33:. ‘

u.

 

. n .i ...“u..

 

 

 

 

 

 

 

 

 

 

 

a: .2; an...
L... flung”;

hams: 3:.

mm,

was!

ma
‘

.rnu: .5
WWW? :fi:mfis. a? _
s w: .1 ‘

A...

«mfi if.

3.: .5.
m3”. . .
H w W».

1..

LI.
fr)... i

a“

. 5
:34!

ca. {fur

......a... fin.»
, M ‘ . v. .3.

IND. O

 

 

40

llolv‘

 

”2. ,
1m 5 _
Wyglfgég,

This is to certify that the
thesis entitled

ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY BASED
SENSORS FOR NDE OF CERF/CONCRETE BOND IN BEAMS

presented by
Sangdo Hong

has been accepted towards fulfillment
of the requirements for the

MS. degree in Civil Enfleering

flaw

Major Professor’s Signature
”FM/ 0? 7, 02/003
5”

Date

MSU is an Affirmative Action/Equal Opportunity Institution

__.__———
v v

ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY BASED
SENSORS FOR NDE OF CFRP/CONCRETE BOND IN BEAMS

By

Sangdo Hong

A THESIS
Submitted to,
Michigan State University

in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Department of Civil and Environmental Engineering

2003

ABSTRACT

ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY BASED SENSORS FOR
NDE OF CFRP/CONCRETE BOND IN BEAMS

By
Sangdo Hong

Electrochemical impedance spectroscopy (EIS) based sensor technology is used for
the NDE of the bond between external CFRP reinforcement and concrete in beams.
Copper tape on the surface of the CFRP sheet, stainless steel ' wire embedded in the
concrete, and reinforcing bars were used as the sensing elements. laboratory experiments
were designed to test the capability of the sensors to detect the debonding of the CFRP
from the concrete and to study the effect of short-term (humidity and temperature
fluctuations and chloride content) and long-term (freeze-thaw and wet-dry exposure and
rebar corrosion) environmental conditions on the measurements. The CFRP sheet was
debonded from the concrete and impedance measurements were taken between various
pairs of electrodes at various interfacial crack lengths. The dependence of the impedance
spectra, and of the parameters obtained from equivalent circuit analysis, on the interfacial
crack length was studied. Capacitance parameters in the equivalent circuit were used to
assess the global state of the bond between CFRP sheets and concrete. Impedance
measurements taken between embedded wire sensors were used to detect the location of

debonded regions.

To my parents, wife and daughters

iii

ACKNOWLEDGMENTS

This research was funded by Michigan State Department of Transportation. This
contribution is gratefully acknowledged.

I would like to express my sincere gratitude to Dr. R. Harichandran whose
supervision and guidance were heightened not only by his friendly nature, but also by his
genuine concern and understanding. I also would like to thank Dr. A. Al-Ostaz and Dr. G.
Davis for their collaboration and the MS guidance committee members, Dr. P.
Soroushian and Dr. A. Vanna for their support. I would like to express my great

appreciation to Mr. S. Ravanbakhsh for his effort and help throughout this research.

iv

TABLE OF CONTENS

 

 

 

 

LIST OF TABLES -- vii
LIST OF FIGURES viii
Chapter 1 Introduction _ 1
1.1 BACKGROUND ........................................................................................................... 2
1.2 ELECTROCHEMICAL IMPEDANCE SPECTROSCOPY — BACKGROUND ........................... 4
1.2.1 Impedance ........................................................................................................ 4

1.2.2 Electrochemical impedance spectroscopy and equivalent circuit
analysis ............................................................................................................... 6

1.3 USE OF EIS FOR DETECTION OF INTERFACIAL DEGRADATION—A
LITERATURE REVIEW ................................................................................................. 8
1.4 HYPOTHESIS AND OBJECTIVES ................................................................................ 16
Chapter 2 Experimental Design 17
2.1 SPECIMEN PREPARATION AND SENSOR CONFIGURATION ........................................ 18
2.1.1 Specimen sizes and preparation ..................................................................... 18
2.1.2 Sensor configuration and installation ............................................................. 21
2.1.3 Sensor combinations for impedance measurements ....................................... 22
2.2 DETECTION OF FRP DEBONDING ............................................................................ 24
2.2.1 Two foot long specimens ............................................................................... 24
2.2.2 Large beam specimen ..................................................................................... 25

2.3 ASSESSMENT OF THE SENSITIVITY OF SENSOR MEASUREMENTS T0
ENVIRONMENTAL EFFECTS ....................................................................................... 26
2.3.1 Humidity ......................................................................................................... 26
2.3.2 Temperature .................................................................................................... 26
2.3.3 Chloride content ............................................................................................. 27
2.3.4 Freeze — thaw .................................................................................................. 29
2.3.5 Wetting and drying ......................................................................................... 29
2.3.6 Corrosion of reinforcing bar ........................................................................... 30
2.4 ELECTRODE SIZE EFFECTS ....................................................................................... 33
Chapter 3 Experimental Results, Equivalent Circuit Analysis, and Discussion ....... 34
3.1 EQUIVALENT CIRCUIT .............................................................................................. 34
3.2 CFRP DEBONDING .................................................................................................. 39
3.2.1 Results for 2-foot long specimens .................................................................. 39
3.2.1.1 Comparison of raw impedance spectra ................................................... 43
3.2.1.2 Results from equivalent circuit analysis ................................................. 43
3.2.2 Results for 8-foot long specimen .................................................................... 48
3.2.2.1 Comparison of raw impedance spectra ................................................... 48
3.2.2.2 Results from equivalent circuit analysis ................................................. 49
3.2.3 Detecting the location of debonded regions ................................................... 54
3.2.3.1 Comparison of raw impedance Spectra ................................................... 54
3.2.3.2 Results from equivalent circuit analysis ................................................. 58

V

3.2.4 Empirical Relationships ................................................................................. 6O

 

3.2.4.1 Raw impedance spectra .......................................................................... 60
3.2.4.2 Equivalent circuit analysis ...................................................................... 63

3.2.5 Summary and Discussion ............................................................................... 65
3.3 SENSITIVITY OF SENSOR MEASUREMENTS To ENVIRONMENTAL EFFECTS .............. 67
3.3.1 Humidity ......................................................................................................... 67
3.3.1.1 Comparison of raw impedance spectra ................................................... 67
3.3.1.2 Results from equivalent circuit analysis ................................................. 68
3.3.1.3 Summary and Discussion ....................................................................... 68

3.3.2 Temperature .................................................................................................... 73
3.3.2.1 Comparison of raw impedance spectra ................................................... 73
3.3.2.2 Results from equivalent circuit analysis ................................................. 77
3.3.2.3 Summary and Discussion ........................................ . .............................. 77

3.3.3 Chloride content ............................................................................................. 80
3.1.3.1 Comparison of raw impedance spectra ................................................... 80
3.3.3.2 Results from equivalent circuit analysis ................................................. 86
3.3.3.3 Summary and Discussion ....................................................................... 86

3.3.4 Freeze — thaw .................................................................................................. 87
3.3.4.1 Comparison of raw impedance spectra ................................................... 87
3.3.4.2 Results from equivalent circuit analysis ................................................. 87
3.3.4.3 Summary and Discussion ....................................................................... 87

3.3.5 Wetting and drying ......................................................................................... 91
3.3.5.1 Comparison of raw impedance spectra ................................................... 91
3.3.5.2 Results from equivalent circuit analysis ................................................. 9]
3.3.5.3 Summary and Discussion ....................................................................... 91

3.3.6 Corrosion of reinforcing bar ........................................................................... 95
3.3.6.1 Comparison of raw impedance spectra ................................................... 95
3.3.6.2 Results from equivalent circuit analysis ................................................. 96
3.3.6.3 Summary and Discussion ....................................................................... 96

3.4 EFFECT OF ELECTRODE SIZE .................................................................................. 100
Chapter 4 Summary, Conclusions, and recommendations for Future Work ......... 102
4.1 SUMMARY ............................................................................................................. 102
4.1.1 Detection of CFRP Debonding ..................................................................... 103
4.1.2 Environmental effects ................................................................................... 105
4.2 CONCLUSIONS ....................................................................................................... 106
4.3 RECOMMENDATIONS FOR FUTURE WORK ............................................................. 108
REFERENCES 109
Appendix—A: Data Collection and Equivalent Circuit Analysis Procedure .......... 111
A-l SOFTWARE PROCEDURE FOR DATA COLLECTION ................................................. 111
A-2 EQUIVALENT CIRCUIT ANALYSIS PROCEDURE ...................................................... 112

vi

LIST OF TABLES

Table 2-1 Number of sensors and their orientation on the specimens .............................. 23

Table 3-1 Parameters of the curve fitted to measured impedance in the high-

frequency range ............................................................................................................ 61
Table 3-2 Parameters of the curve fitted to (C1+C2+C3) from equivalent circuit

analysis ......................................................................................................................... 63
Table 3-3 Specimen weights before and after freeze-thaw testing ................................... 88
Table 3-4 Specimen weights before and after wet-dry testing .......................................... 92
Table 3-5 Specimen weights before and after corrosion testing ....................................... 97

vii

LIST OF FIGURES

Figure 1-1 Typical electrochemical cell ............................................................................ 10

Figure 1-2 Model of the charge transfer process when the rebar and CFRP sheet (via
external sensor) are used as electrodes ......................................................................... 10

Figure 1-3 Equivalent circuit configurations — (a) simple circuit used for a corrosion
reaction (Macdonald 1987) (b) circuit used to study an adhesive bond (Davis et al.
1999) ............................................................................................................................ 11

Figure 14 Typical impedance spectra taken from the wedge test (Source: Davis et al.
1999) ............................................................................................................................ 14

Figure 1-5 Resistance parameters from the equivalent circuit analysis as a function of
time for the wedge tests under wet condition (Source: Davis et al. 1999) ................... 14

Figure 1-6 Variation of capacitance parameter from the equivalent circuit analysis as
a function of bonded area for wedge tests under —(a) dry condition (b) wet

condition (Source: Davis et al. 1999) ........................................................................... 15
Figure 2-1 Small concrete beam configuration and sensor locations ............................... 19
Figure 2-2 Medium concrete beam configuration and sensor locations ........................... 19
Figure 2-3 Large concrete beam configuration and sensor locations ............................... 20
Figure 2-4 Cross-section of large beam .................................................... 23
Figure 2-5 Experimental design flow chart for humidity, temperature and chloride

content effects .............................................................................................................. 28
Figure 2-6 Experimental setup for accelerated corrosion ................................................. 32
Figure 2-7 Concrete slab configuration and sensor locations ........................................... 33
Figure 3-1 Equivalent circuit used to study CFRP debonding .......................................... 37

Figure 3-2 Fitted Bode magnitude and phase angle plots using the two equivalent
circuits. Dots are the actual measured impedance from the rebar to an external
sensor. Solid lines are the theoretical impedance using the equivalent circuit in
Figure 3-1. The dotted lines are the theoretical impedance using the equivalent
circuit in Figure 1-2 (b). ............................................................................................... 37

Figure 3-3 Fitting with the equivalent circuits. Dots are the actual measured

impedances between two internal sensors. Solid lines are the theoretical
impedances using the equivalent circuit in Figure 3-1. The dotted lines are the

viii

theoretical impedances using the equivalent circuit in Figure 1-2 (b). — (a) Bode
magnitude and phase angle plot; (b) Nyquist plot. ...................................................... 38

Figure 3-4 Bode magnitude plot of typical impedance spectra measured from the
rebar to copper tape for wedge test in controlled environment (legend gives the
crack length in inches) ................................................................................................. 40

Figure 3-5 Bode magnitude plot of typical impedance spectra measured from the
rebar to copper tape for wedge test in ambient environment (legend gives the
crack length in inches) ................................................................................................. 40

Figure 3-6 Nyquist plot of typical impedance spectra measured from the rebar to
copper tape for wedge test in controlled environment (legend gives the crack
length in inches) ........................................................................................................... 41

Figure 3-7 Enlargement of Figure 3-6, showing variation of measure impedance in
the low to mid- frequency range (legend gives the crack length in inches) ................. 41

Figure 3-8 Nyquist plot of impedance spectra measured from the rebar to copper tape
for wedge test in ambient environment (legend gives the crack length in inches) ...... 42

Figure 3-9 Variation of typical measured impedance from the rebar to copper tape
for wedge test at 10,000 Hz .......................................................................................... 44

Figure 3-10 Variation of typical measured impedance from the rebar to copper tape
for wedge test at 100 Hz ............................................................................................... 44

Figure 3-11 Variation of typical measured impedance from the rebar to copper tape
for wedge test at 1 Hz ................................................................................................... 45

Figure 3-12 Variation of the CPE parameter for wedge test in the controlled
environment for measurements from the rebar to external sensors ............................. 45

Figure 3-13 Variation of the CPE parameter for wedge test in ambient condition for
measurements from the rebar to external sensors (Note: the CPE parameter for
spec. 2 is measured in controlled environment) ........................................................... 46

Figure 3-14 Variation of the Cl parameter from equivalent circuit analysis for wedge
test from the rebar to external sensors .......................................................................... 46

Figure 3-15 Variation of the C2+C3 parameter from equivalent circuit analysis for
wedge test from the rebar to external sensors .............................................................. 47

Figure 3-16 Variation of the C1+C2+C3 parameter from equivalent circuit analysis
for wedge test from the rebar to external sensors ......................................................... 47

ix

Figure 3-17 Bode magnitude plot of typical impedance spectra measured from the
rebar to copper tape for wedge test of the large beam (legend gives the crack
length in inches) ........................................................................................................... 50

Figure 3-18 Enlargement of Figure 3-17, showing variation of measure impedance in
the low-frequency range (legend gives the crack length in inches) ............................. 50

Figure 3-19 Nyquist plot of impedance spectra measured from the rebar to copper
tape for wedge test of large beam in ambient environment (legend gives the crack
length in inches) ........................................................................................................... 51

Figure 3-20 Enlargement of Figure 3-17 showing variation of measured impedance
in the high-frequency range (legend gives the crack length in inches) ........................ 51

Figure 3-21 Bode magnitude plot of typical impedance Spectra measured from the
rebar to copper tape and the stainless steel bar to cOpper tape for wedge test of the
large beam (legend gives the crack length in inches) ................................................... 52

Figure 3-22 Variation of typical measured impedance from the rebar to copper tape
for wedge test at 100,000 Hz ........................................................................................ 52

Figure 3-23 Variation of the CPE parameter from equivalent circuit analysis for the
large specimen from the rebar to external sensor ......................................................... 53

Figure 3-24 Variation of the capacitance parameters from equivalent circuit analysis
for the large specimen from the rebar to external sensor ............................................. 53

Figure 3-25 Bode magnitude plot of impedance spectra measured between pairs of
internal sensors for eight-foot specimen in ambient condition when crack tip was
between sensors W2 and W3 (legend shows the internal sensors used) ...................... 56

Figure 3-26 Bode magnitude plot of impedance Spectra measured between pairs of
internal sensors for two-foot Specimen in ambient condition when crack tip was
between sensors W3 and W4 (legend shows the internal sensors used) ...................... 56

Figure 3-27 Nyquist plot of impedance spectra measured between pairs of internal
sensors for two-foot specimen in ambient condition when crack tip was between
sensors W3 and W4 (legend shows the internal sensors used) .................................... 57

Figure 3-28 Bode magnitude plot of impedance Spectra measured between pairs of
internal sensors for eight-foot specimen in ambient condition when crack tip was
between sensors W5 and W6 (legend shows the internal sensors used) ...................... 57

Figure 3-29 Variation of the CPE parameter for pairs of internal sensors for the 8-
foot beam when crack tip was between sensors W2 and W3 (name of the bar
shows the internal sensors used) .................................................................................. 59

Figure 3-30 Variation of the CPE parameter for pairs of internal sensors for the 2-
foot long beam when crack tip was between sensors W3 and W4 (name of the bar
shows the internal sensors used) .................................................................................. 59

Figure 3-31 Fit of the measured impedance for 8-foot long beam from rebar to
external sensors at 100,000 Hz: Dots are the measured data and the solid line is
the fitted curve .............................................................................................................. 62

Figure 3-32 Typical fit of the measured impedance for two-foot long beam from
rebar to external sensors at 100,000 Hz: Dots are the measured data and the solid
line is the fitted curve. .................................................................................................. 62

Figure 3-33 Fit for (C1+C2+C3) parameters for 8-foot beam: Dots are the measured
data and the solid line is the fitted curve. ..................................................................... 64

Figure 3-34 Typical fit for (C1+C2+C3) parameters for two-foot beam: Dots are the
measured data and the solid line is the fitted curve. .................................................... 64

Figure 3-35 Bode magnitude plot of typical impedance spectra for specimens with
chloride at different humidity levels ............................................................................ 70

Figure 3-36 Bode magnitude plot of typical impedance spectra for specimens without
chloride at different humidity levels ............................................................................ 70

Figure 3-37 Variation of the CPE parameter from equivalent circuit analysis due to '
different humidity levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C-chloride, N—non-chloride, T-
temperature, H-humidity, Sn-specimen number) ......................................................... 71

Figure 3-38 Variation of the C1 parameter from equivalent circuit analysis due to
different humidity levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C~chloride, N-non-chloride, T-
temperature, H-humidity, Sn-specimen number) ......................................................... 71

Figure 3-39 Variation of the C2+C3 parameter from equivalent circuit analysis due
to different humidity levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C-chloride, N-non-chloride, T-
temperature, H-humidity, Sn-specimen number) ......................................................... 72

Figure 3-40 Variation of the C1+C2+C3 parameter from equivalent circuit analysis
due to different humidity levels for measurements from the rebar to external
sensors (legend gives the name of the specimen: C-chloride, N-non-chloride, T-
temperature, H-humidity, Sn-specimen number) ......................................................... 72

Figure 3-41 Bode magnitude plot of typical impedance spectra for specimens with
chloride at different temperature levels ........................................................................ 74

xi

Figure 3-42 Bode magnitude plot of typical impedance spectra for specimens without
chloride at different temperature levels ........................................................................ 74

Figure 3-43 Variation of typical measured impedance in the high-frequency range at
different temperature levels at 10,000 Hz (legend gives the name of the specimen:
C-chloride, N-non-chloride, Sn-specimen number) ..................................................... 75

Figure 3-44 Variation of typical measured impedance in the mid-frequency range at
different temperature levels at 100 Hz (legend gives the name of the specimen: C-
chloride, N-non-chloride, Sn-specimen number) ......................................................... 75

Figure 3-45 Variation of typical measured impedance in the low-frequency range at
different temperature levels at 1 Hz (legend gives the name of the specimen: C-
chloride, N -non-chloride, Sn-specimen number) ......................................................... 76

Figure 3-46 Variation of the CPE parameter from equivalent circuit analysis at
different temperature levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C-chloride, N-non-chloride, Sn-specimen
number) ........................................................................................................................ 78

Figure 3-47 Variation of the C1 parameter from equivalent circuit analysis at
different temperature levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C-chloride, N-non-chloride, Sn-specimen
number) ........................................................................................................................ 78

Figure 3-48 Variation of the C2+C3 parameter from equivalent circuit analysis at
different temperature levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C-chloride, N-non-chloride, Sn-specimen
number) ........................................................................................................................ 79

Figure 3-49 Variation of the C1+C2+C3 parameter from equivalent circuit analysis
at different temperature levels for measurements from the rebar to external
sensors (legend gives the name of the specimen: C-chloride, N-non-chloride, Sn-
specimen number) ........................................................................................................ 79

Figure 3-50 Bode magnitude plot of typical impedance spectra for specimens
with/without chloride at a relative humidity of 30% (legend gives the name of the
specimen: C-chloride, N-non-chloride, Sn-specimen number) .................................... 81

Figure 3-51 Bode magnitude plot of typical impedance spectra for specimens
with/without chloride at a relative humidity of 60% (legend gives the name of the
specimen: C-chlon'de, N-non-chloride, Sn-specimen number).... ................................ 81

Figure 3-52 Bode magnitude plot of typical impedance spectra for specimens

with/without chloride at a relative humidity of 90% (legend gives the name of the
specimen: C-chloride, N -non-chloride, Sn-specimen number) .................................... 82

xii

Figure 3-53 Variation of typical measured impedance in the high-frequency range at
different humidity levels at 10,000 Hz (legend gives the name of the specimen: C-
chloride, N-non-chloride, Sn-specimen number) ......................................................... 82

Figure 3-54 Variation of typical measured impedance in the mid-frequency range at
different humidity levels at 100 Hz (legend gives the name of the specimen: C-
chloride, N-non-chloride, Sn-Specimen number) ......................................................... 83

Figure 3-55 Variation of typical measured impedance in the low-frequency range at
different humidity levels at 1 Hz (legend gives the name of the specimen: C-
chloride, N-non-chloride, Sn-specimen number) ......................................................... 83

Figure 3-56 Bode magnitude plot of typical impedance spectra for specimens
with/without chloride at a temperature of GOP (legend gives the name of the
specimen: C-chloride, N-non-chloride, Sn-specimen number) .................................... 84

Figure 3-57 Bode magnitude plot of typical impedance spectra for specimens
with/without chloride at a temperature of 70°F (legend gives the name of the
specimen: C-chloride, N-non-chloride, Sn-specimen number) .................................... 84

Figure ‘3-58 Bode magnitude plot of typical impedance spectra for specimens
with/without chloride at a temperature of 100°F (legend gives the name of the
Specimen: C-chloride, N-non-chloride, Sn-specimen number) ................................... 85

Figure 3-59 Bode magnitude plot of typical impedance spectra from the specimens
with chloride for 300 cycles of freeze-thaw (legend gives the name of the
Specimen: B-before, A-after, FT-freeze-thaw, Sn-specimen number) ......................... 88

Figure 3-60 Variation of the CPE parameter from before and after 300 cycles of
freeze-thaw for measurements from the rebar to external sensors (legend gives the
name of the specimen: FT-freeze-thaw, Sn-specimen number) ................................... 89

Figure 3-61 Variation of the C1 parameter before and after 300 cycles of freeze-thaw
for measurements from the rebar to external sensors (legend gives the name of the
Specimen: FT-freeze -thaw, Sn-specimen number) ..................................................... 89

Figure 3-62 Variation of the C2+C3 parameter before and after 300 cycles of freeze-
thaw for measurements from the rebar to external sensors (legend gives the name
of the specimen: FT-freeze -thaw, Sn-specimen number) ........................................... 90

Figure 3-63 Variation of the C1+C2+C3 parameter before and after 300 cycles of
freeze-thaw for measurements from the rebar to external sensors (legend gives the
name of the specimen: FT-freeze -thaw, Sn-specimen number) .................................. 90

Figure 3-64 Bode magnitude plot of typical impedance spectra for specimens with

chloride before and after 300 wet-dry cycles (legend gives the name of the
specimen: B-before, A-after, WD-wet -dry, Sn-specimen number) ............................ 92

xiii

Figure 3-65 Variation of the CPE parameter from before and after 300 cycles of wet-
dry for measurements from the rebar to external sensors (legend gives the name of
the specimen: WD-wet -dry, Sn-specimen number) .................................................... 93

Figure 3-66 Variation of the C1 parameter before and after 300 cycles of wet-dry for
measurements from the rebar to external sensors (legend gives the name of the
specimen: WD-wet -dry, Sn-specimen number) .......................................................... 93

Figure 3-67 Variation of the C2+C3 parameter before and after 300 cycles of wet-dry
for measurements from the rebar to external sensors (legend gives the name of the
Specimen: WD-wet ~dry, Sn—specimen number) .......................................................... 94

Figure 3-68 Variation of the C1+C2+C3 parameter before and after 300 cycles of
wet-dry for measurements from the rebar to external sensors (legend gives the
name of the specimen: WD-wet -dry, Sn-Specimen number) ...................................... 94

Figure 3-69 Bode magnitude plot of typical impedance spectra from the specimens
with chloride for 21 days of corrosion on the rebar (legend gives the name of the
specimen: B-before, A-after, C-corrosion, Sn-specimen number) .............................. 97

Figure 3-70 Variation of the CPE parameter from equivalent circuit analysis for
specimens with chloride before and after 21 days of rebar corrosion for
measurements from the rebar to external sensors (legend gives the name of the
specimen: C-chloride, C-corrosion, Sn-specimen number) ......................................... 98

Figure 3-71 Variation of the C1 parameter from equivalent circuit analysis for
specimens with chloride before and after 21 days of rebar corrosion for
measurements from the rebar to external sensors (legend gives the name of the
specimen: C-chloride, C-corrosion, Sn—specimen number) ........................................ 98

Figure 3-72 Variation of the C2+C3 parameter from equivalent circuit analysis for
specimens with chloride before and after 21 days of rebar corrosion for
measurements from the rebar to external sensors (legend gives the name of the
specimen: C-chloride, C-corrosion, Sn-specimen number) ......................................... 99

Figure 3-73 Variation of the C1+C2+C3 parameter from equivalent circuit analysis
for specimens with chloride before and after 21 days of rebar corrosion for
measurements from the rebar to external sensors (legend gives the name of the
specimen: C-chloride, C-corrosion, Sn-specimen number) .................... 99

Figure 3-74 Bode magnitude plot of impedance spectra from the slab with different
sized sensors (legend gives the name of the sensor: E10034” dia., E2-0.25” dia.,
E3-0.5” dia., E4-0.59” dia., E5-0.75” dia.) ................................................................ 101

Figure A- 1 Measured impedance from rebar to external sensors and impedance of
the equivalent circuit using poor initial values for parameters: Dots are the
measured values, dotted line is the fitted impedance using initial values, and solid
line is the fitted impedance with the final values ....................................................... 114

xiv

Figure A- 2 Measured impedance from rebar to external sensors and impedance of
the equivalent circuit using good initial values for parameters: Dots are the
measured values, dotted line is the fitted impedance using initial values, and solid
line is the fitted impedance with the final value ........................................................ 114

XV

Chapter 1

Introduction

Many structures built in the past need strengthening and retrofitting to overcome
deficiencies caused by increased load demands, environmental deterioration and
structural aging. Thirty-five percent of all bridges in the US. are estimated to be
structurally deficient and require repair, strengthening, widening or replacement
(Karbhari 2000). To overcome structural deficiencies, composite materials such as fiber
reinforced polymers (FRP) are being increasingly used for strengthening and retrofitting.
Carbon-fiber reinforced polymer (CFRP) is a well-known high performance composite
material used to strengthen reinforced concrete structural components.

For flexural strengthening of beams, CFRP plate or sheets are bonded to the tension
face. This is a simple and convenient technique for strengthening and retrofitting
structures (Rahimi et al. 2001). Unlike steel plates, FRP plates do not suffer from
corrosion problems. However, the interfacial bond between FRP and concrete can
deteriorate due to environmental and load-related issues leading to debonding durability
and delamination.

Structural components strengthened or retrofitted with FRP behave as composite
components, and their strengths are calculated by taking this into account. Interfacial
bonding between the adherents plays a significant role in achieving composite behavior
and increasing strength.

Concrete structures with FRP plates can exhibit a brittle failure mode if the FRP

debonds from the concrete. Debonding of the plates and ripping of concrete are common

failure modes in concrete structures rehabilitated with FRP plates (Nguyen et al. 2001).
The ripping of concrete and debonding of the FRP are initiated due to high localized
stress concentration in the interface layer (Buyukozturk 1998).

It is therefore of interest to identify the integrity of interfacial bonding between the
concrete structure and the FRP. Detection of debonding is crucial in characterizing the
strength of the structural composite components, since composite action can only be
achieved with a strong interfacial bond. Nondestructive evaluation (NDE) techniques can
be very useful to detect the debonding between composite materials and concrete

SII'IICIUI’CS.

1.1 Background

Carbon-fiber-reinforced polymer (CFRP) is a common high-performance composite
material used to retrofit and strengthen concrete structures by bonding the CFRP to
concrete. The bond between the CFRP and concrete plays a crucial role in achieving
composite action. Therefore, monitoring the integrity of the bond between two materials
is important. There are several ways to detect or monitor the debonding between FRP and
concrete. The tap test, the acoustic emission (AE) technique, and the ultrasonic pulse
velocity (UPV) inspection method are some of the methods being promoted for this
purpose.

The tap test is the simplest method that can be carried out in the field. It is conducted
with a coin or a special hammer. The inspectors listen to the acoustic sounds generated by
tapping and qualitatively evaluates them for detecting disbonds. The main advantage of
using this technique is that the tap test does not require sophisticated or expensive

equipment. The tap test works because different acoustic sounds are generated depending

on whether the FRP is bonded to the concrete or not. The major disadvantage of using
this technique is that the method depends on subjective interpretation by the inspectors.
Also the inspector must be able to get close to the FRP/concrete surface to conduct the
test. This is often problematic for bridge beams.

Acoustic emission (AE) inspection is based on the detection of sound waves
generated by the structures that are stressed (Dai et al. 1997). When concrete structures
rehabilitated with FRP are subjected to stress, the concrete and FRP materials generate an
AB signal. AE sources include the debonding between FRP and concrete, cracking in
concrete, and plastic deformation and debonding of aggregate in concrete structures
rehabilitated with FRP (Mirmiran et al. 1999). The major advantage of using the AE
technique is that the integrity of structures can be monitored in real time, and the source
of the AE source center can be determined as well. The main disadvantage is that the
structures have to be monitored constantly. One of the main obstacles to having real time
measurements is supplying power to instruments used on civil structures in remote areas.
In addition to supplying the power, the AE technique suffers from the Kaiser effect. The
Kaiser effect is the phenomena that an AB signal is not generated until the previous
maximum load is exceeded. This Kaiser effect leads to the problem that the AE technique
cannot detect pre-existing disbonds.

The ultrasonic pulse velocity (UPV) test is an NDE method Similar to the AE
technique. AE tests are passive tests that analyze the signal generated by concrete
structures under applied loads. The UPV technique uses the characteristics of a pulse
signal that travels through the structure. The Variation of UPV is used to detect the cracks

in the structures (Mirmiran et a1. 2001). Although the UPV technique is a successful

nondestructive evaluation tool, it has one main disadvantage: the UPV measurements are
not easy to interpret. UPV is a function of the stiffness and density of components, and
concrete structures rehabilitated with FRP produce complicated ultrasonic Signals
(Olajide et al. 2000). This complication makes it difficult to isolate disbonds using the
UPV test.

Electrochemical impedance spectrOSCOpy (EIS) is a method that has been used to
study moisture penetration and debonding between two bonded composites (Davis et al.
1999). This work explores the potential of EIS to detect debonding of CFRP from

CODCI'CIC .

1.2 Electrochemical Impedance Spectroscopy - Background

The basic concepts of electrochemical impedance spectroscopy (EIS) are reviewed in
this section. The term “impedance” is a generalization of the term “resistance”. Electrical
resistance is the ability of a material to resist the flow of electrical current. Electrical
resistance is defined in terms of the ratio between voltage (V) and current (I) by Ohm’s

law.

v
R-7 (LU

Current is a measure of the flow of electrical charge. Voltage is the change in energy that
would be experienced by a charge when it travels from location A to location B.
1.2.1 Impedance

Electrochemical impedance (E1) is the resistance to current in an electrochemical
cell. E1 is generally obtained by measuring the current or voltage across a pair of

electrodes due to an applied electrical stimulus (voltage or current). Impedance is most

commonly obtained through the amplitude and phase shift of the sinusoidal response
relative to a sinusoidal input.

Electrical resistance is independent of the frequency of harmonic (sinusoidal) AC
voltage. However, electrical impedance (Z) is dependent on frequency. When the
excitation signal is harmonic, the response signal is also harmonic at the same frequency
with an amplitude and phase angle. The impedance (Z) is a complex-valued quantity and

is defined as the ratio between the excitation voltage V(t), and the response current I(t):

z - KB (1-2)
I(t)
For a harmonic AC excitation signal, V(t) may be expressed as
V (t) - Voe j‘“ - V0 (cos out + jsin wt) (1-3)

where (D is the circular natural frequency of the applied voltage (expressed in

radian/second) and V0 is the amplitude of the excitation signal. The harmonic current

response I(t) may be expressed as

I(t) - Ioej‘m'” - Io[cos((ot - ¢) + jsin(a)t - 42)] (1.4)
where 10 is the amplitude of the harmonic current and d) is the phase shift. The electrical
impedance in Eq. (1-2) is then

Z -- Zoe" = Z0(cos¢ + jsin¢) (1-5)
where 20 = V0 / Io

Impedance, Z, is generally complex-valued. The impedance of a resistor, R, is ZR =

R and is real-valued. The impedance of a capacitor, C, is purely imaginary and is Zc =
1/(ij), where j = \/-_1 and a) is the circular natural frequency in rad/sec of the applied

voltage. The constant phase element (CPE) is often used in an equivalent circuit to

represent the response of real-world electrochemical systems. Electrochemical systems
sometimes do not response ideally because some pIOperties of the electrochemical system
are not homogeneous and often there are distributed elements in addition to lumped

elements in the system. The impedance of the CPE is
2cm - A(iw)’“ (1-6)
When on = 1 (maximum value), the CPE is equivalent to a capacitor with A = 1/C (the
inverse of the capacitance), and when O = 0 (minimum value), the CPE is equivalent to a
resistor with A = R.
For linear impedance elements in series, the equivalent impedance is

Zeq-Z,+ZZ+Z3 (1-7)

For linear impedance elements in parallel, the reciprocal of the impedance is

 

——+—+— (1-8)
Z Z1 Z2 Z3

‘9

1.2.2 Electrochemical impedance spectroscopy and equivalent circuit analysis

Traditionally, electrochemical impedance spectroscopy (EIS) has been used to detect
coating deterioration and substrate corrosion, and the EIS technique is performed in an
electrolyte with counter and reference electrodes (Scully et al. 1989). The counter
electrode works as a supplier to pass sufficient current into the electrolyte. The reference
electrode maintains invariant potential in an electrochemical measurement and serves as a
reference point on the measurements of the working electrode. The working electrode is a
transducer responding to the excitation signal. Figure 1-1 shows the typical
electrochemical cell used in traditional EIS. Use of an external electrolyte is cumbersome

in the field. However, it is possible to use in-situ electrodes to measure the impedance in

the ambient condition without submerging the electrochemical cell using a modified EIS
technique (Davis et al. 2000).

Electrochemical theories have been proposed to explain the charge transfer process
that occurs in electrode/electrolyte systems (Bard et al. 1980). Figure 1-2 shows the
charge transfer process in the electrochemical system. When a metal electrode is
immersed and activated in a solution, it will react with the solution. From this reaction,
positive charges or negative charges are accumulated on the surface of the metal. An
electrical double layer exists at the surface between each electrode and the electrolyte
(pore fluid in the case of concrete). Ions in the electrolyte can only approach the electrode
up to the outer surface of the double layer. Charge transfer between ions and the electrode
involves long-range electrostatic forces, so that their interaction is essentially
independent of the chemical pTOperties of ions. Charge transfer through the electrolyte
occurs by the diffusion of ions. At low excitation frequencies, the impedance to current
flow is controlled by the rate of diffusion, while at high frequencies, the impedance is
controlled by the kinetics of the charge transfer processes at the electrode/electrolyte
interface.

In the modified EIS technique, an AC voltage is applied between the counter and the
reference/working electrode. In this method the reference and working electrodes are
combined into a single electrode. The complex impedance spectrum is then measured as a
function of frequency. Two methods of analyzing the impedance spectra are common.

1. The first method is to simply compare the raw impedance spectra from

different measurements and make inferences. The spectra can be plotted in

many different ways. The conventional presentations are the Bode magnitude

plot (log impedance magnitude vs log frequency), Bode phase angle plot
(phase angle vs log frequency), Nyquist plot (imaginary part of impedance vs
real part of impedance), real impedance plot (real parts of impedance vs log
frequency or log real parts of impedance vs log frequency), and imaginary
impedance plot (imaginary parts of impedance vs log frequency). The
magnitudes of impedance, phase angles, real impedances or imaginary
impedances are compared over the entire frequency range or over specific
frequency ranges. ‘

2. The second method is to analyze the impedance spectra by using an
equivalent circuit model. In this method, the parameters of an electrical circuit
which has a theoretical impedance similar to the measured impedance are
estimated, and spectra are compared based on the differences in the estimated
parameters. Typical equivalent circuits are shown in Figure 1-3. Figure 1-3
(a) shows a circuit used for corrosion studies (Macdonald 1987) and Figure
1-3 (b) shows the circuit used to study an adhesive bond (Davis et al. 1999).
Different equivalent circuits can be used to approximate a measured
impedance spectrum, and some experience is required to select an appropriate

circuit.

1.3 Use of EIS for Detection of Interfacial Degradation—A
Literature Review

In-situ sensors have been adapted to detect moisture ingress and coating
deterioration. Davis et al. (1999) found a strong correlation between low frequency

impedance measurements and moisture ingress into coatings. They found that the initial

coating resistance was very high, but as the coating degrades due to moisture ingress, the
coating resistance decreases in the low frequency region.

An in-situ corrosion sensor previously used in monitoring coating degradation and
substrate corrosion of metals has been used to detect moisture ingress and crack
propagation in structural adhesive bonds. This sensor technology has also been used to
inspect the integrity of composite/composite bonds. Davis et al. (1999) conducted wedge
tests on bonded glass-reinforced and graphite-reinforced composites to assess the
integrity of the bond using EIS sensor technology. The wedge was driven into the
Specimen until the two adherents separated and impedance measurements were taken.
The impedance typically increased in magnitude in the high-frequency region as the

crack propagated. Figure 1-4 Shows the results of the wedge test.

 

 

 

Voltage
source

 

 

 

 

 

 

Counter Reference Working
electrode electrode electrode

 

 

 

 

 

Electrolyte
L

 

 

Current flow

 

 

 

 

Figure 1-1 Typical electrochemical cell

  

Rebar electrode

Electrical double
layer
Concrete with

pore fluid —>
(electrolyte)

I
\ CFRP electrode

Figure 1-2 Model of the charge transfer process when the rebar and CFRP sheet (via
external sensor) are used as electrodes

10

(a)

 

——O

 

 

 

 

 

 

 

 

 

Rs
W—_____,
RP
(b)
Cc
| l
I l
CPE
Rsol
W
Rp0 _/\\/\/\_
R

cpe

Figure 1-3 Equivalent circuit configurations — (a) simple circuit used for a corrosion
reaction (Macdonald 1987) (b) circuit used to study an adhesive bond (Davis et al.
1999)

11

In the simplest method for analyzing the debonding effects, the impedance spectra
can be compared directly (Figure 1-4). However, additional information can be obtained
from the impedance spectra by using equivalent circuit analysis. Davis et al. (1999)
conducted the wedge tests under dry and wet conditions. They used the equivalent circuit
in Figure 1-3 (b), and found that the resistive components in the equivalent circuit they
used were functions of moisture content and the capacitance parameter was a function of
both moisture content and bonded area. The capacitance parameter from the equivalent
circuit analysis varied linearly with the bonded area in the dry wedge test. However, the
relationship was more complex when moisture was introduced during the wedge test. The
resistance parameters from the equivalent circuit analysis correlated with moisture level
in the electrochemical system. However, the resistance parameters did not show any
correlation with bonded area. Figures 1-5 and 1-6 Show the correlations between the
parameters from the equivalent circuit analysis and the bonded area.

Davis et al. (1999) proposed the capacitance of an adhesive was influenced by the
dielectric constant of the adhesive, the thickness of the adhesive and the bonded area. The

capacitance may be expressed as

 

C = 0 (19)

where C is the capacitance of the coating
80 is the permitivity of free space,
a is the dielectric constant of the adhesive (which is dependant
on the moisture content)
A is the bonded area
d is the thickness of the adhesive

The dielectric constant increases as moisture ingresses into the adhesive. The relationship

between the dielectric constant and moisture ingress is

12

8 a 8M +M(5w “5.44) (1'10)

where EM is the dielectric constant of the epoxy adhesive
M is the moisture concentration in the adhesive
8w is the dielectric constant of water

By combining Eq (1-9) and Eq. (1-10), the moisture increase in the adhesive from one

instart of measurement to another may be expressed as

M -£(—£M—) (1-11)

where AC is the difference between the measured capacitance at two
different times

a is the slope of the dry capacitance vs bonded area
relationship

13

 

15+09 ._

 
     

 

 

Dry Wedge Test
1E+08 » FM73 .
E ,
G1 E+07 +
o
I:
'3 .
a1E+06 l
E l
1E+°5 4h Bonded Area !
1E+o4 .L t . t l !
1E-02 1E—01 1E+00 1E+01 1E+02 1E+03 lE+o4

Frequency

 

 

 

Figure 1-4 Typical impedance spectra taken from the wedge test (Source: Davis et al.
1999)

 

 

 

 

 

 

 

1.5m
1.E+10l
9
it
a: 1.E+09 -
+
it
OFPL FM123
1'E+°8 ‘ - FPL FM73
A FPL FMSOOM
1.E+07 . . .
1 10 100 1000 10000
Time (hr)

 

 

Figure 1-5 Resistance parameters from the equivalent circuit analysis as a function of
time for the wedge tests under wet condition (Source: Davis et al. 1999)

(a)

 

2.0E-09

 

 

eFM7

1.5E-09 ‘

 

 

 

Capacitance (F)
2;
m
c':
(O

 

 

 

 

Bonded Area (crib

 

 

 

(b)

 

 

 

1.2E-09 ‘— 1 ‘ "fl
' x FPL FM300M

””3091 o FPL FM123
ll FPL FM73

 

 

8.0E-10

6.0E-10 ‘

Cc (Farad)

4.0E-10 ‘

 

2.0E-10 '

 

 

 

 

0.0E+00

w—‘

0 5 10 15 20 25

Bonded Area (cmz)

 

 

 

Figure 1-6 Variation of capacitance parameter from the equivalent circuit analysis as a
function of bonded area for wedge tests under —(a) dry condition (b) wet
condition (Source: Davis et al. 1999)

15

1.4 Hypothesis and Objectives

The hypothesis investigated in this research is that EIS can be used to detect

debonding of CFRP reinforcement in concrete structures strengthened with CFRP. This

hypothesis is based on the successful use of this method for glass-reinforced

composite/composite bonds by Davis et al. (1999).

The objectives of the research are to:

DeveIOp effective sensor configurations, measurement schemes, and data
interpretation techniques.

Ascertain the effects of short-term environmental conditions such as
temperature and humidity variations on the measurements and
interpretations. S

Ascertain the effect of long term environmental conditions such as freeze-
thaw cycles, wet-dry cycles and corrosion of reinforcement on the

measurements and interpretations

16

Chapter 2

Experimental Design

Debonding of external CFRP reinforcement was investigated by using wedge tests.
In these tests, the CFRP sheet is debonded by driving a thin wedge between it and the
concrete substrate. Impedance measurements are taken periodically and the variation of
impedance with the debonded length or area is studied.

Concrete structures rehabilitated with carbon fiber reinforced polymer (CFRP) are
subjected to various environmental conditions. The environmental conditions can be
divided into short-term and long-term conditions. Short-term environmental conditions
represent the conditions at the time impedance measurement are taken. These conditions
influence the impedance measurements, and include temperature variations, humidity
variations and variations in chloride concentration caused by deicing agents. The long-
term environmental conditions include corrosion of reinforcing bars, freezing and
thawing cycles and wetting and drying cycles. The impact of short term and long term
environmental effects on impedance measurements must be well understood if the latter
are to be used to detect debonding of the CFRP. It also is important to have durable and
reliable sensors to obtain accurate impedance measurements during the service life time
of rehabilitated structures.

Three different humidity levels and three different temperatures were applied to six
reinforced-concrete specimens to study the sensitivity of sensor measurements on these
short term environmental conditions. Of the Six Specimens, three were manufactured with

chloride (18.6 lb NaCl per cubic yard of concrete) and three without chloride. Nine

17

reinforced concrete specimens with chloride were used to study the durability and
reliability of installed sensors. Three specimens were subjected to 300 freeze-thaw
cycles, three were subjected to 300 wet-dry cycles, and three were subjected to
accelerated corrosion conditions to study effects of rebar corrosion.

The size of the concrete specimens and the effective electrode area of installed
sensors can also affect impedance measurements. A large reinforced-concrete beam was

fabricated and used to study size effects.

2.1 Specimen Preparation and Sensor Configuration

2.1.1 Specimen sizes and preparation

Three different sizes of reinforced concrete beams were manufactured and used in
this research. Small reinforced-concrete prismatic beams (6 x 6 x 12 in.) with a #4
reinforcing bar were manufactured to study environmental effects on impedance
measurements. Medium reinforced concrete beams (6 x 6 x 24 in.) with a #4 reinforcing
bar were fabricated for conducting wedge tests to study the effect of CFRP debonding.
Each #4 reinforcing bar was installed near the bottom surface (1.5 in. up from the
bottom). A large reinforced concrete beam (1.5 x 2 x 8 ft) was fabricated to study the
effect of CFRP debonding and specimen size. Figures 2-1 and 2-2 illustrate the size of the
concrete beams and the location of the reinforcing bar.

One large reinforced concrete beam of dimension 8’ x 1.5’ x 2’ was used to study
size effects. The concrete beam had four 8-foot long #6 bars, one 8-foot long quarter inch
by quarter inches quare stainless steel bar, and number 3 stirrups at 10—inch spacing.
Reinforcing bars were planned to be used as internal sensing elements. The stainless steel

bar was included to eliminate the effect of corrosion which might affect measurements. In

18

the large beam, reinforcing bars as well as the stainless steel bar were to be used as
internal sensing elements. Figure 2-3 shows the size of the large concrete beam and the

location of sensors. Figure 2-4 Shows the cross section of the large beam.

 

   

 

 

FRP . Concrete ‘
Composute . 5,3,3.
I I . 6”
I I
a I- ——————— | d ———————— 'II I — — - *1.th

 

 

 

 

 

 

I I I I
I I I
F i
I

 

 

Initially
debonded
area

Figure 2-1 Small concrete beam configuration and sensor locations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

......., TIMI“ l 6”
__i IL: | ' ' I I
r“ I I : I I I g j
I s : I : : :
W7 If?“ E i i I I
— 4:}. .51-...1 _____ I ...... i ..... I _____ i-----.
7%" — - — I — —|- - — — h -p — , i
l ' e .
R I l Initial debonded area E '
I E i ' subsequently lengthened :
l A: . .
I i i " | .
W1 {C1 W2 :C2 W3 ETCB W4 E W5 E C5 W6 i C6
l .i .l I | .
I 2” I 4” I 4” I 4” I 4” 4” 2”

Figure 2-2 Medium concrete beam configuration and sensor locations

19

 

12”

 

 

 

Ia
i2
8
t:

. 8

Qt
5
CL
Q
0

yo

A

a,

N

 

 

 

 

 

 

24”

12”

—>‘

 

H

 

 

 

 

 

 

 

‘
I
I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.25” stainless

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

steel bar
I
:g' " - \Initial debonded
°.- J - — — . area subsequently
_ .. .l _ _ _ , lengthened
- - 1 i -‘;\4 - #6 rebar with

- ' #3 stirrups at 10”
/ on center

\ Stainless steel wire
._. sensors ——
.3 3” on center up to 3’,
6” on center 3’ to 6’, and
12” on center 6’ to 12’

' ' CFRP composite

 

18” I

Figure 23 Large concrete beam configuration and sensor locations

20

2.1.2 Sensor configuration and installation

In the modified EIS technique, the impedance of an electrochemical cell is measured
between two electrode sensors. Three different sensor elements were used in this
research. These were the reinforcing bar, internal sensors and external sensors. Internal
sensors (W5) and external sensors (C,) were installed after the reinforced concrete
specimens were manufactured and grooved (0.25 inch deep) on the bottom surface using
a concrete saw. The grooves were made in the longitudinal and transverse directions.
Internal sensors consisted of 0.625 mm diameter stainless steel wire installed in the
grooves.

External sensors consisted of copper tape with a conductive adhesive backing (Saint-
Gobain, C665) placed over stainless steel wire. These were bonded directly to the carbon
FRP prior to applying the last epoxy layer to create an embedded sensor because of the
excellent conductivity of the CFRP. Initial investigations indicated that external sensors
applied to the outside of the CFRP after the final protective epoxy layer was applied were
ineffective. The thick protective epoxy layer yields impedances from the rebar to the
external sensors that are beyond the range of what can be measured with the Gamry
(PC4-300 Potentiostat board/11125) potentiostat. External sensors were installed only in
the transverse direction on the outmost surface of the CFRP reinforcement. The carbon
fibers are highly conductive in the longitudinal direction. However, because of gaps
between adjacent fibers, the conductivity in the lateral direction is generally poor. By
bonding the copper tape in the transverse direction across all the fibers in the carbon FRP,
good electrical contact can be made with the carbon fibers. The entire FRP then acts as
one of the electrodes in the cell. Since the copper tape is on the outside of the FRP, it

does not affect the bond between the FRP and the concrete or between different FRP
21

layers, and will therefore not compromise the strength of the CFRP reinforcement.
Figures 2-1, 2-2 and 2-3 Show the configuration and direction of the sensors on the
concrete beam specimens. Table 2-1 shows the number of sensors and their directions.
For the large beam, the entire reinforcement cage (four longitudinal bars and stirrups)
acts as an internal sensors and the stainless steel bar was a second internal sensor.

Installation of internal sensors must be done carefully so that the stainless steel wires
do not touch the CFRP. A short circuit can be created between the copper tape on one
side of the CFRP and the wires on the other side. Even though there may be more than
one layer of CFRP used with a thin layer of epoxy between the layers, a short circuit can
be created. A heat-shrinking insulator was used around the internal sensors in the
transverse direction where they overlapped the internal senSOrs in the longitudinal
directions so that electrical contact between the wire electrodes was avoided. A heat-
shrinking insulator was also used on the internal sensors at the edge of the concrete
specimens to avoid contact with exposed carbon fiber in the CFRP.

2.1.3 Sensor combinations for impedance measurements

Three different sensor combinations were used to measure the impedance.
Combinations of internal sensors (stainless steel wires W,) to external sensors (cOpper
tapes Ci), rebar to external sensors, and internal sensors to internal sensors were used to

measure the raw impedance data.

22

24 “

 

 

#3 stirrups
10” o.c

#6 bars

 

 

.6

K #6 bar/

\
m

C

 

 

 

 

18”

D

 

1-0.25”
Stainless steel
bar

Figure 2-4 Cross-section of large beam

Table 2-1 Number of sensors and their orientation on the specimens

 

Small size specimen

Medium size specimen

Large size specimen

 

 

 

 

 

Transverse Longitudinal Transverse Longitudinal Transverse Longitudinal
Internal sensor 3 1 6 1 20 1
External sensor 3 0 6 0 4 O
Rebar O 1 0 1 0 2

 

 

 

 

 

 

 

23

 

2.2 Detection of FRP Debonding

2.2.1 Two foot long specimens

Six 6”x 6” x 24” reinforced concrete specimens with chloride were prepared to
determine whether impedance measurements could be used to detect the debonding of
CFRP from concrete. Each specimen was prepared with an initial crack ( ” in length)
between the CFRP and concrete using a teflon insert. To propagate the interfacial crack
between the CFRP Sheet and the concrete substrate, a wedge (sharpened saw blade) was
driven with a hammer. The crack was propagated by two to four inches each time.
Impedance measurements were made at each extended crack states from the internal
sensors (stainless steel wire or reinforcing bar) to each of the OOpper tape sensors.

Investigation of the debonding of CFRP reinforcement from concrete was conducted
in two different environmental conditions. One was a controlled environment (in a
refrigerator) with a temperature of about 42°F and a constant relative humidity level of
about 35% and the other was an ambient condition in the lab. It was noted during the
study on temperature effects that at high temperatures relatively small variations in
temperature caused significant variations in the impedance measurements. For this
reason, a cold temperature was choSen to reduce the sensitivity of the impedance
measurements to temperature fluctuations. Three concrete specimens were used in the
controlled environment. The ambient condition represented a temperature of about 70°F
to 75°F and a relative humidity that varied from 30% to 60%. Three concrete specimens

were subjected to the wedge test in the ambient condition.

24

2.2.2 Large beam specimen

The effective electrode area of installed sensors and the size of the concrete
specimen are expected to influence impedance measurements. Electrodes with larger
effective surface areas in the electrochemical cell typically decrease the magnitude of
impedance. A large reinforced-concrete beam was fabricated with external and internal
sensor to study the impact of Specimen size. The large specimen was prepared with an
initial one-inch long crack by using a teflon insert. To create the interfacial crack between
the FRP layer and the concrete substrate, a sharpened metal sheet (wedge) was driven
with a hammer. The crack length was increased by two to four inches at each stage until
the crack length was about 25% of the length of the entire beam. Impedance
measurements were made at each extended crack State from internal sensors (stainless
steel wire, stainless steel bar or reinforcing bar) to each of the copper tape sensors and
between pairs of internal stainless steel wire sensors. The beam was constructed in the
laboratory and after 28 days of curing it was moved outdoors. The beam was exposed to
the winter climate from February 21 to April 2, 2003. The original intent was to conduct
the wedge test in an outdoor environment when temperatures were in the 35-450F range.
However, because of the unusually warm winter in 2003, temperature fluctuations were
large. After obtaining impedance measurements at one crack stage (three-inch long crack)
when the beam was outdoors, it was decided that the remainder of the test would be done
indoors. The wedge test was therefore conducted in a normal ambient condition at

temperatures between 65°F and 75°F and a relative humidity between 30% and 60%.

25

2.3 Assessment of the Sensitivity of Sensor Measurements to
Environmental Effects

2.3.1 Humidity

Moisture in the concrete specimens is the medium for transporting charges between
sensors. Since charge is carried in the electrochemical cell (concrete specimen) by the
movement of ions, different levels of moisture should affect the capacity for transporting
charges. Three different humidity levels were considered in the study to assess the
sensitivity of sensor measurements to humidity.

Six small reinforced concrete specimens (6 x 6 x 12 in.) were subjected to relative
humidity levels of 30 percent, 60 percent and 90 percent in a controlled environmental
chamber, while the temperature was held constant at 72°F (room temperature). The range
of relative humidities from 30 to 90 percent covers a wide range from extremely dry air
to highly saturated air.

Each of the six concrete Specimens was subjected to the three different relative
humidity level, until the weight of the specimens stabilized. Of the six specimens, three
were manufactured with chloride (18.6 lb NaCl per cubic yard of concrete) and three
without chloride. Impedance measurements were taken after this exposure.

2.3.2 Temperature

Temperature affects the rate of ion transfer in electrolytes and is therefore expected
to influence the magnitude of impedance. In order to study temperature effects on the
impedance measurements, six small concrete specimens (6 x 6 x 12 in.) were subjected to
three different temperatures. The same six Specimens used to study humidity effects were

reused. The Specimens were subjected to extreme cold (temperature of about 00F), a

26

moderate temperature (about 70°F) and a typical hot summer temperature (about 100°F).
The humidity level was not controlled. For the cold condition, the specimens were kept in
a freezer for a week at a temperature of GOP and the impedance measurements were taken
while specimens were inside the freezer. For the hot summer temperature, the specimens
were placed in an oven for a week at a temperature of 100°F and the impedance
measurements were taken while the specimens were inside the oven. For the moderate
temperature, specimens were exposed to a room temperature of 70°F and a relative
humidity level of 60 percent.

2.3.3 Chloride content

Chloride ions can exist in concrete structures from the time the concrete is cast due
to the use of chemical agents, and the concentration can change with time due to the
diffusion of deicing agents through the concrete. The existence of chloride ions in
concrete provides additional free ions for transport of charges in EIS and is expected to
have an effect on the measurements. An experiment was setup to study the sensitivity of
impedance measurements to the presence of chloride ions. The experiment to study
chloride effects was conducted Simultaneously with the experiment to study humidity and
temperature effects. Six small reinforced concrete specimens (6 x 6 x12 in.) were used.
Of the six specimens, three were manufactured with chloride (18.6 lb NaCl per cubic
yard of concrete) and three without chloride. The specimens were subjected to relative
humidity levels of 30 percent, 60 percent and 90 percent at a temperature of 700 F in a
controlled environmental chamber and the same six reinforced concrete specimens were
also subjected to temperatures of 100°F and GOP without any humidity control. Figure 2-5
illustrates how the same specimens were re-used to study the effects of humidity,

temperature and chloride content.

27

 

Three small reinforced concrete specimens with chloride
Three small reinforced concrete specimens without chloride

 

 

 

 

f

 

 

 

Specimens subjected Specimens subjected Specimens subjected
to RH of 30% at to RH of 60% at to RH of 90% at
temperature of 70°F temperature of 70°F temperature of 70°F

 

 

 

 

 

 

 

 

 

i ii i i

Impedance measurements taken and evaluated with equivalent circuit analysis for
humidity and chloride content effects

 

 

 

 

 

f

 

 

Specimens subjected Specimens subjected
to temperature of 00F to temperature of
100°F

 

 

 

 

 

 

Impedance measurements taken and evaluated with equivalent circuit analysis for
temperature and chloride content effects

 

 

 

 

Figure 2-5 Experimental design flow chart for humidity, temperature and chloride content
effects

28

2.3.4 Freeze - thaw

Concrete bridge structures in Michigan undergo freezing and thawing cycles in
winter which promotes the growth of microcracks in the concrete. This is likely to change
impedance measurements. A freeze-thaw test was setup to study the variability of sensor
measurements. This experiment was setup to determine the change in impedance
measurements before and after 300 freeze-thaw cycles. Three small concrete specimens
(6 x 6 x 12 in.) with chloride (18.6 lb NaCl per cubic yard of concrete) were used to study
freeze — thaw effects.

Initially, the concrete specimens were subjected to a temperature of about 70°F and a
humidity level of 60 percent until their weight stabilized. The baseline impedance
measurements were then taken before the Specimens were subjected to freeze-thaw
cycles. The specimens were then subjected to 300 cycles of freeze-thaw. The ASTM
C666 Procedure B was used in which each freeze -thaw cycle consisted of a two-hour
freezing period and a four-hour thawing period in a freeze-thaw machine. After 300
cycles of freeze-thaw, the specimens were removed from the freeze-thaw machine and
placed in an oven to dry out the excessive water absorbed during the freeze-thaw test.
Once the specimens reached approximately the same weight as before the freeze-thaw
test, they were conditioned at a temperature of about 70°F and impedance measurements
were taken again.

2.3.5 Wetting and drying

Concrete bridge structures experience cycles of wetting and drying on a continuous
basis. The expansion and shrinkage resulting from wetting and drying cycles is likely to
promote the growth of microcracks and cause impedance measurements to change. A

wet-dry experiment was set up to study the changes in impedance measurements after

29

wet-dry cycles. Baseline impedance measurements were taken on three small concrete
specimens (6 x 6 x 12 in.) with chloride at a temperature of about 70°F and a humidity
level of 60 percent. The Specimens were then subjected to 300 wet-dry cycles. Each wet-
dry cycle consisted of a two hour wet period followed by a ten hour dry period. A 2%
NaCl solution was used to wet the Specimens in order to increase the concentration of
chloride ions. After 300 cycles of wetting and drying, the specimens were placed in an
oven at a temperature of about 45°C to dry out the excessive water absorbed during the
experiment. Once the concrete specimens reached approximately the same water content
at which the baseline measurement were taken, impedance measurements were retaken.
2.3.6 Corrosion of reinforcing bar

The reinforcing bar in the specimen was used as an internal sensor. The electrical
properties of the reinforcing bar and surrounding concrete may change due to corrosion
of the reinforcing bar and the presence of corrosion products in the concrete. Cracking of
the concrete surrounding the reinforcement due to the volume expansion associated with
corrosion also is likely to change impedance measurements. An accelerated corrosion
experiment was used to study the effect of corrosion of the reinforcing bar on the
impedance measurements.

For the corrosion study, small concrete specimens (6 x 6 x 12 in.) with two
reinforcing bars, one near the bottom face and the other near the top face, were used.
Baseline impedance measurements were taken at a temperature of 70°F prior to the
accelerated corrosion test. A constant voltage of 12 V was then supplied across the two
reinforcing bars using an external power source. A 2 percent sodium chloride solution (by
weight) was used to wet the concrete specimens for one hour in each 24 hour period. The

moisture in the concrete specimens and the applied current induces corrosion in the

30

anodic reinforcing bar in an accelerated fashion. The anodic bar was located near the
bottom face of the specimens on which the CFRP was bonded. Figure 2-6 shows the
accelerated corrosion experimental setup. The three specimens were subjected to
accelerated corrosion for three weeks (21 days). The applied current was monitored at an
interval of two minutes with a voltage data-logging unit (Omega Engineering Inc.,
AD128-10T2), since this has a direct correlation with the total corrosion in the
reinforcing bar. After 21 days of accelerated corrosion, the specimens were placed in an
oven to dry out the excessive water absorbed during the wetting of the specimens.
Impedance measurements were taken once the concrete specimens reached

approximately the same water content at which the baseline measurement were taken.

31

 

 

External
power source

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I Typical
I 6” x 6” x 12”
R A// specimen with
two #4
reinforcing
1252 Resistor , bar
I \CFRP sheet
I
r r‘ I I
I4 R H '

12Q Resistor

 

 

 

 

 

 

 

 

 

 

_\
_‘
_\

 

 

 

 

 

 

 

i .

 

 

 

 

Voltage data-logging unit

 

 

 

Figure 2-6 Experimental setup for accelerated corrosion

32

2.4 Electrode size effects

The electrode surface area of the sensors installed in concrete may influence
impedance measurements. One reinforced concrete slab of dimension 6” x 2’ x 1’ was
used to study the effect of different electrode sizes. Five steel sensors with diameters of
0.034”, 0.25”, 0.5”, 0.59”, and 0.75” were embedded in the Slab. Two layers of CFRP
sheets were bonded to the face of the slab closest to the embedded sensors. Figure 2-7

shows the concrete slab configuration.

 

CFRP
Sheet

 

 

 

 

 

 

 

 

 

 

 

 

4” 4” 4” 4” 4” 4”

 

 

 

Figure 2-7 Concrete slab configuration and sensor locations

33

Chapter 3

Experimental Results, Equivalent Circuit
Analysis, and Discussion

Three different sensor electrode combinations were used to measure the impedance.
Combinations of internal sensors (stainless steel wires W,) to external sensors (copper
tapes Ci), rebar to external sensors and internal sensors to internal sensors were used to
produce the impedance spectra. The interfacial debonding effects are best observed with
the combinations of rebar to external sensors and internal sensors to internal sensors.

These measurements and their interpretation are discussed in this chapter.

3.1 Equivalent circuit

Impedance spectra were measured between an internal and external sensor, the rebar
and external sensor, and two internal sensors. Once the raw impedance data is obtained,
then equivalent circuit analysis is performed by fitting the data to obtain the circuit
element parameters. Finding an equivalent circuit approximate for the measured EIS
spectra can be a challenge since each equivalent circuit has its own unique characteristics
and impedance spectra. An equivalent circuit is composed of common electrical
components such as resistors and capacitors. Sometimes, specialized elements such as a
constant phase element are added to the equivalent circuit to describe the special
characteristics of the electrochemical cell.

Different equivalent circuits were tried to obtain good fits to the measured
impedance Spectra. The circuit shown in Figure 1-2 (b) which was used by Davis et a1.
(1999) for studying composite/composite bonds was one of the first circuit tried. This

34

circuit fits spectra measured between the rebar and an external sensor reasonably well.
However, it did not produce a good fit in the low-frequency range for impedance
measurements between an internal and external sensor or between two internal sensors. A
new equivalent circuit was developed to provide a better fit of the measured impedance
spectra over the entire frequency range for all sensor combinations. Figure 3-1 shows the
equivalent circuit configuration used in this research. Figure 3-2 shows the measured
spectra between the rebar and an external sensor and the fit by the two circuits shown in
Figures 1-2(b) and 3-1 on the Bode magnitude and phase angle plot. Figure 3-3 shows the
impedance measured between two internal sensors and the fit obtained using the two
equivalent circuits on the Bode magnitude and phase angle plots (Figure 3-3 (a)) and the
Nyquist plot (Figure 3-3 (b)). It can be seen from the figures that the new circuit in Figure
3-1 yields very good fits for all measured spectra even though it is more complex than the
one used by Davis (Figure 1-2(b)). The new circuit has eight independent parameters
(one for each resistor and Capacitor, and two for the CPE), and was used to analyze all the
measurements obtained in this work.

In the equivalent circuit analysis. the R1, C1 and CPE parameters control the
impedance in the low-frequency region (0.1 to 10 Hz) and characterize the diffusion
process in the electrolyte (pore fluid in the concrete). The C2, C3, R2 and R3 parameters
control the impedance in the mid- (10 to 1000 Hz) and high-frequency (1000 to 100,000
Hz) regions and characterize the charge transfer procesSes at the two electrode/electrolyte
interfaces (see Figure 1-2). The impedance of the equivalent circuit in Figure 3-1 can be
expressed analytically using Equations 1-7 and 1-8. The first loop has the equivalent

impedance

35

PY+SQ_ .SP+YQ

 

 

 

 

Z 3-1
‘ P2+Q2 JP2+Q2 ( )
where
P=A*C1*w"" Inn-01275)”
Y = R1 + A *w"“ *cos(a7fl)
Q =C1*R1*w+A*C1*w"“ won-“21)
S - A *w'“ *sin(g§-)
2
The second and third loops have the equivalent impedances
R2 (0"‘C2"‘R22
Z a — . 3‘2
2 1+02 *sz *R22 Jl+w2 *sz *R22 ( )
R3 r0"‘C3"'R32
Z = - ' 3-3
3 1+02 *(332 I1232 11+(o2 *C31 I1232 ( I
The impedance of the entire circuit is therefore
Z,,q=Z,+Z,+Z3 (3-4)

36

I
|

 

 

 

I
I

J\/\fi>— 4W—

R1 CPE R2 R3

 

 

 

 

 

 

 

Figure 3-1 Equivalent circuit used to study CFRP debonding

 

 

 

 

1E9 gm" 5 ' ‘ _ " ”_ w ' ' —”'~“’ ' ""— H 00
1E8 I
3 ~20
1E7§ -
3' * I'sI
A I .-40 “U
E: 1136! . a
u: 2 A
3 1' E?
1E54 I- "3
i I “I:
1E4 g .
3 --80
1E3 ‘
152i . ,..,,,. . ...,... ......... . ........ . .. ......... ......... . ....... -100
1E-2 lE-l 1E0 1E1 1E2 1E3 1E4 1E5 1E6
Freq (Hz)

 

Figure 3-2 Fitted Bode magnitude and phase angle plots using the two equivalent circuits.
Dots are the actual measured impedance from the rebar to an external sensor.
Solid lines are the theoretical impedance using the equivalent circuit in Figure 3-
1. The dotted lines are the theoretical impedance using the equivalent circuit in

Figure 1-2 (b).

37

 

 

 

 

 

 

 

 

1E7 . "W W V SWISWTWV in“ _ 0.0
" -20
1E6 : .
... , -40 v:
E N
.C A
O G.
Z 68
o - -60 a:
E _ a:
NIES ; V
' -80
1E4 .. ,. . . ,..._. . . .' -100
1E-2 1E-1 1E0 1E1 IEZ IE3 IE4 1E5 1E6
Freq (Hz)
(b)
2.5E6'I
’%
2.0E6‘ .
E
81.5136
:3
.§
Is 1.0E6
0.5E63
0.0 .
0.0 5E6
Zreal(ohm)

 

 

 

Figure 3-3 Fitting with the equivalent circuits. Dots are the actual measured impedances
between two internal sensors. Solid lines are the theoretical impedances using the
equivalent circuit in Figure 3-1. The dotted lines are the theoretical impedances
using the equivalent circuit in Figure 1-2 (b). — (a) Bode magnitude and phase
angle plot; (b) Nyquist plot.

38

3.2 CFRP Debonding

3.2.1 Results for 2-foot long specimens

Based on measurements from numerous pairs of sensors, it was found that
measurements from the rebar to external sensors had the strongest promise for revealing a
relationship between impedance measurements and the debonded area of CFRP. Figures
3-4 and 3-5 show the effect of CFRP debonding on the magnitude of the measured
impedance in the controlled environment (in the refrigerator), and in the uncontrolled
environment (ambient condition), respectively. These figures show that the measured
impedance increases over all frequencies due to debonding of the CFRP. The effects of
CFRP debonding on the measured impedance in the controlled environment also can be
seen with the Nyquist plot in Figure 3-6. The radius of the arc in the Nyquist plot
increases as the crack length increases. Figure 3-7 is an enlargement of Figure 3-5 in the
low- to mid-frequency range and shows the interference of humidity and debonding
effects on the measured impedance. The ambient relative humidity fluctuated from 30%
to 60% when the crack lengths increased from 6 in. to 10 in. This increase in relative
humidity caused the impedance magnitude at low frequencies to decrease even though
the interfacial crack length increased. However, the increase in humidity did not cause a
reversal in the measured impedance at mid- and high- frequencies. The interference of
humidity and debonding effects on the measured impedance can be seen more clearly on
the Nyquist plot shown in Figure 3-8. The radii of the arc in the Nyquist plot decreased

significantly when the crack lengths were at 8 and 10 inches.

39

 

 

 

 

1E+8

 

 

1E+7

 

hms)

o
v1E+6 ‘

1E+5

 

 

1E+4

 

Impedance

 

1E+3 ~ ——~w

 

1E+2 r

l l 1

1E-1 1E+O 1E+1 1E+2 1 E+3 1 E+4 1E+5
Log frequency (Hz)

 

 

 

Figure 3-4 Bode magnitude plot of typical impedance spectra measured from the rebar to
copper tape for wedge test in controlled environment (legend gives the crack

 

 

 

        

 

 

 

 

 

 

length in inches)

1E+8 + 4n + 6"
A 1E+7 4 +3" + 10" __
g -o—14" —+—16"
f9, 1E+6 -
8
c 1E+5 4
as
'0
3154.4 —
é

1E+3

1E+2 iv ‘ '" I ' ’ "l I I I I

1E—1 1 E+O 1 E+1 1E+2 ' 1E+3 1 E+4 1 E+5
Log frequency (Hz)

 

 

 

Figure 3-5 Bode magnitude plot of typical impedance spectra measured from the rebar to

copper tape for wedge test in ambient environment (legend gives the crack length
in inches)

40

 

-8.0E+8 1——~ +4" +8"

 

 

 

 

 

 

 

 

 

 

 

 

 

’5 - +10" +12"
2 7.0E+8 .. .. /
0 +16 +18
5 '6.0E+8 “PAL /
c
3
.§
2‘
as
.E
D)
as
E
0.0E+0 I r l
0.0E+0 5.0E+7 1.0E+8 1.5E+8

Real impedance (ohms)

 

 

 

Figure 3-6 Nyquist plot of typical impedance spectra measured from the rebar to copper

tape for wedge test in controlled environment (legend gives the crack length in
inches)

 

 

1E+8

 

 

.3
I11
+

V

A
m
3:

 

 

Impedance (ohms)

 

1E+5-i“ ' f I ~ **I - * I
1 E1 1 E+O 1 E+1 1 E+2

Log frequency (Hz)

 

 

 

Figure 3-7 Enlargement of Figure 3-6, showing variation of measure impedance in the
low to mid- frequency range (legend gives the crack length in inches)

41

 

 

1

Figure 3—8 Nyquist plot of impedance spectra measured from the rebar to copper tape for

Impedance (ohms)

Inary

lmag

 

 

 

 

 

'2.5E+7 + 4"
.A u
-2.0E+7 -\ +6
-1.5E+7 A ~ 7,, , A - . ,, +8
/
/ + 10"
-1.0E+7 + - / p, + -— +
[/I + 14"
// "
-5.0E 6 I
+ . " -—+—16"
0.0E+0 7 l 1’ ’ l l T'mhl’ ,7, _ a 1

0.0E+0 1.0E+7 2.0E+7 3.0E+7 4.0E+7 5.0E+7 6.0E+7

Real impedance (ohms)

wedge test in ambient environment (legend gives the crack length in inches)

42

 

3.2.1.1 Comparison of raw impedance spectra
Figures 3-9 to 3-11 show the variation of the measured impedance at 10,000 Hz, 100

Hz and 1 Hz as a function of the interfacial crack length. These frequencies are arbitrary
representative of high, medium and low frequencies. The impedance magnitude vs. crack
length curves are close together at 10,000 Hz, and spread further and further apart at 100
Hz and 1 Hz. The specimen-to-specimen variation in the measurements is therefore the
least at high frequencies. In general, the measured impedance at a specific frequency
increases exponentially as the crack length increases.

3.2.1.2Jesults from equivalent circuit analysis

The variation of the CPE parameter with crack length is shown in Figures 3-12 and
3-13. Although the CPE parameter exhibits good correlation with the interfacial crack
length it has high specimen-to-specimen variation and is also sensitive to temperature and
humidity. The variations of the capacitance parameters with interfacial crack length are
shown in Figures 3-14 to 3-16. The capacitance parameters exhibit a linear relation with
the crack length and except for specimen 3, the C2+C3 and C1+C2+C3 values for the

other four specimens are quite similar.

43

 

 

 

 

 

 

 

 

 

2.03.47 -——.——‘— _- _-____ ____
+82 +84 +86!
+83 +85
«é: 1.5E+4 ~ _ ,
.C
8
8 10E+4
{is
'U
Q)
G.
E 5.0E+3
0.0E+0~r — . , ,
O 3 6 9 12 15 18 21
Cracklength (inch)

 

 

 

Figure 3-9 Variation of typical measured impedance from the rebar to copper tape for

 

 

 

 

 

 

 

 

 

 

 

 

wedge test at 10,000 Hz
L0E+6fi +82 +84 +86
+83 +85 /
A 8.0E+5
U)
E //
.C
3 6.0E+5 -« .. , _
a:
0
g /
8 4.0E+5
O. ./
E
2.0E+5
0.0E+O T i ——* l' .____ ;---- "T I .. . I I
0 3 6 9 12 15 18 21
Crack length (inch)

 

 

 

Figure 3-10 Variation of typical measured impedance from the rebar to copper tape for
wedge test at 100 Hz

 

 

 

+82 +84 +86

8.0E+7 + 83 + $5 .
a? 7.0E+7 - ~— e
6.0E+7 1'

 

 

 

 

 

 

 

 

 

 

 

0 3 6 9 12 15 18 21
Crack length (inch)

 

 

 

Figure 3-11 Variation of typical measured impedance from the rebar to copper tape for
wedge test at 1 Hz

 

 

 

 

 

 

 

 

 

 

 

 

 

1.6E+9 *' 21
+spec.

3 145.9- +8”. 4 -~—-——-----/
E1.OE+9~ , ~ * — /
0)
Es 8.0E+8 /'
(6 _
a 6.0E+8
O.
LIJ 4.0E+8
O.
0 2.0E+8 4-

0.0E+0 l . I l

0 3 6 9 12 15 18 21
Crack length (inch)

 

 

 

Figure 3-12 Variation of the CPE parameter for wedge test in the controlled environment
for measurements from the rebar to external sensors

45

 

1.2E+8 .. —~* ———~

 

+ spec. 2
+ spec. 3
8.0E+7 ~ + 5990- 5

 

1 .OE+8

 

 

 

 

 

 

CPE parameter (1/Farad)

 

 

 

 

 

6.0E+7

4.0E+7 4

2.0E+7

0.0E+0 i . T T . . .
0 3 9 12 15 18 21

Crack length (inch)

 

 

Figure 3-13 Variation of the CPE parameter for wedge test in ambient condition for
measurements from the rebar to external sensors (Note: the CPE parameter for
spec. 2 is measured in controlled environment)

 

8.0E-9
7.0E-9 1

 
 

- +spec.2+spec.3
M,+spec.4—x—spec.5
+spec.6

 

 

 

 

6.0E-9
5.0E—9 ..
4.0E-9 *
3.0E-9 *

parameter (Farad)

 

 

5 2.0E-9
1.0E-9 * -' -

0.0E+0 " ’ I"
0 3

 

 

 

9 12 15 18 21
Crack length (inch)

 

 

Figure 3-14 Variation of the C1 parameter from equivalent circuit analysis for wedge test
from the rebar to external sensors

46

 

 

 

 

 

 

 

 

 

 

 

3.0E-8 +spec.2 +spec.3
’6 +spec.4 +spec.5
cu 2.5E—8 _
<71 +spec.6
L1:
9 2.0E-8 —
93
3 1.5E-8
9 l
8 !
m 1.0E-8 1
E2
8 5.0E-9

0.0E+0 ~ , , , , . I .

0 3 6 9 12 15 18 21
Crack length (inch)

 

 

 

Figure 3-15 Variation of the C2+C3 parameter from equivalent circuit analysis for wedge
test from the rebar to external sensors

 

 

 

 

 

 

 

 

3.5E-8 — ~ +spec. 2 +spec. 3
:1; +spec. 4 +spec. 5
E 3.0E-8 ~~ +spec.6
5 2.5E-8
‘2
E‘.’ 2.0E-8
8 1.5E-8 ~
8
(3 1.0E-8 —
o
2: 5.0E—9
0
0.0E+0 I T , . , , fl
0 3 6 9 12 15 18 21
Crack length (inch)

 

 

 

Figure 3-16 Variation of the C1+C2+C3 parameter from equivalent circuit analysis for
wedge test from the rebar to external sensors

47

3.2.2 Results for 8-foot long specimen

The large concrete beam was studied in an ambient condition. Temperature
fluctuated from about 65°F to 75°F and the relative humidity fluctuated from about 30%
to 60%.
3.2.2.1 Comparison (3 raw impedance spectra

Figures 3-17 and 3-18 show typical measured impedance spectra for the large beam
as a function of the debonded length of the CFRP reinforcement. The measured
impedances for the large beam are about on order of magnitude lower than those for the
two-foot long beams. This is believed to be due to the larger sensor surface areas in the
beam (reinforcement cage and CFRP sheet). Figure 3-18 is an enlargement of Figure 3-17
in the low-frequency range and Figure 3-19 is the Nyquist plot over the entire frequency
range. The study of the effect of CFRP debonding on the two-foot long specimens
indicated that the radius of the arc in the Nyquist plot increases as the crack length
increases. In Figure 3-19, however, the radius of the arc in the Nyquist plot decreases as
the crack length increases. Figures 3-18 and 3-19 suggest that humidity and temperature
variations interfere with the effect of CFRP debonding in the low-frequency range, and
the impedance magnitude appears to decrease as the debonded length of the CFRP
increases.

The large beam was initially placed outdoors from February to March 2003 and then
moved indoors. It is very likely that the concrete dried out when it was outdoors at very
cold temperatures. After it was moved indoors the concrete would begin absorbing
moisture. Because of its large size, the beam is likely to have absorbed moisture
gradually over a long period of time. The increased moisture would explain the reduction

in impedance at low frequencies.

48

Figure 3-20 is an enlargement of Figure 3-17 in the high-frequency range. In the high
frequency range the magnitude of the impedance increases with the debonded length of
the CFRP although humidity and temperature fluctuated. Figure 3-21 shows typical
measured impedance spectra from the rebar and the stainless steel bar to copper tape
sensor. The measured impedance spectra from the stainless steel bar were very similar to
the measurements from the rebar. Figure 3-22 shows the measured impedance at 100,000
Hz. At this frequency the impedance magnitude shows a linear relation with the
debonded length of CFRP.

3.2.2.2 Results from equivalent circuMaQ/sis

Figures 3-23 and 3-24 show the variation of the parameters from the equivalent
circuit analysis due to CFRP debonding. The CPE parameter decreased as the crack was
propagated. This is a direct result of the low—frequency behavior of the impedance
(Figure 3-19) and is opposite to what is expected. The C1 parameter was almost constant
during the CFRP debonding test. It is speculated that the effect of CFRP debonding on
the C1 parameter was compensated by variations in humidity and temperature. The
C2+C3 parameter and the C1+C2+C3 parameters decreased significantly during the

initial debonding of the CFRP.

49

 

 

j—o—1" +3" +5" +8"

      

 

L—F 1 1 II + 1 6" + 20" _B_ 23“

1E+6 “ "

 

.2
m
+

U"

l

1 E+4

 

 

 

 

Impedance (ohms)
at
.1}.
00

1 E+2

 

 

1E+1 ~ 1 I *7 I .me , 1 .
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5

Log frequency (Hz)

 

 

 

Figure 3-17 Bode magnitude plot of typical impedance spectra measured from the rebar
to copper tape for wedge test of the large beam (legend gives the crack length in

 

 

 

 

inches)

1E+7 cm + 111 + 311 + 5n + 8n —
’0? \ +11" +16" —+—20" +23"
2 \
3 \ \
g \ \ \
(U 1E+6 1* ____ \r—w ” m \" ——”— ___ ”
E- \ \
E \ \
._ \

1E+5 -.-- *6 e h , c, 7 ~ ~ I r- ,, , «n ’

1 E-1 1 E+O 1 E+1
Log frequency (Hz)

 

 

 

Figure 3-18 Enlargement of Figure 3-17, showing variation of measure impedance in the
low-frequency range (legend gives the crack length in inches)

50

 

 

 

 

 

 

 

 

‘6'0E+6i +1" +3" +5" +8" » l
’3‘ +11" +16" —I—20" —-—23" .
.c -5.0E+6
9 1
§ -4.0E+6 « I
m |
E -3.0E+6 ~
.E
E," -2.0E+6
.E
3 -1 0E+6 —

g ' -
0.0E+0 1 l, 1 r’ ~ 1
0.0E+0 5.0E+5 1.0E+6 1.5E+6 2.0E+6 2.5E+6 3.0E+6
Real impedance (ohms)

 

 

 

Figure 3-19 Nyquist plot of impedance spectra measured from the rebar to copper tape
for wedge test of large beam in ambient environment (legend gives the crack

 

 

 

length in inches)
— l + 1" + 3" + 5" mi
4.1E+2 +8" +11" +16"
3.65-1-2 *’ A , ,7 r l __,_ 2011 _a_ 2311

 

 

 

 

 

Magnitude of impedance (ohms)

 

 

 

Log frequency (Hz)

 

 

 

Figure 3—20 Enlargement of Figure 3-17 showing variation of measured impedance in the
high-frequency range (legend gives the crack length in inches)

51

 

 

 

 

 

 

 

 

lE+7 * ’ * “ . Rebar
4

E? 1E+6 1 " * * * ’ — Stainless steel bar
.C
3 1E+5
0)
g \\
<0 1E+4
8 \\
0.
,§ 1E+3
c) \
o

1E+1 ..._....___ _ i _ _ 'l l' ._.___r__ 1 . l

1 E1 1 E+O 1E+1 1E+2 1E+3 1 E+4 1 E+5
Log frequency (Hz)

 

 

 

Figure 3-21 Bode magnitude plot of typical impedance spectra measured from the rebar
to copper tape and the stainless steel bar to copper tape for wedge test of the large
beam (legend gives the crack length in inches)

1

 

 

838
l
|
|

i

8 8
I
l
l
l

 

Magnitude of impedance (ohms)

 

'4 N
O O O
L l
l T
l
I
l
l

 

O
(.0
O)
(D

12 15 18 21 24
Crack length (inch)

 

 

 

Figure 3-22 Variation of typical measured impedance from the rebar to copper tape for
wedge test at 100,000 Hz

52

 

CPE parameter (1/Farad)

 

6.0E+6 -~—— . ,,

5.0E+6 -—

 

 

 

 

1
0
0

 

1 .0E+6

 

0.0E+0 e

l l "”WWTiTii'm l ' l l

9 12 15 18 21 24
Crack length (inch)

 

 

Figure 3-23 Variation of the CPE parameter from equivalent circuit analysis for the large
specimen from the rebar to external sensor

 

Capacitance parameter(Farad)

 

2.0E-7

 

1 .8E-7

—+61 +C2+03 +C1+62+cal

 

 

—<

 

1 .6E-7

 

 

1 .457
1 .2E-7

 

1.0E-7 r
8.0E-8

 

6.0E—8 ‘

 

 

 

4.0E-8

2.0E-8 ~ --

 

0.0E+0

 

O

9 12 15 18 21 24
Crack length (inch)

 

 

Figure 3-24 Variation of the capacitance parameters from equivalent circuit analysis for
the large specimen from the rebar to external sensor

53

3.2.3 Detecting the location of debonded regions

While impedance measurements between the rebar and an external sensor provide a
global assessment of the debonded area, they are incapable of revealing the location of
debonded regions. However, impedance measurements between pairs of internal wire
sensors (W1) can be used to detect debonded regions. Measurements between internal
sensors were taken for the large eight-foot long beam as well as for a two-foot long
specimen as the internal crack between the CF RP and concrete was propagated.

3.2.3.1 Compafrison of raw impedance spectra

Impedance measurements were made between adjacent sensors (such as W1 and W2,
W2 and W3, and so on). Figures 3-25 to 3-28 show the variation of measured impedance
at different crack states. Figure 3-25 shows the measured impedance from the eight-foot
long specimen when the crack between the CFRP and concrete was about 8 inches long
with the crack tip between sensors W2 and W3. At this crack length, W1 and W2 were in
the debonded region and the remaining sensors were in the bonded region. The measured
impedance between sensors in the debonded region (W 1W2) had the highest magnitude
especially in the low-frequency region. The measured impedances between sensors in the
bonded region (W3W4, W4W5, and WSW6) had the lowest magnitudes. The same
distinguishable trends on the impedance measurements were observed with a two-foot
long specimen. Figure 3-26 shows the measured impedance from two-foot long specimen
when the crack was about 12 inches long with the crack tip between sensor W3 and W4.
At this crack length, W1, W2, and W3 were in the debonded region and W4, W5, and W6
were in the bonded region.

Measured impedances between a sensor in the debonded region and a sensor in the

bonded region (eg., W1W4 and W2W3 from Figure 3-25 and W3W4 from Figure 3-26)
54

have magnitudes that fall inbetween those corresponding to sensors in the debonded
region and those corresponding to sensors in the bonded region. Figure 3-27 shows the
Nyquist plot corresponding to Figure 3-26. On the Nyquist plot, the impedance
measurements from the sensors in the debonded region yield the largest arc radii while
the impedance measurements from the sensors in the bonded region produce the smallest
arc radii. When one sensor is in the bonded region and the other in the debonded region
(W3W4), the arc radii is in between those for the other two cases.

Figure 3-28 shows the measured impedances for the two-foot long specimen when
the CFRP was debonded to 16 inches with the crack tip between the W5 and W6 sensors.
Again, impedance magnitudes between sensors in the debonded region (W 1W2, W2W3,
W3W4, and W4W5) are much higher than those in the bonded region (W6W7, W7W8,

and W8W9), especially at low frequencies.

55

 

 

 

 

 

 

 

 

1E+9 +W1W2 +W1W4
+ W2W3 —+—W3W4
g +W4W5 +W5W6
15.
a: ______________ p p -- -
§ =§§§
1: \ \\
8' 1E+6 _ tfihfi— ~. \\__- 72
.§ § §
8’ §\ \
_, 1E+5 6 § \
1E+4 ._.-,_ i l ' l - ._.___.__r - I u 22,]
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5
Log frequency (Hz)

 

 

 

Figure 3-25 Bode magnitude plot of impedance spectra measured between pairs of
internal sensors for eight-foot specimen in ambient condition when crack tip was
between sensors W2 and W3 (legend shows the internal sensors used)

 

    

 

 

 

 

 

 

 

1E+10 _ +w1w2 +w2w3
+ w3w4 -x— w4w5
a 1E+9 + W5W6
E
.C
8
8 1E+8
C
as
8
o. 1E+7 ~ -----
.§
8’
-| 1E+6
1E+5 I I fl I I
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5
Log frequecy (Hz)

 

 

 

Figure 3-26 Bode magnitude plot of impedance spectra measured between pairs of
internal sensors for two-foot specimen in ambient condition when crack tip was
between sensors W3 and W4 (legend shows the internal sensors used)

56

 

-1E+9 l -—— - —————— .. . ,

 

 

 

 

TI?

E /-\\

5

7.: / __ a , , _p,
o—8E+8 . ,_ .

c

cu \

8 \
‘5" +w1w2

a E+8 fl +w2w3

g4 +w3w4 8‘
g _3/ / ' ‘3’ —x—w4w5

_ is? +w5w6

 

 

 

 

 

0E+0 l 1.. l 1 l l l

0.0E+0 5.0E+8 1.0E+9 1.5E+9 2.0E+9 2.5E+9 3.0E+9
Real impedance (ohms)

 

 

 

Figure 3-27 Nyquist plot of impedance spectra measured between pairs of internal
sensors for two-foot specimen in ambient condition when crack tip was between
sensors W3 and W4 (legend shows the internal sensors used)

 

 

 

 

 

 

 

 

1E+1O « WA ,, ,4 +W1W2 +W2W3
+ W3W4 + W4W5
a 1E+9 +W5W6 + W6W7
g + W7W8 —e— W8W9
9, 1E+8 ~ * e W ___,
(D
‘c’
m 1E+7
'0
a:
G.
g 1E+6 ___..-
8’
_.l 1E+5 g ___ .._ . __ __ _ _ __ ___ __ "___.-- _-_
1E+4 “L "T "T T" T i " l ‘ "T " " T" i " ””1
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5

Log frequency (Hz)

 

 

 

Figure 3-28 Bode magnitude plot of impedance spectra measured between pairs of
internal sensors for eight-foot specimen in ambient condition when crack tip was
between sensors W5 and W6 (legend shows the internal sensors used)

57‘

3.2.3.2 Resalts from equivalent circait analysis

While raw impedance measurements are sufficient to detect debonded regions,
equivalent circuit analysis also can be used. The effect of debonding on the impedances
measured between internal sensors is best observed with the CPE parameter which
characterizes the low-frequency behavior of the impedance spectrum. The variation of the
CPE values corresponding to the spectra in Figure 3-25 are shown in Figure 3-29 (crack
tip was between the W2 and W3 sensors). Figure 3-30 shows the variation of the CPE
values corresponding to the spectra in Figure 3-26 (crack tip between the W3 and W4
sensors). The magnitude of the CPE parameter corresponding to measurements between
pairs of sensors in the debonded region are about two orders of magnitude higher than

those corresponding to pairs of sensors in the bonded region.

58

 

1E+10 T’

I
1E+8 Z

1E+7

Log CPE parameter (1/farad)

1E+6

l
1E+5 *—~

 

1E+Qi ,

  

w1w2

 
  

w1 w4 w2w3 w3w4

 

Figure 3-29 Variation of the CPE parameter for pairs of internal sensors for the 8-foot
beam when crack tip was between sensors W2 and W3 (name of the bar shows

the internal sensors used)

 

I 3.0E+9 ,

2.5E+9 1i

2.0E+9 —

1.5E+9 e

CPE parameter (1/Farad)

5.0E+8 _

 

0.0E+O i

1.0E+9 ~L—~-

 

 

 

 

 

 

 

 

 

 

 

 

 

w1w2

T

w2w3 w3w4

 

Figure 3-30 Variation of the CPE parameter for pairs of internal sensors for the 2-foot
long beam when crack tip was between sensors W3 and W4 (name of the bar

shows the internal sensors used)

59

 

 

3.2.4 Empirical Relationships
3.2.4.1 Rm impedance spectra

Measured impedances between the rebar to an external sensor in the high-frequency

region (100,000 - 10,000 Hz) show a remarkable good fit with the empirical formula.

|Z| = al ”1301101“ “2 (3‘5)

where 31, 8, a2 are coefficients obtained by curve fitting
x is the crack length divided by the total length of the beam

Figures 3-31 and 3-32 show curves fitted to the impedance measured at 100,000 Hz.
Table 3-1 shows the parameters for the fits for the eight-foot long beam and two-foot
long beams in the high-frequency range. Unfortunately, the parameters in Equation (3-5)
vary considerably from one fit to another and hence it is difficult to recommend values

that can be used for diagnosis purposes.

60

Table 3-1 Parameters of the curve fitted to measured impedance in the high-frequency

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

range
Frequency a1 [3 a2 a1 8 a;
(KHZ) Eight-foot specimen Two-foot specimen 2
100 10.25 5.29 32.02 26.94 6.09 840.00
64 4.13 8.06 60.07 30.98 6.24 1098.00
40 1.78 11.38 94.56 37.34 6.35 1484.00
25 0.94 14.54 143.16 47.67 6.43 2056.00
16 0.79 16.31 214.78 63.75 6.47 2914.00
10 0.76 17.82 322.67 89.42 6.48 4197.00
Two-foot specimen 4 Two-foot specimen 6
100 1034.00 1.78 -583.00 28.36 6.26 705.29
64 700.00 2.39 45.00 30.39 6.54 928.79
40 619.00 2.79 483.00 36.49 6.68 1257.00
25 661.00 3.01 927.00 49.16 6.68 1729.00
16 815.00 3.11 1483.00 71.00 6.60 2424.50
10 1109.00 3.14 2257.00 109.62 6.46 3443.00
Two-foot specimen 3 Two-foot specimen 5
100 272.00 2.26 263.00 219.00 2.89 369.00
64 389.00 2.17 275.00 294.00 2.85 448.00
40 600.00 2.03 258.00 405.00 2.79 570.10
25 638.00 2.30 606.00 566.00 2.73 763.60
16 1544.00 1.76 101.00 817.13 2.65 1040.00
10 2328.00 1.73 104.00 1157.00 2.62 1509.00

 

61

 

 

1400 _ ,, .-.-____- ,7 .- -.A,_______.__._--,- .. - -
1200 j

 

1000 r ,_ 18-2 ' * ”whims“ ‘
800

 

600 —
400

 

  

200 q ~~~———»~

Magnitude of impedance (ohms)

 

i

 

T l

0.2 0.4 0.6 0.8
Crack length /Total length of the specimen

0

 

 

 

Figure 3-31 Fit of the measured impedance for 8-foot long beam from rebar to external
sensors at 100,000 Hz: Dots are the measured data and the solid line is the fitted
curve.

 

 

000 T

combs-01
001
GOO
GOO

 

 

l

01

awe -~ +~

N

U"

C

o
L
l

 

 

2000
1 500

 

 

 

—L
O
O
O

 

 

 

 

Magnitude of impedance (ohms)

01
O
O

 

O

 

l T l l f T l l

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Crack length / Total length of the specimen

a_mn_ap_p _ pauaflm_a l

O

 

Figure 3-32 Typical fit of the measured impedance for two-foot long beam from rebar to
external sensors at 100,000 Hz: Dots are the measured data and the solid line is
the fitted curve.

62

3.2.4.2 Equivalent circuit a__nalysis

The C1+C2+C3 parameter from equivalent circuit analysis is also fitted with Eq. 3-5
and the fits are shown in Figures 3-33 and 3-34. Table 3-2 shows the parameters in Eq. 3-

5 for the fitted curves for the eight-foot beam and the two-foot beams.

Table 3-2 Parameters of the curve fitted to (C1+C2+C3) from equivalent circuit analysis

 

 

 

 

 

 

 

Specimen a1 B 32
Large 1.75E-07 -39.083 2.43E-08
52 6.44E-08 -0.384 -4.26E-08
S4 3.16E-08 -2.228 -8.39E-10
S6 3.70E-08 1.507 -6.90E-09
S5 3.72E-08 -1.056 -9.28E-09
S3 3.87E-08 -3. 135 8.96E-09

 

 

 

 

 

 

63

 

255-7 I bank“. -2- V ,

2.0E-7 -~

 

1.5E-7 1

1.0E-7 * '

5.0E-8 * W *

CI +C2+CS parameter (Farad)

 

1)
1)
1)

 

l 0.0E+OIW — W W I W
I 0.00 0.05 0.10 0.15 0.20 0.25 0.30

i Crack length /T0ta| length of the specimen

L___~_____ _ WW. _.___--cw "h_ _ a __- ___fi,-.._,-_ -_.___ .

 

Figure 3-33 Fit for (C1+C2+C3) parameters for 8-foot beam: Dots are the measured data
and the solid line is the fitted curve.

 

3.0E-81 ‘7- ~ , - 0

1°

01

II."

m
1

2.05-8 -

 

1.5E-8

 

1 .0E-8

 

 

5.0E-9

C1 +C2+CS parameter (Farad)

 

 

0.0E+0 1 I I n
0 0.2 0.4 0.6 0.8

Crack length / Total length of the specimen

 

 

 

Figure 3-34 Typical fit for (C1+C2+C3) parameters for two-foot beam: Dots are the
measured data and the solid line is the fitted curve.

64

3.2.5 Summary and Discussion

The effect of CFRP debonding was studied using impedance measurements from the
rebar (R) to the copper tape sensors (Ci), and from the internal wire sensors (W5) to the
c0pper tape sensors. Impedances from two-foot (medium) concrete specimens were
measured in a refrigerator and in an ambient condition at the initial crack state (4” crack
length) and at extended crack states. The large concrete beam was subjected to an
ambient condition and the crack was propagated in the ambient environment. The global
effect of interfacial cracking is best observed with the impedance measured from the
rebar to the external sensor (copper tape). The impedances measured between internal
sensors (stainless steel wires) are used to detect the location of debonded regions.
Impedance measurements at high frequencies (100,000 - 1000 Hz) yield data that have
the least sample to sample variation, and display an exponential variation with the
debonded area of CFRP.

Equivalent circuit analysis also is effective in studying CFRP debonding. The
capacitance parameters correlate best with the debonded area of CFRP. It is found that
the CPE, C1, C2+C3 and C1+C2+C3 values from equivalent circuit analysis all correlate
well with the crack length. The relationship between the C2+C3 and C1+C2+C3 values
and the debonded area of CFRP are approximately linear and show low sample-to-sample
variation. These measures have the most promise for detecting the extent of CFRP
debonding. The CPE parameter increases and the capacitance parameters decrease as the
crack is propagated.

Impedance measurements taken between pairs of internal sensors are capable of
revealing the location of regions where the CFRP is debonded from the concrete.

Measurements between pairs of sensors in debonded regions yield high impedances in the

65

low-frequency region, while those between pairs of sensors in bonded regions yield low
impedances. The CPE parameter from equivalent circuit analysis also is a strong
indicator. The CPE value corresponding to measurements between pairs of sensors in the
debonded region is about two orders of magnitude higher than that corresponding to pairs
of sensors in the bonded region. It is postulated that the reason for this distinctive
difference in the magnitude of impedance is the path of current flow in each case. When
the impedance measurement is made between sensors in the debonded area, current must
flow through the concrete, and the impedance is high. On the other hand, when
measurements are made between sensors in the bonded area, current probably flows
predominantly through the highly conductive CFRP sheet, with charge transfer between
the internal sensors and the CFRP sheet, resulting in lower impedances.

The mpedance in the high-frequency range for measurements from the rebar to
external sensors and the C1+C2+C3 values are also fitted with the exponential empirical
equation in Equation (3-5). Although this equation provides an excellent fit to the data,
the parameters of the fitted equation vary widely from specimen-to-specimen and
between the two-foot and eight-foot specimens. therefore, it is difficult to recommend
parameter values that might used for diagnosis of the debonded length without first

having historical data as the debonded progresses.

66

3.3 Sensitivity of Sensor Measurements to Environmental
Effects

In Section 3.2 it was demonstrated that the measured impedances in the high
frequency range and the C1+C2+C3 parameter from the equivalent circuit analysis
correlate well with debonding between the CFRP reinforcement and the concrete
substrate. In order to use these as indicators of the extent of debonding, environmental
effects on the measured impedance were studied.

3.3.1 Humidity

Experiments were performed to determine the effect of humidity on the measured
impedance over the frequency range from 0.1 Hz to 100 kHz.
3.3.1.1 Comparison of raw impedan_ce spectra

Figures 3-35 and 3-36 show typical impedance spectra (log frequency vs log
impedance) at different humidity levels. Different humidity levels affect the magnitude of
the impedance in the low frequency region regardless of the chloride content. It should be
noted that the magnitude of impedance at a relative humidity of 60% is lower than the
magnitude of impedance at a relative humidity of 90% for the samples with chloride
(Figure 3-35). The reason for this was that the controlled environmental chamber was
malfunctioning at the time of these measurements.

Different humidity levels have a significant influence on the impedance
measurements in the low—frequency region (0.1-10 Hz). Moisture in the concrete
specimen serves as an electrolytic medium for transporting charges between sensors. At
low excitation frequencies, the current flow in the electrochemical cell (concrete

specimen) is dominated by the movement of ions (diffusion controlled), and different

67

moisture levels affects the rate of ion transport. However, it is found that different
humidity levels have less of an influence on the impedance measurements in the high-
frequency region (1000-100 kHz). The impedance is governed by the kinetics of the
charge transfer processes at the electrode-electrolyte interface at high frequencies. This
charge transfer process is influenced less by environmental conditions. The impedance
measurements from the mid-frequency range (10-1000 Hz) are somewhat sensitive to
moisture effects because at these frequencies charge transfer occurs both by diffusion and
kinetics at the interfaces.
3.3.1.2 Resultsfrom equivalent circuit analysis

The changes of moisture content in the concrete specimen influence the CPE, C1 and
C2+C3 parameters from the equivalent circuit analysis. The variation of the C1, CPE,
C2+C3 and C1+C2+C3 parameters are shown in Figures 3-37 to 3-40. Different humidity
levels have a significant influence on the CPE parameter and have less influence on the
C1 and C2+C3 parameters in the equivalent circuit analysis. Although both the C1 and
CPE parameters represent the changes of measured impedance in the low-frequency
range, it is found that the C1 parameter is less sensitive and the CPE parameter is more
sensitive to humidity variation. The value of the CPE parameter decreases with
increasing water content in the concrete specimens and the value of capacitance
parameters (C1, C2+C3, and C1+C2+C3) increase with increasing water content. The
C1+C2+C3 parameter also increases slightly as the humidity level increases.
3.3.1.3 Sammary and Discpssion

A change in humidity over a period of time results in a change in the water content in
the concrete specimens. Higher levels of water content lower the magnitude of

impedance, and lower levels of water content increase the magnitude of impedance in the

68

concrete specimen. Since the impedance measurements at low frequencies are governed
by the diffusion of ions which is affected by the water content, the water content in the
concrete specimen most strongly influences the measured impedance in the low-

frequency range.

69

 

+30% +60% +90%

   

 

 

 

1.0E+6 -

1.0E+5 —

Log impedance (ohms)

1.0E+4 W

 

 

1.0E+3 -.. » r I . e 1
1E-1 1 E+0 1E+1 1E+2 1E+3 1 E+4 1 E+5
Log frequency (Hz)

        

l' '* 'I”""' "W” a I

 

 

 

Figure 3-35 Bode magnitude plot of typical impedance spectra for specimens with
chloride at different humidity levels

 

 

 

 

 

 

 

 

 

 

1E+3 +30% +60% —.—90%M§1
A -

‘é’ 1E+7

.C

3

8 1E+6

C

m

E. 1E+5 1—

.§

0)

3 1E+4 -

IE+3 4» I- w l, ~~I ~ - ., 1W —————— 1
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5
Log frequency (Hz)

 

 

 

Figure 3-36 Bode magnitude plot of typical impedance spectra for specimens without
chloride at different humidity levels

70

 

 

 

 

 

 

 

 

 

+CTHS1 +CTH82
3.0E+8 -----. .
+CTH83 +NTHS1
E 2.5E+8 * +NTH82 +NTH83
u‘?
E 2.0E+8 W W
E 15E 8
o . +
5 \~ —x
g 1.0E+8 -‘ WW * WW" W "W * '
Hi
85 5.0E+7 \
0.0E+0 , W ~ PW {WWW I
20 40 60 80 100
Relative humidity (%)

 

 

 

Figure 3-37 Variation of the CPE parameter from equivalent circuit analysis due to
different humidity levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C-chloride, N-non-chloride, T-
temperature, H-humidity, Sn-specimen number)

 

 

4.0E-9

 

 

 

 

 

 

 

 

 

 

 

 

... 3.5E-9

3

LI; 3.0E-9 W

39'. ' '/ /

g 2.5E'9 F / . ‘4

(3.2.0E-9 _.2 ._._ )/ __ _. ___ __ _ _._ _ ___ ._~.__ .__

0 1515-9 +CTHS1 +CTHS2 +CTH83 ~—

—-x— NTHS1 + NTHS2 + NTHSS
1.0E'9 l I I I
20 40 60 80 100

Relative humidity (%)

 

 

 

Figure 3-38 Variation of the C1 parameter from equivalent circuit analysis due to
different humidity levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C-chloride, N-non-chloride, T-
temperature, H-humidity, Sn-specimen number)

71

 

 

 

 

 

 

 

 

 

 

 

 

 

 

' +NTHS1 +NTH32 _._NTH33
’6 3.5E-8
E
I; 3.0E-8 Wt ’-
f1} 2.5E-8 ~
2
9 2.0E-8
8 1.5E-8
(‘0
9 1.0E-8
0 5.0E-9— 4r Wm “___-
0.0E+0 W— -—— I , __ _ _H]
20 40 60 80 100
Relative humidity (%)

 

 

 

 

 

Figure 3-39 Variation of the C2+C3 parameter from equivalent circuit analysis due to
different humidity levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C—chloride, N-non-chloride, T-
temperature, H-humidity, Sn-specimen number)

 

4558 __g +CTHS1 + CTHSZ + CTHS3 __

4.0E-8 _R —-x— NTHSl + NTHSZ + NTHS3
3.5E-8
3.0E-8
2.5E-8
2.0E-8
1.5E-8 _ W WWW
1.0E-8
5.0E-9

0.0E+0 T r v 1
20 40 60 80 100

Relative humidity (%)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

C1 +02+CS parameter (Farad)

 

 

 

 

 

 

Figure 3-40 Variation of the C1+C2+C3 parameter from equivalent circuit analysis due
to different humidity levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C—chloride, N-non-chloride, T -
temperature, H-humidity, Sn-specimen number)

72

3.3.2 Temperature

Experiments were performed to determine the influence of temperature effects on the
measured impedance over the frequency range from 0.1 Hz to 100 kHz. Note, however,
that the humidity level was not kept constant so the moisture content in the specimens are
likely to have been different at the different temperatures.

3.3. 2.1 Comparison of raw impedance spectra

Figures 3-41 and 4-42 show typical impedance spectra at different temperatures. The
slope of the impedance magnitude changes at frequencies below 1000 Hz and becomes
zero in the mid- and low-frequency region at the high temperature of 100°F. At 70°F, the
slope of measured impedance becomes smaller at frequencies below 10 Hz compared to
the measured impedance at 100°F. The impedance measurements are affected in the low-
and mid-frequency regions as the temperature varies from 70 0F to 100 0F.

Measured impedances are also analyzed at specific frequencies. Figures 3-43 to 3-45
show the variation of measured impedance at different temperatures at frequencies of
10,000 Hz, 100Hz, and 1 Hz. 10,000 Hz, 100 Hz, and 1 Hz are arbitrary representative
frequencies for high-(100,000 - 1000Hz), mid- (1000 — 10 Hz), and low-frequency (10 —
0.1 Hz) ranges, respectively. The log of the magnitude at 10,000 Hz and 1 Hz show a
reasonably linear trend with increasing temperature, but the trend is more nonlinear at
100 Hz. At 100 Hz the change in temperature from 00F to 70°F has a much smaller

influence of the impedance magnitude than the change in temperature from 70°F to

100°F.

73

 

1E+9 .._- .--.-———- —-~——~-—- _____ ll + OF + 70F + 10°F }

 

 

_.L
m
+

m

 

Log impedance (ohms)
a
+
O)

rn
+
.b
l
l
II
I/
l

 

1E+3 , e r -- if ~ - . r» T »- TM". ~ ,
1 E-1 1 E+0 1 E+1 1E+2 1E+3 1 E+4 1 E+5
Log frequency (Hz)

 

 

 

Figure 3-41 Bode magnitude plot of typical impedance spectra for specimens with
chloride at different temperature levels

 

 

 

 

 

 

 

 

 

1E+9{ +0F +7OF +1OOF *
7,; 1E+8 4
E
'8 1E+7 ~
I; \\\
8 \\
g 1E+6 , - m—M-w \\ “_-___zw .. a e
O. \\
.E 1E+5 - W ~ , r - \__s ~
8’ ‘§§
"' 1E+4 W W W W WWW -‘ W
‘h
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5
Log frequency (Hz)

 

 

 

Figure 3-42 Bode magnitude plot of typical impedance spectra for specimens without
chloride at different temperature levels

74

 

1.8E+4 Them ---..._ ,. I—O—CS1 +082 +CS3
16E+41r~ _ l—x—NS1 +N82 +N83

1.4E+4 I A
1.2E+4 - -
E 1.0E+4 .
8.0E+3
6.0E+3
4.0E+3
2.0E+3

0.050 r w . w . .
-10 10 so so 70 90 110

Temperature (F)

 

 

ance (ohms)

 

 

 

 

Magnitude of imp

 

 

 

 

 

Figure 3-43 Variation of typical measured impedance in the high-frequency range at
different temperature levels at 10,000 Hz (legend gives the name of the specimen:
C-chloride, N-non-chloride, Sn-specimen number)

 

 

 

 

 

 

 

 

 

 

 

9E+5 I +CS1 +082 +CS3
9" 7‘ 7 * l+NSl +NSZ +NS3
I 7E+5 - 1 ‘W
o 5‘
g 65-1-5 “
g SE+5 “
g 45.5 ~ --
o
a: 3E+5 ___
S
.1: 2E+5 W
C
§15+5~
0E+0 T T Y r f 1
-1O 10 30 50 70 90 110
I Temperature (F)

 

 

Figure 3-44 Variation of typical measured impedance in the mid-frequency range at
different temperature levels at 100 Hz (legend gives the name of the specimen: C-
chloride, N-non-chloride, Sn-specimen number)

75

 

 

35.7 + C81 + C82 + C83
7E+7 '\ __J -x— N81 + N82 + N83

6E+7 §\

5E+7 \

 

 

 

 

4E+7
3E+7 w
2E+7
1E+7-W

0E+0 r Y
-10 1 0 30 50

Temperature (F)

 

 

Magnitude of impedance (ohms)

 

 

 

 

 

 

Figure 3-45 Variation of typical measured impedance in the low-frequency range at
different temperature levels at 1 Hz (legend gives the name of the specimen: C-
chloride, N-non-chloride, Sn-specimen number)

76

3.3.2.2 Results from equivalent circuit gnalysis

The variation of the CPE, C1, the C2+C3 and the C1+C2+C3 parameters with
temperature are shown in Figures 3-46 to 3-49. It should be noted that the change in
temperature from 00F to 700F has a smaller influence than the change in temperature from
70°F to 100°F on all the capacitance parameters. On the other hand, the change in
temperature from 00F to 70°F has a larger influence on the CPE parameter than the
change in temperature from 70°F to 1000F. In general, different temperatures have a
significant influence on the CPE parameter and have less influence on the C1, and the
C2+C3 and C1+C2+C3 parameters in the equivalent circuit analysis. The value of the
CPE parameter decreases with increasing temperature in the concrete specimens and the
value of capacitance parameters (C1, C2+C3) increase with increasing temperature. The
C1+C2+C3 parameter also increases slightly as the humidity level increases.
3.3.2.3 Spmmamnd Discyssion

Change in temperature in the concrete specimen has a significant influence on the
impedance measurements. At very low temperatures 00F, water in the concrete is usually
in a super-cooled state significantly slows down the diffusion of ions. This results in the
magnitude of impedance being very high. On the other hand, high temperatures increase

the mobility of ions and lower the magnitude of impedance.

77

 

 

 

 

 

 

 

 

”5+9 __ _ _ _____ I+031 +032 +css
' L+N81 +N82 +NS3
€1.0E-I-9 "’77 7 ’7 A
(U
E 80E 8 \
v . +
E, : \..
g 6.0E+81 _
‘3. 4.0E+8 \\
LU \.
o. \
0 2.0E+8 \\
\\
0.0E+O -4 . r W \_ - '-
o 20 4o 60 so 100 120
Temperature (F)

 

 

 

Figure 3-46 Variation of the CPE parameter from equivalent circuit analysis at different
temperature levels for measurements from the rebar to external sensors (legend

gives the name of the specimen: C-chloride, N-non-chloride, Sn-specimen
number)

 

+ CS1 + C82 + CS3

1.6E-8 W-
1.4E-8 +N81 +NS2 +N83 A

1.2E-8W ~ - , / ~
1.0E-8

8.0E-9
6.0E-9
4.0E-9
2.0E-9

 

 

 

 

 

 

 

 

 

 

C1 parameter (Farad)

 

 

 

0 20 40 60 80 1 00 1 20
Temperature (F)

 

 

 

Figure 3-47 Variation of the C1 parameter from equivalent circuit analysis at different
temperature levels for measurements from the rebar to external sensors (legend
gives the name of the specimen: C-chloride, N-non-chloride, Sn-specimen
number)

78

 

+C81 +C82 +CS3
+NS1 +N82 +N83

 

3.0E-8

 

 

 

 

2.5E-8 - , W

2.05-3 M
1 .5E—8

 

 

CZ+C3 parameter (Farad)

 

O 20 4O 60 80 1 00 1 20
Temperature (F)

 

 

 

Figure 3-48 Variation of the C2+C3 parameter from equivalent circuit analysis at
different temperature levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C-chloride, N-non-chloride, Sn-specimen
number)

 

+031 +032 +C83 7
-x—N81 +N82 +N83

3.5E-8
3.0E-8
2.5E-8‘
2.0E-8
1.5E-8 _.
1.0E-8 r
5.0E-9 r

0.0E+O ewe—T ~ ,- e. e ,, ,, -. s ., ,
.10 10 so so 70 90 110

Temperature (F)

 

 

 

 

 

 

 

 

C1 +CZ+CS parameters (Farad)

 

 

 

 

Figure 3-49 Variation of the C1+C2+C3 parameter from equivalent circuit analysis at
different temperature levels for measurements from the rebar to external sensors
(legend gives the name of the specimen: C-chloride, N-non-chloride, Sn-specimen
number)

79

3.3.3 Chloride content

The experiment to study the impact of chloride content was conducted
simultaneously with the experiment to study humidity and temperature effects.
3.1.3.1 Comparison of raw impedance spectra

Figures 3-50 to 3-52 show the measured impedance spectra over the entire frequency
range at different humidity levels. Figures 3-53 to 3-55 show the impact of chloride
content at different humidity levels on the measured impedance at specific frequencies.
10,000 Hz, 100 Hz, and 1 Hz are arbitrary frequencies that represent high-(100,000 —
1000 Hz), mid- (1000 - 10 Hz), and low-frequency (10 — 0.1 Hz) ranges, respectively.
The lowering of the magnitude of impedance in the presence of chloride was observed in
the high- and low-frequency range at all three different humidity levels (30, 60 and 90%).
However, in the mid-frequency range, the presence of chloride ions has a smaller
influence on the measured impedance at the three humidity levels.

The combination of chloride content and different temperatures has less influence on
the measured impedance. Figures 3-56 to 3-58 show the measured impedance spectra
over the entire frequency range at different temperature levels. Figures 3-43 to 3-45 show
the impact of chloride content on the measured impedance at three temperatures at
specific frequencies. The presence of chloride shows a clear influence on impedance only
in the high frequency range at 00F and 70°F. Nevertheless, the effect of chloride content

on the measured impedance at different temperatures is relatively minor

80

 

15+9 ~——

 

+ CS1-30% + N81-30%

 

 

 

 

 

 

 

 

 

 

 

 

7,; 1E+8
E
.c
3 1E+7
m
8 E 6
as 1 +
8
:3.
_§ 1E+5
or
o
-’ 1E+4
1E+3 “r
1E-1

 

1E+O

l l f T

1E+1 1E+2 1E+3 1E+4 1E+5

Log frequency (Hz)

 

 

Figure 3-50 Bode magnitude plot of typical impedance spectra for specimens
with/without chloride at a relative humidity of 30% (legend gives the name of the
specimen: C-chloride, N-non-chloride, Sn-specimen number)

 

_L
m .
+
N
A.

_L
m
+
O)
I

1E+5 _ *

1E+4 _

Log impedance (ohms)

_;
m
.1.

OD

    

—+— CS1-60% + N81 -60%

 

W1

l.

l

 

 

 

1E+2

 

15-1

 

l

1E+0

T l I l l

1E+1 1E+2 1E+3 1E+4 1E+5

Log frequency (Hz)

 

 

Figure 3-51 Bode magnitude plot of typical impedance spectra for specimens
with/without chloride at a relative humidity of 60% (legend gives the name of the
specimen: C-chloride, N -non-chloride, Sn-specimen number)

81

 

1E+8 ----—‘ as, +cs1-90% +NS1-90%

 

 

Log impedance (ohms)

 

 

1E-1 1E+0 1E+1 1E+2 "5+3 1E+4 1E+5

Log frequency (Hz)

 

 

 

Figure 3-52 Bode magnitude plot of typical impedance spectra for specimens
with/without chloride at a relative humidity of 90% (legend gives the name of the
specimen: C-chloride, N-non-chloride, Sn-specimen number)

 

 

‘1 co co
”1 ”I "I
+ + +
on w on
4 l l

l

 

 

 

 

 

Magnitude of impedance (ohms)

 

 

 

 

6E+3 1W— ’» —- - — —
55+3_ms f----“-~:~*~ur.z... _a
4E+3 , +081 +C82 +083
—x— N81 + N82 + N83
3E+3 1 ~---—-——~--—---e—-—r~~n— —~ T em .---“T—fim‘ ‘_ _,
20 40 60 80 100
Relative Humidity (%)

 

 

 

Figure 3-53 Variation of typical measured impedance in the high-frequency range at
different humidity levels at 10,000 Hz (legend gives the name of the specimen: C-
chloride, N-non-chloride, Sn-specimen number)

82

 

N
ITI
+

(II

 

 

6E+5

5E+5

 

/;
4E+5 W

W
/

 

 

 

Magnitude of impedance (ohms)

 

 

 

 

 

 

3E+5 + CS1 + 082 + CS3 ._.___...
+ N81 + N82 + N33
2515 1 - W 1 , _,_ , 1 . “To”- - - 1
20 4O 60 80 100
Relative humidity (%)

 

Figure 3-54 Variation of typical measured impedance in the mid-frequency range at
different humidity levels at 100 Hz (legend gives the name of the specimen: C-
chloride, N-non-chloride, Sn-specimen number)

 

 

 

 

4E+7 “
+081 +082 +083

\\;x—NS1 +N82 +N83

1E+7 \ fi/

OE+0 l i l l
20 4O 60 80 1 00

Relative humidity (%)

 

7

 

 

 

N
m
+

N

l .

#3!
n

\

Magnitude of impedance (ohms)

 

 

 

 

 

Figure 3-55 Variation of typical measured impedance in the low-frequency range at
different humidity levels at 1 Hz (legend gives the name of the specimen: C-
chloride, N-non-chloride, Sn-specimen number)

83

 

1E+9 + CSZ-OF + N82-0F

 

 

  
  

 

1E+8 -

 

ohms)

v 1E+7

 

1E+6 ~

 

 

1E+5

Log impedance

1E+4 ~

 

1E+3— I

l"’ ' 'l M l

1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5

Log frequency (Hz)

 

 

 

Figure 3-56 Bode magnitude plot of typical impedance spectra for specimens
with/without chloride at a temperature of 00F (legend gives the name of the
specimen: C-chloride, N-non-chloride, Sn-specimen number)

 

1E+9 ' * F F ‘ * E—o—csz-mF +st-70F

L

1E+8

 

1E+7

 

 

1E+6 e

1E+5

 

 

Log impedance (ohms)

_.L
m
+

.b

 

 

 

1E+3 “i "" "*i' l ' " "l “'I"“" ‘ i
1E-1 1E+0 1 E+1 1E+2 1E+3 1E+4 1E+5

Log frequency (Hz)

 

 

L

Figure 3-57 Bode magnitude plot of typical impedance spectra for specimens
with/without chloride at a temperature of 70°F (legend gives the name of the
specimen: C-chloride, N-non-chloride, Sn-specimen number)

84

 

 

flfimu.

1E+6 - ~~-~—--m-- +osz—100F +N82-100F

 

 

 

 

 

...;
m
+

U1

 

_.L
m
+

.b

 

Log impedance (ohms)
a
+
00

 

 

1E+2 ‘i‘ ’ i l l "‘i

l

1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5

Log frequency (Hz)

 

 

 

Figure 3-58 Bode magnitude plot of typical irn edance spectra for specimens
with/without chloride at a temperature of 100 (legend gives the name of the
specimen: C-chloride, N-non-chloride, Sn-specimen number)

85

3.3.3.2 Results flom equivalent circuit analysis

Figures 3-37 to 3-40 show the effect of chloride content on the equivalent circuit
parameters at different humidity levels. The value of the CPE parameter is decreased by
the presence of chloride ions in the concrete specimens and the value of the capacitance
parameters (C2+C3) is increased. However, there is no clear effect of the presence of
chloride ions on the C1 parameter at different humidity levels. In general, chloride
content decreases the CPE parameter and increases the C2+C3 parameter in the
equivalent circuit analysis.

Figures 3-46 to 3-49 show the variation of the equivalent circuit element parameters
due to chloride content at different temperatures. There is no discernible difference in the
equivalent circuit parameters due to the presence of chloride. The general scatter in the
parameters is within the specimen-to-specimen variation of the parameter values.

3.3.3.3 Summary and Discussion

The chloride ion content in the concrete specimen has a minor influence on the
impedance measurements. The existence of chloride ions in the concrete specimens
affects the mobility of ions. The combination of chloride ions and humidity variation has
a greater impact on the measured impedance than the combination of chloride ions and
temperature variations. In general, the presence chloride ions in the concrete specimens
lowers the magnitude of measured impedance at any humidity level and decreases the

magnitude of the impedance slightly.

86

3.3.4 Freeze — thaw
3.3. 4.1 Compgfison of raw impedancgpectra

The magnitude of impedance before and after 300 freeze-thaw cycles is plotted in
Figure 3-59. The impedance is reduced over the entire frequency range regardless of the
water content in the concrete specimens. However, the differences are most significant in
the low- and mid-frequency ranges. The differences in the weight of the freeze-thaw
specimens before and after freeze-thaw exposure are shown in Table 3-3. The concrete
specimens were put in an oven to dry out the excessive water from the freeze-thaw
experiment. Specimens 1 and 2 contained slightly less water and Specimen 3 contained
slightly more water compared to the time when baseline measurements were taken.
3.3.4.2 Resultsjrom equivalent circuit_gnalysis

Figures 3-60 to 3-63 show the variation of the equivalent circuit parameters
following freeze-thaw exposure. After 300 cycles of freeze-thaw the CPE parameter
dropped significantly-The C1 parameter increased, the C2+C3 parameter decreased, and
the C1+C2+C3 parameter showed little change. The variation on the capacitance
parameters from the freeze-thaw exposure is bigger than that from the short term
environmental. conditions (temperature and humidity). In general, 300 cycles of freeze-
thaw have a significant influence on the CPE parameter and much smaller influences on
the C1, the C2+C3 and the C1+C2+C3 parameter obtained from the equivalent circuit
analysis.
3.3.4.3 Summmnd Discussion

Freeze-thaw cycles generally tend to deteriorate the concrete and increase the width
and density of microcracks. This increases the mobility of ions within the concrete and
hence reduces the magnitude of impedance. The results indicate that the reduction in

87

 

impedance is most significant in the low- to mid-frequency regions. While the CPE, C1

and C2+C3 parameters from the equivalent circuit analysis display significant changes

following freeze-thaw exposure, C1+C2+C3 is generally unchanged.

Table 3-3 Specimen weights before and after freeze-thaw testing

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Weight changes (gram)
Specimen 1 Specimen 2 Specimen 3
Before testing 16724 17938 17117
After testing 16674 17922 17123
Differences -50 -16 6
+ BFI'S1 + AFI'S1

1'0E+9 —+— BFI'SZ + AFT $2
’6?1.0E+87 ‘m—wi -~ 7 ~ +BFIS3 +AFFS3
E
31.0E+7— 11“ e -
8 {I—l I u . _ . ~
g 1.0E+6 ‘ ‘ w R
a) \
o. .
g 1.0E+5
c»
o
-' 1.0E+4

1.0E+3 s e r . , \

1E-1 1E+0 1 E+1 1 E+2 1E+3 1 E+4 1E+5
Log frequency (Hz)

 

 

 

Figure 3-59 Bode magnitude plot of typical impedance spectra from the specimens with
chloride for 300 cycles of freeze-thaw (legend gives the name of the specimen: B-

before, A-after, FT—freeze-thaw, Sn-specimen number)

88

 

 

 

 

** ,1, El Before E] After

 

 

1 .0E-l-9 1

 

 

 

 

 

 

- ————~

 

 

 

1.05.7 J———§g§g§;§;

 

 

 

b
rn
3,

 

Log CPE parameter (Farad)

 

 

 

 

 

 

 

 

 

1.0E+5 ‘i‘535‘3 j ”‘5‘“? .
FI'S1 Frsz

 

 

 

 

Figure 3-60 Variation of the CPE parameter from before and after 300 cycles of freeze-
thaw for measurements from the rebar to external sensors (legend gives the name
of the specimen: FT-freeze-thaw, Sn-specirnen number)

 

_—’_— _—

9E-9 ‘l e k 9* * [flBefore IAfter
8E-9
7E-9 4‘ 4a a ~ 7 . -7 -- - -1 .. A _- __ .
6E-9
55-9 1. -. _- r . - -- _. _, - -1
4E-9
3E-9 - -- ~~~~~ ~ . —-— —- »— __ ———.
1E—9 1.2.2,; 112.2;
0E+0 f 5255533? 1 5323335: I

FTS1 FTSZ FTSS

 

 

 

 

 

 

 

 

 

 

 

C1 parameter (Farad)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3-61 Variation of the C1 parameter before and after 300 cycles of freeze-thaw for
measurements from the rebar to external sensors (legend gives the name of the
specimen: FT-freeze -thaw, Sn-specimen number)

89

 

1 5-3 El Before E: After
9E-9
8E-9 __
7E-9 _, ”2”“
6E-9 4 "
5E-9 -
4E-9 4
3E-9 —:
2E-9 e
1E'9 _ :-:-:-: 5: . ..
OE+0 :Ezfzfz I 3:323:52 I
FI'Sl FTSZ FTSS

 

 

 

 

 

 

 

 

 

 

 

 

 

02+CS parameter (Farad)

     

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3-62 Variation of the C2+C3 parameter before and after 300 cycles of freeze-thaw
for measurements from the rebar to external sensors (legend gives the name of the
specimen: FT-freeze -thaw, Sn-specimen number)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FTS1 FTSZ FTSS

 

Figure 3-63 Variation of the C1+C2+C3 parameter before and after 300 cycles of freeze-
thaw for measurements from the rebar to external sensors (legend gives the name
of the specimen: FT-freeze -thaw, Sn-specimen number)

90

3.3.5 Wetting and drying
3.3.5.1 Comparison oj raw impedance spectra

The magnitude of impedance before and after 300 wet-dry cycles is plotted in Figure
3-64. The concrete specimens were put in an oven to dry out the excessive water after the
wet-dry experiment. The differences in the weight of the wet-dry specimens before and
after wet-dry exposure are shown in Table 3-4. Exposure to 300 wet-dry cycles has a
similar influence on the impedance measurements as exposure to 300 cycles of freeze-
thaw. Although the decrease in impedance is observed over the entire frequency range,
the decrease is largest in the low-frequency range.
3.3.5.2 Results from quvalent circuit analysis

Figures 3-65 to 3-68 show the variation of the equivalent circuit parameters for wet-
dry exposure. After 300 wet-dry cycles the CPE parameter decreased significantly, the
C1 parameter increased, the C2+C3 parameter-decreased and the C1+C2+C3 parameter
did not change significantly. The 300 wet-dry cycles has a similar influence on the
equivalent circuit element parameters as 300 freeze-thaw cycles. One exception was the
WDS3 specimen for which the C2+C3 and C1+C2+C3 parameters were very high before
exposure and decreased after exposure. The reason for this anomaly is unknown.
3.3.5.3 ngmgry and Discussion

Wet-dry cycles increase the width and density of mierocraks due to expansion and
shrinkage of the concrete. This increases the mobility of ions within the concrete and

hence reduces the magnitude of impedance in the mid- and low-frequency regions.

91

Table 3-4 Specimen weights before and after wet-dry testing

 

 

 

 

 

 

Weight changes (gram)
Specimen 1 Specimen 2 Specimen 3
Before testing 17353 16914 16906
After testing 17336 16899 16906
Differences -17 -15 O

 

 

 

 

 

 

+ AWDS1 + BWDS1
+ AWDSZ —o— BWDSZ
+ AWDSS —x— BWDSB

 

     
  

 

 

 

 

 

1E+6

 

Log impedance (ohms)

 

 
 

 

 

—L —L ._L

m "1 III

I: 3 01
1

T T f I I

1E-1 1 E+O 1 E+1 1 E+2 1 [5+3 1 E+4 1 E+5
Log frequency (Hz)

 

 

 

Figure 3-64 Bode magnitude plot of typical impedance spectra for specimens with
chloride before and after 300 wet-dry cycles (legend gives the name of the
specimen: B-before, A-after, WD-wet -dry, Sn-specimen number)

92

 

1E+8 1

El Before

 

 

-L
m
+

m

 

8E+7

6E+7 j --

 

 

CPE parameter (1/Farad)
.5
m
+
\I

25.7 m

 

 

 

 

 

I I I I I

I I I I I
I I I I I

I I I I I
I I I I I

I I I I I
I I I I I I

I I I I I
I I I I I

I I I I I
I I I I I

I I I I g g
I I I I I I

I I I I g g I
I I I I I I

v I O C O . .
I I I I I I I

I I I I I I I I
I I I I I I I I

I I I I I I I I
I I I I p I I I

I I I I I I I I
I I I I I I I

I I I I I I I I
I I I I I I I

I I I O I I I I
I I I I p I I I

I I I I I I I I
I I I I * I I I

a I I - I - I I
I I I I I I I

! I I 4 4 4 l I _

 

B After

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0E+O

 

WDS1

T

j

WDSZ WDS3

 

 

Figure 3-65 Variation of the CPE parameter from before and after 300 cycles of wet-dry
for measurements from the rebar to external sensors (legend gives the name of the
specimen: WD-wet -dry, Sn-specimen number)

7E-9

6E-9

5E-9

4E-9

3E-9

2E-9

6
9
(U
E:
I...
93
Q)
E
E
,—
0

1 E-9

OE+O

 

 

WDS1

 

El Before a After

 

W082 W083

Figure 3-66 Variation of the C1 parameter before and after 300 cycles of wet-dry for
measurements from the rebar to external sensors (legend gives the name of the
specimen: WD-wet -dry, Sn-specimen number)

93

 

4-55-3 ‘ l‘ EflBefore aAfter »- a I
3.5E-8 ~ . v -.----.._ _.
3.0E-8
2.5E-8
2.0E-8 . .
1.5E-8
1'05'8 .

0.0E+0 “535:5 . 525:5‘5‘ , L
WDS1 wosz woss

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

C2+03 parameters (Farad)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3-67 Variation of the C2+C3 parameter before and after 300 cycles of wet-dry for
measurements from the rebar to external sensors (legend gives the name of the
specimen: WD-wet -dry, Sn—specimen number)

 

 

 

l
5E-8 In Before Ki After

 

 

 

 

 

 

4E-8

 

 

3E-8

2E-8 A h »

 

1581 v

01+02+03 parameters (Farad)

 

 

 

 

 

 

 

 

05w :szszsz . 1353..." .
WDS1 wosz wosa

_..-

 

 

 

Figure 3-68 Variation of the C1+C2+C3 parameter before and after 300 cycles of wet-dry
for measurements from the rebar to external sensors (legend gives the name of the
specimen: WD-wet ~dry, Sn-specimen number)

94

3.3.6 Corrosion of reinforcing bar

A constant voltage of 12 V was supplied across the two reinforcing bars to accelerate
the corrosion process using an external power source. The applied current was monitored
during 21 days of the accelerated corrosion test, since this has a direct correlation with
the total corrosion in the reinforcing bar. Steel loss was calculated using the recorded

current using (Philips 1992, PantaZOpoulou 2001).

I*t"'Am
AW -— 3—1
(g) I”? ( )

where AW is the steel loss in grams
I is the total recorded current
t is the duration of the test in seconds
Am is the atomic mass of the metal
2 is valency (assuming that the rust product is mainly
Fe(OH)2, 2 is used for the 2 value)
F is Faraday’s constant
The total steel loss was about 13g for each specimen. This represents an average
corrosion depth of about 0.45mm (3.6% loss) over the bar
3.3.6.1 Compgrison oj rgw impedgnce spear-g
Exposure to accelerated corrosion of the reinforcing bar has a similar influence on
the impedance measurements as exposure to freeze-thaw and wet-dry cycles. The
magnitude of measured impedances before and after accelerated corrosion are plotted in
Figures 3-69. The impedance in the low- and mid-frequency range decreased. The
differences in the weight of the specimens before and after the corrosion test are shown in
Table 3-5. Some of the mass loss is probably due to the diffusion of corrosion products

into the water used to periodically wet the specimens. The concrete specimens were put

in an oven to dry out the excessive water from the corrosion experiment.

95

3.3.6.2 Resylts from equivalent circuit analysis

Figures 3-70 to 3-73 show the changes in the equivalent circuit parameters after the
accelerated corrosion test. The CPE parameter decreased significantly after accelerated
corrosion for 21 days. The capacitance parameters do not change significantly due to
corrosion of the reinforcing bar in the concrete specimens. The C1 parameter increased
slightly, and the C2+C3 and C1+C2+C3 quantities did not change significantly.
3.3.6.3 Summary and Discussion

The electrical properties of the reinforcing bar and surrounding concrete may change
due to corrosion of the reinforcing bar and the presence of corrosion products in the
concrete. The corrosion test was setup to study the effect of corrosion of the reinforcing
bar on the impedance measurements. It is found that the corrosion of the reinforcing bar
significantly lowers the impedance in the low-frequency range, but has a much smaller
influence on the measured impedance in the mid- and high-frequency ranges. As a
consequences, corrosion of the reinforcing bar has a significant influence on the CPE
parameter from the equivalent circuit analysis, and has a much smaller influence on the

capacitance parameters.

96

Table 3-5 Specimen weights before and after corrosion testing

 

 

 

 

 

 

Weight changes (gram)
Sample 1 Sample 2 Sample 3
Before testing 17734 17755 17600
After testing 17723 17681 17564
Differences -1 1 -74 -36

 

 

 

 

 

 

—+— 3081 +ACS1 a

 

 

 

 

 

 

Log impedance (ohms)
a
3,

1E+4

 

 

 

1 E+3 T u T I l 1
1 E1 1 E+0 1 E+1 1 E+2 1E+3 1 E+4 1 E+5

Log frequency (Hz)

 

 

 

Figure 3-69 Bode magnitude plot of typical impedance spectra from the specimens with
chloride for 21 days of corrosion on the rebar (legend gives the name of the
specimen: B-before, A-after, C-corrosion, Sn-specimen number)

97

 

18E+8 '1 7,, 77—7 .2, ,,, 1
ElBefore gAfter
1.6E+8 1
1.4E+8 ~——-—§§ -
1.2E+8 — Z:
1.0E+8 —§—§
8.0E+7 —E:—
6.0E+7
4.0E+7 ———§§ ..
2.0E+7 —§—«§ . .2 :
0.0E+0 ~17 - 53513:???“ 7 Warm
C081 0082

 

 

 

 

 

 

 

 

 

 

 

 

 

CPE parameter (Farad)

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3-70 Variation of the CPE parameter from equivalent circuit analysis for
specimens with chloride before and after 21 days of rebar corrosion for
measurements from the rebar to external sensors (legend gives the name of the
specimen: C-chloride, C-corrosion, Sn-specimen number)

 

l 4.0E-9 '1 l E] Before E3 After ‘

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l ; f
“ 3.5E-9 ‘L_,_._ T -_2_:1:i‘
‘ l
‘ '8 3.0E-9 T___v_ *7 77————- —-————~ J
. a l l
2:3 1 .
1 FE) 2.0E-9 1‘ §
1 “3 I i
1 a 1.5E'9 ' m.‘
l o. f
5 1.0E-9 Lie ll

} 5.0E-10 ~

1 I

‘ 0.0E+0 w

 

1 CCS1

\
l
1

 

Figure 3-71 Variation of the C1 parameter from equivalent circuit analysis for specimens
with chloride before and after 21 days of rebar corrosion for measurements from
the rebar to external sensors (legend gives the name of the specimen: C-chloride,
C-corrosion, Sn—specimen number)

98

 

 

 

 

 
   

4.0E-9 4777—12.

 
  
 
 
    

20.0

00 ,
O.00,00‘ ‘
20 ‘20’0;

0’0’

1
3.0E-9 1
l

 

   

.0
030’ 0’0 0’
0.0

2.0E-9 m1

O O
:‘0’

O
0.6

0
:0
‘02

2

  

02+C3 parameter (Farad)

  

1.0E-9 2-

0
’0

 

rrrr‘w............. [till .... .

 
 
 

00
03:30.
0.0.0.

  

 

 

 

 

 

 

O

O

2

   
 

 

 

 

\ 0.0E+0

CC CCSZ CCS3

(D
_L

 

Figure 3-72 Variation of the C2+C3 parameter from equivalent circuit analysis for
specimens with chloride before and after 21 days of rebar corrosion for
measurements from the rebar to external sensors (legend gives the name of the
specimen: C-chloride, C-corrosion, Sn-specimen number)

 

 

 

 

 

   
   
 

     
    
 
     
  
    
     
 
    
   
   

    
   

 

 

    
    

 
  

 

 

  

 
         
      
     
  

    

    
    
  

 
    
  
  

 
    
 
   
    
  
     
   
  
  

   
 
  

     
  

 

 

   

 
    
   
 
   
   
  

   
  

 
 
       
 
 

  
    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

9E-9 177~— ~77 777 ElBefore KJAfter f .,__ we -7
"5 8E-9
g
9:. 7E-9 "* ’ ’ ’* y ’ * * ’r' .
9
ca : 90¢
5 6E-9 “ ’ <33: ' ’ 1:5,. ’09 ’’’’ ’ 9:9:9 ’
-. . -. 9. ...
‘65 ., , ;. 0.0.0 00
8 =38: —
a :8: 28:.
Q. 4E'9 7 ..... 7 A . ... ,
00 ‘ 0‘0’0
a) 0.0.9 . .9
o 3E-9 - 000 7 7 7 ”’0’
O. ' .9.
+ '0... 5 0’0’0
N ’0’0’ ‘ 0’0‘0
0 2E-9 ’0’0 1 0’0’0 —‘
+ ...Q. : .....
'- 0’... : .0...
O 1E-9 . . . ’0’0’ i’ " 7 1 0’0’0 7
:1.:1:1 9.9.0 : ,,
-:-:-:-: 09 ’9‘0’
0E+0 ""’l"' l l ’ I
CCS1 CCSZ CCS3

 

 

Figure 3-73 Variation of the C1+C2+C3 parameter from equivalent circuit analysis for
specimens with chloride before and after 21 days of rebar corrosion for
measurements from the rebar to external sensors (legend gives the name of the
specimen: C-chloride, C-corrosion, Sn-specimen number)

99

3.4 Effect of electrode size

Figure 374 shows the measured impedance from the CFRP sheet (external sensor)
to the different sized electrodes in the 6” x 2’ x 1’ slab constructed to study the effect of
the electrode size. The different electrode sizes affect only the measured impedance in the
very high frequency region (10,000 - 1000,000 Hz) which are controlled by the kinetics
of the charge transfer processes at the electrode/electrolyte interfaces. A larger effective
electrode area increases the kinetic potential energy. As a result, the measured impedance
decreases in the high frequency range. It should be noted that the measured impedance
from the E3, E4 and E5 sensors to the external sensor yield very similar values over the
entire frequency range. This suggests that increasing the electrode size beyond a certain
limit does not have a significant impact on the measured impedance. This is also reflected
in the measured impedance from the stainless steel bar and the reinforcement cage to the
external sensor for the eight-foot long beam (Figure 3-21). The measured impedance
from the stainless steel bar and the reinforcement cage were very similar to each other

even in the high frequency region.

100

 

—0—E1 +E2

 

+E3 -x-E4

 

 

 

1E+5~ 7 7 7x ,

 

   

Log impedance (ohms)
a
+
.b

_L
m
+
(a)

1

l

l

 

1E+2 +~7 -----w 77 7» 7 ‘1 7- -~ 7177 7~7 7+ 7~ 7~.
1E-1 1E+O 1E+1 1E+2 1E+3 1E+4 1E+5

Log frequency (Hz)

 

 

 

Figure 3-74 Bode magnitude plot of impedance spectra from the slab with different sized
sensors (legend gives the name of the sensor: E1-0.034” dia., E2-0.25” dia., E3-
05” dia., E4-O.59” dia., E5-O.75” dia.)

101

Chapter 4

Summary, Conclusions, and recommendations for
Future Work

4.1 Summary

An experimental study was conducted to assess whether electrochemical impedance
spectroscopy (EIS) could be used to detect the debonding of external CFRP
reinforcement on concrete structures. The EIS-based sensing technology used in this
study was adopted from work done by Davis et al. (1999) on composite/composite bonds.

Wedge tests were conducted with several 6” x 6” x 24” specimens and one 1.5’ x 2’
x 8’ specimen to investigate the detection of CFRP debonding. Sensor elements
(electrodes) consisted of: (a) copper tape with a conductive adhesive applied to the
outside of the CFRP; (b) stainless steel wire placed in grooves on the bottom face of the
concrete prior to bonding the CFRP; and (c) reinforcing bars. Experimental studies were
conducted to also determine the influence of different environmental conditions on
impedance measurements. The environmental conditions included: (1) variation of
humidity levels; (2) variation of temperature levels; (3) chloride content in the concrete
specimen; (4) 300 freeze-thaw cycles; (5) 300 wet-dry cycles; and (6) 21 days of
accelerated corrosion of a reinforcing bar.

Impedance measurements were taken between pairs of sensors using a potentiostat.
The measurements were analyzed by comparing the raw impedance spectra and using
equivalent circuit analysis. The raw impedance measurements for different environmental
conditions and different debonded lengths were compared either over the entire

102

frequency range or at specific frequencies. An equivalent circuit was conceived and fitted
to each measured impedance spectrum and the circuit parameters were estimated. The
fitted circuit parameters were compared for measurements from different environmental
conditions and to different CFRP debond lengths. The raw impedance measurements and
the fitted capacitance parameters (C1+C2+C3) for different debonded lengths are also
fitted to the empirical formula (Equation 3-5)

4.1.1 Detection of CFRP Debonding

Variations in the measured impedance spectra and the equivalent circuit parameters
due to debonding of the CFRP sheet was studied in a controlled environment (controlled
temperature and humidity) and in an ambient condition (fluctuating temperature and
humidity). Impedance measurements taken between the rebar and the extended copper
tape sensors were most effective for assessment of the total debonded area. Impedance
measurements taken between pairs of internal wire sensors were capable of revealing the
location of debonded regions.

CFRP debonding increased the measured impedances over the entire frequency
range when the wedge test was conducted in the controlled environment. However, when
the wedge test was conducted in the ambient condition, the measured impedances from
the low- and mid-frequency range were influenced by the different temperature and
humidity levels. The CFRP debonding effect was most clearly observed with the
impedance at high-frequencies regardless of the environmental condition. The measured
impedance at a given high frequency increases exponentially as the debond length
increases. It is found that different effective electrode areas (size effect) has a significant
influence on the measured impedance. A larger effective electrode area decreases the

magnitude of measured impedance significantly. The debonding of the CFRP was
103

capable of being detected without interference from environmental conditions by
impedances measured in the very high-frequency range from about 10,000 to 100,000
Hz. For the large specimen, even the high frequency impedances were influenced by the
environmental conditions. The increase effective electrode areas for the lorge beam
reduced the impedance magnitudes, and it is postulated that this made the measurements
more sensitive to environmental effects.

The CPE parameter in the equivalent circuit correlated well with debonded area in
the controlled environment but was strongly influenced by variations in temperature and
humidity. Capacitance parameters also correlated well with the debonded area and were
not too sensitive to environmental factors. All capacitance parameters showed a linear
correlation with the debonded area for the wedge test on two-foot long Specimens. The
sum of all capacitances (C1+C2+C3) showed the least sample-to-sample variation. For
the wedge test on the large beam the sum of the capacitances (C1+C2+C3) showed an
exponentially decaying variation with debonded area with high sensitivity for initial
debonding up to about 10% of the total CFRP area.

The raw impedance measurements from the high frequency region and the fitted
capacitance parameters (C1+C2+C3) for different debonded lengths were fitted to an
empirical formula. Although the fit were always excellent, the parameters of the
empirical formula vary significantly from sample to sample and therefore a formula for

diagnosis cannot be readily proposed.

104

4.1.2 Environmental effects

The influence of short-term environmental effects including different humidity
levels, different temperature levels and chloride content in the concrete on the impedance
measurements were studied using small specimens. These environmental factors have a
significant influence on the magnitude of impedance measured in the low- and mid-
frequency range, but have a much smaller influence on the magnitude of impedance in
the high-frequency range. Increased water content due to high humidity lowered the
magnitude of impedance, and reduced water content due to low humidity increased the
magnitude of impedance. High temperature decreased the magnitude of impedance
significantly in the low- and mid-frequency range. The presence of chloride ions in
concrete increased the impedance measurements in the low- and mid-frequency range.

The influence of long-term environmental conditions including freeze-thaw and wet-
dry cycles and corrosion of the reinforcing bar on the measured impedance were also
studied. Impedance measurements were taken before and after small concrete specimens
were subjected to 300 freeze-thaw and wet-dry cycles. Impedance measurements were
also taken before and after one reinforcing bar in each small specimen was subjected to
accelerated corrosion to study the effect of corrosion on the measured impedance,
especially when the rebar was used as a sensing element. These long-term environmental
conditions had a significant influence on the measured impedance in the low-frequency
range and had much less of an influence in the high-frequency range. The magnitude of
impedance in the low-frequency range was reduced after the environmental exposure.

In the equivalent circuit, the R1, C1 and CPE parameters control the impedance in
the low-frequency region and the C2, C3, R2 and R3 parameters control the impedance in

the mid- and high-frequency regions. The different environment conditions had a

105

significant influence on the CPE parameter and had much less of an influence on the
capacitance parameters from the equivalent circuit analysis. High temperature and high
moisture content in concrete decreased the CPE parameter and increase the capacitance
parameters. Low temperature and low moisture content in concrete had the Opposite
effects. Chloride content in concrete decreased the CPE parameter and increased the
capacitance parameters. Freeze-thaw cycles, wet-dry cycles and corrosion of the rebar
increased the CPE and C1 parameters and decreased the C2+C3 values. The sum of all
capacitance parameters (C1+C2+C3) showed the least variation after exposure to the

long-term environmental conditions.

4.2 Conclusions

Impedance measurements taken between steel reinforcement and copper tape sensors
mounted on the outside of external CFRP reinforcement are effective for estimating the
total debonded area of CFRP. Measurements taken in the high-frequency range (1000 —
100,000 Hz) are the most reliable since they are not strongly influenced by environmental
conditions. A more reliable approach is to perform equivalent circuit analysis using the
circuit used in this work, and use the sum of the capacitance parameter (C1+C2+C3) as a
measure of the total debonded area of CFRP. The impedance magnitude at a given high
frequency (say 100,000 Hz), and C1+C2+C3 values correlate well with the debonded
area of CFRP and are not strongly affected by environmental conditions. The impedance
at high frequencies is dominated by the kinetics of the charge transfer processes at the
electrode/electrolyte interfaces, which appear to be influenced strongly by the debonding

of the CFRP.

106

Impedance measurements taken between pairs of internal stainless steel wire sensors
placed in grooves cut into the bottom face of concrete beams before the CFRP is bonded
are effective in locating debonded regions. The magnitude of impedances at low
frequencies taken between a pair of sensors in the debonded region are much higher than
those taken between a pair of sensors in the bonded region. The CPE parameter from the
equivalent circuit analysis is the most effective measure for detecting the location of
debonded regions. The CPE parameter value corresponding to impedance measurements
between sensors in the debonded area is about two orders of magnitude larger than the
value corresponding to impedance measurements between sensors in the bonded area.

Short-term environmental conditions including humidity, temperature, and chloride
content affect the impedance measurements at low frequencies. The short-term
environmental conditions affect the rate of ions diffusion through the concrete, which is
the dominant mechanism for charge transfer at low frequencies. Increased moisture
content, and temperature, and the presence of chloride, increase the charge transfer rate
and hence reduce the impedance at low frequencies.

Long-term environmental conditions including freeze-thaw and wet-dry cycles, and
corrosion of the steel reinforcement, also affect the impedance measurements at low- and
mid-frequencies. The long-term environmental conditions. promote the growth of micro
and macro cracks within the concrete. Repeated expansion and shrinkage due to freeze-
thaw and wet-dry cycles increase micro cracking. The expansion of volume associated
with corrosion of the rebar causes cracking of the concrete. The increase in voids
increases the diffusion rate of ions and thereby reduces the magnitude of impedance at

low frequencies.

107

4.3 Recommendations for Future Work

The EIS-based sensor technology shows strong promise for effective NDE of
concrete structures strengthened with CFRP sheets. However, results from this purely
experimental research indicate that the EIS-based impedance measurements are
influenced significantly by environmental conditions and different effective electrode
areas especially in the low- and mid-frequency range. It is also observed that the
measured impedance can have significant sample-to-sample variation.

A significant contribution to the further enhancement of this work would be the
development of a numerical approach for modeling the charge transfer process through
the specimen. This can be most effectively done at the macro-scale by using a 3-D finite
element model to represent the concrete structure, reinforcement and sensors and
effective electrical properties to represent the current condition through the various
component. The effective electrical properties of concrete would need to be characterized
with respect to temperature, humidity, chloride content, and microcrack width and
density. Once such a model is developed it can be used to study different structureal
components reinforced with CFRP, develop optimal sensor arrangements, and deve10p

nomographs that can be used for diagnosis of CFRP debonding.

108

REFERENCES

Bard, A.J., and Faulkner, LR. (1980). Electrochemical methods: fundamentals and
application. Interscience Publication, New York.

Buyukozturk, 0., and Hearing, B. (1998). “Failure behavior of precracked concrete
beams retrofitting with FRP.” Journal of composites for construction: 138-144.

Dai, ST, and Labuz, J.F. (1998). “Damage and failure analysis of brittle materials by
acoustic emission” Journal of Materials in Civil Engineering, Vol. 9 (4).

Davis, G.D., Dacres, CM. and Krebs, LA. (1999). “In-Situ sensor to detect moisture
intrusion and degradation of coatings, composites, and adhesive bonds.” Proc. T ri-
Services conference on corrosion, Myrtle Beach, South Carolina.

Davis, G.D., Krebs, L.A., Drzal, L.T., Rich, M.J., and Askeland, P. (2000). “Electro-
chemical sensors for nondestructive evaluation of adhesive bonds.” J. Adhes. 72:335-
358.

Gamry Instruments, (1999). Gamry framework system User’s manual revision 3.1.

Gamry Instruments, (April 2003) “Application notes: Equivalent circuit modeling using
the Gamry EIS300 electrochemical impedance spectroscopy software.” Gamry
Instruments, July, 2003. <http://www.gamry.com/G2/Appnotes/Refercnce/
EISModelingEISModelinghtm>

Karbhari, V.M., and Zhao, L. (2000). “Use of composites for 21St century civil
infrastructure.” Computer Method in Applied Mechanics and Engineering 185: 433-
454.

Macdonald, JR. (1987). Impedance spectroscopy: emphasizing solid materials and
systems : Wiley-Interscience Publication, New York.

Mirmiran, A., Shahawy, M., and Echary, J. (1999). “Acoustic emission monitoring of
hybrid FRP-concrete columns.” Journal of Engineering Mechanics, Vol 125 (8).

Mirmiran, A., and Yunmei, W. (2001) “Damage assessment of FRP-encased concrete
using ultrasonic pulse velocity.” Journal of Engineering Mechanics, Vol 127 (2):
126-135.

Nguyen, D.M., Chan, T.K., and Cheong, H.K. (2001). “Brittle failure and bond
development length of CFRP-concrete beams.” Journal of Composites for
Construction. February: 12-17.

Olajide, D., Laurence, J., and Rami M. (2000). “Ultrasonic monitoring of material

degradation in FRP composites.” Journal of Engineering Mechanics Vol. 126 (7):
704-710.

109

Pantazopoulou, S., Bonacci, J, Sheikh, S., Heam, N., and Thomas, M. (2001). “Repair of
corrosion-damaged columns with FRP wraps.” Journal of composites for
construction. Vol. 5 (1): 3-11.

Philips, J. (1992). “The effect of corrosion on the structural performance of new and
repaired one-way slabs.” PhD thesis, Dept. of Civ. Engineering, University of
Toronto.

Rahimi, H., and Hutchinson, A. (2001). “Concrete beams strengthened with externally
bonded FRP plates.” Journal of Composites for Construction. February : 44-56.

Scully, J ., and Kendig, M. (1989). “Basic aspects of the application of electrochemical
impedance for the life prediction of organic coatings on metals.” Proc. Corrosion 89
Paper 32, NACE.

110

Appendix—A: Data Collection and Equivalent
Circuit Analysis Procedure

This section provides a step-by-step description of the procedures used to measure
impedances and perform equivalent circuit analysis. Additional information is provided

in the User’s Manual for the Gamry Framework software.

A-l Software Procedure for Data Collection

The procedure described below is based on the Gamry Framework software version
3.20.

1. Connect the blue, green and white cables on the potentiostat to one sensor and the
red and orange cables to the second sensor.

2. Run the Gamry Framework software.

3. From the Gamry Framework menu bar, select Experiment, click on A
Electrochemical Impedance, and click on Potentiostatic EIS to launch the
Potentiostatic EIS window.

4. Necessary information can be filled out on the Potentiostatic EIS window or
retrieved from a saved template setup file.

5. To use a the saved template setup file, click on the Restore button located at the
top of the Potentiostatic EIS window, to display the Restore Setup window.

6. From the Restore Setup window, select a proper template setup file and select a
stored setup (each template setup file can contain several different setting), and
then click on the Restore button (this brings back to the Potentiostatic window).

7. Specify the output file name and click the OK button.

111

A-2 Equivalent circuit analysis procedure

The procedure described below is based on The Gamry Echem Analyst application

version 1.02.

1. Run the Gamry Echem Analyst application

2. Open the measured EIS data file by clicking File on the menu bar, clicking Open
File, then selecting the EIS file.

3. Click Impedance on the menu bar and then click Fit A Model. This brings up the
Select Model File window. Since the parameters of an electrical circuit which has an
theoretical impedance similar to that of the measured impedance are estimated,
selecting a proper equivalent circuit is very important.

4. Once the proper equivalent circuit model is selected, the Impedance Fitting window
is displayed. The nonlinear least squares fitting procedure requires initial values to be
specified. The initial values for each equivalent circuit are stored with the circuit. To
check whether the initial values are reasonable for the measured data, click Preview
in the Impedance Fitting window.

4.1. If the inipedance generated by the equivalent circuit does not match the measured
impedance, then the initial value should be changed to obtaina better match. The
initial values for the equivalent circuit element parameters used in this research
are listed in Table A-1. Figure A-l shows the fitted impedance using poor initial
values.

4.2. The C1, R1 and CPE parameters controls the fitted impedance in the low-
frequency range and the C2, R2, C3 and R3 parameters controls the fitted

impedance in the mid- and high-frequency ranges. Use the Preview function and

112

adjust the initial value of the parameters. If the initial values are within a factor
of 100 and the shapes of the impedance spectra generated by the initial values are
similar to the measured spectra then the equivalent circuit model will usually
yield a good fit.

4.3. Once reasonable initial values are obtained, the fitting program will converge on
parameters that yield a theoretical impedance close to the measured impedance.
Figure A-2 shows the fitted impedance with the initial values of the parameters
and the fitted impedance from the final converged values of the parameters. The
fitting program uses a Levenberg-Marquardt non-linear least square fitting
algorithm.

5. To finalize the fitting process, click the Calculate button in the Impedance Fitting

window.

Table A- 1. Initial values of equivalent circuit element parameters

 

 

 

 

 

 

 

 

 

 

 

Parameter Rebar to External Internal to Internal Internal to External
Low High Low High Low High
R1 1.0E+03 1.0E+07 1.0E+08 1.0E+08 1.0E+05 1.0E+08
C1 1.0E-07 1.0E-08 1.0E-08 1.0E-10 . 1.0E—08 1.0E-10
R2 1.0E+01 1.0E+03 1.0E+04 1.0E+04 1.0E+04 1.0E+04
R3 1.0E+01 1.0E+04 1.0E+06 1.0E+06 1.0E+06 1.0E+06
C3 1.0E-07 1.0E-08 1.0E-08 1.0E-10 1.0E-08 1.0E-10
CPEl 1.0E+06 1.0E+07 1.0E+08 1.0E+08 1.0E+05 1.0E+08
Alpha 0.6 0.6 0.6 0.1 0.1 0.1
C2 1.0E-07 1.0E-08 1.0E-08 1.0E-10 1.0E-08 1.0E-10

 

 

 

 

 

 

 

113

 

 

 

158% .0.0
157
- -20
1E6 _
A -40 r
g 155; . 5‘
9, i - E
"3 154, ' ’3
E i; :‘6003
N . 8
153 a _ v
.-80
152, .
151' .... .. _. . . -100
15-2 15-1 150 151 152 153 154 155 1E6
Freq (Hz)

 

 

 

Figure A- 1 Measured impedance from rebar to external sensors and impedance of the
equivalent circuit using poor initial values for parameters: Dots are the measured
values, dotted line is the fitted impedance using initial values, and solid line is the
fitted impedance with the final values

 

 

 

 

IE7 “ ’ ' ’ " ' "M c """ 00

1E6 - -20

155: 40 '
E ' a
.2 5'
a 1541 . EL
'6 ' O.
E «600:8:
N 153‘ ’ 3

152 3 '80

15-2 15-1 150 151 152 153 154 155 1E6
Freq (Hz)

 

 

Figure A- 2 Measured impedance from rebar to external sensors and impedance of the
equivalent circuit using good initial values for parameters: Dots are the measured
values, dotted line is the fitted impedance using initial values, and solid line is the
fitted impedance with the final value

114

llllllllllll1Wlsllvllllllllsll I
l 8 1298 62487 88421