. 135

4.5.1. Torque Level Tests ……………………………………………………. 140

4.5.2. Evaluation of Washer Presence………………………………………... 142

4.5.3. Excitation Voltage Evaluation…………………………………………. 143

4.6 Finite Element Analysis ………………………………………………………. 144

4.6.1. Finite Element Model of Test Beam …………………………………... 145

Page 7: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

vi

4.6.2. Finite Element Model of Strain Sensor ……………………………….. 149

4.7 Calibration of Individual Strain Sensors for Field Implementation ………….. 163

4.7.1. Calibration Method and Equipment Used………………………………163

4.7.2. Data Recorded during Load Tests …………………………………….. 166

4.7.3. Individual Calibration Factors…………………………………………. 169

5. Proposed Load Test

5.0 Introduction …………………………………………………………………… 171

5.1 Load Test Objectives and Instruments………………………………………... 171

5.2 Permanently Installed Equipment……………………………………………... 177

5.2.1. Lead Wiring for Instruments…………………………………………… 177

5.2.2. Enclosure Box and Screw Terminals…………………………………... 180

5.2.3. Installation of Strain Sensors…………………………………………... 184

5.3 Load Test Vehicles and Test Configuration…………………………………... 187

5.3.1. Load Test Objectives…………………………………………………... 188

5.3.2. Load Test Configurations……………………………………………… 189

5.4 Data Acquisition System……………………………………………………… 193

5.4.1. Signal Conditioning Modules………………………………………….. 196

5.4.2. Connection to Strain Gage Modules…………………………………… 198

5.4.3. Acquisition Software…………………………………………………... 200

5.4.4. Error Correction in Readings…………………………………………... 201

6. Summary and Conclusions

6.0 Summary……………………………………………………………………… 203

6.1 Conclusions…………………………………………………………………… 204

6.2 Recommendations for Future Research………………………………………. 206

References……………………………………………………………………………… 210

Appendix A…………………………………………………………………………….. 214

Appendix B……………………………………………………………………………..235

Appendix C……………………………………………………………………………..241

Appendix D……………………………………………………………………………..244

Page 8: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

vii

List of Figures

Figure 1.1.1 – Wisconsin Highways 151 before (left) the bypass and after (right) 4

Figure 1.1.2 – The De Neveu Creek Bridge 5

Figure 2.2.1 – The De Neveu Creek Bridge, WI B-20-148 14

Figure 2.2.2 – Cross section of the De Neveu Creek Bridge 15

Figure 2.2.3 – Assembled FRP grillage 15

Figure 2.2.4 – Cross sections of FRP materials used 16

Figure 2.2.5 – Simple-span slab tests 19

Figure 2.2.6 – Restrained end slab tests 19

Figure 2.3.1 – Layout of surveying prisms and strain gages 21

Figure 2.3.2 – Strain gage locations 22

Figure 2.3.3 – Strain gage locations 22

Figure 2.3.4 – Stopped vehicle locations for live-load testing of B-20-148 FRP 25

Figure 2.3.5 – Stopped vehicle locations for live-load testing of

B-20-149 Conventional 27

Figure 2.3.6 – Deflection plot of mid-span girder response in bridge B-20-148 FRP 28

Figure 2.3.7 – Deflection plot of mid-span girder response in bridge B-20-149

Conventional 29

Figure 2.4.1 – Schematic of Ohio Bridge HAM-126-0881 32

Figure 2.4.2 – Static live-load Cases A through L 35

Figure 2.4.3 – Method for calculating internal moments in stingers 37

Figure 2.5.1 – Individual Duraspan® deck panels 39

Figure 2.5.2 – Section of bridge deck and integral grout-filled shear pockets 39

Figure 2.5.3 – Instrument layout of S655 Bridge 41

Figure 2.5.4 – Live-load test cases for S655 Bridge 42

Figure 2.6.1 – Instrumentation layout of the Fairground Road Bridge 46

Figure 2.6.2 – Bridge Diagnostics Strain Transducer 46

Figure 2.6.2 – Method used to locate the neutral axis of stringers 48

Figure 2.6.3 – Comparison of composite nature stress profiles through a typical deck

section 48

Page 9: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

viii

Figure 2.6.4 – Top flange tensile “spike” observed in stringers during live load

testing 49

Figure 2.7.1 – Bridge Street Bridge cross section 51

Figure 2.7.2 – Strain gage location in instrumented spans of Structure B 52

Figure 2.7.3 – Long-term instrumented spans of Structure B 53

Figure 2.7.4 – Live-load test cases for Bridge Street Bridge 54

Figure 3.1.1 – Layout of a typical data acquisition system 60

Figure 3.1.2 – Application of the Nyquist Theorem 63

Figure 3.1.3 – Block diagram of an IOTech DBK43A 66

Figure 3.1.4 – Signal path of the “OFFSET” trimpot 67

Figure 3.1.5 – Signal path during adjustment of the input amplifier gain 68

Figure 3.1.6 – Signal path during adjustment of the scaling amplifier gain 69

Figure 3.1.7 – Typical low-pass frequency response 70

Figure 3.1.8 – Butterworth low-pass filter response 71

Figure 3.1.9 – Chebyshev low-pass filter response 71

Figure 3.1.10 – Comparison of filtered (b & c) and unfiltered data (a) 72

Figure 3.1.11 – (a) standard wave composed of AC and DC signals, (b) AC

Coupled wave 73

Figure 3.2.1 – Typical strain gage 76

Figure 3.2.2 – The Wheatstone bridge 76

Figure 3.2.3 – Typical configurations of the Wheatstone bridge 77

Figure 3.2.4 – Diagram of variables used in calculations 78

Figure 3.2.5 – Simulated strain via shunt calibration 80

Figure 3.3.1 – Quarter bridge strain gage configurations 84

Figure 3.3.2 – Nonlinearity errors for tensile strains in bridge circuits 89

Figure 3.4.1 – Circuit diagram of a typical three-wire transducer 91

Figure 3.4.2 – 30-inch String Potentiometer 92

Figure 3.4.3 – 4-inch Linear Position Sensor 92

Figure 3.4.4 – Linear calibration of sensors 93

Figure 3.5.1 – Data acquisition worksheet 94

Figure 3.5.2 – Logical map of software configuration 95

Page 10: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

ix

Figure 3.5.3 – Hardware configuration window 97

Figure 3.5.4 – Hardware configuration window on the main worksheet 98

Figure 3.5.5 – Expansion of analog inputs in the DBK43A module 99

Figure 3.5.6 – Defining global variables 100

Figure 3.5.7 – Defining filtration properties for each channel within the filter

module dialog box 101

Figure 3.5.8 – ADC and filter modules connected on the worksheet 101

Figure 3.5.9 – Locating a new black box on the main worksheet (left) and opening

the black box (right) 102

Figure 3.5.10 – Modules installed in the black box 104

Figure 3.5.11 – Offset adjust modules and digital meter in black box 105

Figure 3.5.12 – Specification of the Switch module operation 106

Figure 3.5.13 – Specification of the Action module operation 107

Figure 3.5.14 – Specification of the Digital Meter module operation 108

Figure 3.5.15 – Linear scaling from shunt calibration 109

Figure 3.5.16 – Locating the Linear Scaling/Unit Conversion module on the black

box worksheet 110

Figure 3.5.17 – Flow chart depicting the signal path while acquiring simulated

strain voltages from shunt calibration 111

Figure 3.5.18 – Storing global variables for calibration 112

Figure 3.5.19 – Setting linear scaling values for individual strain gages 114

Figure 3.5.20 – Signal path of completed black box worksheet for linear scaling 114

Figure 3.5.21 – Saving the active black box for future applications 115

Figure 3.5.22 – Flow of dialog boxes while saving a black box for future use 116

Figure 3.5.23 – Modified black box for DBK65 transducer channels 117

Figure 3.5.24 – Modifications of the black box for DBK65 transducer channels 119

Figure 3.5.25 – Signal path from the black boxes to the Write Data module 120

Figure 3.5.26 – Specifying data recording options 120

Figure 3.5.27 – Overview of completed worksheet 122

Figure 4.1.1 – Quarter bridge circuit used during laboratory experimentation

with the DaqBook 2000 system 126

Page 11: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

x

Figure 4.1.2 – Completion resistor plug installed in the DBK43A module with top

cover removed 127

Figure 4.2.1 – Strain sensor with quarter bridge strain gage composed of nylon 128

Figure 4.2.2 – Carbon-fiber strain sensor with quarter bridge strain gage (a)

and installation (b) 128

Figure 4.2.3 – Recorded strain levels in carbon-fiber strain sensor and bonded

strain gage 129

Figure 4.3.1 – Final configuration of the strain sensor 131

Figure 4.3.2 – Constructed strain sensors without connection tabs or protective

Coating 132

Figure 4.4.1 – Field installation of the strain sensor to concrete 134

Figure 4.4.2 – Ovalization of a bolt hole under loading 134

Figure 4.5.1 – Four-point bending test used for strain sensor evaluation 135

Figure 4.5.2 – Dimensioned constant-moment beam testing schematic 135

Figure 4.5.3 – Mid-span layout of Strain sensors and complementary strain gages 138

Figure 4.5.4 – Strain sensors installed for the constant-moment beam test 139

Figure 4.5.5 – Complementary strain gages installed on the opposite flange as

Strain sensors for the constant-moment beam test 139

Figure 4.5.6 – Deformed region of washer contact due to tightening the anchorage

nuts to 180 lb-in 142

Figure 4.5.7 – Half bridge temperature compensating circuit 144

Figure 4.6.1 – Illustration of the mapped meshing procedure used 145

Figure 4.6.2 – 20-node brick element used for 3D modeling 146

Figure 4.6.3 – Boundary conditions of the 3D beam model 147

Figure 4.6.4 – The extrusion process uses to build a 3D model of the sensor 150

Figure 4.6.5 – Inadequately restrained model penetrating the steel test beam 152

Figure 4.6.6 – Boundary conditions for the tensile case of sensor model 1 154

Figure 4.6.7 – Boundary conditions for the compression case of sensor model 1 154

Figure 4.6.8 – Boundary conditions for the tensile case of sensor model 2 154

Figure 4.6.9 – Boundary conditions for the compression case of sensor model 2 154

Page 12: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

xi

Figure 4.6.10 – Ring of elements around bolt holes extruded through thickness

of sensor for sensor model 3 155

Figure 4.6.11 – Boundary conditions for the tensile case of sensor model 3 155

Figure 4.6.12 – Boundary conditions for the compression case of sensor model 3 155

Figure 4.6.13 – Boundary conditions for both compression and tensile cases

of sensor model 4 156

Figure 4.6.14 – Detail of boundary conditions imposed to simulate contact with

a washer for sensor model 4 156

Figure 4.6.15 – Boundary conditions for both compression and tensile cases

of sensor models 5 and 6 157

Figure 4.6.16 – Notch for strain relief of wires in sensor 160

Figure 4.6.17 – Longitudinal strain distribution (εy) for tension and compression

cases for the final sensor model 161

Figure 4.7.1 – Weld locations on beam members for the constant-moment load test 164

Figure 4.7.2 – Typical response of strain gages and sensors under applied loading 167

Figure 4.7.3 – Erroneous response of strain gages and sensors under rapid, non-

monotonically increasing loading 168

Figure 5.1.1 – East half of the De Neveu Creek IBRC Bridge indicating

instrument locations 172

Figure 5.1.2 – Section of girder and deck for locating strain sensors 174

Figure 5.1.3 - Composite behavior stress profiles through a typical deck section 174

Figure 5.1.4 – Section view of strain sensors for transverse wheel load distribution 175

Figure 5.1.5 – Plan view of strain sensors to monitor transverse wheel load

distribution 176

Figure 5.1.6 – String potentiometer manufactured by UniMeasure 177

Figure 5.2.1 – Plan of sealed PVC pipes containing lead wire runs for

individual instruments 178

Figure 5.2.2 – PVC piping terminating at the enclosure box 179

Figure 5.2.3 – Installation of the PVC piping housing instrument lead wires

along girder #1 179

Figure 5.2.4 – Detail of a typical screw terminal connection 180

Page 13: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

xii

Figure 5.2.5 – Diagram of lead wires terminated in the screw terminals in the

enclosure box 181

Figure 5.2.6 – Enclosure box housing lead wire connections 181

Figure 5.2.7 – Labels for transverse strain sensors at mid-span 182

Figure 5.2.8 – Labels for girder strain profile sensors at third-span 183

Figure 5.2.9 – Labels for longitudinal strain sensors 184

Figure 5.2.10 – Illustration of the anchorage system for strain sensors 184

Figure 5.2.11 – Approximation of the force acting on the anchor rod. 186

Figure 5.3.1 – Test vehicle dimensions for UM-R/UW-M load test 187

Figure 5.3.2 – Load Test 1 test vehicle locations 190

Figure 5.3.3 – Axle positions of a single girder during Load Tests 1 and 2 191

Figure 5.3.4 – Load Test 2 test vehicle locations 192

Figure 5.3.5 – Load Test 3 test vehicle locations 193

Figure 5.4.1 – Photograph of the data acquisition system used for load testing 194

Figure 5.4.2 – Photograph of the DaqBook 2001 194

Figure 5.4.3 – Photograph of the DBK43A strain gage module with cover

panel removed 195

Figure 5.4.4 – Photograph of the DBK65 transducer module with cover panel

removed 195

Figure 5.4.5 – Analog filter locations on the DBK43A module 197

Figure 5.4.6 – Quarter Bridge circuit and completion resistor configuration for

use with the strain sensors 197

Figure 5.4.7 – Typical pin numbering of a mini-DIN plug 199

Figure 5.4.8 – Pin numbering of mini-DIN plugs used for this project with

typical color-coding 199

Figure 5.4.9 – Location of the excitation voltage jumpers on the DBK 65

circuit board 200

Page 14: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

xiii

List of Tables

Table 2.2.1 –FRP-grillage material properties 17

Table 2.3.1 – Maximum mid-span displacement data from live load tests in bridge

B-20-148 FRP 28

Table 2.3.2 – Maximum deflection data from live load tests in bridge B-20-149

Conventional 29

Table 2.3.3 – Approximate girder distribution factors for load testing of B-20-148

FRP 30

Table 2.5.1 – Load Test #1 - Maximum load test values 43

Table 2.5.2 – Load Test #2 - Maximum load test values 43

Table 3.3.1 – Non-linearity correction factors of Wheatstone bridges 89

Table 4.2.1 – Typical modulus of elasticity values of materials 129

Table 4.5.1 – Torque level test data 141

Table 4.5.2 – Boundary condition test data 142

Table 4.6.1 – Boundary conditions used for the FE of the sensor 153

Table 4.6.2 – Summary of finite element modeling and constant-moment best test

results 161

Table 4.7.1 – Calculation of calibration factors 169

Table 4.7.2 – Calibration factors developed for correction of laboratory

acquired readings 170

Table 5.3.1 – Load Test 1 estimated strain magnitudes 190

Table 5.3.2 – Estimated strains for girder profile sensors installed at third-span of

girders 1 and 2 192

Page 15: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

1

Chapter 1 – Introduction and Summary of Work

1.0 - Introduction

Across the United States a massive network of transportation infrastructure exists.

Starting originally with simple walking trails, this network evolved to include a web of

iron rail lines spurned by the industrial revolution and eventually concrete and asphalt

roads for the automobile. Throughout this progression the bridge has evolved to meet

these demands. From the first wooden structures spanning creeks and rivers, to

incredible displays of steel and concrete that cross gorges, bays and more, the bridge has

become increasingly complex, relying on the development of modern materials, changing

economic conditions and advanced engineering to meet project goals.

Acknowledging the importance of fostering new materials and engineering

methods, the United States Department of Transportation (DOT) created an organization

aimed at continuing the advancement of the tools available for engineers. Initiated under

the Transportation Equity Act for the 21st Century (TEA-21), the Innovative Bridge

Research and Construction Program (IBRC) is a venue for the demonstration of new and

groundbreaking material used in the construction of transportation structures (FHWA

2005). Through this program numerous materials and their applications have been

evaluated for future use in construction. The first installment of funding was allocated

for the period between 1998 and 2004 and accounted for $7 million in research and

development projects and $122 million of construction projects (Conachen 2005). The

Page 16: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

2

program has been extended and funding currently exists for the program through the

2005 fiscal year.

Evaluation of fiber-reinforced polymer (FRP) materials has happened frequently

in the IBRC program. Although the material has been in use for a number of years, its

implementation in infrastructure has been slowed. Sources of this delay stem from

inconsistency in material properties, non-ductile failure mechanisms, and a general

unfamiliarity among designers. FRP composites are composed of oriented fibers,

typically carbon or glass, embedded in a polymeric resin and cured to form a single

material. The matrix of resin and fiber is usually drawn through a die during a process

called pultrusion, pressed into the desired shape prior to the set-up of the resin, or cured

in the final shape intended for the application. Often this process can be costly as the

machinery required may not be readily available to industry and set up of the pultrusion

process can be labor intensive. However, large-scale production can be rapid and very

little preparation is required after the curing process.

Used in lieu of steel reinforcing bars in reinforced concrete, FRP bars or multi-

directional grillages have many advantages. Leading the push in FRP implementation in

infrastructure construction is conventional steel reinforcement’s propensity to corrode. In

2002, 27.1% of the bridges in the United States were classified by the DOT as

structurally deficient or functionally obsolete (ASCE 2005). This statement groups

together structures currently open to traffic that are either not capable of operating at their

design capacities or are being subjected to demands greater than their original design

Page 17: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

3

intended. A major cause of inadequacy for these structures is gradual deterioration of the

steel reinforcing contained within concrete decking. Penetration of water through the

concrete decking in conjunction with high concentrations of chlorides commonly found

in salts used for de-icing and snow removal facilitate this corrosion. Combating this

major source of deterioration, FRP systems are generally not affected by corrosion, and

are immune to the effects of chlorides (Jacobson 2004).

Additionally, FRP materials are capable of developing larger tensile stresses than

steel. Currently, common strengths of steel reinforcing bars reach a maximum of 75 ksi,

while glass-fiber reinforced polymers (GFRP) and carbon-fiber reinforced polymers

(CFRP) have been found to achieve maximum stresses of 230 and 535 ksi, respectively

(Dietsche 2002). These higher stress levels combined with the smaller density of FRP

relative to that of steel, may allow for less material used in design, and in turn offer cost

savings.

1.1 - The De Neveu Creek IBRC Bridge

Completed in April of 2004, the De Neveu Creek IBRC Bridge (WI B-20-148) is

a single-span bridge located on U.S. Highway 151 near Fond du Lac, Wisconsin. A

graphical representation is provided in Figure 1.1.1. The structure carries two lanes of

northbound traffic and contains generous roadside shoulders adjacent to the traffic lanes.

Part of a regional effort to reduce congestion, increase traffic flow and lower crash rates,

Highway 151 has been reconfigured, creating a bypass around Fond du Lac. The De

Neveu Creek Bridge also has a twin immediately to its geographic north, which carries

Page 18: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

4

traffic in the opposite direction. The two bridges are nearly identical with exception

being the reinforcing material of the concrete decks. The northern bridge (WI B-20-149)

has conventional mild-steel reinforcing within its deck, while the southern bridge (WI B-

20-148, the IBRC structure) has a bi-directional FRP grillage.

The De Neveu Creek

IBRC Bridge

Figure 1.1.1 – Wisconsin Highways 151 before (left) the bypass and after (right). Adapted from

Conachen (2005).

The De Neveu Creek IBRC Bridge is of primary importance for this project. The

structure has a 130-foot span and a deck width of nearly 45 feet. The deck is eight inches

thick and supported by seven prestressed concrete stringers, which have been designed

for composite action with the deck. The stringers are set at a uniform spacing of 6’-5”

and are of Wisconsin DOT type 54W. It is of significant interest that the 54W girders

have a top flange width of 48 inches, leaving only 29 inches of unsupported deck.

Page 19: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

5

Figure 1.1.2 – The De Neveu Creek Bridge.

The bi-directional FRP grillage in the deck was fabricated by Strongwell

Incorporated of Chatfield, Minnesota (Jacobson 2004) and was installed in two mats, on

separate levels (Figure 2.2.3). With exception to the exterior deck overhang and parapets

of the bridge, the entirety of the bridge deck is devoid of mild steel reinforcement.

The evaluation of FRP as a construction material was awarded by the IBRC as a

development, construction and long-term monitoring project at the De Neveu Creek

Bridge. The three phases of the project are as follows:

Phase I - Development of a new, cost effective material and necessary design

procedures for such material. The University of Wisconsin-Madison and

design engineers at Alfred-Benesch and Company completed this work.

Phase II – Evaluate the performance of new material for actual implementation in

the De Neveu Bridge. Full material specification was developed, as well

as quality control testing of materials for construction. The material

evaluation and quality control testing was completed at the University of

Page 20: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

6

Wisconsin-Madison (UW). In addition, non-destructive load testing of the

bridge is required to verify design assumptions and distribution of loads

within the deck. The University of Missouri-Rolla (UM-R) conducted a

preliminary load test of both WI-B-20-148/149 in 2004, establishing load

deflection data for comparison with future load tests.

Phase III – Long term monitoring of the bridge. This monitoring is required to

provide an observation of any possible differences between the steel-

reinforced and FRP-reinforced decks exist. Additionally FRP material

properties can be more susceptible to change over time than metallic

material. Observation is thus required to verify the long-term feasibility of

specifying FRP materials in future structures.

The latter segment of Phase II and entirety of Phase III have been retained by Marquette

University and will be carried out between November 2004 and November 2009. An

additional benchmark load test of the bridge is required for long-term analysis and will be

correlated with deflection data produced by UM-R. Subsequent load testing and visual

inspections will provide the basis for long-term evaluation.

1.2 - Objective of Thesis

The objective of this thesis is to develop a cost effective, removable and reliable

strain sensor and data acquisition system to monitor the strain response of WI B-20-148

under live load testing. The portability of the strain sensor is required to limit its

exposure to the outdoor elements, as strain gages are fragile instruments. Additionally,

portability provides the opportunity to re-use the sensor in multiple locations. It is

Page 21: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

7

intended that the sensor will be implemented in a future series of load tests. Additionally,

recommended protocol for load testing is developed for the sensor array with inclusion of

multiple deflection measurements.

1.3 - Scope of Thesis

This thesis presents the development of a strain sensor and data acquisition

system for use in an instrumentation project on the De Neveu Bridge near Fond du Lac,

Wisconsin. A fundamental-level discussion of data acquisition and usage of electrical

sensors is presented within this document as well as the design, laboratory testing and

final development of the removable strain sensors and data acquisition software. Details

pertaining to such topics are discussed herein.

Chapter 2 contains a literature review pertinent to instrumentation and bridge

monitoring, load testing efforts of similar structures and synthesis of such information.

Chapter 3 presents the details of data acquisition hardware, elementary signal processing,

an overview of electrical instrumentation and an introduction to the computer software

used in data acquisition. Chapter 4 discusses the development and calibration of strain

sensors to be used in load testing of the De Neveu Creek IBRC Bridge. Chapter 5

presents recommendations for a future load test, detailing load protocol, installation of

necessary equipment and data acquisition set-up. Finally, Chapter 6 presents

conclusions, recommendations and other observations pertinent to working with

removable strain sensors.

Page 22: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

8

Chapter 2 –Literature Review and Synthesis

2.0 - Introduction

A wide variety of information exists on the multitude of methods for which

diagnostic information can be obtained from structures. The vast amount of equipment is

commercially available and can be daunting to sift through. The selected instrumentation

efforts presented below document the results of a variety of data acquisition systems

operating in differing environments and configurations.

Organization of this chapter begins with general recommendations for the

diagnostic information collection of bridge structures followed by an introduction of the

bridge to be studied in this project. The design, configuration and construction of the De

Neveu Creek IBRC Bridge are documented as well as an overview of preliminary

attempts to obtain benchmark values of strain and deflection data. Four additional load

test projects are also documented, each outlining the data acquisition methods employed

and results of each load test conducted. Finally, a synthesis of the literature documented

in this section is provided, detaining the rationale for development of the strain sensor to

be used in instrumentation and monitoring of the De Neveu Creek IBRC Bridge.

2.1 – Instrumentation Guidelines

As state or federal governments own a majority of bridge structures in the United

States, a number of government agencies have produced documents recommending

procedures for their instrumentation and monitoring. As of recent times, the Federal

Page 23: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

9

Highway Administration (FHWA) produced guidelines for the instrumentation of

bridges, specifically those utilizing high performance concretes in their construction

(FHWA 1996). Similarly, the National Cooperative Highway Research Program

(NCHRP) has developed research initiatives aimed at identifying guidelines for load

testing when rating bridges (NCHRP 1998). Conforming to these guidelines, academia

frequently carries out the load testing of structures. Farhey (2005) provides an excellent

summary documenting the need for diagnostic bridge testing and recommendations for

the instrumentation of structures.

The FHWA publication was created in response to the ever-expanding use of high

performance concretes in practice and the corresponding lack of pertinent research on the

material. The document notes that there are a number of methods available for the

instrumentation of structures; however, this discussion is limited to short-term monitoring

only. For clarity, short-term monitoring is focused on testing that imposes loads on a

structure over a period of a few hours. Specifically, both static and dynamic live load

testing can be considered short-term monitoring. Furthermore, long-term loading

involves monitoring a structure over a significantly longer period, typically months or

years. Long-term monitoring typically focuses on effects due to shrinkage of concrete,

creep of a structure, effects due to cyclic changes in temperature and other time-

dependent effects.

Both the FHWA and NCHRP recommend that short-term strain acquisition be

performed by electrical resistance type gages. Vibrating-wire type gages are not capable

Page 24: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

10

of rapid acquisition but are best suited for long-term monitoring of strains. As in-field

attachment can be difficult, gages to be embedded in concrete should be installed to full-

size sections prior to placement if at all possible. Full-size, instrumented sections of

reinforcement can then be placed in simultaneously with the reinforcement specified

during construction, eliminating the difficulties of field installation. The document also

notes that gages should be adequately protected from both the placement of concrete and

the fresh concrete itself. As each manufacturer produces strain gages of differing

specifications, protection should adhere to the manufacturer’s recommendations.

Furthermore, the FHWA acknowledges that gages can be mounted to exterior surfaces of

hardened concrete. Although more difficult to perform successfully, gages can be

bonded to smooth surfaces, which typically provide an adequate substrate. Troweled,

broom finish and other rough finished surfaces can be more difficult to install gages and

require surface preparation, but have been performed successfully in the past.

Temperature fluctuations are also of importance when obtaining measurements.

Typically electrical resistance strain gages are available with a temperature-compensated

backing to match the intended substrate being monitored. While this backing eliminates

much of the potential thermal effect, no two materials have exactly the same coefficient

of thermal expansion allowing for the possibility of thermal differences between them.

Compensation for these differences is prudent and should be employed for both

measuring instruments and also for any changes in the substrate itself (NCHRP 1998). A

simple solution recommended to address temperature changes is to conduct testing near

sunrise as temperature gradients are at a minimum (FHWA 1996).

Page 25: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

11

Finally, instruments used in any monitoring project require that an appropriate

level of resolution be available. In short-term monitoring values of strain smaller than

100µε are common (FHWA 1996). Usage of high impedance strain gages, typically 350

or 1000 ohms, improves the signal-to-noise ratio of measurements (NCHRP 1998).

Resolution of instruments also requires analysis of region of the substrate to be sampled.

When monitoring a heterogeneous substrate, e.g. reinforced or prestressed concrete, large

gage lengths are required to eliminate local effects (Farhey 2005). Although use of a

larger gage length averages measurements over a region, it also limits local effects that

may omit valuable readings.

At the time of publication a single, reliable method of measuring displacement

was felt to be non-existent for bridge girders (FHWA 1996). However, the use of

calibrated surveying equipment or taut-wire measurement has proven to be successful in

practice. Taut-wire measurements require the installation of a wire, stretched between

two known points of reference with a known tensioning force. Measuring the movement

of girder relative to the wire can produce displacement values. However, utilization of

precise surveying equipment may offer greater flexibility when site conditions limit

physical contact-type measurement of displacements on a bridge. Placement of optical

sensors, prisms, or other similar surveying equipment on the structure allow for it to be

observed from a distance using a calibrated surveying station. Displacements can also be

measured with electrical transducers, e.g. potentiometers, linear variable differential

transformers (LVDT’s) or dial gages but require a stable mounting location. These

Page 26: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

12

methods are typically not practical for displacement monitoring of long-span girders and

best suited for local measurements.

Specific product recommendations are made by the FHWA (1996). The following

instruments are recommended for use in the instrumentation of structures and monitoring

of bridge superstructures and substructures.

Short-term monitoring:

Internal adhered gages on steel reinforcement -

• Micro Measurements CEA-06-250-UW-350 or CEA-06-250-UW-120

• Micro Measurements CEA-06-250-AE-350

External adhered gages on hardened concrete -

• Micro Measurements EA-05-20CBW-120 or EA-06-20CBW-120

• Micro Measurements EA-05-40CBY-120 or EA-06-40CBY-120

External weldable gages on structural steel -

• Texas Measurements TML AWC-8B

Long-Term Monitoring:

Vibrating Wire Gages –

• Geokon VCE-4200 or VCE-4210

• Roctest EM-5

It should be noted that a substantial body of knowledge regarding bridge

monitoring and instrumentation exists in the form of various journal articles, research

papers and other engineering publications. In fact, a substantial portion of mechanical

Page 27: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

13

measurement curricula may be applied to diagnostic bridge monitoring in the form of

displacement and strain measurement. The documents presented in this section are

intended to illustrate that significant efforts focusing on structural bridge monitoring have

previously been performed by a number of agencies and organizations, and those

reviewed are most pertinent to the current effort.

2.2 – The De Neveu Creek IBRC Bridge

As part of the ongoing effort to further develop a new suite of construction

materials available to the engineer, extensive research and development has been

conducted in the state of Wisconsin under the guise of the IBRC program. Details of the

IBRC program can be found section 1.1, which introduces the design and construction of

the De Neveu Creek Bridge near Fond du Lac, Wisconsin.

Conventional in appearance, the De Neveu Creek IBRC Bridge (WI B-20-148) is

located on U.S. Highway 151 south of Fond du Lac, Wisconsin and is part of a new

bypass system. A photograph of the structure can be found in Figure 2.2.1. A partial

plan set of the De Neveu Creek Bridge, as well as its southbound counterpart (B-20-149),

may be found in Appendix A. The bridge is a single-span structure approximately 130’

long and carries two northbound lanes of highway traffic. The structure is skewed

approximately 25 degrees and contains minimal superelevation (0.056 feet/foot). Seven

prestressed concrete stringers, or girders, support the 8” thick FRP-grillage reinforced

concrete deck and are intended to act compositely with the FRP-reinforced deck. Shear

transfer is provided by epoxy-coated #4 mild steel reinforcing bars protruding

Page 28: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

14

approximately 6” from the top flange of each girder and extending transversely 1’-0”,

parallel to the plane of the deck. These shear transfer bars are placed in duplicate, but

opposite transverse directions and longitudinally 1’-6” on center. Stringers are of

WisDOT type 54W and spaced transversely 6’-5” on centers. The 54W girders present

an interesting condition when supporting the deck as the top flange of the stringer is 4’-0”

wide. This large top flange dimension leaves an unsupported deck span of only 2’-5.”

Figure 2.2.2 provides a cross section of the bridge and illustrates this narrow spacing of

the stringers. Also, the concrete specified in the construction of the prestressed stringers

was to be of 9,600 psi minimum compressive strength, which can be considered high

strength concrete (Mindess et al. 2003). The 28-day compressive strength of the deck

concrete was specified as 4,000 psi. Finally, conventional mild-steel reinforcing bars are

used only on the exterior overhangs of the deck and integral parapet guardrails, as

federally approved FRP-reinforced parapets were not available during the design phase

(Jacobson 2004).

Figure 2.2.1 – The De Neveu Creek Bridge, WI B-20-148.

Page 29: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

15

Figure 2.2.2 – Cross section of the De Neveu Creek Bridge.

2.2.1 - FRP Reinforcement in the De Neveu Bridge

The FRP grillage reinforcement merits a detailed description. Initiated as part of

an IBRC directed project through the University of Wisconsin-Madison and Alfred

Benesch & Company, a system of pultruded FRP-reinforced concrete was developed for

implementation at multiple bridge locations. Jacobsen (2004) notes that design of the

FRP-reinforced slab conforms to the recommendations of ACI-440, the State-of-the-art

Report on Fiber Reinforced Plastic Reinforcement for Concrete Structures. The FRP

reinforcement is a bi-directional grating system consisting of two individual layers of

reinforcement, with one layer placed directly over the other layer. Figure 2.2.3 illustrates

the double-mat FRP grillage.

Figure 2.2.3 – Assembled FRP grillage (Conachen 2005)

Page 30: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

16

Each grating layer contains two separate types of pultruded FRP elements. The primary

reinforcing member is a 1.5” high “I” bar spaced at 4” on centers and is positioned in the

transverse direction of the deck, perpendicular to the traffic lane. Running orthogonal to

the I-bars, or parallel to the direction of traffic, are cross-rods spaced at 4” on centers.

Each cross-rod is constructed of three independently pultruded elements, which

are assembled in the manufacturing facility. Figure 2.2.4 illustrates the grillage

components. The center cross-rod section is first notched for the I-bars by a CNC

machine and then threaded into the I-bars. The top and bottom “wedge” pieces are then

pressed onto the center cross-rod and adhered with resin. The wedge pieces lock the I-

bars and cross-rods into a 4” by 4” grillage. A second layer is then affixed to the

previous with shear connectors creating a double-mat grillage (Conachen 2005). The

shear connectors create a constant 2.5” separation between grillage mats and are

positioned 36” transversely (along the I-bars) and 24” longitudinally (Jacobson 2004).

Material properties for the pultruded FRP-grating may be found in Table 2.2.1.

Figure 2.2.4 – Cross sections of FRP materials used (Conachen 2005).

Page 31: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

17

Table 2.2.1 –FRP-grillage material properties (Dietsche 2002).

2.2.2 - Development and Testing of the FRP-grillage Reinforced Deck

Extensive evaluative testing of the FRP-grillage reinforcement for implementation

in reinforced concrete was conducted. Early experiments with commercially available

FRP materials included the testing of concrete slabs reinforced with various “T-bar”

grillages that resulted in rapid shear failure of the systems (Bank et al. 1992a; Bank et al.

1992b). A number of two-span, FRP-reinforced slabs were constructed with double mats

of grillages, subjected to design loads and ultimately tested to failure. Additionally, a

conventionally mild-steel-reinforced slab was built as a comparison specimen. Both

spans of the test were 8’, intending to mimic spans between bridge girders. Test results

indicated that the FRP-grillage systems consistently experienced ultimate failure at loads

Page 32: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

18

greater than that of conventionally reinforced systems. However deflections observed in

the FRP-reinforced systems were greater than their conventional counterpart. While

greater loads were carried, the rapid shear failure of the FRP-reinforced slabs warranted

concern as little warning was observed prior to failure. A follow-up study utilizing

similar materials with shorter span lengths was also conducted further investigating the

shear failure of FRP-reinforced slabs (Bank and Xi 1995). It was identified that the FRP-

reinforced slabs behave very similar to conventionally reinforced concrete up until their

rapid punching shear failure.

Jacobson (2004) performed significant additional testing specific to this bridge

project utilizing similar geometric configurations and materials found in the De Neveu

Creek IBRC Bridge. As expected, the research carried out indicated that punching shear

is the expected strength limit state for the FRP-reinforced deck. Like the tests of Bank, et

al (1992, 1995) this limit state was observed at load levels that were many times greater

than the design load. For clarity, FRP-reinforced slab was designed to accommodate an

AASHTO HS-20 truck (Jacobson 2004). Overall, it was found that when using geometry

representative of the De Neveu Creek IBRC Bridge, the strength requirements would be

satisfied with a factor of safety greater than ten.

Serviceability requirements were also studied in depth, providing deflection

response of the system. Warranted by the greater deflections observed in early testing

(Bank et al. 1992a), two separate types of load tests were conducted during

experimentation by Jacobsen (2004). A series of simply supported slabs were

Page 33: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

19

constructed, each spanning 6’-6” and loaded with a 16-kip “wheel load” representing the

HS-20 design vehicle. Two additional slabs were constructed with end restraints more

closely representing the true behavior of the bridge. These slabs also had a clear span of

6’-6”, however, rotational restraint was provided beyond the end of each support.

Figures 2.2.5 and 2.2.6 illustrate the test configurations. Overall, satisfactory span-to-

deflection ratios were consistently observed with deflection ratios greater than L/1330 for

simple spans and L/7200 for restrained slabs.

Figure 2.2.5 – Simple-span slab tests (Jacobson 2004).

Figure 2.2.6 – Restrained end slab tests (Jacobson 2004).

It is important to note that although performance of the slabs in the laboratory indicated a

satisfactory response, the support conditions of each slab tested as well as the overall

Page 34: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

20

span lengths did not accurately represent the actual configuration of FRP-reinforced slabs

in the De Neveu Creek Bridge.

2.3 - Load Testing of the De Neveu Creek Bridge

Documentation of load testing for the De Neveu Creek Bridge can be found in the

work of Conachen (2005). To identify benchmark values of strain and deflection for the

in-situ structure, the University of Missouri-Rolla was contracted to perform a load test.

The test would involve both bridges at the site (B-20-148 and 149) to provide

comparative data for the FRP-reinforced and traditionally reinforced structures. A

robotic total station was employed to capture deflection data with surveying prisms. The

prisms were located primarily at the mid-span of each bridge with reference prisms

placed at quarter points. Figure 2.3.1 indicates the locations of each surveying prism.

Note that on the De Neveu Creek Bridge (B-20-148) a prism was located on both the east

and west abutments while another was located on the west abutment of B-20-149. These

prisms mounted to the abutments were assumed to remain static during testing and

provided stable reference points for the robotic total station to make its readings.

Page 35: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

21

Figure 2.3.1 – Layout of surveying prisms and strain gages. Adapted from Conachen (2005).

2.3.1 - Strain Gage Instrumentation of the Bridge

A modest array of gages was installed at various locations within both bridges.

Gages were installed directly on the FRP grillage and embedded in the concrete, while

others were applied directly on the surface of the deck. The gages installed on the FRP

were placed in two sub-arrays, each consisting of paired gages in various locations. One

sub-array was located directly over girder number 6 (second girder from the south of B-

20-148) and did not include the lower mat of the FRP grillage. Intent of these gages was

to measure strain caused by negative moment in the deck over the stringer. Surface

mount gages were also provided at this location to produce outer surface strain data in an

attempt to identify the strain profile through the thickness of the deck. These gages were

paired and located longitudinally in three installments. No indication of the longitudinal

spacing can be found in Conachen’s work, however. Illustrations depicting the locations

Page 36: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

22

of strain gages installed on the De Neveu Creek Bridge can be found in Figures 2.3.2 and

2.3.3.

Figure 2.3.2 – Strain gage locations. Adapted from Conachen (2005).

Figure 2.3.3 – Strain gage locations. Adapted from Conachen (2005).

The other sub-array was installed on both FRP-grillages at mid-span between the

number 6 and 7 stringers. Intent of these gages was to obtain strain data relating to

positive moment between girders. As with the negative moment gages, surface mount

gages were installed on both top and underside of the FRP-reinforced concrete deck. The

number of gages installed was not clear, however, it is understood that at least six gages

Page 37: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

23

were installed in three pairs, similar to the layout as the gages over girder number 6.

Again, the longitudinal spacing of the strain gages installed was not provided. .

All gages embedded in concrete were protected, as the instruments are highly

sensitive and placement conditions of fresh concrete near strain gages is an unfriendly

environment. The author notes that a “fast-drying silicone” was used to protect the

gages, contrary to commonly recommended and typically performed methods. Lead

wires for the gages were placed in plastic tubing and terminated in a junction box on the

underside of the deck.

Product numbers for the instruments were as follows:

• Embedded Gages – Micro Measurements CEA-06-250UN-350 (350 ohm,

0.25” gage length)

• Surface Gages – Tokyo Sokki Kenkyujo PL-60-11 (120 ohm, 2.362” gage

length)

2.3.2 - In-Situ Load Test of B-20-148 FRP Bridge

The load testing of the De Neveu Creek Bridge was conducted with a total of six

WisDOT provided dump trucks. Each truck was a three-axle vehicle with total weight of

approximately 75,000 lbs. Four load tests, or “stops,” were conducted in the following

arrangements (Figure 2.3.4):

• Test 1 – A three-truck “train” located along both girders 2 and 6 with a

wheel load directly over girders 2 and 6. Six total trucks.

Page 38: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

24

• Test 2 – A three-trucks “train” located along both girders 3 and 5 with a

wheel load midway between deck spans. Spans directly loaded were

between girders 2 and 3 and between girders 5 and 6. Six total trucks.

• Test 3 – A single four-truck “train” located along girder 6 with wheel

loads midway between girders 6 and 7.

• Test 4 – A single four-truck “train” located along girder 6 with wheel

loads midway between girders 1 and 2.

In each test the trucks were positioned appropriately and stopped for a minimum

of 10 minutes to allows for settling (Bank 2005). Deflection data were obtained in

duplicate for all sensors and then the trucks were removed. It is important to note that

significant strain data was not recorded for the De Neveu Creek Bridge as the strain

equipment was not operating properly during testing (Bank 2005).

Page 39: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

25

Figure 2.3.4 – Stopped vehicle locations for live-load testing of B-20-148 FRP (Bank 2005).

Page 40: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

26

2.3.3 - In-Situ Load Test of B-20-149 Conventional Bridge

The load testing of B-20-149 was conducted with the same six WisDOT dump

trucks provided for testing of the De Neveu Creek Bridge. Three load tests, or “stops,”

were conducted in the following arrangements (Figure 2.3.5):

• Test 1 – A single four-truck “train” was located along girder 8, centered

laterally over the girder.

• Test 2 – Three-truck “trains” were located along both girders 3 and 8 with

a wheel load directly over girders 3 and 8. Six total trucks.

• Test 3 – A single four-truck “train” was located along girder 4, centered

laterally over the girder.

Page 41: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

27

Figure 2.3.5 – Stopped vehicle locations for live-load testing of B-20-149 Conventional (Bank 2005).

2.3.4 - Results of B-148 and B-149 Load Tests

Maximum deflection of the De Neveu Creek Bridge (B-20-148) occurred in

girder 4 of Load Test 2 and was approximately 0.67” downward. A plot of deflection

data can be found in Figure 2.3.6. Maximum strain read from the gages occurred during

Page 42: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

28

load test 3 was + 42 µε in the bottom FRP-grillage between girders. It should be noted

that all strain data recorded in the testing of the De Neveu Creek Bridge is in question as

a number of gages failed during testing. Maximum deflection values listed in Table 2.3.1

were approximated from the graphs presented for each load test.

Figure 2.3.6 – Deflection plot of mid-span girder response in bridge B-20-148 FRP. Adapted from

Conachen (2005).

Load Test Girder ∆ (mm) ∆ (in)

1 4 -14.5 -0.57

2 4 -17.5 -0.69

3 1 -11 -0.43

4 7 -12 -0.47

Maximum Mid-span Displacement

Table 2.3.1 – Maximum mid-span displacement data from live load tests in bridge B-20-148 FRP.

Adapted from Conachen (2005).

Page 43: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

29

Maximum deflection of the conventionally constructed bridge (B-20-149)

occurred in girder 4 of Load Test 2 and was approximately 0.47” downward. This is

expected as the conventional bridge has two additional girders that provide additional

stiffness to the structure. A plot of deflection data can be found in Figure 2.3.7. Table

2.3.2 contains a summary of maximum mid-span deflections during testing.

Figure 2.3.7 – Deflection plot of mid-span girder response in bridge B-20-149 Conventional. Adapted

from Conachen (2005).

Load Test Girder ∆ (mm) ∆ (in)

1 7, 8, 9 -9.5 -0.37

2 7 -12 -0.47

3 3, 4 -8 -0.31

Maximum Mid-span Displacement

Table 2.3.2 – Maximum deflection data from live load tests in bridge B-20-149 Conventional.

Adapted from Conachen (2005).

Page 44: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

30

Overall, the deflection data for both structures provides insight into the

distribution of load between girders. In load cases designed to utilize both lanes of the

bridge (load cases 1 and 2 for the De Neveu Creek Bridge) multiple girders were

activated to resist the loads. Additionally, when attempting to localize loads to a specific

girder (load cases 3 and 4 for the De Neveu Creek Bridge), girders adjacent to the loads

were also activated to distribute them accordingly. Distribution factors for each girder

were computed using the expression, ii

TOTAL

DFδ

δ=∑

where δi indicates deflection of

the i-th girder. A listing of approximate girder distribution factors (GDF) can be found in

Table 2.3.3. Based upon these results, the AASHTO girder distribution factors

(AASHTO Section 4.6.2.6) were observed to be conservative as the prescribed formulas

neglect outside factors that can provide additional stiffness to the structure (AASHTO

1996; Bank 2005). Also, it was acknowledged that the strain data collected is not of any

significant use (Bank 2005).

Girder 1 2 3 4

1 0.11 0.10 0.03 0.22

2 0.13 0.12 0.07 0.22

3 0.16 0.17 0.11 0.20

4 0.17 0.19 0.16 0.16

5 0.16 0.17 0.20 0.10

6 0.13 0.12 0.21 0.07

7 0.11 0.10 0.20 0.03

Interior Girder

Exterior Girder

0.54

0.49

0.38

0.45

Approximate GDF from Load Tests

Load Test

2 Lane Loading 1 Lane Loading

AASHTO GDF

Table 2.3.3 – Approximate girder distribution factors for load testing of B-20-148 FRP (AASHTO

1998; Bank 2005).

Page 45: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

31

2.4 – Ohio Bridge HAM-126-0881 (Lenett et al. 2001)

The structure under consideration for this investigation was a three-span steel

girder bridge with a conventionally reinforced concrete deck. Construction of the bridge

started in 1995 and it was commissioned in 1997. With a goal being to produce a

complete scientific view of the loads typical bridge structures endure over the course of

their service lives, the researchers monitored loads and displacements present in the

bridge for nearly all aspects of the project. Data was recorded during fabrication of the

steel stringers, during transportation to the jobsite and through erection. Long-term

strains and temperature data are still being monitored today through a permanent data

acquisition system. The effort put forth by the researchers for this investigation and

subsequent evaluation was exhaustive and included a multitude of topics related to

conventionally constructed steel stringer bridge structures. For this reason, only aspects

of the project’s instrument evaluation and selection and live load testing were reviewed.

As noted previously, the structure considered for this project was a two-lane, three-span

bridge with a reinforced concrete deck supported by steel wide-flange girders. Figure

2.4.1 illustrates the bridge studied. All components of the bridge were constructed by

conventional means and did not utilize any experimental materials or construction

methods. Five steel stringers support the mild steel reinforced concrete deck that rests on

two intermediate piers and terminate atop the abutments. Exterior spans are

approximately 40’ while the center span is approximately 88’. The length of each span is

noteworthy as the exterior–to-center ratio is only 0.45. The authors make note that the

Ohio Department of Transportation recommend a exterior-to-center ratio more near 0.8 to

Page 46: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

32

equalize the positive bending moments of each span. The exterior stringers were

W36x194 shapes while the center span girders are W36x182 shapes. Spliced connections

are utilized approximately 15’ from each pier in the center span. Additionally, only the

center span stringers are designed for composite action with the deck. This is achieved

by using 1” diameter x 5” tall steel studs spaced along the length of the stringer. No

provisions were made outside of the center span to employ composite action.

40.2’ 88.5’ 40.3’

43’

West Pier

West AbutmentEast Pier

East Abutment

N

4 @ 9’-9”

Figure 2.4.1 – Schematic of Ohio Bridge HAM-126-0881.

2.4.1 - Data Acquisition System Employed

The researchers conducted an extensive evaluation of commercially available

instrumentation equipment citing a number of conclusions. Vibrating wire strain gages

are most suitable for use in long-term monitoring projects. Additionally, integral

temperature sensors are available with this type of gage. Vibrating wire strain gages can

be used for short-term live load testing, however acquisition time of the gages is

significant and requires long vehicle stoppage and settling periods. Electrical resistance

gages are of excellent use for short-term, high speed testing, however, use for long-term

testing is not recommended due to their propensity to lose their datum or drift from their

initial reading.

Page 47: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

33

Reliable data collection is possible from a number of acquisition systems available

commercially. However no system is best suited for all projects. The two systems used

for this project are outlined below:

o The slow-testing devices were read using a rugged and readily mobile

system produced by Campbell Scientific. The CR-10 model could run

continuously on 12-volt battery DC power. The unit could scan only one

channel at a time and obtains data at 64 Hz.

o The high-speed devices were read using a MEGADAC system produced

by Optimum Electronics. The system utilized a high-speed interface (up

to 25 kHz) between the analog-to-digital converter and a computer. This

allowed for ample sampling of data during higher speed testing. Due to

continuous electrical requirements, its susceptibility to damage by

weather, and its high cost, use of this system was limited to the high-speed

devices and installed in a permanent structure located near the bridge.

Displacement transducers used for the project were Celesco PT101-SWP string

potentiometers and Trans-Tek 244 DC-LVDT linearly variable differential transformers.

Electrical resistance gages selected for the high-speed data acquisition varied

according to their installation locations. Gages to be mounted on the steel stringers were

of weldable and manufactured by Texas Electronics, product AWC-8B. Installation

requires spot welding of the stainless steel enclosure directly to the steel substrate. Strain

gages of this type were also located on the transverse diaphragms, or cross-frames, of the

Page 48: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

34

bridge in multiple locations. Gages to be installed in the concrete deck were of

embedded type and cast directly into specified location in the concrete. Special care was

taken during casting of the deck to ensure correct location of each sensor. The embedded

sensors were Micro Measurements EGP series gages. Acknowledging that this project

included near-continuous monitoring of over 600 independent channels of data, the

recommendations conveyed within the documentation are held in high-regard.

2.4.2 - Load Testing and Results

Two live load tests were conducted. Vehicles specified for testing were two

three-axle dump trucks, of which the independent loads were documented at the time of

testing. It was acknowledged that the weight of each truck pair varied from the

benchmark to in-service tests and properly recognized in all following results.

The first test was a static, post-construction test to benchmark the load and

displacement data of the structure prior to traffic loading. Eleven different load cases

were conducted at varying locations to profile the strain response of the structure.

Location of test vehicles is illustrated in figure 2.4.2. Each load case consisted of

locating the test vehicles at points of interest along the spans. The trucks were always

positioned adjacent to each other, or longitudinally in a tailgate-to-tailgate fashion.

Page 49: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

35

Figure 2.4.2 – Static live-load Cases A through L (Lenett et al. 2001)

A follow-up load test was conducted once the structure had been in service for

over one year. Similar truck positions were utilized as the benchmark test; however, the

in-service condition prohibited locating trucks adjacent to each other. In order to conduct

each load case, control measures were installed to limit traffic to only a single lane of the

Page 50: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

36

bridge. To obtain data for each load case, the test vehicle was positioned in the closed

lane next to the open traffic lane. When ready, temporary traffic stops were imposed to

eliminate transient loading from passing vehicles and data collected. As only a single

lane of the bridge was loaded with a test vehicle, as opposed to the twin loading of the

benchmark test, corresponding results were then superimposed for comparison.

Results from the two sets of load tests yielded the following conclusions. The

intermediate cross-frames contributed to the internal redundancy of the structure and

spread the distribution of loads laterally throughout the structure. These frames were

located at 14’ intervals between all stringers and contained weldable strain gages that

were active during all load tests. Composite action of the stringers and deck exists

throughout the center span, which was intended for in design. Partial composite action

was observed in exterior spans during the benchmark load test. This partial composite

behavior, although common in structures of this type, was not intended. However, after

completion of the second load test, the eastern exterior span had lost all indication of

partial composite action while the western exterior span had decreased its degree of this

behavior.

Load distribution factors for each stringer on the structure were calculated based

on the results of each load test. The authors present a method by which the internal

moment of each stringer can be calculated. Load distribution factors were then computed

using:

GirderGirder

AllGirders

MDF

M=

∑ (2.4.1)

Page 51: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

37

The method used to develop internal moments within each girder is as follows:

Figure 2.4.3 – Method for calculating internal moments in stingers (Lenett et al. 2001)

2.5 – South Carolina Route S655 (Turner 2003)

The new Route S655 Bridge over the Norfolk/Southern rail line near Landrum,

South Carolina, replaced an antiquated steel and timber deck structure. The previous

two-lane structure had been in service as early as 1946 and was not in sufficient condition

to safely carry two lanes of modern traffic. Completed in 2001, the new structure spans

60 feet with five steel stringers and a unique glass-fiber reinforced polymer (GFRP) deck.

The W36x150 stringers are located with an 8’-0”center-to-center spacing, which, as

Page 52: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

38

indicated by the author is intended to challenge the limits of the GFRP deck. Also, to

better understand the behavior of the GFRP system, the bridge was instrumented and load

tested. The structure is currently under traffic.

2.5.1 - GFRP Deck Panels

The commercially available deck panels are composed entirely of built up

sections, each consisting of approximately ten pultruded elements. The Duraspan®

panels were produced by Martin Marietta Composites

(www.martinmarietta.com/Products/ composites.asp) and have a number of successful

installations around the country. Each element is 37.2’ wide (corresponding to the bridge

deck width), 7.6” deep and 12” long and is connected to adjacent elements with an

adhesive resin. Pre-assembled panels of these elements were delivered to the site and

installed longitudinally atop each stringer. Each pre-assembled panel was the full width

of the bridge deck and 10’ long. Significant fieldwork was required to install each

section as each individual panel required adhesion to adjacent panels and hand laid FRP

splice panels to prevent moisture penetration. Additionally, each deck panel is designed

to act compositely with the steel stringers and thus significant investigation of the

connection’s shear transfer performance is documented. An illustration of the deck panel

and the installed configuration can be found in Figures 2.5.1 and 2.5.2.

Page 53: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

39

Figure 2.5.1 – Individual Duraspan® deck panels. Adapted from Turner (2003).

Figure 2.5.2 – Section of bridge deck and integral grout-filled shear pockets (Turner 2003)

Shear transfer for composite action of the GFRP deck and steel stringers was

achieved using three 7/8” diameter by 3” tall steel studs encased in a high-strength grout

pocket. The studs were placed along stringer lines and spaced at 2’-0” on centers. To

verify the actual composite response of the deck and stringers, both full-scale load tests

Page 54: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

40

and laboratory testing were by Turner (2003). While the experimental design of the

structure incorporated composite behavior the stringers were designed to support the

bridge in a non-composite manner. This over-designing of the bridge allowed for an

acceptable level of safety in the event that the steel studs did not behave as intended.

2.5.2 - Instrumentation and Load Testing

A variety of instruments were installed on the bridge for the data acquisition

during load tests. Duplicate electrical resistance strain gages were installed at eighth

points along the span. Weldable gages were installed on the steel girders and oriented

longitudinally to obtain strain distribution through the depth of the stringers.

Complementing the weldable gages, adhesive-applied gages were installed on the GFRP

deck in both longitudinal and transverse directions. The transverse gages on the deck

were intended to provide strain data relating to the behavior of the deck in resisting wheel

loads. Longitudinal gages were intended to produce strain data that would relay

information pertinent to the degree of composite behavior of the deck and stringers. In

addition to the strain gages, draw wire transducers (DWT) were installed to measure

vertical deflection of the deck relative to the top of the stringers. Finally, surveying

prisms were installed at locations along the lower flange of the stringers to monitor the

deflection. Figure 2.5.3 illustrates the layout of instruments.

Page 55: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

41

Figure 2.5.3 – Instrument layout of S655 Bridge (Turner 2003)

The load test protocol was repeated in two different test sessions and utilized

three-axle dump trucks classified between an AASHTO HS23-44 and HS25-44 load.

Five load cases were conducted and are illustrated in Figure 2.5.4. The goals for Load

Combination 1 and 2 were to determine behavior in both instrumented and un-

instrumented areas of the structure. Load Combination 3 was used to determine behavior

of the panels under two-lane loading while Load Combination 4 aimed at observing the

response of the GFRP deck over an interior stringer (negative bending behavior).

Finally, Load Combination 5 was used to determine positive moment response of the

GFRP deck between stringers (positive bending behavior).

Page 56: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

42

Figure 2.5.4 – Live-load test cases for S655 Bridge (Turner 2003)

To perform the test for each load case the specified truck(s) were driven to

instrumentation points and stopped for data acquisition. Furthermore, each individual

load case was performed twice.

The magnitudes of strain recorded were reasonably consistent between the two

days of testing and of reasonable magnitude. A summary of maximum deflection and

strain values for both load tests is given in Tables 2.5.1 and 2.5.2.

Page 57: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

43

1 2 3 4 5

Max Girder ∆ (in) -0.144 -0.156 -0.288 -0.156 -0.156

Max Girder uStrain 40 33 78 42 42

Max Deck ∆ (in) -0.019 +0.006 -0.013 -0.017 -0.020

Max Deck uStrain 60 10 95 64 106

Load Case

Load Test #1

Table 2.5.1 – Load Test #1 - Maximum load test values.

1 2 3 4 5

Max Girder ∆ (in) -0.120 -0.108 -0.216 -0.120 -0.132

Max Girder uStrain 40 34 72 40 52

Max Deck ∆ (in) -0.019 +0.006 -0.015 -0.020 -0.018

Max Deck uStrain 72 12 90 100 82

Load Test #2

Load Case

Table 2.5.2 – Load Test #2 - Maximum load test values.

Strain distribution through the depth of the cross-section was analyzed to evaluate

the degree of composite action between girders and GFRP decking. It was noted that the

magnitude of many of the values recorded in these load tests were equal to or smaller

than the accuracy of the data acquisition system (Turner 2003). Thus strain distribution

data from Load Case 3 was used to draw conclusions as the magnitude of its strain data

was generally much higher than the other four load cases. Based on Load Case 3 it was

concluded that partial composite action was present between the girders and deck. This

conclusion was also consistent with behavior observed in laboratory testing of steel stud

specimens. During the laboratory testing, the equivalent shear loads found in a

composite system were not attained by the deck material. Specifically, it was observed

Page 58: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

44

that the decking material continually failed in localized regions before the expected

strength of the grout pocket could be achieved.

Moment distribution factors of the steel stringers were also studied and compared to

design procedures found in the 1996 AASHTO Standard Specification and the 1998

AASHTO LRFD Manual. Experimental distribution factors calculated from strain data

used the following equation:

g

jj

all irders

strainDF

strain=

∑ (2.5.1)

Reviewing the results it was found that the experimental distribution factors were

consistent with the AASHTO predicted values.

2.6 – Fairground Road Bridge

The bridge studied in this document was a three-span, two-lane structure spanning

the Little Miami River in Greene County, Ohio (Bridge Diagnostics Inc. 2002). The

tested structure is composed of FRP deck panels installed on existing steel stringers.

Martin Marietta Composites (www.martinmarietta.com/Products/composites.asp)

manufactured each hollow core FRP panel and is the same product studied by Turner

(2003) in the South Carolina S655 Bridge. Similarly, composite action is achieved by

using three, 7/8” diameter x 3” tall steel studs in a cellular pocket filled with high

strength grout. The pockets are also located at 2’-0” on-centers along the length of each

stringer. The focus of investigation for this project was primarily the analysis of

composite behavior between the FRP deck panels and steel stringers and, while not

pertinent to this project, the load rating of the structure.

Page 59: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

45

2.6.1 - Study of Composite Action and Strain Measurements

To study the composite behavior of the deck system and stringers, strain transducers

manufactured by Bridge Diagnostics Incorporated (www.bridgetest.com/index.htm) were

installed on the stringers of the structure with a small number of transducers installed

directly on the FRP deck panels for verification of results. Four locations along the

length of the bridge were selected as instrumentation points:

• 4’-0” from abutment

• Mid-span of outer, shorter span

• 4’-0” from support pier

• Mid-span of center, long span

These locations provide the opportunity of symmetry to study the structure, saving cost of

installation on the entire bridge. A top and bottom flange longitudinal transducer was

installed on each of the stringers at instrumentation points for a total of 32 units (Figure

2.6.1). For verification of strain distribution through the section of the bridge, two

additional longitudinal transducers were installed on the FRP deck near the top flange of

an interior stringer at mid-span of the outer span. Also, two transducers were installed

transversely on the FRP deck between stringers to monitor the bending behavior of the

FPR deck itself. To monitor vertical displacement of the FRP deck, four linearly varying

differential transformers (LVDT) were installed atop the pier as well.

Page 60: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

46

Figure 2.6.1 – Instrumentation layout of the Fairground Road Bridge

A driving factor in the decision to use strain transducers in lieu of adhered foil

gages is the possibility of removal between test sessions. Also, the speed at which the

equipment can be installed or removed is rapid. Removal of the equipment limits the

possibility of vandalism and degradation due to weather. A photograph of the strain

transducer may be found in Figure 2.6.2.

Figure 2.6.2 – Bridge Diagnostics Strain Transducer (BDI 2005).

2.6.2 - Load Testing and Results

The load test consisted of slowly (less than 5 mph) driving a three-axle dump

truck across the structure in a series of four prescribed paths. The authors did not

Page 61: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

47

disclose detail of load location but did note that duplicate runs were performed to check

consistency of data. Stationary, static load testing of the structure was not performed.

While truck passes were being made, continuous monitoring of the sensors occurred.

Relative distance of the vehicle along the bridge was also monitored. It is of note that

data acquisition of the live load test was sampled at a rate of 40 Hz. A final high-speed

test was also conducted with the test vehicle moving at approximately 45 miles-per-hour

to estimate the impact effect of design vehicles.

The data collected produced a number of interesting results. Using the

assumption that elastic response is observed in the structure, the authors calculated the

neutral axis of each stringer based on the strain readings recorded. Figure 2.6.2 illustrates

the method used for neutral axis computation. The following equation was used to locate

this axis relative to the bottom flange of each stringer:

( )

bottomna

top bottom

dy

εε ε

•=

− (2.6.1)

The distance between reading surfaces of the transducers is noted as “d” in the

above expression (Figure 2.6.2). Based upon these calculations the neutral axis of each

stringer was found to be consistent with others in the structure and also indicated some

degree of composite nature. That is, the neutral axis of each stringer was calculated to be

above mid-depth of each W-section, indicating an imbalance of tensile and compressive

strains within the steel shape. This corresponds to the condition found in Figure 2.6.3.b.

By providing a mechanism for shear transfer, namely the shear stud pockets, some level

of shear load is distributed into the deck. However, as the shear load transferred is less

than what would be present in a fully composite section (Figure 2.6.3.c), the steel

Page 62: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

48

stringers and FRP deck observed in this project are termed partially composite. Verifying

this observation were the strain levels recorded by transducers mounted directly to the

FRP deck. It was found that projected strains based on neutral axis computations were

consistent with the strains recorded by the few transducers mounted to the FRP deck.

Figure 2.6.2 – Method used to locate the neutral axis of stringers. Adapted from Bridge Diagnostics

Inc. (2002).

Figure 2.6.3 – Comparison of composite nature stress profiles through a typical deck section (Lenett

et al. 2001)

Page 63: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

49

It was noted that the response of each longitudinal transducer pair (top and bottom

of individual stringer) was not completely identical. Results indicated that when the test

vehicle was in close proximity of the strain transducer the top flange, which typically

indicated compression, would respond with a tensile “spike” in readings. Figure 2.6.4

provides a plot of longitudinal strain data recorded. When the test vehicle was

significantly far away from the top sensor the transducer readings would return to their

expected values. It is theorized that the flange, when under direct loading, experiences

localized effects thus affecting the recorded data.

Figure 2.6.4 – Top flange tensile “spike” observed in stringers during live load testing Adapted from

Bridge Diagnostics Inc. (2002).

Page 64: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

50

2.7 – The Bridge Street Bridge (Grace et al. 2002; Grace et al. 2005)

Structure B of the Bridge Street Bridge in Southfield, Michigan is unique for a

few reasons. While the double-tee beam system used in the structure is commonly used

in prestressed structure design, the reinforcement used in each girder is quite novel.

Figure 2.7.1 provides a section profile of the bridge. The prestressed concrete stringers

contain CFRP tendons in lieu of conventional steel prestressing tendons. Additionally,

external post-tensioned carbon fiber cables were draped along the lengths of each beam

to provide supplementary longitudinal strength while similar carbon fiber cables were

post-tensioned transversely at each beam diaphragm. The concrete deck is reinforced

with CFRP grids, which is topped with a conventional concrete wearing surface. The

only conventional reinforcement present in each beam is mild steel shear stirrups located

throughout the web of each double-tee. Constructed in tandem with a conventionally

built twin, the bridge is a three-span structure carrying two lanes of traffic. Six of the

beams on Structure B were instrumented for long term monitoring and subjected to an in-

situ load tested to study their behavior relative to AASHTO design procedures (Figures

2.7.2 and 2.7.3).

Page 65: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

51

Figure 2.7.1 – Bridge Street Bridge cross section (Grace et al. 2005)

2.7.1 - Materials Used

Each of the three spans in Structure B consists of four adjacent double-tee beams

each reinforced longitudinally LeadlineTM prestressing tendons

(www.mkagaku.co.jp/english/corporate/008.html) and four external, post-tensioned

carbon-fiber composite cables (CFCCTM, www.tokyorope.co.jp/english/). All four

girders in a span were also post-tensioned transversely with CFCC tendons. A lateral

diaphragm cast into each girder provides anchorage for each tendon. Seven such

diaphragms were located in each girder. Horizontal deck reinforcement is composed of

multiple bi-directional NEFMACTM (www.autoconcomposites.com/index.html) grids of

Page 66: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

52

0.394” diameter carbon fiber reinforcing bars. Specified 28-day concrete strengths were

7,500 psi for the girders and 5,500 psi for concrete deck topping.

2.7.2 - Instrumentation

As monitoring of this structure was conducted from fabrication to construction

and beyond, a majority of all instruments were installed at the precast facility. All twelve

double-tee beams were instrumented initially to monitor stress levels during fabrication

and prestressing. However, only six beams were instrumented with long-term monitoring

equipment for the in-situ observation. Beams to be monitored in the field contained both

internal and external vibrating-wire strain gages installed at the mid and quarter-span

points of each beam, as well as displacement sensors. At each strain monitoring section,

(quarters and mid-span) gages were installed up both webs of the double-tees. Gages

were located near the bottom of each web, at mid-height, near the top in the decking, and

one in the concrete topping. Figure 2.7.2 illustrates locations of strain gages.

Each beam thus contains a total of 30 gages with only the nine concrete topping sensors

installed in the field.

Figure 2.7.2 – Strain gage location in instrumented spans of Structure B.

Page 67: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

53

Positioning of the six long-term instrumented beams was such that the width of

one entire span was instrumented (referred to as the north span) and a single

representative beam was instrumented in the other two spans (Figure 2.7.3). Although

not relevant to the scope of this discussion, it is interesting to note that a load cell was

installed for each longitudinal external post-tensioned cable for the instrumented beams

to monitor their load levels throughout the life of the structure.

Figure 2.7.3 – Long-term instrumented spans of Structure B.

2.7.3 - Load Test and Results

To conduct the load test, two three-axle dump trucks were used in four separate

load cases. The load cases can be classified as interior or exterior beam tests and are

illustrated in Figure 2.7.4. For each test, multiple readings were required, allowing the

vibrating-wire strain gages to settle. Vehicles were located in their desired position and

remained in place for a period no less than five minutes to obtain adequate strain

readings. For every movement during testing settling periods were implemented to allow

the gages to reach a steady state. During the interior beam tests, trucks were positioned

Page 68: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

54

for maximum positive bending moment adjacent to the sidewalk on the span. The

sidewalk limits the distance in which a vehicle may approach the exterior beams. One

test was conducted in the fully instrumented north span of Structure B while another was

carried out in the complimentary south span. For the exterior load test the trucks were

positioned to produce maximum positive bending moment near the exterior parapet of the

span. Similar to the interior beam tests, the exterior load tests were conducted in the fully

instrumented north span and also the middle span for comparison.

Figure 2.7.4 – Live-load test cases for Bridge Street Bridge

To produce distribution factors of the structure, the authors combined the data

from the interior and exterior load tests, superimposing strain readings on each beam for

comparison. Distribution factors were calculated based on total strain in a specific beam

relative to total strain of all beams using equation 2.5.1. Overall, it was found that the

calculated distribution factors agreed very well with distribution factors obtained using

the 1996 AASHTO Standard Specifications and the 1998 AASHTO LRFD methods.

Page 69: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

55

Additionally it was concluded that usage of the AASHTO specifications was appropriate

for design of prestressed concrete beams externally reinforced with carbon-fiber

reinforcement.

2.8 – Synthesis of Literature

As noted previously, a great deal of information exists pertaining to the topic of

bridge monitoring. However, information directly related to the static, live load testing of

structures is not easily obtained. A vast majority of bridges in the United States are

inspected from a visual perspective only as the initial cost of instrumentation often

prohibits the scientific evaluation of them. Structures selected for monitoring are limited

among the bridge inventory but have been proven to provide valuable insight into their

performance. Monitoring efforts such at Ohio HAM, South Carolina S655, the

Fairground Road Bridge and the Bridge Street Bridge offered insight into procedures

used for successful monitoring of structures. Methods of interpreting data relating to the

distribution of vehicle loads among bridge stringers and evaluation of the composite

nature of each different structure are presented in the research carried out, providing a

rational basis for implementation on the IBRC structure of this study.

The successes of these projects provide a proving ground for use of commercially

available instruments. Complying with the recommendations noted in Section 2.1, The

Ohio HAM Bridge, South Carolina S655 and the Fairground Road Bridge illustrate the

preference of electrical-resistance strain gages for short-term load testing, as well as the

use of high-speed data acquisition systems for data collection. Additionally, the testing

Page 70: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

56

of the Bridge Street Bridge illustrates adequacy of using vibrating-wire gages but also

their lengthy acquisition process.

The Fairground Road Bridge illustrates the successful use of removable strain

sensors composed of electrical resistance gages. Extensive amounts of labor were

focuses on attachment of electrical resistance gages to the Ohio HAM Bridge, South

Carolina S655 and the Bridge Street Bridge during fabrication, limiting the requirement

of in-situ installation. Experience of these projects (including the previous, inconclusive

strain gage instrumentation of the De Neveu Creek Bridge) indicates that field bonded

gages is extremely difficult and can limit the usefulness of data obtained. Thus,

removable sensors are preferred for this project to ensure their repeated use over time.

Fabrication of strain sensors in a controlled environment increases consistency among the

sensors and also limits possible damage from peripheral sources, e.g. the environment,

wildlife and possibly vandals.

Cost of instrumentation is also of concern. The suite of equipment utilized in the

four monitoring projects noted incorporated a substantial financial investment. Use of

compact electrical-resistance strain gages bonded directly to the superstructure produces

valuable information at a low cost when the substrate is composed of homogenous

materials such as steel stringers. However, experience has proven that larger, more costly

instruments are required for satisfactory strain data collection on heterogeneous materials

such as concrete. An application of this is noted in the Bridge Street Bridge project, as

larger, internally embedded concrete sensors were utilized. The cost of larger gages or

Page 71: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

57

removable sensors frequently exceeds $500 per instrument, commanding a significant

per-gage investment (MnDOT 2005; Schultz 2005). The instrument array specified for

this project, which will be outlined later in the document, includes 32 locations of strain

gages. Considering the per-instrument cost of commercially available sensors and the

financial capital available for this project development of an alternative, cost effective

instrument is desired.

The instrumentation projects noted also illustrate a wide variety of structural

materials used and differing geometric conditions. The Ohio HAM Bridge utilizes

conventional steel-stringer with mild-steel reinforced concrete deck in an integral multi-

span configuration. The South Carolina S655 project uses a similar steel-stringer with

FRP decking as the Fairground Road Bridge, however S655 contains a single span while

the Fairground Road project is an integral multi-span application. Additionally, the

Bridge Street Bridge is composed of prestressed concrete in a multi-span layout, however

each span acts independently. Based on these differences and the success of their tests,

successful short-term strain monitoring is anticipated for the single span, prestressed De

Neveu Creek IBRC Bridge.

Finally, the previous work conducted on the De Neveu Creek IBRC Bridge

provides a baseline for analysis of data. The load deflection data presented illustrates a

global performance of the structure, indicating an overall performance conforming, albeit

conservatively, to AASHTO design recommendations. However, collection of further

data is requires as a number of performance aspects of the novel structure are not fully

Page 72: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

58

understood. Collection of strain data pertaining to the strain profile of the girders and

concrete deck will allow for assessment of composite action between the superstructure

elements. Documentation of any variation in the strain profile of the structure is

important and provides insight into its performance over time. Observation of the

transverse behavior of the FRP-reinforced concrete decking is also required.

Assumptions made in the design of the concrete deck require verification if the system is

to be implemented elsewhere.

Page 73: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

59

Chapter 3 – Data Acquisition and Strain Measurement

3.0 - Introduction

Usage of electrical instruments to obtain useful mechanical knowledge can take

on many forms. The method in which physical data is most commonly collected for civil

engineering projects is through direct current instruments. Their simplicity and relative

low cost enable researchers to obtain many points of data with an appropriate level of

accuracy. These systems require the usage of a collection instrument, typically strain

gage circuits or other transducer, a data acquisition system and a personal computer to

process the data. A discussion of collection instruments is included later in this chapter.

The data acquisition system and computers used to interpret and record the data collected

are also described in this section.

3.1 – Signal Processing

Within every data acquisition system is a series of modules that translate an

electrical signal into a form that humans can recognize. Early acquisition systems

utilized paper chart recorders or magnetic tape recorders, while today’s systems have

grown into complex networks of analog-to-digital converters (ADCs), high-speed

multiplexers and personal computers (IOTech 2004). Figure 3.1.1 illustrates the layout

of typical data acquisition system. Contained within this section is an introduction to

ADCs, multiplexing of instrument channels, sampling theory, signal amplification and

also noise reduction.

Page 74: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

60

Sensors

Signal Conditioning

(Amplification,

Filtering)

A/D

ConversionComputerMultiplexing

Other Sensor Channels

& Signal Conditioning

Figure 3.1.1 – Layout of a typical data acquisition system

3.1.1 – Analog to Digital Conversion

ADC is a process where an electrical signal is converted into a discrete binary

number for purposes of interpreting it within a digital setting. This conversion process

approximates an instance of the signal and assigns it a binary number, which in turn, is

converted into a digital number. The level of complexity in which the approximation

takes place is dependent upon the size of the binary number assigned to analog signal,

which is quantified by the number of bits used in the conversion. ADCs are capable of

producing a resolution of one part in 2n, where n is the number of bits used (Rizzoni

2003). For example, the IOTech DaqBoard 2000 series used for this project contains a

16-bit ADC. Thus it is capable of a resolution of one in 216, or 1/65,536. If 5 Vdc is input

to the system, the maximum resolution available is 5/65,536, or 0.000076 volts (0.076

mV).

Page 75: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

61

Many types of ADCs exist, with the most basic being the successive-

approximation ADC. A detailed explanation of this type of converter illustrates a very

basic procedure for converting an analog signal to a digital value. Successive-

approximation ADCs compare an analog voltage to a voltage produced by an internal

digital-to-analog converter (DAC). The DAC produces an analog signal that is compared

to the measured signal, which is prescribed by a series of switches. Each switch

represents one bit of resolution. The first switch activated typically produces a voltage

equal to half the full scale of the circuit. This voltage is compared to the analog value

and if the analog voltage is greater than the output voltage, the DAC will increase the

voltage. If the analog voltage is smaller, the DAC will decrease it. This process is

repeated for the second switch and moves the DAC voltage closer to the analog voltage.

Successive steps are performed to close in on the best approximate to analog voltage until

all the switches have been used. This state is the maximum resolution of the DAC.

While the process may seem lengthy, conversion rates in excess of 1MHz are possible

(IOTech 2004).

Error in ADCs can cause difficulties in producing an accurate approximation of

the analog signal. Common sources of error come from the linearity of conversion, lack

of proper digital coding and electrical noise. Manufacturers of data acquisition systems

take great care to properly calibrate their ADCs and provide recommendations for their

accurate usage. Many types of ADCs exist with significantly more complex methods of

approximating an analog signal. The equipment used in this program utilized a

Page 76: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

62

successive-approximation ADC (IOTech 2005a) and peripheral examples of ADCs are

not included within this document.

As multiple sensors are used in large projects, obtaining synchronized data can be

difficult. If an acquisition system samples analog signals at a slow rate, the time

difference between readings of the first and last sensors can be appreciable. For example,

a truck traveling at 45 miles per hour takes only 1.52 seconds to cross a 100’ bridge. If

the sampling rate of the system is 1Hz, only one sensor will be read during the crossing.

The other sensors will not be read during the truck’s crossing. For this reason, high-

speed ADCs are required for rapid acquisition. However, rapid-acquisition ADCs can be

relatively high in cost. To avoid usage of multiple ADCs, a device called a multiplexer is

used. This device rapidly switches between sensor channels and transmits their signals to

a single high-speed ADC for digitizing. While this procedure mimics the usage of

multiple ADCs, it provides a great deal of economy to acquisitions at a suitable sampling

rate (IOTech 2004).

However, it must be recognized that the switching of channels produces data that

are not recorded at the same instance. If time-specific measurements are required Sample

and Hold (S/H) circuits can be used to temporarily store the analog reading from the

sensor. Each S/H is triggered to sample its sensor nearly simultaneously with the other

channels and hold that value until the multiplexer allows the signal through. This allows

for readings to be incredibly close to each other, frequently less than 100 ns apart

(IOTech 2004).

Page 77: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

63

3.1.2 – Sampling Rates

As rapid sampling rates are available through the use of high speed ADCs and

swift multiplexers, a brief discussion of sampling rate is warranted. The goal of ADC is

to best preserve the original signal form, thus one may instinctively assume that the

quicker the sampling the better. Through extensive research it has been found that this is

not necessarily the best answer (Rizzoni 2003). Rather, a simple solution exists – the

Nyquist Sampling Criterion. This criterion states that if all frequencies contained in a

signal are less than a specified value, all information pertinent to that signal may be

obtained by sampling at a minimum of twice that value (IOTech 2004). Application of

the Nyquist theorem provides the most accurate generation of data specific to a measured

signal.

Another phenomenon that must be addressed in signal processing is aliasing.

Aliasing has two facets, both of which facilitate incorrect measurements. When sampling

at too slow a rate, signals much lower in the frequency domain than the true signal may

be produced. Figure 3.1.2 illustrates this with a 5Hz sine wave sampled at a 6Hz rate,

which is less then the recommended rate of 10Hz. The low sampling rate incorrectly

produces a 1 Hz wave as shown below.

1Hz wave observed6Hz sampling points

Time (s)

1.0 sec

Actual 5Hz signal

Figure 3.1.2 – Application of the Nyquist Theorem.

Page 78: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

64

Additionally, it is common to encounter aliases produced by frequencies equal to

half of the desired frequency or greater. Acknowledging the Nyquist Theorem, the

sampling rate should be increased to more than twice the desired frequency, requiring

that any unanticipated frequencies below the desired signal be prevented by other means.

This stage of signal conditioning must be addressed before any digitizing can take place.

Usage of anti-aliasing filters is the most common method of reducing aliasing from a

signal. These filters allow only frequencies smaller than a prescribed level to pass

through the circuit and are termed low-pass filters. Recognizing this issue, a low-pass

filter was utilized on each individual channel for strain gage measurements to remove any

aliasing present. These filters were set very low (less than 4 Hz) as the measured signals

were expected to be constant, with no oscillation.

This issue of aliasing introduces the process of preparing an electrical signal for

digitizing called Signal Conditioning. This conditioning process works to eliminate any

environmental or other sources that may cause misinformation within the desired signal

and consequently invalidate the information processed by the ADC. Signal

amplification, digital and analog filtering, and signal attenuation are topics directly

applicable to strain gage measurements and are discussed herein.

3.1.3 – Signal Amplification

As a large number of strain measurements utilize an electrical circuit called the

Wheatstone bridge, a brief introduction is warranted. The Wheatstone bridge is a basic

electrical circuit whose resistance changes when the base material is strained. This

Page 79: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

65

resistive change can be directly related to the level of strain experienced by the measured

member. However, most applications of the Wheatstone bridge produce very small

electrical changes while in operation, making amplification of the signal relatively

important. Without satisfactory amplification, observation of signal changes becomes

incredibly difficult. Commonly, each sensor channel is provided with its own amplifier

sequence as to preserve customizability among instruments. This enables the user to

obtain information about different types of instruments simultaneously. Many methods

of amplification exist and due to the extreme breadth of the subject, details of electrical

amplification are not contained within this document. However, an overview of signal

amplification and subsequent conditioning used in this project is discussed below.

The IOTech DBK43A module used in strain measurements increases the magnitude

of the measured signal by means of an input amplifier and a scaling amplifier.

Additionally, a signal offset adjustment module is present within the series of amplifiers

to nullify any initial imbalance the circuit may have. The layout of the individual

channels within the DBK43A module is illustrated in Figure 3.1.3. The step-by-step

process is described as follows,

1. The unchanged signal enters the input-gain amplifier from the strain gage sensor

and receives its first increase in magnitude by a multiplier.

2. The signal enters the offset adjust module where any initial imbalance may be

removed.

3. The signal then passes through filtration (to be discussed later), which has a

default amplifier that doubles the signal magnitude.

Page 80: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

66

4. Finally, the scaling amplifier then amplifies the signal by another independent

prescribed multiplier

Once amplification is complete, the signal passes into the multiplexer and eventually into

the ADC.

431

Wheatstone

bridge circuitry

2

Signal

conditioning

circuitry

Figure 3.1.3 – Block diagram of an IOTech DBK43A (IOTech 2005b). Further explanation

regarding signal-conditioning circuitry is found in Figures 3.1.4-6 and its corresponding literature.

To establish known values for each stage of amplification, screw-driven

potentiometers termed “trimpots” are used. Manipulating these trimpots on the DBK43A

may change each stage setting within their available range. Four trimpots are provided

for each channel and are described below,

• EXEC – sets the excitation voltage of the circuit

• OFFSET – adjusts the offset voltage of the circuit

Page 81: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

67

• GAIN – sets the input amplifier gain (100x to 1250x)

• SCALE - sets the scaling amplifier gain (1x to 10x)

The following description of configuring circuit amplification assumes that voltages are

observed at a computer through the use of software compatible with the data acquisition

system. Computer software is utilized to select a signal path illustrated in Figures 3.1.4

through 3.1.6. The equations governing the amplifier output are identified with each

illustration and may be used to establish the desired amplification gains. Once these

values are set, the signal will be ready for ADC.

Initially, if the circuit contains quiescent voltages or significant bridge imbalances

they may be set to zero by adjusting or trimming the “Offset” trimpot. While this offset

has a limited range (-1.25mV to 5.0 mV), it is satisfactory for most applications (IOTech

2005b). The signal path of the offset adjust can be seen in Figure 3.1.4. Note that

multiple inputs are included prior to multiplexer (MUX A). The Wheatstone bridge, a

reference voltage of 5 mV, and an electrical ground may be selected using the software.

Vout

Ground

Figure 3.1.4 – Signal path of the “OFFSET” trimpot (IOTech 2005b).

Page 82: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

68

To set the input amplifier gain (GainINPUT) at a known value, a reference voltage

of 5mV (Vreference) is applied to the signal conditioning circuit while the voltage drop

across each stage is read. To obtain the desired input gain, the “Gain” trimpot is adjusted

until the specified output voltage, Vout, is reached. This voltage is calculated using the

expression below in Figure 3.1.5. Note that in Figure 3.1.5 the signal path does not enter

the scaling amplifier at any time. Furthermore, the signal path does cross through the

filter section, which acts as a small signal amplifier. This stage has a default, but

constant, amplification gain (GainFILTER).

Vout5.0 mV

( )( )( )( )( )( )0.005 2.0

out reference FILTER offset

offset

V V Gain V

V

= −

= −

INPUT

INPUT

Gain

Gain

2.0FILTERGain = 0offsetV ≈,

referenceV =

Figure 3.1.5 – Signal path during adjustment of the input amplifier gain (IOTech 2005b).

Similar to the input gain, the scaling gain (GainSCALING) is set by activating the

signal conditioning circuit and applying a 5mV reference potential, producing an

amplified output. Adjustment of the “Scale” trimpot until the desired output voltage is

reached sets the scaling gain at the specified level. The expression governing this setting

is below. Figure 3.1.6 illustrates the signal path during this process.

Page 83: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

69

Vout5.0 mV

( )( ) ( )( )( )( )( )( )0.005 2.0

out reference INPUT FILTER offset

INPUT offset

V V Gain Gain V

Gain V

= −

= −

SCALE

SCALE

Gain

Gain

referenceV =

Figure 3.1.6 – Signal path during adjustment of the scaling amplifier gain (IOTech 2005b).

As shown in Figure 3.1.3, each of the DBK43As eight individual channels used in

this project contains the sensor’s circuitry, an individual amplifier series, and two

filtering options. There is no interaction between adjacent sensors, nor their amplifiers or

filters.

3.1.4 – Signal Filtering

Filtering a signal can be addressed in both the digital and analog arenas. While

many types of filtration exist, the most common forms are the Butterworth, Chebyshev,

and Bessel filters. Each type may be used for high, low or band-pass applications, but the

discussion herein is limited to low-pass filtering. Low pass filters allow only frequencies

within a signal below a specified cutoff frequency, ωc, to be amplified, or pass through,

while all other frequencies are attenuated (Rizzoni 2003). Figure 3.1.7 illustrates a

typical response of a low-pass filter. It can be seen that preferred frequencies are less

than the cutoff frequency are amplified by a gain of A, acknowledging a tolerance of A±ε.

Page 84: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

70

All frequencies greater than the cutoff frequency are then attenuated to lower gain levels.

This effectively gives preference to those frequencies lesser than the cutoff frequency by

allowing them to be amplified by significantly large gains. The Stopband signifies

frequencies that are amplified by a specified minimum gain of Amin, which is typically

very small causing them to be overshadowed by the Passband frequencies. As Amin and

the lower Passband frequency A-ε are not equal, an attenuation gap is caused where

frequencies are amplified by a gain greater than the minimum, but are not affected by the

maximum amplifier gain A.

Frequency, ω

Stopband

Passband

Filter Gain

ωc

Cutoff Frequency, ωc

A

0

Attenuation of signal

amplificationA-ε

Amin

A+ε

Figure 3.1.7 – Typical low-pass frequency response.

It can be seen that attenuation of undesired frequencies is not instantaneous at the

cutoff frequency, but rather has a varying response. Figures 3.1.8 and 3.1.9 illustrate two

common types of low-pass filters. Figure 3.1.8 shows the characteristics of a Butterworth

filter, which exhibits a flat response across the passband and attenuates unwanted

frequencies quite steeply.

Page 85: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

71

Figure 3.1.8 – Butterworth low-pass filter response (Rizzoni 2003).

Figure 3.1.9 illustrates a Chebyshev filter, which has an even steeper attenuation rate but

not as flat of a response across the passband. Note the movement of the amplitude

response for the second through fourth order Chebyshev filters. A third type of filter is

the Bessel filters, which has a much shallower rate of attenuation (IOTech 2004; Rizzoni

2003).

Figure 3.1.9 – Chebyshev low-pass filter response (Rizzoni 2003).

As noted previously, signal filtering is encountered within the DBK43A module

and is of analog type. Included with the module was a 3.7Hz, third-order Butterworth

low-pass filter, which attenuates signals with frequencies greater than 3.7Hz. As can be

seen in Figure 3.1.3, the filter is located on each channel before the multiplexer, but after

Page 86: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

72

the initial input amplifier. This location is beneficial as it allows the filter to have an

ideal signal-to-noise ratio (IOTech 2004). Usage of this filter eliminates ambient noise

caused by environmental sources (e.g. fluorescent lighting or radio signals) and also

reduces any aliasing caused by these sources. Through testing, it became evident that

usage of these filters is necessary, as elimination of interference sources is extremely

difficult. Figure 3.1.10 illustrates the need for filtration with an unstrained, quarter-

bridge strain gage circuit. Figure 3.1.10(a) shows a “dirty” signal with interference from

radio waves, lights, power systems and aliasing. Figure 3.1.10(b) shows the drastic effect

that the analog filter has on the signal, eliminating nearly all the high-frequency noise

(radio, lighting, power systems). Finally, digital filtering removes the remaining noise,

presumably aliasing from higher frequency signals.

(a) Unfiltered Signal (c) Signal with Analog and

Digital Filtering

(b) Signal with Analog

Filtering Only

Figure 3.1.10 – Comparison of filtered (b & c) and unfiltered data (a) (DASYLab 2004).

Additional filtering can be provided in the digital realm. Very small amounts of

electrical interference were observed after the analog filter was enabled. Thus, use of

digital filtering on all strain channels was utilized to obtain the relatively noise-free signal

Page 87: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

73

illustrated in Figure 3.1.10(c). The digital filter used in this project was a 2 Hz, second-

order Butterworth filter

The DBK43A module offers an additional source of signal conditioning similar to

filtering termed AC Coupling. Employment of this option requires that an internal

electrical jumper located on the modules circuit board be removed, allowing the signal to

pass through a capacitor, which removes the DC component of a combined AC and DC

signal. The signal yielded by enabling this option is a cleaner AC signal, without a shift

due to the DC signal. Figure 3.1.11 graphically illustrates this procedure.

Figure 3.1.11 – (a) standard wave composed of AC and DC signals, (b) AC Coupled wave

(National_Instruments 2005).

Typically, this option is used only for signals in the AC arena or for cyclic DC

measurements with high-frequency signals and was not observed to be effective for this

project (National_Instruments 2005). Additionally, all instruments to be used in this

project are part of DC circuits that output extremely low frequencies, eliminating any true

Page 88: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

74

usefulness of AC coupling. In fact, the only AC signals observed in this project are

caused by environmental noise (e.g. fluorescent lights).

Aside from filtering, other methods of obtaining more accurate signals can be

incorporated into measurements. Measurement sensors are rarely located a small

distance from the data acquisition system warranting the use of lead wires. These lead

wires may only be a few feet long for a laboratory setting. However, large projects such

as bridge monitoring require the usage of several hundred feet of wire. Unfortunately,

long wires allow for significant noise production within a signal and must be addressed.

A common method used to reduce this effect is the installation of shielded or twisted-pair

wire. Shielded wires are preferred for measurement as they eliminate most forms of

interference. If used, however, the shielding for each wire must be connected to a stable

ground near the acquisition system. Electrical grounds are may be found on the chassis

of the ADC, the cable supplying power to the modules, or by connecting directly to the

earth. Grounding should never be at the instrument end nor should the system ever have

two grounds, as these may produce conductive loops and induce an peripheral voltage in

the signal (Rizzoni 2003). Although it exhibits excellent noise attenuation when

grounded properly, shielded wire is expensive and also vulnerable to electrical induction

when located near large current sources (Rizzoni 2003). On the other hand, twisted-pair

wire is excellent in combating noise in large current environments but lacks shielding and

is vulnerable to radiated noise. The twisting path of each wire causes the electromagnetic

field produced in each wire to cross at regular intervals, effectively canceling each other

out (Omega 2000). Additionally, for twisted pair wire to work properly, each wire must

Page 89: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

75

carry a current to achieve canceling. For this project it was not feasible to use either

method prescribed resulting in the need to employ filtering. Lead wires used for this

project were composed of a three-wire Wheatstone bridge circuit (detail may be found in

section 4.1), independently twisted with an inactive wire. It is recommended that for

strain measurements of bridge structures shielded wire be employed if the measuring

environment contains a large amount of radiated noise. It is unexpected to encounter

large currents during bridge monitoring as they are primarily found in industrial

environments and not outdoors.

3.2 – Measurement with Electrical Resistance Strain Gages

Electrical resistance strain gages are devices used to measure strains in a wide

variety of applications. When strained, each gage responds with a change in electrical

resistance that is proportional to the magnitude of strain the gage is subjected to.

Monitoring of this varying resistance value yields information that can be directly

correlated to strains found in the material being studied. Typical electrical resistance

strain gages are composed of a backing material and an electrically resistive network.

When the backing material is deformed, the resistance of the network either decreases or

increases. Decreases in resistance are caused by a compressive strain while a tensile

strain produces an increase in resistance. Typically, these gages are employed as part of

a resistive network designed to produce an electrical output when excited by a constant

voltage source. Figure 3.2.1 illustrates a plan view of a typical strain gage.

Page 90: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

76

Figure 3.2.1 – Typical strain gage (www.vishay.com)

The most common DC circuit used in strain gage measurements is the Wheatstone

bridge. Invented by Samuel Hunter Christie in 1833 and later popularized by Sir Charles

Wheatstone, the bridge contains two arms that when strained produce a relative change in

voltage (Wikipedia.org). This voltage difference is measured between the two mid points

of each arm, denoted in Figure 3.2.2 as nodes a and b. This arrangement of strain gages

and/or resistors has proven to provide excellent sensitivity when measuring the minute

resistive changes that result from strain of the gage.

Figure 3.2.2 – The Wheatstone bridge.

The three most common forms of the Wheatstone bridge are the quarter, half and

full bridge configurations. A quarter bridge configuration has a single strain gage (SG1)

and three additional completion, or dummy, resistors (Ri). During testing the strain

Page 91: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

77

gage’s resistance varies while the completion resistors remain static. A half bridge

network utilizes two strain gages and two completion resistors. In this arrangement, one

entire arm, or half the circuit is variable, thus the moniker, half-bridge. A full bridge

configuration requires the use of four strain gages and zero dummy resistors. Examples

of these configurations can be found in Figure 3.2.3.

Figure 3.2.3 – Typical configurations of the Wheatstone bridge.

Of these three configurations the full bridge is the most sensitive, whereas the

quarter bridge is the least. The high degree of sensitivity of full bridge configurations

and its fewer sources of error make it ideal for strain measuring. However, installation of

four gages can be labor and cost intensive. Half or quarter bridge circuits are typically

used when installation of four gages is not feasible. While quarter bridge circuits offer

the least sensitivity, they can be constructed rapidly. Additionally, quarter bridge circuits

offer reliable measurements so long that the issues of signal noise and error sources are

properly addressed. Signal noise is most commonly eliminated by use of filtering and

shielding while error sources may be compensated by use of mathematical methods.

Both are addressed elsewhere in this chapter.

Page 92: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

78

Recognizing this, the quarter bridge configuration was chosen for this project.

The following is an overview of quarter bridge behavior and the calibration process used

for each gage.

To monitor the electrical response of each circuit, an initial state must be

identified prior to testing. The ideal state is one in which the resistances on each side of

the bridge are equal. If an excitation voltage, Vexec, is applied to the bridge, each resistor

will have an equal resistance drop and the ratios of the upper and lower resistors on both

left and right sides will be unity. Thus, the resulting output voltage across nodes a and b

is zero. This is illustrated by using the resistances found in Figure 3.2.4 and the

following expressions for output voltage,

c d b aab exec exec

a c b d b d a c

R R R RV V V

R R R R R R R R

= − = − + + + +

(3.2.1)

Clarifying, if the ratios Ra/Rc and Rb/Rd both equal 1.0, the bridge is said to be balanced

and yields an output of zero. Note that the ratios of each side of the bridge control the

output of the circuit, not just the individual resistances.

Figure 3.2.4 – Diagram of variables used in calculations.

Page 93: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

79

Data acquisition systems monitor the voltage difference between nodes a and b,

which are directly effected by resistances of each completion resistor or strain gage in the

circuit. As it is not possible to obtain resistors that are exactly equal in resistance and

thus initial states of imbalance will exist in all Wheatstone bridges. This imbalance can

be corrected by either mathematically subtracting the initial voltage differences (Vab) or

by adjusting the offset adjustment trimpot outlined in section 3.1. Constant manipulation

of the offset adjustment trimpot is unreasonable during measurement and thus

mathematical means are typically used. It has been found that low levels of bridge

imbalance do not affect measurements when only a small portion of the strain gages

measurement range is used. If a majority of the strain gages range is required (typically

5% of the gage length), elimination of bridge offset should be addressed. Recall that for

a quarter bridge circuit only one of the four resistors is variable.

If resistor Rb is replaced by an equivalent resistance strain gage and strained by a

magnitude ∆R, the resulting output will be,

c dab exec

a c b d

R RV V

R R R R R

= − + + ∆ +

(3.2.2)

Note that if the gage experiences tension, ∆R will increase, which increases Vab. The

opposite is true for compression.

Often, it is required to calibrate or verify the accuracy of a strain gage within a

resistor network. Both direct and indirect methods of calibration are available to the user.

Direct methods require a known mechanical input to the system to which the output is

Page 94: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

80

compared. For instance, load transducers may be directly calibrated with a known mass.

However, direct calibration of a strain gage circuit by a known strain requires precise

equipment not ordinarily available (Micro-Measurements 2004b). Indirect calibration

can be achieved by electrically simulating the presence of a mechanical input. Most

frequently this is performed by shunt calibration. Shunting is a procedure where a

resistor of large impedance is placed in parallel with the initial resistor and via Ohm’s

Law, a small change in equivalent resistance results. It is important to note that during

shunting, the gage is not strained in any way but when coupled with the large impedance

the two are “seen” electrically as a slight drop in resistance. This causes the Wheatstone

bridge circuit to “feel” a simulated compressive strain in the gage. Recall that a drop in

resistance signifies a compressive strain. Tensile strains may also be induced in the

bridge by shunting a dummy resistor in parallel with the completion resistor after the

gage. Figure 3.2.5 displays two shunting locations and their simulated effect.

+Vexec

-

+Vexec

-

RGage

RGage

RdRc

Ra

RdRc

Ra

a b a b

Figure 3.2.5 – Simulated strain via shunt calibration.

The strain simulated may be obtained by usage of a term describing the sensitivity

of each gage. This term is referred to as the Gage Factor and is typically provided by the

Page 95: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

81

manufacturer of strain gages. The gage factor is defined as the instruments propensity to

change resistance with straining. Symbolically, it is defined as,

g gR R R R

GFL L ε

∆ ∆= =∆

(3.2.3)

where ∆R is the change in resistance of the gage relative to its original, unstrained

resistance, Rg. L is the length of the gages resistive array, while ∆L is the change in

length that occurs simultaneously with ∆R. Recall that strain, ε, is the change in length

relative to its original length, or ∆L/L. This expression may also be re-written in terms of

strain,

gR R

GFε

∆= (3.2.4)

Note that for an increase in resistance the expression yields an increase in strain, or a

positive value indicating tension. The opposite is true for compression.

As noted previously, it is often common to shunt, or place a large impedance in

parallel to the strain gage itself, causing a simulated compression strain in the bridge. It

is prudent to calibrate the gage with a simulated strain roughly equal to the largest

anticipated value. This maximizes the output of the circuit by inducing the full resistive

range of the gage. Manipulation of the gage factor equation produces the following

expression for direct calculation of the required shunt resistance needed.

( )

g

shunt g

simulated

RR R

GF ε= − (3.2.5)

By shunting the strain gage with the shunt resistor, Rshunt, the equivalent resistance of that

specific arm will decrease and simulate a desired compressive strain. The equation below

Page 96: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

82

illustrates the effect that a shunt resistor has on the output voltage, Vab, of the bridge. All

variables are indicated in Figure 3.2.5 where RGage is a strain gage and Ra, Rc and Rd are

dummy resistors of equal resistance.

( )

c dab exec

a c Eq d

R RV V

R R R R

= − + +

(3.2.6)

where the equivalent resistance of the shunted gage is,

1

1 1Eq Gage

Gage Shunt

R RR R

= + <

(3.2.7)

It can be seen that if REq becomes smaller while all other variables remain

constant, Vab will be negative, signifying compression. Electrically, shunting RGage with

RShunt causes decrease in equivalent resistance which is referred to as REq. Satisfying

Ohm’s Law, this decrease in equivalent resistance causes an increase in current across

both RGage and Rd. However, as the resistance of Rd is now greater than RGage more

voltage will drop across Rd. By subtracting the voltage drop across Rd from the

corresponding voltage drop no the opposite side, a negative result is produced, which is

observed as compression.

A Wheatstone bridge may also simulate a tensile strain with by shunting.

Referring to the rightmost illustration in Figure 3.2.5, the gage is not shunted and does

not experience an equivalent resistance change. However, the dummy resistor is placed

in parallel with the shunt resistor and experiences a drop in equivalent resistance. Thus,

the entire shunted arm experiences a greater, but uniform current flow that increases the

voltage drop across the gage, Rb. As noted before, an increase in voltage across the gage

Page 97: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

83

is viewed as tension. Detailed examples of compressive and tensile shunt calibration

calculations can be found in Appendix B.

While shunt calibration is quite simple, addressing measurement errors and signal

attenuation of Wheatstone bridges involves many issues, the most pertinent of which are

discussed in the following sections.

3.3 – Strain Gage Measurement Errors

While calibration of the gage is crucial to accurate measurement, sources of error

exist and must be addressed. Lead wire attenuation, thermal effects of gages and wires,

and strain gage non-linearity are the most common sources of error in measurements and

each are addressed in the following section. It is important to note, however, that error

sources may be addressed by numerous computer software packages available for data

acquisition. Often these packages do not disclose the methods used in addressing error

sources and, thus, the validity of such requires trusting the programming source. The

following section outlines the basics of error correction and only provides numerical

modifiers that may be applied to recorded data for error correction.

The desensitizing of a gage due to lead wire resistance is easily avoided. Figure

3.3.1 illustrates two commonly used quarter bridge strain gage circuits.

Page 98: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

84

Figure 3.3.1 – Quarter bridge strain gage configurations (Micro-Measurements 2005c).

For the two-wire circuit (Figure 3.3.1(a)), it can be seen that the lengthy lead wires act as

small resistors and cause a voltage drop across them. When the gage is strained these

additional potential drops will act in a parasitic manner, effectively desensitizing the

voltage reading of the gage. The effect is similar with a three-wire circuit (Figure

3.3.1(b)), but only a single length of lead wire desensitizes the reading as each arm of the

active side of the circuit experiences the same lead wire resistances, effectively canceling

them out (Micro-Measurements 2005c). Correction of this scenario may be made be the

following relationship.

g Lactual read

g

R R

Rε ε

+ =

(3.3.1)

Page 99: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

85

where Rg is the gage resistance and RL is the lead wire resistance. Also, RL is equal to

two lengths of lead wire for a two-wire circuit and a single length for a three-wire circuit

(Micro-Measurements 2005a). The systems used in laboratory settings often contain

lengths of lead wire short enough that they contribute negligible additional resistance to

the circuit. For example, a circuit with a 350-ohm strain gage in a three-wire

configuration with 30’ of lead wire has 1 ohm of lead wire resistance. Using equation

3.3.1 this length of wire produces only a 0.28% difference. Consideration of this error

source is warranted for field implementation but the effects are typically small.

Thermal effects of strain gage circuits also warrant consideration. Both lead wire

and gage temperature affect readings and both are discussed herein. Gages are typically

manufactured with a backing material that closely mimics the thermal expansion

characteristics of the base material the gage is to be installed on. For most applications

this approach eliminates most sources of thermal expansion/contraction error in the gage.

However, when supplying a large excitation voltage to a gage, the individual strain gage

may act as a heat source if a proper heat-sink is not provided to distribute head generated

away from the strain gage (Micro-Measurements 2005b). The most common example of

a proper heat sink is the substrate that the strain gage is bonded to. For this reason,

measurements on metals normally do not raise concern while measurements on polymers

and other non-conductive materials do.

The following calculation is representative of the heat-sink conditions found in

this project and are provided as an example. The power dissipated by a strain gage and

Page 100: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

86

the power density in a strain gage grid can be estimated using the following equations

(Micro-Measurements 2005b),

Power dissipated in strain gage grid, 2

4exec

Gg

VP

R=

i

[watts] (3.3.2)

Power density in strain gage grid, GD

g

PP

A= [watts/in2] (3.3.3)

If a 350-ohm strain gage grid is 0.25” long and 0.12” wide and nylon is the base material

onto which the gage is adhered (PD = 0.15 watts/in2), the following excitation voltage is

appropriate for the gage (Micro-Measurements 2005b),

( )2 2 350 0.15 0.25 .012 2.51exec g D gV R P A= × × = × × × = [volts] (3.3.4)

This calculation of the excitation voltage (Vexec) demonstrates that an excitation voltage

greater than 2.51 volts will generate more heat than what can be dissipated efficiently in

the base material. This produces elevated temperatures in and also near the gage and

ultimately leads to errors in strain measurements. Proper selection of strain gage

excitation is of significant concern, especially since gages are likely to be active for an

extended period of time during testing.

As mentioned before, elevated temperatures in lead wires can also cause

significant error. Changes in temperature directly affect the resistance of lead wires,

having a similar effect to that of lead wire attenuation. Error produced by large lead wire

resistances in a two or three-wire circuit can be compensated for similarly to that of lead

wire attenuation. To minimize error sources, use of the three-wire quarter bridge circuit

is recommended.

Page 101: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

87

Strain gage non-linearity is another common source of error in strain

measurement. Although primarily effected only at large strain levels, the quarter bridge

arrangement is the most susceptible to non-linearity and must be addressed accordingly.

The non-linear nature of the bridge stems from the electrical behavior of the circuit when

gages are strained. Initially the bridge may be approximated as balanced. Realistically,

the bridge is not perfectly balanced but very near this state. The following example

illustrates the non-linearity of a quarter bridge strain circuit (Micro-Measurements

2004b). The output voltage of the circuit may be expressed in the non-dimensional form

below,

Gageo b a a

exec b d a c Gage d a cinitial

RV R R R

V R R R R R R R R

= − = − + + + +

(3.3.5)

Recall that Rb = RG is a strain gage in a quarter bridge circuit from Figure 3.2.5. When

the circuit is strained, the active arm(s) undergo a resistive change that disrupts the

balanced condition. The output voltage changes as follows,

o G a

exec G d a cstrained

V R R R

V R R R R R

+ ∆= − + ∆ + +

(3.3.6)

where ∆R is the change in resistance of the gage due to straining. To obtain the change in

voltage due to straining, the following expression is used.

o o G G

exec exec G d G dstrained initial

V V R R RV

V V R R R R R

+ ∆∆ = − = − + ∆ + +

(3.3.7)

Note that resistors Ra and Rc do not affect the output voltage in any way. Acknowledging

the assumed initial condition where Ra = Rb = Rc = Rd, This expression can be further

expressed in the following form,

Page 102: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

88

4 2

G

G

RR

VR

R

∆∆ =

∆+

(3.3.8)

By substituting the Gage Factor equation (Eqn. 3.2.3) into the above expression, a

relationship for change in voltage in terms of gage factor and strain is available.

( )

2( )

4 2 4 4(2 )

GF GF GFV

GF GF

ε ε εε ε

∆ = → −+ +

i i i

i i

(3.3.9)

The non-linear nature of this bridge configuration is caused by the fact that as the

gage is strained and brought out of balance, the current running through that arm of the

bridge will change. However, the change in current is not proportional to the change in

resistance. As the gage’s resistance changes, the current through that arm in the circuit

changes in the opposite direction. Fortunately, for applications where strains are small

(less than 1000µε) nonlinearity does not significantly affect readings and allows for

neglect of the non-linear behavior. For this reason nonlinear error correction was not

used in this project. Typically, if error correction is required for measurements, Table

3.3.1 provides expressions to correct strain readings in a number of Wheatstone bridge

configurations. Figure 3.3.2 illustrates the affect that strain gage nonlinearity has on

measurements.

Page 103: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

89

1% error at

10,000 µεµεµεµε

Figure 3.3.2 – Nonlinearity errors for tensile strain in bridge circuits (Micro-Measurements 2004a).

Table 3.3.1 – Non-linearity correction factors of Wheatstone bridges (Micro-Measurements 2004a).

Page 104: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

90

3.4 – String Potentiometers and Linear Position Sensors

Both string potentiometers and linear position sensors fall into a category referred

to as transducers. A transducer is a device that converts one type of energy in to another

type of energy. The transducers used in this project convert mechanical energy, which is

the physical displacement of the material, into an electrical energy via the voltage drop

experienced. A string potentiometer (also referred to as a string “pot”) is composed of a

drum with a string or wire is wound about it, a coil spring to provide resistance and a

sensor to monitor rotation of the drum. As the string displaces, the drum rotates, causing

a change in the output of the transducer. For example, a UniMeasure PA Series string pot

with a 30” range will provide 33mV of output per voltage excitation per inch of

displacement (UniMeasure 2005). If an excitation of 10 VDC is applied to the

transducer, the output voltage at the sensor’s full displacement can be calculated as

follows.

( )( ) ( )33 / / 0.033 10 30 9.9out DCV mV V in mV V in volts= = = (3.4.1)

Note that the output voltage is very near the excitation value. This allows the transducer

to produce the maximum output voltage for the applied range of measurement.

Linear position sensors operate in a similar fashion as string pots. However, in

lieu of a rotating drum, these devices have a solid shaft located inside of a sheath and a

movable resistive contact typically referred to as a “wiper.” The wiper is rigidly affixed

to the shaft and as the shaft displaces the resistance of the sensor changes. This provides

a varying voltage output relative to the shaft’s displacement. A typical circuit is

illustrated in Figure 3.4.1.

Page 105: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

91

While a linear displacement sensors can be more accurate than string pots, their

size often limits their use (IOTech 2004). Due to their need to house a solid displacement

shaft, lengths of linear position sensors are typically more than twice their displacement

range. For large displacement measurements, string potentiometers are the preferred

device. Measurement ranges greater than 60 inches are readily available from a number

of manufacturers. Examples of both can be found in Figures 3.4.2 and 3.4.3.

+VExcitation

+VSense

-VExcitation

-VSense

“Wiper”

Electrical “jumper”

Transducer

Figure 3.4.1 – Circuit diagram of a typical three-wire transducer. An electrical “jumper” is provided

to short the negative voltage excitation signal to the appropriate voltage sense terminal.

Page 106: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

92

Measuring

string

Figure 3.4.2 – 30-inch String Potentiometer. Figure 3.4.3 – 4-inch Linear Position Sensor.

Manufacturers of string pots provide calibration factors for their instruments;

however, it is fairly simple to calibrate a displacement transducer. In the laboratory

setting, it is relatively easy to monitor and verify a displacement transducer’s operation.

For example, all linear position sensors used in laboratory testing for this program were

installed in their intended location and calibrated prior to any testing. Verification of

accuracy was conducted in the following manner. An initial reading was observed on the

computer through the data acquisition system from the sensor in its original, unmoved

location. This value was nulled by the user to obtain a “zero,” or initial point. The

sensor is then displaced a prescribed amount by a machinist gage block and the output

voltage is recorded. Via software, a linear relationship is developed using these two

points (initial and displaced) to translate any sensor movement during testing. Figure

3.4.4 provides a schematic of this linear calibration procedure. From this relationship,

Page 107: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

93

any voltage change of the sensor will correspond to its relative displacement on the line

shown.

Voutput

δdisplaced

Linear calibration line

Initial Point

(0 volts, 0 displacement)

Displaced Point

for calibration

Figure 3.4.4 – Linear calibration of sensors.

3.5 – DASYLab Data Acquisition Software

To interpret the output of the ADC used for this project, a computer program

called DASYLab was utilized. The program’s flexibility in interpreting information sent

from the DaqBook to the computer made it an outstanding choice for real-time data

acquisition. DASYLab is an icon-based program that enables the user to visualize the

various operations being performed on the input signals. Figure 3.5.1 (below) illustrates

a completed DASYLab worksheet ready for data acquisition of four quarter-bridge strain

gage channels and two transducer channels.

Page 108: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

94

Figure 3.5.1 – Data acquisition worksheet. All screen shots are taken from DASYLab (DASYLab

2004).

Each icon is referred to as a module, which may perform a task as simple as an

on/off switch or may perform complex operations such as digital filtering of the signal.

Operation of the modules is performed in a left-to-right manner, starting with the signal

input (typically from the ADC) and onto the user-defined modules. The breadth of

options available for data processing in the program is substantial and discussion herein

is limited to items utilized for this project only.

Page 109: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

95

The following explanation of DASYLab corresponds to construction of this

worksheet, which requires the real-time observation of the input channels mentioned, but

also signal conditioning and user-actuated recording of data. This generic experiment

configuration represents a similar worksheet used in the laboratory experimentation

carried out during the course of this project. Figure 3.5.2 provides a logical map of

operations performed during acquisition.

Figure 3.5.2 – Logical map of software configuration. Signals from the ADC pass through filtering,

calibration and unit conversion and are recorded by the software.

To interpret information being sent from the ADC, the user starts an experiment with

a blank “worksheet” in which modules may be installed and connected in a variety of

ways. However, prior to construction of the worksheet, the program requires that each

exterior signal-conditioning module used in the data acquisition be configured within

DASYLab. The following pieces of data acquisition equipment were used:

• Signal Conditioning – IOTech DBK65 – (2) Transducer channels

• Signal Conditioning – IOTech DBK43A – (4) Strain Gage channels

• Analog-to-Digital Conversion – IOTech DaqBook 2000

Communication between the ADC and the personal computer was established via

direct Ethernet connection in accordance with the manufacturer’s recommendations.

Although the ADC is capable of remote operation on a computer network, the ADC and

Page 110: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

96

computer were directly connected for testing. To configure the signal modules, each

DBK device was added in the program’s Hardware Setup section, located in the

Experiment menu (Figure 3.5.3). While only one DBK43A unit and the DBK65 unit

were used for laboratory testing, Figure 3.5.3 indicates that the four units physically

present the Marquette laboratory have been configured. Once a unit is configured in

DASYLab, it is available for installation on the experiment’s worksheet.

3.5.1 - Installation of ADC Modules

To install modules on the DASYLab worksheet, they may be first selected in the

Modules menu or the provided toolbar. The Modules menu contains all the modules of

DASYLab, organized by their function in to the following categories: Input/Output,

Trigger Functions, Mathematics, Statistics, Signal Analysis, Control, Display, Files, Data

Reduction, Network or Special. More frequently used modules are available on the

toolbar located on the left side of the program window for rapid access. Once a module

is selected (signified by the mouse pointer changing to an oil can) it may be placed by the

user anywhere on the worksheet by left-clicking. Modules may not be located on top of

each other, however.

Page 111: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

97

DBK Modules

connected to the

DaqBook

Figure 3.5.3 – Hardware configuration window.

Following the operation map provided in Figure 3.5.2, the ADC inputs are placed

on the DASYLab worksheet. To obtain transducer signals, the DBK65 was selected,

while the DBK43A will convey strain gage signals. Figure 3.5.4 shows the two ADC

modules located on the worksheet. Note that each unit is given a default system name

that corresponds to its channel on the DaqBook. For example, the strain gage module

placed on the worksheet is on DaqBook channel #0, hence the name DBK43A-10:A1.

Similarly, the transducer unit is on channel #3, therefore the name DBK65-13:A1 is

given.

Page 112: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

98

DBK Modules

expanded as required

Default DBK

Modules

Figure 3.5.4 – Hardware configuration window on the main worksheet.

Once the input modules are located on the worksheet, they must be expanded to

the required number of active channels. Recall that the DBK43A requires four strain

gage channels while the DBK65 needs two transducer channels (Figure 3.5.4). By

double-clicking on the left mouse button of the individual module, each may be expanded

as necessary. Channels may be made active my double clicking; the left mouse button

activates a channel while the right deactivates it. Note that the user must specify the

range of each individual channel. Figure 3.5.5 illustrates the bipolar (±5.00 volts)

specification for strain gage channels. The transducer channels were specified with a

bipolar 10 volt range.

Page 113: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

99

Bipolar Range

Active Channels

Figure 3.5.5 – Expansion of analog inputs in the DBK43A module.

3.5.2 - Installation of Digital Filtering

Following the flow chart for the experiment shown in Figure 3.5.2, filtering is

desired for each channel. Acknowledging that multiple channels of data are required, a

useful feature of DASYLab may be utilized: Global Variables. These variables are static

values that the program retains throughout an experiment. Individual variables may be

read by other modules on the worksheet and can also be modified during the experiment

by the user. These variables may be accessed in the Options menu. For this example

Global variable #1 was defined as 2.00 Hz for use in all channel filters and was given an

appropriate name. Figure 3.5.6 illustrates the dialog box for global variable definitions.

Page 114: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

100

Manually Input

Value

Available Global

Variables

Figure 3.5.6 – Defining global variables.

The Filter module was then placed on the worksheet for each DBK unit and

expanded to the required number of channels. Connection of modules on the worksheet

is performed by clicking on a module output (located on its right side), activating a

connection line. The connection line is then attached to the desired module with another

click. Recall the left-to-right flow of signal through modules mentioned before. Multiple

lines may be established simultaneously by touching the left side of the module to the

right side of the previous module and releasing. Once the filter modules are located and

connected properly, the filter properties of each individual channel must be specified.

Figure 3.5.7 illustrates the specification of global variable #1 for the filter channels while

Figure 3.5.8 illustrates the filter modules connected to the ADC modules on the

Page 115: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

101

worksheet. Additionally, the type, characteristic behavior and order of filter must be

specified.

Figure 3.5.7 – Defining filtration properties for each channel within the filter module dialog box.

Right-clicking the mouse may access the menu.

Global Variable

specified as low-

pass filter

Figure 3.5.8 – ADC and filter modules connected on the worksheet.

Page 116: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

102

3.5.3 - The Black Box Module

Once filter modules are installed for each DBK unit, the calibration of each must

be installed. This series of modules will allow for active calibration of the strain gages

and also the transducers.

To simplify the display and also shorten construction time another feature of

DASYLab will be used. The program has the ability to store a generic series of modules

on a separate, sub-worksheet called a Black Box, which can be imported to worksheets

instead of repetitively re-building similar modules. To illustrate use of this feature, a

black box will be constructed for the DBK43A strain gage unit, stored and then imported

to the worksheet in duplicate for the DBK65 transducer unit. To start, a black box

module is placed on the main worksheet near the DBK43A and opened. Once the black

box is opened, the background of the worksheet changes color to signify the change in

locale that the user is operating in. Figure 3.5.9 illustrates this process. Note that the

black box’s worksheet (right most image) has no modules located on it.

Figure 3.5.9 – Locating a new black box on the main worksheet (left) and opening the black box

(right). Note the color change of the background, distinguishing between the two.

Page 117: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

103

An Export/Import Module is placed on the black box worksheet and, when

prompted, the user specifies the module as an “Import Data” module. This module

establishes communication between the main worksheet and the black box, effectively

conveying the signals from the main worksheet into the black box. This module is

expanded to the appropriate number of channels and closed. As this example currently is

focused on the strain gage module, four channels will be specified for the import module.

Once the signal is available in the black box, a number of modules will be

required to complete the calibration of the gages and the unit conversion from voltage to

strain. Figure 3.5.10 illustrates all the modules to be installed in the black box for

calibration and unit conversion of the signal. The modules in Figure 3.5.10 are not

connected, but descriptors of their function in signal processing are noted.

Page 118: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

104

1

Signals are

sent to the

Offset module…

…then passed

into a

Digital meter…

…which captures

data for

calibration

Global variables are read by

the Scaling module,

converting voltage into strain

and displaying them in

another digital meter

Control switch and

Action module for

Offset Adjust

Figure 3.5.10 – Modules installed in the black box. The signal path and module functions are noted.

3.5.4 - Offset Adjustment of Signals

To complete calibration and unit conversion of the signal, three stages of signal

process are required:

• Removal of quiescent signals,

• Establishment of calibration for linear scaling, and

• Scaling of signals for continuous unit conversion.

Completing the first stage, a scaling module is placed after the import data module,

and specified as an “Offset Adjust” module. This will allow the user to “zero” the signal

Page 119: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

105

prior to acquisition. The module is expanded appropriately, re-named for clarity, closed

and connected to the import data module. Refer to Figure 3.5.11 for locations of

connection wires. It is important to note that the offset adjust module cannot operate

without a control switch. Separate switch and action modules are placed on the

worksheet and connected to each other. Figure 3.5.10 illustrates these control modules

while Figure 3.5.11 indicates their connection wires. To properly configure the switch,

the module is opened by left double-clicking and defined as a “One Shot Switch.” This

type of switch allows the user to instantaneously activate a predefined action. After the

event is complete, the signal returns to normal operation and does not necessarily remain

equal to zero. Additionally, the labels and configuration of the switch “button” may be

modified to the users liking by changing the labels in the “Text” section of the dialog

box. Figure 3.5.12 illustrates the switch module dialog box.

Figure 3.5.11 – Offset adjust modules and digital meter in black box.

Page 120: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

106

Selection of One

Shot Switch

Define button labels

Figure 3.5.12 – Specification of the Switch module operation.

Once the switch is configured, it is ready to operate on the action module (recall

the connections from Figure 3.5.11). The action module requires user input to indicate

the module acted upon and in the manner it will act. For purposes of this worksheet, it is

desired to set the signal value passing through the offset adjust module to zero. Figure

3.5.13 illustrates the action module dialog box, which is accessed by left double-clicking

on the action module. Once the switch and action modules are configured, they are able

to instantaneously zero the offset adjust scaling modules signal at the user’s discretion.

Page 121: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

107

Selection of Offset

Adjust module

Define the start

value as Zero

Figure 3.5.13 – Specification of the Action module operation.

Once the offset adjust modules are located, a digital meter is placed on the

worksheet to provide a visual display of the signal. It is important to note the location of

the meter in Figure 3.5.10-11, as connecting it prior to the offset adjust modules will omit

the effect that the zero switch has on the signal. The digital meter’s units do not have to

be established, as the default unit is volts and defined by the ADC. The characters used

by DASYLab to display units utilized by previous modules is “#0”. However, if the user

elects to modify the units of the signal at any module, any word or abbreviation may be

used. For example, “uStrain” will be input after converting the signal to microstrain later

in this example. Additionally, the inputs of the digital meter must be manually copied to

the outputs. This provides an output signal for additional modules to connect with,

passing the signal after the current modules operation to the outputs. A screen shot of the

digital meter’s dialog box can be seen in Figure 3.5.14.

Page 122: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

108

Module Name

Copy Inputs

Unit Definition

Figure 3.5.14 – Specification of the Digital Meter module operation.

3.5.5 - Establishment of Calibration Modules

To this point, the worksheet outlined previously is capable of obtaining voltage

signals from two ADCs, filtering them with a predefined low-pass filter, transmitting the

signals into a sub-worksheet (termed a black box) and equating the signal to zero at the

users discretion. In order to make tangible sense of the signals, they must be converted

into a unit of measurement. To maintain accuracy of this unit conversion, a calibration

procedure must be carried out. Calibration of the signals for this worksheet is discussed

below.

As quarter bridge strain gages are to be used, shunt calibration of each sensor may

be used to simulate a strain in each gage (details of shunt calibration may be found in

section 3.2). This simulated strain value may be used to define an additional point for

each individual gage. Using this simulated strain value and a zero point (gained by the

offset adjust module discussed prior) a linear relationship describing each strain gage’s

Page 123: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

109

behavior relative to voltage change can be defined. Figure 3.5.15 illustrates the linear

strain-voltage relationship. It is important to note that individual shunt resistors must be

installed on independent channels for this process. When the shunt resistors are activated

each individual strain gage channel will respond differently, as no two resistors are

exactly alike. Thus, prior calculation of the simulated strain is required.

Figure 3.5.15 – Linear scaling from shunt calibration.

In order to accomplish this shunt calibration procedure, a second scaling module

is required for the black box. However, instead of specifying an offset adjust module as

before, this new scaling module requires specification of a “Linear Scaling/Unit

Conversion” module. This module is located to the right, or after, the previous modules

and is depicted in Figure 3.5.16. Once located on the black box worksheet, left double-

clicking on the module opens the scaling dialog box, allowing the user to specify the

required number of channels.

Page 124: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

110

Linear Scaling/Unit

Conversion Module

Figure 3.5.16 – Locating the Linear Scaling/Unit Conversion module on the black box worksheet.

As each strain gage will respond to shunt calibration in a unique manner, the

following procedure is used to obtain calibration values. The signals are zeroed by the

offset adjust modules outlined previously. The shunt resistors are activated, producing a

simulated strain. The shunted output voltages of each signal are recorded as global

variables. These variables are then accessed by the linear scaling module, which creates

the two-point scaling illustrated in Figure 3.5.15. The shunt resistors are then

deactivated, returning the signals to their original state. This process is outlined

graphically in Figure 3.5.17.

Page 125: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

111

Control switch

When the switch is activated,

The signal is passed to the

Global Variable set module…

…and recorded as a

Global Variable…

…which establishes

the linear scaling of

the signal.

Figure 3.5.17 – Flow chart depicting the signal path while acquiring simulated strain voltages from

shunt calibration. These values are recorded as global variables and then read by the scaling module

to produce a linear strain-voltage relationship.

Unlike the filtering process where a single global variable was set manually, this

process involves the active definition of individual global variables, each specific to their

channel. To store each individual global variable, new modules must be located on the

black box worksheet. These modules are illustrated in Figures 3.5.16 and 3.5.17. First a

relay module with a control input is placed on the worksheet and expanded appropriately.

The relay module sends a single signal value through for each channel; in this application

a single voltage is allowed through. Each of the four channels are spliced between the

digital meter and the new scaling module and connected to the relay. A new “one-shot

Page 126: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

112

switch” is placed on the black box worksheet and connected to the relay. This switch

provides a trigger that activates the relay. A Global Variable Set module is then placed

after the relay and expanded to the appropriate number of channels. When actuated by

the switch, the relay passes the present signal values to the global variable set module,

which stores the values as predetermined global variables. The user must define

independent global variables for each channel, which are accessed in the Options menu.

Figure 3.5.6 illustrates the global variables menu. Figure 3.5.18 shows the dialogue box

for global variable set module. Note that individual global variables must be defined by

right clicking over the Global Variable box. The coding used by DASYLab for global

variables consists of $(…) where the ellipsis is a place holder for the global variables

name. In Figure 3.5.18 the global variable used is “STRAIN_CH0”. Subsequent global

variables used in this example were identified as STRAIN_CH1, STRAIN_CH2 and

STRAIN_CH3.

Figure 3.5.18 – Storing global variables for calibration.

Page 127: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

113

3.5.6 - Continuous Unit Conversion

As noted before, global variables are recalled by the linear scaling module, which

develops a linear relationship, converting the input signal into a strain value based on

shunt calibration. To establish this relationship, global variables must be acknowledged

by the scaling module, similar to the way the Global Variable #1 = 2 Hz was for the

filtering module. Left double-clicking can open the dialogue box for the linear scaling

module, which is illustrated in Figure 3.5.19. This screen shot illustrates the linear

scaling module’s dialog box where global variables can be recalled from storage by right

clicking over the x2 region. Additionally, individual simulated strain magnitudes from

shunt calibration must be input to the y1 region noted in Figure 3.5.19. These two inputs

define the “Simulated Strain” point from Figure 3.5.15. The lower point, referred to at

the “Zero Point” by Figure 3.5.15, is defined by [x1,y1] = [0,0] in the linear scaling

module. Additionally, as the linear scaling process converts the signal’s voltage to a

simulated strain value, the units of the signal may be modified from this module onward.

Note that Figure 3.5.19 indicates the location for defining a new unit. Figure 3.5.20

illustrates the overall black box worksheet completed for linear scaling of each individual

signal.

Page 128: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

114

Figure 3.5.19 – Setting linear scaling values for individual strain gages.

Figure 3.5.20 – Signal path of completed black box worksheet for linear scaling.

Page 129: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

115

Once the linear scaling modules are placed on the black box worksheet, an

additional digital meter is installed to observe the scaled strain values and an Export

module is placed to output values back to the main worksheet. The export module

conveys the now converted signals back to the main worksheet.

3.5.7 - Duplicating the Black Box for use with Transducers

The black box must be saved to duplicate it for use with the DBK65 unit. To do

so, the user must activate, or enter, the black box worksheet to be saved. Under the Edit

menu, Save As must be selected, and an ID Tag defined for the black box. The box must

then be saved to the hard drive for future access. Figure 3.5.21 illustrates the menus for

saving a black box while Figure 3.5.22 shows the dialog boxes defining an ID Tag, name

and location of the stored black box.

Selection of the Save

As function for

storing the Black Box

Figure 3.5.21 – Saving the active black box for future applications.

Page 130: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

116

Define the ID Tag for

the Black Box…

…and then save it for

future access.

Figure 3.5.22 – Flow of dialog boxes while saving a black box for future use.

Once saved, the black box must be exited and re-imported to the main worksheet.

The Load function may be found in the same location as Save As function illustrated in

Figure 3.5.21. The previously stored black box (denoted by the *.dbb file extension)

must be located on the hard drive and imported to the main worksheet. Once the “Open”

button is selected in the Load Black Box dialog box, the module will be automatically

placed on the main worksheet. However, the new black box is identical to the one stored

and must be modified (e.g. appropriate number of channels) for the DBK65 unit.

Additionally and unfortunately, as the newly loaded black box was created within the

same main worksheet, all the module names are now presented in duplicate. The user

must enter the imported black box and modify all the names to avoid conflict. Left

Page 131: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

117

double-clicking on individual modules opens them, allowing the user to modify each new

module within the black box. Only module names need to be modified.

3.5.8 - Configuring the Black Box for Use with Different Channels

As the newly loaded black box was configured for four strain gage channels, it

must be further modified to accommodate the DBK65s two separate transducer channels.

Additionally, as the transducers used in strain gage monitoring are often different (e.g. a

load cell and displacement sensor), dissimilar calibration relationships may be present.

Thus, the newly imported black box must be modified to calibrate both of the two

transducer channels independently.

3. Change the module that

the action module operates on.

2. All modules must be

reduced to 2 channels.

1. All modules must be renamed

4. Define new global variables.

5. Redefine the global

variables used by the

linear scaling module

for unit conversion.

Figure 3.5.23 – Modifications of the black box for DBK65 transducer channels. Note that step 6

(below) is not included in the discussion but is depicted in Figure 3.5.24.

Page 132: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

118

Performing the modification of the new black box for transducers entails a

number of details which are outlined as follows and are illustrated in Figure 3.5.23

(above):

1. Renaming all modules contained within the new black box to avoid conflicts.

2. Reducing the signal channels from four (for strain gages) to two for each

transducer. Modules may be simply opened and the number of channels

contracted; connection lines do not have to be modified.

3. Changing the module which the offset adjust action module operates upon.

4. Establishment of two new global variables for independent calibration of the

transducer channels.

5. Redefining the global variables and calibration setting of the linear scaling

module for unit conversion.

6. Copying Offset Adjust modules and Shunt Calibration modules within the black

box for use with an additional transducer channel. Figure 3.5.24 illustrates the

final reconfigured black box.

Figure 3.5.23 illustrates the final configuration of the black box module to be used with

the transducers.

Page 133: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

119

Figure 3.5.24 – Modified black box for DBK65 transducer channels.

Reconfiguration of the newly imported black box does not modify the behavior of

the signal processing outlined previously. The only significant change involved is the

production of two independent Offset Adjust and Shunt Calibration module series,

denoted by Transducer 1 and Transducer 2 in Figure 3.5.22.

Once both black boxes are configured, an additional switch and relay are used to control

the recording of data. An On/Off Switch is connected to a relay, which conveys the

scaled signals to a Write Data module. Figure 3.5.25 illustrates the signal path from both

of the black boxes to the Write Data module.

Page 134: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

120

Switch controls the relay…

…the relay passes the

signals to the Write Data

module…

…which records

data in a specified

location and format

continuously.

Note that all channels

are connected to the

relay.

Black Boxes

Figure 3.5.25 – Signal path from the black boxes to the Write Data module.

Within the write data module, the type of output file and its characteristics can be

defined. Figure 3.5.26 illustrates the write data dialog box.

Figure 3.5.26 – Specifying data recording options.

Page 135: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

121

Once the data collection module is configured the worksheet is able to record

experimental data in the specified file format. The fully configured main worksheet is

shown in Figure 3.5.27. Note the individual zero adjustment and calibration buttons for

the strain gages, transducer 1, and transducer 2 and also the start/stop button for

initializing data recording.

As noted, this worksheet is now adequately prepared for the data acquisition of

strain sensors and transducers. Figure 3.5.27 illustrates the DASYLab main worksheet

for monitoring four strain gages configured in a quarter-bridge circuit with individual

shunt resistors. Also, the main worksheet allows for individual operation and calibration

of two separate transducers. Finally, the user can specify when to start and stop the

continuous recording of the data, which is stored in a file format and location of the users

choosing.

Page 136: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

122

Figure 3.5.27 – Overview of completed worksheet.

Page 137: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

123

Chapter 4 – Development and Testing of a Portable Strain Sensor

4.0 – Introduction

In order to properly monitor the strain response of the De Neveu Creek IBRC

Bridge, selection of a low-cost, highly accurate measuring instrument was required.

From experience gained at Marquette University and also that of other organizations,

electrical resistance gages were selected as the best instruments for measuring short-term

behavior of the structure. However, installation of individual electrical-resistance gages

directly to the structure has a number of drawbacks. Primarily, the labor involved in

properly bonding the gages to a structure is significant. The reliability of field-applied

gages is also questionable. Previous attempts to record strain response of the De Neveu

Creek Bridge testify to this. Additionally, gages installed directly on the structure are not

removable and are vulnerable to environmental degradation and damage from a number

of other sources.

For these reasons, a removable and portable strain sensor was preferred for this

project. After consulting manufacturers of portable strain instrumentation devices within

the industry, it was found that the cost to implement the proposed instrumentation plan

would be prohibitive if pre-manufactured systems were used. As a result, it was decided

that development of a new cost-effective, reliable and removable strain sensor would be

the best option. The following section describes the selection of circuitry and necessary

calibration used for the strain sensors, research conducted toward final selection of the

sensor base material, and a description of the final sensor and its anchorage to the

Page 138: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

124

structural component. Further, the laboratory testing and finite element analysis

conducted to evaluate the performance of the final strain sensor configuration chosen is

detailed and the calibration procedure for each individual sensor is presented.

4.1 – Quarter Bridge Circuit Selection

A noted in section 3.2, the quarter bridge configuration of the Wheatstone bridge

can be constructed rapidly and offers an acceptable degree of precision. It was felt that

the additional sensitivity gained by implementation of a half or full bridge circuit did not

justify the increased expense and labor associated with these configurations. For

example, construction of a full bridge circuit requires more sensor material and

installation of three additional strain gages. While the expense of additional strain gages

is directly offset by the elimination of completion resistors used in a quarter bridge

circuit, the installation of the additional gages incorporates added labor. This stems from

the fact that installation of circuit completion resistors is quite simple relative to the

installation of four gages. Bonding of multiple strain gages in a constrained region

becomes increasingly difficult and leaves significantly less room for error. Thus, both

labor cost and time increases with full and half bridge circuits. If sources of error and

signal conditioning are appropriately addressed, the quarter bridge configuration can

provide satisfactory measurements at a low cost. Using this rationale, the quarter bridge

configuration was selected for the new strain sensor.

To reduce any error sources due to temperature fluctuations, a three-wire

configuration was employed for all gages. Relatively short lead wires were used in

Page 139: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

125

laboratory testing (all wires less than 30’) to reduce any desensitization that may be

present with long wire leads. Also, strain levels produced in all gages utilized only a

small portion of the available range, effectively limiting any non-linear effects.

The strain gages selected were Micro-Measurements CEA-06-250UN-350. These

350-ohm gages offer an increased electrical sensitivity over conventional 120-ohm gages.

Also, a thin coating is installed over the foil resistive array by the manufacturer, adding

increased protection. It is important to note that all strain gages were bonded according

to the procedures outlined by the manufacturer. All gages used for this project were

bonded to their substrate with Micro-Measurements M-Bond 200 Adhesive. Each quarter

bridge circuit used in the development of the sensors is illustrated in Figure 4.1.1.

All data acquisition equipment used was manufactured by IOTech and consisted

of a DaqBook 2000 series analog-to-digital converter, a DBK43A strain gage module and

a DBK65 transducer module. The DBK65 was used exclusively for load and

displacement monitoring and provided an excitation of 10 volts to the transducers.

Excitation provided by the DBK43A strain gage module was approximately 2.50 volts to

all channels with rationale discussed later. All data was captured at a sampling rate of 1

kHz, which is assumed to be much greater than twice any expected frequencies to be

encountered. Wheatstone bridges are intended to be direct current circuits and do not

contain oscillating frequencies. This high data acquisition rate also allows the user to

capture sudden changes in instrument response. However, a rapid sampling rate also

creates enormous amounts of data. Care must be taken to ensure that the space available

Page 140: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

126

for data storage during testing is adequate. A worksheet was constructed in DASYLab

similar to that found in section 3.5, which was used to simultaneously record strain, load

and deflection of all the instruments used during sensor development.

Vexec

RE

RB

RSENSOR

RA

Rshunt

Vo

+

-

Figure 4.1.1 – Quarter bridge circuit used during laboratory experimentation with the DaqBook

2000 system. Note that nomenclature established by the manufacturer is maintained.

Referring to section 3.2, it can be seen that this circuit simulates a tensile strain

when shunted. Resistive completion of this circuit is established by use of three precision

resistors. Each resistor has a value of 350 ohms, ±0.1%, and are referred to as Rn00A,

Rn00B and Rn00E (IOTech 2005). The completion resistors were manually soldiered

into a removable “plug” provided by the manufacturer. Figure 4.1.2 shows the DBK43A

module as used for all laboratory testing. A shunt resistor was installed in each plug for

shunt calibration. Recall equation 3.2.5, which describes the shunt resistance relationship

and may be rearranged to produce the following expression for direct calculation of

simulated strain.

( )

1 gage

simulated

shunt gageshunt gage

gage

R

GF R RR RGF

R

ε = = • ++

(4.1.1)

Page 141: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

127

To ensure accuracy of calibration, the resistance of each individual shunt resistor

was measured with a multimeter. Additionally, the resistance of each sensor and lead

wires was measured. Using these values, the proper tensile strain simulation was

manually calculated and input into the scaling module of the DASYLab worksheet. This

process was repeated for every individual strain sensor and strain gage used. For

example, during laboratory experimentation 64.9 k-ohm resistors were used for shunt

calibration. Using equation 4.1.1 we observe the following for a 350-ohm strain gage

with a gage factor equal to 2.105,

( )

350

2.105(64,900 350)

gage

simulated

shunt gage

R

GF R Rε = = =

+• +2,548µε (4.1.2)

120Ω resistors:Channels 3 & 4

350Ω resistors:Channels 1 & 2

Figure 4.1.2 – Completion resistor plug installed in the DBK43A module with top cover removed.

Note that channels 1-4 are occupied and thus have resistors installed.

4.2 – Material Experimentation and Selection

A number of materials were evaluated before final selection for the strain sensor.

To achieve the objective of developing a removable and portable strain sensor, it was

Page 142: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

128

decided that the quarter bridge strain gage needed to be bonded to a suitable carrier

material. This carrier would then be bolted to the structural component, transmitting any

strain to the carrier then the strain gage. A wide array of materials for embedded and

externally mounted sensors are available, however, they are most often polymer

composites and low modulus metals. Figure 4.2.1 shows an example of a nylon sensor to

be embedded in asphalt or concrete.

Figure 4.2.1 – Strain sensor with quarter bridge strain gage composed of nylon.

Experimentation was carried out with sensors composed of carbon-fiber

composite. The carbon fiber sensors were machined to accept a single strain gage

centered on the sensor and are illustrated in Figure 4.2.2.

Brownie installed on

bottom on concrete slab

(a)

(b)

Figure 4.2.2 – Carbon-fiber strain sensor with quarter bridge strain gage (a) and installation (b).

Page 143: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

129

These sensors were then installed on the top and bottom of an instrumented

concrete slab and bonded using an epoxy recommended by the carbon-fiber manufacturer

(Figure 4.2.2(b)). It was observed that the strain levels produced by the carbon-fiber

strain sensor differed greatly when compared to those produced by a complementary gage

bonded directly to the slab (Figure 4.2.3).

70 80 90 100 110 120

0

50

100

150

200

250

300

350

400

450

uStrain

Time (seconds)

Carbon-Fiber Gage

Bare Gage

Figure 4.2.3 – Recorded strain levels in carbon-fiber strain sensor and bonded strain gage.

The discrepancy in strain readings can be attributed to the fact that the carbon-

fiber sensor has a significantly higher stiffness relative to that of the concrete substrate.

Typical values of material modulus may be found in Table 4.2.1.

Material Modulus of Elasticity (ksi)

Carbon Fiber (fiber only) 33,500

Steel 29,000 - 30,000

Carbon Fiber Composite 22,500

Aluminum 10,000 - 11,000

Concrete 2,500 - 4,500

Nylon 300 - 500

Tyfo S - Epoxy 461

Table 4.2.1 – Typical modulus of elasticity values of materials (Fyfe 2005; Gere 2001).

Page 144: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

130

As the concrete fibers stretch or compress, strain is transferred through the epoxy layer.

Since the modulus of epoxy is much lower than that of the carbon-fiber plate, a large

amount of strain is absorbed by the epoxy layer. Thus, the true strain is never transferred

to the carbon-fiber plate and consequently the strain gage.

Stiffness differences between the measured substrate and sensor material can

affect the measured strains. Farhey (2005) reported that that the presence of a sensor

with greater stiffness than the measured substrate will disturb its natural response. As a

result of these findings, a lower modulus material was required for the sensor. For this

project the target substrate is composed of concrete and FRP materials and therefore the

modulus of the strain gage carrier should be compatible with these materials.

Furthermore, research of bonding materials for sensor attachment was conducted.

It was found that current epoxy and other commercially available adhesives are not

acceptable for bonding strain sensors. The elastic modulus of traditional epoxies is very

low causing the epoxy-adhesive layer to stretch significantly, thus lowered strain levels

are expected in the sensor. Based on this, a mechanical anchorage system is required for

attachment of strain sensors, with a carrier that has an appropriately low modulus of

elasticity.

4.3 – Description of Portable Strain Sensor

Based on the preliminary research conducted, a prototype sensor constructed of

series 6/6 Nylon was manufactured by ROMUS, Incorporated of Milwaukee, WI

Page 145: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

131

(ROMUS 2005). Its low modulus (approximately 400,000 psi) and relatively low cost

was ideal for both performance and mass-production of sensors. The material is also

easily machined allowing for detailed designs to be translated into prototypes. The

original prototype was a rectangular bar 1.00” wide by 4.00” long with a thickness of

0.25.” Figure 4.3.1 illustrates the final geometry of the prototype while Figure 4.3.2

shows completed strain sensors without their protective external coating or electrical

connector tabs. These items are discussed further below.

Two 0.386 in. diameter holes were located 0.50” from each end centered on the

width of the sensor carrier, allowing for mechanical anchorage via epoxy-adhered

threaded studs. These holes define the effective gage length of the sensor to be 3.00.”

Additionally, a central depression 0.50” wide by 1.50” long was machined 0.20” into the

sensor. A secondary depression 0.20” deep and 0.25” wide was machined into a single

end of the main depression to allow for strain relief of the lead wires.

4.00”

1.00”

0.50”

Strain Gage

0.50”

1.50”

0.25”Depression for strain

relief of wires

Oversized Bolt Holes

Figure 4.3.1 – Final configuration of the strain sensor.

Page 146: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

132

Figure 4.3.2 – Constructed strain sensors without connection tabs or protective coating.

Strain relief is achieved by bending the lead wires into the depression and

anchoring them with a quickset epoxy. These depressions allow for the strain gage,

necessary soldering and lead wire adhesive to be below the surface of the sensor,

reducing the risk of accidental damage to the gage. Further geometric detail is provided

in Appendix C.

As the strain sensors are to be installed on the De Neveu Creek Bridge,

environmental protection, rapid connection methods and the ability to remove the gages

at the end of the test are required (sensors will be used for two WisDOT bridges). To

attain a satisfactory level of environmental protection, the central depression each sensor

is filled with a rubber-like compound, M-Coat J, manufactured by Micro-Measurements.

This material a two-part polysulfide liquid polymer that completely seals the gage

(Micro-Measurements 2004). The polymer is relatively soft and will not affect the strain

Page 147: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

133

response of the sensors. Care was taken to isolate the exposed lead wires and gage from

the M-Coat with a Teflon-adhesive tape provided by Micro-Measurements. Additionally,

to ensure rapid deployment of each sensor, individual lead wires exiting the strain sensor

contain an individual, insulated quick-disconnect tab. These connections can be made

quickly and repetitively without an appreciable amount of electrical resistance. Male tabs

were soldiered to the lead wires on the sensor to ensure a durable connection, while the

female tabs will be installed on the lead wires of the bridge by crimping.

4.4 –Anchorage of the Sensor

As noted in section 4.2, a mechanical anchorage system is required for installation

of strain sensors. A simple but effective system is proposed as follows. Mechanical

anchorage for each sensor is to be provided by two 1/4” diameter, 3” long bolts with

standard plain washers on each face of the sensor. Each bolt is to be A307 steel. An

appropriate size nut, torqued to 120 lb-in, confines the washers and sensor. Deformed

washers, also called star washers, are not recommended, as they will significantly scar

and deform the nylon when tightened. Since the De Neveu Creek Bridge is composed

of concrete, each bolt is to be set in a 5/16” diameter hole and adhered with a high-

strength construction epoxy. Each bolt is to be set 1” into the substrate. All anchorage

components are to be manufactured by Powers Fasteners.

Figure 4.4.1 depicts a typical field installation. Transfer of load is accomplished

by friction between the substrate, washers and nylon, and does not rely on bearing of

bolts. Each attachment hole on the sensor is oversized for two primary reasons. The

Page 148: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

134

over-sizing of each hole eliminates the possibility of the bolts bearing directly on the

nylon. Through laboratory testing and numerical modeling it was found that bolt bearing

causes significant local deformations, ovalizing the hole and disrupting strain distribution

through the sensor. Figure 4.4.2 illustrates this effect. Additionally, use of slightly

oversized holes allows for reasonable out-of-plumb tolerances for the field installation of

the threaded studs.

Strain SensorBolt

w/nut

Standard Washers

Drilled hole w/Bolt

set in adhesive

1”

Figure 4.4.1 – Field installation of the strain sensor to concrete.

Bolt

Unloaded hole

Deformed hole under load

Figure 4.4.2 – Ovalization of a bolt hole under loading.

Tests were conducted to evaluate the mechanical anchorage system described

above. Further, a finite element model was developed to provide a comparison to the

data observed. The following two sections outline laboratory tests evaluating the

performance of the sensor and the anchorage system.

Page 149: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

135

4.5 – Laboratory Validation

To provide a consistent venue for evaluating the performance of sensors, a

constant-moment bending test was developed. This configuration produces a constant

curvature over a user-controlled length of the beam thus providing a constant strain at any

fiber along the entire length of the constant moment section. Figure 4.5.1 shows the test

frame and beam configuration used while Figure 4.5.2 provides further detail.

Figure 4.5.1 – Four-point bending test used for strain sensor evaluation. The data acquisition system

can be seen in the background, behind the test frame.

8’- 0”

3’- 0”2’- 6”

Load

Load Cell

W8x31

W6x20

1’- 6”

Figure 4.5.2 – Dimensioned constant-moment beam testing schematic.

Page 150: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

136

The primary bending member was a W6x20 shape, approximately 9’ long, bent

about its minor axis. Minor-axis bending was utilized to eliminate any lateral-torsional

buckling/instability effects when subjecting a segment of the beam to pure bending.

It is interesting to note that one end of the member had been cleanly sawn, while

the other had been cut by an acetylene torch, albeit very cleanly. Roller supports for the

W6 beam were located at 8’-0” with a 6” overhang on each end. Atop the primary beam

was a W8x31 spreader beam 4’-0” long, loaded about its major axis. Roller supports

were located 6” in from each end and centered on the primary beam. As a side note, a 5”

long by 2” deep section of the W8 had been previously removed for material testing. It is

assumed that this removed section had no effect on the distribution of loading during all

tests. Based on the material testing conducted and additional information of the

suppliers, both beams are composed of Grade 50 steel and are expected to maintain a

yield stress of 50yF ksi= .

Load was applied by a hand-actuated hydraulic ram, which was monitored by a

calibrated electronic load cell. Mid-span deflections of the primary beam were monitored

throughout testing. A linear displacement sensor (LDS) and a dial gage were located on

the beam for verification of displacements. The LDS monitored the displacement of the

beam web, 4’ from each support and at mid-span of the W6. Spatial constraints forced

locating the dial gage 4’ from each support but on the bottom exterior flange of the

primary beam. Both the load cell and the LDS were connected to the DBK65 transducer

module for real-time data acquisition synchronized with the strain readings.

Page 151: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

137

For laboratory evaluation, two sets of holes, 21/64” in diameter, were machined

into a single flange of the W6 beam. Each set of holes was offset vertically 1.76” from

the centerline of the web and centered at the mid-span of the beam as shown in Figure

4.5.3. The holes were set at a gage of 3”. Each sensor was attached with two Grade 8

5/16” diameter bolts. The tightening nuts were then gradually torqued in an alternating

fashion to 120 lb-in using a calibrated torque wrench. Special care was given to sensor

location when tightening the nuts. If during the tightening process the sensor moved

from its intended location, the nuts would be loosened and re-tightened with the sensor in

its appropriate location. This was done to ensure that the sensors were oriented parallel

to the flanges of the test beam at the target 1.76” locations.

Complementing these holes for strain sensor attachment were standard strain

gages, bonded directly to the beam on the opposite flange and centered in the same

locations (Figure 4.5.3). Idealizing the W6 as a perfect beam, engineering mechanics

requires that these gages “feel” identical strain as the sensor they are complementing.

This produces a tensile/ compressive pair of readings, each with a bonded gage

complementing a strain sensor for a total of four strain channels.

Page 152: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

138

Standard

Strain Gages

Strain

Sensors

Load equally

distributed to flanges

yb

yt

yt = yb=1.76”

Bolt Holes

Figure 4.5.3 – Mid-span layout of Strain sensors and complementary strain gages.

It is of significant note that the strain gages directly bonded to the main test beam

were Micro-Measurements EA-06-250UN-120. The gages differed from those used in

the sensors as they were only 120 ohms and did not have any protective coating. These

gages were used as they have proven to be very effective in strain measurement in the

laboratory. Also, 120-ohm completion resistors were implemented in the DBK43A for

these channels in lieu of the 350-ohm completion plugs used for the strain sensors.

However, the circuit configuration and shunt calibration process were the same. Figure

4.5.4 illustrates two strain sensors installed while Figure 4.5.5 shows the two bare gages

used for comparison values.

Page 153: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

139

3”

Figure 4.5.4 – Strain sensors installed for the constant-moment beam test.

Top and Bottom strain

gages used for comparison

with Strain sensors

Figure 4.5.5 – Complementary strain gages installed on the opposite flange as Strain sensors for the

constant-moment beam test.

Page 154: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

140

From the discussion noted in section 3.3, heating of the strain gage in absence of

an adequate heat sink will produce errors in measurements. The nylon material used for

the strain sensors may be classified as a poor heat sink and requires a low level of voltage

excitation (Micro-Measurements 2005). Calculations found in section 3.3 (Eqns. 3.3.2 -

3.3.4) are representative of this material and gage geometry and thus may be used for the

strain sensors. For this reason all excitation levels, including those of the bonded strain

gages, were set at approximately 2.5 volts. Those gages bonded directly to steel do have

an adequate heat sink available, but excitation was set at similar levels to maintain

consistency.

4.5.1 - Torque Level Tests

To evaluate the load imposed on the sensor when tightening the fastening bolts

(threaded stud pretension), an experiment evaluating torque levels on each individual

sensor’s mounting bolts was carried out. Two strain sensors were installed on the test

W6 beam and tightened to a pretension corresponding to a torque of 120 lb-in. The test

beam was then loaded to 5 kips and strain, load, and deflection data were recorded. The

frame was unloaded and sensors were removed and re-installed with a pretension

corresponding to 180 lb-in torque. The beam was loaded again to 5 kips. This process

was carried out for three additional pairs of sensors. All specimens were attached to the

beam with a washer only on the outside face of the sensor while the inside face was in

direct contact with the beam. A total of four tests were conducted at the 120 lb-in setting

and four at the 180 lb-in setting.

Page 155: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

141

The tensile and compressive values for the sensors and complementary strain gages were

then averaged and compared using the ratio Sensor BareGageε ε . Table 4.5.1 provides a

summary of data recorded during this test.

Instrument Comp Stdev Tension Stdev Comp Tension

Bare Gage -358 2 369 5 - -

120 lb-in Nylon -433 13 396 26 1.21 1.07

180 lb-in Nylon -418 23 396 21 1.17 1.07

Difference = 4% 0%

Results Nylon/Gage Ratio

Table 4.5.1 – Torque level test data. Compression and tensile values

reported are averages of test results.

It was observed that for tensile readings no change occurred. On the other hand,

compressive readings strain sensor values were 4% closer to the bare gage values at the

higher, 180 lb-in setting than the lower torque setting, albeit with greater uncertainty.

Explanation for this behavior will be provided in the finite element analysis section to

follow. While the higher pretension value did return results closer to the bare gage

values, implementation of this pretension in the field would require an embedment length

greater than 1”. This deeper embedment is not recommended as it may penetrate the

prestressing steel of the girders. Additionally, the higher threaded stud pretension setting

tended to significantly deform the soft nylon sensor material, which may lead to long-

term differences in individual sensor response as the instrument is removed and re-

installed. Figure 4.5.6 illustrates this effect.

Page 156: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

142

Figure 4.5.6 – Deformed region of washer contact due to tightening the anchorage nuts to 180 lb-in.

4.5.2 - Evaluation of Washer Presence

A test was performed to evaluate the effect that various support conditions had on

the strain sensors. Two conditions were selected for field installation: one with a

standard washer on each surface of the nylon strain sensor, and another with only a

washer on the outside of the sensor only. In the latter case, the sensor is closer to the

substrate being monitored, but is also rigidly supported in compression by the substrate

(finite element analysis presented later will illustrate this). Data for this case was

recorded during the torque level load test. Four pairs of sensors were attached to the test

beam, all with 120 lb-in torque levels but with washers on both nylon surfaces. The

beam was then loaded to 5 kips and data recorded. Data from this test are summarized in

Table 4.5.2.

Condition Comp Stdev Tension Stdev Comp Tension

Bare Gage -362 2 371 2 - -

Nylon w/double washers -419 31 399 23 1.16 1.07

Nylon w/single washer -433 13 396 26 1.20 1.07

Difference = -4% 0%

Results Nylon/Gage Ratio

Table 4.5.2 – Boundary condition test data. Compression and tensile values

reported are averages of test results.

Page 157: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

143

It was found that the tensile response of the sensors was nearly identical for each

washer condition. However, when in compression the double washer condition produced

results closer in magnitude to the corresponding tensile case. As having sensors that

behave similarly in tension and compression are desirable, all sensors developed in this

project utilize a washer on all faces of attachment (Figure 4.4.1). Further investigation

into the differences in strain readings is provided in the finite element analysis section to

follow.

4.5.3 - Excitation Voltage Evaluation

Sections 3.3 and 4.5 both contain comments describing the effect that gage

heating can have on measurement. As a precautionary measure, a test was conducted to

evaluate if the excitation voltage level selected for the sensors was in fact appropriate.

Four independent strain sensors were configured in a temperature-compensating half-

bridge circuit and tested using bolt pretensions outlined previously. The half bridge

configuration (Figure 4.5.7) eliminates temperature effects as both sensors experience

equal resistive changes due to any heating. As noted in section 3.2, the output of a

Wheatstone bridge is dependent only upon the ratio of gages on individual sides of the

circuit. The experiments conducted were run for a minimum of 20 minutes to allow long-

term heating to take place in the sensors. Each half bridge circuit contained a sensor

subjected to deformation by the bending test (RSENSOR) while the other sensor (RDummy)

was undisturbed.

Page 158: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

144

Vexec

+

-

350ΩΩΩΩ RSENSOR

350ΩΩΩΩ RDummy

Figure 4.5.7 – Half bridge temperature compensating circuit.

All sensors attached to the test beam were tightened to 120 lb-in and were

installed with only a single washer on the outside face of the sensor, even though double

washers are recommended for field implementation. Strain values recorded during this

test were then compared to data recorded from the other tests that were configured in the

standard, quarter bridge circuit that does not compensate for temperature effects.

Overall, no difference was observed between the temperature compensated half bridge

sensors and the standard quarter bridge sensors. Thus, the excitation level used for this

project is valid and is not expected to produce error in strain readings.

4.6 – Finite Element Analysis

To verify the response of the strain sensors from the constant-moment beam test,

a series of finite element models were constructed using the computer program ANSYS

(ANSYS 2005). Both three-dimensional (3D) models of the W6 test beam and the nylon

strain sensor were constructed.

Page 159: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

145

4.6.1 - Finite Element Model of Test Beam

First, a FE model of the test beam used was constructed. The material model used

for beam included a modulus of elasticity of 28,500 ksi and a Poisson’s ratio of 0.30.

The modulus of elasticity was back-calculated from observed deflection values of the test

beam and assumed cross-sectional properties. Additionally, these values are similar to

typical values for steel material. The geometry of the beam was modeled using

dimensions for a W6x20 found in the AISC manual (AISC 2001). It should be noted that

the values found in the AISC manual for flange and web thickness or overall depth of

section and flange width were not equal to those measured on the test beam, but were

reasonably similar. That is, when calculating the values for moment of inertia (both

major and minor axes) and the sections area they were found to be essentially equal to

those values published by AISC. Using the manual’s values, these dimensions were

input in a two-dimensional (2D) plane and meshed using PLANE42 elements. These 2D

elements were used solely as an initial step in construction of the three-dimensional (3D)

model. All meshing was mapped to lines by defining the number of elements to be

meshed within a defined area. This method of meshing allows the user to have greater

control on the formation of elements. The method of mapped meshing is illustrated in

Figure 4.6.1.

Mapped Meshing

7 divisions specified

per line7 elements per side

are created

Figure 4.6.1 – Illustration of the mapped meshing procedure used.

Page 160: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

146

Once the 2D model of the beam’s cross-section was constructed with an

appropriate arrangement of elements, it was extruded longitudinally, creating the third

dimension of the model. The 3D elements used were of type SOLID95, of which every

element contains 20-nodes has three degrees of freedom (DOF) per node – translation in

the nodal X, Y, and Z directions (Figure 4.6.2). These 20-node “brick” elements were

used throughout the FE model. When completed, the beam model contained 6,798

elements, and 21,320 nodes.

3 DOF: UX, UY, UZ

Z

X

Y

Figure 4.6.2 – 20-node brick element used for 3D modeling.

Boundary conditions of the beam model were provided at individual nodes for

support and loading conditions. Nodes representing the roller supports at beam-ends

were restrained in the vertical (Y) direction. Also, one end of the beam was restrained in

both the horizontal (X) and longitudinal (Z) directions. Figure 4.6.3 illustrates the

boundary conditions imposed on the beam model.

Page 161: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

147

UX = UY = UZ = 0UY = 0

PP

8’- 0”

3’- 0”2’- 6”

Figure 4.6.3 – Boundary conditions of the 3D beam model. The magnitude of load, P, is described in

the text.

Loads were also applied directly to the nodes at locations where the W8 spreader

beam contacts the main test beam. Idealized as roller supports, an aggregate load of

5,000 lbf was applied to the twelve nodes contacting the spreader beam. A force of

416.67 lbf was applied to each node in four groups of three nodes, simulating the total

force applied by the hydraulic ram

After the FE model was constructed and configured, a solution was produced.

Maximum vertical deflection was recorded at mid-span with a magnitude of 0.2019 in

downward. This value agrees well with deflections recorded during laboratory testing at

mid-span of the beam, which had a range of 0.19 to 0.20 in downward.

Page 162: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

148

From the FE beam solution, strains at the level corresponding to the location of

the bonded gages on the test beam were ±350 µ. Positive values are defined as tensile

strains and were located at the bottom of the beam, similar to those strains observed in

testing. It is noted that the strains observed were at ±1.75” from the center of web, while

the strain gages bonded to the test beam were at ±1.76” from center of web. It is felt that

this small difference does not provide reason for concern.

To validate the results of the FE model were computation from beam theory. The

same geometry and loading conditions as the FEM beam model (Figure 4.6.3), and the

same modulus of 28,500 ksi were utilized. Cross-section measurements of the actual

W6x20 test beam were made in the laboratory, yielding a minor axis moment of inertia

equal to 13.387 in4. The fillets of the test beam were omitted for calculation, as their

effect on the moment of inertia is considerably small. Using a distance from the section

centroid to the location of the bonded gages equal to 1.76”, strain at the location of the of

the bonded strain gages was calculated to be ±346 µ. This minute difference between

theoretical calculations and the FE solution indicates that the model is appropriate for this

analysis.

For comparison, strain values recorded by gages bonded directly to the beam in

the laboratory testing had a range of –364 (compression) to 380 µ (tension), which

correspond to a difference of 14 and 20 µ, respectively relative to the results of the beam

model. Overall, the results from the laboratory, theoretical calculations and the FE model

are very close. It was felt that both the strain and deflection values were similar enough

Page 163: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

149

to not warrant any further consideration and validated that the strain gages bonded to the

steel test beam were adequately shunt calibrated and working properly.

4.6.2 - Finite Element Model of Strain Sensor

In order to better understand the behavior of the strain sensor when strained, a

model of the strain sensor was developed. The strain sensors behavior under different

support conditions (washer presence tests, section 4.5.2) and under varying pretensions

(torque level tests, section 4.5.1) was not made clear during laboratory testing. It was of

great importance that a FE model of the sensor be created, providing an alternative venue

for comparison and evaluation. As the geometry of the sensor is asymmetrical, a detailed

FEM was created in a similar manner as the beam model. However, multiple iterations

of modeling were made in attempts to construct the most accurate model.

Acknowledging that strain values reported by the FE model would not exactly match the

values observed from strain gages on the test beam, and also the difficulty in recreating

identical boundary conditions for the sensor, it was decided that the accuracy of the

model would be based upon a comparison of actual values (the actual readings from

bonded strain gages) and FE model values. Boundary conditions would be varied in a

realistic manner until the results produced by the FE model “enveloped” the actual values

observed on the beam. Furthermore, it was expected that strain values for tension and

compression would be equal to each other for the double washer anchorage system

recommended in section 4.5.2. The modeling process is described below.

Page 164: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

150

The material model used for modeling the strain sensor included a modulus of

elasticity of 400 ksi and a Poisson’s ratio of 0.38. These properties were chosen, as they

are common values for the sensor’s base material, Nylon 6/6. For example, the

magnitude of the modulus value was identified as the mean value in wide array of values

for Nylon (Table 4.2.1). Initially, a 2D model of the strain sensor was created using

PLANAR82 elements and mapped meshing. All geometric details to be encountered

within the sensor were incorporated into this 2D model. This was done so that extrusion

of areas could be performed in stages, replicating the geometry of the sensor. An

illustration of the extrusion process is given in Figure 4.6.4. For example, circular lines

for the bolt holes were constructed in the 2D environment. When the planar model was

extruded “upward,” giving the model a defined thickness, the circular lines form a

cylinder that define the boundary of the bolt hole. As with the beam model, all 3D

elements were SOLID95 elements, composed of 20-nodes and of brick shape (Figure

4.6.2). The final sensor model contained 2,164 elements and 9,397 nodes.

(a) (b) (c)

0.00 in0.05 in

0.25 in2D Planar

mesh

Figure 4.6.4 – The extrusion process uses to build a 3D model of the sensor. It can be seen that from

figure (a) the entire 2D planar mesh is extruded 0.05” in figure (b). The center depression is then

created in figure (c) by extruding all areas around the depression.

Page 165: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

151

The boundary conditions of the sensor model varied greatly as it is difficult to

simulate the actual loading of the sensor through the washers and threaded studs. Recall

that all load transfer is to be achieved by the friction between nylon and washer; the

magnitude of such a friction force is challenging to reproduce. In lieu of applying loads

to the sensor model, it was decided that applying a prescribed longitudinal displacement

to the sensor model would accurately simulate the beam test.

A single end would be displaced while the other would be restrained from

displacement. This displacement represents the displacement of the bolts anchored in the

test beam and was calculated using a strain of ±370 µ, which is an approximate

midpoint for the strains experienced by the strain gages bonded to the test beam during

laboratory testing. The following calculation provides explanation to the imposed FE

model displacement of ±0.00111 in.

( )( ) ( )( )63.0 370 10 0.00111Longitudinal AppliedGageLength in inε −∆ = = ± • = ± (4.6.1)

By displacing the nodes in the FE model of the sensor Longitudinal∆ the model could

simulate the sensor deformation seen in the laboratory tests. However, how to apply this

displacement was not originally clear. It was decided that creating a suite of various

boundary conditions could “envelope” the true behavior of the sensor, providing a venue

for comparison. Further, these variations in boundary conditions could help explain the

results seen during the testing done to evaluate the presence of washers on the sensor

(section 4.5.2) and the results of the torque level tests (section 4.6.1).

Page 166: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

152

To envelope the proper boundary conditions of the FE model of the sensor, a

number of trials were conducted focusing on evaluating two primary situations. First, the

regions affected by bolt hole displacement were addressed. By systematically adjusting

the boundary conditions around the bolt holes, an accurate simulation of the contact each

washer has on the sensor was developed. Additionally, the interaction between the

sensor and the steel beam was addressed. Manipulation of the boundary conditions on

the surface between the sensor and the steel beam were varied to study the presence

washers have on the behavior of the sensor. The condition where no washers were

present between the steel beam and the sensor was evaluated, as well as the condition

where they were separated by a washer. Of particular note is the scenario illustrated in

Figure 4.6.5.

Sensor material

penetrating the steel beam

Figure 4.6.5 – Inadequately restrained model penetrating the steel test beam while subjected to

compression.

It was observed that when loaded, the sensor has a tendency to deform either

toward or away from the steel beam. In the case where no washers are present between

the steel beam and sensor (models 1-3 below), compression will cause the central region

of the sensor to deform into the beam, which is not possible. Additionally, when

subjected to tension, the opposite occurs. The central region of the sensor model deform

Page 167: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

153

away from the beam while the outer edges of the model deform into the beam. Thus

appropriate boundary conditions are required that prevent this type of impossible

deformation.

Additionally, it was observed through experimentation with the sensor model that

when the bolt hole regions were subjected to boundary conditions indicative of a large

bolt pretension, specifically the higher 180 lb-in pretension imposed on the threaded stud

in section 4.5.1, the compressive response of the sensor model would become more

similar to the simulated strain of -370µε. It is theorized that these more stringent

boundary conditions produce a more efficient channeling of strain through the sensor

model by providing a greater degree of longitudinal restraint. It was recognized that the

lower pretension value had been selected for reasons outlined in section 4.5.1, and the

imposed during the iterations of modeling were cognizant of this.

Overall, six iterations of the sensor model were developed and are described

below. A table of figure listings is provided in Table 4.6.1 for clarity.

Comp. Tension

1 Figures 4.6.6/7 Single -370 278 1.33

2 Figures 4.6.8/9 Single -378 389 0.97

3 Figures 4.6.10-12 Single -403 413 0.98

4 Figures 4.6.13/14 Double -404 404 1.00

5 Figures 4.6.6/15 Double -404 404 1.00

6 Figures 4.6.6/15 Double -404 404 1.00

Comp/Tension

Ratio

uStrainBoundary

Conditions

Model

Number

Washer

Condition

Table 4.6.1 – Boundary Conditions used for the FE model of the sensor.

Page 168: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

154

Strain Sensor

Stationary Bolt Single Washer

UX = UY = UZ = 0 UY = +0.00111in.

X

Y

Z

Y UZ = 0 UZ = 0

Displaced Bolt

Steel Flange

Figure 4.6.6 – Boundary conditions for the tensile case of the sensor model 1.

UX = UY = UZ = 0 UY = -0.00111in.

X

Y

Z

Y UZ = 0 UZ = 0

Figure 4.6.7 – Boundary conditions for the compression case of sensor model 1.

Strain Sensor

Stationary Bolt Single Washer

UX = UY = UZ = 0 UY = +0.00111in.

X

Y

Z

Y UZ = 0 UZ = 0

Displaced Bolt

Steel Flange

Figure 4.6.8 – Boundary conditions for the tensile case of sensor model 2.

UX = UY = UZ = 0 UY = -0.00111in.

X

Y

Z

Y UZ = 0 UZ = 0

Figure 4.6.9 – Boundary conditions for the compression case of sensor model 2.

Page 169: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

155

Rings around bolt holes to simulate

the presence of washers

Figure 4.6.10 – Ring of elements around bolt holes extruded through thickness of sensor for sensor

model 3.

Strain Sensor

Stationary Bolt Single Washer

UX = UY = UZ = 0 UY = +0.00111in.

X

Y

Z

Y UZ = 0 UZ = 0

Displaced Bolt

Steel Flange

Boundary conditions imposed on

entire cylinder around bolt hole

Figure 4.6.11 – Boundary conditions for the tensile case of sensor model 3.

X

Y

Z

Y UZ = 0 UZ = 0

UX = UY = UZ = 0 UY = +0.00111in.

Boundary conditions imposed on

entire cylinder around bolt hole

Figure 4.6.12 – Boundary conditions for the compression case of sensor model 3.

Page 170: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

156

Strain Sensor

Stationary Bolt Double Washers

UX = UY = UZ = 0 UY = ±0.00111in.

X

Y

Z

Y

UZ = 0

Displaced Bolt

Steel Flange

Only areas contacting the washer

receive boundary conditions indicated

UX = UY = UZ = 0

UY = ±0.00111Figure 4.6.14

Figure 4.6.13 – Boundary conditions for both compression and tensile cases of sensor model 4. Note

the incorporation of Double washers at each bolt.

Figure 4.6.14 – Detail of boundary conditions imposed to simulate contact with a washer for sensor

model 4. The restrained bolt hole is depicted in the figure.

Page 171: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

157

Strain Sensor

Stationary Bolt Double Washers

UX = UY = Uz = 0 UY = ±0.00111in.

X

Y

Z

Y UZ = 0

Displaced Bolt

Steel Flange

Only areas contacting the washer

receive boundary conditions

UY = ±0.00111UX = UY = Uz = 0

Figure 4.6.15 – Boundary conditions for both compression and tensile cases of sensor models 5 and 6.

Model 6 incorporated geometric changes only.

Sensor model 1 was the first trial in modeling the strain sensor. This model

incorporated boundary conditions preventing translation of the sensor into the steel beam,

simulating the condition where no washer exists between the sensor and beam. That is,

the sensor was not allowed to translate into (or away from) the Z-axis as indicated in

Figures 4.6.6-7. The prescribed displacement was applied to half the nodes on the

perimeter of the displaced bolt hole, at both top and bottom surfaces. Elements within

the depth of the bolt hole were did not receive boundary conditions (similar to the

boundary conditions depicted in Figure 4.6.14). Out-of-plane displacement was not

restrained at the displaced bolt hole except where indicated in Figures 4.6.6-7. Also, the

restrained bolt hole (all DOF=0) had boundary conditions imposed at nodes on half the

perimeter of such hole at both top and bottom surfaces. This model resulted in

compressive and tensile strains of -370 and 278 µε, respectively, which produces a

Page 172: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

158

compressive-to-tensile (C/T) ratio of 1.33. As these values are significantly different,

modifications were warranted.

Sensor model 2 was similar to number 1, however, the entire perimeter of the bolt

holes had boundary conditions imposed on them, at both top and bottom surfaces. These

revisions of the boundary conditions imposed were intended to more accurately simulate

the interaction of the washer around the entire perimeter of the bolt holes. This model

resulted in compressive and tensile strains of -378 and 389 µε, respectively. While these

values were more similar than the previous values, the magnitude of tensile strain was

now greater than compressive strain. As a result the C/T ratio became 0.97, which

indicates more similar performance between tension and compression. However, further

investigation was performed to try and improve the performance of the model.

For sensor model 3, a cylinder of elements was constructed surrounding each bolt

hole with an outer diameter of 0.5” (corresponding to the washer’s outer diameter) and

inner diameter of the bolt hole. All elements within the entire cylinder were displaced the

prescribed amount (Uy only) at one bolt hole and restrained (all DOF=0) at the other bolt

hole. Figure 4.6.10 illustrates the “ring” that was extruded through the thickness of the

sensor model to obtain the cylinder. The boundary conditions at the interface between

the steel beam and sensor were maintained similar to model 1 and 2. Strains produced by

the model’s solutions were -403 µε compression and 413 µε tension with a C/T of 0.98.

As these values increased nearly equally, yet maintained nearly the same C/T ratio, it was

felt that using the entire cylinder of the bolt hole only channeled more strain into the

Page 173: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

159

sensor, and did not change how it was distributed in the sensor. Thus, a new set of

boundary conditions were imposed on the model in attempts to attain a similar strain

magnitude in both compression and tension.

Sensor model 4 aimed at modeling a condition where the sensor and steel beam

are separated by a washer, termed “double washers.” The cylinder of elements around

each bolt hole from model 3 was utilized, however, only surface elements received

boundary conditions. For clarity, no elements within the cylinder received boundary

conditions (Figure 4.6..14). At the restrained bolt hole, the ring’s top and bottom

surfaces were restrained in all DOF. At the displaced bolt hole, the ring’s top and bottom

surfaces were displaced the prescribed amount. Additionally, at the bottom surface of the

displaced bolt hole out-of-plane translation (Uz) was restrained. Resulting strains from

this model were ±404 µε. This equality of strain magnitude was expected as the sensor

model was now able to deform freely in either direction. Upon review of the

deformation, the magnitude of out-of-plane deflection was minute (significantly less than

0.01”) and thus was not visible to the eye, or greater than the thickness of the washer.

Sensor model 5 maintained the same boundary conditions as model 4, however,

the out-of-plane translation (Uz) was restrained at all surfaces where washers contact the

sensor (Figure 4.6.15). The same strain magnitudes were produced by the model’s

solution (±404 µε), and were expected since incorporation out-of-plane deformation at

the displaced bolt hole from model 4 had been significantly small.

Page 174: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

160

The final model, sensor model 6, incorporated a geometric change to

accommodate the strain relief notch near the central depression where strain gages are to

be bonded. All boundary conditions imposed in model 5 were maintained. Figure 4.6.16

illustrates the notch added for strain relief.

Notch added in

sensor model 6

only

Figure 4.6.16 – Notch for strain relief of wires in sensor.

The same strain magnitudes were produced by the solution for model 6 (±404 µε)

and it is assumed that no degree of further refinement will produce differing values. A

contour plot of elastic strain distribution is provided in Figure 4.6.17.

Page 175: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

161

Compression Tension

± 404 µε

Notch for strain

relief of wires

Figure 4.6.17 – Longitudinal strain distribution ( )Yε for tension and compression cases for the final

sensor model.

In addition, the magnitude of strain produced by the final sensor model compares

relatively well to that reported during laboratory testing. In tension, the mean strain value

recorded when using washers on all faces of the sensor was 399 µ. The compression

case observed a mean value of –419 µ (Table 4.5.2). While the tension value is quite

similar to the FEM results, the slightly larger difference for compression is deemed

satisfactory given the complex interaction of the sensor and washers under compression.

A summary of results is provided in Table 4.6.1.

Comp Tension Comp Tension

Bare Gage -362 371 0.97

Nylon w/extra washer -419 399 1.05 1.16 1.07

FE Model #6 -404 404 1.00 1.09 1.09

Difference = 7% -2%

C/T

Ratio

Results Nylon/Gage Ratio

Ratios for the nylon sensor are taken with respect to the bare gage strains

while the FE model ratios are relative to a theoretical strain of +/- 370 µε.

Table 4.6.2 – Summary of finite element modeling and constant-moment best test results.

Page 176: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

162

It should be noted that the modeling conducted in this section attempts to simulate

an incredibly complex interaction of multiple events in a very simple manner. From

Figure 4.6.X it can be seen that significant bending occurs in the strain sensor when

under load. Shear lag stemming from the notably larger sides of the sensor relative to the

bottom “tub” of the sensor that houses the strain gage may cause differences in strain

response under tension and compression. Additionally, the notch created for strain relief

in the sensor causes changes in the strain field that may affect the sensors performance

when loaded. Furthermore, the anchorage system used may have differing effects when

subjected to tension or compression. Complexity arises from these issues and may be

superimposed when the instrument is loaded, causing uncertainty in the behavior of the

sensor.

However, these results are deemed appropriate as neither the bare gage values,

nor the strains recorded indicated a true compression to tension ratio of 1.00. Both values

were reasonable close to the desired C/T ratio, however. While it was expected that the

FE sensor model would be capable of attaining this ratio, it is also acknowledged that the

model operates under theoretically ideal conditions and the complexities described

previously lend variance to laboratory results.

As the finite element model of the strain sensor produced consistent results it is

felt that the constant-moment beam test would be an appropriate method to document the

individual behavior of the strain sensors. This documentation is required as they are

intended to record precise measurements of the De Neveu Creek Bridge. Additionally,

Page 177: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

163

while it is expected that all the sensors behave in a similar manner, it is acknowledged

that not every mass-produced sensor is truly identical. Given the small magnitude of

measurements to be conducted with these sensors, minute differences may have

appreciable affects on individual sensor performance.

4.7 – Calibration of Individual Strain Sensors for Field Implementation

To ensure accurate performance of the strain sensors in the field, a calibration

procedure was performed documenting the unique response of each individual sensor

manufactured under tensile and compressive loads. Described within this section is

documentation of the test procedure and methods used, the data recorded from testing,

and analysis undertaken to produce an adjustment factor for individual sensor readings in

the data acquisition system. Individual calibration is required for all of the strain sensors

as irregularities in manufacturing produce behavior distinctive to each specific

instrument.

4.7.1 - Calibration Method and Equipment Used

Given the success observed with the test frame detailed in section 4.5 it was

decided that few modifications were required to utilize the frame for calibration. Other

possible methods of calibration were considered to document the response of the strain

sensors, however, all either proved to lack precision or were too costly. For example,

procedures involving a cantilever bending apparatus do not produce pure axial strains in

the sensors. Also, methods utilizing displacement monitoring, such as optical sensors,

were not available in the laboratory.

Page 178: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

164

To ensure similar performance of the test frame and beams during the many load

tests, each roller support was welded to a primary member. Each roller received two tack

welds per side of the beams using a small wire welder. Welding the roller supports to

beams eliminated the possibility for independent movement but did not provide any

rotational restraint to the system. Figure 4.7.1 illustrates locations of welds.

Additionally, locations of members in the test frame were continuously monitored,

limiting the possibility of any relative movement that could introduce error into the

recorded data.

8’- 0”

3’- 0”2’- 6”

Load

W8x31 W6x20

1’- 6”

Welded roller

supports

Rigid beam

support

Figure 4.7.1 – Weld locations on beam members for the constant-moment load test.

All of the load tests were conducted in the following manner. Two sensors per

test were mounted to the W6x20 test beam, with one in compression and the other tension

as shown in Figure 4.5.3. Each individual nylon strain sensor was installed on the flange

with two Grade 8 5/16” diameter bolts. Washers were placed on the inner and outer

surface of the nylon so that neither the beam nor the fastening nut contacted the sensor

(recall the double washer recommendation of section 4.5.2). Each nut was then tightened

Page 179: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

165

to a torque of 120 lb-in in alternating fashion. The lead wires of each sensor were then

connected to an additional length of wire, which was connected to the data acquisition

module. The other strain sensor was then installed in a similar manner. As was done

with the load tests conducted during development and experimentation of the nylon strain

sensor, complementary strain gages were bonded directly to the opposite flange of the

main test beam. The centerline of these gages were located at the same elevation, which,

in theory, should produce similar magnitudes of strain (Figures 4.5.3, 4.5.5). Also similar

to previous load tests, load and displacement were continuously monitored during testing

by a calibrated load cell, linear position sensor (LPS), and dial gage. The load cell was

located directly under the loading ram on the W8x31 spreader beam while the LPS and

dial gage were located at mid-span of the main test beam.

Prior to load tests, calculated values for shunt calibration of the strain gages and

sensors were produced. Individual resistances of the four strain channels (two 120-ohm

bonded gages, two 350-ohm strain sensors) were read with a multimeter and recorded in

a spreadsheet. The simulated tensile strain was then calculated using the measured

resistance of each shunt resistor and the manufacturer’s gage factor. Simulated tensile

strain magnitudes and resistance for each strain sensor may be found in Appendix D.

Once ready, the data acquisition system was initiated, the sensors and bonded gages shunt

calibrated, and load slowly applied. The hydraulic ram was hand operated, increasing the

load level as uniformly as possible until a maximum load of approximately 3-kips was

reached. Data acquisition was then suspended and the beam slowly unloaded. Strain

sensors were then removed and reinstalled in reverse locations to record their opposite

Page 180: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

166

stain response, or removed entirely for two new sensors to be tested. A total of 35

sensors were tested in both compression and tension.

4.7.2 - Data Recorded during Load Tests

To develop calibration factors for implementation of the strain sensors in the

field, significant post-processing of the data occurred. As noted, each individual

calibration test recorded strain, load and deflection data documenting the response of two

independent strain sensors in tension and compression. Additionally, complementary

bonded gages noted in Figure 4.5.5 were also monitored to provide a baseline value for

comparison. The strain gages bonded directly to the beam are considered to be the actual

value of strain imposed on the W6 test beam and, thus, all calibration was subject to the

accuracy of these gages. Prior laboratory experiments and testing combined with in-situ

and other field installations of similar electrically resistive bonded strain gages have been

very successful and are considered acceptable. Furthermore, the analytical study

conducted previously supports the accuracy of the bonded strain gages (section 4.6.1).

In order to quantify individual strain sensor response relative to the bonded strain

gages, a calibration factor was developed. Given the predominantly linear response of

the strain sensors, it was decided that a simple coefficient multiplier would be

satisfactory. The following expression illustrates the calibration factor used,

( )SENSOR i

i

Gage

CFε

ε= (4.7.1)

Page 181: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

167

where ( )SENSOR iε is the strain recorded in Nylon sensor “i”, and Gageε is the strain

recorded in the corresponding bonded strain gage. The typical measured strain response

of both the bonded strain gages and the portable sensor is shown in Figure 4.7.2.

Figure 4.7.2 – Typical response of strain gages and sensors under applied loading. The load

indicated refers to that applied by the hydraulic ram to the test frame.

It can be seen from this figure that the strain sensors and gages (solid lines) very

nearly match their linear trend lines (dotted), which pass through the origin. Further, the

R-squared values for a linear fit of the measured data are noted, indicating that the trend

lines are very nearly equal. On the other hand, discrepancies exist. These differences

may be attributed to a delay in the strain response of the gages and sensors when loading

is applied rapidly. Note the slightly curved response of the data lines in Figure 4.7.2, and

the pronounced differences in Figure 4.7.3.

Page 182: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

168

A hand-actuated hydraulic pump was utilized for loading and true monotonic

loading increase was not possible. Figure 4.7.3 illustrates this delay in response. It is

important to note that the same sensors and configuration as from Figure 4.7.2 were used,

however, the data depicted in Figure 4.7.3 was disregarded as erroneous due to the rapid

loading.

Figure 4.7.3 – Erroneous response of strain gages and sensors under rapid, non-monotonically

increasing loading.

To produce the calibration factors for each gage, a load level of 2500 lbf was

arbitrarily selected at which strain readings would be analyzed. From figure 4.7.2 it can

be seen that when loaded at an appropriate rate the data forms a nearly linear line (R2 =

0.99), thus any load level would be appropriate to select data from. At this load, three

strain values were sampled from the data, containing values corresponding to both

Page 183: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

169

sensors mounted to the beam and both bonded strain gages, and averaged. The ratios for

compression and tensile response of the strain sensor under loading were computed.

Table 4.7.1 illustrates typical calculations performed for each load test, producing

calibration factors for two gages simultaneously.

Top Nylon [uStrain] Top gage [uStrain] Bot. Nylon [uStrain] Bot. gage [uStrain]

-172.4567 -174.029 183.514 180.1114

-172.4568 -174.0334 183.5069 180.1181

-172.4558 -174.0382 183.4995 180.125

Average = -172.46 -174.03 183.51 180.12

St Dev = 0.00 0.00 0.01 0.01

Calibration

Factor =0.991 1.019

Sensor #005 Sensor #006

Table 4.7.1 – Calculation of calibration factors.

4.7.3 - Individual Calibration Factors

Calibration factors developed for use with field-acquired data are listed in table

4.7.2. If a reading is indicated compressive, multiplication of the recorded strain reading

by the compressive calibration factor unique to that gage will produce the corrected strain

reading. Likewise, the opposite is true for tensile readings. It can be seen that for a

majority of the strain sensors, tensile and compressive response is similar. From the

results of the FE sensor model, calibration factors should theoretically be identical

between tension and compression. However, the anchorage behaves differently in

compression relative to tension, varying the response of the strain sensors. Overall, the

calibration factors for most sensors are relatively similar. It can be seen that in all but

three sensors below, the compression calibration factor was larger than the tensile factor,

indicating a consistently different response.

Page 184: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

170

Sensor Compression Tension

001 1.064 0.964

002 1.093 0.965

003 1.093 1.159

004 1.145 1.045

005 0.991 0.925

006 1.026 1.019

007 1.006 0.975

008 0.877 0.786

009 1.053 1.070

010 1.123 1.091

011 1.080 1.043

012 1.020 0.999

013 1.139 1.028

014 1.151 1.073

015 1.069 1.013

016 1.036 0.999

017 1.069 0.967

018 0.983 0.935

019 1.064 0.972

020 1.129 1.044

021 0.978 0.934

022 0.911 0.851

023 1.079 1.044

024 0.999 0.945

025 1.073 1.026

026 1.103 1.026

027 1.020 0.959

028 1.131 1.033

029 1.049 0.985

030 0.952 0.922

031 0.957 0.940

032 0.979 0.923

033 0.962 0.910

034 1.111 1.026

035 0.989 0.997

Average 1.043 0.988

Table 4.7.2 – Calibration factors developed for correction of laboratory acquired readings.

Page 185: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

171

Chapter 5 – Proposed Load Test

5.0 – Introduction

By conducting a load test of the De Neveu Creek IBRC Bridge, physical data is

produced aiding in the long-term monitoring of the structure. Also, valuable information

pertaining to the physical behavior of the bridge may be produced, allowing for analysis

of its structural performance and also the design methods used. In order to carry out the

load test for the IBRC project, the structure will be instrumented with strain sensors

detailed in previous sections and load tested. The arrangement of sensors and reasoning

for their placement is outlined below. Required equipment is also outlined in this section.

Finally, the load test procedures, vehicles, and expected results are contained herein. The

anticipated time of load test is the spring of 2006.

5.1 – Load Test Objectives and Instruments

The strain sensors outlined previously in section 4 will be used to obtain strain data

during testing and data acquisition focuses on the following objectives:

1. Strain profile of girders and deck (validate composite behavior)

2. Transverse distribution of wheel loads in bridge deck

3. Longitudinal distribution of load between girders

These objectives intend to complement the deflection data recorded during the previous

benchmark load test conducted by the University of Missouri – Rolla. Acquisition of

data periodically over the monitoring period will track degradation of the deck. Figure

5.1.1 illustrates the location of instruments to be used in executing these objectives.

Page 186: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

172

Mid-span

N

Enclosure Box

1

2

3

4

5

6

7

Third-point

of span Longitudinal strain sensor mounted to bottom of girder

Array of (5) strain sensors and string potentiometer for

transverse wheel load distribution

Array of (4) strain sensors for strain profile of girder and

FRP-reinforced deck

Figure 5.1.1 – East half of the De Neveu Creek IBRC Bridge indicating instrument locations. 32 total

strain sensors, two string potentiometers.

Longitudinal distribution of load between the girders will be conducted by

attaching individual strain sensors to the underside of bridge girders at mid-span and a

third-point of the span. As there are seven girders, fourteen individual sensors will be

installed with each sensor being centered transversely on the girder. Data collected

should then be analyzed using a procedure similar to that noted by Turner (2003).

Individual strain values measured in a girder will be divided by the total strain of all

seven girders recorded at similar locations, producing a fraction of total strain

experienced by that individual girder. This allows girder distribution factors to be

generated using the experimental data collected.

Page 187: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

173

Additionally, the global displacement of each girder should be monitored with a

surveying total station. Prisms should be magnetically mounted to the existing steel

plates located on the bottom of each girder from the previous load tests. Observation of

girder deflections during load testing will provide data allowing the results of the

previous load tests to be more directly correlated.

The strain profile of the bridge deck and girders is important to verify that

composite action of exists between the deck and girders. The structure was designed

assuming composite action and verification of such behavior is required. Additionally,

degradation of this composite behavior over time will be evaluated over the monitoring

period of this project. By locating strain sensors at the girder bottom, girder mid-height,

girder top flange, and on the FRP-reinforced deck as indicated in Figure 5.1.2, the strain

variation over the height can be recorded. Two instances of this array will be installed at

third-point of the span, totaling eight strain sensors (Figure 5.1.1). If the loads used in

testing invoke elastic response of the structure, the strain profile for the girder-deck

system will be linear as indicated in Figure 5.1.2, allowing for a rapid and simple

determination of the degree of their composite behavior through locating the neutral axis.

Figure 5.1.3 provides an illustration of differing levels of composite behavior, ranging

from full interaction between the deck and girder (diagram c) to no interaction between

the components (diagram a).

Page 188: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

174

4”

54”

37”

51”

n.a.

8”

25”

Figure 5.1.2 – Section of girder and deck for locating strain sensors. The strain profile on the right in

indicative of a fully composite cross section based on four sensor readings.

Figure 5.1.3 - Composite behavior stress profiles through a typical deck section (Lenett et al. 2001)

The transverse distribution of wheel loads throughout the FRP-reinforced deck is

the final objective of the load test. The FRP-grillage used for primary reinforcement of

the concrete bridge deck is a new material and structural system. Significant laboratory

testing has been conducted to date (Bank et al. 1992a; Bank et al. 1992b; Jacobson 2004),

but in-situ validation is lacking. Work by Conachen (2005) attempted to address the

issue, but due to the failure of instruments, little insight into the transverse behavior of

the bridge’s deck resulted. Thus, two arrays of strain sensors will be installed on the

underside of the bridge deck to evaluate wheel load distribution in the deck. Further,

evaluation of the AASHTO (1998) recommendations can be made.

Page 189: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

175

The upper surface of the bridge deck currently receives traffic, prohibiting the

installation of sensors at the roadway surface. Each array will contain five sensors,

located near the mid-span of the bridge deck (centered between the girders). Sensors will

be spaced at intervals of 18” with the middle sensor located at the target wheel position.

Furthermore, the sensors should be arranged in the staggered manner depicted in Figure

5.1.4.

2

1

5 spaces @ 18” o.c.

Strain sensors

oriented in

transverse direction

String potentiometer

centered between

sensors

3

Centerline of target

wheel load

Figure 5.1.4 – Plan view of strain sensors to monitor transverse wheel load distribution. Sensors are

located approximately at mid-span of bridge.

This staggering of sensors allows for observation of a larger region of deck while

minimizing the number of sensors required. Since the sensor arrays are to be located

significantly far from the abutments, the bridge’s skew is not expected to have any affect

on the behavior of the deck. The longitudinal spacing of the sensors is based in part by

research conducted at the University of Wisconsin-Madison (Dieter 2002) which focused

load tests of FRP-reinforced concrete slabs. Results produced during testing indicated

that the effective distribution region of a single HS-20 wheel load (approximately 20.8

Page 190: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

176

kips) was no more than 36” in either direction from the contact area. Figure 5.1.5

illustrates the lateral location of transverse sensor arrays and the string potentiometer.

21

3’-2.5” 3’-2.5”

Transverse sensor

String potentiometer

mounted on steel tube

Figure 5.1.5 – Section view of strain sensors for transverse wheel load distribution.

In addition to the five strain sensors in each transverse sensor array, a string

potentiometer will be installed to measure relative displacement at mid-span between the

girders and the deck. Each potentiometer unit should be attached to a rigid support,

preferably a steel tube, and secured adequately as shown in Figure 5.1.5. The moveable

end of the potentiometer should be withdrawn from the base unit and attached to the

substrate to be monitored by a hook or other mechanism in-line with the string. String

potentiometer model number PA-30 manufactured by UniMeasure, Inc is recommended

for use. These devices have a measuring string 30” long, which gives flexibility in

attaching the potentiometers. Figure 5.1.6 provides photographs of a string potentiometer.

Displacement data is intended to provide insight into the flexural response of the deck

under load in-situ deflection response.

Page 191: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

177

Moveable end in-line

and free to rotate

Measuring string

Figure 5.1.6 – String potentiometer manufactured by UniMeasure. The right photograph illustrates

a proper connection for measurements.

5.2 - Permanently Installed Equipment

As this study is aimed at evaluating the long-term behavior of the De Neveu

Creek IBRC Bridge, some permanently equipment is required to be installed reducing

effort for future tests and also to provide a greater degree of coherence between data

collected. Currently, protective PVC piping and lead wires for the instruments have been

installed, as well as an electrical enclosure box. The individual sensors have not been

installed on the bridge, nor have their embedded mounting bolts and require installation

anticipated for the spring of 2006.

5.2.1 - Lead Wiring for Instruments

The sensor arrays described above entail a total of 32 strain sensors and two string

potentiometers. Each strain sensor will operate with a three-wire quarter bridge circuit

(section 3.3), while the string potentiometers require three wires also. To provide these

Page 192: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

178

wires for each sensor, an appropriate length of NEMA Category 5e (Cat-5) wire was run

from the sensor location to a weather-proof, non-metallic enclosure box, which is

discussed later. Cat-5 wire is composed of eight individual copper wires, twisted in pairs,

and enclosed with vinyl sheathing. Three wires will be used for the strain sensors,

providing five extra wires in the event of damage or breakage. To protect the wires, they

were fed through conventional 1” PVC pipe, which was then sealed at all joints using a

PVC primer and adhesive. The PVC was rigidly affixed to the bridge girders and deck

using plastic conduit hangers and concrete screws. Figure 5.2.1 illustrates the PVC plan

layout installed on the De Neveu Creek Bridge for this project. Figures 5.2.2 and 5.2.3

provide photographs of the conduit installation. It is of note that all wires were run the

minimum length possible and terminated at close to the sensor location as possible with

appropriate lengths of wire for attachment.

Mid-span

N

Enclosure Box

1

2

3

4

5

6

7

Mid-span longitudinal sensors (7 wires)

Third-point of

span

Third-span longitudinal sensors (7 wires)

Transverse wheel load distribution instruments (10 sensor wires, 2

string potentiometer wires)

Girder and deck strain profile sensors (8 wires)

Figure 5.2.1 – Plan of sealed PVC pipes containing lead wire runs for individual instruments.

Page 193: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

179

Figure 5.2.2 – PVC piping terminating at the enclosure box. Dark regions along pipes are excess

primer used during sealing.

Figure 5.2.3 – Installation of the PVC piping housing instrument lead wires along girder #1. Note

the pipes running laterally from the four main pipes carrying wires to interior girders.

Page 194: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

180

5.2.2 - Enclosure Box and Screw Terminals

The enclosure box was mounted on the north end of the east abutment below the

bridge deck and adjacent to girder number 1. The box is constructed of painted fiberglass

and contains a weather-proof lid that can be locked to prevent unauthorized access.

Additionally, electrical screw terminals were provided inside the enclosure to allow for

termination of the lead wires at the abutment. These terminals are isolated conductors

that retain the lead wires from instruments on the bridge and carry the signal to

connecting wires leading to the data acquisition system. It is intended that the six pin

Mini-DIN connecting wires required by the DBK43A and DBK65 modules of the data

acquisition system will be installed and connected to these terminals prior the load test.

Figure 5.2.4 illustrates a typical screw terminal connection while Figure 5.2.5 illustrates

the connection diagram in the enclosure box for the lead wires as installed in the field.

As an additional note, installation of the screw terminals required penetration of the

enclosure’s fiberglass shell but was thoroughly sealed with a silicon caulk, maintaining

the weather-tight integrity of the enclosure. The enclosure can be seen in Figure 5.2.6

with the screw terminals installed and prior to lead wire connection.

Individual Screw

Terminal

Lead Wire from

Sensor on Bridge

Connecting Wire to Data

Acquisition System

Rear Wall of Enclosure Box

Figure 5.2.4 – Detail of a typical screw terminal connection.

Page 195: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

181

(21) Terminals for mid-span longitudinal strain sensors

Girder

Sensor 1

Girder

Sensor 7

Girder

Sensor 1

Girder

Sensor 7

(21) Terminals for third-span longitudinal strain sensors(12) Terminals for

Girder #2 strain

profile sensors

(12) Terminals for

Girder #1 strain

profile sensors

(15) Terminals for transverse wheel load sensors

and string pot between Girders #2 and #3

LM-1 LM-3LM-2 LM-5LM-4 LM-7LM-6P1-B P1-M P1-T P1-D

LT-1 LT-3LT-2 LT-5LT-4 LT-7LT-6P2-B P2-M P2-T P2-D

(15) Terminals for transverse wheel load sensors

and string pot between Girders #1 and #2

T1-W2 T1-MT1-W1 T1-E2T1-E1 T1-DT2-W2 T2-MT2-W1 T2-E2T2-E1 T2-D

Sensor Farthest

WestSensor Farthest

East

Sensor Farthest

WestSensor Farthest

East

Girder

Bottom

Bottom

of Deck

Girder

Bottom

Bottom

of Deck

Figure 5.2.5 – Diagram of lead wires terminated in the screw terminals in the enclosure box.

Figure 5.2.6 – Enclosure box housing lead wire connections.

Page 196: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

182

Nomenclature used in the enclosure box diagram is as follows:

Transverse Wheel Load Distribution Sensors:

T1-XX indicates the deck span between girders #1 and #2 while T2-XX refers to

the deck span between girders #2 and #3. Figure 5.2.7 illustrates the labels.

• T1-W2 – Transverse strain sensor west most from the middle of span.

• T1-W1 – Transverse strain sensor west most from the middle of span.

• T1-M – Transverse strain sensor located at mid-span.

• T1-E1 – Transverse strain sensor east most from the middle of span.

• T1-E2 – Transverse strain sensor east most from the middle of span.

• T1-D – String Potentiometer displacement transducer located slightly east of mid-

span.

2

1

3

T1-W

1

N

T1-W

2

T1-M

T1-E1T1-E2

T1-D

T2-W

1

T2-W

2

T2-M

T2-E2

T2-E1T2-D

.

Centerline of target

wheel load

Figure 5.2.7 – Labels for transverse strain sensors at mid-span of the bridge.

Girder Strain Profile Sensors:

P1-X indicates sensors installed at third-span of girder #1 while P2-X refers to

sensors installed at third-span of girder #2. Figure 5.2.8 illustrates the labels.

• P1-B – Sensor located on bottom flange of girder, approximately 4” from bottom.

Page 197: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

183

• P1-M – Sensor located near the mid-depth of girder, approximately 37” from

bottom of girder.

• P1-T – Sensor located on the underside top flange of girder, approximately 3”

from top of girder (51” from bottom)

• P1-D – Sensor located on the FRP-reinforced concrete deck atop the girder,

adjacent to the top flange of girder.

P1-D

P1-T

P1-M

P1-B

1

P2-D

P2-T

P2-M

P2-B

2

Figure 5.2.8 – Labels for girder strain profile sensors at third-span.

Longitudinal Sensors

LM –X indicates a strain sensor located at mid-span of the structure while LT-X

refers to strain sensors located at third-span of the bridge. “X” indicates the girder

number the sensor is located on. Figure 5.2.9 illustrates the labels.

Page 198: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

184

Mid-span

N

1

2

3

4

5

6

7

Third-point

of span

LM-1

LM-2

LM-3

LM-4

LM-5

LM-6

LM-7

LT-1

LT-2

LT-3

LT-4

LT-5

LT-6

LT-7

Figure 5.2.9 – Labels for longitudinal strain sensors.

5.2.3 - Installation of Strain Sensors

As noted elsewhere in this document, the strain sensors developed for this project

are to be mounted to the De Neveu Creek Bridge using an anchorage system consisting of

¼” diameter steel bolts embedded in the concrete girders and deck. Two anchor bolts are

required for each sensor and require washers on all exposed nylon surfaces. A diagram

of the anchorage system is provided in Figure 5.2.10.

Strain SensorBolt

w/nut

Standard Washers

Drilled hole w/Bolt

set in adhesive

1”

Figure 5.2.10 – Illustration of the anchorage system for strain sensors.

Page 199: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

185

Each 3” long by ¼” diameter bolt will be set in a 5/16” diameter holes drilled into

the concrete. Embedment should be approximately 1 inch (Powers 2005). After drilling

holes, they should be blown clean and any debris moved prior to injecting the adhesive.

Once the adhesive has been injected to the hole, the anchor rod should be inserted,

twisting the rod as it is embedded. The gel time of adhesive is 15 minutes, requiring that

the anchor rod be temporarily restrained by duct tape or other means while the adhesive

cures. Full cure time for the recommended adhesive is 24 hours (Powers 2005).

However, concern is warranted when drilling into the prestressed concrete girders. The

construction plans for the De Neveu Creek Bridge indicate that a minimum of 1” cover is

provided for all steel reinforcing in the bottom of girders(WiDOT 2003). Penetration of

stressed steel tendons contained in the girders can be extremely dangerous and, thus,

maximum embedment of the anchor bolts will be 1 inch for safety. When the anchor bolt

are set and cured, each sensor should be “sandwiched” with standard washers. The

retaining nut should be tightened to 120 lb-in.

The anchorage system to be used recommended for this project is as follows:

• Anchor Bolts – Powers Fasteners straight anchor rods, ¼” dia. x 3” long of

ASTM A307 steel, product number #07980.

• Washers and Nuts – ½” o.d. x ¼” i.d. plain steel washers and standard ¼”

diameter nuts. Both are included by Powers Fasteners with purchase of anchor

rods.

Page 200: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

186

• Adhesive – Powers Fasteners PowerFast+ adhesive system, Fast Set formula. The

It is of note that at 75 o

F, only 15 minutes of dispensing time is provided by the

epoxy (Powers 2005).

• Drill Bit – 5/16” diameter carbide masonry bit, driven by electrical hammer drill.

It should be noted that when subjected to strain levels anticipated during this load test

(less than 200 µ) the total force required to be sustained by the anchor rods embedded in

the concrete is approximately 12 lbf (Figure 5.2.11).

Strain Sensor

F

( ) ( )20.15 200 400,000 12SensorA E in psi lbfε µε= = =i iF

Anchor Rod

Estimated force in strain sensor:

FF/2

F/2

Friction Force

of Washers

Figure 5.2.11 – Approximation of the force acting on the anchor rod.

Assuming that the anchor rod is rigid, this force is opposed in compression and tension

by the adhesive bonded to the anchor rod present in the hole drilled into the concrete.

From this, a small distribution of stress (both tensile and compressive) is produced in the

epoxy, which has compressive and tensile strengths of 11,125 psi and 7,250 psi,

respectively (Powers 2005). It is felt that these stress levels are not appreciable enough to

Page 201: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

187

warrant further concern and it is anticipated that even at large strain levels, the anchor

bolts and epoxy will be sufficient.

5.3 – Load Test Vehicles and Test Configuration

In order to maintain continuity with the data collected during load testing in

September 2004 by the University of Missouri – Rolla and the University of Wisconsin –

Madison, use of similar test vehicles is recommended. As documented by Conachen

(2005), three-axle dump trucks loaded with loose gravel were used for the load tests and

referred to as a “H-20 Dump Truck.” Figure 5.3.1 shows the approximate dimensions of

the test vehicle.

14’-0”

4’-6”

6’-0”

14’-0”

4’-6”

23’-8”

6’-0” 7’-0”

Figure 5.3.1 – Test vehicle dimensions for UM-R/UW-M load test. Adapted from Conachen (2005).

The H-20 Dump Trucks used in testing of the De Neveu Creek Bridge had an

average front axle weight of 26,013 lbs., an average rear tandem (both rear axles) weight

of 49,913 lbs., and a average total weight of 74,946 lbs. (Conachen 2005). As noted

prior, the use of similar test trucks is recommended for all future testing.

Page 202: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

188

5.3.1 - Load Test Objectives

The objective for each load test is described below.

Load Test Objective Addressed

1. Longitudinal distribution of load between girders and overall

girder deflections.

2. Strain profile of girders and deck.

3. Transverse distribution of wheel loads in bridge deck.

During the previous testing, four load tests were conducted on the De Neveu Creek

Bridge (structure B-20-148) to monitor load distribution between girders. As the bridge

is currently under traffic, load tests similar to those conducted by the University of

Missouri - Rolla are not possible. Rather, it is anticipated that only a single traffic lane of

the bridge be closed to traffic during testing, while the other lane will require temporary

traffic stoppages to conduct each load test.

Continuous data acquisition of each load test should be performed from start to

finish of all tests. To perform each load test, readings of appropriate sensors should be

made before the test vehicles are placed on the structure, providing a baseline for each

individual test. Readings taken from all sensors before the test vehicles are located on the

structure should then be tarred, or “zeroed.” All the test vehicles should then be driven

onto the structure and located in their intended positions. When all vehicles are in their

correct position they should remain idle for five minutes and then driven off the structure.

Once the test vehicles are absent from the bridge, another settling period of five minutes

should be conducted before the data acquisition is stopped. Finally, resting periods

Page 203: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

189

during data acquisition allow for any dynamic effects induced by movement of the test

vehicles to dissipate, providing static data for analysis. However, in the event that any

additional vehicles cross the structure during data acquisition, documentation should take

place to clarify the data when analyzed later.

The protocol for deflection measurements made with surveying equipment is

similar to the data acquisition system. Readings of girder deflection should be made

prior to locating the test vehicles on the bridge, after the vehicles have been located on

the structure and the settling period finished, and finally after the vehicles are removed

from the bridge and another settling period complete.

5.3.2 - Load Test Configurations

Load Test #1 will be conducted similar to the fourth load test conducted by the

University of Missouri-Rolla (Bank 2005). However, as coordinating separate test

vehicles on the structure in a rapid manner may prove difficult, only three test vehicles

are recommended. It is anticipated that use of three test vehicles will produce moments

in the bridge girders large enough to produce significant strain levels. Usage of fewer

vehicles may cause strains in the girders smaller than 30 µε; larger strain levels tend to be

more reliable. The train of test vehicles are be positioned centrally above girder 2, as

indicated in Figure 5.3.2.

During this test the 14 longitudinal strain sensors should be activated to observe

the distribution of load among girders. Based upon an estimated axle load (AASHTO

lane load) of 25-kips, the projected strains to be observed in the sensors are presented in

Page 204: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

190

Table 5.3.1. Additionally, observation of the overall girder deflections should be

recorded during this test.

DWT Device

Strain Gauges

“Rigid” Strut

17.5’

42’

66.5’

130’

(3) Test Vehicles

21 3 4 5 6 7

Figure 5.3.2 – Load Test 1 test vehicle locations.

Sensor Strain (µεµεµεµε)LM-1 145

LM-2 145

LM-3 132

LM-4 105

LM-5 66

LM-6 46

LM-7 20

LT-1 107

LT-2 107

LT-3 97

LT-4 78

LT-5 48

LT-6 34

LT-7 15

Mmid-span = 58800 k-in

M3rd point = 43320 k-in

0.22

0.22

Load Test 1

Distribution Factor

0.20

0.16

0.10

0.07

0.03

0.22

0.22

0.20

0.16

0.10

0.07

0.03

Table 5.3.1 – Load Test 1 estimated strain magnitudes. Distribution numbers are based on a single

lane loading and corresponding load test by UM-R (Bank 2005).

Page 205: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

191

Strains produced in Table 5.3.1 were computed using the following expression,

( ) i bottomi

Composite

M yGDF

E Iε

×=

× (5.3.1)

where IComposite = 589,843 in4, E = 5,583 ksi, ybottom = 36.84” and Mi refers to the

moments at mid-span (65’) and third-point (88’ from left) produced from the loading

illustrated in Figure 5.3.2. Moments were calculated using a 2-dimensional linear-elastic

model in MASTAN2 containing the section and material properties listed previously

(Ziemian and McGuire 2002). GDF in the above expression refers to the distribution

factors of the spring 2004 load test conducted by the University of Missouri – Rolla

(Bank 2005). All properties are that of a fully composite section and compare well with

the properties outlined in Conachen (2005).

17.5’

22’

36’

42’

46.5’

60.5’

66.5’

71’

85’

130’

P ˜ 25 kips, each axle load

Figure 5.3.3 – Axle positions of a single girder during Load Tests 1 and 2. Drawing not to scale.

Load Test 2 uses the same three-truck trains as the previous load test. However,

instead of placing a line of wheel loads directly over a girder, a truck train will be

centered above girders 1 and 2. Figure 5.3.3 provides locations of the test vehicles.

Page 206: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

192

DWT Device

Strain Gauges

“Rigid” Strut

17.5’

42’

66.5’

130’

(3) Test Vehicles

21 3 4 5 6 7

Figure 5.3.4 – Load Test 2 test vehicle locations.

During this test the two arrays of strain profile sensors (Figures 5.1.2 and 5.2.8)

should be active for data acquisition. Results obtained during this test will provide

insight into the composite behavior of the girder and concrete deck. Estimated strain

values using the girder distribution factors reported in Table 5.3.1 for the strain profile

sensors are found in Table 5.3.2 below.

Sensor yi Strain (µεµεµεµε)

P1-B 32.84 95

P1-M 0 0

P1-T -17.16 -50

P1-D -25.16 -73

P2-B 32.84 95

P2-M 0 0

P2-T -17.16 -50

P2-D -25.16 -73

Load Test 2

Table 5.3.2 – Estimated strains for girder profile sensors installed at third-span of girders 1 and 2.

Negative values indicate compression while positive values refer to tension.

Page 207: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

193

Load Test 3 will be conducted to observe the region in which load is distributed in

the FRP-reinforced deck under a single axle load. The front wheels of a test vehicle truck

will be located directly over the sensor arrays installed on the underside of the deck. It is

assumed that the rear tandem axle is of the test vehicle is far enough from the front axle

as to not affect performance of the sensors. A wheel will be positioned directly over the

array between girders 1 and 2, which is centered on the deck span. Figure 5.3.4

illustrates the location of test vehicle wheels at mid-span of the bridge. The lateral

positioning of the test vehicle will be such that both sensor arrays may be activated and

observed simultaneously.

21

EQ.

3

EQ.

Figure 5.3.5 – Load Test 3 test vehicle wheel locations at mid-span of the bridge. Only front wheels

of the test vehicle should be located at mid-span.

5.4 – Data Acquisition System

The data acquisition system to be used for testing consists of an IOTech DaqBook

2001 module, three IOTech DBK43A strain gage modules and an IOTech DBK65

transducer module. The DaqBook acts as an analog-to-digital converter for the system

while and the DBK modules perform signal condition of the strain sensor and transducer

Page 208: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

194

signals. An overview of data acquisition operation is presented in section 3.1.

Photographs of the modules are found in Figure 5.4.1 through 5.4.4.

Figure 5.4.1 – Photograph of the data acquisition system used for load testing. Pictured from top to

bottom are three DBK43A modules, the DBK65 module and the DaqBook 2001 module. Front panel

(left) and rear panel (right).

Figure 5.4.2 – Photograph of the DaqBook 2001. Front panel (left) and rear panel (right).

Page 209: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

195

Figure 5.4.3 – Photograph of the DBK43A strain gage module with cover panel removed. Front

panel (left) and rear panel (right).

Figure 5.4.4 – Photograph of the DBK65 transducer module with cover panel removed. Front panel

(left) and rear panel (right).

Recording of data during the five load test configurations should be conducted in

a continuous manner. As noted previously, the strain readings should be tarred to provide

a starting value of the sensors, test vehicles located on the bridge and allowed to rest, then

driven off the structure and followed by a settling period. The data acquisition system

should be active and recording for this entire process. A sampling rate of 25 Hz is

recommended for acquisition to better document any rapid response of the bridge to the

loads. Previous efforts in bridge monitoring have utilized sampling rates between 25 Hz

and 40 Hz successfully (Bridge Diagnostics Inc. 2002; Lenett et al. 2001). Additionally,

Page 210: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

196

the fact that continuous acquisition of data consumes a large amount of computer storage

space must be addressed. Use of very rapid sampling rates will produce unnecessary

amounts of data and occupy an inordinate amount of hard drive space of which a majority

of such is unnecessary. For these reasons an effort should be made to position vehicles

on the bridge as quickly as possible, limiting the amount of data collected with the

recommended 25 Hz sampling rate.

5.4.1 - Signal Conditioning Modules

The signal conditioning modules used for this load test will be the DBK43A strain

gage module and the DBK65 transducer module. Configuration of each is provided

below.

The DBK43A strain gage module requires few modifications to prepare it for use.

Analog filtering in all strain gage channels will be utilized given significant possibility of

environmental interference produced in the lead wires. The DBK43A modules are

supplied with a 3.7Hz Butterworth filter (section 3.1) and are enabled by changing the

electrical jumpers on the circuit board. The proper orientation of jumpers is given on the

circuit board. Figure 5.4.5 illustrates the location of the filters on the circuit board and

their enabling jumpers.

Page 211: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

197

Front Panel

Rear Panel

Analog

filters and

jumpers

Figure 5.4.5 – Analog filter locations on the DBK43A module.

As a quarter-bridge circuit will be used for all strain sensors, three bridge

completion resistors are required to construct the Wheatstone bridge circuit. An

introduction to Wheatstone bridge measurements is provided in section 3.2. The

DBK43A uses removable “headers” that can be configured for full, half or quarter bridge

circuits. Each sensor channel requires an individual header, which is illustrated in Figure

5.4.6 below.

“A”“B”

“E”Strain

Gage

Completion Resistor

“Header”

Jumper

configuration

Completion

resistors

Figure 5.4.6 – Quarter Bridge circuit and completion resistor configuration for use with the strain

sensors. Adapted from (IOTech 2005).

Page 212: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

198

Note that both completion resistors and jumper settings must be modified as

shown in Figure 5.4.6 to properly configure the quarter bridge strain gage circuit. As

resistance of each strain sensor is 350-ohms (±0.3%), the use of high-precision

completion resistors is required. Thus, 350-ohm (±0.1%) completion resistors were used

to build each resistor plug, completing a nearly balanced Wheatstone bridge circuit.

Location “H” illustrated on the completion header in Figure 5.4.6 is reserved for a shunt

calibration resistor. The shunt calibration resistors recommended for use in all strain

gage channels are 165-kOhm. When used with the strain sensors (Gage Factor = 2.105,

350-ohm), this shunt resistor produces approximately 1000 µ in tension, which is

greater than the expected strain response during the field test. Each shunt resistor should

be measured with a multimeter or other precision instrument so that the simulated strain

may be calculated. Refer to section 3.2 for further discussion on shunt calibration.

5.4.2 – Connection to Strain Gage Modules

The connection interface between the DBK43A module and each strain sensor

requires the use of a mini-DIN, 6-pin plug, similar to PS/2 plugs commonly used for

computer keyboards and mice. Figure 5.4.7 illustrates the pin numbering for

configuration of the mini-DIN plug.

Page 213: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

199

Looking @

Module

Looking @

Plug

12

34

56

1 2

3 4

5 6

Figure 5.4.7 – Typical pin numbering of a mini-DIN plug.

The quarter bridge configuration used for this project requires that the following

pins be connected to the noted wires as follows:

• Pin 5 – positive voltage signals, green wire.

• Pin 4 – negative voltage signals, yellow wire.

• Pin 3 – signal sent to ADC for recording.

Figure 5.4.8 illustrates the connections used for this project.

5 (green) +Vexec

4 (yellow)

3 (orange)

-Vexec

To ADC

Pin Number Strain

Gage

Exterior Interior

Figure 5.4.8 – Pin numbering of mini-DIN plugs used for this project with typical color-coding.

Page 214: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

200

Once the filters are enabled, completion headers configured properly, and plugs attached

to strain sensors, the DBK43A hardware is ready for strain gage measurement.

The DBK65 transducer module only requires specification of the excitation

voltage for each instrument attached to the individual channels. An excitation of 10 volts

is recommended for the UniMeasure string potentiometers used in this project. Selecting

the appropriate jumper setting on the circuit board of the DBK65 module sets the

excitation voltage. Figure 5.4.9 illustrates the location of excitation jumper settings.

Front Panel

Rear Panel

Excitation

voltage

jumpers

Figure 5.4.9 – Location of the excitation voltage jumpers on the DBK 65 circuit board.

5.4.3 - Acquisition Software

The software to be used in this project is DASYLab. Section 3.5 outlines a

general procedure for configuring four strain gage channels and two transducer channels

using the DBK43A and DBK65 signal conditioning modules. The worksheet required

Page 215: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

201

for monitoring the individual strain sensors and string potentiometers is constructed in a

similar manner as described previously, adding the appropriate number of additional

channels for each individual load test. It is recommended that separate worksheets be

used for each individual load test to speed in-field acquisition and allow the user

opportunity to only use those channels necessary for acquisition during load tests.

5.4.4 - Error Correction in Readings

As noted in section 3.3, errors in strain readings can be produced by a number of

sources. However, a number of simple correction methods are available to increase the

accuracy of measurements. Directly pertinent to this project are error caused by lead

wire attenuation, thermal changes and Wheatstone bridge non-linearity.

Attenuation of signal caused by lead wire resistance dampens the electrical

change of a strained Wheatstone bridge circuit. Lead wires installed on the De Neveu

Creek Bridge have a wide array of lengths, with the maximum length of wire installed

being approximately 120.’ However, the wiring used is of relatively low resistance. For

example, a 30’ run of wire has approximately 1-ohm of resistance. If the resistance in

lead wires is assumed to vary linearly with length, the longest lead wires would then

posses approximately 4-ohms of additional resistance. Using the correction expression

noted in section 3.3, the following calculation provides an adjustment coefficient for

strain sensor readings.

( ) ( )350 4 1.01350

g Lactual read read read

g

R R

Rε ε ε ε

+ += = =

(5.4.1)

Page 216: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

202

It can be seen that for the longest lead wire lengths used in this project, only a 1%

error is introduced in the readings. For this reason, lead wire attenuation may be omitted

from analysis.

Thermal changes of the structure, electrical wiring and strain gage itself may also

effect strain measurements. However, it is assumed that straining of the structure due to

thermal effects will not alter measurements taking during load testing as the recording

method is designed to be implemented during a short-term live loading only.

Additionally, care has been taken to avoid any heating of the lead wires during

acquisition. Finally, excitation voltage of the strain gages contained in the strain sensors

has been selected to eliminate any local heating of the sensor. Further detail is provided

in section 3.3.

Wheatstone bridge non-linearity can also effect strain measurements. However,

significantly large strain levels must be experienced for the non-linear nature of the

bridge to effect measurement. It is expected that during this testing, strains caused by the

test vehicles will not be large enough to warrant consideration.

Page 217: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

203

Chapter 6 – Summary and Conclusions

6.0 – Summary

A great deal of information is presented within this document, covering a review

of prior monitoring efforts; data acquisition and strain measurement; the development

and validation of a portable strain sensor; and a proposed load test.

The review of literature presents an overview of recommended monitoring

guidelines, an introduction to the De Neveu Creek IBRC Bridge and results from an

initial live load test conducted in the spring of 2004. Four additional monitoring projects

were also presented, illustrating different methods used to monitor structural response to

live load, and also methods available to analyze data collected during testing. These

sources provide an excellent introduction to the monitoring of bridges for their behavior

under short-term live loading.

As data acquisition and strain measurement is complex, a brief introduction of

basic strain measurement was warranted. Rudimentary discussions on signal processing,

strain gage measurement, errors in measurement and transducers are presented, as well as

a tutorial outlining an application of the DASYLab data acquisition software.

The fourth section of this document presents the rationale for production of a

portable strain sensor to be used at the De Neveu Creek IBRC Bridge, research and

development resulting in the final configuration of the sensor for field implementation

Page 218: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

204

and also laboratory validation of its performance. Finite element analysis models were

also developed to verify the performance of the sensors. Finally, individual calibration

factors for the sensors were determined, with the intention that they to be applied to field

collected data, producing reliable strain data for subsequent analysis.

A proposed load test of the De Neveu Creek IBRC Bridge is also outlined, to be

conducted in the spring of 2006. The objectives of live load testing are presented,

identifying locations of monitoring instruments on the bridge. The permanent equipment

required on the bridge is presented, as well as the proposed methods in which to load the

structure for live load testing. The data acquisition system to be used for the load is

described, with detailed recommendations outlining the operation of the strain gage

modules and transducer module. Recommendations as to the use of acquisition software

and probable errors found during acquisition are also detailed.

6.1 – Conclusions

The successful monitoring projects presented indicate that live load monitoring of

bridge structures is possible and may provide valuable insight into the performance of the

bridge. Identification of electrical resistance strain gages as the preferred live load data

acquisition instrument was completed, noting that vibrating-wire types lack the rapid

response required for short-term testing. Furthermore, it was determined that high-speed

data acquisition systems must be employed to properly capture the true strain response of

the structure. Of particular importance are the conclusions noted in reviewing the

previous load test of the De Neveu Creek IBRC Bridge. Girder distribution factors

Page 219: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

205

(GDFs) based on deflection measurements were presented in section 2.3. These GDFs

are important as they provide a baseline for which to compare results from the proposed

load test. It is theorized that GDFs based on recorded strains from the proposed load test

will be similar to those values presented.

A number of conclusions were drawing in the arena of signal conditioning and

data acquisition. It was identified that high-speed acquisition systems satisfy the Nyquist

sampling theorem, as frequency response of bridge structures are typically quite low.

Use of both digital and analog filtering is warranted, as ambient noise introduced into

measurement signals creates significant uncertainty in measurements. The quarter bridge

Wheatstone bridge configuration was identified as the most accurate and cost-effective

configuration for measurement in this project. Also, shunt calibration was identified as

an adequate method of calibration, ensuring accurate measurements with quarter bridge

strain gages. Common sources of error in measurements were identified and determined

to have negligible effects on measurements proposed.

During development of the portable strain sensor, it was concluded that use of

high-modulus materials is inappropriate as a base material, as the imbalance of relative

stiffness between the sensor and measured structural component can create inaccurate

strain readings in the portable strain sensor configuration proposed. Additionally, the use

of epoxy or other adhesives as an attachment method was determined to be inappropriate

since commercially available adhesives tend to stretch and hinder strain transmission into

Page 220: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

206

the sensor material. Thus, a mechanically anchored, low-modulus material was selected

for the portable strain sensors.

Determination of tensile and compressive calibration factors for each individual

sensor was also performed. It was concluded that individual factors are appropriate given

the predominantly linear behavior of the strain sensors under axial tension and

compression loading. Correction factors for field acquired strains are presented in

section 4.7.3.

6.2 – Recommendations for Future Research

Most notably, the completion of the proposed load test described in section 5 will

produce significant insight into behavior of the De Neveu Creek IBRC Bridge under

loading. Distribution of load to girders, distribution of wheel loads within the FRP-

reinforced deck, and determination of the degree of composite behavior of the bridge are

important topics regarding the structural performance of this bridge type and must be

evaluated. Additionally, completion of a successful load test would validate the in-field

performance of the portable strain sensors developed for this project.

Determination as to what an “appropriately low” stiffness for strain sensor carrier

material noted in section 4.2 is could be performed in a future analysis. Insight into the

behavior of a low-modulus sensor material when attempting to monitor strain on

concrete, steel or other higher modulus substrate is warranted.

Page 221: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

207

While determined appropriate for calibration of the sensors used in this project,

the constant-moment test frame used occupies a large amount of space. Assembly of the

frame also requires a significant investment of time and labor. Exploration into

alternative calibration systems that are more compact and rapidly assembled is warranted

for future construction of sensors.

The modeling used to validate the performance of the constant-moment also has

room for improvement. The beam model constructed was, for all intensive purposes, a

detailed verification of beam theory. Further modeling of the beam used in the laboratory

should incorporate the bolt holes drilled for anchorage of the strain sensors. the use of

these holes in a beam model would provide further detail as to the strain field produced in

the beam model, yielding a model that more closely models the behavior of the test beam.

While commonly accepted material properties values were used for all materials

documented in this project, validation testing of the Nylon material to determine the

actual material properties would be beneficial. Results from this testing would allow for

more detailed finite element modeling of the sensors and produce a greater degree of

confidence in results. However, better modeling of the boundary conditions could be

done as well. For instance, the use of contact elements in the finite element modeling

could increase the accuracy of the models.

Additionally, a model of the constant-moment test frame could be constructed to

provide insight into the total system behavior. This could produce valuable information

Page 222: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

208

as to validation of the behavior of both the test beam and the strain sensors and also

eliminate the occurrence of separate models, as was performed herein.

Investigation into other mechanical anchorage systems is also warranted. While

appropriate, the threaded-bolt system proposed within in this document requires a

significant amount of labor in the field and exploration of alternative systems would be

beneficial.

Further analysis into the behavior of the sensor model is also warranted. As noted

in section 4.6, the behavior of the sensor when under load involves multiple types of

response, which are theorized to interact with each other. Investigation into the behavior

between the washers used in the anchorage system and the Nylon sensors material and

the effect that bending has on the sensor is warranted. Also, analysis pertaining to the

effect that the significantly large sides of the sensor relative to the very thin “tub” of the

sensor bottom could provide additional benefit. Finally, detailed investigation into the

role that the strain relief notch (Figure 4.6.16) plays in sensor behavior would be

reasonable. The degree to which these noted scenarios affect the sensor’s performance is

unknown and thus justifiable.

Although it has been determined to be satisfactory for this project and used

throughout, the IOTech DaqBook data acquisition system with DBK43A and DBK65

modules presented many difficulties. Particularly, the DBK43A modules contain a high

degree of customizability, requiring the user to posses a detailed understanding of

Page 223: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

209

electrical equipment and programming experience. Additionally, the DBK43A modules

purchased for this project did not perform as expected, requiring a great deal of

troubleshooting to resolve hardware issues. For example, voltage regulation of specific

channels the DBK43A modules were not selectable by the user and fixed at inappropriate

levels. Overall, the DaqBook system is a powerful data acquisition tool and use of the

system can provide valuable information, however, a detailed understanding of the

equipment and signal processing is required.

Page 224: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

210

References

AASHTO. (1996). "LRFD Standard Specifications." American Association of State

Highway and Transportation Officials.

AASHTO. (1998). "LRFD Standard Specifications." American Association of State

Highway and Transportation Officials.

AISC. (2001). LRFD Manual of Steel Construction - 3rd Edition, American Institute of

Steel Construction, Chicago, IL.

ANSYS. (2005). "ANSYS University Intermediate, Release 10.0." ANSYS, Inc.,

Canonsburg, PA.

ASCE. (2005). "Report Card for America's Infrastructure." American Society of Civil

Engineers, Reston, VA.

Bank, L. C. (2005). "Personal communication via Email." Milwaukee, WI, Preliminary

Report on Load Testing of WI B-20148/149.

Bank, L. C., and Xi, Z. (1995). "Punching Shear Behavior of Pultruded FRP Grating

Reinforced Concrete Slabs." Non-metallic (FRP) Reinforcement for Concrete

Structures, 361-367.

Bank, L. C., Xi, Z., and Munley, E. "Performance of Doubly-Reinforced Pultruded

Grading/Concrete Slabs." Advanced Composite Materials in Bridges and

Structures, 351-360.

Bank, L. C., Xi, Z., and Munley, E. "Tests of Full-Size Pultruded FRP Grating

Reinforced Concrete Bridge Decks." Materials: Performance and Prevention of

Deficiencies and Failures, Atlanta, Georgia, 618-631.

BDI. (2005). Bridge Diagnostics Incorporated.

Bridge Diagnostics Inc. (2002). "Load Test and Rating Report - Fairground Road Bridge,

Greene County, Ohio."

Conachen, M. J. (2005). "Modular 3-D FRP Reinforcing System for a Bridge Deck in

Fond du Lac, Wisconsin," University of Wisconsin-Madison, Madison,

Wisconsin.

DASYLab. (2004). "DASYLab 8.0." DASYTech USA.

Page 225: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

211

Dieter, D. A. (2002). "Experimental and Analytical Study of Concrete Bridge Decks

Constructed with FRP Stay-in-place Forms and FRP Grid Reinforcing,"

University of Wisconsin-Madison, Madison, WI.

Dietsche, J. S. (2002). "Development of Material Specifications for FRP Structural

Elements for the Reinforcing of a Concrete Bridge Deck," University of

Wisconsin-Madison, Madison, WI.

Farhey, D. N. (2005). "Bridge Instrumentation and Monitoring for Structural

Diagnostics." Structural Health Monitoring, 4(4), 301-318.

FHWA. (1996). "Implementation Program on High Performance Concrete - Guidelines

for Instrumentation of Bridges: FHWA-SA-96-075." Federal Highway

Administration.

FHWA. (2005). "IBRC Program Information,

http://ibrc.fhwa.dot.gov/know/program.cfm." FHWA, ed.

Fyfe. (2005). "Tyfo UC Composite Laminate Strip System." Fyfe Company, L.L.C., San

Diego, CA.

Gere, J. M. (2001). Mechanics of Materials, 5th Edition, Brooks Cole.

Grace, N. F., Navarre, F. C., Nacey, R. B., Bonus, W., and Collavino, L. (2002). "Design-

Construction of Bridge Street Bridge - First CFRP Bridge in the United States."

PCI Journal, 47(5), 20-35.

Grace, N. F., Roller, J. J., Nacey, R. B., Navarre, F. C., and Bonus, W. (2005). "Truck

Load Distribution Behavior of the Bridge St. Bridge, Southfield, Michigan." PCI

Journal, 50(2), 77-89.

IOTech. (2004). Signal Conditioning & PC-Based Data Acquisition Handbook, 3rd Ed.,

IO Tech, Inc., Cleveland, OH.

IOTech. (2005a). "DaqBook/2000 Series User's Manual." p/n 1121-0901, revision 2.0,

IOTech, Inc., Cleveland, OH.

IOTech. (2005b). "DBK Options, User's Manual Part 2 of 2." p/n 457-0912, revision 5.0,

IOTech, Inc., Cleveland, OH.

Jacobson, D. A. (2004). "Experimental and Analytical Study of Fiber Reinforced

Polymer (FRP) Grid-Reinforced Concrete Bridge Decking," University of

Wisconsin-Madison, Madison, WI.

Page 226: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

212

Lenett, M. S., Hunt, V. J., Helmicki, A. J., and Aktan, A. E. (2001). "Instrumentation,

testing, and monitoring of a newly constructed reinforced concrete deck-on-steel

girder bridge- Phase III." UC-CII-01/1, University of Cincinnati, Cincinnati,

Ohio.

Micro-Measurements. (2004a). "Instruction Bulletin B-147-4." Application of M-Coat J

Protective Coating, Vishay Micro-Measurements, Revised March, 1996.

Micro-Measurements. (2004b). "Tech Note TN-507-1." Errors Due to Wheatstone Bridge

Nonlinearity, Vishay Micro-Measurements, Revised January 20, 2005.

Micro-Measurements. (2004c). "Tech Note TN-514." Shunt Calibration of Strain Gage

Instruments, Vishay Micro-Measurements, Revised December 5, 2004.

Micro-Measurements. (2005a). "Errors due to Leadwire Resistance." Vishay

Intertechnology, Inc.

Micro-Measurements. (2005b). "Tech Note TN-502." Optimizing Strain Gage Excitation

Levels, Vishay Micro-Measurements, Revised January 14, 2005.

Micro-Measurements. (2005c). "Tech Note TT-612." The Three-Wire Quarter Bridge

Circuit, Vishay Micro-Measurements, Revised January 20, 2005.

Mindess, S., Young, J. F., and Darwin, D. (2003). Concrete, 2nd Ed., Prentice Hall,

Pearson Education, Inc., Upper Saddle River, New Jersey.

MnDOT. (2005). "Pavement Sensors." Minnesota Department of Transportation.

National_Instruments. (2005). "AC and DC Coupling." National Instruments

Corporation.

NCHRP. (1998). "Research Results Digest #234 - Manual for Bridge Rating Through

Load Testing." National Cooperative Highway Research Program.

Omega. (2000). "Data Acquisition." Transactions in Measurement and Control, Putnam

Publishing Company and OMEGA Press, L.L.C.

Powers. (2005). "Specification and Design Manual - PowerFast+ Epoxy Adhesive

Anchoring System." Powers Fasteners, New Rochelle, NY.

Rizzoni, G. (2003). Principles and Applications of Electrical Engineering, 4th Ed.,

McGraw-Hill, New York.

ROMUS. (2005). "Strain Sensor Shop Drawings." ROMUS Incorporated, Milwaukee,

WI.

Page 227: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

213

Schultz, J. L. (2005). "Personal Communication via Telephone." Milwaukee, WI, Bridge

Diagnostics, Inc. Initical Equipment Estimate.

Turner, M. K. (2003). "In-situ Evaluation of Demonstration GFRP Bridge Deck System

Installed on South Carolina Route S655," University of South Carolina.

UniMeasure. (2005). "PA Series Position Transducer Specifications." UniMeasure, Inc.,

Corvallis, OR.

WiDOT. (2003). "Construction Plans - City of Fond du Lac Bypass, USH 151, State

Project Number 1420-05-71." Structure B-20-148, State of Wisconsin Department

of Transportation.

Wikipedia.org. "Wheatstone Bridge." www.Wikipedia.org.

Ziemian, R. D., and McGuire, W. (2002). "Mastan 2, Version 2.0." John Wiley & Sons,

MATLAB-based Matrix Structural Analysis Program.

Page 228: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

Appendix A – Partial Plan Set of the De Neveu Creek IBRC Bridge

214

Page 229: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

262

215

Page 230: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

263

216

Page 231: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

264

217

Page 232: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

265

218

Page 233: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

266

219

Page 234: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

267

220

Page 235: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

268

221

Page 236: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

269

222

Page 237: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

270

223

Page 238: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

271

224

Page 239: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

272

225

Page 240: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

273

226

Page 241: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

274

227

Page 242: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

275

228

Page 243: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

276

229

Page 244: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

277

230

Page 245: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

278

231

Page 246: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

279

232

Page 247: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

280

233

Page 248: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

281

234

Page 249: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

Appendix B – Compressive and Tensile Shunt Calibration Calculations

235

Page 250: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

Calculation of Voltage Drop Across Each Resistor:

Left Arm - IL

1000Vexec

Ra Rc+( )float 6, 3.57143→:= Right Arm - IR

1000Vexec

Rb Rd+( )float 6, 3.56633→:=

VA

IL Ra⋅

1000float 6, 1.25000→:= VB

IR Rb⋅

1000float 6, 1.25178→:=

VC

IL Rc⋅

1000float 6, 1.25000→:= VD

IR Rd⋅

1000float 6, 1.24822→:=

Initial Imbalance within the Circuit:

VAB 1000 VC VD−( )⋅ float 3, 1.78→:= mV

Also, Output Voltage can be Calculated Directly via:

VO 1000Rc

Ra Rc+

Rd

Rb Rd+−

Vexec⋅ float 3, 1.78→:= mV

mAITOTAL

1000Vexec

REQ

float 6, 7.13776→:=

ohms REQ1

1

Ra Rc+( )1

Rb Rd+( )+

float 8, 350.24982→:=

Tensile Configuration of Shunt ResistorEquivalent and Total Response of Entire Circuit:

Rshunt 64900:=Rb Rgage:=

Rd 350:=Rc 350:=

Rgage 351:=Ra 350:=

volts Vexec 2.500:=

Input Values (all resistances are in ohms):

Quarter Bridge Wheatstone Bridge

236

Page 251: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

- Note that the left arm's voltages do not change as the voltage drop

across it remains constant at the prescribed voltage. However on the

right arm...

VA Va−( ) VC Vc−( )+ 0→

Vc

IlT Rc⋅

1000float 6, 1.25000→:=

Va

IlT Ra⋅

1000float 6, 1.25000→:=

IlT

1000Vexec

Ra Rc+( )float 6, 3.57143→:=Left Arm -

Calculation of Voltage Drop Across Each Resistor including Shunted Arm:

- This positive value makes sense as the decreased

equivalent resistance of the shunted circuit requires

that more current (recall V=IR) to balance the circuit

at the prescribed voltage.

µA, or microamps 1000 ITOTT ITOTAL−( ) float 3, 9.58→

Difference between Initial and Shunted Current Values:

mAITOTT

1000Vexec

REQT

float 6, 7.14734→:=

ohms REQT1

1

Ra Rc+( )1

Rb RCALT+( )+

float 8, 349.78051→:=

Shunted Resistance Values:

RCALT1

1

Rd

1

Rshunt

+

float 8, 348.12261→:=

Equivalent Resistance for the Shunted Arm (shunting of Rd, not the actual strain gage, simulates a

tensile strain):

SHUNT CALIBRATION OF A QUARTER BRIDGE CIRCUIT - TENSILE SIMULATION

237

Page 252: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

...and as noted previously this positive change in voltage indicates a tensile

strain within the Wheatstone bridge.

mV VabT VAB− float 3, 3.36→

Difference between Initial and Shunted Values:

VoT 1000Vexec

Rc

Ra Rc+

RCALT

Rb RCALT+−

⋅ float 3, 5.14→:=

Which can also be Calculated Directly via,

mV VabT 1000 Vc VCALT−( )⋅ float 3, 5.14→:=

Shunted Imbalance within the Circuit:

VD 1.24822→VCALT

IrT RCALT⋅

1000float 6, 1.24486→:=

VB 1.25178→VbT

IrT Rb⋅

1000float 6, 1.25514→:=

SHUNTED VALUE -ORIGINAL VALUE -

IrT

1000Vexec

Rb RCALT+float 6, 3.57591→:=Right Arm -

...voltages do respond to the resistance change

across the shunted arm...

238

Page 253: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

mV VabC 1000 Vc VdC−( )⋅ float 3, 1.59−→:=

Shunted Imbalance within the Circuit:

VD 1.24822→VdC

IrC Rd⋅

1000float 6, 1.25159→:=

VB 1.25178→Vcal

IrC Rcal⋅

1000float 6, 1.24841→:=

SHUNTED VALUE -ORIGINAL VALUE -

IrC

1000Vexec

Rcal Rd+float 6, 3.57597→:=Right Arm -

Calculation of voltages across shunted arm (recall that the non-shunted arm remains static):

- Note that both the compressive and tensile shunt

processes yield the same current flow through the

circuit!

mAITOTT 7.14734→

mAITOTC

1000Vexec

REQC

float 6, 7.14739→:=

ohms REQC1

1

Ra Rc+( )1

Rcal Rd+( )+

float 8, 349.77783→:=

Shunted Resistance Values:

Rcal1

1

Rb

1

Rshunt

+

float 8, 349.11189→:=

Equivalent Resistance for the Shunted Arm (direct shunting of the strain gage, Rb, simulates a

decrease in resistance and thus, a compressive strain):

SHUNT CALIBRATION OF A QUARTER BRIDGE CIRCUIT - COMPRESSIVE SIMULATION

239

Page 254: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

Which can also be Calculated Directly,

VoC 1000Vexec

Rc

Ra Rc+

Rd

Rd Rcal+−

⋅ float 3, 1.59−→:=

Difference between Initial and Shunted Values:

VabC VAB− float 3, 3.37−→ mV

- The negative change in voltage simulates a compressive

strain within the Wheatstone bridge!

-Also, note that the magnitude of both tensile and compressive voltage changes

are approximately equal; the only real difference is their corresponding sign.

240

Page 255: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

Appendix C – Final Configuration of the Portable Strain Sensor

241

Page 256: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

VENDOR:

SHEET 1 OF 3

SCALE: 2:1

0.04

PART NO:

MATERIAL:

DESC:

FILE DATE:

WRITTEN PERMISSION OF ROMUS INC IS PROHIBITED.

5 4 3 2 1

DRAWN BY:WEIGHT:

Project:TITLE:

ROMUS INCROMUS INCROMUS INCROMUS INC

JPS

Wednesday, September 14, 2005 7:10:08 PMstrain-gauge-bisc

STAIN GAUGE BLOCK ASMSTGB

MU FOLEYClient:

FILE NAME:

THE INFORMATION CONTAINED IN THIS DRAWING IS THE SOLE PROPERTY OFROMUS INC. ANY REPRODUCTION IN PART OR AS A WHOLE WITHOUT THE

NO:

242

Page 257: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

0.110

SECTION B-B

0.010

0.200

0.250-0.010

+0.010

VENDOR:

SHEET 2 OF 3

SCALE: 2:1

0.04 NYLON 6/6 BLK

PART NO:

MATERIAL:

DESC:

FILE DATE:

WRITTEN PERMISSION OF ROMUS INC IS PROHIBITED.

5 4 3 2 1

DRAWN BY:WEIGHT:

Project:TITLE:

ROMUS INCROMUS INCROMUS INCROMUS INC

JPS

Wednesday, September 14, 2005 7:11:20 PMstrain-gauge-biscuit

STRAIN GAUGE MNTSTGB

MU FOLEYClient:

FILE NAME:

THE INFORMATION CONTAINED IN THIS DRAWING IS THE SOLE PROPERTY OFROMUS INC. ANY REPRODUCTION IN PART OR AS A WHOLE WITHOUT THE

NO:

0.250

0.500

0.520

1.250

1.5000.386

4.003.000

2 x

1.00

B B

243

Page 258: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

Appendix D – Simulated Tensile Strains used during Sensor Calibration

244

Page 259: Development and Evaluation of a Removable and Portable Strain Sensor for Short-term Live

Sensor # Sensor #Test Top T/Resistance Bottom B/Resistance T/Shunt B/Shunt

1 1 350.4 2 350.8 2563 25592 2 351.0 1 350.9 2568 25603 3 350.8 4 351.0 2566 25614 4 351.0 3 350.8 2568 25595 5 351.1 6 350.9 2568 25606 6 350.6 5 351.4 2565 25637 7 350.9 8 351.3 2567 25638 8 351.4 7 351.2 2571 25629 9 350.8 11 351.0 2566 256110 11 350.7 9 351.0 2565 256111 10 350.6 12 350.7 2565 255812 12 350.7 10 350.5 2565 255713 13 350.8 14 350.7 2566 255814 14 350.8 13 350.5 2566 255715 15 350.8 16 351.2 2566 256216 16 351.3 15 350.8 2570 255917 17 351.1 18 351.2 2568 256218 18 351.3 17 350.8 2570 255919 19 351.0 20 351.1 2568 256120 20 350.5 19 351.1 2564 256121 21 350.6 22 351.3 2565 256322 22 351.4 21 350.7 2571 255823 23 351.0 24 350.6 2568 255824 24 350.8 23 351.2 2566 256225 25 350.6 26 350.5 2565 255726 26 350.6 25 350.6 2565 255827 27 351.4 28 350.9 2571 256028 28 350.9 27 351.2 2567 256229 29 350.6 30 351.3 2565 256330 30 351.3 29 350.7 2570 255831 31 350.7 32 350.6 2565 255832 32 350.8 31 350.8 2566 255933 33 350.7 34 351.1 2565 256134 34 351.2 33 350.6 2569 255835 35 350.8 - - 2566 -36 - - 35 350.7 - 2558

CALIBRATION TESTS - SHUNT VALUES

Sim. uStrain

245